{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0505262732580456,
  "eval_steps": 1000,
  "global_step": 13000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 8.080971332754196e-05,
      "grad_norm": 6.690799236297607,
      "learning_rate": 1.3458950201884254e-08,
      "loss": 1.248,
      "step": 1
    },
    {
      "epoch": 0.00016161942665508393,
      "grad_norm": 5.96121883392334,
      "learning_rate": 2.6917900403768507e-08,
      "loss": 1.1711,
      "step": 2
    },
    {
      "epoch": 0.00024242913998262592,
      "grad_norm": 7.078202247619629,
      "learning_rate": 4.037685060565276e-08,
      "loss": 1.3206,
      "step": 3
    },
    {
      "epoch": 0.00032323885331016786,
      "grad_norm": 13.463970184326172,
      "learning_rate": 5.3835800807537014e-08,
      "loss": 1.1678,
      "step": 4
    },
    {
      "epoch": 0.00040404856663770985,
      "grad_norm": 8.267407417297363,
      "learning_rate": 6.729475100942127e-08,
      "loss": 1.2331,
      "step": 5
    },
    {
      "epoch": 0.00048485827996525184,
      "grad_norm": 7.39365816116333,
      "learning_rate": 8.075370121130552e-08,
      "loss": 1.3377,
      "step": 6
    },
    {
      "epoch": 0.0005656679932927938,
      "grad_norm": 7.1661763191223145,
      "learning_rate": 9.421265141318978e-08,
      "loss": 1.3193,
      "step": 7
    },
    {
      "epoch": 0.0006464777066203357,
      "grad_norm": 15.627093315124512,
      "learning_rate": 1.0767160161507403e-07,
      "loss": 1.3257,
      "step": 8
    },
    {
      "epoch": 0.0007272874199478778,
      "grad_norm": 8.536316871643066,
      "learning_rate": 1.211305518169583e-07,
      "loss": 1.2181,
      "step": 9
    },
    {
      "epoch": 0.0008080971332754197,
      "grad_norm": 11.300825119018555,
      "learning_rate": 1.3458950201884255e-07,
      "loss": 1.3117,
      "step": 10
    },
    {
      "epoch": 0.0008889068466029616,
      "grad_norm": 13.28189754486084,
      "learning_rate": 1.480484522207268e-07,
      "loss": 1.2713,
      "step": 11
    },
    {
      "epoch": 0.0009697165599305037,
      "grad_norm": 10.544150352478027,
      "learning_rate": 1.6150740242261104e-07,
      "loss": 1.1934,
      "step": 12
    },
    {
      "epoch": 0.0010505262732580457,
      "grad_norm": 16.782533645629883,
      "learning_rate": 1.7496635262449528e-07,
      "loss": 1.3952,
      "step": 13
    },
    {
      "epoch": 0.0011313359865855876,
      "grad_norm": 6.385349750518799,
      "learning_rate": 1.8842530282637956e-07,
      "loss": 1.2814,
      "step": 14
    },
    {
      "epoch": 0.0012121456999131296,
      "grad_norm": 12.990242958068848,
      "learning_rate": 2.018842530282638e-07,
      "loss": 1.3054,
      "step": 15
    },
    {
      "epoch": 0.0012929554132406714,
      "grad_norm": 9.732259750366211,
      "learning_rate": 2.1534320323014806e-07,
      "loss": 1.2828,
      "step": 16
    },
    {
      "epoch": 0.0013737651265682135,
      "grad_norm": 21.376018524169922,
      "learning_rate": 2.2880215343203232e-07,
      "loss": 1.2997,
      "step": 17
    },
    {
      "epoch": 0.0014545748398957555,
      "grad_norm": 7.185873031616211,
      "learning_rate": 2.422611036339166e-07,
      "loss": 1.2679,
      "step": 18
    },
    {
      "epoch": 0.0015353845532232973,
      "grad_norm": 5.8356242179870605,
      "learning_rate": 2.557200538358008e-07,
      "loss": 1.2401,
      "step": 19
    },
    {
      "epoch": 0.0016161942665508394,
      "grad_norm": 6.466715335845947,
      "learning_rate": 2.691790040376851e-07,
      "loss": 1.1889,
      "step": 20
    },
    {
      "epoch": 0.0016970039798783814,
      "grad_norm": 6.673277378082275,
      "learning_rate": 2.8263795423956933e-07,
      "loss": 1.2365,
      "step": 21
    },
    {
      "epoch": 0.0017778136932059233,
      "grad_norm": 12.293954849243164,
      "learning_rate": 2.960969044414536e-07,
      "loss": 1.1535,
      "step": 22
    },
    {
      "epoch": 0.0018586234065334653,
      "grad_norm": 6.82350492477417,
      "learning_rate": 3.0955585464333785e-07,
      "loss": 1.3097,
      "step": 23
    },
    {
      "epoch": 0.0019394331198610074,
      "grad_norm": 8.443634986877441,
      "learning_rate": 3.230148048452221e-07,
      "loss": 1.1801,
      "step": 24
    },
    {
      "epoch": 0.002020242833188549,
      "grad_norm": 7.782437324523926,
      "learning_rate": 3.3647375504710637e-07,
      "loss": 1.1593,
      "step": 25
    },
    {
      "epoch": 0.0021010525465160915,
      "grad_norm": 11.640141487121582,
      "learning_rate": 3.4993270524899055e-07,
      "loss": 1.1689,
      "step": 26
    },
    {
      "epoch": 0.0021818622598436333,
      "grad_norm": 6.59988260269165,
      "learning_rate": 3.6339165545087484e-07,
      "loss": 1.2397,
      "step": 27
    },
    {
      "epoch": 0.002262671973171175,
      "grad_norm": 5.805958271026611,
      "learning_rate": 3.768506056527591e-07,
      "loss": 1.2492,
      "step": 28
    },
    {
      "epoch": 0.002343481686498717,
      "grad_norm": 6.94270658493042,
      "learning_rate": 3.903095558546434e-07,
      "loss": 1.1362,
      "step": 29
    },
    {
      "epoch": 0.002424291399826259,
      "grad_norm": 6.733250141143799,
      "learning_rate": 4.037685060565276e-07,
      "loss": 1.2791,
      "step": 30
    },
    {
      "epoch": 0.002505101113153801,
      "grad_norm": 6.11829948425293,
      "learning_rate": 4.172274562584119e-07,
      "loss": 1.2586,
      "step": 31
    },
    {
      "epoch": 0.002585910826481343,
      "grad_norm": 10.662686347961426,
      "learning_rate": 4.306864064602961e-07,
      "loss": 1.31,
      "step": 32
    },
    {
      "epoch": 0.002666720539808885,
      "grad_norm": 6.052074909210205,
      "learning_rate": 4.441453566621804e-07,
      "loss": 1.3037,
      "step": 33
    },
    {
      "epoch": 0.002747530253136427,
      "grad_norm": 5.352566242218018,
      "learning_rate": 4.5760430686406463e-07,
      "loss": 1.2243,
      "step": 34
    },
    {
      "epoch": 0.0028283399664639688,
      "grad_norm": 5.09272575378418,
      "learning_rate": 4.7106325706594887e-07,
      "loss": 1.1706,
      "step": 35
    },
    {
      "epoch": 0.002909149679791511,
      "grad_norm": 4.496048450469971,
      "learning_rate": 4.845222072678332e-07,
      "loss": 1.2023,
      "step": 36
    },
    {
      "epoch": 0.002989959393119053,
      "grad_norm": 4.400938034057617,
      "learning_rate": 4.979811574697174e-07,
      "loss": 1.245,
      "step": 37
    },
    {
      "epoch": 0.0030707691064465947,
      "grad_norm": 5.95123291015625,
      "learning_rate": 5.114401076716016e-07,
      "loss": 1.1342,
      "step": 38
    },
    {
      "epoch": 0.003151578819774137,
      "grad_norm": 5.758495330810547,
      "learning_rate": 5.248990578734859e-07,
      "loss": 1.2793,
      "step": 39
    },
    {
      "epoch": 0.003232388533101679,
      "grad_norm": 4.290956020355225,
      "learning_rate": 5.383580080753702e-07,
      "loss": 1.2222,
      "step": 40
    },
    {
      "epoch": 0.0033131982464292206,
      "grad_norm": 5.054476261138916,
      "learning_rate": 5.518169582772545e-07,
      "loss": 1.2194,
      "step": 41
    },
    {
      "epoch": 0.003394007959756763,
      "grad_norm": 4.21449613571167,
      "learning_rate": 5.652759084791387e-07,
      "loss": 1.168,
      "step": 42
    },
    {
      "epoch": 0.0034748176730843047,
      "grad_norm": 4.471845626831055,
      "learning_rate": 5.78734858681023e-07,
      "loss": 1.139,
      "step": 43
    },
    {
      "epoch": 0.0035556273864118465,
      "grad_norm": 4.471460342407227,
      "learning_rate": 5.921938088829072e-07,
      "loss": 1.1644,
      "step": 44
    },
    {
      "epoch": 0.003636437099739389,
      "grad_norm": 5.968162536621094,
      "learning_rate": 6.056527590847914e-07,
      "loss": 1.2744,
      "step": 45
    },
    {
      "epoch": 0.0037172468130669306,
      "grad_norm": 4.4738664627075195,
      "learning_rate": 6.191117092866757e-07,
      "loss": 1.2473,
      "step": 46
    },
    {
      "epoch": 0.0037980565263944725,
      "grad_norm": 4.90024471282959,
      "learning_rate": 6.3257065948856e-07,
      "loss": 1.1564,
      "step": 47
    },
    {
      "epoch": 0.0038788662397220147,
      "grad_norm": 3.56142520904541,
      "learning_rate": 6.460296096904442e-07,
      "loss": 1.2167,
      "step": 48
    },
    {
      "epoch": 0.003959675953049557,
      "grad_norm": 4.147864818572998,
      "learning_rate": 6.594885598923285e-07,
      "loss": 1.2479,
      "step": 49
    },
    {
      "epoch": 0.004040485666377098,
      "grad_norm": 4.415273189544678,
      "learning_rate": 6.729475100942127e-07,
      "loss": 1.2028,
      "step": 50
    },
    {
      "epoch": 0.004121295379704641,
      "grad_norm": 3.895045757293701,
      "learning_rate": 6.864064602960969e-07,
      "loss": 1.061,
      "step": 51
    },
    {
      "epoch": 0.004202105093032183,
      "grad_norm": 3.7078940868377686,
      "learning_rate": 6.998654104979811e-07,
      "loss": 1.0716,
      "step": 52
    },
    {
      "epoch": 0.004282914806359724,
      "grad_norm": 3.556968927383423,
      "learning_rate": 7.133243606998655e-07,
      "loss": 1.2259,
      "step": 53
    },
    {
      "epoch": 0.0043637245196872666,
      "grad_norm": 4.950558185577393,
      "learning_rate": 7.267833109017497e-07,
      "loss": 1.1629,
      "step": 54
    },
    {
      "epoch": 0.004444534233014808,
      "grad_norm": 4.822227478027344,
      "learning_rate": 7.402422611036341e-07,
      "loss": 1.1893,
      "step": 55
    },
    {
      "epoch": 0.00452534394634235,
      "grad_norm": 4.09580135345459,
      "learning_rate": 7.537012113055183e-07,
      "loss": 1.1445,
      "step": 56
    },
    {
      "epoch": 0.0046061536596698925,
      "grad_norm": 3.3497204780578613,
      "learning_rate": 7.671601615074024e-07,
      "loss": 1.1026,
      "step": 57
    },
    {
      "epoch": 0.004686963372997434,
      "grad_norm": 5.070769309997559,
      "learning_rate": 7.806191117092868e-07,
      "loss": 1.0559,
      "step": 58
    },
    {
      "epoch": 0.004767773086324976,
      "grad_norm": 3.769427537918091,
      "learning_rate": 7.94078061911171e-07,
      "loss": 1.1608,
      "step": 59
    },
    {
      "epoch": 0.004848582799652518,
      "grad_norm": 3.896399736404419,
      "learning_rate": 8.075370121130552e-07,
      "loss": 1.2354,
      "step": 60
    },
    {
      "epoch": 0.00492939251298006,
      "grad_norm": 4.363215446472168,
      "learning_rate": 8.209959623149396e-07,
      "loss": 1.2881,
      "step": 61
    },
    {
      "epoch": 0.005010202226307602,
      "grad_norm": 3.7588884830474854,
      "learning_rate": 8.344549125168238e-07,
      "loss": 1.0485,
      "step": 62
    },
    {
      "epoch": 0.005091011939635144,
      "grad_norm": 4.2055182456970215,
      "learning_rate": 8.47913862718708e-07,
      "loss": 1.0611,
      "step": 63
    },
    {
      "epoch": 0.005171821652962686,
      "grad_norm": 3.6553685665130615,
      "learning_rate": 8.613728129205922e-07,
      "loss": 1.0145,
      "step": 64
    },
    {
      "epoch": 0.005252631366290228,
      "grad_norm": 4.0810699462890625,
      "learning_rate": 8.748317631224765e-07,
      "loss": 1.1216,
      "step": 65
    },
    {
      "epoch": 0.00533344107961777,
      "grad_norm": 3.6002376079559326,
      "learning_rate": 8.882907133243608e-07,
      "loss": 1.1444,
      "step": 66
    },
    {
      "epoch": 0.005414250792945312,
      "grad_norm": 3.773752212524414,
      "learning_rate": 9.01749663526245e-07,
      "loss": 1.1017,
      "step": 67
    },
    {
      "epoch": 0.005495060506272854,
      "grad_norm": 3.276829719543457,
      "learning_rate": 9.152086137281293e-07,
      "loss": 1.1463,
      "step": 68
    },
    {
      "epoch": 0.005575870219600396,
      "grad_norm": 3.6140477657318115,
      "learning_rate": 9.286675639300136e-07,
      "loss": 1.1648,
      "step": 69
    },
    {
      "epoch": 0.0056566799329279376,
      "grad_norm": 4.192599296569824,
      "learning_rate": 9.421265141318977e-07,
      "loss": 1.1997,
      "step": 70
    },
    {
      "epoch": 0.00573748964625548,
      "grad_norm": 3.8302218914031982,
      "learning_rate": 9.55585464333782e-07,
      "loss": 1.0759,
      "step": 71
    },
    {
      "epoch": 0.005818299359583022,
      "grad_norm": 3.8942480087280273,
      "learning_rate": 9.690444145356663e-07,
      "loss": 1.0674,
      "step": 72
    },
    {
      "epoch": 0.0058991090729105635,
      "grad_norm": 3.1199991703033447,
      "learning_rate": 9.825033647375506e-07,
      "loss": 1.1144,
      "step": 73
    },
    {
      "epoch": 0.005979918786238106,
      "grad_norm": 3.5112509727478027,
      "learning_rate": 9.959623149394349e-07,
      "loss": 1.1946,
      "step": 74
    },
    {
      "epoch": 0.006060728499565648,
      "grad_norm": 3.831355333328247,
      "learning_rate": 1.009421265141319e-06,
      "loss": 1.0892,
      "step": 75
    },
    {
      "epoch": 0.006141538212893189,
      "grad_norm": 3.520054340362549,
      "learning_rate": 1.0228802153432032e-06,
      "loss": 1.1438,
      "step": 76
    },
    {
      "epoch": 0.006222347926220732,
      "grad_norm": 3.0068299770355225,
      "learning_rate": 1.0363391655450875e-06,
      "loss": 1.0978,
      "step": 77
    },
    {
      "epoch": 0.006303157639548274,
      "grad_norm": 3.718480110168457,
      "learning_rate": 1.0497981157469718e-06,
      "loss": 1.0832,
      "step": 78
    },
    {
      "epoch": 0.006383967352875815,
      "grad_norm": 3.4276466369628906,
      "learning_rate": 1.063257065948856e-06,
      "loss": 1.243,
      "step": 79
    },
    {
      "epoch": 0.006464777066203358,
      "grad_norm": 4.088370323181152,
      "learning_rate": 1.0767160161507404e-06,
      "loss": 1.1034,
      "step": 80
    },
    {
      "epoch": 0.0065455867795309,
      "grad_norm": 4.904850482940674,
      "learning_rate": 1.0901749663526245e-06,
      "loss": 1.1799,
      "step": 81
    },
    {
      "epoch": 0.006626396492858441,
      "grad_norm": 3.5775904655456543,
      "learning_rate": 1.103633916554509e-06,
      "loss": 0.9865,
      "step": 82
    },
    {
      "epoch": 0.0067072062061859835,
      "grad_norm": 3.5812389850616455,
      "learning_rate": 1.117092866756393e-06,
      "loss": 1.0355,
      "step": 83
    },
    {
      "epoch": 0.006788015919513526,
      "grad_norm": 3.087735891342163,
      "learning_rate": 1.1305518169582773e-06,
      "loss": 1.0554,
      "step": 84
    },
    {
      "epoch": 0.006868825632841067,
      "grad_norm": 3.641606092453003,
      "learning_rate": 1.1440107671601616e-06,
      "loss": 1.199,
      "step": 85
    },
    {
      "epoch": 0.006949635346168609,
      "grad_norm": 3.337639808654785,
      "learning_rate": 1.157469717362046e-06,
      "loss": 0.9961,
      "step": 86
    },
    {
      "epoch": 0.007030445059496152,
      "grad_norm": 3.4842021465301514,
      "learning_rate": 1.1709286675639302e-06,
      "loss": 1.0043,
      "step": 87
    },
    {
      "epoch": 0.007111254772823693,
      "grad_norm": 4.307300090789795,
      "learning_rate": 1.1843876177658145e-06,
      "loss": 1.1515,
      "step": 88
    },
    {
      "epoch": 0.007192064486151235,
      "grad_norm": 3.455139636993408,
      "learning_rate": 1.1978465679676985e-06,
      "loss": 1.14,
      "step": 89
    },
    {
      "epoch": 0.007272874199478778,
      "grad_norm": 3.4302618503570557,
      "learning_rate": 1.2113055181695828e-06,
      "loss": 1.0239,
      "step": 90
    },
    {
      "epoch": 0.007353683912806319,
      "grad_norm": 4.0325188636779785,
      "learning_rate": 1.2247644683714671e-06,
      "loss": 1.1421,
      "step": 91
    },
    {
      "epoch": 0.007434493626133861,
      "grad_norm": 3.5034971237182617,
      "learning_rate": 1.2382234185733514e-06,
      "loss": 1.036,
      "step": 92
    },
    {
      "epoch": 0.0075153033394614035,
      "grad_norm": 3.5903069972991943,
      "learning_rate": 1.2516823687752355e-06,
      "loss": 0.9697,
      "step": 93
    },
    {
      "epoch": 0.007596113052788945,
      "grad_norm": 3.2864511013031006,
      "learning_rate": 1.26514131897712e-06,
      "loss": 1.1045,
      "step": 94
    },
    {
      "epoch": 0.007676922766116487,
      "grad_norm": 3.3390679359436035,
      "learning_rate": 1.2786002691790043e-06,
      "loss": 1.1308,
      "step": 95
    },
    {
      "epoch": 0.0077577324794440294,
      "grad_norm": 3.5311758518218994,
      "learning_rate": 1.2920592193808883e-06,
      "loss": 1.1261,
      "step": 96
    },
    {
      "epoch": 0.007838542192771572,
      "grad_norm": 3.247596025466919,
      "learning_rate": 1.3055181695827726e-06,
      "loss": 1.1322,
      "step": 97
    },
    {
      "epoch": 0.007919351906099114,
      "grad_norm": 4.039790630340576,
      "learning_rate": 1.318977119784657e-06,
      "loss": 1.0643,
      "step": 98
    },
    {
      "epoch": 0.008000161619426654,
      "grad_norm": 3.3263556957244873,
      "learning_rate": 1.3324360699865412e-06,
      "loss": 1.2159,
      "step": 99
    },
    {
      "epoch": 0.008080971332754197,
      "grad_norm": 2.9595468044281006,
      "learning_rate": 1.3458950201884255e-06,
      "loss": 1.0827,
      "step": 100
    },
    {
      "epoch": 0.008161781046081739,
      "grad_norm": 3.3838250637054443,
      "learning_rate": 1.3593539703903098e-06,
      "loss": 1.1956,
      "step": 101
    },
    {
      "epoch": 0.008242590759409281,
      "grad_norm": 3.3292553424835205,
      "learning_rate": 1.3728129205921938e-06,
      "loss": 1.1723,
      "step": 102
    },
    {
      "epoch": 0.008323400472736824,
      "grad_norm": 3.0163986682891846,
      "learning_rate": 1.3862718707940781e-06,
      "loss": 1.1573,
      "step": 103
    },
    {
      "epoch": 0.008404210186064366,
      "grad_norm": 3.532806158065796,
      "learning_rate": 1.3997308209959622e-06,
      "loss": 1.1104,
      "step": 104
    },
    {
      "epoch": 0.008485019899391906,
      "grad_norm": 3.480621337890625,
      "learning_rate": 1.4131897711978467e-06,
      "loss": 1.0676,
      "step": 105
    },
    {
      "epoch": 0.008565829612719449,
      "grad_norm": 3.874357223510742,
      "learning_rate": 1.426648721399731e-06,
      "loss": 1.1004,
      "step": 106
    },
    {
      "epoch": 0.00864663932604699,
      "grad_norm": 3.192427635192871,
      "learning_rate": 1.4401076716016153e-06,
      "loss": 1.1254,
      "step": 107
    },
    {
      "epoch": 0.008727449039374533,
      "grad_norm": 4.162766933441162,
      "learning_rate": 1.4535666218034994e-06,
      "loss": 0.9671,
      "step": 108
    },
    {
      "epoch": 0.008808258752702075,
      "grad_norm": 3.8476390838623047,
      "learning_rate": 1.4670255720053836e-06,
      "loss": 0.9608,
      "step": 109
    },
    {
      "epoch": 0.008889068466029616,
      "grad_norm": 2.9822797775268555,
      "learning_rate": 1.4804845222072681e-06,
      "loss": 1.2083,
      "step": 110
    },
    {
      "epoch": 0.008969878179357158,
      "grad_norm": 3.5006701946258545,
      "learning_rate": 1.4939434724091522e-06,
      "loss": 1.1788,
      "step": 111
    },
    {
      "epoch": 0.0090506878926847,
      "grad_norm": 3.982487678527832,
      "learning_rate": 1.5074024226110365e-06,
      "loss": 1.1051,
      "step": 112
    },
    {
      "epoch": 0.009131497606012243,
      "grad_norm": 3.3723080158233643,
      "learning_rate": 1.5208613728129206e-06,
      "loss": 1.0072,
      "step": 113
    },
    {
      "epoch": 0.009212307319339785,
      "grad_norm": 4.094737529754639,
      "learning_rate": 1.5343203230148049e-06,
      "loss": 1.1149,
      "step": 114
    },
    {
      "epoch": 0.009293117032667327,
      "grad_norm": 3.6751790046691895,
      "learning_rate": 1.5477792732166894e-06,
      "loss": 1.114,
      "step": 115
    },
    {
      "epoch": 0.009373926745994868,
      "grad_norm": 3.659663677215576,
      "learning_rate": 1.5612382234185736e-06,
      "loss": 1.1847,
      "step": 116
    },
    {
      "epoch": 0.00945473645932241,
      "grad_norm": 4.006715297698975,
      "learning_rate": 1.5746971736204577e-06,
      "loss": 1.2139,
      "step": 117
    },
    {
      "epoch": 0.009535546172649952,
      "grad_norm": 4.317342281341553,
      "learning_rate": 1.588156123822342e-06,
      "loss": 1.1472,
      "step": 118
    },
    {
      "epoch": 0.009616355885977495,
      "grad_norm": 3.78163480758667,
      "learning_rate": 1.601615074024226e-06,
      "loss": 1.1659,
      "step": 119
    },
    {
      "epoch": 0.009697165599305037,
      "grad_norm": 3.253537178039551,
      "learning_rate": 1.6150740242261104e-06,
      "loss": 1.1435,
      "step": 120
    },
    {
      "epoch": 0.009777975312632579,
      "grad_norm": 3.281273603439331,
      "learning_rate": 1.6285329744279949e-06,
      "loss": 1.148,
      "step": 121
    },
    {
      "epoch": 0.00985878502596012,
      "grad_norm": 3.442288875579834,
      "learning_rate": 1.6419919246298792e-06,
      "loss": 1.2081,
      "step": 122
    },
    {
      "epoch": 0.009939594739287662,
      "grad_norm": 3.544372797012329,
      "learning_rate": 1.6554508748317632e-06,
      "loss": 1.1631,
      "step": 123
    },
    {
      "epoch": 0.010020404452615204,
      "grad_norm": 3.7572271823883057,
      "learning_rate": 1.6689098250336475e-06,
      "loss": 1.0829,
      "step": 124
    },
    {
      "epoch": 0.010101214165942746,
      "grad_norm": 3.042733907699585,
      "learning_rate": 1.6823687752355316e-06,
      "loss": 1.1737,
      "step": 125
    },
    {
      "epoch": 0.010182023879270289,
      "grad_norm": 2.8813230991363525,
      "learning_rate": 1.695827725437416e-06,
      "loss": 1.0451,
      "step": 126
    },
    {
      "epoch": 0.010262833592597831,
      "grad_norm": 3.256988525390625,
      "learning_rate": 1.7092866756393004e-06,
      "loss": 1.1688,
      "step": 127
    },
    {
      "epoch": 0.010343643305925371,
      "grad_norm": 3.1898910999298096,
      "learning_rate": 1.7227456258411845e-06,
      "loss": 1.08,
      "step": 128
    },
    {
      "epoch": 0.010424453019252914,
      "grad_norm": 3.8104963302612305,
      "learning_rate": 1.7362045760430687e-06,
      "loss": 1.0908,
      "step": 129
    },
    {
      "epoch": 0.010505262732580456,
      "grad_norm": 3.439345121383667,
      "learning_rate": 1.749663526244953e-06,
      "loss": 1.0263,
      "step": 130
    },
    {
      "epoch": 0.010586072445907998,
      "grad_norm": 4.058794021606445,
      "learning_rate": 1.7631224764468375e-06,
      "loss": 1.0531,
      "step": 131
    },
    {
      "epoch": 0.01066688215923554,
      "grad_norm": 2.9343535900115967,
      "learning_rate": 1.7765814266487216e-06,
      "loss": 1.0031,
      "step": 132
    },
    {
      "epoch": 0.010747691872563083,
      "grad_norm": 3.176175832748413,
      "learning_rate": 1.7900403768506059e-06,
      "loss": 1.1362,
      "step": 133
    },
    {
      "epoch": 0.010828501585890623,
      "grad_norm": 3.4748244285583496,
      "learning_rate": 1.80349932705249e-06,
      "loss": 1.0497,
      "step": 134
    },
    {
      "epoch": 0.010909311299218166,
      "grad_norm": 3.0900204181671143,
      "learning_rate": 1.8169582772543742e-06,
      "loss": 1.099,
      "step": 135
    },
    {
      "epoch": 0.010990121012545708,
      "grad_norm": 3.0723509788513184,
      "learning_rate": 1.8304172274562585e-06,
      "loss": 1.0654,
      "step": 136
    },
    {
      "epoch": 0.01107093072587325,
      "grad_norm": 2.9936509132385254,
      "learning_rate": 1.8438761776581428e-06,
      "loss": 1.0557,
      "step": 137
    },
    {
      "epoch": 0.011151740439200792,
      "grad_norm": 2.9151737689971924,
      "learning_rate": 1.8573351278600271e-06,
      "loss": 1.1121,
      "step": 138
    },
    {
      "epoch": 0.011232550152528335,
      "grad_norm": 3.0358784198760986,
      "learning_rate": 1.8707940780619114e-06,
      "loss": 1.1448,
      "step": 139
    },
    {
      "epoch": 0.011313359865855875,
      "grad_norm": 3.3080246448516846,
      "learning_rate": 1.8842530282637955e-06,
      "loss": 1.1314,
      "step": 140
    },
    {
      "epoch": 0.011394169579183417,
      "grad_norm": 3.5009357929229736,
      "learning_rate": 1.8977119784656798e-06,
      "loss": 0.9784,
      "step": 141
    },
    {
      "epoch": 0.01147497929251096,
      "grad_norm": 3.626203775405884,
      "learning_rate": 1.911170928667564e-06,
      "loss": 1.0753,
      "step": 142
    },
    {
      "epoch": 0.011555789005838502,
      "grad_norm": 3.3503258228302,
      "learning_rate": 1.9246298788694483e-06,
      "loss": 1.1486,
      "step": 143
    },
    {
      "epoch": 0.011636598719166044,
      "grad_norm": 3.2445223331451416,
      "learning_rate": 1.9380888290713326e-06,
      "loss": 1.1295,
      "step": 144
    },
    {
      "epoch": 0.011717408432493586,
      "grad_norm": 3.2183837890625,
      "learning_rate": 1.951547779273217e-06,
      "loss": 1.0644,
      "step": 145
    },
    {
      "epoch": 0.011798218145821127,
      "grad_norm": 3.118913173675537,
      "learning_rate": 1.965006729475101e-06,
      "loss": 1.2114,
      "step": 146
    },
    {
      "epoch": 0.01187902785914867,
      "grad_norm": 3.2243826389312744,
      "learning_rate": 1.9784656796769855e-06,
      "loss": 1.205,
      "step": 147
    },
    {
      "epoch": 0.011959837572476211,
      "grad_norm": 2.848964214324951,
      "learning_rate": 1.9919246298788698e-06,
      "loss": 0.9954,
      "step": 148
    },
    {
      "epoch": 0.012040647285803754,
      "grad_norm": 3.7954328060150146,
      "learning_rate": 2.005383580080754e-06,
      "loss": 1.0534,
      "step": 149
    },
    {
      "epoch": 0.012121456999131296,
      "grad_norm": 2.8739662170410156,
      "learning_rate": 2.018842530282638e-06,
      "loss": 1.2154,
      "step": 150
    },
    {
      "epoch": 0.012202266712458838,
      "grad_norm": 3.972341299057007,
      "learning_rate": 2.032301480484522e-06,
      "loss": 1.1231,
      "step": 151
    },
    {
      "epoch": 0.012283076425786379,
      "grad_norm": 3.7189602851867676,
      "learning_rate": 2.0457604306864065e-06,
      "loss": 1.1857,
      "step": 152
    },
    {
      "epoch": 0.012363886139113921,
      "grad_norm": 3.0613625049591064,
      "learning_rate": 2.059219380888291e-06,
      "loss": 1.0634,
      "step": 153
    },
    {
      "epoch": 0.012444695852441463,
      "grad_norm": 3.0153865814208984,
      "learning_rate": 2.072678331090175e-06,
      "loss": 1.0734,
      "step": 154
    },
    {
      "epoch": 0.012525505565769006,
      "grad_norm": 3.1202192306518555,
      "learning_rate": 2.0861372812920593e-06,
      "loss": 1.0793,
      "step": 155
    },
    {
      "epoch": 0.012606315279096548,
      "grad_norm": 2.8822708129882812,
      "learning_rate": 2.0995962314939436e-06,
      "loss": 1.035,
      "step": 156
    },
    {
      "epoch": 0.01268712499242409,
      "grad_norm": 3.4996628761291504,
      "learning_rate": 2.113055181695828e-06,
      "loss": 1.1336,
      "step": 157
    },
    {
      "epoch": 0.01276793470575163,
      "grad_norm": 3.41062331199646,
      "learning_rate": 2.126514131897712e-06,
      "loss": 1.03,
      "step": 158
    },
    {
      "epoch": 0.012848744419079173,
      "grad_norm": 3.3173232078552246,
      "learning_rate": 2.1399730820995965e-06,
      "loss": 1.0768,
      "step": 159
    },
    {
      "epoch": 0.012929554132406715,
      "grad_norm": 3.113736629486084,
      "learning_rate": 2.1534320323014808e-06,
      "loss": 1.2176,
      "step": 160
    },
    {
      "epoch": 0.013010363845734257,
      "grad_norm": 3.7131903171539307,
      "learning_rate": 2.166890982503365e-06,
      "loss": 1.1292,
      "step": 161
    },
    {
      "epoch": 0.0130911735590618,
      "grad_norm": 3.4207205772399902,
      "learning_rate": 2.180349932705249e-06,
      "loss": 1.0778,
      "step": 162
    },
    {
      "epoch": 0.013171983272389342,
      "grad_norm": 3.311591625213623,
      "learning_rate": 2.1938088829071332e-06,
      "loss": 1.0529,
      "step": 163
    },
    {
      "epoch": 0.013252792985716882,
      "grad_norm": 3.1533114910125732,
      "learning_rate": 2.207267833109018e-06,
      "loss": 1.087,
      "step": 164
    },
    {
      "epoch": 0.013333602699044425,
      "grad_norm": 3.94301700592041,
      "learning_rate": 2.2207267833109018e-06,
      "loss": 1.0665,
      "step": 165
    },
    {
      "epoch": 0.013414412412371967,
      "grad_norm": 3.766512632369995,
      "learning_rate": 2.234185733512786e-06,
      "loss": 1.1203,
      "step": 166
    },
    {
      "epoch": 0.01349522212569951,
      "grad_norm": 4.3540263175964355,
      "learning_rate": 2.2476446837146704e-06,
      "loss": 1.1282,
      "step": 167
    },
    {
      "epoch": 0.013576031839027052,
      "grad_norm": 3.223722457885742,
      "learning_rate": 2.2611036339165546e-06,
      "loss": 1.0528,
      "step": 168
    },
    {
      "epoch": 0.013656841552354594,
      "grad_norm": 3.663490056991577,
      "learning_rate": 2.274562584118439e-06,
      "loss": 1.0388,
      "step": 169
    },
    {
      "epoch": 0.013737651265682134,
      "grad_norm": 3.2799320220947266,
      "learning_rate": 2.2880215343203232e-06,
      "loss": 1.2821,
      "step": 170
    },
    {
      "epoch": 0.013818460979009677,
      "grad_norm": 3.058697462081909,
      "learning_rate": 2.3014804845222075e-06,
      "loss": 1.0764,
      "step": 171
    },
    {
      "epoch": 0.013899270692337219,
      "grad_norm": 3.0403807163238525,
      "learning_rate": 2.314939434724092e-06,
      "loss": 1.0531,
      "step": 172
    },
    {
      "epoch": 0.013980080405664761,
      "grad_norm": 3.4602065086364746,
      "learning_rate": 2.3283983849259757e-06,
      "loss": 1.1011,
      "step": 173
    },
    {
      "epoch": 0.014060890118992303,
      "grad_norm": 3.9035913944244385,
      "learning_rate": 2.3418573351278604e-06,
      "loss": 1.1127,
      "step": 174
    },
    {
      "epoch": 0.014141699832319846,
      "grad_norm": 2.8776652812957764,
      "learning_rate": 2.3553162853297447e-06,
      "loss": 1.1733,
      "step": 175
    },
    {
      "epoch": 0.014222509545647386,
      "grad_norm": 3.1856839656829834,
      "learning_rate": 2.368775235531629e-06,
      "loss": 1.0929,
      "step": 176
    },
    {
      "epoch": 0.014303319258974928,
      "grad_norm": 3.479437828063965,
      "learning_rate": 2.382234185733513e-06,
      "loss": 1.1085,
      "step": 177
    },
    {
      "epoch": 0.01438412897230247,
      "grad_norm": 3.5891594886779785,
      "learning_rate": 2.395693135935397e-06,
      "loss": 0.9675,
      "step": 178
    },
    {
      "epoch": 0.014464938685630013,
      "grad_norm": 3.4782447814941406,
      "learning_rate": 2.4091520861372814e-06,
      "loss": 0.9895,
      "step": 179
    },
    {
      "epoch": 0.014545748398957555,
      "grad_norm": 2.508082151412964,
      "learning_rate": 2.4226110363391657e-06,
      "loss": 0.9304,
      "step": 180
    },
    {
      "epoch": 0.014626558112285097,
      "grad_norm": 3.4057435989379883,
      "learning_rate": 2.43606998654105e-06,
      "loss": 1.0716,
      "step": 181
    },
    {
      "epoch": 0.014707367825612638,
      "grad_norm": 3.42026424407959,
      "learning_rate": 2.4495289367429342e-06,
      "loss": 1.1508,
      "step": 182
    },
    {
      "epoch": 0.01478817753894018,
      "grad_norm": 3.884592056274414,
      "learning_rate": 2.4629878869448185e-06,
      "loss": 1.165,
      "step": 183
    },
    {
      "epoch": 0.014868987252267723,
      "grad_norm": 3.5591719150543213,
      "learning_rate": 2.476446837146703e-06,
      "loss": 1.1261,
      "step": 184
    },
    {
      "epoch": 0.014949796965595265,
      "grad_norm": 3.258593797683716,
      "learning_rate": 2.489905787348587e-06,
      "loss": 1.0031,
      "step": 185
    },
    {
      "epoch": 0.015030606678922807,
      "grad_norm": 3.988311767578125,
      "learning_rate": 2.503364737550471e-06,
      "loss": 1.225,
      "step": 186
    },
    {
      "epoch": 0.01511141639225035,
      "grad_norm": 3.750598430633545,
      "learning_rate": 2.5168236877523557e-06,
      "loss": 1.0899,
      "step": 187
    },
    {
      "epoch": 0.01519222610557789,
      "grad_norm": 3.344961404800415,
      "learning_rate": 2.53028263795424e-06,
      "loss": 1.0497,
      "step": 188
    },
    {
      "epoch": 0.015273035818905432,
      "grad_norm": 3.5242574214935303,
      "learning_rate": 2.543741588156124e-06,
      "loss": 0.983,
      "step": 189
    },
    {
      "epoch": 0.015353845532232974,
      "grad_norm": 3.1082167625427246,
      "learning_rate": 2.5572005383580085e-06,
      "loss": 1.0544,
      "step": 190
    },
    {
      "epoch": 0.015434655245560517,
      "grad_norm": 3.4438302516937256,
      "learning_rate": 2.5706594885598924e-06,
      "loss": 1.1623,
      "step": 191
    },
    {
      "epoch": 0.015515464958888059,
      "grad_norm": 3.1369824409484863,
      "learning_rate": 2.5841184387617767e-06,
      "loss": 1.0385,
      "step": 192
    },
    {
      "epoch": 0.015596274672215601,
      "grad_norm": 3.228114366531372,
      "learning_rate": 2.5975773889636614e-06,
      "loss": 0.9485,
      "step": 193
    },
    {
      "epoch": 0.015677084385543143,
      "grad_norm": 3.369588613510132,
      "learning_rate": 2.6110363391655453e-06,
      "loss": 1.2509,
      "step": 194
    },
    {
      "epoch": 0.015757894098870684,
      "grad_norm": 3.242443323135376,
      "learning_rate": 2.6244952893674295e-06,
      "loss": 1.079,
      "step": 195
    },
    {
      "epoch": 0.015838703812198228,
      "grad_norm": 3.1657700538635254,
      "learning_rate": 2.637954239569314e-06,
      "loss": 1.1094,
      "step": 196
    },
    {
      "epoch": 0.01591951352552577,
      "grad_norm": 3.7620697021484375,
      "learning_rate": 2.651413189771198e-06,
      "loss": 1.0819,
      "step": 197
    },
    {
      "epoch": 0.01600032323885331,
      "grad_norm": 3.6370279788970947,
      "learning_rate": 2.6648721399730824e-06,
      "loss": 0.9747,
      "step": 198
    },
    {
      "epoch": 0.016081132952180853,
      "grad_norm": 2.9298436641693115,
      "learning_rate": 2.6783310901749667e-06,
      "loss": 1.1835,
      "step": 199
    },
    {
      "epoch": 0.016161942665508394,
      "grad_norm": 3.335447072982788,
      "learning_rate": 2.691790040376851e-06,
      "loss": 1.0578,
      "step": 200
    },
    {
      "epoch": 0.016242752378835938,
      "grad_norm": 3.341702938079834,
      "learning_rate": 2.705248990578735e-06,
      "loss": 0.9886,
      "step": 201
    },
    {
      "epoch": 0.016323562092163478,
      "grad_norm": 2.7136051654815674,
      "learning_rate": 2.7187079407806195e-06,
      "loss": 1.017,
      "step": 202
    },
    {
      "epoch": 0.01640437180549102,
      "grad_norm": 3.599208354949951,
      "learning_rate": 2.7321668909825034e-06,
      "loss": 1.0553,
      "step": 203
    },
    {
      "epoch": 0.016485181518818563,
      "grad_norm": 2.866802453994751,
      "learning_rate": 2.7456258411843877e-06,
      "loss": 1.1636,
      "step": 204
    },
    {
      "epoch": 0.016565991232146103,
      "grad_norm": 3.0029284954071045,
      "learning_rate": 2.7590847913862724e-06,
      "loss": 1.0556,
      "step": 205
    },
    {
      "epoch": 0.016646800945473647,
      "grad_norm": 3.1328749656677246,
      "learning_rate": 2.7725437415881563e-06,
      "loss": 1.0848,
      "step": 206
    },
    {
      "epoch": 0.016727610658801188,
      "grad_norm": 3.0416531562805176,
      "learning_rate": 2.7860026917900406e-06,
      "loss": 1.0294,
      "step": 207
    },
    {
      "epoch": 0.01680842037212873,
      "grad_norm": 2.908458709716797,
      "learning_rate": 2.7994616419919244e-06,
      "loss": 1.2528,
      "step": 208
    },
    {
      "epoch": 0.016889230085456272,
      "grad_norm": 3.4849448204040527,
      "learning_rate": 2.812920592193809e-06,
      "loss": 1.1045,
      "step": 209
    },
    {
      "epoch": 0.016970039798783813,
      "grad_norm": 3.44240140914917,
      "learning_rate": 2.8263795423956934e-06,
      "loss": 1.0109,
      "step": 210
    },
    {
      "epoch": 0.017050849512111357,
      "grad_norm": 3.6446714401245117,
      "learning_rate": 2.8398384925975777e-06,
      "loss": 1.1597,
      "step": 211
    },
    {
      "epoch": 0.017131659225438897,
      "grad_norm": 3.3532803058624268,
      "learning_rate": 2.853297442799462e-06,
      "loss": 1.012,
      "step": 212
    },
    {
      "epoch": 0.01721246893876644,
      "grad_norm": 2.927619457244873,
      "learning_rate": 2.866756393001346e-06,
      "loss": 1.0018,
      "step": 213
    },
    {
      "epoch": 0.01729327865209398,
      "grad_norm": 3.5168118476867676,
      "learning_rate": 2.8802153432032306e-06,
      "loss": 1.0221,
      "step": 214
    },
    {
      "epoch": 0.017374088365421522,
      "grad_norm": 2.9070892333984375,
      "learning_rate": 2.893674293405115e-06,
      "loss": 1.0964,
      "step": 215
    },
    {
      "epoch": 0.017454898078749066,
      "grad_norm": 3.6018590927124023,
      "learning_rate": 2.9071332436069987e-06,
      "loss": 1.1081,
      "step": 216
    },
    {
      "epoch": 0.017535707792076607,
      "grad_norm": 4.031643390655518,
      "learning_rate": 2.9205921938088834e-06,
      "loss": 1.0165,
      "step": 217
    },
    {
      "epoch": 0.01761651750540415,
      "grad_norm": 3.1344571113586426,
      "learning_rate": 2.9340511440107673e-06,
      "loss": 1.1071,
      "step": 218
    },
    {
      "epoch": 0.01769732721873169,
      "grad_norm": 4.082541465759277,
      "learning_rate": 2.9475100942126516e-06,
      "loss": 1.1359,
      "step": 219
    },
    {
      "epoch": 0.017778136932059232,
      "grad_norm": 3.3356659412384033,
      "learning_rate": 2.9609690444145363e-06,
      "loss": 1.034,
      "step": 220
    },
    {
      "epoch": 0.017858946645386776,
      "grad_norm": 3.1675662994384766,
      "learning_rate": 2.97442799461642e-06,
      "loss": 1.0509,
      "step": 221
    },
    {
      "epoch": 0.017939756358714316,
      "grad_norm": 3.5920846462249756,
      "learning_rate": 2.9878869448183044e-06,
      "loss": 1.0415,
      "step": 222
    },
    {
      "epoch": 0.01802056607204186,
      "grad_norm": 3.178565740585327,
      "learning_rate": 3.0013458950201883e-06,
      "loss": 1.2007,
      "step": 223
    },
    {
      "epoch": 0.0181013757853694,
      "grad_norm": 4.121781349182129,
      "learning_rate": 3.014804845222073e-06,
      "loss": 0.9993,
      "step": 224
    },
    {
      "epoch": 0.018182185498696945,
      "grad_norm": 3.4476571083068848,
      "learning_rate": 3.0282637954239573e-06,
      "loss": 1.0524,
      "step": 225
    },
    {
      "epoch": 0.018262995212024485,
      "grad_norm": 3.460421323776245,
      "learning_rate": 3.041722745625841e-06,
      "loss": 1.0942,
      "step": 226
    },
    {
      "epoch": 0.018343804925352026,
      "grad_norm": 2.8429582118988037,
      "learning_rate": 3.055181695827726e-06,
      "loss": 1.0494,
      "step": 227
    },
    {
      "epoch": 0.01842461463867957,
      "grad_norm": 2.999035120010376,
      "learning_rate": 3.0686406460296097e-06,
      "loss": 1.1238,
      "step": 228
    },
    {
      "epoch": 0.01850542435200711,
      "grad_norm": 4.474518775939941,
      "learning_rate": 3.0820995962314944e-06,
      "loss": 1.085,
      "step": 229
    },
    {
      "epoch": 0.018586234065334654,
      "grad_norm": 3.249032735824585,
      "learning_rate": 3.0955585464333787e-06,
      "loss": 1.083,
      "step": 230
    },
    {
      "epoch": 0.018667043778662195,
      "grad_norm": 3.566030502319336,
      "learning_rate": 3.1090174966352626e-06,
      "loss": 1.2091,
      "step": 231
    },
    {
      "epoch": 0.018747853491989735,
      "grad_norm": 3.5390536785125732,
      "learning_rate": 3.1224764468371473e-06,
      "loss": 1.0893,
      "step": 232
    },
    {
      "epoch": 0.01882866320531728,
      "grad_norm": 3.0526182651519775,
      "learning_rate": 3.135935397039031e-06,
      "loss": 1.0634,
      "step": 233
    },
    {
      "epoch": 0.01890947291864482,
      "grad_norm": 4.032508850097656,
      "learning_rate": 3.1493943472409154e-06,
      "loss": 1.0615,
      "step": 234
    },
    {
      "epoch": 0.018990282631972364,
      "grad_norm": 5.153072834014893,
      "learning_rate": 3.1628532974427993e-06,
      "loss": 1.1304,
      "step": 235
    },
    {
      "epoch": 0.019071092345299905,
      "grad_norm": 3.5427660942077637,
      "learning_rate": 3.176312247644684e-06,
      "loss": 1.1508,
      "step": 236
    },
    {
      "epoch": 0.01915190205862745,
      "grad_norm": 3.172734260559082,
      "learning_rate": 3.1897711978465683e-06,
      "loss": 1.0361,
      "step": 237
    },
    {
      "epoch": 0.01923271177195499,
      "grad_norm": 4.713091850280762,
      "learning_rate": 3.203230148048452e-06,
      "loss": 1.0781,
      "step": 238
    },
    {
      "epoch": 0.01931352148528253,
      "grad_norm": 3.1458215713500977,
      "learning_rate": 3.216689098250337e-06,
      "loss": 1.1755,
      "step": 239
    },
    {
      "epoch": 0.019394331198610074,
      "grad_norm": 3.0068893432617188,
      "learning_rate": 3.2301480484522207e-06,
      "loss": 1.1408,
      "step": 240
    },
    {
      "epoch": 0.019475140911937614,
      "grad_norm": 2.879537343978882,
      "learning_rate": 3.243606998654105e-06,
      "loss": 0.9688,
      "step": 241
    },
    {
      "epoch": 0.019555950625265158,
      "grad_norm": 3.4544901847839355,
      "learning_rate": 3.2570659488559897e-06,
      "loss": 0.9786,
      "step": 242
    },
    {
      "epoch": 0.0196367603385927,
      "grad_norm": 3.8788928985595703,
      "learning_rate": 3.2705248990578736e-06,
      "loss": 1.0054,
      "step": 243
    },
    {
      "epoch": 0.01971757005192024,
      "grad_norm": 3.1861045360565186,
      "learning_rate": 3.2839838492597583e-06,
      "loss": 1.0104,
      "step": 244
    },
    {
      "epoch": 0.019798379765247783,
      "grad_norm": 3.352832078933716,
      "learning_rate": 3.297442799461642e-06,
      "loss": 1.0723,
      "step": 245
    },
    {
      "epoch": 0.019879189478575324,
      "grad_norm": 2.9049644470214844,
      "learning_rate": 3.3109017496635265e-06,
      "loss": 1.108,
      "step": 246
    },
    {
      "epoch": 0.019959999191902868,
      "grad_norm": 3.1392064094543457,
      "learning_rate": 3.324360699865411e-06,
      "loss": 0.9637,
      "step": 247
    },
    {
      "epoch": 0.020040808905230408,
      "grad_norm": 2.9236528873443604,
      "learning_rate": 3.337819650067295e-06,
      "loss": 1.1056,
      "step": 248
    },
    {
      "epoch": 0.020121618618557952,
      "grad_norm": 3.3994264602661133,
      "learning_rate": 3.3512786002691793e-06,
      "loss": 0.9997,
      "step": 249
    },
    {
      "epoch": 0.020202428331885493,
      "grad_norm": 3.403186559677124,
      "learning_rate": 3.364737550471063e-06,
      "loss": 0.9786,
      "step": 250
    },
    {
      "epoch": 0.020283238045213033,
      "grad_norm": 3.129040002822876,
      "learning_rate": 3.378196500672948e-06,
      "loss": 1.2622,
      "step": 251
    },
    {
      "epoch": 0.020364047758540577,
      "grad_norm": 3.049415111541748,
      "learning_rate": 3.391655450874832e-06,
      "loss": 1.2219,
      "step": 252
    },
    {
      "epoch": 0.020444857471868118,
      "grad_norm": 2.960981845855713,
      "learning_rate": 3.405114401076716e-06,
      "loss": 1.0447,
      "step": 253
    },
    {
      "epoch": 0.020525667185195662,
      "grad_norm": 3.1426732540130615,
      "learning_rate": 3.4185733512786008e-06,
      "loss": 1.0372,
      "step": 254
    },
    {
      "epoch": 0.020606476898523202,
      "grad_norm": 2.869678258895874,
      "learning_rate": 3.4320323014804846e-06,
      "loss": 0.9815,
      "step": 255
    },
    {
      "epoch": 0.020687286611850743,
      "grad_norm": 3.1211798191070557,
      "learning_rate": 3.445491251682369e-06,
      "loss": 1.0019,
      "step": 256
    },
    {
      "epoch": 0.020768096325178287,
      "grad_norm": 2.9854209423065186,
      "learning_rate": 3.4589502018842536e-06,
      "loss": 1.0042,
      "step": 257
    },
    {
      "epoch": 0.020848906038505827,
      "grad_norm": 3.7202820777893066,
      "learning_rate": 3.4724091520861375e-06,
      "loss": 1.0727,
      "step": 258
    },
    {
      "epoch": 0.02092971575183337,
      "grad_norm": 3.9204063415527344,
      "learning_rate": 3.4858681022880218e-06,
      "loss": 1.0386,
      "step": 259
    },
    {
      "epoch": 0.021010525465160912,
      "grad_norm": 3.1514883041381836,
      "learning_rate": 3.499327052489906e-06,
      "loss": 1.0543,
      "step": 260
    },
    {
      "epoch": 0.021091335178488456,
      "grad_norm": 3.1727118492126465,
      "learning_rate": 3.5127860026917903e-06,
      "loss": 1.0541,
      "step": 261
    },
    {
      "epoch": 0.021172144891815996,
      "grad_norm": 3.517979860305786,
      "learning_rate": 3.526244952893675e-06,
      "loss": 1.0653,
      "step": 262
    },
    {
      "epoch": 0.021252954605143537,
      "grad_norm": 3.120305061340332,
      "learning_rate": 3.539703903095559e-06,
      "loss": 1.0551,
      "step": 263
    },
    {
      "epoch": 0.02133376431847108,
      "grad_norm": 3.2179126739501953,
      "learning_rate": 3.553162853297443e-06,
      "loss": 0.9822,
      "step": 264
    },
    {
      "epoch": 0.02141457403179862,
      "grad_norm": 3.563760995864868,
      "learning_rate": 3.566621803499327e-06,
      "loss": 0.9401,
      "step": 265
    },
    {
      "epoch": 0.021495383745126165,
      "grad_norm": 3.1966042518615723,
      "learning_rate": 3.5800807537012118e-06,
      "loss": 1.1475,
      "step": 266
    },
    {
      "epoch": 0.021576193458453706,
      "grad_norm": 2.9117894172668457,
      "learning_rate": 3.5935397039030956e-06,
      "loss": 1.0869,
      "step": 267
    },
    {
      "epoch": 0.021657003171781247,
      "grad_norm": 2.9326651096343994,
      "learning_rate": 3.60699865410498e-06,
      "loss": 1.0897,
      "step": 268
    },
    {
      "epoch": 0.02173781288510879,
      "grad_norm": 3.3142008781433105,
      "learning_rate": 3.6204576043068646e-06,
      "loss": 1.1048,
      "step": 269
    },
    {
      "epoch": 0.02181862259843633,
      "grad_norm": 3.220008611679077,
      "learning_rate": 3.6339165545087485e-06,
      "loss": 1.1076,
      "step": 270
    },
    {
      "epoch": 0.021899432311763875,
      "grad_norm": 3.6159610748291016,
      "learning_rate": 3.6473755047106328e-06,
      "loss": 1.0769,
      "step": 271
    },
    {
      "epoch": 0.021980242025091416,
      "grad_norm": 3.3969810009002686,
      "learning_rate": 3.660834454912517e-06,
      "loss": 1.105,
      "step": 272
    },
    {
      "epoch": 0.02206105173841896,
      "grad_norm": 3.580536365509033,
      "learning_rate": 3.6742934051144014e-06,
      "loss": 1.0128,
      "step": 273
    },
    {
      "epoch": 0.0221418614517465,
      "grad_norm": 3.2871317863464355,
      "learning_rate": 3.6877523553162856e-06,
      "loss": 0.9887,
      "step": 274
    },
    {
      "epoch": 0.02222267116507404,
      "grad_norm": 3.7190616130828857,
      "learning_rate": 3.70121130551817e-06,
      "loss": 1.0394,
      "step": 275
    },
    {
      "epoch": 0.022303480878401585,
      "grad_norm": 3.1299095153808594,
      "learning_rate": 3.7146702557200542e-06,
      "loss": 1.1785,
      "step": 276
    },
    {
      "epoch": 0.022384290591729125,
      "grad_norm": 3.4073848724365234,
      "learning_rate": 3.728129205921938e-06,
      "loss": 1.0502,
      "step": 277
    },
    {
      "epoch": 0.02246510030505667,
      "grad_norm": 3.4248945713043213,
      "learning_rate": 3.7415881561238228e-06,
      "loss": 1.0235,
      "step": 278
    },
    {
      "epoch": 0.02254591001838421,
      "grad_norm": 3.900418758392334,
      "learning_rate": 3.755047106325707e-06,
      "loss": 1.0979,
      "step": 279
    },
    {
      "epoch": 0.02262671973171175,
      "grad_norm": 3.830432415008545,
      "learning_rate": 3.768506056527591e-06,
      "loss": 1.1172,
      "step": 280
    },
    {
      "epoch": 0.022707529445039294,
      "grad_norm": 3.209726333618164,
      "learning_rate": 3.7819650067294756e-06,
      "loss": 1.0036,
      "step": 281
    },
    {
      "epoch": 0.022788339158366835,
      "grad_norm": 3.1260225772857666,
      "learning_rate": 3.7954239569313595e-06,
      "loss": 1.1003,
      "step": 282
    },
    {
      "epoch": 0.02286914887169438,
      "grad_norm": 3.050652503967285,
      "learning_rate": 3.808882907133244e-06,
      "loss": 1.0699,
      "step": 283
    },
    {
      "epoch": 0.02294995858502192,
      "grad_norm": 3.4412708282470703,
      "learning_rate": 3.822341857335128e-06,
      "loss": 1.0617,
      "step": 284
    },
    {
      "epoch": 0.023030768298349463,
      "grad_norm": 3.2412991523742676,
      "learning_rate": 3.835800807537012e-06,
      "loss": 1.1161,
      "step": 285
    },
    {
      "epoch": 0.023111578011677004,
      "grad_norm": 3.3610498905181885,
      "learning_rate": 3.849259757738897e-06,
      "loss": 1.0642,
      "step": 286
    },
    {
      "epoch": 0.023192387725004544,
      "grad_norm": 4.764945983886719,
      "learning_rate": 3.8627187079407805e-06,
      "loss": 1.0355,
      "step": 287
    },
    {
      "epoch": 0.02327319743833209,
      "grad_norm": 3.3774383068084717,
      "learning_rate": 3.876177658142665e-06,
      "loss": 0.9947,
      "step": 288
    },
    {
      "epoch": 0.02335400715165963,
      "grad_norm": 3.4327151775360107,
      "learning_rate": 3.88963660834455e-06,
      "loss": 0.9403,
      "step": 289
    },
    {
      "epoch": 0.023434816864987173,
      "grad_norm": 3.2053916454315186,
      "learning_rate": 3.903095558546434e-06,
      "loss": 0.9516,
      "step": 290
    },
    {
      "epoch": 0.023515626578314713,
      "grad_norm": 3.281947612762451,
      "learning_rate": 3.9165545087483185e-06,
      "loss": 1.004,
      "step": 291
    },
    {
      "epoch": 0.023596436291642254,
      "grad_norm": 3.3425886631011963,
      "learning_rate": 3.930013458950202e-06,
      "loss": 1.0692,
      "step": 292
    },
    {
      "epoch": 0.023677246004969798,
      "grad_norm": 3.682798147201538,
      "learning_rate": 3.943472409152086e-06,
      "loss": 0.9881,
      "step": 293
    },
    {
      "epoch": 0.02375805571829734,
      "grad_norm": 4.1106977462768555,
      "learning_rate": 3.956931359353971e-06,
      "loss": 1.029,
      "step": 294
    },
    {
      "epoch": 0.023838865431624882,
      "grad_norm": 2.9321463108062744,
      "learning_rate": 3.970390309555855e-06,
      "loss": 1.0701,
      "step": 295
    },
    {
      "epoch": 0.023919675144952423,
      "grad_norm": 3.840993881225586,
      "learning_rate": 3.9838492597577395e-06,
      "loss": 1.0822,
      "step": 296
    },
    {
      "epoch": 0.024000484858279967,
      "grad_norm": 3.33097505569458,
      "learning_rate": 3.997308209959623e-06,
      "loss": 1.0678,
      "step": 297
    },
    {
      "epoch": 0.024081294571607507,
      "grad_norm": 3.3836615085601807,
      "learning_rate": 4.010767160161508e-06,
      "loss": 0.9405,
      "step": 298
    },
    {
      "epoch": 0.024162104284935048,
      "grad_norm": 3.6300384998321533,
      "learning_rate": 4.024226110363392e-06,
      "loss": 1.0488,
      "step": 299
    },
    {
      "epoch": 0.024242913998262592,
      "grad_norm": 4.789526462554932,
      "learning_rate": 4.037685060565276e-06,
      "loss": 1.1742,
      "step": 300
    },
    {
      "epoch": 0.024323723711590133,
      "grad_norm": 4.001083850860596,
      "learning_rate": 4.0511440107671605e-06,
      "loss": 1.0492,
      "step": 301
    },
    {
      "epoch": 0.024404533424917677,
      "grad_norm": 3.0341265201568604,
      "learning_rate": 4.064602960969044e-06,
      "loss": 1.0646,
      "step": 302
    },
    {
      "epoch": 0.024485343138245217,
      "grad_norm": 3.0529284477233887,
      "learning_rate": 4.078061911170929e-06,
      "loss": 1.0188,
      "step": 303
    },
    {
      "epoch": 0.024566152851572758,
      "grad_norm": 3.5513386726379395,
      "learning_rate": 4.091520861372813e-06,
      "loss": 1.0259,
      "step": 304
    },
    {
      "epoch": 0.0246469625649003,
      "grad_norm": 3.639235734939575,
      "learning_rate": 4.104979811574698e-06,
      "loss": 1.0545,
      "step": 305
    },
    {
      "epoch": 0.024727772278227842,
      "grad_norm": 2.771613597869873,
      "learning_rate": 4.118438761776582e-06,
      "loss": 0.9668,
      "step": 306
    },
    {
      "epoch": 0.024808581991555386,
      "grad_norm": 2.699173927307129,
      "learning_rate": 4.131897711978466e-06,
      "loss": 1.1614,
      "step": 307
    },
    {
      "epoch": 0.024889391704882927,
      "grad_norm": 3.053584098815918,
      "learning_rate": 4.14535666218035e-06,
      "loss": 1.1264,
      "step": 308
    },
    {
      "epoch": 0.02497020141821047,
      "grad_norm": 3.2321035861968994,
      "learning_rate": 4.158815612382234e-06,
      "loss": 1.0321,
      "step": 309
    },
    {
      "epoch": 0.02505101113153801,
      "grad_norm": 3.159599542617798,
      "learning_rate": 4.172274562584119e-06,
      "loss": 1.0271,
      "step": 310
    },
    {
      "epoch": 0.02513182084486555,
      "grad_norm": 3.739326238632202,
      "learning_rate": 4.185733512786003e-06,
      "loss": 1.0775,
      "step": 311
    },
    {
      "epoch": 0.025212630558193096,
      "grad_norm": 2.9522006511688232,
      "learning_rate": 4.199192462987887e-06,
      "loss": 1.16,
      "step": 312
    },
    {
      "epoch": 0.025293440271520636,
      "grad_norm": 4.1226301193237305,
      "learning_rate": 4.212651413189772e-06,
      "loss": 1.0076,
      "step": 313
    },
    {
      "epoch": 0.02537424998484818,
      "grad_norm": 3.438518762588501,
      "learning_rate": 4.226110363391656e-06,
      "loss": 1.1045,
      "step": 314
    },
    {
      "epoch": 0.02545505969817572,
      "grad_norm": 3.4645230770111084,
      "learning_rate": 4.23956931359354e-06,
      "loss": 1.0568,
      "step": 315
    },
    {
      "epoch": 0.02553586941150326,
      "grad_norm": 3.5905091762542725,
      "learning_rate": 4.253028263795424e-06,
      "loss": 1.0134,
      "step": 316
    },
    {
      "epoch": 0.025616679124830805,
      "grad_norm": 3.258754014968872,
      "learning_rate": 4.266487213997308e-06,
      "loss": 0.958,
      "step": 317
    },
    {
      "epoch": 0.025697488838158346,
      "grad_norm": 3.016822338104248,
      "learning_rate": 4.279946164199193e-06,
      "loss": 1.069,
      "step": 318
    },
    {
      "epoch": 0.02577829855148589,
      "grad_norm": 3.46635365486145,
      "learning_rate": 4.293405114401077e-06,
      "loss": 1.0767,
      "step": 319
    },
    {
      "epoch": 0.02585910826481343,
      "grad_norm": 3.2740538120269775,
      "learning_rate": 4.3068640646029616e-06,
      "loss": 1.1643,
      "step": 320
    },
    {
      "epoch": 0.02593991797814097,
      "grad_norm": 2.509807586669922,
      "learning_rate": 4.320323014804845e-06,
      "loss": 1.0419,
      "step": 321
    },
    {
      "epoch": 0.026020727691468515,
      "grad_norm": 3.7150912284851074,
      "learning_rate": 4.33378196500673e-06,
      "loss": 0.9806,
      "step": 322
    },
    {
      "epoch": 0.026101537404796055,
      "grad_norm": 3.265268087387085,
      "learning_rate": 4.347240915208614e-06,
      "loss": 1.0416,
      "step": 323
    },
    {
      "epoch": 0.0261823471181236,
      "grad_norm": 3.423750162124634,
      "learning_rate": 4.360699865410498e-06,
      "loss": 1.0875,
      "step": 324
    },
    {
      "epoch": 0.02626315683145114,
      "grad_norm": 3.0355896949768066,
      "learning_rate": 4.3741588156123826e-06,
      "loss": 1.0706,
      "step": 325
    },
    {
      "epoch": 0.026343966544778684,
      "grad_norm": 3.399970531463623,
      "learning_rate": 4.3876177658142664e-06,
      "loss": 0.9565,
      "step": 326
    },
    {
      "epoch": 0.026424776258106224,
      "grad_norm": 2.427396774291992,
      "learning_rate": 4.401076716016151e-06,
      "loss": 1.1196,
      "step": 327
    },
    {
      "epoch": 0.026505585971433765,
      "grad_norm": 2.8762001991271973,
      "learning_rate": 4.414535666218036e-06,
      "loss": 1.0821,
      "step": 328
    },
    {
      "epoch": 0.02658639568476131,
      "grad_norm": 2.805931329727173,
      "learning_rate": 4.42799461641992e-06,
      "loss": 1.085,
      "step": 329
    },
    {
      "epoch": 0.02666720539808885,
      "grad_norm": 2.9563746452331543,
      "learning_rate": 4.4414535666218036e-06,
      "loss": 1.1011,
      "step": 330
    },
    {
      "epoch": 0.026748015111416393,
      "grad_norm": 3.239896535873413,
      "learning_rate": 4.454912516823688e-06,
      "loss": 1.0285,
      "step": 331
    },
    {
      "epoch": 0.026828824824743934,
      "grad_norm": 3.9557912349700928,
      "learning_rate": 4.468371467025572e-06,
      "loss": 1.0994,
      "step": 332
    },
    {
      "epoch": 0.026909634538071475,
      "grad_norm": 3.006235122680664,
      "learning_rate": 4.481830417227457e-06,
      "loss": 1.0512,
      "step": 333
    },
    {
      "epoch": 0.02699044425139902,
      "grad_norm": 2.7034130096435547,
      "learning_rate": 4.495289367429341e-06,
      "loss": 1.0497,
      "step": 334
    },
    {
      "epoch": 0.02707125396472656,
      "grad_norm": 3.289365530014038,
      "learning_rate": 4.5087483176312254e-06,
      "loss": 1.0073,
      "step": 335
    },
    {
      "epoch": 0.027152063678054103,
      "grad_norm": 2.970745801925659,
      "learning_rate": 4.522207267833109e-06,
      "loss": 1.0532,
      "step": 336
    },
    {
      "epoch": 0.027232873391381644,
      "grad_norm": 3.1716456413269043,
      "learning_rate": 4.535666218034994e-06,
      "loss": 1.1299,
      "step": 337
    },
    {
      "epoch": 0.027313683104709188,
      "grad_norm": 3.3215408325195312,
      "learning_rate": 4.549125168236878e-06,
      "loss": 0.9924,
      "step": 338
    },
    {
      "epoch": 0.027394492818036728,
      "grad_norm": 3.2503867149353027,
      "learning_rate": 4.562584118438762e-06,
      "loss": 1.1627,
      "step": 339
    },
    {
      "epoch": 0.02747530253136427,
      "grad_norm": 3.6553633213043213,
      "learning_rate": 4.5760430686406464e-06,
      "loss": 1.005,
      "step": 340
    },
    {
      "epoch": 0.027556112244691813,
      "grad_norm": 3.136037588119507,
      "learning_rate": 4.58950201884253e-06,
      "loss": 1.0578,
      "step": 341
    },
    {
      "epoch": 0.027636921958019353,
      "grad_norm": 3.279416561126709,
      "learning_rate": 4.602960969044415e-06,
      "loss": 1.1861,
      "step": 342
    },
    {
      "epoch": 0.027717731671346897,
      "grad_norm": 3.5951592922210693,
      "learning_rate": 4.6164199192463e-06,
      "loss": 1.0194,
      "step": 343
    },
    {
      "epoch": 0.027798541384674438,
      "grad_norm": 3.4381399154663086,
      "learning_rate": 4.629878869448184e-06,
      "loss": 1.0519,
      "step": 344
    },
    {
      "epoch": 0.027879351098001978,
      "grad_norm": 3.5402069091796875,
      "learning_rate": 4.6433378196500674e-06,
      "loss": 1.0954,
      "step": 345
    },
    {
      "epoch": 0.027960160811329522,
      "grad_norm": 2.8578295707702637,
      "learning_rate": 4.656796769851951e-06,
      "loss": 1.112,
      "step": 346
    },
    {
      "epoch": 0.028040970524657063,
      "grad_norm": 2.570523977279663,
      "learning_rate": 4.670255720053836e-06,
      "loss": 1.1534,
      "step": 347
    },
    {
      "epoch": 0.028121780237984607,
      "grad_norm": 3.5016822814941406,
      "learning_rate": 4.683714670255721e-06,
      "loss": 1.0122,
      "step": 348
    },
    {
      "epoch": 0.028202589951312147,
      "grad_norm": 3.6116175651550293,
      "learning_rate": 4.697173620457605e-06,
      "loss": 0.9669,
      "step": 349
    },
    {
      "epoch": 0.02828339966463969,
      "grad_norm": 3.697629690170288,
      "learning_rate": 4.710632570659489e-06,
      "loss": 0.9453,
      "step": 350
    },
    {
      "epoch": 0.028364209377967232,
      "grad_norm": 2.860255479812622,
      "learning_rate": 4.724091520861373e-06,
      "loss": 1.0293,
      "step": 351
    },
    {
      "epoch": 0.028445019091294772,
      "grad_norm": 2.7185349464416504,
      "learning_rate": 4.737550471063258e-06,
      "loss": 1.0949,
      "step": 352
    },
    {
      "epoch": 0.028525828804622316,
      "grad_norm": 3.5234262943267822,
      "learning_rate": 4.751009421265142e-06,
      "loss": 1.0291,
      "step": 353
    },
    {
      "epoch": 0.028606638517949857,
      "grad_norm": 3.042196273803711,
      "learning_rate": 4.764468371467026e-06,
      "loss": 1.0282,
      "step": 354
    },
    {
      "epoch": 0.0286874482312774,
      "grad_norm": 3.581164836883545,
      "learning_rate": 4.77792732166891e-06,
      "loss": 1.1342,
      "step": 355
    },
    {
      "epoch": 0.02876825794460494,
      "grad_norm": 4.2634477615356445,
      "learning_rate": 4.791386271870794e-06,
      "loss": 1.0666,
      "step": 356
    },
    {
      "epoch": 0.028849067657932482,
      "grad_norm": 3.3745503425598145,
      "learning_rate": 4.804845222072679e-06,
      "loss": 0.9496,
      "step": 357
    },
    {
      "epoch": 0.028929877371260026,
      "grad_norm": 3.2284529209136963,
      "learning_rate": 4.818304172274563e-06,
      "loss": 1.0532,
      "step": 358
    },
    {
      "epoch": 0.029010687084587566,
      "grad_norm": 3.2654168605804443,
      "learning_rate": 4.8317631224764475e-06,
      "loss": 1.1196,
      "step": 359
    },
    {
      "epoch": 0.02909149679791511,
      "grad_norm": 3.0453786849975586,
      "learning_rate": 4.845222072678331e-06,
      "loss": 0.9026,
      "step": 360
    },
    {
      "epoch": 0.02917230651124265,
      "grad_norm": 3.1820859909057617,
      "learning_rate": 4.858681022880215e-06,
      "loss": 1.0151,
      "step": 361
    },
    {
      "epoch": 0.029253116224570195,
      "grad_norm": 2.892068386077881,
      "learning_rate": 4.8721399730821e-06,
      "loss": 1.0974,
      "step": 362
    },
    {
      "epoch": 0.029333925937897735,
      "grad_norm": 3.9766123294830322,
      "learning_rate": 4.885598923283984e-06,
      "loss": 1.0929,
      "step": 363
    },
    {
      "epoch": 0.029414735651225276,
      "grad_norm": 3.188502550125122,
      "learning_rate": 4.8990578734858685e-06,
      "loss": 1.1204,
      "step": 364
    },
    {
      "epoch": 0.02949554536455282,
      "grad_norm": 2.9032983779907227,
      "learning_rate": 4.912516823687753e-06,
      "loss": 1.1307,
      "step": 365
    },
    {
      "epoch": 0.02957635507788036,
      "grad_norm": 3.1441397666931152,
      "learning_rate": 4.925975773889637e-06,
      "loss": 1.2278,
      "step": 366
    },
    {
      "epoch": 0.029657164791207904,
      "grad_norm": 3.025193929672241,
      "learning_rate": 4.939434724091522e-06,
      "loss": 1.0689,
      "step": 367
    },
    {
      "epoch": 0.029737974504535445,
      "grad_norm": 3.3389151096343994,
      "learning_rate": 4.952893674293406e-06,
      "loss": 1.074,
      "step": 368
    },
    {
      "epoch": 0.029818784217862986,
      "grad_norm": 2.899667263031006,
      "learning_rate": 4.9663526244952895e-06,
      "loss": 1.0223,
      "step": 369
    },
    {
      "epoch": 0.02989959393119053,
      "grad_norm": 3.271374464035034,
      "learning_rate": 4.979811574697174e-06,
      "loss": 1.01,
      "step": 370
    },
    {
      "epoch": 0.02998040364451807,
      "grad_norm": 2.800734043121338,
      "learning_rate": 4.993270524899058e-06,
      "loss": 1.0948,
      "step": 371
    },
    {
      "epoch": 0.030061213357845614,
      "grad_norm": 3.0062313079833984,
      "learning_rate": 5.006729475100942e-06,
      "loss": 1.059,
      "step": 372
    },
    {
      "epoch": 0.030142023071173155,
      "grad_norm": 3.1042842864990234,
      "learning_rate": 5.020188425302827e-06,
      "loss": 1.0002,
      "step": 373
    },
    {
      "epoch": 0.0302228327845007,
      "grad_norm": 3.154513120651245,
      "learning_rate": 5.033647375504711e-06,
      "loss": 0.9859,
      "step": 374
    },
    {
      "epoch": 0.03030364249782824,
      "grad_norm": 2.9707469940185547,
      "learning_rate": 5.047106325706595e-06,
      "loss": 1.0158,
      "step": 375
    },
    {
      "epoch": 0.03038445221115578,
      "grad_norm": 3.7019219398498535,
      "learning_rate": 5.06056527590848e-06,
      "loss": 0.9693,
      "step": 376
    },
    {
      "epoch": 0.030465261924483324,
      "grad_norm": 3.5970447063446045,
      "learning_rate": 5.074024226110364e-06,
      "loss": 1.0817,
      "step": 377
    },
    {
      "epoch": 0.030546071637810864,
      "grad_norm": 3.0415992736816406,
      "learning_rate": 5.087483176312248e-06,
      "loss": 1.0749,
      "step": 378
    },
    {
      "epoch": 0.030626881351138408,
      "grad_norm": 3.1389529705047607,
      "learning_rate": 5.100942126514132e-06,
      "loss": 1.1466,
      "step": 379
    },
    {
      "epoch": 0.03070769106446595,
      "grad_norm": 3.5312931537628174,
      "learning_rate": 5.114401076716017e-06,
      "loss": 1.0444,
      "step": 380
    },
    {
      "epoch": 0.03078850077779349,
      "grad_norm": 2.7313449382781982,
      "learning_rate": 5.127860026917901e-06,
      "loss": 1.0735,
      "step": 381
    },
    {
      "epoch": 0.030869310491121033,
      "grad_norm": 3.8215954303741455,
      "learning_rate": 5.141318977119785e-06,
      "loss": 1.0956,
      "step": 382
    },
    {
      "epoch": 0.030950120204448574,
      "grad_norm": 3.008662223815918,
      "learning_rate": 5.1547779273216695e-06,
      "loss": 1.1227,
      "step": 383
    },
    {
      "epoch": 0.031030929917776118,
      "grad_norm": 3.5137126445770264,
      "learning_rate": 5.168236877523553e-06,
      "loss": 1.0647,
      "step": 384
    },
    {
      "epoch": 0.03111173963110366,
      "grad_norm": 3.1032652854919434,
      "learning_rate": 5.181695827725438e-06,
      "loss": 0.9709,
      "step": 385
    },
    {
      "epoch": 0.031192549344431202,
      "grad_norm": 2.643289089202881,
      "learning_rate": 5.195154777927323e-06,
      "loss": 1.0143,
      "step": 386
    },
    {
      "epoch": 0.03127335905775874,
      "grad_norm": 3.3892714977264404,
      "learning_rate": 5.208613728129206e-06,
      "loss": 1.0897,
      "step": 387
    },
    {
      "epoch": 0.03135416877108629,
      "grad_norm": 3.087230920791626,
      "learning_rate": 5.2220726783310905e-06,
      "loss": 1.0535,
      "step": 388
    },
    {
      "epoch": 0.031434978484413824,
      "grad_norm": 3.1502482891082764,
      "learning_rate": 5.235531628532975e-06,
      "loss": 0.9344,
      "step": 389
    },
    {
      "epoch": 0.03151578819774137,
      "grad_norm": 3.1660571098327637,
      "learning_rate": 5.248990578734859e-06,
      "loss": 0.9898,
      "step": 390
    },
    {
      "epoch": 0.03159659791106891,
      "grad_norm": 3.1750237941741943,
      "learning_rate": 5.262449528936744e-06,
      "loss": 0.9014,
      "step": 391
    },
    {
      "epoch": 0.031677407624396456,
      "grad_norm": 3.0551154613494873,
      "learning_rate": 5.275908479138628e-06,
      "loss": 1.1992,
      "step": 392
    },
    {
      "epoch": 0.03175821733772399,
      "grad_norm": 2.822049617767334,
      "learning_rate": 5.2893674293405115e-06,
      "loss": 1.0354,
      "step": 393
    },
    {
      "epoch": 0.03183902705105154,
      "grad_norm": 3.1479337215423584,
      "learning_rate": 5.302826379542396e-06,
      "loss": 1.0756,
      "step": 394
    },
    {
      "epoch": 0.03191983676437908,
      "grad_norm": 4.29952335357666,
      "learning_rate": 5.316285329744281e-06,
      "loss": 0.9937,
      "step": 395
    },
    {
      "epoch": 0.03200064647770662,
      "grad_norm": 3.460026264190674,
      "learning_rate": 5.329744279946165e-06,
      "loss": 1.0865,
      "step": 396
    },
    {
      "epoch": 0.03208145619103416,
      "grad_norm": 2.976209878921509,
      "learning_rate": 5.343203230148049e-06,
      "loss": 1.0514,
      "step": 397
    },
    {
      "epoch": 0.032162265904361706,
      "grad_norm": 3.255568027496338,
      "learning_rate": 5.356662180349933e-06,
      "loss": 1.0109,
      "step": 398
    },
    {
      "epoch": 0.03224307561768924,
      "grad_norm": 2.8264927864074707,
      "learning_rate": 5.370121130551817e-06,
      "loss": 1.0703,
      "step": 399
    },
    {
      "epoch": 0.03232388533101679,
      "grad_norm": 3.4914638996124268,
      "learning_rate": 5.383580080753702e-06,
      "loss": 1.0132,
      "step": 400
    },
    {
      "epoch": 0.03240469504434433,
      "grad_norm": 3.143792152404785,
      "learning_rate": 5.397039030955587e-06,
      "loss": 1.0457,
      "step": 401
    },
    {
      "epoch": 0.032485504757671875,
      "grad_norm": 3.3566250801086426,
      "learning_rate": 5.41049798115747e-06,
      "loss": 1.0481,
      "step": 402
    },
    {
      "epoch": 0.03256631447099941,
      "grad_norm": 2.8489530086517334,
      "learning_rate": 5.423956931359354e-06,
      "loss": 1.1821,
      "step": 403
    },
    {
      "epoch": 0.032647124184326956,
      "grad_norm": 3.260387897491455,
      "learning_rate": 5.437415881561239e-06,
      "loss": 1.0802,
      "step": 404
    },
    {
      "epoch": 0.0327279338976545,
      "grad_norm": 3.239534378051758,
      "learning_rate": 5.450874831763123e-06,
      "loss": 1.0194,
      "step": 405
    },
    {
      "epoch": 0.03280874361098204,
      "grad_norm": 3.2290117740631104,
      "learning_rate": 5.464333781965007e-06,
      "loss": 1.014,
      "step": 406
    },
    {
      "epoch": 0.03288955332430958,
      "grad_norm": 3.052867889404297,
      "learning_rate": 5.4777927321668915e-06,
      "loss": 1.147,
      "step": 407
    },
    {
      "epoch": 0.032970363037637125,
      "grad_norm": 3.0773065090179443,
      "learning_rate": 5.491251682368775e-06,
      "loss": 1.1505,
      "step": 408
    },
    {
      "epoch": 0.03305117275096467,
      "grad_norm": 4.307579040527344,
      "learning_rate": 5.50471063257066e-06,
      "loss": 1.0423,
      "step": 409
    },
    {
      "epoch": 0.033131982464292206,
      "grad_norm": 2.731487274169922,
      "learning_rate": 5.518169582772545e-06,
      "loss": 1.0521,
      "step": 410
    },
    {
      "epoch": 0.03321279217761975,
      "grad_norm": 2.8557963371276855,
      "learning_rate": 5.531628532974428e-06,
      "loss": 1.1326,
      "step": 411
    },
    {
      "epoch": 0.033293601890947294,
      "grad_norm": 2.9911723136901855,
      "learning_rate": 5.5450874831763125e-06,
      "loss": 1.0449,
      "step": 412
    },
    {
      "epoch": 0.03337441160427483,
      "grad_norm": 3.545870780944824,
      "learning_rate": 5.558546433378197e-06,
      "loss": 0.9643,
      "step": 413
    },
    {
      "epoch": 0.033455221317602375,
      "grad_norm": 3.228506326675415,
      "learning_rate": 5.572005383580081e-06,
      "loss": 0.9888,
      "step": 414
    },
    {
      "epoch": 0.03353603103092992,
      "grad_norm": 3.3360986709594727,
      "learning_rate": 5.585464333781966e-06,
      "loss": 1.1374,
      "step": 415
    },
    {
      "epoch": 0.03361684074425746,
      "grad_norm": 3.142885208129883,
      "learning_rate": 5.598923283983849e-06,
      "loss": 1.0261,
      "step": 416
    },
    {
      "epoch": 0.033697650457585,
      "grad_norm": 3.678441286087036,
      "learning_rate": 5.6123822341857335e-06,
      "loss": 1.0356,
      "step": 417
    },
    {
      "epoch": 0.033778460170912544,
      "grad_norm": 3.1850669384002686,
      "learning_rate": 5.625841184387618e-06,
      "loss": 0.8973,
      "step": 418
    },
    {
      "epoch": 0.03385926988424009,
      "grad_norm": 3.2738733291625977,
      "learning_rate": 5.639300134589503e-06,
      "loss": 1.0297,
      "step": 419
    },
    {
      "epoch": 0.033940079597567625,
      "grad_norm": 3.262432336807251,
      "learning_rate": 5.652759084791387e-06,
      "loss": 1.1121,
      "step": 420
    },
    {
      "epoch": 0.03402088931089517,
      "grad_norm": 3.300443172454834,
      "learning_rate": 5.666218034993271e-06,
      "loss": 1.0872,
      "step": 421
    },
    {
      "epoch": 0.03410169902422271,
      "grad_norm": 3.37052059173584,
      "learning_rate": 5.679676985195155e-06,
      "loss": 0.8631,
      "step": 422
    },
    {
      "epoch": 0.03418250873755025,
      "grad_norm": 3.0574722290039062,
      "learning_rate": 5.693135935397039e-06,
      "loss": 0.8917,
      "step": 423
    },
    {
      "epoch": 0.034263318450877794,
      "grad_norm": 3.1333374977111816,
      "learning_rate": 5.706594885598924e-06,
      "loss": 1.1487,
      "step": 424
    },
    {
      "epoch": 0.03434412816420534,
      "grad_norm": 3.4714949131011963,
      "learning_rate": 5.720053835800809e-06,
      "loss": 1.0668,
      "step": 425
    },
    {
      "epoch": 0.03442493787753288,
      "grad_norm": 3.0593292713165283,
      "learning_rate": 5.733512786002692e-06,
      "loss": 1.0453,
      "step": 426
    },
    {
      "epoch": 0.03450574759086042,
      "grad_norm": 3.006802558898926,
      "learning_rate": 5.746971736204576e-06,
      "loss": 1.0875,
      "step": 427
    },
    {
      "epoch": 0.03458655730418796,
      "grad_norm": 3.3820269107818604,
      "learning_rate": 5.760430686406461e-06,
      "loss": 1.0163,
      "step": 428
    },
    {
      "epoch": 0.03466736701751551,
      "grad_norm": 3.098365306854248,
      "learning_rate": 5.773889636608345e-06,
      "loss": 1.1997,
      "step": 429
    },
    {
      "epoch": 0.034748176730843044,
      "grad_norm": 3.436250686645508,
      "learning_rate": 5.78734858681023e-06,
      "loss": 0.9806,
      "step": 430
    },
    {
      "epoch": 0.03482898644417059,
      "grad_norm": 3.1773383617401123,
      "learning_rate": 5.800807537012113e-06,
      "loss": 0.9911,
      "step": 431
    },
    {
      "epoch": 0.03490979615749813,
      "grad_norm": 3.1867425441741943,
      "learning_rate": 5.814266487213997e-06,
      "loss": 1.1133,
      "step": 432
    },
    {
      "epoch": 0.034990605870825676,
      "grad_norm": 3.287316083908081,
      "learning_rate": 5.827725437415882e-06,
      "loss": 1.0283,
      "step": 433
    },
    {
      "epoch": 0.035071415584153214,
      "grad_norm": 3.760188579559326,
      "learning_rate": 5.841184387617767e-06,
      "loss": 1.1156,
      "step": 434
    },
    {
      "epoch": 0.03515222529748076,
      "grad_norm": 2.9129562377929688,
      "learning_rate": 5.854643337819651e-06,
      "loss": 1.0237,
      "step": 435
    },
    {
      "epoch": 0.0352330350108083,
      "grad_norm": 3.5939934253692627,
      "learning_rate": 5.8681022880215346e-06,
      "loss": 0.9305,
      "step": 436
    },
    {
      "epoch": 0.03531384472413584,
      "grad_norm": 3.105889081954956,
      "learning_rate": 5.8815612382234184e-06,
      "loss": 1.0533,
      "step": 437
    },
    {
      "epoch": 0.03539465443746338,
      "grad_norm": 3.263901948928833,
      "learning_rate": 5.895020188425303e-06,
      "loss": 0.9298,
      "step": 438
    },
    {
      "epoch": 0.03547546415079093,
      "grad_norm": 3.7842049598693848,
      "learning_rate": 5.908479138627188e-06,
      "loss": 1.0523,
      "step": 439
    },
    {
      "epoch": 0.035556273864118464,
      "grad_norm": 2.6877212524414062,
      "learning_rate": 5.9219380888290726e-06,
      "loss": 1.0692,
      "step": 440
    },
    {
      "epoch": 0.03563708357744601,
      "grad_norm": 3.427471399307251,
      "learning_rate": 5.9353970390309556e-06,
      "loss": 0.993,
      "step": 441
    },
    {
      "epoch": 0.03571789329077355,
      "grad_norm": 3.7671279907226562,
      "learning_rate": 5.94885598923284e-06,
      "loss": 1.1378,
      "step": 442
    },
    {
      "epoch": 0.035798703004101096,
      "grad_norm": 3.6590023040771484,
      "learning_rate": 5.962314939434725e-06,
      "loss": 1.0065,
      "step": 443
    },
    {
      "epoch": 0.03587951271742863,
      "grad_norm": 3.3577499389648438,
      "learning_rate": 5.975773889636609e-06,
      "loss": 1.0334,
      "step": 444
    },
    {
      "epoch": 0.03596032243075618,
      "grad_norm": 2.663809061050415,
      "learning_rate": 5.9892328398384936e-06,
      "loss": 1.135,
      "step": 445
    },
    {
      "epoch": 0.03604113214408372,
      "grad_norm": 2.9979307651519775,
      "learning_rate": 6.002691790040377e-06,
      "loss": 0.9237,
      "step": 446
    },
    {
      "epoch": 0.03612194185741126,
      "grad_norm": 3.3452415466308594,
      "learning_rate": 6.016150740242261e-06,
      "loss": 1.0103,
      "step": 447
    },
    {
      "epoch": 0.0362027515707388,
      "grad_norm": 3.129573106765747,
      "learning_rate": 6.029609690444146e-06,
      "loss": 0.9468,
      "step": 448
    },
    {
      "epoch": 0.036283561284066346,
      "grad_norm": 3.246513605117798,
      "learning_rate": 6.043068640646031e-06,
      "loss": 1.0707,
      "step": 449
    },
    {
      "epoch": 0.03636437099739389,
      "grad_norm": 3.0842173099517822,
      "learning_rate": 6.056527590847915e-06,
      "loss": 1.1869,
      "step": 450
    },
    {
      "epoch": 0.03644518071072143,
      "grad_norm": 3.1168720722198486,
      "learning_rate": 6.0699865410497984e-06,
      "loss": 1.1247,
      "step": 451
    },
    {
      "epoch": 0.03652599042404897,
      "grad_norm": 3.1685454845428467,
      "learning_rate": 6.083445491251682e-06,
      "loss": 1.037,
      "step": 452
    },
    {
      "epoch": 0.036606800137376515,
      "grad_norm": 3.101418972015381,
      "learning_rate": 6.096904441453567e-06,
      "loss": 1.0524,
      "step": 453
    },
    {
      "epoch": 0.03668760985070405,
      "grad_norm": 3.458691358566284,
      "learning_rate": 6.110363391655452e-06,
      "loss": 1.0948,
      "step": 454
    },
    {
      "epoch": 0.036768419564031596,
      "grad_norm": 3.8993406295776367,
      "learning_rate": 6.1238223418573364e-06,
      "loss": 1.218,
      "step": 455
    },
    {
      "epoch": 0.03684922927735914,
      "grad_norm": 3.081502676010132,
      "learning_rate": 6.1372812920592195e-06,
      "loss": 1.1364,
      "step": 456
    },
    {
      "epoch": 0.036930038990686684,
      "grad_norm": 3.4621806144714355,
      "learning_rate": 6.150740242261104e-06,
      "loss": 1.0946,
      "step": 457
    },
    {
      "epoch": 0.03701084870401422,
      "grad_norm": 3.7043323516845703,
      "learning_rate": 6.164199192462989e-06,
      "loss": 0.964,
      "step": 458
    },
    {
      "epoch": 0.037091658417341765,
      "grad_norm": 2.8379194736480713,
      "learning_rate": 6.177658142664873e-06,
      "loss": 1.0017,
      "step": 459
    },
    {
      "epoch": 0.03717246813066931,
      "grad_norm": 3.2246439456939697,
      "learning_rate": 6.1911170928667574e-06,
      "loss": 1.0454,
      "step": 460
    },
    {
      "epoch": 0.037253277843996846,
      "grad_norm": 3.018449306488037,
      "learning_rate": 6.2045760430686405e-06,
      "loss": 0.9643,
      "step": 461
    },
    {
      "epoch": 0.03733408755732439,
      "grad_norm": 2.728684425354004,
      "learning_rate": 6.218034993270525e-06,
      "loss": 1.0723,
      "step": 462
    },
    {
      "epoch": 0.037414897270651934,
      "grad_norm": 3.353196382522583,
      "learning_rate": 6.23149394347241e-06,
      "loss": 0.8839,
      "step": 463
    },
    {
      "epoch": 0.03749570698397947,
      "grad_norm": 3.0325539112091064,
      "learning_rate": 6.244952893674295e-06,
      "loss": 1.0293,
      "step": 464
    },
    {
      "epoch": 0.037576516697307015,
      "grad_norm": 3.853480815887451,
      "learning_rate": 6.258411843876178e-06,
      "loss": 0.9792,
      "step": 465
    },
    {
      "epoch": 0.03765732641063456,
      "grad_norm": 3.710716724395752,
      "learning_rate": 6.271870794078062e-06,
      "loss": 1.0661,
      "step": 466
    },
    {
      "epoch": 0.0377381361239621,
      "grad_norm": 3.857067108154297,
      "learning_rate": 6.285329744279946e-06,
      "loss": 1.0816,
      "step": 467
    },
    {
      "epoch": 0.03781894583728964,
      "grad_norm": 3.108438730239868,
      "learning_rate": 6.298788694481831e-06,
      "loss": 0.9808,
      "step": 468
    },
    {
      "epoch": 0.037899755550617184,
      "grad_norm": 2.9933106899261475,
      "learning_rate": 6.312247644683716e-06,
      "loss": 1.072,
      "step": 469
    },
    {
      "epoch": 0.03798056526394473,
      "grad_norm": 2.842607021331787,
      "learning_rate": 6.325706594885599e-06,
      "loss": 1.072,
      "step": 470
    },
    {
      "epoch": 0.038061374977272265,
      "grad_norm": 2.906888723373413,
      "learning_rate": 6.339165545087483e-06,
      "loss": 1.0432,
      "step": 471
    },
    {
      "epoch": 0.03814218469059981,
      "grad_norm": 4.220092296600342,
      "learning_rate": 6.352624495289368e-06,
      "loss": 1.1025,
      "step": 472
    },
    {
      "epoch": 0.03822299440392735,
      "grad_norm": 2.690276622772217,
      "learning_rate": 6.366083445491253e-06,
      "loss": 1.0178,
      "step": 473
    },
    {
      "epoch": 0.0383038041172549,
      "grad_norm": 3.268096685409546,
      "learning_rate": 6.379542395693137e-06,
      "loss": 0.9672,
      "step": 474
    },
    {
      "epoch": 0.038384613830582434,
      "grad_norm": 3.46506404876709,
      "learning_rate": 6.3930013458950205e-06,
      "loss": 1.0341,
      "step": 475
    },
    {
      "epoch": 0.03846542354390998,
      "grad_norm": 2.906332492828369,
      "learning_rate": 6.406460296096904e-06,
      "loss": 1.0948,
      "step": 476
    },
    {
      "epoch": 0.03854623325723752,
      "grad_norm": 3.1585898399353027,
      "learning_rate": 6.419919246298789e-06,
      "loss": 0.9946,
      "step": 477
    },
    {
      "epoch": 0.03862704297056506,
      "grad_norm": 3.1073646545410156,
      "learning_rate": 6.433378196500674e-06,
      "loss": 1.0518,
      "step": 478
    },
    {
      "epoch": 0.0387078526838926,
      "grad_norm": 2.927382230758667,
      "learning_rate": 6.4468371467025585e-06,
      "loss": 1.0208,
      "step": 479
    },
    {
      "epoch": 0.03878866239722015,
      "grad_norm": 3.1290457248687744,
      "learning_rate": 6.4602960969044415e-06,
      "loss": 1.0853,
      "step": 480
    },
    {
      "epoch": 0.03886947211054769,
      "grad_norm": 3.4820735454559326,
      "learning_rate": 6.473755047106326e-06,
      "loss": 1.1244,
      "step": 481
    },
    {
      "epoch": 0.03895028182387523,
      "grad_norm": 2.8432974815368652,
      "learning_rate": 6.48721399730821e-06,
      "loss": 0.965,
      "step": 482
    },
    {
      "epoch": 0.03903109153720277,
      "grad_norm": 3.6052823066711426,
      "learning_rate": 6.500672947510095e-06,
      "loss": 1.0264,
      "step": 483
    },
    {
      "epoch": 0.039111901250530316,
      "grad_norm": 3.049638271331787,
      "learning_rate": 6.5141318977119795e-06,
      "loss": 0.9849,
      "step": 484
    },
    {
      "epoch": 0.03919271096385785,
      "grad_norm": 3.698913812637329,
      "learning_rate": 6.5275908479138625e-06,
      "loss": 1.0338,
      "step": 485
    },
    {
      "epoch": 0.0392735206771854,
      "grad_norm": 3.1109588146209717,
      "learning_rate": 6.541049798115747e-06,
      "loss": 1.0894,
      "step": 486
    },
    {
      "epoch": 0.03935433039051294,
      "grad_norm": 3.6122915744781494,
      "learning_rate": 6.554508748317632e-06,
      "loss": 1.0949,
      "step": 487
    },
    {
      "epoch": 0.03943514010384048,
      "grad_norm": 3.2624239921569824,
      "learning_rate": 6.567967698519517e-06,
      "loss": 1.069,
      "step": 488
    },
    {
      "epoch": 0.03951594981716802,
      "grad_norm": 3.7620856761932373,
      "learning_rate": 6.5814266487214005e-06,
      "loss": 0.9941,
      "step": 489
    },
    {
      "epoch": 0.039596759530495566,
      "grad_norm": 3.5220227241516113,
      "learning_rate": 6.594885598923284e-06,
      "loss": 1.0572,
      "step": 490
    },
    {
      "epoch": 0.03967756924382311,
      "grad_norm": 2.813934564590454,
      "learning_rate": 6.608344549125168e-06,
      "loss": 1.0088,
      "step": 491
    },
    {
      "epoch": 0.03975837895715065,
      "grad_norm": 3.2144017219543457,
      "learning_rate": 6.621803499327053e-06,
      "loss": 1.0113,
      "step": 492
    },
    {
      "epoch": 0.03983918867047819,
      "grad_norm": 3.5524468421936035,
      "learning_rate": 6.635262449528938e-06,
      "loss": 1.0492,
      "step": 493
    },
    {
      "epoch": 0.039919998383805735,
      "grad_norm": 3.4819772243499756,
      "learning_rate": 6.648721399730822e-06,
      "loss": 1.0768,
      "step": 494
    },
    {
      "epoch": 0.04000080809713327,
      "grad_norm": 3.494050979614258,
      "learning_rate": 6.662180349932705e-06,
      "loss": 0.9604,
      "step": 495
    },
    {
      "epoch": 0.040081617810460816,
      "grad_norm": 3.5531656742095947,
      "learning_rate": 6.67563930013459e-06,
      "loss": 1.1349,
      "step": 496
    },
    {
      "epoch": 0.04016242752378836,
      "grad_norm": 3.5784482955932617,
      "learning_rate": 6.689098250336474e-06,
      "loss": 0.995,
      "step": 497
    },
    {
      "epoch": 0.040243237237115904,
      "grad_norm": 2.6178855895996094,
      "learning_rate": 6.702557200538359e-06,
      "loss": 1.1689,
      "step": 498
    },
    {
      "epoch": 0.04032404695044344,
      "grad_norm": 3.002977132797241,
      "learning_rate": 6.716016150740243e-06,
      "loss": 1.1346,
      "step": 499
    },
    {
      "epoch": 0.040404856663770986,
      "grad_norm": 3.476471185684204,
      "learning_rate": 6.729475100942126e-06,
      "loss": 1.0845,
      "step": 500
    },
    {
      "epoch": 0.04048566637709853,
      "grad_norm": 2.8205230236053467,
      "learning_rate": 6.742934051144011e-06,
      "loss": 0.9693,
      "step": 501
    },
    {
      "epoch": 0.04056647609042607,
      "grad_norm": 3.3831284046173096,
      "learning_rate": 6.756393001345896e-06,
      "loss": 1.0258,
      "step": 502
    },
    {
      "epoch": 0.04064728580375361,
      "grad_norm": 3.420717716217041,
      "learning_rate": 6.7698519515477805e-06,
      "loss": 1.0712,
      "step": 503
    },
    {
      "epoch": 0.040728095517081155,
      "grad_norm": 3.0955588817596436,
      "learning_rate": 6.783310901749664e-06,
      "loss": 1.113,
      "step": 504
    },
    {
      "epoch": 0.0408089052304087,
      "grad_norm": 3.0548081398010254,
      "learning_rate": 6.796769851951548e-06,
      "loss": 1.0572,
      "step": 505
    },
    {
      "epoch": 0.040889714943736236,
      "grad_norm": 2.7227015495300293,
      "learning_rate": 6.810228802153432e-06,
      "loss": 1.0681,
      "step": 506
    },
    {
      "epoch": 0.04097052465706378,
      "grad_norm": 3.3277525901794434,
      "learning_rate": 6.823687752355317e-06,
      "loss": 1.0435,
      "step": 507
    },
    {
      "epoch": 0.041051334370391324,
      "grad_norm": 3.2822682857513428,
      "learning_rate": 6.8371467025572015e-06,
      "loss": 1.0552,
      "step": 508
    },
    {
      "epoch": 0.04113214408371886,
      "grad_norm": 3.707235813140869,
      "learning_rate": 6.850605652759086e-06,
      "loss": 1.0892,
      "step": 509
    },
    {
      "epoch": 0.041212953797046405,
      "grad_norm": 3.084048271179199,
      "learning_rate": 6.864064602960969e-06,
      "loss": 1.0666,
      "step": 510
    },
    {
      "epoch": 0.04129376351037395,
      "grad_norm": 3.647761344909668,
      "learning_rate": 6.877523553162854e-06,
      "loss": 1.044,
      "step": 511
    },
    {
      "epoch": 0.041374573223701486,
      "grad_norm": 3.338500499725342,
      "learning_rate": 6.890982503364738e-06,
      "loss": 1.0436,
      "step": 512
    },
    {
      "epoch": 0.04145538293702903,
      "grad_norm": 3.313065767288208,
      "learning_rate": 6.9044414535666225e-06,
      "loss": 1.0777,
      "step": 513
    },
    {
      "epoch": 0.041536192650356574,
      "grad_norm": 3.0511531829833984,
      "learning_rate": 6.917900403768507e-06,
      "loss": 1.0664,
      "step": 514
    },
    {
      "epoch": 0.04161700236368412,
      "grad_norm": 3.0324454307556152,
      "learning_rate": 6.93135935397039e-06,
      "loss": 1.0095,
      "step": 515
    },
    {
      "epoch": 0.041697812077011655,
      "grad_norm": 3.7721328735351562,
      "learning_rate": 6.944818304172275e-06,
      "loss": 1.0013,
      "step": 516
    },
    {
      "epoch": 0.0417786217903392,
      "grad_norm": 2.761353015899658,
      "learning_rate": 6.95827725437416e-06,
      "loss": 1.0539,
      "step": 517
    },
    {
      "epoch": 0.04185943150366674,
      "grad_norm": 2.751193046569824,
      "learning_rate": 6.9717362045760435e-06,
      "loss": 1.0355,
      "step": 518
    },
    {
      "epoch": 0.04194024121699428,
      "grad_norm": 3.0338006019592285,
      "learning_rate": 6.985195154777928e-06,
      "loss": 1.1598,
      "step": 519
    },
    {
      "epoch": 0.042021050930321824,
      "grad_norm": 3.42229962348938,
      "learning_rate": 6.998654104979812e-06,
      "loss": 0.9632,
      "step": 520
    },
    {
      "epoch": 0.04210186064364937,
      "grad_norm": 2.795228958129883,
      "learning_rate": 7.012113055181696e-06,
      "loss": 1.1743,
      "step": 521
    },
    {
      "epoch": 0.04218267035697691,
      "grad_norm": 3.4653124809265137,
      "learning_rate": 7.025572005383581e-06,
      "loss": 1.0084,
      "step": 522
    },
    {
      "epoch": 0.04226348007030445,
      "grad_norm": 3.5253326892852783,
      "learning_rate": 7.039030955585465e-06,
      "loss": 1.0455,
      "step": 523
    },
    {
      "epoch": 0.04234428978363199,
      "grad_norm": 3.0897176265716553,
      "learning_rate": 7.05248990578735e-06,
      "loss": 1.1321,
      "step": 524
    },
    {
      "epoch": 0.04242509949695954,
      "grad_norm": 3.368062973022461,
      "learning_rate": 7.065948855989233e-06,
      "loss": 1.0558,
      "step": 525
    },
    {
      "epoch": 0.042505909210287074,
      "grad_norm": 2.7995309829711914,
      "learning_rate": 7.079407806191118e-06,
      "loss": 1.2026,
      "step": 526
    },
    {
      "epoch": 0.04258671892361462,
      "grad_norm": 2.430469274520874,
      "learning_rate": 7.092866756393002e-06,
      "loss": 1.0671,
      "step": 527
    },
    {
      "epoch": 0.04266752863694216,
      "grad_norm": 3.5027177333831787,
      "learning_rate": 7.106325706594886e-06,
      "loss": 1.0246,
      "step": 528
    },
    {
      "epoch": 0.0427483383502697,
      "grad_norm": 3.2911908626556396,
      "learning_rate": 7.11978465679677e-06,
      "loss": 1.0821,
      "step": 529
    },
    {
      "epoch": 0.04282914806359724,
      "grad_norm": 3.2344489097595215,
      "learning_rate": 7.133243606998654e-06,
      "loss": 0.9794,
      "step": 530
    },
    {
      "epoch": 0.04290995777692479,
      "grad_norm": 3.9854869842529297,
      "learning_rate": 7.146702557200539e-06,
      "loss": 1.0751,
      "step": 531
    },
    {
      "epoch": 0.04299076749025233,
      "grad_norm": 3.5476224422454834,
      "learning_rate": 7.1601615074024235e-06,
      "loss": 1.1167,
      "step": 532
    },
    {
      "epoch": 0.04307157720357987,
      "grad_norm": 3.635493516921997,
      "learning_rate": 7.173620457604307e-06,
      "loss": 1.1054,
      "step": 533
    },
    {
      "epoch": 0.04315238691690741,
      "grad_norm": 3.257753849029541,
      "learning_rate": 7.187079407806191e-06,
      "loss": 1.0461,
      "step": 534
    },
    {
      "epoch": 0.043233196630234956,
      "grad_norm": 3.016845464706421,
      "learning_rate": 7.200538358008076e-06,
      "loss": 1.1186,
      "step": 535
    },
    {
      "epoch": 0.04331400634356249,
      "grad_norm": 3.6017231941223145,
      "learning_rate": 7.21399730820996e-06,
      "loss": 1.1136,
      "step": 536
    },
    {
      "epoch": 0.04339481605689004,
      "grad_norm": 3.1112570762634277,
      "learning_rate": 7.2274562584118446e-06,
      "loss": 0.9677,
      "step": 537
    },
    {
      "epoch": 0.04347562577021758,
      "grad_norm": 3.657864809036255,
      "learning_rate": 7.240915208613729e-06,
      "loss": 1.1158,
      "step": 538
    },
    {
      "epoch": 0.043556435483545125,
      "grad_norm": 3.514461040496826,
      "learning_rate": 7.254374158815612e-06,
      "loss": 1.0667,
      "step": 539
    },
    {
      "epoch": 0.04363724519687266,
      "grad_norm": 2.8252127170562744,
      "learning_rate": 7.267833109017497e-06,
      "loss": 1.0977,
      "step": 540
    },
    {
      "epoch": 0.043718054910200206,
      "grad_norm": 3.032993793487549,
      "learning_rate": 7.281292059219382e-06,
      "loss": 1.0275,
      "step": 541
    },
    {
      "epoch": 0.04379886462352775,
      "grad_norm": 3.2496330738067627,
      "learning_rate": 7.2947510094212656e-06,
      "loss": 1.0126,
      "step": 542
    },
    {
      "epoch": 0.04387967433685529,
      "grad_norm": 4.544018745422363,
      "learning_rate": 7.30820995962315e-06,
      "loss": 1.0042,
      "step": 543
    },
    {
      "epoch": 0.04396048405018283,
      "grad_norm": 2.744112014770508,
      "learning_rate": 7.321668909825034e-06,
      "loss": 1.1303,
      "step": 544
    },
    {
      "epoch": 0.044041293763510375,
      "grad_norm": 3.860858201980591,
      "learning_rate": 7.335127860026918e-06,
      "loss": 1.1068,
      "step": 545
    },
    {
      "epoch": 0.04412210347683792,
      "grad_norm": 3.5854122638702393,
      "learning_rate": 7.348586810228803e-06,
      "loss": 1.0888,
      "step": 546
    },
    {
      "epoch": 0.044202913190165456,
      "grad_norm": 3.399890422821045,
      "learning_rate": 7.362045760430687e-06,
      "loss": 1.1058,
      "step": 547
    },
    {
      "epoch": 0.044283722903493,
      "grad_norm": 3.412484884262085,
      "learning_rate": 7.375504710632571e-06,
      "loss": 1.0304,
      "step": 548
    },
    {
      "epoch": 0.044364532616820544,
      "grad_norm": 3.3257498741149902,
      "learning_rate": 7.388963660834455e-06,
      "loss": 1.0559,
      "step": 549
    },
    {
      "epoch": 0.04444534233014808,
      "grad_norm": 3.3390965461730957,
      "learning_rate": 7.40242261103634e-06,
      "loss": 1.0454,
      "step": 550
    },
    {
      "epoch": 0.044526152043475625,
      "grad_norm": 3.017277479171753,
      "learning_rate": 7.415881561238224e-06,
      "loss": 1.147,
      "step": 551
    },
    {
      "epoch": 0.04460696175680317,
      "grad_norm": 3.9158685207366943,
      "learning_rate": 7.4293405114401084e-06,
      "loss": 1.0479,
      "step": 552
    },
    {
      "epoch": 0.044687771470130706,
      "grad_norm": 3.922313928604126,
      "learning_rate": 7.442799461641993e-06,
      "loss": 0.9658,
      "step": 553
    },
    {
      "epoch": 0.04476858118345825,
      "grad_norm": 3.4209396839141846,
      "learning_rate": 7.456258411843876e-06,
      "loss": 1.0282,
      "step": 554
    },
    {
      "epoch": 0.044849390896785794,
      "grad_norm": 2.8000848293304443,
      "learning_rate": 7.469717362045761e-06,
      "loss": 1.0759,
      "step": 555
    },
    {
      "epoch": 0.04493020061011334,
      "grad_norm": 3.410449504852295,
      "learning_rate": 7.4831763122476456e-06,
      "loss": 1.0831,
      "step": 556
    },
    {
      "epoch": 0.045011010323440875,
      "grad_norm": 2.632046937942505,
      "learning_rate": 7.4966352624495294e-06,
      "loss": 1.1639,
      "step": 557
    },
    {
      "epoch": 0.04509182003676842,
      "grad_norm": 2.8796796798706055,
      "learning_rate": 7.510094212651414e-06,
      "loss": 1.1227,
      "step": 558
    },
    {
      "epoch": 0.04517262975009596,
      "grad_norm": 2.673081636428833,
      "learning_rate": 7.523553162853298e-06,
      "loss": 0.9932,
      "step": 559
    },
    {
      "epoch": 0.0452534394634235,
      "grad_norm": 3.3943419456481934,
      "learning_rate": 7.537012113055182e-06,
      "loss": 0.946,
      "step": 560
    },
    {
      "epoch": 0.045334249176751044,
      "grad_norm": 2.836904764175415,
      "learning_rate": 7.550471063257067e-06,
      "loss": 0.9938,
      "step": 561
    },
    {
      "epoch": 0.04541505889007859,
      "grad_norm": 3.4832441806793213,
      "learning_rate": 7.563930013458951e-06,
      "loss": 1.1204,
      "step": 562
    },
    {
      "epoch": 0.04549586860340613,
      "grad_norm": 3.456580638885498,
      "learning_rate": 7.577388963660835e-06,
      "loss": 1.0397,
      "step": 563
    },
    {
      "epoch": 0.04557667831673367,
      "grad_norm": 3.6279962062835693,
      "learning_rate": 7.590847913862719e-06,
      "loss": 0.9863,
      "step": 564
    },
    {
      "epoch": 0.045657488030061213,
      "grad_norm": 2.8886189460754395,
      "learning_rate": 7.604306864064604e-06,
      "loss": 1.0374,
      "step": 565
    },
    {
      "epoch": 0.04573829774338876,
      "grad_norm": 3.0096817016601562,
      "learning_rate": 7.617765814266488e-06,
      "loss": 1.1097,
      "step": 566
    },
    {
      "epoch": 0.045819107456716295,
      "grad_norm": 3.125135898590088,
      "learning_rate": 7.631224764468373e-06,
      "loss": 1.0381,
      "step": 567
    },
    {
      "epoch": 0.04589991717004384,
      "grad_norm": 3.959364652633667,
      "learning_rate": 7.644683714670256e-06,
      "loss": 1.0374,
      "step": 568
    },
    {
      "epoch": 0.04598072688337138,
      "grad_norm": 3.551934242248535,
      "learning_rate": 7.658142664872141e-06,
      "loss": 1.1002,
      "step": 569
    },
    {
      "epoch": 0.046061536596698927,
      "grad_norm": 4.027700901031494,
      "learning_rate": 7.671601615074024e-06,
      "loss": 1.0443,
      "step": 570
    },
    {
      "epoch": 0.046142346310026464,
      "grad_norm": 3.957592248916626,
      "learning_rate": 7.685060565275909e-06,
      "loss": 1.1657,
      "step": 571
    },
    {
      "epoch": 0.04622315602335401,
      "grad_norm": 3.6092169284820557,
      "learning_rate": 7.698519515477793e-06,
      "loss": 1.0972,
      "step": 572
    },
    {
      "epoch": 0.04630396573668155,
      "grad_norm": 3.7678136825561523,
      "learning_rate": 7.711978465679678e-06,
      "loss": 0.9487,
      "step": 573
    },
    {
      "epoch": 0.04638477545000909,
      "grad_norm": 3.7308578491210938,
      "learning_rate": 7.725437415881561e-06,
      "loss": 1.0842,
      "step": 574
    },
    {
      "epoch": 0.04646558516333663,
      "grad_norm": 3.7531898021698,
      "learning_rate": 7.738896366083446e-06,
      "loss": 1.0672,
      "step": 575
    },
    {
      "epoch": 0.04654639487666418,
      "grad_norm": 3.43570613861084,
      "learning_rate": 7.75235531628533e-06,
      "loss": 0.9758,
      "step": 576
    },
    {
      "epoch": 0.046627204589991714,
      "grad_norm": 3.809967517852783,
      "learning_rate": 7.765814266487215e-06,
      "loss": 1.004,
      "step": 577
    },
    {
      "epoch": 0.04670801430331926,
      "grad_norm": 3.0760293006896973,
      "learning_rate": 7.7792732166891e-06,
      "loss": 1.0434,
      "step": 578
    },
    {
      "epoch": 0.0467888240166468,
      "grad_norm": 3.495680809020996,
      "learning_rate": 7.792732166890983e-06,
      "loss": 1.091,
      "step": 579
    },
    {
      "epoch": 0.046869633729974346,
      "grad_norm": 2.8191490173339844,
      "learning_rate": 7.806191117092868e-06,
      "loss": 1.1569,
      "step": 580
    },
    {
      "epoch": 0.04695044344330188,
      "grad_norm": 3.2043919563293457,
      "learning_rate": 7.819650067294752e-06,
      "loss": 1.0259,
      "step": 581
    },
    {
      "epoch": 0.04703125315662943,
      "grad_norm": 3.280008554458618,
      "learning_rate": 7.833109017496637e-06,
      "loss": 1.1297,
      "step": 582
    },
    {
      "epoch": 0.04711206286995697,
      "grad_norm": 3.444746494293213,
      "learning_rate": 7.84656796769852e-06,
      "loss": 1.0008,
      "step": 583
    },
    {
      "epoch": 0.04719287258328451,
      "grad_norm": 3.302302598953247,
      "learning_rate": 7.860026917900405e-06,
      "loss": 0.9861,
      "step": 584
    },
    {
      "epoch": 0.04727368229661205,
      "grad_norm": 3.2500033378601074,
      "learning_rate": 7.873485868102288e-06,
      "loss": 1.0184,
      "step": 585
    },
    {
      "epoch": 0.047354492009939596,
      "grad_norm": 3.9170641899108887,
      "learning_rate": 7.886944818304172e-06,
      "loss": 1.011,
      "step": 586
    },
    {
      "epoch": 0.04743530172326714,
      "grad_norm": 3.257427453994751,
      "learning_rate": 7.900403768506057e-06,
      "loss": 1.0766,
      "step": 587
    },
    {
      "epoch": 0.04751611143659468,
      "grad_norm": 2.6629252433776855,
      "learning_rate": 7.913862718707942e-06,
      "loss": 1.0352,
      "step": 588
    },
    {
      "epoch": 0.04759692114992222,
      "grad_norm": 3.212484836578369,
      "learning_rate": 7.927321668909825e-06,
      "loss": 0.9738,
      "step": 589
    },
    {
      "epoch": 0.047677730863249765,
      "grad_norm": 2.638335704803467,
      "learning_rate": 7.94078061911171e-06,
      "loss": 0.9539,
      "step": 590
    },
    {
      "epoch": 0.0477585405765773,
      "grad_norm": 2.8916549682617188,
      "learning_rate": 7.954239569313594e-06,
      "loss": 1.0734,
      "step": 591
    },
    {
      "epoch": 0.047839350289904846,
      "grad_norm": 3.582719564437866,
      "learning_rate": 7.967698519515479e-06,
      "loss": 1.1697,
      "step": 592
    },
    {
      "epoch": 0.04792016000323239,
      "grad_norm": 3.850970506668091,
      "learning_rate": 7.981157469717362e-06,
      "loss": 1.1088,
      "step": 593
    },
    {
      "epoch": 0.048000969716559934,
      "grad_norm": 2.4932446479797363,
      "learning_rate": 7.994616419919247e-06,
      "loss": 1.1026,
      "step": 594
    },
    {
      "epoch": 0.04808177942988747,
      "grad_norm": 3.447291374206543,
      "learning_rate": 8.008075370121131e-06,
      "loss": 1.0143,
      "step": 595
    },
    {
      "epoch": 0.048162589143215015,
      "grad_norm": 2.8634376525878906,
      "learning_rate": 8.021534320323016e-06,
      "loss": 1.1517,
      "step": 596
    },
    {
      "epoch": 0.04824339885654256,
      "grad_norm": 3.1959426403045654,
      "learning_rate": 8.034993270524901e-06,
      "loss": 1.0189,
      "step": 597
    },
    {
      "epoch": 0.048324208569870096,
      "grad_norm": 3.149561882019043,
      "learning_rate": 8.048452220726784e-06,
      "loss": 1.0388,
      "step": 598
    },
    {
      "epoch": 0.04840501828319764,
      "grad_norm": 3.0349740982055664,
      "learning_rate": 8.061911170928669e-06,
      "loss": 1.0017,
      "step": 599
    },
    {
      "epoch": 0.048485827996525184,
      "grad_norm": 3.0691850185394287,
      "learning_rate": 8.075370121130552e-06,
      "loss": 1.0108,
      "step": 600
    },
    {
      "epoch": 0.04856663770985272,
      "grad_norm": 3.194122791290283,
      "learning_rate": 8.088829071332436e-06,
      "loss": 1.0434,
      "step": 601
    },
    {
      "epoch": 0.048647447423180265,
      "grad_norm": 2.7453277111053467,
      "learning_rate": 8.102288021534321e-06,
      "loss": 1.0744,
      "step": 602
    },
    {
      "epoch": 0.04872825713650781,
      "grad_norm": 4.147895336151123,
      "learning_rate": 8.115746971736204e-06,
      "loss": 1.1366,
      "step": 603
    },
    {
      "epoch": 0.04880906684983535,
      "grad_norm": 3.354184865951538,
      "learning_rate": 8.129205921938089e-06,
      "loss": 1.0573,
      "step": 604
    },
    {
      "epoch": 0.04888987656316289,
      "grad_norm": 3.0165493488311768,
      "learning_rate": 8.142664872139973e-06,
      "loss": 1.0278,
      "step": 605
    },
    {
      "epoch": 0.048970686276490434,
      "grad_norm": 3.1162426471710205,
      "learning_rate": 8.156123822341858e-06,
      "loss": 1.0117,
      "step": 606
    },
    {
      "epoch": 0.04905149598981798,
      "grad_norm": 3.0014259815216064,
      "learning_rate": 8.169582772543743e-06,
      "loss": 0.9395,
      "step": 607
    },
    {
      "epoch": 0.049132305703145515,
      "grad_norm": 2.954028367996216,
      "learning_rate": 8.183041722745626e-06,
      "loss": 1.0006,
      "step": 608
    },
    {
      "epoch": 0.04921311541647306,
      "grad_norm": 4.1641764640808105,
      "learning_rate": 8.19650067294751e-06,
      "loss": 1.1276,
      "step": 609
    },
    {
      "epoch": 0.0492939251298006,
      "grad_norm": 2.9266011714935303,
      "learning_rate": 8.209959623149395e-06,
      "loss": 1.1714,
      "step": 610
    },
    {
      "epoch": 0.04937473484312815,
      "grad_norm": 3.3506548404693604,
      "learning_rate": 8.22341857335128e-06,
      "loss": 1.0914,
      "step": 611
    },
    {
      "epoch": 0.049455544556455684,
      "grad_norm": 3.3130972385406494,
      "learning_rate": 8.236877523553165e-06,
      "loss": 1.176,
      "step": 612
    },
    {
      "epoch": 0.04953635426978323,
      "grad_norm": 3.0160202980041504,
      "learning_rate": 8.250336473755048e-06,
      "loss": 0.9559,
      "step": 613
    },
    {
      "epoch": 0.04961716398311077,
      "grad_norm": 3.3474361896514893,
      "learning_rate": 8.263795423956933e-06,
      "loss": 0.9843,
      "step": 614
    },
    {
      "epoch": 0.04969797369643831,
      "grad_norm": 3.325251579284668,
      "learning_rate": 8.277254374158816e-06,
      "loss": 0.9848,
      "step": 615
    },
    {
      "epoch": 0.04977878340976585,
      "grad_norm": 3.5325891971588135,
      "learning_rate": 8.2907133243607e-06,
      "loss": 1.071,
      "step": 616
    },
    {
      "epoch": 0.0498595931230934,
      "grad_norm": 3.2238943576812744,
      "learning_rate": 8.304172274562585e-06,
      "loss": 1.0326,
      "step": 617
    },
    {
      "epoch": 0.04994040283642094,
      "grad_norm": 3.123817205429077,
      "learning_rate": 8.317631224764468e-06,
      "loss": 1.0025,
      "step": 618
    },
    {
      "epoch": 0.05002121254974848,
      "grad_norm": 3.456085205078125,
      "learning_rate": 8.331090174966353e-06,
      "loss": 1.0593,
      "step": 619
    },
    {
      "epoch": 0.05010202226307602,
      "grad_norm": 3.784614324569702,
      "learning_rate": 8.344549125168237e-06,
      "loss": 1.0484,
      "step": 620
    },
    {
      "epoch": 0.050182831976403566,
      "grad_norm": 3.1221022605895996,
      "learning_rate": 8.358008075370122e-06,
      "loss": 1.191,
      "step": 621
    },
    {
      "epoch": 0.0502636416897311,
      "grad_norm": 2.8174052238464355,
      "learning_rate": 8.371467025572007e-06,
      "loss": 1.1322,
      "step": 622
    },
    {
      "epoch": 0.05034445140305865,
      "grad_norm": 2.835664749145508,
      "learning_rate": 8.38492597577389e-06,
      "loss": 1.0163,
      "step": 623
    },
    {
      "epoch": 0.05042526111638619,
      "grad_norm": 3.2237017154693604,
      "learning_rate": 8.398384925975775e-06,
      "loss": 1.0373,
      "step": 624
    },
    {
      "epoch": 0.05050607082971373,
      "grad_norm": 3.0693700313568115,
      "learning_rate": 8.41184387617766e-06,
      "loss": 1.0844,
      "step": 625
    },
    {
      "epoch": 0.05058688054304127,
      "grad_norm": 2.6899616718292236,
      "learning_rate": 8.425302826379544e-06,
      "loss": 1.0018,
      "step": 626
    },
    {
      "epoch": 0.050667690256368816,
      "grad_norm": 2.584296226501465,
      "learning_rate": 8.438761776581429e-06,
      "loss": 1.0926,
      "step": 627
    },
    {
      "epoch": 0.05074849996969636,
      "grad_norm": 2.579590082168579,
      "learning_rate": 8.452220726783312e-06,
      "loss": 0.9834,
      "step": 628
    },
    {
      "epoch": 0.0508293096830239,
      "grad_norm": 3.64367938041687,
      "learning_rate": 8.465679676985196e-06,
      "loss": 1.0661,
      "step": 629
    },
    {
      "epoch": 0.05091011939635144,
      "grad_norm": 3.95210862159729,
      "learning_rate": 8.47913862718708e-06,
      "loss": 1.0216,
      "step": 630
    },
    {
      "epoch": 0.050990929109678985,
      "grad_norm": 3.1546990871429443,
      "learning_rate": 8.492597577388964e-06,
      "loss": 1.0854,
      "step": 631
    },
    {
      "epoch": 0.05107173882300652,
      "grad_norm": 2.9738261699676514,
      "learning_rate": 8.506056527590849e-06,
      "loss": 1.0582,
      "step": 632
    },
    {
      "epoch": 0.051152548536334067,
      "grad_norm": 3.137644052505493,
      "learning_rate": 8.519515477792732e-06,
      "loss": 1.2241,
      "step": 633
    },
    {
      "epoch": 0.05123335824966161,
      "grad_norm": 3.27915096282959,
      "learning_rate": 8.532974427994617e-06,
      "loss": 1.0469,
      "step": 634
    },
    {
      "epoch": 0.051314167962989155,
      "grad_norm": 3.2506301403045654,
      "learning_rate": 8.546433378196501e-06,
      "loss": 1.0474,
      "step": 635
    },
    {
      "epoch": 0.05139497767631669,
      "grad_norm": 2.8712782859802246,
      "learning_rate": 8.559892328398386e-06,
      "loss": 1.0206,
      "step": 636
    },
    {
      "epoch": 0.051475787389644236,
      "grad_norm": 2.924260377883911,
      "learning_rate": 8.57335127860027e-06,
      "loss": 1.1007,
      "step": 637
    },
    {
      "epoch": 0.05155659710297178,
      "grad_norm": 3.638122797012329,
      "learning_rate": 8.586810228802154e-06,
      "loss": 1.131,
      "step": 638
    },
    {
      "epoch": 0.05163740681629932,
      "grad_norm": 2.8430850505828857,
      "learning_rate": 8.600269179004038e-06,
      "loss": 1.2306,
      "step": 639
    },
    {
      "epoch": 0.05171821652962686,
      "grad_norm": 2.9707863330841064,
      "learning_rate": 8.613728129205923e-06,
      "loss": 1.0065,
      "step": 640
    },
    {
      "epoch": 0.051799026242954405,
      "grad_norm": 3.260795831680298,
      "learning_rate": 8.627187079407808e-06,
      "loss": 0.9224,
      "step": 641
    },
    {
      "epoch": 0.05187983595628194,
      "grad_norm": 3.043245553970337,
      "learning_rate": 8.64064602960969e-06,
      "loss": 1.0305,
      "step": 642
    },
    {
      "epoch": 0.051960645669609486,
      "grad_norm": 2.996899127960205,
      "learning_rate": 8.654104979811576e-06,
      "loss": 1.0677,
      "step": 643
    },
    {
      "epoch": 0.05204145538293703,
      "grad_norm": 2.490837335586548,
      "learning_rate": 8.66756393001346e-06,
      "loss": 1.0811,
      "step": 644
    },
    {
      "epoch": 0.052122265096264574,
      "grad_norm": 2.932539701461792,
      "learning_rate": 8.681022880215343e-06,
      "loss": 1.1236,
      "step": 645
    },
    {
      "epoch": 0.05220307480959211,
      "grad_norm": 2.9249560832977295,
      "learning_rate": 8.694481830417228e-06,
      "loss": 1.1811,
      "step": 646
    },
    {
      "epoch": 0.052283884522919655,
      "grad_norm": 3.095759153366089,
      "learning_rate": 8.707940780619113e-06,
      "loss": 0.989,
      "step": 647
    },
    {
      "epoch": 0.0523646942362472,
      "grad_norm": 3.0386643409729004,
      "learning_rate": 8.721399730820996e-06,
      "loss": 1.041,
      "step": 648
    },
    {
      "epoch": 0.052445503949574736,
      "grad_norm": 3.2120351791381836,
      "learning_rate": 8.73485868102288e-06,
      "loss": 1.0539,
      "step": 649
    },
    {
      "epoch": 0.05252631366290228,
      "grad_norm": 2.643620491027832,
      "learning_rate": 8.748317631224765e-06,
      "loss": 0.9998,
      "step": 650
    },
    {
      "epoch": 0.052607123376229824,
      "grad_norm": 3.5676777362823486,
      "learning_rate": 8.76177658142665e-06,
      "loss": 1.026,
      "step": 651
    },
    {
      "epoch": 0.05268793308955737,
      "grad_norm": 3.254486322402954,
      "learning_rate": 8.775235531628533e-06,
      "loss": 1.1402,
      "step": 652
    },
    {
      "epoch": 0.052768742802884905,
      "grad_norm": 3.1883718967437744,
      "learning_rate": 8.788694481830418e-06,
      "loss": 1.0265,
      "step": 653
    },
    {
      "epoch": 0.05284955251621245,
      "grad_norm": 2.793278217315674,
      "learning_rate": 8.802153432032302e-06,
      "loss": 1.0114,
      "step": 654
    },
    {
      "epoch": 0.05293036222953999,
      "grad_norm": 2.92417049407959,
      "learning_rate": 8.815612382234187e-06,
      "loss": 1.0035,
      "step": 655
    },
    {
      "epoch": 0.05301117194286753,
      "grad_norm": 3.133819580078125,
      "learning_rate": 8.829071332436072e-06,
      "loss": 1.0065,
      "step": 656
    },
    {
      "epoch": 0.053091981656195074,
      "grad_norm": 3.041851758956909,
      "learning_rate": 8.842530282637955e-06,
      "loss": 1.0923,
      "step": 657
    },
    {
      "epoch": 0.05317279136952262,
      "grad_norm": 3.1926145553588867,
      "learning_rate": 8.85598923283984e-06,
      "loss": 1.0643,
      "step": 658
    },
    {
      "epoch": 0.05325360108285016,
      "grad_norm": 2.8128445148468018,
      "learning_rate": 8.869448183041724e-06,
      "loss": 1.2255,
      "step": 659
    },
    {
      "epoch": 0.0533344107961777,
      "grad_norm": 3.1745128631591797,
      "learning_rate": 8.882907133243607e-06,
      "loss": 0.9263,
      "step": 660
    },
    {
      "epoch": 0.05341522050950524,
      "grad_norm": 2.8548998832702637,
      "learning_rate": 8.896366083445492e-06,
      "loss": 0.9909,
      "step": 661
    },
    {
      "epoch": 0.05349603022283279,
      "grad_norm": 3.219050168991089,
      "learning_rate": 8.909825033647377e-06,
      "loss": 1.1081,
      "step": 662
    },
    {
      "epoch": 0.053576839936160324,
      "grad_norm": 3.200005531311035,
      "learning_rate": 8.92328398384926e-06,
      "loss": 0.9797,
      "step": 663
    },
    {
      "epoch": 0.05365764964948787,
      "grad_norm": 3.417999744415283,
      "learning_rate": 8.936742934051144e-06,
      "loss": 0.9842,
      "step": 664
    },
    {
      "epoch": 0.05373845936281541,
      "grad_norm": 3.4707634449005127,
      "learning_rate": 8.950201884253029e-06,
      "loss": 1.0817,
      "step": 665
    },
    {
      "epoch": 0.05381926907614295,
      "grad_norm": 3.108090400695801,
      "learning_rate": 8.963660834454914e-06,
      "loss": 1.1044,
      "step": 666
    },
    {
      "epoch": 0.05390007878947049,
      "grad_norm": 3.1225290298461914,
      "learning_rate": 8.977119784656797e-06,
      "loss": 0.9562,
      "step": 667
    },
    {
      "epoch": 0.05398088850279804,
      "grad_norm": 3.255286455154419,
      "learning_rate": 8.990578734858681e-06,
      "loss": 0.9713,
      "step": 668
    },
    {
      "epoch": 0.05406169821612558,
      "grad_norm": 3.492190361022949,
      "learning_rate": 9.004037685060566e-06,
      "loss": 1.0841,
      "step": 669
    },
    {
      "epoch": 0.05414250792945312,
      "grad_norm": 3.2213480472564697,
      "learning_rate": 9.017496635262451e-06,
      "loss": 1.0451,
      "step": 670
    },
    {
      "epoch": 0.05422331764278066,
      "grad_norm": 2.792081594467163,
      "learning_rate": 9.030955585464336e-06,
      "loss": 1.005,
      "step": 671
    },
    {
      "epoch": 0.054304127356108206,
      "grad_norm": 2.712742805480957,
      "learning_rate": 9.044414535666219e-06,
      "loss": 1.0522,
      "step": 672
    },
    {
      "epoch": 0.05438493706943574,
      "grad_norm": 2.6193344593048096,
      "learning_rate": 9.057873485868103e-06,
      "loss": 1.0185,
      "step": 673
    },
    {
      "epoch": 0.05446574678276329,
      "grad_norm": 3.2356667518615723,
      "learning_rate": 9.071332436069988e-06,
      "loss": 1.0265,
      "step": 674
    },
    {
      "epoch": 0.05454655649609083,
      "grad_norm": 4.429942607879639,
      "learning_rate": 9.084791386271871e-06,
      "loss": 1.0277,
      "step": 675
    },
    {
      "epoch": 0.054627366209418375,
      "grad_norm": 3.260923385620117,
      "learning_rate": 9.098250336473756e-06,
      "loss": 1.1831,
      "step": 676
    },
    {
      "epoch": 0.05470817592274591,
      "grad_norm": 3.3523643016815186,
      "learning_rate": 9.111709286675639e-06,
      "loss": 1.0104,
      "step": 677
    },
    {
      "epoch": 0.054788985636073456,
      "grad_norm": 3.8985984325408936,
      "learning_rate": 9.125168236877523e-06,
      "loss": 1.014,
      "step": 678
    },
    {
      "epoch": 0.054869795349401,
      "grad_norm": 2.8820083141326904,
      "learning_rate": 9.138627187079408e-06,
      "loss": 1.0623,
      "step": 679
    },
    {
      "epoch": 0.05495060506272854,
      "grad_norm": 3.6060869693756104,
      "learning_rate": 9.152086137281293e-06,
      "loss": 1.0811,
      "step": 680
    },
    {
      "epoch": 0.05503141477605608,
      "grad_norm": 3.030777931213379,
      "learning_rate": 9.165545087483178e-06,
      "loss": 1.1389,
      "step": 681
    },
    {
      "epoch": 0.055112224489383625,
      "grad_norm": 3.050401210784912,
      "learning_rate": 9.17900403768506e-06,
      "loss": 1.0367,
      "step": 682
    },
    {
      "epoch": 0.05519303420271117,
      "grad_norm": 3.2267160415649414,
      "learning_rate": 9.192462987886945e-06,
      "loss": 1.0889,
      "step": 683
    },
    {
      "epoch": 0.055273843916038706,
      "grad_norm": 3.1425304412841797,
      "learning_rate": 9.20592193808883e-06,
      "loss": 0.952,
      "step": 684
    },
    {
      "epoch": 0.05535465362936625,
      "grad_norm": 3.436549425125122,
      "learning_rate": 9.219380888290715e-06,
      "loss": 1.1109,
      "step": 685
    },
    {
      "epoch": 0.055435463342693794,
      "grad_norm": 3.5831665992736816,
      "learning_rate": 9.2328398384926e-06,
      "loss": 1.0488,
      "step": 686
    },
    {
      "epoch": 0.05551627305602133,
      "grad_norm": 3.0877110958099365,
      "learning_rate": 9.246298788694482e-06,
      "loss": 1.1262,
      "step": 687
    },
    {
      "epoch": 0.055597082769348875,
      "grad_norm": 2.9710967540740967,
      "learning_rate": 9.259757738896367e-06,
      "loss": 1.0023,
      "step": 688
    },
    {
      "epoch": 0.05567789248267642,
      "grad_norm": 3.0827548503875732,
      "learning_rate": 9.273216689098252e-06,
      "loss": 1.0158,
      "step": 689
    },
    {
      "epoch": 0.055758702196003956,
      "grad_norm": 3.3601818084716797,
      "learning_rate": 9.286675639300135e-06,
      "loss": 1.1819,
      "step": 690
    },
    {
      "epoch": 0.0558395119093315,
      "grad_norm": 2.9309380054473877,
      "learning_rate": 9.30013458950202e-06,
      "loss": 1.0279,
      "step": 691
    },
    {
      "epoch": 0.055920321622659044,
      "grad_norm": 3.019272804260254,
      "learning_rate": 9.313593539703903e-06,
      "loss": 1.0785,
      "step": 692
    },
    {
      "epoch": 0.05600113133598659,
      "grad_norm": 2.882826089859009,
      "learning_rate": 9.327052489905787e-06,
      "loss": 0.9971,
      "step": 693
    },
    {
      "epoch": 0.056081941049314125,
      "grad_norm": 3.164386510848999,
      "learning_rate": 9.340511440107672e-06,
      "loss": 0.9612,
      "step": 694
    },
    {
      "epoch": 0.05616275076264167,
      "grad_norm": 3.490043878555298,
      "learning_rate": 9.353970390309557e-06,
      "loss": 1.0776,
      "step": 695
    },
    {
      "epoch": 0.05624356047596921,
      "grad_norm": 2.96453595161438,
      "learning_rate": 9.367429340511441e-06,
      "loss": 0.9989,
      "step": 696
    },
    {
      "epoch": 0.05632437018929675,
      "grad_norm": 4.331329822540283,
      "learning_rate": 9.380888290713324e-06,
      "loss": 1.2126,
      "step": 697
    },
    {
      "epoch": 0.056405179902624294,
      "grad_norm": 2.8803155422210693,
      "learning_rate": 9.39434724091521e-06,
      "loss": 1.0638,
      "step": 698
    },
    {
      "epoch": 0.05648598961595184,
      "grad_norm": 3.1566812992095947,
      "learning_rate": 9.407806191117094e-06,
      "loss": 1.0227,
      "step": 699
    },
    {
      "epoch": 0.05656679932927938,
      "grad_norm": 3.2209270000457764,
      "learning_rate": 9.421265141318979e-06,
      "loss": 1.1027,
      "step": 700
    },
    {
      "epoch": 0.05664760904260692,
      "grad_norm": 3.1682584285736084,
      "learning_rate": 9.434724091520863e-06,
      "loss": 1.2391,
      "step": 701
    },
    {
      "epoch": 0.056728418755934464,
      "grad_norm": 3.444187879562378,
      "learning_rate": 9.448183041722746e-06,
      "loss": 1.0747,
      "step": 702
    },
    {
      "epoch": 0.05680922846926201,
      "grad_norm": 3.329221725463867,
      "learning_rate": 9.461641991924631e-06,
      "loss": 1.051,
      "step": 703
    },
    {
      "epoch": 0.056890038182589545,
      "grad_norm": 3.5517077445983887,
      "learning_rate": 9.475100942126516e-06,
      "loss": 1.1431,
      "step": 704
    },
    {
      "epoch": 0.05697084789591709,
      "grad_norm": 2.881688117980957,
      "learning_rate": 9.488559892328399e-06,
      "loss": 1.1095,
      "step": 705
    },
    {
      "epoch": 0.05705165760924463,
      "grad_norm": 3.5473861694335938,
      "learning_rate": 9.502018842530283e-06,
      "loss": 0.9955,
      "step": 706
    },
    {
      "epoch": 0.05713246732257218,
      "grad_norm": 2.839627504348755,
      "learning_rate": 9.515477792732166e-06,
      "loss": 1.1872,
      "step": 707
    },
    {
      "epoch": 0.057213277035899714,
      "grad_norm": 3.1089134216308594,
      "learning_rate": 9.528936742934051e-06,
      "loss": 1.0786,
      "step": 708
    },
    {
      "epoch": 0.05729408674922726,
      "grad_norm": 2.776536703109741,
      "learning_rate": 9.542395693135936e-06,
      "loss": 1.056,
      "step": 709
    },
    {
      "epoch": 0.0573748964625548,
      "grad_norm": 3.4086854457855225,
      "learning_rate": 9.55585464333782e-06,
      "loss": 1.0237,
      "step": 710
    },
    {
      "epoch": 0.05745570617588234,
      "grad_norm": 3.041487455368042,
      "learning_rate": 9.569313593539705e-06,
      "loss": 1.0635,
      "step": 711
    },
    {
      "epoch": 0.05753651588920988,
      "grad_norm": 3.320389986038208,
      "learning_rate": 9.582772543741588e-06,
      "loss": 1.0418,
      "step": 712
    },
    {
      "epoch": 0.05761732560253743,
      "grad_norm": 3.3112857341766357,
      "learning_rate": 9.596231493943473e-06,
      "loss": 1.1571,
      "step": 713
    },
    {
      "epoch": 0.057698135315864964,
      "grad_norm": 3.4754772186279297,
      "learning_rate": 9.609690444145358e-06,
      "loss": 1.1472,
      "step": 714
    },
    {
      "epoch": 0.05777894502919251,
      "grad_norm": 3.2244203090667725,
      "learning_rate": 9.623149394347242e-06,
      "loss": 1.1125,
      "step": 715
    },
    {
      "epoch": 0.05785975474252005,
      "grad_norm": 2.8610832691192627,
      "learning_rate": 9.636608344549126e-06,
      "loss": 1.1578,
      "step": 716
    },
    {
      "epoch": 0.057940564455847596,
      "grad_norm": 3.1100966930389404,
      "learning_rate": 9.65006729475101e-06,
      "loss": 1.0516,
      "step": 717
    },
    {
      "epoch": 0.05802137416917513,
      "grad_norm": 2.962012767791748,
      "learning_rate": 9.663526244952895e-06,
      "loss": 1.084,
      "step": 718
    },
    {
      "epoch": 0.05810218388250268,
      "grad_norm": 3.183809757232666,
      "learning_rate": 9.67698519515478e-06,
      "loss": 1.0381,
      "step": 719
    },
    {
      "epoch": 0.05818299359583022,
      "grad_norm": 3.398799419403076,
      "learning_rate": 9.690444145356663e-06,
      "loss": 1.1571,
      "step": 720
    },
    {
      "epoch": 0.05826380330915776,
      "grad_norm": 3.447977066040039,
      "learning_rate": 9.703903095558547e-06,
      "loss": 1.0165,
      "step": 721
    },
    {
      "epoch": 0.0583446130224853,
      "grad_norm": 3.278087615966797,
      "learning_rate": 9.71736204576043e-06,
      "loss": 1.027,
      "step": 722
    },
    {
      "epoch": 0.058425422735812846,
      "grad_norm": 3.1923909187316895,
      "learning_rate": 9.730820995962315e-06,
      "loss": 0.9763,
      "step": 723
    },
    {
      "epoch": 0.05850623244914039,
      "grad_norm": 3.14595890045166,
      "learning_rate": 9.7442799461642e-06,
      "loss": 1.0987,
      "step": 724
    },
    {
      "epoch": 0.05858704216246793,
      "grad_norm": 2.855151653289795,
      "learning_rate": 9.757738896366085e-06,
      "loss": 1.096,
      "step": 725
    },
    {
      "epoch": 0.05866785187579547,
      "grad_norm": 3.2433483600616455,
      "learning_rate": 9.771197846567968e-06,
      "loss": 1.0199,
      "step": 726
    },
    {
      "epoch": 0.058748661589123015,
      "grad_norm": 3.5929629802703857,
      "learning_rate": 9.784656796769852e-06,
      "loss": 1.0424,
      "step": 727
    },
    {
      "epoch": 0.05882947130245055,
      "grad_norm": 3.2948145866394043,
      "learning_rate": 9.798115746971737e-06,
      "loss": 0.9823,
      "step": 728
    },
    {
      "epoch": 0.058910281015778096,
      "grad_norm": 3.665013313293457,
      "learning_rate": 9.811574697173622e-06,
      "loss": 1.0612,
      "step": 729
    },
    {
      "epoch": 0.05899109072910564,
      "grad_norm": 3.3471338748931885,
      "learning_rate": 9.825033647375506e-06,
      "loss": 1.0525,
      "step": 730
    },
    {
      "epoch": 0.05907190044243318,
      "grad_norm": 3.174917697906494,
      "learning_rate": 9.83849259757739e-06,
      "loss": 1.1074,
      "step": 731
    },
    {
      "epoch": 0.05915271015576072,
      "grad_norm": 3.52289080619812,
      "learning_rate": 9.851951547779274e-06,
      "loss": 1.0424,
      "step": 732
    },
    {
      "epoch": 0.059233519869088265,
      "grad_norm": 2.875234603881836,
      "learning_rate": 9.865410497981159e-06,
      "loss": 0.8862,
      "step": 733
    },
    {
      "epoch": 0.05931432958241581,
      "grad_norm": 3.1018879413604736,
      "learning_rate": 9.878869448183044e-06,
      "loss": 0.9249,
      "step": 734
    },
    {
      "epoch": 0.059395139295743346,
      "grad_norm": 2.894240617752075,
      "learning_rate": 9.892328398384927e-06,
      "loss": 1.0293,
      "step": 735
    },
    {
      "epoch": 0.05947594900907089,
      "grad_norm": 2.658221960067749,
      "learning_rate": 9.905787348586811e-06,
      "loss": 1.024,
      "step": 736
    },
    {
      "epoch": 0.059556758722398434,
      "grad_norm": 3.6054623126983643,
      "learning_rate": 9.919246298788694e-06,
      "loss": 1.0726,
      "step": 737
    },
    {
      "epoch": 0.05963756843572597,
      "grad_norm": 3.583958148956299,
      "learning_rate": 9.932705248990579e-06,
      "loss": 1.0371,
      "step": 738
    },
    {
      "epoch": 0.059718378149053515,
      "grad_norm": 3.387402296066284,
      "learning_rate": 9.946164199192464e-06,
      "loss": 1.0455,
      "step": 739
    },
    {
      "epoch": 0.05979918786238106,
      "grad_norm": 3.2275664806365967,
      "learning_rate": 9.959623149394348e-06,
      "loss": 1.0076,
      "step": 740
    },
    {
      "epoch": 0.0598799975757086,
      "grad_norm": 2.8620572090148926,
      "learning_rate": 9.973082099596231e-06,
      "loss": 0.9865,
      "step": 741
    },
    {
      "epoch": 0.05996080728903614,
      "grad_norm": 2.740565776824951,
      "learning_rate": 9.986541049798116e-06,
      "loss": 0.9939,
      "step": 742
    },
    {
      "epoch": 0.060041617002363684,
      "grad_norm": 2.7126121520996094,
      "learning_rate": 1e-05,
      "loss": 1.1355,
      "step": 743
    },
    {
      "epoch": 0.06012242671569123,
      "grad_norm": 3.2582151889801025,
      "learning_rate": 9.99999995718102e-06,
      "loss": 1.0385,
      "step": 744
    },
    {
      "epoch": 0.060203236429018765,
      "grad_norm": 2.884678840637207,
      "learning_rate": 9.999999828724076e-06,
      "loss": 0.9873,
      "step": 745
    },
    {
      "epoch": 0.06028404614234631,
      "grad_norm": 2.7058234214782715,
      "learning_rate": 9.999999614629171e-06,
      "loss": 1.057,
      "step": 746
    },
    {
      "epoch": 0.06036485585567385,
      "grad_norm": 3.1888225078582764,
      "learning_rate": 9.999999314896312e-06,
      "loss": 1.0373,
      "step": 747
    },
    {
      "epoch": 0.0604456655690014,
      "grad_norm": 3.0204105377197266,
      "learning_rate": 9.9999989295255e-06,
      "loss": 1.0303,
      "step": 748
    },
    {
      "epoch": 0.060526475282328934,
      "grad_norm": 2.9828920364379883,
      "learning_rate": 9.999998458516744e-06,
      "loss": 0.9333,
      "step": 749
    },
    {
      "epoch": 0.06060728499565648,
      "grad_norm": 3.6406376361846924,
      "learning_rate": 9.999997901870051e-06,
      "loss": 1.0704,
      "step": 750
    },
    {
      "epoch": 0.06068809470898402,
      "grad_norm": 3.295814037322998,
      "learning_rate": 9.999997259585433e-06,
      "loss": 1.0499,
      "step": 751
    },
    {
      "epoch": 0.06076890442231156,
      "grad_norm": 3.1430790424346924,
      "learning_rate": 9.999996531662896e-06,
      "loss": 1.0707,
      "step": 752
    },
    {
      "epoch": 0.0608497141356391,
      "grad_norm": 2.681826591491699,
      "learning_rate": 9.999995718102456e-06,
      "loss": 1.0735,
      "step": 753
    },
    {
      "epoch": 0.06093052384896665,
      "grad_norm": 3.966184616088867,
      "learning_rate": 9.999994818904128e-06,
      "loss": 0.9832,
      "step": 754
    },
    {
      "epoch": 0.061011333562294184,
      "grad_norm": 2.9914650917053223,
      "learning_rate": 9.999993834067924e-06,
      "loss": 1.0851,
      "step": 755
    },
    {
      "epoch": 0.06109214327562173,
      "grad_norm": 3.3987998962402344,
      "learning_rate": 9.999992763593863e-06,
      "loss": 1.1394,
      "step": 756
    },
    {
      "epoch": 0.06117295298894927,
      "grad_norm": 3.3105270862579346,
      "learning_rate": 9.999991607481963e-06,
      "loss": 1.0244,
      "step": 757
    },
    {
      "epoch": 0.061253762702276816,
      "grad_norm": 3.0653975009918213,
      "learning_rate": 9.999990365732244e-06,
      "loss": 1.0862,
      "step": 758
    },
    {
      "epoch": 0.06133457241560435,
      "grad_norm": 3.077054262161255,
      "learning_rate": 9.999989038344727e-06,
      "loss": 0.9286,
      "step": 759
    },
    {
      "epoch": 0.0614153821289319,
      "grad_norm": 3.343345880508423,
      "learning_rate": 9.999987625319436e-06,
      "loss": 1.1721,
      "step": 760
    },
    {
      "epoch": 0.06149619184225944,
      "grad_norm": 3.037191152572632,
      "learning_rate": 9.999986126656392e-06,
      "loss": 0.9661,
      "step": 761
    },
    {
      "epoch": 0.06157700155558698,
      "grad_norm": 3.292630195617676,
      "learning_rate": 9.999984542355623e-06,
      "loss": 1.0473,
      "step": 762
    },
    {
      "epoch": 0.06165781126891452,
      "grad_norm": 3.0494720935821533,
      "learning_rate": 9.999982872417156e-06,
      "loss": 1.1973,
      "step": 763
    },
    {
      "epoch": 0.061738620982242066,
      "grad_norm": 2.7087042331695557,
      "learning_rate": 9.99998111684102e-06,
      "loss": 1.1034,
      "step": 764
    },
    {
      "epoch": 0.06181943069556961,
      "grad_norm": 3.176063299179077,
      "learning_rate": 9.999979275627243e-06,
      "loss": 1.1785,
      "step": 765
    },
    {
      "epoch": 0.06190024040889715,
      "grad_norm": 2.815727949142456,
      "learning_rate": 9.99997734877586e-06,
      "loss": 1.1357,
      "step": 766
    },
    {
      "epoch": 0.06198105012222469,
      "grad_norm": 4.608881950378418,
      "learning_rate": 9.9999753362869e-06,
      "loss": 1.0143,
      "step": 767
    },
    {
      "epoch": 0.062061859835552236,
      "grad_norm": 3.1536128520965576,
      "learning_rate": 9.999973238160401e-06,
      "loss": 1.1959,
      "step": 768
    },
    {
      "epoch": 0.06214266954887977,
      "grad_norm": 2.6403682231903076,
      "learning_rate": 9.999971054396396e-06,
      "loss": 0.9694,
      "step": 769
    },
    {
      "epoch": 0.06222347926220732,
      "grad_norm": 2.8894197940826416,
      "learning_rate": 9.999968784994924e-06,
      "loss": 1.0606,
      "step": 770
    },
    {
      "epoch": 0.06230428897553486,
      "grad_norm": 3.4575185775756836,
      "learning_rate": 9.999966429956026e-06,
      "loss": 1.0327,
      "step": 771
    },
    {
      "epoch": 0.062385098688862405,
      "grad_norm": 3.1994760036468506,
      "learning_rate": 9.999963989279737e-06,
      "loss": 0.9079,
      "step": 772
    },
    {
      "epoch": 0.06246590840218994,
      "grad_norm": 2.893705129623413,
      "learning_rate": 9.999961462966104e-06,
      "loss": 1.0912,
      "step": 773
    },
    {
      "epoch": 0.06254671811551749,
      "grad_norm": 3.508070230484009,
      "learning_rate": 9.999958851015165e-06,
      "loss": 1.0183,
      "step": 774
    },
    {
      "epoch": 0.06262752782884502,
      "grad_norm": 3.0492401123046875,
      "learning_rate": 9.999956153426971e-06,
      "loss": 1.024,
      "step": 775
    },
    {
      "epoch": 0.06270833754217257,
      "grad_norm": 3.115318775177002,
      "learning_rate": 9.999953370201564e-06,
      "loss": 1.0539,
      "step": 776
    },
    {
      "epoch": 0.06278914725550011,
      "grad_norm": 2.9921085834503174,
      "learning_rate": 9.999950501338992e-06,
      "loss": 0.8383,
      "step": 777
    },
    {
      "epoch": 0.06286995696882765,
      "grad_norm": 3.039691209793091,
      "learning_rate": 9.999947546839306e-06,
      "loss": 1.0069,
      "step": 778
    },
    {
      "epoch": 0.0629507666821552,
      "grad_norm": 3.0498998165130615,
      "learning_rate": 9.999944506702554e-06,
      "loss": 0.9399,
      "step": 779
    },
    {
      "epoch": 0.06303157639548274,
      "grad_norm": 2.6876518726348877,
      "learning_rate": 9.99994138092879e-06,
      "loss": 1.1121,
      "step": 780
    },
    {
      "epoch": 0.06311238610881027,
      "grad_norm": 3.357180595397949,
      "learning_rate": 9.999938169518067e-06,
      "loss": 1.0398,
      "step": 781
    },
    {
      "epoch": 0.06319319582213782,
      "grad_norm": 3.427842140197754,
      "learning_rate": 9.99993487247044e-06,
      "loss": 1.0485,
      "step": 782
    },
    {
      "epoch": 0.06327400553546536,
      "grad_norm": 3.3528783321380615,
      "learning_rate": 9.999931489785965e-06,
      "loss": 0.9594,
      "step": 783
    },
    {
      "epoch": 0.06335481524879291,
      "grad_norm": 3.1690218448638916,
      "learning_rate": 9.999928021464701e-06,
      "loss": 1.0566,
      "step": 784
    },
    {
      "epoch": 0.06343562496212045,
      "grad_norm": 3.1988813877105713,
      "learning_rate": 9.999924467506707e-06,
      "loss": 1.086,
      "step": 785
    },
    {
      "epoch": 0.06351643467544799,
      "grad_norm": 2.7500905990600586,
      "learning_rate": 9.999920827912044e-06,
      "loss": 1.0032,
      "step": 786
    },
    {
      "epoch": 0.06359724438877554,
      "grad_norm": 3.1908798217773438,
      "learning_rate": 9.999917102680773e-06,
      "loss": 1.1315,
      "step": 787
    },
    {
      "epoch": 0.06367805410210307,
      "grad_norm": 3.889151096343994,
      "learning_rate": 9.99991329181296e-06,
      "loss": 1.0072,
      "step": 788
    },
    {
      "epoch": 0.06375886381543061,
      "grad_norm": 2.5109405517578125,
      "learning_rate": 9.999909395308669e-06,
      "loss": 1.0633,
      "step": 789
    },
    {
      "epoch": 0.06383967352875816,
      "grad_norm": 3.0104565620422363,
      "learning_rate": 9.999905413167965e-06,
      "loss": 1.0833,
      "step": 790
    },
    {
      "epoch": 0.0639204832420857,
      "grad_norm": 2.7210376262664795,
      "learning_rate": 9.99990134539092e-06,
      "loss": 1.0066,
      "step": 791
    },
    {
      "epoch": 0.06400129295541324,
      "grad_norm": 3.518920421600342,
      "learning_rate": 9.9998971919776e-06,
      "loss": 1.0236,
      "step": 792
    },
    {
      "epoch": 0.06408210266874079,
      "grad_norm": 2.496920108795166,
      "learning_rate": 9.99989295292808e-06,
      "loss": 1.133,
      "step": 793
    },
    {
      "epoch": 0.06416291238206832,
      "grad_norm": 3.115713357925415,
      "learning_rate": 9.999888628242429e-06,
      "loss": 1.0996,
      "step": 794
    },
    {
      "epoch": 0.06424372209539586,
      "grad_norm": 2.825153112411499,
      "learning_rate": 9.999884217920724e-06,
      "loss": 1.078,
      "step": 795
    },
    {
      "epoch": 0.06432453180872341,
      "grad_norm": 3.8622965812683105,
      "learning_rate": 9.999879721963037e-06,
      "loss": 1.0097,
      "step": 796
    },
    {
      "epoch": 0.06440534152205095,
      "grad_norm": 3.4580605030059814,
      "learning_rate": 9.999875140369448e-06,
      "loss": 1.0305,
      "step": 797
    },
    {
      "epoch": 0.06448615123537849,
      "grad_norm": 2.898871421813965,
      "learning_rate": 9.999870473140036e-06,
      "loss": 1.0498,
      "step": 798
    },
    {
      "epoch": 0.06456696094870604,
      "grad_norm": 3.083414077758789,
      "learning_rate": 9.999865720274877e-06,
      "loss": 1.1388,
      "step": 799
    },
    {
      "epoch": 0.06464777066203357,
      "grad_norm": 2.934309720993042,
      "learning_rate": 9.999860881774057e-06,
      "loss": 1.0504,
      "step": 800
    },
    {
      "epoch": 0.06472858037536112,
      "grad_norm": 2.991290807723999,
      "learning_rate": 9.999855957637657e-06,
      "loss": 1.1315,
      "step": 801
    },
    {
      "epoch": 0.06480939008868866,
      "grad_norm": 3.165949583053589,
      "learning_rate": 9.999850947865759e-06,
      "loss": 1.0564,
      "step": 802
    },
    {
      "epoch": 0.0648901998020162,
      "grad_norm": 2.813497304916382,
      "learning_rate": 9.999845852458453e-06,
      "loss": 1.1347,
      "step": 803
    },
    {
      "epoch": 0.06497100951534375,
      "grad_norm": 2.7869584560394287,
      "learning_rate": 9.99984067141582e-06,
      "loss": 1.029,
      "step": 804
    },
    {
      "epoch": 0.06505181922867129,
      "grad_norm": 3.336273670196533,
      "learning_rate": 9.99983540473796e-06,
      "loss": 1.0999,
      "step": 805
    },
    {
      "epoch": 0.06513262894199882,
      "grad_norm": 2.916585683822632,
      "learning_rate": 9.99983005242495e-06,
      "loss": 0.9567,
      "step": 806
    },
    {
      "epoch": 0.06521343865532638,
      "grad_norm": 3.108661413192749,
      "learning_rate": 9.99982461447689e-06,
      "loss": 1.1269,
      "step": 807
    },
    {
      "epoch": 0.06529424836865391,
      "grad_norm": 3.390730857849121,
      "learning_rate": 9.999819090893871e-06,
      "loss": 1.0214,
      "step": 808
    },
    {
      "epoch": 0.06537505808198145,
      "grad_norm": 3.4384777545928955,
      "learning_rate": 9.999813481675988e-06,
      "loss": 1.0744,
      "step": 809
    },
    {
      "epoch": 0.065455867795309,
      "grad_norm": 2.9382858276367188,
      "learning_rate": 9.999807786823336e-06,
      "loss": 0.938,
      "step": 810
    },
    {
      "epoch": 0.06553667750863654,
      "grad_norm": 2.9017443656921387,
      "learning_rate": 9.999802006336012e-06,
      "loss": 0.964,
      "step": 811
    },
    {
      "epoch": 0.06561748722196407,
      "grad_norm": 3.442960500717163,
      "learning_rate": 9.999796140214117e-06,
      "loss": 0.9884,
      "step": 812
    },
    {
      "epoch": 0.06569829693529163,
      "grad_norm": 3.5074703693389893,
      "learning_rate": 9.999790188457753e-06,
      "loss": 1.1886,
      "step": 813
    },
    {
      "epoch": 0.06577910664861916,
      "grad_norm": 2.902418613433838,
      "learning_rate": 9.999784151067017e-06,
      "loss": 0.9786,
      "step": 814
    },
    {
      "epoch": 0.0658599163619467,
      "grad_norm": 3.5503032207489014,
      "learning_rate": 9.999778028042015e-06,
      "loss": 1.1564,
      "step": 815
    },
    {
      "epoch": 0.06594072607527425,
      "grad_norm": 3.613726854324341,
      "learning_rate": 9.999771819382854e-06,
      "loss": 0.8653,
      "step": 816
    },
    {
      "epoch": 0.06602153578860179,
      "grad_norm": 3.3173012733459473,
      "learning_rate": 9.999765525089635e-06,
      "loss": 1.0776,
      "step": 817
    },
    {
      "epoch": 0.06610234550192934,
      "grad_norm": 2.9546091556549072,
      "learning_rate": 9.99975914516247e-06,
      "loss": 1.106,
      "step": 818
    },
    {
      "epoch": 0.06618315521525688,
      "grad_norm": 2.750858783721924,
      "learning_rate": 9.999752679601468e-06,
      "loss": 1.0307,
      "step": 819
    },
    {
      "epoch": 0.06626396492858441,
      "grad_norm": 3.2712745666503906,
      "learning_rate": 9.99974612840674e-06,
      "loss": 1.1934,
      "step": 820
    },
    {
      "epoch": 0.06634477464191196,
      "grad_norm": 3.2843518257141113,
      "learning_rate": 9.999739491578395e-06,
      "loss": 1.0981,
      "step": 821
    },
    {
      "epoch": 0.0664255843552395,
      "grad_norm": 3.065136671066284,
      "learning_rate": 9.99973276911655e-06,
      "loss": 1.0886,
      "step": 822
    },
    {
      "epoch": 0.06650639406856704,
      "grad_norm": 3.0093564987182617,
      "learning_rate": 9.99972596102132e-06,
      "loss": 0.9862,
      "step": 823
    },
    {
      "epoch": 0.06658720378189459,
      "grad_norm": 2.9246206283569336,
      "learning_rate": 9.99971906729282e-06,
      "loss": 0.9095,
      "step": 824
    },
    {
      "epoch": 0.06666801349522213,
      "grad_norm": 2.747081995010376,
      "learning_rate": 9.999712087931168e-06,
      "loss": 1.0432,
      "step": 825
    },
    {
      "epoch": 0.06674882320854966,
      "grad_norm": 3.6614139080047607,
      "learning_rate": 9.999705022936484e-06,
      "loss": 0.9202,
      "step": 826
    },
    {
      "epoch": 0.06682963292187721,
      "grad_norm": 2.832754611968994,
      "learning_rate": 9.999697872308892e-06,
      "loss": 1.1262,
      "step": 827
    },
    {
      "epoch": 0.06691044263520475,
      "grad_norm": 3.048327684402466,
      "learning_rate": 9.999690636048508e-06,
      "loss": 1.0015,
      "step": 828
    },
    {
      "epoch": 0.06699125234853229,
      "grad_norm": 2.694953680038452,
      "learning_rate": 9.999683314155462e-06,
      "loss": 1.0099,
      "step": 829
    },
    {
      "epoch": 0.06707206206185984,
      "grad_norm": 3.4013941287994385,
      "learning_rate": 9.999675906629876e-06,
      "loss": 1.0881,
      "step": 830
    },
    {
      "epoch": 0.06715287177518738,
      "grad_norm": 3.1940932273864746,
      "learning_rate": 9.999668413471878e-06,
      "loss": 1.0954,
      "step": 831
    },
    {
      "epoch": 0.06723368148851493,
      "grad_norm": 2.710059404373169,
      "learning_rate": 9.999660834681597e-06,
      "loss": 1.0692,
      "step": 832
    },
    {
      "epoch": 0.06731449120184246,
      "grad_norm": 4.080317974090576,
      "learning_rate": 9.99965317025916e-06,
      "loss": 1.0583,
      "step": 833
    },
    {
      "epoch": 0.06739530091517,
      "grad_norm": 3.435955762863159,
      "learning_rate": 9.999645420204703e-06,
      "loss": 1.1735,
      "step": 834
    },
    {
      "epoch": 0.06747611062849755,
      "grad_norm": 3.6010239124298096,
      "learning_rate": 9.999637584518356e-06,
      "loss": 1.0486,
      "step": 835
    },
    {
      "epoch": 0.06755692034182509,
      "grad_norm": 2.755072832107544,
      "learning_rate": 9.999629663200253e-06,
      "loss": 1.1078,
      "step": 836
    },
    {
      "epoch": 0.06763773005515263,
      "grad_norm": 2.9218761920928955,
      "learning_rate": 9.999621656250528e-06,
      "loss": 0.9966,
      "step": 837
    },
    {
      "epoch": 0.06771853976848018,
      "grad_norm": 2.929852247238159,
      "learning_rate": 9.999613563669322e-06,
      "loss": 1.0675,
      "step": 838
    },
    {
      "epoch": 0.06779934948180771,
      "grad_norm": 2.6583492755889893,
      "learning_rate": 9.999605385456771e-06,
      "loss": 1.0421,
      "step": 839
    },
    {
      "epoch": 0.06788015919513525,
      "grad_norm": 3.123640537261963,
      "learning_rate": 9.999597121613016e-06,
      "loss": 1.0543,
      "step": 840
    },
    {
      "epoch": 0.0679609689084628,
      "grad_norm": 3.2733051776885986,
      "learning_rate": 9.9995887721382e-06,
      "loss": 1.0589,
      "step": 841
    },
    {
      "epoch": 0.06804177862179034,
      "grad_norm": 3.10587215423584,
      "learning_rate": 9.999580337032462e-06,
      "loss": 1.0889,
      "step": 842
    },
    {
      "epoch": 0.06812258833511788,
      "grad_norm": 3.3167688846588135,
      "learning_rate": 9.99957181629595e-06,
      "loss": 1.1007,
      "step": 843
    },
    {
      "epoch": 0.06820339804844543,
      "grad_norm": 3.340961456298828,
      "learning_rate": 9.999563209928807e-06,
      "loss": 1.0252,
      "step": 844
    },
    {
      "epoch": 0.06828420776177296,
      "grad_norm": 3.1415212154388428,
      "learning_rate": 9.999554517931185e-06,
      "loss": 1.1617,
      "step": 845
    },
    {
      "epoch": 0.0683650174751005,
      "grad_norm": 2.381803035736084,
      "learning_rate": 9.999545740303228e-06,
      "loss": 1.071,
      "step": 846
    },
    {
      "epoch": 0.06844582718842805,
      "grad_norm": 3.0057709217071533,
      "learning_rate": 9.999536877045088e-06,
      "loss": 1.085,
      "step": 847
    },
    {
      "epoch": 0.06852663690175559,
      "grad_norm": 3.0332605838775635,
      "learning_rate": 9.99952792815692e-06,
      "loss": 0.9406,
      "step": 848
    },
    {
      "epoch": 0.06860744661508314,
      "grad_norm": 3.0253756046295166,
      "learning_rate": 9.99951889363887e-06,
      "loss": 1.0439,
      "step": 849
    },
    {
      "epoch": 0.06868825632841068,
      "grad_norm": 2.754026174545288,
      "learning_rate": 9.999509773491102e-06,
      "loss": 1.1399,
      "step": 850
    },
    {
      "epoch": 0.06876906604173821,
      "grad_norm": 3.0445125102996826,
      "learning_rate": 9.999500567713765e-06,
      "loss": 1.1194,
      "step": 851
    },
    {
      "epoch": 0.06884987575506576,
      "grad_norm": 2.929490566253662,
      "learning_rate": 9.999491276307018e-06,
      "loss": 1.0489,
      "step": 852
    },
    {
      "epoch": 0.0689306854683933,
      "grad_norm": 3.9951939582824707,
      "learning_rate": 9.999481899271024e-06,
      "loss": 1.1572,
      "step": 853
    },
    {
      "epoch": 0.06901149518172084,
      "grad_norm": 3.125185251235962,
      "learning_rate": 9.99947243660594e-06,
      "loss": 1.0289,
      "step": 854
    },
    {
      "epoch": 0.06909230489504839,
      "grad_norm": 3.0805680751800537,
      "learning_rate": 9.999462888311928e-06,
      "loss": 1.0773,
      "step": 855
    },
    {
      "epoch": 0.06917311460837593,
      "grad_norm": 3.3032066822052,
      "learning_rate": 9.999453254389152e-06,
      "loss": 1.1828,
      "step": 856
    },
    {
      "epoch": 0.06925392432170346,
      "grad_norm": 3.104738712310791,
      "learning_rate": 9.999443534837778e-06,
      "loss": 1.1429,
      "step": 857
    },
    {
      "epoch": 0.06933473403503101,
      "grad_norm": 3.1322181224823,
      "learning_rate": 9.999433729657972e-06,
      "loss": 1.0625,
      "step": 858
    },
    {
      "epoch": 0.06941554374835855,
      "grad_norm": 3.1211037635803223,
      "learning_rate": 9.999423838849902e-06,
      "loss": 1.1511,
      "step": 859
    },
    {
      "epoch": 0.06949635346168609,
      "grad_norm": 2.916612148284912,
      "learning_rate": 9.999413862413738e-06,
      "loss": 1.0347,
      "step": 860
    },
    {
      "epoch": 0.06957716317501364,
      "grad_norm": 2.9557623863220215,
      "learning_rate": 9.999403800349649e-06,
      "loss": 1.011,
      "step": 861
    },
    {
      "epoch": 0.06965797288834118,
      "grad_norm": 3.3609611988067627,
      "learning_rate": 9.999393652657809e-06,
      "loss": 0.9942,
      "step": 862
    },
    {
      "epoch": 0.06973878260166871,
      "grad_norm": 3.2939343452453613,
      "learning_rate": 9.999383419338392e-06,
      "loss": 0.9848,
      "step": 863
    },
    {
      "epoch": 0.06981959231499626,
      "grad_norm": 3.0276999473571777,
      "learning_rate": 9.99937310039157e-06,
      "loss": 1.013,
      "step": 864
    },
    {
      "epoch": 0.0699004020283238,
      "grad_norm": 3.5387251377105713,
      "learning_rate": 9.999362695817524e-06,
      "loss": 1.0187,
      "step": 865
    },
    {
      "epoch": 0.06998121174165135,
      "grad_norm": 2.804373025894165,
      "learning_rate": 9.999352205616431e-06,
      "loss": 1.0709,
      "step": 866
    },
    {
      "epoch": 0.07006202145497889,
      "grad_norm": 2.9923603534698486,
      "learning_rate": 9.999341629788471e-06,
      "loss": 1.1802,
      "step": 867
    },
    {
      "epoch": 0.07014283116830643,
      "grad_norm": 3.1994659900665283,
      "learning_rate": 9.999330968333823e-06,
      "loss": 1.071,
      "step": 868
    },
    {
      "epoch": 0.07022364088163398,
      "grad_norm": 2.899416923522949,
      "learning_rate": 9.99932022125267e-06,
      "loss": 1.2746,
      "step": 869
    },
    {
      "epoch": 0.07030445059496152,
      "grad_norm": 3.0050110816955566,
      "learning_rate": 9.999309388545198e-06,
      "loss": 1.0684,
      "step": 870
    },
    {
      "epoch": 0.07038526030828905,
      "grad_norm": 2.437838315963745,
      "learning_rate": 9.999298470211591e-06,
      "loss": 1.0498,
      "step": 871
    },
    {
      "epoch": 0.0704660700216166,
      "grad_norm": 3.2453885078430176,
      "learning_rate": 9.999287466252037e-06,
      "loss": 0.9166,
      "step": 872
    },
    {
      "epoch": 0.07054687973494414,
      "grad_norm": 2.980861186981201,
      "learning_rate": 9.999276376666724e-06,
      "loss": 1.1694,
      "step": 873
    },
    {
      "epoch": 0.07062768944827168,
      "grad_norm": 3.631786346435547,
      "learning_rate": 9.999265201455841e-06,
      "loss": 1.01,
      "step": 874
    },
    {
      "epoch": 0.07070849916159923,
      "grad_norm": 3.124498128890991,
      "learning_rate": 9.999253940619582e-06,
      "loss": 1.0673,
      "step": 875
    },
    {
      "epoch": 0.07078930887492677,
      "grad_norm": 3.792041301727295,
      "learning_rate": 9.999242594158136e-06,
      "loss": 0.96,
      "step": 876
    },
    {
      "epoch": 0.0708701185882543,
      "grad_norm": 3.1600944995880127,
      "learning_rate": 9.999231162071701e-06,
      "loss": 1.1406,
      "step": 877
    },
    {
      "epoch": 0.07095092830158185,
      "grad_norm": 3.425360918045044,
      "learning_rate": 9.999219644360471e-06,
      "loss": 1.0799,
      "step": 878
    },
    {
      "epoch": 0.07103173801490939,
      "grad_norm": 3.1234705448150635,
      "learning_rate": 9.999208041024644e-06,
      "loss": 1.1135,
      "step": 879
    },
    {
      "epoch": 0.07111254772823693,
      "grad_norm": 3.099431037902832,
      "learning_rate": 9.999196352064418e-06,
      "loss": 0.9932,
      "step": 880
    },
    {
      "epoch": 0.07119335744156448,
      "grad_norm": 4.025658130645752,
      "learning_rate": 9.999184577479994e-06,
      "loss": 0.9866,
      "step": 881
    },
    {
      "epoch": 0.07127416715489202,
      "grad_norm": 3.110438108444214,
      "learning_rate": 9.999172717271573e-06,
      "loss": 0.9624,
      "step": 882
    },
    {
      "epoch": 0.07135497686821957,
      "grad_norm": 2.9473679065704346,
      "learning_rate": 9.99916077143936e-06,
      "loss": 0.9861,
      "step": 883
    },
    {
      "epoch": 0.0714357865815471,
      "grad_norm": 3.1517536640167236,
      "learning_rate": 9.999148739983555e-06,
      "loss": 1.068,
      "step": 884
    },
    {
      "epoch": 0.07151659629487464,
      "grad_norm": 3.280348062515259,
      "learning_rate": 9.99913662290437e-06,
      "loss": 1.0738,
      "step": 885
    },
    {
      "epoch": 0.07159740600820219,
      "grad_norm": 2.6466808319091797,
      "learning_rate": 9.999124420202006e-06,
      "loss": 1.0165,
      "step": 886
    },
    {
      "epoch": 0.07167821572152973,
      "grad_norm": 3.676473617553711,
      "learning_rate": 9.999112131876679e-06,
      "loss": 1.0023,
      "step": 887
    },
    {
      "epoch": 0.07175902543485727,
      "grad_norm": 3.317379951477051,
      "learning_rate": 9.999099757928594e-06,
      "loss": 1.0467,
      "step": 888
    },
    {
      "epoch": 0.07183983514818482,
      "grad_norm": 3.21604061126709,
      "learning_rate": 9.999087298357965e-06,
      "loss": 0.9958,
      "step": 889
    },
    {
      "epoch": 0.07192064486151235,
      "grad_norm": 3.962308168411255,
      "learning_rate": 9.999074753165006e-06,
      "loss": 1.0039,
      "step": 890
    },
    {
      "epoch": 0.07200145457483989,
      "grad_norm": 3.2292165756225586,
      "learning_rate": 9.999062122349931e-06,
      "loss": 1.0584,
      "step": 891
    },
    {
      "epoch": 0.07208226428816744,
      "grad_norm": 3.0409646034240723,
      "learning_rate": 9.999049405912957e-06,
      "loss": 1.0484,
      "step": 892
    },
    {
      "epoch": 0.07216307400149498,
      "grad_norm": 2.9035871028900146,
      "learning_rate": 9.999036603854302e-06,
      "loss": 1.1937,
      "step": 893
    },
    {
      "epoch": 0.07224388371482252,
      "grad_norm": 3.519012451171875,
      "learning_rate": 9.999023716174183e-06,
      "loss": 0.9877,
      "step": 894
    },
    {
      "epoch": 0.07232469342815007,
      "grad_norm": 2.935112953186035,
      "learning_rate": 9.999010742872824e-06,
      "loss": 1.0983,
      "step": 895
    },
    {
      "epoch": 0.0724055031414776,
      "grad_norm": 2.5634124279022217,
      "learning_rate": 9.998997683950445e-06,
      "loss": 1.0543,
      "step": 896
    },
    {
      "epoch": 0.07248631285480515,
      "grad_norm": 3.281818389892578,
      "learning_rate": 9.99898453940727e-06,
      "loss": 1.0396,
      "step": 897
    },
    {
      "epoch": 0.07256712256813269,
      "grad_norm": 3.64463472366333,
      "learning_rate": 9.998971309243524e-06,
      "loss": 1.1441,
      "step": 898
    },
    {
      "epoch": 0.07264793228146023,
      "grad_norm": 2.7865548133850098,
      "learning_rate": 9.998957993459436e-06,
      "loss": 1.1753,
      "step": 899
    },
    {
      "epoch": 0.07272874199478778,
      "grad_norm": 3.0096328258514404,
      "learning_rate": 9.998944592055231e-06,
      "loss": 0.9944,
      "step": 900
    },
    {
      "epoch": 0.07280955170811532,
      "grad_norm": 2.683293581008911,
      "learning_rate": 9.998931105031138e-06,
      "loss": 1.1411,
      "step": 901
    },
    {
      "epoch": 0.07289036142144285,
      "grad_norm": 3.6463422775268555,
      "learning_rate": 9.99891753238739e-06,
      "loss": 1.0012,
      "step": 902
    },
    {
      "epoch": 0.0729711711347704,
      "grad_norm": 2.827143669128418,
      "learning_rate": 9.998903874124222e-06,
      "loss": 1.217,
      "step": 903
    },
    {
      "epoch": 0.07305198084809794,
      "grad_norm": 3.3450632095336914,
      "learning_rate": 9.998890130241862e-06,
      "loss": 0.903,
      "step": 904
    },
    {
      "epoch": 0.07313279056142548,
      "grad_norm": 2.957606792449951,
      "learning_rate": 9.99887630074055e-06,
      "loss": 1.077,
      "step": 905
    },
    {
      "epoch": 0.07321360027475303,
      "grad_norm": 3.0255279541015625,
      "learning_rate": 9.99886238562052e-06,
      "loss": 0.9448,
      "step": 906
    },
    {
      "epoch": 0.07329440998808057,
      "grad_norm": 3.0869853496551514,
      "learning_rate": 9.998848384882011e-06,
      "loss": 1.0614,
      "step": 907
    },
    {
      "epoch": 0.0733752197014081,
      "grad_norm": 3.3012335300445557,
      "learning_rate": 9.998834298525266e-06,
      "loss": 1.0049,
      "step": 908
    },
    {
      "epoch": 0.07345602941473565,
      "grad_norm": 3.3511605262756348,
      "learning_rate": 9.998820126550522e-06,
      "loss": 1.0177,
      "step": 909
    },
    {
      "epoch": 0.07353683912806319,
      "grad_norm": 3.8056230545043945,
      "learning_rate": 9.998805868958024e-06,
      "loss": 1.1596,
      "step": 910
    },
    {
      "epoch": 0.07361764884139073,
      "grad_norm": 3.6114351749420166,
      "learning_rate": 9.998791525748014e-06,
      "loss": 0.986,
      "step": 911
    },
    {
      "epoch": 0.07369845855471828,
      "grad_norm": 3.1725072860717773,
      "learning_rate": 9.998777096920742e-06,
      "loss": 1.0334,
      "step": 912
    },
    {
      "epoch": 0.07377926826804582,
      "grad_norm": 3.106642007827759,
      "learning_rate": 9.99876258247645e-06,
      "loss": 0.9502,
      "step": 913
    },
    {
      "epoch": 0.07386007798137337,
      "grad_norm": 3.1275382041931152,
      "learning_rate": 9.99874798241539e-06,
      "loss": 0.9597,
      "step": 914
    },
    {
      "epoch": 0.0739408876947009,
      "grad_norm": 3.19773006439209,
      "learning_rate": 9.998733296737813e-06,
      "loss": 1.1206,
      "step": 915
    },
    {
      "epoch": 0.07402169740802844,
      "grad_norm": 3.3579866886138916,
      "learning_rate": 9.998718525443965e-06,
      "loss": 1.258,
      "step": 916
    },
    {
      "epoch": 0.07410250712135599,
      "grad_norm": 3.145287275314331,
      "learning_rate": 9.998703668534104e-06,
      "loss": 1.1013,
      "step": 917
    },
    {
      "epoch": 0.07418331683468353,
      "grad_norm": 2.775991678237915,
      "learning_rate": 9.998688726008484e-06,
      "loss": 1.1676,
      "step": 918
    },
    {
      "epoch": 0.07426412654801107,
      "grad_norm": 2.732689619064331,
      "learning_rate": 9.99867369786736e-06,
      "loss": 1.0491,
      "step": 919
    },
    {
      "epoch": 0.07434493626133862,
      "grad_norm": 3.073866605758667,
      "learning_rate": 9.998658584110988e-06,
      "loss": 1.1827,
      "step": 920
    },
    {
      "epoch": 0.07442574597466615,
      "grad_norm": 2.9492454528808594,
      "learning_rate": 9.998643384739628e-06,
      "loss": 1.0028,
      "step": 921
    },
    {
      "epoch": 0.07450655568799369,
      "grad_norm": 2.6749632358551025,
      "learning_rate": 9.998628099753542e-06,
      "loss": 0.9757,
      "step": 922
    },
    {
      "epoch": 0.07458736540132124,
      "grad_norm": 2.9336817264556885,
      "learning_rate": 9.99861272915299e-06,
      "loss": 1.0409,
      "step": 923
    },
    {
      "epoch": 0.07466817511464878,
      "grad_norm": 2.6571335792541504,
      "learning_rate": 9.998597272938235e-06,
      "loss": 0.9513,
      "step": 924
    },
    {
      "epoch": 0.07474898482797632,
      "grad_norm": 2.5291669368743896,
      "learning_rate": 9.998581731109542e-06,
      "loss": 1.1042,
      "step": 925
    },
    {
      "epoch": 0.07482979454130387,
      "grad_norm": 2.519695281982422,
      "learning_rate": 9.998566103667178e-06,
      "loss": 1.0565,
      "step": 926
    },
    {
      "epoch": 0.0749106042546314,
      "grad_norm": 3.283141613006592,
      "learning_rate": 9.998550390611411e-06,
      "loss": 1.1389,
      "step": 927
    },
    {
      "epoch": 0.07499141396795894,
      "grad_norm": 3.3262479305267334,
      "learning_rate": 9.99853459194251e-06,
      "loss": 0.9328,
      "step": 928
    },
    {
      "epoch": 0.07507222368128649,
      "grad_norm": 2.841034173965454,
      "learning_rate": 9.998518707660742e-06,
      "loss": 1.0963,
      "step": 929
    },
    {
      "epoch": 0.07515303339461403,
      "grad_norm": 2.4933481216430664,
      "learning_rate": 9.998502737766385e-06,
      "loss": 1.1025,
      "step": 930
    },
    {
      "epoch": 0.07523384310794158,
      "grad_norm": 3.578246593475342,
      "learning_rate": 9.998486682259707e-06,
      "loss": 1.0138,
      "step": 931
    },
    {
      "epoch": 0.07531465282126912,
      "grad_norm": 3.591609001159668,
      "learning_rate": 9.998470541140988e-06,
      "loss": 0.9914,
      "step": 932
    },
    {
      "epoch": 0.07539546253459666,
      "grad_norm": 3.2323899269104004,
      "learning_rate": 9.9984543144105e-06,
      "loss": 1.1327,
      "step": 933
    },
    {
      "epoch": 0.0754762722479242,
      "grad_norm": 3.5043742656707764,
      "learning_rate": 9.998438002068524e-06,
      "loss": 0.9981,
      "step": 934
    },
    {
      "epoch": 0.07555708196125174,
      "grad_norm": 3.090705156326294,
      "learning_rate": 9.998421604115338e-06,
      "loss": 1.0002,
      "step": 935
    },
    {
      "epoch": 0.07563789167457928,
      "grad_norm": 3.4069063663482666,
      "learning_rate": 9.998405120551223e-06,
      "loss": 1.0914,
      "step": 936
    },
    {
      "epoch": 0.07571870138790683,
      "grad_norm": 3.153970241546631,
      "learning_rate": 9.99838855137646e-06,
      "loss": 1.0971,
      "step": 937
    },
    {
      "epoch": 0.07579951110123437,
      "grad_norm": 2.9036552906036377,
      "learning_rate": 9.998371896591337e-06,
      "loss": 1.1212,
      "step": 938
    },
    {
      "epoch": 0.0758803208145619,
      "grad_norm": 2.978816032409668,
      "learning_rate": 9.998355156196134e-06,
      "loss": 1.0859,
      "step": 939
    },
    {
      "epoch": 0.07596113052788946,
      "grad_norm": 3.104283094406128,
      "learning_rate": 9.998338330191142e-06,
      "loss": 0.9989,
      "step": 940
    },
    {
      "epoch": 0.076041940241217,
      "grad_norm": 2.811476945877075,
      "learning_rate": 9.998321418576647e-06,
      "loss": 1.0216,
      "step": 941
    },
    {
      "epoch": 0.07612274995454453,
      "grad_norm": 3.6909635066986084,
      "learning_rate": 9.998304421352938e-06,
      "loss": 1.0462,
      "step": 942
    },
    {
      "epoch": 0.07620355966787208,
      "grad_norm": 2.850799322128296,
      "learning_rate": 9.998287338520309e-06,
      "loss": 1.1047,
      "step": 943
    },
    {
      "epoch": 0.07628436938119962,
      "grad_norm": 2.847245693206787,
      "learning_rate": 9.998270170079049e-06,
      "loss": 1.0456,
      "step": 944
    },
    {
      "epoch": 0.07636517909452716,
      "grad_norm": 2.992706060409546,
      "learning_rate": 9.998252916029453e-06,
      "loss": 1.063,
      "step": 945
    },
    {
      "epoch": 0.0764459888078547,
      "grad_norm": 3.3464736938476562,
      "learning_rate": 9.99823557637182e-06,
      "loss": 1.1306,
      "step": 946
    },
    {
      "epoch": 0.07652679852118224,
      "grad_norm": 2.7523386478424072,
      "learning_rate": 9.998218151106445e-06,
      "loss": 1.0695,
      "step": 947
    },
    {
      "epoch": 0.0766076082345098,
      "grad_norm": 3.064141035079956,
      "learning_rate": 9.998200640233623e-06,
      "loss": 1.0609,
      "step": 948
    },
    {
      "epoch": 0.07668841794783733,
      "grad_norm": 3.3081955909729004,
      "learning_rate": 9.998183043753657e-06,
      "loss": 0.8739,
      "step": 949
    },
    {
      "epoch": 0.07676922766116487,
      "grad_norm": 2.872645616531372,
      "learning_rate": 9.998165361666849e-06,
      "loss": 1.1083,
      "step": 950
    },
    {
      "epoch": 0.07685003737449242,
      "grad_norm": 3.413339138031006,
      "learning_rate": 9.998147593973501e-06,
      "loss": 0.9652,
      "step": 951
    },
    {
      "epoch": 0.07693084708781996,
      "grad_norm": 3.288818120956421,
      "learning_rate": 9.998129740673918e-06,
      "loss": 1.0153,
      "step": 952
    },
    {
      "epoch": 0.0770116568011475,
      "grad_norm": 3.6216039657592773,
      "learning_rate": 9.998111801768405e-06,
      "loss": 0.9523,
      "step": 953
    },
    {
      "epoch": 0.07709246651447504,
      "grad_norm": 3.0285098552703857,
      "learning_rate": 9.998093777257267e-06,
      "loss": 1.0642,
      "step": 954
    },
    {
      "epoch": 0.07717327622780258,
      "grad_norm": 2.905763864517212,
      "learning_rate": 9.998075667140817e-06,
      "loss": 1.09,
      "step": 955
    },
    {
      "epoch": 0.07725408594113012,
      "grad_norm": 3.0776312351226807,
      "learning_rate": 9.998057471419362e-06,
      "loss": 1.0764,
      "step": 956
    },
    {
      "epoch": 0.07733489565445767,
      "grad_norm": 3.221165418624878,
      "learning_rate": 9.998039190093216e-06,
      "loss": 0.9886,
      "step": 957
    },
    {
      "epoch": 0.0774157053677852,
      "grad_norm": 2.9817917346954346,
      "learning_rate": 9.998020823162691e-06,
      "loss": 1.1292,
      "step": 958
    },
    {
      "epoch": 0.07749651508111274,
      "grad_norm": 3.237248182296753,
      "learning_rate": 9.998002370628101e-06,
      "loss": 1.0642,
      "step": 959
    },
    {
      "epoch": 0.0775773247944403,
      "grad_norm": 2.8197250366210938,
      "learning_rate": 9.99798383248976e-06,
      "loss": 1.0919,
      "step": 960
    },
    {
      "epoch": 0.07765813450776783,
      "grad_norm": 3.466444253921509,
      "learning_rate": 9.997965208747993e-06,
      "loss": 1.1215,
      "step": 961
    },
    {
      "epoch": 0.07773894422109538,
      "grad_norm": 3.26401424407959,
      "learning_rate": 9.997946499403111e-06,
      "loss": 1.0712,
      "step": 962
    },
    {
      "epoch": 0.07781975393442292,
      "grad_norm": 2.941288471221924,
      "learning_rate": 9.997927704455439e-06,
      "loss": 1.0511,
      "step": 963
    },
    {
      "epoch": 0.07790056364775046,
      "grad_norm": 3.2247517108917236,
      "learning_rate": 9.997908823905297e-06,
      "loss": 1.0109,
      "step": 964
    },
    {
      "epoch": 0.07798137336107801,
      "grad_norm": 2.8406410217285156,
      "learning_rate": 9.99788985775301e-06,
      "loss": 1.0049,
      "step": 965
    },
    {
      "epoch": 0.07806218307440554,
      "grad_norm": 3.1028623580932617,
      "learning_rate": 9.9978708059989e-06,
      "loss": 0.9748,
      "step": 966
    },
    {
      "epoch": 0.07814299278773308,
      "grad_norm": 3.500028371810913,
      "learning_rate": 9.997851668643294e-06,
      "loss": 0.9857,
      "step": 967
    },
    {
      "epoch": 0.07822380250106063,
      "grad_norm": 3.2542593479156494,
      "learning_rate": 9.997832445686521e-06,
      "loss": 1.1175,
      "step": 968
    },
    {
      "epoch": 0.07830461221438817,
      "grad_norm": 2.904175281524658,
      "learning_rate": 9.997813137128912e-06,
      "loss": 1.1044,
      "step": 969
    },
    {
      "epoch": 0.0783854219277157,
      "grad_norm": 2.8977391719818115,
      "learning_rate": 9.997793742970794e-06,
      "loss": 1.1186,
      "step": 970
    },
    {
      "epoch": 0.07846623164104326,
      "grad_norm": 2.478787660598755,
      "learning_rate": 9.997774263212503e-06,
      "loss": 1.3754,
      "step": 971
    },
    {
      "epoch": 0.0785470413543708,
      "grad_norm": 2.9613890647888184,
      "learning_rate": 9.997754697854369e-06,
      "loss": 1.0197,
      "step": 972
    },
    {
      "epoch": 0.07862785106769833,
      "grad_norm": 3.194923162460327,
      "learning_rate": 9.997735046896728e-06,
      "loss": 1.0814,
      "step": 973
    },
    {
      "epoch": 0.07870866078102588,
      "grad_norm": 2.8879735469818115,
      "learning_rate": 9.997715310339918e-06,
      "loss": 1.0357,
      "step": 974
    },
    {
      "epoch": 0.07878947049435342,
      "grad_norm": 2.7833099365234375,
      "learning_rate": 9.997695488184275e-06,
      "loss": 1.0096,
      "step": 975
    },
    {
      "epoch": 0.07887028020768096,
      "grad_norm": 2.786844253540039,
      "learning_rate": 9.997675580430141e-06,
      "loss": 1.0949,
      "step": 976
    },
    {
      "epoch": 0.07895108992100851,
      "grad_norm": 3.25620174407959,
      "learning_rate": 9.997655587077858e-06,
      "loss": 1.1849,
      "step": 977
    },
    {
      "epoch": 0.07903189963433604,
      "grad_norm": 3.4656901359558105,
      "learning_rate": 9.997635508127763e-06,
      "loss": 0.9759,
      "step": 978
    },
    {
      "epoch": 0.0791127093476636,
      "grad_norm": 3.0284624099731445,
      "learning_rate": 9.997615343580202e-06,
      "loss": 1.0203,
      "step": 979
    },
    {
      "epoch": 0.07919351906099113,
      "grad_norm": 3.2560715675354004,
      "learning_rate": 9.997595093435525e-06,
      "loss": 1.058,
      "step": 980
    },
    {
      "epoch": 0.07927432877431867,
      "grad_norm": 3.0773305892944336,
      "learning_rate": 9.997574757694073e-06,
      "loss": 1.1351,
      "step": 981
    },
    {
      "epoch": 0.07935513848764622,
      "grad_norm": 3.027306318283081,
      "learning_rate": 9.997554336356197e-06,
      "loss": 1.0502,
      "step": 982
    },
    {
      "epoch": 0.07943594820097376,
      "grad_norm": 3.059410810470581,
      "learning_rate": 9.997533829422247e-06,
      "loss": 0.9773,
      "step": 983
    },
    {
      "epoch": 0.0795167579143013,
      "grad_norm": 3.317594289779663,
      "learning_rate": 9.997513236892573e-06,
      "loss": 1.0683,
      "step": 984
    },
    {
      "epoch": 0.07959756762762885,
      "grad_norm": 2.8482987880706787,
      "learning_rate": 9.997492558767527e-06,
      "loss": 1.162,
      "step": 985
    },
    {
      "epoch": 0.07967837734095638,
      "grad_norm": 3.2453463077545166,
      "learning_rate": 9.997471795047467e-06,
      "loss": 1.0496,
      "step": 986
    },
    {
      "epoch": 0.07975918705428392,
      "grad_norm": 2.988056182861328,
      "learning_rate": 9.997450945732745e-06,
      "loss": 1.1193,
      "step": 987
    },
    {
      "epoch": 0.07983999676761147,
      "grad_norm": 3.4100332260131836,
      "learning_rate": 9.997430010823718e-06,
      "loss": 1.028,
      "step": 988
    },
    {
      "epoch": 0.07992080648093901,
      "grad_norm": 2.7486419677734375,
      "learning_rate": 9.997408990320748e-06,
      "loss": 1.0424,
      "step": 989
    },
    {
      "epoch": 0.08000161619426654,
      "grad_norm": 2.617588520050049,
      "learning_rate": 9.997387884224192e-06,
      "loss": 1.0095,
      "step": 990
    },
    {
      "epoch": 0.0800824259075941,
      "grad_norm": 3.6332125663757324,
      "learning_rate": 9.997366692534411e-06,
      "loss": 1.0239,
      "step": 991
    },
    {
      "epoch": 0.08016323562092163,
      "grad_norm": 2.829860210418701,
      "learning_rate": 9.99734541525177e-06,
      "loss": 1.1689,
      "step": 992
    },
    {
      "epoch": 0.08024404533424917,
      "grad_norm": 2.8051412105560303,
      "learning_rate": 9.997324052376632e-06,
      "loss": 1.1691,
      "step": 993
    },
    {
      "epoch": 0.08032485504757672,
      "grad_norm": 3.1492583751678467,
      "learning_rate": 9.997302603909364e-06,
      "loss": 1.0621,
      "step": 994
    },
    {
      "epoch": 0.08040566476090426,
      "grad_norm": 2.920750141143799,
      "learning_rate": 9.997281069850333e-06,
      "loss": 1.0543,
      "step": 995
    },
    {
      "epoch": 0.08048647447423181,
      "grad_norm": 3.082822322845459,
      "learning_rate": 9.997259450199908e-06,
      "loss": 1.1169,
      "step": 996
    },
    {
      "epoch": 0.08056728418755935,
      "grad_norm": 2.97986102104187,
      "learning_rate": 9.99723774495846e-06,
      "loss": 0.892,
      "step": 997
    },
    {
      "epoch": 0.08064809390088688,
      "grad_norm": 3.309490919113159,
      "learning_rate": 9.997215954126358e-06,
      "loss": 0.9458,
      "step": 998
    },
    {
      "epoch": 0.08072890361421443,
      "grad_norm": 3.0830867290496826,
      "learning_rate": 9.997194077703979e-06,
      "loss": 0.9706,
      "step": 999
    },
    {
      "epoch": 0.08080971332754197,
      "grad_norm": 2.942925214767456,
      "learning_rate": 9.997172115691693e-06,
      "loss": 1.0849,
      "step": 1000
    },
    {
      "epoch": 0.08080971332754197,
      "eval_loss": 0.8867675065994263,
      "eval_runtime": 811.2983,
      "eval_samples_per_second": 102.756,
      "eval_steps_per_second": 12.845,
      "step": 1000
    },
    {
      "epoch": 0.08089052304086951,
      "grad_norm": 3.0301153659820557,
      "learning_rate": 9.99715006808988e-06,
      "loss": 0.9936,
      "step": 1001
    },
    {
      "epoch": 0.08097133275419706,
      "grad_norm": 3.1681180000305176,
      "learning_rate": 9.997127934898917e-06,
      "loss": 0.9121,
      "step": 1002
    },
    {
      "epoch": 0.0810521424675246,
      "grad_norm": 2.993309497833252,
      "learning_rate": 9.997105716119182e-06,
      "loss": 1.0414,
      "step": 1003
    },
    {
      "epoch": 0.08113295218085213,
      "grad_norm": 3.3118903636932373,
      "learning_rate": 9.997083411751057e-06,
      "loss": 1.0731,
      "step": 1004
    },
    {
      "epoch": 0.08121376189417968,
      "grad_norm": 2.8904967308044434,
      "learning_rate": 9.997061021794923e-06,
      "loss": 0.9163,
      "step": 1005
    },
    {
      "epoch": 0.08129457160750722,
      "grad_norm": 3.920422077178955,
      "learning_rate": 9.997038546251163e-06,
      "loss": 1.1656,
      "step": 1006
    },
    {
      "epoch": 0.08137538132083476,
      "grad_norm": 3.122448444366455,
      "learning_rate": 9.997015985120162e-06,
      "loss": 1.0838,
      "step": 1007
    },
    {
      "epoch": 0.08145619103416231,
      "grad_norm": 2.9611315727233887,
      "learning_rate": 9.996993338402307e-06,
      "loss": 1.0386,
      "step": 1008
    },
    {
      "epoch": 0.08153700074748985,
      "grad_norm": 3.181063652038574,
      "learning_rate": 9.996970606097987e-06,
      "loss": 1.0529,
      "step": 1009
    },
    {
      "epoch": 0.0816178104608174,
      "grad_norm": 2.72807240486145,
      "learning_rate": 9.996947788207591e-06,
      "loss": 1.0604,
      "step": 1010
    },
    {
      "epoch": 0.08169862017414493,
      "grad_norm": 3.366835832595825,
      "learning_rate": 9.996924884731507e-06,
      "loss": 1.1703,
      "step": 1011
    },
    {
      "epoch": 0.08177942988747247,
      "grad_norm": 2.6687614917755127,
      "learning_rate": 9.99690189567013e-06,
      "loss": 1.0441,
      "step": 1012
    },
    {
      "epoch": 0.08186023960080002,
      "grad_norm": 2.6495840549468994,
      "learning_rate": 9.996878821023854e-06,
      "loss": 0.9448,
      "step": 1013
    },
    {
      "epoch": 0.08194104931412756,
      "grad_norm": 2.3618621826171875,
      "learning_rate": 9.996855660793071e-06,
      "loss": 1.1143,
      "step": 1014
    },
    {
      "epoch": 0.0820218590274551,
      "grad_norm": 3.137068510055542,
      "learning_rate": 9.996832414978183e-06,
      "loss": 1.1173,
      "step": 1015
    },
    {
      "epoch": 0.08210266874078265,
      "grad_norm": 2.6272051334381104,
      "learning_rate": 9.996809083579584e-06,
      "loss": 0.9598,
      "step": 1016
    },
    {
      "epoch": 0.08218347845411018,
      "grad_norm": 3.519331932067871,
      "learning_rate": 9.996785666597675e-06,
      "loss": 1.0637,
      "step": 1017
    },
    {
      "epoch": 0.08226428816743772,
      "grad_norm": 3.0328478813171387,
      "learning_rate": 9.996762164032857e-06,
      "loss": 1.0441,
      "step": 1018
    },
    {
      "epoch": 0.08234509788076527,
      "grad_norm": 3.2000579833984375,
      "learning_rate": 9.99673857588553e-06,
      "loss": 0.9696,
      "step": 1019
    },
    {
      "epoch": 0.08242590759409281,
      "grad_norm": 2.9278225898742676,
      "learning_rate": 9.996714902156104e-06,
      "loss": 1.0213,
      "step": 1020
    },
    {
      "epoch": 0.08250671730742035,
      "grad_norm": 2.7018179893493652,
      "learning_rate": 9.996691142844977e-06,
      "loss": 1.0357,
      "step": 1021
    },
    {
      "epoch": 0.0825875270207479,
      "grad_norm": 3.73225736618042,
      "learning_rate": 9.996667297952562e-06,
      "loss": 0.9324,
      "step": 1022
    },
    {
      "epoch": 0.08266833673407543,
      "grad_norm": 3.1943392753601074,
      "learning_rate": 9.996643367479264e-06,
      "loss": 0.9327,
      "step": 1023
    },
    {
      "epoch": 0.08274914644740297,
      "grad_norm": 3.605224847793579,
      "learning_rate": 9.996619351425495e-06,
      "loss": 1.0015,
      "step": 1024
    },
    {
      "epoch": 0.08282995616073052,
      "grad_norm": 2.8012125492095947,
      "learning_rate": 9.996595249791666e-06,
      "loss": 1.0836,
      "step": 1025
    },
    {
      "epoch": 0.08291076587405806,
      "grad_norm": 3.3962903022766113,
      "learning_rate": 9.996571062578187e-06,
      "loss": 0.9289,
      "step": 1026
    },
    {
      "epoch": 0.08299157558738561,
      "grad_norm": 4.059617519378662,
      "learning_rate": 9.996546789785476e-06,
      "loss": 1.0981,
      "step": 1027
    },
    {
      "epoch": 0.08307238530071315,
      "grad_norm": 3.8433380126953125,
      "learning_rate": 9.996522431413948e-06,
      "loss": 1.0653,
      "step": 1028
    },
    {
      "epoch": 0.08315319501404068,
      "grad_norm": 2.651620626449585,
      "learning_rate": 9.996497987464019e-06,
      "loss": 1.0522,
      "step": 1029
    },
    {
      "epoch": 0.08323400472736824,
      "grad_norm": 3.1162960529327393,
      "learning_rate": 9.996473457936107e-06,
      "loss": 0.9291,
      "step": 1030
    },
    {
      "epoch": 0.08331481444069577,
      "grad_norm": 3.1897988319396973,
      "learning_rate": 9.996448842830633e-06,
      "loss": 1.0302,
      "step": 1031
    },
    {
      "epoch": 0.08339562415402331,
      "grad_norm": 2.5630762577056885,
      "learning_rate": 9.99642414214802e-06,
      "loss": 1.1536,
      "step": 1032
    },
    {
      "epoch": 0.08347643386735086,
      "grad_norm": 3.242868423461914,
      "learning_rate": 9.99639935588869e-06,
      "loss": 1.0734,
      "step": 1033
    },
    {
      "epoch": 0.0835572435806784,
      "grad_norm": 3.3553194999694824,
      "learning_rate": 9.996374484053065e-06,
      "loss": 0.969,
      "step": 1034
    },
    {
      "epoch": 0.08363805329400593,
      "grad_norm": 3.5512146949768066,
      "learning_rate": 9.996349526641575e-06,
      "loss": 0.9276,
      "step": 1035
    },
    {
      "epoch": 0.08371886300733349,
      "grad_norm": 2.9279568195343018,
      "learning_rate": 9.996324483654646e-06,
      "loss": 1.1008,
      "step": 1036
    },
    {
      "epoch": 0.08379967272066102,
      "grad_norm": 3.434234380722046,
      "learning_rate": 9.996299355092707e-06,
      "loss": 1.1089,
      "step": 1037
    },
    {
      "epoch": 0.08388048243398856,
      "grad_norm": 2.9546749591827393,
      "learning_rate": 9.996274140956188e-06,
      "loss": 1.0548,
      "step": 1038
    },
    {
      "epoch": 0.08396129214731611,
      "grad_norm": 2.6084954738616943,
      "learning_rate": 9.996248841245519e-06,
      "loss": 1.0349,
      "step": 1039
    },
    {
      "epoch": 0.08404210186064365,
      "grad_norm": 3.8449978828430176,
      "learning_rate": 9.996223455961138e-06,
      "loss": 1.1803,
      "step": 1040
    },
    {
      "epoch": 0.08412291157397118,
      "grad_norm": 3.5063529014587402,
      "learning_rate": 9.996197985103476e-06,
      "loss": 0.9196,
      "step": 1041
    },
    {
      "epoch": 0.08420372128729874,
      "grad_norm": 3.098813533782959,
      "learning_rate": 9.99617242867297e-06,
      "loss": 1.0411,
      "step": 1042
    },
    {
      "epoch": 0.08428453100062627,
      "grad_norm": 3.460789442062378,
      "learning_rate": 9.996146786670059e-06,
      "loss": 1.0768,
      "step": 1043
    },
    {
      "epoch": 0.08436534071395382,
      "grad_norm": 3.2159345149993896,
      "learning_rate": 9.996121059095181e-06,
      "loss": 1.0289,
      "step": 1044
    },
    {
      "epoch": 0.08444615042728136,
      "grad_norm": 3.090740203857422,
      "learning_rate": 9.996095245948776e-06,
      "loss": 0.9809,
      "step": 1045
    },
    {
      "epoch": 0.0845269601406089,
      "grad_norm": 3.0718343257904053,
      "learning_rate": 9.996069347231288e-06,
      "loss": 0.9484,
      "step": 1046
    },
    {
      "epoch": 0.08460776985393645,
      "grad_norm": 2.9252374172210693,
      "learning_rate": 9.99604336294316e-06,
      "loss": 0.9557,
      "step": 1047
    },
    {
      "epoch": 0.08468857956726399,
      "grad_norm": 3.0955123901367188,
      "learning_rate": 9.996017293084837e-06,
      "loss": 1.1824,
      "step": 1048
    },
    {
      "epoch": 0.08476938928059152,
      "grad_norm": 2.992337942123413,
      "learning_rate": 9.995991137656763e-06,
      "loss": 1.0964,
      "step": 1049
    },
    {
      "epoch": 0.08485019899391907,
      "grad_norm": 2.545712471008301,
      "learning_rate": 9.99596489665939e-06,
      "loss": 1.0103,
      "step": 1050
    },
    {
      "epoch": 0.08493100870724661,
      "grad_norm": 3.002333402633667,
      "learning_rate": 9.995938570093165e-06,
      "loss": 1.0722,
      "step": 1051
    },
    {
      "epoch": 0.08501181842057415,
      "grad_norm": 2.9183366298675537,
      "learning_rate": 9.995912157958539e-06,
      "loss": 0.9306,
      "step": 1052
    },
    {
      "epoch": 0.0850926281339017,
      "grad_norm": 2.8887429237365723,
      "learning_rate": 9.995885660255966e-06,
      "loss": 1.0188,
      "step": 1053
    },
    {
      "epoch": 0.08517343784722924,
      "grad_norm": 2.960714340209961,
      "learning_rate": 9.9958590769859e-06,
      "loss": 1.017,
      "step": 1054
    },
    {
      "epoch": 0.08525424756055677,
      "grad_norm": 3.45218825340271,
      "learning_rate": 9.995832408148791e-06,
      "loss": 1.0186,
      "step": 1055
    },
    {
      "epoch": 0.08533505727388432,
      "grad_norm": 3.007883071899414,
      "learning_rate": 9.995805653745103e-06,
      "loss": 0.9975,
      "step": 1056
    },
    {
      "epoch": 0.08541586698721186,
      "grad_norm": 2.8929126262664795,
      "learning_rate": 9.99577881377529e-06,
      "loss": 0.993,
      "step": 1057
    },
    {
      "epoch": 0.0854966767005394,
      "grad_norm": 2.9702260494232178,
      "learning_rate": 9.995751888239814e-06,
      "loss": 1.1256,
      "step": 1058
    },
    {
      "epoch": 0.08557748641386695,
      "grad_norm": 2.7810113430023193,
      "learning_rate": 9.995724877139133e-06,
      "loss": 1.0006,
      "step": 1059
    },
    {
      "epoch": 0.08565829612719449,
      "grad_norm": 2.9695522785186768,
      "learning_rate": 9.995697780473711e-06,
      "loss": 1.0215,
      "step": 1060
    },
    {
      "epoch": 0.08573910584052204,
      "grad_norm": 2.9190425872802734,
      "learning_rate": 9.995670598244017e-06,
      "loss": 1.0327,
      "step": 1061
    },
    {
      "epoch": 0.08581991555384957,
      "grad_norm": 3.2689478397369385,
      "learning_rate": 9.995643330450508e-06,
      "loss": 1.0618,
      "step": 1062
    },
    {
      "epoch": 0.08590072526717711,
      "grad_norm": 3.0045993328094482,
      "learning_rate": 9.995615977093656e-06,
      "loss": 1.0528,
      "step": 1063
    },
    {
      "epoch": 0.08598153498050466,
      "grad_norm": 3.6134586334228516,
      "learning_rate": 9.99558853817393e-06,
      "loss": 1.0186,
      "step": 1064
    },
    {
      "epoch": 0.0860623446938322,
      "grad_norm": 2.8262598514556885,
      "learning_rate": 9.9955610136918e-06,
      "loss": 1.1177,
      "step": 1065
    },
    {
      "epoch": 0.08614315440715974,
      "grad_norm": 2.9657886028289795,
      "learning_rate": 9.995533403647733e-06,
      "loss": 0.964,
      "step": 1066
    },
    {
      "epoch": 0.08622396412048729,
      "grad_norm": 2.782003402709961,
      "learning_rate": 9.995505708042206e-06,
      "loss": 1.0412,
      "step": 1067
    },
    {
      "epoch": 0.08630477383381482,
      "grad_norm": 3.0706048011779785,
      "learning_rate": 9.995477926875692e-06,
      "loss": 1.042,
      "step": 1068
    },
    {
      "epoch": 0.08638558354714236,
      "grad_norm": 3.0264039039611816,
      "learning_rate": 9.995450060148668e-06,
      "loss": 1.0444,
      "step": 1069
    },
    {
      "epoch": 0.08646639326046991,
      "grad_norm": 3.288853406906128,
      "learning_rate": 9.995422107861612e-06,
      "loss": 0.9645,
      "step": 1070
    },
    {
      "epoch": 0.08654720297379745,
      "grad_norm": 2.773499011993408,
      "learning_rate": 9.995394070015e-06,
      "loss": 1.1742,
      "step": 1071
    },
    {
      "epoch": 0.08662801268712499,
      "grad_norm": 3.0497007369995117,
      "learning_rate": 9.995365946609312e-06,
      "loss": 1.0304,
      "step": 1072
    },
    {
      "epoch": 0.08670882240045254,
      "grad_norm": 3.416323661804199,
      "learning_rate": 9.995337737645034e-06,
      "loss": 1.1184,
      "step": 1073
    },
    {
      "epoch": 0.08678963211378007,
      "grad_norm": 2.6845903396606445,
      "learning_rate": 9.995309443122644e-06,
      "loss": 1.0445,
      "step": 1074
    },
    {
      "epoch": 0.08687044182710763,
      "grad_norm": 3.158496141433716,
      "learning_rate": 9.99528106304263e-06,
      "loss": 1.0373,
      "step": 1075
    },
    {
      "epoch": 0.08695125154043516,
      "grad_norm": 2.750772714614868,
      "learning_rate": 9.995252597405478e-06,
      "loss": 1.1135,
      "step": 1076
    },
    {
      "epoch": 0.0870320612537627,
      "grad_norm": 3.0192668437957764,
      "learning_rate": 9.995224046211672e-06,
      "loss": 0.9874,
      "step": 1077
    },
    {
      "epoch": 0.08711287096709025,
      "grad_norm": 3.3132712841033936,
      "learning_rate": 9.995195409461705e-06,
      "loss": 0.9698,
      "step": 1078
    },
    {
      "epoch": 0.08719368068041779,
      "grad_norm": 2.6165432929992676,
      "learning_rate": 9.995166687156065e-06,
      "loss": 0.9945,
      "step": 1079
    },
    {
      "epoch": 0.08727449039374532,
      "grad_norm": 3.113999128341675,
      "learning_rate": 9.995137879295246e-06,
      "loss": 0.9752,
      "step": 1080
    },
    {
      "epoch": 0.08735530010707288,
      "grad_norm": 3.393367052078247,
      "learning_rate": 9.995108985879742e-06,
      "loss": 1.1017,
      "step": 1081
    },
    {
      "epoch": 0.08743610982040041,
      "grad_norm": 3.375389575958252,
      "learning_rate": 9.995080006910044e-06,
      "loss": 0.983,
      "step": 1082
    },
    {
      "epoch": 0.08751691953372795,
      "grad_norm": 3.390043258666992,
      "learning_rate": 9.995050942386653e-06,
      "loss": 1.0365,
      "step": 1083
    },
    {
      "epoch": 0.0875977292470555,
      "grad_norm": 3.4741687774658203,
      "learning_rate": 9.995021792310063e-06,
      "loss": 1.0178,
      "step": 1084
    },
    {
      "epoch": 0.08767853896038304,
      "grad_norm": 2.5700831413269043,
      "learning_rate": 9.994992556680774e-06,
      "loss": 1.1193,
      "step": 1085
    },
    {
      "epoch": 0.08775934867371057,
      "grad_norm": 3.1229987144470215,
      "learning_rate": 9.994963235499288e-06,
      "loss": 1.0312,
      "step": 1086
    },
    {
      "epoch": 0.08784015838703813,
      "grad_norm": 2.904766082763672,
      "learning_rate": 9.994933828766108e-06,
      "loss": 0.9871,
      "step": 1087
    },
    {
      "epoch": 0.08792096810036566,
      "grad_norm": 2.8027563095092773,
      "learning_rate": 9.994904336481735e-06,
      "loss": 1.0225,
      "step": 1088
    },
    {
      "epoch": 0.0880017778136932,
      "grad_norm": 2.766493558883667,
      "learning_rate": 9.994874758646676e-06,
      "loss": 1.041,
      "step": 1089
    },
    {
      "epoch": 0.08808258752702075,
      "grad_norm": 3.0756635665893555,
      "learning_rate": 9.994845095261436e-06,
      "loss": 1.1068,
      "step": 1090
    },
    {
      "epoch": 0.08816339724034829,
      "grad_norm": 3.4534332752227783,
      "learning_rate": 9.994815346326524e-06,
      "loss": 1.0894,
      "step": 1091
    },
    {
      "epoch": 0.08824420695367584,
      "grad_norm": 2.903165817260742,
      "learning_rate": 9.99478551184245e-06,
      "loss": 0.9894,
      "step": 1092
    },
    {
      "epoch": 0.08832501666700338,
      "grad_norm": 3.0376791954040527,
      "learning_rate": 9.994755591809726e-06,
      "loss": 0.8919,
      "step": 1093
    },
    {
      "epoch": 0.08840582638033091,
      "grad_norm": 3.136165142059326,
      "learning_rate": 9.994725586228861e-06,
      "loss": 1.065,
      "step": 1094
    },
    {
      "epoch": 0.08848663609365846,
      "grad_norm": 3.06294322013855,
      "learning_rate": 9.994695495100372e-06,
      "loss": 0.993,
      "step": 1095
    },
    {
      "epoch": 0.088567445806986,
      "grad_norm": 2.9708285331726074,
      "learning_rate": 9.994665318424774e-06,
      "loss": 0.9118,
      "step": 1096
    },
    {
      "epoch": 0.08864825552031354,
      "grad_norm": 3.305607557296753,
      "learning_rate": 9.994635056202584e-06,
      "loss": 1.047,
      "step": 1097
    },
    {
      "epoch": 0.08872906523364109,
      "grad_norm": 2.799258232116699,
      "learning_rate": 9.994604708434318e-06,
      "loss": 0.9901,
      "step": 1098
    },
    {
      "epoch": 0.08880987494696863,
      "grad_norm": 3.0276477336883545,
      "learning_rate": 9.994574275120497e-06,
      "loss": 1.0149,
      "step": 1099
    },
    {
      "epoch": 0.08889068466029616,
      "grad_norm": 2.715073823928833,
      "learning_rate": 9.994543756261644e-06,
      "loss": 1.0994,
      "step": 1100
    },
    {
      "epoch": 0.08897149437362371,
      "grad_norm": 2.9331631660461426,
      "learning_rate": 9.99451315185828e-06,
      "loss": 1.0786,
      "step": 1101
    },
    {
      "epoch": 0.08905230408695125,
      "grad_norm": 3.1864383220672607,
      "learning_rate": 9.99448246191093e-06,
      "loss": 0.9758,
      "step": 1102
    },
    {
      "epoch": 0.08913311380027879,
      "grad_norm": 3.592210054397583,
      "learning_rate": 9.994451686420117e-06,
      "loss": 0.9905,
      "step": 1103
    },
    {
      "epoch": 0.08921392351360634,
      "grad_norm": 2.7969889640808105,
      "learning_rate": 9.994420825386373e-06,
      "loss": 0.9889,
      "step": 1104
    },
    {
      "epoch": 0.08929473322693388,
      "grad_norm": 3.270756721496582,
      "learning_rate": 9.994389878810222e-06,
      "loss": 1.0471,
      "step": 1105
    },
    {
      "epoch": 0.08937554294026141,
      "grad_norm": 3.086674213409424,
      "learning_rate": 9.994358846692197e-06,
      "loss": 1.0376,
      "step": 1106
    },
    {
      "epoch": 0.08945635265358896,
      "grad_norm": 4.83221960067749,
      "learning_rate": 9.994327729032827e-06,
      "loss": 1.1289,
      "step": 1107
    },
    {
      "epoch": 0.0895371623669165,
      "grad_norm": 2.469668388366699,
      "learning_rate": 9.994296525832647e-06,
      "loss": 1.1292,
      "step": 1108
    },
    {
      "epoch": 0.08961797208024405,
      "grad_norm": 2.915151357650757,
      "learning_rate": 9.99426523709219e-06,
      "loss": 0.9702,
      "step": 1109
    },
    {
      "epoch": 0.08969878179357159,
      "grad_norm": 2.6539580821990967,
      "learning_rate": 9.994233862811996e-06,
      "loss": 1.0733,
      "step": 1110
    },
    {
      "epoch": 0.08977959150689913,
      "grad_norm": 3.2237725257873535,
      "learning_rate": 9.994202402992595e-06,
      "loss": 1.0379,
      "step": 1111
    },
    {
      "epoch": 0.08986040122022668,
      "grad_norm": 2.849095344543457,
      "learning_rate": 9.994170857634531e-06,
      "loss": 1.1208,
      "step": 1112
    },
    {
      "epoch": 0.08994121093355421,
      "grad_norm": 2.9023728370666504,
      "learning_rate": 9.994139226738345e-06,
      "loss": 1.1777,
      "step": 1113
    },
    {
      "epoch": 0.09002202064688175,
      "grad_norm": 3.650582790374756,
      "learning_rate": 9.994107510304576e-06,
      "loss": 1.0579,
      "step": 1114
    },
    {
      "epoch": 0.0901028303602093,
      "grad_norm": 3.201936721801758,
      "learning_rate": 9.994075708333767e-06,
      "loss": 1.166,
      "step": 1115
    },
    {
      "epoch": 0.09018364007353684,
      "grad_norm": 3.080941915512085,
      "learning_rate": 9.994043820826465e-06,
      "loss": 0.891,
      "step": 1116
    },
    {
      "epoch": 0.09026444978686438,
      "grad_norm": 3.0385451316833496,
      "learning_rate": 9.994011847783213e-06,
      "loss": 1.0791,
      "step": 1117
    },
    {
      "epoch": 0.09034525950019193,
      "grad_norm": 3.177011251449585,
      "learning_rate": 9.993979789204565e-06,
      "loss": 1.0181,
      "step": 1118
    },
    {
      "epoch": 0.09042606921351946,
      "grad_norm": 3.9674606323242188,
      "learning_rate": 9.993947645091063e-06,
      "loss": 1.1167,
      "step": 1119
    },
    {
      "epoch": 0.090506878926847,
      "grad_norm": 2.9916577339172363,
      "learning_rate": 9.993915415443259e-06,
      "loss": 0.989,
      "step": 1120
    },
    {
      "epoch": 0.09058768864017455,
      "grad_norm": 2.9214396476745605,
      "learning_rate": 9.993883100261707e-06,
      "loss": 0.9741,
      "step": 1121
    },
    {
      "epoch": 0.09066849835350209,
      "grad_norm": 3.219503164291382,
      "learning_rate": 9.993850699546962e-06,
      "loss": 1.0711,
      "step": 1122
    },
    {
      "epoch": 0.09074930806682964,
      "grad_norm": 3.2714686393737793,
      "learning_rate": 9.993818213299574e-06,
      "loss": 0.9782,
      "step": 1123
    },
    {
      "epoch": 0.09083011778015718,
      "grad_norm": 3.129592180252075,
      "learning_rate": 9.993785641520104e-06,
      "loss": 0.9803,
      "step": 1124
    },
    {
      "epoch": 0.09091092749348471,
      "grad_norm": 3.368413209915161,
      "learning_rate": 9.993752984209106e-06,
      "loss": 0.9238,
      "step": 1125
    },
    {
      "epoch": 0.09099173720681226,
      "grad_norm": 3.0969696044921875,
      "learning_rate": 9.993720241367144e-06,
      "loss": 1.0342,
      "step": 1126
    },
    {
      "epoch": 0.0910725469201398,
      "grad_norm": 3.028538942337036,
      "learning_rate": 9.993687412994774e-06,
      "loss": 1.059,
      "step": 1127
    },
    {
      "epoch": 0.09115335663346734,
      "grad_norm": 2.712517023086548,
      "learning_rate": 9.993654499092563e-06,
      "loss": 0.9761,
      "step": 1128
    },
    {
      "epoch": 0.09123416634679489,
      "grad_norm": 3.0142993927001953,
      "learning_rate": 9.993621499661069e-06,
      "loss": 1.0237,
      "step": 1129
    },
    {
      "epoch": 0.09131497606012243,
      "grad_norm": 3.3147966861724854,
      "learning_rate": 9.993588414700862e-06,
      "loss": 1.0907,
      "step": 1130
    },
    {
      "epoch": 0.09139578577344996,
      "grad_norm": 3.217470407485962,
      "learning_rate": 9.993555244212508e-06,
      "loss": 1.0408,
      "step": 1131
    },
    {
      "epoch": 0.09147659548677751,
      "grad_norm": 3.2648322582244873,
      "learning_rate": 9.993521988196572e-06,
      "loss": 1.0677,
      "step": 1132
    },
    {
      "epoch": 0.09155740520010505,
      "grad_norm": 3.304222583770752,
      "learning_rate": 9.993488646653626e-06,
      "loss": 0.8679,
      "step": 1133
    },
    {
      "epoch": 0.09163821491343259,
      "grad_norm": 2.758460521697998,
      "learning_rate": 9.993455219584242e-06,
      "loss": 1.0116,
      "step": 1134
    },
    {
      "epoch": 0.09171902462676014,
      "grad_norm": 2.5346457958221436,
      "learning_rate": 9.993421706988991e-06,
      "loss": 1.1609,
      "step": 1135
    },
    {
      "epoch": 0.09179983434008768,
      "grad_norm": 3.143198251724243,
      "learning_rate": 9.993388108868447e-06,
      "loss": 0.945,
      "step": 1136
    },
    {
      "epoch": 0.09188064405341521,
      "grad_norm": 3.185825824737549,
      "learning_rate": 9.993354425223186e-06,
      "loss": 1.0233,
      "step": 1137
    },
    {
      "epoch": 0.09196145376674277,
      "grad_norm": 2.7583227157592773,
      "learning_rate": 9.993320656053785e-06,
      "loss": 1.1102,
      "step": 1138
    },
    {
      "epoch": 0.0920422634800703,
      "grad_norm": 2.8564047813415527,
      "learning_rate": 9.993286801360822e-06,
      "loss": 0.9633,
      "step": 1139
    },
    {
      "epoch": 0.09212307319339785,
      "grad_norm": 3.02797269821167,
      "learning_rate": 9.993252861144875e-06,
      "loss": 0.9511,
      "step": 1140
    },
    {
      "epoch": 0.09220388290672539,
      "grad_norm": 2.9829790592193604,
      "learning_rate": 9.993218835406531e-06,
      "loss": 0.9639,
      "step": 1141
    },
    {
      "epoch": 0.09228469262005293,
      "grad_norm": 2.6532115936279297,
      "learning_rate": 9.993184724146367e-06,
      "loss": 0.988,
      "step": 1142
    },
    {
      "epoch": 0.09236550233338048,
      "grad_norm": 3.163055896759033,
      "learning_rate": 9.993150527364969e-06,
      "loss": 1.0345,
      "step": 1143
    },
    {
      "epoch": 0.09244631204670802,
      "grad_norm": 2.615417957305908,
      "learning_rate": 9.993116245062923e-06,
      "loss": 1.1074,
      "step": 1144
    },
    {
      "epoch": 0.09252712176003555,
      "grad_norm": 2.909055471420288,
      "learning_rate": 9.993081877240816e-06,
      "loss": 0.9334,
      "step": 1145
    },
    {
      "epoch": 0.0926079314733631,
      "grad_norm": 2.8035972118377686,
      "learning_rate": 9.993047423899239e-06,
      "loss": 1.0726,
      "step": 1146
    },
    {
      "epoch": 0.09268874118669064,
      "grad_norm": 4.139641761779785,
      "learning_rate": 9.993012885038777e-06,
      "loss": 1.0193,
      "step": 1147
    },
    {
      "epoch": 0.09276955090001818,
      "grad_norm": 2.9986088275909424,
      "learning_rate": 9.992978260660024e-06,
      "loss": 0.9637,
      "step": 1148
    },
    {
      "epoch": 0.09285036061334573,
      "grad_norm": 2.904437303543091,
      "learning_rate": 9.992943550763577e-06,
      "loss": 1.1149,
      "step": 1149
    },
    {
      "epoch": 0.09293117032667327,
      "grad_norm": 2.809866428375244,
      "learning_rate": 9.992908755350024e-06,
      "loss": 1.145,
      "step": 1150
    },
    {
      "epoch": 0.0930119800400008,
      "grad_norm": 3.3679118156433105,
      "learning_rate": 9.992873874419965e-06,
      "loss": 1.0692,
      "step": 1151
    },
    {
      "epoch": 0.09309278975332835,
      "grad_norm": 3.463749647140503,
      "learning_rate": 9.992838907973996e-06,
      "loss": 0.952,
      "step": 1152
    },
    {
      "epoch": 0.09317359946665589,
      "grad_norm": 2.9335153102874756,
      "learning_rate": 9.992803856012718e-06,
      "loss": 1.0042,
      "step": 1153
    },
    {
      "epoch": 0.09325440917998343,
      "grad_norm": 3.3813300132751465,
      "learning_rate": 9.992768718536727e-06,
      "loss": 0.9115,
      "step": 1154
    },
    {
      "epoch": 0.09333521889331098,
      "grad_norm": 3.05812406539917,
      "learning_rate": 9.99273349554663e-06,
      "loss": 1.0089,
      "step": 1155
    },
    {
      "epoch": 0.09341602860663852,
      "grad_norm": 2.7985036373138428,
      "learning_rate": 9.992698187043026e-06,
      "loss": 0.9933,
      "step": 1156
    },
    {
      "epoch": 0.09349683831996607,
      "grad_norm": 2.7855563163757324,
      "learning_rate": 9.992662793026522e-06,
      "loss": 1.0122,
      "step": 1157
    },
    {
      "epoch": 0.0935776480332936,
      "grad_norm": 2.9177639484405518,
      "learning_rate": 9.992627313497724e-06,
      "loss": 0.9769,
      "step": 1158
    },
    {
      "epoch": 0.09365845774662114,
      "grad_norm": 2.8957858085632324,
      "learning_rate": 9.992591748457239e-06,
      "loss": 1.0274,
      "step": 1159
    },
    {
      "epoch": 0.09373926745994869,
      "grad_norm": 2.8816301822662354,
      "learning_rate": 9.992556097905677e-06,
      "loss": 1.0627,
      "step": 1160
    },
    {
      "epoch": 0.09382007717327623,
      "grad_norm": 3.1168038845062256,
      "learning_rate": 9.992520361843647e-06,
      "loss": 1.0046,
      "step": 1161
    },
    {
      "epoch": 0.09390088688660377,
      "grad_norm": 2.815091609954834,
      "learning_rate": 9.992484540271764e-06,
      "loss": 1.0452,
      "step": 1162
    },
    {
      "epoch": 0.09398169659993132,
      "grad_norm": 2.5361690521240234,
      "learning_rate": 9.992448633190637e-06,
      "loss": 1.0325,
      "step": 1163
    },
    {
      "epoch": 0.09406250631325885,
      "grad_norm": 3.326178789138794,
      "learning_rate": 9.992412640600886e-06,
      "loss": 1.016,
      "step": 1164
    },
    {
      "epoch": 0.09414331602658639,
      "grad_norm": 2.790809392929077,
      "learning_rate": 9.992376562503125e-06,
      "loss": 1.0654,
      "step": 1165
    },
    {
      "epoch": 0.09422412573991394,
      "grad_norm": 2.718125104904175,
      "learning_rate": 9.992340398897971e-06,
      "loss": 1.0797,
      "step": 1166
    },
    {
      "epoch": 0.09430493545324148,
      "grad_norm": 2.9559214115142822,
      "learning_rate": 9.992304149786045e-06,
      "loss": 1.0271,
      "step": 1167
    },
    {
      "epoch": 0.09438574516656902,
      "grad_norm": 2.895761489868164,
      "learning_rate": 9.992267815167968e-06,
      "loss": 1.1303,
      "step": 1168
    },
    {
      "epoch": 0.09446655487989657,
      "grad_norm": 2.5579259395599365,
      "learning_rate": 9.992231395044363e-06,
      "loss": 0.9703,
      "step": 1169
    },
    {
      "epoch": 0.0945473645932241,
      "grad_norm": 3.430900812149048,
      "learning_rate": 9.992194889415851e-06,
      "loss": 1.0025,
      "step": 1170
    },
    {
      "epoch": 0.09462817430655164,
      "grad_norm": 2.4399726390838623,
      "learning_rate": 9.992158298283058e-06,
      "loss": 1.0111,
      "step": 1171
    },
    {
      "epoch": 0.09470898401987919,
      "grad_norm": 2.5215682983398438,
      "learning_rate": 9.992121621646612e-06,
      "loss": 1.1092,
      "step": 1172
    },
    {
      "epoch": 0.09478979373320673,
      "grad_norm": 2.789412021636963,
      "learning_rate": 9.99208485950714e-06,
      "loss": 1.11,
      "step": 1173
    },
    {
      "epoch": 0.09487060344653428,
      "grad_norm": 3.4019277095794678,
      "learning_rate": 9.992048011865275e-06,
      "loss": 1.1689,
      "step": 1174
    },
    {
      "epoch": 0.09495141315986182,
      "grad_norm": 3.0252597332000732,
      "learning_rate": 9.992011078721643e-06,
      "loss": 1.0104,
      "step": 1175
    },
    {
      "epoch": 0.09503222287318935,
      "grad_norm": 3.443145751953125,
      "learning_rate": 9.991974060076878e-06,
      "loss": 0.9915,
      "step": 1176
    },
    {
      "epoch": 0.0951130325865169,
      "grad_norm": 3.1933462619781494,
      "learning_rate": 9.991936955931618e-06,
      "loss": 0.9152,
      "step": 1177
    },
    {
      "epoch": 0.09519384229984444,
      "grad_norm": 3.098817825317383,
      "learning_rate": 9.991899766286495e-06,
      "loss": 1.11,
      "step": 1178
    },
    {
      "epoch": 0.09527465201317198,
      "grad_norm": 2.6892173290252686,
      "learning_rate": 9.991862491142145e-06,
      "loss": 1.0359,
      "step": 1179
    },
    {
      "epoch": 0.09535546172649953,
      "grad_norm": 2.912292003631592,
      "learning_rate": 9.991825130499208e-06,
      "loss": 1.0121,
      "step": 1180
    },
    {
      "epoch": 0.09543627143982707,
      "grad_norm": 2.7129578590393066,
      "learning_rate": 9.991787684358326e-06,
      "loss": 0.989,
      "step": 1181
    },
    {
      "epoch": 0.0955170811531546,
      "grad_norm": 3.12666654586792,
      "learning_rate": 9.991750152720135e-06,
      "loss": 1.0979,
      "step": 1182
    },
    {
      "epoch": 0.09559789086648215,
      "grad_norm": 2.8867383003234863,
      "learning_rate": 9.991712535585283e-06,
      "loss": 1.1116,
      "step": 1183
    },
    {
      "epoch": 0.09567870057980969,
      "grad_norm": 4.017873287200928,
      "learning_rate": 9.991674832954413e-06,
      "loss": 1.0397,
      "step": 1184
    },
    {
      "epoch": 0.09575951029313723,
      "grad_norm": 2.715961217880249,
      "learning_rate": 9.991637044828169e-06,
      "loss": 1.0186,
      "step": 1185
    },
    {
      "epoch": 0.09584032000646478,
      "grad_norm": 3.015178918838501,
      "learning_rate": 9.991599171207198e-06,
      "loss": 0.9089,
      "step": 1186
    },
    {
      "epoch": 0.09592112971979232,
      "grad_norm": 3.0427072048187256,
      "learning_rate": 9.991561212092152e-06,
      "loss": 1.0281,
      "step": 1187
    },
    {
      "epoch": 0.09600193943311987,
      "grad_norm": 2.909169912338257,
      "learning_rate": 9.99152316748368e-06,
      "loss": 0.9984,
      "step": 1188
    },
    {
      "epoch": 0.0960827491464474,
      "grad_norm": 3.5129024982452393,
      "learning_rate": 9.99148503738243e-06,
      "loss": 1.0376,
      "step": 1189
    },
    {
      "epoch": 0.09616355885977494,
      "grad_norm": 3.431941270828247,
      "learning_rate": 9.99144682178906e-06,
      "loss": 0.919,
      "step": 1190
    },
    {
      "epoch": 0.09624436857310249,
      "grad_norm": 3.0185487270355225,
      "learning_rate": 9.99140852070422e-06,
      "loss": 0.9442,
      "step": 1191
    },
    {
      "epoch": 0.09632517828643003,
      "grad_norm": 3.1691038608551025,
      "learning_rate": 9.99137013412857e-06,
      "loss": 1.0793,
      "step": 1192
    },
    {
      "epoch": 0.09640598799975757,
      "grad_norm": 3.392519235610962,
      "learning_rate": 9.991331662062766e-06,
      "loss": 1.0241,
      "step": 1193
    },
    {
      "epoch": 0.09648679771308512,
      "grad_norm": 3.272472381591797,
      "learning_rate": 9.991293104507467e-06,
      "loss": 1.1483,
      "step": 1194
    },
    {
      "epoch": 0.09656760742641265,
      "grad_norm": 2.778557300567627,
      "learning_rate": 9.991254461463332e-06,
      "loss": 0.9363,
      "step": 1195
    },
    {
      "epoch": 0.09664841713974019,
      "grad_norm": 3.2703022956848145,
      "learning_rate": 9.991215732931024e-06,
      "loss": 1.1747,
      "step": 1196
    },
    {
      "epoch": 0.09672922685306774,
      "grad_norm": 2.765848398208618,
      "learning_rate": 9.991176918911207e-06,
      "loss": 1.0589,
      "step": 1197
    },
    {
      "epoch": 0.09681003656639528,
      "grad_norm": 3.1882576942443848,
      "learning_rate": 9.991138019404545e-06,
      "loss": 1.0079,
      "step": 1198
    },
    {
      "epoch": 0.09689084627972282,
      "grad_norm": 2.969113826751709,
      "learning_rate": 9.991099034411705e-06,
      "loss": 1.0743,
      "step": 1199
    },
    {
      "epoch": 0.09697165599305037,
      "grad_norm": 3.393944501876831,
      "learning_rate": 9.991059963933355e-06,
      "loss": 1.0501,
      "step": 1200
    },
    {
      "epoch": 0.0970524657063779,
      "grad_norm": 2.5616579055786133,
      "learning_rate": 9.99102080797016e-06,
      "loss": 1.1143,
      "step": 1201
    },
    {
      "epoch": 0.09713327541970544,
      "grad_norm": 2.8260860443115234,
      "learning_rate": 9.990981566522797e-06,
      "loss": 1.0439,
      "step": 1202
    },
    {
      "epoch": 0.09721408513303299,
      "grad_norm": 2.793118953704834,
      "learning_rate": 9.990942239591934e-06,
      "loss": 1.0708,
      "step": 1203
    },
    {
      "epoch": 0.09729489484636053,
      "grad_norm": 2.6585116386413574,
      "learning_rate": 9.990902827178246e-06,
      "loss": 1.1239,
      "step": 1204
    },
    {
      "epoch": 0.09737570455968808,
      "grad_norm": 2.647916555404663,
      "learning_rate": 9.990863329282406e-06,
      "loss": 1.0526,
      "step": 1205
    },
    {
      "epoch": 0.09745651427301562,
      "grad_norm": 2.7440760135650635,
      "learning_rate": 9.990823745905095e-06,
      "loss": 1.0276,
      "step": 1206
    },
    {
      "epoch": 0.09753732398634316,
      "grad_norm": 3.040421962738037,
      "learning_rate": 9.990784077046985e-06,
      "loss": 1.0668,
      "step": 1207
    },
    {
      "epoch": 0.0976181336996707,
      "grad_norm": 3.0546274185180664,
      "learning_rate": 9.990744322708761e-06,
      "loss": 0.913,
      "step": 1208
    },
    {
      "epoch": 0.09769894341299824,
      "grad_norm": 3.5525715351104736,
      "learning_rate": 9.9907044828911e-06,
      "loss": 0.9561,
      "step": 1209
    },
    {
      "epoch": 0.09777975312632578,
      "grad_norm": 3.1630806922912598,
      "learning_rate": 9.990664557594687e-06,
      "loss": 1.1618,
      "step": 1210
    },
    {
      "epoch": 0.09786056283965333,
      "grad_norm": 3.20241379737854,
      "learning_rate": 9.990624546820204e-06,
      "loss": 1.1931,
      "step": 1211
    },
    {
      "epoch": 0.09794137255298087,
      "grad_norm": 2.860574722290039,
      "learning_rate": 9.990584450568338e-06,
      "loss": 0.958,
      "step": 1212
    },
    {
      "epoch": 0.0980221822663084,
      "grad_norm": 3.0059783458709717,
      "learning_rate": 9.990544268839773e-06,
      "loss": 1.0053,
      "step": 1213
    },
    {
      "epoch": 0.09810299197963596,
      "grad_norm": 3.187241315841675,
      "learning_rate": 9.9905040016352e-06,
      "loss": 0.9463,
      "step": 1214
    },
    {
      "epoch": 0.0981838016929635,
      "grad_norm": 3.665228843688965,
      "learning_rate": 9.990463648955306e-06,
      "loss": 1.0178,
      "step": 1215
    },
    {
      "epoch": 0.09826461140629103,
      "grad_norm": 2.5117361545562744,
      "learning_rate": 9.990423210800786e-06,
      "loss": 0.9685,
      "step": 1216
    },
    {
      "epoch": 0.09834542111961858,
      "grad_norm": 3.453538417816162,
      "learning_rate": 9.99038268717233e-06,
      "loss": 1.0154,
      "step": 1217
    },
    {
      "epoch": 0.09842623083294612,
      "grad_norm": 3.6725711822509766,
      "learning_rate": 9.990342078070632e-06,
      "loss": 1.0863,
      "step": 1218
    },
    {
      "epoch": 0.09850704054627366,
      "grad_norm": 2.6737985610961914,
      "learning_rate": 9.990301383496389e-06,
      "loss": 1.1508,
      "step": 1219
    },
    {
      "epoch": 0.0985878502596012,
      "grad_norm": 3.180436134338379,
      "learning_rate": 9.990260603450294e-06,
      "loss": 1.0313,
      "step": 1220
    },
    {
      "epoch": 0.09866865997292874,
      "grad_norm": 3.468973159790039,
      "learning_rate": 9.99021973793305e-06,
      "loss": 1.0211,
      "step": 1221
    },
    {
      "epoch": 0.0987494696862563,
      "grad_norm": 3.042269468307495,
      "learning_rate": 9.990178786945356e-06,
      "loss": 0.9764,
      "step": 1222
    },
    {
      "epoch": 0.09883027939958383,
      "grad_norm": 2.8079960346221924,
      "learning_rate": 9.990137750487912e-06,
      "loss": 1.1182,
      "step": 1223
    },
    {
      "epoch": 0.09891108911291137,
      "grad_norm": 2.746098279953003,
      "learning_rate": 9.990096628561422e-06,
      "loss": 0.958,
      "step": 1224
    },
    {
      "epoch": 0.09899189882623892,
      "grad_norm": 2.608017683029175,
      "learning_rate": 9.99005542116659e-06,
      "loss": 1.0234,
      "step": 1225
    },
    {
      "epoch": 0.09907270853956646,
      "grad_norm": 3.6061453819274902,
      "learning_rate": 9.990014128304122e-06,
      "loss": 1.0367,
      "step": 1226
    },
    {
      "epoch": 0.099153518252894,
      "grad_norm": 2.522488832473755,
      "learning_rate": 9.989972749974724e-06,
      "loss": 1.0161,
      "step": 1227
    },
    {
      "epoch": 0.09923432796622154,
      "grad_norm": 3.319269895553589,
      "learning_rate": 9.989931286179106e-06,
      "loss": 1.1268,
      "step": 1228
    },
    {
      "epoch": 0.09931513767954908,
      "grad_norm": 2.809621572494507,
      "learning_rate": 9.989889736917979e-06,
      "loss": 0.9355,
      "step": 1229
    },
    {
      "epoch": 0.09939594739287662,
      "grad_norm": 3.0901947021484375,
      "learning_rate": 9.989848102192052e-06,
      "loss": 1.1745,
      "step": 1230
    },
    {
      "epoch": 0.09947675710620417,
      "grad_norm": 2.6698098182678223,
      "learning_rate": 9.989806382002039e-06,
      "loss": 1.0672,
      "step": 1231
    },
    {
      "epoch": 0.0995575668195317,
      "grad_norm": 2.8612308502197266,
      "learning_rate": 9.989764576348656e-06,
      "loss": 0.9378,
      "step": 1232
    },
    {
      "epoch": 0.09963837653285924,
      "grad_norm": 3.5653903484344482,
      "learning_rate": 9.98972268523262e-06,
      "loss": 1.0345,
      "step": 1233
    },
    {
      "epoch": 0.0997191862461868,
      "grad_norm": 2.6366031169891357,
      "learning_rate": 9.989680708654644e-06,
      "loss": 1.011,
      "step": 1234
    },
    {
      "epoch": 0.09979999595951433,
      "grad_norm": 3.223850965499878,
      "learning_rate": 9.989638646615452e-06,
      "loss": 0.9727,
      "step": 1235
    },
    {
      "epoch": 0.09988080567284188,
      "grad_norm": 3.5510125160217285,
      "learning_rate": 9.989596499115759e-06,
      "loss": 1.0507,
      "step": 1236
    },
    {
      "epoch": 0.09996161538616942,
      "grad_norm": 3.2256150245666504,
      "learning_rate": 9.989554266156291e-06,
      "loss": 0.9427,
      "step": 1237
    },
    {
      "epoch": 0.10004242509949696,
      "grad_norm": 2.9781923294067383,
      "learning_rate": 9.989511947737772e-06,
      "loss": 0.9778,
      "step": 1238
    },
    {
      "epoch": 0.10012323481282451,
      "grad_norm": 2.9848222732543945,
      "learning_rate": 9.989469543860924e-06,
      "loss": 1.0727,
      "step": 1239
    },
    {
      "epoch": 0.10020404452615204,
      "grad_norm": 2.984651803970337,
      "learning_rate": 9.989427054526476e-06,
      "loss": 1.0833,
      "step": 1240
    },
    {
      "epoch": 0.10028485423947958,
      "grad_norm": 2.8715219497680664,
      "learning_rate": 9.98938447973515e-06,
      "loss": 1.0833,
      "step": 1241
    },
    {
      "epoch": 0.10036566395280713,
      "grad_norm": 2.879469156265259,
      "learning_rate": 9.989341819487683e-06,
      "loss": 1.1641,
      "step": 1242
    },
    {
      "epoch": 0.10044647366613467,
      "grad_norm": 3.4816324710845947,
      "learning_rate": 9.989299073784801e-06,
      "loss": 1.0091,
      "step": 1243
    },
    {
      "epoch": 0.1005272833794622,
      "grad_norm": 3.2763583660125732,
      "learning_rate": 9.989256242627237e-06,
      "loss": 1.104,
      "step": 1244
    },
    {
      "epoch": 0.10060809309278976,
      "grad_norm": 3.425124406814575,
      "learning_rate": 9.989213326015724e-06,
      "loss": 1.0348,
      "step": 1245
    },
    {
      "epoch": 0.1006889028061173,
      "grad_norm": 2.846920967102051,
      "learning_rate": 9.989170323950999e-06,
      "loss": 0.967,
      "step": 1246
    },
    {
      "epoch": 0.10076971251944483,
      "grad_norm": 2.907984495162964,
      "learning_rate": 9.989127236433795e-06,
      "loss": 1.0871,
      "step": 1247
    },
    {
      "epoch": 0.10085052223277238,
      "grad_norm": 2.9998886585235596,
      "learning_rate": 9.989084063464855e-06,
      "loss": 1.004,
      "step": 1248
    },
    {
      "epoch": 0.10093133194609992,
      "grad_norm": 2.672520637512207,
      "learning_rate": 9.989040805044914e-06,
      "loss": 1.1658,
      "step": 1249
    },
    {
      "epoch": 0.10101214165942746,
      "grad_norm": 2.842155694961548,
      "learning_rate": 9.988997461174717e-06,
      "loss": 1.0719,
      "step": 1250
    },
    {
      "epoch": 0.10109295137275501,
      "grad_norm": 3.131115198135376,
      "learning_rate": 9.988954031855001e-06,
      "loss": 1.0328,
      "step": 1251
    },
    {
      "epoch": 0.10117376108608254,
      "grad_norm": 3.119647264480591,
      "learning_rate": 9.988910517086514e-06,
      "loss": 0.9574,
      "step": 1252
    },
    {
      "epoch": 0.1012545707994101,
      "grad_norm": 2.5935230255126953,
      "learning_rate": 9.988866916870001e-06,
      "loss": 1.0931,
      "step": 1253
    },
    {
      "epoch": 0.10133538051273763,
      "grad_norm": 3.033775806427002,
      "learning_rate": 9.988823231206208e-06,
      "loss": 1.0586,
      "step": 1254
    },
    {
      "epoch": 0.10141619022606517,
      "grad_norm": 2.881053924560547,
      "learning_rate": 9.98877946009588e-06,
      "loss": 1.1503,
      "step": 1255
    },
    {
      "epoch": 0.10149699993939272,
      "grad_norm": 2.9397523403167725,
      "learning_rate": 9.988735603539775e-06,
      "loss": 0.9481,
      "step": 1256
    },
    {
      "epoch": 0.10157780965272026,
      "grad_norm": 3.23587965965271,
      "learning_rate": 9.988691661538634e-06,
      "loss": 1.1529,
      "step": 1257
    },
    {
      "epoch": 0.1016586193660478,
      "grad_norm": 3.011920690536499,
      "learning_rate": 9.988647634093218e-06,
      "loss": 0.9915,
      "step": 1258
    },
    {
      "epoch": 0.10173942907937535,
      "grad_norm": 3.658949136734009,
      "learning_rate": 9.988603521204276e-06,
      "loss": 0.9997,
      "step": 1259
    },
    {
      "epoch": 0.10182023879270288,
      "grad_norm": 3.2877025604248047,
      "learning_rate": 9.988559322872567e-06,
      "loss": 1.0461,
      "step": 1260
    },
    {
      "epoch": 0.10190104850603042,
      "grad_norm": 3.080794334411621,
      "learning_rate": 9.988515039098845e-06,
      "loss": 0.9342,
      "step": 1261
    },
    {
      "epoch": 0.10198185821935797,
      "grad_norm": 3.1314265727996826,
      "learning_rate": 9.988470669883869e-06,
      "loss": 1.0981,
      "step": 1262
    },
    {
      "epoch": 0.10206266793268551,
      "grad_norm": 2.952272415161133,
      "learning_rate": 9.9884262152284e-06,
      "loss": 1.0115,
      "step": 1263
    },
    {
      "epoch": 0.10214347764601305,
      "grad_norm": 3.143523693084717,
      "learning_rate": 9.988381675133202e-06,
      "loss": 1.0682,
      "step": 1264
    },
    {
      "epoch": 0.1022242873593406,
      "grad_norm": 2.6488254070281982,
      "learning_rate": 9.98833704959903e-06,
      "loss": 1.2389,
      "step": 1265
    },
    {
      "epoch": 0.10230509707266813,
      "grad_norm": 3.380514144897461,
      "learning_rate": 9.988292338626658e-06,
      "loss": 1.0109,
      "step": 1266
    },
    {
      "epoch": 0.10238590678599567,
      "grad_norm": 2.8261666297912598,
      "learning_rate": 9.988247542216844e-06,
      "loss": 1.057,
      "step": 1267
    },
    {
      "epoch": 0.10246671649932322,
      "grad_norm": 3.0670905113220215,
      "learning_rate": 9.988202660370358e-06,
      "loss": 1.1282,
      "step": 1268
    },
    {
      "epoch": 0.10254752621265076,
      "grad_norm": 2.9264938831329346,
      "learning_rate": 9.988157693087971e-06,
      "loss": 0.971,
      "step": 1269
    },
    {
      "epoch": 0.10262833592597831,
      "grad_norm": 2.8461177349090576,
      "learning_rate": 9.98811264037045e-06,
      "loss": 1.0776,
      "step": 1270
    },
    {
      "epoch": 0.10270914563930585,
      "grad_norm": 2.623901844024658,
      "learning_rate": 9.988067502218569e-06,
      "loss": 1.096,
      "step": 1271
    },
    {
      "epoch": 0.10278995535263338,
      "grad_norm": 2.9775118827819824,
      "learning_rate": 9.988022278633097e-06,
      "loss": 0.9979,
      "step": 1272
    },
    {
      "epoch": 0.10287076506596093,
      "grad_norm": 3.2639927864074707,
      "learning_rate": 9.987976969614816e-06,
      "loss": 1.0962,
      "step": 1273
    },
    {
      "epoch": 0.10295157477928847,
      "grad_norm": 2.864597797393799,
      "learning_rate": 9.987931575164495e-06,
      "loss": 0.9859,
      "step": 1274
    },
    {
      "epoch": 0.10303238449261601,
      "grad_norm": 2.9163618087768555,
      "learning_rate": 9.987886095282916e-06,
      "loss": 1.0499,
      "step": 1275
    },
    {
      "epoch": 0.10311319420594356,
      "grad_norm": 3.827516794204712,
      "learning_rate": 9.987840529970853e-06,
      "loss": 0.9777,
      "step": 1276
    },
    {
      "epoch": 0.1031940039192711,
      "grad_norm": 2.858494758605957,
      "learning_rate": 9.987794879229091e-06,
      "loss": 0.9267,
      "step": 1277
    },
    {
      "epoch": 0.10327481363259863,
      "grad_norm": 2.7237348556518555,
      "learning_rate": 9.98774914305841e-06,
      "loss": 1.0345,
      "step": 1278
    },
    {
      "epoch": 0.10335562334592618,
      "grad_norm": 2.8835742473602295,
      "learning_rate": 9.987703321459594e-06,
      "loss": 1.0166,
      "step": 1279
    },
    {
      "epoch": 0.10343643305925372,
      "grad_norm": 3.4978060722351074,
      "learning_rate": 9.987657414433428e-06,
      "loss": 0.992,
      "step": 1280
    },
    {
      "epoch": 0.10351724277258126,
      "grad_norm": 2.936861753463745,
      "learning_rate": 9.987611421980697e-06,
      "loss": 1.01,
      "step": 1281
    },
    {
      "epoch": 0.10359805248590881,
      "grad_norm": 3.074481725692749,
      "learning_rate": 9.98756534410219e-06,
      "loss": 0.9462,
      "step": 1282
    },
    {
      "epoch": 0.10367886219923635,
      "grad_norm": 3.292534828186035,
      "learning_rate": 9.987519180798696e-06,
      "loss": 1.0477,
      "step": 1283
    },
    {
      "epoch": 0.10375967191256388,
      "grad_norm": 2.9260239601135254,
      "learning_rate": 9.987472932071004e-06,
      "loss": 1.0362,
      "step": 1284
    },
    {
      "epoch": 0.10384048162589143,
      "grad_norm": 2.984226942062378,
      "learning_rate": 9.987426597919908e-06,
      "loss": 1.0361,
      "step": 1285
    },
    {
      "epoch": 0.10392129133921897,
      "grad_norm": 3.5397887229919434,
      "learning_rate": 9.987380178346203e-06,
      "loss": 1.0109,
      "step": 1286
    },
    {
      "epoch": 0.10400210105254652,
      "grad_norm": 3.0161139965057373,
      "learning_rate": 9.98733367335068e-06,
      "loss": 1.0292,
      "step": 1287
    },
    {
      "epoch": 0.10408291076587406,
      "grad_norm": 2.910106658935547,
      "learning_rate": 9.987287082934139e-06,
      "loss": 1.1038,
      "step": 1288
    },
    {
      "epoch": 0.1041637204792016,
      "grad_norm": 2.628661870956421,
      "learning_rate": 9.987240407097376e-06,
      "loss": 0.9265,
      "step": 1289
    },
    {
      "epoch": 0.10424453019252915,
      "grad_norm": 3.4557712078094482,
      "learning_rate": 9.987193645841191e-06,
      "loss": 1.0006,
      "step": 1290
    },
    {
      "epoch": 0.10432533990585668,
      "grad_norm": 2.7255899906158447,
      "learning_rate": 9.987146799166386e-06,
      "loss": 0.909,
      "step": 1291
    },
    {
      "epoch": 0.10440614961918422,
      "grad_norm": 2.7850592136383057,
      "learning_rate": 9.98709986707376e-06,
      "loss": 0.9859,
      "step": 1292
    },
    {
      "epoch": 0.10448695933251177,
      "grad_norm": 2.9532461166381836,
      "learning_rate": 9.98705284956412e-06,
      "loss": 1.1035,
      "step": 1293
    },
    {
      "epoch": 0.10456776904583931,
      "grad_norm": 2.9681034088134766,
      "learning_rate": 9.987005746638272e-06,
      "loss": 1.1082,
      "step": 1294
    },
    {
      "epoch": 0.10464857875916685,
      "grad_norm": 2.5660200119018555,
      "learning_rate": 9.986958558297021e-06,
      "loss": 1.0165,
      "step": 1295
    },
    {
      "epoch": 0.1047293884724944,
      "grad_norm": 3.059093952178955,
      "learning_rate": 9.986911284541177e-06,
      "loss": 0.9627,
      "step": 1296
    },
    {
      "epoch": 0.10481019818582193,
      "grad_norm": 3.7301979064941406,
      "learning_rate": 9.986863925371545e-06,
      "loss": 1.1762,
      "step": 1297
    },
    {
      "epoch": 0.10489100789914947,
      "grad_norm": 3.185795545578003,
      "learning_rate": 9.986816480788941e-06,
      "loss": 1.1375,
      "step": 1298
    },
    {
      "epoch": 0.10497181761247702,
      "grad_norm": 2.9829509258270264,
      "learning_rate": 9.986768950794176e-06,
      "loss": 0.986,
      "step": 1299
    },
    {
      "epoch": 0.10505262732580456,
      "grad_norm": 2.644713878631592,
      "learning_rate": 9.986721335388064e-06,
      "loss": 1.0705,
      "step": 1300
    },
    {
      "epoch": 0.10513343703913211,
      "grad_norm": 2.5542094707489014,
      "learning_rate": 9.98667363457142e-06,
      "loss": 1.0634,
      "step": 1301
    },
    {
      "epoch": 0.10521424675245965,
      "grad_norm": 3.101796865463257,
      "learning_rate": 9.986625848345063e-06,
      "loss": 1.0187,
      "step": 1302
    },
    {
      "epoch": 0.10529505646578718,
      "grad_norm": 3.5837759971618652,
      "learning_rate": 9.986577976709808e-06,
      "loss": 0.8596,
      "step": 1303
    },
    {
      "epoch": 0.10537586617911474,
      "grad_norm": 3.0948948860168457,
      "learning_rate": 9.986530019666477e-06,
      "loss": 1.1609,
      "step": 1304
    },
    {
      "epoch": 0.10545667589244227,
      "grad_norm": 3.398942232131958,
      "learning_rate": 9.986481977215892e-06,
      "loss": 1.0362,
      "step": 1305
    },
    {
      "epoch": 0.10553748560576981,
      "grad_norm": 2.6761116981506348,
      "learning_rate": 9.986433849358876e-06,
      "loss": 1.1106,
      "step": 1306
    },
    {
      "epoch": 0.10561829531909736,
      "grad_norm": 3.135209560394287,
      "learning_rate": 9.986385636096252e-06,
      "loss": 1.0177,
      "step": 1307
    },
    {
      "epoch": 0.1056991050324249,
      "grad_norm": 2.8185837268829346,
      "learning_rate": 9.986337337428847e-06,
      "loss": 1.0082,
      "step": 1308
    },
    {
      "epoch": 0.10577991474575243,
      "grad_norm": 2.7826640605926514,
      "learning_rate": 9.986288953357486e-06,
      "loss": 0.9948,
      "step": 1309
    },
    {
      "epoch": 0.10586072445907999,
      "grad_norm": 2.922977924346924,
      "learning_rate": 9.986240483883e-06,
      "loss": 1.0352,
      "step": 1310
    },
    {
      "epoch": 0.10594153417240752,
      "grad_norm": 2.787425994873047,
      "learning_rate": 9.986191929006217e-06,
      "loss": 1.0207,
      "step": 1311
    },
    {
      "epoch": 0.10602234388573506,
      "grad_norm": 3.1789329051971436,
      "learning_rate": 9.986143288727972e-06,
      "loss": 0.9805,
      "step": 1312
    },
    {
      "epoch": 0.10610315359906261,
      "grad_norm": 3.885017156600952,
      "learning_rate": 9.986094563049096e-06,
      "loss": 0.9988,
      "step": 1313
    },
    {
      "epoch": 0.10618396331239015,
      "grad_norm": 2.5836968421936035,
      "learning_rate": 9.986045751970423e-06,
      "loss": 0.9195,
      "step": 1314
    },
    {
      "epoch": 0.10626477302571768,
      "grad_norm": 2.9889109134674072,
      "learning_rate": 9.98599685549279e-06,
      "loss": 0.8965,
      "step": 1315
    },
    {
      "epoch": 0.10634558273904524,
      "grad_norm": 2.8345937728881836,
      "learning_rate": 9.985947873617033e-06,
      "loss": 1.0412,
      "step": 1316
    },
    {
      "epoch": 0.10642639245237277,
      "grad_norm": 2.8975749015808105,
      "learning_rate": 9.985898806343995e-06,
      "loss": 0.9263,
      "step": 1317
    },
    {
      "epoch": 0.10650720216570032,
      "grad_norm": 3.2546446323394775,
      "learning_rate": 9.985849653674512e-06,
      "loss": 1.1148,
      "step": 1318
    },
    {
      "epoch": 0.10658801187902786,
      "grad_norm": 2.811129093170166,
      "learning_rate": 9.985800415609426e-06,
      "loss": 1.1397,
      "step": 1319
    },
    {
      "epoch": 0.1066688215923554,
      "grad_norm": 2.553903102874756,
      "learning_rate": 9.985751092149581e-06,
      "loss": 0.939,
      "step": 1320
    },
    {
      "epoch": 0.10674963130568295,
      "grad_norm": 2.546623468399048,
      "learning_rate": 9.985701683295825e-06,
      "loss": 1.0799,
      "step": 1321
    },
    {
      "epoch": 0.10683044101901049,
      "grad_norm": 3.3942575454711914,
      "learning_rate": 9.985652189049001e-06,
      "loss": 1.028,
      "step": 1322
    },
    {
      "epoch": 0.10691125073233802,
      "grad_norm": 2.967158079147339,
      "learning_rate": 9.985602609409957e-06,
      "loss": 1.1134,
      "step": 1323
    },
    {
      "epoch": 0.10699206044566557,
      "grad_norm": 3.029263734817505,
      "learning_rate": 9.985552944379544e-06,
      "loss": 1.0863,
      "step": 1324
    },
    {
      "epoch": 0.10707287015899311,
      "grad_norm": 2.7675929069519043,
      "learning_rate": 9.98550319395861e-06,
      "loss": 1.0656,
      "step": 1325
    },
    {
      "epoch": 0.10715367987232065,
      "grad_norm": 2.4967398643493652,
      "learning_rate": 9.985453358148008e-06,
      "loss": 0.9626,
      "step": 1326
    },
    {
      "epoch": 0.1072344895856482,
      "grad_norm": 3.1739935874938965,
      "learning_rate": 9.985403436948593e-06,
      "loss": 0.8729,
      "step": 1327
    },
    {
      "epoch": 0.10731529929897574,
      "grad_norm": 2.97395658493042,
      "learning_rate": 9.985353430361219e-06,
      "loss": 0.9491,
      "step": 1328
    },
    {
      "epoch": 0.10739610901230327,
      "grad_norm": 2.934847116470337,
      "learning_rate": 9.985303338386743e-06,
      "loss": 1.0922,
      "step": 1329
    },
    {
      "epoch": 0.10747691872563082,
      "grad_norm": 2.6325595378875732,
      "learning_rate": 9.98525316102602e-06,
      "loss": 1.0173,
      "step": 1330
    },
    {
      "epoch": 0.10755772843895836,
      "grad_norm": 2.7959065437316895,
      "learning_rate": 9.985202898279914e-06,
      "loss": 1.0029,
      "step": 1331
    },
    {
      "epoch": 0.1076385381522859,
      "grad_norm": 2.5871143341064453,
      "learning_rate": 9.985152550149283e-06,
      "loss": 1.0173,
      "step": 1332
    },
    {
      "epoch": 0.10771934786561345,
      "grad_norm": 3.379809617996216,
      "learning_rate": 9.985102116634991e-06,
      "loss": 1.0963,
      "step": 1333
    },
    {
      "epoch": 0.10780015757894099,
      "grad_norm": 3.1764838695526123,
      "learning_rate": 9.9850515977379e-06,
      "loss": 1.1185,
      "step": 1334
    },
    {
      "epoch": 0.10788096729226854,
      "grad_norm": 2.8184707164764404,
      "learning_rate": 9.985000993458876e-06,
      "loss": 0.8538,
      "step": 1335
    },
    {
      "epoch": 0.10796177700559607,
      "grad_norm": 3.082862377166748,
      "learning_rate": 9.984950303798787e-06,
      "loss": 1.0536,
      "step": 1336
    },
    {
      "epoch": 0.10804258671892361,
      "grad_norm": 3.090818405151367,
      "learning_rate": 9.984899528758498e-06,
      "loss": 1.0958,
      "step": 1337
    },
    {
      "epoch": 0.10812339643225116,
      "grad_norm": 3.066433906555176,
      "learning_rate": 9.984848668338883e-06,
      "loss": 1.0357,
      "step": 1338
    },
    {
      "epoch": 0.1082042061455787,
      "grad_norm": 2.9226813316345215,
      "learning_rate": 9.984797722540808e-06,
      "loss": 0.9513,
      "step": 1339
    },
    {
      "epoch": 0.10828501585890624,
      "grad_norm": 2.9141769409179688,
      "learning_rate": 9.984746691365153e-06,
      "loss": 0.8497,
      "step": 1340
    },
    {
      "epoch": 0.10836582557223379,
      "grad_norm": 2.980806589126587,
      "learning_rate": 9.984695574812785e-06,
      "loss": 0.9877,
      "step": 1341
    },
    {
      "epoch": 0.10844663528556132,
      "grad_norm": 2.9618945121765137,
      "learning_rate": 9.984644372884579e-06,
      "loss": 1.0403,
      "step": 1342
    },
    {
      "epoch": 0.10852744499888886,
      "grad_norm": 2.344740152359009,
      "learning_rate": 9.984593085581419e-06,
      "loss": 1.1414,
      "step": 1343
    },
    {
      "epoch": 0.10860825471221641,
      "grad_norm": 3.0067477226257324,
      "learning_rate": 9.984541712904178e-06,
      "loss": 1.0102,
      "step": 1344
    },
    {
      "epoch": 0.10868906442554395,
      "grad_norm": 2.7962794303894043,
      "learning_rate": 9.984490254853737e-06,
      "loss": 1.0444,
      "step": 1345
    },
    {
      "epoch": 0.10876987413887149,
      "grad_norm": 2.8900249004364014,
      "learning_rate": 9.984438711430978e-06,
      "loss": 1.0275,
      "step": 1346
    },
    {
      "epoch": 0.10885068385219904,
      "grad_norm": 3.3476011753082275,
      "learning_rate": 9.984387082636783e-06,
      "loss": 1.0067,
      "step": 1347
    },
    {
      "epoch": 0.10893149356552657,
      "grad_norm": 2.934443473815918,
      "learning_rate": 9.984335368472039e-06,
      "loss": 1.0258,
      "step": 1348
    },
    {
      "epoch": 0.10901230327885413,
      "grad_norm": 3.3023133277893066,
      "learning_rate": 9.984283568937625e-06,
      "loss": 0.9624,
      "step": 1349
    },
    {
      "epoch": 0.10909311299218166,
      "grad_norm": 3.7269654273986816,
      "learning_rate": 9.984231684034436e-06,
      "loss": 1.0803,
      "step": 1350
    },
    {
      "epoch": 0.1091739227055092,
      "grad_norm": 3.2603659629821777,
      "learning_rate": 9.984179713763356e-06,
      "loss": 1.1566,
      "step": 1351
    },
    {
      "epoch": 0.10925473241883675,
      "grad_norm": 3.2887766361236572,
      "learning_rate": 9.984127658125278e-06,
      "loss": 0.994,
      "step": 1352
    },
    {
      "epoch": 0.10933554213216429,
      "grad_norm": 3.1544361114501953,
      "learning_rate": 9.98407551712109e-06,
      "loss": 1.0248,
      "step": 1353
    },
    {
      "epoch": 0.10941635184549182,
      "grad_norm": 2.8217058181762695,
      "learning_rate": 9.984023290751688e-06,
      "loss": 1.1021,
      "step": 1354
    },
    {
      "epoch": 0.10949716155881938,
      "grad_norm": 3.799391269683838,
      "learning_rate": 9.983970979017966e-06,
      "loss": 1.0307,
      "step": 1355
    },
    {
      "epoch": 0.10957797127214691,
      "grad_norm": 3.279231071472168,
      "learning_rate": 9.983918581920817e-06,
      "loss": 1.011,
      "step": 1356
    },
    {
      "epoch": 0.10965878098547445,
      "grad_norm": 3.18418288230896,
      "learning_rate": 9.983866099461144e-06,
      "loss": 1.014,
      "step": 1357
    },
    {
      "epoch": 0.109739590698802,
      "grad_norm": 2.627591371536255,
      "learning_rate": 9.983813531639843e-06,
      "loss": 1.1113,
      "step": 1358
    },
    {
      "epoch": 0.10982040041212954,
      "grad_norm": 2.9131274223327637,
      "learning_rate": 9.983760878457812e-06,
      "loss": 1.0794,
      "step": 1359
    },
    {
      "epoch": 0.10990121012545707,
      "grad_norm": 2.730416774749756,
      "learning_rate": 9.983708139915956e-06,
      "loss": 0.9852,
      "step": 1360
    },
    {
      "epoch": 0.10998201983878463,
      "grad_norm": 2.822190761566162,
      "learning_rate": 9.983655316015178e-06,
      "loss": 0.9026,
      "step": 1361
    },
    {
      "epoch": 0.11006282955211216,
      "grad_norm": 2.740520715713501,
      "learning_rate": 9.983602406756381e-06,
      "loss": 1.0623,
      "step": 1362
    },
    {
      "epoch": 0.1101436392654397,
      "grad_norm": 3.0149307250976562,
      "learning_rate": 9.983549412140475e-06,
      "loss": 1.0159,
      "step": 1363
    },
    {
      "epoch": 0.11022444897876725,
      "grad_norm": 2.778958797454834,
      "learning_rate": 9.98349633216836e-06,
      "loss": 1.0258,
      "step": 1364
    },
    {
      "epoch": 0.11030525869209479,
      "grad_norm": 2.7317237854003906,
      "learning_rate": 9.983443166840953e-06,
      "loss": 1.0187,
      "step": 1365
    },
    {
      "epoch": 0.11038606840542234,
      "grad_norm": 2.9775919914245605,
      "learning_rate": 9.983389916159164e-06,
      "loss": 0.9955,
      "step": 1366
    },
    {
      "epoch": 0.11046687811874988,
      "grad_norm": 3.141310214996338,
      "learning_rate": 9.983336580123899e-06,
      "loss": 1.0282,
      "step": 1367
    },
    {
      "epoch": 0.11054768783207741,
      "grad_norm": 3.1330995559692383,
      "learning_rate": 9.983283158736077e-06,
      "loss": 1.1138,
      "step": 1368
    },
    {
      "epoch": 0.11062849754540496,
      "grad_norm": 2.7490227222442627,
      "learning_rate": 9.98322965199661e-06,
      "loss": 1.0073,
      "step": 1369
    },
    {
      "epoch": 0.1107093072587325,
      "grad_norm": 3.0610904693603516,
      "learning_rate": 9.98317605990642e-06,
      "loss": 1.0441,
      "step": 1370
    },
    {
      "epoch": 0.11079011697206004,
      "grad_norm": 2.916208505630493,
      "learning_rate": 9.983122382466416e-06,
      "loss": 0.974,
      "step": 1371
    },
    {
      "epoch": 0.11087092668538759,
      "grad_norm": 3.0176069736480713,
      "learning_rate": 9.983068619677522e-06,
      "loss": 1.1055,
      "step": 1372
    },
    {
      "epoch": 0.11095173639871513,
      "grad_norm": 3.2212870121002197,
      "learning_rate": 9.98301477154066e-06,
      "loss": 1.0239,
      "step": 1373
    },
    {
      "epoch": 0.11103254611204266,
      "grad_norm": 2.967841386795044,
      "learning_rate": 9.982960838056752e-06,
      "loss": 0.976,
      "step": 1374
    },
    {
      "epoch": 0.11111335582537021,
      "grad_norm": 3.0756235122680664,
      "learning_rate": 9.98290681922672e-06,
      "loss": 1.0716,
      "step": 1375
    },
    {
      "epoch": 0.11119416553869775,
      "grad_norm": 2.6591484546661377,
      "learning_rate": 9.98285271505149e-06,
      "loss": 1.1386,
      "step": 1376
    },
    {
      "epoch": 0.11127497525202529,
      "grad_norm": 2.74874210357666,
      "learning_rate": 9.982798525531988e-06,
      "loss": 1.0843,
      "step": 1377
    },
    {
      "epoch": 0.11135578496535284,
      "grad_norm": 3.1984851360321045,
      "learning_rate": 9.982744250669144e-06,
      "loss": 1.0687,
      "step": 1378
    },
    {
      "epoch": 0.11143659467868038,
      "grad_norm": 2.9788684844970703,
      "learning_rate": 9.982689890463886e-06,
      "loss": 1.1124,
      "step": 1379
    },
    {
      "epoch": 0.11151740439200791,
      "grad_norm": 2.97041392326355,
      "learning_rate": 9.982635444917146e-06,
      "loss": 0.976,
      "step": 1380
    },
    {
      "epoch": 0.11159821410533546,
      "grad_norm": 2.7469606399536133,
      "learning_rate": 9.982580914029855e-06,
      "loss": 1.0973,
      "step": 1381
    },
    {
      "epoch": 0.111679023818663,
      "grad_norm": 3.545305013656616,
      "learning_rate": 9.982526297802948e-06,
      "loss": 1.0933,
      "step": 1382
    },
    {
      "epoch": 0.11175983353199055,
      "grad_norm": 2.8980941772460938,
      "learning_rate": 9.982471596237363e-06,
      "loss": 0.9828,
      "step": 1383
    },
    {
      "epoch": 0.11184064324531809,
      "grad_norm": 2.698004722595215,
      "learning_rate": 9.982416809334031e-06,
      "loss": 0.9321,
      "step": 1384
    },
    {
      "epoch": 0.11192145295864563,
      "grad_norm": 3.128380298614502,
      "learning_rate": 9.982361937093896e-06,
      "loss": 1.0612,
      "step": 1385
    },
    {
      "epoch": 0.11200226267197318,
      "grad_norm": 2.8916828632354736,
      "learning_rate": 9.982306979517895e-06,
      "loss": 1.0464,
      "step": 1386
    },
    {
      "epoch": 0.11208307238530071,
      "grad_norm": 3.5671095848083496,
      "learning_rate": 9.982251936606968e-06,
      "loss": 1.0209,
      "step": 1387
    },
    {
      "epoch": 0.11216388209862825,
      "grad_norm": 3.4572086334228516,
      "learning_rate": 9.982196808362061e-06,
      "loss": 1.0191,
      "step": 1388
    },
    {
      "epoch": 0.1122446918119558,
      "grad_norm": 2.9889228343963623,
      "learning_rate": 9.982141594784117e-06,
      "loss": 1.0063,
      "step": 1389
    },
    {
      "epoch": 0.11232550152528334,
      "grad_norm": 2.919299602508545,
      "learning_rate": 9.982086295874083e-06,
      "loss": 0.8683,
      "step": 1390
    },
    {
      "epoch": 0.11240631123861088,
      "grad_norm": 2.9756603240966797,
      "learning_rate": 9.982030911632903e-06,
      "loss": 0.9901,
      "step": 1391
    },
    {
      "epoch": 0.11248712095193843,
      "grad_norm": 2.9206433296203613,
      "learning_rate": 9.981975442061527e-06,
      "loss": 0.9961,
      "step": 1392
    },
    {
      "epoch": 0.11256793066526596,
      "grad_norm": 2.865624189376831,
      "learning_rate": 9.981919887160907e-06,
      "loss": 1.0601,
      "step": 1393
    },
    {
      "epoch": 0.1126487403785935,
      "grad_norm": 2.8166418075561523,
      "learning_rate": 9.98186424693199e-06,
      "loss": 0.9622,
      "step": 1394
    },
    {
      "epoch": 0.11272955009192105,
      "grad_norm": 2.8043010234832764,
      "learning_rate": 9.981808521375733e-06,
      "loss": 1.0692,
      "step": 1395
    },
    {
      "epoch": 0.11281035980524859,
      "grad_norm": 3.0252106189727783,
      "learning_rate": 9.98175271049309e-06,
      "loss": 1.0087,
      "step": 1396
    },
    {
      "epoch": 0.11289116951857613,
      "grad_norm": 3.2231223583221436,
      "learning_rate": 9.981696814285016e-06,
      "loss": 1.0396,
      "step": 1397
    },
    {
      "epoch": 0.11297197923190368,
      "grad_norm": 3.1894619464874268,
      "learning_rate": 9.981640832752469e-06,
      "loss": 1.0445,
      "step": 1398
    },
    {
      "epoch": 0.11305278894523121,
      "grad_norm": 2.6307520866394043,
      "learning_rate": 9.981584765896406e-06,
      "loss": 1.0222,
      "step": 1399
    },
    {
      "epoch": 0.11313359865855876,
      "grad_norm": 3.258815050125122,
      "learning_rate": 9.981528613717789e-06,
      "loss": 1.1528,
      "step": 1400
    },
    {
      "epoch": 0.1132144083718863,
      "grad_norm": 2.789964199066162,
      "learning_rate": 9.981472376217579e-06,
      "loss": 1.01,
      "step": 1401
    },
    {
      "epoch": 0.11329521808521384,
      "grad_norm": 2.6406161785125732,
      "learning_rate": 9.981416053396741e-06,
      "loss": 0.9244,
      "step": 1402
    },
    {
      "epoch": 0.11337602779854139,
      "grad_norm": 2.730431318283081,
      "learning_rate": 9.981359645256237e-06,
      "loss": 1.0405,
      "step": 1403
    },
    {
      "epoch": 0.11345683751186893,
      "grad_norm": 2.996708631515503,
      "learning_rate": 9.981303151797036e-06,
      "loss": 1.098,
      "step": 1404
    },
    {
      "epoch": 0.11353764722519646,
      "grad_norm": 2.910529375076294,
      "learning_rate": 9.981246573020102e-06,
      "loss": 0.9583,
      "step": 1405
    },
    {
      "epoch": 0.11361845693852402,
      "grad_norm": 2.94097638130188,
      "learning_rate": 9.981189908926407e-06,
      "loss": 1.0203,
      "step": 1406
    },
    {
      "epoch": 0.11369926665185155,
      "grad_norm": 2.6606483459472656,
      "learning_rate": 9.981133159516921e-06,
      "loss": 0.9185,
      "step": 1407
    },
    {
      "epoch": 0.11378007636517909,
      "grad_norm": 2.9483907222747803,
      "learning_rate": 9.981076324792615e-06,
      "loss": 1.0317,
      "step": 1408
    },
    {
      "epoch": 0.11386088607850664,
      "grad_norm": 3.210864543914795,
      "learning_rate": 9.981019404754462e-06,
      "loss": 0.9826,
      "step": 1409
    },
    {
      "epoch": 0.11394169579183418,
      "grad_norm": 2.6373202800750732,
      "learning_rate": 9.98096239940344e-06,
      "loss": 1.0134,
      "step": 1410
    },
    {
      "epoch": 0.11402250550516171,
      "grad_norm": 2.9394280910491943,
      "learning_rate": 9.980905308740521e-06,
      "loss": 1.1652,
      "step": 1411
    },
    {
      "epoch": 0.11410331521848927,
      "grad_norm": 3.0923264026641846,
      "learning_rate": 9.980848132766688e-06,
      "loss": 1.0251,
      "step": 1412
    },
    {
      "epoch": 0.1141841249318168,
      "grad_norm": 3.204538583755493,
      "learning_rate": 9.980790871482914e-06,
      "loss": 1.0652,
      "step": 1413
    },
    {
      "epoch": 0.11426493464514435,
      "grad_norm": 2.6705994606018066,
      "learning_rate": 9.980733524890186e-06,
      "loss": 1.1874,
      "step": 1414
    },
    {
      "epoch": 0.11434574435847189,
      "grad_norm": 2.97686505317688,
      "learning_rate": 9.980676092989481e-06,
      "loss": 0.906,
      "step": 1415
    },
    {
      "epoch": 0.11442655407179943,
      "grad_norm": 2.7322659492492676,
      "learning_rate": 9.980618575781788e-06,
      "loss": 1.0015,
      "step": 1416
    },
    {
      "epoch": 0.11450736378512698,
      "grad_norm": 3.1375131607055664,
      "learning_rate": 9.980560973268088e-06,
      "loss": 0.9007,
      "step": 1417
    },
    {
      "epoch": 0.11458817349845452,
      "grad_norm": 2.9137816429138184,
      "learning_rate": 9.980503285449368e-06,
      "loss": 1.0052,
      "step": 1418
    },
    {
      "epoch": 0.11466898321178205,
      "grad_norm": 2.709336280822754,
      "learning_rate": 9.980445512326616e-06,
      "loss": 1.0539,
      "step": 1419
    },
    {
      "epoch": 0.1147497929251096,
      "grad_norm": 3.1966333389282227,
      "learning_rate": 9.980387653900822e-06,
      "loss": 0.9864,
      "step": 1420
    },
    {
      "epoch": 0.11483060263843714,
      "grad_norm": 2.9226574897766113,
      "learning_rate": 9.98032971017298e-06,
      "loss": 1.1161,
      "step": 1421
    },
    {
      "epoch": 0.11491141235176468,
      "grad_norm": 3.04280161857605,
      "learning_rate": 9.980271681144078e-06,
      "loss": 1.0011,
      "step": 1422
    },
    {
      "epoch": 0.11499222206509223,
      "grad_norm": 2.8626012802124023,
      "learning_rate": 9.980213566815111e-06,
      "loss": 1.1177,
      "step": 1423
    },
    {
      "epoch": 0.11507303177841977,
      "grad_norm": 2.996809720993042,
      "learning_rate": 9.980155367187077e-06,
      "loss": 1.014,
      "step": 1424
    },
    {
      "epoch": 0.1151538414917473,
      "grad_norm": 2.845885753631592,
      "learning_rate": 9.980097082260968e-06,
      "loss": 1.04,
      "step": 1425
    },
    {
      "epoch": 0.11523465120507485,
      "grad_norm": 2.825157403945923,
      "learning_rate": 9.980038712037788e-06,
      "loss": 1.13,
      "step": 1426
    },
    {
      "epoch": 0.11531546091840239,
      "grad_norm": 3.3714938163757324,
      "learning_rate": 9.979980256518534e-06,
      "loss": 1.0191,
      "step": 1427
    },
    {
      "epoch": 0.11539627063172993,
      "grad_norm": 3.208986520767212,
      "learning_rate": 9.979921715704204e-06,
      "loss": 1.1193,
      "step": 1428
    },
    {
      "epoch": 0.11547708034505748,
      "grad_norm": 3.0360562801361084,
      "learning_rate": 9.979863089595804e-06,
      "loss": 1.0445,
      "step": 1429
    },
    {
      "epoch": 0.11555789005838502,
      "grad_norm": 2.9771480560302734,
      "learning_rate": 9.979804378194339e-06,
      "loss": 0.9945,
      "step": 1430
    },
    {
      "epoch": 0.11563869977171257,
      "grad_norm": 2.920020341873169,
      "learning_rate": 9.979745581500815e-06,
      "loss": 1.0168,
      "step": 1431
    },
    {
      "epoch": 0.1157195094850401,
      "grad_norm": 3.5531094074249268,
      "learning_rate": 9.979686699516236e-06,
      "loss": 1.1163,
      "step": 1432
    },
    {
      "epoch": 0.11580031919836764,
      "grad_norm": 2.9046502113342285,
      "learning_rate": 9.979627732241613e-06,
      "loss": 0.9846,
      "step": 1433
    },
    {
      "epoch": 0.11588112891169519,
      "grad_norm": 3.326899528503418,
      "learning_rate": 9.979568679677952e-06,
      "loss": 1.0666,
      "step": 1434
    },
    {
      "epoch": 0.11596193862502273,
      "grad_norm": 2.7189176082611084,
      "learning_rate": 9.97950954182627e-06,
      "loss": 1.0358,
      "step": 1435
    },
    {
      "epoch": 0.11604274833835027,
      "grad_norm": 2.8128626346588135,
      "learning_rate": 9.979450318687576e-06,
      "loss": 1.0221,
      "step": 1436
    },
    {
      "epoch": 0.11612355805167782,
      "grad_norm": 2.7262070178985596,
      "learning_rate": 9.979391010262885e-06,
      "loss": 0.9747,
      "step": 1437
    },
    {
      "epoch": 0.11620436776500535,
      "grad_norm": 2.7805073261260986,
      "learning_rate": 9.979331616553215e-06,
      "loss": 0.9004,
      "step": 1438
    },
    {
      "epoch": 0.11628517747833289,
      "grad_norm": 2.975998878479004,
      "learning_rate": 9.979272137559582e-06,
      "loss": 1.0997,
      "step": 1439
    },
    {
      "epoch": 0.11636598719166044,
      "grad_norm": 3.03562068939209,
      "learning_rate": 9.979212573283002e-06,
      "loss": 0.9495,
      "step": 1440
    },
    {
      "epoch": 0.11644679690498798,
      "grad_norm": 3.038222074508667,
      "learning_rate": 9.9791529237245e-06,
      "loss": 1.0314,
      "step": 1441
    },
    {
      "epoch": 0.11652760661831552,
      "grad_norm": 3.4328672885894775,
      "learning_rate": 9.979093188885095e-06,
      "loss": 0.9923,
      "step": 1442
    },
    {
      "epoch": 0.11660841633164307,
      "grad_norm": 2.62213134765625,
      "learning_rate": 9.979033368765806e-06,
      "loss": 1.1393,
      "step": 1443
    },
    {
      "epoch": 0.1166892260449706,
      "grad_norm": 2.790879726409912,
      "learning_rate": 9.978973463367668e-06,
      "loss": 1.103,
      "step": 1444
    },
    {
      "epoch": 0.11677003575829814,
      "grad_norm": 3.236301898956299,
      "learning_rate": 9.978913472691697e-06,
      "loss": 0.986,
      "step": 1445
    },
    {
      "epoch": 0.11685084547162569,
      "grad_norm": 3.7407853603363037,
      "learning_rate": 9.978853396738926e-06,
      "loss": 1.0055,
      "step": 1446
    },
    {
      "epoch": 0.11693165518495323,
      "grad_norm": 2.775300979614258,
      "learning_rate": 9.978793235510382e-06,
      "loss": 1.0833,
      "step": 1447
    },
    {
      "epoch": 0.11701246489828078,
      "grad_norm": 2.7946441173553467,
      "learning_rate": 9.978732989007096e-06,
      "loss": 1.1405,
      "step": 1448
    },
    {
      "epoch": 0.11709327461160832,
      "grad_norm": 2.7497661113739014,
      "learning_rate": 9.978672657230099e-06,
      "loss": 1.0585,
      "step": 1449
    },
    {
      "epoch": 0.11717408432493585,
      "grad_norm": 3.4076762199401855,
      "learning_rate": 9.978612240180428e-06,
      "loss": 1.0371,
      "step": 1450
    },
    {
      "epoch": 0.1172548940382634,
      "grad_norm": 2.9828991889953613,
      "learning_rate": 9.978551737859112e-06,
      "loss": 1.0987,
      "step": 1451
    },
    {
      "epoch": 0.11733570375159094,
      "grad_norm": 2.623058795928955,
      "learning_rate": 9.97849115026719e-06,
      "loss": 1.0565,
      "step": 1452
    },
    {
      "epoch": 0.11741651346491848,
      "grad_norm": 3.308063268661499,
      "learning_rate": 9.9784304774057e-06,
      "loss": 0.9084,
      "step": 1453
    },
    {
      "epoch": 0.11749732317824603,
      "grad_norm": 2.9729201793670654,
      "learning_rate": 9.978369719275682e-06,
      "loss": 1.1879,
      "step": 1454
    },
    {
      "epoch": 0.11757813289157357,
      "grad_norm": 3.2391316890716553,
      "learning_rate": 9.978308875878176e-06,
      "loss": 1.0461,
      "step": 1455
    },
    {
      "epoch": 0.1176589426049011,
      "grad_norm": 3.392163038253784,
      "learning_rate": 9.978247947214223e-06,
      "loss": 1.0646,
      "step": 1456
    },
    {
      "epoch": 0.11773975231822865,
      "grad_norm": 2.89571213722229,
      "learning_rate": 9.978186933284868e-06,
      "loss": 0.9876,
      "step": 1457
    },
    {
      "epoch": 0.11782056203155619,
      "grad_norm": 2.7080161571502686,
      "learning_rate": 9.978125834091155e-06,
      "loss": 1.0437,
      "step": 1458
    },
    {
      "epoch": 0.11790137174488373,
      "grad_norm": 3.3069257736206055,
      "learning_rate": 9.97806464963413e-06,
      "loss": 1.0233,
      "step": 1459
    },
    {
      "epoch": 0.11798218145821128,
      "grad_norm": 2.981694221496582,
      "learning_rate": 9.978003379914843e-06,
      "loss": 1.0159,
      "step": 1460
    },
    {
      "epoch": 0.11806299117153882,
      "grad_norm": 2.967569351196289,
      "learning_rate": 9.977942024934339e-06,
      "loss": 0.9759,
      "step": 1461
    },
    {
      "epoch": 0.11814380088486635,
      "grad_norm": 3.4966399669647217,
      "learning_rate": 9.977880584693676e-06,
      "loss": 1.0371,
      "step": 1462
    },
    {
      "epoch": 0.1182246105981939,
      "grad_norm": 2.8589589595794678,
      "learning_rate": 9.977819059193901e-06,
      "loss": 1.002,
      "step": 1463
    },
    {
      "epoch": 0.11830542031152144,
      "grad_norm": 2.8677189350128174,
      "learning_rate": 9.977757448436068e-06,
      "loss": 0.9869,
      "step": 1464
    },
    {
      "epoch": 0.11838623002484899,
      "grad_norm": 3.4240634441375732,
      "learning_rate": 9.977695752421235e-06,
      "loss": 0.9439,
      "step": 1465
    },
    {
      "epoch": 0.11846703973817653,
      "grad_norm": 3.0775208473205566,
      "learning_rate": 9.977633971150455e-06,
      "loss": 0.9236,
      "step": 1466
    },
    {
      "epoch": 0.11854784945150407,
      "grad_norm": 2.7149441242218018,
      "learning_rate": 9.97757210462479e-06,
      "loss": 0.8936,
      "step": 1467
    },
    {
      "epoch": 0.11862865916483162,
      "grad_norm": 3.633910894393921,
      "learning_rate": 9.977510152845298e-06,
      "loss": 1.1891,
      "step": 1468
    },
    {
      "epoch": 0.11870946887815916,
      "grad_norm": 2.748262405395508,
      "learning_rate": 9.97744811581304e-06,
      "loss": 0.9211,
      "step": 1469
    },
    {
      "epoch": 0.11879027859148669,
      "grad_norm": 2.662281036376953,
      "learning_rate": 9.977385993529077e-06,
      "loss": 1.0569,
      "step": 1470
    },
    {
      "epoch": 0.11887108830481424,
      "grad_norm": 2.6894032955169678,
      "learning_rate": 9.977323785994475e-06,
      "loss": 1.033,
      "step": 1471
    },
    {
      "epoch": 0.11895189801814178,
      "grad_norm": 2.809366226196289,
      "learning_rate": 9.977261493210299e-06,
      "loss": 1.005,
      "step": 1472
    },
    {
      "epoch": 0.11903270773146932,
      "grad_norm": 3.1424038410186768,
      "learning_rate": 9.977199115177616e-06,
      "loss": 1.0223,
      "step": 1473
    },
    {
      "epoch": 0.11911351744479687,
      "grad_norm": 2.9231789112091064,
      "learning_rate": 9.977136651897495e-06,
      "loss": 0.9585,
      "step": 1474
    },
    {
      "epoch": 0.1191943271581244,
      "grad_norm": 3.0695743560791016,
      "learning_rate": 9.977074103371005e-06,
      "loss": 1.0148,
      "step": 1475
    },
    {
      "epoch": 0.11927513687145194,
      "grad_norm": 3.018730640411377,
      "learning_rate": 9.977011469599218e-06,
      "loss": 1.01,
      "step": 1476
    },
    {
      "epoch": 0.1193559465847795,
      "grad_norm": 2.963622570037842,
      "learning_rate": 9.976948750583204e-06,
      "loss": 0.9736,
      "step": 1477
    },
    {
      "epoch": 0.11943675629810703,
      "grad_norm": 2.6269237995147705,
      "learning_rate": 9.976885946324042e-06,
      "loss": 0.9666,
      "step": 1478
    },
    {
      "epoch": 0.11951756601143458,
      "grad_norm": 2.6247920989990234,
      "learning_rate": 9.976823056822802e-06,
      "loss": 1.0616,
      "step": 1479
    },
    {
      "epoch": 0.11959837572476212,
      "grad_norm": 2.672311782836914,
      "learning_rate": 9.976760082080567e-06,
      "loss": 1.0102,
      "step": 1480
    },
    {
      "epoch": 0.11967918543808966,
      "grad_norm": 3.7789974212646484,
      "learning_rate": 9.976697022098411e-06,
      "loss": 0.9777,
      "step": 1481
    },
    {
      "epoch": 0.1197599951514172,
      "grad_norm": 2.782094955444336,
      "learning_rate": 9.976633876877417e-06,
      "loss": 1.1142,
      "step": 1482
    },
    {
      "epoch": 0.11984080486474474,
      "grad_norm": 2.689985513687134,
      "learning_rate": 9.976570646418665e-06,
      "loss": 1.038,
      "step": 1483
    },
    {
      "epoch": 0.11992161457807228,
      "grad_norm": 2.7190632820129395,
      "learning_rate": 9.97650733072324e-06,
      "loss": 0.999,
      "step": 1484
    },
    {
      "epoch": 0.12000242429139983,
      "grad_norm": 2.327003240585327,
      "learning_rate": 9.976443929792223e-06,
      "loss": 1.0793,
      "step": 1485
    },
    {
      "epoch": 0.12008323400472737,
      "grad_norm": 3.1247661113739014,
      "learning_rate": 9.976380443626701e-06,
      "loss": 1.0358,
      "step": 1486
    },
    {
      "epoch": 0.1201640437180549,
      "grad_norm": 3.1908297538757324,
      "learning_rate": 9.976316872227765e-06,
      "loss": 1.0745,
      "step": 1487
    },
    {
      "epoch": 0.12024485343138246,
      "grad_norm": 2.9428727626800537,
      "learning_rate": 9.976253215596498e-06,
      "loss": 1.0319,
      "step": 1488
    },
    {
      "epoch": 0.12032566314471,
      "grad_norm": 3.5103609561920166,
      "learning_rate": 9.976189473733995e-06,
      "loss": 0.9165,
      "step": 1489
    },
    {
      "epoch": 0.12040647285803753,
      "grad_norm": 2.866847276687622,
      "learning_rate": 9.976125646641346e-06,
      "loss": 0.9826,
      "step": 1490
    },
    {
      "epoch": 0.12048728257136508,
      "grad_norm": 2.668189287185669,
      "learning_rate": 9.976061734319644e-06,
      "loss": 1.0166,
      "step": 1491
    },
    {
      "epoch": 0.12056809228469262,
      "grad_norm": 2.6091597080230713,
      "learning_rate": 9.975997736769984e-06,
      "loss": 1.0109,
      "step": 1492
    },
    {
      "epoch": 0.12064890199802016,
      "grad_norm": 3.3840620517730713,
      "learning_rate": 9.975933653993462e-06,
      "loss": 1.1303,
      "step": 1493
    },
    {
      "epoch": 0.1207297117113477,
      "grad_norm": 2.5798120498657227,
      "learning_rate": 9.975869485991175e-06,
      "loss": 1.1408,
      "step": 1494
    },
    {
      "epoch": 0.12081052142467524,
      "grad_norm": 3.8495700359344482,
      "learning_rate": 9.975805232764223e-06,
      "loss": 1.1363,
      "step": 1495
    },
    {
      "epoch": 0.1208913311380028,
      "grad_norm": 2.622189998626709,
      "learning_rate": 9.975740894313706e-06,
      "loss": 1.0252,
      "step": 1496
    },
    {
      "epoch": 0.12097214085133033,
      "grad_norm": 2.8880317211151123,
      "learning_rate": 9.975676470640724e-06,
      "loss": 0.9853,
      "step": 1497
    },
    {
      "epoch": 0.12105295056465787,
      "grad_norm": 3.6159539222717285,
      "learning_rate": 9.975611961746387e-06,
      "loss": 1.0694,
      "step": 1498
    },
    {
      "epoch": 0.12113376027798542,
      "grad_norm": 4.053859710693359,
      "learning_rate": 9.975547367631793e-06,
      "loss": 1.0925,
      "step": 1499
    },
    {
      "epoch": 0.12121456999131296,
      "grad_norm": 2.7435827255249023,
      "learning_rate": 9.97548268829805e-06,
      "loss": 1.0758,
      "step": 1500
    },
    {
      "epoch": 0.1212953797046405,
      "grad_norm": 2.9291670322418213,
      "learning_rate": 9.975417923746268e-06,
      "loss": 1.0237,
      "step": 1501
    },
    {
      "epoch": 0.12137618941796804,
      "grad_norm": 2.6226656436920166,
      "learning_rate": 9.975353073977555e-06,
      "loss": 1.0267,
      "step": 1502
    },
    {
      "epoch": 0.12145699913129558,
      "grad_norm": 2.6459484100341797,
      "learning_rate": 9.97528813899302e-06,
      "loss": 0.9548,
      "step": 1503
    },
    {
      "epoch": 0.12153780884462312,
      "grad_norm": 3.0570766925811768,
      "learning_rate": 9.975223118793776e-06,
      "loss": 0.9643,
      "step": 1504
    },
    {
      "epoch": 0.12161861855795067,
      "grad_norm": 3.1126012802124023,
      "learning_rate": 9.97515801338094e-06,
      "loss": 1.0619,
      "step": 1505
    },
    {
      "epoch": 0.1216994282712782,
      "grad_norm": 3.191077947616577,
      "learning_rate": 9.975092822755623e-06,
      "loss": 0.8965,
      "step": 1506
    },
    {
      "epoch": 0.12178023798460574,
      "grad_norm": 2.781688690185547,
      "learning_rate": 9.975027546918943e-06,
      "loss": 1.0474,
      "step": 1507
    },
    {
      "epoch": 0.1218610476979333,
      "grad_norm": 2.710775375366211,
      "learning_rate": 9.974962185872017e-06,
      "loss": 1.0024,
      "step": 1508
    },
    {
      "epoch": 0.12194185741126083,
      "grad_norm": 3.1338086128234863,
      "learning_rate": 9.974896739615967e-06,
      "loss": 1.0516,
      "step": 1509
    },
    {
      "epoch": 0.12202266712458837,
      "grad_norm": 2.9669113159179688,
      "learning_rate": 9.974831208151912e-06,
      "loss": 0.9887,
      "step": 1510
    },
    {
      "epoch": 0.12210347683791592,
      "grad_norm": 3.1898107528686523,
      "learning_rate": 9.974765591480975e-06,
      "loss": 1.129,
      "step": 1511
    },
    {
      "epoch": 0.12218428655124346,
      "grad_norm": 2.8128304481506348,
      "learning_rate": 9.97469988960428e-06,
      "loss": 0.9935,
      "step": 1512
    },
    {
      "epoch": 0.12226509626457101,
      "grad_norm": 2.8253588676452637,
      "learning_rate": 9.974634102522951e-06,
      "loss": 1.022,
      "step": 1513
    },
    {
      "epoch": 0.12234590597789854,
      "grad_norm": 3.3674936294555664,
      "learning_rate": 9.974568230238116e-06,
      "loss": 1.0668,
      "step": 1514
    },
    {
      "epoch": 0.12242671569122608,
      "grad_norm": 2.8994176387786865,
      "learning_rate": 9.974502272750904e-06,
      "loss": 0.952,
      "step": 1515
    },
    {
      "epoch": 0.12250752540455363,
      "grad_norm": 3.3196377754211426,
      "learning_rate": 9.974436230062443e-06,
      "loss": 1.0472,
      "step": 1516
    },
    {
      "epoch": 0.12258833511788117,
      "grad_norm": 2.662877321243286,
      "learning_rate": 9.974370102173864e-06,
      "loss": 1.0556,
      "step": 1517
    },
    {
      "epoch": 0.1226691448312087,
      "grad_norm": 2.990156888961792,
      "learning_rate": 9.974303889086302e-06,
      "loss": 0.9752,
      "step": 1518
    },
    {
      "epoch": 0.12274995454453626,
      "grad_norm": 3.2955334186553955,
      "learning_rate": 9.974237590800888e-06,
      "loss": 0.9966,
      "step": 1519
    },
    {
      "epoch": 0.1228307642578638,
      "grad_norm": 2.9709582328796387,
      "learning_rate": 9.974171207318762e-06,
      "loss": 0.9343,
      "step": 1520
    },
    {
      "epoch": 0.12291157397119133,
      "grad_norm": 3.0959041118621826,
      "learning_rate": 9.974104738641056e-06,
      "loss": 1.076,
      "step": 1521
    },
    {
      "epoch": 0.12299238368451888,
      "grad_norm": 2.8134162425994873,
      "learning_rate": 9.974038184768908e-06,
      "loss": 0.911,
      "step": 1522
    },
    {
      "epoch": 0.12307319339784642,
      "grad_norm": 2.486924171447754,
      "learning_rate": 9.973971545703464e-06,
      "loss": 0.94,
      "step": 1523
    },
    {
      "epoch": 0.12315400311117396,
      "grad_norm": 3.090559244155884,
      "learning_rate": 9.97390482144586e-06,
      "loss": 0.9848,
      "step": 1524
    },
    {
      "epoch": 0.12323481282450151,
      "grad_norm": 2.6111233234405518,
      "learning_rate": 9.97383801199724e-06,
      "loss": 1.077,
      "step": 1525
    },
    {
      "epoch": 0.12331562253782904,
      "grad_norm": 2.474102020263672,
      "learning_rate": 9.973771117358751e-06,
      "loss": 1.0163,
      "step": 1526
    },
    {
      "epoch": 0.1233964322511566,
      "grad_norm": 3.4042325019836426,
      "learning_rate": 9.973704137531537e-06,
      "loss": 1.054,
      "step": 1527
    },
    {
      "epoch": 0.12347724196448413,
      "grad_norm": 2.7256107330322266,
      "learning_rate": 9.973637072516742e-06,
      "loss": 0.957,
      "step": 1528
    },
    {
      "epoch": 0.12355805167781167,
      "grad_norm": 3.280165910720825,
      "learning_rate": 9.97356992231552e-06,
      "loss": 1.0552,
      "step": 1529
    },
    {
      "epoch": 0.12363886139113922,
      "grad_norm": 2.8168840408325195,
      "learning_rate": 9.973502686929018e-06,
      "loss": 0.8983,
      "step": 1530
    },
    {
      "epoch": 0.12371967110446676,
      "grad_norm": 2.5056023597717285,
      "learning_rate": 9.973435366358388e-06,
      "loss": 1.1126,
      "step": 1531
    },
    {
      "epoch": 0.1238004808177943,
      "grad_norm": 3.222533702850342,
      "learning_rate": 9.973367960604783e-06,
      "loss": 0.9352,
      "step": 1532
    },
    {
      "epoch": 0.12388129053112185,
      "grad_norm": 3.013413906097412,
      "learning_rate": 9.973300469669357e-06,
      "loss": 0.9878,
      "step": 1533
    },
    {
      "epoch": 0.12396210024444938,
      "grad_norm": 2.886183977127075,
      "learning_rate": 9.973232893553268e-06,
      "loss": 1.0524,
      "step": 1534
    },
    {
      "epoch": 0.12404290995777692,
      "grad_norm": 2.5032222270965576,
      "learning_rate": 9.973165232257671e-06,
      "loss": 0.9094,
      "step": 1535
    },
    {
      "epoch": 0.12412371967110447,
      "grad_norm": 3.167069673538208,
      "learning_rate": 9.973097485783727e-06,
      "loss": 0.9374,
      "step": 1536
    },
    {
      "epoch": 0.12420452938443201,
      "grad_norm": 3.163844108581543,
      "learning_rate": 9.973029654132595e-06,
      "loss": 0.8943,
      "step": 1537
    },
    {
      "epoch": 0.12428533909775955,
      "grad_norm": 2.601196765899658,
      "learning_rate": 9.972961737305437e-06,
      "loss": 1.0731,
      "step": 1538
    },
    {
      "epoch": 0.1243661488110871,
      "grad_norm": 3.303682565689087,
      "learning_rate": 9.972893735303414e-06,
      "loss": 1.0653,
      "step": 1539
    },
    {
      "epoch": 0.12444695852441463,
      "grad_norm": 3.289884328842163,
      "learning_rate": 9.972825648127697e-06,
      "loss": 1.0373,
      "step": 1540
    },
    {
      "epoch": 0.12452776823774217,
      "grad_norm": 2.7413132190704346,
      "learning_rate": 9.972757475779446e-06,
      "loss": 0.9449,
      "step": 1541
    },
    {
      "epoch": 0.12460857795106972,
      "grad_norm": 2.809582233428955,
      "learning_rate": 9.972689218259831e-06,
      "loss": 1.0173,
      "step": 1542
    },
    {
      "epoch": 0.12468938766439726,
      "grad_norm": 2.9432179927825928,
      "learning_rate": 9.972620875570022e-06,
      "loss": 1.0912,
      "step": 1543
    },
    {
      "epoch": 0.12477019737772481,
      "grad_norm": 3.4051713943481445,
      "learning_rate": 9.972552447711188e-06,
      "loss": 1.1274,
      "step": 1544
    },
    {
      "epoch": 0.12485100709105235,
      "grad_norm": 3.025392770767212,
      "learning_rate": 9.972483934684503e-06,
      "loss": 1.0697,
      "step": 1545
    },
    {
      "epoch": 0.12493181680437988,
      "grad_norm": 3.9957685470581055,
      "learning_rate": 9.972415336491137e-06,
      "loss": 0.9908,
      "step": 1546
    },
    {
      "epoch": 0.12501262651770742,
      "grad_norm": 2.8916802406311035,
      "learning_rate": 9.972346653132266e-06,
      "loss": 0.942,
      "step": 1547
    },
    {
      "epoch": 0.12509343623103497,
      "grad_norm": 3.4179534912109375,
      "learning_rate": 9.97227788460907e-06,
      "loss": 1.1929,
      "step": 1548
    },
    {
      "epoch": 0.12517424594436252,
      "grad_norm": 3.1534507274627686,
      "learning_rate": 9.972209030922722e-06,
      "loss": 0.9994,
      "step": 1549
    },
    {
      "epoch": 0.12525505565769005,
      "grad_norm": 3.0198779106140137,
      "learning_rate": 9.972140092074404e-06,
      "loss": 1.015,
      "step": 1550
    },
    {
      "epoch": 0.1253358653710176,
      "grad_norm": 3.012791395187378,
      "learning_rate": 9.972071068065297e-06,
      "loss": 1.1126,
      "step": 1551
    },
    {
      "epoch": 0.12541667508434515,
      "grad_norm": 2.9998300075531006,
      "learning_rate": 9.97200195889658e-06,
      "loss": 0.9682,
      "step": 1552
    },
    {
      "epoch": 0.12549748479767267,
      "grad_norm": 2.617751121520996,
      "learning_rate": 9.97193276456944e-06,
      "loss": 1.1222,
      "step": 1553
    },
    {
      "epoch": 0.12557829451100022,
      "grad_norm": 2.919160842895508,
      "learning_rate": 9.971863485085063e-06,
      "loss": 1.104,
      "step": 1554
    },
    {
      "epoch": 0.12565910422432777,
      "grad_norm": 3.4130806922912598,
      "learning_rate": 9.971794120444633e-06,
      "loss": 1.0856,
      "step": 1555
    },
    {
      "epoch": 0.1257399139376553,
      "grad_norm": 2.5439846515655518,
      "learning_rate": 9.97172467064934e-06,
      "loss": 0.9971,
      "step": 1556
    },
    {
      "epoch": 0.12582072365098285,
      "grad_norm": 2.987593650817871,
      "learning_rate": 9.971655135700369e-06,
      "loss": 0.9025,
      "step": 1557
    },
    {
      "epoch": 0.1259015333643104,
      "grad_norm": 2.4838151931762695,
      "learning_rate": 9.971585515598916e-06,
      "loss": 1.0237,
      "step": 1558
    },
    {
      "epoch": 0.12598234307763792,
      "grad_norm": 3.056072473526001,
      "learning_rate": 9.971515810346172e-06,
      "loss": 0.982,
      "step": 1559
    },
    {
      "epoch": 0.12606315279096547,
      "grad_norm": 2.6471681594848633,
      "learning_rate": 9.97144601994333e-06,
      "loss": 1.081,
      "step": 1560
    },
    {
      "epoch": 0.12614396250429302,
      "grad_norm": 2.814272880554199,
      "learning_rate": 9.971376144391587e-06,
      "loss": 1.0383,
      "step": 1561
    },
    {
      "epoch": 0.12622477221762055,
      "grad_norm": 3.0984888076782227,
      "learning_rate": 9.971306183692138e-06,
      "loss": 1.0266,
      "step": 1562
    },
    {
      "epoch": 0.1263055819309481,
      "grad_norm": 2.8970961570739746,
      "learning_rate": 9.971236137846181e-06,
      "loss": 1.0187,
      "step": 1563
    },
    {
      "epoch": 0.12638639164427565,
      "grad_norm": 3.058929920196533,
      "learning_rate": 9.971166006854918e-06,
      "loss": 1.0151,
      "step": 1564
    },
    {
      "epoch": 0.12646720135760317,
      "grad_norm": 3.1027820110321045,
      "learning_rate": 9.971095790719549e-06,
      "loss": 1.0664,
      "step": 1565
    },
    {
      "epoch": 0.12654801107093072,
      "grad_norm": 2.7299392223358154,
      "learning_rate": 9.971025489441277e-06,
      "loss": 1.0323,
      "step": 1566
    },
    {
      "epoch": 0.12662882078425827,
      "grad_norm": 2.9159555435180664,
      "learning_rate": 9.970955103021304e-06,
      "loss": 1.003,
      "step": 1567
    },
    {
      "epoch": 0.12670963049758582,
      "grad_norm": 2.9966683387756348,
      "learning_rate": 9.970884631460837e-06,
      "loss": 1.0543,
      "step": 1568
    },
    {
      "epoch": 0.12679044021091335,
      "grad_norm": 2.942044258117676,
      "learning_rate": 9.970814074761086e-06,
      "loss": 1.0306,
      "step": 1569
    },
    {
      "epoch": 0.1268712499242409,
      "grad_norm": 3.0210185050964355,
      "learning_rate": 9.970743432923254e-06,
      "loss": 1.01,
      "step": 1570
    },
    {
      "epoch": 0.12695205963756845,
      "grad_norm": 3.9895670413970947,
      "learning_rate": 9.970672705948554e-06,
      "loss": 0.9611,
      "step": 1571
    },
    {
      "epoch": 0.12703286935089597,
      "grad_norm": 3.2037484645843506,
      "learning_rate": 9.970601893838199e-06,
      "loss": 1.0619,
      "step": 1572
    },
    {
      "epoch": 0.12711367906422352,
      "grad_norm": 2.9033043384552,
      "learning_rate": 9.970530996593396e-06,
      "loss": 0.9644,
      "step": 1573
    },
    {
      "epoch": 0.12719448877755107,
      "grad_norm": 2.760676383972168,
      "learning_rate": 9.970460014215365e-06,
      "loss": 1.0322,
      "step": 1574
    },
    {
      "epoch": 0.1272752984908786,
      "grad_norm": 2.8878512382507324,
      "learning_rate": 9.970388946705322e-06,
      "loss": 1.1337,
      "step": 1575
    },
    {
      "epoch": 0.12735610820420615,
      "grad_norm": 3.3586654663085938,
      "learning_rate": 9.97031779406448e-06,
      "loss": 1.0103,
      "step": 1576
    },
    {
      "epoch": 0.1274369179175337,
      "grad_norm": 2.880124807357788,
      "learning_rate": 9.970246556294059e-06,
      "loss": 1.1382,
      "step": 1577
    },
    {
      "epoch": 0.12751772763086122,
      "grad_norm": 2.892253875732422,
      "learning_rate": 9.970175233395282e-06,
      "loss": 1.0529,
      "step": 1578
    },
    {
      "epoch": 0.12759853734418877,
      "grad_norm": 2.8811302185058594,
      "learning_rate": 9.970103825369368e-06,
      "loss": 1.0315,
      "step": 1579
    },
    {
      "epoch": 0.12767934705751632,
      "grad_norm": 2.897580623626709,
      "learning_rate": 9.970032332217539e-06,
      "loss": 0.9286,
      "step": 1580
    },
    {
      "epoch": 0.12776015677084385,
      "grad_norm": 3.135737895965576,
      "learning_rate": 9.969960753941021e-06,
      "loss": 1.0004,
      "step": 1581
    },
    {
      "epoch": 0.1278409664841714,
      "grad_norm": 2.7766566276550293,
      "learning_rate": 9.96988909054104e-06,
      "loss": 0.9378,
      "step": 1582
    },
    {
      "epoch": 0.12792177619749895,
      "grad_norm": 2.9192938804626465,
      "learning_rate": 9.969817342018826e-06,
      "loss": 1.0221,
      "step": 1583
    },
    {
      "epoch": 0.12800258591082647,
      "grad_norm": 2.775981903076172,
      "learning_rate": 9.969745508375604e-06,
      "loss": 0.9592,
      "step": 1584
    },
    {
      "epoch": 0.12808339562415402,
      "grad_norm": 3.3218469619750977,
      "learning_rate": 9.969673589612604e-06,
      "loss": 1.1219,
      "step": 1585
    },
    {
      "epoch": 0.12816420533748157,
      "grad_norm": 2.779749631881714,
      "learning_rate": 9.96960158573106e-06,
      "loss": 1.0264,
      "step": 1586
    },
    {
      "epoch": 0.1282450150508091,
      "grad_norm": 2.96001935005188,
      "learning_rate": 9.969529496732205e-06,
      "loss": 1.1207,
      "step": 1587
    },
    {
      "epoch": 0.12832582476413665,
      "grad_norm": 3.388425588607788,
      "learning_rate": 9.969457322617273e-06,
      "loss": 0.9705,
      "step": 1588
    },
    {
      "epoch": 0.1284066344774642,
      "grad_norm": 2.4653358459472656,
      "learning_rate": 9.969385063387503e-06,
      "loss": 0.9421,
      "step": 1589
    },
    {
      "epoch": 0.12848744419079172,
      "grad_norm": 2.5082638263702393,
      "learning_rate": 9.969312719044127e-06,
      "loss": 1.0129,
      "step": 1590
    },
    {
      "epoch": 0.12856825390411927,
      "grad_norm": 3.0804243087768555,
      "learning_rate": 9.969240289588388e-06,
      "loss": 1.0081,
      "step": 1591
    },
    {
      "epoch": 0.12864906361744682,
      "grad_norm": 2.7551684379577637,
      "learning_rate": 9.969167775021527e-06,
      "loss": 0.9329,
      "step": 1592
    },
    {
      "epoch": 0.12872987333077435,
      "grad_norm": 2.7766008377075195,
      "learning_rate": 9.969095175344784e-06,
      "loss": 0.9295,
      "step": 1593
    },
    {
      "epoch": 0.1288106830441019,
      "grad_norm": 3.1291987895965576,
      "learning_rate": 9.969022490559403e-06,
      "loss": 1.052,
      "step": 1594
    },
    {
      "epoch": 0.12889149275742945,
      "grad_norm": 3.7330756187438965,
      "learning_rate": 9.96894972066663e-06,
      "loss": 1.0948,
      "step": 1595
    },
    {
      "epoch": 0.12897230247075697,
      "grad_norm": 2.907026529312134,
      "learning_rate": 9.968876865667709e-06,
      "loss": 0.9207,
      "step": 1596
    },
    {
      "epoch": 0.12905311218408452,
      "grad_norm": 2.8110721111297607,
      "learning_rate": 9.968803925563891e-06,
      "loss": 0.9706,
      "step": 1597
    },
    {
      "epoch": 0.12913392189741207,
      "grad_norm": 3.5314857959747314,
      "learning_rate": 9.968730900356423e-06,
      "loss": 0.9355,
      "step": 1598
    },
    {
      "epoch": 0.1292147316107396,
      "grad_norm": 3.0395474433898926,
      "learning_rate": 9.968657790046557e-06,
      "loss": 1.0052,
      "step": 1599
    },
    {
      "epoch": 0.12929554132406715,
      "grad_norm": 2.651458263397217,
      "learning_rate": 9.968584594635544e-06,
      "loss": 0.9559,
      "step": 1600
    },
    {
      "epoch": 0.1293763510373947,
      "grad_norm": 2.9642333984375,
      "learning_rate": 9.968511314124638e-06,
      "loss": 1.007,
      "step": 1601
    },
    {
      "epoch": 0.12945716075072225,
      "grad_norm": 2.743347644805908,
      "learning_rate": 9.968437948515094e-06,
      "loss": 1.0504,
      "step": 1602
    },
    {
      "epoch": 0.12953797046404977,
      "grad_norm": 2.830263614654541,
      "learning_rate": 9.96836449780817e-06,
      "loss": 1.003,
      "step": 1603
    },
    {
      "epoch": 0.12961878017737732,
      "grad_norm": 2.768022060394287,
      "learning_rate": 9.968290962005122e-06,
      "loss": 0.9653,
      "step": 1604
    },
    {
      "epoch": 0.12969958989070487,
      "grad_norm": 2.78212833404541,
      "learning_rate": 9.968217341107212e-06,
      "loss": 1.0442,
      "step": 1605
    },
    {
      "epoch": 0.1297803996040324,
      "grad_norm": 3.129190444946289,
      "learning_rate": 9.968143635115698e-06,
      "loss": 1.1285,
      "step": 1606
    },
    {
      "epoch": 0.12986120931735995,
      "grad_norm": 3.335914134979248,
      "learning_rate": 9.968069844031846e-06,
      "loss": 1.1157,
      "step": 1607
    },
    {
      "epoch": 0.1299420190306875,
      "grad_norm": 3.0782394409179688,
      "learning_rate": 9.967995967856917e-06,
      "loss": 1.0001,
      "step": 1608
    },
    {
      "epoch": 0.13002282874401502,
      "grad_norm": 2.9339089393615723,
      "learning_rate": 9.967922006592175e-06,
      "loss": 1.0452,
      "step": 1609
    },
    {
      "epoch": 0.13010363845734257,
      "grad_norm": 2.6225640773773193,
      "learning_rate": 9.96784796023889e-06,
      "loss": 1.1259,
      "step": 1610
    },
    {
      "epoch": 0.13018444817067013,
      "grad_norm": 2.4689254760742188,
      "learning_rate": 9.967773828798328e-06,
      "loss": 1.0334,
      "step": 1611
    },
    {
      "epoch": 0.13026525788399765,
      "grad_norm": 3.2011399269104004,
      "learning_rate": 9.967699612271762e-06,
      "loss": 1.0487,
      "step": 1612
    },
    {
      "epoch": 0.1303460675973252,
      "grad_norm": 3.048624277114868,
      "learning_rate": 9.967625310660461e-06,
      "loss": 1.0496,
      "step": 1613
    },
    {
      "epoch": 0.13042687731065275,
      "grad_norm": 2.8851726055145264,
      "learning_rate": 9.967550923965695e-06,
      "loss": 1.0354,
      "step": 1614
    },
    {
      "epoch": 0.13050768702398027,
      "grad_norm": 3.3419952392578125,
      "learning_rate": 9.967476452188742e-06,
      "loss": 1.0365,
      "step": 1615
    },
    {
      "epoch": 0.13058849673730782,
      "grad_norm": 2.9978623390197754,
      "learning_rate": 9.967401895330874e-06,
      "loss": 1.0527,
      "step": 1616
    },
    {
      "epoch": 0.13066930645063538,
      "grad_norm": 2.8629448413848877,
      "learning_rate": 9.967327253393373e-06,
      "loss": 1.0362,
      "step": 1617
    },
    {
      "epoch": 0.1307501161639629,
      "grad_norm": 2.6528220176696777,
      "learning_rate": 9.967252526377513e-06,
      "loss": 1.0123,
      "step": 1618
    },
    {
      "epoch": 0.13083092587729045,
      "grad_norm": 3.220398187637329,
      "learning_rate": 9.967177714284577e-06,
      "loss": 1.0968,
      "step": 1619
    },
    {
      "epoch": 0.130911735590618,
      "grad_norm": 3.31400728225708,
      "learning_rate": 9.967102817115844e-06,
      "loss": 1.04,
      "step": 1620
    },
    {
      "epoch": 0.13099254530394552,
      "grad_norm": 2.9584708213806152,
      "learning_rate": 9.967027834872595e-06,
      "loss": 1.0777,
      "step": 1621
    },
    {
      "epoch": 0.13107335501727307,
      "grad_norm": 2.9647164344787598,
      "learning_rate": 9.96695276755612e-06,
      "loss": 0.9648,
      "step": 1622
    },
    {
      "epoch": 0.13115416473060063,
      "grad_norm": 2.6207494735717773,
      "learning_rate": 9.9668776151677e-06,
      "loss": 1.0751,
      "step": 1623
    },
    {
      "epoch": 0.13123497444392815,
      "grad_norm": 3.3089005947113037,
      "learning_rate": 9.966802377708625e-06,
      "loss": 1.0214,
      "step": 1624
    },
    {
      "epoch": 0.1313157841572557,
      "grad_norm": 3.3470988273620605,
      "learning_rate": 9.966727055180183e-06,
      "loss": 1.0695,
      "step": 1625
    },
    {
      "epoch": 0.13139659387058325,
      "grad_norm": 2.7479801177978516,
      "learning_rate": 9.966651647583661e-06,
      "loss": 1.0169,
      "step": 1626
    },
    {
      "epoch": 0.13147740358391077,
      "grad_norm": 2.6470906734466553,
      "learning_rate": 9.966576154920354e-06,
      "loss": 0.998,
      "step": 1627
    },
    {
      "epoch": 0.13155821329723832,
      "grad_norm": 2.62980318069458,
      "learning_rate": 9.966500577191554e-06,
      "loss": 1.0047,
      "step": 1628
    },
    {
      "epoch": 0.13163902301056588,
      "grad_norm": 3.066617488861084,
      "learning_rate": 9.966424914398556e-06,
      "loss": 1.0842,
      "step": 1629
    },
    {
      "epoch": 0.1317198327238934,
      "grad_norm": 2.5637338161468506,
      "learning_rate": 9.966349166542655e-06,
      "loss": 1.093,
      "step": 1630
    },
    {
      "epoch": 0.13180064243722095,
      "grad_norm": 3.0110042095184326,
      "learning_rate": 9.966273333625149e-06,
      "loss": 1.0627,
      "step": 1631
    },
    {
      "epoch": 0.1318814521505485,
      "grad_norm": 2.872307777404785,
      "learning_rate": 9.966197415647336e-06,
      "loss": 0.9796,
      "step": 1632
    },
    {
      "epoch": 0.13196226186387605,
      "grad_norm": 2.9434759616851807,
      "learning_rate": 9.966121412610516e-06,
      "loss": 0.9855,
      "step": 1633
    },
    {
      "epoch": 0.13204307157720357,
      "grad_norm": 2.70803165435791,
      "learning_rate": 9.966045324515993e-06,
      "loss": 0.9988,
      "step": 1634
    },
    {
      "epoch": 0.13212388129053113,
      "grad_norm": 2.9613752365112305,
      "learning_rate": 9.96596915136507e-06,
      "loss": 1.0091,
      "step": 1635
    },
    {
      "epoch": 0.13220469100385868,
      "grad_norm": 2.6390221118927,
      "learning_rate": 9.965892893159048e-06,
      "loss": 1.196,
      "step": 1636
    },
    {
      "epoch": 0.1322855007171862,
      "grad_norm": 2.928917169570923,
      "learning_rate": 9.965816549899237e-06,
      "loss": 0.9174,
      "step": 1637
    },
    {
      "epoch": 0.13236631043051375,
      "grad_norm": 2.783160448074341,
      "learning_rate": 9.965740121586942e-06,
      "loss": 1.1806,
      "step": 1638
    },
    {
      "epoch": 0.1324471201438413,
      "grad_norm": 2.8499670028686523,
      "learning_rate": 9.965663608223476e-06,
      "loss": 1.0064,
      "step": 1639
    },
    {
      "epoch": 0.13252792985716882,
      "grad_norm": 3.1209301948547363,
      "learning_rate": 9.965587009810145e-06,
      "loss": 0.9955,
      "step": 1640
    },
    {
      "epoch": 0.13260873957049638,
      "grad_norm": 3.3921396732330322,
      "learning_rate": 9.965510326348263e-06,
      "loss": 1.0438,
      "step": 1641
    },
    {
      "epoch": 0.13268954928382393,
      "grad_norm": 2.538139820098877,
      "learning_rate": 9.965433557839143e-06,
      "loss": 1.0475,
      "step": 1642
    },
    {
      "epoch": 0.13277035899715145,
      "grad_norm": 3.514256238937378,
      "learning_rate": 9.9653567042841e-06,
      "loss": 0.9934,
      "step": 1643
    },
    {
      "epoch": 0.132851168710479,
      "grad_norm": 2.601105213165283,
      "learning_rate": 9.965279765684449e-06,
      "loss": 0.9303,
      "step": 1644
    },
    {
      "epoch": 0.13293197842380655,
      "grad_norm": 3.2366483211517334,
      "learning_rate": 9.96520274204151e-06,
      "loss": 1.024,
      "step": 1645
    },
    {
      "epoch": 0.13301278813713407,
      "grad_norm": 3.0515966415405273,
      "learning_rate": 9.965125633356602e-06,
      "loss": 1.0211,
      "step": 1646
    },
    {
      "epoch": 0.13309359785046163,
      "grad_norm": 3.162769317626953,
      "learning_rate": 9.965048439631045e-06,
      "loss": 1.0201,
      "step": 1647
    },
    {
      "epoch": 0.13317440756378918,
      "grad_norm": 2.630096197128296,
      "learning_rate": 9.964971160866159e-06,
      "loss": 1.1113,
      "step": 1648
    },
    {
      "epoch": 0.1332552172771167,
      "grad_norm": 2.9393413066864014,
      "learning_rate": 9.964893797063271e-06,
      "loss": 1.1372,
      "step": 1649
    },
    {
      "epoch": 0.13333602699044425,
      "grad_norm": 2.98698091506958,
      "learning_rate": 9.964816348223705e-06,
      "loss": 1.0106,
      "step": 1650
    },
    {
      "epoch": 0.1334168367037718,
      "grad_norm": 2.547166585922241,
      "learning_rate": 9.964738814348787e-06,
      "loss": 1.0499,
      "step": 1651
    },
    {
      "epoch": 0.13349764641709932,
      "grad_norm": 3.0125157833099365,
      "learning_rate": 9.964661195439847e-06,
      "loss": 1.0054,
      "step": 1652
    },
    {
      "epoch": 0.13357845613042688,
      "grad_norm": 2.9701647758483887,
      "learning_rate": 9.96458349149821e-06,
      "loss": 1.0462,
      "step": 1653
    },
    {
      "epoch": 0.13365926584375443,
      "grad_norm": 2.3569045066833496,
      "learning_rate": 9.964505702525211e-06,
      "loss": 1.1098,
      "step": 1654
    },
    {
      "epoch": 0.13374007555708195,
      "grad_norm": 2.789321184158325,
      "learning_rate": 9.96442782852218e-06,
      "loss": 1.0091,
      "step": 1655
    },
    {
      "epoch": 0.1338208852704095,
      "grad_norm": 2.524693727493286,
      "learning_rate": 9.964349869490453e-06,
      "loss": 1.0318,
      "step": 1656
    },
    {
      "epoch": 0.13390169498373705,
      "grad_norm": 3.0347893238067627,
      "learning_rate": 9.964271825431362e-06,
      "loss": 1.0054,
      "step": 1657
    },
    {
      "epoch": 0.13398250469706457,
      "grad_norm": 2.8926568031311035,
      "learning_rate": 9.964193696346248e-06,
      "loss": 1.0277,
      "step": 1658
    },
    {
      "epoch": 0.13406331441039213,
      "grad_norm": 2.7216997146606445,
      "learning_rate": 9.964115482236444e-06,
      "loss": 1.1218,
      "step": 1659
    },
    {
      "epoch": 0.13414412412371968,
      "grad_norm": 2.5974133014678955,
      "learning_rate": 9.964037183103295e-06,
      "loss": 1.0529,
      "step": 1660
    },
    {
      "epoch": 0.1342249338370472,
      "grad_norm": 3.0064709186553955,
      "learning_rate": 9.963958798948137e-06,
      "loss": 1.0208,
      "step": 1661
    },
    {
      "epoch": 0.13430574355037475,
      "grad_norm": 2.6379201412200928,
      "learning_rate": 9.963880329772317e-06,
      "loss": 1.0222,
      "step": 1662
    },
    {
      "epoch": 0.1343865532637023,
      "grad_norm": 3.0081305503845215,
      "learning_rate": 9.963801775577175e-06,
      "loss": 1.0505,
      "step": 1663
    },
    {
      "epoch": 0.13446736297702985,
      "grad_norm": 3.289186954498291,
      "learning_rate": 9.96372313636406e-06,
      "loss": 0.9699,
      "step": 1664
    },
    {
      "epoch": 0.13454817269035738,
      "grad_norm": 3.1908047199249268,
      "learning_rate": 9.963644412134319e-06,
      "loss": 0.9822,
      "step": 1665
    },
    {
      "epoch": 0.13462898240368493,
      "grad_norm": 2.66094970703125,
      "learning_rate": 9.963565602889295e-06,
      "loss": 1.0759,
      "step": 1666
    },
    {
      "epoch": 0.13470979211701248,
      "grad_norm": 2.412898540496826,
      "learning_rate": 9.963486708630344e-06,
      "loss": 1.021,
      "step": 1667
    },
    {
      "epoch": 0.13479060183034,
      "grad_norm": 2.9810945987701416,
      "learning_rate": 9.963407729358813e-06,
      "loss": 0.9571,
      "step": 1668
    },
    {
      "epoch": 0.13487141154366755,
      "grad_norm": 2.7914586067199707,
      "learning_rate": 9.96332866507606e-06,
      "loss": 1.0353,
      "step": 1669
    },
    {
      "epoch": 0.1349522212569951,
      "grad_norm": 2.8356821537017822,
      "learning_rate": 9.963249515783433e-06,
      "loss": 0.9546,
      "step": 1670
    },
    {
      "epoch": 0.13503303097032263,
      "grad_norm": 2.4492247104644775,
      "learning_rate": 9.96317028148229e-06,
      "loss": 1.0215,
      "step": 1671
    },
    {
      "epoch": 0.13511384068365018,
      "grad_norm": 3.152137279510498,
      "learning_rate": 9.963090962173989e-06,
      "loss": 0.935,
      "step": 1672
    },
    {
      "epoch": 0.13519465039697773,
      "grad_norm": 3.0612106323242188,
      "learning_rate": 9.963011557859888e-06,
      "loss": 0.914,
      "step": 1673
    },
    {
      "epoch": 0.13527546011030525,
      "grad_norm": 2.58404278755188,
      "learning_rate": 9.962932068541347e-06,
      "loss": 0.9286,
      "step": 1674
    },
    {
      "epoch": 0.1353562698236328,
      "grad_norm": 2.5726115703582764,
      "learning_rate": 9.962852494219728e-06,
      "loss": 1.057,
      "step": 1675
    },
    {
      "epoch": 0.13543707953696035,
      "grad_norm": 2.853208541870117,
      "learning_rate": 9.962772834896392e-06,
      "loss": 0.9748,
      "step": 1676
    },
    {
      "epoch": 0.13551788925028788,
      "grad_norm": 3.302896738052368,
      "learning_rate": 9.962693090572706e-06,
      "loss": 1.0241,
      "step": 1677
    },
    {
      "epoch": 0.13559869896361543,
      "grad_norm": 3.2174758911132812,
      "learning_rate": 9.962613261250034e-06,
      "loss": 1.0237,
      "step": 1678
    },
    {
      "epoch": 0.13567950867694298,
      "grad_norm": 3.20485782623291,
      "learning_rate": 9.962533346929744e-06,
      "loss": 0.9673,
      "step": 1679
    },
    {
      "epoch": 0.1357603183902705,
      "grad_norm": 2.799534320831299,
      "learning_rate": 9.962453347613205e-06,
      "loss": 0.9905,
      "step": 1680
    },
    {
      "epoch": 0.13584112810359805,
      "grad_norm": 3.0652966499328613,
      "learning_rate": 9.962373263301785e-06,
      "loss": 1.0848,
      "step": 1681
    },
    {
      "epoch": 0.1359219378169256,
      "grad_norm": 2.4975082874298096,
      "learning_rate": 9.96229309399686e-06,
      "loss": 1.0285,
      "step": 1682
    },
    {
      "epoch": 0.13600274753025313,
      "grad_norm": 2.763394355773926,
      "learning_rate": 9.962212839699799e-06,
      "loss": 0.9335,
      "step": 1683
    },
    {
      "epoch": 0.13608355724358068,
      "grad_norm": 2.9514853954315186,
      "learning_rate": 9.962132500411978e-06,
      "loss": 0.9796,
      "step": 1684
    },
    {
      "epoch": 0.13616436695690823,
      "grad_norm": 2.668872356414795,
      "learning_rate": 9.962052076134773e-06,
      "loss": 1.0644,
      "step": 1685
    },
    {
      "epoch": 0.13624517667023575,
      "grad_norm": 3.2854628562927246,
      "learning_rate": 9.961971566869564e-06,
      "loss": 1.1409,
      "step": 1686
    },
    {
      "epoch": 0.1363259863835633,
      "grad_norm": 2.632746458053589,
      "learning_rate": 9.961890972617724e-06,
      "loss": 0.9844,
      "step": 1687
    },
    {
      "epoch": 0.13640679609689085,
      "grad_norm": 2.752347707748413,
      "learning_rate": 9.96181029338064e-06,
      "loss": 1.0505,
      "step": 1688
    },
    {
      "epoch": 0.13648760581021838,
      "grad_norm": 2.6630587577819824,
      "learning_rate": 9.961729529159689e-06,
      "loss": 1.0518,
      "step": 1689
    },
    {
      "epoch": 0.13656841552354593,
      "grad_norm": 2.814940929412842,
      "learning_rate": 9.961648679956257e-06,
      "loss": 1.0219,
      "step": 1690
    },
    {
      "epoch": 0.13664922523687348,
      "grad_norm": 3.119725227355957,
      "learning_rate": 9.961567745771725e-06,
      "loss": 0.9951,
      "step": 1691
    },
    {
      "epoch": 0.136730034950201,
      "grad_norm": 2.685208797454834,
      "learning_rate": 9.961486726607486e-06,
      "loss": 0.946,
      "step": 1692
    },
    {
      "epoch": 0.13681084466352855,
      "grad_norm": 3.2550692558288574,
      "learning_rate": 9.96140562246492e-06,
      "loss": 0.915,
      "step": 1693
    },
    {
      "epoch": 0.1368916543768561,
      "grad_norm": 2.9896762371063232,
      "learning_rate": 9.961324433345423e-06,
      "loss": 0.9649,
      "step": 1694
    },
    {
      "epoch": 0.13697246409018363,
      "grad_norm": 2.7193186283111572,
      "learning_rate": 9.96124315925038e-06,
      "loss": 1.0714,
      "step": 1695
    },
    {
      "epoch": 0.13705327380351118,
      "grad_norm": 3.505295991897583,
      "learning_rate": 9.961161800181187e-06,
      "loss": 0.9097,
      "step": 1696
    },
    {
      "epoch": 0.13713408351683873,
      "grad_norm": 2.7523858547210693,
      "learning_rate": 9.961080356139235e-06,
      "loss": 1.0551,
      "step": 1697
    },
    {
      "epoch": 0.13721489323016628,
      "grad_norm": 3.00246262550354,
      "learning_rate": 9.960998827125921e-06,
      "loss": 1.0762,
      "step": 1698
    },
    {
      "epoch": 0.1372957029434938,
      "grad_norm": 2.5749502182006836,
      "learning_rate": 9.960917213142637e-06,
      "loss": 0.925,
      "step": 1699
    },
    {
      "epoch": 0.13737651265682135,
      "grad_norm": 3.183603525161743,
      "learning_rate": 9.960835514190787e-06,
      "loss": 1.0033,
      "step": 1700
    },
    {
      "epoch": 0.1374573223701489,
      "grad_norm": 2.928327798843384,
      "learning_rate": 9.960753730271766e-06,
      "loss": 0.9592,
      "step": 1701
    },
    {
      "epoch": 0.13753813208347643,
      "grad_norm": 2.5657637119293213,
      "learning_rate": 9.960671861386978e-06,
      "loss": 0.916,
      "step": 1702
    },
    {
      "epoch": 0.13761894179680398,
      "grad_norm": 2.9228098392486572,
      "learning_rate": 9.960589907537821e-06,
      "loss": 1.1493,
      "step": 1703
    },
    {
      "epoch": 0.13769975151013153,
      "grad_norm": 2.7308812141418457,
      "learning_rate": 9.960507868725703e-06,
      "loss": 1.0686,
      "step": 1704
    },
    {
      "epoch": 0.13778056122345905,
      "grad_norm": 2.9457664489746094,
      "learning_rate": 9.960425744952027e-06,
      "loss": 0.9022,
      "step": 1705
    },
    {
      "epoch": 0.1378613709367866,
      "grad_norm": 2.8565680980682373,
      "learning_rate": 9.960343536218198e-06,
      "loss": 0.947,
      "step": 1706
    },
    {
      "epoch": 0.13794218065011415,
      "grad_norm": 3.288144826889038,
      "learning_rate": 9.960261242525626e-06,
      "loss": 0.9386,
      "step": 1707
    },
    {
      "epoch": 0.13802299036344168,
      "grad_norm": 2.7543015480041504,
      "learning_rate": 9.960178863875723e-06,
      "loss": 1.0274,
      "step": 1708
    },
    {
      "epoch": 0.13810380007676923,
      "grad_norm": 3.6321394443511963,
      "learning_rate": 9.960096400269894e-06,
      "loss": 1.0048,
      "step": 1709
    },
    {
      "epoch": 0.13818460979009678,
      "grad_norm": 3.537015199661255,
      "learning_rate": 9.960013851709555e-06,
      "loss": 0.9272,
      "step": 1710
    },
    {
      "epoch": 0.1382654195034243,
      "grad_norm": 2.508254289627075,
      "learning_rate": 9.959931218196122e-06,
      "loss": 1.0029,
      "step": 1711
    },
    {
      "epoch": 0.13834622921675185,
      "grad_norm": 2.9683449268341064,
      "learning_rate": 9.959848499731004e-06,
      "loss": 1.0596,
      "step": 1712
    },
    {
      "epoch": 0.1384270389300794,
      "grad_norm": 3.078261375427246,
      "learning_rate": 9.959765696315625e-06,
      "loss": 1.0411,
      "step": 1713
    },
    {
      "epoch": 0.13850784864340693,
      "grad_norm": 2.7142679691314697,
      "learning_rate": 9.959682807951399e-06,
      "loss": 1.0267,
      "step": 1714
    },
    {
      "epoch": 0.13858865835673448,
      "grad_norm": 2.784114122390747,
      "learning_rate": 9.959599834639745e-06,
      "loss": 1.0803,
      "step": 1715
    },
    {
      "epoch": 0.13866946807006203,
      "grad_norm": 3.481841564178467,
      "learning_rate": 9.959516776382086e-06,
      "loss": 0.9428,
      "step": 1716
    },
    {
      "epoch": 0.13875027778338955,
      "grad_norm": 3.4312241077423096,
      "learning_rate": 9.959433633179844e-06,
      "loss": 0.9275,
      "step": 1717
    },
    {
      "epoch": 0.1388310874967171,
      "grad_norm": 2.9053895473480225,
      "learning_rate": 9.959350405034445e-06,
      "loss": 0.956,
      "step": 1718
    },
    {
      "epoch": 0.13891189721004465,
      "grad_norm": 3.082746744155884,
      "learning_rate": 9.959267091947311e-06,
      "loss": 0.9316,
      "step": 1719
    },
    {
      "epoch": 0.13899270692337218,
      "grad_norm": 2.992299795150757,
      "learning_rate": 9.959183693919871e-06,
      "loss": 1.0388,
      "step": 1720
    },
    {
      "epoch": 0.13907351663669973,
      "grad_norm": 2.9082751274108887,
      "learning_rate": 9.959100210953555e-06,
      "loss": 0.9716,
      "step": 1721
    },
    {
      "epoch": 0.13915432635002728,
      "grad_norm": 2.7955381870269775,
      "learning_rate": 9.959016643049788e-06,
      "loss": 0.9659,
      "step": 1722
    },
    {
      "epoch": 0.1392351360633548,
      "grad_norm": 2.9297831058502197,
      "learning_rate": 9.958932990210006e-06,
      "loss": 1.0348,
      "step": 1723
    },
    {
      "epoch": 0.13931594577668235,
      "grad_norm": 2.9167332649230957,
      "learning_rate": 9.958849252435638e-06,
      "loss": 1.0126,
      "step": 1724
    },
    {
      "epoch": 0.1393967554900099,
      "grad_norm": 3.501805543899536,
      "learning_rate": 9.958765429728121e-06,
      "loss": 1.0629,
      "step": 1725
    },
    {
      "epoch": 0.13947756520333743,
      "grad_norm": 3.1738104820251465,
      "learning_rate": 9.95868152208889e-06,
      "loss": 1.0639,
      "step": 1726
    },
    {
      "epoch": 0.13955837491666498,
      "grad_norm": 3.3682796955108643,
      "learning_rate": 9.958597529519384e-06,
      "loss": 1.0865,
      "step": 1727
    },
    {
      "epoch": 0.13963918462999253,
      "grad_norm": 2.4615824222564697,
      "learning_rate": 9.958513452021038e-06,
      "loss": 1.0323,
      "step": 1728
    },
    {
      "epoch": 0.13971999434332008,
      "grad_norm": 2.5694899559020996,
      "learning_rate": 9.958429289595295e-06,
      "loss": 1.1025,
      "step": 1729
    },
    {
      "epoch": 0.1398008040566476,
      "grad_norm": 2.9231653213500977,
      "learning_rate": 9.958345042243594e-06,
      "loss": 1.057,
      "step": 1730
    },
    {
      "epoch": 0.13988161376997515,
      "grad_norm": 3.7526228427886963,
      "learning_rate": 9.95826070996738e-06,
      "loss": 1.1402,
      "step": 1731
    },
    {
      "epoch": 0.1399624234833027,
      "grad_norm": 2.602592945098877,
      "learning_rate": 9.958176292768095e-06,
      "loss": 0.9878,
      "step": 1732
    },
    {
      "epoch": 0.14004323319663023,
      "grad_norm": 3.1426117420196533,
      "learning_rate": 9.958091790647188e-06,
      "loss": 0.9382,
      "step": 1733
    },
    {
      "epoch": 0.14012404290995778,
      "grad_norm": 3.071408271789551,
      "learning_rate": 9.958007203606103e-06,
      "loss": 0.9991,
      "step": 1734
    },
    {
      "epoch": 0.14020485262328533,
      "grad_norm": 3.2682929039001465,
      "learning_rate": 9.957922531646293e-06,
      "loss": 1.1086,
      "step": 1735
    },
    {
      "epoch": 0.14028566233661285,
      "grad_norm": 3.3437681198120117,
      "learning_rate": 9.957837774769203e-06,
      "loss": 1.0144,
      "step": 1736
    },
    {
      "epoch": 0.1403664720499404,
      "grad_norm": 2.6054043769836426,
      "learning_rate": 9.95775293297629e-06,
      "loss": 1.02,
      "step": 1737
    },
    {
      "epoch": 0.14044728176326796,
      "grad_norm": 2.789842367172241,
      "learning_rate": 9.957668006269003e-06,
      "loss": 0.9584,
      "step": 1738
    },
    {
      "epoch": 0.14052809147659548,
      "grad_norm": 3.1631407737731934,
      "learning_rate": 9.957582994648798e-06,
      "loss": 1.0068,
      "step": 1739
    },
    {
      "epoch": 0.14060890118992303,
      "grad_norm": 3.041994094848633,
      "learning_rate": 9.957497898117133e-06,
      "loss": 1.0425,
      "step": 1740
    },
    {
      "epoch": 0.14068971090325058,
      "grad_norm": 3.3951337337493896,
      "learning_rate": 9.957412716675461e-06,
      "loss": 0.919,
      "step": 1741
    },
    {
      "epoch": 0.1407705206165781,
      "grad_norm": 3.176448106765747,
      "learning_rate": 9.957327450325245e-06,
      "loss": 1.0793,
      "step": 1742
    },
    {
      "epoch": 0.14085133032990566,
      "grad_norm": 3.1270318031311035,
      "learning_rate": 9.957242099067945e-06,
      "loss": 0.9797,
      "step": 1743
    },
    {
      "epoch": 0.1409321400432332,
      "grad_norm": 3.135282039642334,
      "learning_rate": 9.957156662905022e-06,
      "loss": 0.9498,
      "step": 1744
    },
    {
      "epoch": 0.14101294975656073,
      "grad_norm": 3.351174831390381,
      "learning_rate": 9.957071141837938e-06,
      "loss": 1.1038,
      "step": 1745
    },
    {
      "epoch": 0.14109375946988828,
      "grad_norm": 3.098735809326172,
      "learning_rate": 9.956985535868157e-06,
      "loss": 1.0687,
      "step": 1746
    },
    {
      "epoch": 0.14117456918321583,
      "grad_norm": 2.904134750366211,
      "learning_rate": 9.956899844997151e-06,
      "loss": 0.9037,
      "step": 1747
    },
    {
      "epoch": 0.14125537889654335,
      "grad_norm": 2.7765417098999023,
      "learning_rate": 9.95681406922638e-06,
      "loss": 1.1331,
      "step": 1748
    },
    {
      "epoch": 0.1413361886098709,
      "grad_norm": 3.543769121170044,
      "learning_rate": 9.956728208557322e-06,
      "loss": 1.0291,
      "step": 1749
    },
    {
      "epoch": 0.14141699832319846,
      "grad_norm": 2.637230157852173,
      "learning_rate": 9.95664226299144e-06,
      "loss": 0.931,
      "step": 1750
    },
    {
      "epoch": 0.14149780803652598,
      "grad_norm": 3.0778005123138428,
      "learning_rate": 9.956556232530207e-06,
      "loss": 0.9302,
      "step": 1751
    },
    {
      "epoch": 0.14157861774985353,
      "grad_norm": 2.739237070083618,
      "learning_rate": 9.9564701171751e-06,
      "loss": 1.0227,
      "step": 1752
    },
    {
      "epoch": 0.14165942746318108,
      "grad_norm": 2.5961148738861084,
      "learning_rate": 9.956383916927594e-06,
      "loss": 1.1691,
      "step": 1753
    },
    {
      "epoch": 0.1417402371765086,
      "grad_norm": 2.5436360836029053,
      "learning_rate": 9.95629763178916e-06,
      "loss": 0.9643,
      "step": 1754
    },
    {
      "epoch": 0.14182104688983616,
      "grad_norm": 2.729128122329712,
      "learning_rate": 9.956211261761282e-06,
      "loss": 1.0151,
      "step": 1755
    },
    {
      "epoch": 0.1419018566031637,
      "grad_norm": 2.8905763626098633,
      "learning_rate": 9.956124806845435e-06,
      "loss": 1.0167,
      "step": 1756
    },
    {
      "epoch": 0.14198266631649123,
      "grad_norm": 2.813722610473633,
      "learning_rate": 9.956038267043101e-06,
      "loss": 1.0791,
      "step": 1757
    },
    {
      "epoch": 0.14206347602981878,
      "grad_norm": 3.6439390182495117,
      "learning_rate": 9.955951642355765e-06,
      "loss": 1.0461,
      "step": 1758
    },
    {
      "epoch": 0.14214428574314633,
      "grad_norm": 2.759126663208008,
      "learning_rate": 9.955864932784907e-06,
      "loss": 0.956,
      "step": 1759
    },
    {
      "epoch": 0.14222509545647385,
      "grad_norm": 2.8389487266540527,
      "learning_rate": 9.955778138332012e-06,
      "loss": 1.007,
      "step": 1760
    },
    {
      "epoch": 0.1423059051698014,
      "grad_norm": 3.0590078830718994,
      "learning_rate": 9.955691258998571e-06,
      "loss": 1.0354,
      "step": 1761
    },
    {
      "epoch": 0.14238671488312896,
      "grad_norm": 2.698124885559082,
      "learning_rate": 9.955604294786067e-06,
      "loss": 0.9418,
      "step": 1762
    },
    {
      "epoch": 0.1424675245964565,
      "grad_norm": 3.084228515625,
      "learning_rate": 9.955517245695992e-06,
      "loss": 1.0172,
      "step": 1763
    },
    {
      "epoch": 0.14254833430978403,
      "grad_norm": 2.997697114944458,
      "learning_rate": 9.955430111729838e-06,
      "loss": 0.9607,
      "step": 1764
    },
    {
      "epoch": 0.14262914402311158,
      "grad_norm": 3.1823668479919434,
      "learning_rate": 9.955342892889093e-06,
      "loss": 1.0731,
      "step": 1765
    },
    {
      "epoch": 0.14270995373643913,
      "grad_norm": 3.112485885620117,
      "learning_rate": 9.955255589175255e-06,
      "loss": 0.9316,
      "step": 1766
    },
    {
      "epoch": 0.14279076344976666,
      "grad_norm": 3.300565004348755,
      "learning_rate": 9.95516820058982e-06,
      "loss": 0.9639,
      "step": 1767
    },
    {
      "epoch": 0.1428715731630942,
      "grad_norm": 3.437588691711426,
      "learning_rate": 9.95508072713428e-06,
      "loss": 1.0657,
      "step": 1768
    },
    {
      "epoch": 0.14295238287642176,
      "grad_norm": 2.999753713607788,
      "learning_rate": 9.954993168810137e-06,
      "loss": 1.0074,
      "step": 1769
    },
    {
      "epoch": 0.14303319258974928,
      "grad_norm": 2.9702138900756836,
      "learning_rate": 9.95490552561889e-06,
      "loss": 1.143,
      "step": 1770
    },
    {
      "epoch": 0.14311400230307683,
      "grad_norm": 2.9902632236480713,
      "learning_rate": 9.95481779756204e-06,
      "loss": 1.0102,
      "step": 1771
    },
    {
      "epoch": 0.14319481201640438,
      "grad_norm": 2.7242865562438965,
      "learning_rate": 9.954729984641089e-06,
      "loss": 1.0243,
      "step": 1772
    },
    {
      "epoch": 0.1432756217297319,
      "grad_norm": 2.6519744396209717,
      "learning_rate": 9.954642086857541e-06,
      "loss": 0.9326,
      "step": 1773
    },
    {
      "epoch": 0.14335643144305946,
      "grad_norm": 2.8660805225372314,
      "learning_rate": 9.954554104212902e-06,
      "loss": 1.1151,
      "step": 1774
    },
    {
      "epoch": 0.143437241156387,
      "grad_norm": 3.0777528285980225,
      "learning_rate": 9.954466036708678e-06,
      "loss": 1.0236,
      "step": 1775
    },
    {
      "epoch": 0.14351805086971453,
      "grad_norm": 2.6604461669921875,
      "learning_rate": 9.95437788434638e-06,
      "loss": 1.2005,
      "step": 1776
    },
    {
      "epoch": 0.14359886058304208,
      "grad_norm": 2.9123761653900146,
      "learning_rate": 9.954289647127516e-06,
      "loss": 1.046,
      "step": 1777
    },
    {
      "epoch": 0.14367967029636963,
      "grad_norm": 2.609844923019409,
      "learning_rate": 9.954201325053596e-06,
      "loss": 1.0015,
      "step": 1778
    },
    {
      "epoch": 0.14376048000969716,
      "grad_norm": 2.655297040939331,
      "learning_rate": 9.954112918126135e-06,
      "loss": 0.9992,
      "step": 1779
    },
    {
      "epoch": 0.1438412897230247,
      "grad_norm": 2.788531541824341,
      "learning_rate": 9.954024426346645e-06,
      "loss": 0.9095,
      "step": 1780
    },
    {
      "epoch": 0.14392209943635226,
      "grad_norm": 2.9971466064453125,
      "learning_rate": 9.953935849716645e-06,
      "loss": 1.011,
      "step": 1781
    },
    {
      "epoch": 0.14400290914967978,
      "grad_norm": 3.2861266136169434,
      "learning_rate": 9.95384718823765e-06,
      "loss": 0.8697,
      "step": 1782
    },
    {
      "epoch": 0.14408371886300733,
      "grad_norm": 2.6745035648345947,
      "learning_rate": 9.953758441911176e-06,
      "loss": 1.005,
      "step": 1783
    },
    {
      "epoch": 0.14416452857633488,
      "grad_norm": 2.98370623588562,
      "learning_rate": 9.953669610738747e-06,
      "loss": 0.9531,
      "step": 1784
    },
    {
      "epoch": 0.1442453382896624,
      "grad_norm": 2.832010269165039,
      "learning_rate": 9.953580694721882e-06,
      "loss": 0.8957,
      "step": 1785
    },
    {
      "epoch": 0.14432614800298996,
      "grad_norm": 2.6120967864990234,
      "learning_rate": 9.953491693862107e-06,
      "loss": 1.1036,
      "step": 1786
    },
    {
      "epoch": 0.1444069577163175,
      "grad_norm": 2.9595401287078857,
      "learning_rate": 9.953402608160943e-06,
      "loss": 1.0677,
      "step": 1787
    },
    {
      "epoch": 0.14448776742964503,
      "grad_norm": 2.821009397506714,
      "learning_rate": 9.953313437619915e-06,
      "loss": 0.9509,
      "step": 1788
    },
    {
      "epoch": 0.14456857714297258,
      "grad_norm": 2.1895339488983154,
      "learning_rate": 9.953224182240557e-06,
      "loss": 1.1377,
      "step": 1789
    },
    {
      "epoch": 0.14464938685630013,
      "grad_norm": 2.7714715003967285,
      "learning_rate": 9.95313484202439e-06,
      "loss": 1.0252,
      "step": 1790
    },
    {
      "epoch": 0.14473019656962766,
      "grad_norm": 2.9314582347869873,
      "learning_rate": 9.953045416972948e-06,
      "loss": 0.8948,
      "step": 1791
    },
    {
      "epoch": 0.1448110062829552,
      "grad_norm": 3.025941848754883,
      "learning_rate": 9.952955907087762e-06,
      "loss": 0.9994,
      "step": 1792
    },
    {
      "epoch": 0.14489181599628276,
      "grad_norm": 3.127885580062866,
      "learning_rate": 9.952866312370367e-06,
      "loss": 1.0634,
      "step": 1793
    },
    {
      "epoch": 0.1449726257096103,
      "grad_norm": 2.982555866241455,
      "learning_rate": 9.952776632822293e-06,
      "loss": 0.9424,
      "step": 1794
    },
    {
      "epoch": 0.14505343542293783,
      "grad_norm": 2.7474989891052246,
      "learning_rate": 9.95268686844508e-06,
      "loss": 0.8819,
      "step": 1795
    },
    {
      "epoch": 0.14513424513626538,
      "grad_norm": 2.7660906314849854,
      "learning_rate": 9.952597019240264e-06,
      "loss": 1.095,
      "step": 1796
    },
    {
      "epoch": 0.14521505484959293,
      "grad_norm": 2.556553602218628,
      "learning_rate": 9.952507085209382e-06,
      "loss": 0.958,
      "step": 1797
    },
    {
      "epoch": 0.14529586456292046,
      "grad_norm": 3.200990676879883,
      "learning_rate": 9.952417066353979e-06,
      "loss": 1.182,
      "step": 1798
    },
    {
      "epoch": 0.145376674276248,
      "grad_norm": 2.2909719944000244,
      "learning_rate": 9.952326962675593e-06,
      "loss": 1.0834,
      "step": 1799
    },
    {
      "epoch": 0.14545748398957556,
      "grad_norm": 3.0186989307403564,
      "learning_rate": 9.952236774175767e-06,
      "loss": 1.0537,
      "step": 1800
    },
    {
      "epoch": 0.14553829370290308,
      "grad_norm": 2.985985040664673,
      "learning_rate": 9.95214650085605e-06,
      "loss": 0.9396,
      "step": 1801
    },
    {
      "epoch": 0.14561910341623063,
      "grad_norm": 3.10266375541687,
      "learning_rate": 9.952056142717983e-06,
      "loss": 1.0285,
      "step": 1802
    },
    {
      "epoch": 0.14569991312955818,
      "grad_norm": 2.8633596897125244,
      "learning_rate": 9.951965699763118e-06,
      "loss": 1.0462,
      "step": 1803
    },
    {
      "epoch": 0.1457807228428857,
      "grad_norm": 3.117130994796753,
      "learning_rate": 9.951875171993e-06,
      "loss": 0.962,
      "step": 1804
    },
    {
      "epoch": 0.14586153255621326,
      "grad_norm": 2.4210338592529297,
      "learning_rate": 9.951784559409181e-06,
      "loss": 1.1702,
      "step": 1805
    },
    {
      "epoch": 0.1459423422695408,
      "grad_norm": 3.194343090057373,
      "learning_rate": 9.951693862013214e-06,
      "loss": 0.9176,
      "step": 1806
    },
    {
      "epoch": 0.14602315198286833,
      "grad_norm": 2.8605997562408447,
      "learning_rate": 9.951603079806653e-06,
      "loss": 0.9523,
      "step": 1807
    },
    {
      "epoch": 0.14610396169619588,
      "grad_norm": 2.895078182220459,
      "learning_rate": 9.95151221279105e-06,
      "loss": 0.9703,
      "step": 1808
    },
    {
      "epoch": 0.14618477140952343,
      "grad_norm": 3.0536563396453857,
      "learning_rate": 9.951421260967964e-06,
      "loss": 0.9933,
      "step": 1809
    },
    {
      "epoch": 0.14626558112285096,
      "grad_norm": 2.785574197769165,
      "learning_rate": 9.951330224338952e-06,
      "loss": 1.0731,
      "step": 1810
    },
    {
      "epoch": 0.1463463908361785,
      "grad_norm": 2.6607959270477295,
      "learning_rate": 9.951239102905573e-06,
      "loss": 1.0461,
      "step": 1811
    },
    {
      "epoch": 0.14642720054950606,
      "grad_norm": 2.855102300643921,
      "learning_rate": 9.951147896669389e-06,
      "loss": 1.1034,
      "step": 1812
    },
    {
      "epoch": 0.14650801026283358,
      "grad_norm": 2.478165864944458,
      "learning_rate": 9.951056605631959e-06,
      "loss": 0.9691,
      "step": 1813
    },
    {
      "epoch": 0.14658881997616113,
      "grad_norm": 2.6855709552764893,
      "learning_rate": 9.950965229794849e-06,
      "loss": 1.0317,
      "step": 1814
    },
    {
      "epoch": 0.14666962968948868,
      "grad_norm": 2.644298791885376,
      "learning_rate": 9.950873769159624e-06,
      "loss": 1.0109,
      "step": 1815
    },
    {
      "epoch": 0.1467504394028162,
      "grad_norm": 3.3271265029907227,
      "learning_rate": 9.950782223727851e-06,
      "loss": 1.0195,
      "step": 1816
    },
    {
      "epoch": 0.14683124911614376,
      "grad_norm": 2.890562057495117,
      "learning_rate": 9.950690593501096e-06,
      "loss": 1.1402,
      "step": 1817
    },
    {
      "epoch": 0.1469120588294713,
      "grad_norm": 2.3465516567230225,
      "learning_rate": 9.950598878480928e-06,
      "loss": 1.0696,
      "step": 1818
    },
    {
      "epoch": 0.14699286854279883,
      "grad_norm": 2.366276741027832,
      "learning_rate": 9.95050707866892e-06,
      "loss": 0.9195,
      "step": 1819
    },
    {
      "epoch": 0.14707367825612638,
      "grad_norm": 2.5351967811584473,
      "learning_rate": 9.950415194066646e-06,
      "loss": 1.016,
      "step": 1820
    },
    {
      "epoch": 0.14715448796945393,
      "grad_norm": 2.9258339405059814,
      "learning_rate": 9.950323224675675e-06,
      "loss": 0.9287,
      "step": 1821
    },
    {
      "epoch": 0.14723529768278146,
      "grad_norm": 3.2073264122009277,
      "learning_rate": 9.950231170497585e-06,
      "loss": 1.0693,
      "step": 1822
    },
    {
      "epoch": 0.147316107396109,
      "grad_norm": 3.120288133621216,
      "learning_rate": 9.950139031533952e-06,
      "loss": 1.0181,
      "step": 1823
    },
    {
      "epoch": 0.14739691710943656,
      "grad_norm": 2.8186001777648926,
      "learning_rate": 9.950046807786355e-06,
      "loss": 0.909,
      "step": 1824
    },
    {
      "epoch": 0.14747772682276408,
      "grad_norm": 2.494920015335083,
      "learning_rate": 9.949954499256373e-06,
      "loss": 1.1287,
      "step": 1825
    },
    {
      "epoch": 0.14755853653609163,
      "grad_norm": 2.6874499320983887,
      "learning_rate": 9.949862105945587e-06,
      "loss": 0.9388,
      "step": 1826
    },
    {
      "epoch": 0.14763934624941918,
      "grad_norm": 2.686732292175293,
      "learning_rate": 9.949769627855579e-06,
      "loss": 0.9626,
      "step": 1827
    },
    {
      "epoch": 0.14772015596274674,
      "grad_norm": 3.2707130908966064,
      "learning_rate": 9.949677064987933e-06,
      "loss": 1.0166,
      "step": 1828
    },
    {
      "epoch": 0.14780096567607426,
      "grad_norm": 2.8146145343780518,
      "learning_rate": 9.949584417344236e-06,
      "loss": 0.9725,
      "step": 1829
    },
    {
      "epoch": 0.1478817753894018,
      "grad_norm": 2.8295071125030518,
      "learning_rate": 9.949491684926071e-06,
      "loss": 1.0094,
      "step": 1830
    },
    {
      "epoch": 0.14796258510272936,
      "grad_norm": 3.5557024478912354,
      "learning_rate": 9.949398867735032e-06,
      "loss": 0.967,
      "step": 1831
    },
    {
      "epoch": 0.14804339481605688,
      "grad_norm": 3.3532896041870117,
      "learning_rate": 9.949305965772704e-06,
      "loss": 1.1545,
      "step": 1832
    },
    {
      "epoch": 0.14812420452938443,
      "grad_norm": 3.946514844894409,
      "learning_rate": 9.94921297904068e-06,
      "loss": 0.9121,
      "step": 1833
    },
    {
      "epoch": 0.14820501424271199,
      "grad_norm": 2.8444509506225586,
      "learning_rate": 9.949119907540552e-06,
      "loss": 1.0807,
      "step": 1834
    },
    {
      "epoch": 0.1482858239560395,
      "grad_norm": 3.132686138153076,
      "learning_rate": 9.949026751273916e-06,
      "loss": 1.1495,
      "step": 1835
    },
    {
      "epoch": 0.14836663366936706,
      "grad_norm": 2.9394941329956055,
      "learning_rate": 9.948933510242365e-06,
      "loss": 0.9223,
      "step": 1836
    },
    {
      "epoch": 0.1484474433826946,
      "grad_norm": 3.1135542392730713,
      "learning_rate": 9.948840184447497e-06,
      "loss": 1.0792,
      "step": 1837
    },
    {
      "epoch": 0.14852825309602213,
      "grad_norm": 3.050935745239258,
      "learning_rate": 9.94874677389091e-06,
      "loss": 1.0063,
      "step": 1838
    },
    {
      "epoch": 0.14860906280934968,
      "grad_norm": 2.676487684249878,
      "learning_rate": 9.948653278574205e-06,
      "loss": 0.9533,
      "step": 1839
    },
    {
      "epoch": 0.14868987252267724,
      "grad_norm": 2.9276177883148193,
      "learning_rate": 9.948559698498984e-06,
      "loss": 0.9597,
      "step": 1840
    },
    {
      "epoch": 0.14877068223600476,
      "grad_norm": 2.749194383621216,
      "learning_rate": 9.948466033666846e-06,
      "loss": 1.0334,
      "step": 1841
    },
    {
      "epoch": 0.1488514919493323,
      "grad_norm": 2.471069812774658,
      "learning_rate": 9.948372284079398e-06,
      "loss": 1.0528,
      "step": 1842
    },
    {
      "epoch": 0.14893230166265986,
      "grad_norm": 3.0131027698516846,
      "learning_rate": 9.948278449738246e-06,
      "loss": 0.9566,
      "step": 1843
    },
    {
      "epoch": 0.14901311137598738,
      "grad_norm": 3.0999159812927246,
      "learning_rate": 9.948184530644998e-06,
      "loss": 1.0947,
      "step": 1844
    },
    {
      "epoch": 0.14909392108931493,
      "grad_norm": 2.7991368770599365,
      "learning_rate": 9.948090526801259e-06,
      "loss": 0.9532,
      "step": 1845
    },
    {
      "epoch": 0.14917473080264249,
      "grad_norm": 2.8065764904022217,
      "learning_rate": 9.947996438208644e-06,
      "loss": 1.0142,
      "step": 1846
    },
    {
      "epoch": 0.14925554051597,
      "grad_norm": 2.9444782733917236,
      "learning_rate": 9.94790226486876e-06,
      "loss": 1.0238,
      "step": 1847
    },
    {
      "epoch": 0.14933635022929756,
      "grad_norm": 2.598299980163574,
      "learning_rate": 9.947808006783223e-06,
      "loss": 1.0269,
      "step": 1848
    },
    {
      "epoch": 0.1494171599426251,
      "grad_norm": 3.035118579864502,
      "learning_rate": 9.947713663953644e-06,
      "loss": 1.0333,
      "step": 1849
    },
    {
      "epoch": 0.14949796965595263,
      "grad_norm": 3.401883840560913,
      "learning_rate": 9.947619236381644e-06,
      "loss": 0.9576,
      "step": 1850
    },
    {
      "epoch": 0.14957877936928018,
      "grad_norm": 2.4581234455108643,
      "learning_rate": 9.947524724068835e-06,
      "loss": 1.0626,
      "step": 1851
    },
    {
      "epoch": 0.14965958908260774,
      "grad_norm": 2.7806568145751953,
      "learning_rate": 9.94743012701684e-06,
      "loss": 1.0083,
      "step": 1852
    },
    {
      "epoch": 0.14974039879593526,
      "grad_norm": 2.992265224456787,
      "learning_rate": 9.947335445227276e-06,
      "loss": 1.1236,
      "step": 1853
    },
    {
      "epoch": 0.1498212085092628,
      "grad_norm": 2.996896982192993,
      "learning_rate": 9.947240678701766e-06,
      "loss": 0.9795,
      "step": 1854
    },
    {
      "epoch": 0.14990201822259036,
      "grad_norm": 2.843867778778076,
      "learning_rate": 9.947145827441934e-06,
      "loss": 1.0279,
      "step": 1855
    },
    {
      "epoch": 0.14998282793591788,
      "grad_norm": 3.454298734664917,
      "learning_rate": 9.947050891449403e-06,
      "loss": 1.1464,
      "step": 1856
    },
    {
      "epoch": 0.15006363764924543,
      "grad_norm": 2.8720662593841553,
      "learning_rate": 9.9469558707258e-06,
      "loss": 0.9398,
      "step": 1857
    },
    {
      "epoch": 0.15014444736257299,
      "grad_norm": 2.728085994720459,
      "learning_rate": 9.946860765272753e-06,
      "loss": 0.9881,
      "step": 1858
    },
    {
      "epoch": 0.15022525707590054,
      "grad_norm": 2.709182024002075,
      "learning_rate": 9.94676557509189e-06,
      "loss": 1.1373,
      "step": 1859
    },
    {
      "epoch": 0.15030606678922806,
      "grad_norm": 3.1405465602874756,
      "learning_rate": 9.946670300184841e-06,
      "loss": 1.0212,
      "step": 1860
    },
    {
      "epoch": 0.1503868765025556,
      "grad_norm": 4.029134750366211,
      "learning_rate": 9.946574940553238e-06,
      "loss": 0.9502,
      "step": 1861
    },
    {
      "epoch": 0.15046768621588316,
      "grad_norm": 3.1117727756500244,
      "learning_rate": 9.946479496198715e-06,
      "loss": 1.0837,
      "step": 1862
    },
    {
      "epoch": 0.15054849592921069,
      "grad_norm": 2.859290361404419,
      "learning_rate": 9.946383967122907e-06,
      "loss": 1.1341,
      "step": 1863
    },
    {
      "epoch": 0.15062930564253824,
      "grad_norm": 2.6758170127868652,
      "learning_rate": 9.94628835332745e-06,
      "loss": 1.1002,
      "step": 1864
    },
    {
      "epoch": 0.1507101153558658,
      "grad_norm": 3.0312492847442627,
      "learning_rate": 9.94619265481398e-06,
      "loss": 0.939,
      "step": 1865
    },
    {
      "epoch": 0.1507909250691933,
      "grad_norm": 2.5172882080078125,
      "learning_rate": 9.946096871584138e-06,
      "loss": 1.0346,
      "step": 1866
    },
    {
      "epoch": 0.15087173478252086,
      "grad_norm": 3.1140804290771484,
      "learning_rate": 9.946001003639562e-06,
      "loss": 1.0218,
      "step": 1867
    },
    {
      "epoch": 0.1509525444958484,
      "grad_norm": 2.838855266571045,
      "learning_rate": 9.9459050509819e-06,
      "loss": 0.9093,
      "step": 1868
    },
    {
      "epoch": 0.15103335420917594,
      "grad_norm": 2.6162962913513184,
      "learning_rate": 9.945809013612787e-06,
      "loss": 0.9472,
      "step": 1869
    },
    {
      "epoch": 0.15111416392250349,
      "grad_norm": 4.642756938934326,
      "learning_rate": 9.945712891533874e-06,
      "loss": 1.0032,
      "step": 1870
    },
    {
      "epoch": 0.15119497363583104,
      "grad_norm": 3.521791696548462,
      "learning_rate": 9.945616684746805e-06,
      "loss": 1.053,
      "step": 1871
    },
    {
      "epoch": 0.15127578334915856,
      "grad_norm": 3.1945931911468506,
      "learning_rate": 9.945520393253228e-06,
      "loss": 1.0093,
      "step": 1872
    },
    {
      "epoch": 0.1513565930624861,
      "grad_norm": 3.0325980186462402,
      "learning_rate": 9.945424017054794e-06,
      "loss": 0.9768,
      "step": 1873
    },
    {
      "epoch": 0.15143740277581366,
      "grad_norm": 2.7389962673187256,
      "learning_rate": 9.945327556153151e-06,
      "loss": 1.1918,
      "step": 1874
    },
    {
      "epoch": 0.15151821248914119,
      "grad_norm": 2.963046073913574,
      "learning_rate": 9.945231010549952e-06,
      "loss": 0.9698,
      "step": 1875
    },
    {
      "epoch": 0.15159902220246874,
      "grad_norm": 3.454117774963379,
      "learning_rate": 9.945134380246853e-06,
      "loss": 1.0315,
      "step": 1876
    },
    {
      "epoch": 0.1516798319157963,
      "grad_norm": 3.3037338256835938,
      "learning_rate": 9.945037665245504e-06,
      "loss": 0.9624,
      "step": 1877
    },
    {
      "epoch": 0.1517606416291238,
      "grad_norm": 2.6708765029907227,
      "learning_rate": 9.944940865547566e-06,
      "loss": 0.9077,
      "step": 1878
    },
    {
      "epoch": 0.15184145134245136,
      "grad_norm": 2.526172399520874,
      "learning_rate": 9.944843981154696e-06,
      "loss": 1.041,
      "step": 1879
    },
    {
      "epoch": 0.1519222610557789,
      "grad_norm": 2.736255407333374,
      "learning_rate": 9.944747012068553e-06,
      "loss": 1.0002,
      "step": 1880
    },
    {
      "epoch": 0.15200307076910644,
      "grad_norm": 2.9772162437438965,
      "learning_rate": 9.944649958290796e-06,
      "loss": 0.9891,
      "step": 1881
    },
    {
      "epoch": 0.152083880482434,
      "grad_norm": 2.9495186805725098,
      "learning_rate": 9.94455281982309e-06,
      "loss": 1.118,
      "step": 1882
    },
    {
      "epoch": 0.15216469019576154,
      "grad_norm": 2.974454879760742,
      "learning_rate": 9.944455596667097e-06,
      "loss": 0.9662,
      "step": 1883
    },
    {
      "epoch": 0.15224549990908906,
      "grad_norm": 2.9383349418640137,
      "learning_rate": 9.944358288824485e-06,
      "loss": 1.0998,
      "step": 1884
    },
    {
      "epoch": 0.1523263096224166,
      "grad_norm": 2.7243707180023193,
      "learning_rate": 9.944260896296917e-06,
      "loss": 1.0039,
      "step": 1885
    },
    {
      "epoch": 0.15240711933574416,
      "grad_norm": 3.1253015995025635,
      "learning_rate": 9.944163419086062e-06,
      "loss": 1.0838,
      "step": 1886
    },
    {
      "epoch": 0.15248792904907169,
      "grad_norm": 3.283964157104492,
      "learning_rate": 9.944065857193591e-06,
      "loss": 0.9866,
      "step": 1887
    },
    {
      "epoch": 0.15256873876239924,
      "grad_norm": 3.455512523651123,
      "learning_rate": 9.943968210621174e-06,
      "loss": 1.0921,
      "step": 1888
    },
    {
      "epoch": 0.1526495484757268,
      "grad_norm": 2.915229320526123,
      "learning_rate": 9.943870479370485e-06,
      "loss": 1.0121,
      "step": 1889
    },
    {
      "epoch": 0.1527303581890543,
      "grad_norm": 2.8723795413970947,
      "learning_rate": 9.943772663443194e-06,
      "loss": 1.0359,
      "step": 1890
    },
    {
      "epoch": 0.15281116790238186,
      "grad_norm": 3.028423547744751,
      "learning_rate": 9.94367476284098e-06,
      "loss": 0.9814,
      "step": 1891
    },
    {
      "epoch": 0.1528919776157094,
      "grad_norm": 2.655653238296509,
      "learning_rate": 9.94357677756552e-06,
      "loss": 1.1435,
      "step": 1892
    },
    {
      "epoch": 0.15297278732903696,
      "grad_norm": 2.7452893257141113,
      "learning_rate": 9.94347870761849e-06,
      "loss": 1.0701,
      "step": 1893
    },
    {
      "epoch": 0.1530535970423645,
      "grad_norm": 2.8322525024414062,
      "learning_rate": 9.943380553001571e-06,
      "loss": 0.9368,
      "step": 1894
    },
    {
      "epoch": 0.15313440675569204,
      "grad_norm": 2.974698305130005,
      "learning_rate": 9.943282313716444e-06,
      "loss": 1.0267,
      "step": 1895
    },
    {
      "epoch": 0.1532152164690196,
      "grad_norm": 3.2014925479888916,
      "learning_rate": 9.94318398976479e-06,
      "loss": 1.0498,
      "step": 1896
    },
    {
      "epoch": 0.1532960261823471,
      "grad_norm": 2.808967351913452,
      "learning_rate": 9.943085581148296e-06,
      "loss": 0.9781,
      "step": 1897
    },
    {
      "epoch": 0.15337683589567466,
      "grad_norm": 2.5994133949279785,
      "learning_rate": 9.942987087868646e-06,
      "loss": 1.033,
      "step": 1898
    },
    {
      "epoch": 0.1534576456090022,
      "grad_norm": 2.7472386360168457,
      "learning_rate": 9.942888509927525e-06,
      "loss": 1.0477,
      "step": 1899
    },
    {
      "epoch": 0.15353845532232974,
      "grad_norm": 3.5498239994049072,
      "learning_rate": 9.942789847326626e-06,
      "loss": 1.0185,
      "step": 1900
    },
    {
      "epoch": 0.1536192650356573,
      "grad_norm": 2.7498979568481445,
      "learning_rate": 9.942691100067635e-06,
      "loss": 1.0287,
      "step": 1901
    },
    {
      "epoch": 0.15370007474898484,
      "grad_norm": 2.7843916416168213,
      "learning_rate": 9.942592268152244e-06,
      "loss": 1.0318,
      "step": 1902
    },
    {
      "epoch": 0.15378088446231236,
      "grad_norm": 2.5021860599517822,
      "learning_rate": 9.942493351582147e-06,
      "loss": 1.0759,
      "step": 1903
    },
    {
      "epoch": 0.1538616941756399,
      "grad_norm": 2.946166753768921,
      "learning_rate": 9.942394350359038e-06,
      "loss": 0.9596,
      "step": 1904
    },
    {
      "epoch": 0.15394250388896746,
      "grad_norm": 2.8323495388031006,
      "learning_rate": 9.942295264484612e-06,
      "loss": 1.1329,
      "step": 1905
    },
    {
      "epoch": 0.154023313602295,
      "grad_norm": 2.8073818683624268,
      "learning_rate": 9.942196093960564e-06,
      "loss": 1.066,
      "step": 1906
    },
    {
      "epoch": 0.15410412331562254,
      "grad_norm": 2.6901755332946777,
      "learning_rate": 9.942096838788598e-06,
      "loss": 1.0574,
      "step": 1907
    },
    {
      "epoch": 0.1541849330289501,
      "grad_norm": 3.3798248767852783,
      "learning_rate": 9.941997498970408e-06,
      "loss": 1.1475,
      "step": 1908
    },
    {
      "epoch": 0.1542657427422776,
      "grad_norm": 2.586766004562378,
      "learning_rate": 9.941898074507698e-06,
      "loss": 1.0121,
      "step": 1909
    },
    {
      "epoch": 0.15434655245560516,
      "grad_norm": 2.8129591941833496,
      "learning_rate": 9.941798565402175e-06,
      "loss": 1.1243,
      "step": 1910
    },
    {
      "epoch": 0.1544273621689327,
      "grad_norm": 2.8571395874023438,
      "learning_rate": 9.941698971655536e-06,
      "loss": 0.9355,
      "step": 1911
    },
    {
      "epoch": 0.15450817188226024,
      "grad_norm": 2.6519362926483154,
      "learning_rate": 9.94159929326949e-06,
      "loss": 0.9789,
      "step": 1912
    },
    {
      "epoch": 0.1545889815955878,
      "grad_norm": 2.305694818496704,
      "learning_rate": 9.941499530245746e-06,
      "loss": 0.8932,
      "step": 1913
    },
    {
      "epoch": 0.15466979130891534,
      "grad_norm": 2.8709194660186768,
      "learning_rate": 9.94139968258601e-06,
      "loss": 1.0991,
      "step": 1914
    },
    {
      "epoch": 0.15475060102224286,
      "grad_norm": 3.5704281330108643,
      "learning_rate": 9.941299750291994e-06,
      "loss": 1.0547,
      "step": 1915
    },
    {
      "epoch": 0.1548314107355704,
      "grad_norm": 2.815941095352173,
      "learning_rate": 9.94119973336541e-06,
      "loss": 1.1171,
      "step": 1916
    },
    {
      "epoch": 0.15491222044889796,
      "grad_norm": 2.738818407058716,
      "learning_rate": 9.941099631807968e-06,
      "loss": 1.1006,
      "step": 1917
    },
    {
      "epoch": 0.1549930301622255,
      "grad_norm": 2.942873239517212,
      "learning_rate": 9.940999445621387e-06,
      "loss": 1.0785,
      "step": 1918
    },
    {
      "epoch": 0.15507383987555304,
      "grad_norm": 2.959956407546997,
      "learning_rate": 9.940899174807379e-06,
      "loss": 1.051,
      "step": 1919
    },
    {
      "epoch": 0.1551546495888806,
      "grad_norm": 3.177203893661499,
      "learning_rate": 9.940798819367663e-06,
      "loss": 1.0827,
      "step": 1920
    },
    {
      "epoch": 0.1552354593022081,
      "grad_norm": 2.9764657020568848,
      "learning_rate": 9.940698379303958e-06,
      "loss": 0.9899,
      "step": 1921
    },
    {
      "epoch": 0.15531626901553566,
      "grad_norm": 3.004303455352783,
      "learning_rate": 9.940597854617984e-06,
      "loss": 1.0142,
      "step": 1922
    },
    {
      "epoch": 0.1553970787288632,
      "grad_norm": 2.973814010620117,
      "learning_rate": 9.940497245311462e-06,
      "loss": 0.9586,
      "step": 1923
    },
    {
      "epoch": 0.15547788844219076,
      "grad_norm": 2.8607869148254395,
      "learning_rate": 9.940396551386117e-06,
      "loss": 0.9894,
      "step": 1924
    },
    {
      "epoch": 0.1555586981555183,
      "grad_norm": 3.0081584453582764,
      "learning_rate": 9.940295772843673e-06,
      "loss": 1.0247,
      "step": 1925
    },
    {
      "epoch": 0.15563950786884584,
      "grad_norm": 3.7671658992767334,
      "learning_rate": 9.940194909685854e-06,
      "loss": 1.0165,
      "step": 1926
    },
    {
      "epoch": 0.1557203175821734,
      "grad_norm": 2.9663097858428955,
      "learning_rate": 9.94009396191439e-06,
      "loss": 0.9678,
      "step": 1927
    },
    {
      "epoch": 0.1558011272955009,
      "grad_norm": 3.5792739391326904,
      "learning_rate": 9.93999292953101e-06,
      "loss": 0.9796,
      "step": 1928
    },
    {
      "epoch": 0.15588193700882846,
      "grad_norm": 2.8195860385894775,
      "learning_rate": 9.939891812537442e-06,
      "loss": 0.961,
      "step": 1929
    },
    {
      "epoch": 0.15596274672215601,
      "grad_norm": 2.773045539855957,
      "learning_rate": 9.93979061093542e-06,
      "loss": 1.0076,
      "step": 1930
    },
    {
      "epoch": 0.15604355643548354,
      "grad_norm": 3.2170090675354004,
      "learning_rate": 9.939689324726678e-06,
      "loss": 1.0341,
      "step": 1931
    },
    {
      "epoch": 0.1561243661488111,
      "grad_norm": 2.630054235458374,
      "learning_rate": 9.939587953912949e-06,
      "loss": 0.9949,
      "step": 1932
    },
    {
      "epoch": 0.15620517586213864,
      "grad_norm": 2.7837579250335693,
      "learning_rate": 9.93948649849597e-06,
      "loss": 1.1105,
      "step": 1933
    },
    {
      "epoch": 0.15628598557546616,
      "grad_norm": 3.19868803024292,
      "learning_rate": 9.939384958477478e-06,
      "loss": 1.0349,
      "step": 1934
    },
    {
      "epoch": 0.15636679528879371,
      "grad_norm": 2.537768840789795,
      "learning_rate": 9.939283333859214e-06,
      "loss": 0.9077,
      "step": 1935
    },
    {
      "epoch": 0.15644760500212126,
      "grad_norm": 3.3658368587493896,
      "learning_rate": 9.939181624642917e-06,
      "loss": 0.9721,
      "step": 1936
    },
    {
      "epoch": 0.1565284147154488,
      "grad_norm": 3.2723824977874756,
      "learning_rate": 9.939079830830329e-06,
      "loss": 0.9994,
      "step": 1937
    },
    {
      "epoch": 0.15660922442877634,
      "grad_norm": 3.204620838165283,
      "learning_rate": 9.938977952423193e-06,
      "loss": 1.0439,
      "step": 1938
    },
    {
      "epoch": 0.1566900341421039,
      "grad_norm": 3.06638240814209,
      "learning_rate": 9.938875989423255e-06,
      "loss": 1.0653,
      "step": 1939
    },
    {
      "epoch": 0.1567708438554314,
      "grad_norm": 2.6601171493530273,
      "learning_rate": 9.938773941832263e-06,
      "loss": 0.9215,
      "step": 1940
    },
    {
      "epoch": 0.15685165356875896,
      "grad_norm": 3.1091103553771973,
      "learning_rate": 9.938671809651961e-06,
      "loss": 0.9964,
      "step": 1941
    },
    {
      "epoch": 0.15693246328208652,
      "grad_norm": 3.090252637863159,
      "learning_rate": 9.938569592884101e-06,
      "loss": 1.011,
      "step": 1942
    },
    {
      "epoch": 0.15701327299541404,
      "grad_norm": 3.389678716659546,
      "learning_rate": 9.938467291530434e-06,
      "loss": 1.0345,
      "step": 1943
    },
    {
      "epoch": 0.1570940827087416,
      "grad_norm": 3.044147491455078,
      "learning_rate": 9.93836490559271e-06,
      "loss": 1.0974,
      "step": 1944
    },
    {
      "epoch": 0.15717489242206914,
      "grad_norm": 2.801713228225708,
      "learning_rate": 9.938262435072683e-06,
      "loss": 1.0745,
      "step": 1945
    },
    {
      "epoch": 0.15725570213539666,
      "grad_norm": 3.206634283065796,
      "learning_rate": 9.93815987997211e-06,
      "loss": 1.0328,
      "step": 1946
    },
    {
      "epoch": 0.15733651184872421,
      "grad_norm": 2.9063611030578613,
      "learning_rate": 9.938057240292747e-06,
      "loss": 1.0866,
      "step": 1947
    },
    {
      "epoch": 0.15741732156205177,
      "grad_norm": 2.8502037525177,
      "learning_rate": 9.93795451603635e-06,
      "loss": 0.9957,
      "step": 1948
    },
    {
      "epoch": 0.1574981312753793,
      "grad_norm": 2.766782760620117,
      "learning_rate": 9.937851707204682e-06,
      "loss": 0.9233,
      "step": 1949
    },
    {
      "epoch": 0.15757894098870684,
      "grad_norm": 2.7361745834350586,
      "learning_rate": 9.937748813799499e-06,
      "loss": 0.9254,
      "step": 1950
    },
    {
      "epoch": 0.1576597507020344,
      "grad_norm": 2.6297945976257324,
      "learning_rate": 9.937645835822567e-06,
      "loss": 0.9656,
      "step": 1951
    },
    {
      "epoch": 0.1577405604153619,
      "grad_norm": 2.5020108222961426,
      "learning_rate": 9.937542773275648e-06,
      "loss": 1.1277,
      "step": 1952
    },
    {
      "epoch": 0.15782137012868946,
      "grad_norm": 2.625633955001831,
      "learning_rate": 9.93743962616051e-06,
      "loss": 1.0259,
      "step": 1953
    },
    {
      "epoch": 0.15790217984201702,
      "grad_norm": 2.857710123062134,
      "learning_rate": 9.937336394478916e-06,
      "loss": 1.0568,
      "step": 1954
    },
    {
      "epoch": 0.15798298955534457,
      "grad_norm": 2.824298143386841,
      "learning_rate": 9.937233078232636e-06,
      "loss": 0.8799,
      "step": 1955
    },
    {
      "epoch": 0.1580637992686721,
      "grad_norm": 2.856640338897705,
      "learning_rate": 9.937129677423439e-06,
      "loss": 1.0913,
      "step": 1956
    },
    {
      "epoch": 0.15814460898199964,
      "grad_norm": 3.206674098968506,
      "learning_rate": 9.937026192053097e-06,
      "loss": 0.9916,
      "step": 1957
    },
    {
      "epoch": 0.1582254186953272,
      "grad_norm": 2.5003178119659424,
      "learning_rate": 9.936922622123382e-06,
      "loss": 1.0594,
      "step": 1958
    },
    {
      "epoch": 0.15830622840865471,
      "grad_norm": 2.9067587852478027,
      "learning_rate": 9.936818967636065e-06,
      "loss": 0.9746,
      "step": 1959
    },
    {
      "epoch": 0.15838703812198227,
      "grad_norm": 3.135910749435425,
      "learning_rate": 9.936715228592927e-06,
      "loss": 1.0536,
      "step": 1960
    },
    {
      "epoch": 0.15846784783530982,
      "grad_norm": 2.9614381790161133,
      "learning_rate": 9.93661140499574e-06,
      "loss": 1.0545,
      "step": 1961
    },
    {
      "epoch": 0.15854865754863734,
      "grad_norm": 2.7072744369506836,
      "learning_rate": 9.936507496846286e-06,
      "loss": 0.9803,
      "step": 1962
    },
    {
      "epoch": 0.1586294672619649,
      "grad_norm": 3.2678558826446533,
      "learning_rate": 9.936403504146341e-06,
      "loss": 1.0455,
      "step": 1963
    },
    {
      "epoch": 0.15871027697529244,
      "grad_norm": 2.2548952102661133,
      "learning_rate": 9.93629942689769e-06,
      "loss": 1.0457,
      "step": 1964
    },
    {
      "epoch": 0.15879108668861996,
      "grad_norm": 2.572094440460205,
      "learning_rate": 9.936195265102111e-06,
      "loss": 1.0572,
      "step": 1965
    },
    {
      "epoch": 0.15887189640194752,
      "grad_norm": 2.6914725303649902,
      "learning_rate": 9.936091018761392e-06,
      "loss": 0.9983,
      "step": 1966
    },
    {
      "epoch": 0.15895270611527507,
      "grad_norm": 2.4944095611572266,
      "learning_rate": 9.935986687877314e-06,
      "loss": 1.0104,
      "step": 1967
    },
    {
      "epoch": 0.1590335158286026,
      "grad_norm": 2.5117979049682617,
      "learning_rate": 9.93588227245167e-06,
      "loss": 1.1069,
      "step": 1968
    },
    {
      "epoch": 0.15911432554193014,
      "grad_norm": 2.5497148036956787,
      "learning_rate": 9.935777772486244e-06,
      "loss": 1.1017,
      "step": 1969
    },
    {
      "epoch": 0.1591951352552577,
      "grad_norm": 3.289396286010742,
      "learning_rate": 9.935673187982828e-06,
      "loss": 1.0039,
      "step": 1970
    },
    {
      "epoch": 0.15927594496858521,
      "grad_norm": 2.688300609588623,
      "learning_rate": 9.935568518943213e-06,
      "loss": 0.8942,
      "step": 1971
    },
    {
      "epoch": 0.15935675468191277,
      "grad_norm": 2.555457353591919,
      "learning_rate": 9.935463765369192e-06,
      "loss": 1.0607,
      "step": 1972
    },
    {
      "epoch": 0.15943756439524032,
      "grad_norm": 2.710364580154419,
      "learning_rate": 9.935358927262554e-06,
      "loss": 1.1108,
      "step": 1973
    },
    {
      "epoch": 0.15951837410856784,
      "grad_norm": 2.998103380203247,
      "learning_rate": 9.935254004625104e-06,
      "loss": 1.0177,
      "step": 1974
    },
    {
      "epoch": 0.1595991838218954,
      "grad_norm": 2.5873477458953857,
      "learning_rate": 9.93514899745863e-06,
      "loss": 0.9205,
      "step": 1975
    },
    {
      "epoch": 0.15967999353522294,
      "grad_norm": 2.698263645172119,
      "learning_rate": 9.935043905764936e-06,
      "loss": 0.8586,
      "step": 1976
    },
    {
      "epoch": 0.15976080324855046,
      "grad_norm": 2.9557156562805176,
      "learning_rate": 9.934938729545823e-06,
      "loss": 1.0399,
      "step": 1977
    },
    {
      "epoch": 0.15984161296187802,
      "grad_norm": 3.3453972339630127,
      "learning_rate": 9.934833468803087e-06,
      "loss": 0.9828,
      "step": 1978
    },
    {
      "epoch": 0.15992242267520557,
      "grad_norm": 3.246471643447876,
      "learning_rate": 9.934728123538534e-06,
      "loss": 0.9561,
      "step": 1979
    },
    {
      "epoch": 0.1600032323885331,
      "grad_norm": 2.8648548126220703,
      "learning_rate": 9.934622693753968e-06,
      "loss": 1.0334,
      "step": 1980
    },
    {
      "epoch": 0.16008404210186064,
      "grad_norm": 2.9262137413024902,
      "learning_rate": 9.934517179451197e-06,
      "loss": 1.1668,
      "step": 1981
    },
    {
      "epoch": 0.1601648518151882,
      "grad_norm": 2.7647500038146973,
      "learning_rate": 9.934411580632025e-06,
      "loss": 0.9749,
      "step": 1982
    },
    {
      "epoch": 0.16024566152851571,
      "grad_norm": 2.793349266052246,
      "learning_rate": 9.93430589729826e-06,
      "loss": 1.0755,
      "step": 1983
    },
    {
      "epoch": 0.16032647124184327,
      "grad_norm": 3.2601189613342285,
      "learning_rate": 9.934200129451716e-06,
      "loss": 1.0347,
      "step": 1984
    },
    {
      "epoch": 0.16040728095517082,
      "grad_norm": 3.3131632804870605,
      "learning_rate": 9.934094277094202e-06,
      "loss": 1.1393,
      "step": 1985
    },
    {
      "epoch": 0.16048809066849834,
      "grad_norm": 2.795083999633789,
      "learning_rate": 9.93398834022753e-06,
      "loss": 1.0861,
      "step": 1986
    },
    {
      "epoch": 0.1605689003818259,
      "grad_norm": 3.098545551300049,
      "learning_rate": 9.933882318853517e-06,
      "loss": 1.0253,
      "step": 1987
    },
    {
      "epoch": 0.16064971009515344,
      "grad_norm": 2.473449468612671,
      "learning_rate": 9.933776212973977e-06,
      "loss": 0.9149,
      "step": 1988
    },
    {
      "epoch": 0.160730519808481,
      "grad_norm": 2.8285419940948486,
      "learning_rate": 9.933670022590729e-06,
      "loss": 0.9828,
      "step": 1989
    },
    {
      "epoch": 0.16081132952180852,
      "grad_norm": 2.5255420207977295,
      "learning_rate": 9.93356374770559e-06,
      "loss": 1.0115,
      "step": 1990
    },
    {
      "epoch": 0.16089213923513607,
      "grad_norm": 2.6797611713409424,
      "learning_rate": 9.933457388320382e-06,
      "loss": 0.9577,
      "step": 1991
    },
    {
      "epoch": 0.16097294894846362,
      "grad_norm": 2.7080936431884766,
      "learning_rate": 9.933350944436925e-06,
      "loss": 0.9191,
      "step": 1992
    },
    {
      "epoch": 0.16105375866179114,
      "grad_norm": 2.987670660018921,
      "learning_rate": 9.933244416057044e-06,
      "loss": 0.9964,
      "step": 1993
    },
    {
      "epoch": 0.1611345683751187,
      "grad_norm": 2.8873672485351562,
      "learning_rate": 9.93313780318256e-06,
      "loss": 0.9755,
      "step": 1994
    },
    {
      "epoch": 0.16121537808844624,
      "grad_norm": 2.718186855316162,
      "learning_rate": 9.933031105815304e-06,
      "loss": 0.909,
      "step": 1995
    },
    {
      "epoch": 0.16129618780177377,
      "grad_norm": 2.6171813011169434,
      "learning_rate": 9.9329243239571e-06,
      "loss": 1.0523,
      "step": 1996
    },
    {
      "epoch": 0.16137699751510132,
      "grad_norm": 2.519484043121338,
      "learning_rate": 9.932817457609777e-06,
      "loss": 1.0525,
      "step": 1997
    },
    {
      "epoch": 0.16145780722842887,
      "grad_norm": 2.6761474609375,
      "learning_rate": 9.932710506775169e-06,
      "loss": 0.9614,
      "step": 1998
    },
    {
      "epoch": 0.1615386169417564,
      "grad_norm": 3.2018771171569824,
      "learning_rate": 9.932603471455101e-06,
      "loss": 0.898,
      "step": 1999
    },
    {
      "epoch": 0.16161942665508394,
      "grad_norm": 3.4662911891937256,
      "learning_rate": 9.932496351651413e-06,
      "loss": 0.9578,
      "step": 2000
    },
    {
      "epoch": 0.16161942665508394,
      "eval_loss": 0.8565676212310791,
      "eval_runtime": 811.8928,
      "eval_samples_per_second": 102.681,
      "eval_steps_per_second": 12.835,
      "step": 2000
    },
    {
      "epoch": 0.1617002363684115,
      "grad_norm": 3.3392117023468018,
      "learning_rate": 9.932389147365937e-06,
      "loss": 0.9734,
      "step": 2001
    },
    {
      "epoch": 0.16178104608173902,
      "grad_norm": 3.312257766723633,
      "learning_rate": 9.932281858600508e-06,
      "loss": 1.0206,
      "step": 2002
    },
    {
      "epoch": 0.16186185579506657,
      "grad_norm": 2.7282278537750244,
      "learning_rate": 9.932174485356965e-06,
      "loss": 1.0603,
      "step": 2003
    },
    {
      "epoch": 0.16194266550839412,
      "grad_norm": 2.7596275806427,
      "learning_rate": 9.932067027637148e-06,
      "loss": 0.9617,
      "step": 2004
    },
    {
      "epoch": 0.16202347522172164,
      "grad_norm": 3.511976480484009,
      "learning_rate": 9.931959485442895e-06,
      "loss": 1.0428,
      "step": 2005
    },
    {
      "epoch": 0.1621042849350492,
      "grad_norm": 2.7216062545776367,
      "learning_rate": 9.93185185877605e-06,
      "loss": 1.0315,
      "step": 2006
    },
    {
      "epoch": 0.16218509464837674,
      "grad_norm": 2.992203712463379,
      "learning_rate": 9.931744147638456e-06,
      "loss": 0.9996,
      "step": 2007
    },
    {
      "epoch": 0.16226590436170427,
      "grad_norm": 2.978989839553833,
      "learning_rate": 9.931636352031957e-06,
      "loss": 1.0715,
      "step": 2008
    },
    {
      "epoch": 0.16234671407503182,
      "grad_norm": 3.601834535598755,
      "learning_rate": 9.931528471958398e-06,
      "loss": 0.9904,
      "step": 2009
    },
    {
      "epoch": 0.16242752378835937,
      "grad_norm": 3.0366740226745605,
      "learning_rate": 9.93142050741963e-06,
      "loss": 0.9713,
      "step": 2010
    },
    {
      "epoch": 0.1625083335016869,
      "grad_norm": 3.2024919986724854,
      "learning_rate": 9.931312458417501e-06,
      "loss": 1.0013,
      "step": 2011
    },
    {
      "epoch": 0.16258914321501444,
      "grad_norm": 3.1184558868408203,
      "learning_rate": 9.93120432495386e-06,
      "loss": 0.9458,
      "step": 2012
    },
    {
      "epoch": 0.162669952928342,
      "grad_norm": 3.003923177719116,
      "learning_rate": 9.931096107030561e-06,
      "loss": 1.0266,
      "step": 2013
    },
    {
      "epoch": 0.16275076264166952,
      "grad_norm": 3.309068441390991,
      "learning_rate": 9.930987804649456e-06,
      "loss": 0.9442,
      "step": 2014
    },
    {
      "epoch": 0.16283157235499707,
      "grad_norm": 2.885678768157959,
      "learning_rate": 9.930879417812402e-06,
      "loss": 1.1082,
      "step": 2015
    },
    {
      "epoch": 0.16291238206832462,
      "grad_norm": 2.6922783851623535,
      "learning_rate": 9.930770946521254e-06,
      "loss": 1.027,
      "step": 2016
    },
    {
      "epoch": 0.16299319178165214,
      "grad_norm": 3.1714210510253906,
      "learning_rate": 9.93066239077787e-06,
      "loss": 1.0827,
      "step": 2017
    },
    {
      "epoch": 0.1630740014949797,
      "grad_norm": 2.744509220123291,
      "learning_rate": 9.930553750584108e-06,
      "loss": 0.9794,
      "step": 2018
    },
    {
      "epoch": 0.16315481120830724,
      "grad_norm": 3.1168100833892822,
      "learning_rate": 9.930445025941833e-06,
      "loss": 1.064,
      "step": 2019
    },
    {
      "epoch": 0.1632356209216348,
      "grad_norm": 2.8695626258850098,
      "learning_rate": 9.930336216852902e-06,
      "loss": 1.0854,
      "step": 2020
    },
    {
      "epoch": 0.16331643063496232,
      "grad_norm": 2.774207592010498,
      "learning_rate": 9.930227323319182e-06,
      "loss": 0.9408,
      "step": 2021
    },
    {
      "epoch": 0.16339724034828987,
      "grad_norm": 2.645362377166748,
      "learning_rate": 9.930118345342535e-06,
      "loss": 1.0234,
      "step": 2022
    },
    {
      "epoch": 0.16347805006161742,
      "grad_norm": 2.7598698139190674,
      "learning_rate": 9.930009282924831e-06,
      "loss": 1.0593,
      "step": 2023
    },
    {
      "epoch": 0.16355885977494494,
      "grad_norm": 2.9732730388641357,
      "learning_rate": 9.929900136067936e-06,
      "loss": 1.1,
      "step": 2024
    },
    {
      "epoch": 0.1636396694882725,
      "grad_norm": 3.9611868858337402,
      "learning_rate": 9.929790904773722e-06,
      "loss": 0.9501,
      "step": 2025
    },
    {
      "epoch": 0.16372047920160004,
      "grad_norm": 2.8488450050354004,
      "learning_rate": 9.929681589044056e-06,
      "loss": 0.992,
      "step": 2026
    },
    {
      "epoch": 0.16380128891492757,
      "grad_norm": 2.877012014389038,
      "learning_rate": 9.929572188880811e-06,
      "loss": 1.1216,
      "step": 2027
    },
    {
      "epoch": 0.16388209862825512,
      "grad_norm": 2.778301954269409,
      "learning_rate": 9.929462704285864e-06,
      "loss": 1.0419,
      "step": 2028
    },
    {
      "epoch": 0.16396290834158267,
      "grad_norm": 3.2158327102661133,
      "learning_rate": 9.929353135261085e-06,
      "loss": 1.1185,
      "step": 2029
    },
    {
      "epoch": 0.1640437180549102,
      "grad_norm": 2.8992562294006348,
      "learning_rate": 9.929243481808357e-06,
      "loss": 0.9015,
      "step": 2030
    },
    {
      "epoch": 0.16412452776823774,
      "grad_norm": 3.4684958457946777,
      "learning_rate": 9.929133743929554e-06,
      "loss": 0.9631,
      "step": 2031
    },
    {
      "epoch": 0.1642053374815653,
      "grad_norm": 2.9608821868896484,
      "learning_rate": 9.929023921626555e-06,
      "loss": 0.9267,
      "step": 2032
    },
    {
      "epoch": 0.16428614719489282,
      "grad_norm": 3.2251927852630615,
      "learning_rate": 9.928914014901245e-06,
      "loss": 0.9074,
      "step": 2033
    },
    {
      "epoch": 0.16436695690822037,
      "grad_norm": 2.771134376525879,
      "learning_rate": 9.928804023755501e-06,
      "loss": 0.9306,
      "step": 2034
    },
    {
      "epoch": 0.16444776662154792,
      "grad_norm": 3.039679765701294,
      "learning_rate": 9.928693948191212e-06,
      "loss": 1.0715,
      "step": 2035
    },
    {
      "epoch": 0.16452857633487544,
      "grad_norm": 2.6662399768829346,
      "learning_rate": 9.92858378821026e-06,
      "loss": 1.045,
      "step": 2036
    },
    {
      "epoch": 0.164609386048203,
      "grad_norm": 2.980207681655884,
      "learning_rate": 9.928473543814532e-06,
      "loss": 1.0713,
      "step": 2037
    },
    {
      "epoch": 0.16469019576153054,
      "grad_norm": 3.3169238567352295,
      "learning_rate": 9.928363215005919e-06,
      "loss": 0.9875,
      "step": 2038
    },
    {
      "epoch": 0.16477100547485807,
      "grad_norm": 3.3401217460632324,
      "learning_rate": 9.928252801786307e-06,
      "loss": 1.0844,
      "step": 2039
    },
    {
      "epoch": 0.16485181518818562,
      "grad_norm": 2.9196760654449463,
      "learning_rate": 9.928142304157589e-06,
      "loss": 0.9477,
      "step": 2040
    },
    {
      "epoch": 0.16493262490151317,
      "grad_norm": 2.6891627311706543,
      "learning_rate": 9.928031722121658e-06,
      "loss": 0.9829,
      "step": 2041
    },
    {
      "epoch": 0.1650134346148407,
      "grad_norm": 2.8249528408050537,
      "learning_rate": 9.927921055680405e-06,
      "loss": 0.9991,
      "step": 2042
    },
    {
      "epoch": 0.16509424432816824,
      "grad_norm": 2.618173599243164,
      "learning_rate": 9.92781030483573e-06,
      "loss": 0.9323,
      "step": 2043
    },
    {
      "epoch": 0.1651750540414958,
      "grad_norm": 3.1562328338623047,
      "learning_rate": 9.927699469589528e-06,
      "loss": 0.9882,
      "step": 2044
    },
    {
      "epoch": 0.16525586375482332,
      "grad_norm": 2.5101442337036133,
      "learning_rate": 9.927588549943697e-06,
      "loss": 0.9527,
      "step": 2045
    },
    {
      "epoch": 0.16533667346815087,
      "grad_norm": 2.8681554794311523,
      "learning_rate": 9.927477545900136e-06,
      "loss": 1.057,
      "step": 2046
    },
    {
      "epoch": 0.16541748318147842,
      "grad_norm": 2.7133429050445557,
      "learning_rate": 9.927366457460748e-06,
      "loss": 0.9862,
      "step": 2047
    },
    {
      "epoch": 0.16549829289480594,
      "grad_norm": 2.907148838043213,
      "learning_rate": 9.927255284627434e-06,
      "loss": 0.9719,
      "step": 2048
    },
    {
      "epoch": 0.1655791026081335,
      "grad_norm": 3.191343307495117,
      "learning_rate": 9.927144027402097e-06,
      "loss": 0.8926,
      "step": 2049
    },
    {
      "epoch": 0.16565991232146104,
      "grad_norm": 2.7720816135406494,
      "learning_rate": 9.927032685786647e-06,
      "loss": 1.134,
      "step": 2050
    },
    {
      "epoch": 0.16574072203478857,
      "grad_norm": 3.098902702331543,
      "learning_rate": 9.926921259782988e-06,
      "loss": 0.9979,
      "step": 2051
    },
    {
      "epoch": 0.16582153174811612,
      "grad_norm": 2.7182741165161133,
      "learning_rate": 9.926809749393028e-06,
      "loss": 0.9128,
      "step": 2052
    },
    {
      "epoch": 0.16590234146144367,
      "grad_norm": 3.1051368713378906,
      "learning_rate": 9.926698154618679e-06,
      "loss": 1.1124,
      "step": 2053
    },
    {
      "epoch": 0.16598315117477122,
      "grad_norm": 3.2548975944519043,
      "learning_rate": 9.92658647546185e-06,
      "loss": 1.0029,
      "step": 2054
    },
    {
      "epoch": 0.16606396088809874,
      "grad_norm": 3.4755003452301025,
      "learning_rate": 9.926474711924456e-06,
      "loss": 1.0163,
      "step": 2055
    },
    {
      "epoch": 0.1661447706014263,
      "grad_norm": 2.8650968074798584,
      "learning_rate": 9.92636286400841e-06,
      "loss": 1.0742,
      "step": 2056
    },
    {
      "epoch": 0.16622558031475385,
      "grad_norm": 2.8151135444641113,
      "learning_rate": 9.926250931715627e-06,
      "loss": 0.9316,
      "step": 2057
    },
    {
      "epoch": 0.16630639002808137,
      "grad_norm": 2.9163153171539307,
      "learning_rate": 9.926138915048026e-06,
      "loss": 0.9452,
      "step": 2058
    },
    {
      "epoch": 0.16638719974140892,
      "grad_norm": 2.589385986328125,
      "learning_rate": 9.926026814007525e-06,
      "loss": 1.1494,
      "step": 2059
    },
    {
      "epoch": 0.16646800945473647,
      "grad_norm": 3.025043249130249,
      "learning_rate": 9.925914628596043e-06,
      "loss": 0.9743,
      "step": 2060
    },
    {
      "epoch": 0.166548819168064,
      "grad_norm": 3.18224835395813,
      "learning_rate": 9.925802358815502e-06,
      "loss": 1.0212,
      "step": 2061
    },
    {
      "epoch": 0.16662962888139154,
      "grad_norm": 2.94707989692688,
      "learning_rate": 9.925690004667824e-06,
      "loss": 1.0495,
      "step": 2062
    },
    {
      "epoch": 0.1667104385947191,
      "grad_norm": 3.2346737384796143,
      "learning_rate": 9.925577566154935e-06,
      "loss": 1.031,
      "step": 2063
    },
    {
      "epoch": 0.16679124830804662,
      "grad_norm": 2.933335304260254,
      "learning_rate": 9.92546504327876e-06,
      "loss": 0.9792,
      "step": 2064
    },
    {
      "epoch": 0.16687205802137417,
      "grad_norm": 2.6725218296051025,
      "learning_rate": 9.925352436041226e-06,
      "loss": 1.063,
      "step": 2065
    },
    {
      "epoch": 0.16695286773470172,
      "grad_norm": 2.6440131664276123,
      "learning_rate": 9.925239744444263e-06,
      "loss": 1.0466,
      "step": 2066
    },
    {
      "epoch": 0.16703367744802924,
      "grad_norm": 2.71238374710083,
      "learning_rate": 9.925126968489802e-06,
      "loss": 0.9376,
      "step": 2067
    },
    {
      "epoch": 0.1671144871613568,
      "grad_norm": 2.830446243286133,
      "learning_rate": 9.925014108179769e-06,
      "loss": 0.9454,
      "step": 2068
    },
    {
      "epoch": 0.16719529687468435,
      "grad_norm": 2.73677659034729,
      "learning_rate": 9.924901163516104e-06,
      "loss": 1.1244,
      "step": 2069
    },
    {
      "epoch": 0.16727610658801187,
      "grad_norm": 3.2142064571380615,
      "learning_rate": 9.924788134500735e-06,
      "loss": 1.0707,
      "step": 2070
    },
    {
      "epoch": 0.16735691630133942,
      "grad_norm": 3.0720906257629395,
      "learning_rate": 9.924675021135603e-06,
      "loss": 1.159,
      "step": 2071
    },
    {
      "epoch": 0.16743772601466697,
      "grad_norm": 3.0425429344177246,
      "learning_rate": 9.924561823422646e-06,
      "loss": 1.0203,
      "step": 2072
    },
    {
      "epoch": 0.1675185357279945,
      "grad_norm": 3.232916831970215,
      "learning_rate": 9.924448541363797e-06,
      "loss": 1.0186,
      "step": 2073
    },
    {
      "epoch": 0.16759934544132205,
      "grad_norm": 3.0700509548187256,
      "learning_rate": 9.924335174961e-06,
      "loss": 0.9727,
      "step": 2074
    },
    {
      "epoch": 0.1676801551546496,
      "grad_norm": 2.597489833831787,
      "learning_rate": 9.924221724216197e-06,
      "loss": 0.8967,
      "step": 2075
    },
    {
      "epoch": 0.16776096486797712,
      "grad_norm": 2.4284236431121826,
      "learning_rate": 9.924108189131331e-06,
      "loss": 0.9514,
      "step": 2076
    },
    {
      "epoch": 0.16784177458130467,
      "grad_norm": 2.5934653282165527,
      "learning_rate": 9.923994569708345e-06,
      "loss": 1.0835,
      "step": 2077
    },
    {
      "epoch": 0.16792258429463222,
      "grad_norm": 2.8386030197143555,
      "learning_rate": 9.923880865949187e-06,
      "loss": 0.9402,
      "step": 2078
    },
    {
      "epoch": 0.16800339400795974,
      "grad_norm": 2.592698574066162,
      "learning_rate": 9.923767077855802e-06,
      "loss": 0.9582,
      "step": 2079
    },
    {
      "epoch": 0.1680842037212873,
      "grad_norm": 3.1584062576293945,
      "learning_rate": 9.923653205430141e-06,
      "loss": 0.9662,
      "step": 2080
    },
    {
      "epoch": 0.16816501343461485,
      "grad_norm": 2.8627047538757324,
      "learning_rate": 9.923539248674154e-06,
      "loss": 1.0484,
      "step": 2081
    },
    {
      "epoch": 0.16824582314794237,
      "grad_norm": 2.56826114654541,
      "learning_rate": 9.923425207589793e-06,
      "loss": 1.1137,
      "step": 2082
    },
    {
      "epoch": 0.16832663286126992,
      "grad_norm": 2.89048433303833,
      "learning_rate": 9.923311082179012e-06,
      "loss": 0.9849,
      "step": 2083
    },
    {
      "epoch": 0.16840744257459747,
      "grad_norm": 2.874666452407837,
      "learning_rate": 9.923196872443764e-06,
      "loss": 1.1006,
      "step": 2084
    },
    {
      "epoch": 0.16848825228792502,
      "grad_norm": 3.502878427505493,
      "learning_rate": 9.923082578386003e-06,
      "loss": 1.1278,
      "step": 2085
    },
    {
      "epoch": 0.16856906200125255,
      "grad_norm": 2.9675581455230713,
      "learning_rate": 9.922968200007691e-06,
      "loss": 1.0886,
      "step": 2086
    },
    {
      "epoch": 0.1686498717145801,
      "grad_norm": 2.957794427871704,
      "learning_rate": 9.922853737310787e-06,
      "loss": 0.9379,
      "step": 2087
    },
    {
      "epoch": 0.16873068142790765,
      "grad_norm": 2.7929487228393555,
      "learning_rate": 9.922739190297248e-06,
      "loss": 0.9304,
      "step": 2088
    },
    {
      "epoch": 0.16881149114123517,
      "grad_norm": 3.1866209506988525,
      "learning_rate": 9.922624558969037e-06,
      "loss": 0.9009,
      "step": 2089
    },
    {
      "epoch": 0.16889230085456272,
      "grad_norm": 3.1909749507904053,
      "learning_rate": 9.922509843328118e-06,
      "loss": 1.0454,
      "step": 2090
    },
    {
      "epoch": 0.16897311056789027,
      "grad_norm": 2.9721570014953613,
      "learning_rate": 9.922395043376459e-06,
      "loss": 0.9353,
      "step": 2091
    },
    {
      "epoch": 0.1690539202812178,
      "grad_norm": 2.662489414215088,
      "learning_rate": 9.92228015911602e-06,
      "loss": 0.8749,
      "step": 2092
    },
    {
      "epoch": 0.16913472999454535,
      "grad_norm": 2.497319459915161,
      "learning_rate": 9.922165190548773e-06,
      "loss": 1.0896,
      "step": 2093
    },
    {
      "epoch": 0.1692155397078729,
      "grad_norm": 3.0421512126922607,
      "learning_rate": 9.922050137676685e-06,
      "loss": 0.9221,
      "step": 2094
    },
    {
      "epoch": 0.16929634942120042,
      "grad_norm": 3.1248340606689453,
      "learning_rate": 9.921935000501728e-06,
      "loss": 1.049,
      "step": 2095
    },
    {
      "epoch": 0.16937715913452797,
      "grad_norm": 2.7764015197753906,
      "learning_rate": 9.921819779025874e-06,
      "loss": 1.0277,
      "step": 2096
    },
    {
      "epoch": 0.16945796884785552,
      "grad_norm": 3.0548336505889893,
      "learning_rate": 9.921704473251095e-06,
      "loss": 1.0134,
      "step": 2097
    },
    {
      "epoch": 0.16953877856118305,
      "grad_norm": 2.8793864250183105,
      "learning_rate": 9.921589083179369e-06,
      "loss": 1.0039,
      "step": 2098
    },
    {
      "epoch": 0.1696195882745106,
      "grad_norm": 2.764636516571045,
      "learning_rate": 9.921473608812669e-06,
      "loss": 0.9687,
      "step": 2099
    },
    {
      "epoch": 0.16970039798783815,
      "grad_norm": 2.684819459915161,
      "learning_rate": 9.921358050152973e-06,
      "loss": 1.0376,
      "step": 2100
    },
    {
      "epoch": 0.16978120770116567,
      "grad_norm": 2.733719825744629,
      "learning_rate": 9.921242407202262e-06,
      "loss": 0.9721,
      "step": 2101
    },
    {
      "epoch": 0.16986201741449322,
      "grad_norm": 2.589684009552002,
      "learning_rate": 9.921126679962515e-06,
      "loss": 0.9558,
      "step": 2102
    },
    {
      "epoch": 0.16994282712782077,
      "grad_norm": 2.8476974964141846,
      "learning_rate": 9.921010868435716e-06,
      "loss": 1.0327,
      "step": 2103
    },
    {
      "epoch": 0.1700236368411483,
      "grad_norm": 3.0370452404022217,
      "learning_rate": 9.92089497262385e-06,
      "loss": 0.9495,
      "step": 2104
    },
    {
      "epoch": 0.17010444655447585,
      "grad_norm": 2.854360342025757,
      "learning_rate": 9.920778992528896e-06,
      "loss": 1.087,
      "step": 2105
    },
    {
      "epoch": 0.1701852562678034,
      "grad_norm": 3.350029945373535,
      "learning_rate": 9.920662928152846e-06,
      "loss": 0.928,
      "step": 2106
    },
    {
      "epoch": 0.17026606598113092,
      "grad_norm": 2.5941176414489746,
      "learning_rate": 9.920546779497686e-06,
      "loss": 1.0268,
      "step": 2107
    },
    {
      "epoch": 0.17034687569445847,
      "grad_norm": 2.824829339981079,
      "learning_rate": 9.920430546565405e-06,
      "loss": 1.0249,
      "step": 2108
    },
    {
      "epoch": 0.17042768540778602,
      "grad_norm": 3.5965116024017334,
      "learning_rate": 9.920314229357995e-06,
      "loss": 1.0276,
      "step": 2109
    },
    {
      "epoch": 0.17050849512111355,
      "grad_norm": 2.5723979473114014,
      "learning_rate": 9.920197827877445e-06,
      "loss": 1.0343,
      "step": 2110
    },
    {
      "epoch": 0.1705893048344411,
      "grad_norm": 3.0630664825439453,
      "learning_rate": 9.920081342125753e-06,
      "loss": 0.963,
      "step": 2111
    },
    {
      "epoch": 0.17067011454776865,
      "grad_norm": 2.9649055004119873,
      "learning_rate": 9.919964772104912e-06,
      "loss": 0.9868,
      "step": 2112
    },
    {
      "epoch": 0.17075092426109617,
      "grad_norm": 2.8002421855926514,
      "learning_rate": 9.919848117816919e-06,
      "loss": 1.0483,
      "step": 2113
    },
    {
      "epoch": 0.17083173397442372,
      "grad_norm": 3.262807607650757,
      "learning_rate": 9.919731379263772e-06,
      "loss": 1.1254,
      "step": 2114
    },
    {
      "epoch": 0.17091254368775127,
      "grad_norm": 2.4717869758605957,
      "learning_rate": 9.91961455644747e-06,
      "loss": 0.9715,
      "step": 2115
    },
    {
      "epoch": 0.1709933534010788,
      "grad_norm": 2.941643476486206,
      "learning_rate": 9.919497649370014e-06,
      "loss": 0.9739,
      "step": 2116
    },
    {
      "epoch": 0.17107416311440635,
      "grad_norm": 2.670825481414795,
      "learning_rate": 9.919380658033405e-06,
      "loss": 1.0739,
      "step": 2117
    },
    {
      "epoch": 0.1711549728277339,
      "grad_norm": 3.208672285079956,
      "learning_rate": 9.91926358243965e-06,
      "loss": 1.0586,
      "step": 2118
    },
    {
      "epoch": 0.17123578254106145,
      "grad_norm": 2.5170207023620605,
      "learning_rate": 9.919146422590753e-06,
      "loss": 0.9602,
      "step": 2119
    },
    {
      "epoch": 0.17131659225438897,
      "grad_norm": 2.994489908218384,
      "learning_rate": 9.91902917848872e-06,
      "loss": 1.0003,
      "step": 2120
    },
    {
      "epoch": 0.17139740196771652,
      "grad_norm": 3.4159748554229736,
      "learning_rate": 9.918911850135557e-06,
      "loss": 1.0409,
      "step": 2121
    },
    {
      "epoch": 0.17147821168104407,
      "grad_norm": 2.351083993911743,
      "learning_rate": 9.918794437533279e-06,
      "loss": 1.0347,
      "step": 2122
    },
    {
      "epoch": 0.1715590213943716,
      "grad_norm": 2.8775289058685303,
      "learning_rate": 9.918676940683891e-06,
      "loss": 1.0373,
      "step": 2123
    },
    {
      "epoch": 0.17163983110769915,
      "grad_norm": 2.7405753135681152,
      "learning_rate": 9.918559359589411e-06,
      "loss": 0.9521,
      "step": 2124
    },
    {
      "epoch": 0.1717206408210267,
      "grad_norm": 2.8221731185913086,
      "learning_rate": 9.918441694251848e-06,
      "loss": 1.1207,
      "step": 2125
    },
    {
      "epoch": 0.17180145053435422,
      "grad_norm": 2.7699766159057617,
      "learning_rate": 9.918323944673221e-06,
      "loss": 0.8984,
      "step": 2126
    },
    {
      "epoch": 0.17188226024768177,
      "grad_norm": 3.1191518306732178,
      "learning_rate": 9.918206110855543e-06,
      "loss": 1.0376,
      "step": 2127
    },
    {
      "epoch": 0.17196306996100932,
      "grad_norm": 2.816222667694092,
      "learning_rate": 9.918088192800836e-06,
      "loss": 1.0366,
      "step": 2128
    },
    {
      "epoch": 0.17204387967433685,
      "grad_norm": 2.8271803855895996,
      "learning_rate": 9.917970190511117e-06,
      "loss": 0.9186,
      "step": 2129
    },
    {
      "epoch": 0.1721246893876644,
      "grad_norm": 2.88246488571167,
      "learning_rate": 9.91785210398841e-06,
      "loss": 0.9728,
      "step": 2130
    },
    {
      "epoch": 0.17220549910099195,
      "grad_norm": 2.6959381103515625,
      "learning_rate": 9.917733933234733e-06,
      "loss": 0.9794,
      "step": 2131
    },
    {
      "epoch": 0.17228630881431947,
      "grad_norm": 2.781506299972534,
      "learning_rate": 9.917615678252114e-06,
      "loss": 0.9592,
      "step": 2132
    },
    {
      "epoch": 0.17236711852764702,
      "grad_norm": 3.295612335205078,
      "learning_rate": 9.917497339042579e-06,
      "loss": 1.0035,
      "step": 2133
    },
    {
      "epoch": 0.17244792824097457,
      "grad_norm": 3.5183417797088623,
      "learning_rate": 9.917378915608151e-06,
      "loss": 1.1322,
      "step": 2134
    },
    {
      "epoch": 0.1725287379543021,
      "grad_norm": 2.9559192657470703,
      "learning_rate": 9.917260407950859e-06,
      "loss": 1.0715,
      "step": 2135
    },
    {
      "epoch": 0.17260954766762965,
      "grad_norm": 2.6079885959625244,
      "learning_rate": 9.917141816072737e-06,
      "loss": 0.9281,
      "step": 2136
    },
    {
      "epoch": 0.1726903573809572,
      "grad_norm": 3.3428955078125,
      "learning_rate": 9.917023139975813e-06,
      "loss": 0.9539,
      "step": 2137
    },
    {
      "epoch": 0.17277116709428472,
      "grad_norm": 2.4502480030059814,
      "learning_rate": 9.916904379662118e-06,
      "loss": 0.8888,
      "step": 2138
    },
    {
      "epoch": 0.17285197680761227,
      "grad_norm": 2.9971001148223877,
      "learning_rate": 9.916785535133688e-06,
      "loss": 1.1042,
      "step": 2139
    },
    {
      "epoch": 0.17293278652093982,
      "grad_norm": 2.7164599895477295,
      "learning_rate": 9.91666660639256e-06,
      "loss": 1.0137,
      "step": 2140
    },
    {
      "epoch": 0.17301359623426735,
      "grad_norm": 2.8608083724975586,
      "learning_rate": 9.916547593440769e-06,
      "loss": 1.0781,
      "step": 2141
    },
    {
      "epoch": 0.1730944059475949,
      "grad_norm": 2.838046073913574,
      "learning_rate": 9.916428496280353e-06,
      "loss": 1.0676,
      "step": 2142
    },
    {
      "epoch": 0.17317521566092245,
      "grad_norm": 2.646768808364868,
      "learning_rate": 9.916309314913353e-06,
      "loss": 0.9184,
      "step": 2143
    },
    {
      "epoch": 0.17325602537424997,
      "grad_norm": 3.8071377277374268,
      "learning_rate": 9.91619004934181e-06,
      "loss": 1.0174,
      "step": 2144
    },
    {
      "epoch": 0.17333683508757752,
      "grad_norm": 2.8915224075317383,
      "learning_rate": 9.916070699567767e-06,
      "loss": 0.9417,
      "step": 2145
    },
    {
      "epoch": 0.17341764480090507,
      "grad_norm": 2.7951903343200684,
      "learning_rate": 9.915951265593266e-06,
      "loss": 0.9639,
      "step": 2146
    },
    {
      "epoch": 0.1734984545142326,
      "grad_norm": 3.3803536891937256,
      "learning_rate": 9.915831747420357e-06,
      "loss": 1.0123,
      "step": 2147
    },
    {
      "epoch": 0.17357926422756015,
      "grad_norm": 3.14410662651062,
      "learning_rate": 9.915712145051084e-06,
      "loss": 1.0791,
      "step": 2148
    },
    {
      "epoch": 0.1736600739408877,
      "grad_norm": 2.457718849182129,
      "learning_rate": 9.915592458487495e-06,
      "loss": 1.0325,
      "step": 2149
    },
    {
      "epoch": 0.17374088365421525,
      "grad_norm": 2.9668896198272705,
      "learning_rate": 9.915472687731642e-06,
      "loss": 0.9524,
      "step": 2150
    },
    {
      "epoch": 0.17382169336754277,
      "grad_norm": 2.822756290435791,
      "learning_rate": 9.915352832785574e-06,
      "loss": 1.0502,
      "step": 2151
    },
    {
      "epoch": 0.17390250308087032,
      "grad_norm": 2.743389129638672,
      "learning_rate": 9.915232893651347e-06,
      "loss": 1.0771,
      "step": 2152
    },
    {
      "epoch": 0.17398331279419788,
      "grad_norm": 2.876563787460327,
      "learning_rate": 9.915112870331012e-06,
      "loss": 0.975,
      "step": 2153
    },
    {
      "epoch": 0.1740641225075254,
      "grad_norm": 2.6534552574157715,
      "learning_rate": 9.914992762826628e-06,
      "loss": 0.9492,
      "step": 2154
    },
    {
      "epoch": 0.17414493222085295,
      "grad_norm": 3.0383739471435547,
      "learning_rate": 9.914872571140248e-06,
      "loss": 0.9855,
      "step": 2155
    },
    {
      "epoch": 0.1742257419341805,
      "grad_norm": 2.998772144317627,
      "learning_rate": 9.914752295273934e-06,
      "loss": 0.9629,
      "step": 2156
    },
    {
      "epoch": 0.17430655164750802,
      "grad_norm": 2.6088712215423584,
      "learning_rate": 9.914631935229746e-06,
      "loss": 1.0546,
      "step": 2157
    },
    {
      "epoch": 0.17438736136083557,
      "grad_norm": 2.937267303466797,
      "learning_rate": 9.914511491009744e-06,
      "loss": 0.9758,
      "step": 2158
    },
    {
      "epoch": 0.17446817107416313,
      "grad_norm": 2.9102063179016113,
      "learning_rate": 9.914390962615992e-06,
      "loss": 0.955,
      "step": 2159
    },
    {
      "epoch": 0.17454898078749065,
      "grad_norm": 2.8595845699310303,
      "learning_rate": 9.914270350050552e-06,
      "loss": 0.9489,
      "step": 2160
    },
    {
      "epoch": 0.1746297905008182,
      "grad_norm": 2.782841205596924,
      "learning_rate": 9.914149653315492e-06,
      "loss": 1.1525,
      "step": 2161
    },
    {
      "epoch": 0.17471060021414575,
      "grad_norm": 2.7954177856445312,
      "learning_rate": 9.914028872412879e-06,
      "loss": 0.9337,
      "step": 2162
    },
    {
      "epoch": 0.17479140992747327,
      "grad_norm": 2.7344326972961426,
      "learning_rate": 9.91390800734478e-06,
      "loss": 0.9292,
      "step": 2163
    },
    {
      "epoch": 0.17487221964080082,
      "grad_norm": 2.9386417865753174,
      "learning_rate": 9.91378705811327e-06,
      "loss": 0.9851,
      "step": 2164
    },
    {
      "epoch": 0.17495302935412838,
      "grad_norm": 2.8945112228393555,
      "learning_rate": 9.913666024720414e-06,
      "loss": 1.0128,
      "step": 2165
    },
    {
      "epoch": 0.1750338390674559,
      "grad_norm": 3.150853395462036,
      "learning_rate": 9.91354490716829e-06,
      "loss": 1.0122,
      "step": 2166
    },
    {
      "epoch": 0.17511464878078345,
      "grad_norm": 2.9394326210021973,
      "learning_rate": 9.91342370545897e-06,
      "loss": 0.9914,
      "step": 2167
    },
    {
      "epoch": 0.175195458494111,
      "grad_norm": 2.774982213973999,
      "learning_rate": 9.91330241959453e-06,
      "loss": 1.0258,
      "step": 2168
    },
    {
      "epoch": 0.17527626820743852,
      "grad_norm": 2.781067132949829,
      "learning_rate": 9.913181049577048e-06,
      "loss": 1.0331,
      "step": 2169
    },
    {
      "epoch": 0.17535707792076607,
      "grad_norm": 2.9024343490600586,
      "learning_rate": 9.913059595408603e-06,
      "loss": 0.9953,
      "step": 2170
    },
    {
      "epoch": 0.17543788763409363,
      "grad_norm": 3.012073516845703,
      "learning_rate": 9.912938057091274e-06,
      "loss": 1.0043,
      "step": 2171
    },
    {
      "epoch": 0.17551869734742115,
      "grad_norm": 2.716157913208008,
      "learning_rate": 9.912816434627147e-06,
      "loss": 1.0252,
      "step": 2172
    },
    {
      "epoch": 0.1755995070607487,
      "grad_norm": 3.0797886848449707,
      "learning_rate": 9.912694728018297e-06,
      "loss": 0.9114,
      "step": 2173
    },
    {
      "epoch": 0.17568031677407625,
      "grad_norm": 2.655963182449341,
      "learning_rate": 9.912572937266816e-06,
      "loss": 1.1196,
      "step": 2174
    },
    {
      "epoch": 0.17576112648740377,
      "grad_norm": 3.1143031120300293,
      "learning_rate": 9.912451062374786e-06,
      "loss": 1.0057,
      "step": 2175
    },
    {
      "epoch": 0.17584193620073132,
      "grad_norm": 2.9792284965515137,
      "learning_rate": 9.912329103344295e-06,
      "loss": 0.9854,
      "step": 2176
    },
    {
      "epoch": 0.17592274591405888,
      "grad_norm": 2.852971315383911,
      "learning_rate": 9.912207060177433e-06,
      "loss": 0.9972,
      "step": 2177
    },
    {
      "epoch": 0.1760035556273864,
      "grad_norm": 2.5394182205200195,
      "learning_rate": 9.91208493287629e-06,
      "loss": 0.9824,
      "step": 2178
    },
    {
      "epoch": 0.17608436534071395,
      "grad_norm": 2.984800338745117,
      "learning_rate": 9.911962721442957e-06,
      "loss": 1.1021,
      "step": 2179
    },
    {
      "epoch": 0.1761651750540415,
      "grad_norm": 2.647099733352661,
      "learning_rate": 9.91184042587953e-06,
      "loss": 0.93,
      "step": 2180
    },
    {
      "epoch": 0.17624598476736905,
      "grad_norm": 2.4311182498931885,
      "learning_rate": 9.911718046188096e-06,
      "loss": 1.074,
      "step": 2181
    },
    {
      "epoch": 0.17632679448069657,
      "grad_norm": 2.8348121643066406,
      "learning_rate": 9.91159558237076e-06,
      "loss": 0.9774,
      "step": 2182
    },
    {
      "epoch": 0.17640760419402413,
      "grad_norm": 2.905052423477173,
      "learning_rate": 9.911473034429617e-06,
      "loss": 0.9155,
      "step": 2183
    },
    {
      "epoch": 0.17648841390735168,
      "grad_norm": 2.745105504989624,
      "learning_rate": 9.911350402366764e-06,
      "loss": 0.9996,
      "step": 2184
    },
    {
      "epoch": 0.1765692236206792,
      "grad_norm": 3.183624267578125,
      "learning_rate": 9.911227686184299e-06,
      "loss": 0.927,
      "step": 2185
    },
    {
      "epoch": 0.17665003333400675,
      "grad_norm": 2.996196746826172,
      "learning_rate": 9.911104885884331e-06,
      "loss": 0.9375,
      "step": 2186
    },
    {
      "epoch": 0.1767308430473343,
      "grad_norm": 3.10860276222229,
      "learning_rate": 9.910982001468958e-06,
      "loss": 1.0341,
      "step": 2187
    },
    {
      "epoch": 0.17681165276066182,
      "grad_norm": 3.0041086673736572,
      "learning_rate": 9.910859032940286e-06,
      "loss": 1.028,
      "step": 2188
    },
    {
      "epoch": 0.17689246247398938,
      "grad_norm": 3.4713857173919678,
      "learning_rate": 9.91073598030042e-06,
      "loss": 1.0096,
      "step": 2189
    },
    {
      "epoch": 0.17697327218731693,
      "grad_norm": 3.2745108604431152,
      "learning_rate": 9.91061284355147e-06,
      "loss": 1.0437,
      "step": 2190
    },
    {
      "epoch": 0.17705408190064445,
      "grad_norm": 3.0695979595184326,
      "learning_rate": 9.910489622695542e-06,
      "loss": 0.9703,
      "step": 2191
    },
    {
      "epoch": 0.177134891613972,
      "grad_norm": 3.1220107078552246,
      "learning_rate": 9.910366317734752e-06,
      "loss": 1.1641,
      "step": 2192
    },
    {
      "epoch": 0.17721570132729955,
      "grad_norm": 2.478935956954956,
      "learning_rate": 9.910242928671206e-06,
      "loss": 0.9982,
      "step": 2193
    },
    {
      "epoch": 0.17729651104062708,
      "grad_norm": 2.8023176193237305,
      "learning_rate": 9.91011945550702e-06,
      "loss": 1.08,
      "step": 2194
    },
    {
      "epoch": 0.17737732075395463,
      "grad_norm": 2.962545394897461,
      "learning_rate": 9.909995898244306e-06,
      "loss": 1.052,
      "step": 2195
    },
    {
      "epoch": 0.17745813046728218,
      "grad_norm": 3.106247663497925,
      "learning_rate": 9.909872256885184e-06,
      "loss": 0.9073,
      "step": 2196
    },
    {
      "epoch": 0.1775389401806097,
      "grad_norm": 2.672226667404175,
      "learning_rate": 9.909748531431772e-06,
      "loss": 0.9634,
      "step": 2197
    },
    {
      "epoch": 0.17761974989393725,
      "grad_norm": 2.9921627044677734,
      "learning_rate": 9.909624721886186e-06,
      "loss": 0.8934,
      "step": 2198
    },
    {
      "epoch": 0.1777005596072648,
      "grad_norm": 2.4521498680114746,
      "learning_rate": 9.90950082825055e-06,
      "loss": 1.0277,
      "step": 2199
    },
    {
      "epoch": 0.17778136932059233,
      "grad_norm": 2.8029232025146484,
      "learning_rate": 9.90937685052698e-06,
      "loss": 0.8953,
      "step": 2200
    },
    {
      "epoch": 0.17786217903391988,
      "grad_norm": 2.7503063678741455,
      "learning_rate": 9.909252788717606e-06,
      "loss": 1.0325,
      "step": 2201
    },
    {
      "epoch": 0.17794298874724743,
      "grad_norm": 3.0869476795196533,
      "learning_rate": 9.90912864282455e-06,
      "loss": 0.9307,
      "step": 2202
    },
    {
      "epoch": 0.17802379846057495,
      "grad_norm": 2.852287530899048,
      "learning_rate": 9.909004412849939e-06,
      "loss": 0.9603,
      "step": 2203
    },
    {
      "epoch": 0.1781046081739025,
      "grad_norm": 2.903380870819092,
      "learning_rate": 9.908880098795898e-06,
      "loss": 1.0392,
      "step": 2204
    },
    {
      "epoch": 0.17818541788723005,
      "grad_norm": 2.9859073162078857,
      "learning_rate": 9.90875570066456e-06,
      "loss": 1.0383,
      "step": 2205
    },
    {
      "epoch": 0.17826622760055758,
      "grad_norm": 2.5537900924682617,
      "learning_rate": 9.908631218458056e-06,
      "loss": 1.1067,
      "step": 2206
    },
    {
      "epoch": 0.17834703731388513,
      "grad_norm": 2.6442184448242188,
      "learning_rate": 9.908506652178513e-06,
      "loss": 1.0278,
      "step": 2207
    },
    {
      "epoch": 0.17842784702721268,
      "grad_norm": 2.952118158340454,
      "learning_rate": 9.90838200182807e-06,
      "loss": 0.9521,
      "step": 2208
    },
    {
      "epoch": 0.1785086567405402,
      "grad_norm": 2.5400750637054443,
      "learning_rate": 9.908257267408861e-06,
      "loss": 0.9087,
      "step": 2209
    },
    {
      "epoch": 0.17858946645386775,
      "grad_norm": 2.7366065979003906,
      "learning_rate": 9.908132448923019e-06,
      "loss": 0.9954,
      "step": 2210
    },
    {
      "epoch": 0.1786702761671953,
      "grad_norm": 2.8425841331481934,
      "learning_rate": 9.908007546372685e-06,
      "loss": 0.9866,
      "step": 2211
    },
    {
      "epoch": 0.17875108588052283,
      "grad_norm": 3.0371780395507812,
      "learning_rate": 9.907882559759996e-06,
      "loss": 0.9368,
      "step": 2212
    },
    {
      "epoch": 0.17883189559385038,
      "grad_norm": 2.5463290214538574,
      "learning_rate": 9.907757489087094e-06,
      "loss": 1.0368,
      "step": 2213
    },
    {
      "epoch": 0.17891270530717793,
      "grad_norm": 2.4215946197509766,
      "learning_rate": 9.907632334356123e-06,
      "loss": 1.0487,
      "step": 2214
    },
    {
      "epoch": 0.17899351502050548,
      "grad_norm": 2.9008519649505615,
      "learning_rate": 9.907507095569222e-06,
      "loss": 1.0081,
      "step": 2215
    },
    {
      "epoch": 0.179074324733833,
      "grad_norm": 2.6641945838928223,
      "learning_rate": 9.907381772728541e-06,
      "loss": 1.0243,
      "step": 2216
    },
    {
      "epoch": 0.17915513444716055,
      "grad_norm": 2.735846996307373,
      "learning_rate": 9.907256365836224e-06,
      "loss": 0.9198,
      "step": 2217
    },
    {
      "epoch": 0.1792359441604881,
      "grad_norm": 3.134108066558838,
      "learning_rate": 9.90713087489442e-06,
      "loss": 1.0603,
      "step": 2218
    },
    {
      "epoch": 0.17931675387381563,
      "grad_norm": 2.7960338592529297,
      "learning_rate": 9.907005299905275e-06,
      "loss": 0.8058,
      "step": 2219
    },
    {
      "epoch": 0.17939756358714318,
      "grad_norm": 3.0357117652893066,
      "learning_rate": 9.906879640870943e-06,
      "loss": 1.0705,
      "step": 2220
    },
    {
      "epoch": 0.17947837330047073,
      "grad_norm": 2.984736442565918,
      "learning_rate": 9.906753897793578e-06,
      "loss": 1.009,
      "step": 2221
    },
    {
      "epoch": 0.17955918301379825,
      "grad_norm": 2.8583996295928955,
      "learning_rate": 9.906628070675329e-06,
      "loss": 1.0643,
      "step": 2222
    },
    {
      "epoch": 0.1796399927271258,
      "grad_norm": 3.1405558586120605,
      "learning_rate": 9.906502159518353e-06,
      "loss": 1.118,
      "step": 2223
    },
    {
      "epoch": 0.17972080244045335,
      "grad_norm": 2.5613033771514893,
      "learning_rate": 9.906376164324808e-06,
      "loss": 1.0275,
      "step": 2224
    },
    {
      "epoch": 0.17980161215378088,
      "grad_norm": 3.179513454437256,
      "learning_rate": 9.906250085096852e-06,
      "loss": 1.1959,
      "step": 2225
    },
    {
      "epoch": 0.17988242186710843,
      "grad_norm": 3.293401002883911,
      "learning_rate": 9.90612392183664e-06,
      "loss": 1.0808,
      "step": 2226
    },
    {
      "epoch": 0.17996323158043598,
      "grad_norm": 2.9560415744781494,
      "learning_rate": 9.90599767454634e-06,
      "loss": 0.9355,
      "step": 2227
    },
    {
      "epoch": 0.1800440412937635,
      "grad_norm": 3.093355655670166,
      "learning_rate": 9.905871343228108e-06,
      "loss": 0.9234,
      "step": 2228
    },
    {
      "epoch": 0.18012485100709105,
      "grad_norm": 3.24405574798584,
      "learning_rate": 9.905744927884112e-06,
      "loss": 0.991,
      "step": 2229
    },
    {
      "epoch": 0.1802056607204186,
      "grad_norm": 2.381044864654541,
      "learning_rate": 9.905618428516514e-06,
      "loss": 1.0116,
      "step": 2230
    },
    {
      "epoch": 0.18028647043374613,
      "grad_norm": 2.3708693981170654,
      "learning_rate": 9.905491845127485e-06,
      "loss": 1.156,
      "step": 2231
    },
    {
      "epoch": 0.18036728014707368,
      "grad_norm": 2.8890292644500732,
      "learning_rate": 9.905365177719187e-06,
      "loss": 0.9778,
      "step": 2232
    },
    {
      "epoch": 0.18044808986040123,
      "grad_norm": 2.637432098388672,
      "learning_rate": 9.905238426293793e-06,
      "loss": 0.9777,
      "step": 2233
    },
    {
      "epoch": 0.18052889957372875,
      "grad_norm": 2.81315016746521,
      "learning_rate": 9.905111590853475e-06,
      "loss": 1.0677,
      "step": 2234
    },
    {
      "epoch": 0.1806097092870563,
      "grad_norm": 2.9049456119537354,
      "learning_rate": 9.904984671400403e-06,
      "loss": 1.0104,
      "step": 2235
    },
    {
      "epoch": 0.18069051900038385,
      "grad_norm": 2.824903964996338,
      "learning_rate": 9.904857667936753e-06,
      "loss": 1.0085,
      "step": 2236
    },
    {
      "epoch": 0.18077132871371138,
      "grad_norm": 2.541304588317871,
      "learning_rate": 9.904730580464698e-06,
      "loss": 1.0775,
      "step": 2237
    },
    {
      "epoch": 0.18085213842703893,
      "grad_norm": 3.0248048305511475,
      "learning_rate": 9.904603408986416e-06,
      "loss": 0.946,
      "step": 2238
    },
    {
      "epoch": 0.18093294814036648,
      "grad_norm": 2.661759853363037,
      "learning_rate": 9.904476153504085e-06,
      "loss": 1.1053,
      "step": 2239
    },
    {
      "epoch": 0.181013757853694,
      "grad_norm": 2.5051047801971436,
      "learning_rate": 9.904348814019885e-06,
      "loss": 0.9926,
      "step": 2240
    },
    {
      "epoch": 0.18109456756702155,
      "grad_norm": 2.8634889125823975,
      "learning_rate": 9.904221390535996e-06,
      "loss": 0.9587,
      "step": 2241
    },
    {
      "epoch": 0.1811753772803491,
      "grad_norm": 2.776017904281616,
      "learning_rate": 9.904093883054602e-06,
      "loss": 0.9272,
      "step": 2242
    },
    {
      "epoch": 0.18125618699367663,
      "grad_norm": 2.577670097351074,
      "learning_rate": 9.903966291577884e-06,
      "loss": 0.9992,
      "step": 2243
    },
    {
      "epoch": 0.18133699670700418,
      "grad_norm": 2.838353395462036,
      "learning_rate": 9.90383861610803e-06,
      "loss": 1.012,
      "step": 2244
    },
    {
      "epoch": 0.18141780642033173,
      "grad_norm": 2.815898895263672,
      "learning_rate": 9.903710856647227e-06,
      "loss": 1.0239,
      "step": 2245
    },
    {
      "epoch": 0.18149861613365928,
      "grad_norm": 2.680938243865967,
      "learning_rate": 9.903583013197662e-06,
      "loss": 1.0676,
      "step": 2246
    },
    {
      "epoch": 0.1815794258469868,
      "grad_norm": 2.8625309467315674,
      "learning_rate": 9.903455085761525e-06,
      "loss": 1.0337,
      "step": 2247
    },
    {
      "epoch": 0.18166023556031435,
      "grad_norm": 2.6191186904907227,
      "learning_rate": 9.903327074341006e-06,
      "loss": 1.0447,
      "step": 2248
    },
    {
      "epoch": 0.1817410452736419,
      "grad_norm": 2.5081663131713867,
      "learning_rate": 9.9031989789383e-06,
      "loss": 1.0414,
      "step": 2249
    },
    {
      "epoch": 0.18182185498696943,
      "grad_norm": 3.1408021450042725,
      "learning_rate": 9.903070799555598e-06,
      "loss": 1.0195,
      "step": 2250
    },
    {
      "epoch": 0.18190266470029698,
      "grad_norm": 2.689897298812866,
      "learning_rate": 9.902942536195098e-06,
      "loss": 1.0318,
      "step": 2251
    },
    {
      "epoch": 0.18198347441362453,
      "grad_norm": 3.066641092300415,
      "learning_rate": 9.902814188858994e-06,
      "loss": 1.0626,
      "step": 2252
    },
    {
      "epoch": 0.18206428412695205,
      "grad_norm": 3.338888645172119,
      "learning_rate": 9.902685757549486e-06,
      "loss": 0.9944,
      "step": 2253
    },
    {
      "epoch": 0.1821450938402796,
      "grad_norm": 2.9678893089294434,
      "learning_rate": 9.902557242268775e-06,
      "loss": 1.0364,
      "step": 2254
    },
    {
      "epoch": 0.18222590355360715,
      "grad_norm": 3.2926056385040283,
      "learning_rate": 9.90242864301906e-06,
      "loss": 1.0177,
      "step": 2255
    },
    {
      "epoch": 0.18230671326693468,
      "grad_norm": 2.818373680114746,
      "learning_rate": 9.902299959802546e-06,
      "loss": 1.2125,
      "step": 2256
    },
    {
      "epoch": 0.18238752298026223,
      "grad_norm": 2.7557077407836914,
      "learning_rate": 9.902171192621435e-06,
      "loss": 0.9113,
      "step": 2257
    },
    {
      "epoch": 0.18246833269358978,
      "grad_norm": 2.6193900108337402,
      "learning_rate": 9.902042341477932e-06,
      "loss": 1.1363,
      "step": 2258
    },
    {
      "epoch": 0.1825491424069173,
      "grad_norm": 2.5538594722747803,
      "learning_rate": 9.901913406374246e-06,
      "loss": 1.0446,
      "step": 2259
    },
    {
      "epoch": 0.18262995212024485,
      "grad_norm": 2.68630051612854,
      "learning_rate": 9.901784387312583e-06,
      "loss": 1.0794,
      "step": 2260
    },
    {
      "epoch": 0.1827107618335724,
      "grad_norm": 2.8566384315490723,
      "learning_rate": 9.901655284295156e-06,
      "loss": 1.0573,
      "step": 2261
    },
    {
      "epoch": 0.18279157154689993,
      "grad_norm": 2.828277111053467,
      "learning_rate": 9.901526097324171e-06,
      "loss": 0.9004,
      "step": 2262
    },
    {
      "epoch": 0.18287238126022748,
      "grad_norm": 2.6929445266723633,
      "learning_rate": 9.901396826401846e-06,
      "loss": 1.0182,
      "step": 2263
    },
    {
      "epoch": 0.18295319097355503,
      "grad_norm": 2.9283299446105957,
      "learning_rate": 9.901267471530393e-06,
      "loss": 0.9546,
      "step": 2264
    },
    {
      "epoch": 0.18303400068688255,
      "grad_norm": 2.531330108642578,
      "learning_rate": 9.901138032712028e-06,
      "loss": 0.958,
      "step": 2265
    },
    {
      "epoch": 0.1831148104002101,
      "grad_norm": 2.7807202339172363,
      "learning_rate": 9.901008509948967e-06,
      "loss": 1.0124,
      "step": 2266
    },
    {
      "epoch": 0.18319562011353765,
      "grad_norm": 2.8674800395965576,
      "learning_rate": 9.900878903243428e-06,
      "loss": 1.034,
      "step": 2267
    },
    {
      "epoch": 0.18327642982686518,
      "grad_norm": 2.8897581100463867,
      "learning_rate": 9.900749212597632e-06,
      "loss": 0.9783,
      "step": 2268
    },
    {
      "epoch": 0.18335723954019273,
      "grad_norm": 2.7262604236602783,
      "learning_rate": 9.900619438013802e-06,
      "loss": 0.971,
      "step": 2269
    },
    {
      "epoch": 0.18343804925352028,
      "grad_norm": 3.7016351222991943,
      "learning_rate": 9.900489579494156e-06,
      "loss": 0.9499,
      "step": 2270
    },
    {
      "epoch": 0.1835188589668478,
      "grad_norm": 2.412543535232544,
      "learning_rate": 9.900359637040922e-06,
      "loss": 1.0185,
      "step": 2271
    },
    {
      "epoch": 0.18359966868017535,
      "grad_norm": 2.899360179901123,
      "learning_rate": 9.900229610656324e-06,
      "loss": 0.9437,
      "step": 2272
    },
    {
      "epoch": 0.1836804783935029,
      "grad_norm": 2.7710652351379395,
      "learning_rate": 9.90009950034259e-06,
      "loss": 0.9944,
      "step": 2273
    },
    {
      "epoch": 0.18376128810683043,
      "grad_norm": 3.9818215370178223,
      "learning_rate": 9.89996930610195e-06,
      "loss": 1.0571,
      "step": 2274
    },
    {
      "epoch": 0.18384209782015798,
      "grad_norm": 4.321252822875977,
      "learning_rate": 9.89983902793663e-06,
      "loss": 0.9093,
      "step": 2275
    },
    {
      "epoch": 0.18392290753348553,
      "grad_norm": 2.7052674293518066,
      "learning_rate": 9.899708665848864e-06,
      "loss": 0.9636,
      "step": 2276
    },
    {
      "epoch": 0.18400371724681305,
      "grad_norm": 2.865767240524292,
      "learning_rate": 9.899578219840883e-06,
      "loss": 1.0194,
      "step": 2277
    },
    {
      "epoch": 0.1840845269601406,
      "grad_norm": 3.1455867290496826,
      "learning_rate": 9.899447689914924e-06,
      "loss": 1.0515,
      "step": 2278
    },
    {
      "epoch": 0.18416533667346816,
      "grad_norm": 2.658369302749634,
      "learning_rate": 9.89931707607322e-06,
      "loss": 1.0933,
      "step": 2279
    },
    {
      "epoch": 0.1842461463867957,
      "grad_norm": 3.026167392730713,
      "learning_rate": 9.899186378318008e-06,
      "loss": 0.9614,
      "step": 2280
    },
    {
      "epoch": 0.18432695610012323,
      "grad_norm": 3.9894533157348633,
      "learning_rate": 9.89905559665153e-06,
      "loss": 0.9487,
      "step": 2281
    },
    {
      "epoch": 0.18440776581345078,
      "grad_norm": 3.4070656299591064,
      "learning_rate": 9.898924731076022e-06,
      "loss": 0.9747,
      "step": 2282
    },
    {
      "epoch": 0.18448857552677833,
      "grad_norm": 2.44861102104187,
      "learning_rate": 9.89879378159373e-06,
      "loss": 0.9783,
      "step": 2283
    },
    {
      "epoch": 0.18456938524010585,
      "grad_norm": 2.5936949253082275,
      "learning_rate": 9.898662748206891e-06,
      "loss": 0.9835,
      "step": 2284
    },
    {
      "epoch": 0.1846501949534334,
      "grad_norm": 2.9348199367523193,
      "learning_rate": 9.89853163091775e-06,
      "loss": 0.957,
      "step": 2285
    },
    {
      "epoch": 0.18473100466676096,
      "grad_norm": 2.6662724018096924,
      "learning_rate": 9.898400429728559e-06,
      "loss": 0.9759,
      "step": 2286
    },
    {
      "epoch": 0.18481181438008848,
      "grad_norm": 3.1823620796203613,
      "learning_rate": 9.89826914464156e-06,
      "loss": 1.082,
      "step": 2287
    },
    {
      "epoch": 0.18489262409341603,
      "grad_norm": 2.6004109382629395,
      "learning_rate": 9.898137775659002e-06,
      "loss": 1.0556,
      "step": 2288
    },
    {
      "epoch": 0.18497343380674358,
      "grad_norm": 2.8404204845428467,
      "learning_rate": 9.898006322783137e-06,
      "loss": 0.9183,
      "step": 2289
    },
    {
      "epoch": 0.1850542435200711,
      "grad_norm": 2.837385892868042,
      "learning_rate": 9.897874786016213e-06,
      "loss": 0.9642,
      "step": 2290
    },
    {
      "epoch": 0.18513505323339866,
      "grad_norm": 2.9858784675598145,
      "learning_rate": 9.897743165360487e-06,
      "loss": 1.1122,
      "step": 2291
    },
    {
      "epoch": 0.1852158629467262,
      "grad_norm": 2.650075912475586,
      "learning_rate": 9.89761146081821e-06,
      "loss": 1.142,
      "step": 2292
    },
    {
      "epoch": 0.18529667266005373,
      "grad_norm": 3.637911796569824,
      "learning_rate": 9.89747967239164e-06,
      "loss": 1.0368,
      "step": 2293
    },
    {
      "epoch": 0.18537748237338128,
      "grad_norm": 2.865354537963867,
      "learning_rate": 9.897347800083034e-06,
      "loss": 1.0131,
      "step": 2294
    },
    {
      "epoch": 0.18545829208670883,
      "grad_norm": 2.903250217437744,
      "learning_rate": 9.89721584389465e-06,
      "loss": 0.9644,
      "step": 2295
    },
    {
      "epoch": 0.18553910180003635,
      "grad_norm": 2.552626848220825,
      "learning_rate": 9.897083803828747e-06,
      "loss": 1.0422,
      "step": 2296
    },
    {
      "epoch": 0.1856199115133639,
      "grad_norm": 2.6443898677825928,
      "learning_rate": 9.896951679887588e-06,
      "loss": 1.0194,
      "step": 2297
    },
    {
      "epoch": 0.18570072122669146,
      "grad_norm": 2.6982264518737793,
      "learning_rate": 9.896819472073435e-06,
      "loss": 0.9968,
      "step": 2298
    },
    {
      "epoch": 0.18578153094001898,
      "grad_norm": 3.444505453109741,
      "learning_rate": 9.896687180388555e-06,
      "loss": 0.9242,
      "step": 2299
    },
    {
      "epoch": 0.18586234065334653,
      "grad_norm": 2.814103364944458,
      "learning_rate": 9.89655480483521e-06,
      "loss": 1.0171,
      "step": 2300
    },
    {
      "epoch": 0.18594315036667408,
      "grad_norm": 2.8541147708892822,
      "learning_rate": 9.896422345415671e-06,
      "loss": 1.121,
      "step": 2301
    },
    {
      "epoch": 0.1860239600800016,
      "grad_norm": 2.724806070327759,
      "learning_rate": 9.896289802132204e-06,
      "loss": 1.0631,
      "step": 2302
    },
    {
      "epoch": 0.18610476979332916,
      "grad_norm": 2.483346700668335,
      "learning_rate": 9.896157174987079e-06,
      "loss": 1.0567,
      "step": 2303
    },
    {
      "epoch": 0.1861855795066567,
      "grad_norm": 2.560619592666626,
      "learning_rate": 9.89602446398257e-06,
      "loss": 1.0899,
      "step": 2304
    },
    {
      "epoch": 0.18626638921998423,
      "grad_norm": 2.711629629135132,
      "learning_rate": 9.895891669120948e-06,
      "loss": 0.9556,
      "step": 2305
    },
    {
      "epoch": 0.18634719893331178,
      "grad_norm": 3.1584885120391846,
      "learning_rate": 9.895758790404488e-06,
      "loss": 1.0342,
      "step": 2306
    },
    {
      "epoch": 0.18642800864663933,
      "grad_norm": 2.9955644607543945,
      "learning_rate": 9.895625827835466e-06,
      "loss": 0.9627,
      "step": 2307
    },
    {
      "epoch": 0.18650881835996685,
      "grad_norm": 2.802232503890991,
      "learning_rate": 9.89549278141616e-06,
      "loss": 1.0367,
      "step": 2308
    },
    {
      "epoch": 0.1865896280732944,
      "grad_norm": 2.8121204376220703,
      "learning_rate": 9.895359651148848e-06,
      "loss": 1.0337,
      "step": 2309
    },
    {
      "epoch": 0.18667043778662196,
      "grad_norm": 2.481471538543701,
      "learning_rate": 9.895226437035808e-06,
      "loss": 0.9112,
      "step": 2310
    },
    {
      "epoch": 0.1867512474999495,
      "grad_norm": 2.9287850856781006,
      "learning_rate": 9.895093139079326e-06,
      "loss": 0.9825,
      "step": 2311
    },
    {
      "epoch": 0.18683205721327703,
      "grad_norm": 2.7301418781280518,
      "learning_rate": 9.894959757281684e-06,
      "loss": 1.0598,
      "step": 2312
    },
    {
      "epoch": 0.18691286692660458,
      "grad_norm": 2.687197685241699,
      "learning_rate": 9.894826291645163e-06,
      "loss": 0.9043,
      "step": 2313
    },
    {
      "epoch": 0.18699367663993213,
      "grad_norm": 2.7087533473968506,
      "learning_rate": 9.894692742172052e-06,
      "loss": 1.0148,
      "step": 2314
    },
    {
      "epoch": 0.18707448635325966,
      "grad_norm": 2.8972854614257812,
      "learning_rate": 9.89455910886464e-06,
      "loss": 1.0323,
      "step": 2315
    },
    {
      "epoch": 0.1871552960665872,
      "grad_norm": 2.723487138748169,
      "learning_rate": 9.894425391725211e-06,
      "loss": 0.9708,
      "step": 2316
    },
    {
      "epoch": 0.18723610577991476,
      "grad_norm": 2.984046220779419,
      "learning_rate": 9.89429159075606e-06,
      "loss": 1.0942,
      "step": 2317
    },
    {
      "epoch": 0.18731691549324228,
      "grad_norm": 2.903895854949951,
      "learning_rate": 9.894157705959474e-06,
      "loss": 1.0002,
      "step": 2318
    },
    {
      "epoch": 0.18739772520656983,
      "grad_norm": 3.094198703765869,
      "learning_rate": 9.89402373733775e-06,
      "loss": 0.9538,
      "step": 2319
    },
    {
      "epoch": 0.18747853491989738,
      "grad_norm": 2.7821521759033203,
      "learning_rate": 9.893889684893182e-06,
      "loss": 0.97,
      "step": 2320
    },
    {
      "epoch": 0.1875593446332249,
      "grad_norm": 3.355893135070801,
      "learning_rate": 9.893755548628065e-06,
      "loss": 0.9173,
      "step": 2321
    },
    {
      "epoch": 0.18764015434655246,
      "grad_norm": 2.6395037174224854,
      "learning_rate": 9.893621328544697e-06,
      "loss": 0.8974,
      "step": 2322
    },
    {
      "epoch": 0.18772096405988,
      "grad_norm": 2.7599050998687744,
      "learning_rate": 9.893487024645376e-06,
      "loss": 1.1345,
      "step": 2323
    },
    {
      "epoch": 0.18780177377320753,
      "grad_norm": 3.1787357330322266,
      "learning_rate": 9.893352636932403e-06,
      "loss": 1.05,
      "step": 2324
    },
    {
      "epoch": 0.18788258348653508,
      "grad_norm": 2.8817572593688965,
      "learning_rate": 9.89321816540808e-06,
      "loss": 0.9488,
      "step": 2325
    },
    {
      "epoch": 0.18796339319986263,
      "grad_norm": 2.432938814163208,
      "learning_rate": 9.893083610074708e-06,
      "loss": 0.9587,
      "step": 2326
    },
    {
      "epoch": 0.18804420291319016,
      "grad_norm": 2.720282793045044,
      "learning_rate": 9.892948970934595e-06,
      "loss": 1.1109,
      "step": 2327
    },
    {
      "epoch": 0.1881250126265177,
      "grad_norm": 2.891789436340332,
      "learning_rate": 9.892814247990045e-06,
      "loss": 0.976,
      "step": 2328
    },
    {
      "epoch": 0.18820582233984526,
      "grad_norm": 2.7646126747131348,
      "learning_rate": 9.892679441243367e-06,
      "loss": 1.0885,
      "step": 2329
    },
    {
      "epoch": 0.18828663205317278,
      "grad_norm": 2.6306145191192627,
      "learning_rate": 9.892544550696867e-06,
      "loss": 1.1034,
      "step": 2330
    },
    {
      "epoch": 0.18836744176650033,
      "grad_norm": 2.8713583946228027,
      "learning_rate": 9.892409576352859e-06,
      "loss": 1.0409,
      "step": 2331
    },
    {
      "epoch": 0.18844825147982788,
      "grad_norm": 3.1042556762695312,
      "learning_rate": 9.892274518213652e-06,
      "loss": 1.0268,
      "step": 2332
    },
    {
      "epoch": 0.1885290611931554,
      "grad_norm": 2.764892578125,
      "learning_rate": 9.892139376281559e-06,
      "loss": 0.9407,
      "step": 2333
    },
    {
      "epoch": 0.18860987090648296,
      "grad_norm": 2.5440847873687744,
      "learning_rate": 9.892004150558897e-06,
      "loss": 1.0254,
      "step": 2334
    },
    {
      "epoch": 0.1886906806198105,
      "grad_norm": 2.648404121398926,
      "learning_rate": 9.89186884104798e-06,
      "loss": 1.0697,
      "step": 2335
    },
    {
      "epoch": 0.18877149033313803,
      "grad_norm": 2.939295530319214,
      "learning_rate": 9.891733447751129e-06,
      "loss": 0.9549,
      "step": 2336
    },
    {
      "epoch": 0.18885230004646558,
      "grad_norm": 2.9315342903137207,
      "learning_rate": 9.891597970670657e-06,
      "loss": 0.8858,
      "step": 2337
    },
    {
      "epoch": 0.18893310975979313,
      "grad_norm": 2.5525410175323486,
      "learning_rate": 9.89146240980889e-06,
      "loss": 0.9254,
      "step": 2338
    },
    {
      "epoch": 0.18901391947312066,
      "grad_norm": 2.603762626647949,
      "learning_rate": 9.891326765168147e-06,
      "loss": 1.0947,
      "step": 2339
    },
    {
      "epoch": 0.1890947291864482,
      "grad_norm": 2.6096351146698,
      "learning_rate": 9.891191036750752e-06,
      "loss": 0.9084,
      "step": 2340
    },
    {
      "epoch": 0.18917553889977576,
      "grad_norm": 2.9336111545562744,
      "learning_rate": 9.89105522455903e-06,
      "loss": 0.9864,
      "step": 2341
    },
    {
      "epoch": 0.18925634861310328,
      "grad_norm": 2.792168140411377,
      "learning_rate": 9.890919328595306e-06,
      "loss": 0.9951,
      "step": 2342
    },
    {
      "epoch": 0.18933715832643083,
      "grad_norm": 2.763575792312622,
      "learning_rate": 9.890783348861909e-06,
      "loss": 1.0332,
      "step": 2343
    },
    {
      "epoch": 0.18941796803975838,
      "grad_norm": 2.647984027862549,
      "learning_rate": 9.890647285361166e-06,
      "loss": 1.0565,
      "step": 2344
    },
    {
      "epoch": 0.18949877775308593,
      "grad_norm": 2.5773236751556396,
      "learning_rate": 9.890511138095411e-06,
      "loss": 1.0024,
      "step": 2345
    },
    {
      "epoch": 0.18957958746641346,
      "grad_norm": 3.2868947982788086,
      "learning_rate": 9.890374907066971e-06,
      "loss": 1.0944,
      "step": 2346
    },
    {
      "epoch": 0.189660397179741,
      "grad_norm": 3.268430709838867,
      "learning_rate": 9.890238592278184e-06,
      "loss": 1.0724,
      "step": 2347
    },
    {
      "epoch": 0.18974120689306856,
      "grad_norm": 3.127488136291504,
      "learning_rate": 9.890102193731381e-06,
      "loss": 1.0143,
      "step": 2348
    },
    {
      "epoch": 0.18982201660639608,
      "grad_norm": 2.9193356037139893,
      "learning_rate": 9.889965711428901e-06,
      "loss": 0.9556,
      "step": 2349
    },
    {
      "epoch": 0.18990282631972363,
      "grad_norm": 3.0478930473327637,
      "learning_rate": 9.88982914537308e-06,
      "loss": 1.0065,
      "step": 2350
    },
    {
      "epoch": 0.18998363603305118,
      "grad_norm": 2.723362922668457,
      "learning_rate": 9.88969249556626e-06,
      "loss": 0.9885,
      "step": 2351
    },
    {
      "epoch": 0.1900644457463787,
      "grad_norm": 2.417820930480957,
      "learning_rate": 9.889555762010776e-06,
      "loss": 0.9351,
      "step": 2352
    },
    {
      "epoch": 0.19014525545970626,
      "grad_norm": 2.9843337535858154,
      "learning_rate": 9.889418944708973e-06,
      "loss": 0.999,
      "step": 2353
    },
    {
      "epoch": 0.1902260651730338,
      "grad_norm": 3.093397855758667,
      "learning_rate": 9.889282043663196e-06,
      "loss": 1.0169,
      "step": 2354
    },
    {
      "epoch": 0.19030687488636133,
      "grad_norm": 2.4472341537475586,
      "learning_rate": 9.889145058875786e-06,
      "loss": 0.8608,
      "step": 2355
    },
    {
      "epoch": 0.19038768459968888,
      "grad_norm": 3.399470806121826,
      "learning_rate": 9.889007990349093e-06,
      "loss": 1.1322,
      "step": 2356
    },
    {
      "epoch": 0.19046849431301643,
      "grad_norm": 2.760270833969116,
      "learning_rate": 9.888870838085463e-06,
      "loss": 1.0778,
      "step": 2357
    },
    {
      "epoch": 0.19054930402634396,
      "grad_norm": 2.8980255126953125,
      "learning_rate": 9.888733602087244e-06,
      "loss": 1.0017,
      "step": 2358
    },
    {
      "epoch": 0.1906301137396715,
      "grad_norm": 3.068690776824951,
      "learning_rate": 9.888596282356788e-06,
      "loss": 1.057,
      "step": 2359
    },
    {
      "epoch": 0.19071092345299906,
      "grad_norm": 2.7030603885650635,
      "learning_rate": 9.888458878896445e-06,
      "loss": 1.0547,
      "step": 2360
    },
    {
      "epoch": 0.19079173316632658,
      "grad_norm": 2.6108741760253906,
      "learning_rate": 9.888321391708571e-06,
      "loss": 0.9859,
      "step": 2361
    },
    {
      "epoch": 0.19087254287965413,
      "grad_norm": 2.7236034870147705,
      "learning_rate": 9.88818382079552e-06,
      "loss": 1.0942,
      "step": 2362
    },
    {
      "epoch": 0.19095335259298168,
      "grad_norm": 3.056166410446167,
      "learning_rate": 9.888046166159648e-06,
      "loss": 1.0482,
      "step": 2363
    },
    {
      "epoch": 0.1910341623063092,
      "grad_norm": 3.031191349029541,
      "learning_rate": 9.887908427803313e-06,
      "loss": 1.0582,
      "step": 2364
    },
    {
      "epoch": 0.19111497201963676,
      "grad_norm": 2.9865663051605225,
      "learning_rate": 9.887770605728873e-06,
      "loss": 1.1339,
      "step": 2365
    },
    {
      "epoch": 0.1911957817329643,
      "grad_norm": 2.5968196392059326,
      "learning_rate": 9.88763269993869e-06,
      "loss": 1.0259,
      "step": 2366
    },
    {
      "epoch": 0.19127659144629183,
      "grad_norm": 3.1886796951293945,
      "learning_rate": 9.887494710435125e-06,
      "loss": 0.9362,
      "step": 2367
    },
    {
      "epoch": 0.19135740115961938,
      "grad_norm": 3.991478443145752,
      "learning_rate": 9.887356637220543e-06,
      "loss": 0.9187,
      "step": 2368
    },
    {
      "epoch": 0.19143821087294693,
      "grad_norm": 2.774512767791748,
      "learning_rate": 9.887218480297305e-06,
      "loss": 0.9443,
      "step": 2369
    },
    {
      "epoch": 0.19151902058627446,
      "grad_norm": 3.140108108520508,
      "learning_rate": 9.887080239667782e-06,
      "loss": 0.9853,
      "step": 2370
    },
    {
      "epoch": 0.191599830299602,
      "grad_norm": 2.8573155403137207,
      "learning_rate": 9.886941915334339e-06,
      "loss": 0.9862,
      "step": 2371
    },
    {
      "epoch": 0.19168064001292956,
      "grad_norm": 3.353161334991455,
      "learning_rate": 9.886803507299347e-06,
      "loss": 1.0625,
      "step": 2372
    },
    {
      "epoch": 0.19176144972625708,
      "grad_norm": 3.1049435138702393,
      "learning_rate": 9.886665015565173e-06,
      "loss": 0.9729,
      "step": 2373
    },
    {
      "epoch": 0.19184225943958463,
      "grad_norm": 2.4401090145111084,
      "learning_rate": 9.886526440134195e-06,
      "loss": 0.9366,
      "step": 2374
    },
    {
      "epoch": 0.19192306915291218,
      "grad_norm": 2.4159469604492188,
      "learning_rate": 9.886387781008779e-06,
      "loss": 0.8985,
      "step": 2375
    },
    {
      "epoch": 0.19200387886623974,
      "grad_norm": 2.9945261478424072,
      "learning_rate": 9.886249038191305e-06,
      "loss": 1.0135,
      "step": 2376
    },
    {
      "epoch": 0.19208468857956726,
      "grad_norm": 2.855586528778076,
      "learning_rate": 9.88611021168415e-06,
      "loss": 1.0542,
      "step": 2377
    },
    {
      "epoch": 0.1921654982928948,
      "grad_norm": 3.1779487133026123,
      "learning_rate": 9.885971301489687e-06,
      "loss": 0.8935,
      "step": 2378
    },
    {
      "epoch": 0.19224630800622236,
      "grad_norm": 2.762692451477051,
      "learning_rate": 9.8858323076103e-06,
      "loss": 1.0176,
      "step": 2379
    },
    {
      "epoch": 0.19232711771954988,
      "grad_norm": 3.1234843730926514,
      "learning_rate": 9.885693230048368e-06,
      "loss": 0.9998,
      "step": 2380
    },
    {
      "epoch": 0.19240792743287743,
      "grad_norm": 3.9722297191619873,
      "learning_rate": 9.885554068806272e-06,
      "loss": 0.9857,
      "step": 2381
    },
    {
      "epoch": 0.19248873714620499,
      "grad_norm": 2.6323139667510986,
      "learning_rate": 9.885414823886397e-06,
      "loss": 0.9864,
      "step": 2382
    },
    {
      "epoch": 0.1925695468595325,
      "grad_norm": 3.1694223880767822,
      "learning_rate": 9.885275495291127e-06,
      "loss": 0.9757,
      "step": 2383
    },
    {
      "epoch": 0.19265035657286006,
      "grad_norm": 2.5772705078125,
      "learning_rate": 9.885136083022847e-06,
      "loss": 1.0262,
      "step": 2384
    },
    {
      "epoch": 0.1927311662861876,
      "grad_norm": 2.823486566543579,
      "learning_rate": 9.884996587083948e-06,
      "loss": 1.0096,
      "step": 2385
    },
    {
      "epoch": 0.19281197599951513,
      "grad_norm": 2.7001473903656006,
      "learning_rate": 9.884857007476817e-06,
      "loss": 0.9953,
      "step": 2386
    },
    {
      "epoch": 0.19289278571284268,
      "grad_norm": 2.8641977310180664,
      "learning_rate": 9.884717344203846e-06,
      "loss": 1.0912,
      "step": 2387
    },
    {
      "epoch": 0.19297359542617024,
      "grad_norm": 3.891497850418091,
      "learning_rate": 9.884577597267426e-06,
      "loss": 0.9944,
      "step": 2388
    },
    {
      "epoch": 0.19305440513949776,
      "grad_norm": 2.898452043533325,
      "learning_rate": 9.88443776666995e-06,
      "loss": 0.9517,
      "step": 2389
    },
    {
      "epoch": 0.1931352148528253,
      "grad_norm": 3.1403660774230957,
      "learning_rate": 9.884297852413815e-06,
      "loss": 1.0849,
      "step": 2390
    },
    {
      "epoch": 0.19321602456615286,
      "grad_norm": 2.761983871459961,
      "learning_rate": 9.884157854501416e-06,
      "loss": 1.0986,
      "step": 2391
    },
    {
      "epoch": 0.19329683427948038,
      "grad_norm": 3.095810651779175,
      "learning_rate": 9.884017772935151e-06,
      "loss": 0.9498,
      "step": 2392
    },
    {
      "epoch": 0.19337764399280793,
      "grad_norm": 2.5449116230010986,
      "learning_rate": 9.883877607717421e-06,
      "loss": 0.998,
      "step": 2393
    },
    {
      "epoch": 0.19345845370613549,
      "grad_norm": 2.822758436203003,
      "learning_rate": 9.883737358850622e-06,
      "loss": 0.9593,
      "step": 2394
    },
    {
      "epoch": 0.193539263419463,
      "grad_norm": 2.54616379737854,
      "learning_rate": 9.883597026337161e-06,
      "loss": 1.0445,
      "step": 2395
    },
    {
      "epoch": 0.19362007313279056,
      "grad_norm": 2.7621335983276367,
      "learning_rate": 9.883456610179437e-06,
      "loss": 1.0078,
      "step": 2396
    },
    {
      "epoch": 0.1937008828461181,
      "grad_norm": 2.8157503604888916,
      "learning_rate": 9.883316110379861e-06,
      "loss": 0.9763,
      "step": 2397
    },
    {
      "epoch": 0.19378169255944563,
      "grad_norm": 3.068323850631714,
      "learning_rate": 9.883175526940835e-06,
      "loss": 0.9221,
      "step": 2398
    },
    {
      "epoch": 0.19386250227277319,
      "grad_norm": 3.1782143115997314,
      "learning_rate": 9.883034859864768e-06,
      "loss": 0.9957,
      "step": 2399
    },
    {
      "epoch": 0.19394331198610074,
      "grad_norm": 2.9651782512664795,
      "learning_rate": 9.882894109154071e-06,
      "loss": 0.9781,
      "step": 2400
    },
    {
      "epoch": 0.19402412169942826,
      "grad_norm": 3.156034469604492,
      "learning_rate": 9.88275327481115e-06,
      "loss": 1.0522,
      "step": 2401
    },
    {
      "epoch": 0.1941049314127558,
      "grad_norm": 2.7413623332977295,
      "learning_rate": 9.882612356838422e-06,
      "loss": 1.0363,
      "step": 2402
    },
    {
      "epoch": 0.19418574112608336,
      "grad_norm": 2.7400834560394287,
      "learning_rate": 9.8824713552383e-06,
      "loss": 1.0021,
      "step": 2403
    },
    {
      "epoch": 0.19426655083941088,
      "grad_norm": 2.8538806438446045,
      "learning_rate": 9.882330270013194e-06,
      "loss": 1.0255,
      "step": 2404
    },
    {
      "epoch": 0.19434736055273844,
      "grad_norm": 3.811596393585205,
      "learning_rate": 9.882189101165527e-06,
      "loss": 1.0438,
      "step": 2405
    },
    {
      "epoch": 0.19442817026606599,
      "grad_norm": 3.5310895442962646,
      "learning_rate": 9.882047848697714e-06,
      "loss": 1.0074,
      "step": 2406
    },
    {
      "epoch": 0.1945089799793935,
      "grad_norm": 3.50319504737854,
      "learning_rate": 9.881906512612172e-06,
      "loss": 1.13,
      "step": 2407
    },
    {
      "epoch": 0.19458978969272106,
      "grad_norm": 2.4688737392425537,
      "learning_rate": 9.881765092911327e-06,
      "loss": 0.9055,
      "step": 2408
    },
    {
      "epoch": 0.1946705994060486,
      "grad_norm": 2.4967446327209473,
      "learning_rate": 9.881623589597596e-06,
      "loss": 0.8877,
      "step": 2409
    },
    {
      "epoch": 0.19475140911937616,
      "grad_norm": 2.6067020893096924,
      "learning_rate": 9.881482002673406e-06,
      "loss": 1.0201,
      "step": 2410
    },
    {
      "epoch": 0.19483221883270369,
      "grad_norm": 2.8096611499786377,
      "learning_rate": 9.881340332141183e-06,
      "loss": 1.0239,
      "step": 2411
    },
    {
      "epoch": 0.19491302854603124,
      "grad_norm": 2.951375722885132,
      "learning_rate": 9.881198578003348e-06,
      "loss": 0.9808,
      "step": 2412
    },
    {
      "epoch": 0.1949938382593588,
      "grad_norm": 3.2135069370269775,
      "learning_rate": 9.881056740262334e-06,
      "loss": 1.0019,
      "step": 2413
    },
    {
      "epoch": 0.1950746479726863,
      "grad_norm": 2.837573528289795,
      "learning_rate": 9.880914818920568e-06,
      "loss": 0.9657,
      "step": 2414
    },
    {
      "epoch": 0.19515545768601386,
      "grad_norm": 2.8844478130340576,
      "learning_rate": 9.880772813980484e-06,
      "loss": 1.0598,
      "step": 2415
    },
    {
      "epoch": 0.1952362673993414,
      "grad_norm": 2.8207285404205322,
      "learning_rate": 9.880630725444509e-06,
      "loss": 0.9335,
      "step": 2416
    },
    {
      "epoch": 0.19531707711266894,
      "grad_norm": 2.8908140659332275,
      "learning_rate": 9.88048855331508e-06,
      "loss": 0.9494,
      "step": 2417
    },
    {
      "epoch": 0.1953978868259965,
      "grad_norm": 2.654698610305786,
      "learning_rate": 9.880346297594631e-06,
      "loss": 1.0921,
      "step": 2418
    },
    {
      "epoch": 0.19547869653932404,
      "grad_norm": 2.965447187423706,
      "learning_rate": 9.8802039582856e-06,
      "loss": 0.9454,
      "step": 2419
    },
    {
      "epoch": 0.19555950625265156,
      "grad_norm": 2.5932366847991943,
      "learning_rate": 9.880061535390424e-06,
      "loss": 1.0111,
      "step": 2420
    },
    {
      "epoch": 0.1956403159659791,
      "grad_norm": 2.62595796585083,
      "learning_rate": 9.87991902891154e-06,
      "loss": 0.9679,
      "step": 2421
    },
    {
      "epoch": 0.19572112567930666,
      "grad_norm": 2.9708480834960938,
      "learning_rate": 9.879776438851393e-06,
      "loss": 1.071,
      "step": 2422
    },
    {
      "epoch": 0.19580193539263419,
      "grad_norm": 2.690122604370117,
      "learning_rate": 9.879633765212422e-06,
      "loss": 0.9928,
      "step": 2423
    },
    {
      "epoch": 0.19588274510596174,
      "grad_norm": 3.028838872909546,
      "learning_rate": 9.879491007997073e-06,
      "loss": 0.9819,
      "step": 2424
    },
    {
      "epoch": 0.1959635548192893,
      "grad_norm": 2.915705919265747,
      "learning_rate": 9.87934816720779e-06,
      "loss": 1.0966,
      "step": 2425
    },
    {
      "epoch": 0.1960443645326168,
      "grad_norm": 2.9188296794891357,
      "learning_rate": 9.879205242847018e-06,
      "loss": 1.1146,
      "step": 2426
    },
    {
      "epoch": 0.19612517424594436,
      "grad_norm": 2.9386608600616455,
      "learning_rate": 9.879062234917208e-06,
      "loss": 0.864,
      "step": 2427
    },
    {
      "epoch": 0.1962059839592719,
      "grad_norm": 2.9816863536834717,
      "learning_rate": 9.878919143420806e-06,
      "loss": 0.9519,
      "step": 2428
    },
    {
      "epoch": 0.19628679367259944,
      "grad_norm": 2.8826448917388916,
      "learning_rate": 9.878775968360265e-06,
      "loss": 1.0976,
      "step": 2429
    },
    {
      "epoch": 0.196367603385927,
      "grad_norm": 3.5601329803466797,
      "learning_rate": 9.878632709738036e-06,
      "loss": 1.0267,
      "step": 2430
    },
    {
      "epoch": 0.19644841309925454,
      "grad_norm": 3.157382011413574,
      "learning_rate": 9.878489367556576e-06,
      "loss": 0.9642,
      "step": 2431
    },
    {
      "epoch": 0.19652922281258206,
      "grad_norm": 3.0705273151397705,
      "learning_rate": 9.878345941818338e-06,
      "loss": 0.9957,
      "step": 2432
    },
    {
      "epoch": 0.1966100325259096,
      "grad_norm": 2.779900074005127,
      "learning_rate": 9.878202432525774e-06,
      "loss": 0.9627,
      "step": 2433
    },
    {
      "epoch": 0.19669084223923716,
      "grad_norm": 3.0926270484924316,
      "learning_rate": 9.87805883968135e-06,
      "loss": 0.9306,
      "step": 2434
    },
    {
      "epoch": 0.19677165195256469,
      "grad_norm": 2.7956552505493164,
      "learning_rate": 9.877915163287519e-06,
      "loss": 0.9934,
      "step": 2435
    },
    {
      "epoch": 0.19685246166589224,
      "grad_norm": 3.262789011001587,
      "learning_rate": 9.877771403346747e-06,
      "loss": 0.951,
      "step": 2436
    },
    {
      "epoch": 0.1969332713792198,
      "grad_norm": 2.491626501083374,
      "learning_rate": 9.87762755986149e-06,
      "loss": 1.0969,
      "step": 2437
    },
    {
      "epoch": 0.1970140810925473,
      "grad_norm": 3.3042078018188477,
      "learning_rate": 9.877483632834219e-06,
      "loss": 0.9696,
      "step": 2438
    },
    {
      "epoch": 0.19709489080587486,
      "grad_norm": 2.8220388889312744,
      "learning_rate": 9.877339622267394e-06,
      "loss": 0.9866,
      "step": 2439
    },
    {
      "epoch": 0.1971757005192024,
      "grad_norm": 2.9727158546447754,
      "learning_rate": 9.877195528163483e-06,
      "loss": 1.0571,
      "step": 2440
    },
    {
      "epoch": 0.19725651023252996,
      "grad_norm": 3.236030101776123,
      "learning_rate": 9.877051350524953e-06,
      "loss": 1.0211,
      "step": 2441
    },
    {
      "epoch": 0.1973373199458575,
      "grad_norm": 2.409712076187134,
      "learning_rate": 9.876907089354276e-06,
      "loss": 1.0282,
      "step": 2442
    },
    {
      "epoch": 0.19741812965918504,
      "grad_norm": 2.8064048290252686,
      "learning_rate": 9.876762744653921e-06,
      "loss": 0.9514,
      "step": 2443
    },
    {
      "epoch": 0.1974989393725126,
      "grad_norm": 3.2507739067077637,
      "learning_rate": 9.87661831642636e-06,
      "loss": 1.0725,
      "step": 2444
    },
    {
      "epoch": 0.1975797490858401,
      "grad_norm": 2.829425096511841,
      "learning_rate": 9.876473804674067e-06,
      "loss": 0.9631,
      "step": 2445
    },
    {
      "epoch": 0.19766055879916766,
      "grad_norm": 2.590425491333008,
      "learning_rate": 9.876329209399518e-06,
      "loss": 0.9723,
      "step": 2446
    },
    {
      "epoch": 0.1977413685124952,
      "grad_norm": 2.5433905124664307,
      "learning_rate": 9.876184530605189e-06,
      "loss": 1.0544,
      "step": 2447
    },
    {
      "epoch": 0.19782217822582274,
      "grad_norm": 2.9684863090515137,
      "learning_rate": 9.876039768293557e-06,
      "loss": 0.9622,
      "step": 2448
    },
    {
      "epoch": 0.1979029879391503,
      "grad_norm": 2.6978437900543213,
      "learning_rate": 9.875894922467101e-06,
      "loss": 0.9914,
      "step": 2449
    },
    {
      "epoch": 0.19798379765247784,
      "grad_norm": 2.679476499557495,
      "learning_rate": 9.875749993128306e-06,
      "loss": 1.0121,
      "step": 2450
    },
    {
      "epoch": 0.19806460736580536,
      "grad_norm": 2.7482845783233643,
      "learning_rate": 9.875604980279651e-06,
      "loss": 1.0252,
      "step": 2451
    },
    {
      "epoch": 0.1981454170791329,
      "grad_norm": 2.545858860015869,
      "learning_rate": 9.875459883923619e-06,
      "loss": 1.0051,
      "step": 2452
    },
    {
      "epoch": 0.19822622679246046,
      "grad_norm": 2.84562349319458,
      "learning_rate": 9.875314704062697e-06,
      "loss": 1.0222,
      "step": 2453
    },
    {
      "epoch": 0.198307036505788,
      "grad_norm": 3.056476354598999,
      "learning_rate": 9.875169440699372e-06,
      "loss": 1.0069,
      "step": 2454
    },
    {
      "epoch": 0.19838784621911554,
      "grad_norm": 3.218858242034912,
      "learning_rate": 9.87502409383613e-06,
      "loss": 0.9731,
      "step": 2455
    },
    {
      "epoch": 0.1984686559324431,
      "grad_norm": 2.7342875003814697,
      "learning_rate": 9.874878663475462e-06,
      "loss": 1.0858,
      "step": 2456
    },
    {
      "epoch": 0.1985494656457706,
      "grad_norm": 2.640138626098633,
      "learning_rate": 9.874733149619857e-06,
      "loss": 1.086,
      "step": 2457
    },
    {
      "epoch": 0.19863027535909816,
      "grad_norm": 2.712902784347534,
      "learning_rate": 9.87458755227181e-06,
      "loss": 0.9231,
      "step": 2458
    },
    {
      "epoch": 0.1987110850724257,
      "grad_norm": 2.5488665103912354,
      "learning_rate": 9.874441871433814e-06,
      "loss": 0.9176,
      "step": 2459
    },
    {
      "epoch": 0.19879189478575324,
      "grad_norm": 2.6249120235443115,
      "learning_rate": 9.874296107108362e-06,
      "loss": 0.9626,
      "step": 2460
    },
    {
      "epoch": 0.1988727044990808,
      "grad_norm": 2.878230333328247,
      "learning_rate": 9.874150259297952e-06,
      "loss": 0.8884,
      "step": 2461
    },
    {
      "epoch": 0.19895351421240834,
      "grad_norm": 2.5039255619049072,
      "learning_rate": 9.87400432800508e-06,
      "loss": 1.0489,
      "step": 2462
    },
    {
      "epoch": 0.19903432392573586,
      "grad_norm": 2.8095619678497314,
      "learning_rate": 9.87385831323225e-06,
      "loss": 1.084,
      "step": 2463
    },
    {
      "epoch": 0.1991151336390634,
      "grad_norm": 2.7729170322418213,
      "learning_rate": 9.87371221498196e-06,
      "loss": 0.9587,
      "step": 2464
    },
    {
      "epoch": 0.19919594335239096,
      "grad_norm": 2.798548936843872,
      "learning_rate": 9.873566033256714e-06,
      "loss": 1.0125,
      "step": 2465
    },
    {
      "epoch": 0.1992767530657185,
      "grad_norm": 2.673734188079834,
      "learning_rate": 9.873419768059014e-06,
      "loss": 1.0097,
      "step": 2466
    },
    {
      "epoch": 0.19935756277904604,
      "grad_norm": 2.5640718936920166,
      "learning_rate": 9.873273419391364e-06,
      "loss": 1.0343,
      "step": 2467
    },
    {
      "epoch": 0.1994383724923736,
      "grad_norm": 2.601259469985962,
      "learning_rate": 9.873126987256273e-06,
      "loss": 0.9153,
      "step": 2468
    },
    {
      "epoch": 0.1995191822057011,
      "grad_norm": 2.698262929916382,
      "learning_rate": 9.87298047165625e-06,
      "loss": 1.0388,
      "step": 2469
    },
    {
      "epoch": 0.19959999191902866,
      "grad_norm": 2.803069591522217,
      "learning_rate": 9.872833872593801e-06,
      "loss": 0.9826,
      "step": 2470
    },
    {
      "epoch": 0.19968080163235621,
      "grad_norm": 2.310145616531372,
      "learning_rate": 9.87268719007144e-06,
      "loss": 1.0419,
      "step": 2471
    },
    {
      "epoch": 0.19976161134568376,
      "grad_norm": 2.7354466915130615,
      "learning_rate": 9.872540424091677e-06,
      "loss": 1.0474,
      "step": 2472
    },
    {
      "epoch": 0.1998424210590113,
      "grad_norm": 2.915072202682495,
      "learning_rate": 9.872393574657026e-06,
      "loss": 1.0912,
      "step": 2473
    },
    {
      "epoch": 0.19992323077233884,
      "grad_norm": 2.9000625610351562,
      "learning_rate": 9.872246641770004e-06,
      "loss": 1.1786,
      "step": 2474
    },
    {
      "epoch": 0.2000040404856664,
      "grad_norm": 2.6486880779266357,
      "learning_rate": 9.872099625433127e-06,
      "loss": 0.9718,
      "step": 2475
    },
    {
      "epoch": 0.2000848501989939,
      "grad_norm": 2.8417553901672363,
      "learning_rate": 9.871952525648911e-06,
      "loss": 1.14,
      "step": 2476
    },
    {
      "epoch": 0.20016565991232146,
      "grad_norm": 3.3232791423797607,
      "learning_rate": 9.871805342419879e-06,
      "loss": 1.0333,
      "step": 2477
    },
    {
      "epoch": 0.20024646962564902,
      "grad_norm": 2.7784183025360107,
      "learning_rate": 9.871658075748546e-06,
      "loss": 0.9657,
      "step": 2478
    },
    {
      "epoch": 0.20032727933897654,
      "grad_norm": 2.562795639038086,
      "learning_rate": 9.871510725637442e-06,
      "loss": 1.0903,
      "step": 2479
    },
    {
      "epoch": 0.2004080890523041,
      "grad_norm": 3.0607187747955322,
      "learning_rate": 9.871363292089085e-06,
      "loss": 0.9937,
      "step": 2480
    },
    {
      "epoch": 0.20048889876563164,
      "grad_norm": 2.629852533340454,
      "learning_rate": 9.871215775106003e-06,
      "loss": 0.9726,
      "step": 2481
    },
    {
      "epoch": 0.20056970847895916,
      "grad_norm": 2.361461639404297,
      "learning_rate": 9.871068174690722e-06,
      "loss": 1.0028,
      "step": 2482
    },
    {
      "epoch": 0.20065051819228671,
      "grad_norm": 2.896855592727661,
      "learning_rate": 9.87092049084577e-06,
      "loss": 0.9409,
      "step": 2483
    },
    {
      "epoch": 0.20073132790561427,
      "grad_norm": 2.9578440189361572,
      "learning_rate": 9.870772723573674e-06,
      "loss": 1.018,
      "step": 2484
    },
    {
      "epoch": 0.2008121376189418,
      "grad_norm": 2.7431399822235107,
      "learning_rate": 9.87062487287697e-06,
      "loss": 0.9153,
      "step": 2485
    },
    {
      "epoch": 0.20089294733226934,
      "grad_norm": 2.8610525131225586,
      "learning_rate": 9.870476938758185e-06,
      "loss": 1.0967,
      "step": 2486
    },
    {
      "epoch": 0.2009737570455969,
      "grad_norm": 2.564570665359497,
      "learning_rate": 9.870328921219856e-06,
      "loss": 0.9596,
      "step": 2487
    },
    {
      "epoch": 0.2010545667589244,
      "grad_norm": 3.0416998863220215,
      "learning_rate": 9.870180820264518e-06,
      "loss": 1.1215,
      "step": 2488
    },
    {
      "epoch": 0.20113537647225196,
      "grad_norm": 2.994410276412964,
      "learning_rate": 9.870032635894708e-06,
      "loss": 0.952,
      "step": 2489
    },
    {
      "epoch": 0.20121618618557952,
      "grad_norm": 3.3901889324188232,
      "learning_rate": 9.869884368112961e-06,
      "loss": 1.0788,
      "step": 2490
    },
    {
      "epoch": 0.20129699589890704,
      "grad_norm": 2.755059242248535,
      "learning_rate": 9.86973601692182e-06,
      "loss": 0.9808,
      "step": 2491
    },
    {
      "epoch": 0.2013778056122346,
      "grad_norm": 2.62873911857605,
      "learning_rate": 9.869587582323824e-06,
      "loss": 0.9143,
      "step": 2492
    },
    {
      "epoch": 0.20145861532556214,
      "grad_norm": 2.8190078735351562,
      "learning_rate": 9.869439064321516e-06,
      "loss": 1.0799,
      "step": 2493
    },
    {
      "epoch": 0.20153942503888966,
      "grad_norm": 2.5255749225616455,
      "learning_rate": 9.86929046291744e-06,
      "loss": 1.0696,
      "step": 2494
    },
    {
      "epoch": 0.20162023475221721,
      "grad_norm": 2.4826602935791016,
      "learning_rate": 9.86914177811414e-06,
      "loss": 1.0309,
      "step": 2495
    },
    {
      "epoch": 0.20170104446554477,
      "grad_norm": 2.7080788612365723,
      "learning_rate": 9.868993009914162e-06,
      "loss": 1.0622,
      "step": 2496
    },
    {
      "epoch": 0.2017818541788723,
      "grad_norm": 2.994687557220459,
      "learning_rate": 9.868844158320056e-06,
      "loss": 1.0052,
      "step": 2497
    },
    {
      "epoch": 0.20186266389219984,
      "grad_norm": 2.9315123558044434,
      "learning_rate": 9.868695223334372e-06,
      "loss": 0.9804,
      "step": 2498
    },
    {
      "epoch": 0.2019434736055274,
      "grad_norm": 2.8823635578155518,
      "learning_rate": 9.868546204959659e-06,
      "loss": 0.9873,
      "step": 2499
    },
    {
      "epoch": 0.2020242833188549,
      "grad_norm": 2.6396167278289795,
      "learning_rate": 9.868397103198471e-06,
      "loss": 0.9844,
      "step": 2500
    },
    {
      "epoch": 0.20210509303218246,
      "grad_norm": 2.803288221359253,
      "learning_rate": 9.86824791805336e-06,
      "loss": 0.8829,
      "step": 2501
    },
    {
      "epoch": 0.20218590274551002,
      "grad_norm": 2.912961483001709,
      "learning_rate": 9.86809864952688e-06,
      "loss": 0.9903,
      "step": 2502
    },
    {
      "epoch": 0.20226671245883754,
      "grad_norm": 2.793362855911255,
      "learning_rate": 9.867949297621592e-06,
      "loss": 1.0127,
      "step": 2503
    },
    {
      "epoch": 0.2023475221721651,
      "grad_norm": 2.580033779144287,
      "learning_rate": 9.867799862340054e-06,
      "loss": 0.9637,
      "step": 2504
    },
    {
      "epoch": 0.20242833188549264,
      "grad_norm": 3.100097179412842,
      "learning_rate": 9.867650343684818e-06,
      "loss": 0.9721,
      "step": 2505
    },
    {
      "epoch": 0.2025091415988202,
      "grad_norm": 2.967200756072998,
      "learning_rate": 9.867500741658454e-06,
      "loss": 0.9854,
      "step": 2506
    },
    {
      "epoch": 0.20258995131214771,
      "grad_norm": 2.929659843444824,
      "learning_rate": 9.867351056263517e-06,
      "loss": 0.9923,
      "step": 2507
    },
    {
      "epoch": 0.20267076102547527,
      "grad_norm": 2.7155697345733643,
      "learning_rate": 9.867201287502576e-06,
      "loss": 1.0235,
      "step": 2508
    },
    {
      "epoch": 0.20275157073880282,
      "grad_norm": 3.458383321762085,
      "learning_rate": 9.867051435378194e-06,
      "loss": 0.9887,
      "step": 2509
    },
    {
      "epoch": 0.20283238045213034,
      "grad_norm": 2.467725992202759,
      "learning_rate": 9.866901499892938e-06,
      "loss": 0.9697,
      "step": 2510
    },
    {
      "epoch": 0.2029131901654579,
      "grad_norm": 2.7580583095550537,
      "learning_rate": 9.866751481049377e-06,
      "loss": 1.0378,
      "step": 2511
    },
    {
      "epoch": 0.20299399987878544,
      "grad_norm": 3.0809414386749268,
      "learning_rate": 9.866601378850077e-06,
      "loss": 0.9444,
      "step": 2512
    },
    {
      "epoch": 0.20307480959211296,
      "grad_norm": 2.972541093826294,
      "learning_rate": 9.866451193297613e-06,
      "loss": 0.9478,
      "step": 2513
    },
    {
      "epoch": 0.20315561930544052,
      "grad_norm": 2.998645067214966,
      "learning_rate": 9.866300924394556e-06,
      "loss": 1.0539,
      "step": 2514
    },
    {
      "epoch": 0.20323642901876807,
      "grad_norm": 3.317265033721924,
      "learning_rate": 9.866150572143477e-06,
      "loss": 0.9376,
      "step": 2515
    },
    {
      "epoch": 0.2033172387320956,
      "grad_norm": 3.172424554824829,
      "learning_rate": 9.866000136546954e-06,
      "loss": 0.8793,
      "step": 2516
    },
    {
      "epoch": 0.20339804844542314,
      "grad_norm": 2.846748113632202,
      "learning_rate": 9.865849617607565e-06,
      "loss": 1.0539,
      "step": 2517
    },
    {
      "epoch": 0.2034788581587507,
      "grad_norm": 2.9173882007598877,
      "learning_rate": 9.865699015327885e-06,
      "loss": 1.0106,
      "step": 2518
    },
    {
      "epoch": 0.20355966787207821,
      "grad_norm": 2.671884059906006,
      "learning_rate": 9.865548329710496e-06,
      "loss": 1.1,
      "step": 2519
    },
    {
      "epoch": 0.20364047758540577,
      "grad_norm": 2.9467697143554688,
      "learning_rate": 9.865397560757975e-06,
      "loss": 1.0092,
      "step": 2520
    },
    {
      "epoch": 0.20372128729873332,
      "grad_norm": 2.752096176147461,
      "learning_rate": 9.865246708472907e-06,
      "loss": 0.9468,
      "step": 2521
    },
    {
      "epoch": 0.20380209701206084,
      "grad_norm": 3.0249135494232178,
      "learning_rate": 9.865095772857875e-06,
      "loss": 0.949,
      "step": 2522
    },
    {
      "epoch": 0.2038829067253884,
      "grad_norm": 3.1122334003448486,
      "learning_rate": 9.864944753915466e-06,
      "loss": 0.969,
      "step": 2523
    },
    {
      "epoch": 0.20396371643871594,
      "grad_norm": 2.7139110565185547,
      "learning_rate": 9.864793651648266e-06,
      "loss": 0.9581,
      "step": 2524
    },
    {
      "epoch": 0.20404452615204346,
      "grad_norm": 2.9858901500701904,
      "learning_rate": 9.864642466058861e-06,
      "loss": 0.9883,
      "step": 2525
    },
    {
      "epoch": 0.20412533586537102,
      "grad_norm": 2.9757955074310303,
      "learning_rate": 9.864491197149841e-06,
      "loss": 1.1164,
      "step": 2526
    },
    {
      "epoch": 0.20420614557869857,
      "grad_norm": 3.0603973865509033,
      "learning_rate": 9.864339844923801e-06,
      "loss": 0.9447,
      "step": 2527
    },
    {
      "epoch": 0.2042869552920261,
      "grad_norm": 2.7169153690338135,
      "learning_rate": 9.864188409383326e-06,
      "loss": 0.9277,
      "step": 2528
    },
    {
      "epoch": 0.20436776500535364,
      "grad_norm": 3.0862722396850586,
      "learning_rate": 9.864036890531014e-06,
      "loss": 0.9635,
      "step": 2529
    },
    {
      "epoch": 0.2044485747186812,
      "grad_norm": 2.611670970916748,
      "learning_rate": 9.863885288369461e-06,
      "loss": 0.9959,
      "step": 2530
    },
    {
      "epoch": 0.20452938443200872,
      "grad_norm": 2.7444217205047607,
      "learning_rate": 9.863733602901262e-06,
      "loss": 0.9811,
      "step": 2531
    },
    {
      "epoch": 0.20461019414533627,
      "grad_norm": 2.7754032611846924,
      "learning_rate": 9.863581834129017e-06,
      "loss": 0.9322,
      "step": 2532
    },
    {
      "epoch": 0.20469100385866382,
      "grad_norm": 3.2942535877227783,
      "learning_rate": 9.863429982055322e-06,
      "loss": 1.0804,
      "step": 2533
    },
    {
      "epoch": 0.20477181357199134,
      "grad_norm": 2.810506820678711,
      "learning_rate": 9.86327804668278e-06,
      "loss": 1.0103,
      "step": 2534
    },
    {
      "epoch": 0.2048526232853189,
      "grad_norm": 3.491856575012207,
      "learning_rate": 9.863126028013993e-06,
      "loss": 0.9455,
      "step": 2535
    },
    {
      "epoch": 0.20493343299864644,
      "grad_norm": 2.792609214782715,
      "learning_rate": 9.862973926051565e-06,
      "loss": 0.959,
      "step": 2536
    },
    {
      "epoch": 0.205014242711974,
      "grad_norm": 2.7215230464935303,
      "learning_rate": 9.8628217407981e-06,
      "loss": 0.9344,
      "step": 2537
    },
    {
      "epoch": 0.20509505242530152,
      "grad_norm": 2.9850406646728516,
      "learning_rate": 9.862669472256206e-06,
      "loss": 1.0273,
      "step": 2538
    },
    {
      "epoch": 0.20517586213862907,
      "grad_norm": 2.7064993381500244,
      "learning_rate": 9.86251712042849e-06,
      "loss": 0.9229,
      "step": 2539
    },
    {
      "epoch": 0.20525667185195662,
      "grad_norm": 2.823521375656128,
      "learning_rate": 9.86236468531756e-06,
      "loss": 1.105,
      "step": 2540
    },
    {
      "epoch": 0.20533748156528414,
      "grad_norm": 3.5803730487823486,
      "learning_rate": 9.862212166926031e-06,
      "loss": 0.9772,
      "step": 2541
    },
    {
      "epoch": 0.2054182912786117,
      "grad_norm": 2.4767980575561523,
      "learning_rate": 9.862059565256512e-06,
      "loss": 1.0041,
      "step": 2542
    },
    {
      "epoch": 0.20549910099193924,
      "grad_norm": 3.0558948516845703,
      "learning_rate": 9.861906880311617e-06,
      "loss": 0.996,
      "step": 2543
    },
    {
      "epoch": 0.20557991070526677,
      "grad_norm": 2.6973979473114014,
      "learning_rate": 9.861754112093964e-06,
      "loss": 1.0557,
      "step": 2544
    },
    {
      "epoch": 0.20566072041859432,
      "grad_norm": 3.0480377674102783,
      "learning_rate": 9.861601260606166e-06,
      "loss": 1.0027,
      "step": 2545
    },
    {
      "epoch": 0.20574153013192187,
      "grad_norm": 2.9968481063842773,
      "learning_rate": 9.861448325850842e-06,
      "loss": 1.0336,
      "step": 2546
    },
    {
      "epoch": 0.2058223398452494,
      "grad_norm": 2.68129825592041,
      "learning_rate": 9.861295307830612e-06,
      "loss": 1.0041,
      "step": 2547
    },
    {
      "epoch": 0.20590314955857694,
      "grad_norm": 4.269932270050049,
      "learning_rate": 9.861142206548096e-06,
      "loss": 1.1087,
      "step": 2548
    },
    {
      "epoch": 0.2059839592719045,
      "grad_norm": 2.564828634262085,
      "learning_rate": 9.860989022005915e-06,
      "loss": 1.023,
      "step": 2549
    },
    {
      "epoch": 0.20606476898523202,
      "grad_norm": 3.118055582046509,
      "learning_rate": 9.860835754206698e-06,
      "loss": 1.0049,
      "step": 2550
    },
    {
      "epoch": 0.20614557869855957,
      "grad_norm": 2.6097300052642822,
      "learning_rate": 9.860682403153064e-06,
      "loss": 1.0637,
      "step": 2551
    },
    {
      "epoch": 0.20622638841188712,
      "grad_norm": 2.9691739082336426,
      "learning_rate": 9.860528968847642e-06,
      "loss": 1.036,
      "step": 2552
    },
    {
      "epoch": 0.20630719812521464,
      "grad_norm": 3.2187507152557373,
      "learning_rate": 9.86037545129306e-06,
      "loss": 1.0471,
      "step": 2553
    },
    {
      "epoch": 0.2063880078385422,
      "grad_norm": 2.5262982845306396,
      "learning_rate": 9.860221850491949e-06,
      "loss": 1.1136,
      "step": 2554
    },
    {
      "epoch": 0.20646881755186974,
      "grad_norm": 2.7065389156341553,
      "learning_rate": 9.860068166446938e-06,
      "loss": 0.9593,
      "step": 2555
    },
    {
      "epoch": 0.20654962726519727,
      "grad_norm": 2.8696558475494385,
      "learning_rate": 9.85991439916066e-06,
      "loss": 1.0267,
      "step": 2556
    },
    {
      "epoch": 0.20663043697852482,
      "grad_norm": 2.757690906524658,
      "learning_rate": 9.859760548635746e-06,
      "loss": 0.95,
      "step": 2557
    },
    {
      "epoch": 0.20671124669185237,
      "grad_norm": 2.6395418643951416,
      "learning_rate": 9.859606614874834e-06,
      "loss": 0.947,
      "step": 2558
    },
    {
      "epoch": 0.2067920564051799,
      "grad_norm": 2.605659008026123,
      "learning_rate": 9.859452597880559e-06,
      "loss": 0.8962,
      "step": 2559
    },
    {
      "epoch": 0.20687286611850744,
      "grad_norm": 2.7953250408172607,
      "learning_rate": 9.85929849765556e-06,
      "loss": 1.0655,
      "step": 2560
    },
    {
      "epoch": 0.206953675831835,
      "grad_norm": 2.9792885780334473,
      "learning_rate": 9.859144314202478e-06,
      "loss": 1.0137,
      "step": 2561
    },
    {
      "epoch": 0.20703448554516252,
      "grad_norm": 3.3989391326904297,
      "learning_rate": 9.85899004752395e-06,
      "loss": 0.8957,
      "step": 2562
    },
    {
      "epoch": 0.20711529525849007,
      "grad_norm": 2.6138968467712402,
      "learning_rate": 9.858835697622619e-06,
      "loss": 1.1482,
      "step": 2563
    },
    {
      "epoch": 0.20719610497181762,
      "grad_norm": 2.423694133758545,
      "learning_rate": 9.858681264501133e-06,
      "loss": 1.0465,
      "step": 2564
    },
    {
      "epoch": 0.20727691468514514,
      "grad_norm": 3.0095713138580322,
      "learning_rate": 9.858526748162132e-06,
      "loss": 0.9462,
      "step": 2565
    },
    {
      "epoch": 0.2073577243984727,
      "grad_norm": 3.7489736080169678,
      "learning_rate": 9.858372148608263e-06,
      "loss": 1.0188,
      "step": 2566
    },
    {
      "epoch": 0.20743853411180024,
      "grad_norm": 2.8325319290161133,
      "learning_rate": 9.858217465842178e-06,
      "loss": 0.9389,
      "step": 2567
    },
    {
      "epoch": 0.20751934382512777,
      "grad_norm": 2.903128147125244,
      "learning_rate": 9.85806269986652e-06,
      "loss": 1.1131,
      "step": 2568
    },
    {
      "epoch": 0.20760015353845532,
      "grad_norm": 3.4537770748138428,
      "learning_rate": 9.857907850683946e-06,
      "loss": 1.1703,
      "step": 2569
    },
    {
      "epoch": 0.20768096325178287,
      "grad_norm": 3.0188677310943604,
      "learning_rate": 9.857752918297103e-06,
      "loss": 0.9049,
      "step": 2570
    },
    {
      "epoch": 0.20776177296511042,
      "grad_norm": 3.0577592849731445,
      "learning_rate": 9.857597902708649e-06,
      "loss": 1.1445,
      "step": 2571
    },
    {
      "epoch": 0.20784258267843794,
      "grad_norm": 2.568509340286255,
      "learning_rate": 9.857442803921235e-06,
      "loss": 0.9156,
      "step": 2572
    },
    {
      "epoch": 0.2079233923917655,
      "grad_norm": 2.7146859169006348,
      "learning_rate": 9.857287621937522e-06,
      "loss": 1.0497,
      "step": 2573
    },
    {
      "epoch": 0.20800420210509304,
      "grad_norm": 3.3684639930725098,
      "learning_rate": 9.857132356760164e-06,
      "loss": 0.9339,
      "step": 2574
    },
    {
      "epoch": 0.20808501181842057,
      "grad_norm": 2.9014923572540283,
      "learning_rate": 9.856977008391824e-06,
      "loss": 1.0074,
      "step": 2575
    },
    {
      "epoch": 0.20816582153174812,
      "grad_norm": 2.680986166000366,
      "learning_rate": 9.856821576835159e-06,
      "loss": 1.1141,
      "step": 2576
    },
    {
      "epoch": 0.20824663124507567,
      "grad_norm": 2.9888999462127686,
      "learning_rate": 9.856666062092833e-06,
      "loss": 0.9345,
      "step": 2577
    },
    {
      "epoch": 0.2083274409584032,
      "grad_norm": 2.7942521572113037,
      "learning_rate": 9.856510464167508e-06,
      "loss": 1.0688,
      "step": 2578
    },
    {
      "epoch": 0.20840825067173074,
      "grad_norm": 2.224989414215088,
      "learning_rate": 9.856354783061851e-06,
      "loss": 1.0542,
      "step": 2579
    },
    {
      "epoch": 0.2084890603850583,
      "grad_norm": 2.730905294418335,
      "learning_rate": 9.856199018778527e-06,
      "loss": 0.9543,
      "step": 2580
    },
    {
      "epoch": 0.20856987009838582,
      "grad_norm": 2.5238234996795654,
      "learning_rate": 9.856043171320206e-06,
      "loss": 1.0149,
      "step": 2581
    },
    {
      "epoch": 0.20865067981171337,
      "grad_norm": 2.5379068851470947,
      "learning_rate": 9.855887240689556e-06,
      "loss": 0.9269,
      "step": 2582
    },
    {
      "epoch": 0.20873148952504092,
      "grad_norm": 3.115297317504883,
      "learning_rate": 9.855731226889246e-06,
      "loss": 0.9183,
      "step": 2583
    },
    {
      "epoch": 0.20881229923836844,
      "grad_norm": 2.9481358528137207,
      "learning_rate": 9.855575129921953e-06,
      "loss": 1.0207,
      "step": 2584
    },
    {
      "epoch": 0.208893108951696,
      "grad_norm": 2.897611141204834,
      "learning_rate": 9.855418949790346e-06,
      "loss": 1.1211,
      "step": 2585
    },
    {
      "epoch": 0.20897391866502354,
      "grad_norm": 3.3586490154266357,
      "learning_rate": 9.8552626864971e-06,
      "loss": 0.9698,
      "step": 2586
    },
    {
      "epoch": 0.20905472837835107,
      "grad_norm": 2.8079240322113037,
      "learning_rate": 9.855106340044893e-06,
      "loss": 1.0102,
      "step": 2587
    },
    {
      "epoch": 0.20913553809167862,
      "grad_norm": 2.4567348957061768,
      "learning_rate": 9.854949910436403e-06,
      "loss": 0.9974,
      "step": 2588
    },
    {
      "epoch": 0.20921634780500617,
      "grad_norm": 2.6109845638275146,
      "learning_rate": 9.85479339767431e-06,
      "loss": 0.977,
      "step": 2589
    },
    {
      "epoch": 0.2092971575183337,
      "grad_norm": 3.109159231185913,
      "learning_rate": 9.854636801761292e-06,
      "loss": 0.9796,
      "step": 2590
    },
    {
      "epoch": 0.20937796723166124,
      "grad_norm": 2.863799571990967,
      "learning_rate": 9.854480122700031e-06,
      "loss": 1.036,
      "step": 2591
    },
    {
      "epoch": 0.2094587769449888,
      "grad_norm": 3.403641939163208,
      "learning_rate": 9.854323360493215e-06,
      "loss": 1.1061,
      "step": 2592
    },
    {
      "epoch": 0.20953958665831632,
      "grad_norm": 2.7065038681030273,
      "learning_rate": 9.854166515143526e-06,
      "loss": 0.9701,
      "step": 2593
    },
    {
      "epoch": 0.20962039637164387,
      "grad_norm": 2.658090829849243,
      "learning_rate": 9.85400958665365e-06,
      "loss": 0.9804,
      "step": 2594
    },
    {
      "epoch": 0.20970120608497142,
      "grad_norm": 3.0017261505126953,
      "learning_rate": 9.853852575026274e-06,
      "loss": 1.1035,
      "step": 2595
    },
    {
      "epoch": 0.20978201579829894,
      "grad_norm": 3.070401430130005,
      "learning_rate": 9.853695480264091e-06,
      "loss": 1.0101,
      "step": 2596
    },
    {
      "epoch": 0.2098628255116265,
      "grad_norm": 2.946901559829712,
      "learning_rate": 9.853538302369787e-06,
      "loss": 0.932,
      "step": 2597
    },
    {
      "epoch": 0.20994363522495404,
      "grad_norm": 2.6623542308807373,
      "learning_rate": 9.853381041346058e-06,
      "loss": 1.0965,
      "step": 2598
    },
    {
      "epoch": 0.21002444493828157,
      "grad_norm": 2.513234853744507,
      "learning_rate": 9.853223697195596e-06,
      "loss": 0.952,
      "step": 2599
    },
    {
      "epoch": 0.21010525465160912,
      "grad_norm": 2.526477813720703,
      "learning_rate": 9.853066269921095e-06,
      "loss": 1.1357,
      "step": 2600
    },
    {
      "epoch": 0.21018606436493667,
      "grad_norm": 3.1761679649353027,
      "learning_rate": 9.85290875952525e-06,
      "loss": 1.0726,
      "step": 2601
    },
    {
      "epoch": 0.21026687407826422,
      "grad_norm": 2.8965203762054443,
      "learning_rate": 9.852751166010764e-06,
      "loss": 1.0015,
      "step": 2602
    },
    {
      "epoch": 0.21034768379159174,
      "grad_norm": 2.646193027496338,
      "learning_rate": 9.852593489380331e-06,
      "loss": 0.9826,
      "step": 2603
    },
    {
      "epoch": 0.2104284935049193,
      "grad_norm": 2.5496766567230225,
      "learning_rate": 9.852435729636656e-06,
      "loss": 1.0915,
      "step": 2604
    },
    {
      "epoch": 0.21050930321824685,
      "grad_norm": 2.8119606971740723,
      "learning_rate": 9.852277886782436e-06,
      "loss": 0.9898,
      "step": 2605
    },
    {
      "epoch": 0.21059011293157437,
      "grad_norm": 2.8318567276000977,
      "learning_rate": 9.852119960820379e-06,
      "loss": 1.0825,
      "step": 2606
    },
    {
      "epoch": 0.21067092264490192,
      "grad_norm": 3.3942651748657227,
      "learning_rate": 9.851961951753186e-06,
      "loss": 0.962,
      "step": 2607
    },
    {
      "epoch": 0.21075173235822947,
      "grad_norm": 2.667219400405884,
      "learning_rate": 9.85180385958357e-06,
      "loss": 1.0618,
      "step": 2608
    },
    {
      "epoch": 0.210832542071557,
      "grad_norm": 2.84959077835083,
      "learning_rate": 9.851645684314229e-06,
      "loss": 0.9492,
      "step": 2609
    },
    {
      "epoch": 0.21091335178488455,
      "grad_norm": 2.9073519706726074,
      "learning_rate": 9.851487425947878e-06,
      "loss": 1.1176,
      "step": 2610
    },
    {
      "epoch": 0.2109941614982121,
      "grad_norm": 2.5956332683563232,
      "learning_rate": 9.85132908448723e-06,
      "loss": 1.1036,
      "step": 2611
    },
    {
      "epoch": 0.21107497121153962,
      "grad_norm": 2.601874589920044,
      "learning_rate": 9.851170659934992e-06,
      "loss": 1.0242,
      "step": 2612
    },
    {
      "epoch": 0.21115578092486717,
      "grad_norm": 3.020754337310791,
      "learning_rate": 9.851012152293878e-06,
      "loss": 0.9522,
      "step": 2613
    },
    {
      "epoch": 0.21123659063819472,
      "grad_norm": 2.9108903408050537,
      "learning_rate": 9.850853561566607e-06,
      "loss": 1.1153,
      "step": 2614
    },
    {
      "epoch": 0.21131740035152224,
      "grad_norm": 3.2650146484375,
      "learning_rate": 9.85069488775589e-06,
      "loss": 1.0406,
      "step": 2615
    },
    {
      "epoch": 0.2113982100648498,
      "grad_norm": 3.147453546524048,
      "learning_rate": 9.850536130864447e-06,
      "loss": 0.9551,
      "step": 2616
    },
    {
      "epoch": 0.21147901977817735,
      "grad_norm": 2.9640023708343506,
      "learning_rate": 9.850377290894999e-06,
      "loss": 1.0614,
      "step": 2617
    },
    {
      "epoch": 0.21155982949150487,
      "grad_norm": 2.782986879348755,
      "learning_rate": 9.850218367850263e-06,
      "loss": 0.9692,
      "step": 2618
    },
    {
      "epoch": 0.21164063920483242,
      "grad_norm": 2.605253219604492,
      "learning_rate": 9.850059361732966e-06,
      "loss": 1.0636,
      "step": 2619
    },
    {
      "epoch": 0.21172144891815997,
      "grad_norm": 3.21453857421875,
      "learning_rate": 9.849900272545824e-06,
      "loss": 1.047,
      "step": 2620
    },
    {
      "epoch": 0.2118022586314875,
      "grad_norm": 2.6399147510528564,
      "learning_rate": 9.84974110029157e-06,
      "loss": 0.9074,
      "step": 2621
    },
    {
      "epoch": 0.21188306834481505,
      "grad_norm": 3.2476913928985596,
      "learning_rate": 9.849581844972924e-06,
      "loss": 0.9601,
      "step": 2622
    },
    {
      "epoch": 0.2119638780581426,
      "grad_norm": 2.878459930419922,
      "learning_rate": 9.849422506592616e-06,
      "loss": 1.0309,
      "step": 2623
    },
    {
      "epoch": 0.21204468777147012,
      "grad_norm": 2.9090170860290527,
      "learning_rate": 9.849263085153375e-06,
      "loss": 0.9745,
      "step": 2624
    },
    {
      "epoch": 0.21212549748479767,
      "grad_norm": 2.7304728031158447,
      "learning_rate": 9.849103580657933e-06,
      "loss": 1.0822,
      "step": 2625
    },
    {
      "epoch": 0.21220630719812522,
      "grad_norm": 2.804781436920166,
      "learning_rate": 9.848943993109018e-06,
      "loss": 0.9415,
      "step": 2626
    },
    {
      "epoch": 0.21228711691145274,
      "grad_norm": 2.578540563583374,
      "learning_rate": 9.848784322509366e-06,
      "loss": 1.0419,
      "step": 2627
    },
    {
      "epoch": 0.2123679266247803,
      "grad_norm": 2.940869092941284,
      "learning_rate": 9.848624568861713e-06,
      "loss": 0.9309,
      "step": 2628
    },
    {
      "epoch": 0.21244873633810785,
      "grad_norm": 2.9244909286499023,
      "learning_rate": 9.848464732168794e-06,
      "loss": 0.7676,
      "step": 2629
    },
    {
      "epoch": 0.21252954605143537,
      "grad_norm": 2.6605122089385986,
      "learning_rate": 9.848304812433345e-06,
      "loss": 0.9459,
      "step": 2630
    },
    {
      "epoch": 0.21261035576476292,
      "grad_norm": 2.712827444076538,
      "learning_rate": 9.848144809658106e-06,
      "loss": 0.8843,
      "step": 2631
    },
    {
      "epoch": 0.21269116547809047,
      "grad_norm": 2.9642374515533447,
      "learning_rate": 9.84798472384582e-06,
      "loss": 1.0906,
      "step": 2632
    },
    {
      "epoch": 0.212771975191418,
      "grad_norm": 2.5326364040374756,
      "learning_rate": 9.847824554999224e-06,
      "loss": 0.9623,
      "step": 2633
    },
    {
      "epoch": 0.21285278490474555,
      "grad_norm": 2.921952247619629,
      "learning_rate": 9.847664303121064e-06,
      "loss": 0.9162,
      "step": 2634
    },
    {
      "epoch": 0.2129335946180731,
      "grad_norm": 2.930405616760254,
      "learning_rate": 9.847503968214087e-06,
      "loss": 1.0664,
      "step": 2635
    },
    {
      "epoch": 0.21301440433140065,
      "grad_norm": 2.6176633834838867,
      "learning_rate": 9.847343550281037e-06,
      "loss": 0.9447,
      "step": 2636
    },
    {
      "epoch": 0.21309521404472817,
      "grad_norm": 2.7375478744506836,
      "learning_rate": 9.84718304932466e-06,
      "loss": 1.0987,
      "step": 2637
    },
    {
      "epoch": 0.21317602375805572,
      "grad_norm": 2.5312304496765137,
      "learning_rate": 9.847022465347708e-06,
      "loss": 0.9693,
      "step": 2638
    },
    {
      "epoch": 0.21325683347138327,
      "grad_norm": 2.684300422668457,
      "learning_rate": 9.84686179835293e-06,
      "loss": 0.9798,
      "step": 2639
    },
    {
      "epoch": 0.2133376431847108,
      "grad_norm": 2.656810998916626,
      "learning_rate": 9.846701048343075e-06,
      "loss": 0.9271,
      "step": 2640
    },
    {
      "epoch": 0.21341845289803835,
      "grad_norm": 3.095665693283081,
      "learning_rate": 9.8465402153209e-06,
      "loss": 0.9004,
      "step": 2641
    },
    {
      "epoch": 0.2134992626113659,
      "grad_norm": 2.690553665161133,
      "learning_rate": 9.84637929928916e-06,
      "loss": 1.0095,
      "step": 2642
    },
    {
      "epoch": 0.21358007232469342,
      "grad_norm": 2.515697479248047,
      "learning_rate": 9.84621830025061e-06,
      "loss": 0.9953,
      "step": 2643
    },
    {
      "epoch": 0.21366088203802097,
      "grad_norm": 2.9535646438598633,
      "learning_rate": 9.846057218208004e-06,
      "loss": 1.021,
      "step": 2644
    },
    {
      "epoch": 0.21374169175134852,
      "grad_norm": 3.273070812225342,
      "learning_rate": 9.845896053164108e-06,
      "loss": 1.047,
      "step": 2645
    },
    {
      "epoch": 0.21382250146467605,
      "grad_norm": 2.442079544067383,
      "learning_rate": 9.845734805121678e-06,
      "loss": 1.0215,
      "step": 2646
    },
    {
      "epoch": 0.2139033111780036,
      "grad_norm": 2.7628297805786133,
      "learning_rate": 9.845573474083477e-06,
      "loss": 1.1598,
      "step": 2647
    },
    {
      "epoch": 0.21398412089133115,
      "grad_norm": 3.064302921295166,
      "learning_rate": 9.845412060052264e-06,
      "loss": 0.9836,
      "step": 2648
    },
    {
      "epoch": 0.21406493060465867,
      "grad_norm": 2.716076135635376,
      "learning_rate": 9.84525056303081e-06,
      "loss": 1.1441,
      "step": 2649
    },
    {
      "epoch": 0.21414574031798622,
      "grad_norm": 2.504394292831421,
      "learning_rate": 9.845088983021878e-06,
      "loss": 0.9482,
      "step": 2650
    },
    {
      "epoch": 0.21422655003131377,
      "grad_norm": 2.709070920944214,
      "learning_rate": 9.844927320028236e-06,
      "loss": 1.0309,
      "step": 2651
    },
    {
      "epoch": 0.2143073597446413,
      "grad_norm": 2.7059810161590576,
      "learning_rate": 9.844765574052653e-06,
      "loss": 0.9578,
      "step": 2652
    },
    {
      "epoch": 0.21438816945796885,
      "grad_norm": 2.5697734355926514,
      "learning_rate": 9.844603745097898e-06,
      "loss": 0.9499,
      "step": 2653
    },
    {
      "epoch": 0.2144689791712964,
      "grad_norm": 2.6361827850341797,
      "learning_rate": 9.844441833166744e-06,
      "loss": 1.1609,
      "step": 2654
    },
    {
      "epoch": 0.21454978888462392,
      "grad_norm": 2.6841630935668945,
      "learning_rate": 9.844279838261966e-06,
      "loss": 0.9998,
      "step": 2655
    },
    {
      "epoch": 0.21463059859795147,
      "grad_norm": 2.805133581161499,
      "learning_rate": 9.844117760386333e-06,
      "loss": 0.9384,
      "step": 2656
    },
    {
      "epoch": 0.21471140831127902,
      "grad_norm": 2.73477840423584,
      "learning_rate": 9.843955599542627e-06,
      "loss": 1.1571,
      "step": 2657
    },
    {
      "epoch": 0.21479221802460655,
      "grad_norm": 2.5829696655273438,
      "learning_rate": 9.843793355733622e-06,
      "loss": 0.9096,
      "step": 2658
    },
    {
      "epoch": 0.2148730277379341,
      "grad_norm": 3.528594493865967,
      "learning_rate": 9.843631028962098e-06,
      "loss": 0.9122,
      "step": 2659
    },
    {
      "epoch": 0.21495383745126165,
      "grad_norm": 2.6276795864105225,
      "learning_rate": 9.843468619230833e-06,
      "loss": 1.0149,
      "step": 2660
    },
    {
      "epoch": 0.21503464716458917,
      "grad_norm": 2.650006055831909,
      "learning_rate": 9.843306126542613e-06,
      "loss": 0.9662,
      "step": 2661
    },
    {
      "epoch": 0.21511545687791672,
      "grad_norm": 3.024837017059326,
      "learning_rate": 9.843143550900219e-06,
      "loss": 1.0055,
      "step": 2662
    },
    {
      "epoch": 0.21519626659124427,
      "grad_norm": 2.8011395931243896,
      "learning_rate": 9.842980892306436e-06,
      "loss": 1.0391,
      "step": 2663
    },
    {
      "epoch": 0.2152770763045718,
      "grad_norm": 2.5469000339508057,
      "learning_rate": 9.842818150764047e-06,
      "loss": 1.1026,
      "step": 2664
    },
    {
      "epoch": 0.21535788601789935,
      "grad_norm": 2.7628118991851807,
      "learning_rate": 9.842655326275843e-06,
      "loss": 0.9756,
      "step": 2665
    },
    {
      "epoch": 0.2154386957312269,
      "grad_norm": 2.6595170497894287,
      "learning_rate": 9.842492418844612e-06,
      "loss": 1.032,
      "step": 2666
    },
    {
      "epoch": 0.21551950544455445,
      "grad_norm": 2.702587604522705,
      "learning_rate": 9.842329428473143e-06,
      "loss": 0.972,
      "step": 2667
    },
    {
      "epoch": 0.21560031515788197,
      "grad_norm": 2.8801355361938477,
      "learning_rate": 9.842166355164227e-06,
      "loss": 1.0336,
      "step": 2668
    },
    {
      "epoch": 0.21568112487120952,
      "grad_norm": 3.040523052215576,
      "learning_rate": 9.84200319892066e-06,
      "loss": 0.9853,
      "step": 2669
    },
    {
      "epoch": 0.21576193458453707,
      "grad_norm": 3.4121859073638916,
      "learning_rate": 9.841839959745236e-06,
      "loss": 1.0504,
      "step": 2670
    },
    {
      "epoch": 0.2158427442978646,
      "grad_norm": 2.914210796356201,
      "learning_rate": 9.841676637640747e-06,
      "loss": 0.9771,
      "step": 2671
    },
    {
      "epoch": 0.21592355401119215,
      "grad_norm": 2.971970796585083,
      "learning_rate": 9.841513232609994e-06,
      "loss": 1.0135,
      "step": 2672
    },
    {
      "epoch": 0.2160043637245197,
      "grad_norm": 2.7145802974700928,
      "learning_rate": 9.841349744655776e-06,
      "loss": 1.046,
      "step": 2673
    },
    {
      "epoch": 0.21608517343784722,
      "grad_norm": 2.629023790359497,
      "learning_rate": 9.84118617378089e-06,
      "loss": 0.9498,
      "step": 2674
    },
    {
      "epoch": 0.21616598315117477,
      "grad_norm": 2.536773920059204,
      "learning_rate": 9.841022519988142e-06,
      "loss": 1.0527,
      "step": 2675
    },
    {
      "epoch": 0.21624679286450232,
      "grad_norm": 2.9043614864349365,
      "learning_rate": 9.84085878328033e-06,
      "loss": 0.9488,
      "step": 2676
    },
    {
      "epoch": 0.21632760257782985,
      "grad_norm": 2.8709604740142822,
      "learning_rate": 9.840694963660262e-06,
      "loss": 0.9625,
      "step": 2677
    },
    {
      "epoch": 0.2164084122911574,
      "grad_norm": 2.7993056774139404,
      "learning_rate": 9.840531061130742e-06,
      "loss": 1.1074,
      "step": 2678
    },
    {
      "epoch": 0.21648922200448495,
      "grad_norm": 3.800844430923462,
      "learning_rate": 9.84036707569458e-06,
      "loss": 1.053,
      "step": 2679
    },
    {
      "epoch": 0.21657003171781247,
      "grad_norm": 2.661588430404663,
      "learning_rate": 9.840203007354581e-06,
      "loss": 0.9514,
      "step": 2680
    },
    {
      "epoch": 0.21665084143114002,
      "grad_norm": 2.9842541217803955,
      "learning_rate": 9.840038856113558e-06,
      "loss": 0.9228,
      "step": 2681
    },
    {
      "epoch": 0.21673165114446757,
      "grad_norm": 2.817148447036743,
      "learning_rate": 9.83987462197432e-06,
      "loss": 0.9725,
      "step": 2682
    },
    {
      "epoch": 0.2168124608577951,
      "grad_norm": 3.2969770431518555,
      "learning_rate": 9.839710304939683e-06,
      "loss": 1.0276,
      "step": 2683
    },
    {
      "epoch": 0.21689327057112265,
      "grad_norm": 2.8415517807006836,
      "learning_rate": 9.839545905012457e-06,
      "loss": 1.0284,
      "step": 2684
    },
    {
      "epoch": 0.2169740802844502,
      "grad_norm": 3.089386224746704,
      "learning_rate": 9.839381422195464e-06,
      "loss": 1.0254,
      "step": 2685
    },
    {
      "epoch": 0.21705488999777772,
      "grad_norm": 3.7873218059539795,
      "learning_rate": 9.839216856491514e-06,
      "loss": 1.0733,
      "step": 2686
    },
    {
      "epoch": 0.21713569971110527,
      "grad_norm": 2.827165365219116,
      "learning_rate": 9.839052207903431e-06,
      "loss": 0.918,
      "step": 2687
    },
    {
      "epoch": 0.21721650942443282,
      "grad_norm": 3.022129535675049,
      "learning_rate": 9.838887476434033e-06,
      "loss": 0.9254,
      "step": 2688
    },
    {
      "epoch": 0.21729731913776035,
      "grad_norm": 2.853858470916748,
      "learning_rate": 9.838722662086142e-06,
      "loss": 1.0868,
      "step": 2689
    },
    {
      "epoch": 0.2173781288510879,
      "grad_norm": 2.8158044815063477,
      "learning_rate": 9.83855776486258e-06,
      "loss": 0.9851,
      "step": 2690
    },
    {
      "epoch": 0.21745893856441545,
      "grad_norm": 2.678738594055176,
      "learning_rate": 9.838392784766172e-06,
      "loss": 1.1053,
      "step": 2691
    },
    {
      "epoch": 0.21753974827774297,
      "grad_norm": 2.6264524459838867,
      "learning_rate": 9.838227721799742e-06,
      "loss": 0.978,
      "step": 2692
    },
    {
      "epoch": 0.21762055799107052,
      "grad_norm": 2.5500595569610596,
      "learning_rate": 9.83806257596612e-06,
      "loss": 0.9456,
      "step": 2693
    },
    {
      "epoch": 0.21770136770439807,
      "grad_norm": 2.5118589401245117,
      "learning_rate": 9.837897347268134e-06,
      "loss": 0.8679,
      "step": 2694
    },
    {
      "epoch": 0.2177821774177256,
      "grad_norm": 2.5622048377990723,
      "learning_rate": 9.837732035708613e-06,
      "loss": 1.1071,
      "step": 2695
    },
    {
      "epoch": 0.21786298713105315,
      "grad_norm": 3.4003055095672607,
      "learning_rate": 9.837566641290388e-06,
      "loss": 1.0472,
      "step": 2696
    },
    {
      "epoch": 0.2179437968443807,
      "grad_norm": 2.976776599884033,
      "learning_rate": 9.837401164016293e-06,
      "loss": 1.0828,
      "step": 2697
    },
    {
      "epoch": 0.21802460655770825,
      "grad_norm": 2.4960176944732666,
      "learning_rate": 9.837235603889162e-06,
      "loss": 0.9775,
      "step": 2698
    },
    {
      "epoch": 0.21810541627103577,
      "grad_norm": 2.8212971687316895,
      "learning_rate": 9.837069960911829e-06,
      "loss": 0.9227,
      "step": 2699
    },
    {
      "epoch": 0.21818622598436332,
      "grad_norm": 3.3433587551116943,
      "learning_rate": 9.836904235087132e-06,
      "loss": 0.9447,
      "step": 2700
    },
    {
      "epoch": 0.21826703569769088,
      "grad_norm": 3.5738041400909424,
      "learning_rate": 9.836738426417911e-06,
      "loss": 0.9704,
      "step": 2701
    },
    {
      "epoch": 0.2183478454110184,
      "grad_norm": 2.5765058994293213,
      "learning_rate": 9.836572534907005e-06,
      "loss": 1.0977,
      "step": 2702
    },
    {
      "epoch": 0.21842865512434595,
      "grad_norm": 2.7243411540985107,
      "learning_rate": 9.836406560557254e-06,
      "loss": 1.0175,
      "step": 2703
    },
    {
      "epoch": 0.2185094648376735,
      "grad_norm": 2.91159725189209,
      "learning_rate": 9.836240503371503e-06,
      "loss": 1.0504,
      "step": 2704
    },
    {
      "epoch": 0.21859027455100102,
      "grad_norm": 2.647514581680298,
      "learning_rate": 9.836074363352594e-06,
      "loss": 1.0958,
      "step": 2705
    },
    {
      "epoch": 0.21867108426432857,
      "grad_norm": 2.7375965118408203,
      "learning_rate": 9.835908140503374e-06,
      "loss": 1.0491,
      "step": 2706
    },
    {
      "epoch": 0.21875189397765613,
      "grad_norm": 2.7068612575531006,
      "learning_rate": 9.83574183482669e-06,
      "loss": 1.0139,
      "step": 2707
    },
    {
      "epoch": 0.21883270369098365,
      "grad_norm": 3.1376380920410156,
      "learning_rate": 9.835575446325386e-06,
      "loss": 0.9426,
      "step": 2708
    },
    {
      "epoch": 0.2189135134043112,
      "grad_norm": 3.265387773513794,
      "learning_rate": 9.83540897500232e-06,
      "loss": 1.0667,
      "step": 2709
    },
    {
      "epoch": 0.21899432311763875,
      "grad_norm": 2.5371029376983643,
      "learning_rate": 9.835242420860338e-06,
      "loss": 0.9974,
      "step": 2710
    },
    {
      "epoch": 0.21907513283096627,
      "grad_norm": 2.32127046585083,
      "learning_rate": 9.835075783902294e-06,
      "loss": 0.9514,
      "step": 2711
    },
    {
      "epoch": 0.21915594254429382,
      "grad_norm": 2.689162254333496,
      "learning_rate": 9.834909064131042e-06,
      "loss": 0.8845,
      "step": 2712
    },
    {
      "epoch": 0.21923675225762138,
      "grad_norm": 2.634633779525757,
      "learning_rate": 9.834742261549436e-06,
      "loss": 1.0544,
      "step": 2713
    },
    {
      "epoch": 0.2193175619709489,
      "grad_norm": 2.6736345291137695,
      "learning_rate": 9.834575376160336e-06,
      "loss": 0.9929,
      "step": 2714
    },
    {
      "epoch": 0.21939837168427645,
      "grad_norm": 2.721517562866211,
      "learning_rate": 9.834408407966597e-06,
      "loss": 1.113,
      "step": 2715
    },
    {
      "epoch": 0.219479181397604,
      "grad_norm": 2.985367774963379,
      "learning_rate": 9.834241356971082e-06,
      "loss": 0.9398,
      "step": 2716
    },
    {
      "epoch": 0.21955999111093152,
      "grad_norm": 2.432053804397583,
      "learning_rate": 9.834074223176648e-06,
      "loss": 1.0002,
      "step": 2717
    },
    {
      "epoch": 0.21964080082425907,
      "grad_norm": 2.7274587154388428,
      "learning_rate": 9.833907006586162e-06,
      "loss": 1.1054,
      "step": 2718
    },
    {
      "epoch": 0.21972161053758663,
      "grad_norm": 2.675016403198242,
      "learning_rate": 9.833739707202485e-06,
      "loss": 0.973,
      "step": 2719
    },
    {
      "epoch": 0.21980242025091415,
      "grad_norm": 2.9005885124206543,
      "learning_rate": 9.833572325028485e-06,
      "loss": 1.0865,
      "step": 2720
    },
    {
      "epoch": 0.2198832299642417,
      "grad_norm": 3.1062748432159424,
      "learning_rate": 9.833404860067027e-06,
      "loss": 0.9241,
      "step": 2721
    },
    {
      "epoch": 0.21996403967756925,
      "grad_norm": 2.929474353790283,
      "learning_rate": 9.833237312320979e-06,
      "loss": 0.9104,
      "step": 2722
    },
    {
      "epoch": 0.22004484939089677,
      "grad_norm": 2.922116279602051,
      "learning_rate": 9.833069681793212e-06,
      "loss": 0.9415,
      "step": 2723
    },
    {
      "epoch": 0.22012565910422432,
      "grad_norm": 2.6386096477508545,
      "learning_rate": 9.832901968486597e-06,
      "loss": 1.0628,
      "step": 2724
    },
    {
      "epoch": 0.22020646881755188,
      "grad_norm": 2.8515048027038574,
      "learning_rate": 9.832734172404003e-06,
      "loss": 0.8596,
      "step": 2725
    },
    {
      "epoch": 0.2202872785308794,
      "grad_norm": 2.7838757038116455,
      "learning_rate": 9.83256629354831e-06,
      "loss": 0.987,
      "step": 2726
    },
    {
      "epoch": 0.22036808824420695,
      "grad_norm": 2.8564083576202393,
      "learning_rate": 9.83239833192239e-06,
      "loss": 1.0477,
      "step": 2727
    },
    {
      "epoch": 0.2204488979575345,
      "grad_norm": 2.657640218734741,
      "learning_rate": 9.83223028752912e-06,
      "loss": 0.9471,
      "step": 2728
    },
    {
      "epoch": 0.22052970767086202,
      "grad_norm": 2.1479780673980713,
      "learning_rate": 9.832062160371378e-06,
      "loss": 1.0779,
      "step": 2729
    },
    {
      "epoch": 0.22061051738418958,
      "grad_norm": 2.413940906524658,
      "learning_rate": 9.831893950452044e-06,
      "loss": 1.0194,
      "step": 2730
    },
    {
      "epoch": 0.22069132709751713,
      "grad_norm": 2.5537819862365723,
      "learning_rate": 9.831725657773999e-06,
      "loss": 1.0628,
      "step": 2731
    },
    {
      "epoch": 0.22077213681084468,
      "grad_norm": 2.848949909210205,
      "learning_rate": 9.831557282340125e-06,
      "loss": 1.0231,
      "step": 2732
    },
    {
      "epoch": 0.2208529465241722,
      "grad_norm": 3.080414295196533,
      "learning_rate": 9.831388824153306e-06,
      "loss": 1.0486,
      "step": 2733
    },
    {
      "epoch": 0.22093375623749975,
      "grad_norm": 3.120030641555786,
      "learning_rate": 9.831220283216428e-06,
      "loss": 1.0214,
      "step": 2734
    },
    {
      "epoch": 0.2210145659508273,
      "grad_norm": 3.1183664798736572,
      "learning_rate": 9.831051659532378e-06,
      "loss": 1.0401,
      "step": 2735
    },
    {
      "epoch": 0.22109537566415483,
      "grad_norm": 3.092597246170044,
      "learning_rate": 9.830882953104042e-06,
      "loss": 0.9186,
      "step": 2736
    },
    {
      "epoch": 0.22117618537748238,
      "grad_norm": 3.042254686355591,
      "learning_rate": 9.830714163934312e-06,
      "loss": 0.9569,
      "step": 2737
    },
    {
      "epoch": 0.22125699509080993,
      "grad_norm": 2.5840420722961426,
      "learning_rate": 9.830545292026077e-06,
      "loss": 1.0692,
      "step": 2738
    },
    {
      "epoch": 0.22133780480413745,
      "grad_norm": 3.314822196960449,
      "learning_rate": 9.83037633738223e-06,
      "loss": 0.9808,
      "step": 2739
    },
    {
      "epoch": 0.221418614517465,
      "grad_norm": 2.7320046424865723,
      "learning_rate": 9.830207300005665e-06,
      "loss": 0.9668,
      "step": 2740
    },
    {
      "epoch": 0.22149942423079255,
      "grad_norm": 2.76070237159729,
      "learning_rate": 9.830038179899278e-06,
      "loss": 0.9549,
      "step": 2741
    },
    {
      "epoch": 0.22158023394412008,
      "grad_norm": 2.71372389793396,
      "learning_rate": 9.829868977065964e-06,
      "loss": 0.9483,
      "step": 2742
    },
    {
      "epoch": 0.22166104365744763,
      "grad_norm": 3.2915704250335693,
      "learning_rate": 9.829699691508624e-06,
      "loss": 0.9798,
      "step": 2743
    },
    {
      "epoch": 0.22174185337077518,
      "grad_norm": 2.8336358070373535,
      "learning_rate": 9.829530323230151e-06,
      "loss": 1.0364,
      "step": 2744
    },
    {
      "epoch": 0.2218226630841027,
      "grad_norm": 3.036170721054077,
      "learning_rate": 9.829360872233455e-06,
      "loss": 1.034,
      "step": 2745
    },
    {
      "epoch": 0.22190347279743025,
      "grad_norm": 2.8102879524230957,
      "learning_rate": 9.829191338521431e-06,
      "loss": 1.0109,
      "step": 2746
    },
    {
      "epoch": 0.2219842825107578,
      "grad_norm": 3.2442543506622314,
      "learning_rate": 9.829021722096984e-06,
      "loss": 1.0299,
      "step": 2747
    },
    {
      "epoch": 0.22206509222408533,
      "grad_norm": 2.845066547393799,
      "learning_rate": 9.828852022963023e-06,
      "loss": 0.9285,
      "step": 2748
    },
    {
      "epoch": 0.22214590193741288,
      "grad_norm": 2.9424619674682617,
      "learning_rate": 9.828682241122452e-06,
      "loss": 1.0555,
      "step": 2749
    },
    {
      "epoch": 0.22222671165074043,
      "grad_norm": 2.4247968196868896,
      "learning_rate": 9.828512376578177e-06,
      "loss": 0.9813,
      "step": 2750
    },
    {
      "epoch": 0.22230752136406795,
      "grad_norm": 3.497758626937866,
      "learning_rate": 9.828342429333108e-06,
      "loss": 0.9966,
      "step": 2751
    },
    {
      "epoch": 0.2223883310773955,
      "grad_norm": 2.899437427520752,
      "learning_rate": 9.828172399390158e-06,
      "loss": 1.0316,
      "step": 2752
    },
    {
      "epoch": 0.22246914079072305,
      "grad_norm": 2.879088878631592,
      "learning_rate": 9.82800228675224e-06,
      "loss": 1.031,
      "step": 2753
    },
    {
      "epoch": 0.22254995050405058,
      "grad_norm": 2.6835739612579346,
      "learning_rate": 9.827832091422265e-06,
      "loss": 0.9471,
      "step": 2754
    },
    {
      "epoch": 0.22263076021737813,
      "grad_norm": 2.815185308456421,
      "learning_rate": 9.827661813403148e-06,
      "loss": 1.058,
      "step": 2755
    },
    {
      "epoch": 0.22271156993070568,
      "grad_norm": 2.932528018951416,
      "learning_rate": 9.827491452697806e-06,
      "loss": 0.9813,
      "step": 2756
    },
    {
      "epoch": 0.2227923796440332,
      "grad_norm": 2.6303751468658447,
      "learning_rate": 9.827321009309159e-06,
      "loss": 0.9364,
      "step": 2757
    },
    {
      "epoch": 0.22287318935736075,
      "grad_norm": 2.613288640975952,
      "learning_rate": 9.827150483240123e-06,
      "loss": 0.9016,
      "step": 2758
    },
    {
      "epoch": 0.2229539990706883,
      "grad_norm": 2.978224039077759,
      "learning_rate": 9.826979874493618e-06,
      "loss": 0.9855,
      "step": 2759
    },
    {
      "epoch": 0.22303480878401583,
      "grad_norm": 2.6431267261505127,
      "learning_rate": 9.826809183072572e-06,
      "loss": 0.9776,
      "step": 2760
    },
    {
      "epoch": 0.22311561849734338,
      "grad_norm": 2.6241977214813232,
      "learning_rate": 9.826638408979903e-06,
      "loss": 1.0613,
      "step": 2761
    },
    {
      "epoch": 0.22319642821067093,
      "grad_norm": 2.6231777667999268,
      "learning_rate": 9.826467552218537e-06,
      "loss": 1.0571,
      "step": 2762
    },
    {
      "epoch": 0.22327723792399848,
      "grad_norm": 3.046290397644043,
      "learning_rate": 9.826296612791403e-06,
      "loss": 0.9999,
      "step": 2763
    },
    {
      "epoch": 0.223358047637326,
      "grad_norm": 3.1164638996124268,
      "learning_rate": 9.826125590701425e-06,
      "loss": 1.0017,
      "step": 2764
    },
    {
      "epoch": 0.22343885735065355,
      "grad_norm": 2.63796067237854,
      "learning_rate": 9.825954485951536e-06,
      "loss": 1.073,
      "step": 2765
    },
    {
      "epoch": 0.2235196670639811,
      "grad_norm": 2.663356065750122,
      "learning_rate": 9.825783298544662e-06,
      "loss": 0.9145,
      "step": 2766
    },
    {
      "epoch": 0.22360047677730863,
      "grad_norm": 3.481412887573242,
      "learning_rate": 9.82561202848374e-06,
      "loss": 0.8822,
      "step": 2767
    },
    {
      "epoch": 0.22368128649063618,
      "grad_norm": 2.5267741680145264,
      "learning_rate": 9.8254406757717e-06,
      "loss": 0.9521,
      "step": 2768
    },
    {
      "epoch": 0.22376209620396373,
      "grad_norm": 2.999199151992798,
      "learning_rate": 9.825269240411478e-06,
      "loss": 0.9925,
      "step": 2769
    },
    {
      "epoch": 0.22384290591729125,
      "grad_norm": 2.903404474258423,
      "learning_rate": 9.825097722406012e-06,
      "loss": 0.9252,
      "step": 2770
    },
    {
      "epoch": 0.2239237156306188,
      "grad_norm": 2.6118807792663574,
      "learning_rate": 9.824926121758236e-06,
      "loss": 0.857,
      "step": 2771
    },
    {
      "epoch": 0.22400452534394635,
      "grad_norm": 2.513718366622925,
      "learning_rate": 9.824754438471091e-06,
      "loss": 1.0926,
      "step": 2772
    },
    {
      "epoch": 0.22408533505727388,
      "grad_norm": 2.6014599800109863,
      "learning_rate": 9.82458267254752e-06,
      "loss": 0.9563,
      "step": 2773
    },
    {
      "epoch": 0.22416614477060143,
      "grad_norm": 2.4804561138153076,
      "learning_rate": 9.82441082399046e-06,
      "loss": 1.1281,
      "step": 2774
    },
    {
      "epoch": 0.22424695448392898,
      "grad_norm": 2.6518805027008057,
      "learning_rate": 9.824238892802858e-06,
      "loss": 0.9119,
      "step": 2775
    },
    {
      "epoch": 0.2243277641972565,
      "grad_norm": 2.772603750228882,
      "learning_rate": 9.824066878987657e-06,
      "loss": 1.0272,
      "step": 2776
    },
    {
      "epoch": 0.22440857391058405,
      "grad_norm": 3.0260932445526123,
      "learning_rate": 9.823894782547803e-06,
      "loss": 1.1759,
      "step": 2777
    },
    {
      "epoch": 0.2244893836239116,
      "grad_norm": 2.77278208732605,
      "learning_rate": 9.823722603486247e-06,
      "loss": 0.9999,
      "step": 2778
    },
    {
      "epoch": 0.22457019333723913,
      "grad_norm": 2.864107370376587,
      "learning_rate": 9.823550341805933e-06,
      "loss": 1.0211,
      "step": 2779
    },
    {
      "epoch": 0.22465100305056668,
      "grad_norm": 2.401287078857422,
      "learning_rate": 9.823377997509816e-06,
      "loss": 0.9441,
      "step": 2780
    },
    {
      "epoch": 0.22473181276389423,
      "grad_norm": 2.9367048740386963,
      "learning_rate": 9.823205570600844e-06,
      "loss": 0.9651,
      "step": 2781
    },
    {
      "epoch": 0.22481262247722175,
      "grad_norm": 2.9967193603515625,
      "learning_rate": 9.823033061081973e-06,
      "loss": 1.0323,
      "step": 2782
    },
    {
      "epoch": 0.2248934321905493,
      "grad_norm": 2.6746819019317627,
      "learning_rate": 9.822860468956155e-06,
      "loss": 1.0986,
      "step": 2783
    },
    {
      "epoch": 0.22497424190387685,
      "grad_norm": 2.1297669410705566,
      "learning_rate": 9.822687794226348e-06,
      "loss": 1.1349,
      "step": 2784
    },
    {
      "epoch": 0.22505505161720438,
      "grad_norm": 2.794849395751953,
      "learning_rate": 9.82251503689551e-06,
      "loss": 0.9071,
      "step": 2785
    },
    {
      "epoch": 0.22513586133053193,
      "grad_norm": 2.7644519805908203,
      "learning_rate": 9.822342196966601e-06,
      "loss": 1.0522,
      "step": 2786
    },
    {
      "epoch": 0.22521667104385948,
      "grad_norm": 2.584132194519043,
      "learning_rate": 9.822169274442577e-06,
      "loss": 1.0595,
      "step": 2787
    },
    {
      "epoch": 0.225297480757187,
      "grad_norm": 2.742427349090576,
      "learning_rate": 9.821996269326403e-06,
      "loss": 0.9886,
      "step": 2788
    },
    {
      "epoch": 0.22537829047051455,
      "grad_norm": 2.970609188079834,
      "learning_rate": 9.821823181621043e-06,
      "loss": 0.9794,
      "step": 2789
    },
    {
      "epoch": 0.2254591001838421,
      "grad_norm": 3.2143619060516357,
      "learning_rate": 9.821650011329458e-06,
      "loss": 1.0037,
      "step": 2790
    },
    {
      "epoch": 0.22553990989716963,
      "grad_norm": 3.3506712913513184,
      "learning_rate": 9.821476758454616e-06,
      "loss": 1.0079,
      "step": 2791
    },
    {
      "epoch": 0.22562071961049718,
      "grad_norm": 2.8092939853668213,
      "learning_rate": 9.821303422999484e-06,
      "loss": 0.9338,
      "step": 2792
    },
    {
      "epoch": 0.22570152932382473,
      "grad_norm": 2.6990749835968018,
      "learning_rate": 9.821130004967032e-06,
      "loss": 1.0369,
      "step": 2793
    },
    {
      "epoch": 0.22578233903715225,
      "grad_norm": 2.468820810317993,
      "learning_rate": 9.82095650436023e-06,
      "loss": 1.1118,
      "step": 2794
    },
    {
      "epoch": 0.2258631487504798,
      "grad_norm": 2.9115161895751953,
      "learning_rate": 9.820782921182049e-06,
      "loss": 0.9823,
      "step": 2795
    },
    {
      "epoch": 0.22594395846380735,
      "grad_norm": 2.591421365737915,
      "learning_rate": 9.82060925543546e-06,
      "loss": 1.1116,
      "step": 2796
    },
    {
      "epoch": 0.2260247681771349,
      "grad_norm": 2.7503275871276855,
      "learning_rate": 9.82043550712344e-06,
      "loss": 1.0216,
      "step": 2797
    },
    {
      "epoch": 0.22610557789046243,
      "grad_norm": 2.9692800045013428,
      "learning_rate": 9.820261676248969e-06,
      "loss": 0.8514,
      "step": 2798
    },
    {
      "epoch": 0.22618638760378998,
      "grad_norm": 2.3442323207855225,
      "learning_rate": 9.820087762815013e-06,
      "loss": 1.0024,
      "step": 2799
    },
    {
      "epoch": 0.22626719731711753,
      "grad_norm": 2.81729793548584,
      "learning_rate": 9.819913766824563e-06,
      "loss": 1.0367,
      "step": 2800
    },
    {
      "epoch": 0.22634800703044505,
      "grad_norm": 2.569409132003784,
      "learning_rate": 9.81973968828059e-06,
      "loss": 1.1571,
      "step": 2801
    },
    {
      "epoch": 0.2264288167437726,
      "grad_norm": 3.1350796222686768,
      "learning_rate": 9.819565527186082e-06,
      "loss": 0.9435,
      "step": 2802
    },
    {
      "epoch": 0.22650962645710015,
      "grad_norm": 2.7625088691711426,
      "learning_rate": 9.819391283544018e-06,
      "loss": 0.9354,
      "step": 2803
    },
    {
      "epoch": 0.22659043617042768,
      "grad_norm": 2.755223035812378,
      "learning_rate": 9.819216957357382e-06,
      "loss": 0.962,
      "step": 2804
    },
    {
      "epoch": 0.22667124588375523,
      "grad_norm": 2.7333672046661377,
      "learning_rate": 9.819042548629163e-06,
      "loss": 0.9441,
      "step": 2805
    },
    {
      "epoch": 0.22675205559708278,
      "grad_norm": 2.9572391510009766,
      "learning_rate": 9.818868057362346e-06,
      "loss": 1.0607,
      "step": 2806
    },
    {
      "epoch": 0.2268328653104103,
      "grad_norm": 2.7844395637512207,
      "learning_rate": 9.81869348355992e-06,
      "loss": 0.9216,
      "step": 2807
    },
    {
      "epoch": 0.22691367502373785,
      "grad_norm": 2.7145252227783203,
      "learning_rate": 9.818518827224877e-06,
      "loss": 1.0075,
      "step": 2808
    },
    {
      "epoch": 0.2269944847370654,
      "grad_norm": 3.09114670753479,
      "learning_rate": 9.818344088360204e-06,
      "loss": 1.1068,
      "step": 2809
    },
    {
      "epoch": 0.22707529445039293,
      "grad_norm": 2.9854092597961426,
      "learning_rate": 9.818169266968899e-06,
      "loss": 1.0327,
      "step": 2810
    },
    {
      "epoch": 0.22715610416372048,
      "grad_norm": 3.1989142894744873,
      "learning_rate": 9.81799436305395e-06,
      "loss": 1.0527,
      "step": 2811
    },
    {
      "epoch": 0.22723691387704803,
      "grad_norm": 3.338559150695801,
      "learning_rate": 9.81781937661836e-06,
      "loss": 1.0737,
      "step": 2812
    },
    {
      "epoch": 0.22731772359037555,
      "grad_norm": 3.4588310718536377,
      "learning_rate": 9.81764430766512e-06,
      "loss": 0.9356,
      "step": 2813
    },
    {
      "epoch": 0.2273985333037031,
      "grad_norm": 2.61423397064209,
      "learning_rate": 9.817469156197232e-06,
      "loss": 1.0299,
      "step": 2814
    },
    {
      "epoch": 0.22747934301703066,
      "grad_norm": 3.1423604488372803,
      "learning_rate": 9.817293922217697e-06,
      "loss": 0.9767,
      "step": 2815
    },
    {
      "epoch": 0.22756015273035818,
      "grad_norm": 2.9998934268951416,
      "learning_rate": 9.817118605729512e-06,
      "loss": 1.0646,
      "step": 2816
    },
    {
      "epoch": 0.22764096244368573,
      "grad_norm": 2.620575428009033,
      "learning_rate": 9.816943206735682e-06,
      "loss": 0.9874,
      "step": 2817
    },
    {
      "epoch": 0.22772177215701328,
      "grad_norm": 3.3366506099700928,
      "learning_rate": 9.816767725239212e-06,
      "loss": 1.1476,
      "step": 2818
    },
    {
      "epoch": 0.2278025818703408,
      "grad_norm": 2.993959903717041,
      "learning_rate": 9.816592161243106e-06,
      "loss": 1.1764,
      "step": 2819
    },
    {
      "epoch": 0.22788339158366835,
      "grad_norm": 2.9595839977264404,
      "learning_rate": 9.816416514750372e-06,
      "loss": 0.9691,
      "step": 2820
    },
    {
      "epoch": 0.2279642012969959,
      "grad_norm": 2.701143741607666,
      "learning_rate": 9.816240785764019e-06,
      "loss": 1.0103,
      "step": 2821
    },
    {
      "epoch": 0.22804501101032343,
      "grad_norm": 3.1361124515533447,
      "learning_rate": 9.816064974287055e-06,
      "loss": 0.9805,
      "step": 2822
    },
    {
      "epoch": 0.22812582072365098,
      "grad_norm": 2.8855795860290527,
      "learning_rate": 9.815889080322491e-06,
      "loss": 1.0092,
      "step": 2823
    },
    {
      "epoch": 0.22820663043697853,
      "grad_norm": 2.637803316116333,
      "learning_rate": 9.815713103873343e-06,
      "loss": 0.9422,
      "step": 2824
    },
    {
      "epoch": 0.22828744015030605,
      "grad_norm": 2.517380475997925,
      "learning_rate": 9.815537044942622e-06,
      "loss": 1.0371,
      "step": 2825
    },
    {
      "epoch": 0.2283682498636336,
      "grad_norm": 2.720059633255005,
      "learning_rate": 9.815360903533345e-06,
      "loss": 0.9992,
      "step": 2826
    },
    {
      "epoch": 0.22844905957696116,
      "grad_norm": 2.635488986968994,
      "learning_rate": 9.815184679648529e-06,
      "loss": 1.0112,
      "step": 2827
    },
    {
      "epoch": 0.2285298692902887,
      "grad_norm": 3.504739284515381,
      "learning_rate": 9.815008373291188e-06,
      "loss": 1.0923,
      "step": 2828
    },
    {
      "epoch": 0.22861067900361623,
      "grad_norm": 3.1671524047851562,
      "learning_rate": 9.814831984464347e-06,
      "loss": 0.9475,
      "step": 2829
    },
    {
      "epoch": 0.22869148871694378,
      "grad_norm": 2.455099105834961,
      "learning_rate": 9.814655513171028e-06,
      "loss": 1.089,
      "step": 2830
    },
    {
      "epoch": 0.22877229843027133,
      "grad_norm": 2.755807876586914,
      "learning_rate": 9.814478959414248e-06,
      "loss": 0.9854,
      "step": 2831
    },
    {
      "epoch": 0.22885310814359885,
      "grad_norm": 3.0457632541656494,
      "learning_rate": 9.814302323197033e-06,
      "loss": 0.982,
      "step": 2832
    },
    {
      "epoch": 0.2289339178569264,
      "grad_norm": 2.8635318279266357,
      "learning_rate": 9.814125604522412e-06,
      "loss": 0.9566,
      "step": 2833
    },
    {
      "epoch": 0.22901472757025396,
      "grad_norm": 2.9081757068634033,
      "learning_rate": 9.813948803393407e-06,
      "loss": 0.9664,
      "step": 2834
    },
    {
      "epoch": 0.22909553728358148,
      "grad_norm": 2.9280664920806885,
      "learning_rate": 9.813771919813049e-06,
      "loss": 1.0221,
      "step": 2835
    },
    {
      "epoch": 0.22917634699690903,
      "grad_norm": 2.643336534500122,
      "learning_rate": 9.813594953784366e-06,
      "loss": 0.923,
      "step": 2836
    },
    {
      "epoch": 0.22925715671023658,
      "grad_norm": 2.497655153274536,
      "learning_rate": 9.813417905310391e-06,
      "loss": 0.9823,
      "step": 2837
    },
    {
      "epoch": 0.2293379664235641,
      "grad_norm": 2.8711609840393066,
      "learning_rate": 9.813240774394153e-06,
      "loss": 1.0016,
      "step": 2838
    },
    {
      "epoch": 0.22941877613689166,
      "grad_norm": 2.8473143577575684,
      "learning_rate": 9.81306356103869e-06,
      "loss": 1.0128,
      "step": 2839
    },
    {
      "epoch": 0.2294995858502192,
      "grad_norm": 2.6120336055755615,
      "learning_rate": 9.812886265247035e-06,
      "loss": 0.9938,
      "step": 2840
    },
    {
      "epoch": 0.22958039556354673,
      "grad_norm": 2.9384772777557373,
      "learning_rate": 9.812708887022223e-06,
      "loss": 0.9876,
      "step": 2841
    },
    {
      "epoch": 0.22966120527687428,
      "grad_norm": 3.0501091480255127,
      "learning_rate": 9.812531426367296e-06,
      "loss": 1.0746,
      "step": 2842
    },
    {
      "epoch": 0.22974201499020183,
      "grad_norm": 2.7544689178466797,
      "learning_rate": 9.81235388328529e-06,
      "loss": 1.0758,
      "step": 2843
    },
    {
      "epoch": 0.22982282470352935,
      "grad_norm": 3.162550210952759,
      "learning_rate": 9.812176257779248e-06,
      "loss": 0.9563,
      "step": 2844
    },
    {
      "epoch": 0.2299036344168569,
      "grad_norm": 2.8001787662506104,
      "learning_rate": 9.81199854985221e-06,
      "loss": 0.9562,
      "step": 2845
    },
    {
      "epoch": 0.22998444413018446,
      "grad_norm": 3.1681456565856934,
      "learning_rate": 9.811820759507223e-06,
      "loss": 0.8944,
      "step": 2846
    },
    {
      "epoch": 0.23006525384351198,
      "grad_norm": 2.618809461593628,
      "learning_rate": 9.81164288674733e-06,
      "loss": 1.0327,
      "step": 2847
    },
    {
      "epoch": 0.23014606355683953,
      "grad_norm": 3.1844985485076904,
      "learning_rate": 9.81146493157558e-06,
      "loss": 1.1014,
      "step": 2848
    },
    {
      "epoch": 0.23022687327016708,
      "grad_norm": 2.4997401237487793,
      "learning_rate": 9.811286893995014e-06,
      "loss": 1.0001,
      "step": 2849
    },
    {
      "epoch": 0.2303076829834946,
      "grad_norm": 2.8787453174591064,
      "learning_rate": 9.811108774008689e-06,
      "loss": 0.9882,
      "step": 2850
    },
    {
      "epoch": 0.23038849269682216,
      "grad_norm": 2.824585437774658,
      "learning_rate": 9.810930571619652e-06,
      "loss": 0.9349,
      "step": 2851
    },
    {
      "epoch": 0.2304693024101497,
      "grad_norm": 3.053377151489258,
      "learning_rate": 9.810752286830958e-06,
      "loss": 0.8595,
      "step": 2852
    },
    {
      "epoch": 0.23055011212347723,
      "grad_norm": 2.454207420349121,
      "learning_rate": 9.810573919645658e-06,
      "loss": 0.9653,
      "step": 2853
    },
    {
      "epoch": 0.23063092183680478,
      "grad_norm": 3.3735809326171875,
      "learning_rate": 9.810395470066807e-06,
      "loss": 0.978,
      "step": 2854
    },
    {
      "epoch": 0.23071173155013233,
      "grad_norm": 2.8116884231567383,
      "learning_rate": 9.810216938097463e-06,
      "loss": 1.046,
      "step": 2855
    },
    {
      "epoch": 0.23079254126345985,
      "grad_norm": 3.1904733180999756,
      "learning_rate": 9.810038323740683e-06,
      "loss": 1.0383,
      "step": 2856
    },
    {
      "epoch": 0.2308733509767874,
      "grad_norm": 3.200434923171997,
      "learning_rate": 9.809859626999526e-06,
      "loss": 0.9868,
      "step": 2857
    },
    {
      "epoch": 0.23095416069011496,
      "grad_norm": 2.598670721054077,
      "learning_rate": 9.809680847877052e-06,
      "loss": 0.9273,
      "step": 2858
    },
    {
      "epoch": 0.23103497040344248,
      "grad_norm": 3.1879770755767822,
      "learning_rate": 9.809501986376324e-06,
      "loss": 1.1089,
      "step": 2859
    },
    {
      "epoch": 0.23111578011677003,
      "grad_norm": 2.736104726791382,
      "learning_rate": 9.809323042500406e-06,
      "loss": 1.1119,
      "step": 2860
    },
    {
      "epoch": 0.23119658983009758,
      "grad_norm": 2.963679075241089,
      "learning_rate": 9.809144016252361e-06,
      "loss": 0.9761,
      "step": 2861
    },
    {
      "epoch": 0.23127739954342513,
      "grad_norm": 2.912889242172241,
      "learning_rate": 9.808964907635258e-06,
      "loss": 1.0861,
      "step": 2862
    },
    {
      "epoch": 0.23135820925675266,
      "grad_norm": 2.7533481121063232,
      "learning_rate": 9.808785716652163e-06,
      "loss": 1.0043,
      "step": 2863
    },
    {
      "epoch": 0.2314390189700802,
      "grad_norm": 2.922305107116699,
      "learning_rate": 9.808606443306146e-06,
      "loss": 0.9635,
      "step": 2864
    },
    {
      "epoch": 0.23151982868340776,
      "grad_norm": 3.466017961502075,
      "learning_rate": 9.808427087600276e-06,
      "loss": 1.0228,
      "step": 2865
    },
    {
      "epoch": 0.23160063839673528,
      "grad_norm": 2.7406976222991943,
      "learning_rate": 9.808247649537626e-06,
      "loss": 1.0092,
      "step": 2866
    },
    {
      "epoch": 0.23168144811006283,
      "grad_norm": 2.6772921085357666,
      "learning_rate": 9.808068129121268e-06,
      "loss": 1.0089,
      "step": 2867
    },
    {
      "epoch": 0.23176225782339038,
      "grad_norm": 2.930373191833496,
      "learning_rate": 9.80788852635428e-06,
      "loss": 0.9548,
      "step": 2868
    },
    {
      "epoch": 0.2318430675367179,
      "grad_norm": 2.7349531650543213,
      "learning_rate": 9.807708841239734e-06,
      "loss": 1.0062,
      "step": 2869
    },
    {
      "epoch": 0.23192387725004546,
      "grad_norm": 2.849152088165283,
      "learning_rate": 9.807529073780712e-06,
      "loss": 0.9175,
      "step": 2870
    },
    {
      "epoch": 0.232004686963373,
      "grad_norm": 2.6046605110168457,
      "learning_rate": 9.80734922398029e-06,
      "loss": 1.0255,
      "step": 2871
    },
    {
      "epoch": 0.23208549667670053,
      "grad_norm": 2.920684337615967,
      "learning_rate": 9.807169291841548e-06,
      "loss": 0.9472,
      "step": 2872
    },
    {
      "epoch": 0.23216630639002808,
      "grad_norm": 3.0803701877593994,
      "learning_rate": 9.806989277367569e-06,
      "loss": 1.1611,
      "step": 2873
    },
    {
      "epoch": 0.23224711610335563,
      "grad_norm": 3.029608964920044,
      "learning_rate": 9.806809180561436e-06,
      "loss": 0.9328,
      "step": 2874
    },
    {
      "epoch": 0.23232792581668316,
      "grad_norm": 2.641021966934204,
      "learning_rate": 9.806629001426234e-06,
      "loss": 0.9521,
      "step": 2875
    },
    {
      "epoch": 0.2324087355300107,
      "grad_norm": 2.828425884246826,
      "learning_rate": 9.806448739965048e-06,
      "loss": 0.9299,
      "step": 2876
    },
    {
      "epoch": 0.23248954524333826,
      "grad_norm": 2.696329355239868,
      "learning_rate": 9.806268396180967e-06,
      "loss": 0.9952,
      "step": 2877
    },
    {
      "epoch": 0.23257035495666578,
      "grad_norm": 2.6079211235046387,
      "learning_rate": 9.806087970077079e-06,
      "loss": 0.9596,
      "step": 2878
    },
    {
      "epoch": 0.23265116466999333,
      "grad_norm": 3.1535024642944336,
      "learning_rate": 9.805907461656473e-06,
      "loss": 0.8422,
      "step": 2879
    },
    {
      "epoch": 0.23273197438332088,
      "grad_norm": 2.7458996772766113,
      "learning_rate": 9.805726870922244e-06,
      "loss": 1.0438,
      "step": 2880
    },
    {
      "epoch": 0.2328127840966484,
      "grad_norm": 2.8035130500793457,
      "learning_rate": 9.80554619787748e-06,
      "loss": 1.1018,
      "step": 2881
    },
    {
      "epoch": 0.23289359380997596,
      "grad_norm": 2.688476085662842,
      "learning_rate": 9.80536544252528e-06,
      "loss": 0.9787,
      "step": 2882
    },
    {
      "epoch": 0.2329744035233035,
      "grad_norm": 3.001877784729004,
      "learning_rate": 9.80518460486874e-06,
      "loss": 1.12,
      "step": 2883
    },
    {
      "epoch": 0.23305521323663103,
      "grad_norm": 2.7390987873077393,
      "learning_rate": 9.805003684910955e-06,
      "loss": 0.9534,
      "step": 2884
    },
    {
      "epoch": 0.23313602294995858,
      "grad_norm": 3.121856451034546,
      "learning_rate": 9.804822682655023e-06,
      "loss": 1.0815,
      "step": 2885
    },
    {
      "epoch": 0.23321683266328613,
      "grad_norm": 2.5415515899658203,
      "learning_rate": 9.804641598104048e-06,
      "loss": 1.1259,
      "step": 2886
    },
    {
      "epoch": 0.23329764237661366,
      "grad_norm": 2.9589216709136963,
      "learning_rate": 9.804460431261128e-06,
      "loss": 0.9394,
      "step": 2887
    },
    {
      "epoch": 0.2333784520899412,
      "grad_norm": 2.749288320541382,
      "learning_rate": 9.804279182129366e-06,
      "loss": 0.9642,
      "step": 2888
    },
    {
      "epoch": 0.23345926180326876,
      "grad_norm": 2.368316650390625,
      "learning_rate": 9.804097850711867e-06,
      "loss": 1.0122,
      "step": 2889
    },
    {
      "epoch": 0.23354007151659628,
      "grad_norm": 3.158442258834839,
      "learning_rate": 9.80391643701174e-06,
      "loss": 0.9345,
      "step": 2890
    },
    {
      "epoch": 0.23362088122992383,
      "grad_norm": 2.4102766513824463,
      "learning_rate": 9.803734941032087e-06,
      "loss": 1.0051,
      "step": 2891
    },
    {
      "epoch": 0.23370169094325138,
      "grad_norm": 2.6964380741119385,
      "learning_rate": 9.803553362776019e-06,
      "loss": 1.1063,
      "step": 2892
    },
    {
      "epoch": 0.23378250065657893,
      "grad_norm": 3.018733024597168,
      "learning_rate": 9.803371702246647e-06,
      "loss": 0.9383,
      "step": 2893
    },
    {
      "epoch": 0.23386331036990646,
      "grad_norm": 2.8816728591918945,
      "learning_rate": 9.803189959447082e-06,
      "loss": 0.9539,
      "step": 2894
    },
    {
      "epoch": 0.233944120083234,
      "grad_norm": 2.8543145656585693,
      "learning_rate": 9.803008134380435e-06,
      "loss": 0.9333,
      "step": 2895
    },
    {
      "epoch": 0.23402492979656156,
      "grad_norm": 2.6977081298828125,
      "learning_rate": 9.802826227049822e-06,
      "loss": 1.0031,
      "step": 2896
    },
    {
      "epoch": 0.23410573950988908,
      "grad_norm": 2.8548684120178223,
      "learning_rate": 9.802644237458357e-06,
      "loss": 1.045,
      "step": 2897
    },
    {
      "epoch": 0.23418654922321663,
      "grad_norm": 3.118274688720703,
      "learning_rate": 9.802462165609159e-06,
      "loss": 1.0571,
      "step": 2898
    },
    {
      "epoch": 0.23426735893654418,
      "grad_norm": 2.4857306480407715,
      "learning_rate": 9.802280011505345e-06,
      "loss": 1.016,
      "step": 2899
    },
    {
      "epoch": 0.2343481686498717,
      "grad_norm": 3.076098680496216,
      "learning_rate": 9.802097775150037e-06,
      "loss": 1.011,
      "step": 2900
    },
    {
      "epoch": 0.23442897836319926,
      "grad_norm": 2.630444288253784,
      "learning_rate": 9.801915456546353e-06,
      "loss": 0.976,
      "step": 2901
    },
    {
      "epoch": 0.2345097880765268,
      "grad_norm": 2.6252129077911377,
      "learning_rate": 9.801733055697417e-06,
      "loss": 0.99,
      "step": 2902
    },
    {
      "epoch": 0.23459059778985433,
      "grad_norm": 2.6402323246002197,
      "learning_rate": 9.801550572606355e-06,
      "loss": 0.9066,
      "step": 2903
    },
    {
      "epoch": 0.23467140750318188,
      "grad_norm": 2.880779981613159,
      "learning_rate": 9.80136800727629e-06,
      "loss": 0.9364,
      "step": 2904
    },
    {
      "epoch": 0.23475221721650943,
      "grad_norm": 2.7739346027374268,
      "learning_rate": 9.801185359710352e-06,
      "loss": 1.0044,
      "step": 2905
    },
    {
      "epoch": 0.23483302692983696,
      "grad_norm": 2.8313372135162354,
      "learning_rate": 9.801002629911664e-06,
      "loss": 0.9938,
      "step": 2906
    },
    {
      "epoch": 0.2349138366431645,
      "grad_norm": 3.302762508392334,
      "learning_rate": 9.800819817883362e-06,
      "loss": 1.0321,
      "step": 2907
    },
    {
      "epoch": 0.23499464635649206,
      "grad_norm": 2.7084765434265137,
      "learning_rate": 9.800636923628572e-06,
      "loss": 1.0704,
      "step": 2908
    },
    {
      "epoch": 0.23507545606981958,
      "grad_norm": 2.904644727706909,
      "learning_rate": 9.800453947150427e-06,
      "loss": 0.9729,
      "step": 2909
    },
    {
      "epoch": 0.23515626578314713,
      "grad_norm": 2.632418394088745,
      "learning_rate": 9.800270888452065e-06,
      "loss": 1.0213,
      "step": 2910
    },
    {
      "epoch": 0.23523707549647468,
      "grad_norm": 3.0656416416168213,
      "learning_rate": 9.80008774753662e-06,
      "loss": 0.9299,
      "step": 2911
    },
    {
      "epoch": 0.2353178852098022,
      "grad_norm": 2.863532781600952,
      "learning_rate": 9.799904524407224e-06,
      "loss": 0.9577,
      "step": 2912
    },
    {
      "epoch": 0.23539869492312976,
      "grad_norm": 2.7166807651519775,
      "learning_rate": 9.799721219067023e-06,
      "loss": 0.9382,
      "step": 2913
    },
    {
      "epoch": 0.2354795046364573,
      "grad_norm": 2.9613704681396484,
      "learning_rate": 9.799537831519149e-06,
      "loss": 1.0444,
      "step": 2914
    },
    {
      "epoch": 0.23556031434978483,
      "grad_norm": 3.0312464237213135,
      "learning_rate": 9.799354361766746e-06,
      "loss": 1.0073,
      "step": 2915
    },
    {
      "epoch": 0.23564112406311238,
      "grad_norm": 2.47131609916687,
      "learning_rate": 9.79917080981296e-06,
      "loss": 0.9809,
      "step": 2916
    },
    {
      "epoch": 0.23572193377643993,
      "grad_norm": 2.8032500743865967,
      "learning_rate": 9.798987175660928e-06,
      "loss": 1.1325,
      "step": 2917
    },
    {
      "epoch": 0.23580274348976746,
      "grad_norm": 2.947420358657837,
      "learning_rate": 9.798803459313802e-06,
      "loss": 1.1373,
      "step": 2918
    },
    {
      "epoch": 0.235883553203095,
      "grad_norm": 3.7523951530456543,
      "learning_rate": 9.798619660774724e-06,
      "loss": 1.089,
      "step": 2919
    },
    {
      "epoch": 0.23596436291642256,
      "grad_norm": 2.727415084838867,
      "learning_rate": 9.798435780046842e-06,
      "loss": 1.1135,
      "step": 2920
    },
    {
      "epoch": 0.23604517262975008,
      "grad_norm": 2.7972569465637207,
      "learning_rate": 9.79825181713331e-06,
      "loss": 0.9596,
      "step": 2921
    },
    {
      "epoch": 0.23612598234307763,
      "grad_norm": 2.521152973175049,
      "learning_rate": 9.798067772037272e-06,
      "loss": 0.985,
      "step": 2922
    },
    {
      "epoch": 0.23620679205640518,
      "grad_norm": 2.7290236949920654,
      "learning_rate": 9.797883644761886e-06,
      "loss": 1.0603,
      "step": 2923
    },
    {
      "epoch": 0.2362876017697327,
      "grad_norm": 2.937466621398926,
      "learning_rate": 9.797699435310305e-06,
      "loss": 0.9835,
      "step": 2924
    },
    {
      "epoch": 0.23636841148306026,
      "grad_norm": 2.8509199619293213,
      "learning_rate": 9.79751514368568e-06,
      "loss": 0.9966,
      "step": 2925
    },
    {
      "epoch": 0.2364492211963878,
      "grad_norm": 2.9798872470855713,
      "learning_rate": 9.79733076989117e-06,
      "loss": 0.9335,
      "step": 2926
    },
    {
      "epoch": 0.23653003090971536,
      "grad_norm": 2.9888782501220703,
      "learning_rate": 9.797146313929935e-06,
      "loss": 1.0645,
      "step": 2927
    },
    {
      "epoch": 0.23661084062304288,
      "grad_norm": 2.9092607498168945,
      "learning_rate": 9.796961775805131e-06,
      "loss": 0.9685,
      "step": 2928
    },
    {
      "epoch": 0.23669165033637043,
      "grad_norm": 2.904297113418579,
      "learning_rate": 9.796777155519921e-06,
      "loss": 1.1148,
      "step": 2929
    },
    {
      "epoch": 0.23677246004969799,
      "grad_norm": 2.948443651199341,
      "learning_rate": 9.796592453077466e-06,
      "loss": 1.0204,
      "step": 2930
    },
    {
      "epoch": 0.2368532697630255,
      "grad_norm": 2.9255568981170654,
      "learning_rate": 9.79640766848093e-06,
      "loss": 0.9652,
      "step": 2931
    },
    {
      "epoch": 0.23693407947635306,
      "grad_norm": 2.946317195892334,
      "learning_rate": 9.796222801733476e-06,
      "loss": 0.9634,
      "step": 2932
    },
    {
      "epoch": 0.2370148891896806,
      "grad_norm": 2.8127524852752686,
      "learning_rate": 9.79603785283827e-06,
      "loss": 0.9098,
      "step": 2933
    },
    {
      "epoch": 0.23709569890300813,
      "grad_norm": 2.9615817070007324,
      "learning_rate": 9.795852821798486e-06,
      "loss": 0.9135,
      "step": 2934
    },
    {
      "epoch": 0.23717650861633569,
      "grad_norm": 2.7446794509887695,
      "learning_rate": 9.795667708617287e-06,
      "loss": 1.1187,
      "step": 2935
    },
    {
      "epoch": 0.23725731832966324,
      "grad_norm": 2.8557610511779785,
      "learning_rate": 9.795482513297845e-06,
      "loss": 1.1194,
      "step": 2936
    },
    {
      "epoch": 0.23733812804299076,
      "grad_norm": 2.5263028144836426,
      "learning_rate": 9.795297235843333e-06,
      "loss": 0.9522,
      "step": 2937
    },
    {
      "epoch": 0.2374189377563183,
      "grad_norm": 2.5753190517425537,
      "learning_rate": 9.795111876256921e-06,
      "loss": 1.0271,
      "step": 2938
    },
    {
      "epoch": 0.23749974746964586,
      "grad_norm": 2.882373809814453,
      "learning_rate": 9.79492643454179e-06,
      "loss": 0.9694,
      "step": 2939
    },
    {
      "epoch": 0.23758055718297338,
      "grad_norm": 2.4024150371551514,
      "learning_rate": 9.794740910701111e-06,
      "loss": 1.0339,
      "step": 2940
    },
    {
      "epoch": 0.23766136689630094,
      "grad_norm": 3.4500327110290527,
      "learning_rate": 9.794555304738063e-06,
      "loss": 0.9981,
      "step": 2941
    },
    {
      "epoch": 0.23774217660962849,
      "grad_norm": 2.7431914806365967,
      "learning_rate": 9.794369616655823e-06,
      "loss": 1.0637,
      "step": 2942
    },
    {
      "epoch": 0.237822986322956,
      "grad_norm": 2.808762550354004,
      "learning_rate": 9.794183846457577e-06,
      "loss": 0.9553,
      "step": 2943
    },
    {
      "epoch": 0.23790379603628356,
      "grad_norm": 2.5411794185638428,
      "learning_rate": 9.7939979941465e-06,
      "loss": 0.9299,
      "step": 2944
    },
    {
      "epoch": 0.2379846057496111,
      "grad_norm": 2.707524299621582,
      "learning_rate": 9.793812059725781e-06,
      "loss": 1.028,
      "step": 2945
    },
    {
      "epoch": 0.23806541546293863,
      "grad_norm": 2.967235565185547,
      "learning_rate": 9.7936260431986e-06,
      "loss": 1.0642,
      "step": 2946
    },
    {
      "epoch": 0.23814622517626619,
      "grad_norm": 2.7679848670959473,
      "learning_rate": 9.793439944568146e-06,
      "loss": 0.9546,
      "step": 2947
    },
    {
      "epoch": 0.23822703488959374,
      "grad_norm": 2.782672166824341,
      "learning_rate": 9.793253763837606e-06,
      "loss": 1.0416,
      "step": 2948
    },
    {
      "epoch": 0.23830784460292126,
      "grad_norm": 3.0185306072235107,
      "learning_rate": 9.793067501010167e-06,
      "loss": 1.0117,
      "step": 2949
    },
    {
      "epoch": 0.2383886543162488,
      "grad_norm": 2.802870035171509,
      "learning_rate": 9.792881156089023e-06,
      "loss": 1.0281,
      "step": 2950
    },
    {
      "epoch": 0.23846946402957636,
      "grad_norm": 2.8822507858276367,
      "learning_rate": 9.79269472907736e-06,
      "loss": 1.0046,
      "step": 2951
    },
    {
      "epoch": 0.23855027374290388,
      "grad_norm": 3.2563014030456543,
      "learning_rate": 9.792508219978377e-06,
      "loss": 1.0214,
      "step": 2952
    },
    {
      "epoch": 0.23863108345623144,
      "grad_norm": 2.619318962097168,
      "learning_rate": 9.792321628795264e-06,
      "loss": 1.0024,
      "step": 2953
    },
    {
      "epoch": 0.238711893169559,
      "grad_norm": 2.830854654312134,
      "learning_rate": 9.792134955531219e-06,
      "loss": 0.944,
      "step": 2954
    },
    {
      "epoch": 0.2387927028828865,
      "grad_norm": 3.4874989986419678,
      "learning_rate": 9.791948200189439e-06,
      "loss": 1.0637,
      "step": 2955
    },
    {
      "epoch": 0.23887351259621406,
      "grad_norm": 2.825983762741089,
      "learning_rate": 9.791761362773122e-06,
      "loss": 0.8994,
      "step": 2956
    },
    {
      "epoch": 0.2389543223095416,
      "grad_norm": 2.640371799468994,
      "learning_rate": 9.791574443285469e-06,
      "loss": 0.9939,
      "step": 2957
    },
    {
      "epoch": 0.23903513202286916,
      "grad_norm": 2.5753226280212402,
      "learning_rate": 9.791387441729681e-06,
      "loss": 1.192,
      "step": 2958
    },
    {
      "epoch": 0.23911594173619669,
      "grad_norm": 3.058668375015259,
      "learning_rate": 9.79120035810896e-06,
      "loss": 0.9996,
      "step": 2959
    },
    {
      "epoch": 0.23919675144952424,
      "grad_norm": 2.856213331222534,
      "learning_rate": 9.791013192426513e-06,
      "loss": 1.0295,
      "step": 2960
    },
    {
      "epoch": 0.2392775611628518,
      "grad_norm": 2.7181801795959473,
      "learning_rate": 9.790825944685542e-06,
      "loss": 0.9916,
      "step": 2961
    },
    {
      "epoch": 0.2393583708761793,
      "grad_norm": 2.5401663780212402,
      "learning_rate": 9.790638614889256e-06,
      "loss": 0.9128,
      "step": 2962
    },
    {
      "epoch": 0.23943918058950686,
      "grad_norm": 2.5197436809539795,
      "learning_rate": 9.790451203040865e-06,
      "loss": 0.9002,
      "step": 2963
    },
    {
      "epoch": 0.2395199903028344,
      "grad_norm": 2.8694779872894287,
      "learning_rate": 9.790263709143577e-06,
      "loss": 0.8323,
      "step": 2964
    },
    {
      "epoch": 0.23960080001616194,
      "grad_norm": 2.646378755569458,
      "learning_rate": 9.790076133200604e-06,
      "loss": 1.0607,
      "step": 2965
    },
    {
      "epoch": 0.2396816097294895,
      "grad_norm": 2.950759172439575,
      "learning_rate": 9.789888475215158e-06,
      "loss": 0.8371,
      "step": 2966
    },
    {
      "epoch": 0.23976241944281704,
      "grad_norm": 2.921673059463501,
      "learning_rate": 9.789700735190453e-06,
      "loss": 0.9896,
      "step": 2967
    },
    {
      "epoch": 0.23984322915614456,
      "grad_norm": 3.2484123706817627,
      "learning_rate": 9.789512913129706e-06,
      "loss": 0.8615,
      "step": 2968
    },
    {
      "epoch": 0.2399240388694721,
      "grad_norm": 2.5600643157958984,
      "learning_rate": 9.789325009036134e-06,
      "loss": 0.9869,
      "step": 2969
    },
    {
      "epoch": 0.24000484858279966,
      "grad_norm": 2.513324499130249,
      "learning_rate": 9.789137022912953e-06,
      "loss": 1.014,
      "step": 2970
    },
    {
      "epoch": 0.24008565829612719,
      "grad_norm": 3.612118721008301,
      "learning_rate": 9.788948954763385e-06,
      "loss": 0.9303,
      "step": 2971
    },
    {
      "epoch": 0.24016646800945474,
      "grad_norm": 2.5807790756225586,
      "learning_rate": 9.78876080459065e-06,
      "loss": 1.0551,
      "step": 2972
    },
    {
      "epoch": 0.2402472777227823,
      "grad_norm": 2.8657569885253906,
      "learning_rate": 9.788572572397969e-06,
      "loss": 0.8713,
      "step": 2973
    },
    {
      "epoch": 0.2403280874361098,
      "grad_norm": 3.181068181991577,
      "learning_rate": 9.78838425818857e-06,
      "loss": 0.917,
      "step": 2974
    },
    {
      "epoch": 0.24040889714943736,
      "grad_norm": 3.0996499061584473,
      "learning_rate": 9.788195861965678e-06,
      "loss": 0.9985,
      "step": 2975
    },
    {
      "epoch": 0.2404897068627649,
      "grad_norm": 2.7109224796295166,
      "learning_rate": 9.788007383732514e-06,
      "loss": 0.995,
      "step": 2976
    },
    {
      "epoch": 0.24057051657609244,
      "grad_norm": 2.801605463027954,
      "learning_rate": 9.787818823492312e-06,
      "loss": 0.961,
      "step": 2977
    },
    {
      "epoch": 0.24065132628942,
      "grad_norm": 2.9663140773773193,
      "learning_rate": 9.7876301812483e-06,
      "loss": 0.9846,
      "step": 2978
    },
    {
      "epoch": 0.24073213600274754,
      "grad_norm": 2.6364505290985107,
      "learning_rate": 9.787441457003709e-06,
      "loss": 1.1276,
      "step": 2979
    },
    {
      "epoch": 0.24081294571607506,
      "grad_norm": 2.5775394439697266,
      "learning_rate": 9.78725265076177e-06,
      "loss": 0.9062,
      "step": 2980
    },
    {
      "epoch": 0.2408937554294026,
      "grad_norm": 3.2745320796966553,
      "learning_rate": 9.787063762525717e-06,
      "loss": 0.9921,
      "step": 2981
    },
    {
      "epoch": 0.24097456514273016,
      "grad_norm": 2.4596545696258545,
      "learning_rate": 9.786874792298788e-06,
      "loss": 0.9798,
      "step": 2982
    },
    {
      "epoch": 0.24105537485605769,
      "grad_norm": 2.857215166091919,
      "learning_rate": 9.786685740084219e-06,
      "loss": 0.9625,
      "step": 2983
    },
    {
      "epoch": 0.24113618456938524,
      "grad_norm": 2.645327568054199,
      "learning_rate": 9.786496605885245e-06,
      "loss": 0.9486,
      "step": 2984
    },
    {
      "epoch": 0.2412169942827128,
      "grad_norm": 2.879042148590088,
      "learning_rate": 9.786307389705108e-06,
      "loss": 0.9191,
      "step": 2985
    },
    {
      "epoch": 0.2412978039960403,
      "grad_norm": 3.1000540256500244,
      "learning_rate": 9.786118091547045e-06,
      "loss": 1.0089,
      "step": 2986
    },
    {
      "epoch": 0.24137861370936786,
      "grad_norm": 3.00303316116333,
      "learning_rate": 9.785928711414306e-06,
      "loss": 0.9263,
      "step": 2987
    },
    {
      "epoch": 0.2414594234226954,
      "grad_norm": 2.8019394874572754,
      "learning_rate": 9.785739249310126e-06,
      "loss": 1.0141,
      "step": 2988
    },
    {
      "epoch": 0.24154023313602296,
      "grad_norm": 2.7068896293640137,
      "learning_rate": 9.785549705237755e-06,
      "loss": 0.9381,
      "step": 2989
    },
    {
      "epoch": 0.2416210428493505,
      "grad_norm": 2.8227250576019287,
      "learning_rate": 9.785360079200439e-06,
      "loss": 0.9671,
      "step": 2990
    },
    {
      "epoch": 0.24170185256267804,
      "grad_norm": 3.5661299228668213,
      "learning_rate": 9.785170371201424e-06,
      "loss": 0.9556,
      "step": 2991
    },
    {
      "epoch": 0.2417826622760056,
      "grad_norm": 2.823042392730713,
      "learning_rate": 9.784980581243962e-06,
      "loss": 1.0224,
      "step": 2992
    },
    {
      "epoch": 0.2418634719893331,
      "grad_norm": 2.84657621383667,
      "learning_rate": 9.7847907093313e-06,
      "loss": 0.8715,
      "step": 2993
    },
    {
      "epoch": 0.24194428170266066,
      "grad_norm": 3.0351145267486572,
      "learning_rate": 9.784600755466693e-06,
      "loss": 0.9789,
      "step": 2994
    },
    {
      "epoch": 0.2420250914159882,
      "grad_norm": 2.847142219543457,
      "learning_rate": 9.784410719653395e-06,
      "loss": 1.0276,
      "step": 2995
    },
    {
      "epoch": 0.24210590112931574,
      "grad_norm": 3.054438352584839,
      "learning_rate": 9.784220601894656e-06,
      "loss": 0.9257,
      "step": 2996
    },
    {
      "epoch": 0.2421867108426433,
      "grad_norm": 2.8259174823760986,
      "learning_rate": 9.784030402193737e-06,
      "loss": 0.9961,
      "step": 2997
    },
    {
      "epoch": 0.24226752055597084,
      "grad_norm": 2.6278364658355713,
      "learning_rate": 9.783840120553895e-06,
      "loss": 0.9693,
      "step": 2998
    },
    {
      "epoch": 0.24234833026929836,
      "grad_norm": 2.6503517627716064,
      "learning_rate": 9.78364975697839e-06,
      "loss": 1.0962,
      "step": 2999
    },
    {
      "epoch": 0.2424291399826259,
      "grad_norm": 2.714247941970825,
      "learning_rate": 9.783459311470478e-06,
      "loss": 0.9403,
      "step": 3000
    },
    {
      "epoch": 0.2424291399826259,
      "eval_loss": 0.8388580083847046,
      "eval_runtime": 814.4038,
      "eval_samples_per_second": 102.364,
      "eval_steps_per_second": 12.796,
      "step": 3000
    },
    {
      "epoch": 0.24250994969595346,
      "grad_norm": 2.9535956382751465,
      "learning_rate": 9.783268784033426e-06,
      "loss": 0.9662,
      "step": 3001
    },
    {
      "epoch": 0.242590759409281,
      "grad_norm": 2.770203113555908,
      "learning_rate": 9.783078174670492e-06,
      "loss": 0.9747,
      "step": 3002
    },
    {
      "epoch": 0.24267156912260854,
      "grad_norm": 2.815793514251709,
      "learning_rate": 9.782887483384946e-06,
      "loss": 1.0265,
      "step": 3003
    },
    {
      "epoch": 0.2427523788359361,
      "grad_norm": 2.5195326805114746,
      "learning_rate": 9.782696710180051e-06,
      "loss": 1.0203,
      "step": 3004
    },
    {
      "epoch": 0.2428331885492636,
      "grad_norm": 3.260784387588501,
      "learning_rate": 9.782505855059076e-06,
      "loss": 0.9573,
      "step": 3005
    },
    {
      "epoch": 0.24291399826259116,
      "grad_norm": 2.7619948387145996,
      "learning_rate": 9.782314918025289e-06,
      "loss": 1.0384,
      "step": 3006
    },
    {
      "epoch": 0.24299480797591871,
      "grad_norm": 2.631403684616089,
      "learning_rate": 9.782123899081958e-06,
      "loss": 0.984,
      "step": 3007
    },
    {
      "epoch": 0.24307561768924624,
      "grad_norm": 3.0094573497772217,
      "learning_rate": 9.781932798232362e-06,
      "loss": 1.1815,
      "step": 3008
    },
    {
      "epoch": 0.2431564274025738,
      "grad_norm": 2.775972366333008,
      "learning_rate": 9.781741615479764e-06,
      "loss": 0.9415,
      "step": 3009
    },
    {
      "epoch": 0.24323723711590134,
      "grad_norm": 2.6309051513671875,
      "learning_rate": 9.781550350827446e-06,
      "loss": 0.9621,
      "step": 3010
    },
    {
      "epoch": 0.24331804682922886,
      "grad_norm": 2.8019237518310547,
      "learning_rate": 9.78135900427868e-06,
      "loss": 0.9563,
      "step": 3011
    },
    {
      "epoch": 0.2433988565425564,
      "grad_norm": 3.1115059852600098,
      "learning_rate": 9.781167575836747e-06,
      "loss": 0.9576,
      "step": 3012
    },
    {
      "epoch": 0.24347966625588396,
      "grad_norm": 2.668731212615967,
      "learning_rate": 9.780976065504923e-06,
      "loss": 0.9748,
      "step": 3013
    },
    {
      "epoch": 0.2435604759692115,
      "grad_norm": 3.3178939819335938,
      "learning_rate": 9.78078447328649e-06,
      "loss": 0.9354,
      "step": 3014
    },
    {
      "epoch": 0.24364128568253904,
      "grad_norm": 3.220385789871216,
      "learning_rate": 9.780592799184728e-06,
      "loss": 0.9475,
      "step": 3015
    },
    {
      "epoch": 0.2437220953958666,
      "grad_norm": 2.9032740592956543,
      "learning_rate": 9.780401043202919e-06,
      "loss": 0.949,
      "step": 3016
    },
    {
      "epoch": 0.2438029051091941,
      "grad_norm": 3.000532627105713,
      "learning_rate": 9.780209205344347e-06,
      "loss": 1.0839,
      "step": 3017
    },
    {
      "epoch": 0.24388371482252166,
      "grad_norm": 2.539182424545288,
      "learning_rate": 9.780017285612303e-06,
      "loss": 0.9495,
      "step": 3018
    },
    {
      "epoch": 0.24396452453584921,
      "grad_norm": 2.2839114665985107,
      "learning_rate": 9.779825284010067e-06,
      "loss": 0.974,
      "step": 3019
    },
    {
      "epoch": 0.24404533424917674,
      "grad_norm": 2.8400840759277344,
      "learning_rate": 9.779633200540933e-06,
      "loss": 0.9371,
      "step": 3020
    },
    {
      "epoch": 0.2441261439625043,
      "grad_norm": 2.7170376777648926,
      "learning_rate": 9.779441035208185e-06,
      "loss": 1.0756,
      "step": 3021
    },
    {
      "epoch": 0.24420695367583184,
      "grad_norm": 2.4907898902893066,
      "learning_rate": 9.779248788015123e-06,
      "loss": 1.0196,
      "step": 3022
    },
    {
      "epoch": 0.2442877633891594,
      "grad_norm": 3.175201654434204,
      "learning_rate": 9.779056458965032e-06,
      "loss": 1.061,
      "step": 3023
    },
    {
      "epoch": 0.2443685731024869,
      "grad_norm": 2.583282709121704,
      "learning_rate": 9.778864048061209e-06,
      "loss": 0.9611,
      "step": 3024
    },
    {
      "epoch": 0.24444938281581446,
      "grad_norm": 2.6285994052886963,
      "learning_rate": 9.77867155530695e-06,
      "loss": 1.0399,
      "step": 3025
    },
    {
      "epoch": 0.24453019252914202,
      "grad_norm": 3.0164005756378174,
      "learning_rate": 9.778478980705552e-06,
      "loss": 0.9734,
      "step": 3026
    },
    {
      "epoch": 0.24461100224246954,
      "grad_norm": 2.634148359298706,
      "learning_rate": 9.778286324260314e-06,
      "loss": 1.073,
      "step": 3027
    },
    {
      "epoch": 0.2446918119557971,
      "grad_norm": 2.6194915771484375,
      "learning_rate": 9.778093585974531e-06,
      "loss": 0.9263,
      "step": 3028
    },
    {
      "epoch": 0.24477262166912464,
      "grad_norm": 2.734647512435913,
      "learning_rate": 9.77790076585151e-06,
      "loss": 1.0295,
      "step": 3029
    },
    {
      "epoch": 0.24485343138245216,
      "grad_norm": 3.3778486251831055,
      "learning_rate": 9.777707863894551e-06,
      "loss": 1.0255,
      "step": 3030
    },
    {
      "epoch": 0.24493424109577971,
      "grad_norm": 2.509401798248291,
      "learning_rate": 9.777514880106957e-06,
      "loss": 0.9847,
      "step": 3031
    },
    {
      "epoch": 0.24501505080910727,
      "grad_norm": 3.0848164558410645,
      "learning_rate": 9.777321814492036e-06,
      "loss": 0.9488,
      "step": 3032
    },
    {
      "epoch": 0.2450958605224348,
      "grad_norm": 2.637641191482544,
      "learning_rate": 9.777128667053093e-06,
      "loss": 0.8775,
      "step": 3033
    },
    {
      "epoch": 0.24517667023576234,
      "grad_norm": 2.3294782638549805,
      "learning_rate": 9.776935437793436e-06,
      "loss": 1.1534,
      "step": 3034
    },
    {
      "epoch": 0.2452574799490899,
      "grad_norm": 2.6865875720977783,
      "learning_rate": 9.776742126716374e-06,
      "loss": 0.846,
      "step": 3035
    },
    {
      "epoch": 0.2453382896624174,
      "grad_norm": 2.6133646965026855,
      "learning_rate": 9.77654873382522e-06,
      "loss": 1.0109,
      "step": 3036
    },
    {
      "epoch": 0.24541909937574496,
      "grad_norm": 2.9119057655334473,
      "learning_rate": 9.776355259123286e-06,
      "loss": 1.0162,
      "step": 3037
    },
    {
      "epoch": 0.24549990908907252,
      "grad_norm": 3.213521718978882,
      "learning_rate": 9.776161702613884e-06,
      "loss": 0.8763,
      "step": 3038
    },
    {
      "epoch": 0.24558071880240004,
      "grad_norm": 2.7100210189819336,
      "learning_rate": 9.775968064300331e-06,
      "loss": 1.001,
      "step": 3039
    },
    {
      "epoch": 0.2456615285157276,
      "grad_norm": 2.9769105911254883,
      "learning_rate": 9.775774344185942e-06,
      "loss": 1.0154,
      "step": 3040
    },
    {
      "epoch": 0.24574233822905514,
      "grad_norm": 3.063699960708618,
      "learning_rate": 9.775580542274035e-06,
      "loss": 0.9613,
      "step": 3041
    },
    {
      "epoch": 0.24582314794238266,
      "grad_norm": 2.744769334793091,
      "learning_rate": 9.775386658567931e-06,
      "loss": 1.007,
      "step": 3042
    },
    {
      "epoch": 0.24590395765571021,
      "grad_norm": 2.8555009365081787,
      "learning_rate": 9.775192693070949e-06,
      "loss": 0.9691,
      "step": 3043
    },
    {
      "epoch": 0.24598476736903777,
      "grad_norm": 2.6923105716705322,
      "learning_rate": 9.774998645786413e-06,
      "loss": 1.0184,
      "step": 3044
    },
    {
      "epoch": 0.2460655770823653,
      "grad_norm": 2.7845826148986816,
      "learning_rate": 9.774804516717646e-06,
      "loss": 0.8735,
      "step": 3045
    },
    {
      "epoch": 0.24614638679569284,
      "grad_norm": 2.641279935836792,
      "learning_rate": 9.774610305867972e-06,
      "loss": 1.0847,
      "step": 3046
    },
    {
      "epoch": 0.2462271965090204,
      "grad_norm": 2.825593948364258,
      "learning_rate": 9.774416013240717e-06,
      "loss": 1.0203,
      "step": 3047
    },
    {
      "epoch": 0.2463080062223479,
      "grad_norm": 2.7793221473693848,
      "learning_rate": 9.77422163883921e-06,
      "loss": 0.9459,
      "step": 3048
    },
    {
      "epoch": 0.24638881593567546,
      "grad_norm": 2.340348482131958,
      "learning_rate": 9.77402718266678e-06,
      "loss": 0.9582,
      "step": 3049
    },
    {
      "epoch": 0.24646962564900302,
      "grad_norm": 2.321822166442871,
      "learning_rate": 9.773832644726757e-06,
      "loss": 1.0461,
      "step": 3050
    },
    {
      "epoch": 0.24655043536233054,
      "grad_norm": 2.6112189292907715,
      "learning_rate": 9.773638025022474e-06,
      "loss": 1.0322,
      "step": 3051
    },
    {
      "epoch": 0.2466312450756581,
      "grad_norm": 2.735830545425415,
      "learning_rate": 9.773443323557263e-06,
      "loss": 0.972,
      "step": 3052
    },
    {
      "epoch": 0.24671205478898564,
      "grad_norm": 2.960653781890869,
      "learning_rate": 9.773248540334461e-06,
      "loss": 1.0007,
      "step": 3053
    },
    {
      "epoch": 0.2467928645023132,
      "grad_norm": 2.744239330291748,
      "learning_rate": 9.7730536753574e-06,
      "loss": 1.0094,
      "step": 3054
    },
    {
      "epoch": 0.24687367421564071,
      "grad_norm": 3.0212156772613525,
      "learning_rate": 9.772858728629421e-06,
      "loss": 0.9539,
      "step": 3055
    },
    {
      "epoch": 0.24695448392896827,
      "grad_norm": 2.6391782760620117,
      "learning_rate": 9.772663700153864e-06,
      "loss": 1.0186,
      "step": 3056
    },
    {
      "epoch": 0.24703529364229582,
      "grad_norm": 2.6837496757507324,
      "learning_rate": 9.772468589934066e-06,
      "loss": 1.0038,
      "step": 3057
    },
    {
      "epoch": 0.24711610335562334,
      "grad_norm": 2.7959649562835693,
      "learning_rate": 9.77227339797337e-06,
      "loss": 1.0742,
      "step": 3058
    },
    {
      "epoch": 0.2471969130689509,
      "grad_norm": 2.8792660236358643,
      "learning_rate": 9.772078124275121e-06,
      "loss": 0.9766,
      "step": 3059
    },
    {
      "epoch": 0.24727772278227844,
      "grad_norm": 2.823575496673584,
      "learning_rate": 9.771882768842663e-06,
      "loss": 0.9364,
      "step": 3060
    },
    {
      "epoch": 0.24735853249560597,
      "grad_norm": 2.5848450660705566,
      "learning_rate": 9.771687331679338e-06,
      "loss": 0.9675,
      "step": 3061
    },
    {
      "epoch": 0.24743934220893352,
      "grad_norm": 2.9905905723571777,
      "learning_rate": 9.771491812788498e-06,
      "loss": 1.0847,
      "step": 3062
    },
    {
      "epoch": 0.24752015192226107,
      "grad_norm": 3.0971758365631104,
      "learning_rate": 9.77129621217349e-06,
      "loss": 0.9738,
      "step": 3063
    },
    {
      "epoch": 0.2476009616355886,
      "grad_norm": 2.6731889247894287,
      "learning_rate": 9.771100529837662e-06,
      "loss": 0.9633,
      "step": 3064
    },
    {
      "epoch": 0.24768177134891614,
      "grad_norm": 2.5933165550231934,
      "learning_rate": 9.770904765784372e-06,
      "loss": 1.0145,
      "step": 3065
    },
    {
      "epoch": 0.2477625810622437,
      "grad_norm": 2.8988571166992188,
      "learning_rate": 9.770708920016967e-06,
      "loss": 1.002,
      "step": 3066
    },
    {
      "epoch": 0.24784339077557122,
      "grad_norm": 2.518685817718506,
      "learning_rate": 9.770512992538801e-06,
      "loss": 0.9592,
      "step": 3067
    },
    {
      "epoch": 0.24792420048889877,
      "grad_norm": 2.4646313190460205,
      "learning_rate": 9.770316983353235e-06,
      "loss": 0.9018,
      "step": 3068
    },
    {
      "epoch": 0.24800501020222632,
      "grad_norm": 2.8748724460601807,
      "learning_rate": 9.770120892463622e-06,
      "loss": 0.9996,
      "step": 3069
    },
    {
      "epoch": 0.24808581991555384,
      "grad_norm": 2.6587514877319336,
      "learning_rate": 9.769924719873322e-06,
      "loss": 0.9461,
      "step": 3070
    },
    {
      "epoch": 0.2481666296288814,
      "grad_norm": 2.981292247772217,
      "learning_rate": 9.769728465585694e-06,
      "loss": 0.9517,
      "step": 3071
    },
    {
      "epoch": 0.24824743934220894,
      "grad_norm": 2.7922825813293457,
      "learning_rate": 9.7695321296041e-06,
      "loss": 1.0152,
      "step": 3072
    },
    {
      "epoch": 0.24832824905553647,
      "grad_norm": 2.7855329513549805,
      "learning_rate": 9.769335711931904e-06,
      "loss": 0.9409,
      "step": 3073
    },
    {
      "epoch": 0.24840905876886402,
      "grad_norm": 2.9992854595184326,
      "learning_rate": 9.769139212572469e-06,
      "loss": 1.0594,
      "step": 3074
    },
    {
      "epoch": 0.24848986848219157,
      "grad_norm": 2.942552089691162,
      "learning_rate": 9.768942631529158e-06,
      "loss": 0.9937,
      "step": 3075
    },
    {
      "epoch": 0.2485706781955191,
      "grad_norm": 2.8532662391662598,
      "learning_rate": 9.768745968805343e-06,
      "loss": 0.9324,
      "step": 3076
    },
    {
      "epoch": 0.24865148790884664,
      "grad_norm": 3.196516752243042,
      "learning_rate": 9.768549224404388e-06,
      "loss": 0.9785,
      "step": 3077
    },
    {
      "epoch": 0.2487322976221742,
      "grad_norm": 2.676330089569092,
      "learning_rate": 9.768352398329668e-06,
      "loss": 1.0158,
      "step": 3078
    },
    {
      "epoch": 0.24881310733550172,
      "grad_norm": 2.9280941486358643,
      "learning_rate": 9.768155490584548e-06,
      "loss": 1.0447,
      "step": 3079
    },
    {
      "epoch": 0.24889391704882927,
      "grad_norm": 3.0791561603546143,
      "learning_rate": 9.767958501172404e-06,
      "loss": 1.0058,
      "step": 3080
    },
    {
      "epoch": 0.24897472676215682,
      "grad_norm": 2.6139445304870605,
      "learning_rate": 9.767761430096608e-06,
      "loss": 0.9489,
      "step": 3081
    },
    {
      "epoch": 0.24905553647548434,
      "grad_norm": 3.2888574600219727,
      "learning_rate": 9.767564277360538e-06,
      "loss": 1.0186,
      "step": 3082
    },
    {
      "epoch": 0.2491363461888119,
      "grad_norm": 2.849696636199951,
      "learning_rate": 9.767367042967568e-06,
      "loss": 1.0644,
      "step": 3083
    },
    {
      "epoch": 0.24921715590213944,
      "grad_norm": 2.7495691776275635,
      "learning_rate": 9.76716972692108e-06,
      "loss": 1.0258,
      "step": 3084
    },
    {
      "epoch": 0.24929796561546697,
      "grad_norm": 2.598660945892334,
      "learning_rate": 9.766972329224449e-06,
      "loss": 0.9113,
      "step": 3085
    },
    {
      "epoch": 0.24937877532879452,
      "grad_norm": 2.9330737590789795,
      "learning_rate": 9.766774849881058e-06,
      "loss": 0.9168,
      "step": 3086
    },
    {
      "epoch": 0.24945958504212207,
      "grad_norm": 3.1084072589874268,
      "learning_rate": 9.766577288894291e-06,
      "loss": 1.0031,
      "step": 3087
    },
    {
      "epoch": 0.24954039475544962,
      "grad_norm": 2.482382297515869,
      "learning_rate": 9.76637964626753e-06,
      "loss": 0.9676,
      "step": 3088
    },
    {
      "epoch": 0.24962120446877714,
      "grad_norm": 3.176751136779785,
      "learning_rate": 9.766181922004158e-06,
      "loss": 0.9304,
      "step": 3089
    },
    {
      "epoch": 0.2497020141821047,
      "grad_norm": 2.7606074810028076,
      "learning_rate": 9.765984116107565e-06,
      "loss": 1.0468,
      "step": 3090
    },
    {
      "epoch": 0.24978282389543224,
      "grad_norm": 2.808696985244751,
      "learning_rate": 9.765786228581138e-06,
      "loss": 0.8697,
      "step": 3091
    },
    {
      "epoch": 0.24986363360875977,
      "grad_norm": 2.8495774269104004,
      "learning_rate": 9.765588259428267e-06,
      "loss": 1.0248,
      "step": 3092
    },
    {
      "epoch": 0.24994444332208732,
      "grad_norm": 2.910490036010742,
      "learning_rate": 9.76539020865234e-06,
      "loss": 1.0361,
      "step": 3093
    },
    {
      "epoch": 0.25002525303541484,
      "grad_norm": 2.546539306640625,
      "learning_rate": 9.765192076256752e-06,
      "loss": 0.9902,
      "step": 3094
    },
    {
      "epoch": 0.2501060627487424,
      "grad_norm": 2.936716079711914,
      "learning_rate": 9.764993862244895e-06,
      "loss": 1.1006,
      "step": 3095
    },
    {
      "epoch": 0.25018687246206994,
      "grad_norm": 2.944744110107422,
      "learning_rate": 9.764795566620164e-06,
      "loss": 0.9621,
      "step": 3096
    },
    {
      "epoch": 0.25026768217539747,
      "grad_norm": 2.513930559158325,
      "learning_rate": 9.764597189385957e-06,
      "loss": 0.9737,
      "step": 3097
    },
    {
      "epoch": 0.25034849188872504,
      "grad_norm": 2.5775349140167236,
      "learning_rate": 9.76439873054567e-06,
      "loss": 0.9102,
      "step": 3098
    },
    {
      "epoch": 0.25042930160205257,
      "grad_norm": 2.507645845413208,
      "learning_rate": 9.764200190102702e-06,
      "loss": 0.9838,
      "step": 3099
    },
    {
      "epoch": 0.2505101113153801,
      "grad_norm": 2.95524263381958,
      "learning_rate": 9.764001568060455e-06,
      "loss": 0.9593,
      "step": 3100
    },
    {
      "epoch": 0.25059092102870767,
      "grad_norm": 2.742894172668457,
      "learning_rate": 9.763802864422329e-06,
      "loss": 0.9644,
      "step": 3101
    },
    {
      "epoch": 0.2506717307420352,
      "grad_norm": 2.268688201904297,
      "learning_rate": 9.76360407919173e-06,
      "loss": 1.0229,
      "step": 3102
    },
    {
      "epoch": 0.2507525404553627,
      "grad_norm": 2.7283692359924316,
      "learning_rate": 9.763405212372059e-06,
      "loss": 1.0102,
      "step": 3103
    },
    {
      "epoch": 0.2508333501686903,
      "grad_norm": 2.77974534034729,
      "learning_rate": 9.763206263966725e-06,
      "loss": 1.0688,
      "step": 3104
    },
    {
      "epoch": 0.2509141598820178,
      "grad_norm": 2.8854501247406006,
      "learning_rate": 9.763007233979133e-06,
      "loss": 1.1202,
      "step": 3105
    },
    {
      "epoch": 0.25099496959534534,
      "grad_norm": 2.7231879234313965,
      "learning_rate": 9.762808122412694e-06,
      "loss": 0.9909,
      "step": 3106
    },
    {
      "epoch": 0.2510757793086729,
      "grad_norm": 2.7498559951782227,
      "learning_rate": 9.762608929270821e-06,
      "loss": 0.9976,
      "step": 3107
    },
    {
      "epoch": 0.25115658902200044,
      "grad_norm": 2.8472721576690674,
      "learning_rate": 9.762409654556919e-06,
      "loss": 0.9627,
      "step": 3108
    },
    {
      "epoch": 0.25123739873532797,
      "grad_norm": 2.599015474319458,
      "learning_rate": 9.762210298274408e-06,
      "loss": 0.9065,
      "step": 3109
    },
    {
      "epoch": 0.25131820844865554,
      "grad_norm": 2.8924973011016846,
      "learning_rate": 9.762010860426696e-06,
      "loss": 1.0123,
      "step": 3110
    },
    {
      "epoch": 0.25139901816198307,
      "grad_norm": 2.4286463260650635,
      "learning_rate": 9.761811341017205e-06,
      "loss": 1.1075,
      "step": 3111
    },
    {
      "epoch": 0.2514798278753106,
      "grad_norm": 2.6996352672576904,
      "learning_rate": 9.761611740049345e-06,
      "loss": 1.0744,
      "step": 3112
    },
    {
      "epoch": 0.25156063758863817,
      "grad_norm": 3.5628347396850586,
      "learning_rate": 9.761412057526543e-06,
      "loss": 0.8973,
      "step": 3113
    },
    {
      "epoch": 0.2516414473019657,
      "grad_norm": 2.604649305343628,
      "learning_rate": 9.761212293452213e-06,
      "loss": 0.9428,
      "step": 3114
    },
    {
      "epoch": 0.2517222570152932,
      "grad_norm": 2.881993293762207,
      "learning_rate": 9.76101244782978e-06,
      "loss": 0.8902,
      "step": 3115
    },
    {
      "epoch": 0.2518030667286208,
      "grad_norm": 2.4835922718048096,
      "learning_rate": 9.760812520662665e-06,
      "loss": 0.9009,
      "step": 3116
    },
    {
      "epoch": 0.2518838764419483,
      "grad_norm": 2.4867897033691406,
      "learning_rate": 9.760612511954293e-06,
      "loss": 1.0544,
      "step": 3117
    },
    {
      "epoch": 0.25196468615527584,
      "grad_norm": 2.5339856147766113,
      "learning_rate": 9.760412421708087e-06,
      "loss": 0.851,
      "step": 3118
    },
    {
      "epoch": 0.2520454958686034,
      "grad_norm": 2.5214011669158936,
      "learning_rate": 9.760212249927479e-06,
      "loss": 1.0445,
      "step": 3119
    },
    {
      "epoch": 0.25212630558193094,
      "grad_norm": 2.8076467514038086,
      "learning_rate": 9.760011996615894e-06,
      "loss": 1.1723,
      "step": 3120
    },
    {
      "epoch": 0.25220711529525847,
      "grad_norm": 2.591118812561035,
      "learning_rate": 9.759811661776763e-06,
      "loss": 0.9127,
      "step": 3121
    },
    {
      "epoch": 0.25228792500858604,
      "grad_norm": 2.901501417160034,
      "learning_rate": 9.759611245413518e-06,
      "loss": 0.8978,
      "step": 3122
    },
    {
      "epoch": 0.25236873472191357,
      "grad_norm": 3.079751968383789,
      "learning_rate": 9.759410747529589e-06,
      "loss": 1.037,
      "step": 3123
    },
    {
      "epoch": 0.2524495444352411,
      "grad_norm": 2.789341926574707,
      "learning_rate": 9.759210168128412e-06,
      "loss": 0.9348,
      "step": 3124
    },
    {
      "epoch": 0.25253035414856867,
      "grad_norm": 2.325867176055908,
      "learning_rate": 9.759009507213423e-06,
      "loss": 0.8928,
      "step": 3125
    },
    {
      "epoch": 0.2526111638618962,
      "grad_norm": 3.3124213218688965,
      "learning_rate": 9.758808764788056e-06,
      "loss": 1.0175,
      "step": 3126
    },
    {
      "epoch": 0.2526919735752237,
      "grad_norm": 2.7578797340393066,
      "learning_rate": 9.758607940855754e-06,
      "loss": 0.9931,
      "step": 3127
    },
    {
      "epoch": 0.2527727832885513,
      "grad_norm": 2.998295307159424,
      "learning_rate": 9.758407035419952e-06,
      "loss": 0.9904,
      "step": 3128
    },
    {
      "epoch": 0.2528535930018788,
      "grad_norm": 2.877265453338623,
      "learning_rate": 9.758206048484094e-06,
      "loss": 1.0652,
      "step": 3129
    },
    {
      "epoch": 0.25293440271520634,
      "grad_norm": 2.6761255264282227,
      "learning_rate": 9.758004980051621e-06,
      "loss": 1.0342,
      "step": 3130
    },
    {
      "epoch": 0.2530152124285339,
      "grad_norm": 2.4134268760681152,
      "learning_rate": 9.757803830125976e-06,
      "loss": 1.0312,
      "step": 3131
    },
    {
      "epoch": 0.25309602214186144,
      "grad_norm": 2.7035300731658936,
      "learning_rate": 9.757602598710607e-06,
      "loss": 0.9359,
      "step": 3132
    },
    {
      "epoch": 0.25317683185518897,
      "grad_norm": 2.915403366088867,
      "learning_rate": 9.757401285808957e-06,
      "loss": 1.0197,
      "step": 3133
    },
    {
      "epoch": 0.25325764156851654,
      "grad_norm": 2.634140968322754,
      "learning_rate": 9.757199891424478e-06,
      "loss": 0.9837,
      "step": 3134
    },
    {
      "epoch": 0.25333845128184407,
      "grad_norm": 2.884725570678711,
      "learning_rate": 9.756998415560616e-06,
      "loss": 1.068,
      "step": 3135
    },
    {
      "epoch": 0.25341926099517165,
      "grad_norm": 2.804107666015625,
      "learning_rate": 9.756796858220823e-06,
      "loss": 0.9709,
      "step": 3136
    },
    {
      "epoch": 0.25350007070849917,
      "grad_norm": 2.7716612815856934,
      "learning_rate": 9.756595219408552e-06,
      "loss": 0.9399,
      "step": 3137
    },
    {
      "epoch": 0.2535808804218267,
      "grad_norm": 2.914701461791992,
      "learning_rate": 9.756393499127257e-06,
      "loss": 1.0254,
      "step": 3138
    },
    {
      "epoch": 0.25366169013515427,
      "grad_norm": 2.6848130226135254,
      "learning_rate": 9.756191697380391e-06,
      "loss": 1.0057,
      "step": 3139
    },
    {
      "epoch": 0.2537424998484818,
      "grad_norm": 2.1119136810302734,
      "learning_rate": 9.755989814171409e-06,
      "loss": 0.9878,
      "step": 3140
    },
    {
      "epoch": 0.2538233095618093,
      "grad_norm": 2.6999025344848633,
      "learning_rate": 9.755787849503775e-06,
      "loss": 0.9924,
      "step": 3141
    },
    {
      "epoch": 0.2539041192751369,
      "grad_norm": 2.9252264499664307,
      "learning_rate": 9.755585803380941e-06,
      "loss": 0.9457,
      "step": 3142
    },
    {
      "epoch": 0.2539849289884644,
      "grad_norm": 2.814427137374878,
      "learning_rate": 9.755383675806372e-06,
      "loss": 0.9491,
      "step": 3143
    },
    {
      "epoch": 0.25406573870179194,
      "grad_norm": 2.658046007156372,
      "learning_rate": 9.75518146678353e-06,
      "loss": 1.0038,
      "step": 3144
    },
    {
      "epoch": 0.2541465484151195,
      "grad_norm": 2.743593454360962,
      "learning_rate": 9.754979176315876e-06,
      "loss": 1.1438,
      "step": 3145
    },
    {
      "epoch": 0.25422735812844705,
      "grad_norm": 3.0818870067596436,
      "learning_rate": 9.754776804406876e-06,
      "loss": 1.0256,
      "step": 3146
    },
    {
      "epoch": 0.25430816784177457,
      "grad_norm": 2.808840036392212,
      "learning_rate": 9.754574351059995e-06,
      "loss": 0.9556,
      "step": 3147
    },
    {
      "epoch": 0.25438897755510215,
      "grad_norm": 2.8346235752105713,
      "learning_rate": 9.754371816278702e-06,
      "loss": 0.9963,
      "step": 3148
    },
    {
      "epoch": 0.25446978726842967,
      "grad_norm": 2.8395915031433105,
      "learning_rate": 9.754169200066466e-06,
      "loss": 0.9539,
      "step": 3149
    },
    {
      "epoch": 0.2545505969817572,
      "grad_norm": 3.3882341384887695,
      "learning_rate": 9.753966502426756e-06,
      "loss": 1.0115,
      "step": 3150
    },
    {
      "epoch": 0.25463140669508477,
      "grad_norm": 3.1990177631378174,
      "learning_rate": 9.753763723363045e-06,
      "loss": 0.989,
      "step": 3151
    },
    {
      "epoch": 0.2547122164084123,
      "grad_norm": 2.8756611347198486,
      "learning_rate": 9.753560862878806e-06,
      "loss": 0.9778,
      "step": 3152
    },
    {
      "epoch": 0.2547930261217398,
      "grad_norm": 2.808302879333496,
      "learning_rate": 9.753357920977514e-06,
      "loss": 1.048,
      "step": 3153
    },
    {
      "epoch": 0.2548738358350674,
      "grad_norm": 2.6227824687957764,
      "learning_rate": 9.753154897662642e-06,
      "loss": 0.9811,
      "step": 3154
    },
    {
      "epoch": 0.2549546455483949,
      "grad_norm": 2.6170687675476074,
      "learning_rate": 9.75295179293767e-06,
      "loss": 0.9553,
      "step": 3155
    },
    {
      "epoch": 0.25503545526172244,
      "grad_norm": 2.8382341861724854,
      "learning_rate": 9.752748606806077e-06,
      "loss": 0.9198,
      "step": 3156
    },
    {
      "epoch": 0.25511626497505,
      "grad_norm": 2.7465264797210693,
      "learning_rate": 9.752545339271342e-06,
      "loss": 0.9956,
      "step": 3157
    },
    {
      "epoch": 0.25519707468837755,
      "grad_norm": 2.976637125015259,
      "learning_rate": 9.752341990336946e-06,
      "loss": 1.0486,
      "step": 3158
    },
    {
      "epoch": 0.25527788440170507,
      "grad_norm": 2.98663592338562,
      "learning_rate": 9.752138560006372e-06,
      "loss": 1.0955,
      "step": 3159
    },
    {
      "epoch": 0.25535869411503265,
      "grad_norm": 2.647614002227783,
      "learning_rate": 9.751935048283105e-06,
      "loss": 1.0881,
      "step": 3160
    },
    {
      "epoch": 0.25543950382836017,
      "grad_norm": 2.572448968887329,
      "learning_rate": 9.751731455170632e-06,
      "loss": 0.9703,
      "step": 3161
    },
    {
      "epoch": 0.2555203135416877,
      "grad_norm": 2.944451332092285,
      "learning_rate": 9.751527780672438e-06,
      "loss": 0.811,
      "step": 3162
    },
    {
      "epoch": 0.2556011232550153,
      "grad_norm": 2.7639005184173584,
      "learning_rate": 9.75132402479201e-06,
      "loss": 1.0385,
      "step": 3163
    },
    {
      "epoch": 0.2556819329683428,
      "grad_norm": 2.619258165359497,
      "learning_rate": 9.75112018753284e-06,
      "loss": 0.9642,
      "step": 3164
    },
    {
      "epoch": 0.2557627426816703,
      "grad_norm": 3.1263070106506348,
      "learning_rate": 9.750916268898423e-06,
      "loss": 0.9945,
      "step": 3165
    },
    {
      "epoch": 0.2558435523949979,
      "grad_norm": 3.069749116897583,
      "learning_rate": 9.750712268892245e-06,
      "loss": 0.965,
      "step": 3166
    },
    {
      "epoch": 0.2559243621083254,
      "grad_norm": 2.712395429611206,
      "learning_rate": 9.750508187517802e-06,
      "loss": 1.0585,
      "step": 3167
    },
    {
      "epoch": 0.25600517182165294,
      "grad_norm": 2.7137744426727295,
      "learning_rate": 9.75030402477859e-06,
      "loss": 1.1198,
      "step": 3168
    },
    {
      "epoch": 0.2560859815349805,
      "grad_norm": 3.896778106689453,
      "learning_rate": 9.750099780678106e-06,
      "loss": 1.0689,
      "step": 3169
    },
    {
      "epoch": 0.25616679124830805,
      "grad_norm": 2.7930946350097656,
      "learning_rate": 9.749895455219849e-06,
      "loss": 0.9992,
      "step": 3170
    },
    {
      "epoch": 0.25624760096163557,
      "grad_norm": 2.665447473526001,
      "learning_rate": 9.749691048407318e-06,
      "loss": 0.9512,
      "step": 3171
    },
    {
      "epoch": 0.25632841067496315,
      "grad_norm": 2.651581287384033,
      "learning_rate": 9.749486560244014e-06,
      "loss": 0.9462,
      "step": 3172
    },
    {
      "epoch": 0.25640922038829067,
      "grad_norm": 2.931328535079956,
      "learning_rate": 9.749281990733438e-06,
      "loss": 0.9202,
      "step": 3173
    },
    {
      "epoch": 0.2564900301016182,
      "grad_norm": 2.4715116024017334,
      "learning_rate": 9.749077339879095e-06,
      "loss": 1.0637,
      "step": 3174
    },
    {
      "epoch": 0.2565708398149458,
      "grad_norm": 2.9267590045928955,
      "learning_rate": 9.74887260768449e-06,
      "loss": 1.1934,
      "step": 3175
    },
    {
      "epoch": 0.2566516495282733,
      "grad_norm": 2.7444472312927246,
      "learning_rate": 9.748667794153131e-06,
      "loss": 0.8217,
      "step": 3176
    },
    {
      "epoch": 0.2567324592416008,
      "grad_norm": 2.5331554412841797,
      "learning_rate": 9.748462899288523e-06,
      "loss": 1.0896,
      "step": 3177
    },
    {
      "epoch": 0.2568132689549284,
      "grad_norm": 2.6985771656036377,
      "learning_rate": 9.748257923094177e-06,
      "loss": 0.9791,
      "step": 3178
    },
    {
      "epoch": 0.2568940786682559,
      "grad_norm": 2.4230868816375732,
      "learning_rate": 9.748052865573605e-06,
      "loss": 1.0133,
      "step": 3179
    },
    {
      "epoch": 0.25697488838158344,
      "grad_norm": 2.8683249950408936,
      "learning_rate": 9.747847726730318e-06,
      "loss": 0.9224,
      "step": 3180
    },
    {
      "epoch": 0.257055698094911,
      "grad_norm": 3.0334200859069824,
      "learning_rate": 9.74764250656783e-06,
      "loss": 0.8719,
      "step": 3181
    },
    {
      "epoch": 0.25713650780823855,
      "grad_norm": 3.434133768081665,
      "learning_rate": 9.747437205089654e-06,
      "loss": 0.9158,
      "step": 3182
    },
    {
      "epoch": 0.25721731752156607,
      "grad_norm": 2.634556770324707,
      "learning_rate": 9.747231822299308e-06,
      "loss": 1.0091,
      "step": 3183
    },
    {
      "epoch": 0.25729812723489365,
      "grad_norm": 2.6447670459747314,
      "learning_rate": 9.747026358200309e-06,
      "loss": 0.9963,
      "step": 3184
    },
    {
      "epoch": 0.25737893694822117,
      "grad_norm": 2.687317132949829,
      "learning_rate": 9.746820812796176e-06,
      "loss": 1.1168,
      "step": 3185
    },
    {
      "epoch": 0.2574597466615487,
      "grad_norm": 2.8211278915405273,
      "learning_rate": 9.746615186090432e-06,
      "loss": 1.0444,
      "step": 3186
    },
    {
      "epoch": 0.2575405563748763,
      "grad_norm": 2.3882555961608887,
      "learning_rate": 9.746409478086593e-06,
      "loss": 1.0406,
      "step": 3187
    },
    {
      "epoch": 0.2576213660882038,
      "grad_norm": 2.952209711074829,
      "learning_rate": 9.74620368878819e-06,
      "loss": 0.9646,
      "step": 3188
    },
    {
      "epoch": 0.2577021758015313,
      "grad_norm": 2.738934278488159,
      "learning_rate": 9.745997818198743e-06,
      "loss": 0.9903,
      "step": 3189
    },
    {
      "epoch": 0.2577829855148589,
      "grad_norm": 3.0450305938720703,
      "learning_rate": 9.74579186632178e-06,
      "loss": 1.0351,
      "step": 3190
    },
    {
      "epoch": 0.2578637952281864,
      "grad_norm": 2.916048049926758,
      "learning_rate": 9.745585833160824e-06,
      "loss": 1.0677,
      "step": 3191
    },
    {
      "epoch": 0.25794460494151394,
      "grad_norm": 2.883650541305542,
      "learning_rate": 9.745379718719408e-06,
      "loss": 0.8588,
      "step": 3192
    },
    {
      "epoch": 0.2580254146548415,
      "grad_norm": 2.514897584915161,
      "learning_rate": 9.745173523001063e-06,
      "loss": 1.0435,
      "step": 3193
    },
    {
      "epoch": 0.25810622436816905,
      "grad_norm": 3.411410331726074,
      "learning_rate": 9.744967246009319e-06,
      "loss": 1.0101,
      "step": 3194
    },
    {
      "epoch": 0.25818703408149657,
      "grad_norm": 2.801797389984131,
      "learning_rate": 9.744760887747708e-06,
      "loss": 0.9591,
      "step": 3195
    },
    {
      "epoch": 0.25826784379482415,
      "grad_norm": 2.578493118286133,
      "learning_rate": 9.744554448219767e-06,
      "loss": 1.1203,
      "step": 3196
    },
    {
      "epoch": 0.25834865350815167,
      "grad_norm": 2.7883083820343018,
      "learning_rate": 9.74434792742903e-06,
      "loss": 1.0737,
      "step": 3197
    },
    {
      "epoch": 0.2584294632214792,
      "grad_norm": 3.0916290283203125,
      "learning_rate": 9.744141325379032e-06,
      "loss": 0.8681,
      "step": 3198
    },
    {
      "epoch": 0.2585102729348068,
      "grad_norm": 2.6086947917938232,
      "learning_rate": 9.743934642073318e-06,
      "loss": 0.9693,
      "step": 3199
    },
    {
      "epoch": 0.2585910826481343,
      "grad_norm": 3.25252628326416,
      "learning_rate": 9.743727877515422e-06,
      "loss": 1.0934,
      "step": 3200
    },
    {
      "epoch": 0.2586718923614619,
      "grad_norm": 2.8991661071777344,
      "learning_rate": 9.743521031708888e-06,
      "loss": 1.045,
      "step": 3201
    },
    {
      "epoch": 0.2587527020747894,
      "grad_norm": 2.863893985748291,
      "learning_rate": 9.743314104657258e-06,
      "loss": 1.0193,
      "step": 3202
    },
    {
      "epoch": 0.2588335117881169,
      "grad_norm": 2.706991672515869,
      "learning_rate": 9.743107096364078e-06,
      "loss": 1.0053,
      "step": 3203
    },
    {
      "epoch": 0.2589143215014445,
      "grad_norm": 3.0327980518341064,
      "learning_rate": 9.742900006832889e-06,
      "loss": 0.9308,
      "step": 3204
    },
    {
      "epoch": 0.258995131214772,
      "grad_norm": 2.8023879528045654,
      "learning_rate": 9.742692836067242e-06,
      "loss": 1.0627,
      "step": 3205
    },
    {
      "epoch": 0.25907594092809955,
      "grad_norm": 2.7891502380371094,
      "learning_rate": 9.742485584070687e-06,
      "loss": 1.0111,
      "step": 3206
    },
    {
      "epoch": 0.2591567506414271,
      "grad_norm": 2.743758201599121,
      "learning_rate": 9.742278250846769e-06,
      "loss": 1.001,
      "step": 3207
    },
    {
      "epoch": 0.25923756035475465,
      "grad_norm": 3.3825297355651855,
      "learning_rate": 9.74207083639904e-06,
      "loss": 1.0003,
      "step": 3208
    },
    {
      "epoch": 0.25931837006808217,
      "grad_norm": 2.479440927505493,
      "learning_rate": 9.741863340731054e-06,
      "loss": 1.0394,
      "step": 3209
    },
    {
      "epoch": 0.25939917978140975,
      "grad_norm": 3.305624485015869,
      "learning_rate": 9.741655763846365e-06,
      "loss": 0.9289,
      "step": 3210
    },
    {
      "epoch": 0.2594799894947373,
      "grad_norm": 3.0319902896881104,
      "learning_rate": 9.741448105748529e-06,
      "loss": 0.9316,
      "step": 3211
    },
    {
      "epoch": 0.2595607992080648,
      "grad_norm": 3.279977560043335,
      "learning_rate": 9.7412403664411e-06,
      "loss": 1.0275,
      "step": 3212
    },
    {
      "epoch": 0.2596416089213924,
      "grad_norm": 3.2768304347991943,
      "learning_rate": 9.741032545927639e-06,
      "loss": 0.9945,
      "step": 3213
    },
    {
      "epoch": 0.2597224186347199,
      "grad_norm": 3.07078218460083,
      "learning_rate": 9.740824644211703e-06,
      "loss": 0.9448,
      "step": 3214
    },
    {
      "epoch": 0.2598032283480474,
      "grad_norm": 2.890254259109497,
      "learning_rate": 9.740616661296853e-06,
      "loss": 1.0775,
      "step": 3215
    },
    {
      "epoch": 0.259884038061375,
      "grad_norm": 3.059772491455078,
      "learning_rate": 9.740408597186655e-06,
      "loss": 0.9619,
      "step": 3216
    },
    {
      "epoch": 0.2599648477747025,
      "grad_norm": 2.7392866611480713,
      "learning_rate": 9.740200451884668e-06,
      "loss": 0.8956,
      "step": 3217
    },
    {
      "epoch": 0.26004565748803005,
      "grad_norm": 3.0844335556030273,
      "learning_rate": 9.739992225394459e-06,
      "loss": 0.9921,
      "step": 3218
    },
    {
      "epoch": 0.2601264672013576,
      "grad_norm": 2.98893404006958,
      "learning_rate": 9.739783917719595e-06,
      "loss": 1.0277,
      "step": 3219
    },
    {
      "epoch": 0.26020727691468515,
      "grad_norm": 2.8196518421173096,
      "learning_rate": 9.73957552886364e-06,
      "loss": 1.0273,
      "step": 3220
    },
    {
      "epoch": 0.26028808662801267,
      "grad_norm": 2.75636887550354,
      "learning_rate": 9.739367058830169e-06,
      "loss": 1.0258,
      "step": 3221
    },
    {
      "epoch": 0.26036889634134025,
      "grad_norm": 2.8138818740844727,
      "learning_rate": 9.73915850762275e-06,
      "loss": 1.0349,
      "step": 3222
    },
    {
      "epoch": 0.2604497060546678,
      "grad_norm": 3.6661946773529053,
      "learning_rate": 9.738949875244953e-06,
      "loss": 0.9743,
      "step": 3223
    },
    {
      "epoch": 0.2605305157679953,
      "grad_norm": 2.9072020053863525,
      "learning_rate": 9.738741161700356e-06,
      "loss": 0.9443,
      "step": 3224
    },
    {
      "epoch": 0.2606113254813229,
      "grad_norm": 3.0198700428009033,
      "learning_rate": 9.738532366992528e-06,
      "loss": 1.0106,
      "step": 3225
    },
    {
      "epoch": 0.2606921351946504,
      "grad_norm": 2.8815252780914307,
      "learning_rate": 9.73832349112505e-06,
      "loss": 1.0824,
      "step": 3226
    },
    {
      "epoch": 0.2607729449079779,
      "grad_norm": 2.8785507678985596,
      "learning_rate": 9.738114534101498e-06,
      "loss": 1.004,
      "step": 3227
    },
    {
      "epoch": 0.2608537546213055,
      "grad_norm": 3.012033224105835,
      "learning_rate": 9.737905495925448e-06,
      "loss": 0.912,
      "step": 3228
    },
    {
      "epoch": 0.260934564334633,
      "grad_norm": 3.8269202709198,
      "learning_rate": 9.737696376600485e-06,
      "loss": 0.9791,
      "step": 3229
    },
    {
      "epoch": 0.26101537404796055,
      "grad_norm": 2.977001190185547,
      "learning_rate": 9.737487176130189e-06,
      "loss": 1.0622,
      "step": 3230
    },
    {
      "epoch": 0.2610961837612881,
      "grad_norm": 2.966339349746704,
      "learning_rate": 9.73727789451814e-06,
      "loss": 1.0245,
      "step": 3231
    },
    {
      "epoch": 0.26117699347461565,
      "grad_norm": 2.9824018478393555,
      "learning_rate": 9.737068531767927e-06,
      "loss": 0.9502,
      "step": 3232
    },
    {
      "epoch": 0.26125780318794317,
      "grad_norm": 3.0262484550476074,
      "learning_rate": 9.736859087883135e-06,
      "loss": 1.1588,
      "step": 3233
    },
    {
      "epoch": 0.26133861290127075,
      "grad_norm": 2.919766664505005,
      "learning_rate": 9.736649562867349e-06,
      "loss": 0.9691,
      "step": 3234
    },
    {
      "epoch": 0.2614194226145983,
      "grad_norm": 2.6849515438079834,
      "learning_rate": 9.73643995672416e-06,
      "loss": 0.9459,
      "step": 3235
    },
    {
      "epoch": 0.2615002323279258,
      "grad_norm": 2.936185121536255,
      "learning_rate": 9.736230269457156e-06,
      "loss": 0.9981,
      "step": 3236
    },
    {
      "epoch": 0.2615810420412534,
      "grad_norm": 2.6898229122161865,
      "learning_rate": 9.73602050106993e-06,
      "loss": 1.0971,
      "step": 3237
    },
    {
      "epoch": 0.2616618517545809,
      "grad_norm": 2.7416326999664307,
      "learning_rate": 9.735810651566076e-06,
      "loss": 1.0667,
      "step": 3238
    },
    {
      "epoch": 0.2617426614679084,
      "grad_norm": 2.762446880340576,
      "learning_rate": 9.735600720949183e-06,
      "loss": 0.8635,
      "step": 3239
    },
    {
      "epoch": 0.261823471181236,
      "grad_norm": 2.7108333110809326,
      "learning_rate": 9.735390709222853e-06,
      "loss": 0.9456,
      "step": 3240
    },
    {
      "epoch": 0.2619042808945635,
      "grad_norm": 2.640698194503784,
      "learning_rate": 9.735180616390678e-06,
      "loss": 0.9913,
      "step": 3241
    },
    {
      "epoch": 0.26198509060789105,
      "grad_norm": 2.8711159229278564,
      "learning_rate": 9.734970442456261e-06,
      "loss": 0.9699,
      "step": 3242
    },
    {
      "epoch": 0.2620659003212186,
      "grad_norm": 2.8066492080688477,
      "learning_rate": 9.734760187423198e-06,
      "loss": 1.0387,
      "step": 3243
    },
    {
      "epoch": 0.26214671003454615,
      "grad_norm": 3.017329692840576,
      "learning_rate": 9.73454985129509e-06,
      "loss": 0.8811,
      "step": 3244
    },
    {
      "epoch": 0.26222751974787367,
      "grad_norm": 2.606670379638672,
      "learning_rate": 9.734339434075543e-06,
      "loss": 0.9169,
      "step": 3245
    },
    {
      "epoch": 0.26230832946120125,
      "grad_norm": 2.583763360977173,
      "learning_rate": 9.734128935768158e-06,
      "loss": 1.0191,
      "step": 3246
    },
    {
      "epoch": 0.2623891391745288,
      "grad_norm": 3.545943260192871,
      "learning_rate": 9.733918356376542e-06,
      "loss": 1.1157,
      "step": 3247
    },
    {
      "epoch": 0.2624699488878563,
      "grad_norm": 2.810558319091797,
      "learning_rate": 9.733707695904301e-06,
      "loss": 0.9794,
      "step": 3248
    },
    {
      "epoch": 0.2625507586011839,
      "grad_norm": 2.942075252532959,
      "learning_rate": 9.733496954355042e-06,
      "loss": 0.9981,
      "step": 3249
    },
    {
      "epoch": 0.2626315683145114,
      "grad_norm": 2.93169903755188,
      "learning_rate": 9.733286131732377e-06,
      "loss": 0.9009,
      "step": 3250
    },
    {
      "epoch": 0.2627123780278389,
      "grad_norm": 2.745704174041748,
      "learning_rate": 9.733075228039914e-06,
      "loss": 0.8353,
      "step": 3251
    },
    {
      "epoch": 0.2627931877411665,
      "grad_norm": 2.6289989948272705,
      "learning_rate": 9.732864243281269e-06,
      "loss": 1.0759,
      "step": 3252
    },
    {
      "epoch": 0.262873997454494,
      "grad_norm": 2.8611576557159424,
      "learning_rate": 9.732653177460052e-06,
      "loss": 1.0328,
      "step": 3253
    },
    {
      "epoch": 0.26295480716782155,
      "grad_norm": 2.9202113151550293,
      "learning_rate": 9.73244203057988e-06,
      "loss": 1.0254,
      "step": 3254
    },
    {
      "epoch": 0.2630356168811491,
      "grad_norm": 2.776247501373291,
      "learning_rate": 9.732230802644367e-06,
      "loss": 0.9879,
      "step": 3255
    },
    {
      "epoch": 0.26311642659447665,
      "grad_norm": 3.0972414016723633,
      "learning_rate": 9.732019493657134e-06,
      "loss": 0.9659,
      "step": 3256
    },
    {
      "epoch": 0.26319723630780417,
      "grad_norm": 2.735015392303467,
      "learning_rate": 9.7318081036218e-06,
      "loss": 0.9677,
      "step": 3257
    },
    {
      "epoch": 0.26327804602113175,
      "grad_norm": 2.8691771030426025,
      "learning_rate": 9.731596632541985e-06,
      "loss": 1.1455,
      "step": 3258
    },
    {
      "epoch": 0.2633588557344593,
      "grad_norm": 2.6793711185455322,
      "learning_rate": 9.731385080421308e-06,
      "loss": 0.9582,
      "step": 3259
    },
    {
      "epoch": 0.2634396654477868,
      "grad_norm": 2.3230631351470947,
      "learning_rate": 9.731173447263395e-06,
      "loss": 0.9922,
      "step": 3260
    },
    {
      "epoch": 0.2635204751611144,
      "grad_norm": 3.0409085750579834,
      "learning_rate": 9.730961733071873e-06,
      "loss": 0.9773,
      "step": 3261
    },
    {
      "epoch": 0.2636012848744419,
      "grad_norm": 2.7706222534179688,
      "learning_rate": 9.730749937850365e-06,
      "loss": 1.031,
      "step": 3262
    },
    {
      "epoch": 0.2636820945877694,
      "grad_norm": 3.1750383377075195,
      "learning_rate": 9.730538061602497e-06,
      "loss": 1.0103,
      "step": 3263
    },
    {
      "epoch": 0.263762904301097,
      "grad_norm": 2.8183891773223877,
      "learning_rate": 9.730326104331904e-06,
      "loss": 1.0243,
      "step": 3264
    },
    {
      "epoch": 0.2638437140144245,
      "grad_norm": 3.4168620109558105,
      "learning_rate": 9.73011406604221e-06,
      "loss": 0.9172,
      "step": 3265
    },
    {
      "epoch": 0.2639245237277521,
      "grad_norm": 2.2862253189086914,
      "learning_rate": 9.72990194673705e-06,
      "loss": 1.0757,
      "step": 3266
    },
    {
      "epoch": 0.2640053334410796,
      "grad_norm": 2.386305570602417,
      "learning_rate": 9.729689746420057e-06,
      "loss": 0.9463,
      "step": 3267
    },
    {
      "epoch": 0.26408614315440715,
      "grad_norm": 2.6689112186431885,
      "learning_rate": 9.729477465094866e-06,
      "loss": 0.9309,
      "step": 3268
    },
    {
      "epoch": 0.26416695286773473,
      "grad_norm": 2.803179979324341,
      "learning_rate": 9.729265102765108e-06,
      "loss": 0.9962,
      "step": 3269
    },
    {
      "epoch": 0.26424776258106225,
      "grad_norm": 2.9772040843963623,
      "learning_rate": 9.729052659434428e-06,
      "loss": 1.0425,
      "step": 3270
    },
    {
      "epoch": 0.2643285722943898,
      "grad_norm": 3.799191951751709,
      "learning_rate": 9.728840135106458e-06,
      "loss": 0.9622,
      "step": 3271
    },
    {
      "epoch": 0.26440938200771735,
      "grad_norm": 3.263028383255005,
      "learning_rate": 9.728627529784842e-06,
      "loss": 0.98,
      "step": 3272
    },
    {
      "epoch": 0.2644901917210449,
      "grad_norm": 3.127373695373535,
      "learning_rate": 9.72841484347322e-06,
      "loss": 1.0495,
      "step": 3273
    },
    {
      "epoch": 0.2645710014343724,
      "grad_norm": 2.6552889347076416,
      "learning_rate": 9.728202076175235e-06,
      "loss": 0.993,
      "step": 3274
    },
    {
      "epoch": 0.2646518111477,
      "grad_norm": 2.61242938041687,
      "learning_rate": 9.727989227894532e-06,
      "loss": 0.914,
      "step": 3275
    },
    {
      "epoch": 0.2647326208610275,
      "grad_norm": 2.606858253479004,
      "learning_rate": 9.727776298634755e-06,
      "loss": 0.9245,
      "step": 3276
    },
    {
      "epoch": 0.264813430574355,
      "grad_norm": 2.996605634689331,
      "learning_rate": 9.72756328839955e-06,
      "loss": 1.0037,
      "step": 3277
    },
    {
      "epoch": 0.2648942402876826,
      "grad_norm": 2.7778725624084473,
      "learning_rate": 9.72735019719257e-06,
      "loss": 0.9636,
      "step": 3278
    },
    {
      "epoch": 0.2649750500010101,
      "grad_norm": 3.145533323287964,
      "learning_rate": 9.727137025017459e-06,
      "loss": 0.8808,
      "step": 3279
    },
    {
      "epoch": 0.26505585971433765,
      "grad_norm": 3.605461835861206,
      "learning_rate": 9.726923771877872e-06,
      "loss": 1.1387,
      "step": 3280
    },
    {
      "epoch": 0.26513666942766523,
      "grad_norm": 3.2222542762756348,
      "learning_rate": 9.72671043777746e-06,
      "loss": 0.986,
      "step": 3281
    },
    {
      "epoch": 0.26521747914099275,
      "grad_norm": 2.7679476737976074,
      "learning_rate": 9.726497022719878e-06,
      "loss": 1.0436,
      "step": 3282
    },
    {
      "epoch": 0.2652982888543203,
      "grad_norm": 2.496774196624756,
      "learning_rate": 9.72628352670878e-06,
      "loss": 0.9885,
      "step": 3283
    },
    {
      "epoch": 0.26537909856764785,
      "grad_norm": 2.3951852321624756,
      "learning_rate": 9.726069949747823e-06,
      "loss": 1.0944,
      "step": 3284
    },
    {
      "epoch": 0.2654599082809754,
      "grad_norm": 2.877875566482544,
      "learning_rate": 9.725856291840666e-06,
      "loss": 1.0931,
      "step": 3285
    },
    {
      "epoch": 0.2655407179943029,
      "grad_norm": 2.9747419357299805,
      "learning_rate": 9.725642552990967e-06,
      "loss": 1.045,
      "step": 3286
    },
    {
      "epoch": 0.2656215277076305,
      "grad_norm": 3.114614963531494,
      "learning_rate": 9.725428733202388e-06,
      "loss": 1.0449,
      "step": 3287
    },
    {
      "epoch": 0.265702337420958,
      "grad_norm": 2.7330870628356934,
      "learning_rate": 9.725214832478591e-06,
      "loss": 0.9713,
      "step": 3288
    },
    {
      "epoch": 0.2657831471342855,
      "grad_norm": 2.685214042663574,
      "learning_rate": 9.725000850823241e-06,
      "loss": 0.9988,
      "step": 3289
    },
    {
      "epoch": 0.2658639568476131,
      "grad_norm": 2.7034835815429688,
      "learning_rate": 9.72478678824e-06,
      "loss": 1.1093,
      "step": 3290
    },
    {
      "epoch": 0.2659447665609406,
      "grad_norm": 2.813911199569702,
      "learning_rate": 9.724572644732535e-06,
      "loss": 0.9647,
      "step": 3291
    },
    {
      "epoch": 0.26602557627426815,
      "grad_norm": 2.4258036613464355,
      "learning_rate": 9.724358420304514e-06,
      "loss": 0.949,
      "step": 3292
    },
    {
      "epoch": 0.26610638598759573,
      "grad_norm": 2.711528778076172,
      "learning_rate": 9.724144114959609e-06,
      "loss": 0.9055,
      "step": 3293
    },
    {
      "epoch": 0.26618719570092325,
      "grad_norm": 3.055825710296631,
      "learning_rate": 9.723929728701487e-06,
      "loss": 0.9706,
      "step": 3294
    },
    {
      "epoch": 0.2662680054142508,
      "grad_norm": 2.645800828933716,
      "learning_rate": 9.723715261533819e-06,
      "loss": 1.0238,
      "step": 3295
    },
    {
      "epoch": 0.26634881512757835,
      "grad_norm": 2.7595863342285156,
      "learning_rate": 9.723500713460282e-06,
      "loss": 1.0574,
      "step": 3296
    },
    {
      "epoch": 0.2664296248409059,
      "grad_norm": 2.8845608234405518,
      "learning_rate": 9.72328608448455e-06,
      "loss": 1.0352,
      "step": 3297
    },
    {
      "epoch": 0.2665104345542334,
      "grad_norm": 2.635479211807251,
      "learning_rate": 9.7230713746103e-06,
      "loss": 1.0123,
      "step": 3298
    },
    {
      "epoch": 0.266591244267561,
      "grad_norm": 2.7326478958129883,
      "learning_rate": 9.722856583841204e-06,
      "loss": 1.0065,
      "step": 3299
    },
    {
      "epoch": 0.2666720539808885,
      "grad_norm": 2.9225542545318604,
      "learning_rate": 9.722641712180946e-06,
      "loss": 0.9945,
      "step": 3300
    },
    {
      "epoch": 0.266752863694216,
      "grad_norm": 3.000427007675171,
      "learning_rate": 9.722426759633206e-06,
      "loss": 0.9869,
      "step": 3301
    },
    {
      "epoch": 0.2668336734075436,
      "grad_norm": 2.937549352645874,
      "learning_rate": 9.722211726201663e-06,
      "loss": 1.0905,
      "step": 3302
    },
    {
      "epoch": 0.2669144831208711,
      "grad_norm": 3.3800082206726074,
      "learning_rate": 9.721996611890001e-06,
      "loss": 0.9916,
      "step": 3303
    },
    {
      "epoch": 0.26699529283419865,
      "grad_norm": 2.5424742698669434,
      "learning_rate": 9.721781416701906e-06,
      "loss": 0.8998,
      "step": 3304
    },
    {
      "epoch": 0.26707610254752623,
      "grad_norm": 2.817131280899048,
      "learning_rate": 9.721566140641061e-06,
      "loss": 1.0004,
      "step": 3305
    },
    {
      "epoch": 0.26715691226085375,
      "grad_norm": 2.5327131748199463,
      "learning_rate": 9.721350783711156e-06,
      "loss": 0.9162,
      "step": 3306
    },
    {
      "epoch": 0.2672377219741813,
      "grad_norm": 2.8354663848876953,
      "learning_rate": 9.72113534591588e-06,
      "loss": 0.9002,
      "step": 3307
    },
    {
      "epoch": 0.26731853168750885,
      "grad_norm": 2.798231363296509,
      "learning_rate": 9.72091982725892e-06,
      "loss": 1.0898,
      "step": 3308
    },
    {
      "epoch": 0.2673993414008364,
      "grad_norm": 2.914630174636841,
      "learning_rate": 9.720704227743967e-06,
      "loss": 0.9134,
      "step": 3309
    },
    {
      "epoch": 0.2674801511141639,
      "grad_norm": 2.7227044105529785,
      "learning_rate": 9.720488547374715e-06,
      "loss": 0.8023,
      "step": 3310
    },
    {
      "epoch": 0.2675609608274915,
      "grad_norm": 2.8826520442962646,
      "learning_rate": 9.720272786154859e-06,
      "loss": 1.0412,
      "step": 3311
    },
    {
      "epoch": 0.267641770540819,
      "grad_norm": 3.4115779399871826,
      "learning_rate": 9.720056944088095e-06,
      "loss": 0.9443,
      "step": 3312
    },
    {
      "epoch": 0.2677225802541465,
      "grad_norm": 2.8294053077697754,
      "learning_rate": 9.719841021178118e-06,
      "loss": 0.961,
      "step": 3313
    },
    {
      "epoch": 0.2678033899674741,
      "grad_norm": 2.9594686031341553,
      "learning_rate": 9.719625017428624e-06,
      "loss": 0.9638,
      "step": 3314
    },
    {
      "epoch": 0.2678841996808016,
      "grad_norm": 2.814594030380249,
      "learning_rate": 9.71940893284332e-06,
      "loss": 0.9381,
      "step": 3315
    },
    {
      "epoch": 0.26796500939412915,
      "grad_norm": 2.804126262664795,
      "learning_rate": 9.7191927674259e-06,
      "loss": 0.9781,
      "step": 3316
    },
    {
      "epoch": 0.26804581910745673,
      "grad_norm": 2.8507261276245117,
      "learning_rate": 9.718976521180068e-06,
      "loss": 1.0205,
      "step": 3317
    },
    {
      "epoch": 0.26812662882078425,
      "grad_norm": 3.2076685428619385,
      "learning_rate": 9.718760194109531e-06,
      "loss": 1.0617,
      "step": 3318
    },
    {
      "epoch": 0.2682074385341118,
      "grad_norm": 3.3740994930267334,
      "learning_rate": 9.71854378621799e-06,
      "loss": 0.9181,
      "step": 3319
    },
    {
      "epoch": 0.26828824824743935,
      "grad_norm": 3.2359611988067627,
      "learning_rate": 9.718327297509154e-06,
      "loss": 1.0289,
      "step": 3320
    },
    {
      "epoch": 0.2683690579607669,
      "grad_norm": 2.7341487407684326,
      "learning_rate": 9.718110727986732e-06,
      "loss": 1.0355,
      "step": 3321
    },
    {
      "epoch": 0.2684498676740944,
      "grad_norm": 2.5096347332000732,
      "learning_rate": 9.71789407765443e-06,
      "loss": 1.0184,
      "step": 3322
    },
    {
      "epoch": 0.268530677387422,
      "grad_norm": 2.4363651275634766,
      "learning_rate": 9.71767734651596e-06,
      "loss": 1.0611,
      "step": 3323
    },
    {
      "epoch": 0.2686114871007495,
      "grad_norm": 2.4556515216827393,
      "learning_rate": 9.717460534575034e-06,
      "loss": 0.997,
      "step": 3324
    },
    {
      "epoch": 0.268692296814077,
      "grad_norm": 2.579828977584839,
      "learning_rate": 9.717243641835367e-06,
      "loss": 1.0561,
      "step": 3325
    },
    {
      "epoch": 0.2687731065274046,
      "grad_norm": 2.8750548362731934,
      "learning_rate": 9.717026668300674e-06,
      "loss": 1.0476,
      "step": 3326
    },
    {
      "epoch": 0.2688539162407321,
      "grad_norm": 2.656571865081787,
      "learning_rate": 9.716809613974667e-06,
      "loss": 0.9635,
      "step": 3327
    },
    {
      "epoch": 0.2689347259540597,
      "grad_norm": 2.723982572555542,
      "learning_rate": 9.716592478861067e-06,
      "loss": 0.9942,
      "step": 3328
    },
    {
      "epoch": 0.26901553566738723,
      "grad_norm": 2.554208755493164,
      "learning_rate": 9.716375262963595e-06,
      "loss": 0.9859,
      "step": 3329
    },
    {
      "epoch": 0.26909634538071475,
      "grad_norm": 2.954664707183838,
      "learning_rate": 9.716157966285966e-06,
      "loss": 0.8633,
      "step": 3330
    },
    {
      "epoch": 0.26917715509404233,
      "grad_norm": 2.7281293869018555,
      "learning_rate": 9.715940588831906e-06,
      "loss": 0.9219,
      "step": 3331
    },
    {
      "epoch": 0.26925796480736985,
      "grad_norm": 2.283665180206299,
      "learning_rate": 9.715723130605139e-06,
      "loss": 0.9913,
      "step": 3332
    },
    {
      "epoch": 0.2693387745206974,
      "grad_norm": 2.8884527683258057,
      "learning_rate": 9.715505591609383e-06,
      "loss": 0.9713,
      "step": 3333
    },
    {
      "epoch": 0.26941958423402496,
      "grad_norm": 3.3616936206817627,
      "learning_rate": 9.715287971848373e-06,
      "loss": 0.9262,
      "step": 3334
    },
    {
      "epoch": 0.2695003939473525,
      "grad_norm": 2.526184558868408,
      "learning_rate": 9.715070271325828e-06,
      "loss": 0.9996,
      "step": 3335
    },
    {
      "epoch": 0.26958120366068,
      "grad_norm": 3.252199411392212,
      "learning_rate": 9.714852490045483e-06,
      "loss": 0.9224,
      "step": 3336
    },
    {
      "epoch": 0.2696620133740076,
      "grad_norm": 2.5886433124542236,
      "learning_rate": 9.714634628011064e-06,
      "loss": 1.0188,
      "step": 3337
    },
    {
      "epoch": 0.2697428230873351,
      "grad_norm": 2.8822431564331055,
      "learning_rate": 9.714416685226305e-06,
      "loss": 1.0581,
      "step": 3338
    },
    {
      "epoch": 0.2698236328006626,
      "grad_norm": 2.8415160179138184,
      "learning_rate": 9.714198661694936e-06,
      "loss": 1.0685,
      "step": 3339
    },
    {
      "epoch": 0.2699044425139902,
      "grad_norm": 3.07682728767395,
      "learning_rate": 9.713980557420693e-06,
      "loss": 1.0587,
      "step": 3340
    },
    {
      "epoch": 0.26998525222731773,
      "grad_norm": 2.992180824279785,
      "learning_rate": 9.713762372407311e-06,
      "loss": 1.1087,
      "step": 3341
    },
    {
      "epoch": 0.27006606194064525,
      "grad_norm": 2.7979042530059814,
      "learning_rate": 9.71354410665853e-06,
      "loss": 0.979,
      "step": 3342
    },
    {
      "epoch": 0.27014687165397283,
      "grad_norm": 2.861024856567383,
      "learning_rate": 9.713325760178085e-06,
      "loss": 1.041,
      "step": 3343
    },
    {
      "epoch": 0.27022768136730035,
      "grad_norm": 2.712601900100708,
      "learning_rate": 9.713107332969715e-06,
      "loss": 0.957,
      "step": 3344
    },
    {
      "epoch": 0.2703084910806279,
      "grad_norm": 2.7607052326202393,
      "learning_rate": 9.712888825037164e-06,
      "loss": 1.0124,
      "step": 3345
    },
    {
      "epoch": 0.27038930079395546,
      "grad_norm": 2.5137217044830322,
      "learning_rate": 9.712670236384172e-06,
      "loss": 0.9161,
      "step": 3346
    },
    {
      "epoch": 0.270470110507283,
      "grad_norm": 3.13071870803833,
      "learning_rate": 9.712451567014485e-06,
      "loss": 1.0568,
      "step": 3347
    },
    {
      "epoch": 0.2705509202206105,
      "grad_norm": 2.989260673522949,
      "learning_rate": 9.712232816931848e-06,
      "loss": 0.9237,
      "step": 3348
    },
    {
      "epoch": 0.2706317299339381,
      "grad_norm": 2.614856004714966,
      "learning_rate": 9.712013986140006e-06,
      "loss": 1.0629,
      "step": 3349
    },
    {
      "epoch": 0.2707125396472656,
      "grad_norm": 2.7321884632110596,
      "learning_rate": 9.711795074642709e-06,
      "loss": 0.9444,
      "step": 3350
    },
    {
      "epoch": 0.2707933493605931,
      "grad_norm": 2.5214805603027344,
      "learning_rate": 9.711576082443705e-06,
      "loss": 0.9694,
      "step": 3351
    },
    {
      "epoch": 0.2708741590739207,
      "grad_norm": 2.974591016769409,
      "learning_rate": 9.711357009546746e-06,
      "loss": 1.0546,
      "step": 3352
    },
    {
      "epoch": 0.27095496878724823,
      "grad_norm": 2.881301164627075,
      "learning_rate": 9.711137855955584e-06,
      "loss": 0.9214,
      "step": 3353
    },
    {
      "epoch": 0.27103577850057575,
      "grad_norm": 3.0097289085388184,
      "learning_rate": 9.71091862167397e-06,
      "loss": 1.0195,
      "step": 3354
    },
    {
      "epoch": 0.27111658821390333,
      "grad_norm": 2.7289974689483643,
      "learning_rate": 9.710699306705664e-06,
      "loss": 0.9618,
      "step": 3355
    },
    {
      "epoch": 0.27119739792723085,
      "grad_norm": 2.989780902862549,
      "learning_rate": 9.710479911054417e-06,
      "loss": 1.045,
      "step": 3356
    },
    {
      "epoch": 0.2712782076405584,
      "grad_norm": 2.8268089294433594,
      "learning_rate": 9.71026043472399e-06,
      "loss": 1.0552,
      "step": 3357
    },
    {
      "epoch": 0.27135901735388596,
      "grad_norm": 2.8823611736297607,
      "learning_rate": 9.710040877718142e-06,
      "loss": 0.9612,
      "step": 3358
    },
    {
      "epoch": 0.2714398270672135,
      "grad_norm": 2.8212215900421143,
      "learning_rate": 9.709821240040632e-06,
      "loss": 0.9574,
      "step": 3359
    },
    {
      "epoch": 0.271520636780541,
      "grad_norm": 2.992547035217285,
      "learning_rate": 9.709601521695223e-06,
      "loss": 0.9661,
      "step": 3360
    },
    {
      "epoch": 0.2716014464938686,
      "grad_norm": 2.427157402038574,
      "learning_rate": 9.709381722685675e-06,
      "loss": 0.9108,
      "step": 3361
    },
    {
      "epoch": 0.2716822562071961,
      "grad_norm": 3.3962388038635254,
      "learning_rate": 9.70916184301576e-06,
      "loss": 0.9932,
      "step": 3362
    },
    {
      "epoch": 0.2717630659205236,
      "grad_norm": 2.3203012943267822,
      "learning_rate": 9.708941882689236e-06,
      "loss": 1.1635,
      "step": 3363
    },
    {
      "epoch": 0.2718438756338512,
      "grad_norm": 2.8734757900238037,
      "learning_rate": 9.708721841709875e-06,
      "loss": 0.9125,
      "step": 3364
    },
    {
      "epoch": 0.27192468534717873,
      "grad_norm": 3.199871778488159,
      "learning_rate": 9.708501720081445e-06,
      "loss": 1.0013,
      "step": 3365
    },
    {
      "epoch": 0.27200549506050625,
      "grad_norm": 2.4899682998657227,
      "learning_rate": 9.708281517807717e-06,
      "loss": 0.9772,
      "step": 3366
    },
    {
      "epoch": 0.27208630477383383,
      "grad_norm": 2.588430643081665,
      "learning_rate": 9.70806123489246e-06,
      "loss": 1.0641,
      "step": 3367
    },
    {
      "epoch": 0.27216711448716135,
      "grad_norm": 2.7496261596679688,
      "learning_rate": 9.707840871339447e-06,
      "loss": 1.0542,
      "step": 3368
    },
    {
      "epoch": 0.2722479242004889,
      "grad_norm": 2.8912103176116943,
      "learning_rate": 9.707620427152455e-06,
      "loss": 1.1012,
      "step": 3369
    },
    {
      "epoch": 0.27232873391381646,
      "grad_norm": 2.703918695449829,
      "learning_rate": 9.707399902335258e-06,
      "loss": 0.997,
      "step": 3370
    },
    {
      "epoch": 0.272409543627144,
      "grad_norm": 2.8338420391082764,
      "learning_rate": 9.707179296891633e-06,
      "loss": 1.0285,
      "step": 3371
    },
    {
      "epoch": 0.2724903533404715,
      "grad_norm": 2.996202230453491,
      "learning_rate": 9.706958610825359e-06,
      "loss": 0.8792,
      "step": 3372
    },
    {
      "epoch": 0.2725711630537991,
      "grad_norm": 3.0629310607910156,
      "learning_rate": 9.706737844140216e-06,
      "loss": 0.9494,
      "step": 3373
    },
    {
      "epoch": 0.2726519727671266,
      "grad_norm": 2.534231424331665,
      "learning_rate": 9.706516996839983e-06,
      "loss": 0.9461,
      "step": 3374
    },
    {
      "epoch": 0.27273278248045413,
      "grad_norm": 2.7926554679870605,
      "learning_rate": 9.706296068928446e-06,
      "loss": 0.9521,
      "step": 3375
    },
    {
      "epoch": 0.2728135921937817,
      "grad_norm": 2.740150213241577,
      "learning_rate": 9.706075060409387e-06,
      "loss": 1.0774,
      "step": 3376
    },
    {
      "epoch": 0.27289440190710923,
      "grad_norm": 2.7504961490631104,
      "learning_rate": 9.705853971286592e-06,
      "loss": 1.0194,
      "step": 3377
    },
    {
      "epoch": 0.27297521162043675,
      "grad_norm": 2.932079315185547,
      "learning_rate": 9.705632801563846e-06,
      "loss": 0.9249,
      "step": 3378
    },
    {
      "epoch": 0.27305602133376433,
      "grad_norm": 3.284442901611328,
      "learning_rate": 9.70541155124494e-06,
      "loss": 0.9577,
      "step": 3379
    },
    {
      "epoch": 0.27313683104709185,
      "grad_norm": 2.3821094036102295,
      "learning_rate": 9.70519022033366e-06,
      "loss": 1.0973,
      "step": 3380
    },
    {
      "epoch": 0.2732176407604194,
      "grad_norm": 2.890260934829712,
      "learning_rate": 9.7049688088338e-06,
      "loss": 0.96,
      "step": 3381
    },
    {
      "epoch": 0.27329845047374696,
      "grad_norm": 2.68121337890625,
      "learning_rate": 9.704747316749152e-06,
      "loss": 1.0355,
      "step": 3382
    },
    {
      "epoch": 0.2733792601870745,
      "grad_norm": 2.871030330657959,
      "learning_rate": 9.704525744083506e-06,
      "loss": 1.0175,
      "step": 3383
    },
    {
      "epoch": 0.273460069900402,
      "grad_norm": 2.8889451026916504,
      "learning_rate": 9.704304090840662e-06,
      "loss": 1.0256,
      "step": 3384
    },
    {
      "epoch": 0.2735408796137296,
      "grad_norm": 2.9536449909210205,
      "learning_rate": 9.704082357024414e-06,
      "loss": 1.0409,
      "step": 3385
    },
    {
      "epoch": 0.2736216893270571,
      "grad_norm": 3.0326268672943115,
      "learning_rate": 9.703860542638558e-06,
      "loss": 0.8903,
      "step": 3386
    },
    {
      "epoch": 0.27370249904038463,
      "grad_norm": 2.743579149246216,
      "learning_rate": 9.703638647686898e-06,
      "loss": 1.0098,
      "step": 3387
    },
    {
      "epoch": 0.2737833087537122,
      "grad_norm": 2.934258460998535,
      "learning_rate": 9.703416672173229e-06,
      "loss": 1.0523,
      "step": 3388
    },
    {
      "epoch": 0.27386411846703973,
      "grad_norm": 2.703131675720215,
      "learning_rate": 9.703194616101356e-06,
      "loss": 0.9418,
      "step": 3389
    },
    {
      "epoch": 0.27394492818036725,
      "grad_norm": 2.9281959533691406,
      "learning_rate": 9.702972479475082e-06,
      "loss": 0.9658,
      "step": 3390
    },
    {
      "epoch": 0.27402573789369483,
      "grad_norm": 3.2979321479797363,
      "learning_rate": 9.702750262298212e-06,
      "loss": 0.9057,
      "step": 3391
    },
    {
      "epoch": 0.27410654760702235,
      "grad_norm": 3.229180335998535,
      "learning_rate": 9.70252796457455e-06,
      "loss": 0.8603,
      "step": 3392
    },
    {
      "epoch": 0.27418735732034993,
      "grad_norm": 2.976534843444824,
      "learning_rate": 9.702305586307906e-06,
      "loss": 0.9375,
      "step": 3393
    },
    {
      "epoch": 0.27426816703367746,
      "grad_norm": 2.718949317932129,
      "learning_rate": 9.702083127502087e-06,
      "loss": 1.0614,
      "step": 3394
    },
    {
      "epoch": 0.274348976747005,
      "grad_norm": 2.5181586742401123,
      "learning_rate": 9.701860588160903e-06,
      "loss": 0.9732,
      "step": 3395
    },
    {
      "epoch": 0.27442978646033256,
      "grad_norm": 2.5775434970855713,
      "learning_rate": 9.701637968288168e-06,
      "loss": 1.0289,
      "step": 3396
    },
    {
      "epoch": 0.2745105961736601,
      "grad_norm": 2.569443464279175,
      "learning_rate": 9.701415267887693e-06,
      "loss": 1.0845,
      "step": 3397
    },
    {
      "epoch": 0.2745914058869876,
      "grad_norm": 2.777613639831543,
      "learning_rate": 9.701192486963293e-06,
      "loss": 0.9458,
      "step": 3398
    },
    {
      "epoch": 0.2746722156003152,
      "grad_norm": 2.5597851276397705,
      "learning_rate": 9.700969625518784e-06,
      "loss": 1.0283,
      "step": 3399
    },
    {
      "epoch": 0.2747530253136427,
      "grad_norm": 2.84451961517334,
      "learning_rate": 9.70074668355798e-06,
      "loss": 1.0286,
      "step": 3400
    },
    {
      "epoch": 0.27483383502697023,
      "grad_norm": 3.0637969970703125,
      "learning_rate": 9.700523661084703e-06,
      "loss": 1.0789,
      "step": 3401
    },
    {
      "epoch": 0.2749146447402978,
      "grad_norm": 2.773308753967285,
      "learning_rate": 9.700300558102773e-06,
      "loss": 0.9973,
      "step": 3402
    },
    {
      "epoch": 0.27499545445362533,
      "grad_norm": 2.556149482727051,
      "learning_rate": 9.700077374616009e-06,
      "loss": 0.8536,
      "step": 3403
    },
    {
      "epoch": 0.27507626416695286,
      "grad_norm": 3.3209409713745117,
      "learning_rate": 9.699854110628233e-06,
      "loss": 0.8942,
      "step": 3404
    },
    {
      "epoch": 0.27515707388028043,
      "grad_norm": 2.7973313331604004,
      "learning_rate": 9.699630766143273e-06,
      "loss": 1.044,
      "step": 3405
    },
    {
      "epoch": 0.27523788359360796,
      "grad_norm": 2.9914135932922363,
      "learning_rate": 9.69940734116495e-06,
      "loss": 1.0301,
      "step": 3406
    },
    {
      "epoch": 0.2753186933069355,
      "grad_norm": 2.7159347534179688,
      "learning_rate": 9.699183835697092e-06,
      "loss": 0.9963,
      "step": 3407
    },
    {
      "epoch": 0.27539950302026306,
      "grad_norm": 2.775402069091797,
      "learning_rate": 9.69896024974353e-06,
      "loss": 1.0171,
      "step": 3408
    },
    {
      "epoch": 0.2754803127335906,
      "grad_norm": 3.0971827507019043,
      "learning_rate": 9.69873658330809e-06,
      "loss": 1.025,
      "step": 3409
    },
    {
      "epoch": 0.2755611224469181,
      "grad_norm": 2.6890649795532227,
      "learning_rate": 9.698512836394605e-06,
      "loss": 0.9742,
      "step": 3410
    },
    {
      "epoch": 0.2756419321602457,
      "grad_norm": 2.641848564147949,
      "learning_rate": 9.698289009006904e-06,
      "loss": 0.9539,
      "step": 3411
    },
    {
      "epoch": 0.2757227418735732,
      "grad_norm": 2.791677713394165,
      "learning_rate": 9.698065101148826e-06,
      "loss": 0.9554,
      "step": 3412
    },
    {
      "epoch": 0.27580355158690073,
      "grad_norm": 2.483377695083618,
      "learning_rate": 9.697841112824202e-06,
      "loss": 1.0689,
      "step": 3413
    },
    {
      "epoch": 0.2758843613002283,
      "grad_norm": 3.3745176792144775,
      "learning_rate": 9.697617044036868e-06,
      "loss": 1.0128,
      "step": 3414
    },
    {
      "epoch": 0.27596517101355583,
      "grad_norm": 2.8752527236938477,
      "learning_rate": 9.697392894790665e-06,
      "loss": 1.1637,
      "step": 3415
    },
    {
      "epoch": 0.27604598072688336,
      "grad_norm": 2.905606985092163,
      "learning_rate": 9.69716866508943e-06,
      "loss": 0.9819,
      "step": 3416
    },
    {
      "epoch": 0.27612679044021093,
      "grad_norm": 3.0775539875030518,
      "learning_rate": 9.696944354937e-06,
      "loss": 0.9637,
      "step": 3417
    },
    {
      "epoch": 0.27620760015353846,
      "grad_norm": 3.1826045513153076,
      "learning_rate": 9.696719964337224e-06,
      "loss": 0.9855,
      "step": 3418
    },
    {
      "epoch": 0.276288409866866,
      "grad_norm": 2.7648861408233643,
      "learning_rate": 9.696495493293942e-06,
      "loss": 0.9567,
      "step": 3419
    },
    {
      "epoch": 0.27636921958019356,
      "grad_norm": 2.7583377361297607,
      "learning_rate": 9.696270941811e-06,
      "loss": 1.007,
      "step": 3420
    },
    {
      "epoch": 0.2764500292935211,
      "grad_norm": 2.6436402797698975,
      "learning_rate": 9.69604630989224e-06,
      "loss": 1.01,
      "step": 3421
    },
    {
      "epoch": 0.2765308390068486,
      "grad_norm": 3.6561508178710938,
      "learning_rate": 9.695821597541512e-06,
      "loss": 0.9828,
      "step": 3422
    },
    {
      "epoch": 0.2766116487201762,
      "grad_norm": 2.9194488525390625,
      "learning_rate": 9.695596804762666e-06,
      "loss": 0.8796,
      "step": 3423
    },
    {
      "epoch": 0.2766924584335037,
      "grad_norm": 2.4754891395568848,
      "learning_rate": 9.69537193155955e-06,
      "loss": 1.0016,
      "step": 3424
    },
    {
      "epoch": 0.27677326814683123,
      "grad_norm": 2.4715240001678467,
      "learning_rate": 9.695146977936016e-06,
      "loss": 0.8923,
      "step": 3425
    },
    {
      "epoch": 0.2768540778601588,
      "grad_norm": 2.36635422706604,
      "learning_rate": 9.694921943895918e-06,
      "loss": 1.0358,
      "step": 3426
    },
    {
      "epoch": 0.27693488757348633,
      "grad_norm": 2.7679364681243896,
      "learning_rate": 9.694696829443112e-06,
      "loss": 1.0525,
      "step": 3427
    },
    {
      "epoch": 0.27701569728681386,
      "grad_norm": 2.478515625,
      "learning_rate": 9.694471634581447e-06,
      "loss": 0.9719,
      "step": 3428
    },
    {
      "epoch": 0.27709650700014143,
      "grad_norm": 2.82259202003479,
      "learning_rate": 9.694246359314787e-06,
      "loss": 1.0163,
      "step": 3429
    },
    {
      "epoch": 0.27717731671346896,
      "grad_norm": 2.9287900924682617,
      "learning_rate": 9.694021003646987e-06,
      "loss": 1.013,
      "step": 3430
    },
    {
      "epoch": 0.2772581264267965,
      "grad_norm": 2.914734363555908,
      "learning_rate": 9.693795567581907e-06,
      "loss": 0.9487,
      "step": 3431
    },
    {
      "epoch": 0.27733893614012406,
      "grad_norm": 2.637640953063965,
      "learning_rate": 9.693570051123412e-06,
      "loss": 1.0951,
      "step": 3432
    },
    {
      "epoch": 0.2774197458534516,
      "grad_norm": 2.958834171295166,
      "learning_rate": 9.693344454275358e-06,
      "loss": 0.9708,
      "step": 3433
    },
    {
      "epoch": 0.2775005555667791,
      "grad_norm": 2.638725757598877,
      "learning_rate": 9.693118777041612e-06,
      "loss": 1.0823,
      "step": 3434
    },
    {
      "epoch": 0.2775813652801067,
      "grad_norm": 2.6947219371795654,
      "learning_rate": 9.692893019426042e-06,
      "loss": 1.0269,
      "step": 3435
    },
    {
      "epoch": 0.2776621749934342,
      "grad_norm": 2.7380142211914062,
      "learning_rate": 9.692667181432512e-06,
      "loss": 0.9439,
      "step": 3436
    },
    {
      "epoch": 0.27774298470676173,
      "grad_norm": 2.9021363258361816,
      "learning_rate": 9.692441263064889e-06,
      "loss": 0.9586,
      "step": 3437
    },
    {
      "epoch": 0.2778237944200893,
      "grad_norm": 3.3157687187194824,
      "learning_rate": 9.692215264327042e-06,
      "loss": 1.0625,
      "step": 3438
    },
    {
      "epoch": 0.27790460413341683,
      "grad_norm": 2.8448753356933594,
      "learning_rate": 9.691989185222847e-06,
      "loss": 1.0707,
      "step": 3439
    },
    {
      "epoch": 0.27798541384674436,
      "grad_norm": 3.3725030422210693,
      "learning_rate": 9.691763025756171e-06,
      "loss": 0.9813,
      "step": 3440
    },
    {
      "epoch": 0.27806622356007193,
      "grad_norm": 2.9721672534942627,
      "learning_rate": 9.691536785930891e-06,
      "loss": 0.9135,
      "step": 3441
    },
    {
      "epoch": 0.27814703327339946,
      "grad_norm": 2.909263849258423,
      "learning_rate": 9.691310465750879e-06,
      "loss": 1.0183,
      "step": 3442
    },
    {
      "epoch": 0.278227842986727,
      "grad_norm": 2.8733043670654297,
      "learning_rate": 9.691084065220013e-06,
      "loss": 1.2139,
      "step": 3443
    },
    {
      "epoch": 0.27830865270005456,
      "grad_norm": 2.779677152633667,
      "learning_rate": 9.69085758434217e-06,
      "loss": 1.0551,
      "step": 3444
    },
    {
      "epoch": 0.2783894624133821,
      "grad_norm": 2.7430167198181152,
      "learning_rate": 9.690631023121228e-06,
      "loss": 0.9888,
      "step": 3445
    },
    {
      "epoch": 0.2784702721267096,
      "grad_norm": 3.428687810897827,
      "learning_rate": 9.690404381561072e-06,
      "loss": 0.979,
      "step": 3446
    },
    {
      "epoch": 0.2785510818400372,
      "grad_norm": 2.6249406337738037,
      "learning_rate": 9.690177659665578e-06,
      "loss": 1.055,
      "step": 3447
    },
    {
      "epoch": 0.2786318915533647,
      "grad_norm": 2.9783623218536377,
      "learning_rate": 9.689950857438632e-06,
      "loss": 0.9662,
      "step": 3448
    },
    {
      "epoch": 0.27871270126669223,
      "grad_norm": 2.5768818855285645,
      "learning_rate": 9.68972397488412e-06,
      "loss": 0.9955,
      "step": 3449
    },
    {
      "epoch": 0.2787935109800198,
      "grad_norm": 2.944002389907837,
      "learning_rate": 9.689497012005924e-06,
      "loss": 1.1916,
      "step": 3450
    },
    {
      "epoch": 0.27887432069334733,
      "grad_norm": 3.232891321182251,
      "learning_rate": 9.689269968807936e-06,
      "loss": 1.0031,
      "step": 3451
    },
    {
      "epoch": 0.27895513040667486,
      "grad_norm": 2.5034186840057373,
      "learning_rate": 9.689042845294041e-06,
      "loss": 1.0152,
      "step": 3452
    },
    {
      "epoch": 0.27903594012000243,
      "grad_norm": 3.2794289588928223,
      "learning_rate": 9.688815641468131e-06,
      "loss": 1.1153,
      "step": 3453
    },
    {
      "epoch": 0.27911674983332996,
      "grad_norm": 2.6757853031158447,
      "learning_rate": 9.688588357334096e-06,
      "loss": 0.921,
      "step": 3454
    },
    {
      "epoch": 0.2791975595466575,
      "grad_norm": 3.212266445159912,
      "learning_rate": 9.688360992895832e-06,
      "loss": 0.9788,
      "step": 3455
    },
    {
      "epoch": 0.27927836925998506,
      "grad_norm": 2.9902377128601074,
      "learning_rate": 9.68813354815723e-06,
      "loss": 1.1899,
      "step": 3456
    },
    {
      "epoch": 0.2793591789733126,
      "grad_norm": 2.3348207473754883,
      "learning_rate": 9.687906023122184e-06,
      "loss": 0.9246,
      "step": 3457
    },
    {
      "epoch": 0.27943998868664016,
      "grad_norm": 3.160766363143921,
      "learning_rate": 9.687678417794597e-06,
      "loss": 1.1292,
      "step": 3458
    },
    {
      "epoch": 0.2795207983999677,
      "grad_norm": 2.848845958709717,
      "learning_rate": 9.687450732178363e-06,
      "loss": 0.8698,
      "step": 3459
    },
    {
      "epoch": 0.2796016081132952,
      "grad_norm": 3.4590060710906982,
      "learning_rate": 9.687222966277381e-06,
      "loss": 1.089,
      "step": 3460
    },
    {
      "epoch": 0.2796824178266228,
      "grad_norm": 2.869417667388916,
      "learning_rate": 9.686995120095555e-06,
      "loss": 0.915,
      "step": 3461
    },
    {
      "epoch": 0.2797632275399503,
      "grad_norm": 2.422532081604004,
      "learning_rate": 9.686767193636785e-06,
      "loss": 0.9609,
      "step": 3462
    },
    {
      "epoch": 0.27984403725327783,
      "grad_norm": 3.6336019039154053,
      "learning_rate": 9.686539186904977e-06,
      "loss": 0.9827,
      "step": 3463
    },
    {
      "epoch": 0.2799248469666054,
      "grad_norm": 3.015155076980591,
      "learning_rate": 9.686311099904034e-06,
      "loss": 1.1245,
      "step": 3464
    },
    {
      "epoch": 0.28000565667993293,
      "grad_norm": 2.3626017570495605,
      "learning_rate": 9.686082932637864e-06,
      "loss": 0.9372,
      "step": 3465
    },
    {
      "epoch": 0.28008646639326046,
      "grad_norm": 2.927370309829712,
      "learning_rate": 9.685854685110376e-06,
      "loss": 1.0378,
      "step": 3466
    },
    {
      "epoch": 0.28016727610658804,
      "grad_norm": 2.8526339530944824,
      "learning_rate": 9.685626357325477e-06,
      "loss": 0.9278,
      "step": 3467
    },
    {
      "epoch": 0.28024808581991556,
      "grad_norm": 2.72507643699646,
      "learning_rate": 9.685397949287079e-06,
      "loss": 1.044,
      "step": 3468
    },
    {
      "epoch": 0.2803288955332431,
      "grad_norm": 2.673955202102661,
      "learning_rate": 9.685169460999093e-06,
      "loss": 1.0791,
      "step": 3469
    },
    {
      "epoch": 0.28040970524657066,
      "grad_norm": 2.734114646911621,
      "learning_rate": 9.684940892465434e-06,
      "loss": 0.9674,
      "step": 3470
    },
    {
      "epoch": 0.2804905149598982,
      "grad_norm": 2.4035141468048096,
      "learning_rate": 9.684712243690015e-06,
      "loss": 0.941,
      "step": 3471
    },
    {
      "epoch": 0.2805713246732257,
      "grad_norm": 3.046450138092041,
      "learning_rate": 9.684483514676755e-06,
      "loss": 1.0485,
      "step": 3472
    },
    {
      "epoch": 0.2806521343865533,
      "grad_norm": 2.5257482528686523,
      "learning_rate": 9.684254705429568e-06,
      "loss": 0.881,
      "step": 3473
    },
    {
      "epoch": 0.2807329440998808,
      "grad_norm": 3.0747008323669434,
      "learning_rate": 9.684025815952375e-06,
      "loss": 1.0863,
      "step": 3474
    },
    {
      "epoch": 0.28081375381320833,
      "grad_norm": 2.855036497116089,
      "learning_rate": 9.683796846249097e-06,
      "loss": 1.0607,
      "step": 3475
    },
    {
      "epoch": 0.2808945635265359,
      "grad_norm": 2.654651403427124,
      "learning_rate": 9.683567796323654e-06,
      "loss": 0.9433,
      "step": 3476
    },
    {
      "epoch": 0.28097537323986344,
      "grad_norm": 2.6321558952331543,
      "learning_rate": 9.683338666179971e-06,
      "loss": 0.9739,
      "step": 3477
    },
    {
      "epoch": 0.28105618295319096,
      "grad_norm": 2.620388984680176,
      "learning_rate": 9.683109455821972e-06,
      "loss": 1.0461,
      "step": 3478
    },
    {
      "epoch": 0.28113699266651854,
      "grad_norm": 2.774127244949341,
      "learning_rate": 9.68288016525358e-06,
      "loss": 1.0158,
      "step": 3479
    },
    {
      "epoch": 0.28121780237984606,
      "grad_norm": 3.1548571586608887,
      "learning_rate": 9.682650794478725e-06,
      "loss": 0.8939,
      "step": 3480
    },
    {
      "epoch": 0.2812986120931736,
      "grad_norm": 2.7580490112304688,
      "learning_rate": 9.682421343501335e-06,
      "loss": 1.0513,
      "step": 3481
    },
    {
      "epoch": 0.28137942180650116,
      "grad_norm": 2.82216215133667,
      "learning_rate": 9.68219181232534e-06,
      "loss": 0.9678,
      "step": 3482
    },
    {
      "epoch": 0.2814602315198287,
      "grad_norm": 2.7549870014190674,
      "learning_rate": 9.681962200954671e-06,
      "loss": 0.9741,
      "step": 3483
    },
    {
      "epoch": 0.2815410412331562,
      "grad_norm": 2.6894516944885254,
      "learning_rate": 9.68173250939326e-06,
      "loss": 0.9422,
      "step": 3484
    },
    {
      "epoch": 0.2816218509464838,
      "grad_norm": 2.5008151531219482,
      "learning_rate": 9.681502737645043e-06,
      "loss": 0.9906,
      "step": 3485
    },
    {
      "epoch": 0.2817026606598113,
      "grad_norm": 2.5604753494262695,
      "learning_rate": 9.681272885713955e-06,
      "loss": 0.9815,
      "step": 3486
    },
    {
      "epoch": 0.28178347037313883,
      "grad_norm": 3.2388126850128174,
      "learning_rate": 9.681042953603932e-06,
      "loss": 0.9839,
      "step": 3487
    },
    {
      "epoch": 0.2818642800864664,
      "grad_norm": 2.8115487098693848,
      "learning_rate": 9.68081294131891e-06,
      "loss": 0.9874,
      "step": 3488
    },
    {
      "epoch": 0.28194508979979394,
      "grad_norm": 2.4315402507781982,
      "learning_rate": 9.680582848862834e-06,
      "loss": 0.8547,
      "step": 3489
    },
    {
      "epoch": 0.28202589951312146,
      "grad_norm": 2.8511130809783936,
      "learning_rate": 9.680352676239641e-06,
      "loss": 1.012,
      "step": 3490
    },
    {
      "epoch": 0.28210670922644904,
      "grad_norm": 3.0524845123291016,
      "learning_rate": 9.680122423453272e-06,
      "loss": 0.9357,
      "step": 3491
    },
    {
      "epoch": 0.28218751893977656,
      "grad_norm": 3.1882684230804443,
      "learning_rate": 9.679892090507676e-06,
      "loss": 1.0719,
      "step": 3492
    },
    {
      "epoch": 0.2822683286531041,
      "grad_norm": 2.989455461502075,
      "learning_rate": 9.679661677406793e-06,
      "loss": 1.0933,
      "step": 3493
    },
    {
      "epoch": 0.28234913836643166,
      "grad_norm": 2.903139114379883,
      "learning_rate": 9.679431184154572e-06,
      "loss": 0.8652,
      "step": 3494
    },
    {
      "epoch": 0.2824299480797592,
      "grad_norm": 2.8893330097198486,
      "learning_rate": 9.67920061075496e-06,
      "loss": 0.9554,
      "step": 3495
    },
    {
      "epoch": 0.2825107577930867,
      "grad_norm": 2.6712095737457275,
      "learning_rate": 9.678969957211905e-06,
      "loss": 0.9762,
      "step": 3496
    },
    {
      "epoch": 0.2825915675064143,
      "grad_norm": 2.954951763153076,
      "learning_rate": 9.67873922352936e-06,
      "loss": 0.964,
      "step": 3497
    },
    {
      "epoch": 0.2826723772197418,
      "grad_norm": 2.700484037399292,
      "learning_rate": 9.678508409711276e-06,
      "loss": 1.0128,
      "step": 3498
    },
    {
      "epoch": 0.28275318693306933,
      "grad_norm": 2.642106533050537,
      "learning_rate": 9.678277515761605e-06,
      "loss": 0.9558,
      "step": 3499
    },
    {
      "epoch": 0.2828339966463969,
      "grad_norm": 2.848428964614868,
      "learning_rate": 9.678046541684302e-06,
      "loss": 1.0457,
      "step": 3500
    },
    {
      "epoch": 0.28291480635972444,
      "grad_norm": 2.7813591957092285,
      "learning_rate": 9.677815487483326e-06,
      "loss": 0.9187,
      "step": 3501
    },
    {
      "epoch": 0.28299561607305196,
      "grad_norm": 2.9099740982055664,
      "learning_rate": 9.67758435316263e-06,
      "loss": 1.1141,
      "step": 3502
    },
    {
      "epoch": 0.28307642578637954,
      "grad_norm": 2.7438671588897705,
      "learning_rate": 9.677353138726177e-06,
      "loss": 0.9629,
      "step": 3503
    },
    {
      "epoch": 0.28315723549970706,
      "grad_norm": 2.780836820602417,
      "learning_rate": 9.677121844177923e-06,
      "loss": 0.9671,
      "step": 3504
    },
    {
      "epoch": 0.2832380452130346,
      "grad_norm": 2.482347011566162,
      "learning_rate": 9.676890469521833e-06,
      "loss": 1.0432,
      "step": 3505
    },
    {
      "epoch": 0.28331885492636216,
      "grad_norm": 2.5355114936828613,
      "learning_rate": 9.676659014761868e-06,
      "loss": 1.0369,
      "step": 3506
    },
    {
      "epoch": 0.2833996646396897,
      "grad_norm": 2.961168050765991,
      "learning_rate": 9.676427479901991e-06,
      "loss": 1.0697,
      "step": 3507
    },
    {
      "epoch": 0.2834804743530172,
      "grad_norm": 2.63830828666687,
      "learning_rate": 9.676195864946171e-06,
      "loss": 1.0377,
      "step": 3508
    },
    {
      "epoch": 0.2835612840663448,
      "grad_norm": 3.5168895721435547,
      "learning_rate": 9.675964169898373e-06,
      "loss": 0.9061,
      "step": 3509
    },
    {
      "epoch": 0.2836420937796723,
      "grad_norm": 2.7696313858032227,
      "learning_rate": 9.675732394762567e-06,
      "loss": 0.9532,
      "step": 3510
    },
    {
      "epoch": 0.28372290349299983,
      "grad_norm": 2.9334309101104736,
      "learning_rate": 9.675500539542719e-06,
      "loss": 1.0098,
      "step": 3511
    },
    {
      "epoch": 0.2838037132063274,
      "grad_norm": 2.6671390533447266,
      "learning_rate": 9.675268604242804e-06,
      "loss": 1.0216,
      "step": 3512
    },
    {
      "epoch": 0.28388452291965494,
      "grad_norm": 2.7944910526275635,
      "learning_rate": 9.675036588866793e-06,
      "loss": 0.9303,
      "step": 3513
    },
    {
      "epoch": 0.28396533263298246,
      "grad_norm": 2.7365944385528564,
      "learning_rate": 9.674804493418659e-06,
      "loss": 0.9002,
      "step": 3514
    },
    {
      "epoch": 0.28404614234631004,
      "grad_norm": 3.015563726425171,
      "learning_rate": 9.67457231790238e-06,
      "loss": 0.9412,
      "step": 3515
    },
    {
      "epoch": 0.28412695205963756,
      "grad_norm": 2.741112470626831,
      "learning_rate": 9.674340062321929e-06,
      "loss": 0.9404,
      "step": 3516
    },
    {
      "epoch": 0.2842077617729651,
      "grad_norm": 2.7046263217926025,
      "learning_rate": 9.674107726681285e-06,
      "loss": 0.9861,
      "step": 3517
    },
    {
      "epoch": 0.28428857148629266,
      "grad_norm": 2.813877582550049,
      "learning_rate": 9.67387531098443e-06,
      "loss": 1.0162,
      "step": 3518
    },
    {
      "epoch": 0.2843693811996202,
      "grad_norm": 2.9360432624816895,
      "learning_rate": 9.673642815235342e-06,
      "loss": 1.0211,
      "step": 3519
    },
    {
      "epoch": 0.2844501909129477,
      "grad_norm": 2.8569142818450928,
      "learning_rate": 9.673410239438007e-06,
      "loss": 1.0572,
      "step": 3520
    },
    {
      "epoch": 0.2845310006262753,
      "grad_norm": 2.4324591159820557,
      "learning_rate": 9.6731775835964e-06,
      "loss": 0.9912,
      "step": 3521
    },
    {
      "epoch": 0.2846118103396028,
      "grad_norm": 2.9998257160186768,
      "learning_rate": 9.672944847714515e-06,
      "loss": 0.9571,
      "step": 3522
    },
    {
      "epoch": 0.2846926200529304,
      "grad_norm": 2.362875461578369,
      "learning_rate": 9.672712031796332e-06,
      "loss": 0.9492,
      "step": 3523
    },
    {
      "epoch": 0.2847734297662579,
      "grad_norm": 2.6290156841278076,
      "learning_rate": 9.672479135845843e-06,
      "loss": 0.9707,
      "step": 3524
    },
    {
      "epoch": 0.28485423947958544,
      "grad_norm": 2.2613868713378906,
      "learning_rate": 9.672246159867033e-06,
      "loss": 0.9548,
      "step": 3525
    },
    {
      "epoch": 0.284935049192913,
      "grad_norm": 2.750572443008423,
      "learning_rate": 9.672013103863895e-06,
      "loss": 1.036,
      "step": 3526
    },
    {
      "epoch": 0.28501585890624054,
      "grad_norm": 2.6883738040924072,
      "learning_rate": 9.671779967840422e-06,
      "loss": 0.9813,
      "step": 3527
    },
    {
      "epoch": 0.28509666861956806,
      "grad_norm": 3.0548415184020996,
      "learning_rate": 9.671546751800602e-06,
      "loss": 1.0667,
      "step": 3528
    },
    {
      "epoch": 0.28517747833289564,
      "grad_norm": 2.81595516204834,
      "learning_rate": 9.671313455748434e-06,
      "loss": 0.8933,
      "step": 3529
    },
    {
      "epoch": 0.28525828804622316,
      "grad_norm": 2.566643238067627,
      "learning_rate": 9.671080079687913e-06,
      "loss": 1.1025,
      "step": 3530
    },
    {
      "epoch": 0.2853390977595507,
      "grad_norm": 3.4577713012695312,
      "learning_rate": 9.670846623623033e-06,
      "loss": 0.96,
      "step": 3531
    },
    {
      "epoch": 0.28541990747287826,
      "grad_norm": 3.4869630336761475,
      "learning_rate": 9.670613087557797e-06,
      "loss": 1.1067,
      "step": 3532
    },
    {
      "epoch": 0.2855007171862058,
      "grad_norm": 2.777127981185913,
      "learning_rate": 9.670379471496203e-06,
      "loss": 0.9792,
      "step": 3533
    },
    {
      "epoch": 0.2855815268995333,
      "grad_norm": 2.815192699432373,
      "learning_rate": 9.67014577544225e-06,
      "loss": 1.097,
      "step": 3534
    },
    {
      "epoch": 0.2856623366128609,
      "grad_norm": 3.0062084197998047,
      "learning_rate": 9.669911999399945e-06,
      "loss": 1.0119,
      "step": 3535
    },
    {
      "epoch": 0.2857431463261884,
      "grad_norm": 3.3104159832000732,
      "learning_rate": 9.669678143373289e-06,
      "loss": 0.8744,
      "step": 3536
    },
    {
      "epoch": 0.28582395603951594,
      "grad_norm": 2.381932497024536,
      "learning_rate": 9.669444207366288e-06,
      "loss": 0.996,
      "step": 3537
    },
    {
      "epoch": 0.2859047657528435,
      "grad_norm": 2.455183506011963,
      "learning_rate": 9.669210191382949e-06,
      "loss": 1.0022,
      "step": 3538
    },
    {
      "epoch": 0.28598557546617104,
      "grad_norm": 2.8171143531799316,
      "learning_rate": 9.66897609542728e-06,
      "loss": 0.918,
      "step": 3539
    },
    {
      "epoch": 0.28606638517949856,
      "grad_norm": 2.676234006881714,
      "learning_rate": 9.66874191950329e-06,
      "loss": 0.9389,
      "step": 3540
    },
    {
      "epoch": 0.28614719489282614,
      "grad_norm": 2.7240962982177734,
      "learning_rate": 9.668507663614993e-06,
      "loss": 0.9976,
      "step": 3541
    },
    {
      "epoch": 0.28622800460615366,
      "grad_norm": 3.261559009552002,
      "learning_rate": 9.668273327766395e-06,
      "loss": 0.9515,
      "step": 3542
    },
    {
      "epoch": 0.2863088143194812,
      "grad_norm": 3.1276698112487793,
      "learning_rate": 9.668038911961516e-06,
      "loss": 1.0021,
      "step": 3543
    },
    {
      "epoch": 0.28638962403280877,
      "grad_norm": 2.76226544380188,
      "learning_rate": 9.667804416204367e-06,
      "loss": 0.949,
      "step": 3544
    },
    {
      "epoch": 0.2864704337461363,
      "grad_norm": 2.9888455867767334,
      "learning_rate": 9.667569840498966e-06,
      "loss": 0.9896,
      "step": 3545
    },
    {
      "epoch": 0.2865512434594638,
      "grad_norm": 2.889864206314087,
      "learning_rate": 9.667335184849332e-06,
      "loss": 1.0067,
      "step": 3546
    },
    {
      "epoch": 0.2866320531727914,
      "grad_norm": 2.665231466293335,
      "learning_rate": 9.66710044925948e-06,
      "loss": 0.9624,
      "step": 3547
    },
    {
      "epoch": 0.2867128628861189,
      "grad_norm": 2.893362045288086,
      "learning_rate": 9.666865633733434e-06,
      "loss": 0.8477,
      "step": 3548
    },
    {
      "epoch": 0.28679367259944644,
      "grad_norm": 2.84060001373291,
      "learning_rate": 9.666630738275213e-06,
      "loss": 1.0926,
      "step": 3549
    },
    {
      "epoch": 0.286874482312774,
      "grad_norm": 2.3701794147491455,
      "learning_rate": 9.666395762888844e-06,
      "loss": 0.9602,
      "step": 3550
    },
    {
      "epoch": 0.28695529202610154,
      "grad_norm": 2.820270299911499,
      "learning_rate": 9.666160707578349e-06,
      "loss": 1.0241,
      "step": 3551
    },
    {
      "epoch": 0.28703610173942906,
      "grad_norm": 2.936659336090088,
      "learning_rate": 9.665925572347754e-06,
      "loss": 0.9598,
      "step": 3552
    },
    {
      "epoch": 0.28711691145275664,
      "grad_norm": 2.8876888751983643,
      "learning_rate": 9.665690357201087e-06,
      "loss": 0.9806,
      "step": 3553
    },
    {
      "epoch": 0.28719772116608416,
      "grad_norm": 2.5708611011505127,
      "learning_rate": 9.665455062142377e-06,
      "loss": 1.006,
      "step": 3554
    },
    {
      "epoch": 0.2872785308794117,
      "grad_norm": 2.5681517124176025,
      "learning_rate": 9.665219687175652e-06,
      "loss": 1.0493,
      "step": 3555
    },
    {
      "epoch": 0.28735934059273927,
      "grad_norm": 2.6039462089538574,
      "learning_rate": 9.664984232304946e-06,
      "loss": 0.9338,
      "step": 3556
    },
    {
      "epoch": 0.2874401503060668,
      "grad_norm": 2.749727964401245,
      "learning_rate": 9.66474869753429e-06,
      "loss": 0.9797,
      "step": 3557
    },
    {
      "epoch": 0.2875209600193943,
      "grad_norm": 2.9040639400482178,
      "learning_rate": 9.66451308286772e-06,
      "loss": 1.0086,
      "step": 3558
    },
    {
      "epoch": 0.2876017697327219,
      "grad_norm": 2.893544912338257,
      "learning_rate": 9.664277388309268e-06,
      "loss": 0.9645,
      "step": 3559
    },
    {
      "epoch": 0.2876825794460494,
      "grad_norm": 2.9192464351654053,
      "learning_rate": 9.664041613862973e-06,
      "loss": 0.9184,
      "step": 3560
    },
    {
      "epoch": 0.28776338915937694,
      "grad_norm": 2.6505277156829834,
      "learning_rate": 9.663805759532876e-06,
      "loss": 0.9097,
      "step": 3561
    },
    {
      "epoch": 0.2878441988727045,
      "grad_norm": 2.486187219619751,
      "learning_rate": 9.663569825323012e-06,
      "loss": 1.0441,
      "step": 3562
    },
    {
      "epoch": 0.28792500858603204,
      "grad_norm": 2.4875733852386475,
      "learning_rate": 9.663333811237426e-06,
      "loss": 0.9103,
      "step": 3563
    },
    {
      "epoch": 0.28800581829935956,
      "grad_norm": 2.6382365226745605,
      "learning_rate": 9.663097717280157e-06,
      "loss": 0.8666,
      "step": 3564
    },
    {
      "epoch": 0.28808662801268714,
      "grad_norm": 3.0614850521087646,
      "learning_rate": 9.662861543455248e-06,
      "loss": 1.0001,
      "step": 3565
    },
    {
      "epoch": 0.28816743772601466,
      "grad_norm": 2.5728049278259277,
      "learning_rate": 9.662625289766749e-06,
      "loss": 0.9767,
      "step": 3566
    },
    {
      "epoch": 0.2882482474393422,
      "grad_norm": 2.496704339981079,
      "learning_rate": 9.662388956218702e-06,
      "loss": 1.0522,
      "step": 3567
    },
    {
      "epoch": 0.28832905715266977,
      "grad_norm": 2.6987080574035645,
      "learning_rate": 9.662152542815158e-06,
      "loss": 0.8845,
      "step": 3568
    },
    {
      "epoch": 0.2884098668659973,
      "grad_norm": 2.947056770324707,
      "learning_rate": 9.661916049560162e-06,
      "loss": 1.0148,
      "step": 3569
    },
    {
      "epoch": 0.2884906765793248,
      "grad_norm": 3.203179359436035,
      "learning_rate": 9.661679476457771e-06,
      "loss": 0.9221,
      "step": 3570
    },
    {
      "epoch": 0.2885714862926524,
      "grad_norm": 3.0656864643096924,
      "learning_rate": 9.66144282351203e-06,
      "loss": 1.0925,
      "step": 3571
    },
    {
      "epoch": 0.2886522960059799,
      "grad_norm": 3.0500292778015137,
      "learning_rate": 9.661206090726996e-06,
      "loss": 0.9946,
      "step": 3572
    },
    {
      "epoch": 0.28873310571930744,
      "grad_norm": 2.9077112674713135,
      "learning_rate": 9.660969278106724e-06,
      "loss": 0.8432,
      "step": 3573
    },
    {
      "epoch": 0.288813915432635,
      "grad_norm": 3.2144157886505127,
      "learning_rate": 9.66073238565527e-06,
      "loss": 1.082,
      "step": 3574
    },
    {
      "epoch": 0.28889472514596254,
      "grad_norm": 2.6139988899230957,
      "learning_rate": 9.660495413376688e-06,
      "loss": 0.9484,
      "step": 3575
    },
    {
      "epoch": 0.28897553485929006,
      "grad_norm": 3.0894522666931152,
      "learning_rate": 9.66025836127504e-06,
      "loss": 0.8584,
      "step": 3576
    },
    {
      "epoch": 0.28905634457261764,
      "grad_norm": 2.5344035625457764,
      "learning_rate": 9.660021229354386e-06,
      "loss": 0.9804,
      "step": 3577
    },
    {
      "epoch": 0.28913715428594516,
      "grad_norm": 2.615623712539673,
      "learning_rate": 9.659784017618787e-06,
      "loss": 0.9411,
      "step": 3578
    },
    {
      "epoch": 0.2892179639992727,
      "grad_norm": 2.9990971088409424,
      "learning_rate": 9.659546726072306e-06,
      "loss": 1.0582,
      "step": 3579
    },
    {
      "epoch": 0.28929877371260027,
      "grad_norm": 2.8509674072265625,
      "learning_rate": 9.659309354719005e-06,
      "loss": 0.9297,
      "step": 3580
    },
    {
      "epoch": 0.2893795834259278,
      "grad_norm": 2.2842657566070557,
      "learning_rate": 9.659071903562953e-06,
      "loss": 1.0806,
      "step": 3581
    },
    {
      "epoch": 0.2894603931392553,
      "grad_norm": 2.9615931510925293,
      "learning_rate": 9.658834372608216e-06,
      "loss": 0.9645,
      "step": 3582
    },
    {
      "epoch": 0.2895412028525829,
      "grad_norm": 2.4293406009674072,
      "learning_rate": 9.65859676185886e-06,
      "loss": 1.0036,
      "step": 3583
    },
    {
      "epoch": 0.2896220125659104,
      "grad_norm": 3.4946937561035156,
      "learning_rate": 9.65835907131896e-06,
      "loss": 0.844,
      "step": 3584
    },
    {
      "epoch": 0.28970282227923794,
      "grad_norm": 2.9931979179382324,
      "learning_rate": 9.65812130099258e-06,
      "loss": 0.8941,
      "step": 3585
    },
    {
      "epoch": 0.2897836319925655,
      "grad_norm": 3.216002941131592,
      "learning_rate": 9.657883450883798e-06,
      "loss": 1.0303,
      "step": 3586
    },
    {
      "epoch": 0.28986444170589304,
      "grad_norm": 2.9423599243164062,
      "learning_rate": 9.657645520996686e-06,
      "loss": 0.9116,
      "step": 3587
    },
    {
      "epoch": 0.2899452514192206,
      "grad_norm": 2.7516708374023438,
      "learning_rate": 9.657407511335319e-06,
      "loss": 0.9291,
      "step": 3588
    },
    {
      "epoch": 0.29002606113254814,
      "grad_norm": 3.6603939533233643,
      "learning_rate": 9.657169421903772e-06,
      "loss": 1.0703,
      "step": 3589
    },
    {
      "epoch": 0.29010687084587566,
      "grad_norm": 2.783923625946045,
      "learning_rate": 9.656931252706126e-06,
      "loss": 1.0308,
      "step": 3590
    },
    {
      "epoch": 0.29018768055920324,
      "grad_norm": 2.6937367916107178,
      "learning_rate": 9.656693003746458e-06,
      "loss": 0.8576,
      "step": 3591
    },
    {
      "epoch": 0.29026849027253077,
      "grad_norm": 2.9233529567718506,
      "learning_rate": 9.65645467502885e-06,
      "loss": 1.0536,
      "step": 3592
    },
    {
      "epoch": 0.2903492999858583,
      "grad_norm": 2.8620266914367676,
      "learning_rate": 9.656216266557384e-06,
      "loss": 0.9495,
      "step": 3593
    },
    {
      "epoch": 0.29043010969918587,
      "grad_norm": 2.6702728271484375,
      "learning_rate": 9.655977778336142e-06,
      "loss": 1.0955,
      "step": 3594
    },
    {
      "epoch": 0.2905109194125134,
      "grad_norm": 2.7964835166931152,
      "learning_rate": 9.655739210369208e-06,
      "loss": 0.961,
      "step": 3595
    },
    {
      "epoch": 0.2905917291258409,
      "grad_norm": 3.0626680850982666,
      "learning_rate": 9.65550056266067e-06,
      "loss": 0.8856,
      "step": 3596
    },
    {
      "epoch": 0.2906725388391685,
      "grad_norm": 3.4128012657165527,
      "learning_rate": 9.655261835214617e-06,
      "loss": 0.9221,
      "step": 3597
    },
    {
      "epoch": 0.290753348552496,
      "grad_norm": 2.408466339111328,
      "learning_rate": 9.655023028035135e-06,
      "loss": 1.0366,
      "step": 3598
    },
    {
      "epoch": 0.29083415826582354,
      "grad_norm": 2.675818920135498,
      "learning_rate": 9.654784141126315e-06,
      "loss": 0.9772,
      "step": 3599
    },
    {
      "epoch": 0.2909149679791511,
      "grad_norm": 2.5891449451446533,
      "learning_rate": 9.654545174492248e-06,
      "loss": 1.0234,
      "step": 3600
    },
    {
      "epoch": 0.29099577769247864,
      "grad_norm": 2.5514161586761475,
      "learning_rate": 9.654306128137028e-06,
      "loss": 1.1006,
      "step": 3601
    },
    {
      "epoch": 0.29107658740580616,
      "grad_norm": 2.697810173034668,
      "learning_rate": 9.65406700206475e-06,
      "loss": 1.0898,
      "step": 3602
    },
    {
      "epoch": 0.29115739711913374,
      "grad_norm": 2.2984206676483154,
      "learning_rate": 9.653827796279507e-06,
      "loss": 1.0286,
      "step": 3603
    },
    {
      "epoch": 0.29123820683246127,
      "grad_norm": 2.5531578063964844,
      "learning_rate": 9.653588510785398e-06,
      "loss": 1.0689,
      "step": 3604
    },
    {
      "epoch": 0.2913190165457888,
      "grad_norm": 2.9822959899902344,
      "learning_rate": 9.65334914558652e-06,
      "loss": 0.9697,
      "step": 3605
    },
    {
      "epoch": 0.29139982625911637,
      "grad_norm": 2.4546971321105957,
      "learning_rate": 9.653109700686974e-06,
      "loss": 1.0312,
      "step": 3606
    },
    {
      "epoch": 0.2914806359724439,
      "grad_norm": 2.8869683742523193,
      "learning_rate": 9.652870176090862e-06,
      "loss": 1.0018,
      "step": 3607
    },
    {
      "epoch": 0.2915614456857714,
      "grad_norm": 2.554905891418457,
      "learning_rate": 9.652630571802283e-06,
      "loss": 0.8601,
      "step": 3608
    },
    {
      "epoch": 0.291642255399099,
      "grad_norm": 2.709362030029297,
      "learning_rate": 9.652390887825344e-06,
      "loss": 0.961,
      "step": 3609
    },
    {
      "epoch": 0.2917230651124265,
      "grad_norm": 3.170376777648926,
      "learning_rate": 9.65215112416415e-06,
      "loss": 1.0732,
      "step": 3610
    },
    {
      "epoch": 0.29180387482575404,
      "grad_norm": 2.6856372356414795,
      "learning_rate": 9.651911280822806e-06,
      "loss": 0.9184,
      "step": 3611
    },
    {
      "epoch": 0.2918846845390816,
      "grad_norm": 2.742732048034668,
      "learning_rate": 9.651671357805421e-06,
      "loss": 0.8924,
      "step": 3612
    },
    {
      "epoch": 0.29196549425240914,
      "grad_norm": 3.003530740737915,
      "learning_rate": 9.651431355116105e-06,
      "loss": 1.0086,
      "step": 3613
    },
    {
      "epoch": 0.29204630396573666,
      "grad_norm": 2.8207788467407227,
      "learning_rate": 9.651191272758967e-06,
      "loss": 1.0997,
      "step": 3614
    },
    {
      "epoch": 0.29212711367906424,
      "grad_norm": 3.3507986068725586,
      "learning_rate": 9.65095111073812e-06,
      "loss": 0.8922,
      "step": 3615
    },
    {
      "epoch": 0.29220792339239177,
      "grad_norm": 2.6607301235198975,
      "learning_rate": 9.650710869057675e-06,
      "loss": 1.0341,
      "step": 3616
    },
    {
      "epoch": 0.2922887331057193,
      "grad_norm": 2.9044384956359863,
      "learning_rate": 9.650470547721753e-06,
      "loss": 0.9735,
      "step": 3617
    },
    {
      "epoch": 0.29236954281904687,
      "grad_norm": 2.7206690311431885,
      "learning_rate": 9.650230146734463e-06,
      "loss": 0.9322,
      "step": 3618
    },
    {
      "epoch": 0.2924503525323744,
      "grad_norm": 2.93376088142395,
      "learning_rate": 9.649989666099926e-06,
      "loss": 0.9781,
      "step": 3619
    },
    {
      "epoch": 0.2925311622457019,
      "grad_norm": 2.40251088142395,
      "learning_rate": 9.649749105822261e-06,
      "loss": 1.1007,
      "step": 3620
    },
    {
      "epoch": 0.2926119719590295,
      "grad_norm": 2.493821859359741,
      "learning_rate": 9.649508465905589e-06,
      "loss": 1.0024,
      "step": 3621
    },
    {
      "epoch": 0.292692781672357,
      "grad_norm": 2.7107715606689453,
      "learning_rate": 9.649267746354027e-06,
      "loss": 1.0928,
      "step": 3622
    },
    {
      "epoch": 0.29277359138568454,
      "grad_norm": 3.0153157711029053,
      "learning_rate": 9.649026947171703e-06,
      "loss": 1.0024,
      "step": 3623
    },
    {
      "epoch": 0.2928544010990121,
      "grad_norm": 2.727877378463745,
      "learning_rate": 9.64878606836274e-06,
      "loss": 1.0563,
      "step": 3624
    },
    {
      "epoch": 0.29293521081233964,
      "grad_norm": 2.567570686340332,
      "learning_rate": 9.648545109931262e-06,
      "loss": 0.9565,
      "step": 3625
    },
    {
      "epoch": 0.29301602052566716,
      "grad_norm": 3.3060126304626465,
      "learning_rate": 9.648304071881398e-06,
      "loss": 1.1092,
      "step": 3626
    },
    {
      "epoch": 0.29309683023899474,
      "grad_norm": 2.8973357677459717,
      "learning_rate": 9.648062954217275e-06,
      "loss": 0.9373,
      "step": 3627
    },
    {
      "epoch": 0.29317763995232227,
      "grad_norm": 2.6033413410186768,
      "learning_rate": 9.647821756943023e-06,
      "loss": 1.1346,
      "step": 3628
    },
    {
      "epoch": 0.2932584496656498,
      "grad_norm": 2.7342493534088135,
      "learning_rate": 9.647580480062775e-06,
      "loss": 0.9699,
      "step": 3629
    },
    {
      "epoch": 0.29333925937897737,
      "grad_norm": 2.461733818054199,
      "learning_rate": 9.647339123580662e-06,
      "loss": 1.0773,
      "step": 3630
    },
    {
      "epoch": 0.2934200690923049,
      "grad_norm": 3.093608856201172,
      "learning_rate": 9.647097687500815e-06,
      "loss": 1.1015,
      "step": 3631
    },
    {
      "epoch": 0.2935008788056324,
      "grad_norm": 2.7451560497283936,
      "learning_rate": 9.646856171827374e-06,
      "loss": 0.9776,
      "step": 3632
    },
    {
      "epoch": 0.29358168851896,
      "grad_norm": 3.1740598678588867,
      "learning_rate": 9.646614576564475e-06,
      "loss": 0.8478,
      "step": 3633
    },
    {
      "epoch": 0.2936624982322875,
      "grad_norm": 2.531581401824951,
      "learning_rate": 9.646372901716252e-06,
      "loss": 1.043,
      "step": 3634
    },
    {
      "epoch": 0.29374330794561504,
      "grad_norm": 2.956531047821045,
      "learning_rate": 9.646131147286848e-06,
      "loss": 0.9862,
      "step": 3635
    },
    {
      "epoch": 0.2938241176589426,
      "grad_norm": 2.895075798034668,
      "learning_rate": 9.645889313280403e-06,
      "loss": 0.892,
      "step": 3636
    },
    {
      "epoch": 0.29390492737227014,
      "grad_norm": 3.3034794330596924,
      "learning_rate": 9.645647399701058e-06,
      "loss": 0.8616,
      "step": 3637
    },
    {
      "epoch": 0.29398573708559766,
      "grad_norm": 2.8100292682647705,
      "learning_rate": 9.645405406552956e-06,
      "loss": 0.9875,
      "step": 3638
    },
    {
      "epoch": 0.29406654679892524,
      "grad_norm": 2.9189043045043945,
      "learning_rate": 9.645163333840244e-06,
      "loss": 0.9983,
      "step": 3639
    },
    {
      "epoch": 0.29414735651225277,
      "grad_norm": 2.8886024951934814,
      "learning_rate": 9.644921181567068e-06,
      "loss": 0.9779,
      "step": 3640
    },
    {
      "epoch": 0.2942281662255803,
      "grad_norm": 3.025773048400879,
      "learning_rate": 9.644678949737573e-06,
      "loss": 1.0725,
      "step": 3641
    },
    {
      "epoch": 0.29430897593890787,
      "grad_norm": 2.791072130203247,
      "learning_rate": 9.64443663835591e-06,
      "loss": 1.0175,
      "step": 3642
    },
    {
      "epoch": 0.2943897856522354,
      "grad_norm": 2.844024896621704,
      "learning_rate": 9.644194247426227e-06,
      "loss": 0.9882,
      "step": 3643
    },
    {
      "epoch": 0.2944705953655629,
      "grad_norm": 2.7305147647857666,
      "learning_rate": 9.643951776952677e-06,
      "loss": 0.943,
      "step": 3644
    },
    {
      "epoch": 0.2945514050788905,
      "grad_norm": 2.6960883140563965,
      "learning_rate": 9.643709226939414e-06,
      "loss": 0.9377,
      "step": 3645
    },
    {
      "epoch": 0.294632214792218,
      "grad_norm": 2.5656726360321045,
      "learning_rate": 9.643466597390591e-06,
      "loss": 0.9167,
      "step": 3646
    },
    {
      "epoch": 0.29471302450554554,
      "grad_norm": 2.8714895248413086,
      "learning_rate": 9.643223888310363e-06,
      "loss": 0.9791,
      "step": 3647
    },
    {
      "epoch": 0.2947938342188731,
      "grad_norm": 3.0841047763824463,
      "learning_rate": 9.642981099702888e-06,
      "loss": 1.0111,
      "step": 3648
    },
    {
      "epoch": 0.29487464393220064,
      "grad_norm": 3.318530321121216,
      "learning_rate": 9.642738231572327e-06,
      "loss": 0.9167,
      "step": 3649
    },
    {
      "epoch": 0.29495545364552817,
      "grad_norm": 3.0374815464019775,
      "learning_rate": 9.642495283922834e-06,
      "loss": 0.9001,
      "step": 3650
    },
    {
      "epoch": 0.29503626335885574,
      "grad_norm": 2.714785575866699,
      "learning_rate": 9.642252256758573e-06,
      "loss": 0.9983,
      "step": 3651
    },
    {
      "epoch": 0.29511707307218327,
      "grad_norm": 2.6203105449676514,
      "learning_rate": 9.64200915008371e-06,
      "loss": 1.1392,
      "step": 3652
    },
    {
      "epoch": 0.29519788278551085,
      "grad_norm": 3.2234745025634766,
      "learning_rate": 9.6417659639024e-06,
      "loss": 1.1179,
      "step": 3653
    },
    {
      "epoch": 0.29527869249883837,
      "grad_norm": 3.3086729049682617,
      "learning_rate": 9.641522698218817e-06,
      "loss": 0.9776,
      "step": 3654
    },
    {
      "epoch": 0.2953595022121659,
      "grad_norm": 2.2480361461639404,
      "learning_rate": 9.641279353037125e-06,
      "loss": 0.9454,
      "step": 3655
    },
    {
      "epoch": 0.29544031192549347,
      "grad_norm": 2.897298812866211,
      "learning_rate": 9.64103592836149e-06,
      "loss": 0.9071,
      "step": 3656
    },
    {
      "epoch": 0.295521121638821,
      "grad_norm": 3.1581902503967285,
      "learning_rate": 9.640792424196081e-06,
      "loss": 0.9598,
      "step": 3657
    },
    {
      "epoch": 0.2956019313521485,
      "grad_norm": 3.033461332321167,
      "learning_rate": 9.640548840545071e-06,
      "loss": 0.9439,
      "step": 3658
    },
    {
      "epoch": 0.2956827410654761,
      "grad_norm": 2.8116235733032227,
      "learning_rate": 9.640305177412633e-06,
      "loss": 0.9067,
      "step": 3659
    },
    {
      "epoch": 0.2957635507788036,
      "grad_norm": 2.388702869415283,
      "learning_rate": 9.640061434802936e-06,
      "loss": 1.0109,
      "step": 3660
    },
    {
      "epoch": 0.29584436049213114,
      "grad_norm": 2.5147347450256348,
      "learning_rate": 9.63981761272016e-06,
      "loss": 1.0931,
      "step": 3661
    },
    {
      "epoch": 0.2959251702054587,
      "grad_norm": 2.6969945430755615,
      "learning_rate": 9.639573711168476e-06,
      "loss": 1.0101,
      "step": 3662
    },
    {
      "epoch": 0.29600597991878624,
      "grad_norm": 2.7561285495758057,
      "learning_rate": 9.639329730152062e-06,
      "loss": 0.9888,
      "step": 3663
    },
    {
      "epoch": 0.29608678963211377,
      "grad_norm": 2.8578975200653076,
      "learning_rate": 9.639085669675102e-06,
      "loss": 0.9464,
      "step": 3664
    },
    {
      "epoch": 0.29616759934544135,
      "grad_norm": 2.514591932296753,
      "learning_rate": 9.63884152974177e-06,
      "loss": 0.8707,
      "step": 3665
    },
    {
      "epoch": 0.29624840905876887,
      "grad_norm": 2.9842793941497803,
      "learning_rate": 9.638597310356251e-06,
      "loss": 1.0779,
      "step": 3666
    },
    {
      "epoch": 0.2963292187720964,
      "grad_norm": 2.5405080318450928,
      "learning_rate": 9.638353011522727e-06,
      "loss": 1.0123,
      "step": 3667
    },
    {
      "epoch": 0.29641002848542397,
      "grad_norm": 3.033236265182495,
      "learning_rate": 9.638108633245382e-06,
      "loss": 0.9378,
      "step": 3668
    },
    {
      "epoch": 0.2964908381987515,
      "grad_norm": 2.6469950675964355,
      "learning_rate": 9.637864175528403e-06,
      "loss": 1.0793,
      "step": 3669
    },
    {
      "epoch": 0.296571647912079,
      "grad_norm": 3.0484907627105713,
      "learning_rate": 9.637619638375975e-06,
      "loss": 1.0042,
      "step": 3670
    },
    {
      "epoch": 0.2966524576254066,
      "grad_norm": 2.8458609580993652,
      "learning_rate": 9.637375021792288e-06,
      "loss": 0.985,
      "step": 3671
    },
    {
      "epoch": 0.2967332673387341,
      "grad_norm": 2.721964120864868,
      "learning_rate": 9.63713032578153e-06,
      "loss": 1.0156,
      "step": 3672
    },
    {
      "epoch": 0.29681407705206164,
      "grad_norm": 2.86923885345459,
      "learning_rate": 9.636885550347892e-06,
      "loss": 0.931,
      "step": 3673
    },
    {
      "epoch": 0.2968948867653892,
      "grad_norm": 2.955361843109131,
      "learning_rate": 9.63664069549557e-06,
      "loss": 1.0084,
      "step": 3674
    },
    {
      "epoch": 0.29697569647871674,
      "grad_norm": 2.169773817062378,
      "learning_rate": 9.636395761228753e-06,
      "loss": 0.8846,
      "step": 3675
    },
    {
      "epoch": 0.29705650619204427,
      "grad_norm": 3.340684652328491,
      "learning_rate": 9.636150747551637e-06,
      "loss": 0.8681,
      "step": 3676
    },
    {
      "epoch": 0.29713731590537185,
      "grad_norm": 2.6499557495117188,
      "learning_rate": 9.635905654468424e-06,
      "loss": 0.9807,
      "step": 3677
    },
    {
      "epoch": 0.29721812561869937,
      "grad_norm": 2.8025360107421875,
      "learning_rate": 9.635660481983304e-06,
      "loss": 1.0047,
      "step": 3678
    },
    {
      "epoch": 0.2972989353320269,
      "grad_norm": 2.656595468521118,
      "learning_rate": 9.635415230100481e-06,
      "loss": 0.9095,
      "step": 3679
    },
    {
      "epoch": 0.29737974504535447,
      "grad_norm": 3.083083391189575,
      "learning_rate": 9.635169898824156e-06,
      "loss": 1.1252,
      "step": 3680
    },
    {
      "epoch": 0.297460554758682,
      "grad_norm": 2.9544124603271484,
      "learning_rate": 9.634924488158529e-06,
      "loss": 0.949,
      "step": 3681
    },
    {
      "epoch": 0.2975413644720095,
      "grad_norm": 2.7282896041870117,
      "learning_rate": 9.634678998107802e-06,
      "loss": 0.9059,
      "step": 3682
    },
    {
      "epoch": 0.2976221741853371,
      "grad_norm": 2.920116424560547,
      "learning_rate": 9.634433428676182e-06,
      "loss": 0.9557,
      "step": 3683
    },
    {
      "epoch": 0.2977029838986646,
      "grad_norm": 2.758039712905884,
      "learning_rate": 9.634187779867874e-06,
      "loss": 0.9961,
      "step": 3684
    },
    {
      "epoch": 0.29778379361199214,
      "grad_norm": 2.424414873123169,
      "learning_rate": 9.633942051687086e-06,
      "loss": 1.0828,
      "step": 3685
    },
    {
      "epoch": 0.2978646033253197,
      "grad_norm": 2.638777017593384,
      "learning_rate": 9.633696244138026e-06,
      "loss": 0.9923,
      "step": 3686
    },
    {
      "epoch": 0.29794541303864724,
      "grad_norm": 2.3497443199157715,
      "learning_rate": 9.633450357224905e-06,
      "loss": 0.9937,
      "step": 3687
    },
    {
      "epoch": 0.29802622275197477,
      "grad_norm": 2.947047472000122,
      "learning_rate": 9.633204390951933e-06,
      "loss": 1.0171,
      "step": 3688
    },
    {
      "epoch": 0.29810703246530235,
      "grad_norm": 2.8731634616851807,
      "learning_rate": 9.632958345323324e-06,
      "loss": 0.9742,
      "step": 3689
    },
    {
      "epoch": 0.29818784217862987,
      "grad_norm": 2.6260106563568115,
      "learning_rate": 9.632712220343293e-06,
      "loss": 1.0433,
      "step": 3690
    },
    {
      "epoch": 0.2982686518919574,
      "grad_norm": 2.646087169647217,
      "learning_rate": 9.632466016016055e-06,
      "loss": 1.0002,
      "step": 3691
    },
    {
      "epoch": 0.29834946160528497,
      "grad_norm": 2.8434994220733643,
      "learning_rate": 9.632219732345824e-06,
      "loss": 1.0329,
      "step": 3692
    },
    {
      "epoch": 0.2984302713186125,
      "grad_norm": 2.7054171562194824,
      "learning_rate": 9.631973369336822e-06,
      "loss": 0.985,
      "step": 3693
    },
    {
      "epoch": 0.29851108103194,
      "grad_norm": 2.8496406078338623,
      "learning_rate": 9.631726926993268e-06,
      "loss": 0.9409,
      "step": 3694
    },
    {
      "epoch": 0.2985918907452676,
      "grad_norm": 2.6724507808685303,
      "learning_rate": 9.631480405319381e-06,
      "loss": 1.023,
      "step": 3695
    },
    {
      "epoch": 0.2986727004585951,
      "grad_norm": 2.8490123748779297,
      "learning_rate": 9.631233804319384e-06,
      "loss": 1.1671,
      "step": 3696
    },
    {
      "epoch": 0.29875351017192264,
      "grad_norm": 2.7283239364624023,
      "learning_rate": 9.630987123997503e-06,
      "loss": 1.0061,
      "step": 3697
    },
    {
      "epoch": 0.2988343198852502,
      "grad_norm": 2.5398035049438477,
      "learning_rate": 9.63074036435796e-06,
      "loss": 0.9462,
      "step": 3698
    },
    {
      "epoch": 0.29891512959857774,
      "grad_norm": 2.722632884979248,
      "learning_rate": 9.630493525404982e-06,
      "loss": 1.0019,
      "step": 3699
    },
    {
      "epoch": 0.29899593931190527,
      "grad_norm": 2.939347743988037,
      "learning_rate": 9.630246607142799e-06,
      "loss": 1.0087,
      "step": 3700
    },
    {
      "epoch": 0.29907674902523285,
      "grad_norm": 2.5201287269592285,
      "learning_rate": 9.629999609575638e-06,
      "loss": 1.0367,
      "step": 3701
    },
    {
      "epoch": 0.29915755873856037,
      "grad_norm": 2.288954019546509,
      "learning_rate": 9.629752532707729e-06,
      "loss": 0.9458,
      "step": 3702
    },
    {
      "epoch": 0.2992383684518879,
      "grad_norm": 2.856987476348877,
      "learning_rate": 9.629505376543306e-06,
      "loss": 0.9537,
      "step": 3703
    },
    {
      "epoch": 0.29931917816521547,
      "grad_norm": 3.162282705307007,
      "learning_rate": 9.6292581410866e-06,
      "loss": 0.9306,
      "step": 3704
    },
    {
      "epoch": 0.299399987878543,
      "grad_norm": 2.8928518295288086,
      "learning_rate": 9.629010826341846e-06,
      "loss": 1.0406,
      "step": 3705
    },
    {
      "epoch": 0.2994807975918705,
      "grad_norm": 2.788278818130493,
      "learning_rate": 9.628763432313282e-06,
      "loss": 0.9235,
      "step": 3706
    },
    {
      "epoch": 0.2995616073051981,
      "grad_norm": 2.2690505981445312,
      "learning_rate": 9.628515959005142e-06,
      "loss": 0.9959,
      "step": 3707
    },
    {
      "epoch": 0.2996424170185256,
      "grad_norm": 2.923549175262451,
      "learning_rate": 9.628268406421668e-06,
      "loss": 0.9595,
      "step": 3708
    },
    {
      "epoch": 0.29972322673185314,
      "grad_norm": 2.489567279815674,
      "learning_rate": 9.628020774567098e-06,
      "loss": 1.0296,
      "step": 3709
    },
    {
      "epoch": 0.2998040364451807,
      "grad_norm": 3.2813069820404053,
      "learning_rate": 9.627773063445674e-06,
      "loss": 0.9688,
      "step": 3710
    },
    {
      "epoch": 0.29988484615850824,
      "grad_norm": 2.8532261848449707,
      "learning_rate": 9.627525273061637e-06,
      "loss": 1.0838,
      "step": 3711
    },
    {
      "epoch": 0.29996565587183577,
      "grad_norm": 2.5741493701934814,
      "learning_rate": 9.627277403419233e-06,
      "loss": 1.0229,
      "step": 3712
    },
    {
      "epoch": 0.30004646558516335,
      "grad_norm": 3.150580406188965,
      "learning_rate": 9.627029454522706e-06,
      "loss": 1.0542,
      "step": 3713
    },
    {
      "epoch": 0.30012727529849087,
      "grad_norm": 2.7101001739501953,
      "learning_rate": 9.626781426376305e-06,
      "loss": 0.8599,
      "step": 3714
    },
    {
      "epoch": 0.3002080850118184,
      "grad_norm": 3.4060416221618652,
      "learning_rate": 9.626533318984275e-06,
      "loss": 0.8764,
      "step": 3715
    },
    {
      "epoch": 0.30028889472514597,
      "grad_norm": 3.1740713119506836,
      "learning_rate": 9.62628513235087e-06,
      "loss": 1.0542,
      "step": 3716
    },
    {
      "epoch": 0.3003697044384735,
      "grad_norm": 2.790522336959839,
      "learning_rate": 9.626036866480335e-06,
      "loss": 1.0059,
      "step": 3717
    },
    {
      "epoch": 0.3004505141518011,
      "grad_norm": 2.853146553039551,
      "learning_rate": 9.625788521376927e-06,
      "loss": 0.9635,
      "step": 3718
    },
    {
      "epoch": 0.3005313238651286,
      "grad_norm": 2.8684794902801514,
      "learning_rate": 9.625540097044896e-06,
      "loss": 1.0199,
      "step": 3719
    },
    {
      "epoch": 0.3006121335784561,
      "grad_norm": 2.624752998352051,
      "learning_rate": 9.625291593488501e-06,
      "loss": 0.9774,
      "step": 3720
    },
    {
      "epoch": 0.3006929432917837,
      "grad_norm": 2.7984793186187744,
      "learning_rate": 9.625043010711995e-06,
      "loss": 0.9907,
      "step": 3721
    },
    {
      "epoch": 0.3007737530051112,
      "grad_norm": 2.6331899166107178,
      "learning_rate": 9.624794348719636e-06,
      "loss": 1.0337,
      "step": 3722
    },
    {
      "epoch": 0.30085456271843874,
      "grad_norm": 2.8247597217559814,
      "learning_rate": 9.624545607515685e-06,
      "loss": 0.8952,
      "step": 3723
    },
    {
      "epoch": 0.3009353724317663,
      "grad_norm": 2.4679906368255615,
      "learning_rate": 9.624296787104398e-06,
      "loss": 1.1373,
      "step": 3724
    },
    {
      "epoch": 0.30101618214509385,
      "grad_norm": 2.802913188934326,
      "learning_rate": 9.624047887490043e-06,
      "loss": 1.0895,
      "step": 3725
    },
    {
      "epoch": 0.30109699185842137,
      "grad_norm": 3.0468852519989014,
      "learning_rate": 9.623798908676877e-06,
      "loss": 1.0068,
      "step": 3726
    },
    {
      "epoch": 0.30117780157174895,
      "grad_norm": 2.6949493885040283,
      "learning_rate": 9.623549850669168e-06,
      "loss": 0.9805,
      "step": 3727
    },
    {
      "epoch": 0.30125861128507647,
      "grad_norm": 2.809337615966797,
      "learning_rate": 9.623300713471181e-06,
      "loss": 1.0349,
      "step": 3728
    },
    {
      "epoch": 0.301339420998404,
      "grad_norm": 2.325697898864746,
      "learning_rate": 9.623051497087183e-06,
      "loss": 1.058,
      "step": 3729
    },
    {
      "epoch": 0.3014202307117316,
      "grad_norm": 2.797555685043335,
      "learning_rate": 9.622802201521441e-06,
      "loss": 0.8398,
      "step": 3730
    },
    {
      "epoch": 0.3015010404250591,
      "grad_norm": 3.0890145301818848,
      "learning_rate": 9.622552826778228e-06,
      "loss": 1.0047,
      "step": 3731
    },
    {
      "epoch": 0.3015818501383866,
      "grad_norm": 2.638610363006592,
      "learning_rate": 9.622303372861812e-06,
      "loss": 0.9964,
      "step": 3732
    },
    {
      "epoch": 0.3016626598517142,
      "grad_norm": 2.8396639823913574,
      "learning_rate": 9.622053839776469e-06,
      "loss": 0.972,
      "step": 3733
    },
    {
      "epoch": 0.3017434695650417,
      "grad_norm": 3.0043957233428955,
      "learning_rate": 9.62180422752647e-06,
      "loss": 0.9412,
      "step": 3734
    },
    {
      "epoch": 0.30182427927836925,
      "grad_norm": 3.0119330883026123,
      "learning_rate": 9.62155453611609e-06,
      "loss": 0.9555,
      "step": 3735
    },
    {
      "epoch": 0.3019050889916968,
      "grad_norm": 2.6366915702819824,
      "learning_rate": 9.621304765549607e-06,
      "loss": 0.9421,
      "step": 3736
    },
    {
      "epoch": 0.30198589870502435,
      "grad_norm": 2.7750697135925293,
      "learning_rate": 9.621054915831299e-06,
      "loss": 0.8742,
      "step": 3737
    },
    {
      "epoch": 0.30206670841835187,
      "grad_norm": 2.836611032485962,
      "learning_rate": 9.620804986965447e-06,
      "loss": 0.9479,
      "step": 3738
    },
    {
      "epoch": 0.30214751813167945,
      "grad_norm": 2.514456033706665,
      "learning_rate": 9.620554978956326e-06,
      "loss": 0.9826,
      "step": 3739
    },
    {
      "epoch": 0.30222832784500697,
      "grad_norm": 2.808971643447876,
      "learning_rate": 9.620304891808225e-06,
      "loss": 0.9684,
      "step": 3740
    },
    {
      "epoch": 0.3023091375583345,
      "grad_norm": 2.664354085922241,
      "learning_rate": 9.620054725525423e-06,
      "loss": 0.9621,
      "step": 3741
    },
    {
      "epoch": 0.3023899472716621,
      "grad_norm": 3.2560720443725586,
      "learning_rate": 9.619804480112205e-06,
      "loss": 0.9162,
      "step": 3742
    },
    {
      "epoch": 0.3024707569849896,
      "grad_norm": 2.7616066932678223,
      "learning_rate": 9.619554155572859e-06,
      "loss": 1.0174,
      "step": 3743
    },
    {
      "epoch": 0.3025515666983171,
      "grad_norm": 2.6008918285369873,
      "learning_rate": 9.61930375191167e-06,
      "loss": 0.9481,
      "step": 3744
    },
    {
      "epoch": 0.3026323764116447,
      "grad_norm": 2.4243476390838623,
      "learning_rate": 9.61905326913293e-06,
      "loss": 1.0277,
      "step": 3745
    },
    {
      "epoch": 0.3027131861249722,
      "grad_norm": 3.0201876163482666,
      "learning_rate": 9.618802707240926e-06,
      "loss": 1.0801,
      "step": 3746
    },
    {
      "epoch": 0.30279399583829975,
      "grad_norm": 2.391294240951538,
      "learning_rate": 9.618552066239952e-06,
      "loss": 0.9288,
      "step": 3747
    },
    {
      "epoch": 0.3028748055516273,
      "grad_norm": 2.787870407104492,
      "learning_rate": 9.6183013461343e-06,
      "loss": 1.0832,
      "step": 3748
    },
    {
      "epoch": 0.30295561526495485,
      "grad_norm": 2.917569637298584,
      "learning_rate": 9.618050546928265e-06,
      "loss": 1.1066,
      "step": 3749
    },
    {
      "epoch": 0.30303642497828237,
      "grad_norm": 2.6563382148742676,
      "learning_rate": 9.617799668626138e-06,
      "loss": 1.0291,
      "step": 3750
    },
    {
      "epoch": 0.30311723469160995,
      "grad_norm": 2.652367115020752,
      "learning_rate": 9.617548711232223e-06,
      "loss": 1.1087,
      "step": 3751
    },
    {
      "epoch": 0.3031980444049375,
      "grad_norm": 2.404257297515869,
      "learning_rate": 9.617297674750813e-06,
      "loss": 1.0194,
      "step": 3752
    },
    {
      "epoch": 0.303278854118265,
      "grad_norm": 2.656034469604492,
      "learning_rate": 9.617046559186209e-06,
      "loss": 1.1243,
      "step": 3753
    },
    {
      "epoch": 0.3033596638315926,
      "grad_norm": 2.990955114364624,
      "learning_rate": 9.616795364542715e-06,
      "loss": 0.9551,
      "step": 3754
    },
    {
      "epoch": 0.3034404735449201,
      "grad_norm": 2.93247127532959,
      "learning_rate": 9.61654409082463e-06,
      "loss": 1.0054,
      "step": 3755
    },
    {
      "epoch": 0.3035212832582476,
      "grad_norm": 2.9435315132141113,
      "learning_rate": 9.616292738036258e-06,
      "loss": 0.9341,
      "step": 3756
    },
    {
      "epoch": 0.3036020929715752,
      "grad_norm": 3.014838695526123,
      "learning_rate": 9.616041306181905e-06,
      "loss": 0.8704,
      "step": 3757
    },
    {
      "epoch": 0.3036829026849027,
      "grad_norm": 2.5602517127990723,
      "learning_rate": 9.615789795265877e-06,
      "loss": 0.981,
      "step": 3758
    },
    {
      "epoch": 0.30376371239823025,
      "grad_norm": 3.1778597831726074,
      "learning_rate": 9.61553820529248e-06,
      "loss": 1.0666,
      "step": 3759
    },
    {
      "epoch": 0.3038445221115578,
      "grad_norm": 2.844055414199829,
      "learning_rate": 9.615286536266028e-06,
      "loss": 1.0498,
      "step": 3760
    },
    {
      "epoch": 0.30392533182488535,
      "grad_norm": 2.639004707336426,
      "learning_rate": 9.615034788190827e-06,
      "loss": 1.0543,
      "step": 3761
    },
    {
      "epoch": 0.30400614153821287,
      "grad_norm": 2.979391098022461,
      "learning_rate": 9.61478296107119e-06,
      "loss": 1.0619,
      "step": 3762
    },
    {
      "epoch": 0.30408695125154045,
      "grad_norm": 2.64461088180542,
      "learning_rate": 9.61453105491143e-06,
      "loss": 0.889,
      "step": 3763
    },
    {
      "epoch": 0.304167760964868,
      "grad_norm": 2.8508715629577637,
      "learning_rate": 9.614279069715865e-06,
      "loss": 0.9243,
      "step": 3764
    },
    {
      "epoch": 0.3042485706781955,
      "grad_norm": 2.7899696826934814,
      "learning_rate": 9.614027005488806e-06,
      "loss": 1.0311,
      "step": 3765
    },
    {
      "epoch": 0.3043293803915231,
      "grad_norm": 2.9851186275482178,
      "learning_rate": 9.613774862234573e-06,
      "loss": 1.039,
      "step": 3766
    },
    {
      "epoch": 0.3044101901048506,
      "grad_norm": 2.874389171600342,
      "learning_rate": 9.613522639957482e-06,
      "loss": 1.0424,
      "step": 3767
    },
    {
      "epoch": 0.3044909998181781,
      "grad_norm": 2.706479072570801,
      "learning_rate": 9.613270338661856e-06,
      "loss": 1.0294,
      "step": 3768
    },
    {
      "epoch": 0.3045718095315057,
      "grad_norm": 2.365277051925659,
      "learning_rate": 9.613017958352015e-06,
      "loss": 0.9595,
      "step": 3769
    },
    {
      "epoch": 0.3046526192448332,
      "grad_norm": 3.055814266204834,
      "learning_rate": 9.612765499032281e-06,
      "loss": 1.0066,
      "step": 3770
    },
    {
      "epoch": 0.30473342895816075,
      "grad_norm": 2.9857189655303955,
      "learning_rate": 9.61251296070698e-06,
      "loss": 0.9575,
      "step": 3771
    },
    {
      "epoch": 0.3048142386714883,
      "grad_norm": 2.737423896789551,
      "learning_rate": 9.612260343380438e-06,
      "loss": 0.9298,
      "step": 3772
    },
    {
      "epoch": 0.30489504838481585,
      "grad_norm": 2.642305850982666,
      "learning_rate": 9.612007647056976e-06,
      "loss": 0.9823,
      "step": 3773
    },
    {
      "epoch": 0.30497585809814337,
      "grad_norm": 2.365912914276123,
      "learning_rate": 9.611754871740928e-06,
      "loss": 1.0307,
      "step": 3774
    },
    {
      "epoch": 0.30505666781147095,
      "grad_norm": 2.7018494606018066,
      "learning_rate": 9.61150201743662e-06,
      "loss": 0.901,
      "step": 3775
    },
    {
      "epoch": 0.3051374775247985,
      "grad_norm": 2.776804208755493,
      "learning_rate": 9.611249084148386e-06,
      "loss": 1.0134,
      "step": 3776
    },
    {
      "epoch": 0.305218287238126,
      "grad_norm": 2.5004022121429443,
      "learning_rate": 9.610996071880557e-06,
      "loss": 0.9594,
      "step": 3777
    },
    {
      "epoch": 0.3052990969514536,
      "grad_norm": 3.125434637069702,
      "learning_rate": 9.610742980637462e-06,
      "loss": 1.0563,
      "step": 3778
    },
    {
      "epoch": 0.3053799066647811,
      "grad_norm": 2.4795141220092773,
      "learning_rate": 9.610489810423442e-06,
      "loss": 0.984,
      "step": 3779
    },
    {
      "epoch": 0.3054607163781086,
      "grad_norm": 3.0191361904144287,
      "learning_rate": 9.610236561242832e-06,
      "loss": 0.9131,
      "step": 3780
    },
    {
      "epoch": 0.3055415260914362,
      "grad_norm": 2.7647979259490967,
      "learning_rate": 9.609983233099967e-06,
      "loss": 0.8676,
      "step": 3781
    },
    {
      "epoch": 0.3056223358047637,
      "grad_norm": 2.794858455657959,
      "learning_rate": 9.609729825999188e-06,
      "loss": 1.0217,
      "step": 3782
    },
    {
      "epoch": 0.3057031455180913,
      "grad_norm": 2.7970950603485107,
      "learning_rate": 9.609476339944833e-06,
      "loss": 1.1258,
      "step": 3783
    },
    {
      "epoch": 0.3057839552314188,
      "grad_norm": 3.3518712520599365,
      "learning_rate": 9.609222774941248e-06,
      "loss": 0.9881,
      "step": 3784
    },
    {
      "epoch": 0.30586476494474635,
      "grad_norm": 2.4890379905700684,
      "learning_rate": 9.608969130992769e-06,
      "loss": 1.1094,
      "step": 3785
    },
    {
      "epoch": 0.3059455746580739,
      "grad_norm": 3.038357973098755,
      "learning_rate": 9.608715408103748e-06,
      "loss": 1.0204,
      "step": 3786
    },
    {
      "epoch": 0.30602638437140145,
      "grad_norm": 3.140078067779541,
      "learning_rate": 9.608461606278526e-06,
      "loss": 0.8341,
      "step": 3787
    },
    {
      "epoch": 0.306107194084729,
      "grad_norm": 2.7476046085357666,
      "learning_rate": 9.60820772552145e-06,
      "loss": 0.9423,
      "step": 3788
    },
    {
      "epoch": 0.30618800379805655,
      "grad_norm": 2.845513105392456,
      "learning_rate": 9.60795376583687e-06,
      "loss": 1.1315,
      "step": 3789
    },
    {
      "epoch": 0.3062688135113841,
      "grad_norm": 2.850813388824463,
      "learning_rate": 9.607699727229136e-06,
      "loss": 1.0409,
      "step": 3790
    },
    {
      "epoch": 0.3063496232247116,
      "grad_norm": 2.919067144393921,
      "learning_rate": 9.607445609702598e-06,
      "loss": 0.881,
      "step": 3791
    },
    {
      "epoch": 0.3064304329380392,
      "grad_norm": 2.876399517059326,
      "learning_rate": 9.607191413261609e-06,
      "loss": 0.9883,
      "step": 3792
    },
    {
      "epoch": 0.3065112426513667,
      "grad_norm": 2.606030225753784,
      "learning_rate": 9.606937137910522e-06,
      "loss": 0.9093,
      "step": 3793
    },
    {
      "epoch": 0.3065920523646942,
      "grad_norm": 2.95466947555542,
      "learning_rate": 9.606682783653692e-06,
      "loss": 1.09,
      "step": 3794
    },
    {
      "epoch": 0.3066728620780218,
      "grad_norm": 2.9630496501922607,
      "learning_rate": 9.606428350495476e-06,
      "loss": 1.0455,
      "step": 3795
    },
    {
      "epoch": 0.3067536717913493,
      "grad_norm": 2.888209104537964,
      "learning_rate": 9.606173838440234e-06,
      "loss": 0.9933,
      "step": 3796
    },
    {
      "epoch": 0.30683448150467685,
      "grad_norm": 2.905669927597046,
      "learning_rate": 9.605919247492322e-06,
      "loss": 1.0449,
      "step": 3797
    },
    {
      "epoch": 0.3069152912180044,
      "grad_norm": 2.7727410793304443,
      "learning_rate": 9.605664577656099e-06,
      "loss": 0.9082,
      "step": 3798
    },
    {
      "epoch": 0.30699610093133195,
      "grad_norm": 2.158062696456909,
      "learning_rate": 9.605409828935932e-06,
      "loss": 1.0205,
      "step": 3799
    },
    {
      "epoch": 0.3070769106446595,
      "grad_norm": 2.29494047164917,
      "learning_rate": 9.605155001336182e-06,
      "loss": 1.0388,
      "step": 3800
    },
    {
      "epoch": 0.30715772035798705,
      "grad_norm": 2.8616902828216553,
      "learning_rate": 9.604900094861212e-06,
      "loss": 0.9782,
      "step": 3801
    },
    {
      "epoch": 0.3072385300713146,
      "grad_norm": 2.527134895324707,
      "learning_rate": 9.60464510951539e-06,
      "loss": 1.0172,
      "step": 3802
    },
    {
      "epoch": 0.3073193397846421,
      "grad_norm": 2.363241195678711,
      "learning_rate": 9.604390045303083e-06,
      "loss": 1.1337,
      "step": 3803
    },
    {
      "epoch": 0.3074001494979697,
      "grad_norm": 2.7523727416992188,
      "learning_rate": 9.604134902228658e-06,
      "loss": 0.8518,
      "step": 3804
    },
    {
      "epoch": 0.3074809592112972,
      "grad_norm": 2.749645709991455,
      "learning_rate": 9.603879680296486e-06,
      "loss": 1.0404,
      "step": 3805
    },
    {
      "epoch": 0.3075617689246247,
      "grad_norm": 3.2840802669525146,
      "learning_rate": 9.603624379510938e-06,
      "loss": 1.1761,
      "step": 3806
    },
    {
      "epoch": 0.3076425786379523,
      "grad_norm": 2.6806998252868652,
      "learning_rate": 9.60336899987639e-06,
      "loss": 1.004,
      "step": 3807
    },
    {
      "epoch": 0.3077233883512798,
      "grad_norm": 2.5842466354370117,
      "learning_rate": 9.60311354139721e-06,
      "loss": 0.8935,
      "step": 3808
    },
    {
      "epoch": 0.30780419806460735,
      "grad_norm": 2.7936301231384277,
      "learning_rate": 9.602858004077778e-06,
      "loss": 0.9205,
      "step": 3809
    },
    {
      "epoch": 0.3078850077779349,
      "grad_norm": 2.390162944793701,
      "learning_rate": 9.602602387922471e-06,
      "loss": 1.1122,
      "step": 3810
    },
    {
      "epoch": 0.30796581749126245,
      "grad_norm": 2.725221633911133,
      "learning_rate": 9.602346692935662e-06,
      "loss": 1.0698,
      "step": 3811
    },
    {
      "epoch": 0.30804662720459,
      "grad_norm": 3.072129964828491,
      "learning_rate": 9.602090919121736e-06,
      "loss": 0.8672,
      "step": 3812
    },
    {
      "epoch": 0.30812743691791755,
      "grad_norm": 3.481904983520508,
      "learning_rate": 9.60183506648507e-06,
      "loss": 0.9182,
      "step": 3813
    },
    {
      "epoch": 0.3082082466312451,
      "grad_norm": 2.9256131649017334,
      "learning_rate": 9.601579135030051e-06,
      "loss": 0.9397,
      "step": 3814
    },
    {
      "epoch": 0.3082890563445726,
      "grad_norm": 2.864955425262451,
      "learning_rate": 9.601323124761057e-06,
      "loss": 0.9037,
      "step": 3815
    },
    {
      "epoch": 0.3083698660579002,
      "grad_norm": 2.587109088897705,
      "learning_rate": 9.601067035682474e-06,
      "loss": 0.9215,
      "step": 3816
    },
    {
      "epoch": 0.3084506757712277,
      "grad_norm": 3.277244806289673,
      "learning_rate": 9.60081086779869e-06,
      "loss": 0.9555,
      "step": 3817
    },
    {
      "epoch": 0.3085314854845552,
      "grad_norm": 2.660778760910034,
      "learning_rate": 9.600554621114093e-06,
      "loss": 0.9269,
      "step": 3818
    },
    {
      "epoch": 0.3086122951978828,
      "grad_norm": 2.7236886024475098,
      "learning_rate": 9.60029829563307e-06,
      "loss": 0.9754,
      "step": 3819
    },
    {
      "epoch": 0.3086931049112103,
      "grad_norm": 2.7859270572662354,
      "learning_rate": 9.600041891360013e-06,
      "loss": 0.9567,
      "step": 3820
    },
    {
      "epoch": 0.30877391462453785,
      "grad_norm": 2.2497758865356445,
      "learning_rate": 9.599785408299311e-06,
      "loss": 0.8751,
      "step": 3821
    },
    {
      "epoch": 0.3088547243378654,
      "grad_norm": 2.236384630203247,
      "learning_rate": 9.599528846455359e-06,
      "loss": 1.0906,
      "step": 3822
    },
    {
      "epoch": 0.30893553405119295,
      "grad_norm": 2.7493271827697754,
      "learning_rate": 9.599272205832553e-06,
      "loss": 1.0709,
      "step": 3823
    },
    {
      "epoch": 0.3090163437645205,
      "grad_norm": 2.8007590770721436,
      "learning_rate": 9.599015486435284e-06,
      "loss": 1.0544,
      "step": 3824
    },
    {
      "epoch": 0.30909715347784805,
      "grad_norm": 3.1281704902648926,
      "learning_rate": 9.59875868826795e-06,
      "loss": 0.9405,
      "step": 3825
    },
    {
      "epoch": 0.3091779631911756,
      "grad_norm": 2.5548365116119385,
      "learning_rate": 9.598501811334955e-06,
      "loss": 0.8974,
      "step": 3826
    },
    {
      "epoch": 0.3092587729045031,
      "grad_norm": 2.692622661590576,
      "learning_rate": 9.59824485564069e-06,
      "loss": 1.0373,
      "step": 3827
    },
    {
      "epoch": 0.3093395826178307,
      "grad_norm": 3.0443758964538574,
      "learning_rate": 9.597987821189563e-06,
      "loss": 0.9423,
      "step": 3828
    },
    {
      "epoch": 0.3094203923311582,
      "grad_norm": 2.748542070388794,
      "learning_rate": 9.597730707985972e-06,
      "loss": 0.9463,
      "step": 3829
    },
    {
      "epoch": 0.3095012020444857,
      "grad_norm": 2.822598695755005,
      "learning_rate": 9.597473516034325e-06,
      "loss": 1.0552,
      "step": 3830
    },
    {
      "epoch": 0.3095820117578133,
      "grad_norm": 2.4813973903656006,
      "learning_rate": 9.597216245339023e-06,
      "loss": 1.0568,
      "step": 3831
    },
    {
      "epoch": 0.3096628214711408,
      "grad_norm": 2.863154172897339,
      "learning_rate": 9.596958895904475e-06,
      "loss": 1.1406,
      "step": 3832
    },
    {
      "epoch": 0.30974363118446835,
      "grad_norm": 3.0755743980407715,
      "learning_rate": 9.596701467735087e-06,
      "loss": 1.0006,
      "step": 3833
    },
    {
      "epoch": 0.3098244408977959,
      "grad_norm": 2.7633132934570312,
      "learning_rate": 9.596443960835269e-06,
      "loss": 0.9465,
      "step": 3834
    },
    {
      "epoch": 0.30990525061112345,
      "grad_norm": 2.625476598739624,
      "learning_rate": 9.59618637520943e-06,
      "loss": 0.8874,
      "step": 3835
    },
    {
      "epoch": 0.309986060324451,
      "grad_norm": 2.920628309249878,
      "learning_rate": 9.595928710861987e-06,
      "loss": 0.9775,
      "step": 3836
    },
    {
      "epoch": 0.31006687003777855,
      "grad_norm": 2.6560709476470947,
      "learning_rate": 9.595670967797347e-06,
      "loss": 0.9767,
      "step": 3837
    },
    {
      "epoch": 0.3101476797511061,
      "grad_norm": 2.8147332668304443,
      "learning_rate": 9.595413146019927e-06,
      "loss": 0.9863,
      "step": 3838
    },
    {
      "epoch": 0.3102284894644336,
      "grad_norm": 2.8549447059631348,
      "learning_rate": 9.595155245534143e-06,
      "loss": 0.8773,
      "step": 3839
    },
    {
      "epoch": 0.3103092991777612,
      "grad_norm": 3.3137104511260986,
      "learning_rate": 9.594897266344411e-06,
      "loss": 1.023,
      "step": 3840
    },
    {
      "epoch": 0.3103901088910887,
      "grad_norm": 2.49528431892395,
      "learning_rate": 9.594639208455154e-06,
      "loss": 0.9012,
      "step": 3841
    },
    {
      "epoch": 0.3104709186044162,
      "grad_norm": 3.0016326904296875,
      "learning_rate": 9.594381071870785e-06,
      "loss": 1.0405,
      "step": 3842
    },
    {
      "epoch": 0.3105517283177438,
      "grad_norm": 2.6793394088745117,
      "learning_rate": 9.594122856595731e-06,
      "loss": 0.9916,
      "step": 3843
    },
    {
      "epoch": 0.3106325380310713,
      "grad_norm": 2.6459403038024902,
      "learning_rate": 9.593864562634411e-06,
      "loss": 0.993,
      "step": 3844
    },
    {
      "epoch": 0.3107133477443989,
      "grad_norm": 3.2174904346466064,
      "learning_rate": 9.59360618999125e-06,
      "loss": 0.9395,
      "step": 3845
    },
    {
      "epoch": 0.3107941574577264,
      "grad_norm": 2.563730001449585,
      "learning_rate": 9.593347738670676e-06,
      "loss": 0.9565,
      "step": 3846
    },
    {
      "epoch": 0.31087496717105395,
      "grad_norm": 2.956404209136963,
      "learning_rate": 9.593089208677112e-06,
      "loss": 0.9685,
      "step": 3847
    },
    {
      "epoch": 0.31095577688438153,
      "grad_norm": 2.6325273513793945,
      "learning_rate": 9.592830600014985e-06,
      "loss": 0.8934,
      "step": 3848
    },
    {
      "epoch": 0.31103658659770905,
      "grad_norm": 2.8246214389801025,
      "learning_rate": 9.592571912688728e-06,
      "loss": 0.9795,
      "step": 3849
    },
    {
      "epoch": 0.3111173963110366,
      "grad_norm": 2.808764934539795,
      "learning_rate": 9.592313146702773e-06,
      "loss": 1.0248,
      "step": 3850
    },
    {
      "epoch": 0.31119820602436415,
      "grad_norm": 2.5618157386779785,
      "learning_rate": 9.592054302061546e-06,
      "loss": 1.0312,
      "step": 3851
    },
    {
      "epoch": 0.3112790157376917,
      "grad_norm": 2.673449993133545,
      "learning_rate": 9.591795378769485e-06,
      "loss": 0.8822,
      "step": 3852
    },
    {
      "epoch": 0.3113598254510192,
      "grad_norm": 2.4913876056671143,
      "learning_rate": 9.591536376831023e-06,
      "loss": 0.9974,
      "step": 3853
    },
    {
      "epoch": 0.3114406351643468,
      "grad_norm": 2.7856106758117676,
      "learning_rate": 9.591277296250596e-06,
      "loss": 0.9446,
      "step": 3854
    },
    {
      "epoch": 0.3115214448776743,
      "grad_norm": 3.2890031337738037,
      "learning_rate": 9.591018137032642e-06,
      "loss": 0.9611,
      "step": 3855
    },
    {
      "epoch": 0.3116022545910018,
      "grad_norm": 2.4133715629577637,
      "learning_rate": 9.590758899181601e-06,
      "loss": 0.9354,
      "step": 3856
    },
    {
      "epoch": 0.3116830643043294,
      "grad_norm": 3.1188809871673584,
      "learning_rate": 9.590499582701913e-06,
      "loss": 1.1055,
      "step": 3857
    },
    {
      "epoch": 0.31176387401765693,
      "grad_norm": 2.61584210395813,
      "learning_rate": 9.590240187598016e-06,
      "loss": 0.9736,
      "step": 3858
    },
    {
      "epoch": 0.31184468373098445,
      "grad_norm": 2.536604166030884,
      "learning_rate": 9.589980713874357e-06,
      "loss": 1.044,
      "step": 3859
    },
    {
      "epoch": 0.31192549344431203,
      "grad_norm": 2.9626662731170654,
      "learning_rate": 9.589721161535375e-06,
      "loss": 0.8921,
      "step": 3860
    },
    {
      "epoch": 0.31200630315763955,
      "grad_norm": 3.1650912761688232,
      "learning_rate": 9.589461530585523e-06,
      "loss": 1.1349,
      "step": 3861
    },
    {
      "epoch": 0.3120871128709671,
      "grad_norm": 2.8416025638580322,
      "learning_rate": 9.58920182102924e-06,
      "loss": 0.8835,
      "step": 3862
    },
    {
      "epoch": 0.31216792258429465,
      "grad_norm": 3.3654446601867676,
      "learning_rate": 9.58894203287098e-06,
      "loss": 0.9788,
      "step": 3863
    },
    {
      "epoch": 0.3122487322976222,
      "grad_norm": 2.6304450035095215,
      "learning_rate": 9.58868216611519e-06,
      "loss": 0.974,
      "step": 3864
    },
    {
      "epoch": 0.3123295420109497,
      "grad_norm": 2.8896827697753906,
      "learning_rate": 9.58842222076632e-06,
      "loss": 0.9549,
      "step": 3865
    },
    {
      "epoch": 0.3124103517242773,
      "grad_norm": 2.5072031021118164,
      "learning_rate": 9.588162196828826e-06,
      "loss": 0.9045,
      "step": 3866
    },
    {
      "epoch": 0.3124911614376048,
      "grad_norm": 2.752204418182373,
      "learning_rate": 9.587902094307158e-06,
      "loss": 1.0282,
      "step": 3867
    },
    {
      "epoch": 0.3125719711509323,
      "grad_norm": 2.5259454250335693,
      "learning_rate": 9.58764191320577e-06,
      "loss": 1.0006,
      "step": 3868
    },
    {
      "epoch": 0.3126527808642599,
      "grad_norm": 2.7212178707122803,
      "learning_rate": 9.587381653529123e-06,
      "loss": 1.0307,
      "step": 3869
    },
    {
      "epoch": 0.31273359057758743,
      "grad_norm": 3.102100133895874,
      "learning_rate": 9.587121315281671e-06,
      "loss": 0.9911,
      "step": 3870
    },
    {
      "epoch": 0.31281440029091495,
      "grad_norm": 3.042722463607788,
      "learning_rate": 9.586860898467875e-06,
      "loss": 1.0282,
      "step": 3871
    },
    {
      "epoch": 0.31289521000424253,
      "grad_norm": 3.0969510078430176,
      "learning_rate": 9.586600403092192e-06,
      "loss": 1.0182,
      "step": 3872
    },
    {
      "epoch": 0.31297601971757005,
      "grad_norm": 2.870457172393799,
      "learning_rate": 9.586339829159088e-06,
      "loss": 0.9135,
      "step": 3873
    },
    {
      "epoch": 0.3130568294308976,
      "grad_norm": 2.727113962173462,
      "learning_rate": 9.586079176673021e-06,
      "loss": 1.1281,
      "step": 3874
    },
    {
      "epoch": 0.31313763914422515,
      "grad_norm": 2.603419065475464,
      "learning_rate": 9.585818445638462e-06,
      "loss": 0.822,
      "step": 3875
    },
    {
      "epoch": 0.3132184488575527,
      "grad_norm": 2.8258495330810547,
      "learning_rate": 9.585557636059872e-06,
      "loss": 0.9585,
      "step": 3876
    },
    {
      "epoch": 0.3132992585708802,
      "grad_norm": 2.852052688598633,
      "learning_rate": 9.585296747941717e-06,
      "loss": 1.1353,
      "step": 3877
    },
    {
      "epoch": 0.3133800682842078,
      "grad_norm": 2.827730178833008,
      "learning_rate": 9.58503578128847e-06,
      "loss": 1.0242,
      "step": 3878
    },
    {
      "epoch": 0.3134608779975353,
      "grad_norm": 3.0042948722839355,
      "learning_rate": 9.584774736104597e-06,
      "loss": 1.1395,
      "step": 3879
    },
    {
      "epoch": 0.3135416877108628,
      "grad_norm": 2.697857618331909,
      "learning_rate": 9.584513612394568e-06,
      "loss": 1.045,
      "step": 3880
    },
    {
      "epoch": 0.3136224974241904,
      "grad_norm": 2.605607748031616,
      "learning_rate": 9.58425241016286e-06,
      "loss": 1.0711,
      "step": 3881
    },
    {
      "epoch": 0.31370330713751793,
      "grad_norm": 2.865797281265259,
      "learning_rate": 9.583991129413943e-06,
      "loss": 1.1232,
      "step": 3882
    },
    {
      "epoch": 0.31378411685084545,
      "grad_norm": 2.542365550994873,
      "learning_rate": 9.583729770152295e-06,
      "loss": 0.9739,
      "step": 3883
    },
    {
      "epoch": 0.31386492656417303,
      "grad_norm": 2.798555850982666,
      "learning_rate": 9.58346833238239e-06,
      "loss": 0.9233,
      "step": 3884
    },
    {
      "epoch": 0.31394573627750055,
      "grad_norm": 2.568464756011963,
      "learning_rate": 9.583206816108706e-06,
      "loss": 1.0478,
      "step": 3885
    },
    {
      "epoch": 0.3140265459908281,
      "grad_norm": 2.5329174995422363,
      "learning_rate": 9.582945221335724e-06,
      "loss": 1.0034,
      "step": 3886
    },
    {
      "epoch": 0.31410735570415566,
      "grad_norm": 3.2005858421325684,
      "learning_rate": 9.582683548067924e-06,
      "loss": 0.9865,
      "step": 3887
    },
    {
      "epoch": 0.3141881654174832,
      "grad_norm": 2.885324001312256,
      "learning_rate": 9.582421796309786e-06,
      "loss": 1.0066,
      "step": 3888
    },
    {
      "epoch": 0.3142689751308107,
      "grad_norm": 3.1991405487060547,
      "learning_rate": 9.582159966065793e-06,
      "loss": 1.0843,
      "step": 3889
    },
    {
      "epoch": 0.3143497848441383,
      "grad_norm": 2.9450576305389404,
      "learning_rate": 9.581898057340432e-06,
      "loss": 1.0509,
      "step": 3890
    },
    {
      "epoch": 0.3144305945574658,
      "grad_norm": 2.807943105697632,
      "learning_rate": 9.581636070138189e-06,
      "loss": 0.8537,
      "step": 3891
    },
    {
      "epoch": 0.3145114042707933,
      "grad_norm": 2.774735450744629,
      "learning_rate": 9.58137400446355e-06,
      "loss": 1.0605,
      "step": 3892
    },
    {
      "epoch": 0.3145922139841209,
      "grad_norm": 2.735790491104126,
      "learning_rate": 9.581111860321002e-06,
      "loss": 1.0951,
      "step": 3893
    },
    {
      "epoch": 0.31467302369744843,
      "grad_norm": 2.945537567138672,
      "learning_rate": 9.580849637715037e-06,
      "loss": 0.9333,
      "step": 3894
    },
    {
      "epoch": 0.31475383341077595,
      "grad_norm": 2.72914457321167,
      "learning_rate": 9.580587336650147e-06,
      "loss": 1.0557,
      "step": 3895
    },
    {
      "epoch": 0.31483464312410353,
      "grad_norm": 3.2410850524902344,
      "learning_rate": 9.580324957130823e-06,
      "loss": 1.0085,
      "step": 3896
    },
    {
      "epoch": 0.31491545283743105,
      "grad_norm": 3.114778757095337,
      "learning_rate": 9.580062499161557e-06,
      "loss": 1.0151,
      "step": 3897
    },
    {
      "epoch": 0.3149962625507586,
      "grad_norm": 2.4615917205810547,
      "learning_rate": 9.579799962746848e-06,
      "loss": 0.974,
      "step": 3898
    },
    {
      "epoch": 0.31507707226408616,
      "grad_norm": 2.4671497344970703,
      "learning_rate": 9.57953734789119e-06,
      "loss": 0.9907,
      "step": 3899
    },
    {
      "epoch": 0.3151578819774137,
      "grad_norm": 2.4573681354522705,
      "learning_rate": 9.579274654599082e-06,
      "loss": 0.9598,
      "step": 3900
    },
    {
      "epoch": 0.3152386916907412,
      "grad_norm": 2.579907178878784,
      "learning_rate": 9.579011882875024e-06,
      "loss": 0.9427,
      "step": 3901
    },
    {
      "epoch": 0.3153195014040688,
      "grad_norm": 3.1489217281341553,
      "learning_rate": 9.578749032723516e-06,
      "loss": 0.9931,
      "step": 3902
    },
    {
      "epoch": 0.3154003111173963,
      "grad_norm": 3.128610849380493,
      "learning_rate": 9.57848610414906e-06,
      "loss": 0.9354,
      "step": 3903
    },
    {
      "epoch": 0.3154811208307238,
      "grad_norm": 2.8404691219329834,
      "learning_rate": 9.57822309715616e-06,
      "loss": 0.908,
      "step": 3904
    },
    {
      "epoch": 0.3155619305440514,
      "grad_norm": 2.4756977558135986,
      "learning_rate": 9.577960011749319e-06,
      "loss": 1.0025,
      "step": 3905
    },
    {
      "epoch": 0.31564274025737893,
      "grad_norm": 2.6812450885772705,
      "learning_rate": 9.577696847933044e-06,
      "loss": 0.9053,
      "step": 3906
    },
    {
      "epoch": 0.31572354997070645,
      "grad_norm": 2.4922304153442383,
      "learning_rate": 9.577433605711842e-06,
      "loss": 0.999,
      "step": 3907
    },
    {
      "epoch": 0.31580435968403403,
      "grad_norm": 2.8330764770507812,
      "learning_rate": 9.577170285090223e-06,
      "loss": 0.9833,
      "step": 3908
    },
    {
      "epoch": 0.31588516939736155,
      "grad_norm": 2.674743890762329,
      "learning_rate": 9.576906886072695e-06,
      "loss": 0.9638,
      "step": 3909
    },
    {
      "epoch": 0.31596597911068913,
      "grad_norm": 2.507943630218506,
      "learning_rate": 9.576643408663771e-06,
      "loss": 0.9634,
      "step": 3910
    },
    {
      "epoch": 0.31604678882401666,
      "grad_norm": 2.446462869644165,
      "learning_rate": 9.576379852867964e-06,
      "loss": 1.045,
      "step": 3911
    },
    {
      "epoch": 0.3161275985373442,
      "grad_norm": 2.408454179763794,
      "learning_rate": 9.576116218689785e-06,
      "loss": 1.0675,
      "step": 3912
    },
    {
      "epoch": 0.31620840825067176,
      "grad_norm": 2.7893428802490234,
      "learning_rate": 9.575852506133756e-06,
      "loss": 0.9539,
      "step": 3913
    },
    {
      "epoch": 0.3162892179639993,
      "grad_norm": 2.7513561248779297,
      "learning_rate": 9.575588715204387e-06,
      "loss": 0.9783,
      "step": 3914
    },
    {
      "epoch": 0.3163700276773268,
      "grad_norm": 3.0103142261505127,
      "learning_rate": 9.575324845906197e-06,
      "loss": 0.9148,
      "step": 3915
    },
    {
      "epoch": 0.3164508373906544,
      "grad_norm": 2.799314260482788,
      "learning_rate": 9.575060898243709e-06,
      "loss": 1.0869,
      "step": 3916
    },
    {
      "epoch": 0.3165316471039819,
      "grad_norm": 2.6418752670288086,
      "learning_rate": 9.574796872221441e-06,
      "loss": 0.9344,
      "step": 3917
    },
    {
      "epoch": 0.31661245681730943,
      "grad_norm": 2.895109176635742,
      "learning_rate": 9.574532767843917e-06,
      "loss": 0.9376,
      "step": 3918
    },
    {
      "epoch": 0.316693266530637,
      "grad_norm": 2.6166465282440186,
      "learning_rate": 9.574268585115659e-06,
      "loss": 0.9595,
      "step": 3919
    },
    {
      "epoch": 0.31677407624396453,
      "grad_norm": 2.743903875350952,
      "learning_rate": 9.574004324041191e-06,
      "loss": 1.0445,
      "step": 3920
    },
    {
      "epoch": 0.31685488595729205,
      "grad_norm": 2.5423619747161865,
      "learning_rate": 9.573739984625041e-06,
      "loss": 1.0441,
      "step": 3921
    },
    {
      "epoch": 0.31693569567061963,
      "grad_norm": 2.3177995681762695,
      "learning_rate": 9.573475566871737e-06,
      "loss": 1.0372,
      "step": 3922
    },
    {
      "epoch": 0.31701650538394716,
      "grad_norm": 2.443727731704712,
      "learning_rate": 9.573211070785807e-06,
      "loss": 0.9468,
      "step": 3923
    },
    {
      "epoch": 0.3170973150972747,
      "grad_norm": 2.727766990661621,
      "learning_rate": 9.572946496371778e-06,
      "loss": 0.9571,
      "step": 3924
    },
    {
      "epoch": 0.31717812481060226,
      "grad_norm": 2.215959310531616,
      "learning_rate": 9.572681843634187e-06,
      "loss": 0.9412,
      "step": 3925
    },
    {
      "epoch": 0.3172589345239298,
      "grad_norm": 2.926909923553467,
      "learning_rate": 9.572417112577563e-06,
      "loss": 1.1068,
      "step": 3926
    },
    {
      "epoch": 0.3173397442372573,
      "grad_norm": 2.5570342540740967,
      "learning_rate": 9.572152303206443e-06,
      "loss": 0.9862,
      "step": 3927
    },
    {
      "epoch": 0.3174205539505849,
      "grad_norm": 2.497966766357422,
      "learning_rate": 9.57188741552536e-06,
      "loss": 0.9851,
      "step": 3928
    },
    {
      "epoch": 0.3175013636639124,
      "grad_norm": 3.5437493324279785,
      "learning_rate": 9.571622449538852e-06,
      "loss": 1.023,
      "step": 3929
    },
    {
      "epoch": 0.31758217337723993,
      "grad_norm": 3.100724697113037,
      "learning_rate": 9.571357405251459e-06,
      "loss": 1.1043,
      "step": 3930
    },
    {
      "epoch": 0.3176629830905675,
      "grad_norm": 2.3053576946258545,
      "learning_rate": 9.571092282667716e-06,
      "loss": 1.0503,
      "step": 3931
    },
    {
      "epoch": 0.31774379280389503,
      "grad_norm": 2.488769054412842,
      "learning_rate": 9.570827081792167e-06,
      "loss": 1.0722,
      "step": 3932
    },
    {
      "epoch": 0.31782460251722255,
      "grad_norm": 2.8296937942504883,
      "learning_rate": 9.570561802629355e-06,
      "loss": 1.0153,
      "step": 3933
    },
    {
      "epoch": 0.31790541223055013,
      "grad_norm": 2.601266384124756,
      "learning_rate": 9.570296445183822e-06,
      "loss": 1.0516,
      "step": 3934
    },
    {
      "epoch": 0.31798622194387766,
      "grad_norm": 2.7215843200683594,
      "learning_rate": 9.570031009460114e-06,
      "loss": 0.8979,
      "step": 3935
    },
    {
      "epoch": 0.3180670316572052,
      "grad_norm": 2.969320297241211,
      "learning_rate": 9.569765495462777e-06,
      "loss": 0.993,
      "step": 3936
    },
    {
      "epoch": 0.31814784137053276,
      "grad_norm": 2.622695207595825,
      "learning_rate": 9.569499903196357e-06,
      "loss": 0.9333,
      "step": 3937
    },
    {
      "epoch": 0.3182286510838603,
      "grad_norm": 3.369231700897217,
      "learning_rate": 9.569234232665405e-06,
      "loss": 0.9316,
      "step": 3938
    },
    {
      "epoch": 0.3183094607971878,
      "grad_norm": 2.9470860958099365,
      "learning_rate": 9.56896848387447e-06,
      "loss": 1.0316,
      "step": 3939
    },
    {
      "epoch": 0.3183902705105154,
      "grad_norm": 3.0356106758117676,
      "learning_rate": 9.568702656828103e-06,
      "loss": 0.8713,
      "step": 3940
    },
    {
      "epoch": 0.3184710802238429,
      "grad_norm": 2.354508399963379,
      "learning_rate": 9.568436751530862e-06,
      "loss": 1.0344,
      "step": 3941
    },
    {
      "epoch": 0.31855188993717043,
      "grad_norm": 2.9620916843414307,
      "learning_rate": 9.568170767987294e-06,
      "loss": 0.8858,
      "step": 3942
    },
    {
      "epoch": 0.318632699650498,
      "grad_norm": 3.0026702880859375,
      "learning_rate": 9.567904706201961e-06,
      "loss": 0.9729,
      "step": 3943
    },
    {
      "epoch": 0.31871350936382553,
      "grad_norm": 2.5581581592559814,
      "learning_rate": 9.567638566179414e-06,
      "loss": 0.9812,
      "step": 3944
    },
    {
      "epoch": 0.31879431907715305,
      "grad_norm": 2.487456798553467,
      "learning_rate": 9.567372347924217e-06,
      "loss": 0.9962,
      "step": 3945
    },
    {
      "epoch": 0.31887512879048063,
      "grad_norm": 2.9502642154693604,
      "learning_rate": 9.567106051440926e-06,
      "loss": 0.9709,
      "step": 3946
    },
    {
      "epoch": 0.31895593850380816,
      "grad_norm": 3.1015729904174805,
      "learning_rate": 9.566839676734103e-06,
      "loss": 0.9733,
      "step": 3947
    },
    {
      "epoch": 0.3190367482171357,
      "grad_norm": 2.408635377883911,
      "learning_rate": 9.566573223808313e-06,
      "loss": 1.0635,
      "step": 3948
    },
    {
      "epoch": 0.31911755793046326,
      "grad_norm": 3.367497444152832,
      "learning_rate": 9.566306692668115e-06,
      "loss": 0.9283,
      "step": 3949
    },
    {
      "epoch": 0.3191983676437908,
      "grad_norm": 3.2404932975769043,
      "learning_rate": 9.566040083318076e-06,
      "loss": 1.0313,
      "step": 3950
    },
    {
      "epoch": 0.3192791773571183,
      "grad_norm": 2.889040946960449,
      "learning_rate": 9.565773395762763e-06,
      "loss": 0.9841,
      "step": 3951
    },
    {
      "epoch": 0.3193599870704459,
      "grad_norm": 2.901196241378784,
      "learning_rate": 9.565506630006745e-06,
      "loss": 1.0316,
      "step": 3952
    },
    {
      "epoch": 0.3194407967837734,
      "grad_norm": 2.811598062515259,
      "learning_rate": 9.565239786054587e-06,
      "loss": 1.0213,
      "step": 3953
    },
    {
      "epoch": 0.31952160649710093,
      "grad_norm": 3.011934757232666,
      "learning_rate": 9.564972863910862e-06,
      "loss": 1.0583,
      "step": 3954
    },
    {
      "epoch": 0.3196024162104285,
      "grad_norm": 3.0957868099212646,
      "learning_rate": 9.564705863580145e-06,
      "loss": 0.9961,
      "step": 3955
    },
    {
      "epoch": 0.31968322592375603,
      "grad_norm": 2.9000260829925537,
      "learning_rate": 9.564438785067002e-06,
      "loss": 1.0179,
      "step": 3956
    },
    {
      "epoch": 0.31976403563708355,
      "grad_norm": 2.7005488872528076,
      "learning_rate": 9.564171628376013e-06,
      "loss": 0.9293,
      "step": 3957
    },
    {
      "epoch": 0.31984484535041113,
      "grad_norm": 2.8162786960601807,
      "learning_rate": 9.563904393511749e-06,
      "loss": 0.9181,
      "step": 3958
    },
    {
      "epoch": 0.31992565506373866,
      "grad_norm": 2.960033416748047,
      "learning_rate": 9.563637080478793e-06,
      "loss": 0.9736,
      "step": 3959
    },
    {
      "epoch": 0.3200064647770662,
      "grad_norm": 3.206284523010254,
      "learning_rate": 9.563369689281718e-06,
      "loss": 0.8939,
      "step": 3960
    },
    {
      "epoch": 0.32008727449039376,
      "grad_norm": 2.616919755935669,
      "learning_rate": 9.563102219925109e-06,
      "loss": 0.971,
      "step": 3961
    },
    {
      "epoch": 0.3201680842037213,
      "grad_norm": 3.0238308906555176,
      "learning_rate": 9.562834672413542e-06,
      "loss": 0.9845,
      "step": 3962
    },
    {
      "epoch": 0.3202488939170488,
      "grad_norm": 2.5969364643096924,
      "learning_rate": 9.562567046751603e-06,
      "loss": 1.0305,
      "step": 3963
    },
    {
      "epoch": 0.3203297036303764,
      "grad_norm": 2.2757885456085205,
      "learning_rate": 9.562299342943873e-06,
      "loss": 0.999,
      "step": 3964
    },
    {
      "epoch": 0.3204105133437039,
      "grad_norm": 2.4738149642944336,
      "learning_rate": 9.56203156099494e-06,
      "loss": 1.0553,
      "step": 3965
    },
    {
      "epoch": 0.32049132305703143,
      "grad_norm": 2.243032217025757,
      "learning_rate": 9.561763700909387e-06,
      "loss": 1.1218,
      "step": 3966
    },
    {
      "epoch": 0.320572132770359,
      "grad_norm": 2.5168676376342773,
      "learning_rate": 9.561495762691804e-06,
      "loss": 1.1254,
      "step": 3967
    },
    {
      "epoch": 0.32065294248368653,
      "grad_norm": 2.716259717941284,
      "learning_rate": 9.561227746346783e-06,
      "loss": 0.8839,
      "step": 3968
    },
    {
      "epoch": 0.32073375219701405,
      "grad_norm": 2.474407434463501,
      "learning_rate": 9.560959651878908e-06,
      "loss": 1.0295,
      "step": 3969
    },
    {
      "epoch": 0.32081456191034163,
      "grad_norm": 2.8599421977996826,
      "learning_rate": 9.560691479292777e-06,
      "loss": 0.9559,
      "step": 3970
    },
    {
      "epoch": 0.32089537162366916,
      "grad_norm": 2.726964235305786,
      "learning_rate": 9.560423228592978e-06,
      "loss": 0.8889,
      "step": 3971
    },
    {
      "epoch": 0.3209761813369967,
      "grad_norm": 3.0214600563049316,
      "learning_rate": 9.560154899784109e-06,
      "loss": 1.0466,
      "step": 3972
    },
    {
      "epoch": 0.32105699105032426,
      "grad_norm": 3.2783267498016357,
      "learning_rate": 9.559886492870764e-06,
      "loss": 1.0275,
      "step": 3973
    },
    {
      "epoch": 0.3211378007636518,
      "grad_norm": 2.6123223304748535,
      "learning_rate": 9.55961800785754e-06,
      "loss": 0.9725,
      "step": 3974
    },
    {
      "epoch": 0.32121861047697936,
      "grad_norm": 2.3589444160461426,
      "learning_rate": 9.55934944474904e-06,
      "loss": 0.9172,
      "step": 3975
    },
    {
      "epoch": 0.3212994201903069,
      "grad_norm": 2.607569456100464,
      "learning_rate": 9.559080803549857e-06,
      "loss": 0.9393,
      "step": 3976
    },
    {
      "epoch": 0.3213802299036344,
      "grad_norm": 2.7016165256500244,
      "learning_rate": 9.558812084264595e-06,
      "loss": 1.0372,
      "step": 3977
    },
    {
      "epoch": 0.321461039616962,
      "grad_norm": 2.495267629623413,
      "learning_rate": 9.55854328689786e-06,
      "loss": 1.0041,
      "step": 3978
    },
    {
      "epoch": 0.3215418493302895,
      "grad_norm": 2.5262086391448975,
      "learning_rate": 9.558274411454249e-06,
      "loss": 1.0064,
      "step": 3979
    },
    {
      "epoch": 0.32162265904361703,
      "grad_norm": 2.77862286567688,
      "learning_rate": 9.558005457938372e-06,
      "loss": 1.0388,
      "step": 3980
    },
    {
      "epoch": 0.3217034687569446,
      "grad_norm": 2.4823708534240723,
      "learning_rate": 9.557736426354837e-06,
      "loss": 0.9786,
      "step": 3981
    },
    {
      "epoch": 0.32178427847027213,
      "grad_norm": 2.639963388442993,
      "learning_rate": 9.557467316708246e-06,
      "loss": 0.9069,
      "step": 3982
    },
    {
      "epoch": 0.32186508818359966,
      "grad_norm": 3.4461371898651123,
      "learning_rate": 9.557198129003213e-06,
      "loss": 1.0156,
      "step": 3983
    },
    {
      "epoch": 0.32194589789692724,
      "grad_norm": 2.6586742401123047,
      "learning_rate": 9.556928863244348e-06,
      "loss": 1.0775,
      "step": 3984
    },
    {
      "epoch": 0.32202670761025476,
      "grad_norm": 2.1365299224853516,
      "learning_rate": 9.556659519436262e-06,
      "loss": 0.8579,
      "step": 3985
    },
    {
      "epoch": 0.3221075173235823,
      "grad_norm": 2.56663179397583,
      "learning_rate": 9.556390097583567e-06,
      "loss": 1.0838,
      "step": 3986
    },
    {
      "epoch": 0.32218832703690986,
      "grad_norm": 2.783458948135376,
      "learning_rate": 9.556120597690879e-06,
      "loss": 0.9394,
      "step": 3987
    },
    {
      "epoch": 0.3222691367502374,
      "grad_norm": 2.74609112739563,
      "learning_rate": 9.555851019762812e-06,
      "loss": 1.0268,
      "step": 3988
    },
    {
      "epoch": 0.3223499464635649,
      "grad_norm": 2.665937662124634,
      "learning_rate": 9.555581363803987e-06,
      "loss": 1.0699,
      "step": 3989
    },
    {
      "epoch": 0.3224307561768925,
      "grad_norm": 2.3670506477355957,
      "learning_rate": 9.55531162981902e-06,
      "loss": 1.014,
      "step": 3990
    },
    {
      "epoch": 0.32251156589022,
      "grad_norm": 2.883831739425659,
      "learning_rate": 9.555041817812531e-06,
      "loss": 0.9548,
      "step": 3991
    },
    {
      "epoch": 0.32259237560354753,
      "grad_norm": 2.978527307510376,
      "learning_rate": 9.554771927789142e-06,
      "loss": 0.984,
      "step": 3992
    },
    {
      "epoch": 0.3226731853168751,
      "grad_norm": 2.601973295211792,
      "learning_rate": 9.554501959753472e-06,
      "loss": 1.0509,
      "step": 3993
    },
    {
      "epoch": 0.32275399503020263,
      "grad_norm": 3.092226266860962,
      "learning_rate": 9.554231913710153e-06,
      "loss": 0.995,
      "step": 3994
    },
    {
      "epoch": 0.32283480474353016,
      "grad_norm": 2.800023078918457,
      "learning_rate": 9.5539617896638e-06,
      "loss": 1.015,
      "step": 3995
    },
    {
      "epoch": 0.32291561445685774,
      "grad_norm": 2.28165340423584,
      "learning_rate": 9.553691587619048e-06,
      "loss": 0.9769,
      "step": 3996
    },
    {
      "epoch": 0.32299642417018526,
      "grad_norm": 2.8008766174316406,
      "learning_rate": 9.553421307580521e-06,
      "loss": 1.0006,
      "step": 3997
    },
    {
      "epoch": 0.3230772338835128,
      "grad_norm": 3.051344394683838,
      "learning_rate": 9.55315094955285e-06,
      "loss": 0.9896,
      "step": 3998
    },
    {
      "epoch": 0.32315804359684036,
      "grad_norm": 2.8846046924591064,
      "learning_rate": 9.552880513540663e-06,
      "loss": 1.0331,
      "step": 3999
    },
    {
      "epoch": 0.3232388533101679,
      "grad_norm": 2.6979334354400635,
      "learning_rate": 9.552609999548594e-06,
      "loss": 0.8969,
      "step": 4000
    },
    {
      "epoch": 0.3232388533101679,
      "eval_loss": 0.8279703855514526,
      "eval_runtime": 814.558,
      "eval_samples_per_second": 102.345,
      "eval_steps_per_second": 12.793,
      "step": 4000
    },
    {
      "epoch": 0.3233196630234954,
      "grad_norm": 2.881255865097046,
      "learning_rate": 9.552339407581275e-06,
      "loss": 0.9799,
      "step": 4001
    },
    {
      "epoch": 0.323400472736823,
      "grad_norm": 2.7741382122039795,
      "learning_rate": 9.55206873764334e-06,
      "loss": 0.9929,
      "step": 4002
    },
    {
      "epoch": 0.3234812824501505,
      "grad_norm": 2.676832914352417,
      "learning_rate": 9.55179798973943e-06,
      "loss": 0.9728,
      "step": 4003
    },
    {
      "epoch": 0.32356209216347803,
      "grad_norm": 2.584409475326538,
      "learning_rate": 9.551527163874174e-06,
      "loss": 0.9822,
      "step": 4004
    },
    {
      "epoch": 0.3236429018768056,
      "grad_norm": 2.607215166091919,
      "learning_rate": 9.55125626005222e-06,
      "loss": 0.9944,
      "step": 4005
    },
    {
      "epoch": 0.32372371159013313,
      "grad_norm": 2.788327217102051,
      "learning_rate": 9.5509852782782e-06,
      "loss": 0.9474,
      "step": 4006
    },
    {
      "epoch": 0.32380452130346066,
      "grad_norm": 3.0161447525024414,
      "learning_rate": 9.550714218556759e-06,
      "loss": 0.9448,
      "step": 4007
    },
    {
      "epoch": 0.32388533101678824,
      "grad_norm": 2.7205970287323,
      "learning_rate": 9.550443080892538e-06,
      "loss": 1.0268,
      "step": 4008
    },
    {
      "epoch": 0.32396614073011576,
      "grad_norm": 2.9318370819091797,
      "learning_rate": 9.550171865290181e-06,
      "loss": 0.9557,
      "step": 4009
    },
    {
      "epoch": 0.3240469504434433,
      "grad_norm": 2.945824384689331,
      "learning_rate": 9.549900571754336e-06,
      "loss": 0.927,
      "step": 4010
    },
    {
      "epoch": 0.32412776015677086,
      "grad_norm": 2.685911178588867,
      "learning_rate": 9.549629200289646e-06,
      "loss": 0.9796,
      "step": 4011
    },
    {
      "epoch": 0.3242085698700984,
      "grad_norm": 2.4019417762756348,
      "learning_rate": 9.549357750900762e-06,
      "loss": 0.9273,
      "step": 4012
    },
    {
      "epoch": 0.3242893795834259,
      "grad_norm": 3.1122894287109375,
      "learning_rate": 9.54908622359233e-06,
      "loss": 0.9409,
      "step": 4013
    },
    {
      "epoch": 0.3243701892967535,
      "grad_norm": 2.6726021766662598,
      "learning_rate": 9.548814618369004e-06,
      "loss": 1.0427,
      "step": 4014
    },
    {
      "epoch": 0.324450999010081,
      "grad_norm": 2.7439935207366943,
      "learning_rate": 9.548542935235433e-06,
      "loss": 0.8824,
      "step": 4015
    },
    {
      "epoch": 0.32453180872340853,
      "grad_norm": 2.644510269165039,
      "learning_rate": 9.548271174196275e-06,
      "loss": 0.9333,
      "step": 4016
    },
    {
      "epoch": 0.3246126184367361,
      "grad_norm": 3.181877613067627,
      "learning_rate": 9.547999335256179e-06,
      "loss": 0.9233,
      "step": 4017
    },
    {
      "epoch": 0.32469342815006363,
      "grad_norm": 2.7948482036590576,
      "learning_rate": 9.547727418419802e-06,
      "loss": 1.0905,
      "step": 4018
    },
    {
      "epoch": 0.32477423786339116,
      "grad_norm": 2.7870097160339355,
      "learning_rate": 9.547455423691804e-06,
      "loss": 0.9834,
      "step": 4019
    },
    {
      "epoch": 0.32485504757671874,
      "grad_norm": 2.678520679473877,
      "learning_rate": 9.547183351076843e-06,
      "loss": 0.919,
      "step": 4020
    },
    {
      "epoch": 0.32493585729004626,
      "grad_norm": 2.41336727142334,
      "learning_rate": 9.546911200579579e-06,
      "loss": 1.0095,
      "step": 4021
    },
    {
      "epoch": 0.3250166670033738,
      "grad_norm": 2.888129949569702,
      "learning_rate": 9.546638972204669e-06,
      "loss": 0.9891,
      "step": 4022
    },
    {
      "epoch": 0.32509747671670136,
      "grad_norm": 2.523606538772583,
      "learning_rate": 9.54636666595678e-06,
      "loss": 1.0194,
      "step": 4023
    },
    {
      "epoch": 0.3251782864300289,
      "grad_norm": 2.9586949348449707,
      "learning_rate": 9.546094281840576e-06,
      "loss": 1.0903,
      "step": 4024
    },
    {
      "epoch": 0.3252590961433564,
      "grad_norm": 3.0866212844848633,
      "learning_rate": 9.545821819860722e-06,
      "loss": 0.9632,
      "step": 4025
    },
    {
      "epoch": 0.325339905856684,
      "grad_norm": 2.503249168395996,
      "learning_rate": 9.545549280021882e-06,
      "loss": 0.9814,
      "step": 4026
    },
    {
      "epoch": 0.3254207155700115,
      "grad_norm": 3.371840000152588,
      "learning_rate": 9.545276662328727e-06,
      "loss": 0.9856,
      "step": 4027
    },
    {
      "epoch": 0.32550152528333903,
      "grad_norm": 2.9147255420684814,
      "learning_rate": 9.545003966785922e-06,
      "loss": 0.9717,
      "step": 4028
    },
    {
      "epoch": 0.3255823349966666,
      "grad_norm": 2.737259864807129,
      "learning_rate": 9.544731193398144e-06,
      "loss": 1.047,
      "step": 4029
    },
    {
      "epoch": 0.32566314470999413,
      "grad_norm": 2.514043092727661,
      "learning_rate": 9.54445834217006e-06,
      "loss": 0.979,
      "step": 4030
    },
    {
      "epoch": 0.32574395442332166,
      "grad_norm": 2.676771402359009,
      "learning_rate": 9.544185413106345e-06,
      "loss": 0.9492,
      "step": 4031
    },
    {
      "epoch": 0.32582476413664924,
      "grad_norm": 2.657778024673462,
      "learning_rate": 9.543912406211677e-06,
      "loss": 1.0499,
      "step": 4032
    },
    {
      "epoch": 0.32590557384997676,
      "grad_norm": 2.5434489250183105,
      "learning_rate": 9.543639321490725e-06,
      "loss": 0.9831,
      "step": 4033
    },
    {
      "epoch": 0.3259863835633043,
      "grad_norm": 2.455796003341675,
      "learning_rate": 9.543366158948171e-06,
      "loss": 1.0414,
      "step": 4034
    },
    {
      "epoch": 0.32606719327663186,
      "grad_norm": 2.7317655086517334,
      "learning_rate": 9.543092918588691e-06,
      "loss": 0.9999,
      "step": 4035
    },
    {
      "epoch": 0.3261480029899594,
      "grad_norm": 2.511050224304199,
      "learning_rate": 9.542819600416968e-06,
      "loss": 0.9754,
      "step": 4036
    },
    {
      "epoch": 0.3262288127032869,
      "grad_norm": 2.6154658794403076,
      "learning_rate": 9.54254620443768e-06,
      "loss": 0.912,
      "step": 4037
    },
    {
      "epoch": 0.3263096224166145,
      "grad_norm": 2.6074345111846924,
      "learning_rate": 9.542272730655513e-06,
      "loss": 1.0228,
      "step": 4038
    },
    {
      "epoch": 0.326390432129942,
      "grad_norm": 2.535208225250244,
      "learning_rate": 9.541999179075147e-06,
      "loss": 0.9436,
      "step": 4039
    },
    {
      "epoch": 0.3264712418432696,
      "grad_norm": 2.6358659267425537,
      "learning_rate": 9.541725549701273e-06,
      "loss": 0.8877,
      "step": 4040
    },
    {
      "epoch": 0.3265520515565971,
      "grad_norm": 3.4987316131591797,
      "learning_rate": 9.541451842538571e-06,
      "loss": 0.9417,
      "step": 4041
    },
    {
      "epoch": 0.32663286126992463,
      "grad_norm": 2.594028949737549,
      "learning_rate": 9.541178057591733e-06,
      "loss": 1.0146,
      "step": 4042
    },
    {
      "epoch": 0.3267136709832522,
      "grad_norm": 2.839911460876465,
      "learning_rate": 9.540904194865448e-06,
      "loss": 0.9229,
      "step": 4043
    },
    {
      "epoch": 0.32679448069657974,
      "grad_norm": 2.946542263031006,
      "learning_rate": 9.540630254364404e-06,
      "loss": 1.0548,
      "step": 4044
    },
    {
      "epoch": 0.32687529040990726,
      "grad_norm": 2.7420732975006104,
      "learning_rate": 9.540356236093296e-06,
      "loss": 0.9801,
      "step": 4045
    },
    {
      "epoch": 0.32695610012323484,
      "grad_norm": 2.915590286254883,
      "learning_rate": 9.540082140056816e-06,
      "loss": 0.9744,
      "step": 4046
    },
    {
      "epoch": 0.32703690983656236,
      "grad_norm": 2.9443070888519287,
      "learning_rate": 9.53980796625966e-06,
      "loss": 0.9807,
      "step": 4047
    },
    {
      "epoch": 0.3271177195498899,
      "grad_norm": 3.117426872253418,
      "learning_rate": 9.539533714706522e-06,
      "loss": 1.0068,
      "step": 4048
    },
    {
      "epoch": 0.32719852926321746,
      "grad_norm": 3.088529348373413,
      "learning_rate": 9.5392593854021e-06,
      "loss": 1.0154,
      "step": 4049
    },
    {
      "epoch": 0.327279338976545,
      "grad_norm": 2.6547067165374756,
      "learning_rate": 9.538984978351092e-06,
      "loss": 1.0103,
      "step": 4050
    },
    {
      "epoch": 0.3273601486898725,
      "grad_norm": 3.094447135925293,
      "learning_rate": 9.538710493558199e-06,
      "loss": 0.975,
      "step": 4051
    },
    {
      "epoch": 0.3274409584032001,
      "grad_norm": 2.459456443786621,
      "learning_rate": 9.53843593102812e-06,
      "loss": 1.0546,
      "step": 4052
    },
    {
      "epoch": 0.3275217681165276,
      "grad_norm": 2.74337100982666,
      "learning_rate": 9.538161290765561e-06,
      "loss": 0.9193,
      "step": 4053
    },
    {
      "epoch": 0.32760257782985513,
      "grad_norm": 2.9267196655273438,
      "learning_rate": 9.537886572775225e-06,
      "loss": 0.9215,
      "step": 4054
    },
    {
      "epoch": 0.3276833875431827,
      "grad_norm": 3.0507187843322754,
      "learning_rate": 9.537611777061815e-06,
      "loss": 0.9681,
      "step": 4055
    },
    {
      "epoch": 0.32776419725651024,
      "grad_norm": 3.12231707572937,
      "learning_rate": 9.537336903630038e-06,
      "loss": 0.9909,
      "step": 4056
    },
    {
      "epoch": 0.32784500696983776,
      "grad_norm": 2.637436866760254,
      "learning_rate": 9.537061952484606e-06,
      "loss": 0.8506,
      "step": 4057
    },
    {
      "epoch": 0.32792581668316534,
      "grad_norm": 2.2959401607513428,
      "learning_rate": 9.536786923630223e-06,
      "loss": 0.9894,
      "step": 4058
    },
    {
      "epoch": 0.32800662639649286,
      "grad_norm": 3.118029832839966,
      "learning_rate": 9.536511817071602e-06,
      "loss": 1.0285,
      "step": 4059
    },
    {
      "epoch": 0.3280874361098204,
      "grad_norm": 2.8827004432678223,
      "learning_rate": 9.536236632813458e-06,
      "loss": 1.0687,
      "step": 4060
    },
    {
      "epoch": 0.32816824582314796,
      "grad_norm": 3.069139242172241,
      "learning_rate": 9.535961370860497e-06,
      "loss": 1.1305,
      "step": 4061
    },
    {
      "epoch": 0.3282490555364755,
      "grad_norm": 2.61059832572937,
      "learning_rate": 9.53568603121744e-06,
      "loss": 0.9611,
      "step": 4062
    },
    {
      "epoch": 0.328329865249803,
      "grad_norm": 2.6590816974639893,
      "learning_rate": 9.535410613889e-06,
      "loss": 0.9605,
      "step": 4063
    },
    {
      "epoch": 0.3284106749631306,
      "grad_norm": 2.606452465057373,
      "learning_rate": 9.535135118879895e-06,
      "loss": 0.9853,
      "step": 4064
    },
    {
      "epoch": 0.3284914846764581,
      "grad_norm": 2.8682236671447754,
      "learning_rate": 9.534859546194843e-06,
      "loss": 1.0149,
      "step": 4065
    },
    {
      "epoch": 0.32857229438978564,
      "grad_norm": 3.171236515045166,
      "learning_rate": 9.534583895838562e-06,
      "loss": 0.9628,
      "step": 4066
    },
    {
      "epoch": 0.3286531041031132,
      "grad_norm": 2.8859357833862305,
      "learning_rate": 9.534308167815778e-06,
      "loss": 1.0213,
      "step": 4067
    },
    {
      "epoch": 0.32873391381644074,
      "grad_norm": 2.733534097671509,
      "learning_rate": 9.534032362131211e-06,
      "loss": 0.9466,
      "step": 4068
    },
    {
      "epoch": 0.32881472352976826,
      "grad_norm": 2.582289218902588,
      "learning_rate": 9.533756478789585e-06,
      "loss": 0.9773,
      "step": 4069
    },
    {
      "epoch": 0.32889553324309584,
      "grad_norm": 3.1076979637145996,
      "learning_rate": 9.533480517795623e-06,
      "loss": 0.9909,
      "step": 4070
    },
    {
      "epoch": 0.32897634295642336,
      "grad_norm": 3.08135724067688,
      "learning_rate": 9.533204479154056e-06,
      "loss": 0.9729,
      "step": 4071
    },
    {
      "epoch": 0.3290571526697509,
      "grad_norm": 2.463721513748169,
      "learning_rate": 9.532928362869609e-06,
      "loss": 1.0492,
      "step": 4072
    },
    {
      "epoch": 0.32913796238307846,
      "grad_norm": 2.6856791973114014,
      "learning_rate": 9.532652168947011e-06,
      "loss": 0.9851,
      "step": 4073
    },
    {
      "epoch": 0.329218772096406,
      "grad_norm": 3.083483934402466,
      "learning_rate": 9.532375897390993e-06,
      "loss": 0.9419,
      "step": 4074
    },
    {
      "epoch": 0.3292995818097335,
      "grad_norm": 2.5787622928619385,
      "learning_rate": 9.532099548206288e-06,
      "loss": 0.9431,
      "step": 4075
    },
    {
      "epoch": 0.3293803915230611,
      "grad_norm": 2.474858283996582,
      "learning_rate": 9.531823121397628e-06,
      "loss": 1.0828,
      "step": 4076
    },
    {
      "epoch": 0.3294612012363886,
      "grad_norm": 2.7180659770965576,
      "learning_rate": 9.531546616969747e-06,
      "loss": 0.8964,
      "step": 4077
    },
    {
      "epoch": 0.32954201094971614,
      "grad_norm": 2.3666176795959473,
      "learning_rate": 9.531270034927383e-06,
      "loss": 0.9429,
      "step": 4078
    },
    {
      "epoch": 0.3296228206630437,
      "grad_norm": 2.6842448711395264,
      "learning_rate": 9.530993375275272e-06,
      "loss": 0.9373,
      "step": 4079
    },
    {
      "epoch": 0.32970363037637124,
      "grad_norm": 2.9368889331817627,
      "learning_rate": 9.53071663801815e-06,
      "loss": 0.9983,
      "step": 4080
    },
    {
      "epoch": 0.32978444008969876,
      "grad_norm": 2.569736957550049,
      "learning_rate": 9.530439823160761e-06,
      "loss": 0.9224,
      "step": 4081
    },
    {
      "epoch": 0.32986524980302634,
      "grad_norm": 2.9519383907318115,
      "learning_rate": 9.530162930707846e-06,
      "loss": 0.9483,
      "step": 4082
    },
    {
      "epoch": 0.32994605951635386,
      "grad_norm": 2.475437641143799,
      "learning_rate": 9.529885960664144e-06,
      "loss": 1.0032,
      "step": 4083
    },
    {
      "epoch": 0.3300268692296814,
      "grad_norm": 2.5936172008514404,
      "learning_rate": 9.529608913034401e-06,
      "loss": 1.1032,
      "step": 4084
    },
    {
      "epoch": 0.33010767894300896,
      "grad_norm": 2.4394097328186035,
      "learning_rate": 9.529331787823362e-06,
      "loss": 0.9851,
      "step": 4085
    },
    {
      "epoch": 0.3301884886563365,
      "grad_norm": 2.895510673522949,
      "learning_rate": 9.529054585035774e-06,
      "loss": 0.9855,
      "step": 4086
    },
    {
      "epoch": 0.330269298369664,
      "grad_norm": 3.4800808429718018,
      "learning_rate": 9.528777304676384e-06,
      "loss": 0.9552,
      "step": 4087
    },
    {
      "epoch": 0.3303501080829916,
      "grad_norm": 2.9893293380737305,
      "learning_rate": 9.528499946749941e-06,
      "loss": 1.0569,
      "step": 4088
    },
    {
      "epoch": 0.3304309177963191,
      "grad_norm": 2.711333990097046,
      "learning_rate": 9.528222511261196e-06,
      "loss": 0.9181,
      "step": 4089
    },
    {
      "epoch": 0.33051172750964664,
      "grad_norm": 3.0602447986602783,
      "learning_rate": 9.527944998214899e-06,
      "loss": 0.9545,
      "step": 4090
    },
    {
      "epoch": 0.3305925372229742,
      "grad_norm": 2.645359516143799,
      "learning_rate": 9.527667407615807e-06,
      "loss": 1.0609,
      "step": 4091
    },
    {
      "epoch": 0.33067334693630174,
      "grad_norm": 2.887587785720825,
      "learning_rate": 9.527389739468672e-06,
      "loss": 1.0834,
      "step": 4092
    },
    {
      "epoch": 0.33075415664962926,
      "grad_norm": 2.4695920944213867,
      "learning_rate": 9.527111993778248e-06,
      "loss": 0.8339,
      "step": 4093
    },
    {
      "epoch": 0.33083496636295684,
      "grad_norm": 2.296949863433838,
      "learning_rate": 9.526834170549298e-06,
      "loss": 1.0393,
      "step": 4094
    },
    {
      "epoch": 0.33091577607628436,
      "grad_norm": 3.023770332336426,
      "learning_rate": 9.526556269786573e-06,
      "loss": 0.9751,
      "step": 4095
    },
    {
      "epoch": 0.3309965857896119,
      "grad_norm": 2.6461973190307617,
      "learning_rate": 9.526278291494837e-06,
      "loss": 1.1395,
      "step": 4096
    },
    {
      "epoch": 0.33107739550293946,
      "grad_norm": 2.568086624145508,
      "learning_rate": 9.526000235678852e-06,
      "loss": 0.9888,
      "step": 4097
    },
    {
      "epoch": 0.331158205216267,
      "grad_norm": 3.141347885131836,
      "learning_rate": 9.525722102343377e-06,
      "loss": 0.9531,
      "step": 4098
    },
    {
      "epoch": 0.3312390149295945,
      "grad_norm": 2.9465346336364746,
      "learning_rate": 9.525443891493178e-06,
      "loss": 0.979,
      "step": 4099
    },
    {
      "epoch": 0.3313198246429221,
      "grad_norm": 3.0027921199798584,
      "learning_rate": 9.52516560313302e-06,
      "loss": 0.9878,
      "step": 4100
    },
    {
      "epoch": 0.3314006343562496,
      "grad_norm": 2.8586180210113525,
      "learning_rate": 9.524887237267671e-06,
      "loss": 0.9823,
      "step": 4101
    },
    {
      "epoch": 0.33148144406957714,
      "grad_norm": 3.216400384902954,
      "learning_rate": 9.524608793901893e-06,
      "loss": 1.0316,
      "step": 4102
    },
    {
      "epoch": 0.3315622537829047,
      "grad_norm": 2.8101775646209717,
      "learning_rate": 9.524330273040462e-06,
      "loss": 0.9666,
      "step": 4103
    },
    {
      "epoch": 0.33164306349623224,
      "grad_norm": 3.338576078414917,
      "learning_rate": 9.524051674688145e-06,
      "loss": 1.0002,
      "step": 4104
    },
    {
      "epoch": 0.3317238732095598,
      "grad_norm": 2.9538190364837646,
      "learning_rate": 9.523772998849712e-06,
      "loss": 0.9646,
      "step": 4105
    },
    {
      "epoch": 0.33180468292288734,
      "grad_norm": 2.605074167251587,
      "learning_rate": 9.52349424552994e-06,
      "loss": 0.9202,
      "step": 4106
    },
    {
      "epoch": 0.33188549263621486,
      "grad_norm": 2.7699732780456543,
      "learning_rate": 9.5232154147336e-06,
      "loss": 0.9016,
      "step": 4107
    },
    {
      "epoch": 0.33196630234954244,
      "grad_norm": 2.916870355606079,
      "learning_rate": 9.52293650646547e-06,
      "loss": 0.8376,
      "step": 4108
    },
    {
      "epoch": 0.33204711206286996,
      "grad_norm": 2.7381067276000977,
      "learning_rate": 9.522657520730327e-06,
      "loss": 0.9252,
      "step": 4109
    },
    {
      "epoch": 0.3321279217761975,
      "grad_norm": 2.4672820568084717,
      "learning_rate": 9.522378457532948e-06,
      "loss": 1.0367,
      "step": 4110
    },
    {
      "epoch": 0.33220873148952507,
      "grad_norm": 2.7581498622894287,
      "learning_rate": 9.522099316878113e-06,
      "loss": 1.0315,
      "step": 4111
    },
    {
      "epoch": 0.3322895412028526,
      "grad_norm": 2.723599433898926,
      "learning_rate": 9.521820098770602e-06,
      "loss": 0.9118,
      "step": 4112
    },
    {
      "epoch": 0.3323703509161801,
      "grad_norm": 2.8597822189331055,
      "learning_rate": 9.521540803215199e-06,
      "loss": 1.0321,
      "step": 4113
    },
    {
      "epoch": 0.3324511606295077,
      "grad_norm": 2.4662044048309326,
      "learning_rate": 9.521261430216689e-06,
      "loss": 0.9933,
      "step": 4114
    },
    {
      "epoch": 0.3325319703428352,
      "grad_norm": 2.8004307746887207,
      "learning_rate": 9.520981979779853e-06,
      "loss": 1.0135,
      "step": 4115
    },
    {
      "epoch": 0.33261278005616274,
      "grad_norm": 2.5971570014953613,
      "learning_rate": 9.520702451909481e-06,
      "loss": 0.92,
      "step": 4116
    },
    {
      "epoch": 0.3326935897694903,
      "grad_norm": 2.7834208011627197,
      "learning_rate": 9.520422846610359e-06,
      "loss": 1.0115,
      "step": 4117
    },
    {
      "epoch": 0.33277439948281784,
      "grad_norm": 2.5413529872894287,
      "learning_rate": 9.520143163887277e-06,
      "loss": 0.9328,
      "step": 4118
    },
    {
      "epoch": 0.33285520919614536,
      "grad_norm": 2.7738726139068604,
      "learning_rate": 9.519863403745023e-06,
      "loss": 1.0739,
      "step": 4119
    },
    {
      "epoch": 0.33293601890947294,
      "grad_norm": 2.608670949935913,
      "learning_rate": 9.519583566188389e-06,
      "loss": 1.0701,
      "step": 4120
    },
    {
      "epoch": 0.33301682862280046,
      "grad_norm": 2.788914442062378,
      "learning_rate": 9.519303651222171e-06,
      "loss": 1.0519,
      "step": 4121
    },
    {
      "epoch": 0.333097638336128,
      "grad_norm": 3.1283442974090576,
      "learning_rate": 9.51902365885116e-06,
      "loss": 1.0496,
      "step": 4122
    },
    {
      "epoch": 0.33317844804945557,
      "grad_norm": 2.777540922164917,
      "learning_rate": 9.518743589080153e-06,
      "loss": 0.9842,
      "step": 4123
    },
    {
      "epoch": 0.3332592577627831,
      "grad_norm": 2.6435718536376953,
      "learning_rate": 9.518463441913947e-06,
      "loss": 0.948,
      "step": 4124
    },
    {
      "epoch": 0.3333400674761106,
      "grad_norm": 2.7920753955841064,
      "learning_rate": 9.51818321735734e-06,
      "loss": 1.032,
      "step": 4125
    },
    {
      "epoch": 0.3334208771894382,
      "grad_norm": 2.519684076309204,
      "learning_rate": 9.517902915415131e-06,
      "loss": 0.9893,
      "step": 4126
    },
    {
      "epoch": 0.3335016869027657,
      "grad_norm": 2.717278480529785,
      "learning_rate": 9.517622536092123e-06,
      "loss": 1.0427,
      "step": 4127
    },
    {
      "epoch": 0.33358249661609324,
      "grad_norm": 2.4553167819976807,
      "learning_rate": 9.517342079393114e-06,
      "loss": 0.992,
      "step": 4128
    },
    {
      "epoch": 0.3336633063294208,
      "grad_norm": 2.500657320022583,
      "learning_rate": 9.517061545322912e-06,
      "loss": 1.003,
      "step": 4129
    },
    {
      "epoch": 0.33374411604274834,
      "grad_norm": 2.820107936859131,
      "learning_rate": 9.51678093388632e-06,
      "loss": 1.0727,
      "step": 4130
    },
    {
      "epoch": 0.33382492575607586,
      "grad_norm": 2.645263910293579,
      "learning_rate": 9.516500245088144e-06,
      "loss": 0.9722,
      "step": 4131
    },
    {
      "epoch": 0.33390573546940344,
      "grad_norm": 2.426103353500366,
      "learning_rate": 9.516219478933193e-06,
      "loss": 1.1225,
      "step": 4132
    },
    {
      "epoch": 0.33398654518273097,
      "grad_norm": 2.993973970413208,
      "learning_rate": 9.515938635426274e-06,
      "loss": 0.9664,
      "step": 4133
    },
    {
      "epoch": 0.3340673548960585,
      "grad_norm": 2.6964805126190186,
      "learning_rate": 9.515657714572197e-06,
      "loss": 0.9502,
      "step": 4134
    },
    {
      "epoch": 0.33414816460938607,
      "grad_norm": 2.846302032470703,
      "learning_rate": 9.515376716375777e-06,
      "loss": 1.0977,
      "step": 4135
    },
    {
      "epoch": 0.3342289743227136,
      "grad_norm": 2.6280717849731445,
      "learning_rate": 9.515095640841822e-06,
      "loss": 0.925,
      "step": 4136
    },
    {
      "epoch": 0.3343097840360411,
      "grad_norm": 3.2251338958740234,
      "learning_rate": 9.514814487975148e-06,
      "loss": 1.063,
      "step": 4137
    },
    {
      "epoch": 0.3343905937493687,
      "grad_norm": 2.701317310333252,
      "learning_rate": 9.514533257780571e-06,
      "loss": 0.967,
      "step": 4138
    },
    {
      "epoch": 0.3344714034626962,
      "grad_norm": 2.729400396347046,
      "learning_rate": 9.514251950262908e-06,
      "loss": 1.0842,
      "step": 4139
    },
    {
      "epoch": 0.33455221317602374,
      "grad_norm": 2.852540969848633,
      "learning_rate": 9.513970565426978e-06,
      "loss": 0.9324,
      "step": 4140
    },
    {
      "epoch": 0.3346330228893513,
      "grad_norm": 2.7535130977630615,
      "learning_rate": 9.513689103277597e-06,
      "loss": 0.9181,
      "step": 4141
    },
    {
      "epoch": 0.33471383260267884,
      "grad_norm": 2.5309736728668213,
      "learning_rate": 9.513407563819589e-06,
      "loss": 0.9963,
      "step": 4142
    },
    {
      "epoch": 0.33479464231600636,
      "grad_norm": 2.8194198608398438,
      "learning_rate": 9.513125947057776e-06,
      "loss": 1.0348,
      "step": 4143
    },
    {
      "epoch": 0.33487545202933394,
      "grad_norm": 2.8604578971862793,
      "learning_rate": 9.512844252996978e-06,
      "loss": 1.1125,
      "step": 4144
    },
    {
      "epoch": 0.33495626174266147,
      "grad_norm": 2.880497932434082,
      "learning_rate": 9.512562481642023e-06,
      "loss": 0.9588,
      "step": 4145
    },
    {
      "epoch": 0.335037071455989,
      "grad_norm": 2.5192902088165283,
      "learning_rate": 9.512280632997737e-06,
      "loss": 1.0429,
      "step": 4146
    },
    {
      "epoch": 0.33511788116931657,
      "grad_norm": 2.6330995559692383,
      "learning_rate": 9.511998707068946e-06,
      "loss": 0.9643,
      "step": 4147
    },
    {
      "epoch": 0.3351986908826441,
      "grad_norm": 2.7292113304138184,
      "learning_rate": 9.511716703860479e-06,
      "loss": 0.9461,
      "step": 4148
    },
    {
      "epoch": 0.3352795005959716,
      "grad_norm": 2.478864908218384,
      "learning_rate": 9.511434623377167e-06,
      "loss": 0.9124,
      "step": 4149
    },
    {
      "epoch": 0.3353603103092992,
      "grad_norm": 2.4361627101898193,
      "learning_rate": 9.51115246562384e-06,
      "loss": 0.9846,
      "step": 4150
    },
    {
      "epoch": 0.3354411200226267,
      "grad_norm": 2.6726222038269043,
      "learning_rate": 9.510870230605333e-06,
      "loss": 1.0666,
      "step": 4151
    },
    {
      "epoch": 0.33552192973595424,
      "grad_norm": 3.023792266845703,
      "learning_rate": 9.510587918326477e-06,
      "loss": 1.0163,
      "step": 4152
    },
    {
      "epoch": 0.3356027394492818,
      "grad_norm": 3.5809295177459717,
      "learning_rate": 9.51030552879211e-06,
      "loss": 0.9217,
      "step": 4153
    },
    {
      "epoch": 0.33568354916260934,
      "grad_norm": 2.224635601043701,
      "learning_rate": 9.510023062007067e-06,
      "loss": 0.9508,
      "step": 4154
    },
    {
      "epoch": 0.33576435887593686,
      "grad_norm": 2.5898799896240234,
      "learning_rate": 9.509740517976186e-06,
      "loss": 0.9548,
      "step": 4155
    },
    {
      "epoch": 0.33584516858926444,
      "grad_norm": 2.8230583667755127,
      "learning_rate": 9.509457896704306e-06,
      "loss": 0.9562,
      "step": 4156
    },
    {
      "epoch": 0.33592597830259197,
      "grad_norm": 2.452807664871216,
      "learning_rate": 9.50917519819627e-06,
      "loss": 1.0768,
      "step": 4157
    },
    {
      "epoch": 0.3360067880159195,
      "grad_norm": 2.7506778240203857,
      "learning_rate": 9.508892422456916e-06,
      "loss": 1.0678,
      "step": 4158
    },
    {
      "epoch": 0.33608759772924707,
      "grad_norm": 2.8394293785095215,
      "learning_rate": 9.508609569491091e-06,
      "loss": 0.9876,
      "step": 4159
    },
    {
      "epoch": 0.3361684074425746,
      "grad_norm": 3.0543792247772217,
      "learning_rate": 9.508326639303639e-06,
      "loss": 0.9489,
      "step": 4160
    },
    {
      "epoch": 0.3362492171559021,
      "grad_norm": 2.382162570953369,
      "learning_rate": 9.508043631899405e-06,
      "loss": 1.0068,
      "step": 4161
    },
    {
      "epoch": 0.3363300268692297,
      "grad_norm": 2.6702747344970703,
      "learning_rate": 9.507760547283233e-06,
      "loss": 0.94,
      "step": 4162
    },
    {
      "epoch": 0.3364108365825572,
      "grad_norm": 2.6312103271484375,
      "learning_rate": 9.507477385459978e-06,
      "loss": 1.0088,
      "step": 4163
    },
    {
      "epoch": 0.33649164629588474,
      "grad_norm": 3.097381353378296,
      "learning_rate": 9.507194146434486e-06,
      "loss": 0.8685,
      "step": 4164
    },
    {
      "epoch": 0.3365724560092123,
      "grad_norm": 2.7595736980438232,
      "learning_rate": 9.50691083021161e-06,
      "loss": 1.0228,
      "step": 4165
    },
    {
      "epoch": 0.33665326572253984,
      "grad_norm": 2.5723166465759277,
      "learning_rate": 9.506627436796199e-06,
      "loss": 0.8889,
      "step": 4166
    },
    {
      "epoch": 0.33673407543586736,
      "grad_norm": 3.3848299980163574,
      "learning_rate": 9.50634396619311e-06,
      "loss": 1.0653,
      "step": 4167
    },
    {
      "epoch": 0.33681488514919494,
      "grad_norm": 2.8883447647094727,
      "learning_rate": 9.506060418407197e-06,
      "loss": 0.8401,
      "step": 4168
    },
    {
      "epoch": 0.33689569486252247,
      "grad_norm": 2.591768980026245,
      "learning_rate": 9.505776793443318e-06,
      "loss": 1.0239,
      "step": 4169
    },
    {
      "epoch": 0.33697650457585004,
      "grad_norm": 2.822874069213867,
      "learning_rate": 9.50549309130633e-06,
      "loss": 1.0274,
      "step": 4170
    },
    {
      "epoch": 0.33705731428917757,
      "grad_norm": 2.902327537536621,
      "learning_rate": 9.505209312001091e-06,
      "loss": 1.1014,
      "step": 4171
    },
    {
      "epoch": 0.3371381240025051,
      "grad_norm": 2.242377996444702,
      "learning_rate": 9.504925455532463e-06,
      "loss": 0.9613,
      "step": 4172
    },
    {
      "epoch": 0.33721893371583267,
      "grad_norm": 2.6212575435638428,
      "learning_rate": 9.504641521905306e-06,
      "loss": 0.8804,
      "step": 4173
    },
    {
      "epoch": 0.3372997434291602,
      "grad_norm": 2.40804123878479,
      "learning_rate": 9.504357511124487e-06,
      "loss": 0.9385,
      "step": 4174
    },
    {
      "epoch": 0.3373805531424877,
      "grad_norm": 2.574183940887451,
      "learning_rate": 9.504073423194864e-06,
      "loss": 0.9373,
      "step": 4175
    },
    {
      "epoch": 0.3374613628558153,
      "grad_norm": 2.52545428276062,
      "learning_rate": 9.503789258121309e-06,
      "loss": 1.1001,
      "step": 4176
    },
    {
      "epoch": 0.3375421725691428,
      "grad_norm": 2.7722158432006836,
      "learning_rate": 9.503505015908685e-06,
      "loss": 0.9674,
      "step": 4177
    },
    {
      "epoch": 0.33762298228247034,
      "grad_norm": 2.7113494873046875,
      "learning_rate": 9.503220696561863e-06,
      "loss": 1.0573,
      "step": 4178
    },
    {
      "epoch": 0.3377037919957979,
      "grad_norm": 2.638622999191284,
      "learning_rate": 9.50293630008571e-06,
      "loss": 0.8603,
      "step": 4179
    },
    {
      "epoch": 0.33778460170912544,
      "grad_norm": 2.8496179580688477,
      "learning_rate": 9.5026518264851e-06,
      "loss": 0.9971,
      "step": 4180
    },
    {
      "epoch": 0.33786541142245297,
      "grad_norm": 3.526613473892212,
      "learning_rate": 9.502367275764904e-06,
      "loss": 0.8981,
      "step": 4181
    },
    {
      "epoch": 0.33794622113578054,
      "grad_norm": 2.460406541824341,
      "learning_rate": 9.502082647929993e-06,
      "loss": 0.9297,
      "step": 4182
    },
    {
      "epoch": 0.33802703084910807,
      "grad_norm": 2.441196918487549,
      "learning_rate": 9.501797942985247e-06,
      "loss": 1.0186,
      "step": 4183
    },
    {
      "epoch": 0.3381078405624356,
      "grad_norm": 2.971407651901245,
      "learning_rate": 9.50151316093554e-06,
      "loss": 1.1108,
      "step": 4184
    },
    {
      "epoch": 0.33818865027576317,
      "grad_norm": 2.512982130050659,
      "learning_rate": 9.501228301785748e-06,
      "loss": 0.9974,
      "step": 4185
    },
    {
      "epoch": 0.3382694599890907,
      "grad_norm": 2.587165117263794,
      "learning_rate": 9.500943365540753e-06,
      "loss": 1.0078,
      "step": 4186
    },
    {
      "epoch": 0.3383502697024182,
      "grad_norm": 2.699758529663086,
      "learning_rate": 9.500658352205433e-06,
      "loss": 1.0141,
      "step": 4187
    },
    {
      "epoch": 0.3384310794157458,
      "grad_norm": 2.9663798809051514,
      "learning_rate": 9.50037326178467e-06,
      "loss": 0.8985,
      "step": 4188
    },
    {
      "epoch": 0.3385118891290733,
      "grad_norm": 2.920869827270508,
      "learning_rate": 9.500088094283347e-06,
      "loss": 0.9699,
      "step": 4189
    },
    {
      "epoch": 0.33859269884240084,
      "grad_norm": 2.5547428131103516,
      "learning_rate": 9.499802849706348e-06,
      "loss": 0.8421,
      "step": 4190
    },
    {
      "epoch": 0.3386735085557284,
      "grad_norm": 2.5543925762176514,
      "learning_rate": 9.499517528058562e-06,
      "loss": 0.966,
      "step": 4191
    },
    {
      "epoch": 0.33875431826905594,
      "grad_norm": 2.707092046737671,
      "learning_rate": 9.49923212934487e-06,
      "loss": 0.9476,
      "step": 4192
    },
    {
      "epoch": 0.33883512798238347,
      "grad_norm": 2.891502618789673,
      "learning_rate": 9.498946653570164e-06,
      "loss": 0.9965,
      "step": 4193
    },
    {
      "epoch": 0.33891593769571104,
      "grad_norm": 2.6104674339294434,
      "learning_rate": 9.498661100739332e-06,
      "loss": 1.0823,
      "step": 4194
    },
    {
      "epoch": 0.33899674740903857,
      "grad_norm": 2.9352173805236816,
      "learning_rate": 9.498375470857266e-06,
      "loss": 1.0509,
      "step": 4195
    },
    {
      "epoch": 0.3390775571223661,
      "grad_norm": 2.6980602741241455,
      "learning_rate": 9.498089763928857e-06,
      "loss": 0.9992,
      "step": 4196
    },
    {
      "epoch": 0.33915836683569367,
      "grad_norm": 2.7117533683776855,
      "learning_rate": 9.497803979959e-06,
      "loss": 1.0155,
      "step": 4197
    },
    {
      "epoch": 0.3392391765490212,
      "grad_norm": 2.6956875324249268,
      "learning_rate": 9.497518118952588e-06,
      "loss": 0.9609,
      "step": 4198
    },
    {
      "epoch": 0.3393199862623487,
      "grad_norm": 2.6220474243164062,
      "learning_rate": 9.49723218091452e-06,
      "loss": 0.9975,
      "step": 4199
    },
    {
      "epoch": 0.3394007959756763,
      "grad_norm": 2.730402708053589,
      "learning_rate": 9.49694616584969e-06,
      "loss": 1.1123,
      "step": 4200
    },
    {
      "epoch": 0.3394816056890038,
      "grad_norm": 2.820833206176758,
      "learning_rate": 9.496660073762998e-06,
      "loss": 1.023,
      "step": 4201
    },
    {
      "epoch": 0.33956241540233134,
      "grad_norm": 2.9102706909179688,
      "learning_rate": 9.496373904659344e-06,
      "loss": 0.903,
      "step": 4202
    },
    {
      "epoch": 0.3396432251156589,
      "grad_norm": 3.058046340942383,
      "learning_rate": 9.496087658543629e-06,
      "loss": 0.9479,
      "step": 4203
    },
    {
      "epoch": 0.33972403482898644,
      "grad_norm": 2.2796804904937744,
      "learning_rate": 9.495801335420757e-06,
      "loss": 0.9136,
      "step": 4204
    },
    {
      "epoch": 0.33980484454231397,
      "grad_norm": 2.5576324462890625,
      "learning_rate": 9.495514935295631e-06,
      "loss": 1.024,
      "step": 4205
    },
    {
      "epoch": 0.33988565425564154,
      "grad_norm": 2.435377597808838,
      "learning_rate": 9.495228458173159e-06,
      "loss": 0.8844,
      "step": 4206
    },
    {
      "epoch": 0.33996646396896907,
      "grad_norm": 2.8297946453094482,
      "learning_rate": 9.494941904058241e-06,
      "loss": 0.9269,
      "step": 4207
    },
    {
      "epoch": 0.3400472736822966,
      "grad_norm": 2.8731467723846436,
      "learning_rate": 9.494655272955792e-06,
      "loss": 1.0457,
      "step": 4208
    },
    {
      "epoch": 0.34012808339562417,
      "grad_norm": 2.9491989612579346,
      "learning_rate": 9.49436856487072e-06,
      "loss": 0.8883,
      "step": 4209
    },
    {
      "epoch": 0.3402088931089517,
      "grad_norm": 2.664522647857666,
      "learning_rate": 9.494081779807931e-06,
      "loss": 0.9569,
      "step": 4210
    },
    {
      "epoch": 0.3402897028222792,
      "grad_norm": 2.518174886703491,
      "learning_rate": 9.493794917772342e-06,
      "loss": 1.0473,
      "step": 4211
    },
    {
      "epoch": 0.3403705125356068,
      "grad_norm": 2.750535249710083,
      "learning_rate": 9.493507978768865e-06,
      "loss": 0.9221,
      "step": 4212
    },
    {
      "epoch": 0.3404513222489343,
      "grad_norm": 2.493407964706421,
      "learning_rate": 9.493220962802413e-06,
      "loss": 1.0743,
      "step": 4213
    },
    {
      "epoch": 0.34053213196226184,
      "grad_norm": 2.778715133666992,
      "learning_rate": 9.492933869877902e-06,
      "loss": 0.9011,
      "step": 4214
    },
    {
      "epoch": 0.3406129416755894,
      "grad_norm": 3.035158634185791,
      "learning_rate": 9.492646700000252e-06,
      "loss": 0.9334,
      "step": 4215
    },
    {
      "epoch": 0.34069375138891694,
      "grad_norm": 2.4467737674713135,
      "learning_rate": 9.492359453174377e-06,
      "loss": 0.8573,
      "step": 4216
    },
    {
      "epoch": 0.34077456110224447,
      "grad_norm": 2.7152509689331055,
      "learning_rate": 9.492072129405203e-06,
      "loss": 0.9816,
      "step": 4217
    },
    {
      "epoch": 0.34085537081557205,
      "grad_norm": 3.1276209354400635,
      "learning_rate": 9.491784728697646e-06,
      "loss": 0.9982,
      "step": 4218
    },
    {
      "epoch": 0.34093618052889957,
      "grad_norm": 2.524350166320801,
      "learning_rate": 9.49149725105663e-06,
      "loss": 0.8906,
      "step": 4219
    },
    {
      "epoch": 0.3410169902422271,
      "grad_norm": 2.934344530105591,
      "learning_rate": 9.491209696487077e-06,
      "loss": 0.9201,
      "step": 4220
    },
    {
      "epoch": 0.34109779995555467,
      "grad_norm": 2.8160815238952637,
      "learning_rate": 9.490922064993917e-06,
      "loss": 0.8818,
      "step": 4221
    },
    {
      "epoch": 0.3411786096688822,
      "grad_norm": 2.7600460052490234,
      "learning_rate": 9.490634356582072e-06,
      "loss": 1.055,
      "step": 4222
    },
    {
      "epoch": 0.3412594193822097,
      "grad_norm": 3.295536518096924,
      "learning_rate": 9.490346571256472e-06,
      "loss": 0.9552,
      "step": 4223
    },
    {
      "epoch": 0.3413402290955373,
      "grad_norm": 2.6677610874176025,
      "learning_rate": 9.490058709022045e-06,
      "loss": 0.9235,
      "step": 4224
    },
    {
      "epoch": 0.3414210388088648,
      "grad_norm": 3.1460516452789307,
      "learning_rate": 9.48977076988372e-06,
      "loss": 0.9745,
      "step": 4225
    },
    {
      "epoch": 0.34150184852219234,
      "grad_norm": 2.595522165298462,
      "learning_rate": 9.489482753846435e-06,
      "loss": 1.0154,
      "step": 4226
    },
    {
      "epoch": 0.3415826582355199,
      "grad_norm": 2.8672518730163574,
      "learning_rate": 9.489194660915115e-06,
      "loss": 0.9013,
      "step": 4227
    },
    {
      "epoch": 0.34166346794884744,
      "grad_norm": 3.376835584640503,
      "learning_rate": 9.488906491094698e-06,
      "loss": 1.0286,
      "step": 4228
    },
    {
      "epoch": 0.34174427766217497,
      "grad_norm": 2.7237389087677,
      "learning_rate": 9.48861824439012e-06,
      "loss": 1.034,
      "step": 4229
    },
    {
      "epoch": 0.34182508737550255,
      "grad_norm": 3.1918201446533203,
      "learning_rate": 9.488329920806316e-06,
      "loss": 0.8988,
      "step": 4230
    },
    {
      "epoch": 0.34190589708883007,
      "grad_norm": 3.385815382003784,
      "learning_rate": 9.488041520348228e-06,
      "loss": 0.924,
      "step": 4231
    },
    {
      "epoch": 0.3419867068021576,
      "grad_norm": 2.591743230819702,
      "learning_rate": 9.48775304302079e-06,
      "loss": 0.9091,
      "step": 4232
    },
    {
      "epoch": 0.34206751651548517,
      "grad_norm": 2.834713935852051,
      "learning_rate": 9.48746448882895e-06,
      "loss": 0.8816,
      "step": 4233
    },
    {
      "epoch": 0.3421483262288127,
      "grad_norm": 2.700610637664795,
      "learning_rate": 9.487175857777644e-06,
      "loss": 1.0109,
      "step": 4234
    },
    {
      "epoch": 0.3422291359421403,
      "grad_norm": 2.9804446697235107,
      "learning_rate": 9.48688714987182e-06,
      "loss": 0.971,
      "step": 4235
    },
    {
      "epoch": 0.3423099456554678,
      "grad_norm": 2.7324395179748535,
      "learning_rate": 9.486598365116418e-06,
      "loss": 1.0875,
      "step": 4236
    },
    {
      "epoch": 0.3423907553687953,
      "grad_norm": 2.8630380630493164,
      "learning_rate": 9.486309503516388e-06,
      "loss": 0.8134,
      "step": 4237
    },
    {
      "epoch": 0.3424715650821229,
      "grad_norm": 2.923279047012329,
      "learning_rate": 9.486020565076677e-06,
      "loss": 0.8796,
      "step": 4238
    },
    {
      "epoch": 0.3425523747954504,
      "grad_norm": 3.012637138366699,
      "learning_rate": 9.485731549802235e-06,
      "loss": 1.0446,
      "step": 4239
    },
    {
      "epoch": 0.34263318450877794,
      "grad_norm": 2.2806928157806396,
      "learning_rate": 9.48544245769801e-06,
      "loss": 1.0663,
      "step": 4240
    },
    {
      "epoch": 0.3427139942221055,
      "grad_norm": 2.2977986335754395,
      "learning_rate": 9.485153288768951e-06,
      "loss": 0.9665,
      "step": 4241
    },
    {
      "epoch": 0.34279480393543305,
      "grad_norm": 3.0221078395843506,
      "learning_rate": 9.484864043020017e-06,
      "loss": 0.8723,
      "step": 4242
    },
    {
      "epoch": 0.34287561364876057,
      "grad_norm": 2.7185659408569336,
      "learning_rate": 9.484574720456156e-06,
      "loss": 0.9093,
      "step": 4243
    },
    {
      "epoch": 0.34295642336208815,
      "grad_norm": 2.2894680500030518,
      "learning_rate": 9.484285321082329e-06,
      "loss": 0.9966,
      "step": 4244
    },
    {
      "epoch": 0.34303723307541567,
      "grad_norm": 2.9871702194213867,
      "learning_rate": 9.483995844903488e-06,
      "loss": 0.9714,
      "step": 4245
    },
    {
      "epoch": 0.3431180427887432,
      "grad_norm": 2.6651320457458496,
      "learning_rate": 9.483706291924593e-06,
      "loss": 1.0343,
      "step": 4246
    },
    {
      "epoch": 0.3431988525020708,
      "grad_norm": 2.907731294631958,
      "learning_rate": 9.483416662150604e-06,
      "loss": 0.9251,
      "step": 4247
    },
    {
      "epoch": 0.3432796622153983,
      "grad_norm": 2.44343638420105,
      "learning_rate": 9.483126955586481e-06,
      "loss": 1.0846,
      "step": 4248
    },
    {
      "epoch": 0.3433604719287258,
      "grad_norm": 2.701967477798462,
      "learning_rate": 9.482837172237185e-06,
      "loss": 1.0523,
      "step": 4249
    },
    {
      "epoch": 0.3434412816420534,
      "grad_norm": 3.1003775596618652,
      "learning_rate": 9.482547312107682e-06,
      "loss": 0.932,
      "step": 4250
    },
    {
      "epoch": 0.3435220913553809,
      "grad_norm": 2.4994635581970215,
      "learning_rate": 9.482257375202934e-06,
      "loss": 0.944,
      "step": 4251
    },
    {
      "epoch": 0.34360290106870844,
      "grad_norm": 2.710099697113037,
      "learning_rate": 9.481967361527907e-06,
      "loss": 1.012,
      "step": 4252
    },
    {
      "epoch": 0.343683710782036,
      "grad_norm": 3.0543906688690186,
      "learning_rate": 9.48167727108757e-06,
      "loss": 0.9846,
      "step": 4253
    },
    {
      "epoch": 0.34376452049536355,
      "grad_norm": 2.9311554431915283,
      "learning_rate": 9.48138710388689e-06,
      "loss": 0.8771,
      "step": 4254
    },
    {
      "epoch": 0.34384533020869107,
      "grad_norm": 2.8884940147399902,
      "learning_rate": 9.481096859930839e-06,
      "loss": 0.8946,
      "step": 4255
    },
    {
      "epoch": 0.34392613992201865,
      "grad_norm": 2.545696258544922,
      "learning_rate": 9.480806539224383e-06,
      "loss": 1.0564,
      "step": 4256
    },
    {
      "epoch": 0.34400694963534617,
      "grad_norm": 3.0662972927093506,
      "learning_rate": 9.480516141772501e-06,
      "loss": 1.0557,
      "step": 4257
    },
    {
      "epoch": 0.3440877593486737,
      "grad_norm": 2.9301185607910156,
      "learning_rate": 9.480225667580164e-06,
      "loss": 0.9476,
      "step": 4258
    },
    {
      "epoch": 0.3441685690620013,
      "grad_norm": 2.5623271465301514,
      "learning_rate": 9.479935116652345e-06,
      "loss": 1.0103,
      "step": 4259
    },
    {
      "epoch": 0.3442493787753288,
      "grad_norm": 2.6191351413726807,
      "learning_rate": 9.479644488994025e-06,
      "loss": 0.8592,
      "step": 4260
    },
    {
      "epoch": 0.3443301884886563,
      "grad_norm": 3.001798629760742,
      "learning_rate": 9.479353784610177e-06,
      "loss": 1.0849,
      "step": 4261
    },
    {
      "epoch": 0.3444109982019839,
      "grad_norm": 2.927638530731201,
      "learning_rate": 9.479063003505782e-06,
      "loss": 1.06,
      "step": 4262
    },
    {
      "epoch": 0.3444918079153114,
      "grad_norm": 2.7032415866851807,
      "learning_rate": 9.478772145685821e-06,
      "loss": 1.0905,
      "step": 4263
    },
    {
      "epoch": 0.34457261762863894,
      "grad_norm": 3.0599684715270996,
      "learning_rate": 9.478481211155277e-06,
      "loss": 0.9663,
      "step": 4264
    },
    {
      "epoch": 0.3446534273419665,
      "grad_norm": 2.296513319015503,
      "learning_rate": 9.478190199919131e-06,
      "loss": 0.9691,
      "step": 4265
    },
    {
      "epoch": 0.34473423705529405,
      "grad_norm": 2.9048807621002197,
      "learning_rate": 9.477899111982367e-06,
      "loss": 1.1204,
      "step": 4266
    },
    {
      "epoch": 0.34481504676862157,
      "grad_norm": 2.9777579307556152,
      "learning_rate": 9.477607947349971e-06,
      "loss": 0.9684,
      "step": 4267
    },
    {
      "epoch": 0.34489585648194915,
      "grad_norm": 3.0115413665771484,
      "learning_rate": 9.47731670602693e-06,
      "loss": 0.8818,
      "step": 4268
    },
    {
      "epoch": 0.34497666619527667,
      "grad_norm": 2.541673183441162,
      "learning_rate": 9.477025388018235e-06,
      "loss": 0.987,
      "step": 4269
    },
    {
      "epoch": 0.3450574759086042,
      "grad_norm": 2.505735397338867,
      "learning_rate": 9.47673399332887e-06,
      "loss": 0.9997,
      "step": 4270
    },
    {
      "epoch": 0.3451382856219318,
      "grad_norm": 2.5219671726226807,
      "learning_rate": 9.476442521963831e-06,
      "loss": 1.0012,
      "step": 4271
    },
    {
      "epoch": 0.3452190953352593,
      "grad_norm": 2.983621597290039,
      "learning_rate": 9.476150973928107e-06,
      "loss": 0.8977,
      "step": 4272
    },
    {
      "epoch": 0.3452999050485868,
      "grad_norm": 3.414088726043701,
      "learning_rate": 9.475859349226693e-06,
      "loss": 0.8983,
      "step": 4273
    },
    {
      "epoch": 0.3453807147619144,
      "grad_norm": 2.7984800338745117,
      "learning_rate": 9.475567647864584e-06,
      "loss": 0.9156,
      "step": 4274
    },
    {
      "epoch": 0.3454615244752419,
      "grad_norm": 2.5534207820892334,
      "learning_rate": 9.475275869846776e-06,
      "loss": 1.0151,
      "step": 4275
    },
    {
      "epoch": 0.34554233418856944,
      "grad_norm": 2.717841625213623,
      "learning_rate": 9.474984015178266e-06,
      "loss": 0.935,
      "step": 4276
    },
    {
      "epoch": 0.345623143901897,
      "grad_norm": 2.970533847808838,
      "learning_rate": 9.474692083864052e-06,
      "loss": 0.9799,
      "step": 4277
    },
    {
      "epoch": 0.34570395361522455,
      "grad_norm": 2.94915771484375,
      "learning_rate": 9.474400075909136e-06,
      "loss": 0.8858,
      "step": 4278
    },
    {
      "epoch": 0.34578476332855207,
      "grad_norm": 2.6709272861480713,
      "learning_rate": 9.474107991318517e-06,
      "loss": 1.0326,
      "step": 4279
    },
    {
      "epoch": 0.34586557304187965,
      "grad_norm": 2.8930013179779053,
      "learning_rate": 9.4738158300972e-06,
      "loss": 0.9573,
      "step": 4280
    },
    {
      "epoch": 0.34594638275520717,
      "grad_norm": 2.6665401458740234,
      "learning_rate": 9.473523592250188e-06,
      "loss": 1.0777,
      "step": 4281
    },
    {
      "epoch": 0.3460271924685347,
      "grad_norm": 2.5457189083099365,
      "learning_rate": 9.473231277782486e-06,
      "loss": 1.1171,
      "step": 4282
    },
    {
      "epoch": 0.3461080021818623,
      "grad_norm": 3.2128570079803467,
      "learning_rate": 9.472938886699103e-06,
      "loss": 1.0744,
      "step": 4283
    },
    {
      "epoch": 0.3461888118951898,
      "grad_norm": 2.9156112670898438,
      "learning_rate": 9.472646419005043e-06,
      "loss": 0.9827,
      "step": 4284
    },
    {
      "epoch": 0.3462696216085173,
      "grad_norm": 2.5341570377349854,
      "learning_rate": 9.472353874705318e-06,
      "loss": 1.0247,
      "step": 4285
    },
    {
      "epoch": 0.3463504313218449,
      "grad_norm": 2.9804251194000244,
      "learning_rate": 9.472061253804937e-06,
      "loss": 0.9844,
      "step": 4286
    },
    {
      "epoch": 0.3464312410351724,
      "grad_norm": 2.4154458045959473,
      "learning_rate": 9.471768556308914e-06,
      "loss": 1.0119,
      "step": 4287
    },
    {
      "epoch": 0.34651205074849994,
      "grad_norm": 2.570214033126831,
      "learning_rate": 9.471475782222261e-06,
      "loss": 0.9532,
      "step": 4288
    },
    {
      "epoch": 0.3465928604618275,
      "grad_norm": 2.7691421508789062,
      "learning_rate": 9.471182931549992e-06,
      "loss": 1.0611,
      "step": 4289
    },
    {
      "epoch": 0.34667367017515505,
      "grad_norm": 2.6109275817871094,
      "learning_rate": 9.470890004297122e-06,
      "loss": 1.0118,
      "step": 4290
    },
    {
      "epoch": 0.34675447988848257,
      "grad_norm": 2.7187349796295166,
      "learning_rate": 9.47059700046867e-06,
      "loss": 0.9936,
      "step": 4291
    },
    {
      "epoch": 0.34683528960181015,
      "grad_norm": 3.0581231117248535,
      "learning_rate": 9.470303920069655e-06,
      "loss": 1.0772,
      "step": 4292
    },
    {
      "epoch": 0.34691609931513767,
      "grad_norm": 2.838987350463867,
      "learning_rate": 9.470010763105096e-06,
      "loss": 1.0029,
      "step": 4293
    },
    {
      "epoch": 0.3469969090284652,
      "grad_norm": 2.7598118782043457,
      "learning_rate": 9.469717529580013e-06,
      "loss": 0.8415,
      "step": 4294
    },
    {
      "epoch": 0.3470777187417928,
      "grad_norm": 2.6195731163024902,
      "learning_rate": 9.469424219499429e-06,
      "loss": 0.9175,
      "step": 4295
    },
    {
      "epoch": 0.3471585284551203,
      "grad_norm": 3.4818670749664307,
      "learning_rate": 9.469130832868369e-06,
      "loss": 0.9711,
      "step": 4296
    },
    {
      "epoch": 0.3472393381684478,
      "grad_norm": 2.504044771194458,
      "learning_rate": 9.468837369691854e-06,
      "loss": 1.0015,
      "step": 4297
    },
    {
      "epoch": 0.3473201478817754,
      "grad_norm": 3.460052251815796,
      "learning_rate": 9.468543829974917e-06,
      "loss": 0.9326,
      "step": 4298
    },
    {
      "epoch": 0.3474009575951029,
      "grad_norm": 2.8199450969696045,
      "learning_rate": 9.468250213722578e-06,
      "loss": 0.9231,
      "step": 4299
    },
    {
      "epoch": 0.3474817673084305,
      "grad_norm": 2.3318564891815186,
      "learning_rate": 9.467956520939872e-06,
      "loss": 0.9933,
      "step": 4300
    },
    {
      "epoch": 0.347562577021758,
      "grad_norm": 2.505892276763916,
      "learning_rate": 9.467662751631827e-06,
      "loss": 0.9749,
      "step": 4301
    },
    {
      "epoch": 0.34764338673508555,
      "grad_norm": 2.5387418270111084,
      "learning_rate": 9.467368905803474e-06,
      "loss": 1.0967,
      "step": 4302
    },
    {
      "epoch": 0.3477241964484131,
      "grad_norm": 3.123910665512085,
      "learning_rate": 9.467074983459845e-06,
      "loss": 0.9769,
      "step": 4303
    },
    {
      "epoch": 0.34780500616174065,
      "grad_norm": 2.621479034423828,
      "learning_rate": 9.466780984605978e-06,
      "loss": 0.9603,
      "step": 4304
    },
    {
      "epoch": 0.34788581587506817,
      "grad_norm": 2.7805023193359375,
      "learning_rate": 9.466486909246904e-06,
      "loss": 0.9352,
      "step": 4305
    },
    {
      "epoch": 0.34796662558839575,
      "grad_norm": 2.935696601867676,
      "learning_rate": 9.466192757387665e-06,
      "loss": 0.9463,
      "step": 4306
    },
    {
      "epoch": 0.3480474353017233,
      "grad_norm": 3.6778948307037354,
      "learning_rate": 9.465898529033292e-06,
      "loss": 0.9605,
      "step": 4307
    },
    {
      "epoch": 0.3481282450150508,
      "grad_norm": 2.4616496562957764,
      "learning_rate": 9.46560422418883e-06,
      "loss": 1.0901,
      "step": 4308
    },
    {
      "epoch": 0.3482090547283784,
      "grad_norm": 3.0213122367858887,
      "learning_rate": 9.46530984285932e-06,
      "loss": 1.0312,
      "step": 4309
    },
    {
      "epoch": 0.3482898644417059,
      "grad_norm": 2.616098165512085,
      "learning_rate": 9.4650153850498e-06,
      "loss": 1.048,
      "step": 4310
    },
    {
      "epoch": 0.3483706741550334,
      "grad_norm": 2.6346166133880615,
      "learning_rate": 9.464720850765317e-06,
      "loss": 1.0417,
      "step": 4311
    },
    {
      "epoch": 0.348451483868361,
      "grad_norm": 3.287877321243286,
      "learning_rate": 9.464426240010912e-06,
      "loss": 0.9386,
      "step": 4312
    },
    {
      "epoch": 0.3485322935816885,
      "grad_norm": 2.364901304244995,
      "learning_rate": 9.464131552791634e-06,
      "loss": 0.9209,
      "step": 4313
    },
    {
      "epoch": 0.34861310329501605,
      "grad_norm": 2.60255765914917,
      "learning_rate": 9.46383678911253e-06,
      "loss": 0.9576,
      "step": 4314
    },
    {
      "epoch": 0.3486939130083436,
      "grad_norm": 2.7048897743225098,
      "learning_rate": 9.46354194897865e-06,
      "loss": 0.9808,
      "step": 4315
    },
    {
      "epoch": 0.34877472272167115,
      "grad_norm": 2.939545154571533,
      "learning_rate": 9.463247032395039e-06,
      "loss": 0.9164,
      "step": 4316
    },
    {
      "epoch": 0.34885553243499867,
      "grad_norm": 2.8341784477233887,
      "learning_rate": 9.462952039366752e-06,
      "loss": 0.9076,
      "step": 4317
    },
    {
      "epoch": 0.34893634214832625,
      "grad_norm": 2.6567914485931396,
      "learning_rate": 9.46265696989884e-06,
      "loss": 0.9411,
      "step": 4318
    },
    {
      "epoch": 0.3490171518616538,
      "grad_norm": 2.7168097496032715,
      "learning_rate": 9.46236182399636e-06,
      "loss": 0.9037,
      "step": 4319
    },
    {
      "epoch": 0.3490979615749813,
      "grad_norm": 3.1529085636138916,
      "learning_rate": 9.46206660166436e-06,
      "loss": 0.9303,
      "step": 4320
    },
    {
      "epoch": 0.3491787712883089,
      "grad_norm": 3.0032570362091064,
      "learning_rate": 9.461771302907907e-06,
      "loss": 1.1699,
      "step": 4321
    },
    {
      "epoch": 0.3492595810016364,
      "grad_norm": 3.1368443965911865,
      "learning_rate": 9.461475927732049e-06,
      "loss": 0.9446,
      "step": 4322
    },
    {
      "epoch": 0.3493403907149639,
      "grad_norm": 2.594129800796509,
      "learning_rate": 9.461180476141848e-06,
      "loss": 1.0802,
      "step": 4323
    },
    {
      "epoch": 0.3494212004282915,
      "grad_norm": 2.818218946456909,
      "learning_rate": 9.460884948142368e-06,
      "loss": 1.0315,
      "step": 4324
    },
    {
      "epoch": 0.349502010141619,
      "grad_norm": 2.2989888191223145,
      "learning_rate": 9.460589343738669e-06,
      "loss": 1.0713,
      "step": 4325
    },
    {
      "epoch": 0.34958281985494655,
      "grad_norm": 2.951605796813965,
      "learning_rate": 9.46029366293581e-06,
      "loss": 1.0521,
      "step": 4326
    },
    {
      "epoch": 0.3496636295682741,
      "grad_norm": 2.722334146499634,
      "learning_rate": 9.45999790573886e-06,
      "loss": 1.039,
      "step": 4327
    },
    {
      "epoch": 0.34974443928160165,
      "grad_norm": 2.9131970405578613,
      "learning_rate": 9.459702072152883e-06,
      "loss": 1.0311,
      "step": 4328
    },
    {
      "epoch": 0.34982524899492917,
      "grad_norm": 2.7194559574127197,
      "learning_rate": 9.459406162182944e-06,
      "loss": 0.9912,
      "step": 4329
    },
    {
      "epoch": 0.34990605870825675,
      "grad_norm": 2.382632255554199,
      "learning_rate": 9.459110175834114e-06,
      "loss": 0.8718,
      "step": 4330
    },
    {
      "epoch": 0.3499868684215843,
      "grad_norm": 2.467538356781006,
      "learning_rate": 9.45881411311146e-06,
      "loss": 1.0031,
      "step": 4331
    },
    {
      "epoch": 0.3500676781349118,
      "grad_norm": 2.790538787841797,
      "learning_rate": 9.458517974020058e-06,
      "loss": 0.9404,
      "step": 4332
    },
    {
      "epoch": 0.3501484878482394,
      "grad_norm": 3.0771515369415283,
      "learning_rate": 9.458221758564973e-06,
      "loss": 1.0456,
      "step": 4333
    },
    {
      "epoch": 0.3502292975615669,
      "grad_norm": 2.63175630569458,
      "learning_rate": 9.457925466751285e-06,
      "loss": 0.9279,
      "step": 4334
    },
    {
      "epoch": 0.3503101072748944,
      "grad_norm": 2.870114803314209,
      "learning_rate": 9.457629098584064e-06,
      "loss": 0.9564,
      "step": 4335
    },
    {
      "epoch": 0.350390916988222,
      "grad_norm": 2.5363028049468994,
      "learning_rate": 9.457332654068389e-06,
      "loss": 1.0778,
      "step": 4336
    },
    {
      "epoch": 0.3504717267015495,
      "grad_norm": 2.3083391189575195,
      "learning_rate": 9.457036133209334e-06,
      "loss": 0.9804,
      "step": 4337
    },
    {
      "epoch": 0.35055253641487705,
      "grad_norm": 2.6185781955718994,
      "learning_rate": 9.456739536011982e-06,
      "loss": 0.9889,
      "step": 4338
    },
    {
      "epoch": 0.3506333461282046,
      "grad_norm": 3.330862045288086,
      "learning_rate": 9.45644286248141e-06,
      "loss": 0.9267,
      "step": 4339
    },
    {
      "epoch": 0.35071415584153215,
      "grad_norm": 2.8113012313842773,
      "learning_rate": 9.456146112622702e-06,
      "loss": 1.0139,
      "step": 4340
    },
    {
      "epoch": 0.3507949655548597,
      "grad_norm": 2.7219343185424805,
      "learning_rate": 9.455849286440936e-06,
      "loss": 0.8668,
      "step": 4341
    },
    {
      "epoch": 0.35087577526818725,
      "grad_norm": 3.1652705669403076,
      "learning_rate": 9.4555523839412e-06,
      "loss": 0.9704,
      "step": 4342
    },
    {
      "epoch": 0.3509565849815148,
      "grad_norm": 2.5455594062805176,
      "learning_rate": 9.455255405128579e-06,
      "loss": 0.9233,
      "step": 4343
    },
    {
      "epoch": 0.3510373946948423,
      "grad_norm": 2.779808282852173,
      "learning_rate": 9.454958350008156e-06,
      "loss": 1.086,
      "step": 4344
    },
    {
      "epoch": 0.3511182044081699,
      "grad_norm": 2.620594024658203,
      "learning_rate": 9.454661218585024e-06,
      "loss": 0.9271,
      "step": 4345
    },
    {
      "epoch": 0.3511990141214974,
      "grad_norm": 2.7019081115722656,
      "learning_rate": 9.454364010864267e-06,
      "loss": 0.9576,
      "step": 4346
    },
    {
      "epoch": 0.3512798238348249,
      "grad_norm": 3.306880474090576,
      "learning_rate": 9.45406672685098e-06,
      "loss": 0.9455,
      "step": 4347
    },
    {
      "epoch": 0.3513606335481525,
      "grad_norm": 2.6243791580200195,
      "learning_rate": 9.453769366550251e-06,
      "loss": 0.9925,
      "step": 4348
    },
    {
      "epoch": 0.35144144326148,
      "grad_norm": 3.0641255378723145,
      "learning_rate": 9.453471929967177e-06,
      "loss": 1.0466,
      "step": 4349
    },
    {
      "epoch": 0.35152225297480755,
      "grad_norm": 3.116964817047119,
      "learning_rate": 9.45317441710685e-06,
      "loss": 0.9735,
      "step": 4350
    },
    {
      "epoch": 0.3516030626881351,
      "grad_norm": 2.6487467288970947,
      "learning_rate": 9.452876827974364e-06,
      "loss": 0.9188,
      "step": 4351
    },
    {
      "epoch": 0.35168387240146265,
      "grad_norm": 3.2928214073181152,
      "learning_rate": 9.452579162574817e-06,
      "loss": 0.9692,
      "step": 4352
    },
    {
      "epoch": 0.3517646821147902,
      "grad_norm": 2.889279842376709,
      "learning_rate": 9.45228142091331e-06,
      "loss": 1.0466,
      "step": 4353
    },
    {
      "epoch": 0.35184549182811775,
      "grad_norm": 2.446531057357788,
      "learning_rate": 9.451983602994941e-06,
      "loss": 1.088,
      "step": 4354
    },
    {
      "epoch": 0.3519263015414453,
      "grad_norm": 2.785648822784424,
      "learning_rate": 9.45168570882481e-06,
      "loss": 0.8018,
      "step": 4355
    },
    {
      "epoch": 0.3520071112547728,
      "grad_norm": 2.1402995586395264,
      "learning_rate": 9.45138773840802e-06,
      "loss": 1.0743,
      "step": 4356
    },
    {
      "epoch": 0.3520879209681004,
      "grad_norm": 2.4489831924438477,
      "learning_rate": 9.451089691749673e-06,
      "loss": 0.9196,
      "step": 4357
    },
    {
      "epoch": 0.3521687306814279,
      "grad_norm": 3.200695514678955,
      "learning_rate": 9.450791568854876e-06,
      "loss": 1.0073,
      "step": 4358
    },
    {
      "epoch": 0.3522495403947554,
      "grad_norm": 3.0508291721343994,
      "learning_rate": 9.450493369728734e-06,
      "loss": 0.9789,
      "step": 4359
    },
    {
      "epoch": 0.352330350108083,
      "grad_norm": 2.3306634426116943,
      "learning_rate": 9.450195094376356e-06,
      "loss": 0.9437,
      "step": 4360
    },
    {
      "epoch": 0.3524111598214105,
      "grad_norm": 2.770146369934082,
      "learning_rate": 9.44989674280285e-06,
      "loss": 0.9436,
      "step": 4361
    },
    {
      "epoch": 0.3524919695347381,
      "grad_norm": 2.736323595046997,
      "learning_rate": 9.449598315013321e-06,
      "loss": 1.0182,
      "step": 4362
    },
    {
      "epoch": 0.3525727792480656,
      "grad_norm": 2.631716012954712,
      "learning_rate": 9.44929981101289e-06,
      "loss": 0.9786,
      "step": 4363
    },
    {
      "epoch": 0.35265358896139315,
      "grad_norm": 3.045142412185669,
      "learning_rate": 9.449001230806663e-06,
      "loss": 1.0003,
      "step": 4364
    },
    {
      "epoch": 0.35273439867472073,
      "grad_norm": 2.553821325302124,
      "learning_rate": 9.448702574399752e-06,
      "loss": 0.9904,
      "step": 4365
    },
    {
      "epoch": 0.35281520838804825,
      "grad_norm": 2.679896831512451,
      "learning_rate": 9.44840384179728e-06,
      "loss": 1.0476,
      "step": 4366
    },
    {
      "epoch": 0.3528960181013758,
      "grad_norm": 3.0261433124542236,
      "learning_rate": 9.448105033004358e-06,
      "loss": 0.9751,
      "step": 4367
    },
    {
      "epoch": 0.35297682781470335,
      "grad_norm": 2.6874983310699463,
      "learning_rate": 9.447806148026103e-06,
      "loss": 0.9185,
      "step": 4368
    },
    {
      "epoch": 0.3530576375280309,
      "grad_norm": 2.4449284076690674,
      "learning_rate": 9.44750718686764e-06,
      "loss": 1.0648,
      "step": 4369
    },
    {
      "epoch": 0.3531384472413584,
      "grad_norm": 2.977109670639038,
      "learning_rate": 9.447208149534084e-06,
      "loss": 1.0043,
      "step": 4370
    },
    {
      "epoch": 0.353219256954686,
      "grad_norm": 2.5412518978118896,
      "learning_rate": 9.446909036030558e-06,
      "loss": 0.9824,
      "step": 4371
    },
    {
      "epoch": 0.3533000666680135,
      "grad_norm": 2.513836145401001,
      "learning_rate": 9.446609846362187e-06,
      "loss": 1.1681,
      "step": 4372
    },
    {
      "epoch": 0.353380876381341,
      "grad_norm": 3.205918550491333,
      "learning_rate": 9.446310580534094e-06,
      "loss": 0.935,
      "step": 4373
    },
    {
      "epoch": 0.3534616860946686,
      "grad_norm": 2.459001302719116,
      "learning_rate": 9.446011238551404e-06,
      "loss": 1.0462,
      "step": 4374
    },
    {
      "epoch": 0.3535424958079961,
      "grad_norm": 2.2630016803741455,
      "learning_rate": 9.445711820419245e-06,
      "loss": 0.9954,
      "step": 4375
    },
    {
      "epoch": 0.35362330552132365,
      "grad_norm": 2.9816484451293945,
      "learning_rate": 9.445412326142747e-06,
      "loss": 1.0124,
      "step": 4376
    },
    {
      "epoch": 0.35370411523465123,
      "grad_norm": 2.439490556716919,
      "learning_rate": 9.445112755727036e-06,
      "loss": 1.0331,
      "step": 4377
    },
    {
      "epoch": 0.35378492494797875,
      "grad_norm": 2.5915751457214355,
      "learning_rate": 9.444813109177245e-06,
      "loss": 0.973,
      "step": 4378
    },
    {
      "epoch": 0.3538657346613063,
      "grad_norm": 2.94569730758667,
      "learning_rate": 9.444513386498504e-06,
      "loss": 0.9762,
      "step": 4379
    },
    {
      "epoch": 0.35394654437463385,
      "grad_norm": 2.8117334842681885,
      "learning_rate": 9.444213587695949e-06,
      "loss": 0.9397,
      "step": 4380
    },
    {
      "epoch": 0.3540273540879614,
      "grad_norm": 2.5008766651153564,
      "learning_rate": 9.443913712774717e-06,
      "loss": 1.052,
      "step": 4381
    },
    {
      "epoch": 0.3541081638012889,
      "grad_norm": 2.7154204845428467,
      "learning_rate": 9.443613761739939e-06,
      "loss": 0.9143,
      "step": 4382
    },
    {
      "epoch": 0.3541889735146165,
      "grad_norm": 3.495941162109375,
      "learning_rate": 9.443313734596756e-06,
      "loss": 1.0291,
      "step": 4383
    },
    {
      "epoch": 0.354269783227944,
      "grad_norm": 2.525376796722412,
      "learning_rate": 9.443013631350303e-06,
      "loss": 1.0136,
      "step": 4384
    },
    {
      "epoch": 0.3543505929412715,
      "grad_norm": 3.2318382263183594,
      "learning_rate": 9.442713452005728e-06,
      "loss": 0.9272,
      "step": 4385
    },
    {
      "epoch": 0.3544314026545991,
      "grad_norm": 2.618738889694214,
      "learning_rate": 9.442413196568161e-06,
      "loss": 0.9367,
      "step": 4386
    },
    {
      "epoch": 0.3545122123679266,
      "grad_norm": 3.1385159492492676,
      "learning_rate": 9.442112865042753e-06,
      "loss": 1.0642,
      "step": 4387
    },
    {
      "epoch": 0.35459302208125415,
      "grad_norm": 2.933108329772949,
      "learning_rate": 9.441812457434647e-06,
      "loss": 0.9823,
      "step": 4388
    },
    {
      "epoch": 0.35467383179458173,
      "grad_norm": 2.355842351913452,
      "learning_rate": 9.441511973748987e-06,
      "loss": 0.9344,
      "step": 4389
    },
    {
      "epoch": 0.35475464150790925,
      "grad_norm": 2.866903305053711,
      "learning_rate": 9.441211413990918e-06,
      "loss": 0.9452,
      "step": 4390
    },
    {
      "epoch": 0.3548354512212368,
      "grad_norm": 2.942039728164673,
      "learning_rate": 9.440910778165588e-06,
      "loss": 1.0805,
      "step": 4391
    },
    {
      "epoch": 0.35491626093456435,
      "grad_norm": 3.402123212814331,
      "learning_rate": 9.44061006627815e-06,
      "loss": 1.1258,
      "step": 4392
    },
    {
      "epoch": 0.3549970706478919,
      "grad_norm": 2.882495641708374,
      "learning_rate": 9.44030927833375e-06,
      "loss": 0.9454,
      "step": 4393
    },
    {
      "epoch": 0.3550778803612194,
      "grad_norm": 2.3867785930633545,
      "learning_rate": 9.440008414337543e-06,
      "loss": 0.9132,
      "step": 4394
    },
    {
      "epoch": 0.355158690074547,
      "grad_norm": 2.6597611904144287,
      "learning_rate": 9.439707474294679e-06,
      "loss": 0.8589,
      "step": 4395
    },
    {
      "epoch": 0.3552394997878745,
      "grad_norm": 3.0939173698425293,
      "learning_rate": 9.439406458210316e-06,
      "loss": 0.9932,
      "step": 4396
    },
    {
      "epoch": 0.355320309501202,
      "grad_norm": 2.8719091415405273,
      "learning_rate": 9.439105366089606e-06,
      "loss": 0.9933,
      "step": 4397
    },
    {
      "epoch": 0.3554011192145296,
      "grad_norm": 2.700834035873413,
      "learning_rate": 9.43880419793771e-06,
      "loss": 1.1072,
      "step": 4398
    },
    {
      "epoch": 0.3554819289278571,
      "grad_norm": 3.0498509407043457,
      "learning_rate": 9.438502953759783e-06,
      "loss": 0.9062,
      "step": 4399
    },
    {
      "epoch": 0.35556273864118465,
      "grad_norm": 2.5735740661621094,
      "learning_rate": 9.438201633560983e-06,
      "loss": 0.999,
      "step": 4400
    },
    {
      "epoch": 0.35564354835451223,
      "grad_norm": 2.5071537494659424,
      "learning_rate": 9.437900237346479e-06,
      "loss": 0.9051,
      "step": 4401
    },
    {
      "epoch": 0.35572435806783975,
      "grad_norm": 2.5384161472320557,
      "learning_rate": 9.437598765121423e-06,
      "loss": 1.1434,
      "step": 4402
    },
    {
      "epoch": 0.3558051677811673,
      "grad_norm": 2.3582170009613037,
      "learning_rate": 9.437297216890986e-06,
      "loss": 0.9889,
      "step": 4403
    },
    {
      "epoch": 0.35588597749449485,
      "grad_norm": 2.8007912635803223,
      "learning_rate": 9.436995592660328e-06,
      "loss": 0.8858,
      "step": 4404
    },
    {
      "epoch": 0.3559667872078224,
      "grad_norm": 2.6563291549682617,
      "learning_rate": 9.436693892434618e-06,
      "loss": 0.9193,
      "step": 4405
    },
    {
      "epoch": 0.3560475969211499,
      "grad_norm": 2.6463418006896973,
      "learning_rate": 9.436392116219024e-06,
      "loss": 1.0564,
      "step": 4406
    },
    {
      "epoch": 0.3561284066344775,
      "grad_norm": 3.1819608211517334,
      "learning_rate": 9.436090264018711e-06,
      "loss": 0.9396,
      "step": 4407
    },
    {
      "epoch": 0.356209216347805,
      "grad_norm": 2.84258770942688,
      "learning_rate": 9.435788335838852e-06,
      "loss": 1.0478,
      "step": 4408
    },
    {
      "epoch": 0.3562900260611325,
      "grad_norm": 3.5766983032226562,
      "learning_rate": 9.435486331684617e-06,
      "loss": 1.224,
      "step": 4409
    },
    {
      "epoch": 0.3563708357744601,
      "grad_norm": 2.5731217861175537,
      "learning_rate": 9.435184251561181e-06,
      "loss": 0.9477,
      "step": 4410
    },
    {
      "epoch": 0.3564516454877876,
      "grad_norm": 2.6633787155151367,
      "learning_rate": 9.434882095473714e-06,
      "loss": 0.9986,
      "step": 4411
    },
    {
      "epoch": 0.35653245520111515,
      "grad_norm": 2.7791190147399902,
      "learning_rate": 9.434579863427396e-06,
      "loss": 0.9405,
      "step": 4412
    },
    {
      "epoch": 0.35661326491444273,
      "grad_norm": 2.5482404232025146,
      "learning_rate": 9.434277555427397e-06,
      "loss": 0.896,
      "step": 4413
    },
    {
      "epoch": 0.35669407462777025,
      "grad_norm": 2.590073585510254,
      "learning_rate": 9.433975171478901e-06,
      "loss": 0.7863,
      "step": 4414
    },
    {
      "epoch": 0.3567748843410978,
      "grad_norm": 2.9536354541778564,
      "learning_rate": 9.433672711587086e-06,
      "loss": 1.1034,
      "step": 4415
    },
    {
      "epoch": 0.35685569405442535,
      "grad_norm": 2.2706964015960693,
      "learning_rate": 9.433370175757129e-06,
      "loss": 1.0198,
      "step": 4416
    },
    {
      "epoch": 0.3569365037677529,
      "grad_norm": 2.293123245239258,
      "learning_rate": 9.433067563994214e-06,
      "loss": 0.9699,
      "step": 4417
    },
    {
      "epoch": 0.3570173134810804,
      "grad_norm": 2.8644261360168457,
      "learning_rate": 9.432764876303523e-06,
      "loss": 0.9393,
      "step": 4418
    },
    {
      "epoch": 0.357098123194408,
      "grad_norm": 2.9875433444976807,
      "learning_rate": 9.432462112690242e-06,
      "loss": 0.9763,
      "step": 4419
    },
    {
      "epoch": 0.3571789329077355,
      "grad_norm": 2.342803478240967,
      "learning_rate": 9.432159273159556e-06,
      "loss": 0.9794,
      "step": 4420
    },
    {
      "epoch": 0.357259742621063,
      "grad_norm": 2.90524959564209,
      "learning_rate": 9.43185635771665e-06,
      "loss": 0.8631,
      "step": 4421
    },
    {
      "epoch": 0.3573405523343906,
      "grad_norm": 2.5036685466766357,
      "learning_rate": 9.431553366366716e-06,
      "loss": 0.9362,
      "step": 4422
    },
    {
      "epoch": 0.3574213620477181,
      "grad_norm": 3.7688751220703125,
      "learning_rate": 9.43125029911494e-06,
      "loss": 1.0821,
      "step": 4423
    },
    {
      "epoch": 0.35750217176104565,
      "grad_norm": 2.795793056488037,
      "learning_rate": 9.430947155966514e-06,
      "loss": 1.0631,
      "step": 4424
    },
    {
      "epoch": 0.35758298147437323,
      "grad_norm": 2.783707857131958,
      "learning_rate": 9.430643936926631e-06,
      "loss": 0.9612,
      "step": 4425
    },
    {
      "epoch": 0.35766379118770075,
      "grad_norm": 3.1343765258789062,
      "learning_rate": 9.430340642000484e-06,
      "loss": 0.8432,
      "step": 4426
    },
    {
      "epoch": 0.35774460090102833,
      "grad_norm": 2.3770313262939453,
      "learning_rate": 9.430037271193267e-06,
      "loss": 1.0099,
      "step": 4427
    },
    {
      "epoch": 0.35782541061435585,
      "grad_norm": 2.72056245803833,
      "learning_rate": 9.429733824510176e-06,
      "loss": 0.9592,
      "step": 4428
    },
    {
      "epoch": 0.3579062203276834,
      "grad_norm": 2.875227451324463,
      "learning_rate": 9.42943030195641e-06,
      "loss": 0.969,
      "step": 4429
    },
    {
      "epoch": 0.35798703004101096,
      "grad_norm": 2.701589584350586,
      "learning_rate": 9.429126703537165e-06,
      "loss": 0.9963,
      "step": 4430
    },
    {
      "epoch": 0.3580678397543385,
      "grad_norm": 2.879420518875122,
      "learning_rate": 9.428823029257643e-06,
      "loss": 1.0575,
      "step": 4431
    },
    {
      "epoch": 0.358148649467666,
      "grad_norm": 2.6203296184539795,
      "learning_rate": 9.428519279123045e-06,
      "loss": 1.008,
      "step": 4432
    },
    {
      "epoch": 0.3582294591809936,
      "grad_norm": 2.648378849029541,
      "learning_rate": 9.428215453138571e-06,
      "loss": 1.0841,
      "step": 4433
    },
    {
      "epoch": 0.3583102688943211,
      "grad_norm": 2.7141456604003906,
      "learning_rate": 9.42791155130943e-06,
      "loss": 1.0489,
      "step": 4434
    },
    {
      "epoch": 0.3583910786076486,
      "grad_norm": 2.451988935470581,
      "learning_rate": 9.42760757364082e-06,
      "loss": 0.978,
      "step": 4435
    },
    {
      "epoch": 0.3584718883209762,
      "grad_norm": 3.4383907318115234,
      "learning_rate": 9.427303520137954e-06,
      "loss": 0.9989,
      "step": 4436
    },
    {
      "epoch": 0.35855269803430373,
      "grad_norm": 3.1784305572509766,
      "learning_rate": 9.426999390806037e-06,
      "loss": 1.0339,
      "step": 4437
    },
    {
      "epoch": 0.35863350774763125,
      "grad_norm": 2.6991403102874756,
      "learning_rate": 9.426695185650276e-06,
      "loss": 0.9995,
      "step": 4438
    },
    {
      "epoch": 0.35871431746095883,
      "grad_norm": 2.82651424407959,
      "learning_rate": 9.426390904675887e-06,
      "loss": 0.9791,
      "step": 4439
    },
    {
      "epoch": 0.35879512717428635,
      "grad_norm": 2.4426968097686768,
      "learning_rate": 9.426086547888074e-06,
      "loss": 0.9886,
      "step": 4440
    },
    {
      "epoch": 0.3588759368876139,
      "grad_norm": 2.8267159461975098,
      "learning_rate": 9.425782115292054e-06,
      "loss": 1.0125,
      "step": 4441
    },
    {
      "epoch": 0.35895674660094146,
      "grad_norm": 2.4940249919891357,
      "learning_rate": 9.425477606893044e-06,
      "loss": 1.0049,
      "step": 4442
    },
    {
      "epoch": 0.359037556314269,
      "grad_norm": 3.048516035079956,
      "learning_rate": 9.425173022696255e-06,
      "loss": 0.8767,
      "step": 4443
    },
    {
      "epoch": 0.3591183660275965,
      "grad_norm": 2.5132670402526855,
      "learning_rate": 9.424868362706905e-06,
      "loss": 0.8913,
      "step": 4444
    },
    {
      "epoch": 0.3591991757409241,
      "grad_norm": 2.2747271060943604,
      "learning_rate": 9.424563626930213e-06,
      "loss": 1.0056,
      "step": 4445
    },
    {
      "epoch": 0.3592799854542516,
      "grad_norm": 2.6004068851470947,
      "learning_rate": 9.424258815371397e-06,
      "loss": 0.9968,
      "step": 4446
    },
    {
      "epoch": 0.35936079516757913,
      "grad_norm": 2.3039145469665527,
      "learning_rate": 9.423953928035678e-06,
      "loss": 0.8961,
      "step": 4447
    },
    {
      "epoch": 0.3594416048809067,
      "grad_norm": 2.539527654647827,
      "learning_rate": 9.423648964928279e-06,
      "loss": 0.9564,
      "step": 4448
    },
    {
      "epoch": 0.35952241459423423,
      "grad_norm": 2.3412609100341797,
      "learning_rate": 9.423343926054422e-06,
      "loss": 1.0276,
      "step": 4449
    },
    {
      "epoch": 0.35960322430756175,
      "grad_norm": 2.665891170501709,
      "learning_rate": 9.423038811419334e-06,
      "loss": 0.9202,
      "step": 4450
    },
    {
      "epoch": 0.35968403402088933,
      "grad_norm": 2.923896312713623,
      "learning_rate": 9.422733621028239e-06,
      "loss": 0.9785,
      "step": 4451
    },
    {
      "epoch": 0.35976484373421685,
      "grad_norm": 2.6648497581481934,
      "learning_rate": 9.422428354886364e-06,
      "loss": 1.0934,
      "step": 4452
    },
    {
      "epoch": 0.3598456534475444,
      "grad_norm": 2.780925989151001,
      "learning_rate": 9.422123012998936e-06,
      "loss": 0.8999,
      "step": 4453
    },
    {
      "epoch": 0.35992646316087196,
      "grad_norm": 2.895123243331909,
      "learning_rate": 9.421817595371188e-06,
      "loss": 0.9942,
      "step": 4454
    },
    {
      "epoch": 0.3600072728741995,
      "grad_norm": 2.727386951446533,
      "learning_rate": 9.42151210200835e-06,
      "loss": 1.0444,
      "step": 4455
    },
    {
      "epoch": 0.360088082587527,
      "grad_norm": 2.6367311477661133,
      "learning_rate": 9.421206532915655e-06,
      "loss": 0.9106,
      "step": 4456
    },
    {
      "epoch": 0.3601688923008546,
      "grad_norm": 2.836529016494751,
      "learning_rate": 9.420900888098334e-06,
      "loss": 0.9969,
      "step": 4457
    },
    {
      "epoch": 0.3602497020141821,
      "grad_norm": 2.6362826824188232,
      "learning_rate": 9.420595167561624e-06,
      "loss": 0.9138,
      "step": 4458
    },
    {
      "epoch": 0.36033051172750963,
      "grad_norm": 2.9698569774627686,
      "learning_rate": 9.420289371310762e-06,
      "loss": 1.0344,
      "step": 4459
    },
    {
      "epoch": 0.3604113214408372,
      "grad_norm": 2.665956735610962,
      "learning_rate": 9.419983499350982e-06,
      "loss": 0.9462,
      "step": 4460
    },
    {
      "epoch": 0.36049213115416473,
      "grad_norm": 2.300938129425049,
      "learning_rate": 9.419677551687528e-06,
      "loss": 0.9502,
      "step": 4461
    },
    {
      "epoch": 0.36057294086749225,
      "grad_norm": 2.4960052967071533,
      "learning_rate": 9.419371528325638e-06,
      "loss": 0.9395,
      "step": 4462
    },
    {
      "epoch": 0.36065375058081983,
      "grad_norm": 2.689157485961914,
      "learning_rate": 9.41906542927055e-06,
      "loss": 0.9432,
      "step": 4463
    },
    {
      "epoch": 0.36073456029414736,
      "grad_norm": 2.563105821609497,
      "learning_rate": 9.418759254527512e-06,
      "loss": 1.0336,
      "step": 4464
    },
    {
      "epoch": 0.3608153700074749,
      "grad_norm": 2.5278499126434326,
      "learning_rate": 9.418453004101763e-06,
      "loss": 1.1189,
      "step": 4465
    },
    {
      "epoch": 0.36089617972080246,
      "grad_norm": 2.824989080429077,
      "learning_rate": 9.418146677998554e-06,
      "loss": 1.0469,
      "step": 4466
    },
    {
      "epoch": 0.36097698943413,
      "grad_norm": 2.2645199298858643,
      "learning_rate": 9.417840276223127e-06,
      "loss": 1.0926,
      "step": 4467
    },
    {
      "epoch": 0.3610577991474575,
      "grad_norm": 3.170844554901123,
      "learning_rate": 9.417533798780732e-06,
      "loss": 0.956,
      "step": 4468
    },
    {
      "epoch": 0.3611386088607851,
      "grad_norm": 2.6823577880859375,
      "learning_rate": 9.417227245676618e-06,
      "loss": 1.0137,
      "step": 4469
    },
    {
      "epoch": 0.3612194185741126,
      "grad_norm": 2.3987109661102295,
      "learning_rate": 9.416920616916035e-06,
      "loss": 0.9045,
      "step": 4470
    },
    {
      "epoch": 0.36130022828744013,
      "grad_norm": 2.764575719833374,
      "learning_rate": 9.416613912504235e-06,
      "loss": 1.0281,
      "step": 4471
    },
    {
      "epoch": 0.3613810380007677,
      "grad_norm": 2.5729925632476807,
      "learning_rate": 9.416307132446474e-06,
      "loss": 0.9318,
      "step": 4472
    },
    {
      "epoch": 0.36146184771409523,
      "grad_norm": 2.812591075897217,
      "learning_rate": 9.416000276748e-06,
      "loss": 1.0329,
      "step": 4473
    },
    {
      "epoch": 0.36154265742742275,
      "grad_norm": 2.8140928745269775,
      "learning_rate": 9.415693345414072e-06,
      "loss": 1.0318,
      "step": 4474
    },
    {
      "epoch": 0.36162346714075033,
      "grad_norm": 2.4964237213134766,
      "learning_rate": 9.41538633844995e-06,
      "loss": 0.9337,
      "step": 4475
    },
    {
      "epoch": 0.36170427685407786,
      "grad_norm": 2.8262624740600586,
      "learning_rate": 9.415079255860888e-06,
      "loss": 0.9315,
      "step": 4476
    },
    {
      "epoch": 0.3617850865674054,
      "grad_norm": 2.7971997261047363,
      "learning_rate": 9.414772097652148e-06,
      "loss": 0.9649,
      "step": 4477
    },
    {
      "epoch": 0.36186589628073296,
      "grad_norm": 2.835785388946533,
      "learning_rate": 9.41446486382899e-06,
      "loss": 1.0176,
      "step": 4478
    },
    {
      "epoch": 0.3619467059940605,
      "grad_norm": 2.54367995262146,
      "learning_rate": 9.414157554396677e-06,
      "loss": 0.9148,
      "step": 4479
    },
    {
      "epoch": 0.362027515707388,
      "grad_norm": 3.091097116470337,
      "learning_rate": 9.41385016936047e-06,
      "loss": 0.9647,
      "step": 4480
    },
    {
      "epoch": 0.3621083254207156,
      "grad_norm": 2.828221321105957,
      "learning_rate": 9.413542708725635e-06,
      "loss": 0.9779,
      "step": 4481
    },
    {
      "epoch": 0.3621891351340431,
      "grad_norm": 2.577510118484497,
      "learning_rate": 9.413235172497442e-06,
      "loss": 0.9412,
      "step": 4482
    },
    {
      "epoch": 0.36226994484737063,
      "grad_norm": 2.7678253650665283,
      "learning_rate": 9.412927560681154e-06,
      "loss": 1.0555,
      "step": 4483
    },
    {
      "epoch": 0.3623507545606982,
      "grad_norm": 2.514561176300049,
      "learning_rate": 9.412619873282038e-06,
      "loss": 1.0715,
      "step": 4484
    },
    {
      "epoch": 0.36243156427402573,
      "grad_norm": 2.812615394592285,
      "learning_rate": 9.412312110305368e-06,
      "loss": 0.8406,
      "step": 4485
    },
    {
      "epoch": 0.36251237398735325,
      "grad_norm": 2.455737590789795,
      "learning_rate": 9.412004271756415e-06,
      "loss": 0.8643,
      "step": 4486
    },
    {
      "epoch": 0.36259318370068083,
      "grad_norm": 2.6049814224243164,
      "learning_rate": 9.411696357640447e-06,
      "loss": 0.914,
      "step": 4487
    },
    {
      "epoch": 0.36267399341400836,
      "grad_norm": 3.2906692028045654,
      "learning_rate": 9.411388367962744e-06,
      "loss": 1.0165,
      "step": 4488
    },
    {
      "epoch": 0.3627548031273359,
      "grad_norm": 2.6364407539367676,
      "learning_rate": 9.411080302728577e-06,
      "loss": 0.9579,
      "step": 4489
    },
    {
      "epoch": 0.36283561284066346,
      "grad_norm": 2.5739123821258545,
      "learning_rate": 9.410772161943224e-06,
      "loss": 0.9622,
      "step": 4490
    },
    {
      "epoch": 0.362916422553991,
      "grad_norm": 2.5797173976898193,
      "learning_rate": 9.410463945611963e-06,
      "loss": 0.8997,
      "step": 4491
    },
    {
      "epoch": 0.36299723226731856,
      "grad_norm": 2.528426170349121,
      "learning_rate": 9.410155653740071e-06,
      "loss": 0.9644,
      "step": 4492
    },
    {
      "epoch": 0.3630780419806461,
      "grad_norm": 2.6368141174316406,
      "learning_rate": 9.409847286332831e-06,
      "loss": 1.1497,
      "step": 4493
    },
    {
      "epoch": 0.3631588516939736,
      "grad_norm": 2.983921766281128,
      "learning_rate": 9.409538843395523e-06,
      "loss": 0.9232,
      "step": 4494
    },
    {
      "epoch": 0.3632396614073012,
      "grad_norm": 2.966362476348877,
      "learning_rate": 9.40923032493343e-06,
      "loss": 0.7913,
      "step": 4495
    },
    {
      "epoch": 0.3633204711206287,
      "grad_norm": 2.6679775714874268,
      "learning_rate": 9.408921730951835e-06,
      "loss": 0.8781,
      "step": 4496
    },
    {
      "epoch": 0.36340128083395623,
      "grad_norm": 2.9152774810791016,
      "learning_rate": 9.408613061456027e-06,
      "loss": 1.0007,
      "step": 4497
    },
    {
      "epoch": 0.3634820905472838,
      "grad_norm": 3.973459243774414,
      "learning_rate": 9.40830431645129e-06,
      "loss": 0.9284,
      "step": 4498
    },
    {
      "epoch": 0.36356290026061133,
      "grad_norm": 2.953460931777954,
      "learning_rate": 9.40799549594291e-06,
      "loss": 1.046,
      "step": 4499
    },
    {
      "epoch": 0.36364370997393886,
      "grad_norm": 2.8357183933258057,
      "learning_rate": 9.407686599936182e-06,
      "loss": 0.9262,
      "step": 4500
    },
    {
      "epoch": 0.36372451968726643,
      "grad_norm": 2.6355485916137695,
      "learning_rate": 9.407377628436394e-06,
      "loss": 0.9544,
      "step": 4501
    },
    {
      "epoch": 0.36380532940059396,
      "grad_norm": 2.738980770111084,
      "learning_rate": 9.407068581448836e-06,
      "loss": 1.0368,
      "step": 4502
    },
    {
      "epoch": 0.3638861391139215,
      "grad_norm": 2.5949649810791016,
      "learning_rate": 9.406759458978803e-06,
      "loss": 0.9707,
      "step": 4503
    },
    {
      "epoch": 0.36396694882724906,
      "grad_norm": 2.6837685108184814,
      "learning_rate": 9.406450261031589e-06,
      "loss": 1.0089,
      "step": 4504
    },
    {
      "epoch": 0.3640477585405766,
      "grad_norm": 2.6699249744415283,
      "learning_rate": 9.40614098761249e-06,
      "loss": 1.0019,
      "step": 4505
    },
    {
      "epoch": 0.3641285682539041,
      "grad_norm": 3.1839182376861572,
      "learning_rate": 9.405831638726804e-06,
      "loss": 0.9515,
      "step": 4506
    },
    {
      "epoch": 0.3642093779672317,
      "grad_norm": 2.8554553985595703,
      "learning_rate": 9.405522214379828e-06,
      "loss": 0.9493,
      "step": 4507
    },
    {
      "epoch": 0.3642901876805592,
      "grad_norm": 2.585563898086548,
      "learning_rate": 9.405212714576863e-06,
      "loss": 1.0017,
      "step": 4508
    },
    {
      "epoch": 0.36437099739388673,
      "grad_norm": 2.9683499336242676,
      "learning_rate": 9.40490313932321e-06,
      "loss": 1.1998,
      "step": 4509
    },
    {
      "epoch": 0.3644518071072143,
      "grad_norm": 3.171738624572754,
      "learning_rate": 9.404593488624168e-06,
      "loss": 1.0149,
      "step": 4510
    },
    {
      "epoch": 0.36453261682054183,
      "grad_norm": 2.4399123191833496,
      "learning_rate": 9.404283762485045e-06,
      "loss": 1.0968,
      "step": 4511
    },
    {
      "epoch": 0.36461342653386936,
      "grad_norm": 2.8575239181518555,
      "learning_rate": 9.403973960911143e-06,
      "loss": 1.0486,
      "step": 4512
    },
    {
      "epoch": 0.36469423624719693,
      "grad_norm": 3.0597331523895264,
      "learning_rate": 9.40366408390777e-06,
      "loss": 0.9539,
      "step": 4513
    },
    {
      "epoch": 0.36477504596052446,
      "grad_norm": 2.4745028018951416,
      "learning_rate": 9.403354131480233e-06,
      "loss": 0.8326,
      "step": 4514
    },
    {
      "epoch": 0.364855855673852,
      "grad_norm": 2.9149515628814697,
      "learning_rate": 9.40304410363384e-06,
      "loss": 0.983,
      "step": 4515
    },
    {
      "epoch": 0.36493666538717956,
      "grad_norm": 2.8048317432403564,
      "learning_rate": 9.402734000373903e-06,
      "loss": 0.9749,
      "step": 4516
    },
    {
      "epoch": 0.3650174751005071,
      "grad_norm": 2.515558958053589,
      "learning_rate": 9.402423821705728e-06,
      "loss": 0.9942,
      "step": 4517
    },
    {
      "epoch": 0.3650982848138346,
      "grad_norm": 2.7846484184265137,
      "learning_rate": 9.402113567634633e-06,
      "loss": 0.9612,
      "step": 4518
    },
    {
      "epoch": 0.3651790945271622,
      "grad_norm": 2.86734676361084,
      "learning_rate": 9.401803238165933e-06,
      "loss": 1.0884,
      "step": 4519
    },
    {
      "epoch": 0.3652599042404897,
      "grad_norm": 2.6100146770477295,
      "learning_rate": 9.401492833304936e-06,
      "loss": 0.8563,
      "step": 4520
    },
    {
      "epoch": 0.36534071395381723,
      "grad_norm": 2.7019171714782715,
      "learning_rate": 9.401182353056966e-06,
      "loss": 0.9184,
      "step": 4521
    },
    {
      "epoch": 0.3654215236671448,
      "grad_norm": 3.166868209838867,
      "learning_rate": 9.400871797427338e-06,
      "loss": 0.992,
      "step": 4522
    },
    {
      "epoch": 0.36550233338047233,
      "grad_norm": 2.637887477874756,
      "learning_rate": 9.400561166421369e-06,
      "loss": 0.984,
      "step": 4523
    },
    {
      "epoch": 0.36558314309379986,
      "grad_norm": 2.8214378356933594,
      "learning_rate": 9.400250460044382e-06,
      "loss": 1.0293,
      "step": 4524
    },
    {
      "epoch": 0.36566395280712743,
      "grad_norm": 3.0553970336914062,
      "learning_rate": 9.399939678301697e-06,
      "loss": 1.0487,
      "step": 4525
    },
    {
      "epoch": 0.36574476252045496,
      "grad_norm": 2.739971160888672,
      "learning_rate": 9.39962882119864e-06,
      "loss": 1.1134,
      "step": 4526
    },
    {
      "epoch": 0.3658255722337825,
      "grad_norm": 2.5887255668640137,
      "learning_rate": 9.39931788874053e-06,
      "loss": 0.9873,
      "step": 4527
    },
    {
      "epoch": 0.36590638194711006,
      "grad_norm": 2.8833765983581543,
      "learning_rate": 9.399006880932696e-06,
      "loss": 0.9344,
      "step": 4528
    },
    {
      "epoch": 0.3659871916604376,
      "grad_norm": 2.879179000854492,
      "learning_rate": 9.398695797780465e-06,
      "loss": 0.9708,
      "step": 4529
    },
    {
      "epoch": 0.3660680013737651,
      "grad_norm": 3.0080201625823975,
      "learning_rate": 9.398384639289165e-06,
      "loss": 0.9233,
      "step": 4530
    },
    {
      "epoch": 0.3661488110870927,
      "grad_norm": 2.843886137008667,
      "learning_rate": 9.398073405464123e-06,
      "loss": 1.0434,
      "step": 4531
    },
    {
      "epoch": 0.3662296208004202,
      "grad_norm": 2.586495876312256,
      "learning_rate": 9.397762096310673e-06,
      "loss": 1.1622,
      "step": 4532
    },
    {
      "epoch": 0.36631043051374773,
      "grad_norm": 2.517352819442749,
      "learning_rate": 9.397450711834145e-06,
      "loss": 1.0387,
      "step": 4533
    },
    {
      "epoch": 0.3663912402270753,
      "grad_norm": 2.4357497692108154,
      "learning_rate": 9.397139252039873e-06,
      "loss": 1.0196,
      "step": 4534
    },
    {
      "epoch": 0.36647204994040283,
      "grad_norm": 2.8926069736480713,
      "learning_rate": 9.396827716933191e-06,
      "loss": 1.017,
      "step": 4535
    },
    {
      "epoch": 0.36655285965373036,
      "grad_norm": 2.7496743202209473,
      "learning_rate": 9.396516106519436e-06,
      "loss": 1.0057,
      "step": 4536
    },
    {
      "epoch": 0.36663366936705793,
      "grad_norm": 2.624067783355713,
      "learning_rate": 9.396204420803943e-06,
      "loss": 1.075,
      "step": 4537
    },
    {
      "epoch": 0.36671447908038546,
      "grad_norm": 2.9119105339050293,
      "learning_rate": 9.395892659792053e-06,
      "loss": 1.0004,
      "step": 4538
    },
    {
      "epoch": 0.366795288793713,
      "grad_norm": 2.5376782417297363,
      "learning_rate": 9.395580823489103e-06,
      "loss": 1.0453,
      "step": 4539
    },
    {
      "epoch": 0.36687609850704056,
      "grad_norm": 2.7258594036102295,
      "learning_rate": 9.395268911900437e-06,
      "loss": 1.0709,
      "step": 4540
    },
    {
      "epoch": 0.3669569082203681,
      "grad_norm": 2.718148946762085,
      "learning_rate": 9.394956925031394e-06,
      "loss": 0.943,
      "step": 4541
    },
    {
      "epoch": 0.3670377179336956,
      "grad_norm": 2.7300665378570557,
      "learning_rate": 9.39464486288732e-06,
      "loss": 0.9242,
      "step": 4542
    },
    {
      "epoch": 0.3671185276470232,
      "grad_norm": 2.771728277206421,
      "learning_rate": 9.39433272547356e-06,
      "loss": 0.998,
      "step": 4543
    },
    {
      "epoch": 0.3671993373603507,
      "grad_norm": 2.4261975288391113,
      "learning_rate": 9.394020512795459e-06,
      "loss": 0.8755,
      "step": 4544
    },
    {
      "epoch": 0.36728014707367823,
      "grad_norm": 2.893710136413574,
      "learning_rate": 9.393708224858365e-06,
      "loss": 0.9709,
      "step": 4545
    },
    {
      "epoch": 0.3673609567870058,
      "grad_norm": 2.779921293258667,
      "learning_rate": 9.393395861667625e-06,
      "loss": 0.9192,
      "step": 4546
    },
    {
      "epoch": 0.36744176650033333,
      "grad_norm": 2.4235339164733887,
      "learning_rate": 9.393083423228591e-06,
      "loss": 1.103,
      "step": 4547
    },
    {
      "epoch": 0.36752257621366086,
      "grad_norm": 3.1893932819366455,
      "learning_rate": 9.392770909546615e-06,
      "loss": 1.0178,
      "step": 4548
    },
    {
      "epoch": 0.36760338592698844,
      "grad_norm": 2.8173398971557617,
      "learning_rate": 9.392458320627046e-06,
      "loss": 0.9715,
      "step": 4549
    },
    {
      "epoch": 0.36768419564031596,
      "grad_norm": 2.9895713329315186,
      "learning_rate": 9.392145656475245e-06,
      "loss": 1.1446,
      "step": 4550
    },
    {
      "epoch": 0.3677650053536435,
      "grad_norm": 2.971907138824463,
      "learning_rate": 9.39183291709656e-06,
      "loss": 0.9728,
      "step": 4551
    },
    {
      "epoch": 0.36784581506697106,
      "grad_norm": 2.853642225265503,
      "learning_rate": 9.39152010249635e-06,
      "loss": 1.0509,
      "step": 4552
    },
    {
      "epoch": 0.3679266247802986,
      "grad_norm": 3.2043793201446533,
      "learning_rate": 9.391207212679971e-06,
      "loss": 0.9442,
      "step": 4553
    },
    {
      "epoch": 0.3680074344936261,
      "grad_norm": 2.5248141288757324,
      "learning_rate": 9.390894247652786e-06,
      "loss": 1.0295,
      "step": 4554
    },
    {
      "epoch": 0.3680882442069537,
      "grad_norm": 2.902416229248047,
      "learning_rate": 9.390581207420153e-06,
      "loss": 0.9146,
      "step": 4555
    },
    {
      "epoch": 0.3681690539202812,
      "grad_norm": 2.6413331031799316,
      "learning_rate": 9.390268091987434e-06,
      "loss": 0.9069,
      "step": 4556
    },
    {
      "epoch": 0.3682498636336088,
      "grad_norm": 2.557710886001587,
      "learning_rate": 9.38995490135999e-06,
      "loss": 0.9081,
      "step": 4557
    },
    {
      "epoch": 0.3683306733469363,
      "grad_norm": 2.6280970573425293,
      "learning_rate": 9.389641635543189e-06,
      "loss": 0.9767,
      "step": 4558
    },
    {
      "epoch": 0.36841148306026383,
      "grad_norm": 2.41762113571167,
      "learning_rate": 9.389328294542392e-06,
      "loss": 1.0447,
      "step": 4559
    },
    {
      "epoch": 0.3684922927735914,
      "grad_norm": 2.6446521282196045,
      "learning_rate": 9.38901487836297e-06,
      "loss": 1.044,
      "step": 4560
    },
    {
      "epoch": 0.36857310248691894,
      "grad_norm": 3.1756978034973145,
      "learning_rate": 9.38870138701029e-06,
      "loss": 0.8964,
      "step": 4561
    },
    {
      "epoch": 0.36865391220024646,
      "grad_norm": 2.9383111000061035,
      "learning_rate": 9.388387820489719e-06,
      "loss": 0.9009,
      "step": 4562
    },
    {
      "epoch": 0.36873472191357404,
      "grad_norm": 2.624556064605713,
      "learning_rate": 9.38807417880663e-06,
      "loss": 0.9675,
      "step": 4563
    },
    {
      "epoch": 0.36881553162690156,
      "grad_norm": 2.8243956565856934,
      "learning_rate": 9.387760461966395e-06,
      "loss": 0.9539,
      "step": 4564
    },
    {
      "epoch": 0.3688963413402291,
      "grad_norm": 2.3740501403808594,
      "learning_rate": 9.387446669974384e-06,
      "loss": 0.9306,
      "step": 4565
    },
    {
      "epoch": 0.36897715105355666,
      "grad_norm": 2.642275094985962,
      "learning_rate": 9.387132802835977e-06,
      "loss": 0.9699,
      "step": 4566
    },
    {
      "epoch": 0.3690579607668842,
      "grad_norm": 3.44392728805542,
      "learning_rate": 9.386818860556545e-06,
      "loss": 0.9687,
      "step": 4567
    },
    {
      "epoch": 0.3691387704802117,
      "grad_norm": 2.5872433185577393,
      "learning_rate": 9.386504843141466e-06,
      "loss": 0.8814,
      "step": 4568
    },
    {
      "epoch": 0.3692195801935393,
      "grad_norm": 2.684056520462036,
      "learning_rate": 9.38619075059612e-06,
      "loss": 0.9708,
      "step": 4569
    },
    {
      "epoch": 0.3693003899068668,
      "grad_norm": 2.7787370681762695,
      "learning_rate": 9.385876582925886e-06,
      "loss": 0.8949,
      "step": 4570
    },
    {
      "epoch": 0.36938119962019433,
      "grad_norm": 2.595571756362915,
      "learning_rate": 9.385562340136144e-06,
      "loss": 1.0831,
      "step": 4571
    },
    {
      "epoch": 0.3694620093335219,
      "grad_norm": 2.7705674171447754,
      "learning_rate": 9.385248022232278e-06,
      "loss": 1.0678,
      "step": 4572
    },
    {
      "epoch": 0.36954281904684944,
      "grad_norm": 2.8611176013946533,
      "learning_rate": 9.384933629219669e-06,
      "loss": 0.9156,
      "step": 4573
    },
    {
      "epoch": 0.36962362876017696,
      "grad_norm": 2.8933324813842773,
      "learning_rate": 9.384619161103703e-06,
      "loss": 1.2177,
      "step": 4574
    },
    {
      "epoch": 0.36970443847350454,
      "grad_norm": 2.9562392234802246,
      "learning_rate": 9.384304617889768e-06,
      "loss": 0.9406,
      "step": 4575
    },
    {
      "epoch": 0.36978524818683206,
      "grad_norm": 2.4091672897338867,
      "learning_rate": 9.38398999958325e-06,
      "loss": 1.052,
      "step": 4576
    },
    {
      "epoch": 0.3698660579001596,
      "grad_norm": 2.5455174446105957,
      "learning_rate": 9.383675306189535e-06,
      "loss": 1.003,
      "step": 4577
    },
    {
      "epoch": 0.36994686761348716,
      "grad_norm": 2.325300931930542,
      "learning_rate": 9.383360537714018e-06,
      "loss": 1.0479,
      "step": 4578
    },
    {
      "epoch": 0.3700276773268147,
      "grad_norm": 2.7085652351379395,
      "learning_rate": 9.383045694162085e-06,
      "loss": 0.9682,
      "step": 4579
    },
    {
      "epoch": 0.3701084870401422,
      "grad_norm": 3.875587224960327,
      "learning_rate": 9.382730775539133e-06,
      "loss": 0.9901,
      "step": 4580
    },
    {
      "epoch": 0.3701892967534698,
      "grad_norm": 2.634938955307007,
      "learning_rate": 9.382415781850553e-06,
      "loss": 0.9529,
      "step": 4581
    },
    {
      "epoch": 0.3702701064667973,
      "grad_norm": 2.4124062061309814,
      "learning_rate": 9.38210071310174e-06,
      "loss": 0.9837,
      "step": 4582
    },
    {
      "epoch": 0.37035091618012483,
      "grad_norm": 2.9303414821624756,
      "learning_rate": 9.381785569298092e-06,
      "loss": 0.9598,
      "step": 4583
    },
    {
      "epoch": 0.3704317258934524,
      "grad_norm": 3.144341468811035,
      "learning_rate": 9.381470350445006e-06,
      "loss": 1.0144,
      "step": 4584
    },
    {
      "epoch": 0.37051253560677994,
      "grad_norm": 2.8793258666992188,
      "learning_rate": 9.38115505654788e-06,
      "loss": 0.908,
      "step": 4585
    },
    {
      "epoch": 0.37059334532010746,
      "grad_norm": 2.8264477252960205,
      "learning_rate": 9.380839687612116e-06,
      "loss": 0.9622,
      "step": 4586
    },
    {
      "epoch": 0.37067415503343504,
      "grad_norm": 2.8271288871765137,
      "learning_rate": 9.380524243643115e-06,
      "loss": 1.0603,
      "step": 4587
    },
    {
      "epoch": 0.37075496474676256,
      "grad_norm": 2.6553587913513184,
      "learning_rate": 9.380208724646279e-06,
      "loss": 0.9144,
      "step": 4588
    },
    {
      "epoch": 0.3708357744600901,
      "grad_norm": 2.6913859844207764,
      "learning_rate": 9.379893130627011e-06,
      "loss": 1.0163,
      "step": 4589
    },
    {
      "epoch": 0.37091658417341766,
      "grad_norm": 3.0093512535095215,
      "learning_rate": 9.37957746159072e-06,
      "loss": 0.9999,
      "step": 4590
    },
    {
      "epoch": 0.3709973938867452,
      "grad_norm": 3.0305333137512207,
      "learning_rate": 9.37926171754281e-06,
      "loss": 0.9112,
      "step": 4591
    },
    {
      "epoch": 0.3710782036000727,
      "grad_norm": 2.818683624267578,
      "learning_rate": 9.37894589848869e-06,
      "loss": 0.9946,
      "step": 4592
    },
    {
      "epoch": 0.3711590133134003,
      "grad_norm": 2.397392749786377,
      "learning_rate": 9.378630004433766e-06,
      "loss": 1.0275,
      "step": 4593
    },
    {
      "epoch": 0.3712398230267278,
      "grad_norm": 2.944654703140259,
      "learning_rate": 9.378314035383454e-06,
      "loss": 0.973,
      "step": 4594
    },
    {
      "epoch": 0.37132063274005533,
      "grad_norm": 2.380833148956299,
      "learning_rate": 9.377997991343163e-06,
      "loss": 0.9979,
      "step": 4595
    },
    {
      "epoch": 0.3714014424533829,
      "grad_norm": 2.939668655395508,
      "learning_rate": 9.377681872318303e-06,
      "loss": 0.9139,
      "step": 4596
    },
    {
      "epoch": 0.37148225216671044,
      "grad_norm": 3.133352279663086,
      "learning_rate": 9.377365678314293e-06,
      "loss": 0.9466,
      "step": 4597
    },
    {
      "epoch": 0.37156306188003796,
      "grad_norm": 2.810354471206665,
      "learning_rate": 9.377049409336547e-06,
      "loss": 0.9573,
      "step": 4598
    },
    {
      "epoch": 0.37164387159336554,
      "grad_norm": 2.642275810241699,
      "learning_rate": 9.376733065390483e-06,
      "loss": 1.0413,
      "step": 4599
    },
    {
      "epoch": 0.37172468130669306,
      "grad_norm": 3.47493314743042,
      "learning_rate": 9.376416646481516e-06,
      "loss": 0.8108,
      "step": 4600
    },
    {
      "epoch": 0.3718054910200206,
      "grad_norm": 2.652791976928711,
      "learning_rate": 9.37610015261507e-06,
      "loss": 0.9259,
      "step": 4601
    },
    {
      "epoch": 0.37188630073334816,
      "grad_norm": 3.052654981613159,
      "learning_rate": 9.375783583796562e-06,
      "loss": 0.9279,
      "step": 4602
    },
    {
      "epoch": 0.3719671104466757,
      "grad_norm": 2.550945997238159,
      "learning_rate": 9.375466940031416e-06,
      "loss": 0.9951,
      "step": 4603
    },
    {
      "epoch": 0.3720479201600032,
      "grad_norm": 2.9834840297698975,
      "learning_rate": 9.375150221325053e-06,
      "loss": 0.9975,
      "step": 4604
    },
    {
      "epoch": 0.3721287298733308,
      "grad_norm": 2.732625961303711,
      "learning_rate": 9.3748334276829e-06,
      "loss": 1.0384,
      "step": 4605
    },
    {
      "epoch": 0.3722095395866583,
      "grad_norm": 2.907871961593628,
      "learning_rate": 9.374516559110386e-06,
      "loss": 1.1045,
      "step": 4606
    },
    {
      "epoch": 0.37229034929998583,
      "grad_norm": 2.7010860443115234,
      "learning_rate": 9.37419961561293e-06,
      "loss": 0.9371,
      "step": 4607
    },
    {
      "epoch": 0.3723711590133134,
      "grad_norm": 2.9858970642089844,
      "learning_rate": 9.373882597195969e-06,
      "loss": 0.9703,
      "step": 4608
    },
    {
      "epoch": 0.37245196872664094,
      "grad_norm": 2.719649076461792,
      "learning_rate": 9.373565503864925e-06,
      "loss": 0.9681,
      "step": 4609
    },
    {
      "epoch": 0.37253277843996846,
      "grad_norm": 2.775792121887207,
      "learning_rate": 9.373248335625237e-06,
      "loss": 1.1137,
      "step": 4610
    },
    {
      "epoch": 0.37261358815329604,
      "grad_norm": 3.075058698654175,
      "learning_rate": 9.37293109248233e-06,
      "loss": 1.0127,
      "step": 4611
    },
    {
      "epoch": 0.37269439786662356,
      "grad_norm": 3.0514657497406006,
      "learning_rate": 9.37261377444164e-06,
      "loss": 1.0136,
      "step": 4612
    },
    {
      "epoch": 0.3727752075799511,
      "grad_norm": 2.7548375129699707,
      "learning_rate": 9.372296381508606e-06,
      "loss": 0.9435,
      "step": 4613
    },
    {
      "epoch": 0.37285601729327866,
      "grad_norm": 2.669076919555664,
      "learning_rate": 9.37197891368866e-06,
      "loss": 0.7948,
      "step": 4614
    },
    {
      "epoch": 0.3729368270066062,
      "grad_norm": 3.102717876434326,
      "learning_rate": 9.371661370987238e-06,
      "loss": 0.9228,
      "step": 4615
    },
    {
      "epoch": 0.3730176367199337,
      "grad_norm": 2.722560167312622,
      "learning_rate": 9.371343753409783e-06,
      "loss": 0.9516,
      "step": 4616
    },
    {
      "epoch": 0.3730984464332613,
      "grad_norm": 2.6848738193511963,
      "learning_rate": 9.371026060961732e-06,
      "loss": 1.0751,
      "step": 4617
    },
    {
      "epoch": 0.3731792561465888,
      "grad_norm": 2.5699706077575684,
      "learning_rate": 9.370708293648528e-06,
      "loss": 1.1555,
      "step": 4618
    },
    {
      "epoch": 0.37326006585991633,
      "grad_norm": 2.504427671432495,
      "learning_rate": 9.370390451475614e-06,
      "loss": 0.8811,
      "step": 4619
    },
    {
      "epoch": 0.3733408755732439,
      "grad_norm": 2.705014228820801,
      "learning_rate": 9.370072534448432e-06,
      "loss": 1.0209,
      "step": 4620
    },
    {
      "epoch": 0.37342168528657144,
      "grad_norm": 2.680250644683838,
      "learning_rate": 9.369754542572427e-06,
      "loss": 0.9363,
      "step": 4621
    },
    {
      "epoch": 0.373502494999899,
      "grad_norm": 2.696732759475708,
      "learning_rate": 9.369436475853048e-06,
      "loss": 0.9178,
      "step": 4622
    },
    {
      "epoch": 0.37358330471322654,
      "grad_norm": 2.756539821624756,
      "learning_rate": 9.36911833429574e-06,
      "loss": 0.9903,
      "step": 4623
    },
    {
      "epoch": 0.37366411442655406,
      "grad_norm": 2.860239267349243,
      "learning_rate": 9.368800117905954e-06,
      "loss": 1.0033,
      "step": 4624
    },
    {
      "epoch": 0.37374492413988164,
      "grad_norm": 2.2197537422180176,
      "learning_rate": 9.368481826689138e-06,
      "loss": 0.8965,
      "step": 4625
    },
    {
      "epoch": 0.37382573385320916,
      "grad_norm": 2.4059908390045166,
      "learning_rate": 9.368163460650747e-06,
      "loss": 0.9164,
      "step": 4626
    },
    {
      "epoch": 0.3739065435665367,
      "grad_norm": 2.822862386703491,
      "learning_rate": 9.36784501979623e-06,
      "loss": 0.9616,
      "step": 4627
    },
    {
      "epoch": 0.37398735327986427,
      "grad_norm": 2.7833025455474854,
      "learning_rate": 9.367526504131043e-06,
      "loss": 1.0955,
      "step": 4628
    },
    {
      "epoch": 0.3740681629931918,
      "grad_norm": 2.7099392414093018,
      "learning_rate": 9.367207913660643e-06,
      "loss": 0.9954,
      "step": 4629
    },
    {
      "epoch": 0.3741489727065193,
      "grad_norm": 2.4273574352264404,
      "learning_rate": 9.366889248390486e-06,
      "loss": 0.9942,
      "step": 4630
    },
    {
      "epoch": 0.3742297824198469,
      "grad_norm": 2.7751688957214355,
      "learning_rate": 9.366570508326026e-06,
      "loss": 1.0044,
      "step": 4631
    },
    {
      "epoch": 0.3743105921331744,
      "grad_norm": 2.7348666191101074,
      "learning_rate": 9.366251693472728e-06,
      "loss": 0.9199,
      "step": 4632
    },
    {
      "epoch": 0.37439140184650194,
      "grad_norm": 2.558270215988159,
      "learning_rate": 9.365932803836046e-06,
      "loss": 1.117,
      "step": 4633
    },
    {
      "epoch": 0.3744722115598295,
      "grad_norm": 2.744283676147461,
      "learning_rate": 9.36561383942145e-06,
      "loss": 0.9424,
      "step": 4634
    },
    {
      "epoch": 0.37455302127315704,
      "grad_norm": 2.5696918964385986,
      "learning_rate": 9.365294800234397e-06,
      "loss": 1.0267,
      "step": 4635
    },
    {
      "epoch": 0.37463383098648456,
      "grad_norm": 2.5888736248016357,
      "learning_rate": 9.364975686280352e-06,
      "loss": 1.038,
      "step": 4636
    },
    {
      "epoch": 0.37471464069981214,
      "grad_norm": 2.646268129348755,
      "learning_rate": 9.364656497564782e-06,
      "loss": 1.0989,
      "step": 4637
    },
    {
      "epoch": 0.37479545041313966,
      "grad_norm": 2.6140968799591064,
      "learning_rate": 9.364337234093155e-06,
      "loss": 0.9057,
      "step": 4638
    },
    {
      "epoch": 0.3748762601264672,
      "grad_norm": 2.7762510776519775,
      "learning_rate": 9.364017895870938e-06,
      "loss": 0.9697,
      "step": 4639
    },
    {
      "epoch": 0.37495706983979477,
      "grad_norm": 2.621048927307129,
      "learning_rate": 9.363698482903598e-06,
      "loss": 0.8835,
      "step": 4640
    },
    {
      "epoch": 0.3750378795531223,
      "grad_norm": 2.8719687461853027,
      "learning_rate": 9.36337899519661e-06,
      "loss": 0.9315,
      "step": 4641
    },
    {
      "epoch": 0.3751186892664498,
      "grad_norm": 2.5091278553009033,
      "learning_rate": 9.363059432755443e-06,
      "loss": 0.988,
      "step": 4642
    },
    {
      "epoch": 0.3751994989797774,
      "grad_norm": 2.474213123321533,
      "learning_rate": 9.362739795585573e-06,
      "loss": 0.9952,
      "step": 4643
    },
    {
      "epoch": 0.3752803086931049,
      "grad_norm": 3.510223865509033,
      "learning_rate": 9.362420083692474e-06,
      "loss": 0.9259,
      "step": 4644
    },
    {
      "epoch": 0.37536111840643244,
      "grad_norm": 2.7654919624328613,
      "learning_rate": 9.36210029708162e-06,
      "loss": 1.0706,
      "step": 4645
    },
    {
      "epoch": 0.37544192811976,
      "grad_norm": 2.7717068195343018,
      "learning_rate": 9.361780435758488e-06,
      "loss": 0.8909,
      "step": 4646
    },
    {
      "epoch": 0.37552273783308754,
      "grad_norm": 2.8184854984283447,
      "learning_rate": 9.361460499728558e-06,
      "loss": 0.9658,
      "step": 4647
    },
    {
      "epoch": 0.37560354754641506,
      "grad_norm": 2.530444860458374,
      "learning_rate": 9.361140488997311e-06,
      "loss": 0.7889,
      "step": 4648
    },
    {
      "epoch": 0.37568435725974264,
      "grad_norm": 2.4376254081726074,
      "learning_rate": 9.360820403570225e-06,
      "loss": 1.007,
      "step": 4649
    },
    {
      "epoch": 0.37576516697307016,
      "grad_norm": 2.770446538925171,
      "learning_rate": 9.360500243452785e-06,
      "loss": 1.0088,
      "step": 4650
    },
    {
      "epoch": 0.3758459766863977,
      "grad_norm": 2.4954068660736084,
      "learning_rate": 9.360180008650472e-06,
      "loss": 1.1051,
      "step": 4651
    },
    {
      "epoch": 0.37592678639972527,
      "grad_norm": 2.4725492000579834,
      "learning_rate": 9.359859699168773e-06,
      "loss": 1.0519,
      "step": 4652
    },
    {
      "epoch": 0.3760075961130528,
      "grad_norm": 2.686007499694824,
      "learning_rate": 9.359539315013173e-06,
      "loss": 1.0196,
      "step": 4653
    },
    {
      "epoch": 0.3760884058263803,
      "grad_norm": 2.589627504348755,
      "learning_rate": 9.359218856189161e-06,
      "loss": 0.945,
      "step": 4654
    },
    {
      "epoch": 0.3761692155397079,
      "grad_norm": 3.184671401977539,
      "learning_rate": 9.358898322702222e-06,
      "loss": 0.9778,
      "step": 4655
    },
    {
      "epoch": 0.3762500252530354,
      "grad_norm": 2.420375108718872,
      "learning_rate": 9.358577714557849e-06,
      "loss": 0.9061,
      "step": 4656
    },
    {
      "epoch": 0.37633083496636294,
      "grad_norm": 2.8433828353881836,
      "learning_rate": 9.358257031761532e-06,
      "loss": 1.1095,
      "step": 4657
    },
    {
      "epoch": 0.3764116446796905,
      "grad_norm": 3.1400113105773926,
      "learning_rate": 9.357936274318766e-06,
      "loss": 1.0581,
      "step": 4658
    },
    {
      "epoch": 0.37649245439301804,
      "grad_norm": 2.6416494846343994,
      "learning_rate": 9.357615442235042e-06,
      "loss": 0.984,
      "step": 4659
    },
    {
      "epoch": 0.37657326410634556,
      "grad_norm": 2.5612497329711914,
      "learning_rate": 9.357294535515857e-06,
      "loss": 0.9338,
      "step": 4660
    },
    {
      "epoch": 0.37665407381967314,
      "grad_norm": 3.2765259742736816,
      "learning_rate": 9.356973554166704e-06,
      "loss": 1.1653,
      "step": 4661
    },
    {
      "epoch": 0.37673488353300066,
      "grad_norm": 2.5079116821289062,
      "learning_rate": 9.356652498193085e-06,
      "loss": 1.0834,
      "step": 4662
    },
    {
      "epoch": 0.3768156932463282,
      "grad_norm": 2.9052252769470215,
      "learning_rate": 9.356331367600497e-06,
      "loss": 0.8942,
      "step": 4663
    },
    {
      "epoch": 0.37689650295965577,
      "grad_norm": 2.5440428256988525,
      "learning_rate": 9.35601016239444e-06,
      "loss": 0.9333,
      "step": 4664
    },
    {
      "epoch": 0.3769773126729833,
      "grad_norm": 2.6236190795898438,
      "learning_rate": 9.355688882580414e-06,
      "loss": 0.9764,
      "step": 4665
    },
    {
      "epoch": 0.3770581223863108,
      "grad_norm": 2.6265389919281006,
      "learning_rate": 9.355367528163925e-06,
      "loss": 0.8844,
      "step": 4666
    },
    {
      "epoch": 0.3771389320996384,
      "grad_norm": 2.5625059604644775,
      "learning_rate": 9.355046099150475e-06,
      "loss": 0.9659,
      "step": 4667
    },
    {
      "epoch": 0.3772197418129659,
      "grad_norm": 2.58971905708313,
      "learning_rate": 9.35472459554557e-06,
      "loss": 0.9519,
      "step": 4668
    },
    {
      "epoch": 0.37730055152629344,
      "grad_norm": 2.975165843963623,
      "learning_rate": 9.354403017354715e-06,
      "loss": 0.9435,
      "step": 4669
    },
    {
      "epoch": 0.377381361239621,
      "grad_norm": 2.81882905960083,
      "learning_rate": 9.35408136458342e-06,
      "loss": 0.9213,
      "step": 4670
    },
    {
      "epoch": 0.37746217095294854,
      "grad_norm": 2.43086314201355,
      "learning_rate": 9.353759637237192e-06,
      "loss": 1.0836,
      "step": 4671
    },
    {
      "epoch": 0.37754298066627606,
      "grad_norm": 2.7550718784332275,
      "learning_rate": 9.353437835321543e-06,
      "loss": 0.9825,
      "step": 4672
    },
    {
      "epoch": 0.37762379037960364,
      "grad_norm": 2.6080548763275146,
      "learning_rate": 9.353115958841987e-06,
      "loss": 0.9459,
      "step": 4673
    },
    {
      "epoch": 0.37770460009293116,
      "grad_norm": 2.7612552642822266,
      "learning_rate": 9.352794007804033e-06,
      "loss": 0.9331,
      "step": 4674
    },
    {
      "epoch": 0.3777854098062587,
      "grad_norm": 2.4627575874328613,
      "learning_rate": 9.352471982213195e-06,
      "loss": 0.8379,
      "step": 4675
    },
    {
      "epoch": 0.37786621951958627,
      "grad_norm": 2.686471700668335,
      "learning_rate": 9.35214988207499e-06,
      "loss": 1.0379,
      "step": 4676
    },
    {
      "epoch": 0.3779470292329138,
      "grad_norm": 2.488630771636963,
      "learning_rate": 9.351827707394937e-06,
      "loss": 1.0181,
      "step": 4677
    },
    {
      "epoch": 0.3780278389462413,
      "grad_norm": 2.6898245811462402,
      "learning_rate": 9.351505458178551e-06,
      "loss": 1.0643,
      "step": 4678
    },
    {
      "epoch": 0.3781086486595689,
      "grad_norm": 2.7466509342193604,
      "learning_rate": 9.351183134431352e-06,
      "loss": 0.9447,
      "step": 4679
    },
    {
      "epoch": 0.3781894583728964,
      "grad_norm": 2.5569167137145996,
      "learning_rate": 9.350860736158861e-06,
      "loss": 1.016,
      "step": 4680
    },
    {
      "epoch": 0.37827026808622394,
      "grad_norm": 2.2112040519714355,
      "learning_rate": 9.3505382633666e-06,
      "loss": 0.8968,
      "step": 4681
    },
    {
      "epoch": 0.3783510777995515,
      "grad_norm": 3.138573169708252,
      "learning_rate": 9.350215716060093e-06,
      "loss": 1.0001,
      "step": 4682
    },
    {
      "epoch": 0.37843188751287904,
      "grad_norm": 2.741068124771118,
      "learning_rate": 9.349893094244863e-06,
      "loss": 1.0164,
      "step": 4683
    },
    {
      "epoch": 0.37851269722620656,
      "grad_norm": 2.4641666412353516,
      "learning_rate": 9.349570397926435e-06,
      "loss": 1.0531,
      "step": 4684
    },
    {
      "epoch": 0.37859350693953414,
      "grad_norm": 2.799994468688965,
      "learning_rate": 9.349247627110338e-06,
      "loss": 1.0897,
      "step": 4685
    },
    {
      "epoch": 0.37867431665286166,
      "grad_norm": 2.553637742996216,
      "learning_rate": 9.3489247818021e-06,
      "loss": 0.9916,
      "step": 4686
    },
    {
      "epoch": 0.37875512636618924,
      "grad_norm": 2.3900206089019775,
      "learning_rate": 9.34860186200725e-06,
      "loss": 0.9037,
      "step": 4687
    },
    {
      "epoch": 0.37883593607951677,
      "grad_norm": 2.382781505584717,
      "learning_rate": 9.348278867731317e-06,
      "loss": 0.9374,
      "step": 4688
    },
    {
      "epoch": 0.3789167457928443,
      "grad_norm": 2.3783068656921387,
      "learning_rate": 9.347955798979838e-06,
      "loss": 1.0351,
      "step": 4689
    },
    {
      "epoch": 0.37899755550617187,
      "grad_norm": 2.8407225608825684,
      "learning_rate": 9.347632655758341e-06,
      "loss": 1.0577,
      "step": 4690
    },
    {
      "epoch": 0.3790783652194994,
      "grad_norm": 3.121370553970337,
      "learning_rate": 9.347309438072365e-06,
      "loss": 0.9437,
      "step": 4691
    },
    {
      "epoch": 0.3791591749328269,
      "grad_norm": 2.5054781436920166,
      "learning_rate": 9.346986145927443e-06,
      "loss": 0.9386,
      "step": 4692
    },
    {
      "epoch": 0.3792399846461545,
      "grad_norm": 2.5887506008148193,
      "learning_rate": 9.346662779329115e-06,
      "loss": 0.9107,
      "step": 4693
    },
    {
      "epoch": 0.379320794359482,
      "grad_norm": 2.8974032402038574,
      "learning_rate": 9.346339338282915e-06,
      "loss": 0.923,
      "step": 4694
    },
    {
      "epoch": 0.37940160407280954,
      "grad_norm": 2.7863271236419678,
      "learning_rate": 9.346015822794387e-06,
      "loss": 0.9532,
      "step": 4695
    },
    {
      "epoch": 0.3794824137861371,
      "grad_norm": 2.574018716812134,
      "learning_rate": 9.34569223286907e-06,
      "loss": 0.8615,
      "step": 4696
    },
    {
      "epoch": 0.37956322349946464,
      "grad_norm": 2.5277822017669678,
      "learning_rate": 9.345368568512508e-06,
      "loss": 1.028,
      "step": 4697
    },
    {
      "epoch": 0.37964403321279216,
      "grad_norm": 3.112009048461914,
      "learning_rate": 9.345044829730243e-06,
      "loss": 0.929,
      "step": 4698
    },
    {
      "epoch": 0.37972484292611974,
      "grad_norm": 2.4879302978515625,
      "learning_rate": 9.34472101652782e-06,
      "loss": 1.0871,
      "step": 4699
    },
    {
      "epoch": 0.37980565263944727,
      "grad_norm": 2.7445104122161865,
      "learning_rate": 9.344397128910784e-06,
      "loss": 1.0077,
      "step": 4700
    },
    {
      "epoch": 0.3798864623527748,
      "grad_norm": 2.4609758853912354,
      "learning_rate": 9.344073166884686e-06,
      "loss": 1.0116,
      "step": 4701
    },
    {
      "epoch": 0.37996727206610237,
      "grad_norm": 3.099045753479004,
      "learning_rate": 9.343749130455074e-06,
      "loss": 0.9865,
      "step": 4702
    },
    {
      "epoch": 0.3800480817794299,
      "grad_norm": 2.8231124877929688,
      "learning_rate": 9.343425019627493e-06,
      "loss": 1.1225,
      "step": 4703
    },
    {
      "epoch": 0.3801288914927574,
      "grad_norm": 2.8193726539611816,
      "learning_rate": 9.3431008344075e-06,
      "loss": 0.9882,
      "step": 4704
    },
    {
      "epoch": 0.380209701206085,
      "grad_norm": 2.245363473892212,
      "learning_rate": 9.342776574800645e-06,
      "loss": 0.9592,
      "step": 4705
    },
    {
      "epoch": 0.3802905109194125,
      "grad_norm": 2.7241601943969727,
      "learning_rate": 9.342452240812481e-06,
      "loss": 0.831,
      "step": 4706
    },
    {
      "epoch": 0.38037132063274004,
      "grad_norm": 2.3399782180786133,
      "learning_rate": 9.342127832448565e-06,
      "loss": 1.0248,
      "step": 4707
    },
    {
      "epoch": 0.3804521303460676,
      "grad_norm": 2.310903310775757,
      "learning_rate": 9.341803349714453e-06,
      "loss": 1.0077,
      "step": 4708
    },
    {
      "epoch": 0.38053294005939514,
      "grad_norm": 2.90739369392395,
      "learning_rate": 9.341478792615702e-06,
      "loss": 1.059,
      "step": 4709
    },
    {
      "epoch": 0.38061374977272266,
      "grad_norm": 2.5221786499023438,
      "learning_rate": 9.341154161157868e-06,
      "loss": 0.9381,
      "step": 4710
    },
    {
      "epoch": 0.38069455948605024,
      "grad_norm": 2.8505871295928955,
      "learning_rate": 9.340829455346518e-06,
      "loss": 1.0079,
      "step": 4711
    },
    {
      "epoch": 0.38077536919937777,
      "grad_norm": 2.158698558807373,
      "learning_rate": 9.340504675187207e-06,
      "loss": 1.0045,
      "step": 4712
    },
    {
      "epoch": 0.3808561789127053,
      "grad_norm": 3.18208909034729,
      "learning_rate": 9.340179820685503e-06,
      "loss": 1.1025,
      "step": 4713
    },
    {
      "epoch": 0.38093698862603287,
      "grad_norm": 2.932091474533081,
      "learning_rate": 9.339854891846964e-06,
      "loss": 1.005,
      "step": 4714
    },
    {
      "epoch": 0.3810177983393604,
      "grad_norm": 2.5078186988830566,
      "learning_rate": 9.339529888677161e-06,
      "loss": 1.0195,
      "step": 4715
    },
    {
      "epoch": 0.3810986080526879,
      "grad_norm": 3.0139050483703613,
      "learning_rate": 9.339204811181657e-06,
      "loss": 0.8981,
      "step": 4716
    },
    {
      "epoch": 0.3811794177660155,
      "grad_norm": 2.5075485706329346,
      "learning_rate": 9.33887965936602e-06,
      "loss": 0.9846,
      "step": 4717
    },
    {
      "epoch": 0.381260227479343,
      "grad_norm": 2.7496562004089355,
      "learning_rate": 9.33855443323582e-06,
      "loss": 0.8334,
      "step": 4718
    },
    {
      "epoch": 0.38134103719267054,
      "grad_norm": 2.6395156383514404,
      "learning_rate": 9.338229132796629e-06,
      "loss": 1.0034,
      "step": 4719
    },
    {
      "epoch": 0.3814218469059981,
      "grad_norm": 2.8209757804870605,
      "learning_rate": 9.337903758054016e-06,
      "loss": 1.0862,
      "step": 4720
    },
    {
      "epoch": 0.38150265661932564,
      "grad_norm": 2.652390718460083,
      "learning_rate": 9.337578309013554e-06,
      "loss": 0.8791,
      "step": 4721
    },
    {
      "epoch": 0.38158346633265317,
      "grad_norm": 3.160367012023926,
      "learning_rate": 9.337252785680818e-06,
      "loss": 1.1129,
      "step": 4722
    },
    {
      "epoch": 0.38166427604598074,
      "grad_norm": 2.3652641773223877,
      "learning_rate": 9.336927188061385e-06,
      "loss": 1.0013,
      "step": 4723
    },
    {
      "epoch": 0.38174508575930827,
      "grad_norm": 2.477011203765869,
      "learning_rate": 9.336601516160828e-06,
      "loss": 1.0117,
      "step": 4724
    },
    {
      "epoch": 0.3818258954726358,
      "grad_norm": 3.6358301639556885,
      "learning_rate": 9.336275769984727e-06,
      "loss": 1.0446,
      "step": 4725
    },
    {
      "epoch": 0.38190670518596337,
      "grad_norm": 3.075953483581543,
      "learning_rate": 9.335949949538663e-06,
      "loss": 0.9108,
      "step": 4726
    },
    {
      "epoch": 0.3819875148992909,
      "grad_norm": 2.7538113594055176,
      "learning_rate": 9.335624054828212e-06,
      "loss": 0.9995,
      "step": 4727
    },
    {
      "epoch": 0.3820683246126184,
      "grad_norm": 2.8855652809143066,
      "learning_rate": 9.335298085858959e-06,
      "loss": 0.9995,
      "step": 4728
    },
    {
      "epoch": 0.382149134325946,
      "grad_norm": 3.38128399848938,
      "learning_rate": 9.334972042636489e-06,
      "loss": 1.0016,
      "step": 4729
    },
    {
      "epoch": 0.3822299440392735,
      "grad_norm": 2.4725615978240967,
      "learning_rate": 9.334645925166382e-06,
      "loss": 0.9524,
      "step": 4730
    },
    {
      "epoch": 0.38231075375260104,
      "grad_norm": 2.950460433959961,
      "learning_rate": 9.334319733454227e-06,
      "loss": 0.8953,
      "step": 4731
    },
    {
      "epoch": 0.3823915634659286,
      "grad_norm": 3.0027737617492676,
      "learning_rate": 9.333993467505608e-06,
      "loss": 0.9027,
      "step": 4732
    },
    {
      "epoch": 0.38247237317925614,
      "grad_norm": 2.9997451305389404,
      "learning_rate": 9.333667127326114e-06,
      "loss": 1.1046,
      "step": 4733
    },
    {
      "epoch": 0.38255318289258367,
      "grad_norm": 2.69627046585083,
      "learning_rate": 9.333340712921337e-06,
      "loss": 0.9887,
      "step": 4734
    },
    {
      "epoch": 0.38263399260591124,
      "grad_norm": 2.633528232574463,
      "learning_rate": 9.333014224296864e-06,
      "loss": 0.9717,
      "step": 4735
    },
    {
      "epoch": 0.38271480231923877,
      "grad_norm": 3.3414008617401123,
      "learning_rate": 9.33268766145829e-06,
      "loss": 0.9609,
      "step": 4736
    },
    {
      "epoch": 0.3827956120325663,
      "grad_norm": 2.7071304321289062,
      "learning_rate": 9.332361024411206e-06,
      "loss": 0.9961,
      "step": 4737
    },
    {
      "epoch": 0.38287642174589387,
      "grad_norm": 2.9471144676208496,
      "learning_rate": 9.332034313161207e-06,
      "loss": 0.9456,
      "step": 4738
    },
    {
      "epoch": 0.3829572314592214,
      "grad_norm": 2.5740861892700195,
      "learning_rate": 9.331707527713891e-06,
      "loss": 0.9751,
      "step": 4739
    },
    {
      "epoch": 0.3830380411725489,
      "grad_norm": 2.7647855281829834,
      "learning_rate": 9.331380668074852e-06,
      "loss": 0.9046,
      "step": 4740
    },
    {
      "epoch": 0.3831188508858765,
      "grad_norm": 3.026322364807129,
      "learning_rate": 9.331053734249688e-06,
      "loss": 0.904,
      "step": 4741
    },
    {
      "epoch": 0.383199660599204,
      "grad_norm": 3.0095396041870117,
      "learning_rate": 9.330726726244002e-06,
      "loss": 0.9377,
      "step": 4742
    },
    {
      "epoch": 0.38328047031253154,
      "grad_norm": 2.852733850479126,
      "learning_rate": 9.330399644063392e-06,
      "loss": 0.9513,
      "step": 4743
    },
    {
      "epoch": 0.3833612800258591,
      "grad_norm": 3.252817153930664,
      "learning_rate": 9.330072487713462e-06,
      "loss": 0.9426,
      "step": 4744
    },
    {
      "epoch": 0.38344208973918664,
      "grad_norm": 2.7273945808410645,
      "learning_rate": 9.329745257199816e-06,
      "loss": 1.0306,
      "step": 4745
    },
    {
      "epoch": 0.38352289945251417,
      "grad_norm": 2.3329997062683105,
      "learning_rate": 9.329417952528055e-06,
      "loss": 1.0341,
      "step": 4746
    },
    {
      "epoch": 0.38360370916584174,
      "grad_norm": 2.8117733001708984,
      "learning_rate": 9.329090573703787e-06,
      "loss": 0.8849,
      "step": 4747
    },
    {
      "epoch": 0.38368451887916927,
      "grad_norm": 2.821118116378784,
      "learning_rate": 9.32876312073262e-06,
      "loss": 0.9057,
      "step": 4748
    },
    {
      "epoch": 0.3837653285924968,
      "grad_norm": 2.883186101913452,
      "learning_rate": 9.328435593620162e-06,
      "loss": 1.0717,
      "step": 4749
    },
    {
      "epoch": 0.38384613830582437,
      "grad_norm": 2.41925048828125,
      "learning_rate": 9.328107992372023e-06,
      "loss": 1.0482,
      "step": 4750
    },
    {
      "epoch": 0.3839269480191519,
      "grad_norm": 2.7774710655212402,
      "learning_rate": 9.327780316993811e-06,
      "loss": 0.9736,
      "step": 4751
    },
    {
      "epoch": 0.38400775773247947,
      "grad_norm": 3.132113218307495,
      "learning_rate": 9.327452567491143e-06,
      "loss": 0.9372,
      "step": 4752
    },
    {
      "epoch": 0.384088567445807,
      "grad_norm": 2.61859130859375,
      "learning_rate": 9.327124743869631e-06,
      "loss": 1.05,
      "step": 4753
    },
    {
      "epoch": 0.3841693771591345,
      "grad_norm": 3.047009229660034,
      "learning_rate": 9.326796846134888e-06,
      "loss": 0.9907,
      "step": 4754
    },
    {
      "epoch": 0.3842501868724621,
      "grad_norm": 2.372396945953369,
      "learning_rate": 9.326468874292531e-06,
      "loss": 1.0379,
      "step": 4755
    },
    {
      "epoch": 0.3843309965857896,
      "grad_norm": 2.8072192668914795,
      "learning_rate": 9.32614082834818e-06,
      "loss": 0.9341,
      "step": 4756
    },
    {
      "epoch": 0.38441180629911714,
      "grad_norm": 2.7737817764282227,
      "learning_rate": 9.325812708307449e-06,
      "loss": 0.9666,
      "step": 4757
    },
    {
      "epoch": 0.3844926160124447,
      "grad_norm": 2.4324991703033447,
      "learning_rate": 9.32548451417596e-06,
      "loss": 0.9549,
      "step": 4758
    },
    {
      "epoch": 0.38457342572577224,
      "grad_norm": 2.6352956295013428,
      "learning_rate": 9.325156245959336e-06,
      "loss": 1.0462,
      "step": 4759
    },
    {
      "epoch": 0.38465423543909977,
      "grad_norm": 2.7103171348571777,
      "learning_rate": 9.324827903663198e-06,
      "loss": 1.0216,
      "step": 4760
    },
    {
      "epoch": 0.38473504515242735,
      "grad_norm": 2.5112156867980957,
      "learning_rate": 9.32449948729317e-06,
      "loss": 1.0214,
      "step": 4761
    },
    {
      "epoch": 0.38481585486575487,
      "grad_norm": 2.653942823410034,
      "learning_rate": 9.324170996854875e-06,
      "loss": 0.9748,
      "step": 4762
    },
    {
      "epoch": 0.3848966645790824,
      "grad_norm": 2.8191370964050293,
      "learning_rate": 9.323842432353943e-06,
      "loss": 0.9905,
      "step": 4763
    },
    {
      "epoch": 0.38497747429240997,
      "grad_norm": 2.8964755535125732,
      "learning_rate": 9.323513793795997e-06,
      "loss": 1.0662,
      "step": 4764
    },
    {
      "epoch": 0.3850582840057375,
      "grad_norm": 2.9598093032836914,
      "learning_rate": 9.32318508118667e-06,
      "loss": 1.0305,
      "step": 4765
    },
    {
      "epoch": 0.385139093719065,
      "grad_norm": 2.8310484886169434,
      "learning_rate": 9.322856294531589e-06,
      "loss": 0.9283,
      "step": 4766
    },
    {
      "epoch": 0.3852199034323926,
      "grad_norm": 3.1651182174682617,
      "learning_rate": 9.322527433836386e-06,
      "loss": 0.975,
      "step": 4767
    },
    {
      "epoch": 0.3853007131457201,
      "grad_norm": 2.526073932647705,
      "learning_rate": 9.322198499106693e-06,
      "loss": 0.8425,
      "step": 4768
    },
    {
      "epoch": 0.38538152285904764,
      "grad_norm": 3.729663848876953,
      "learning_rate": 9.321869490348147e-06,
      "loss": 0.8551,
      "step": 4769
    },
    {
      "epoch": 0.3854623325723752,
      "grad_norm": 2.6058268547058105,
      "learning_rate": 9.321540407566382e-06,
      "loss": 1.0275,
      "step": 4770
    },
    {
      "epoch": 0.38554314228570274,
      "grad_norm": 2.9478402137756348,
      "learning_rate": 9.321211250767033e-06,
      "loss": 1.0433,
      "step": 4771
    },
    {
      "epoch": 0.38562395199903027,
      "grad_norm": 2.6278555393218994,
      "learning_rate": 9.320882019955737e-06,
      "loss": 1.0271,
      "step": 4772
    },
    {
      "epoch": 0.38570476171235785,
      "grad_norm": 2.607504367828369,
      "learning_rate": 9.320552715138136e-06,
      "loss": 1.0778,
      "step": 4773
    },
    {
      "epoch": 0.38578557142568537,
      "grad_norm": 2.555558919906616,
      "learning_rate": 9.320223336319865e-06,
      "loss": 0.9599,
      "step": 4774
    },
    {
      "epoch": 0.3858663811390129,
      "grad_norm": 3.114999294281006,
      "learning_rate": 9.319893883506572e-06,
      "loss": 0.8927,
      "step": 4775
    },
    {
      "epoch": 0.38594719085234047,
      "grad_norm": 2.8525390625,
      "learning_rate": 9.319564356703895e-06,
      "loss": 1.0061,
      "step": 4776
    },
    {
      "epoch": 0.386028000565668,
      "grad_norm": 2.945138454437256,
      "learning_rate": 9.31923475591748e-06,
      "loss": 1.0231,
      "step": 4777
    },
    {
      "epoch": 0.3861088102789955,
      "grad_norm": 2.8288087844848633,
      "learning_rate": 9.318905081152972e-06,
      "loss": 1.0277,
      "step": 4778
    },
    {
      "epoch": 0.3861896199923231,
      "grad_norm": 3.0563604831695557,
      "learning_rate": 9.318575332416016e-06,
      "loss": 0.921,
      "step": 4779
    },
    {
      "epoch": 0.3862704297056506,
      "grad_norm": 2.299703598022461,
      "learning_rate": 9.318245509712262e-06,
      "loss": 1.0233,
      "step": 4780
    },
    {
      "epoch": 0.38635123941897814,
      "grad_norm": 2.6213436126708984,
      "learning_rate": 9.317915613047358e-06,
      "loss": 1.0653,
      "step": 4781
    },
    {
      "epoch": 0.3864320491323057,
      "grad_norm": 2.7528960704803467,
      "learning_rate": 9.317585642426954e-06,
      "loss": 0.961,
      "step": 4782
    },
    {
      "epoch": 0.38651285884563324,
      "grad_norm": 2.405949354171753,
      "learning_rate": 9.317255597856703e-06,
      "loss": 1.0507,
      "step": 4783
    },
    {
      "epoch": 0.38659366855896077,
      "grad_norm": 2.5849359035491943,
      "learning_rate": 9.316925479342258e-06,
      "loss": 0.9822,
      "step": 4784
    },
    {
      "epoch": 0.38667447827228835,
      "grad_norm": 3.079529285430908,
      "learning_rate": 9.316595286889271e-06,
      "loss": 0.8935,
      "step": 4785
    },
    {
      "epoch": 0.38675528798561587,
      "grad_norm": 3.638134717941284,
      "learning_rate": 9.316265020503398e-06,
      "loss": 0.9915,
      "step": 4786
    },
    {
      "epoch": 0.3868360976989434,
      "grad_norm": 2.217405080795288,
      "learning_rate": 9.315934680190296e-06,
      "loss": 0.9402,
      "step": 4787
    },
    {
      "epoch": 0.38691690741227097,
      "grad_norm": 2.814835548400879,
      "learning_rate": 9.315604265955625e-06,
      "loss": 0.9646,
      "step": 4788
    },
    {
      "epoch": 0.3869977171255985,
      "grad_norm": 2.7332446575164795,
      "learning_rate": 9.315273777805041e-06,
      "loss": 1.0287,
      "step": 4789
    },
    {
      "epoch": 0.387078526838926,
      "grad_norm": 2.6318838596343994,
      "learning_rate": 9.314943215744205e-06,
      "loss": 0.9873,
      "step": 4790
    },
    {
      "epoch": 0.3871593365522536,
      "grad_norm": 2.76735782623291,
      "learning_rate": 9.31461257977878e-06,
      "loss": 0.8939,
      "step": 4791
    },
    {
      "epoch": 0.3872401462655811,
      "grad_norm": 2.2776925563812256,
      "learning_rate": 9.314281869914429e-06,
      "loss": 1.1241,
      "step": 4792
    },
    {
      "epoch": 0.38732095597890864,
      "grad_norm": 2.7892775535583496,
      "learning_rate": 9.313951086156815e-06,
      "loss": 1.0662,
      "step": 4793
    },
    {
      "epoch": 0.3874017656922362,
      "grad_norm": 3.429621696472168,
      "learning_rate": 9.313620228511605e-06,
      "loss": 0.9103,
      "step": 4794
    },
    {
      "epoch": 0.38748257540556375,
      "grad_norm": 2.6924426555633545,
      "learning_rate": 9.313289296984465e-06,
      "loss": 1.0738,
      "step": 4795
    },
    {
      "epoch": 0.38756338511889127,
      "grad_norm": 2.7512428760528564,
      "learning_rate": 9.312958291581064e-06,
      "loss": 1.023,
      "step": 4796
    },
    {
      "epoch": 0.38764419483221885,
      "grad_norm": 2.5741891860961914,
      "learning_rate": 9.312627212307069e-06,
      "loss": 0.8525,
      "step": 4797
    },
    {
      "epoch": 0.38772500454554637,
      "grad_norm": 2.840163469314575,
      "learning_rate": 9.312296059168153e-06,
      "loss": 1.0139,
      "step": 4798
    },
    {
      "epoch": 0.3878058142588739,
      "grad_norm": 2.5960631370544434,
      "learning_rate": 9.311964832169987e-06,
      "loss": 1.0073,
      "step": 4799
    },
    {
      "epoch": 0.38788662397220147,
      "grad_norm": 2.8192176818847656,
      "learning_rate": 9.311633531318243e-06,
      "loss": 1.0632,
      "step": 4800
    },
    {
      "epoch": 0.387967433685529,
      "grad_norm": 3.081470251083374,
      "learning_rate": 9.311302156618597e-06,
      "loss": 0.8643,
      "step": 4801
    },
    {
      "epoch": 0.3880482433988565,
      "grad_norm": 2.318772077560425,
      "learning_rate": 9.310970708076724e-06,
      "loss": 0.9243,
      "step": 4802
    },
    {
      "epoch": 0.3881290531121841,
      "grad_norm": 2.716639757156372,
      "learning_rate": 9.310639185698301e-06,
      "loss": 1.0154,
      "step": 4803
    },
    {
      "epoch": 0.3882098628255116,
      "grad_norm": 2.643319606781006,
      "learning_rate": 9.310307589489007e-06,
      "loss": 1.0741,
      "step": 4804
    },
    {
      "epoch": 0.38829067253883914,
      "grad_norm": 2.5733048915863037,
      "learning_rate": 9.309975919454519e-06,
      "loss": 1.0569,
      "step": 4805
    },
    {
      "epoch": 0.3883714822521667,
      "grad_norm": 2.561190128326416,
      "learning_rate": 9.309644175600521e-06,
      "loss": 1.0505,
      "step": 4806
    },
    {
      "epoch": 0.38845229196549425,
      "grad_norm": 3.104463577270508,
      "learning_rate": 9.309312357932693e-06,
      "loss": 0.9595,
      "step": 4807
    },
    {
      "epoch": 0.38853310167882177,
      "grad_norm": 2.5482091903686523,
      "learning_rate": 9.308980466456718e-06,
      "loss": 0.9709,
      "step": 4808
    },
    {
      "epoch": 0.38861391139214935,
      "grad_norm": 2.562335252761841,
      "learning_rate": 9.30864850117828e-06,
      "loss": 0.91,
      "step": 4809
    },
    {
      "epoch": 0.38869472110547687,
      "grad_norm": 2.939732074737549,
      "learning_rate": 9.308316462103069e-06,
      "loss": 0.9,
      "step": 4810
    },
    {
      "epoch": 0.3887755308188044,
      "grad_norm": 2.47993803024292,
      "learning_rate": 9.307984349236767e-06,
      "loss": 1.0863,
      "step": 4811
    },
    {
      "epoch": 0.38885634053213197,
      "grad_norm": 2.6861395835876465,
      "learning_rate": 9.307652162585063e-06,
      "loss": 0.9657,
      "step": 4812
    },
    {
      "epoch": 0.3889371502454595,
      "grad_norm": 2.1660261154174805,
      "learning_rate": 9.30731990215365e-06,
      "loss": 0.9754,
      "step": 4813
    },
    {
      "epoch": 0.389017959958787,
      "grad_norm": 2.4602410793304443,
      "learning_rate": 9.306987567948216e-06,
      "loss": 0.9003,
      "step": 4814
    },
    {
      "epoch": 0.3890987696721146,
      "grad_norm": 2.805319309234619,
      "learning_rate": 9.306655159974451e-06,
      "loss": 0.9549,
      "step": 4815
    },
    {
      "epoch": 0.3891795793854421,
      "grad_norm": 3.027040719985962,
      "learning_rate": 9.306322678238054e-06,
      "loss": 0.9587,
      "step": 4816
    },
    {
      "epoch": 0.3892603890987697,
      "grad_norm": 2.4003875255584717,
      "learning_rate": 9.305990122744716e-06,
      "loss": 0.8755,
      "step": 4817
    },
    {
      "epoch": 0.3893411988120972,
      "grad_norm": 2.4520020484924316,
      "learning_rate": 9.305657493500134e-06,
      "loss": 0.9611,
      "step": 4818
    },
    {
      "epoch": 0.38942200852542475,
      "grad_norm": 2.8413140773773193,
      "learning_rate": 9.305324790510001e-06,
      "loss": 0.9743,
      "step": 4819
    },
    {
      "epoch": 0.3895028182387523,
      "grad_norm": 2.7833945751190186,
      "learning_rate": 9.304992013780023e-06,
      "loss": 1.025,
      "step": 4820
    },
    {
      "epoch": 0.38958362795207985,
      "grad_norm": 2.8623552322387695,
      "learning_rate": 9.304659163315894e-06,
      "loss": 1.0315,
      "step": 4821
    },
    {
      "epoch": 0.38966443766540737,
      "grad_norm": 3.116002321243286,
      "learning_rate": 9.304326239123316e-06,
      "loss": 0.9782,
      "step": 4822
    },
    {
      "epoch": 0.38974524737873495,
      "grad_norm": 2.8525900840759277,
      "learning_rate": 9.303993241207994e-06,
      "loss": 0.8915,
      "step": 4823
    },
    {
      "epoch": 0.3898260570920625,
      "grad_norm": 2.838837146759033,
      "learning_rate": 9.303660169575626e-06,
      "loss": 1.0472,
      "step": 4824
    },
    {
      "epoch": 0.38990686680539,
      "grad_norm": 3.3665411472320557,
      "learning_rate": 9.303327024231924e-06,
      "loss": 1.0027,
      "step": 4825
    },
    {
      "epoch": 0.3899876765187176,
      "grad_norm": 2.8950564861297607,
      "learning_rate": 9.302993805182586e-06,
      "loss": 0.8665,
      "step": 4826
    },
    {
      "epoch": 0.3900684862320451,
      "grad_norm": 2.328845500946045,
      "learning_rate": 9.302660512433324e-06,
      "loss": 1.0294,
      "step": 4827
    },
    {
      "epoch": 0.3901492959453726,
      "grad_norm": 2.6346685886383057,
      "learning_rate": 9.302327145989846e-06,
      "loss": 0.9807,
      "step": 4828
    },
    {
      "epoch": 0.3902301056587002,
      "grad_norm": 2.6841721534729004,
      "learning_rate": 9.301993705857864e-06,
      "loss": 0.994,
      "step": 4829
    },
    {
      "epoch": 0.3903109153720277,
      "grad_norm": 2.8069612979888916,
      "learning_rate": 9.301660192043082e-06,
      "loss": 0.9295,
      "step": 4830
    },
    {
      "epoch": 0.39039172508535525,
      "grad_norm": 3.122180223464966,
      "learning_rate": 9.301326604551219e-06,
      "loss": 0.994,
      "step": 4831
    },
    {
      "epoch": 0.3904725347986828,
      "grad_norm": 2.763744354248047,
      "learning_rate": 9.300992943387988e-06,
      "loss": 0.8689,
      "step": 4832
    },
    {
      "epoch": 0.39055334451201035,
      "grad_norm": 2.752152442932129,
      "learning_rate": 9.3006592085591e-06,
      "loss": 1.0883,
      "step": 4833
    },
    {
      "epoch": 0.39063415422533787,
      "grad_norm": 3.1523523330688477,
      "learning_rate": 9.300325400070274e-06,
      "loss": 1.0242,
      "step": 4834
    },
    {
      "epoch": 0.39071496393866545,
      "grad_norm": 2.9127399921417236,
      "learning_rate": 9.299991517927224e-06,
      "loss": 1.2013,
      "step": 4835
    },
    {
      "epoch": 0.390795773651993,
      "grad_norm": 2.768834352493286,
      "learning_rate": 9.299657562135676e-06,
      "loss": 1.1064,
      "step": 4836
    },
    {
      "epoch": 0.3908765833653205,
      "grad_norm": 2.5925710201263428,
      "learning_rate": 9.29932353270134e-06,
      "loss": 1.0339,
      "step": 4837
    },
    {
      "epoch": 0.3909573930786481,
      "grad_norm": 3.155369281768799,
      "learning_rate": 9.298989429629946e-06,
      "loss": 0.9888,
      "step": 4838
    },
    {
      "epoch": 0.3910382027919756,
      "grad_norm": 2.962634801864624,
      "learning_rate": 9.298655252927211e-06,
      "loss": 1.1091,
      "step": 4839
    },
    {
      "epoch": 0.3911190125053031,
      "grad_norm": 2.7181291580200195,
      "learning_rate": 9.298321002598858e-06,
      "loss": 0.9446,
      "step": 4840
    },
    {
      "epoch": 0.3911998222186307,
      "grad_norm": 2.699556350708008,
      "learning_rate": 9.297986678650617e-06,
      "loss": 1.0881,
      "step": 4841
    },
    {
      "epoch": 0.3912806319319582,
      "grad_norm": 2.432003974914551,
      "learning_rate": 9.29765228108821e-06,
      "loss": 0.9296,
      "step": 4842
    },
    {
      "epoch": 0.39136144164528575,
      "grad_norm": 2.3125052452087402,
      "learning_rate": 9.297317809917366e-06,
      "loss": 0.9134,
      "step": 4843
    },
    {
      "epoch": 0.3914422513586133,
      "grad_norm": 2.6786537170410156,
      "learning_rate": 9.296983265143812e-06,
      "loss": 0.905,
      "step": 4844
    },
    {
      "epoch": 0.39152306107194085,
      "grad_norm": 2.5250349044799805,
      "learning_rate": 9.296648646773279e-06,
      "loss": 0.9933,
      "step": 4845
    },
    {
      "epoch": 0.39160387078526837,
      "grad_norm": 3.2397947311401367,
      "learning_rate": 9.2963139548115e-06,
      "loss": 0.9712,
      "step": 4846
    },
    {
      "epoch": 0.39168468049859595,
      "grad_norm": 2.5650792121887207,
      "learning_rate": 9.295979189264206e-06,
      "loss": 1.1292,
      "step": 4847
    },
    {
      "epoch": 0.3917654902119235,
      "grad_norm": 3.4991955757141113,
      "learning_rate": 9.29564435013713e-06,
      "loss": 0.9921,
      "step": 4848
    },
    {
      "epoch": 0.391846299925251,
      "grad_norm": 2.7665810585021973,
      "learning_rate": 9.295309437436007e-06,
      "loss": 1.0006,
      "step": 4849
    },
    {
      "epoch": 0.3919271096385786,
      "grad_norm": 3.329380512237549,
      "learning_rate": 9.294974451166576e-06,
      "loss": 0.973,
      "step": 4850
    },
    {
      "epoch": 0.3920079193519061,
      "grad_norm": 2.8061275482177734,
      "learning_rate": 9.29463939133457e-06,
      "loss": 0.9337,
      "step": 4851
    },
    {
      "epoch": 0.3920887290652336,
      "grad_norm": 2.5380403995513916,
      "learning_rate": 9.294304257945732e-06,
      "loss": 0.8911,
      "step": 4852
    },
    {
      "epoch": 0.3921695387785612,
      "grad_norm": 2.8728508949279785,
      "learning_rate": 9.293969051005798e-06,
      "loss": 0.9115,
      "step": 4853
    },
    {
      "epoch": 0.3922503484918887,
      "grad_norm": 3.0287833213806152,
      "learning_rate": 9.293633770520514e-06,
      "loss": 0.8794,
      "step": 4854
    },
    {
      "epoch": 0.39233115820521625,
      "grad_norm": 3.8823435306549072,
      "learning_rate": 9.293298416495618e-06,
      "loss": 0.9241,
      "step": 4855
    },
    {
      "epoch": 0.3924119679185438,
      "grad_norm": 2.7386648654937744,
      "learning_rate": 9.292962988936856e-06,
      "loss": 0.9388,
      "step": 4856
    },
    {
      "epoch": 0.39249277763187135,
      "grad_norm": 2.5009844303131104,
      "learning_rate": 9.292627487849975e-06,
      "loss": 0.994,
      "step": 4857
    },
    {
      "epoch": 0.39257358734519887,
      "grad_norm": 2.9496538639068604,
      "learning_rate": 9.292291913240716e-06,
      "loss": 0.9525,
      "step": 4858
    },
    {
      "epoch": 0.39265439705852645,
      "grad_norm": 2.7968921661376953,
      "learning_rate": 9.291956265114832e-06,
      "loss": 0.918,
      "step": 4859
    },
    {
      "epoch": 0.392735206771854,
      "grad_norm": 3.234957695007324,
      "learning_rate": 9.29162054347807e-06,
      "loss": 0.8922,
      "step": 4860
    },
    {
      "epoch": 0.3928160164851815,
      "grad_norm": 2.7150919437408447,
      "learning_rate": 9.291284748336179e-06,
      "loss": 0.8799,
      "step": 4861
    },
    {
      "epoch": 0.3928968261985091,
      "grad_norm": 2.8099822998046875,
      "learning_rate": 9.29094887969491e-06,
      "loss": 0.9143,
      "step": 4862
    },
    {
      "epoch": 0.3929776359118366,
      "grad_norm": 2.824078321456909,
      "learning_rate": 9.290612937560017e-06,
      "loss": 0.9023,
      "step": 4863
    },
    {
      "epoch": 0.3930584456251641,
      "grad_norm": 2.730456590652466,
      "learning_rate": 9.290276921937256e-06,
      "loss": 1.0412,
      "step": 4864
    },
    {
      "epoch": 0.3931392553384917,
      "grad_norm": 2.659019947052002,
      "learning_rate": 9.289940832832377e-06,
      "loss": 1.0026,
      "step": 4865
    },
    {
      "epoch": 0.3932200650518192,
      "grad_norm": 3.3061633110046387,
      "learning_rate": 9.28960467025114e-06,
      "loss": 0.9082,
      "step": 4866
    },
    {
      "epoch": 0.39330087476514675,
      "grad_norm": 3.18700909614563,
      "learning_rate": 9.289268434199302e-06,
      "loss": 0.9735,
      "step": 4867
    },
    {
      "epoch": 0.3933816844784743,
      "grad_norm": 2.674553632736206,
      "learning_rate": 9.28893212468262e-06,
      "loss": 1.0538,
      "step": 4868
    },
    {
      "epoch": 0.39346249419180185,
      "grad_norm": 2.7704389095306396,
      "learning_rate": 9.28859574170686e-06,
      "loss": 0.9991,
      "step": 4869
    },
    {
      "epoch": 0.39354330390512937,
      "grad_norm": 2.8787927627563477,
      "learning_rate": 9.288259285277776e-06,
      "loss": 0.9363,
      "step": 4870
    },
    {
      "epoch": 0.39362411361845695,
      "grad_norm": 2.7347707748413086,
      "learning_rate": 9.287922755401135e-06,
      "loss": 0.8876,
      "step": 4871
    },
    {
      "epoch": 0.3937049233317845,
      "grad_norm": 2.5701942443847656,
      "learning_rate": 9.2875861520827e-06,
      "loss": 0.9612,
      "step": 4872
    },
    {
      "epoch": 0.393785733045112,
      "grad_norm": 2.8623557090759277,
      "learning_rate": 9.287249475328236e-06,
      "loss": 1.0507,
      "step": 4873
    },
    {
      "epoch": 0.3938665427584396,
      "grad_norm": 2.523864269256592,
      "learning_rate": 9.28691272514351e-06,
      "loss": 1.0098,
      "step": 4874
    },
    {
      "epoch": 0.3939473524717671,
      "grad_norm": 2.9024195671081543,
      "learning_rate": 9.28657590153429e-06,
      "loss": 1.0518,
      "step": 4875
    },
    {
      "epoch": 0.3940281621850946,
      "grad_norm": 2.553307056427002,
      "learning_rate": 9.286239004506342e-06,
      "loss": 0.8807,
      "step": 4876
    },
    {
      "epoch": 0.3941089718984222,
      "grad_norm": 2.453803777694702,
      "learning_rate": 9.28590203406544e-06,
      "loss": 0.9068,
      "step": 4877
    },
    {
      "epoch": 0.3941897816117497,
      "grad_norm": 2.7090606689453125,
      "learning_rate": 9.285564990217355e-06,
      "loss": 0.9219,
      "step": 4878
    },
    {
      "epoch": 0.3942705913250773,
      "grad_norm": 2.4958701133728027,
      "learning_rate": 9.285227872967857e-06,
      "loss": 1.0385,
      "step": 4879
    },
    {
      "epoch": 0.3943514010384048,
      "grad_norm": 2.777512550354004,
      "learning_rate": 9.284890682322723e-06,
      "loss": 0.9917,
      "step": 4880
    },
    {
      "epoch": 0.39443221075173235,
      "grad_norm": 2.6221389770507812,
      "learning_rate": 9.284553418287725e-06,
      "loss": 1.0376,
      "step": 4881
    },
    {
      "epoch": 0.3945130204650599,
      "grad_norm": 2.436344861984253,
      "learning_rate": 9.284216080868645e-06,
      "loss": 1.0367,
      "step": 4882
    },
    {
      "epoch": 0.39459383017838745,
      "grad_norm": 2.7965519428253174,
      "learning_rate": 9.283878670071255e-06,
      "loss": 0.8879,
      "step": 4883
    },
    {
      "epoch": 0.394674639891715,
      "grad_norm": 2.6184844970703125,
      "learning_rate": 9.283541185901337e-06,
      "loss": 0.9681,
      "step": 4884
    },
    {
      "epoch": 0.39475544960504255,
      "grad_norm": 2.4829251766204834,
      "learning_rate": 9.28320362836467e-06,
      "loss": 0.9471,
      "step": 4885
    },
    {
      "epoch": 0.3948362593183701,
      "grad_norm": 2.3925609588623047,
      "learning_rate": 9.282865997467037e-06,
      "loss": 0.9715,
      "step": 4886
    },
    {
      "epoch": 0.3949170690316976,
      "grad_norm": 2.9331045150756836,
      "learning_rate": 9.282528293214219e-06,
      "loss": 1.0717,
      "step": 4887
    },
    {
      "epoch": 0.3949978787450252,
      "grad_norm": 2.435237169265747,
      "learning_rate": 9.282190515612003e-06,
      "loss": 0.9683,
      "step": 4888
    },
    {
      "epoch": 0.3950786884583527,
      "grad_norm": 2.5075645446777344,
      "learning_rate": 9.281852664666171e-06,
      "loss": 0.9484,
      "step": 4889
    },
    {
      "epoch": 0.3951594981716802,
      "grad_norm": 2.5758745670318604,
      "learning_rate": 9.281514740382511e-06,
      "loss": 0.9597,
      "step": 4890
    },
    {
      "epoch": 0.3952403078850078,
      "grad_norm": 2.3525283336639404,
      "learning_rate": 9.281176742766811e-06,
      "loss": 1.0151,
      "step": 4891
    },
    {
      "epoch": 0.3953211175983353,
      "grad_norm": 2.4200117588043213,
      "learning_rate": 9.280838671824861e-06,
      "loss": 0.9881,
      "step": 4892
    },
    {
      "epoch": 0.39540192731166285,
      "grad_norm": 2.7584762573242188,
      "learning_rate": 9.280500527562449e-06,
      "loss": 0.8927,
      "step": 4893
    },
    {
      "epoch": 0.3954827370249904,
      "grad_norm": 2.8647139072418213,
      "learning_rate": 9.280162309985369e-06,
      "loss": 1.0694,
      "step": 4894
    },
    {
      "epoch": 0.39556354673831795,
      "grad_norm": 3.1318676471710205,
      "learning_rate": 9.279824019099412e-06,
      "loss": 1.0155,
      "step": 4895
    },
    {
      "epoch": 0.3956443564516455,
      "grad_norm": 2.8281431198120117,
      "learning_rate": 9.279485654910371e-06,
      "loss": 0.9599,
      "step": 4896
    },
    {
      "epoch": 0.39572516616497305,
      "grad_norm": 2.5201003551483154,
      "learning_rate": 9.279147217424046e-06,
      "loss": 1.0393,
      "step": 4897
    },
    {
      "epoch": 0.3958059758783006,
      "grad_norm": 2.6277709007263184,
      "learning_rate": 9.27880870664623e-06,
      "loss": 1.1643,
      "step": 4898
    },
    {
      "epoch": 0.3958867855916281,
      "grad_norm": 2.8465261459350586,
      "learning_rate": 9.27847012258272e-06,
      "loss": 0.9927,
      "step": 4899
    },
    {
      "epoch": 0.3959675953049557,
      "grad_norm": 2.7331273555755615,
      "learning_rate": 9.27813146523932e-06,
      "loss": 0.9616,
      "step": 4900
    },
    {
      "epoch": 0.3960484050182832,
      "grad_norm": 3.133472442626953,
      "learning_rate": 9.277792734621825e-06,
      "loss": 1.0276,
      "step": 4901
    },
    {
      "epoch": 0.3961292147316107,
      "grad_norm": 2.812513828277588,
      "learning_rate": 9.277453930736039e-06,
      "loss": 0.923,
      "step": 4902
    },
    {
      "epoch": 0.3962100244449383,
      "grad_norm": 3.0874874591827393,
      "learning_rate": 9.277115053587764e-06,
      "loss": 0.9529,
      "step": 4903
    },
    {
      "epoch": 0.3962908341582658,
      "grad_norm": 2.806426525115967,
      "learning_rate": 9.276776103182806e-06,
      "loss": 0.9119,
      "step": 4904
    },
    {
      "epoch": 0.39637164387159335,
      "grad_norm": 2.516444206237793,
      "learning_rate": 9.276437079526969e-06,
      "loss": 0.8762,
      "step": 4905
    },
    {
      "epoch": 0.3964524535849209,
      "grad_norm": 3.095404863357544,
      "learning_rate": 9.27609798262606e-06,
      "loss": 0.9768,
      "step": 4906
    },
    {
      "epoch": 0.39653326329824845,
      "grad_norm": 2.437582015991211,
      "learning_rate": 9.275758812485887e-06,
      "loss": 0.9226,
      "step": 4907
    },
    {
      "epoch": 0.396614073011576,
      "grad_norm": 2.405050754547119,
      "learning_rate": 9.275419569112258e-06,
      "loss": 1.0769,
      "step": 4908
    },
    {
      "epoch": 0.39669488272490355,
      "grad_norm": 2.7021193504333496,
      "learning_rate": 9.275080252510986e-06,
      "loss": 0.9472,
      "step": 4909
    },
    {
      "epoch": 0.3967756924382311,
      "grad_norm": 3.1435816287994385,
      "learning_rate": 9.27474086268788e-06,
      "loss": 0.9132,
      "step": 4910
    },
    {
      "epoch": 0.3968565021515586,
      "grad_norm": 3.182762384414673,
      "learning_rate": 9.274401399648755e-06,
      "loss": 0.902,
      "step": 4911
    },
    {
      "epoch": 0.3969373118648862,
      "grad_norm": 2.7405760288238525,
      "learning_rate": 9.274061863399424e-06,
      "loss": 0.924,
      "step": 4912
    },
    {
      "epoch": 0.3970181215782137,
      "grad_norm": 2.6312766075134277,
      "learning_rate": 9.273722253945701e-06,
      "loss": 1.1052,
      "step": 4913
    },
    {
      "epoch": 0.3970989312915412,
      "grad_norm": 2.851205825805664,
      "learning_rate": 9.273382571293407e-06,
      "loss": 0.9171,
      "step": 4914
    },
    {
      "epoch": 0.3971797410048688,
      "grad_norm": 2.6783201694488525,
      "learning_rate": 9.273042815448357e-06,
      "loss": 1.0252,
      "step": 4915
    },
    {
      "epoch": 0.3972605507181963,
      "grad_norm": 2.60844087600708,
      "learning_rate": 9.272702986416368e-06,
      "loss": 0.9343,
      "step": 4916
    },
    {
      "epoch": 0.39734136043152385,
      "grad_norm": 2.7676470279693604,
      "learning_rate": 9.272363084203264e-06,
      "loss": 0.9558,
      "step": 4917
    },
    {
      "epoch": 0.3974221701448514,
      "grad_norm": 3.101247549057007,
      "learning_rate": 9.272023108814867e-06,
      "loss": 0.8752,
      "step": 4918
    },
    {
      "epoch": 0.39750297985817895,
      "grad_norm": 3.0404317378997803,
      "learning_rate": 9.271683060256997e-06,
      "loss": 1.0014,
      "step": 4919
    },
    {
      "epoch": 0.3975837895715065,
      "grad_norm": 2.6339497566223145,
      "learning_rate": 9.271342938535481e-06,
      "loss": 0.983,
      "step": 4920
    },
    {
      "epoch": 0.39766459928483405,
      "grad_norm": 2.664785623550415,
      "learning_rate": 9.271002743656142e-06,
      "loss": 0.993,
      "step": 4921
    },
    {
      "epoch": 0.3977454089981616,
      "grad_norm": 3.0166149139404297,
      "learning_rate": 9.270662475624809e-06,
      "loss": 0.874,
      "step": 4922
    },
    {
      "epoch": 0.3978262187114891,
      "grad_norm": 2.7920243740081787,
      "learning_rate": 9.270322134447309e-06,
      "loss": 1.0232,
      "step": 4923
    },
    {
      "epoch": 0.3979070284248167,
      "grad_norm": 2.591841220855713,
      "learning_rate": 9.26998172012947e-06,
      "loss": 0.9601,
      "step": 4924
    },
    {
      "epoch": 0.3979878381381442,
      "grad_norm": 3.2867536544799805,
      "learning_rate": 9.269641232677126e-06,
      "loss": 1.0158,
      "step": 4925
    },
    {
      "epoch": 0.3980686478514717,
      "grad_norm": 2.7288825511932373,
      "learning_rate": 9.269300672096105e-06,
      "loss": 0.9982,
      "step": 4926
    },
    {
      "epoch": 0.3981494575647993,
      "grad_norm": 2.49446702003479,
      "learning_rate": 9.268960038392242e-06,
      "loss": 0.9894,
      "step": 4927
    },
    {
      "epoch": 0.3982302672781268,
      "grad_norm": 2.6400210857391357,
      "learning_rate": 9.268619331571369e-06,
      "loss": 0.9767,
      "step": 4928
    },
    {
      "epoch": 0.39831107699145435,
      "grad_norm": 2.6287052631378174,
      "learning_rate": 9.268278551639325e-06,
      "loss": 0.9216,
      "step": 4929
    },
    {
      "epoch": 0.39839188670478193,
      "grad_norm": 2.7632782459259033,
      "learning_rate": 9.267937698601946e-06,
      "loss": 0.9815,
      "step": 4930
    },
    {
      "epoch": 0.39847269641810945,
      "grad_norm": 2.5979580879211426,
      "learning_rate": 9.267596772465066e-06,
      "loss": 0.9465,
      "step": 4931
    },
    {
      "epoch": 0.398553506131437,
      "grad_norm": 2.6885600090026855,
      "learning_rate": 9.267255773234526e-06,
      "loss": 0.9149,
      "step": 4932
    },
    {
      "epoch": 0.39863431584476455,
      "grad_norm": 2.767235517501831,
      "learning_rate": 9.266914700916172e-06,
      "loss": 0.9113,
      "step": 4933
    },
    {
      "epoch": 0.3987151255580921,
      "grad_norm": 2.7402396202087402,
      "learning_rate": 9.266573555515838e-06,
      "loss": 0.9815,
      "step": 4934
    },
    {
      "epoch": 0.3987959352714196,
      "grad_norm": 2.782092332839966,
      "learning_rate": 9.266232337039372e-06,
      "loss": 0.8879,
      "step": 4935
    },
    {
      "epoch": 0.3988767449847472,
      "grad_norm": 2.624999523162842,
      "learning_rate": 9.265891045492616e-06,
      "loss": 0.9257,
      "step": 4936
    },
    {
      "epoch": 0.3989575546980747,
      "grad_norm": 2.7730746269226074,
      "learning_rate": 9.265549680881416e-06,
      "loss": 1.0308,
      "step": 4937
    },
    {
      "epoch": 0.3990383644114022,
      "grad_norm": 2.442979335784912,
      "learning_rate": 9.26520824321162e-06,
      "loss": 0.9566,
      "step": 4938
    },
    {
      "epoch": 0.3991191741247298,
      "grad_norm": 3.1745896339416504,
      "learning_rate": 9.264866732489073e-06,
      "loss": 1.0078,
      "step": 4939
    },
    {
      "epoch": 0.3991999838380573,
      "grad_norm": 2.7653796672821045,
      "learning_rate": 9.264525148719628e-06,
      "loss": 0.8728,
      "step": 4940
    },
    {
      "epoch": 0.39928079355138485,
      "grad_norm": 2.759915828704834,
      "learning_rate": 9.264183491909133e-06,
      "loss": 1.0393,
      "step": 4941
    },
    {
      "epoch": 0.39936160326471243,
      "grad_norm": 2.7928683757781982,
      "learning_rate": 9.263841762063438e-06,
      "loss": 0.9586,
      "step": 4942
    },
    {
      "epoch": 0.39944241297803995,
      "grad_norm": 2.7925634384155273,
      "learning_rate": 9.263499959188403e-06,
      "loss": 1.0601,
      "step": 4943
    },
    {
      "epoch": 0.39952322269136753,
      "grad_norm": 2.689061164855957,
      "learning_rate": 9.263158083289874e-06,
      "loss": 1.0542,
      "step": 4944
    },
    {
      "epoch": 0.39960403240469505,
      "grad_norm": 2.5203075408935547,
      "learning_rate": 9.262816134373711e-06,
      "loss": 0.8948,
      "step": 4945
    },
    {
      "epoch": 0.3996848421180226,
      "grad_norm": 3.390443801879883,
      "learning_rate": 9.26247411244577e-06,
      "loss": 0.9546,
      "step": 4946
    },
    {
      "epoch": 0.39976565183135016,
      "grad_norm": 2.86378812789917,
      "learning_rate": 9.26213201751191e-06,
      "loss": 1.1114,
      "step": 4947
    },
    {
      "epoch": 0.3998464615446777,
      "grad_norm": 3.285104274749756,
      "learning_rate": 9.261789849577988e-06,
      "loss": 0.968,
      "step": 4948
    },
    {
      "epoch": 0.3999272712580052,
      "grad_norm": 2.772576332092285,
      "learning_rate": 9.261447608649866e-06,
      "loss": 1.0612,
      "step": 4949
    },
    {
      "epoch": 0.4000080809713328,
      "grad_norm": 2.4840214252471924,
      "learning_rate": 9.261105294733405e-06,
      "loss": 0.9996,
      "step": 4950
    },
    {
      "epoch": 0.4000888906846603,
      "grad_norm": 2.6143908500671387,
      "learning_rate": 9.26076290783447e-06,
      "loss": 0.8929,
      "step": 4951
    },
    {
      "epoch": 0.4001697003979878,
      "grad_norm": 3.0794944763183594,
      "learning_rate": 9.260420447958922e-06,
      "loss": 0.9006,
      "step": 4952
    },
    {
      "epoch": 0.4002505101113154,
      "grad_norm": 2.430558204650879,
      "learning_rate": 9.26007791511263e-06,
      "loss": 1.0531,
      "step": 4953
    },
    {
      "epoch": 0.40033131982464293,
      "grad_norm": 2.5490753650665283,
      "learning_rate": 9.259735309301458e-06,
      "loss": 0.9219,
      "step": 4954
    },
    {
      "epoch": 0.40041212953797045,
      "grad_norm": 2.9990432262420654,
      "learning_rate": 9.259392630531275e-06,
      "loss": 1.0364,
      "step": 4955
    },
    {
      "epoch": 0.40049293925129803,
      "grad_norm": 2.576552629470825,
      "learning_rate": 9.259049878807951e-06,
      "loss": 0.9433,
      "step": 4956
    },
    {
      "epoch": 0.40057374896462555,
      "grad_norm": 3.066596508026123,
      "learning_rate": 9.258707054137354e-06,
      "loss": 0.9373,
      "step": 4957
    },
    {
      "epoch": 0.4006545586779531,
      "grad_norm": 2.7827351093292236,
      "learning_rate": 9.258364156525359e-06,
      "loss": 0.9872,
      "step": 4958
    },
    {
      "epoch": 0.40073536839128066,
      "grad_norm": 2.649155378341675,
      "learning_rate": 9.258021185977838e-06,
      "loss": 0.9655,
      "step": 4959
    },
    {
      "epoch": 0.4008161781046082,
      "grad_norm": 2.7922468185424805,
      "learning_rate": 9.257678142500663e-06,
      "loss": 0.9581,
      "step": 4960
    },
    {
      "epoch": 0.4008969878179357,
      "grad_norm": 3.1652913093566895,
      "learning_rate": 9.257335026099714e-06,
      "loss": 1.0054,
      "step": 4961
    },
    {
      "epoch": 0.4009777975312633,
      "grad_norm": 2.7333192825317383,
      "learning_rate": 9.256991836780864e-06,
      "loss": 1.0715,
      "step": 4962
    },
    {
      "epoch": 0.4010586072445908,
      "grad_norm": 2.778630018234253,
      "learning_rate": 9.256648574549992e-06,
      "loss": 0.9657,
      "step": 4963
    },
    {
      "epoch": 0.4011394169579183,
      "grad_norm": 2.9172372817993164,
      "learning_rate": 9.256305239412977e-06,
      "loss": 0.9402,
      "step": 4964
    },
    {
      "epoch": 0.4012202266712459,
      "grad_norm": 2.408510446548462,
      "learning_rate": 9.2559618313757e-06,
      "loss": 1.058,
      "step": 4965
    },
    {
      "epoch": 0.40130103638457343,
      "grad_norm": 2.9090781211853027,
      "learning_rate": 9.255618350444042e-06,
      "loss": 0.9233,
      "step": 4966
    },
    {
      "epoch": 0.40138184609790095,
      "grad_norm": 2.568174123764038,
      "learning_rate": 9.255274796623887e-06,
      "loss": 0.9663,
      "step": 4967
    },
    {
      "epoch": 0.40146265581122853,
      "grad_norm": 3.020400285720825,
      "learning_rate": 9.254931169921121e-06,
      "loss": 0.9937,
      "step": 4968
    },
    {
      "epoch": 0.40154346552455605,
      "grad_norm": 2.8069992065429688,
      "learning_rate": 9.254587470341624e-06,
      "loss": 0.9923,
      "step": 4969
    },
    {
      "epoch": 0.4016242752378836,
      "grad_norm": 2.922452211380005,
      "learning_rate": 9.25424369789129e-06,
      "loss": 1.0148,
      "step": 4970
    },
    {
      "epoch": 0.40170508495121116,
      "grad_norm": 2.6371235847473145,
      "learning_rate": 9.253899852576e-06,
      "loss": 1.0003,
      "step": 4971
    },
    {
      "epoch": 0.4017858946645387,
      "grad_norm": 2.6325368881225586,
      "learning_rate": 9.253555934401647e-06,
      "loss": 1.0359,
      "step": 4972
    },
    {
      "epoch": 0.4018667043778662,
      "grad_norm": 2.802062511444092,
      "learning_rate": 9.253211943374122e-06,
      "loss": 0.877,
      "step": 4973
    },
    {
      "epoch": 0.4019475140911938,
      "grad_norm": 2.4746196269989014,
      "learning_rate": 9.252867879499314e-06,
      "loss": 0.9746,
      "step": 4974
    },
    {
      "epoch": 0.4020283238045213,
      "grad_norm": 2.960315227508545,
      "learning_rate": 9.25252374278312e-06,
      "loss": 0.9999,
      "step": 4975
    },
    {
      "epoch": 0.4021091335178488,
      "grad_norm": 3.069545269012451,
      "learning_rate": 9.252179533231428e-06,
      "loss": 1.0284,
      "step": 4976
    },
    {
      "epoch": 0.4021899432311764,
      "grad_norm": 2.5236165523529053,
      "learning_rate": 9.251835250850141e-06,
      "loss": 0.88,
      "step": 4977
    },
    {
      "epoch": 0.40227075294450393,
      "grad_norm": 2.598947525024414,
      "learning_rate": 9.25149089564515e-06,
      "loss": 1.0208,
      "step": 4978
    },
    {
      "epoch": 0.40235156265783145,
      "grad_norm": 2.8756184577941895,
      "learning_rate": 9.251146467622356e-06,
      "loss": 0.9926,
      "step": 4979
    },
    {
      "epoch": 0.40243237237115903,
      "grad_norm": 2.3564445972442627,
      "learning_rate": 9.250801966787657e-06,
      "loss": 0.9608,
      "step": 4980
    },
    {
      "epoch": 0.40251318208448655,
      "grad_norm": 2.262138605117798,
      "learning_rate": 9.250457393146954e-06,
      "loss": 1.012,
      "step": 4981
    },
    {
      "epoch": 0.4025939917978141,
      "grad_norm": 2.7162976264953613,
      "learning_rate": 9.250112746706148e-06,
      "loss": 0.9554,
      "step": 4982
    },
    {
      "epoch": 0.40267480151114166,
      "grad_norm": 2.8713443279266357,
      "learning_rate": 9.249768027471142e-06,
      "loss": 0.9066,
      "step": 4983
    },
    {
      "epoch": 0.4027556112244692,
      "grad_norm": 2.5405688285827637,
      "learning_rate": 9.24942323544784e-06,
      "loss": 1.043,
      "step": 4984
    },
    {
      "epoch": 0.4028364209377967,
      "grad_norm": 2.867854595184326,
      "learning_rate": 9.249078370642149e-06,
      "loss": 0.9656,
      "step": 4985
    },
    {
      "epoch": 0.4029172306511243,
      "grad_norm": 2.8119189739227295,
      "learning_rate": 9.248733433059976e-06,
      "loss": 1.0146,
      "step": 4986
    },
    {
      "epoch": 0.4029980403644518,
      "grad_norm": 2.9293620586395264,
      "learning_rate": 9.248388422707227e-06,
      "loss": 1.0749,
      "step": 4987
    },
    {
      "epoch": 0.4030788500777793,
      "grad_norm": 2.5957796573638916,
      "learning_rate": 9.24804333958981e-06,
      "loss": 1.0848,
      "step": 4988
    },
    {
      "epoch": 0.4031596597911069,
      "grad_norm": 2.9666616916656494,
      "learning_rate": 9.247698183713637e-06,
      "loss": 0.9397,
      "step": 4989
    },
    {
      "epoch": 0.40324046950443443,
      "grad_norm": 2.509471893310547,
      "learning_rate": 9.247352955084623e-06,
      "loss": 1.0488,
      "step": 4990
    },
    {
      "epoch": 0.40332127921776195,
      "grad_norm": 2.3868610858917236,
      "learning_rate": 9.247007653708677e-06,
      "loss": 1.0358,
      "step": 4991
    },
    {
      "epoch": 0.40340208893108953,
      "grad_norm": 2.387303352355957,
      "learning_rate": 9.246662279591713e-06,
      "loss": 0.9579,
      "step": 4992
    },
    {
      "epoch": 0.40348289864441705,
      "grad_norm": 2.869910717010498,
      "learning_rate": 9.24631683273965e-06,
      "loss": 1.0516,
      "step": 4993
    },
    {
      "epoch": 0.4035637083577446,
      "grad_norm": 2.5866353511810303,
      "learning_rate": 9.245971313158399e-06,
      "loss": 0.8982,
      "step": 4994
    },
    {
      "epoch": 0.40364451807107216,
      "grad_norm": 2.5346262454986572,
      "learning_rate": 9.245625720853883e-06,
      "loss": 1.0128,
      "step": 4995
    },
    {
      "epoch": 0.4037253277843997,
      "grad_norm": 2.6702122688293457,
      "learning_rate": 9.24528005583202e-06,
      "loss": 0.857,
      "step": 4996
    },
    {
      "epoch": 0.4038061374977272,
      "grad_norm": 3.1742382049560547,
      "learning_rate": 9.244934318098729e-06,
      "loss": 0.9896,
      "step": 4997
    },
    {
      "epoch": 0.4038869472110548,
      "grad_norm": 3.3246395587921143,
      "learning_rate": 9.24458850765993e-06,
      "loss": 1.0591,
      "step": 4998
    },
    {
      "epoch": 0.4039677569243823,
      "grad_norm": 2.796469211578369,
      "learning_rate": 9.244242624521551e-06,
      "loss": 0.8671,
      "step": 4999
    },
    {
      "epoch": 0.4040485666377098,
      "grad_norm": 2.476184606552124,
      "learning_rate": 9.243896668689514e-06,
      "loss": 1.1087,
      "step": 5000
    },
    {
      "epoch": 0.4040485666377098,
      "eval_loss": 0.8171849846839905,
      "eval_runtime": 813.6456,
      "eval_samples_per_second": 102.46,
      "eval_steps_per_second": 12.808,
      "step": 5000
    },
    {
      "epoch": 0.4041293763510374,
      "grad_norm": 2.734010934829712,
      "learning_rate": 9.243550640169743e-06,
      "loss": 0.9336,
      "step": 5001
    },
    {
      "epoch": 0.40421018606436493,
      "grad_norm": 2.4164507389068604,
      "learning_rate": 9.243204538968165e-06,
      "loss": 0.9413,
      "step": 5002
    },
    {
      "epoch": 0.40429099577769245,
      "grad_norm": 2.576552152633667,
      "learning_rate": 9.242858365090708e-06,
      "loss": 0.9995,
      "step": 5003
    },
    {
      "epoch": 0.40437180549102003,
      "grad_norm": 2.5718369483947754,
      "learning_rate": 9.242512118543302e-06,
      "loss": 0.8484,
      "step": 5004
    },
    {
      "epoch": 0.40445261520434755,
      "grad_norm": 2.749690532684326,
      "learning_rate": 9.242165799331877e-06,
      "loss": 0.9282,
      "step": 5005
    },
    {
      "epoch": 0.4045334249176751,
      "grad_norm": 2.779601812362671,
      "learning_rate": 9.241819407462364e-06,
      "loss": 0.9441,
      "step": 5006
    },
    {
      "epoch": 0.40461423463100266,
      "grad_norm": 2.642305374145508,
      "learning_rate": 9.241472942940697e-06,
      "loss": 1.0112,
      "step": 5007
    },
    {
      "epoch": 0.4046950443443302,
      "grad_norm": 2.582791805267334,
      "learning_rate": 9.241126405772809e-06,
      "loss": 0.9899,
      "step": 5008
    },
    {
      "epoch": 0.40477585405765776,
      "grad_norm": 2.3769354820251465,
      "learning_rate": 9.240779795964637e-06,
      "loss": 0.9363,
      "step": 5009
    },
    {
      "epoch": 0.4048566637709853,
      "grad_norm": 3.000255823135376,
      "learning_rate": 9.240433113522114e-06,
      "loss": 0.899,
      "step": 5010
    },
    {
      "epoch": 0.4049374734843128,
      "grad_norm": 2.493863821029663,
      "learning_rate": 9.240086358451182e-06,
      "loss": 1.0402,
      "step": 5011
    },
    {
      "epoch": 0.4050182831976404,
      "grad_norm": 2.9411659240722656,
      "learning_rate": 9.239739530757776e-06,
      "loss": 1.0769,
      "step": 5012
    },
    {
      "epoch": 0.4050990929109679,
      "grad_norm": 3.193103790283203,
      "learning_rate": 9.23939263044784e-06,
      "loss": 1.161,
      "step": 5013
    },
    {
      "epoch": 0.40517990262429543,
      "grad_norm": 2.7817883491516113,
      "learning_rate": 9.239045657527315e-06,
      "loss": 1.0976,
      "step": 5014
    },
    {
      "epoch": 0.405260712337623,
      "grad_norm": 2.5293238162994385,
      "learning_rate": 9.238698612002143e-06,
      "loss": 1.0325,
      "step": 5015
    },
    {
      "epoch": 0.40534152205095053,
      "grad_norm": 2.5327422618865967,
      "learning_rate": 9.238351493878268e-06,
      "loss": 0.885,
      "step": 5016
    },
    {
      "epoch": 0.40542233176427805,
      "grad_norm": 2.972574472427368,
      "learning_rate": 9.238004303161635e-06,
      "loss": 0.9468,
      "step": 5017
    },
    {
      "epoch": 0.40550314147760563,
      "grad_norm": 2.9105494022369385,
      "learning_rate": 9.23765703985819e-06,
      "loss": 0.9391,
      "step": 5018
    },
    {
      "epoch": 0.40558395119093316,
      "grad_norm": 2.900505542755127,
      "learning_rate": 9.237309703973882e-06,
      "loss": 1.1088,
      "step": 5019
    },
    {
      "epoch": 0.4056647609042607,
      "grad_norm": 2.674318790435791,
      "learning_rate": 9.23696229551466e-06,
      "loss": 0.8911,
      "step": 5020
    },
    {
      "epoch": 0.40574557061758826,
      "grad_norm": 3.3266994953155518,
      "learning_rate": 9.236614814486473e-06,
      "loss": 1.0359,
      "step": 5021
    },
    {
      "epoch": 0.4058263803309158,
      "grad_norm": 3.1881871223449707,
      "learning_rate": 9.236267260895275e-06,
      "loss": 0.9984,
      "step": 5022
    },
    {
      "epoch": 0.4059071900442433,
      "grad_norm": 3.207288980484009,
      "learning_rate": 9.235919634747017e-06,
      "loss": 0.9254,
      "step": 5023
    },
    {
      "epoch": 0.4059879997575709,
      "grad_norm": 2.8666768074035645,
      "learning_rate": 9.235571936047652e-06,
      "loss": 0.9434,
      "step": 5024
    },
    {
      "epoch": 0.4060688094708984,
      "grad_norm": 2.9539434909820557,
      "learning_rate": 9.235224164803138e-06,
      "loss": 1.001,
      "step": 5025
    },
    {
      "epoch": 0.40614961918422593,
      "grad_norm": 2.580145835876465,
      "learning_rate": 9.234876321019429e-06,
      "loss": 1.102,
      "step": 5026
    },
    {
      "epoch": 0.4062304288975535,
      "grad_norm": 2.855926036834717,
      "learning_rate": 9.234528404702484e-06,
      "loss": 0.9665,
      "step": 5027
    },
    {
      "epoch": 0.40631123861088103,
      "grad_norm": 2.862095594406128,
      "learning_rate": 9.23418041585826e-06,
      "loss": 0.9499,
      "step": 5028
    },
    {
      "epoch": 0.40639204832420855,
      "grad_norm": 2.802712917327881,
      "learning_rate": 9.233832354492721e-06,
      "loss": 0.9587,
      "step": 5029
    },
    {
      "epoch": 0.40647285803753613,
      "grad_norm": 2.902677297592163,
      "learning_rate": 9.233484220611825e-06,
      "loss": 1.161,
      "step": 5030
    },
    {
      "epoch": 0.40655366775086366,
      "grad_norm": 2.7959911823272705,
      "learning_rate": 9.233136014221537e-06,
      "loss": 0.9348,
      "step": 5031
    },
    {
      "epoch": 0.4066344774641912,
      "grad_norm": 2.9856088161468506,
      "learning_rate": 9.232787735327821e-06,
      "loss": 0.9717,
      "step": 5032
    },
    {
      "epoch": 0.40671528717751876,
      "grad_norm": 3.0933585166931152,
      "learning_rate": 9.232439383936638e-06,
      "loss": 0.8442,
      "step": 5033
    },
    {
      "epoch": 0.4067960968908463,
      "grad_norm": 2.7330124378204346,
      "learning_rate": 9.23209096005396e-06,
      "loss": 0.8693,
      "step": 5034
    },
    {
      "epoch": 0.4068769066041738,
      "grad_norm": 2.3398380279541016,
      "learning_rate": 9.23174246368575e-06,
      "loss": 1.0293,
      "step": 5035
    },
    {
      "epoch": 0.4069577163175014,
      "grad_norm": 2.698263168334961,
      "learning_rate": 9.231393894837983e-06,
      "loss": 0.8785,
      "step": 5036
    },
    {
      "epoch": 0.4070385260308289,
      "grad_norm": 2.6033833026885986,
      "learning_rate": 9.231045253516622e-06,
      "loss": 1.0154,
      "step": 5037
    },
    {
      "epoch": 0.40711933574415643,
      "grad_norm": 2.1953933238983154,
      "learning_rate": 9.230696539727641e-06,
      "loss": 1.0248,
      "step": 5038
    },
    {
      "epoch": 0.407200145457484,
      "grad_norm": 2.8867642879486084,
      "learning_rate": 9.230347753477015e-06,
      "loss": 0.9788,
      "step": 5039
    },
    {
      "epoch": 0.40728095517081153,
      "grad_norm": 3.269437551498413,
      "learning_rate": 9.229998894770717e-06,
      "loss": 0.953,
      "step": 5040
    },
    {
      "epoch": 0.40736176488413905,
      "grad_norm": 2.780695676803589,
      "learning_rate": 9.22964996361472e-06,
      "loss": 0.96,
      "step": 5041
    },
    {
      "epoch": 0.40744257459746663,
      "grad_norm": 2.9803736209869385,
      "learning_rate": 9.229300960015003e-06,
      "loss": 1.0653,
      "step": 5042
    },
    {
      "epoch": 0.40752338431079416,
      "grad_norm": 2.8825063705444336,
      "learning_rate": 9.22895188397754e-06,
      "loss": 1.0618,
      "step": 5043
    },
    {
      "epoch": 0.4076041940241217,
      "grad_norm": 2.861132860183716,
      "learning_rate": 9.228602735508312e-06,
      "loss": 1.0021,
      "step": 5044
    },
    {
      "epoch": 0.40768500373744926,
      "grad_norm": 2.679664373397827,
      "learning_rate": 9.2282535146133e-06,
      "loss": 0.9712,
      "step": 5045
    },
    {
      "epoch": 0.4077658134507768,
      "grad_norm": 2.448105812072754,
      "learning_rate": 9.227904221298485e-06,
      "loss": 0.8787,
      "step": 5046
    },
    {
      "epoch": 0.4078466231641043,
      "grad_norm": 3.302600145339966,
      "learning_rate": 9.227554855569847e-06,
      "loss": 0.9055,
      "step": 5047
    },
    {
      "epoch": 0.4079274328774319,
      "grad_norm": 2.812666654586792,
      "learning_rate": 9.227205417433373e-06,
      "loss": 0.8513,
      "step": 5048
    },
    {
      "epoch": 0.4080082425907594,
      "grad_norm": 2.6948466300964355,
      "learning_rate": 9.226855906895047e-06,
      "loss": 0.9003,
      "step": 5049
    },
    {
      "epoch": 0.40808905230408693,
      "grad_norm": 2.5601093769073486,
      "learning_rate": 9.226506323960856e-06,
      "loss": 0.9442,
      "step": 5050
    },
    {
      "epoch": 0.4081698620174145,
      "grad_norm": 2.8308019638061523,
      "learning_rate": 9.226156668636785e-06,
      "loss": 0.8969,
      "step": 5051
    },
    {
      "epoch": 0.40825067173074203,
      "grad_norm": 3.942269802093506,
      "learning_rate": 9.225806940928825e-06,
      "loss": 1.0799,
      "step": 5052
    },
    {
      "epoch": 0.40833148144406956,
      "grad_norm": 2.7755072116851807,
      "learning_rate": 9.225457140842964e-06,
      "loss": 1.0485,
      "step": 5053
    },
    {
      "epoch": 0.40841229115739713,
      "grad_norm": 2.5427701473236084,
      "learning_rate": 9.225107268385196e-06,
      "loss": 1.0699,
      "step": 5054
    },
    {
      "epoch": 0.40849310087072466,
      "grad_norm": 2.619122266769409,
      "learning_rate": 9.224757323561511e-06,
      "loss": 0.9209,
      "step": 5055
    },
    {
      "epoch": 0.4085739105840522,
      "grad_norm": 2.6481258869171143,
      "learning_rate": 9.224407306377906e-06,
      "loss": 0.8791,
      "step": 5056
    },
    {
      "epoch": 0.40865472029737976,
      "grad_norm": 2.825996160507202,
      "learning_rate": 9.224057216840371e-06,
      "loss": 0.9755,
      "step": 5057
    },
    {
      "epoch": 0.4087355300107073,
      "grad_norm": 2.170696496963501,
      "learning_rate": 9.223707054954905e-06,
      "loss": 1.1268,
      "step": 5058
    },
    {
      "epoch": 0.4088163397240348,
      "grad_norm": 2.748819589614868,
      "learning_rate": 9.223356820727507e-06,
      "loss": 0.958,
      "step": 5059
    },
    {
      "epoch": 0.4088971494373624,
      "grad_norm": 2.8398070335388184,
      "learning_rate": 9.223006514164174e-06,
      "loss": 0.9628,
      "step": 5060
    },
    {
      "epoch": 0.4089779591506899,
      "grad_norm": 2.6341593265533447,
      "learning_rate": 9.222656135270904e-06,
      "loss": 0.9338,
      "step": 5061
    },
    {
      "epoch": 0.40905876886401743,
      "grad_norm": 2.7226674556732178,
      "learning_rate": 9.2223056840537e-06,
      "loss": 0.9315,
      "step": 5062
    },
    {
      "epoch": 0.409139578577345,
      "grad_norm": 2.8942036628723145,
      "learning_rate": 9.221955160518567e-06,
      "loss": 0.8611,
      "step": 5063
    },
    {
      "epoch": 0.40922038829067253,
      "grad_norm": 2.848767042160034,
      "learning_rate": 9.221604564671505e-06,
      "loss": 1.0153,
      "step": 5064
    },
    {
      "epoch": 0.40930119800400006,
      "grad_norm": 2.7654354572296143,
      "learning_rate": 9.221253896518519e-06,
      "loss": 0.9566,
      "step": 5065
    },
    {
      "epoch": 0.40938200771732763,
      "grad_norm": 3.298430919647217,
      "learning_rate": 9.220903156065617e-06,
      "loss": 0.9473,
      "step": 5066
    },
    {
      "epoch": 0.40946281743065516,
      "grad_norm": 2.4986090660095215,
      "learning_rate": 9.220552343318804e-06,
      "loss": 0.9936,
      "step": 5067
    },
    {
      "epoch": 0.4095436271439827,
      "grad_norm": 2.3295962810516357,
      "learning_rate": 9.220201458284091e-06,
      "loss": 1.0303,
      "step": 5068
    },
    {
      "epoch": 0.40962443685731026,
      "grad_norm": 2.584465980529785,
      "learning_rate": 9.219850500967487e-06,
      "loss": 0.8988,
      "step": 5069
    },
    {
      "epoch": 0.4097052465706378,
      "grad_norm": 2.6033895015716553,
      "learning_rate": 9.219499471375002e-06,
      "loss": 1.0548,
      "step": 5070
    },
    {
      "epoch": 0.4097860562839653,
      "grad_norm": 2.4835457801818848,
      "learning_rate": 9.219148369512649e-06,
      "loss": 0.9786,
      "step": 5071
    },
    {
      "epoch": 0.4098668659972929,
      "grad_norm": 2.6273036003112793,
      "learning_rate": 9.218797195386443e-06,
      "loss": 0.9294,
      "step": 5072
    },
    {
      "epoch": 0.4099476757106204,
      "grad_norm": 3.011119842529297,
      "learning_rate": 9.218445949002395e-06,
      "loss": 0.9466,
      "step": 5073
    },
    {
      "epoch": 0.410028485423948,
      "grad_norm": 2.3667774200439453,
      "learning_rate": 9.218094630366525e-06,
      "loss": 0.9471,
      "step": 5074
    },
    {
      "epoch": 0.4101092951372755,
      "grad_norm": 2.8165531158447266,
      "learning_rate": 9.217743239484848e-06,
      "loss": 0.9382,
      "step": 5075
    },
    {
      "epoch": 0.41019010485060303,
      "grad_norm": 2.526662588119507,
      "learning_rate": 9.217391776363385e-06,
      "loss": 0.9283,
      "step": 5076
    },
    {
      "epoch": 0.4102709145639306,
      "grad_norm": 3.033874750137329,
      "learning_rate": 9.217040241008152e-06,
      "loss": 0.9426,
      "step": 5077
    },
    {
      "epoch": 0.41035172427725813,
      "grad_norm": 2.5313334465026855,
      "learning_rate": 9.216688633425172e-06,
      "loss": 1.0562,
      "step": 5078
    },
    {
      "epoch": 0.41043253399058566,
      "grad_norm": 3.0072054862976074,
      "learning_rate": 9.216336953620467e-06,
      "loss": 0.8155,
      "step": 5079
    },
    {
      "epoch": 0.41051334370391324,
      "grad_norm": 2.9152071475982666,
      "learning_rate": 9.215985201600059e-06,
      "loss": 0.8833,
      "step": 5080
    },
    {
      "epoch": 0.41059415341724076,
      "grad_norm": 2.9390978813171387,
      "learning_rate": 9.215633377369977e-06,
      "loss": 0.8395,
      "step": 5081
    },
    {
      "epoch": 0.4106749631305683,
      "grad_norm": 2.6304209232330322,
      "learning_rate": 9.215281480936242e-06,
      "loss": 0.9726,
      "step": 5082
    },
    {
      "epoch": 0.41075577284389586,
      "grad_norm": 2.772843599319458,
      "learning_rate": 9.214929512304884e-06,
      "loss": 0.8998,
      "step": 5083
    },
    {
      "epoch": 0.4108365825572234,
      "grad_norm": 2.7768871784210205,
      "learning_rate": 9.214577471481929e-06,
      "loss": 1.0487,
      "step": 5084
    },
    {
      "epoch": 0.4109173922705509,
      "grad_norm": 3.6414031982421875,
      "learning_rate": 9.21422535847341e-06,
      "loss": 1.0651,
      "step": 5085
    },
    {
      "epoch": 0.4109982019838785,
      "grad_norm": 3.065429449081421,
      "learning_rate": 9.213873173285354e-06,
      "loss": 1.0332,
      "step": 5086
    },
    {
      "epoch": 0.411079011697206,
      "grad_norm": 2.226344585418701,
      "learning_rate": 9.213520915923798e-06,
      "loss": 1.1063,
      "step": 5087
    },
    {
      "epoch": 0.41115982141053353,
      "grad_norm": 2.8055105209350586,
      "learning_rate": 9.21316858639477e-06,
      "loss": 0.9426,
      "step": 5088
    },
    {
      "epoch": 0.4112406311238611,
      "grad_norm": 3.0142786502838135,
      "learning_rate": 9.212816184704307e-06,
      "loss": 1.1111,
      "step": 5089
    },
    {
      "epoch": 0.41132144083718863,
      "grad_norm": 3.043494701385498,
      "learning_rate": 9.212463710858446e-06,
      "loss": 0.9475,
      "step": 5090
    },
    {
      "epoch": 0.41140225055051616,
      "grad_norm": 2.598656415939331,
      "learning_rate": 9.212111164863223e-06,
      "loss": 0.9964,
      "step": 5091
    },
    {
      "epoch": 0.41148306026384374,
      "grad_norm": 2.5568392276763916,
      "learning_rate": 9.211758546724674e-06,
      "loss": 0.9552,
      "step": 5092
    },
    {
      "epoch": 0.41156386997717126,
      "grad_norm": 2.725928544998169,
      "learning_rate": 9.21140585644884e-06,
      "loss": 0.8849,
      "step": 5093
    },
    {
      "epoch": 0.4116446796904988,
      "grad_norm": 2.8262064456939697,
      "learning_rate": 9.211053094041764e-06,
      "loss": 1.1436,
      "step": 5094
    },
    {
      "epoch": 0.41172548940382636,
      "grad_norm": 2.905388832092285,
      "learning_rate": 9.210700259509487e-06,
      "loss": 0.9966,
      "step": 5095
    },
    {
      "epoch": 0.4118062991171539,
      "grad_norm": 3.468975782394409,
      "learning_rate": 9.210347352858048e-06,
      "loss": 1.01,
      "step": 5096
    },
    {
      "epoch": 0.4118871088304814,
      "grad_norm": 2.6812238693237305,
      "learning_rate": 9.209994374093499e-06,
      "loss": 0.9068,
      "step": 5097
    },
    {
      "epoch": 0.411967918543809,
      "grad_norm": 2.638923406600952,
      "learning_rate": 9.209641323221879e-06,
      "loss": 1.0169,
      "step": 5098
    },
    {
      "epoch": 0.4120487282571365,
      "grad_norm": 2.6545894145965576,
      "learning_rate": 9.209288200249238e-06,
      "loss": 0.961,
      "step": 5099
    },
    {
      "epoch": 0.41212953797046403,
      "grad_norm": 2.69412899017334,
      "learning_rate": 9.208935005181622e-06,
      "loss": 0.9719,
      "step": 5100
    },
    {
      "epoch": 0.4122103476837916,
      "grad_norm": 2.4850449562072754,
      "learning_rate": 9.208581738025084e-06,
      "loss": 0.934,
      "step": 5101
    },
    {
      "epoch": 0.41229115739711913,
      "grad_norm": 2.5823187828063965,
      "learning_rate": 9.208228398785672e-06,
      "loss": 0.9177,
      "step": 5102
    },
    {
      "epoch": 0.41237196711044666,
      "grad_norm": 3.0760159492492676,
      "learning_rate": 9.207874987469439e-06,
      "loss": 0.8992,
      "step": 5103
    },
    {
      "epoch": 0.41245277682377424,
      "grad_norm": 2.7605202198028564,
      "learning_rate": 9.207521504082438e-06,
      "loss": 0.8759,
      "step": 5104
    },
    {
      "epoch": 0.41253358653710176,
      "grad_norm": 2.959604501724243,
      "learning_rate": 9.207167948630721e-06,
      "loss": 0.8524,
      "step": 5105
    },
    {
      "epoch": 0.4126143962504293,
      "grad_norm": 3.298809289932251,
      "learning_rate": 9.206814321120346e-06,
      "loss": 0.986,
      "step": 5106
    },
    {
      "epoch": 0.41269520596375686,
      "grad_norm": 2.623382329940796,
      "learning_rate": 9.206460621557369e-06,
      "loss": 0.9225,
      "step": 5107
    },
    {
      "epoch": 0.4127760156770844,
      "grad_norm": 2.648101806640625,
      "learning_rate": 9.20610684994785e-06,
      "loss": 0.9735,
      "step": 5108
    },
    {
      "epoch": 0.4128568253904119,
      "grad_norm": 2.6998050212860107,
      "learning_rate": 9.205753006297845e-06,
      "loss": 0.8762,
      "step": 5109
    },
    {
      "epoch": 0.4129376351037395,
      "grad_norm": 3.2479190826416016,
      "learning_rate": 9.205399090613415e-06,
      "loss": 0.8798,
      "step": 5110
    },
    {
      "epoch": 0.413018444817067,
      "grad_norm": 2.274042844772339,
      "learning_rate": 9.205045102900624e-06,
      "loss": 0.9613,
      "step": 5111
    },
    {
      "epoch": 0.41309925453039453,
      "grad_norm": 2.6246562004089355,
      "learning_rate": 9.204691043165533e-06,
      "loss": 1.0374,
      "step": 5112
    },
    {
      "epoch": 0.4131800642437221,
      "grad_norm": 2.179845094680786,
      "learning_rate": 9.204336911414207e-06,
      "loss": 1.0956,
      "step": 5113
    },
    {
      "epoch": 0.41326087395704963,
      "grad_norm": 2.4248814582824707,
      "learning_rate": 9.203982707652711e-06,
      "loss": 0.8883,
      "step": 5114
    },
    {
      "epoch": 0.41334168367037716,
      "grad_norm": 2.8744332790374756,
      "learning_rate": 9.203628431887113e-06,
      "loss": 1.0765,
      "step": 5115
    },
    {
      "epoch": 0.41342249338370474,
      "grad_norm": 2.878135919570923,
      "learning_rate": 9.20327408412348e-06,
      "loss": 0.9488,
      "step": 5116
    },
    {
      "epoch": 0.41350330309703226,
      "grad_norm": 2.7689199447631836,
      "learning_rate": 9.202919664367878e-06,
      "loss": 0.9663,
      "step": 5117
    },
    {
      "epoch": 0.4135841128103598,
      "grad_norm": 3.0066959857940674,
      "learning_rate": 9.202565172626383e-06,
      "loss": 1.0099,
      "step": 5118
    },
    {
      "epoch": 0.41366492252368736,
      "grad_norm": 2.8451433181762695,
      "learning_rate": 9.202210608905062e-06,
      "loss": 0.9581,
      "step": 5119
    },
    {
      "epoch": 0.4137457322370149,
      "grad_norm": 2.792628288269043,
      "learning_rate": 9.201855973209992e-06,
      "loss": 1.0241,
      "step": 5120
    },
    {
      "epoch": 0.4138265419503424,
      "grad_norm": 3.0069565773010254,
      "learning_rate": 9.201501265547242e-06,
      "loss": 0.938,
      "step": 5121
    },
    {
      "epoch": 0.41390735166367,
      "grad_norm": 2.465031862258911,
      "learning_rate": 9.201146485922891e-06,
      "loss": 1.0827,
      "step": 5122
    },
    {
      "epoch": 0.4139881613769975,
      "grad_norm": 2.852816343307495,
      "learning_rate": 9.200791634343015e-06,
      "loss": 0.9214,
      "step": 5123
    },
    {
      "epoch": 0.41406897109032503,
      "grad_norm": 2.762470006942749,
      "learning_rate": 9.20043671081369e-06,
      "loss": 1.1156,
      "step": 5124
    },
    {
      "epoch": 0.4141497808036526,
      "grad_norm": 2.724828004837036,
      "learning_rate": 9.200081715341001e-06,
      "loss": 0.9464,
      "step": 5125
    },
    {
      "epoch": 0.41423059051698013,
      "grad_norm": 2.659640073776245,
      "learning_rate": 9.19972664793102e-06,
      "loss": 0.9812,
      "step": 5126
    },
    {
      "epoch": 0.41431140023030766,
      "grad_norm": 2.647005081176758,
      "learning_rate": 9.199371508589831e-06,
      "loss": 1.0383,
      "step": 5127
    },
    {
      "epoch": 0.41439220994363524,
      "grad_norm": 3.1099772453308105,
      "learning_rate": 9.199016297323518e-06,
      "loss": 1.0125,
      "step": 5128
    },
    {
      "epoch": 0.41447301965696276,
      "grad_norm": 2.5530290603637695,
      "learning_rate": 9.198661014138166e-06,
      "loss": 1.0799,
      "step": 5129
    },
    {
      "epoch": 0.4145538293702903,
      "grad_norm": 2.73724102973938,
      "learning_rate": 9.198305659039858e-06,
      "loss": 0.9411,
      "step": 5130
    },
    {
      "epoch": 0.41463463908361786,
      "grad_norm": 2.540130615234375,
      "learning_rate": 9.19795023203468e-06,
      "loss": 0.8835,
      "step": 5131
    },
    {
      "epoch": 0.4147154487969454,
      "grad_norm": 2.7805604934692383,
      "learning_rate": 9.197594733128724e-06,
      "loss": 0.8499,
      "step": 5132
    },
    {
      "epoch": 0.4147962585102729,
      "grad_norm": 3.1604864597320557,
      "learning_rate": 9.197239162328071e-06,
      "loss": 0.9373,
      "step": 5133
    },
    {
      "epoch": 0.4148770682236005,
      "grad_norm": 2.4651713371276855,
      "learning_rate": 9.196883519638818e-06,
      "loss": 0.9582,
      "step": 5134
    },
    {
      "epoch": 0.414957877936928,
      "grad_norm": 2.41593074798584,
      "learning_rate": 9.196527805067054e-06,
      "loss": 0.9802,
      "step": 5135
    },
    {
      "epoch": 0.41503868765025553,
      "grad_norm": 2.9213290214538574,
      "learning_rate": 9.19617201861887e-06,
      "loss": 0.8652,
      "step": 5136
    },
    {
      "epoch": 0.4151194973635831,
      "grad_norm": 2.4444069862365723,
      "learning_rate": 9.195816160300363e-06,
      "loss": 1.009,
      "step": 5137
    },
    {
      "epoch": 0.41520030707691064,
      "grad_norm": 2.550560235977173,
      "learning_rate": 9.195460230117626e-06,
      "loss": 1.154,
      "step": 5138
    },
    {
      "epoch": 0.4152811167902382,
      "grad_norm": 2.699474334716797,
      "learning_rate": 9.195104228076754e-06,
      "loss": 0.9999,
      "step": 5139
    },
    {
      "epoch": 0.41536192650356574,
      "grad_norm": 2.7403440475463867,
      "learning_rate": 9.194748154183849e-06,
      "loss": 1.0574,
      "step": 5140
    },
    {
      "epoch": 0.41544273621689326,
      "grad_norm": 2.607996702194214,
      "learning_rate": 9.194392008445003e-06,
      "loss": 0.9134,
      "step": 5141
    },
    {
      "epoch": 0.41552354593022084,
      "grad_norm": 2.9673011302948,
      "learning_rate": 9.19403579086632e-06,
      "loss": 1.0222,
      "step": 5142
    },
    {
      "epoch": 0.41560435564354836,
      "grad_norm": 2.496568202972412,
      "learning_rate": 9.193679501453902e-06,
      "loss": 0.9699,
      "step": 5143
    },
    {
      "epoch": 0.4156851653568759,
      "grad_norm": 2.9500949382781982,
      "learning_rate": 9.19332314021385e-06,
      "loss": 1.034,
      "step": 5144
    },
    {
      "epoch": 0.41576597507020346,
      "grad_norm": 2.675473690032959,
      "learning_rate": 9.192966707152266e-06,
      "loss": 1.0035,
      "step": 5145
    },
    {
      "epoch": 0.415846784783531,
      "grad_norm": 2.7029106616973877,
      "learning_rate": 9.192610202275259e-06,
      "loss": 0.9405,
      "step": 5146
    },
    {
      "epoch": 0.4159275944968585,
      "grad_norm": 2.850595235824585,
      "learning_rate": 9.19225362558893e-06,
      "loss": 0.8701,
      "step": 5147
    },
    {
      "epoch": 0.4160084042101861,
      "grad_norm": 3.1219558715820312,
      "learning_rate": 9.19189697709939e-06,
      "loss": 1.0485,
      "step": 5148
    },
    {
      "epoch": 0.4160892139235136,
      "grad_norm": 2.5429701805114746,
      "learning_rate": 9.191540256812745e-06,
      "loss": 0.9133,
      "step": 5149
    },
    {
      "epoch": 0.41617002363684114,
      "grad_norm": 2.516554832458496,
      "learning_rate": 9.191183464735107e-06,
      "loss": 1.0231,
      "step": 5150
    },
    {
      "epoch": 0.4162508333501687,
      "grad_norm": 2.6769444942474365,
      "learning_rate": 9.190826600872587e-06,
      "loss": 1.0348,
      "step": 5151
    },
    {
      "epoch": 0.41633164306349624,
      "grad_norm": 2.447317123413086,
      "learning_rate": 9.190469665231296e-06,
      "loss": 0.9809,
      "step": 5152
    },
    {
      "epoch": 0.41641245277682376,
      "grad_norm": 2.7418107986450195,
      "learning_rate": 9.190112657817347e-06,
      "loss": 1.014,
      "step": 5153
    },
    {
      "epoch": 0.41649326249015134,
      "grad_norm": 2.7991063594818115,
      "learning_rate": 9.189755578636856e-06,
      "loss": 0.9121,
      "step": 5154
    },
    {
      "epoch": 0.41657407220347886,
      "grad_norm": 2.72627592086792,
      "learning_rate": 9.18939842769594e-06,
      "loss": 0.9702,
      "step": 5155
    },
    {
      "epoch": 0.4166548819168064,
      "grad_norm": 2.372669219970703,
      "learning_rate": 9.189041205000713e-06,
      "loss": 1.1127,
      "step": 5156
    },
    {
      "epoch": 0.41673569163013396,
      "grad_norm": 3.2202353477478027,
      "learning_rate": 9.188683910557294e-06,
      "loss": 1.0525,
      "step": 5157
    },
    {
      "epoch": 0.4168165013434615,
      "grad_norm": 2.886657238006592,
      "learning_rate": 9.188326544371805e-06,
      "loss": 0.9962,
      "step": 5158
    },
    {
      "epoch": 0.416897311056789,
      "grad_norm": 2.703826904296875,
      "learning_rate": 9.187969106450364e-06,
      "loss": 0.9504,
      "step": 5159
    },
    {
      "epoch": 0.4169781207701166,
      "grad_norm": 2.624884843826294,
      "learning_rate": 9.187611596799094e-06,
      "loss": 1.0139,
      "step": 5160
    },
    {
      "epoch": 0.4170589304834441,
      "grad_norm": 2.1545822620391846,
      "learning_rate": 9.18725401542412e-06,
      "loss": 1.0799,
      "step": 5161
    },
    {
      "epoch": 0.41713974019677164,
      "grad_norm": 2.6193768978118896,
      "learning_rate": 9.186896362331564e-06,
      "loss": 0.8731,
      "step": 5162
    },
    {
      "epoch": 0.4172205499100992,
      "grad_norm": 2.7223222255706787,
      "learning_rate": 9.186538637527554e-06,
      "loss": 1.049,
      "step": 5163
    },
    {
      "epoch": 0.41730135962342674,
      "grad_norm": 2.860353708267212,
      "learning_rate": 9.186180841018216e-06,
      "loss": 0.8993,
      "step": 5164
    },
    {
      "epoch": 0.41738216933675426,
      "grad_norm": 3.40059757232666,
      "learning_rate": 9.185822972809677e-06,
      "loss": 0.9874,
      "step": 5165
    },
    {
      "epoch": 0.41746297905008184,
      "grad_norm": 3.3224143981933594,
      "learning_rate": 9.185465032908068e-06,
      "loss": 1.0188,
      "step": 5166
    },
    {
      "epoch": 0.41754378876340936,
      "grad_norm": 2.659998893737793,
      "learning_rate": 9.185107021319516e-06,
      "loss": 1.0582,
      "step": 5167
    },
    {
      "epoch": 0.4176245984767369,
      "grad_norm": 2.666018486022949,
      "learning_rate": 9.184748938050161e-06,
      "loss": 0.8759,
      "step": 5168
    },
    {
      "epoch": 0.41770540819006446,
      "grad_norm": 2.5733697414398193,
      "learning_rate": 9.184390783106128e-06,
      "loss": 1.017,
      "step": 5169
    },
    {
      "epoch": 0.417786217903392,
      "grad_norm": 2.5420942306518555,
      "learning_rate": 9.184032556493555e-06,
      "loss": 0.9655,
      "step": 5170
    },
    {
      "epoch": 0.4178670276167195,
      "grad_norm": 2.6693360805511475,
      "learning_rate": 9.183674258218577e-06,
      "loss": 0.9056,
      "step": 5171
    },
    {
      "epoch": 0.4179478373300471,
      "grad_norm": 2.8547558784484863,
      "learning_rate": 9.183315888287331e-06,
      "loss": 0.9444,
      "step": 5172
    },
    {
      "epoch": 0.4180286470433746,
      "grad_norm": 2.659933090209961,
      "learning_rate": 9.182957446705956e-06,
      "loss": 0.8854,
      "step": 5173
    },
    {
      "epoch": 0.41810945675670214,
      "grad_norm": 2.836608409881592,
      "learning_rate": 9.182598933480588e-06,
      "loss": 1.0334,
      "step": 5174
    },
    {
      "epoch": 0.4181902664700297,
      "grad_norm": 2.6722848415374756,
      "learning_rate": 9.18224034861737e-06,
      "loss": 0.9776,
      "step": 5175
    },
    {
      "epoch": 0.41827107618335724,
      "grad_norm": 2.8575034141540527,
      "learning_rate": 9.181881692122443e-06,
      "loss": 0.9902,
      "step": 5176
    },
    {
      "epoch": 0.41835188589668476,
      "grad_norm": 2.6821653842926025,
      "learning_rate": 9.18152296400195e-06,
      "loss": 0.955,
      "step": 5177
    },
    {
      "epoch": 0.41843269561001234,
      "grad_norm": 2.884547710418701,
      "learning_rate": 9.181164164262036e-06,
      "loss": 1.0227,
      "step": 5178
    },
    {
      "epoch": 0.41851350532333986,
      "grad_norm": 2.6616051197052,
      "learning_rate": 9.180805292908846e-06,
      "loss": 1.0137,
      "step": 5179
    },
    {
      "epoch": 0.4185943150366674,
      "grad_norm": 2.635457754135132,
      "learning_rate": 9.180446349948523e-06,
      "loss": 1.0195,
      "step": 5180
    },
    {
      "epoch": 0.41867512474999496,
      "grad_norm": 2.7012946605682373,
      "learning_rate": 9.180087335387222e-06,
      "loss": 1.0158,
      "step": 5181
    },
    {
      "epoch": 0.4187559344633225,
      "grad_norm": 2.358668327331543,
      "learning_rate": 9.179728249231086e-06,
      "loss": 1.0879,
      "step": 5182
    },
    {
      "epoch": 0.41883674417665,
      "grad_norm": 2.769690752029419,
      "learning_rate": 9.179369091486268e-06,
      "loss": 0.9497,
      "step": 5183
    },
    {
      "epoch": 0.4189175538899776,
      "grad_norm": 2.7684693336486816,
      "learning_rate": 9.179009862158919e-06,
      "loss": 0.9457,
      "step": 5184
    },
    {
      "epoch": 0.4189983636033051,
      "grad_norm": 2.4689581394195557,
      "learning_rate": 9.178650561255192e-06,
      "loss": 1.1107,
      "step": 5185
    },
    {
      "epoch": 0.41907917331663264,
      "grad_norm": 2.504530668258667,
      "learning_rate": 9.178291188781238e-06,
      "loss": 1.1336,
      "step": 5186
    },
    {
      "epoch": 0.4191599830299602,
      "grad_norm": 2.4523117542266846,
      "learning_rate": 9.177931744743218e-06,
      "loss": 0.944,
      "step": 5187
    },
    {
      "epoch": 0.41924079274328774,
      "grad_norm": 3.249720335006714,
      "learning_rate": 9.177572229147283e-06,
      "loss": 0.9814,
      "step": 5188
    },
    {
      "epoch": 0.41932160245661526,
      "grad_norm": 2.5293381214141846,
      "learning_rate": 9.177212641999595e-06,
      "loss": 0.9223,
      "step": 5189
    },
    {
      "epoch": 0.41940241216994284,
      "grad_norm": 2.4914333820343018,
      "learning_rate": 9.176852983306309e-06,
      "loss": 0.8903,
      "step": 5190
    },
    {
      "epoch": 0.41948322188327036,
      "grad_norm": 3.0650694370269775,
      "learning_rate": 9.176493253073587e-06,
      "loss": 0.9382,
      "step": 5191
    },
    {
      "epoch": 0.4195640315965979,
      "grad_norm": 2.7988922595977783,
      "learning_rate": 9.17613345130759e-06,
      "loss": 0.9149,
      "step": 5192
    },
    {
      "epoch": 0.41964484130992546,
      "grad_norm": 2.7266433238983154,
      "learning_rate": 9.175773578014483e-06,
      "loss": 1.1189,
      "step": 5193
    },
    {
      "epoch": 0.419725651023253,
      "grad_norm": 3.0123531818389893,
      "learning_rate": 9.175413633200422e-06,
      "loss": 1.0823,
      "step": 5194
    },
    {
      "epoch": 0.4198064607365805,
      "grad_norm": 2.556122303009033,
      "learning_rate": 9.175053616871582e-06,
      "loss": 0.9105,
      "step": 5195
    },
    {
      "epoch": 0.4198872704499081,
      "grad_norm": 2.5047812461853027,
      "learning_rate": 9.174693529034122e-06,
      "loss": 1.0527,
      "step": 5196
    },
    {
      "epoch": 0.4199680801632356,
      "grad_norm": 2.6202611923217773,
      "learning_rate": 9.174333369694214e-06,
      "loss": 0.9677,
      "step": 5197
    },
    {
      "epoch": 0.42004888987656314,
      "grad_norm": 2.7027087211608887,
      "learning_rate": 9.173973138858023e-06,
      "loss": 1.0207,
      "step": 5198
    },
    {
      "epoch": 0.4201296995898907,
      "grad_norm": 2.796079397201538,
      "learning_rate": 9.173612836531722e-06,
      "loss": 0.9337,
      "step": 5199
    },
    {
      "epoch": 0.42021050930321824,
      "grad_norm": 2.7054691314697266,
      "learning_rate": 9.173252462721481e-06,
      "loss": 0.9401,
      "step": 5200
    },
    {
      "epoch": 0.42029131901654576,
      "grad_norm": 2.5890510082244873,
      "learning_rate": 9.17289201743347e-06,
      "loss": 0.9646,
      "step": 5201
    },
    {
      "epoch": 0.42037212872987334,
      "grad_norm": 2.548583507537842,
      "learning_rate": 9.172531500673866e-06,
      "loss": 0.9433,
      "step": 5202
    },
    {
      "epoch": 0.42045293844320086,
      "grad_norm": 2.728844165802002,
      "learning_rate": 9.17217091244884e-06,
      "loss": 0.8917,
      "step": 5203
    },
    {
      "epoch": 0.42053374815652844,
      "grad_norm": 3.138631582260132,
      "learning_rate": 9.171810252764575e-06,
      "loss": 1.0581,
      "step": 5204
    },
    {
      "epoch": 0.42061455786985597,
      "grad_norm": 2.84704327583313,
      "learning_rate": 9.17144952162724e-06,
      "loss": 0.9265,
      "step": 5205
    },
    {
      "epoch": 0.4206953675831835,
      "grad_norm": 2.7900187969207764,
      "learning_rate": 9.171088719043018e-06,
      "loss": 1.0159,
      "step": 5206
    },
    {
      "epoch": 0.42077617729651107,
      "grad_norm": 2.5694427490234375,
      "learning_rate": 9.170727845018089e-06,
      "loss": 0.9761,
      "step": 5207
    },
    {
      "epoch": 0.4208569870098386,
      "grad_norm": 2.68499493598938,
      "learning_rate": 9.17036689955863e-06,
      "loss": 0.8771,
      "step": 5208
    },
    {
      "epoch": 0.4209377967231661,
      "grad_norm": 2.537735939025879,
      "learning_rate": 9.170005882670827e-06,
      "loss": 0.8706,
      "step": 5209
    },
    {
      "epoch": 0.4210186064364937,
      "grad_norm": 2.7610819339752197,
      "learning_rate": 9.169644794360862e-06,
      "loss": 1.0456,
      "step": 5210
    },
    {
      "epoch": 0.4210994161498212,
      "grad_norm": 2.8448081016540527,
      "learning_rate": 9.16928363463492e-06,
      "loss": 0.926,
      "step": 5211
    },
    {
      "epoch": 0.42118022586314874,
      "grad_norm": 2.9562947750091553,
      "learning_rate": 9.168922403499187e-06,
      "loss": 0.9373,
      "step": 5212
    },
    {
      "epoch": 0.4212610355764763,
      "grad_norm": 2.7735419273376465,
      "learning_rate": 9.16856110095985e-06,
      "loss": 1.0635,
      "step": 5213
    },
    {
      "epoch": 0.42134184528980384,
      "grad_norm": 2.9197466373443604,
      "learning_rate": 9.168199727023095e-06,
      "loss": 0.9387,
      "step": 5214
    },
    {
      "epoch": 0.42142265500313136,
      "grad_norm": 2.7584502696990967,
      "learning_rate": 9.167838281695114e-06,
      "loss": 0.9189,
      "step": 5215
    },
    {
      "epoch": 0.42150346471645894,
      "grad_norm": 2.570056676864624,
      "learning_rate": 9.167476764982096e-06,
      "loss": 0.973,
      "step": 5216
    },
    {
      "epoch": 0.42158427442978647,
      "grad_norm": 2.8928418159484863,
      "learning_rate": 9.167115176890234e-06,
      "loss": 0.8473,
      "step": 5217
    },
    {
      "epoch": 0.421665084143114,
      "grad_norm": 2.551988124847412,
      "learning_rate": 9.166753517425722e-06,
      "loss": 0.9271,
      "step": 5218
    },
    {
      "epoch": 0.42174589385644157,
      "grad_norm": 2.578200578689575,
      "learning_rate": 9.166391786594752e-06,
      "loss": 0.9274,
      "step": 5219
    },
    {
      "epoch": 0.4218267035697691,
      "grad_norm": 2.5528528690338135,
      "learning_rate": 9.166029984403522e-06,
      "loss": 1.0344,
      "step": 5220
    },
    {
      "epoch": 0.4219075132830966,
      "grad_norm": 2.59887433052063,
      "learning_rate": 9.165668110858227e-06,
      "loss": 1.023,
      "step": 5221
    },
    {
      "epoch": 0.4219883229964242,
      "grad_norm": 2.60282564163208,
      "learning_rate": 9.165306165965067e-06,
      "loss": 0.9438,
      "step": 5222
    },
    {
      "epoch": 0.4220691327097517,
      "grad_norm": 2.800067901611328,
      "learning_rate": 9.164944149730239e-06,
      "loss": 0.9983,
      "step": 5223
    },
    {
      "epoch": 0.42214994242307924,
      "grad_norm": 3.0861284732818604,
      "learning_rate": 9.164582062159944e-06,
      "loss": 0.9616,
      "step": 5224
    },
    {
      "epoch": 0.4222307521364068,
      "grad_norm": 2.7225136756896973,
      "learning_rate": 9.164219903260385e-06,
      "loss": 1.0283,
      "step": 5225
    },
    {
      "epoch": 0.42231156184973434,
      "grad_norm": 2.9631736278533936,
      "learning_rate": 9.163857673037763e-06,
      "loss": 0.9806,
      "step": 5226
    },
    {
      "epoch": 0.42239237156306186,
      "grad_norm": 2.812352180480957,
      "learning_rate": 9.163495371498284e-06,
      "loss": 0.9563,
      "step": 5227
    },
    {
      "epoch": 0.42247318127638944,
      "grad_norm": 2.5974884033203125,
      "learning_rate": 9.163132998648151e-06,
      "loss": 1.0102,
      "step": 5228
    },
    {
      "epoch": 0.42255399098971697,
      "grad_norm": 2.6368088722229004,
      "learning_rate": 9.162770554493574e-06,
      "loss": 1.0636,
      "step": 5229
    },
    {
      "epoch": 0.4226348007030445,
      "grad_norm": 2.7173140048980713,
      "learning_rate": 9.162408039040757e-06,
      "loss": 0.9587,
      "step": 5230
    },
    {
      "epoch": 0.42271561041637207,
      "grad_norm": 2.772019386291504,
      "learning_rate": 9.162045452295912e-06,
      "loss": 0.9674,
      "step": 5231
    },
    {
      "epoch": 0.4227964201296996,
      "grad_norm": 2.743579626083374,
      "learning_rate": 9.161682794265249e-06,
      "loss": 0.8895,
      "step": 5232
    },
    {
      "epoch": 0.4228772298430271,
      "grad_norm": 2.71034836769104,
      "learning_rate": 9.161320064954977e-06,
      "loss": 0.9423,
      "step": 5233
    },
    {
      "epoch": 0.4229580395563547,
      "grad_norm": 2.274339437484741,
      "learning_rate": 9.16095726437131e-06,
      "loss": 0.952,
      "step": 5234
    },
    {
      "epoch": 0.4230388492696822,
      "grad_norm": 2.940167188644409,
      "learning_rate": 9.160594392520464e-06,
      "loss": 1.0118,
      "step": 5235
    },
    {
      "epoch": 0.42311965898300974,
      "grad_norm": 2.619039297103882,
      "learning_rate": 9.160231449408652e-06,
      "loss": 1.0813,
      "step": 5236
    },
    {
      "epoch": 0.4232004686963373,
      "grad_norm": 2.540062427520752,
      "learning_rate": 9.15986843504209e-06,
      "loss": 0.8947,
      "step": 5237
    },
    {
      "epoch": 0.42328127840966484,
      "grad_norm": 2.7829084396362305,
      "learning_rate": 9.159505349426996e-06,
      "loss": 0.9161,
      "step": 5238
    },
    {
      "epoch": 0.42336208812299236,
      "grad_norm": 3.0055229663848877,
      "learning_rate": 9.15914219256959e-06,
      "loss": 0.9497,
      "step": 5239
    },
    {
      "epoch": 0.42344289783631994,
      "grad_norm": 2.353222608566284,
      "learning_rate": 9.158778964476089e-06,
      "loss": 1.1137,
      "step": 5240
    },
    {
      "epoch": 0.42352370754964747,
      "grad_norm": 2.474656820297241,
      "learning_rate": 9.158415665152716e-06,
      "loss": 0.9872,
      "step": 5241
    },
    {
      "epoch": 0.423604517262975,
      "grad_norm": 2.186896324157715,
      "learning_rate": 9.158052294605696e-06,
      "loss": 0.8645,
      "step": 5242
    },
    {
      "epoch": 0.42368532697630257,
      "grad_norm": 2.6836929321289062,
      "learning_rate": 9.15768885284125e-06,
      "loss": 0.9984,
      "step": 5243
    },
    {
      "epoch": 0.4237661366896301,
      "grad_norm": 2.5624165534973145,
      "learning_rate": 9.157325339865602e-06,
      "loss": 0.9355,
      "step": 5244
    },
    {
      "epoch": 0.4238469464029576,
      "grad_norm": 2.7690236568450928,
      "learning_rate": 9.15696175568498e-06,
      "loss": 0.9173,
      "step": 5245
    },
    {
      "epoch": 0.4239277561162852,
      "grad_norm": 3.2101879119873047,
      "learning_rate": 9.156598100305609e-06,
      "loss": 1.0797,
      "step": 5246
    },
    {
      "epoch": 0.4240085658296127,
      "grad_norm": 2.504683494567871,
      "learning_rate": 9.156234373733722e-06,
      "loss": 1.0471,
      "step": 5247
    },
    {
      "epoch": 0.42408937554294024,
      "grad_norm": 2.4415435791015625,
      "learning_rate": 9.155870575975543e-06,
      "loss": 1.0734,
      "step": 5248
    },
    {
      "epoch": 0.4241701852562678,
      "grad_norm": 2.7042734622955322,
      "learning_rate": 9.155506707037307e-06,
      "loss": 0.9438,
      "step": 5249
    },
    {
      "epoch": 0.42425099496959534,
      "grad_norm": 2.7948741912841797,
      "learning_rate": 9.155142766925245e-06,
      "loss": 0.9362,
      "step": 5250
    },
    {
      "epoch": 0.42433180468292286,
      "grad_norm": 2.6793696880340576,
      "learning_rate": 9.15477875564559e-06,
      "loss": 0.9816,
      "step": 5251
    },
    {
      "epoch": 0.42441261439625044,
      "grad_norm": 3.6548120975494385,
      "learning_rate": 9.15441467320458e-06,
      "loss": 0.9709,
      "step": 5252
    },
    {
      "epoch": 0.42449342410957797,
      "grad_norm": 2.507627487182617,
      "learning_rate": 9.154050519608444e-06,
      "loss": 0.959,
      "step": 5253
    },
    {
      "epoch": 0.4245742338229055,
      "grad_norm": 2.5162367820739746,
      "learning_rate": 9.153686294863424e-06,
      "loss": 1.1674,
      "step": 5254
    },
    {
      "epoch": 0.42465504353623307,
      "grad_norm": 2.562340497970581,
      "learning_rate": 9.153321998975759e-06,
      "loss": 1.0999,
      "step": 5255
    },
    {
      "epoch": 0.4247358532495606,
      "grad_norm": 2.9593281745910645,
      "learning_rate": 9.152957631951686e-06,
      "loss": 0.9948,
      "step": 5256
    },
    {
      "epoch": 0.4248166629628881,
      "grad_norm": 2.4677393436431885,
      "learning_rate": 9.152593193797447e-06,
      "loss": 1.0488,
      "step": 5257
    },
    {
      "epoch": 0.4248974726762157,
      "grad_norm": 2.6863720417022705,
      "learning_rate": 9.152228684519285e-06,
      "loss": 1.0273,
      "step": 5258
    },
    {
      "epoch": 0.4249782823895432,
      "grad_norm": 2.892573118209839,
      "learning_rate": 9.151864104123439e-06,
      "loss": 0.9474,
      "step": 5259
    },
    {
      "epoch": 0.42505909210287074,
      "grad_norm": 2.885190010070801,
      "learning_rate": 9.151499452616158e-06,
      "loss": 1.0603,
      "step": 5260
    },
    {
      "epoch": 0.4251399018161983,
      "grad_norm": 2.7721941471099854,
      "learning_rate": 9.151134730003683e-06,
      "loss": 1.0027,
      "step": 5261
    },
    {
      "epoch": 0.42522071152952584,
      "grad_norm": 2.463444709777832,
      "learning_rate": 9.150769936292267e-06,
      "loss": 0.8949,
      "step": 5262
    },
    {
      "epoch": 0.42530152124285336,
      "grad_norm": 2.463534355163574,
      "learning_rate": 9.150405071488153e-06,
      "loss": 1.0008,
      "step": 5263
    },
    {
      "epoch": 0.42538233095618094,
      "grad_norm": 2.7191054821014404,
      "learning_rate": 9.150040135597591e-06,
      "loss": 1.0674,
      "step": 5264
    },
    {
      "epoch": 0.42546314066950847,
      "grad_norm": 2.578448534011841,
      "learning_rate": 9.149675128626833e-06,
      "loss": 1.0489,
      "step": 5265
    },
    {
      "epoch": 0.425543950382836,
      "grad_norm": 2.3779213428497314,
      "learning_rate": 9.149310050582129e-06,
      "loss": 0.8996,
      "step": 5266
    },
    {
      "epoch": 0.42562476009616357,
      "grad_norm": 3.0681824684143066,
      "learning_rate": 9.148944901469736e-06,
      "loss": 0.8802,
      "step": 5267
    },
    {
      "epoch": 0.4257055698094911,
      "grad_norm": 3.037339210510254,
      "learning_rate": 9.148579681295901e-06,
      "loss": 0.9154,
      "step": 5268
    },
    {
      "epoch": 0.42578637952281867,
      "grad_norm": 3.0940604209899902,
      "learning_rate": 9.148214390066885e-06,
      "loss": 0.8757,
      "step": 5269
    },
    {
      "epoch": 0.4258671892361462,
      "grad_norm": 2.309497833251953,
      "learning_rate": 9.147849027788943e-06,
      "loss": 0.843,
      "step": 5270
    },
    {
      "epoch": 0.4259479989494737,
      "grad_norm": 2.5686209201812744,
      "learning_rate": 9.147483594468334e-06,
      "loss": 0.9062,
      "step": 5271
    },
    {
      "epoch": 0.4260288086628013,
      "grad_norm": 3.318406820297241,
      "learning_rate": 9.147118090111316e-06,
      "loss": 0.9611,
      "step": 5272
    },
    {
      "epoch": 0.4261096183761288,
      "grad_norm": 2.794062852859497,
      "learning_rate": 9.146752514724147e-06,
      "loss": 1.0105,
      "step": 5273
    },
    {
      "epoch": 0.42619042808945634,
      "grad_norm": 2.572922468185425,
      "learning_rate": 9.146386868313091e-06,
      "loss": 0.9437,
      "step": 5274
    },
    {
      "epoch": 0.4262712378027839,
      "grad_norm": 2.977999210357666,
      "learning_rate": 9.14602115088441e-06,
      "loss": 0.9243,
      "step": 5275
    },
    {
      "epoch": 0.42635204751611144,
      "grad_norm": 2.479283094406128,
      "learning_rate": 9.145655362444366e-06,
      "loss": 0.8979,
      "step": 5276
    },
    {
      "epoch": 0.42643285722943897,
      "grad_norm": 3.1657676696777344,
      "learning_rate": 9.145289502999228e-06,
      "loss": 0.9842,
      "step": 5277
    },
    {
      "epoch": 0.42651366694276655,
      "grad_norm": 3.2508034706115723,
      "learning_rate": 9.14492357255526e-06,
      "loss": 1.0106,
      "step": 5278
    },
    {
      "epoch": 0.42659447665609407,
      "grad_norm": 2.5886282920837402,
      "learning_rate": 9.144557571118729e-06,
      "loss": 1.1737,
      "step": 5279
    },
    {
      "epoch": 0.4266752863694216,
      "grad_norm": 3.0532402992248535,
      "learning_rate": 9.144191498695904e-06,
      "loss": 0.9356,
      "step": 5280
    },
    {
      "epoch": 0.42675609608274917,
      "grad_norm": 3.0943808555603027,
      "learning_rate": 9.143825355293058e-06,
      "loss": 0.954,
      "step": 5281
    },
    {
      "epoch": 0.4268369057960767,
      "grad_norm": 2.315962314605713,
      "learning_rate": 9.143459140916456e-06,
      "loss": 0.9484,
      "step": 5282
    },
    {
      "epoch": 0.4269177155094042,
      "grad_norm": 2.507899522781372,
      "learning_rate": 9.143092855572375e-06,
      "loss": 0.9995,
      "step": 5283
    },
    {
      "epoch": 0.4269985252227318,
      "grad_norm": 2.505897045135498,
      "learning_rate": 9.14272649926709e-06,
      "loss": 0.917,
      "step": 5284
    },
    {
      "epoch": 0.4270793349360593,
      "grad_norm": 2.5773026943206787,
      "learning_rate": 9.14236007200687e-06,
      "loss": 0.8165,
      "step": 5285
    },
    {
      "epoch": 0.42716014464938684,
      "grad_norm": 2.3543965816497803,
      "learning_rate": 9.141993573797997e-06,
      "loss": 1.1064,
      "step": 5286
    },
    {
      "epoch": 0.4272409543627144,
      "grad_norm": 2.9803483486175537,
      "learning_rate": 9.141627004646743e-06,
      "loss": 1.0708,
      "step": 5287
    },
    {
      "epoch": 0.42732176407604194,
      "grad_norm": 2.6834347248077393,
      "learning_rate": 9.14126036455939e-06,
      "loss": 1.0106,
      "step": 5288
    },
    {
      "epoch": 0.42740257378936947,
      "grad_norm": 2.5566248893737793,
      "learning_rate": 9.140893653542216e-06,
      "loss": 1.0089,
      "step": 5289
    },
    {
      "epoch": 0.42748338350269705,
      "grad_norm": 3.1436350345611572,
      "learning_rate": 9.140526871601503e-06,
      "loss": 0.9301,
      "step": 5290
    },
    {
      "epoch": 0.42756419321602457,
      "grad_norm": 2.8564767837524414,
      "learning_rate": 9.140160018743533e-06,
      "loss": 1.0127,
      "step": 5291
    },
    {
      "epoch": 0.4276450029293521,
      "grad_norm": 2.2093665599823,
      "learning_rate": 9.13979309497459e-06,
      "loss": 1.0052,
      "step": 5292
    },
    {
      "epoch": 0.42772581264267967,
      "grad_norm": 2.702512264251709,
      "learning_rate": 9.139426100300956e-06,
      "loss": 0.9751,
      "step": 5293
    },
    {
      "epoch": 0.4278066223560072,
      "grad_norm": 2.3560664653778076,
      "learning_rate": 9.139059034728918e-06,
      "loss": 0.9461,
      "step": 5294
    },
    {
      "epoch": 0.4278874320693347,
      "grad_norm": 2.5384583473205566,
      "learning_rate": 9.138691898264762e-06,
      "loss": 0.9219,
      "step": 5295
    },
    {
      "epoch": 0.4279682417826623,
      "grad_norm": 2.374020576477051,
      "learning_rate": 9.13832469091478e-06,
      "loss": 1.0367,
      "step": 5296
    },
    {
      "epoch": 0.4280490514959898,
      "grad_norm": 2.316286563873291,
      "learning_rate": 9.137957412685257e-06,
      "loss": 0.9588,
      "step": 5297
    },
    {
      "epoch": 0.42812986120931734,
      "grad_norm": 3.2077367305755615,
      "learning_rate": 9.137590063582486e-06,
      "loss": 1.0781,
      "step": 5298
    },
    {
      "epoch": 0.4282106709226449,
      "grad_norm": 2.6625146865844727,
      "learning_rate": 9.137222643612757e-06,
      "loss": 0.9954,
      "step": 5299
    },
    {
      "epoch": 0.42829148063597244,
      "grad_norm": 2.5936596393585205,
      "learning_rate": 9.136855152782364e-06,
      "loss": 0.8905,
      "step": 5300
    },
    {
      "epoch": 0.42837229034929997,
      "grad_norm": 2.6800479888916016,
      "learning_rate": 9.136487591097603e-06,
      "loss": 0.9877,
      "step": 5301
    },
    {
      "epoch": 0.42845310006262755,
      "grad_norm": 2.628873348236084,
      "learning_rate": 9.136119958564766e-06,
      "loss": 0.984,
      "step": 5302
    },
    {
      "epoch": 0.42853390977595507,
      "grad_norm": 2.3615784645080566,
      "learning_rate": 9.135752255190153e-06,
      "loss": 1.0313,
      "step": 5303
    },
    {
      "epoch": 0.4286147194892826,
      "grad_norm": 3.0959222316741943,
      "learning_rate": 9.13538448098006e-06,
      "loss": 0.9252,
      "step": 5304
    },
    {
      "epoch": 0.42869552920261017,
      "grad_norm": 2.471327304840088,
      "learning_rate": 9.135016635940785e-06,
      "loss": 1.0587,
      "step": 5305
    },
    {
      "epoch": 0.4287763389159377,
      "grad_norm": 3.0087757110595703,
      "learning_rate": 9.134648720078631e-06,
      "loss": 0.8549,
      "step": 5306
    },
    {
      "epoch": 0.4288571486292652,
      "grad_norm": 3.0245444774627686,
      "learning_rate": 9.134280733399898e-06,
      "loss": 0.9498,
      "step": 5307
    },
    {
      "epoch": 0.4289379583425928,
      "grad_norm": 2.6729087829589844,
      "learning_rate": 9.13391267591089e-06,
      "loss": 0.9431,
      "step": 5308
    },
    {
      "epoch": 0.4290187680559203,
      "grad_norm": 2.839160919189453,
      "learning_rate": 9.133544547617907e-06,
      "loss": 0.8525,
      "step": 5309
    },
    {
      "epoch": 0.42909957776924784,
      "grad_norm": 2.6436760425567627,
      "learning_rate": 9.133176348527258e-06,
      "loss": 0.9851,
      "step": 5310
    },
    {
      "epoch": 0.4291803874825754,
      "grad_norm": 2.847926378250122,
      "learning_rate": 9.13280807864525e-06,
      "loss": 0.9391,
      "step": 5311
    },
    {
      "epoch": 0.42926119719590294,
      "grad_norm": 2.718937397003174,
      "learning_rate": 9.132439737978186e-06,
      "loss": 0.902,
      "step": 5312
    },
    {
      "epoch": 0.42934200690923047,
      "grad_norm": 2.855121612548828,
      "learning_rate": 9.132071326532381e-06,
      "loss": 1.0452,
      "step": 5313
    },
    {
      "epoch": 0.42942281662255805,
      "grad_norm": 2.9336607456207275,
      "learning_rate": 9.131702844314139e-06,
      "loss": 0.8561,
      "step": 5314
    },
    {
      "epoch": 0.42950362633588557,
      "grad_norm": 2.779228448867798,
      "learning_rate": 9.131334291329777e-06,
      "loss": 1.0408,
      "step": 5315
    },
    {
      "epoch": 0.4295844360492131,
      "grad_norm": 2.922060251235962,
      "learning_rate": 9.130965667585603e-06,
      "loss": 0.935,
      "step": 5316
    },
    {
      "epoch": 0.42966524576254067,
      "grad_norm": 2.9307093620300293,
      "learning_rate": 9.13059697308793e-06,
      "loss": 1.0235,
      "step": 5317
    },
    {
      "epoch": 0.4297460554758682,
      "grad_norm": 2.2900094985961914,
      "learning_rate": 9.130228207843077e-06,
      "loss": 0.9313,
      "step": 5318
    },
    {
      "epoch": 0.4298268651891957,
      "grad_norm": 2.5423219203948975,
      "learning_rate": 9.129859371857357e-06,
      "loss": 0.9774,
      "step": 5319
    },
    {
      "epoch": 0.4299076749025233,
      "grad_norm": 3.076057195663452,
      "learning_rate": 9.129490465137088e-06,
      "loss": 0.9863,
      "step": 5320
    },
    {
      "epoch": 0.4299884846158508,
      "grad_norm": 2.477181911468506,
      "learning_rate": 9.12912148768859e-06,
      "loss": 0.8986,
      "step": 5321
    },
    {
      "epoch": 0.43006929432917834,
      "grad_norm": 2.6529858112335205,
      "learning_rate": 9.12875243951818e-06,
      "loss": 0.9301,
      "step": 5322
    },
    {
      "epoch": 0.4301501040425059,
      "grad_norm": 2.9500670433044434,
      "learning_rate": 9.128383320632182e-06,
      "loss": 1.0576,
      "step": 5323
    },
    {
      "epoch": 0.43023091375583344,
      "grad_norm": 3.4155526161193848,
      "learning_rate": 9.128014131036915e-06,
      "loss": 1.0253,
      "step": 5324
    },
    {
      "epoch": 0.43031172346916097,
      "grad_norm": 2.9432568550109863,
      "learning_rate": 9.127644870738703e-06,
      "loss": 0.9619,
      "step": 5325
    },
    {
      "epoch": 0.43039253318248855,
      "grad_norm": 2.53903865814209,
      "learning_rate": 9.127275539743873e-06,
      "loss": 1.0469,
      "step": 5326
    },
    {
      "epoch": 0.43047334289581607,
      "grad_norm": 2.6842873096466064,
      "learning_rate": 9.12690613805875e-06,
      "loss": 0.97,
      "step": 5327
    },
    {
      "epoch": 0.4305541526091436,
      "grad_norm": 2.5515427589416504,
      "learning_rate": 9.126536665689656e-06,
      "loss": 0.96,
      "step": 5328
    },
    {
      "epoch": 0.43063496232247117,
      "grad_norm": 2.4751298427581787,
      "learning_rate": 9.126167122642926e-06,
      "loss": 0.976,
      "step": 5329
    },
    {
      "epoch": 0.4307157720357987,
      "grad_norm": 2.612642765045166,
      "learning_rate": 9.125797508924886e-06,
      "loss": 0.9744,
      "step": 5330
    },
    {
      "epoch": 0.4307965817491262,
      "grad_norm": 3.090175151824951,
      "learning_rate": 9.125427824541867e-06,
      "loss": 0.9953,
      "step": 5331
    },
    {
      "epoch": 0.4308773914624538,
      "grad_norm": 3.0229883193969727,
      "learning_rate": 9.1250580695002e-06,
      "loss": 1.028,
      "step": 5332
    },
    {
      "epoch": 0.4309582011757813,
      "grad_norm": 2.398843765258789,
      "learning_rate": 9.124688243806221e-06,
      "loss": 1.0018,
      "step": 5333
    },
    {
      "epoch": 0.4310390108891089,
      "grad_norm": 2.757558584213257,
      "learning_rate": 9.124318347466262e-06,
      "loss": 0.8998,
      "step": 5334
    },
    {
      "epoch": 0.4311198206024364,
      "grad_norm": 3.0316805839538574,
      "learning_rate": 9.123948380486657e-06,
      "loss": 1.109,
      "step": 5335
    },
    {
      "epoch": 0.43120063031576394,
      "grad_norm": 3.376316547393799,
      "learning_rate": 9.123578342873745e-06,
      "loss": 1.1187,
      "step": 5336
    },
    {
      "epoch": 0.4312814400290915,
      "grad_norm": 2.987802505493164,
      "learning_rate": 9.123208234633862e-06,
      "loss": 1.0732,
      "step": 5337
    },
    {
      "epoch": 0.43136224974241905,
      "grad_norm": 3.3925554752349854,
      "learning_rate": 9.12283805577335e-06,
      "loss": 0.9825,
      "step": 5338
    },
    {
      "epoch": 0.43144305945574657,
      "grad_norm": 2.5645298957824707,
      "learning_rate": 9.122467806298546e-06,
      "loss": 1.018,
      "step": 5339
    },
    {
      "epoch": 0.43152386916907415,
      "grad_norm": 2.698145866394043,
      "learning_rate": 9.122097486215793e-06,
      "loss": 0.9105,
      "step": 5340
    },
    {
      "epoch": 0.43160467888240167,
      "grad_norm": 2.5226173400878906,
      "learning_rate": 9.121727095531435e-06,
      "loss": 0.9614,
      "step": 5341
    },
    {
      "epoch": 0.4316854885957292,
      "grad_norm": 3.4001269340515137,
      "learning_rate": 9.121356634251813e-06,
      "loss": 0.9106,
      "step": 5342
    },
    {
      "epoch": 0.4317662983090568,
      "grad_norm": 2.5968616008758545,
      "learning_rate": 9.120986102383274e-06,
      "loss": 0.976,
      "step": 5343
    },
    {
      "epoch": 0.4318471080223843,
      "grad_norm": 2.5773744583129883,
      "learning_rate": 9.120615499932166e-06,
      "loss": 0.9412,
      "step": 5344
    },
    {
      "epoch": 0.4319279177357118,
      "grad_norm": 2.390458583831787,
      "learning_rate": 9.120244826904832e-06,
      "loss": 1.0266,
      "step": 5345
    },
    {
      "epoch": 0.4320087274490394,
      "grad_norm": 2.7882652282714844,
      "learning_rate": 9.119874083307624e-06,
      "loss": 1.0087,
      "step": 5346
    },
    {
      "epoch": 0.4320895371623669,
      "grad_norm": 2.5810043811798096,
      "learning_rate": 9.11950326914689e-06,
      "loss": 0.9264,
      "step": 5347
    },
    {
      "epoch": 0.43217034687569444,
      "grad_norm": 2.6967110633850098,
      "learning_rate": 9.119132384428984e-06,
      "loss": 0.9351,
      "step": 5348
    },
    {
      "epoch": 0.432251156589022,
      "grad_norm": 2.826792001724243,
      "learning_rate": 9.118761429160256e-06,
      "loss": 1.0107,
      "step": 5349
    },
    {
      "epoch": 0.43233196630234955,
      "grad_norm": 2.733144760131836,
      "learning_rate": 9.118390403347059e-06,
      "loss": 1.1316,
      "step": 5350
    },
    {
      "epoch": 0.43241277601567707,
      "grad_norm": 2.64494252204895,
      "learning_rate": 9.118019306995752e-06,
      "loss": 0.9722,
      "step": 5351
    },
    {
      "epoch": 0.43249358572900465,
      "grad_norm": 2.6184942722320557,
      "learning_rate": 9.117648140112685e-06,
      "loss": 0.9555,
      "step": 5352
    },
    {
      "epoch": 0.43257439544233217,
      "grad_norm": 3.931938409805298,
      "learning_rate": 9.11727690270422e-06,
      "loss": 0.9374,
      "step": 5353
    },
    {
      "epoch": 0.4326552051556597,
      "grad_norm": 2.304018497467041,
      "learning_rate": 9.116905594776713e-06,
      "loss": 0.9416,
      "step": 5354
    },
    {
      "epoch": 0.4327360148689873,
      "grad_norm": 2.753777265548706,
      "learning_rate": 9.116534216336524e-06,
      "loss": 1.0664,
      "step": 5355
    },
    {
      "epoch": 0.4328168245823148,
      "grad_norm": 2.5695817470550537,
      "learning_rate": 9.116162767390014e-06,
      "loss": 0.9207,
      "step": 5356
    },
    {
      "epoch": 0.4328976342956423,
      "grad_norm": 2.5012948513031006,
      "learning_rate": 9.115791247943546e-06,
      "loss": 0.9799,
      "step": 5357
    },
    {
      "epoch": 0.4329784440089699,
      "grad_norm": 3.0362002849578857,
      "learning_rate": 9.115419658003482e-06,
      "loss": 1.0229,
      "step": 5358
    },
    {
      "epoch": 0.4330592537222974,
      "grad_norm": 3.1641323566436768,
      "learning_rate": 9.115047997576186e-06,
      "loss": 0.8914,
      "step": 5359
    },
    {
      "epoch": 0.43314006343562494,
      "grad_norm": 2.7601094245910645,
      "learning_rate": 9.114676266668024e-06,
      "loss": 0.8356,
      "step": 5360
    },
    {
      "epoch": 0.4332208731489525,
      "grad_norm": 2.494013786315918,
      "learning_rate": 9.114304465285363e-06,
      "loss": 0.9652,
      "step": 5361
    },
    {
      "epoch": 0.43330168286228005,
      "grad_norm": 2.597374677658081,
      "learning_rate": 9.113932593434573e-06,
      "loss": 0.9342,
      "step": 5362
    },
    {
      "epoch": 0.43338249257560757,
      "grad_norm": 2.848135232925415,
      "learning_rate": 9.11356065112202e-06,
      "loss": 0.9799,
      "step": 5363
    },
    {
      "epoch": 0.43346330228893515,
      "grad_norm": 2.5346763134002686,
      "learning_rate": 9.113188638354078e-06,
      "loss": 1.0089,
      "step": 5364
    },
    {
      "epoch": 0.43354411200226267,
      "grad_norm": 2.888031244277954,
      "learning_rate": 9.112816555137115e-06,
      "loss": 0.8894,
      "step": 5365
    },
    {
      "epoch": 0.4336249217155902,
      "grad_norm": 2.9543514251708984,
      "learning_rate": 9.112444401477506e-06,
      "loss": 0.8563,
      "step": 5366
    },
    {
      "epoch": 0.4337057314289178,
      "grad_norm": 2.573028326034546,
      "learning_rate": 9.112072177381625e-06,
      "loss": 1.01,
      "step": 5367
    },
    {
      "epoch": 0.4337865411422453,
      "grad_norm": 2.6583313941955566,
      "learning_rate": 9.111699882855846e-06,
      "loss": 0.9755,
      "step": 5368
    },
    {
      "epoch": 0.4338673508555728,
      "grad_norm": 2.7696800231933594,
      "learning_rate": 9.111327517906548e-06,
      "loss": 0.9147,
      "step": 5369
    },
    {
      "epoch": 0.4339481605689004,
      "grad_norm": 2.383000373840332,
      "learning_rate": 9.110955082540108e-06,
      "loss": 0.9502,
      "step": 5370
    },
    {
      "epoch": 0.4340289702822279,
      "grad_norm": 2.257112979888916,
      "learning_rate": 9.1105825767629e-06,
      "loss": 0.9889,
      "step": 5371
    },
    {
      "epoch": 0.43410977999555544,
      "grad_norm": 2.764824628829956,
      "learning_rate": 9.110210000581312e-06,
      "loss": 1.05,
      "step": 5372
    },
    {
      "epoch": 0.434190589708883,
      "grad_norm": 2.6023004055023193,
      "learning_rate": 9.109837354001721e-06,
      "loss": 1.1121,
      "step": 5373
    },
    {
      "epoch": 0.43427139942221055,
      "grad_norm": 2.582038640975952,
      "learning_rate": 9.10946463703051e-06,
      "loss": 0.948,
      "step": 5374
    },
    {
      "epoch": 0.43435220913553807,
      "grad_norm": 3.0894858837127686,
      "learning_rate": 9.109091849674063e-06,
      "loss": 0.9489,
      "step": 5375
    },
    {
      "epoch": 0.43443301884886565,
      "grad_norm": 2.3062944412231445,
      "learning_rate": 9.108718991938764e-06,
      "loss": 1.1621,
      "step": 5376
    },
    {
      "epoch": 0.43451382856219317,
      "grad_norm": 2.9417057037353516,
      "learning_rate": 9.108346063831002e-06,
      "loss": 0.7992,
      "step": 5377
    },
    {
      "epoch": 0.4345946382755207,
      "grad_norm": 2.533569097518921,
      "learning_rate": 9.10797306535716e-06,
      "loss": 0.9554,
      "step": 5378
    },
    {
      "epoch": 0.4346754479888483,
      "grad_norm": 2.499107599258423,
      "learning_rate": 9.10759999652363e-06,
      "loss": 1.0426,
      "step": 5379
    },
    {
      "epoch": 0.4347562577021758,
      "grad_norm": 2.9666755199432373,
      "learning_rate": 9.1072268573368e-06,
      "loss": 0.9048,
      "step": 5380
    },
    {
      "epoch": 0.4348370674155033,
      "grad_norm": 2.65811824798584,
      "learning_rate": 9.106853647803062e-06,
      "loss": 0.9305,
      "step": 5381
    },
    {
      "epoch": 0.4349178771288309,
      "grad_norm": 3.227161169052124,
      "learning_rate": 9.106480367928808e-06,
      "loss": 1.0654,
      "step": 5382
    },
    {
      "epoch": 0.4349986868421584,
      "grad_norm": 2.56229567527771,
      "learning_rate": 9.10610701772043e-06,
      "loss": 1.1111,
      "step": 5383
    },
    {
      "epoch": 0.43507949655548595,
      "grad_norm": 2.882650136947632,
      "learning_rate": 9.105733597184327e-06,
      "loss": 0.9176,
      "step": 5384
    },
    {
      "epoch": 0.4351603062688135,
      "grad_norm": 2.5286736488342285,
      "learning_rate": 9.10536010632689e-06,
      "loss": 0.9715,
      "step": 5385
    },
    {
      "epoch": 0.43524111598214105,
      "grad_norm": 2.5907533168792725,
      "learning_rate": 9.104986545154516e-06,
      "loss": 1.0406,
      "step": 5386
    },
    {
      "epoch": 0.43532192569546857,
      "grad_norm": 2.4601783752441406,
      "learning_rate": 9.104612913673607e-06,
      "loss": 0.9509,
      "step": 5387
    },
    {
      "epoch": 0.43540273540879615,
      "grad_norm": 2.891664981842041,
      "learning_rate": 9.10423921189056e-06,
      "loss": 1.0152,
      "step": 5388
    },
    {
      "epoch": 0.43548354512212367,
      "grad_norm": 2.612304925918579,
      "learning_rate": 9.103865439811775e-06,
      "loss": 0.9979,
      "step": 5389
    },
    {
      "epoch": 0.4355643548354512,
      "grad_norm": 2.445415735244751,
      "learning_rate": 9.103491597443656e-06,
      "loss": 0.9951,
      "step": 5390
    },
    {
      "epoch": 0.4356451645487788,
      "grad_norm": 2.978505849838257,
      "learning_rate": 9.103117684792605e-06,
      "loss": 1.0978,
      "step": 5391
    },
    {
      "epoch": 0.4357259742621063,
      "grad_norm": 2.5591397285461426,
      "learning_rate": 9.102743701865023e-06,
      "loss": 0.9738,
      "step": 5392
    },
    {
      "epoch": 0.4358067839754338,
      "grad_norm": 2.7000041007995605,
      "learning_rate": 9.102369648667319e-06,
      "loss": 0.9148,
      "step": 5393
    },
    {
      "epoch": 0.4358875936887614,
      "grad_norm": 2.5955159664154053,
      "learning_rate": 9.101995525205901e-06,
      "loss": 0.9332,
      "step": 5394
    },
    {
      "epoch": 0.4359684034020889,
      "grad_norm": 2.9734086990356445,
      "learning_rate": 9.101621331487174e-06,
      "loss": 1.0102,
      "step": 5395
    },
    {
      "epoch": 0.4360492131154165,
      "grad_norm": 2.4886577129364014,
      "learning_rate": 9.101247067517547e-06,
      "loss": 0.9779,
      "step": 5396
    },
    {
      "epoch": 0.436130022828744,
      "grad_norm": 2.8928043842315674,
      "learning_rate": 9.100872733303432e-06,
      "loss": 0.909,
      "step": 5397
    },
    {
      "epoch": 0.43621083254207155,
      "grad_norm": 2.512948751449585,
      "learning_rate": 9.10049832885124e-06,
      "loss": 0.9903,
      "step": 5398
    },
    {
      "epoch": 0.4362916422553991,
      "grad_norm": 2.312678575515747,
      "learning_rate": 9.100123854167381e-06,
      "loss": 1.0323,
      "step": 5399
    },
    {
      "epoch": 0.43637245196872665,
      "grad_norm": 2.5982494354248047,
      "learning_rate": 9.099749309258273e-06,
      "loss": 1.0086,
      "step": 5400
    },
    {
      "epoch": 0.43645326168205417,
      "grad_norm": 2.693448781967163,
      "learning_rate": 9.099374694130329e-06,
      "loss": 0.9778,
      "step": 5401
    },
    {
      "epoch": 0.43653407139538175,
      "grad_norm": 2.573061227798462,
      "learning_rate": 9.099000008789965e-06,
      "loss": 1.0132,
      "step": 5402
    },
    {
      "epoch": 0.4366148811087093,
      "grad_norm": 3.0626840591430664,
      "learning_rate": 9.098625253243598e-06,
      "loss": 0.9552,
      "step": 5403
    },
    {
      "epoch": 0.4366956908220368,
      "grad_norm": 2.446549654006958,
      "learning_rate": 9.098250427497648e-06,
      "loss": 0.8148,
      "step": 5404
    },
    {
      "epoch": 0.4367765005353644,
      "grad_norm": 2.5462405681610107,
      "learning_rate": 9.097875531558534e-06,
      "loss": 0.9274,
      "step": 5405
    },
    {
      "epoch": 0.4368573102486919,
      "grad_norm": 2.6182477474212646,
      "learning_rate": 9.097500565432677e-06,
      "loss": 0.8131,
      "step": 5406
    },
    {
      "epoch": 0.4369381199620194,
      "grad_norm": 2.399460792541504,
      "learning_rate": 9.097125529126501e-06,
      "loss": 0.994,
      "step": 5407
    },
    {
      "epoch": 0.437018929675347,
      "grad_norm": 2.7606916427612305,
      "learning_rate": 9.096750422646427e-06,
      "loss": 1.0381,
      "step": 5408
    },
    {
      "epoch": 0.4370997393886745,
      "grad_norm": 2.4011104106903076,
      "learning_rate": 9.096375245998883e-06,
      "loss": 0.9902,
      "step": 5409
    },
    {
      "epoch": 0.43718054910200205,
      "grad_norm": 2.7909364700317383,
      "learning_rate": 9.09599999919029e-06,
      "loss": 0.8746,
      "step": 5410
    },
    {
      "epoch": 0.4372613588153296,
      "grad_norm": 2.861340045928955,
      "learning_rate": 9.095624682227079e-06,
      "loss": 1.0243,
      "step": 5411
    },
    {
      "epoch": 0.43734216852865715,
      "grad_norm": 2.7483179569244385,
      "learning_rate": 9.095249295115677e-06,
      "loss": 0.843,
      "step": 5412
    },
    {
      "epoch": 0.4374229782419847,
      "grad_norm": 2.6070587635040283,
      "learning_rate": 9.094873837862512e-06,
      "loss": 0.8971,
      "step": 5413
    },
    {
      "epoch": 0.43750378795531225,
      "grad_norm": 2.604743719100952,
      "learning_rate": 9.094498310474018e-06,
      "loss": 1.0212,
      "step": 5414
    },
    {
      "epoch": 0.4375845976686398,
      "grad_norm": 2.5673041343688965,
      "learning_rate": 9.094122712956624e-06,
      "loss": 0.8909,
      "step": 5415
    },
    {
      "epoch": 0.4376654073819673,
      "grad_norm": 2.5342977046966553,
      "learning_rate": 9.093747045316765e-06,
      "loss": 0.8853,
      "step": 5416
    },
    {
      "epoch": 0.4377462170952949,
      "grad_norm": 2.943723440170288,
      "learning_rate": 9.093371307560874e-06,
      "loss": 0.9714,
      "step": 5417
    },
    {
      "epoch": 0.4378270268086224,
      "grad_norm": 2.836357355117798,
      "learning_rate": 9.092995499695387e-06,
      "loss": 1.0187,
      "step": 5418
    },
    {
      "epoch": 0.4379078365219499,
      "grad_norm": 2.8180837631225586,
      "learning_rate": 9.092619621726739e-06,
      "loss": 0.954,
      "step": 5419
    },
    {
      "epoch": 0.4379886462352775,
      "grad_norm": 2.809494972229004,
      "learning_rate": 9.092243673661371e-06,
      "loss": 1.0528,
      "step": 5420
    },
    {
      "epoch": 0.438069455948605,
      "grad_norm": 2.688429117202759,
      "learning_rate": 9.091867655505721e-06,
      "loss": 0.9568,
      "step": 5421
    },
    {
      "epoch": 0.43815026566193255,
      "grad_norm": 2.4685428142547607,
      "learning_rate": 9.091491567266228e-06,
      "loss": 0.9759,
      "step": 5422
    },
    {
      "epoch": 0.4382310753752601,
      "grad_norm": 2.688058614730835,
      "learning_rate": 9.091115408949334e-06,
      "loss": 1.0722,
      "step": 5423
    },
    {
      "epoch": 0.43831188508858765,
      "grad_norm": 2.458096742630005,
      "learning_rate": 9.090739180561482e-06,
      "loss": 0.8386,
      "step": 5424
    },
    {
      "epoch": 0.4383926948019152,
      "grad_norm": 3.4665586948394775,
      "learning_rate": 9.090362882109118e-06,
      "loss": 0.8698,
      "step": 5425
    },
    {
      "epoch": 0.43847350451524275,
      "grad_norm": 2.9267451763153076,
      "learning_rate": 9.089986513598683e-06,
      "loss": 0.8727,
      "step": 5426
    },
    {
      "epoch": 0.4385543142285703,
      "grad_norm": 2.9240384101867676,
      "learning_rate": 9.089610075036625e-06,
      "loss": 1.1244,
      "step": 5427
    },
    {
      "epoch": 0.4386351239418978,
      "grad_norm": 2.644961357116699,
      "learning_rate": 9.089233566429393e-06,
      "loss": 0.9061,
      "step": 5428
    },
    {
      "epoch": 0.4387159336552254,
      "grad_norm": 3.047579526901245,
      "learning_rate": 9.088856987783435e-06,
      "loss": 0.9648,
      "step": 5429
    },
    {
      "epoch": 0.4387967433685529,
      "grad_norm": 2.378953695297241,
      "learning_rate": 9.088480339105198e-06,
      "loss": 0.915,
      "step": 5430
    },
    {
      "epoch": 0.4388775530818804,
      "grad_norm": 2.9312777519226074,
      "learning_rate": 9.088103620401136e-06,
      "loss": 0.9505,
      "step": 5431
    },
    {
      "epoch": 0.438958362795208,
      "grad_norm": 3.608285665512085,
      "learning_rate": 9.087726831677702e-06,
      "loss": 0.9127,
      "step": 5432
    },
    {
      "epoch": 0.4390391725085355,
      "grad_norm": 2.4746925830841064,
      "learning_rate": 9.087349972941348e-06,
      "loss": 0.882,
      "step": 5433
    },
    {
      "epoch": 0.43911998222186305,
      "grad_norm": 2.8620457649230957,
      "learning_rate": 9.086973044198529e-06,
      "loss": 0.944,
      "step": 5434
    },
    {
      "epoch": 0.4392007919351906,
      "grad_norm": 3.0637123584747314,
      "learning_rate": 9.086596045455699e-06,
      "loss": 1.0807,
      "step": 5435
    },
    {
      "epoch": 0.43928160164851815,
      "grad_norm": 2.63630747795105,
      "learning_rate": 9.086218976719318e-06,
      "loss": 0.9779,
      "step": 5436
    },
    {
      "epoch": 0.4393624113618457,
      "grad_norm": 2.4325077533721924,
      "learning_rate": 9.085841837995843e-06,
      "loss": 1.0537,
      "step": 5437
    },
    {
      "epoch": 0.43944322107517325,
      "grad_norm": 3.2006518840789795,
      "learning_rate": 9.085464629291733e-06,
      "loss": 0.9899,
      "step": 5438
    },
    {
      "epoch": 0.4395240307885008,
      "grad_norm": 2.8442983627319336,
      "learning_rate": 9.08508735061345e-06,
      "loss": 0.9736,
      "step": 5439
    },
    {
      "epoch": 0.4396048405018283,
      "grad_norm": 2.949164867401123,
      "learning_rate": 9.084710001967455e-06,
      "loss": 0.9632,
      "step": 5440
    },
    {
      "epoch": 0.4396856502151559,
      "grad_norm": 2.717405080795288,
      "learning_rate": 9.084332583360211e-06,
      "loss": 0.9216,
      "step": 5441
    },
    {
      "epoch": 0.4397664599284834,
      "grad_norm": 2.8537914752960205,
      "learning_rate": 9.083955094798183e-06,
      "loss": 0.9337,
      "step": 5442
    },
    {
      "epoch": 0.4398472696418109,
      "grad_norm": 2.6488921642303467,
      "learning_rate": 9.083577536287836e-06,
      "loss": 0.9685,
      "step": 5443
    },
    {
      "epoch": 0.4399280793551385,
      "grad_norm": 2.626530408859253,
      "learning_rate": 9.083199907835636e-06,
      "loss": 1.0865,
      "step": 5444
    },
    {
      "epoch": 0.440008889068466,
      "grad_norm": 2.767813205718994,
      "learning_rate": 9.082822209448052e-06,
      "loss": 0.9972,
      "step": 5445
    },
    {
      "epoch": 0.44008969878179355,
      "grad_norm": 2.9953784942626953,
      "learning_rate": 9.082444441131552e-06,
      "loss": 0.961,
      "step": 5446
    },
    {
      "epoch": 0.4401705084951211,
      "grad_norm": 2.9615743160247803,
      "learning_rate": 9.082066602892606e-06,
      "loss": 1.048,
      "step": 5447
    },
    {
      "epoch": 0.44025131820844865,
      "grad_norm": 2.9631004333496094,
      "learning_rate": 9.081688694737687e-06,
      "loss": 0.9233,
      "step": 5448
    },
    {
      "epoch": 0.4403321279217762,
      "grad_norm": 3.297285795211792,
      "learning_rate": 9.081310716673268e-06,
      "loss": 0.8017,
      "step": 5449
    },
    {
      "epoch": 0.44041293763510375,
      "grad_norm": 2.679313898086548,
      "learning_rate": 9.08093266870582e-06,
      "loss": 0.9945,
      "step": 5450
    },
    {
      "epoch": 0.4404937473484313,
      "grad_norm": 2.6820168495178223,
      "learning_rate": 9.08055455084182e-06,
      "loss": 0.9967,
      "step": 5451
    },
    {
      "epoch": 0.4405745570617588,
      "grad_norm": 2.633715867996216,
      "learning_rate": 9.080176363087746e-06,
      "loss": 1.0125,
      "step": 5452
    },
    {
      "epoch": 0.4406553667750864,
      "grad_norm": 2.7603092193603516,
      "learning_rate": 9.079798105450073e-06,
      "loss": 0.9295,
      "step": 5453
    },
    {
      "epoch": 0.4407361764884139,
      "grad_norm": 2.804338216781616,
      "learning_rate": 9.07941977793528e-06,
      "loss": 0.9907,
      "step": 5454
    },
    {
      "epoch": 0.4408169862017414,
      "grad_norm": 2.5634734630584717,
      "learning_rate": 9.079041380549846e-06,
      "loss": 0.8885,
      "step": 5455
    },
    {
      "epoch": 0.440897795915069,
      "grad_norm": 2.8201401233673096,
      "learning_rate": 9.078662913300254e-06,
      "loss": 0.8524,
      "step": 5456
    },
    {
      "epoch": 0.4409786056283965,
      "grad_norm": 2.7683045864105225,
      "learning_rate": 9.078284376192985e-06,
      "loss": 0.9688,
      "step": 5457
    },
    {
      "epoch": 0.44105941534172405,
      "grad_norm": 2.4153308868408203,
      "learning_rate": 9.077905769234521e-06,
      "loss": 0.9944,
      "step": 5458
    },
    {
      "epoch": 0.4411402250550516,
      "grad_norm": 2.8334877490997314,
      "learning_rate": 9.07752709243135e-06,
      "loss": 1.0495,
      "step": 5459
    },
    {
      "epoch": 0.44122103476837915,
      "grad_norm": 2.699657678604126,
      "learning_rate": 9.077148345789957e-06,
      "loss": 0.9845,
      "step": 5460
    },
    {
      "epoch": 0.44130184448170673,
      "grad_norm": 3.2312214374542236,
      "learning_rate": 9.076769529316828e-06,
      "loss": 0.9549,
      "step": 5461
    },
    {
      "epoch": 0.44138265419503425,
      "grad_norm": 2.6487627029418945,
      "learning_rate": 9.07639064301845e-06,
      "loss": 1.0104,
      "step": 5462
    },
    {
      "epoch": 0.4414634639083618,
      "grad_norm": 2.9800760746002197,
      "learning_rate": 9.076011686901314e-06,
      "loss": 1.0576,
      "step": 5463
    },
    {
      "epoch": 0.44154427362168935,
      "grad_norm": 2.6835033893585205,
      "learning_rate": 9.075632660971912e-06,
      "loss": 0.9986,
      "step": 5464
    },
    {
      "epoch": 0.4416250833350169,
      "grad_norm": 3.2633471488952637,
      "learning_rate": 9.075253565236733e-06,
      "loss": 1.005,
      "step": 5465
    },
    {
      "epoch": 0.4417058930483444,
      "grad_norm": 2.7454686164855957,
      "learning_rate": 9.07487439970227e-06,
      "loss": 1.0678,
      "step": 5466
    },
    {
      "epoch": 0.441786702761672,
      "grad_norm": 3.266718864440918,
      "learning_rate": 9.07449516437502e-06,
      "loss": 0.9595,
      "step": 5467
    },
    {
      "epoch": 0.4418675124749995,
      "grad_norm": 2.8997883796691895,
      "learning_rate": 9.074115859261477e-06,
      "loss": 0.8997,
      "step": 5468
    },
    {
      "epoch": 0.441948322188327,
      "grad_norm": 2.9768147468566895,
      "learning_rate": 9.073736484368136e-06,
      "loss": 1.0163,
      "step": 5469
    },
    {
      "epoch": 0.4420291319016546,
      "grad_norm": 2.6398308277130127,
      "learning_rate": 9.073357039701497e-06,
      "loss": 0.9092,
      "step": 5470
    },
    {
      "epoch": 0.4421099416149821,
      "grad_norm": 2.6662533283233643,
      "learning_rate": 9.072977525268058e-06,
      "loss": 0.9313,
      "step": 5471
    },
    {
      "epoch": 0.44219075132830965,
      "grad_norm": 2.9424004554748535,
      "learning_rate": 9.07259794107432e-06,
      "loss": 1.0139,
      "step": 5472
    },
    {
      "epoch": 0.44227156104163723,
      "grad_norm": 2.9376235008239746,
      "learning_rate": 9.072218287126781e-06,
      "loss": 0.8727,
      "step": 5473
    },
    {
      "epoch": 0.44235237075496475,
      "grad_norm": 2.708012342453003,
      "learning_rate": 9.07183856343195e-06,
      "loss": 1.0118,
      "step": 5474
    },
    {
      "epoch": 0.4424331804682923,
      "grad_norm": 2.6264543533325195,
      "learning_rate": 9.071458769996323e-06,
      "loss": 1.0235,
      "step": 5475
    },
    {
      "epoch": 0.44251399018161985,
      "grad_norm": 2.2916948795318604,
      "learning_rate": 9.071078906826413e-06,
      "loss": 0.9985,
      "step": 5476
    },
    {
      "epoch": 0.4425947998949474,
      "grad_norm": 2.640254259109497,
      "learning_rate": 9.07069897392872e-06,
      "loss": 0.9172,
      "step": 5477
    },
    {
      "epoch": 0.4426756096082749,
      "grad_norm": 2.9489004611968994,
      "learning_rate": 9.070318971309753e-06,
      "loss": 1.0118,
      "step": 5478
    },
    {
      "epoch": 0.4427564193216025,
      "grad_norm": 2.5072970390319824,
      "learning_rate": 9.069938898976021e-06,
      "loss": 0.9398,
      "step": 5479
    },
    {
      "epoch": 0.44283722903493,
      "grad_norm": 2.9356276988983154,
      "learning_rate": 9.069558756934035e-06,
      "loss": 0.9477,
      "step": 5480
    },
    {
      "epoch": 0.4429180387482575,
      "grad_norm": 2.9397525787353516,
      "learning_rate": 9.069178545190303e-06,
      "loss": 0.863,
      "step": 5481
    },
    {
      "epoch": 0.4429988484615851,
      "grad_norm": 2.6035776138305664,
      "learning_rate": 9.06879826375134e-06,
      "loss": 0.9092,
      "step": 5482
    },
    {
      "epoch": 0.4430796581749126,
      "grad_norm": 2.726120948791504,
      "learning_rate": 9.068417912623658e-06,
      "loss": 0.9508,
      "step": 5483
    },
    {
      "epoch": 0.44316046788824015,
      "grad_norm": 2.9055328369140625,
      "learning_rate": 9.06803749181377e-06,
      "loss": 0.9201,
      "step": 5484
    },
    {
      "epoch": 0.44324127760156773,
      "grad_norm": 2.7960684299468994,
      "learning_rate": 9.067657001328192e-06,
      "loss": 1.0273,
      "step": 5485
    },
    {
      "epoch": 0.44332208731489525,
      "grad_norm": 2.568148374557495,
      "learning_rate": 9.067276441173444e-06,
      "loss": 0.9686,
      "step": 5486
    },
    {
      "epoch": 0.4434028970282228,
      "grad_norm": 2.7288520336151123,
      "learning_rate": 9.066895811356042e-06,
      "loss": 1.0194,
      "step": 5487
    },
    {
      "epoch": 0.44348370674155035,
      "grad_norm": 2.4051589965820312,
      "learning_rate": 9.066515111882506e-06,
      "loss": 1.0438,
      "step": 5488
    },
    {
      "epoch": 0.4435645164548779,
      "grad_norm": 2.540262222290039,
      "learning_rate": 9.066134342759355e-06,
      "loss": 0.9808,
      "step": 5489
    },
    {
      "epoch": 0.4436453261682054,
      "grad_norm": 2.9549736976623535,
      "learning_rate": 9.065753503993111e-06,
      "loss": 0.9709,
      "step": 5490
    },
    {
      "epoch": 0.443726135881533,
      "grad_norm": 2.5096869468688965,
      "learning_rate": 9.0653725955903e-06,
      "loss": 0.9466,
      "step": 5491
    },
    {
      "epoch": 0.4438069455948605,
      "grad_norm": 3.1507253646850586,
      "learning_rate": 9.064991617557442e-06,
      "loss": 1.0093,
      "step": 5492
    },
    {
      "epoch": 0.443887755308188,
      "grad_norm": 2.5430915355682373,
      "learning_rate": 9.064610569901062e-06,
      "loss": 1.017,
      "step": 5493
    },
    {
      "epoch": 0.4439685650215156,
      "grad_norm": 2.9738543033599854,
      "learning_rate": 9.06422945262769e-06,
      "loss": 0.9594,
      "step": 5494
    },
    {
      "epoch": 0.4440493747348431,
      "grad_norm": 3.2727372646331787,
      "learning_rate": 9.06384826574385e-06,
      "loss": 0.9934,
      "step": 5495
    },
    {
      "epoch": 0.44413018444817065,
      "grad_norm": 2.7948172092437744,
      "learning_rate": 9.063467009256075e-06,
      "loss": 1.0145,
      "step": 5496
    },
    {
      "epoch": 0.44421099416149823,
      "grad_norm": 2.4964559078216553,
      "learning_rate": 9.063085683170892e-06,
      "loss": 1.1203,
      "step": 5497
    },
    {
      "epoch": 0.44429180387482575,
      "grad_norm": 3.0438883304595947,
      "learning_rate": 9.06270428749483e-06,
      "loss": 1.0103,
      "step": 5498
    },
    {
      "epoch": 0.4443726135881533,
      "grad_norm": 2.4090576171875,
      "learning_rate": 9.062322822234426e-06,
      "loss": 0.9486,
      "step": 5499
    },
    {
      "epoch": 0.44445342330148085,
      "grad_norm": 2.6331546306610107,
      "learning_rate": 9.061941287396211e-06,
      "loss": 0.8768,
      "step": 5500
    },
    {
      "epoch": 0.4445342330148084,
      "grad_norm": 2.906010627746582,
      "learning_rate": 9.061559682986722e-06,
      "loss": 0.9696,
      "step": 5501
    },
    {
      "epoch": 0.4446150427281359,
      "grad_norm": 2.735092878341675,
      "learning_rate": 9.061178009012492e-06,
      "loss": 0.9804,
      "step": 5502
    },
    {
      "epoch": 0.4446958524414635,
      "grad_norm": 2.8308145999908447,
      "learning_rate": 9.06079626548006e-06,
      "loss": 1.0371,
      "step": 5503
    },
    {
      "epoch": 0.444776662154791,
      "grad_norm": 2.8933494091033936,
      "learning_rate": 9.060414452395964e-06,
      "loss": 0.9639,
      "step": 5504
    },
    {
      "epoch": 0.4448574718681185,
      "grad_norm": 2.6455814838409424,
      "learning_rate": 9.060032569766746e-06,
      "loss": 0.9895,
      "step": 5505
    },
    {
      "epoch": 0.4449382815814461,
      "grad_norm": 2.737823963165283,
      "learning_rate": 9.059650617598941e-06,
      "loss": 0.9732,
      "step": 5506
    },
    {
      "epoch": 0.4450190912947736,
      "grad_norm": 2.6841917037963867,
      "learning_rate": 9.059268595899095e-06,
      "loss": 0.9628,
      "step": 5507
    },
    {
      "epoch": 0.44509990100810115,
      "grad_norm": 2.488337516784668,
      "learning_rate": 9.05888650467375e-06,
      "loss": 1.0677,
      "step": 5508
    },
    {
      "epoch": 0.44518071072142873,
      "grad_norm": 3.0691475868225098,
      "learning_rate": 9.05850434392945e-06,
      "loss": 0.994,
      "step": 5509
    },
    {
      "epoch": 0.44526152043475625,
      "grad_norm": 2.6210825443267822,
      "learning_rate": 9.058122113672742e-06,
      "loss": 0.9076,
      "step": 5510
    },
    {
      "epoch": 0.4453423301480838,
      "grad_norm": 3.26194429397583,
      "learning_rate": 9.05773981391017e-06,
      "loss": 0.9735,
      "step": 5511
    },
    {
      "epoch": 0.44542313986141135,
      "grad_norm": 3.310988664627075,
      "learning_rate": 9.057357444648287e-06,
      "loss": 1.0247,
      "step": 5512
    },
    {
      "epoch": 0.4455039495747389,
      "grad_norm": 3.0622739791870117,
      "learning_rate": 9.056975005893638e-06,
      "loss": 1.01,
      "step": 5513
    },
    {
      "epoch": 0.4455847592880664,
      "grad_norm": 3.1104676723480225,
      "learning_rate": 9.056592497652772e-06,
      "loss": 0.8911,
      "step": 5514
    },
    {
      "epoch": 0.445665569001394,
      "grad_norm": 2.6094138622283936,
      "learning_rate": 9.056209919932243e-06,
      "loss": 0.889,
      "step": 5515
    },
    {
      "epoch": 0.4457463787147215,
      "grad_norm": 2.6728122234344482,
      "learning_rate": 9.055827272738601e-06,
      "loss": 1.0073,
      "step": 5516
    },
    {
      "epoch": 0.445827188428049,
      "grad_norm": 2.468600034713745,
      "learning_rate": 9.055444556078406e-06,
      "loss": 1.021,
      "step": 5517
    },
    {
      "epoch": 0.4459079981413766,
      "grad_norm": 2.840611457824707,
      "learning_rate": 9.055061769958206e-06,
      "loss": 0.9477,
      "step": 5518
    },
    {
      "epoch": 0.44598880785470413,
      "grad_norm": 3.3472161293029785,
      "learning_rate": 9.05467891438456e-06,
      "loss": 0.9948,
      "step": 5519
    },
    {
      "epoch": 0.44606961756803165,
      "grad_norm": 3.0067176818847656,
      "learning_rate": 9.054295989364027e-06,
      "loss": 0.9886,
      "step": 5520
    },
    {
      "epoch": 0.44615042728135923,
      "grad_norm": 3.0566020011901855,
      "learning_rate": 9.053912994903163e-06,
      "loss": 0.9347,
      "step": 5521
    },
    {
      "epoch": 0.44623123699468675,
      "grad_norm": 2.7558786869049072,
      "learning_rate": 9.053529931008529e-06,
      "loss": 0.9629,
      "step": 5522
    },
    {
      "epoch": 0.4463120467080143,
      "grad_norm": 2.394063711166382,
      "learning_rate": 9.053146797686685e-06,
      "loss": 0.9981,
      "step": 5523
    },
    {
      "epoch": 0.44639285642134185,
      "grad_norm": 2.907478094100952,
      "learning_rate": 9.052763594944197e-06,
      "loss": 0.8697,
      "step": 5524
    },
    {
      "epoch": 0.4464736661346694,
      "grad_norm": 3.145693778991699,
      "learning_rate": 9.052380322787622e-06,
      "loss": 0.9103,
      "step": 5525
    },
    {
      "epoch": 0.44655447584799696,
      "grad_norm": 2.6979012489318848,
      "learning_rate": 9.051996981223527e-06,
      "loss": 1.0811,
      "step": 5526
    },
    {
      "epoch": 0.4466352855613245,
      "grad_norm": 2.6951351165771484,
      "learning_rate": 9.051613570258481e-06,
      "loss": 1.0288,
      "step": 5527
    },
    {
      "epoch": 0.446716095274652,
      "grad_norm": 2.644151210784912,
      "learning_rate": 9.051230089899048e-06,
      "loss": 0.9223,
      "step": 5528
    },
    {
      "epoch": 0.4467969049879796,
      "grad_norm": 2.6776485443115234,
      "learning_rate": 9.050846540151796e-06,
      "loss": 1.0171,
      "step": 5529
    },
    {
      "epoch": 0.4468777147013071,
      "grad_norm": 3.01741361618042,
      "learning_rate": 9.050462921023295e-06,
      "loss": 0.9938,
      "step": 5530
    },
    {
      "epoch": 0.44695852441463463,
      "grad_norm": 3.307741403579712,
      "learning_rate": 9.050079232520115e-06,
      "loss": 0.9213,
      "step": 5531
    },
    {
      "epoch": 0.4470393341279622,
      "grad_norm": 2.58425235748291,
      "learning_rate": 9.04969547464883e-06,
      "loss": 1.0016,
      "step": 5532
    },
    {
      "epoch": 0.44712014384128973,
      "grad_norm": 3.055528402328491,
      "learning_rate": 9.049311647416006e-06,
      "loss": 0.9972,
      "step": 5533
    },
    {
      "epoch": 0.44720095355461725,
      "grad_norm": 2.4644417762756348,
      "learning_rate": 9.048927750828225e-06,
      "loss": 1.0032,
      "step": 5534
    },
    {
      "epoch": 0.44728176326794483,
      "grad_norm": 2.6862640380859375,
      "learning_rate": 9.048543784892058e-06,
      "loss": 0.9424,
      "step": 5535
    },
    {
      "epoch": 0.44736257298127236,
      "grad_norm": 2.4994943141937256,
      "learning_rate": 9.048159749614084e-06,
      "loss": 1.0303,
      "step": 5536
    },
    {
      "epoch": 0.4474433826945999,
      "grad_norm": 2.360243797302246,
      "learning_rate": 9.047775645000878e-06,
      "loss": 1.034,
      "step": 5537
    },
    {
      "epoch": 0.44752419240792746,
      "grad_norm": 2.4282078742980957,
      "learning_rate": 9.047391471059021e-06,
      "loss": 1.1387,
      "step": 5538
    },
    {
      "epoch": 0.447605002121255,
      "grad_norm": 2.479712724685669,
      "learning_rate": 9.04700722779509e-06,
      "loss": 0.9426,
      "step": 5539
    },
    {
      "epoch": 0.4476858118345825,
      "grad_norm": 2.675340414047241,
      "learning_rate": 9.046622915215668e-06,
      "loss": 1.0598,
      "step": 5540
    },
    {
      "epoch": 0.4477666215479101,
      "grad_norm": 2.6983983516693115,
      "learning_rate": 9.046238533327338e-06,
      "loss": 1.0777,
      "step": 5541
    },
    {
      "epoch": 0.4478474312612376,
      "grad_norm": 2.520012855529785,
      "learning_rate": 9.045854082136683e-06,
      "loss": 1.0144,
      "step": 5542
    },
    {
      "epoch": 0.44792824097456513,
      "grad_norm": 2.781796455383301,
      "learning_rate": 9.045469561650288e-06,
      "loss": 0.9603,
      "step": 5543
    },
    {
      "epoch": 0.4480090506878927,
      "grad_norm": 2.2349886894226074,
      "learning_rate": 9.045084971874738e-06,
      "loss": 1.0555,
      "step": 5544
    },
    {
      "epoch": 0.44808986040122023,
      "grad_norm": 2.447571039199829,
      "learning_rate": 9.044700312816621e-06,
      "loss": 0.9906,
      "step": 5545
    },
    {
      "epoch": 0.44817067011454775,
      "grad_norm": 2.8241076469421387,
      "learning_rate": 9.044315584482524e-06,
      "loss": 0.8948,
      "step": 5546
    },
    {
      "epoch": 0.44825147982787533,
      "grad_norm": 2.5972695350646973,
      "learning_rate": 9.043930786879038e-06,
      "loss": 0.9589,
      "step": 5547
    },
    {
      "epoch": 0.44833228954120286,
      "grad_norm": 2.873607873916626,
      "learning_rate": 9.043545920012753e-06,
      "loss": 0.8906,
      "step": 5548
    },
    {
      "epoch": 0.4484130992545304,
      "grad_norm": 3.0557618141174316,
      "learning_rate": 9.04316098389026e-06,
      "loss": 0.9086,
      "step": 5549
    },
    {
      "epoch": 0.44849390896785796,
      "grad_norm": 2.5151772499084473,
      "learning_rate": 9.042775978518152e-06,
      "loss": 0.8784,
      "step": 5550
    },
    {
      "epoch": 0.4485747186811855,
      "grad_norm": 2.53061580657959,
      "learning_rate": 9.042390903903027e-06,
      "loss": 0.9832,
      "step": 5551
    },
    {
      "epoch": 0.448655528394513,
      "grad_norm": 3.1286022663116455,
      "learning_rate": 9.042005760051476e-06,
      "loss": 0.9332,
      "step": 5552
    },
    {
      "epoch": 0.4487363381078406,
      "grad_norm": 2.4694337844848633,
      "learning_rate": 9.041620546970096e-06,
      "loss": 0.9306,
      "step": 5553
    },
    {
      "epoch": 0.4488171478211681,
      "grad_norm": 2.8130273818969727,
      "learning_rate": 9.041235264665487e-06,
      "loss": 0.9732,
      "step": 5554
    },
    {
      "epoch": 0.44889795753449563,
      "grad_norm": 3.2897021770477295,
      "learning_rate": 9.040849913144245e-06,
      "loss": 0.9042,
      "step": 5555
    },
    {
      "epoch": 0.4489787672478232,
      "grad_norm": 3.095824718475342,
      "learning_rate": 9.040464492412974e-06,
      "loss": 0.9953,
      "step": 5556
    },
    {
      "epoch": 0.44905957696115073,
      "grad_norm": 2.5027883052825928,
      "learning_rate": 9.040079002478274e-06,
      "loss": 0.9512,
      "step": 5557
    },
    {
      "epoch": 0.44914038667447825,
      "grad_norm": 2.522536039352417,
      "learning_rate": 9.039693443346745e-06,
      "loss": 0.9921,
      "step": 5558
    },
    {
      "epoch": 0.44922119638780583,
      "grad_norm": 2.464492082595825,
      "learning_rate": 9.039307815024994e-06,
      "loss": 1.0918,
      "step": 5559
    },
    {
      "epoch": 0.44930200610113336,
      "grad_norm": 2.9177801609039307,
      "learning_rate": 9.038922117519622e-06,
      "loss": 0.7919,
      "step": 5560
    },
    {
      "epoch": 0.4493828158144609,
      "grad_norm": 2.7841975688934326,
      "learning_rate": 9.038536350837239e-06,
      "loss": 1.029,
      "step": 5561
    },
    {
      "epoch": 0.44946362552778846,
      "grad_norm": 2.675217628479004,
      "learning_rate": 9.038150514984452e-06,
      "loss": 0.8656,
      "step": 5562
    },
    {
      "epoch": 0.449544435241116,
      "grad_norm": 2.4436745643615723,
      "learning_rate": 9.037764609967865e-06,
      "loss": 0.9462,
      "step": 5563
    },
    {
      "epoch": 0.4496252449544435,
      "grad_norm": 2.2391719818115234,
      "learning_rate": 9.037378635794093e-06,
      "loss": 0.9355,
      "step": 5564
    },
    {
      "epoch": 0.4497060546677711,
      "grad_norm": 3.132204294204712,
      "learning_rate": 9.036992592469744e-06,
      "loss": 0.9168,
      "step": 5565
    },
    {
      "epoch": 0.4497868643810986,
      "grad_norm": 2.589966297149658,
      "learning_rate": 9.03660648000143e-06,
      "loss": 1.015,
      "step": 5566
    },
    {
      "epoch": 0.44986767409442613,
      "grad_norm": 2.939164876937866,
      "learning_rate": 9.036220298395767e-06,
      "loss": 0.9294,
      "step": 5567
    },
    {
      "epoch": 0.4499484838077537,
      "grad_norm": 2.3374087810516357,
      "learning_rate": 9.035834047659365e-06,
      "loss": 0.9356,
      "step": 5568
    },
    {
      "epoch": 0.45002929352108123,
      "grad_norm": 3.0965662002563477,
      "learning_rate": 9.03544772779884e-06,
      "loss": 0.94,
      "step": 5569
    },
    {
      "epoch": 0.45011010323440875,
      "grad_norm": 2.566420793533325,
      "learning_rate": 9.035061338820815e-06,
      "loss": 1.0781,
      "step": 5570
    },
    {
      "epoch": 0.45019091294773633,
      "grad_norm": 2.7284440994262695,
      "learning_rate": 9.0346748807319e-06,
      "loss": 0.9242,
      "step": 5571
    },
    {
      "epoch": 0.45027172266106386,
      "grad_norm": 2.9208481311798096,
      "learning_rate": 9.034288353538716e-06,
      "loss": 0.9111,
      "step": 5572
    },
    {
      "epoch": 0.4503525323743914,
      "grad_norm": 2.588937282562256,
      "learning_rate": 9.033901757247888e-06,
      "loss": 1.0383,
      "step": 5573
    },
    {
      "epoch": 0.45043334208771896,
      "grad_norm": 2.595132350921631,
      "learning_rate": 9.033515091866033e-06,
      "loss": 1.0119,
      "step": 5574
    },
    {
      "epoch": 0.4505141518010465,
      "grad_norm": 2.634265899658203,
      "learning_rate": 9.033128357399774e-06,
      "loss": 0.9153,
      "step": 5575
    },
    {
      "epoch": 0.450594961514374,
      "grad_norm": 2.527857542037964,
      "learning_rate": 9.032741553855736e-06,
      "loss": 1.0447,
      "step": 5576
    },
    {
      "epoch": 0.4506757712277016,
      "grad_norm": 2.6825711727142334,
      "learning_rate": 9.032354681240543e-06,
      "loss": 0.9012,
      "step": 5577
    },
    {
      "epoch": 0.4507565809410291,
      "grad_norm": 3.1356825828552246,
      "learning_rate": 9.031967739560823e-06,
      "loss": 0.9397,
      "step": 5578
    },
    {
      "epoch": 0.45083739065435663,
      "grad_norm": 3.129138946533203,
      "learning_rate": 9.031580728823201e-06,
      "loss": 0.9757,
      "step": 5579
    },
    {
      "epoch": 0.4509182003676842,
      "grad_norm": 3.3961873054504395,
      "learning_rate": 9.031193649034308e-06,
      "loss": 0.9768,
      "step": 5580
    },
    {
      "epoch": 0.45099901008101173,
      "grad_norm": 2.80542254447937,
      "learning_rate": 9.030806500200773e-06,
      "loss": 0.9601,
      "step": 5581
    },
    {
      "epoch": 0.45107981979433925,
      "grad_norm": 2.480135679244995,
      "learning_rate": 9.030419282329225e-06,
      "loss": 1.1052,
      "step": 5582
    },
    {
      "epoch": 0.45116062950766683,
      "grad_norm": 2.8546664714813232,
      "learning_rate": 9.030031995426299e-06,
      "loss": 0.9511,
      "step": 5583
    },
    {
      "epoch": 0.45124143922099436,
      "grad_norm": 2.225099563598633,
      "learning_rate": 9.029644639498624e-06,
      "loss": 0.9371,
      "step": 5584
    },
    {
      "epoch": 0.4513222489343219,
      "grad_norm": 2.7203240394592285,
      "learning_rate": 9.02925721455284e-06,
      "loss": 1.0717,
      "step": 5585
    },
    {
      "epoch": 0.45140305864764946,
      "grad_norm": 2.8199462890625,
      "learning_rate": 9.02886972059558e-06,
      "loss": 1.0173,
      "step": 5586
    },
    {
      "epoch": 0.451483868360977,
      "grad_norm": 2.914057970046997,
      "learning_rate": 9.02848215763348e-06,
      "loss": 0.8989,
      "step": 5587
    },
    {
      "epoch": 0.4515646780743045,
      "grad_norm": 2.5518972873687744,
      "learning_rate": 9.028094525673176e-06,
      "loss": 1.0232,
      "step": 5588
    },
    {
      "epoch": 0.4516454877876321,
      "grad_norm": 2.5966525077819824,
      "learning_rate": 9.027706824721315e-06,
      "loss": 0.9939,
      "step": 5589
    },
    {
      "epoch": 0.4517262975009596,
      "grad_norm": 2.928921699523926,
      "learning_rate": 9.027319054784529e-06,
      "loss": 1.003,
      "step": 5590
    },
    {
      "epoch": 0.4518071072142872,
      "grad_norm": 2.5647921562194824,
      "learning_rate": 9.026931215869465e-06,
      "loss": 0.9058,
      "step": 5591
    },
    {
      "epoch": 0.4518879169276147,
      "grad_norm": 3.163996696472168,
      "learning_rate": 9.026543307982762e-06,
      "loss": 0.8761,
      "step": 5592
    },
    {
      "epoch": 0.45196872664094223,
      "grad_norm": 2.940082550048828,
      "learning_rate": 9.026155331131066e-06,
      "loss": 1.0036,
      "step": 5593
    },
    {
      "epoch": 0.4520495363542698,
      "grad_norm": 2.8263580799102783,
      "learning_rate": 9.025767285321023e-06,
      "loss": 0.8245,
      "step": 5594
    },
    {
      "epoch": 0.45213034606759733,
      "grad_norm": 2.4488768577575684,
      "learning_rate": 9.025379170559277e-06,
      "loss": 0.9657,
      "step": 5595
    },
    {
      "epoch": 0.45221115578092486,
      "grad_norm": 2.892960548400879,
      "learning_rate": 9.024990986852476e-06,
      "loss": 1.0057,
      "step": 5596
    },
    {
      "epoch": 0.45229196549425243,
      "grad_norm": 4.108729362487793,
      "learning_rate": 9.024602734207271e-06,
      "loss": 0.9929,
      "step": 5597
    },
    {
      "epoch": 0.45237277520757996,
      "grad_norm": 2.586036205291748,
      "learning_rate": 9.024214412630307e-06,
      "loss": 0.9764,
      "step": 5598
    },
    {
      "epoch": 0.4524535849209075,
      "grad_norm": 2.725705862045288,
      "learning_rate": 9.023826022128242e-06,
      "loss": 1.0102,
      "step": 5599
    },
    {
      "epoch": 0.45253439463423506,
      "grad_norm": 3.0589475631713867,
      "learning_rate": 9.023437562707721e-06,
      "loss": 0.9423,
      "step": 5600
    },
    {
      "epoch": 0.4526152043475626,
      "grad_norm": 2.645115852355957,
      "learning_rate": 9.023049034375401e-06,
      "loss": 0.8723,
      "step": 5601
    },
    {
      "epoch": 0.4526960140608901,
      "grad_norm": 2.4714303016662598,
      "learning_rate": 9.022660437137937e-06,
      "loss": 1.0745,
      "step": 5602
    },
    {
      "epoch": 0.4527768237742177,
      "grad_norm": 2.32243275642395,
      "learning_rate": 9.022271771001985e-06,
      "loss": 1.0224,
      "step": 5603
    },
    {
      "epoch": 0.4528576334875452,
      "grad_norm": 2.6025161743164062,
      "learning_rate": 9.021883035974198e-06,
      "loss": 1.0014,
      "step": 5604
    },
    {
      "epoch": 0.45293844320087273,
      "grad_norm": 2.8429667949676514,
      "learning_rate": 9.021494232061239e-06,
      "loss": 0.8874,
      "step": 5605
    },
    {
      "epoch": 0.4530192529142003,
      "grad_norm": 2.6361501216888428,
      "learning_rate": 9.021105359269764e-06,
      "loss": 1.0371,
      "step": 5606
    },
    {
      "epoch": 0.45310006262752783,
      "grad_norm": 2.868471384048462,
      "learning_rate": 9.020716417606435e-06,
      "loss": 1.0487,
      "step": 5607
    },
    {
      "epoch": 0.45318087234085536,
      "grad_norm": 2.3944990634918213,
      "learning_rate": 9.020327407077913e-06,
      "loss": 1.0229,
      "step": 5608
    },
    {
      "epoch": 0.45326168205418293,
      "grad_norm": 2.7799999713897705,
      "learning_rate": 9.019938327690863e-06,
      "loss": 1.0133,
      "step": 5609
    },
    {
      "epoch": 0.45334249176751046,
      "grad_norm": 3.00563907623291,
      "learning_rate": 9.019549179451946e-06,
      "loss": 0.9378,
      "step": 5610
    },
    {
      "epoch": 0.453423301480838,
      "grad_norm": 2.709092378616333,
      "learning_rate": 9.019159962367826e-06,
      "loss": 0.9811,
      "step": 5611
    },
    {
      "epoch": 0.45350411119416556,
      "grad_norm": 2.7960126399993896,
      "learning_rate": 9.018770676445174e-06,
      "loss": 0.9504,
      "step": 5612
    },
    {
      "epoch": 0.4535849209074931,
      "grad_norm": 2.468398094177246,
      "learning_rate": 9.018381321690655e-06,
      "loss": 0.9909,
      "step": 5613
    },
    {
      "epoch": 0.4536657306208206,
      "grad_norm": 2.8399434089660645,
      "learning_rate": 9.017991898110936e-06,
      "loss": 1.0006,
      "step": 5614
    },
    {
      "epoch": 0.4537465403341482,
      "grad_norm": 2.856843948364258,
      "learning_rate": 9.01760240571269e-06,
      "loss": 0.9189,
      "step": 5615
    },
    {
      "epoch": 0.4538273500474757,
      "grad_norm": 3.444997549057007,
      "learning_rate": 9.017212844502587e-06,
      "loss": 1.087,
      "step": 5616
    },
    {
      "epoch": 0.45390815976080323,
      "grad_norm": 2.8802638053894043,
      "learning_rate": 9.016823214487298e-06,
      "loss": 0.9089,
      "step": 5617
    },
    {
      "epoch": 0.4539889694741308,
      "grad_norm": 2.468212604522705,
      "learning_rate": 9.016433515673498e-06,
      "loss": 1.0928,
      "step": 5618
    },
    {
      "epoch": 0.45406977918745833,
      "grad_norm": 2.814220905303955,
      "learning_rate": 9.016043748067861e-06,
      "loss": 0.9596,
      "step": 5619
    },
    {
      "epoch": 0.45415058890078586,
      "grad_norm": 2.472062110900879,
      "learning_rate": 9.015653911677064e-06,
      "loss": 1.0257,
      "step": 5620
    },
    {
      "epoch": 0.45423139861411344,
      "grad_norm": 2.6337456703186035,
      "learning_rate": 9.015264006507781e-06,
      "loss": 0.9351,
      "step": 5621
    },
    {
      "epoch": 0.45431220832744096,
      "grad_norm": 2.545419931411743,
      "learning_rate": 9.014874032566694e-06,
      "loss": 1.0553,
      "step": 5622
    },
    {
      "epoch": 0.4543930180407685,
      "grad_norm": 2.675389528274536,
      "learning_rate": 9.014483989860478e-06,
      "loss": 0.9357,
      "step": 5623
    },
    {
      "epoch": 0.45447382775409606,
      "grad_norm": 3.3827435970306396,
      "learning_rate": 9.014093878395816e-06,
      "loss": 0.9865,
      "step": 5624
    },
    {
      "epoch": 0.4545546374674236,
      "grad_norm": 2.7035064697265625,
      "learning_rate": 9.01370369817939e-06,
      "loss": 0.9868,
      "step": 5625
    },
    {
      "epoch": 0.4546354471807511,
      "grad_norm": 2.7121737003326416,
      "learning_rate": 9.013313449217884e-06,
      "loss": 0.9845,
      "step": 5626
    },
    {
      "epoch": 0.4547162568940787,
      "grad_norm": 2.6461739540100098,
      "learning_rate": 9.012923131517978e-06,
      "loss": 0.908,
      "step": 5627
    },
    {
      "epoch": 0.4547970666074062,
      "grad_norm": 2.8254449367523193,
      "learning_rate": 9.01253274508636e-06,
      "loss": 0.7409,
      "step": 5628
    },
    {
      "epoch": 0.45487787632073373,
      "grad_norm": 2.7166452407836914,
      "learning_rate": 9.012142289929714e-06,
      "loss": 0.9015,
      "step": 5629
    },
    {
      "epoch": 0.4549586860340613,
      "grad_norm": 2.6406447887420654,
      "learning_rate": 9.011751766054732e-06,
      "loss": 0.931,
      "step": 5630
    },
    {
      "epoch": 0.45503949574738883,
      "grad_norm": 2.7557570934295654,
      "learning_rate": 9.0113611734681e-06,
      "loss": 0.7792,
      "step": 5631
    },
    {
      "epoch": 0.45512030546071636,
      "grad_norm": 2.6907272338867188,
      "learning_rate": 9.010970512176509e-06,
      "loss": 1.0254,
      "step": 5632
    },
    {
      "epoch": 0.45520111517404394,
      "grad_norm": 2.2805144786834717,
      "learning_rate": 9.010579782186647e-06,
      "loss": 1.0791,
      "step": 5633
    },
    {
      "epoch": 0.45528192488737146,
      "grad_norm": 3.235880136489868,
      "learning_rate": 9.010188983505208e-06,
      "loss": 0.9581,
      "step": 5634
    },
    {
      "epoch": 0.455362734600699,
      "grad_norm": 3.049842119216919,
      "learning_rate": 9.009798116138889e-06,
      "loss": 0.9207,
      "step": 5635
    },
    {
      "epoch": 0.45544354431402656,
      "grad_norm": 2.42728590965271,
      "learning_rate": 9.009407180094378e-06,
      "loss": 0.8828,
      "step": 5636
    },
    {
      "epoch": 0.4555243540273541,
      "grad_norm": 2.411924362182617,
      "learning_rate": 9.009016175378375e-06,
      "loss": 0.9356,
      "step": 5637
    },
    {
      "epoch": 0.4556051637406816,
      "grad_norm": 2.7910783290863037,
      "learning_rate": 9.008625101997577e-06,
      "loss": 0.9046,
      "step": 5638
    },
    {
      "epoch": 0.4556859734540092,
      "grad_norm": 2.5463552474975586,
      "learning_rate": 9.008233959958682e-06,
      "loss": 0.834,
      "step": 5639
    },
    {
      "epoch": 0.4557667831673367,
      "grad_norm": 2.5790765285491943,
      "learning_rate": 9.007842749268388e-06,
      "loss": 0.9421,
      "step": 5640
    },
    {
      "epoch": 0.45584759288066423,
      "grad_norm": 2.989309787750244,
      "learning_rate": 9.007451469933395e-06,
      "loss": 0.9489,
      "step": 5641
    },
    {
      "epoch": 0.4559284025939918,
      "grad_norm": 2.7662150859832764,
      "learning_rate": 9.007060121960408e-06,
      "loss": 1.1282,
      "step": 5642
    },
    {
      "epoch": 0.45600921230731933,
      "grad_norm": 2.5617337226867676,
      "learning_rate": 9.006668705356128e-06,
      "loss": 1.0608,
      "step": 5643
    },
    {
      "epoch": 0.45609002202064686,
      "grad_norm": 3.037203550338745,
      "learning_rate": 9.006277220127257e-06,
      "loss": 0.9584,
      "step": 5644
    },
    {
      "epoch": 0.45617083173397444,
      "grad_norm": 2.7384161949157715,
      "learning_rate": 9.0058856662805e-06,
      "loss": 1.0149,
      "step": 5645
    },
    {
      "epoch": 0.45625164144730196,
      "grad_norm": 2.435323476791382,
      "learning_rate": 9.00549404382257e-06,
      "loss": 0.9534,
      "step": 5646
    },
    {
      "epoch": 0.4563324511606295,
      "grad_norm": 2.4581611156463623,
      "learning_rate": 9.005102352760166e-06,
      "loss": 0.866,
      "step": 5647
    },
    {
      "epoch": 0.45641326087395706,
      "grad_norm": 2.6652679443359375,
      "learning_rate": 9.004710593100003e-06,
      "loss": 0.9682,
      "step": 5648
    },
    {
      "epoch": 0.4564940705872846,
      "grad_norm": 2.8415911197662354,
      "learning_rate": 9.004318764848787e-06,
      "loss": 0.996,
      "step": 5649
    },
    {
      "epoch": 0.4565748803006121,
      "grad_norm": 2.6814072132110596,
      "learning_rate": 9.003926868013231e-06,
      "loss": 0.8604,
      "step": 5650
    },
    {
      "epoch": 0.4566556900139397,
      "grad_norm": 3.0876829624176025,
      "learning_rate": 9.00353490260005e-06,
      "loss": 1.0219,
      "step": 5651
    },
    {
      "epoch": 0.4567364997272672,
      "grad_norm": 2.7799458503723145,
      "learning_rate": 9.003142868615948e-06,
      "loss": 0.9927,
      "step": 5652
    },
    {
      "epoch": 0.45681730944059473,
      "grad_norm": 9.848539352416992,
      "learning_rate": 9.002750766067649e-06,
      "loss": 0.9253,
      "step": 5653
    },
    {
      "epoch": 0.4568981191539223,
      "grad_norm": 3.060499668121338,
      "learning_rate": 9.002358594961867e-06,
      "loss": 0.8231,
      "step": 5654
    },
    {
      "epoch": 0.45697892886724983,
      "grad_norm": 2.563527822494507,
      "learning_rate": 9.001966355305317e-06,
      "loss": 1.005,
      "step": 5655
    },
    {
      "epoch": 0.4570597385805774,
      "grad_norm": 2.9491777420043945,
      "learning_rate": 9.001574047104716e-06,
      "loss": 1.0011,
      "step": 5656
    },
    {
      "epoch": 0.45714054829390494,
      "grad_norm": 3.1052157878875732,
      "learning_rate": 9.001181670366787e-06,
      "loss": 0.893,
      "step": 5657
    },
    {
      "epoch": 0.45722135800723246,
      "grad_norm": 2.521304130554199,
      "learning_rate": 9.000789225098247e-06,
      "loss": 0.9368,
      "step": 5658
    },
    {
      "epoch": 0.45730216772056004,
      "grad_norm": 2.575448989868164,
      "learning_rate": 9.00039671130582e-06,
      "loss": 0.9664,
      "step": 5659
    },
    {
      "epoch": 0.45738297743388756,
      "grad_norm": 2.4491307735443115,
      "learning_rate": 9.000004128996226e-06,
      "loss": 1.0269,
      "step": 5660
    },
    {
      "epoch": 0.4574637871472151,
      "grad_norm": 2.6399776935577393,
      "learning_rate": 8.999611478176192e-06,
      "loss": 1.001,
      "step": 5661
    },
    {
      "epoch": 0.45754459686054266,
      "grad_norm": 2.724689245223999,
      "learning_rate": 8.999218758852443e-06,
      "loss": 0.9712,
      "step": 5662
    },
    {
      "epoch": 0.4576254065738702,
      "grad_norm": 2.584888219833374,
      "learning_rate": 8.998825971031704e-06,
      "loss": 0.9636,
      "step": 5663
    },
    {
      "epoch": 0.4577062162871977,
      "grad_norm": 3.007192373275757,
      "learning_rate": 8.998433114720701e-06,
      "loss": 0.9979,
      "step": 5664
    },
    {
      "epoch": 0.4577870260005253,
      "grad_norm": 2.7164647579193115,
      "learning_rate": 8.998040189926168e-06,
      "loss": 0.8697,
      "step": 5665
    },
    {
      "epoch": 0.4578678357138528,
      "grad_norm": 2.5364456176757812,
      "learning_rate": 8.997647196654828e-06,
      "loss": 0.9978,
      "step": 5666
    },
    {
      "epoch": 0.45794864542718033,
      "grad_norm": 2.646486282348633,
      "learning_rate": 8.997254134913418e-06,
      "loss": 0.9778,
      "step": 5667
    },
    {
      "epoch": 0.4580294551405079,
      "grad_norm": 2.6803393363952637,
      "learning_rate": 8.996861004708667e-06,
      "loss": 1.1445,
      "step": 5668
    },
    {
      "epoch": 0.45811026485383544,
      "grad_norm": 2.4635229110717773,
      "learning_rate": 8.996467806047309e-06,
      "loss": 1.0915,
      "step": 5669
    },
    {
      "epoch": 0.45819107456716296,
      "grad_norm": 2.5180959701538086,
      "learning_rate": 8.996074538936077e-06,
      "loss": 0.9426,
      "step": 5670
    },
    {
      "epoch": 0.45827188428049054,
      "grad_norm": 2.726747989654541,
      "learning_rate": 8.99568120338171e-06,
      "loss": 1.0266,
      "step": 5671
    },
    {
      "epoch": 0.45835269399381806,
      "grad_norm": 3.1313908100128174,
      "learning_rate": 8.995287799390943e-06,
      "loss": 0.915,
      "step": 5672
    },
    {
      "epoch": 0.4584335037071456,
      "grad_norm": 3.0397677421569824,
      "learning_rate": 8.994894326970514e-06,
      "loss": 0.9487,
      "step": 5673
    },
    {
      "epoch": 0.45851431342047316,
      "grad_norm": 2.721773147583008,
      "learning_rate": 8.994500786127163e-06,
      "loss": 0.9255,
      "step": 5674
    },
    {
      "epoch": 0.4585951231338007,
      "grad_norm": 2.7440669536590576,
      "learning_rate": 8.994107176867628e-06,
      "loss": 1.0435,
      "step": 5675
    },
    {
      "epoch": 0.4586759328471282,
      "grad_norm": 2.8031744956970215,
      "learning_rate": 8.993713499198655e-06,
      "loss": 0.9442,
      "step": 5676
    },
    {
      "epoch": 0.4587567425604558,
      "grad_norm": 2.5146827697753906,
      "learning_rate": 8.993319753126983e-06,
      "loss": 0.9099,
      "step": 5677
    },
    {
      "epoch": 0.4588375522737833,
      "grad_norm": 2.8717031478881836,
      "learning_rate": 8.992925938659357e-06,
      "loss": 1.0509,
      "step": 5678
    },
    {
      "epoch": 0.45891836198711083,
      "grad_norm": 2.7194910049438477,
      "learning_rate": 8.992532055802524e-06,
      "loss": 0.9336,
      "step": 5679
    },
    {
      "epoch": 0.4589991717004384,
      "grad_norm": 2.8881373405456543,
      "learning_rate": 8.992138104563226e-06,
      "loss": 0.9019,
      "step": 5680
    },
    {
      "epoch": 0.45907998141376594,
      "grad_norm": 2.431675910949707,
      "learning_rate": 8.991744084948214e-06,
      "loss": 1.0839,
      "step": 5681
    },
    {
      "epoch": 0.45916079112709346,
      "grad_norm": 2.25947642326355,
      "learning_rate": 8.991349996964236e-06,
      "loss": 0.9717,
      "step": 5682
    },
    {
      "epoch": 0.45924160084042104,
      "grad_norm": 2.704808473587036,
      "learning_rate": 8.990955840618041e-06,
      "loss": 0.948,
      "step": 5683
    },
    {
      "epoch": 0.45932241055374856,
      "grad_norm": 2.8153810501098633,
      "learning_rate": 8.99056161591638e-06,
      "loss": 0.895,
      "step": 5684
    },
    {
      "epoch": 0.4594032202670761,
      "grad_norm": 2.8578639030456543,
      "learning_rate": 8.990167322866005e-06,
      "loss": 1.1398,
      "step": 5685
    },
    {
      "epoch": 0.45948402998040366,
      "grad_norm": 2.7915093898773193,
      "learning_rate": 8.989772961473671e-06,
      "loss": 0.894,
      "step": 5686
    },
    {
      "epoch": 0.4595648396937312,
      "grad_norm": 2.6754133701324463,
      "learning_rate": 8.989378531746131e-06,
      "loss": 1.0066,
      "step": 5687
    },
    {
      "epoch": 0.4596456494070587,
      "grad_norm": 2.9381282329559326,
      "learning_rate": 8.98898403369014e-06,
      "loss": 0.9623,
      "step": 5688
    },
    {
      "epoch": 0.4597264591203863,
      "grad_norm": 3.558082342147827,
      "learning_rate": 8.988589467312455e-06,
      "loss": 0.9892,
      "step": 5689
    },
    {
      "epoch": 0.4598072688337138,
      "grad_norm": 2.568845748901367,
      "learning_rate": 8.988194832619835e-06,
      "loss": 0.9607,
      "step": 5690
    },
    {
      "epoch": 0.45988807854704133,
      "grad_norm": 2.9882869720458984,
      "learning_rate": 8.98780012961904e-06,
      "loss": 0.9085,
      "step": 5691
    },
    {
      "epoch": 0.4599688882603689,
      "grad_norm": 2.629019260406494,
      "learning_rate": 8.987405358316827e-06,
      "loss": 1.0403,
      "step": 5692
    },
    {
      "epoch": 0.46004969797369644,
      "grad_norm": 2.489562511444092,
      "learning_rate": 8.987010518719961e-06,
      "loss": 0.8797,
      "step": 5693
    },
    {
      "epoch": 0.46013050768702396,
      "grad_norm": 2.858153820037842,
      "learning_rate": 8.986615610835203e-06,
      "loss": 1.046,
      "step": 5694
    },
    {
      "epoch": 0.46021131740035154,
      "grad_norm": 2.669794797897339,
      "learning_rate": 8.986220634669318e-06,
      "loss": 0.9201,
      "step": 5695
    },
    {
      "epoch": 0.46029212711367906,
      "grad_norm": 2.7267401218414307,
      "learning_rate": 8.985825590229068e-06,
      "loss": 0.9804,
      "step": 5696
    },
    {
      "epoch": 0.4603729368270066,
      "grad_norm": 2.7089977264404297,
      "learning_rate": 8.985430477521222e-06,
      "loss": 0.9896,
      "step": 5697
    },
    {
      "epoch": 0.46045374654033416,
      "grad_norm": 2.598404884338379,
      "learning_rate": 8.985035296552546e-06,
      "loss": 0.989,
      "step": 5698
    },
    {
      "epoch": 0.4605345562536617,
      "grad_norm": 3.1748383045196533,
      "learning_rate": 8.984640047329809e-06,
      "loss": 0.9024,
      "step": 5699
    },
    {
      "epoch": 0.4606153659669892,
      "grad_norm": 2.6499054431915283,
      "learning_rate": 8.984244729859781e-06,
      "loss": 1.0047,
      "step": 5700
    },
    {
      "epoch": 0.4606961756803168,
      "grad_norm": 3.3875343799591064,
      "learning_rate": 8.983849344149232e-06,
      "loss": 0.9978,
      "step": 5701
    },
    {
      "epoch": 0.4607769853936443,
      "grad_norm": 2.834674835205078,
      "learning_rate": 8.983453890204935e-06,
      "loss": 0.9273,
      "step": 5702
    },
    {
      "epoch": 0.46085779510697183,
      "grad_norm": 2.6363673210144043,
      "learning_rate": 8.983058368033663e-06,
      "loss": 0.9971,
      "step": 5703
    },
    {
      "epoch": 0.4609386048202994,
      "grad_norm": 2.9036641120910645,
      "learning_rate": 8.982662777642188e-06,
      "loss": 0.9269,
      "step": 5704
    },
    {
      "epoch": 0.46101941453362694,
      "grad_norm": 2.6789112091064453,
      "learning_rate": 8.982267119037289e-06,
      "loss": 0.9112,
      "step": 5705
    },
    {
      "epoch": 0.46110022424695446,
      "grad_norm": 2.800849437713623,
      "learning_rate": 8.981871392225742e-06,
      "loss": 0.9262,
      "step": 5706
    },
    {
      "epoch": 0.46118103396028204,
      "grad_norm": 2.4979593753814697,
      "learning_rate": 8.981475597214324e-06,
      "loss": 0.9378,
      "step": 5707
    },
    {
      "epoch": 0.46126184367360956,
      "grad_norm": 2.5978217124938965,
      "learning_rate": 8.981079734009813e-06,
      "loss": 1.0223,
      "step": 5708
    },
    {
      "epoch": 0.4613426533869371,
      "grad_norm": 2.449002981185913,
      "learning_rate": 8.980683802618989e-06,
      "loss": 0.9522,
      "step": 5709
    },
    {
      "epoch": 0.46142346310026466,
      "grad_norm": 2.9029011726379395,
      "learning_rate": 8.980287803048636e-06,
      "loss": 0.9524,
      "step": 5710
    },
    {
      "epoch": 0.4615042728135922,
      "grad_norm": 2.6404201984405518,
      "learning_rate": 8.979891735305534e-06,
      "loss": 0.9854,
      "step": 5711
    },
    {
      "epoch": 0.4615850825269197,
      "grad_norm": 3.219568967819214,
      "learning_rate": 8.97949559939647e-06,
      "loss": 0.8591,
      "step": 5712
    },
    {
      "epoch": 0.4616658922402473,
      "grad_norm": 3.0073201656341553,
      "learning_rate": 8.979099395328226e-06,
      "loss": 1.0269,
      "step": 5713
    },
    {
      "epoch": 0.4617467019535748,
      "grad_norm": 2.707247734069824,
      "learning_rate": 8.978703123107588e-06,
      "loss": 1.0136,
      "step": 5714
    },
    {
      "epoch": 0.46182751166690234,
      "grad_norm": 2.714320659637451,
      "learning_rate": 8.978306782741344e-06,
      "loss": 0.9909,
      "step": 5715
    },
    {
      "epoch": 0.4619083213802299,
      "grad_norm": 3.021829605102539,
      "learning_rate": 8.977910374236281e-06,
      "loss": 0.9573,
      "step": 5716
    },
    {
      "epoch": 0.46198913109355744,
      "grad_norm": 2.5574440956115723,
      "learning_rate": 8.97751389759919e-06,
      "loss": 0.8751,
      "step": 5717
    },
    {
      "epoch": 0.46206994080688496,
      "grad_norm": 2.4902138710021973,
      "learning_rate": 8.977117352836864e-06,
      "loss": 0.8953,
      "step": 5718
    },
    {
      "epoch": 0.46215075052021254,
      "grad_norm": 2.8296992778778076,
      "learning_rate": 8.97672073995609e-06,
      "loss": 1.0601,
      "step": 5719
    },
    {
      "epoch": 0.46223156023354006,
      "grad_norm": 2.5583040714263916,
      "learning_rate": 8.976324058963664e-06,
      "loss": 0.9975,
      "step": 5720
    },
    {
      "epoch": 0.46231236994686764,
      "grad_norm": 2.419672966003418,
      "learning_rate": 8.975927309866379e-06,
      "loss": 1.051,
      "step": 5721
    },
    {
      "epoch": 0.46239317966019516,
      "grad_norm": 2.8298377990722656,
      "learning_rate": 8.975530492671031e-06,
      "loss": 1.0306,
      "step": 5722
    },
    {
      "epoch": 0.4624739893735227,
      "grad_norm": 2.5772652626037598,
      "learning_rate": 8.975133607384416e-06,
      "loss": 0.8895,
      "step": 5723
    },
    {
      "epoch": 0.46255479908685027,
      "grad_norm": 2.7519946098327637,
      "learning_rate": 8.974736654013333e-06,
      "loss": 0.9955,
      "step": 5724
    },
    {
      "epoch": 0.4626356088001778,
      "grad_norm": 3.088197708129883,
      "learning_rate": 8.97433963256458e-06,
      "loss": 0.9018,
      "step": 5725
    },
    {
      "epoch": 0.4627164185135053,
      "grad_norm": 2.506464719772339,
      "learning_rate": 8.973942543044956e-06,
      "loss": 0.9529,
      "step": 5726
    },
    {
      "epoch": 0.4627972282268329,
      "grad_norm": 3.6072235107421875,
      "learning_rate": 8.973545385461261e-06,
      "loss": 0.9706,
      "step": 5727
    },
    {
      "epoch": 0.4628780379401604,
      "grad_norm": 3.1062591075897217,
      "learning_rate": 8.973148159820304e-06,
      "loss": 0.9377,
      "step": 5728
    },
    {
      "epoch": 0.46295884765348794,
      "grad_norm": 2.4257314205169678,
      "learning_rate": 8.972750866128881e-06,
      "loss": 1.0533,
      "step": 5729
    },
    {
      "epoch": 0.4630396573668155,
      "grad_norm": 2.609692335128784,
      "learning_rate": 8.9723535043938e-06,
      "loss": 0.9097,
      "step": 5730
    },
    {
      "epoch": 0.46312046708014304,
      "grad_norm": 3.091174602508545,
      "learning_rate": 8.971956074621867e-06,
      "loss": 0.8879,
      "step": 5731
    },
    {
      "epoch": 0.46320127679347056,
      "grad_norm": 2.596283435821533,
      "learning_rate": 8.97155857681989e-06,
      "loss": 1.0411,
      "step": 5732
    },
    {
      "epoch": 0.46328208650679814,
      "grad_norm": 2.7031702995300293,
      "learning_rate": 8.971161010994674e-06,
      "loss": 0.9502,
      "step": 5733
    },
    {
      "epoch": 0.46336289622012566,
      "grad_norm": 3.0437660217285156,
      "learning_rate": 8.97076337715303e-06,
      "loss": 0.9302,
      "step": 5734
    },
    {
      "epoch": 0.4634437059334532,
      "grad_norm": 2.7243452072143555,
      "learning_rate": 8.970365675301768e-06,
      "loss": 0.9279,
      "step": 5735
    },
    {
      "epoch": 0.46352451564678077,
      "grad_norm": 2.4927990436553955,
      "learning_rate": 8.969967905447702e-06,
      "loss": 1.0101,
      "step": 5736
    },
    {
      "epoch": 0.4636053253601083,
      "grad_norm": 2.4401707649230957,
      "learning_rate": 8.969570067597641e-06,
      "loss": 0.8932,
      "step": 5737
    },
    {
      "epoch": 0.4636861350734358,
      "grad_norm": 2.596168279647827,
      "learning_rate": 8.969172161758404e-06,
      "loss": 0.8357,
      "step": 5738
    },
    {
      "epoch": 0.4637669447867634,
      "grad_norm": 2.832498550415039,
      "learning_rate": 8.968774187936802e-06,
      "loss": 0.9875,
      "step": 5739
    },
    {
      "epoch": 0.4638477545000909,
      "grad_norm": 2.958401918411255,
      "learning_rate": 8.968376146139653e-06,
      "loss": 0.9529,
      "step": 5740
    },
    {
      "epoch": 0.46392856421341844,
      "grad_norm": 2.814657211303711,
      "learning_rate": 8.967978036373773e-06,
      "loss": 1.0036,
      "step": 5741
    },
    {
      "epoch": 0.464009373926746,
      "grad_norm": 2.820213794708252,
      "learning_rate": 8.96757985864598e-06,
      "loss": 0.9722,
      "step": 5742
    },
    {
      "epoch": 0.46409018364007354,
      "grad_norm": 2.4376022815704346,
      "learning_rate": 8.967181612963098e-06,
      "loss": 1.0275,
      "step": 5743
    },
    {
      "epoch": 0.46417099335340106,
      "grad_norm": 2.5551934242248535,
      "learning_rate": 8.966783299331945e-06,
      "loss": 0.9647,
      "step": 5744
    },
    {
      "epoch": 0.46425180306672864,
      "grad_norm": 2.318618059158325,
      "learning_rate": 8.966384917759345e-06,
      "loss": 1.0881,
      "step": 5745
    },
    {
      "epoch": 0.46433261278005616,
      "grad_norm": 2.677311420440674,
      "learning_rate": 8.96598646825212e-06,
      "loss": 1.0155,
      "step": 5746
    },
    {
      "epoch": 0.4644134224933837,
      "grad_norm": 2.4367010593414307,
      "learning_rate": 8.965587950817091e-06,
      "loss": 0.9527,
      "step": 5747
    },
    {
      "epoch": 0.46449423220671127,
      "grad_norm": 2.6899735927581787,
      "learning_rate": 8.965189365461091e-06,
      "loss": 0.8498,
      "step": 5748
    },
    {
      "epoch": 0.4645750419200388,
      "grad_norm": 2.9822683334350586,
      "learning_rate": 8.96479071219094e-06,
      "loss": 0.9261,
      "step": 5749
    },
    {
      "epoch": 0.4646558516333663,
      "grad_norm": 2.724426746368408,
      "learning_rate": 8.964391991013473e-06,
      "loss": 1.0494,
      "step": 5750
    },
    {
      "epoch": 0.4647366613466939,
      "grad_norm": 3.3723952770233154,
      "learning_rate": 8.963993201935513e-06,
      "loss": 1.054,
      "step": 5751
    },
    {
      "epoch": 0.4648174710600214,
      "grad_norm": 2.668436050415039,
      "learning_rate": 8.96359434496389e-06,
      "loss": 0.8919,
      "step": 5752
    },
    {
      "epoch": 0.46489828077334894,
      "grad_norm": 2.999967098236084,
      "learning_rate": 8.96319542010544e-06,
      "loss": 0.8965,
      "step": 5753
    },
    {
      "epoch": 0.4649790904866765,
      "grad_norm": 2.8097987174987793,
      "learning_rate": 8.962796427366993e-06,
      "loss": 0.9619,
      "step": 5754
    },
    {
      "epoch": 0.46505990020000404,
      "grad_norm": 2.3243496417999268,
      "learning_rate": 8.962397366755384e-06,
      "loss": 0.971,
      "step": 5755
    },
    {
      "epoch": 0.46514070991333156,
      "grad_norm": 2.707468032836914,
      "learning_rate": 8.961998238277447e-06,
      "loss": 0.9702,
      "step": 5756
    },
    {
      "epoch": 0.46522151962665914,
      "grad_norm": 2.9703903198242188,
      "learning_rate": 8.961599041940018e-06,
      "loss": 1.0181,
      "step": 5757
    },
    {
      "epoch": 0.46530232933998666,
      "grad_norm": 2.3622820377349854,
      "learning_rate": 8.961199777749935e-06,
      "loss": 1.0661,
      "step": 5758
    },
    {
      "epoch": 0.4653831390533142,
      "grad_norm": 2.6456966400146484,
      "learning_rate": 8.960800445714035e-06,
      "loss": 1.0032,
      "step": 5759
    },
    {
      "epoch": 0.46546394876664177,
      "grad_norm": 3.1365106105804443,
      "learning_rate": 8.96040104583916e-06,
      "loss": 0.9849,
      "step": 5760
    },
    {
      "epoch": 0.4655447584799693,
      "grad_norm": 2.591796636581421,
      "learning_rate": 8.960001578132148e-06,
      "loss": 0.9843,
      "step": 5761
    },
    {
      "epoch": 0.4656255681932968,
      "grad_norm": 2.7226526737213135,
      "learning_rate": 8.959602042599843e-06,
      "loss": 0.9802,
      "step": 5762
    },
    {
      "epoch": 0.4657063779066244,
      "grad_norm": 2.3909027576446533,
      "learning_rate": 8.959202439249087e-06,
      "loss": 0.9329,
      "step": 5763
    },
    {
      "epoch": 0.4657871876199519,
      "grad_norm": 2.6812145709991455,
      "learning_rate": 8.958802768086726e-06,
      "loss": 1.0069,
      "step": 5764
    },
    {
      "epoch": 0.46586799733327944,
      "grad_norm": 3.149233818054199,
      "learning_rate": 8.958403029119602e-06,
      "loss": 0.9595,
      "step": 5765
    },
    {
      "epoch": 0.465948807046607,
      "grad_norm": 2.8651227951049805,
      "learning_rate": 8.958003222354566e-06,
      "loss": 0.9259,
      "step": 5766
    },
    {
      "epoch": 0.46602961675993454,
      "grad_norm": 2.6721975803375244,
      "learning_rate": 8.95760334779846e-06,
      "loss": 0.9779,
      "step": 5767
    },
    {
      "epoch": 0.46611042647326206,
      "grad_norm": 2.89190936088562,
      "learning_rate": 8.957203405458139e-06,
      "loss": 0.868,
      "step": 5768
    },
    {
      "epoch": 0.46619123618658964,
      "grad_norm": 2.8196425437927246,
      "learning_rate": 8.956803395340448e-06,
      "loss": 1.0964,
      "step": 5769
    },
    {
      "epoch": 0.46627204589991716,
      "grad_norm": 2.3658134937286377,
      "learning_rate": 8.956403317452242e-06,
      "loss": 0.828,
      "step": 5770
    },
    {
      "epoch": 0.4663528556132447,
      "grad_norm": 2.655977725982666,
      "learning_rate": 8.95600317180037e-06,
      "loss": 0.9586,
      "step": 5771
    },
    {
      "epoch": 0.46643366532657227,
      "grad_norm": 3.4540302753448486,
      "learning_rate": 8.955602958391691e-06,
      "loss": 1.0272,
      "step": 5772
    },
    {
      "epoch": 0.4665144750398998,
      "grad_norm": 2.621208906173706,
      "learning_rate": 8.955202677233052e-06,
      "loss": 0.9203,
      "step": 5773
    },
    {
      "epoch": 0.4665952847532273,
      "grad_norm": 2.926393747329712,
      "learning_rate": 8.954802328331315e-06,
      "loss": 1.0637,
      "step": 5774
    },
    {
      "epoch": 0.4666760944665549,
      "grad_norm": 2.921645402908325,
      "learning_rate": 8.954401911693336e-06,
      "loss": 1.1009,
      "step": 5775
    },
    {
      "epoch": 0.4667569041798824,
      "grad_norm": 2.4261834621429443,
      "learning_rate": 8.95400142732597e-06,
      "loss": 1.0117,
      "step": 5776
    },
    {
      "epoch": 0.46683771389320994,
      "grad_norm": 3.024146556854248,
      "learning_rate": 8.95360087523608e-06,
      "loss": 0.9313,
      "step": 5777
    },
    {
      "epoch": 0.4669185236065375,
      "grad_norm": 2.688607931137085,
      "learning_rate": 8.953200255430523e-06,
      "loss": 0.9076,
      "step": 5778
    },
    {
      "epoch": 0.46699933331986504,
      "grad_norm": 2.5541460514068604,
      "learning_rate": 8.952799567916164e-06,
      "loss": 0.9786,
      "step": 5779
    },
    {
      "epoch": 0.46708014303319256,
      "grad_norm": 2.9037370681762695,
      "learning_rate": 8.952398812699865e-06,
      "loss": 0.8789,
      "step": 5780
    },
    {
      "epoch": 0.46716095274652014,
      "grad_norm": 3.2507524490356445,
      "learning_rate": 8.951997989788487e-06,
      "loss": 0.9646,
      "step": 5781
    },
    {
      "epoch": 0.46724176245984766,
      "grad_norm": 2.625180959701538,
      "learning_rate": 8.9515970991889e-06,
      "loss": 0.8997,
      "step": 5782
    },
    {
      "epoch": 0.4673225721731752,
      "grad_norm": 2.617286443710327,
      "learning_rate": 8.951196140907967e-06,
      "loss": 0.91,
      "step": 5783
    },
    {
      "epoch": 0.46740338188650277,
      "grad_norm": 2.7201478481292725,
      "learning_rate": 8.950795114952554e-06,
      "loss": 0.9564,
      "step": 5784
    },
    {
      "epoch": 0.4674841915998303,
      "grad_norm": 3.4661874771118164,
      "learning_rate": 8.950394021329535e-06,
      "loss": 0.9577,
      "step": 5785
    },
    {
      "epoch": 0.46756500131315787,
      "grad_norm": 2.390563726425171,
      "learning_rate": 8.949992860045776e-06,
      "loss": 0.958,
      "step": 5786
    },
    {
      "epoch": 0.4676458110264854,
      "grad_norm": 2.7613518238067627,
      "learning_rate": 8.949591631108147e-06,
      "loss": 0.9444,
      "step": 5787
    },
    {
      "epoch": 0.4677266207398129,
      "grad_norm": 3.0035452842712402,
      "learning_rate": 8.949190334523523e-06,
      "loss": 1.0011,
      "step": 5788
    },
    {
      "epoch": 0.4678074304531405,
      "grad_norm": 2.6570937633514404,
      "learning_rate": 8.948788970298774e-06,
      "loss": 0.8661,
      "step": 5789
    },
    {
      "epoch": 0.467888240166468,
      "grad_norm": 2.3081705570220947,
      "learning_rate": 8.948387538440777e-06,
      "loss": 0.9587,
      "step": 5790
    },
    {
      "epoch": 0.46796904987979554,
      "grad_norm": 2.375067710876465,
      "learning_rate": 8.947986038956409e-06,
      "loss": 1.1291,
      "step": 5791
    },
    {
      "epoch": 0.4680498595931231,
      "grad_norm": 2.3014063835144043,
      "learning_rate": 8.947584471852541e-06,
      "loss": 0.9375,
      "step": 5792
    },
    {
      "epoch": 0.46813066930645064,
      "grad_norm": 2.9732635021209717,
      "learning_rate": 8.947182837136057e-06,
      "loss": 0.8834,
      "step": 5793
    },
    {
      "epoch": 0.46821147901977817,
      "grad_norm": 2.5396475791931152,
      "learning_rate": 8.946781134813833e-06,
      "loss": 0.9785,
      "step": 5794
    },
    {
      "epoch": 0.46829228873310574,
      "grad_norm": 2.2900161743164062,
      "learning_rate": 8.94637936489275e-06,
      "loss": 0.8655,
      "step": 5795
    },
    {
      "epoch": 0.46837309844643327,
      "grad_norm": 2.8865981101989746,
      "learning_rate": 8.945977527379688e-06,
      "loss": 0.9418,
      "step": 5796
    },
    {
      "epoch": 0.4684539081597608,
      "grad_norm": 2.7252612113952637,
      "learning_rate": 8.945575622281531e-06,
      "loss": 0.9861,
      "step": 5797
    },
    {
      "epoch": 0.46853471787308837,
      "grad_norm": 2.5759341716766357,
      "learning_rate": 8.945173649605163e-06,
      "loss": 1.0446,
      "step": 5798
    },
    {
      "epoch": 0.4686155275864159,
      "grad_norm": 2.8114752769470215,
      "learning_rate": 8.944771609357466e-06,
      "loss": 0.9281,
      "step": 5799
    },
    {
      "epoch": 0.4686963372997434,
      "grad_norm": 2.4970602989196777,
      "learning_rate": 8.944369501545329e-06,
      "loss": 0.8837,
      "step": 5800
    },
    {
      "epoch": 0.468777147013071,
      "grad_norm": 4.049400329589844,
      "learning_rate": 8.94396732617564e-06,
      "loss": 0.8989,
      "step": 5801
    },
    {
      "epoch": 0.4688579567263985,
      "grad_norm": 2.4196088314056396,
      "learning_rate": 8.943565083255283e-06,
      "loss": 0.9404,
      "step": 5802
    },
    {
      "epoch": 0.46893876643972604,
      "grad_norm": 2.362985610961914,
      "learning_rate": 8.94316277279115e-06,
      "loss": 0.8586,
      "step": 5803
    },
    {
      "epoch": 0.4690195761530536,
      "grad_norm": 2.5610573291778564,
      "learning_rate": 8.942760394790131e-06,
      "loss": 0.8932,
      "step": 5804
    },
    {
      "epoch": 0.46910038586638114,
      "grad_norm": 2.748135805130005,
      "learning_rate": 8.942357949259121e-06,
      "loss": 0.9569,
      "step": 5805
    },
    {
      "epoch": 0.46918119557970867,
      "grad_norm": 2.7132761478424072,
      "learning_rate": 8.941955436205007e-06,
      "loss": 0.9156,
      "step": 5806
    },
    {
      "epoch": 0.46926200529303624,
      "grad_norm": 2.4545814990997314,
      "learning_rate": 8.941552855634688e-06,
      "loss": 0.9154,
      "step": 5807
    },
    {
      "epoch": 0.46934281500636377,
      "grad_norm": 2.44968318939209,
      "learning_rate": 8.941150207555058e-06,
      "loss": 0.9353,
      "step": 5808
    },
    {
      "epoch": 0.4694236247196913,
      "grad_norm": 2.758939266204834,
      "learning_rate": 8.940747491973012e-06,
      "loss": 1.0059,
      "step": 5809
    },
    {
      "epoch": 0.46950443443301887,
      "grad_norm": 2.6677908897399902,
      "learning_rate": 8.940344708895448e-06,
      "loss": 0.9244,
      "step": 5810
    },
    {
      "epoch": 0.4695852441463464,
      "grad_norm": 2.568324089050293,
      "learning_rate": 8.939941858329266e-06,
      "loss": 1.1052,
      "step": 5811
    },
    {
      "epoch": 0.4696660538596739,
      "grad_norm": 2.742100477218628,
      "learning_rate": 8.939538940281365e-06,
      "loss": 0.9692,
      "step": 5812
    },
    {
      "epoch": 0.4697468635730015,
      "grad_norm": 2.638664960861206,
      "learning_rate": 8.939135954758645e-06,
      "loss": 0.9043,
      "step": 5813
    },
    {
      "epoch": 0.469827673286329,
      "grad_norm": 2.781904935836792,
      "learning_rate": 8.93873290176801e-06,
      "loss": 0.9713,
      "step": 5814
    },
    {
      "epoch": 0.46990848299965654,
      "grad_norm": 2.575031280517578,
      "learning_rate": 8.938329781316362e-06,
      "loss": 0.9987,
      "step": 5815
    },
    {
      "epoch": 0.4699892927129841,
      "grad_norm": 2.5054314136505127,
      "learning_rate": 8.937926593410606e-06,
      "loss": 0.9191,
      "step": 5816
    },
    {
      "epoch": 0.47007010242631164,
      "grad_norm": 2.7720255851745605,
      "learning_rate": 8.937523338057648e-06,
      "loss": 0.9486,
      "step": 5817
    },
    {
      "epoch": 0.47015091213963917,
      "grad_norm": 2.47377872467041,
      "learning_rate": 8.937120015264394e-06,
      "loss": 1.0157,
      "step": 5818
    },
    {
      "epoch": 0.47023172185296674,
      "grad_norm": 2.3373284339904785,
      "learning_rate": 8.936716625037752e-06,
      "loss": 1.0121,
      "step": 5819
    },
    {
      "epoch": 0.47031253156629427,
      "grad_norm": 2.8756234645843506,
      "learning_rate": 8.936313167384632e-06,
      "loss": 1.0173,
      "step": 5820
    },
    {
      "epoch": 0.4703933412796218,
      "grad_norm": 3.0548393726348877,
      "learning_rate": 8.935909642311945e-06,
      "loss": 0.8297,
      "step": 5821
    },
    {
      "epoch": 0.47047415099294937,
      "grad_norm": 2.876857042312622,
      "learning_rate": 8.935506049826599e-06,
      "loss": 1.0103,
      "step": 5822
    },
    {
      "epoch": 0.4705549607062769,
      "grad_norm": 2.4455296993255615,
      "learning_rate": 8.935102389935511e-06,
      "loss": 1.02,
      "step": 5823
    },
    {
      "epoch": 0.4706357704196044,
      "grad_norm": 2.428161144256592,
      "learning_rate": 8.93469866264559e-06,
      "loss": 1.0236,
      "step": 5824
    },
    {
      "epoch": 0.470716580132932,
      "grad_norm": 2.6065218448638916,
      "learning_rate": 8.934294867963755e-06,
      "loss": 0.9891,
      "step": 5825
    },
    {
      "epoch": 0.4707973898462595,
      "grad_norm": 2.5996830463409424,
      "learning_rate": 8.93389100589692e-06,
      "loss": 1.0679,
      "step": 5826
    },
    {
      "epoch": 0.47087819955958704,
      "grad_norm": 3.0469918251037598,
      "learning_rate": 8.933487076452002e-06,
      "loss": 0.9552,
      "step": 5827
    },
    {
      "epoch": 0.4709590092729146,
      "grad_norm": 2.888056993484497,
      "learning_rate": 8.93308307963592e-06,
      "loss": 0.9379,
      "step": 5828
    },
    {
      "epoch": 0.47103981898624214,
      "grad_norm": 2.7134406566619873,
      "learning_rate": 8.932679015455594e-06,
      "loss": 0.8655,
      "step": 5829
    },
    {
      "epoch": 0.47112062869956967,
      "grad_norm": 2.543226718902588,
      "learning_rate": 8.932274883917944e-06,
      "loss": 1.1304,
      "step": 5830
    },
    {
      "epoch": 0.47120143841289724,
      "grad_norm": 2.520210027694702,
      "learning_rate": 8.931870685029891e-06,
      "loss": 0.9515,
      "step": 5831
    },
    {
      "epoch": 0.47128224812622477,
      "grad_norm": 2.9487199783325195,
      "learning_rate": 8.931466418798357e-06,
      "loss": 1.0249,
      "step": 5832
    },
    {
      "epoch": 0.4713630578395523,
      "grad_norm": 2.8173553943634033,
      "learning_rate": 8.93106208523027e-06,
      "loss": 0.9164,
      "step": 5833
    },
    {
      "epoch": 0.47144386755287987,
      "grad_norm": 2.3859004974365234,
      "learning_rate": 8.930657684332555e-06,
      "loss": 0.8828,
      "step": 5834
    },
    {
      "epoch": 0.4715246772662074,
      "grad_norm": 2.8427581787109375,
      "learning_rate": 8.930253216112135e-06,
      "loss": 0.8944,
      "step": 5835
    },
    {
      "epoch": 0.4716054869795349,
      "grad_norm": 2.501901865005493,
      "learning_rate": 8.929848680575938e-06,
      "loss": 1.0132,
      "step": 5836
    },
    {
      "epoch": 0.4716862966928625,
      "grad_norm": 2.772751808166504,
      "learning_rate": 8.929444077730894e-06,
      "loss": 1.0731,
      "step": 5837
    },
    {
      "epoch": 0.47176710640619,
      "grad_norm": 2.6158859729766846,
      "learning_rate": 8.929039407583933e-06,
      "loss": 0.998,
      "step": 5838
    },
    {
      "epoch": 0.47184791611951754,
      "grad_norm": 2.7519757747650146,
      "learning_rate": 8.928634670141987e-06,
      "loss": 0.8945,
      "step": 5839
    },
    {
      "epoch": 0.4719287258328451,
      "grad_norm": 2.8849856853485107,
      "learning_rate": 8.928229865411986e-06,
      "loss": 0.9832,
      "step": 5840
    },
    {
      "epoch": 0.47200953554617264,
      "grad_norm": 2.746818780899048,
      "learning_rate": 8.927824993400864e-06,
      "loss": 1.0956,
      "step": 5841
    },
    {
      "epoch": 0.47209034525950017,
      "grad_norm": 2.6034724712371826,
      "learning_rate": 8.927420054115556e-06,
      "loss": 0.9588,
      "step": 5842
    },
    {
      "epoch": 0.47217115497282774,
      "grad_norm": 2.913607597351074,
      "learning_rate": 8.927015047562998e-06,
      "loss": 1.0609,
      "step": 5843
    },
    {
      "epoch": 0.47225196468615527,
      "grad_norm": 2.7796878814697266,
      "learning_rate": 8.926609973750125e-06,
      "loss": 0.9498,
      "step": 5844
    },
    {
      "epoch": 0.4723327743994828,
      "grad_norm": 2.9315896034240723,
      "learning_rate": 8.926204832683876e-06,
      "loss": 1.0,
      "step": 5845
    },
    {
      "epoch": 0.47241358411281037,
      "grad_norm": 3.195502996444702,
      "learning_rate": 8.92579962437119e-06,
      "loss": 1.0401,
      "step": 5846
    },
    {
      "epoch": 0.4724943938261379,
      "grad_norm": 3.0949106216430664,
      "learning_rate": 8.925394348819008e-06,
      "loss": 0.9566,
      "step": 5847
    },
    {
      "epoch": 0.4725752035394654,
      "grad_norm": 2.629986047744751,
      "learning_rate": 8.92498900603427e-06,
      "loss": 1.0411,
      "step": 5848
    },
    {
      "epoch": 0.472656013252793,
      "grad_norm": 2.9056739807128906,
      "learning_rate": 8.924583596023921e-06,
      "loss": 0.9203,
      "step": 5849
    },
    {
      "epoch": 0.4727368229661205,
      "grad_norm": 3.0195114612579346,
      "learning_rate": 8.924178118794902e-06,
      "loss": 1.005,
      "step": 5850
    },
    {
      "epoch": 0.4728176326794481,
      "grad_norm": 2.76839542388916,
      "learning_rate": 8.923772574354159e-06,
      "loss": 0.8678,
      "step": 5851
    },
    {
      "epoch": 0.4728984423927756,
      "grad_norm": 2.5567920207977295,
      "learning_rate": 8.923366962708639e-06,
      "loss": 0.964,
      "step": 5852
    },
    {
      "epoch": 0.47297925210610314,
      "grad_norm": 2.298448085784912,
      "learning_rate": 8.922961283865285e-06,
      "loss": 0.9843,
      "step": 5853
    },
    {
      "epoch": 0.4730600618194307,
      "grad_norm": 2.7807490825653076,
      "learning_rate": 8.922555537831053e-06,
      "loss": 0.8826,
      "step": 5854
    },
    {
      "epoch": 0.47314087153275824,
      "grad_norm": 2.535998582839966,
      "learning_rate": 8.922149724612884e-06,
      "loss": 1.0366,
      "step": 5855
    },
    {
      "epoch": 0.47322168124608577,
      "grad_norm": 3.046980857849121,
      "learning_rate": 8.921743844217734e-06,
      "loss": 0.9763,
      "step": 5856
    },
    {
      "epoch": 0.47330249095941335,
      "grad_norm": 2.562838077545166,
      "learning_rate": 8.921337896652552e-06,
      "loss": 0.9475,
      "step": 5857
    },
    {
      "epoch": 0.47338330067274087,
      "grad_norm": 3.1323447227478027,
      "learning_rate": 8.920931881924294e-06,
      "loss": 1.0673,
      "step": 5858
    },
    {
      "epoch": 0.4734641103860684,
      "grad_norm": 2.4061942100524902,
      "learning_rate": 8.92052580003991e-06,
      "loss": 0.9865,
      "step": 5859
    },
    {
      "epoch": 0.47354492009939597,
      "grad_norm": 2.6887903213500977,
      "learning_rate": 8.920119651006358e-06,
      "loss": 1.0016,
      "step": 5860
    },
    {
      "epoch": 0.4736257298127235,
      "grad_norm": 2.7243714332580566,
      "learning_rate": 8.919713434830595e-06,
      "loss": 1.0094,
      "step": 5861
    },
    {
      "epoch": 0.473706539526051,
      "grad_norm": 2.7244873046875,
      "learning_rate": 8.919307151519576e-06,
      "loss": 0.8577,
      "step": 5862
    },
    {
      "epoch": 0.4737873492393786,
      "grad_norm": 2.487475633621216,
      "learning_rate": 8.91890080108026e-06,
      "loss": 1.0437,
      "step": 5863
    },
    {
      "epoch": 0.4738681589527061,
      "grad_norm": 2.6981966495513916,
      "learning_rate": 8.91849438351961e-06,
      "loss": 0.9076,
      "step": 5864
    },
    {
      "epoch": 0.47394896866603364,
      "grad_norm": 2.518085479736328,
      "learning_rate": 8.918087898844583e-06,
      "loss": 0.8898,
      "step": 5865
    },
    {
      "epoch": 0.4740297783793612,
      "grad_norm": 2.9908382892608643,
      "learning_rate": 8.917681347062142e-06,
      "loss": 0.9356,
      "step": 5866
    },
    {
      "epoch": 0.47411058809268875,
      "grad_norm": 2.8750083446502686,
      "learning_rate": 8.917274728179253e-06,
      "loss": 0.8916,
      "step": 5867
    },
    {
      "epoch": 0.47419139780601627,
      "grad_norm": 2.3316619396209717,
      "learning_rate": 8.916868042202876e-06,
      "loss": 0.9936,
      "step": 5868
    },
    {
      "epoch": 0.47427220751934385,
      "grad_norm": 2.837986707687378,
      "learning_rate": 8.91646128913998e-06,
      "loss": 0.9301,
      "step": 5869
    },
    {
      "epoch": 0.47435301723267137,
      "grad_norm": 2.7421464920043945,
      "learning_rate": 8.916054468997532e-06,
      "loss": 0.8151,
      "step": 5870
    },
    {
      "epoch": 0.4744338269459989,
      "grad_norm": 2.631556272506714,
      "learning_rate": 8.915647581782496e-06,
      "loss": 0.9708,
      "step": 5871
    },
    {
      "epoch": 0.47451463665932647,
      "grad_norm": 2.9722564220428467,
      "learning_rate": 8.915240627501845e-06,
      "loss": 0.9472,
      "step": 5872
    },
    {
      "epoch": 0.474595446372654,
      "grad_norm": 2.663175106048584,
      "learning_rate": 8.914833606162547e-06,
      "loss": 1.0013,
      "step": 5873
    },
    {
      "epoch": 0.4746762560859815,
      "grad_norm": 2.532076120376587,
      "learning_rate": 8.914426517771574e-06,
      "loss": 0.9357,
      "step": 5874
    },
    {
      "epoch": 0.4747570657993091,
      "grad_norm": 2.539815902709961,
      "learning_rate": 8.914019362335899e-06,
      "loss": 1.0391,
      "step": 5875
    },
    {
      "epoch": 0.4748378755126366,
      "grad_norm": 2.487412452697754,
      "learning_rate": 8.913612139862495e-06,
      "loss": 1.1073,
      "step": 5876
    },
    {
      "epoch": 0.47491868522596414,
      "grad_norm": 2.6713690757751465,
      "learning_rate": 8.913204850358337e-06,
      "loss": 0.9434,
      "step": 5877
    },
    {
      "epoch": 0.4749994949392917,
      "grad_norm": 2.9342000484466553,
      "learning_rate": 8.912797493830399e-06,
      "loss": 1.0425,
      "step": 5878
    },
    {
      "epoch": 0.47508030465261925,
      "grad_norm": 2.4762911796569824,
      "learning_rate": 8.91239007028566e-06,
      "loss": 0.8584,
      "step": 5879
    },
    {
      "epoch": 0.47516111436594677,
      "grad_norm": 2.757404327392578,
      "learning_rate": 8.911982579731097e-06,
      "loss": 0.8604,
      "step": 5880
    },
    {
      "epoch": 0.47524192407927435,
      "grad_norm": 2.7208292484283447,
      "learning_rate": 8.911575022173692e-06,
      "loss": 0.9445,
      "step": 5881
    },
    {
      "epoch": 0.47532273379260187,
      "grad_norm": 2.792088031768799,
      "learning_rate": 8.911167397620423e-06,
      "loss": 1.0644,
      "step": 5882
    },
    {
      "epoch": 0.4754035435059294,
      "grad_norm": 2.9895739555358887,
      "learning_rate": 8.910759706078273e-06,
      "loss": 1.0276,
      "step": 5883
    },
    {
      "epoch": 0.47548435321925697,
      "grad_norm": 2.301095962524414,
      "learning_rate": 8.910351947554223e-06,
      "loss": 1.0377,
      "step": 5884
    },
    {
      "epoch": 0.4755651629325845,
      "grad_norm": 2.9060397148132324,
      "learning_rate": 8.909944122055259e-06,
      "loss": 0.9816,
      "step": 5885
    },
    {
      "epoch": 0.475645972645912,
      "grad_norm": 2.6341824531555176,
      "learning_rate": 8.909536229588362e-06,
      "loss": 0.9556,
      "step": 5886
    },
    {
      "epoch": 0.4757267823592396,
      "grad_norm": 2.6049411296844482,
      "learning_rate": 8.909128270160522e-06,
      "loss": 1.0321,
      "step": 5887
    },
    {
      "epoch": 0.4758075920725671,
      "grad_norm": 2.5819296836853027,
      "learning_rate": 8.90872024377873e-06,
      "loss": 0.9837,
      "step": 5888
    },
    {
      "epoch": 0.47588840178589464,
      "grad_norm": 2.6825339794158936,
      "learning_rate": 8.908312150449965e-06,
      "loss": 0.9934,
      "step": 5889
    },
    {
      "epoch": 0.4759692114992222,
      "grad_norm": 2.4855387210845947,
      "learning_rate": 8.907903990181224e-06,
      "loss": 0.9458,
      "step": 5890
    },
    {
      "epoch": 0.47605002121254975,
      "grad_norm": 2.2605373859405518,
      "learning_rate": 8.907495762979495e-06,
      "loss": 0.9588,
      "step": 5891
    },
    {
      "epoch": 0.47613083092587727,
      "grad_norm": 3.199838161468506,
      "learning_rate": 8.907087468851772e-06,
      "loss": 0.8523,
      "step": 5892
    },
    {
      "epoch": 0.47621164063920485,
      "grad_norm": 2.5428595542907715,
      "learning_rate": 8.906679107805046e-06,
      "loss": 0.929,
      "step": 5893
    },
    {
      "epoch": 0.47629245035253237,
      "grad_norm": 2.752619743347168,
      "learning_rate": 8.90627067984631e-06,
      "loss": 0.9855,
      "step": 5894
    },
    {
      "epoch": 0.4763732600658599,
      "grad_norm": 2.589021921157837,
      "learning_rate": 8.905862184982561e-06,
      "loss": 0.9226,
      "step": 5895
    },
    {
      "epoch": 0.4764540697791875,
      "grad_norm": 2.7154488563537598,
      "learning_rate": 8.905453623220797e-06,
      "loss": 0.9373,
      "step": 5896
    },
    {
      "epoch": 0.476534879492515,
      "grad_norm": 2.9029784202575684,
      "learning_rate": 8.905044994568015e-06,
      "loss": 1.1735,
      "step": 5897
    },
    {
      "epoch": 0.4766156892058425,
      "grad_norm": 2.720885992050171,
      "learning_rate": 8.904636299031212e-06,
      "loss": 1.0186,
      "step": 5898
    },
    {
      "epoch": 0.4766964989191701,
      "grad_norm": 2.810302257537842,
      "learning_rate": 8.90422753661739e-06,
      "loss": 0.8561,
      "step": 5899
    },
    {
      "epoch": 0.4767773086324976,
      "grad_norm": 2.8350160121917725,
      "learning_rate": 8.90381870733355e-06,
      "loss": 0.9832,
      "step": 5900
    },
    {
      "epoch": 0.47685811834582514,
      "grad_norm": 3.249141216278076,
      "learning_rate": 8.903409811186694e-06,
      "loss": 0.97,
      "step": 5901
    },
    {
      "epoch": 0.4769389280591527,
      "grad_norm": 2.7863097190856934,
      "learning_rate": 8.903000848183822e-06,
      "loss": 0.9245,
      "step": 5902
    },
    {
      "epoch": 0.47701973777248025,
      "grad_norm": 2.48960542678833,
      "learning_rate": 8.902591818331944e-06,
      "loss": 0.916,
      "step": 5903
    },
    {
      "epoch": 0.47710054748580777,
      "grad_norm": 2.8880462646484375,
      "learning_rate": 8.902182721638064e-06,
      "loss": 0.9901,
      "step": 5904
    },
    {
      "epoch": 0.47718135719913535,
      "grad_norm": 2.901137351989746,
      "learning_rate": 8.901773558109185e-06,
      "loss": 1.0175,
      "step": 5905
    },
    {
      "epoch": 0.47726216691246287,
      "grad_norm": 2.259371280670166,
      "learning_rate": 8.90136432775232e-06,
      "loss": 1.0321,
      "step": 5906
    },
    {
      "epoch": 0.4773429766257904,
      "grad_norm": 2.7619972229003906,
      "learning_rate": 8.900955030574478e-06,
      "loss": 0.9332,
      "step": 5907
    },
    {
      "epoch": 0.477423786339118,
      "grad_norm": 2.9904873371124268,
      "learning_rate": 8.900545666582665e-06,
      "loss": 0.9263,
      "step": 5908
    },
    {
      "epoch": 0.4775045960524455,
      "grad_norm": 2.4642465114593506,
      "learning_rate": 8.900136235783896e-06,
      "loss": 0.9581,
      "step": 5909
    },
    {
      "epoch": 0.477585405765773,
      "grad_norm": 3.090346336364746,
      "learning_rate": 8.899726738185182e-06,
      "loss": 0.9824,
      "step": 5910
    },
    {
      "epoch": 0.4776662154791006,
      "grad_norm": 2.9905292987823486,
      "learning_rate": 8.899317173793537e-06,
      "loss": 0.981,
      "step": 5911
    },
    {
      "epoch": 0.4777470251924281,
      "grad_norm": 2.654024600982666,
      "learning_rate": 8.898907542615975e-06,
      "loss": 0.9052,
      "step": 5912
    },
    {
      "epoch": 0.4778278349057557,
      "grad_norm": 2.6835696697235107,
      "learning_rate": 8.898497844659515e-06,
      "loss": 1.0652,
      "step": 5913
    },
    {
      "epoch": 0.4779086446190832,
      "grad_norm": 2.551361560821533,
      "learning_rate": 8.898088079931171e-06,
      "loss": 0.9675,
      "step": 5914
    },
    {
      "epoch": 0.47798945433241075,
      "grad_norm": 2.444711685180664,
      "learning_rate": 8.897678248437965e-06,
      "loss": 0.9254,
      "step": 5915
    },
    {
      "epoch": 0.4780702640457383,
      "grad_norm": 2.4407832622528076,
      "learning_rate": 8.89726835018691e-06,
      "loss": 0.967,
      "step": 5916
    },
    {
      "epoch": 0.47815107375906585,
      "grad_norm": 2.499941110610962,
      "learning_rate": 8.896858385185032e-06,
      "loss": 0.9801,
      "step": 5917
    },
    {
      "epoch": 0.47823188347239337,
      "grad_norm": 2.9067742824554443,
      "learning_rate": 8.896448353439352e-06,
      "loss": 0.9183,
      "step": 5918
    },
    {
      "epoch": 0.47831269318572095,
      "grad_norm": 2.7358031272888184,
      "learning_rate": 8.896038254956892e-06,
      "loss": 0.9574,
      "step": 5919
    },
    {
      "epoch": 0.4783935028990485,
      "grad_norm": 2.9418129920959473,
      "learning_rate": 8.895628089744674e-06,
      "loss": 1.0692,
      "step": 5920
    },
    {
      "epoch": 0.478474312612376,
      "grad_norm": 2.660916328430176,
      "learning_rate": 8.895217857809728e-06,
      "loss": 1.1609,
      "step": 5921
    },
    {
      "epoch": 0.4785551223257036,
      "grad_norm": 2.3367667198181152,
      "learning_rate": 8.894807559159075e-06,
      "loss": 1.0068,
      "step": 5922
    },
    {
      "epoch": 0.4786359320390311,
      "grad_norm": 2.8970179557800293,
      "learning_rate": 8.894397193799747e-06,
      "loss": 1.054,
      "step": 5923
    },
    {
      "epoch": 0.4787167417523586,
      "grad_norm": 2.2323038578033447,
      "learning_rate": 8.893986761738769e-06,
      "loss": 1.0777,
      "step": 5924
    },
    {
      "epoch": 0.4787975514656862,
      "grad_norm": 2.432058572769165,
      "learning_rate": 8.893576262983173e-06,
      "loss": 1.0511,
      "step": 5925
    },
    {
      "epoch": 0.4788783611790137,
      "grad_norm": 2.663792848587036,
      "learning_rate": 8.893165697539988e-06,
      "loss": 1.0488,
      "step": 5926
    },
    {
      "epoch": 0.47895917089234125,
      "grad_norm": 2.9636640548706055,
      "learning_rate": 8.892755065416247e-06,
      "loss": 1.048,
      "step": 5927
    },
    {
      "epoch": 0.4790399806056688,
      "grad_norm": 2.5020740032196045,
      "learning_rate": 8.892344366618985e-06,
      "loss": 1.122,
      "step": 5928
    },
    {
      "epoch": 0.47912079031899635,
      "grad_norm": 2.7484488487243652,
      "learning_rate": 8.891933601155233e-06,
      "loss": 0.9464,
      "step": 5929
    },
    {
      "epoch": 0.47920160003232387,
      "grad_norm": 2.557628631591797,
      "learning_rate": 8.891522769032029e-06,
      "loss": 0.9269,
      "step": 5930
    },
    {
      "epoch": 0.47928240974565145,
      "grad_norm": 2.434396266937256,
      "learning_rate": 8.891111870256406e-06,
      "loss": 0.9905,
      "step": 5931
    },
    {
      "epoch": 0.479363219458979,
      "grad_norm": 2.8306679725646973,
      "learning_rate": 8.890700904835405e-06,
      "loss": 0.959,
      "step": 5932
    },
    {
      "epoch": 0.4794440291723065,
      "grad_norm": 3.045585870742798,
      "learning_rate": 8.890289872776066e-06,
      "loss": 1.0857,
      "step": 5933
    },
    {
      "epoch": 0.4795248388856341,
      "grad_norm": 2.8493340015411377,
      "learning_rate": 8.889878774085425e-06,
      "loss": 0.996,
      "step": 5934
    },
    {
      "epoch": 0.4796056485989616,
      "grad_norm": 2.993567705154419,
      "learning_rate": 8.889467608770526e-06,
      "loss": 1.0702,
      "step": 5935
    },
    {
      "epoch": 0.4796864583122891,
      "grad_norm": 2.2887275218963623,
      "learning_rate": 8.88905637683841e-06,
      "loss": 0.9916,
      "step": 5936
    },
    {
      "epoch": 0.4797672680256167,
      "grad_norm": 2.474585771560669,
      "learning_rate": 8.88864507829612e-06,
      "loss": 1.0589,
      "step": 5937
    },
    {
      "epoch": 0.4798480777389442,
      "grad_norm": 2.6525509357452393,
      "learning_rate": 8.888233713150702e-06,
      "loss": 1.0499,
      "step": 5938
    },
    {
      "epoch": 0.47992888745227175,
      "grad_norm": 2.752514600753784,
      "learning_rate": 8.887822281409202e-06,
      "loss": 1.0005,
      "step": 5939
    },
    {
      "epoch": 0.4800096971655993,
      "grad_norm": 2.515058755874634,
      "learning_rate": 8.887410783078664e-06,
      "loss": 0.9086,
      "step": 5940
    },
    {
      "epoch": 0.48009050687892685,
      "grad_norm": 2.3969075679779053,
      "learning_rate": 8.88699921816614e-06,
      "loss": 0.9103,
      "step": 5941
    },
    {
      "epoch": 0.48017131659225437,
      "grad_norm": 2.6233177185058594,
      "learning_rate": 8.886587586678675e-06,
      "loss": 1.0183,
      "step": 5942
    },
    {
      "epoch": 0.48025212630558195,
      "grad_norm": 2.835303783416748,
      "learning_rate": 8.886175888623323e-06,
      "loss": 0.9628,
      "step": 5943
    },
    {
      "epoch": 0.4803329360189095,
      "grad_norm": 3.0450258255004883,
      "learning_rate": 8.885764124007132e-06,
      "loss": 0.8766,
      "step": 5944
    },
    {
      "epoch": 0.480413745732237,
      "grad_norm": 2.686596155166626,
      "learning_rate": 8.885352292837157e-06,
      "loss": 0.9347,
      "step": 5945
    },
    {
      "epoch": 0.4804945554455646,
      "grad_norm": 2.518345355987549,
      "learning_rate": 8.884940395120451e-06,
      "loss": 1.0938,
      "step": 5946
    },
    {
      "epoch": 0.4805753651588921,
      "grad_norm": 2.5679078102111816,
      "learning_rate": 8.884528430864067e-06,
      "loss": 1.114,
      "step": 5947
    },
    {
      "epoch": 0.4806561748722196,
      "grad_norm": 2.5896337032318115,
      "learning_rate": 8.884116400075064e-06,
      "loss": 0.9511,
      "step": 5948
    },
    {
      "epoch": 0.4807369845855472,
      "grad_norm": 2.400423288345337,
      "learning_rate": 8.883704302760499e-06,
      "loss": 0.9513,
      "step": 5949
    },
    {
      "epoch": 0.4808177942988747,
      "grad_norm": 3.165663242340088,
      "learning_rate": 8.883292138927427e-06,
      "loss": 0.9374,
      "step": 5950
    },
    {
      "epoch": 0.48089860401220225,
      "grad_norm": 3.0608668327331543,
      "learning_rate": 8.88287990858291e-06,
      "loss": 0.9592,
      "step": 5951
    },
    {
      "epoch": 0.4809794137255298,
      "grad_norm": 2.5450081825256348,
      "learning_rate": 8.882467611734006e-06,
      "loss": 0.9344,
      "step": 5952
    },
    {
      "epoch": 0.48106022343885735,
      "grad_norm": 2.836158037185669,
      "learning_rate": 8.882055248387781e-06,
      "loss": 0.8362,
      "step": 5953
    },
    {
      "epoch": 0.48114103315218487,
      "grad_norm": 2.946566104888916,
      "learning_rate": 8.881642818551295e-06,
      "loss": 0.8943,
      "step": 5954
    },
    {
      "epoch": 0.48122184286551245,
      "grad_norm": 2.8138973712921143,
      "learning_rate": 8.881230322231612e-06,
      "loss": 1.0198,
      "step": 5955
    },
    {
      "epoch": 0.48130265257884,
      "grad_norm": 2.5750601291656494,
      "learning_rate": 8.880817759435796e-06,
      "loss": 1.0705,
      "step": 5956
    },
    {
      "epoch": 0.4813834622921675,
      "grad_norm": 2.8209667205810547,
      "learning_rate": 8.880405130170916e-06,
      "loss": 1.1097,
      "step": 5957
    },
    {
      "epoch": 0.4814642720054951,
      "grad_norm": 3.044922113418579,
      "learning_rate": 8.879992434444037e-06,
      "loss": 0.9697,
      "step": 5958
    },
    {
      "epoch": 0.4815450817188226,
      "grad_norm": 2.602264642715454,
      "learning_rate": 8.879579672262228e-06,
      "loss": 1.042,
      "step": 5959
    },
    {
      "epoch": 0.4816258914321501,
      "grad_norm": 2.904402017593384,
      "learning_rate": 8.879166843632559e-06,
      "loss": 0.9123,
      "step": 5960
    },
    {
      "epoch": 0.4817067011454777,
      "grad_norm": 2.8899734020233154,
      "learning_rate": 8.878753948562103e-06,
      "loss": 0.9966,
      "step": 5961
    },
    {
      "epoch": 0.4817875108588052,
      "grad_norm": 3.0337719917297363,
      "learning_rate": 8.878340987057926e-06,
      "loss": 0.9003,
      "step": 5962
    },
    {
      "epoch": 0.48186832057213275,
      "grad_norm": 2.6741790771484375,
      "learning_rate": 8.877927959127106e-06,
      "loss": 1.1136,
      "step": 5963
    },
    {
      "epoch": 0.4819491302854603,
      "grad_norm": 2.7310738563537598,
      "learning_rate": 8.877514864776718e-06,
      "loss": 0.9247,
      "step": 5964
    },
    {
      "epoch": 0.48202993999878785,
      "grad_norm": 2.645674467086792,
      "learning_rate": 8.877101704013832e-06,
      "loss": 0.9974,
      "step": 5965
    },
    {
      "epoch": 0.48211074971211537,
      "grad_norm": 2.878068208694458,
      "learning_rate": 8.876688476845527e-06,
      "loss": 1.1114,
      "step": 5966
    },
    {
      "epoch": 0.48219155942544295,
      "grad_norm": 2.9921646118164062,
      "learning_rate": 8.876275183278883e-06,
      "loss": 0.9797,
      "step": 5967
    },
    {
      "epoch": 0.4822723691387705,
      "grad_norm": 2.4352478981018066,
      "learning_rate": 8.875861823320977e-06,
      "loss": 0.8394,
      "step": 5968
    },
    {
      "epoch": 0.482353178852098,
      "grad_norm": 2.6279549598693848,
      "learning_rate": 8.87544839697889e-06,
      "loss": 1.0402,
      "step": 5969
    },
    {
      "epoch": 0.4824339885654256,
      "grad_norm": 2.7221219539642334,
      "learning_rate": 8.8750349042597e-06,
      "loss": 0.8392,
      "step": 5970
    },
    {
      "epoch": 0.4825147982787531,
      "grad_norm": 2.767946481704712,
      "learning_rate": 8.87462134517049e-06,
      "loss": 0.9744,
      "step": 5971
    },
    {
      "epoch": 0.4825956079920806,
      "grad_norm": 2.7020928859710693,
      "learning_rate": 8.874207719718345e-06,
      "loss": 0.8544,
      "step": 5972
    },
    {
      "epoch": 0.4826764177054082,
      "grad_norm": 3.2032032012939453,
      "learning_rate": 8.873794027910349e-06,
      "loss": 0.8749,
      "step": 5973
    },
    {
      "epoch": 0.4827572274187357,
      "grad_norm": 2.381051778793335,
      "learning_rate": 8.873380269753586e-06,
      "loss": 0.9447,
      "step": 5974
    },
    {
      "epoch": 0.48283803713206325,
      "grad_norm": 2.5544629096984863,
      "learning_rate": 8.872966445255144e-06,
      "loss": 0.9809,
      "step": 5975
    },
    {
      "epoch": 0.4829188468453908,
      "grad_norm": 2.5966579914093018,
      "learning_rate": 8.872552554422111e-06,
      "loss": 1.0221,
      "step": 5976
    },
    {
      "epoch": 0.48299965655871835,
      "grad_norm": 2.7803614139556885,
      "learning_rate": 8.872138597261578e-06,
      "loss": 0.8829,
      "step": 5977
    },
    {
      "epoch": 0.4830804662720459,
      "grad_norm": 2.755143880844116,
      "learning_rate": 8.87172457378063e-06,
      "loss": 0.9079,
      "step": 5978
    },
    {
      "epoch": 0.48316127598537345,
      "grad_norm": 2.575796604156494,
      "learning_rate": 8.871310483986359e-06,
      "loss": 0.9363,
      "step": 5979
    },
    {
      "epoch": 0.483242085698701,
      "grad_norm": 2.5812630653381348,
      "learning_rate": 8.870896327885863e-06,
      "loss": 0.8756,
      "step": 5980
    },
    {
      "epoch": 0.48332289541202855,
      "grad_norm": 2.4501044750213623,
      "learning_rate": 8.870482105486229e-06,
      "loss": 1.0085,
      "step": 5981
    },
    {
      "epoch": 0.4834037051253561,
      "grad_norm": 2.852572202682495,
      "learning_rate": 8.870067816794557e-06,
      "loss": 0.9326,
      "step": 5982
    },
    {
      "epoch": 0.4834845148386836,
      "grad_norm": 2.9296329021453857,
      "learning_rate": 8.869653461817937e-06,
      "loss": 0.8954,
      "step": 5983
    },
    {
      "epoch": 0.4835653245520112,
      "grad_norm": 3.000253915786743,
      "learning_rate": 8.86923904056347e-06,
      "loss": 1.0676,
      "step": 5984
    },
    {
      "epoch": 0.4836461342653387,
      "grad_norm": 2.5141992568969727,
      "learning_rate": 8.868824553038255e-06,
      "loss": 0.8814,
      "step": 5985
    },
    {
      "epoch": 0.4837269439786662,
      "grad_norm": 2.6345553398132324,
      "learning_rate": 8.868409999249387e-06,
      "loss": 0.8525,
      "step": 5986
    },
    {
      "epoch": 0.4838077536919938,
      "grad_norm": 2.6637556552886963,
      "learning_rate": 8.867995379203969e-06,
      "loss": 1.0274,
      "step": 5987
    },
    {
      "epoch": 0.4838885634053213,
      "grad_norm": 2.3642663955688477,
      "learning_rate": 8.867580692909102e-06,
      "loss": 0.938,
      "step": 5988
    },
    {
      "epoch": 0.48396937311864885,
      "grad_norm": 2.8194892406463623,
      "learning_rate": 8.867165940371888e-06,
      "loss": 0.9672,
      "step": 5989
    },
    {
      "epoch": 0.4840501828319764,
      "grad_norm": 3.4492805004119873,
      "learning_rate": 8.866751121599432e-06,
      "loss": 0.8927,
      "step": 5990
    },
    {
      "epoch": 0.48413099254530395,
      "grad_norm": 2.597614049911499,
      "learning_rate": 8.866336236598839e-06,
      "loss": 0.8299,
      "step": 5991
    },
    {
      "epoch": 0.4842118022586315,
      "grad_norm": 2.568424701690674,
      "learning_rate": 8.865921285377214e-06,
      "loss": 1.0524,
      "step": 5992
    },
    {
      "epoch": 0.48429261197195905,
      "grad_norm": 3.424942970275879,
      "learning_rate": 8.865506267941663e-06,
      "loss": 0.9346,
      "step": 5993
    },
    {
      "epoch": 0.4843734216852866,
      "grad_norm": 2.7126035690307617,
      "learning_rate": 8.865091184299295e-06,
      "loss": 0.9374,
      "step": 5994
    },
    {
      "epoch": 0.4844542313986141,
      "grad_norm": 2.644007682800293,
      "learning_rate": 8.864676034457222e-06,
      "loss": 1.0067,
      "step": 5995
    },
    {
      "epoch": 0.4845350411119417,
      "grad_norm": 2.539374828338623,
      "learning_rate": 8.864260818422549e-06,
      "loss": 0.9168,
      "step": 5996
    },
    {
      "epoch": 0.4846158508252692,
      "grad_norm": 2.574723243713379,
      "learning_rate": 8.863845536202394e-06,
      "loss": 0.9339,
      "step": 5997
    },
    {
      "epoch": 0.4846966605385967,
      "grad_norm": 2.717312812805176,
      "learning_rate": 8.863430187803867e-06,
      "loss": 0.9301,
      "step": 5998
    },
    {
      "epoch": 0.4847774702519243,
      "grad_norm": 2.5925450325012207,
      "learning_rate": 8.86301477323408e-06,
      "loss": 0.9089,
      "step": 5999
    },
    {
      "epoch": 0.4848582799652518,
      "grad_norm": 2.3955633640289307,
      "learning_rate": 8.862599292500151e-06,
      "loss": 0.9023,
      "step": 6000
    },
    {
      "epoch": 0.4848582799652518,
      "eval_loss": 0.8077141046524048,
      "eval_runtime": 812.2973,
      "eval_samples_per_second": 102.63,
      "eval_steps_per_second": 12.829,
      "step": 6000
    },
    {
      "epoch": 0.48493908967857935,
      "grad_norm": 2.3458774089813232,
      "learning_rate": 8.862183745609195e-06,
      "loss": 0.9074,
      "step": 6001
    },
    {
      "epoch": 0.48501989939190693,
      "grad_norm": 2.666863441467285,
      "learning_rate": 8.861768132568327e-06,
      "loss": 1.0235,
      "step": 6002
    },
    {
      "epoch": 0.48510070910523445,
      "grad_norm": 2.8406612873077393,
      "learning_rate": 8.86135245338467e-06,
      "loss": 1.088,
      "step": 6003
    },
    {
      "epoch": 0.485181518818562,
      "grad_norm": 2.791804075241089,
      "learning_rate": 8.86093670806534e-06,
      "loss": 0.9759,
      "step": 6004
    },
    {
      "epoch": 0.48526232853188955,
      "grad_norm": 3.0780794620513916,
      "learning_rate": 8.860520896617459e-06,
      "loss": 1.0224,
      "step": 6005
    },
    {
      "epoch": 0.4853431382452171,
      "grad_norm": 3.0164146423339844,
      "learning_rate": 8.86010501904815e-06,
      "loss": 0.8652,
      "step": 6006
    },
    {
      "epoch": 0.4854239479585446,
      "grad_norm": 3.068770170211792,
      "learning_rate": 8.859689075364535e-06,
      "loss": 1.0249,
      "step": 6007
    },
    {
      "epoch": 0.4855047576718722,
      "grad_norm": 2.6718451976776123,
      "learning_rate": 8.859273065573736e-06,
      "loss": 0.9633,
      "step": 6008
    },
    {
      "epoch": 0.4855855673851997,
      "grad_norm": 2.410693645477295,
      "learning_rate": 8.858856989682883e-06,
      "loss": 0.967,
      "step": 6009
    },
    {
      "epoch": 0.4856663770985272,
      "grad_norm": 2.8103816509246826,
      "learning_rate": 8.858440847699097e-06,
      "loss": 0.8983,
      "step": 6010
    },
    {
      "epoch": 0.4857471868118548,
      "grad_norm": 2.8238930702209473,
      "learning_rate": 8.85802463962951e-06,
      "loss": 1.0157,
      "step": 6011
    },
    {
      "epoch": 0.4858279965251823,
      "grad_norm": 2.3492908477783203,
      "learning_rate": 8.857608365481247e-06,
      "loss": 1.0542,
      "step": 6012
    },
    {
      "epoch": 0.48590880623850985,
      "grad_norm": 2.9035751819610596,
      "learning_rate": 8.85719202526144e-06,
      "loss": 0.9303,
      "step": 6013
    },
    {
      "epoch": 0.48598961595183743,
      "grad_norm": 2.6763007640838623,
      "learning_rate": 8.85677561897722e-06,
      "loss": 0.8904,
      "step": 6014
    },
    {
      "epoch": 0.48607042566516495,
      "grad_norm": 2.706169605255127,
      "learning_rate": 8.85635914663572e-06,
      "loss": 1.0327,
      "step": 6015
    },
    {
      "epoch": 0.4861512353784925,
      "grad_norm": 2.4542858600616455,
      "learning_rate": 8.855942608244069e-06,
      "loss": 0.9041,
      "step": 6016
    },
    {
      "epoch": 0.48623204509182005,
      "grad_norm": 2.3222744464874268,
      "learning_rate": 8.855526003809405e-06,
      "loss": 1.0501,
      "step": 6017
    },
    {
      "epoch": 0.4863128548051476,
      "grad_norm": 2.7569198608398438,
      "learning_rate": 8.855109333338863e-06,
      "loss": 0.9584,
      "step": 6018
    },
    {
      "epoch": 0.4863936645184751,
      "grad_norm": 2.7721915245056152,
      "learning_rate": 8.854692596839577e-06,
      "loss": 1.0452,
      "step": 6019
    },
    {
      "epoch": 0.4864744742318027,
      "grad_norm": 2.8172824382781982,
      "learning_rate": 8.854275794318688e-06,
      "loss": 1.1104,
      "step": 6020
    },
    {
      "epoch": 0.4865552839451302,
      "grad_norm": 2.705747365951538,
      "learning_rate": 8.853858925783334e-06,
      "loss": 0.9695,
      "step": 6021
    },
    {
      "epoch": 0.4866360936584577,
      "grad_norm": 2.884138822555542,
      "learning_rate": 8.853441991240652e-06,
      "loss": 0.8577,
      "step": 6022
    },
    {
      "epoch": 0.4867169033717853,
      "grad_norm": 2.4505674839019775,
      "learning_rate": 8.853024990697787e-06,
      "loss": 0.9479,
      "step": 6023
    },
    {
      "epoch": 0.4867977130851128,
      "grad_norm": 2.5882649421691895,
      "learning_rate": 8.85260792416188e-06,
      "loss": 1.0449,
      "step": 6024
    },
    {
      "epoch": 0.48687852279844035,
      "grad_norm": 2.8596298694610596,
      "learning_rate": 8.852190791640075e-06,
      "loss": 0.9191,
      "step": 6025
    },
    {
      "epoch": 0.48695933251176793,
      "grad_norm": 2.4536993503570557,
      "learning_rate": 8.851773593139514e-06,
      "loss": 1.0858,
      "step": 6026
    },
    {
      "epoch": 0.48704014222509545,
      "grad_norm": 3.342395305633545,
      "learning_rate": 8.851356328667343e-06,
      "loss": 0.995,
      "step": 6027
    },
    {
      "epoch": 0.487120951938423,
      "grad_norm": 2.5988099575042725,
      "learning_rate": 8.850938998230711e-06,
      "loss": 1.0465,
      "step": 6028
    },
    {
      "epoch": 0.48720176165175055,
      "grad_norm": 2.6186883449554443,
      "learning_rate": 8.850521601836765e-06,
      "loss": 0.9064,
      "step": 6029
    },
    {
      "epoch": 0.4872825713650781,
      "grad_norm": 3.1623666286468506,
      "learning_rate": 8.850104139492655e-06,
      "loss": 0.9117,
      "step": 6030
    },
    {
      "epoch": 0.4873633810784056,
      "grad_norm": 2.673964500427246,
      "learning_rate": 8.849686611205528e-06,
      "loss": 1.0292,
      "step": 6031
    },
    {
      "epoch": 0.4874441907917332,
      "grad_norm": 2.733144521713257,
      "learning_rate": 8.849269016982537e-06,
      "loss": 0.9927,
      "step": 6032
    },
    {
      "epoch": 0.4875250005050607,
      "grad_norm": 2.493893623352051,
      "learning_rate": 8.848851356830834e-06,
      "loss": 0.9182,
      "step": 6033
    },
    {
      "epoch": 0.4876058102183882,
      "grad_norm": 2.8255133628845215,
      "learning_rate": 8.848433630757575e-06,
      "loss": 1.0022,
      "step": 6034
    },
    {
      "epoch": 0.4876866199317158,
      "grad_norm": 3.0941851139068604,
      "learning_rate": 8.848015838769912e-06,
      "loss": 0.9175,
      "step": 6035
    },
    {
      "epoch": 0.4877674296450433,
      "grad_norm": 2.5062406063079834,
      "learning_rate": 8.847597980875e-06,
      "loss": 0.8803,
      "step": 6036
    },
    {
      "epoch": 0.48784823935837085,
      "grad_norm": 2.7179102897644043,
      "learning_rate": 8.84718005708e-06,
      "loss": 0.901,
      "step": 6037
    },
    {
      "epoch": 0.48792904907169843,
      "grad_norm": 2.4092929363250732,
      "learning_rate": 8.846762067392065e-06,
      "loss": 1.114,
      "step": 6038
    },
    {
      "epoch": 0.48800985878502595,
      "grad_norm": 2.510439395904541,
      "learning_rate": 8.846344011818357e-06,
      "loss": 1.1026,
      "step": 6039
    },
    {
      "epoch": 0.4880906684983535,
      "grad_norm": 2.695366621017456,
      "learning_rate": 8.845925890366036e-06,
      "loss": 1.1017,
      "step": 6040
    },
    {
      "epoch": 0.48817147821168105,
      "grad_norm": 2.9203357696533203,
      "learning_rate": 8.845507703042263e-06,
      "loss": 0.9382,
      "step": 6041
    },
    {
      "epoch": 0.4882522879250086,
      "grad_norm": 2.4055418968200684,
      "learning_rate": 8.8450894498542e-06,
      "loss": 1.0106,
      "step": 6042
    },
    {
      "epoch": 0.48833309763833616,
      "grad_norm": 2.5554018020629883,
      "learning_rate": 8.844671130809013e-06,
      "loss": 1.0267,
      "step": 6043
    },
    {
      "epoch": 0.4884139073516637,
      "grad_norm": 3.0125513076782227,
      "learning_rate": 8.844252745913866e-06,
      "loss": 0.9283,
      "step": 6044
    },
    {
      "epoch": 0.4884947170649912,
      "grad_norm": 2.96811842918396,
      "learning_rate": 8.843834295175921e-06,
      "loss": 1.0757,
      "step": 6045
    },
    {
      "epoch": 0.4885755267783188,
      "grad_norm": 2.77514910697937,
      "learning_rate": 8.843415778602352e-06,
      "loss": 0.8695,
      "step": 6046
    },
    {
      "epoch": 0.4886563364916463,
      "grad_norm": 2.5683491230010986,
      "learning_rate": 8.842997196200318e-06,
      "loss": 0.9588,
      "step": 6047
    },
    {
      "epoch": 0.4887371462049738,
      "grad_norm": 2.290745973587036,
      "learning_rate": 8.842578547976998e-06,
      "loss": 0.9693,
      "step": 6048
    },
    {
      "epoch": 0.4888179559183014,
      "grad_norm": 2.7583298683166504,
      "learning_rate": 8.842159833939557e-06,
      "loss": 0.8398,
      "step": 6049
    },
    {
      "epoch": 0.48889876563162893,
      "grad_norm": 2.3874542713165283,
      "learning_rate": 8.841741054095167e-06,
      "loss": 0.9163,
      "step": 6050
    },
    {
      "epoch": 0.48897957534495645,
      "grad_norm": 2.66827392578125,
      "learning_rate": 8.841322208451003e-06,
      "loss": 0.9906,
      "step": 6051
    },
    {
      "epoch": 0.48906038505828403,
      "grad_norm": 2.6189911365509033,
      "learning_rate": 8.840903297014236e-06,
      "loss": 0.9406,
      "step": 6052
    },
    {
      "epoch": 0.48914119477161155,
      "grad_norm": 2.418142080307007,
      "learning_rate": 8.840484319792042e-06,
      "loss": 0.9687,
      "step": 6053
    },
    {
      "epoch": 0.4892220044849391,
      "grad_norm": 2.5195775032043457,
      "learning_rate": 8.840065276791598e-06,
      "loss": 1.0489,
      "step": 6054
    },
    {
      "epoch": 0.48930281419826666,
      "grad_norm": 2.4295156002044678,
      "learning_rate": 8.83964616802008e-06,
      "loss": 1.0801,
      "step": 6055
    },
    {
      "epoch": 0.4893836239115942,
      "grad_norm": 3.047612428665161,
      "learning_rate": 8.839226993484667e-06,
      "loss": 0.87,
      "step": 6056
    },
    {
      "epoch": 0.4894644336249217,
      "grad_norm": 2.496460437774658,
      "learning_rate": 8.838807753192537e-06,
      "loss": 1.0319,
      "step": 6057
    },
    {
      "epoch": 0.4895452433382493,
      "grad_norm": 2.630291700363159,
      "learning_rate": 8.838388447150872e-06,
      "loss": 0.9838,
      "step": 6058
    },
    {
      "epoch": 0.4896260530515768,
      "grad_norm": 2.7976157665252686,
      "learning_rate": 8.837969075366855e-06,
      "loss": 0.9455,
      "step": 6059
    },
    {
      "epoch": 0.4897068627649043,
      "grad_norm": 2.8019280433654785,
      "learning_rate": 8.837549637847665e-06,
      "loss": 0.9119,
      "step": 6060
    },
    {
      "epoch": 0.4897876724782319,
      "grad_norm": 2.3282132148742676,
      "learning_rate": 8.837130134600489e-06,
      "loss": 0.9218,
      "step": 6061
    },
    {
      "epoch": 0.48986848219155943,
      "grad_norm": 2.7614595890045166,
      "learning_rate": 8.83671056563251e-06,
      "loss": 1.082,
      "step": 6062
    },
    {
      "epoch": 0.48994929190488695,
      "grad_norm": 3.011939525604248,
      "learning_rate": 8.836290930950918e-06,
      "loss": 0.9681,
      "step": 6063
    },
    {
      "epoch": 0.49003010161821453,
      "grad_norm": 3.2438557147979736,
      "learning_rate": 8.835871230562899e-06,
      "loss": 0.8931,
      "step": 6064
    },
    {
      "epoch": 0.49011091133154205,
      "grad_norm": 2.61104679107666,
      "learning_rate": 8.835451464475637e-06,
      "loss": 0.9651,
      "step": 6065
    },
    {
      "epoch": 0.4901917210448696,
      "grad_norm": 2.612506628036499,
      "learning_rate": 8.835031632696328e-06,
      "loss": 0.9069,
      "step": 6066
    },
    {
      "epoch": 0.49027253075819716,
      "grad_norm": 2.769911050796509,
      "learning_rate": 8.834611735232157e-06,
      "loss": 0.9735,
      "step": 6067
    },
    {
      "epoch": 0.4903533404715247,
      "grad_norm": 2.936298131942749,
      "learning_rate": 8.83419177209032e-06,
      "loss": 0.9326,
      "step": 6068
    },
    {
      "epoch": 0.4904341501848522,
      "grad_norm": 2.64540696144104,
      "learning_rate": 8.833771743278007e-06,
      "loss": 0.9809,
      "step": 6069
    },
    {
      "epoch": 0.4905149598981798,
      "grad_norm": 3.032196283340454,
      "learning_rate": 8.833351648802413e-06,
      "loss": 0.9453,
      "step": 6070
    },
    {
      "epoch": 0.4905957696115073,
      "grad_norm": 2.9307849407196045,
      "learning_rate": 8.832931488670735e-06,
      "loss": 1.0201,
      "step": 6071
    },
    {
      "epoch": 0.4906765793248348,
      "grad_norm": 2.5332212448120117,
      "learning_rate": 8.832511262890169e-06,
      "loss": 0.8267,
      "step": 6072
    },
    {
      "epoch": 0.4907573890381624,
      "grad_norm": 2.430079460144043,
      "learning_rate": 8.832090971467909e-06,
      "loss": 0.956,
      "step": 6073
    },
    {
      "epoch": 0.49083819875148993,
      "grad_norm": 2.6602840423583984,
      "learning_rate": 8.831670614411157e-06,
      "loss": 0.9903,
      "step": 6074
    },
    {
      "epoch": 0.49091900846481745,
      "grad_norm": 3.0974788665771484,
      "learning_rate": 8.831250191727112e-06,
      "loss": 1.0623,
      "step": 6075
    },
    {
      "epoch": 0.49099981817814503,
      "grad_norm": 2.710204601287842,
      "learning_rate": 8.830829703422976e-06,
      "loss": 0.9669,
      "step": 6076
    },
    {
      "epoch": 0.49108062789147255,
      "grad_norm": 3.272197961807251,
      "learning_rate": 8.830409149505947e-06,
      "loss": 0.8662,
      "step": 6077
    },
    {
      "epoch": 0.4911614376048001,
      "grad_norm": 2.55169415473938,
      "learning_rate": 8.829988529983232e-06,
      "loss": 0.96,
      "step": 6078
    },
    {
      "epoch": 0.49124224731812766,
      "grad_norm": 2.7996485233306885,
      "learning_rate": 8.829567844862033e-06,
      "loss": 0.9724,
      "step": 6079
    },
    {
      "epoch": 0.4913230570314552,
      "grad_norm": 2.742427110671997,
      "learning_rate": 8.829147094149557e-06,
      "loss": 1.0475,
      "step": 6080
    },
    {
      "epoch": 0.4914038667447827,
      "grad_norm": 2.507977247238159,
      "learning_rate": 8.82872627785301e-06,
      "loss": 1.0049,
      "step": 6081
    },
    {
      "epoch": 0.4914846764581103,
      "grad_norm": 2.621269702911377,
      "learning_rate": 8.828305395979597e-06,
      "loss": 0.8754,
      "step": 6082
    },
    {
      "epoch": 0.4915654861714378,
      "grad_norm": 2.5403740406036377,
      "learning_rate": 8.827884448536531e-06,
      "loss": 0.9419,
      "step": 6083
    },
    {
      "epoch": 0.4916462958847653,
      "grad_norm": 2.73575496673584,
      "learning_rate": 8.827463435531018e-06,
      "loss": 0.8716,
      "step": 6084
    },
    {
      "epoch": 0.4917271055980929,
      "grad_norm": 2.153745412826538,
      "learning_rate": 8.827042356970272e-06,
      "loss": 0.9779,
      "step": 6085
    },
    {
      "epoch": 0.49180791531142043,
      "grad_norm": 2.304722785949707,
      "learning_rate": 8.826621212861504e-06,
      "loss": 1.0142,
      "step": 6086
    },
    {
      "epoch": 0.49188872502474795,
      "grad_norm": 2.536217212677002,
      "learning_rate": 8.826200003211924e-06,
      "loss": 0.8366,
      "step": 6087
    },
    {
      "epoch": 0.49196953473807553,
      "grad_norm": 2.6360926628112793,
      "learning_rate": 8.825778728028753e-06,
      "loss": 0.8163,
      "step": 6088
    },
    {
      "epoch": 0.49205034445140305,
      "grad_norm": 2.456071376800537,
      "learning_rate": 8.8253573873192e-06,
      "loss": 0.8741,
      "step": 6089
    },
    {
      "epoch": 0.4921311541647306,
      "grad_norm": 2.8125364780426025,
      "learning_rate": 8.824935981090485e-06,
      "loss": 0.9358,
      "step": 6090
    },
    {
      "epoch": 0.49221196387805816,
      "grad_norm": 2.7256171703338623,
      "learning_rate": 8.824514509349824e-06,
      "loss": 0.9223,
      "step": 6091
    },
    {
      "epoch": 0.4922927735913857,
      "grad_norm": 2.9064652919769287,
      "learning_rate": 8.824092972104437e-06,
      "loss": 1.1134,
      "step": 6092
    },
    {
      "epoch": 0.4923735833047132,
      "grad_norm": 2.519120454788208,
      "learning_rate": 8.823671369361545e-06,
      "loss": 1.0679,
      "step": 6093
    },
    {
      "epoch": 0.4924543930180408,
      "grad_norm": 2.463228940963745,
      "learning_rate": 8.823249701128366e-06,
      "loss": 0.9919,
      "step": 6094
    },
    {
      "epoch": 0.4925352027313683,
      "grad_norm": 3.5522258281707764,
      "learning_rate": 8.822827967412123e-06,
      "loss": 0.8532,
      "step": 6095
    },
    {
      "epoch": 0.4926160124446958,
      "grad_norm": 2.7985095977783203,
      "learning_rate": 8.82240616822004e-06,
      "loss": 0.9936,
      "step": 6096
    },
    {
      "epoch": 0.4926968221580234,
      "grad_norm": 3.1773319244384766,
      "learning_rate": 8.821984303559343e-06,
      "loss": 1.0097,
      "step": 6097
    },
    {
      "epoch": 0.49277763187135093,
      "grad_norm": 3.1122496128082275,
      "learning_rate": 8.821562373437256e-06,
      "loss": 0.9816,
      "step": 6098
    },
    {
      "epoch": 0.49285844158467845,
      "grad_norm": 3.0248422622680664,
      "learning_rate": 8.821140377861005e-06,
      "loss": 0.9274,
      "step": 6099
    },
    {
      "epoch": 0.49293925129800603,
      "grad_norm": 2.9675469398498535,
      "learning_rate": 8.820718316837818e-06,
      "loss": 0.9304,
      "step": 6100
    },
    {
      "epoch": 0.49302006101133355,
      "grad_norm": 2.454986572265625,
      "learning_rate": 8.820296190374924e-06,
      "loss": 1.0008,
      "step": 6101
    },
    {
      "epoch": 0.4931008707246611,
      "grad_norm": 2.251755714416504,
      "learning_rate": 8.819873998479554e-06,
      "loss": 1.1383,
      "step": 6102
    },
    {
      "epoch": 0.49318168043798866,
      "grad_norm": 3.004356622695923,
      "learning_rate": 8.819451741158938e-06,
      "loss": 1.0893,
      "step": 6103
    },
    {
      "epoch": 0.4932624901513162,
      "grad_norm": 2.6525914669036865,
      "learning_rate": 8.819029418420309e-06,
      "loss": 0.9572,
      "step": 6104
    },
    {
      "epoch": 0.4933432998646437,
      "grad_norm": 2.451263189315796,
      "learning_rate": 8.8186070302709e-06,
      "loss": 0.9217,
      "step": 6105
    },
    {
      "epoch": 0.4934241095779713,
      "grad_norm": 3.0996081829071045,
      "learning_rate": 8.818184576717945e-06,
      "loss": 1.0521,
      "step": 6106
    },
    {
      "epoch": 0.4935049192912988,
      "grad_norm": 2.725372314453125,
      "learning_rate": 8.81776205776868e-06,
      "loss": 0.9312,
      "step": 6107
    },
    {
      "epoch": 0.4935857290046264,
      "grad_norm": 2.304767370223999,
      "learning_rate": 8.817339473430342e-06,
      "loss": 1.0237,
      "step": 6108
    },
    {
      "epoch": 0.4936665387179539,
      "grad_norm": 2.740532636642456,
      "learning_rate": 8.816916823710168e-06,
      "loss": 0.9926,
      "step": 6109
    },
    {
      "epoch": 0.49374734843128143,
      "grad_norm": 2.321873426437378,
      "learning_rate": 8.8164941086154e-06,
      "loss": 0.9443,
      "step": 6110
    },
    {
      "epoch": 0.493828158144609,
      "grad_norm": 2.3607263565063477,
      "learning_rate": 8.816071328153275e-06,
      "loss": 1.0647,
      "step": 6111
    },
    {
      "epoch": 0.49390896785793653,
      "grad_norm": 2.1246426105499268,
      "learning_rate": 8.815648482331033e-06,
      "loss": 1.006,
      "step": 6112
    },
    {
      "epoch": 0.49398977757126405,
      "grad_norm": 2.4910242557525635,
      "learning_rate": 8.81522557115592e-06,
      "loss": 1.0039,
      "step": 6113
    },
    {
      "epoch": 0.49407058728459163,
      "grad_norm": 3.2152459621429443,
      "learning_rate": 8.814802594635177e-06,
      "loss": 0.9698,
      "step": 6114
    },
    {
      "epoch": 0.49415139699791916,
      "grad_norm": 2.749821186065674,
      "learning_rate": 8.81437955277605e-06,
      "loss": 1.0242,
      "step": 6115
    },
    {
      "epoch": 0.4942322067112467,
      "grad_norm": 2.736800193786621,
      "learning_rate": 8.813956445585784e-06,
      "loss": 0.9153,
      "step": 6116
    },
    {
      "epoch": 0.49431301642457426,
      "grad_norm": 3.20119309425354,
      "learning_rate": 8.813533273071625e-06,
      "loss": 0.9608,
      "step": 6117
    },
    {
      "epoch": 0.4943938261379018,
      "grad_norm": 2.3296122550964355,
      "learning_rate": 8.813110035240822e-06,
      "loss": 0.975,
      "step": 6118
    },
    {
      "epoch": 0.4944746358512293,
      "grad_norm": 2.2238399982452393,
      "learning_rate": 8.812686732100623e-06,
      "loss": 0.925,
      "step": 6119
    },
    {
      "epoch": 0.4945554455645569,
      "grad_norm": 3.024690866470337,
      "learning_rate": 8.81226336365828e-06,
      "loss": 0.9206,
      "step": 6120
    },
    {
      "epoch": 0.4946362552778844,
      "grad_norm": 2.7818827629089355,
      "learning_rate": 8.811839929921045e-06,
      "loss": 0.8784,
      "step": 6121
    },
    {
      "epoch": 0.49471706499121193,
      "grad_norm": 2.881441354751587,
      "learning_rate": 8.811416430896166e-06,
      "loss": 0.9981,
      "step": 6122
    },
    {
      "epoch": 0.4947978747045395,
      "grad_norm": 2.6555793285369873,
      "learning_rate": 8.8109928665909e-06,
      "loss": 0.9033,
      "step": 6123
    },
    {
      "epoch": 0.49487868441786703,
      "grad_norm": 2.7477972507476807,
      "learning_rate": 8.8105692370125e-06,
      "loss": 0.8881,
      "step": 6124
    },
    {
      "epoch": 0.49495949413119456,
      "grad_norm": 2.9489376544952393,
      "learning_rate": 8.810145542168224e-06,
      "loss": 0.9328,
      "step": 6125
    },
    {
      "epoch": 0.49504030384452213,
      "grad_norm": 2.777998685836792,
      "learning_rate": 8.809721782065326e-06,
      "loss": 0.946,
      "step": 6126
    },
    {
      "epoch": 0.49512111355784966,
      "grad_norm": 3.1331918239593506,
      "learning_rate": 8.809297956711067e-06,
      "loss": 0.9587,
      "step": 6127
    },
    {
      "epoch": 0.4952019232711772,
      "grad_norm": 3.153442621231079,
      "learning_rate": 8.808874066112702e-06,
      "loss": 1.0215,
      "step": 6128
    },
    {
      "epoch": 0.49528273298450476,
      "grad_norm": 2.8020200729370117,
      "learning_rate": 8.808450110277497e-06,
      "loss": 1.0223,
      "step": 6129
    },
    {
      "epoch": 0.4953635426978323,
      "grad_norm": 2.3491687774658203,
      "learning_rate": 8.808026089212707e-06,
      "loss": 0.8781,
      "step": 6130
    },
    {
      "epoch": 0.4954443524111598,
      "grad_norm": 2.952775478363037,
      "learning_rate": 8.8076020029256e-06,
      "loss": 0.8275,
      "step": 6131
    },
    {
      "epoch": 0.4955251621244874,
      "grad_norm": 2.6388676166534424,
      "learning_rate": 8.807177851423436e-06,
      "loss": 1.0038,
      "step": 6132
    },
    {
      "epoch": 0.4956059718378149,
      "grad_norm": 2.2663233280181885,
      "learning_rate": 8.806753634713482e-06,
      "loss": 1.0661,
      "step": 6133
    },
    {
      "epoch": 0.49568678155114243,
      "grad_norm": 2.8605127334594727,
      "learning_rate": 8.806329352803e-06,
      "loss": 0.8744,
      "step": 6134
    },
    {
      "epoch": 0.49576759126447,
      "grad_norm": 2.6653573513031006,
      "learning_rate": 8.80590500569926e-06,
      "loss": 0.987,
      "step": 6135
    },
    {
      "epoch": 0.49584840097779753,
      "grad_norm": 2.4701249599456787,
      "learning_rate": 8.805480593409532e-06,
      "loss": 0.95,
      "step": 6136
    },
    {
      "epoch": 0.49592921069112506,
      "grad_norm": 3.3145911693573,
      "learning_rate": 8.805056115941081e-06,
      "loss": 0.8803,
      "step": 6137
    },
    {
      "epoch": 0.49601002040445263,
      "grad_norm": 3.1727418899536133,
      "learning_rate": 8.804631573301179e-06,
      "loss": 1.1275,
      "step": 6138
    },
    {
      "epoch": 0.49609083011778016,
      "grad_norm": 2.9942028522491455,
      "learning_rate": 8.8042069654971e-06,
      "loss": 0.8496,
      "step": 6139
    },
    {
      "epoch": 0.4961716398311077,
      "grad_norm": 2.2804789543151855,
      "learning_rate": 8.80378229253611e-06,
      "loss": 1.0268,
      "step": 6140
    },
    {
      "epoch": 0.49625244954443526,
      "grad_norm": 2.989638328552246,
      "learning_rate": 8.803357554425489e-06,
      "loss": 0.9654,
      "step": 6141
    },
    {
      "epoch": 0.4963332592577628,
      "grad_norm": 2.3394556045532227,
      "learning_rate": 8.802932751172508e-06,
      "loss": 0.9999,
      "step": 6142
    },
    {
      "epoch": 0.4964140689710903,
      "grad_norm": 2.64165997505188,
      "learning_rate": 8.802507882784444e-06,
      "loss": 0.9509,
      "step": 6143
    },
    {
      "epoch": 0.4964948786844179,
      "grad_norm": 2.64670729637146,
      "learning_rate": 8.802082949268576e-06,
      "loss": 1.0073,
      "step": 6144
    },
    {
      "epoch": 0.4965756883977454,
      "grad_norm": 2.7997217178344727,
      "learning_rate": 8.801657950632178e-06,
      "loss": 0.8719,
      "step": 6145
    },
    {
      "epoch": 0.49665649811107293,
      "grad_norm": 2.3776872158050537,
      "learning_rate": 8.801232886882534e-06,
      "loss": 0.9945,
      "step": 6146
    },
    {
      "epoch": 0.4967373078244005,
      "grad_norm": 2.469510793685913,
      "learning_rate": 8.80080775802692e-06,
      "loss": 0.9354,
      "step": 6147
    },
    {
      "epoch": 0.49681811753772803,
      "grad_norm": 3.029223918914795,
      "learning_rate": 8.80038256407262e-06,
      "loss": 0.9409,
      "step": 6148
    },
    {
      "epoch": 0.49689892725105556,
      "grad_norm": 2.9083776473999023,
      "learning_rate": 8.799957305026915e-06,
      "loss": 0.8882,
      "step": 6149
    },
    {
      "epoch": 0.49697973696438313,
      "grad_norm": 2.4205853939056396,
      "learning_rate": 8.79953198089709e-06,
      "loss": 0.9442,
      "step": 6150
    },
    {
      "epoch": 0.49706054667771066,
      "grad_norm": 2.592292308807373,
      "learning_rate": 8.799106591690427e-06,
      "loss": 0.9678,
      "step": 6151
    },
    {
      "epoch": 0.4971413563910382,
      "grad_norm": 2.641314744949341,
      "learning_rate": 8.798681137414215e-06,
      "loss": 0.8714,
      "step": 6152
    },
    {
      "epoch": 0.49722216610436576,
      "grad_norm": 3.0298564434051514,
      "learning_rate": 8.798255618075742e-06,
      "loss": 1.0515,
      "step": 6153
    },
    {
      "epoch": 0.4973029758176933,
      "grad_norm": 2.943699598312378,
      "learning_rate": 8.797830033682293e-06,
      "loss": 1.1118,
      "step": 6154
    },
    {
      "epoch": 0.4973837855310208,
      "grad_norm": 2.3978610038757324,
      "learning_rate": 8.79740438424116e-06,
      "loss": 0.9146,
      "step": 6155
    },
    {
      "epoch": 0.4974645952443484,
      "grad_norm": 3.341064214706421,
      "learning_rate": 8.79697866975963e-06,
      "loss": 0.9141,
      "step": 6156
    },
    {
      "epoch": 0.4975454049576759,
      "grad_norm": 2.4609732627868652,
      "learning_rate": 8.796552890244996e-06,
      "loss": 1.1024,
      "step": 6157
    },
    {
      "epoch": 0.49762621467100343,
      "grad_norm": 2.2896673679351807,
      "learning_rate": 8.79612704570455e-06,
      "loss": 0.893,
      "step": 6158
    },
    {
      "epoch": 0.497707024384331,
      "grad_norm": 2.6929233074188232,
      "learning_rate": 8.795701136145588e-06,
      "loss": 0.9694,
      "step": 6159
    },
    {
      "epoch": 0.49778783409765853,
      "grad_norm": 2.648365020751953,
      "learning_rate": 8.795275161575404e-06,
      "loss": 1.099,
      "step": 6160
    },
    {
      "epoch": 0.49786864381098606,
      "grad_norm": 2.5364372730255127,
      "learning_rate": 8.794849122001293e-06,
      "loss": 0.8936,
      "step": 6161
    },
    {
      "epoch": 0.49794945352431363,
      "grad_norm": 3.0253398418426514,
      "learning_rate": 8.794423017430552e-06,
      "loss": 1.0144,
      "step": 6162
    },
    {
      "epoch": 0.49803026323764116,
      "grad_norm": 2.4398863315582275,
      "learning_rate": 8.793996847870478e-06,
      "loss": 0.9247,
      "step": 6163
    },
    {
      "epoch": 0.4981110729509687,
      "grad_norm": 2.418994426727295,
      "learning_rate": 8.793570613328373e-06,
      "loss": 1.0,
      "step": 6164
    },
    {
      "epoch": 0.49819188266429626,
      "grad_norm": 2.9876675605773926,
      "learning_rate": 8.793144313811535e-06,
      "loss": 1.0659,
      "step": 6165
    },
    {
      "epoch": 0.4982726923776238,
      "grad_norm": 2.9962754249572754,
      "learning_rate": 8.792717949327268e-06,
      "loss": 0.8499,
      "step": 6166
    },
    {
      "epoch": 0.4983535020909513,
      "grad_norm": 2.7783010005950928,
      "learning_rate": 8.792291519882873e-06,
      "loss": 0.9165,
      "step": 6167
    },
    {
      "epoch": 0.4984343118042789,
      "grad_norm": 2.6788246631622314,
      "learning_rate": 8.791865025485653e-06,
      "loss": 0.899,
      "step": 6168
    },
    {
      "epoch": 0.4985151215176064,
      "grad_norm": 3.0774552822113037,
      "learning_rate": 8.791438466142915e-06,
      "loss": 1.0353,
      "step": 6169
    },
    {
      "epoch": 0.49859593123093393,
      "grad_norm": 2.8979735374450684,
      "learning_rate": 8.791011841861961e-06,
      "loss": 1.0816,
      "step": 6170
    },
    {
      "epoch": 0.4986767409442615,
      "grad_norm": 2.9480981826782227,
      "learning_rate": 8.790585152650102e-06,
      "loss": 0.9721,
      "step": 6171
    },
    {
      "epoch": 0.49875755065758903,
      "grad_norm": 2.411952257156372,
      "learning_rate": 8.790158398514646e-06,
      "loss": 0.9462,
      "step": 6172
    },
    {
      "epoch": 0.4988383603709166,
      "grad_norm": 2.7789766788482666,
      "learning_rate": 8.7897315794629e-06,
      "loss": 0.9406,
      "step": 6173
    },
    {
      "epoch": 0.49891917008424413,
      "grad_norm": 2.9420881271362305,
      "learning_rate": 8.789304695502175e-06,
      "loss": 0.9678,
      "step": 6174
    },
    {
      "epoch": 0.49899997979757166,
      "grad_norm": 2.5352845191955566,
      "learning_rate": 8.788877746639784e-06,
      "loss": 0.8404,
      "step": 6175
    },
    {
      "epoch": 0.49908078951089924,
      "grad_norm": 2.512543201446533,
      "learning_rate": 8.788450732883037e-06,
      "loss": 1.0372,
      "step": 6176
    },
    {
      "epoch": 0.49916159922422676,
      "grad_norm": 2.5715396404266357,
      "learning_rate": 8.78802365423925e-06,
      "loss": 0.9153,
      "step": 6177
    },
    {
      "epoch": 0.4992424089375543,
      "grad_norm": 2.588712692260742,
      "learning_rate": 8.787596510715737e-06,
      "loss": 0.9617,
      "step": 6178
    },
    {
      "epoch": 0.49932321865088186,
      "grad_norm": 2.6835100650787354,
      "learning_rate": 8.787169302319816e-06,
      "loss": 0.9415,
      "step": 6179
    },
    {
      "epoch": 0.4994040283642094,
      "grad_norm": 2.8224241733551025,
      "learning_rate": 8.786742029058798e-06,
      "loss": 0.9496,
      "step": 6180
    },
    {
      "epoch": 0.4994848380775369,
      "grad_norm": 3.0012919902801514,
      "learning_rate": 8.78631469094001e-06,
      "loss": 0.9521,
      "step": 6181
    },
    {
      "epoch": 0.4995656477908645,
      "grad_norm": 2.628826856613159,
      "learning_rate": 8.785887287970764e-06,
      "loss": 1.0576,
      "step": 6182
    },
    {
      "epoch": 0.499646457504192,
      "grad_norm": 2.7362473011016846,
      "learning_rate": 8.785459820158381e-06,
      "loss": 0.9859,
      "step": 6183
    },
    {
      "epoch": 0.49972726721751953,
      "grad_norm": 3.128894567489624,
      "learning_rate": 8.785032287510188e-06,
      "loss": 1.0043,
      "step": 6184
    },
    {
      "epoch": 0.4998080769308471,
      "grad_norm": 2.8722219467163086,
      "learning_rate": 8.784604690033503e-06,
      "loss": 0.9722,
      "step": 6185
    },
    {
      "epoch": 0.49988888664417463,
      "grad_norm": 2.324587821960449,
      "learning_rate": 8.78417702773565e-06,
      "loss": 0.9709,
      "step": 6186
    },
    {
      "epoch": 0.49996969635750216,
      "grad_norm": 2.820197820663452,
      "learning_rate": 8.783749300623954e-06,
      "loss": 0.9434,
      "step": 6187
    },
    {
      "epoch": 0.5000505060708297,
      "grad_norm": 2.7109031677246094,
      "learning_rate": 8.783321508705744e-06,
      "loss": 1.0395,
      "step": 6188
    },
    {
      "epoch": 0.5001313157841573,
      "grad_norm": 2.534446954727173,
      "learning_rate": 8.782893651988342e-06,
      "loss": 0.8489,
      "step": 6189
    },
    {
      "epoch": 0.5002121254974848,
      "grad_norm": 3.0825867652893066,
      "learning_rate": 8.78246573047908e-06,
      "loss": 1.0358,
      "step": 6190
    },
    {
      "epoch": 0.5002929352108123,
      "grad_norm": 2.6846885681152344,
      "learning_rate": 8.782037744185285e-06,
      "loss": 0.961,
      "step": 6191
    },
    {
      "epoch": 0.5003737449241399,
      "grad_norm": 2.543034076690674,
      "learning_rate": 8.781609693114288e-06,
      "loss": 0.9305,
      "step": 6192
    },
    {
      "epoch": 0.5004545546374675,
      "grad_norm": 2.4893667697906494,
      "learning_rate": 8.781181577273423e-06,
      "loss": 0.8925,
      "step": 6193
    },
    {
      "epoch": 0.5005353643507949,
      "grad_norm": 2.456136465072632,
      "learning_rate": 8.780753396670019e-06,
      "loss": 0.8987,
      "step": 6194
    },
    {
      "epoch": 0.5006161740641225,
      "grad_norm": 2.920161247253418,
      "learning_rate": 8.78032515131141e-06,
      "loss": 0.9138,
      "step": 6195
    },
    {
      "epoch": 0.5006969837774501,
      "grad_norm": 2.6189510822296143,
      "learning_rate": 8.779896841204933e-06,
      "loss": 0.9793,
      "step": 6196
    },
    {
      "epoch": 0.5007777934907776,
      "grad_norm": 2.9476518630981445,
      "learning_rate": 8.779468466357923e-06,
      "loss": 0.9536,
      "step": 6197
    },
    {
      "epoch": 0.5008586032041051,
      "grad_norm": 2.7596771717071533,
      "learning_rate": 8.779040026777716e-06,
      "loss": 1.036,
      "step": 6198
    },
    {
      "epoch": 0.5009394129174327,
      "grad_norm": 2.8270766735076904,
      "learning_rate": 8.778611522471653e-06,
      "loss": 0.954,
      "step": 6199
    },
    {
      "epoch": 0.5010202226307602,
      "grad_norm": 2.723810911178589,
      "learning_rate": 8.77818295344707e-06,
      "loss": 0.9046,
      "step": 6200
    },
    {
      "epoch": 0.5011010323440878,
      "grad_norm": 2.7161903381347656,
      "learning_rate": 8.777754319711309e-06,
      "loss": 0.979,
      "step": 6201
    },
    {
      "epoch": 0.5011818420574153,
      "grad_norm": 2.6303982734680176,
      "learning_rate": 8.77732562127171e-06,
      "loss": 1.0112,
      "step": 6202
    },
    {
      "epoch": 0.5012626517707428,
      "grad_norm": 2.57110595703125,
      "learning_rate": 8.776896858135618e-06,
      "loss": 1.1584,
      "step": 6203
    },
    {
      "epoch": 0.5013434614840704,
      "grad_norm": 2.7270331382751465,
      "learning_rate": 8.776468030310375e-06,
      "loss": 0.8973,
      "step": 6204
    },
    {
      "epoch": 0.501424271197398,
      "grad_norm": 2.3623132705688477,
      "learning_rate": 8.776039137803325e-06,
      "loss": 0.9521,
      "step": 6205
    },
    {
      "epoch": 0.5015050809107254,
      "grad_norm": 2.399603843688965,
      "learning_rate": 8.775610180621816e-06,
      "loss": 0.9881,
      "step": 6206
    },
    {
      "epoch": 0.501585890624053,
      "grad_norm": 2.3874268531799316,
      "learning_rate": 8.775181158773194e-06,
      "loss": 1.0545,
      "step": 6207
    },
    {
      "epoch": 0.5016667003373806,
      "grad_norm": 2.8882362842559814,
      "learning_rate": 8.774752072264807e-06,
      "loss": 1.0007,
      "step": 6208
    },
    {
      "epoch": 0.5017475100507081,
      "grad_norm": 2.8591010570526123,
      "learning_rate": 8.774322921104003e-06,
      "loss": 0.919,
      "step": 6209
    },
    {
      "epoch": 0.5018283197640356,
      "grad_norm": 2.402244806289673,
      "learning_rate": 8.773893705298135e-06,
      "loss": 0.8777,
      "step": 6210
    },
    {
      "epoch": 0.5019091294773632,
      "grad_norm": 2.6203484535217285,
      "learning_rate": 8.773464424854553e-06,
      "loss": 0.9995,
      "step": 6211
    },
    {
      "epoch": 0.5019899391906907,
      "grad_norm": 2.413853883743286,
      "learning_rate": 8.773035079780612e-06,
      "loss": 0.8767,
      "step": 6212
    },
    {
      "epoch": 0.5020707489040183,
      "grad_norm": 2.554950714111328,
      "learning_rate": 8.77260567008366e-06,
      "loss": 0.9528,
      "step": 6213
    },
    {
      "epoch": 0.5021515586173458,
      "grad_norm": 2.642920970916748,
      "learning_rate": 8.772176195771056e-06,
      "loss": 1.0379,
      "step": 6214
    },
    {
      "epoch": 0.5022323683306733,
      "grad_norm": 2.720839023590088,
      "learning_rate": 8.771746656850156e-06,
      "loss": 1.0416,
      "step": 6215
    },
    {
      "epoch": 0.5023131780440009,
      "grad_norm": 2.906888008117676,
      "learning_rate": 8.771317053328313e-06,
      "loss": 0.9075,
      "step": 6216
    },
    {
      "epoch": 0.5023939877573285,
      "grad_norm": 2.529552698135376,
      "learning_rate": 8.77088738521289e-06,
      "loss": 0.9825,
      "step": 6217
    },
    {
      "epoch": 0.5024747974706559,
      "grad_norm": 3.1397318840026855,
      "learning_rate": 8.770457652511244e-06,
      "loss": 0.9345,
      "step": 6218
    },
    {
      "epoch": 0.5025556071839835,
      "grad_norm": 3.220461130142212,
      "learning_rate": 8.770027855230737e-06,
      "loss": 0.8796,
      "step": 6219
    },
    {
      "epoch": 0.5026364168973111,
      "grad_norm": 2.9072651863098145,
      "learning_rate": 8.769597993378728e-06,
      "loss": 0.9466,
      "step": 6220
    },
    {
      "epoch": 0.5027172266106386,
      "grad_norm": 2.3401410579681396,
      "learning_rate": 8.769168066962577e-06,
      "loss": 0.8891,
      "step": 6221
    },
    {
      "epoch": 0.5027980363239661,
      "grad_norm": 3.1132638454437256,
      "learning_rate": 8.768738075989654e-06,
      "loss": 1.0079,
      "step": 6222
    },
    {
      "epoch": 0.5028788460372937,
      "grad_norm": 2.5749056339263916,
      "learning_rate": 8.76830802046732e-06,
      "loss": 0.9403,
      "step": 6223
    },
    {
      "epoch": 0.5029596557506212,
      "grad_norm": 2.546679735183716,
      "learning_rate": 8.767877900402941e-06,
      "loss": 0.8923,
      "step": 6224
    },
    {
      "epoch": 0.5030404654639488,
      "grad_norm": 2.497368812561035,
      "learning_rate": 8.767447715803885e-06,
      "loss": 0.9375,
      "step": 6225
    },
    {
      "epoch": 0.5031212751772763,
      "grad_norm": 2.727546453475952,
      "learning_rate": 8.76701746667752e-06,
      "loss": 1.0558,
      "step": 6226
    },
    {
      "epoch": 0.5032020848906038,
      "grad_norm": 2.8279621601104736,
      "learning_rate": 8.766587153031214e-06,
      "loss": 0.8809,
      "step": 6227
    },
    {
      "epoch": 0.5032828946039314,
      "grad_norm": 2.2303953170776367,
      "learning_rate": 8.766156774872336e-06,
      "loss": 0.8804,
      "step": 6228
    },
    {
      "epoch": 0.503363704317259,
      "grad_norm": 3.113374948501587,
      "learning_rate": 8.765726332208263e-06,
      "loss": 0.9811,
      "step": 6229
    },
    {
      "epoch": 0.5034445140305864,
      "grad_norm": 3.204686164855957,
      "learning_rate": 8.765295825046359e-06,
      "loss": 0.8846,
      "step": 6230
    },
    {
      "epoch": 0.503525323743914,
      "grad_norm": 2.4541518688201904,
      "learning_rate": 8.764865253394005e-06,
      "loss": 1.006,
      "step": 6231
    },
    {
      "epoch": 0.5036061334572416,
      "grad_norm": 2.583739995956421,
      "learning_rate": 8.764434617258572e-06,
      "loss": 1.0605,
      "step": 6232
    },
    {
      "epoch": 0.5036869431705691,
      "grad_norm": 2.5617687702178955,
      "learning_rate": 8.764003916647437e-06,
      "loss": 0.9262,
      "step": 6233
    },
    {
      "epoch": 0.5037677528838966,
      "grad_norm": 2.6554017066955566,
      "learning_rate": 8.763573151567974e-06,
      "loss": 0.911,
      "step": 6234
    },
    {
      "epoch": 0.5038485625972242,
      "grad_norm": 2.6688547134399414,
      "learning_rate": 8.763142322027567e-06,
      "loss": 1.0141,
      "step": 6235
    },
    {
      "epoch": 0.5039293723105517,
      "grad_norm": 2.570295810699463,
      "learning_rate": 8.762711428033589e-06,
      "loss": 0.9379,
      "step": 6236
    },
    {
      "epoch": 0.5040101820238793,
      "grad_norm": 2.693105697631836,
      "learning_rate": 8.762280469593422e-06,
      "loss": 0.9292,
      "step": 6237
    },
    {
      "epoch": 0.5040909917372068,
      "grad_norm": 2.7524890899658203,
      "learning_rate": 8.76184944671445e-06,
      "loss": 0.9364,
      "step": 6238
    },
    {
      "epoch": 0.5041718014505343,
      "grad_norm": 2.74165940284729,
      "learning_rate": 8.761418359404053e-06,
      "loss": 0.9026,
      "step": 6239
    },
    {
      "epoch": 0.5042526111638619,
      "grad_norm": 2.6603362560272217,
      "learning_rate": 8.760987207669613e-06,
      "loss": 0.996,
      "step": 6240
    },
    {
      "epoch": 0.5043334208771895,
      "grad_norm": 2.772826671600342,
      "learning_rate": 8.760555991518519e-06,
      "loss": 1.0076,
      "step": 6241
    },
    {
      "epoch": 0.5044142305905169,
      "grad_norm": 2.823110580444336,
      "learning_rate": 8.760124710958151e-06,
      "loss": 0.9509,
      "step": 6242
    },
    {
      "epoch": 0.5044950403038445,
      "grad_norm": 2.750408172607422,
      "learning_rate": 8.7596933659959e-06,
      "loss": 0.9475,
      "step": 6243
    },
    {
      "epoch": 0.5045758500171721,
      "grad_norm": 2.5096611976623535,
      "learning_rate": 8.759261956639154e-06,
      "loss": 1.0581,
      "step": 6244
    },
    {
      "epoch": 0.5046566597304996,
      "grad_norm": 2.499408006668091,
      "learning_rate": 8.7588304828953e-06,
      "loss": 1.0803,
      "step": 6245
    },
    {
      "epoch": 0.5047374694438271,
      "grad_norm": 2.963024139404297,
      "learning_rate": 8.758398944771729e-06,
      "loss": 0.9131,
      "step": 6246
    },
    {
      "epoch": 0.5048182791571547,
      "grad_norm": 2.4039483070373535,
      "learning_rate": 8.757967342275832e-06,
      "loss": 0.9971,
      "step": 6247
    },
    {
      "epoch": 0.5048990888704822,
      "grad_norm": 2.4837539196014404,
      "learning_rate": 8.757535675415002e-06,
      "loss": 0.9081,
      "step": 6248
    },
    {
      "epoch": 0.5049798985838098,
      "grad_norm": 2.674001455307007,
      "learning_rate": 8.75710394419663e-06,
      "loss": 1.0918,
      "step": 6249
    },
    {
      "epoch": 0.5050607082971373,
      "grad_norm": 2.8628764152526855,
      "learning_rate": 8.756672148628113e-06,
      "loss": 0.9556,
      "step": 6250
    },
    {
      "epoch": 0.5051415180104648,
      "grad_norm": 3.219918727874756,
      "learning_rate": 8.756240288716845e-06,
      "loss": 0.9388,
      "step": 6251
    },
    {
      "epoch": 0.5052223277237924,
      "grad_norm": 2.65590238571167,
      "learning_rate": 8.755808364470226e-06,
      "loss": 0.977,
      "step": 6252
    },
    {
      "epoch": 0.50530313743712,
      "grad_norm": 2.408907651901245,
      "learning_rate": 8.75537637589565e-06,
      "loss": 1.1253,
      "step": 6253
    },
    {
      "epoch": 0.5053839471504474,
      "grad_norm": 2.954531192779541,
      "learning_rate": 8.754944323000516e-06,
      "loss": 0.9744,
      "step": 6254
    },
    {
      "epoch": 0.505464756863775,
      "grad_norm": 3.1942427158355713,
      "learning_rate": 8.754512205792228e-06,
      "loss": 0.8807,
      "step": 6255
    },
    {
      "epoch": 0.5055455665771026,
      "grad_norm": 2.611370801925659,
      "learning_rate": 8.754080024278184e-06,
      "loss": 0.9943,
      "step": 6256
    },
    {
      "epoch": 0.5056263762904301,
      "grad_norm": 2.339695692062378,
      "learning_rate": 8.753647778465787e-06,
      "loss": 0.8966,
      "step": 6257
    },
    {
      "epoch": 0.5057071860037576,
      "grad_norm": 2.845473051071167,
      "learning_rate": 8.753215468362437e-06,
      "loss": 0.947,
      "step": 6258
    },
    {
      "epoch": 0.5057879957170852,
      "grad_norm": 3.000074625015259,
      "learning_rate": 8.752783093975545e-06,
      "loss": 0.9843,
      "step": 6259
    },
    {
      "epoch": 0.5058688054304127,
      "grad_norm": 2.9566633701324463,
      "learning_rate": 8.75235065531251e-06,
      "loss": 0.9323,
      "step": 6260
    },
    {
      "epoch": 0.5059496151437403,
      "grad_norm": 2.8121120929718018,
      "learning_rate": 8.751918152380745e-06,
      "loss": 1.0533,
      "step": 6261
    },
    {
      "epoch": 0.5060304248570678,
      "grad_norm": 2.6486778259277344,
      "learning_rate": 8.751485585187653e-06,
      "loss": 0.9038,
      "step": 6262
    },
    {
      "epoch": 0.5061112345703953,
      "grad_norm": 2.8989672660827637,
      "learning_rate": 8.751052953740644e-06,
      "loss": 0.9592,
      "step": 6263
    },
    {
      "epoch": 0.5061920442837229,
      "grad_norm": 2.752859592437744,
      "learning_rate": 8.750620258047129e-06,
      "loss": 0.8946,
      "step": 6264
    },
    {
      "epoch": 0.5062728539970505,
      "grad_norm": 2.588942289352417,
      "learning_rate": 8.750187498114517e-06,
      "loss": 0.9424,
      "step": 6265
    },
    {
      "epoch": 0.5063536637103779,
      "grad_norm": 2.766521453857422,
      "learning_rate": 8.749754673950224e-06,
      "loss": 1.0519,
      "step": 6266
    },
    {
      "epoch": 0.5064344734237055,
      "grad_norm": 2.765406847000122,
      "learning_rate": 8.749321785561657e-06,
      "loss": 0.9219,
      "step": 6267
    },
    {
      "epoch": 0.5065152831370331,
      "grad_norm": 2.811837911605835,
      "learning_rate": 8.748888832956236e-06,
      "loss": 1.0876,
      "step": 6268
    },
    {
      "epoch": 0.5065960928503607,
      "grad_norm": 3.11313796043396,
      "learning_rate": 8.748455816141374e-06,
      "loss": 0.9263,
      "step": 6269
    },
    {
      "epoch": 0.5066769025636881,
      "grad_norm": 2.6542294025421143,
      "learning_rate": 8.74802273512449e-06,
      "loss": 0.7961,
      "step": 6270
    },
    {
      "epoch": 0.5067577122770157,
      "grad_norm": 2.3677380084991455,
      "learning_rate": 8.747589589912995e-06,
      "loss": 1.0259,
      "step": 6271
    },
    {
      "epoch": 0.5068385219903433,
      "grad_norm": 2.921304225921631,
      "learning_rate": 8.747156380514315e-06,
      "loss": 1.0164,
      "step": 6272
    },
    {
      "epoch": 0.5069193317036708,
      "grad_norm": 2.4303314685821533,
      "learning_rate": 8.746723106935867e-06,
      "loss": 0.9315,
      "step": 6273
    },
    {
      "epoch": 0.5070001414169983,
      "grad_norm": 2.752507448196411,
      "learning_rate": 8.746289769185073e-06,
      "loss": 0.9311,
      "step": 6274
    },
    {
      "epoch": 0.5070809511303259,
      "grad_norm": 3.3525657653808594,
      "learning_rate": 8.745856367269352e-06,
      "loss": 1.0205,
      "step": 6275
    },
    {
      "epoch": 0.5071617608436534,
      "grad_norm": 2.606055974960327,
      "learning_rate": 8.74542290119613e-06,
      "loss": 0.941,
      "step": 6276
    },
    {
      "epoch": 0.507242570556981,
      "grad_norm": 2.3623929023742676,
      "learning_rate": 8.744989370972831e-06,
      "loss": 0.991,
      "step": 6277
    },
    {
      "epoch": 0.5073233802703085,
      "grad_norm": 3.1393396854400635,
      "learning_rate": 8.744555776606879e-06,
      "loss": 0.9919,
      "step": 6278
    },
    {
      "epoch": 0.507404189983636,
      "grad_norm": 2.143432855606079,
      "learning_rate": 8.744122118105702e-06,
      "loss": 1.0498,
      "step": 6279
    },
    {
      "epoch": 0.5074849996969636,
      "grad_norm": 2.5479798316955566,
      "learning_rate": 8.743688395476726e-06,
      "loss": 0.8121,
      "step": 6280
    },
    {
      "epoch": 0.5075658094102912,
      "grad_norm": 2.4630141258239746,
      "learning_rate": 8.74325460872738e-06,
      "loss": 1.031,
      "step": 6281
    },
    {
      "epoch": 0.5076466191236186,
      "grad_norm": 2.621180772781372,
      "learning_rate": 8.742820757865094e-06,
      "loss": 0.892,
      "step": 6282
    },
    {
      "epoch": 0.5077274288369462,
      "grad_norm": 2.5219764709472656,
      "learning_rate": 8.742386842897302e-06,
      "loss": 1.064,
      "step": 6283
    },
    {
      "epoch": 0.5078082385502738,
      "grad_norm": 2.7024383544921875,
      "learning_rate": 8.741952863831429e-06,
      "loss": 0.9999,
      "step": 6284
    },
    {
      "epoch": 0.5078890482636013,
      "grad_norm": 3.3070693016052246,
      "learning_rate": 8.741518820674912e-06,
      "loss": 0.9797,
      "step": 6285
    },
    {
      "epoch": 0.5079698579769288,
      "grad_norm": 2.8247246742248535,
      "learning_rate": 8.741084713435187e-06,
      "loss": 0.9545,
      "step": 6286
    },
    {
      "epoch": 0.5080506676902564,
      "grad_norm": 3.0412349700927734,
      "learning_rate": 8.740650542119686e-06,
      "loss": 0.9883,
      "step": 6287
    },
    {
      "epoch": 0.5081314774035839,
      "grad_norm": 2.3945281505584717,
      "learning_rate": 8.740216306735847e-06,
      "loss": 0.9806,
      "step": 6288
    },
    {
      "epoch": 0.5082122871169115,
      "grad_norm": 2.5237603187561035,
      "learning_rate": 8.739782007291107e-06,
      "loss": 0.908,
      "step": 6289
    },
    {
      "epoch": 0.508293096830239,
      "grad_norm": 2.7784175872802734,
      "learning_rate": 8.739347643792904e-06,
      "loss": 0.9167,
      "step": 6290
    },
    {
      "epoch": 0.5083739065435665,
      "grad_norm": 2.8435075283050537,
      "learning_rate": 8.738913216248678e-06,
      "loss": 1.0108,
      "step": 6291
    },
    {
      "epoch": 0.5084547162568941,
      "grad_norm": 2.382512092590332,
      "learning_rate": 8.73847872466587e-06,
      "loss": 1.0241,
      "step": 6292
    },
    {
      "epoch": 0.5085355259702217,
      "grad_norm": 2.925574541091919,
      "learning_rate": 8.73804416905192e-06,
      "loss": 0.9335,
      "step": 6293
    },
    {
      "epoch": 0.5086163356835491,
      "grad_norm": 2.9449074268341064,
      "learning_rate": 8.737609549414274e-06,
      "loss": 0.955,
      "step": 6294
    },
    {
      "epoch": 0.5086971453968767,
      "grad_norm": 2.4350380897521973,
      "learning_rate": 8.737174865760374e-06,
      "loss": 0.9562,
      "step": 6295
    },
    {
      "epoch": 0.5087779551102043,
      "grad_norm": 2.7059905529022217,
      "learning_rate": 8.736740118097665e-06,
      "loss": 1.0349,
      "step": 6296
    },
    {
      "epoch": 0.5088587648235318,
      "grad_norm": 3.114983558654785,
      "learning_rate": 8.736305306433595e-06,
      "loss": 1.0846,
      "step": 6297
    },
    {
      "epoch": 0.5089395745368593,
      "grad_norm": 3.2726974487304688,
      "learning_rate": 8.735870430775609e-06,
      "loss": 0.9581,
      "step": 6298
    },
    {
      "epoch": 0.5090203842501869,
      "grad_norm": 2.6871676445007324,
      "learning_rate": 8.735435491131155e-06,
      "loss": 0.8718,
      "step": 6299
    },
    {
      "epoch": 0.5091011939635144,
      "grad_norm": 2.542513608932495,
      "learning_rate": 8.735000487507684e-06,
      "loss": 0.929,
      "step": 6300
    },
    {
      "epoch": 0.509182003676842,
      "grad_norm": 2.7970104217529297,
      "learning_rate": 8.734565419912649e-06,
      "loss": 0.9334,
      "step": 6301
    },
    {
      "epoch": 0.5092628133901695,
      "grad_norm": 2.6967902183532715,
      "learning_rate": 8.734130288353495e-06,
      "loss": 1.0328,
      "step": 6302
    },
    {
      "epoch": 0.509343623103497,
      "grad_norm": 2.9185562133789062,
      "learning_rate": 8.733695092837681e-06,
      "loss": 0.9334,
      "step": 6303
    },
    {
      "epoch": 0.5094244328168246,
      "grad_norm": 2.716060161590576,
      "learning_rate": 8.73325983337266e-06,
      "loss": 1.0498,
      "step": 6304
    },
    {
      "epoch": 0.5095052425301522,
      "grad_norm": 2.8402271270751953,
      "learning_rate": 8.732824509965882e-06,
      "loss": 1.0054,
      "step": 6305
    },
    {
      "epoch": 0.5095860522434796,
      "grad_norm": 2.672893762588501,
      "learning_rate": 8.732389122624809e-06,
      "loss": 1.0344,
      "step": 6306
    },
    {
      "epoch": 0.5096668619568072,
      "grad_norm": 2.799960136413574,
      "learning_rate": 8.731953671356895e-06,
      "loss": 0.9002,
      "step": 6307
    },
    {
      "epoch": 0.5097476716701348,
      "grad_norm": 2.4315521717071533,
      "learning_rate": 8.7315181561696e-06,
      "loss": 0.9639,
      "step": 6308
    },
    {
      "epoch": 0.5098284813834623,
      "grad_norm": 2.9446537494659424,
      "learning_rate": 8.73108257707038e-06,
      "loss": 0.9541,
      "step": 6309
    },
    {
      "epoch": 0.5099092910967898,
      "grad_norm": 2.6569461822509766,
      "learning_rate": 8.730646934066699e-06,
      "loss": 0.9573,
      "step": 6310
    },
    {
      "epoch": 0.5099901008101174,
      "grad_norm": 3.246030807495117,
      "learning_rate": 8.730211227166017e-06,
      "loss": 1.0095,
      "step": 6311
    },
    {
      "epoch": 0.5100709105234449,
      "grad_norm": 2.408226490020752,
      "learning_rate": 8.729775456375798e-06,
      "loss": 0.9738,
      "step": 6312
    },
    {
      "epoch": 0.5101517202367725,
      "grad_norm": 3.040785789489746,
      "learning_rate": 8.729339621703502e-06,
      "loss": 0.9502,
      "step": 6313
    },
    {
      "epoch": 0.5102325299501,
      "grad_norm": 2.7272884845733643,
      "learning_rate": 8.728903723156598e-06,
      "loss": 0.9586,
      "step": 6314
    },
    {
      "epoch": 0.5103133396634275,
      "grad_norm": 2.6391963958740234,
      "learning_rate": 8.72846776074255e-06,
      "loss": 0.9756,
      "step": 6315
    },
    {
      "epoch": 0.5103941493767551,
      "grad_norm": 2.7884223461151123,
      "learning_rate": 8.728031734468825e-06,
      "loss": 0.9851,
      "step": 6316
    },
    {
      "epoch": 0.5104749590900827,
      "grad_norm": 2.6884872913360596,
      "learning_rate": 8.727595644342892e-06,
      "loss": 0.8943,
      "step": 6317
    },
    {
      "epoch": 0.5105557688034101,
      "grad_norm": 2.61759614944458,
      "learning_rate": 8.72715949037222e-06,
      "loss": 1.0132,
      "step": 6318
    },
    {
      "epoch": 0.5106365785167377,
      "grad_norm": 2.98608660697937,
      "learning_rate": 8.726723272564274e-06,
      "loss": 0.9991,
      "step": 6319
    },
    {
      "epoch": 0.5107173882300653,
      "grad_norm": 2.5214684009552,
      "learning_rate": 8.726286990926537e-06,
      "loss": 0.979,
      "step": 6320
    },
    {
      "epoch": 0.5107981979433928,
      "grad_norm": 3.056746482849121,
      "learning_rate": 8.725850645466469e-06,
      "loss": 0.9909,
      "step": 6321
    },
    {
      "epoch": 0.5108790076567203,
      "grad_norm": 2.549048662185669,
      "learning_rate": 8.725414236191552e-06,
      "loss": 0.9043,
      "step": 6322
    },
    {
      "epoch": 0.5109598173700479,
      "grad_norm": 3.2244904041290283,
      "learning_rate": 8.724977763109256e-06,
      "loss": 0.8554,
      "step": 6323
    },
    {
      "epoch": 0.5110406270833754,
      "grad_norm": 2.591453790664673,
      "learning_rate": 8.724541226227059e-06,
      "loss": 0.9035,
      "step": 6324
    },
    {
      "epoch": 0.511121436796703,
      "grad_norm": 2.7734375,
      "learning_rate": 8.724104625552437e-06,
      "loss": 0.9175,
      "step": 6325
    },
    {
      "epoch": 0.5112022465100305,
      "grad_norm": 2.62668776512146,
      "learning_rate": 8.72366796109287e-06,
      "loss": 0.9148,
      "step": 6326
    },
    {
      "epoch": 0.511283056223358,
      "grad_norm": 2.4754507541656494,
      "learning_rate": 8.723231232855833e-06,
      "loss": 1.0398,
      "step": 6327
    },
    {
      "epoch": 0.5113638659366856,
      "grad_norm": 2.5738584995269775,
      "learning_rate": 8.72279444084881e-06,
      "loss": 0.9858,
      "step": 6328
    },
    {
      "epoch": 0.5114446756500132,
      "grad_norm": 2.735138416290283,
      "learning_rate": 8.72235758507928e-06,
      "loss": 0.9875,
      "step": 6329
    },
    {
      "epoch": 0.5115254853633406,
      "grad_norm": 2.754800796508789,
      "learning_rate": 8.721920665554724e-06,
      "loss": 0.8941,
      "step": 6330
    },
    {
      "epoch": 0.5116062950766682,
      "grad_norm": 2.9557249546051025,
      "learning_rate": 8.721483682282628e-06,
      "loss": 0.8747,
      "step": 6331
    },
    {
      "epoch": 0.5116871047899958,
      "grad_norm": 2.5468697547912598,
      "learning_rate": 8.721046635270478e-06,
      "loss": 0.8815,
      "step": 6332
    },
    {
      "epoch": 0.5117679145033233,
      "grad_norm": 2.746835947036743,
      "learning_rate": 8.720609524525754e-06,
      "loss": 1.001,
      "step": 6333
    },
    {
      "epoch": 0.5118487242166508,
      "grad_norm": 3.068387985229492,
      "learning_rate": 8.720172350055947e-06,
      "loss": 0.9632,
      "step": 6334
    },
    {
      "epoch": 0.5119295339299784,
      "grad_norm": 3.0410642623901367,
      "learning_rate": 8.719735111868544e-06,
      "loss": 0.9902,
      "step": 6335
    },
    {
      "epoch": 0.5120103436433059,
      "grad_norm": 3.010000467300415,
      "learning_rate": 8.719297809971034e-06,
      "loss": 1.0023,
      "step": 6336
    },
    {
      "epoch": 0.5120911533566335,
      "grad_norm": 2.6686947345733643,
      "learning_rate": 8.718860444370905e-06,
      "loss": 0.896,
      "step": 6337
    },
    {
      "epoch": 0.512171963069961,
      "grad_norm": 2.974932909011841,
      "learning_rate": 8.71842301507565e-06,
      "loss": 0.8491,
      "step": 6338
    },
    {
      "epoch": 0.5122527727832885,
      "grad_norm": 3.2390475273132324,
      "learning_rate": 8.71798552209276e-06,
      "loss": 0.9092,
      "step": 6339
    },
    {
      "epoch": 0.5123335824966161,
      "grad_norm": 2.7157886028289795,
      "learning_rate": 8.71754796542973e-06,
      "loss": 0.9994,
      "step": 6340
    },
    {
      "epoch": 0.5124143922099437,
      "grad_norm": 3.2451345920562744,
      "learning_rate": 8.717110345094053e-06,
      "loss": 0.9487,
      "step": 6341
    },
    {
      "epoch": 0.5124952019232711,
      "grad_norm": 2.884882926940918,
      "learning_rate": 8.716672661093222e-06,
      "loss": 1.073,
      "step": 6342
    },
    {
      "epoch": 0.5125760116365987,
      "grad_norm": 2.51513409614563,
      "learning_rate": 8.716234913434738e-06,
      "loss": 1.0276,
      "step": 6343
    },
    {
      "epoch": 0.5126568213499263,
      "grad_norm": 2.5609960556030273,
      "learning_rate": 8.715797102126096e-06,
      "loss": 0.9558,
      "step": 6344
    },
    {
      "epoch": 0.5127376310632538,
      "grad_norm": 2.093696355819702,
      "learning_rate": 8.715359227174795e-06,
      "loss": 1.0238,
      "step": 6345
    },
    {
      "epoch": 0.5128184407765813,
      "grad_norm": 2.7014880180358887,
      "learning_rate": 8.714921288588334e-06,
      "loss": 0.8161,
      "step": 6346
    },
    {
      "epoch": 0.5128992504899089,
      "grad_norm": 2.665351390838623,
      "learning_rate": 8.714483286374216e-06,
      "loss": 1.0244,
      "step": 6347
    },
    {
      "epoch": 0.5129800602032364,
      "grad_norm": 2.641115665435791,
      "learning_rate": 8.714045220539939e-06,
      "loss": 0.9913,
      "step": 6348
    },
    {
      "epoch": 0.513060869916564,
      "grad_norm": 2.4953019618988037,
      "learning_rate": 8.713607091093011e-06,
      "loss": 0.9176,
      "step": 6349
    },
    {
      "epoch": 0.5131416796298915,
      "grad_norm": 3.101696014404297,
      "learning_rate": 8.713168898040933e-06,
      "loss": 0.8918,
      "step": 6350
    },
    {
      "epoch": 0.513222489343219,
      "grad_norm": 3.100341320037842,
      "learning_rate": 8.712730641391212e-06,
      "loss": 0.8482,
      "step": 6351
    },
    {
      "epoch": 0.5133032990565466,
      "grad_norm": 2.8223681449890137,
      "learning_rate": 8.712292321151352e-06,
      "loss": 0.9577,
      "step": 6352
    },
    {
      "epoch": 0.5133841087698742,
      "grad_norm": 2.632502794265747,
      "learning_rate": 8.711853937328862e-06,
      "loss": 0.9707,
      "step": 6353
    },
    {
      "epoch": 0.5134649184832016,
      "grad_norm": 2.548308849334717,
      "learning_rate": 8.71141548993125e-06,
      "loss": 0.9325,
      "step": 6354
    },
    {
      "epoch": 0.5135457281965292,
      "grad_norm": 2.7708146572113037,
      "learning_rate": 8.710976978966024e-06,
      "loss": 0.9812,
      "step": 6355
    },
    {
      "epoch": 0.5136265379098568,
      "grad_norm": 2.337926149368286,
      "learning_rate": 8.710538404440697e-06,
      "loss": 0.9608,
      "step": 6356
    },
    {
      "epoch": 0.5137073476231843,
      "grad_norm": 2.679847240447998,
      "learning_rate": 8.71009976636278e-06,
      "loss": 0.9442,
      "step": 6357
    },
    {
      "epoch": 0.5137881573365118,
      "grad_norm": 2.399090528488159,
      "learning_rate": 8.709661064739786e-06,
      "loss": 0.9894,
      "step": 6358
    },
    {
      "epoch": 0.5138689670498394,
      "grad_norm": 2.6882615089416504,
      "learning_rate": 8.70922229957923e-06,
      "loss": 0.9342,
      "step": 6359
    },
    {
      "epoch": 0.5139497767631669,
      "grad_norm": 3.1053476333618164,
      "learning_rate": 8.708783470888621e-06,
      "loss": 0.9774,
      "step": 6360
    },
    {
      "epoch": 0.5140305864764945,
      "grad_norm": 2.697941780090332,
      "learning_rate": 8.708344578675486e-06,
      "loss": 0.9726,
      "step": 6361
    },
    {
      "epoch": 0.514111396189822,
      "grad_norm": 3.2440953254699707,
      "learning_rate": 8.70790562294733e-06,
      "loss": 1.0524,
      "step": 6362
    },
    {
      "epoch": 0.5141922059031495,
      "grad_norm": 2.412670135498047,
      "learning_rate": 8.70746660371168e-06,
      "loss": 0.9276,
      "step": 6363
    },
    {
      "epoch": 0.5142730156164771,
      "grad_norm": 2.728304624557495,
      "learning_rate": 8.707027520976053e-06,
      "loss": 0.9257,
      "step": 6364
    },
    {
      "epoch": 0.5143538253298047,
      "grad_norm": 2.4933555126190186,
      "learning_rate": 8.706588374747967e-06,
      "loss": 0.8722,
      "step": 6365
    },
    {
      "epoch": 0.5144346350431321,
      "grad_norm": 2.3907010555267334,
      "learning_rate": 8.706149165034948e-06,
      "loss": 0.9703,
      "step": 6366
    },
    {
      "epoch": 0.5145154447564597,
      "grad_norm": 2.875530958175659,
      "learning_rate": 8.705709891844514e-06,
      "loss": 0.9495,
      "step": 6367
    },
    {
      "epoch": 0.5145962544697873,
      "grad_norm": 2.625635862350464,
      "learning_rate": 8.70527055518419e-06,
      "loss": 0.8828,
      "step": 6368
    },
    {
      "epoch": 0.5146770641831148,
      "grad_norm": 2.691448211669922,
      "learning_rate": 8.704831155061504e-06,
      "loss": 0.9231,
      "step": 6369
    },
    {
      "epoch": 0.5147578738964423,
      "grad_norm": 2.423642873764038,
      "learning_rate": 8.704391691483977e-06,
      "loss": 0.9891,
      "step": 6370
    },
    {
      "epoch": 0.5148386836097699,
      "grad_norm": 2.7718937397003174,
      "learning_rate": 8.70395216445914e-06,
      "loss": 1.0474,
      "step": 6371
    },
    {
      "epoch": 0.5149194933230974,
      "grad_norm": 2.699124574661255,
      "learning_rate": 8.703512573994516e-06,
      "loss": 0.9036,
      "step": 6372
    },
    {
      "epoch": 0.515000303036425,
      "grad_norm": 2.8082897663116455,
      "learning_rate": 8.703072920097641e-06,
      "loss": 0.9693,
      "step": 6373
    },
    {
      "epoch": 0.5150811127497525,
      "grad_norm": 2.2477915287017822,
      "learning_rate": 8.702633202776041e-06,
      "loss": 0.9756,
      "step": 6374
    },
    {
      "epoch": 0.51516192246308,
      "grad_norm": 2.8851675987243652,
      "learning_rate": 8.702193422037248e-06,
      "loss": 0.926,
      "step": 6375
    },
    {
      "epoch": 0.5152427321764076,
      "grad_norm": 2.42008376121521,
      "learning_rate": 8.701753577888792e-06,
      "loss": 0.9037,
      "step": 6376
    },
    {
      "epoch": 0.5153235418897352,
      "grad_norm": 2.554056167602539,
      "learning_rate": 8.701313670338212e-06,
      "loss": 0.9475,
      "step": 6377
    },
    {
      "epoch": 0.5154043516030626,
      "grad_norm": 3.2706005573272705,
      "learning_rate": 8.700873699393037e-06,
      "loss": 1.0666,
      "step": 6378
    },
    {
      "epoch": 0.5154851613163902,
      "grad_norm": 2.3901360034942627,
      "learning_rate": 8.700433665060806e-06,
      "loss": 0.9394,
      "step": 6379
    },
    {
      "epoch": 0.5155659710297178,
      "grad_norm": 2.8380510807037354,
      "learning_rate": 8.699993567349055e-06,
      "loss": 0.9402,
      "step": 6380
    },
    {
      "epoch": 0.5156467807430453,
      "grad_norm": 2.676969528198242,
      "learning_rate": 8.699553406265321e-06,
      "loss": 0.9905,
      "step": 6381
    },
    {
      "epoch": 0.5157275904563728,
      "grad_norm": 2.6981847286224365,
      "learning_rate": 8.699113181817145e-06,
      "loss": 0.8519,
      "step": 6382
    },
    {
      "epoch": 0.5158084001697004,
      "grad_norm": 2.890380382537842,
      "learning_rate": 8.698672894012063e-06,
      "loss": 0.9418,
      "step": 6383
    },
    {
      "epoch": 0.5158892098830279,
      "grad_norm": 2.671825408935547,
      "learning_rate": 8.69823254285762e-06,
      "loss": 0.9275,
      "step": 6384
    },
    {
      "epoch": 0.5159700195963555,
      "grad_norm": 2.573758602142334,
      "learning_rate": 8.697792128361358e-06,
      "loss": 1.0287,
      "step": 6385
    },
    {
      "epoch": 0.516050829309683,
      "grad_norm": 2.424605131149292,
      "learning_rate": 8.697351650530816e-06,
      "loss": 0.9835,
      "step": 6386
    },
    {
      "epoch": 0.5161316390230105,
      "grad_norm": 2.4024574756622314,
      "learning_rate": 8.696911109373544e-06,
      "loss": 0.9429,
      "step": 6387
    },
    {
      "epoch": 0.5162124487363381,
      "grad_norm": 2.7372288703918457,
      "learning_rate": 8.696470504897084e-06,
      "loss": 0.986,
      "step": 6388
    },
    {
      "epoch": 0.5162932584496657,
      "grad_norm": 2.214829683303833,
      "learning_rate": 8.696029837108981e-06,
      "loss": 0.9655,
      "step": 6389
    },
    {
      "epoch": 0.5163740681629931,
      "grad_norm": 2.0110013484954834,
      "learning_rate": 8.695589106016787e-06,
      "loss": 0.9664,
      "step": 6390
    },
    {
      "epoch": 0.5164548778763207,
      "grad_norm": 2.6689956188201904,
      "learning_rate": 8.695148311628047e-06,
      "loss": 0.9042,
      "step": 6391
    },
    {
      "epoch": 0.5165356875896483,
      "grad_norm": 3.2499120235443115,
      "learning_rate": 8.694707453950312e-06,
      "loss": 1.0567,
      "step": 6392
    },
    {
      "epoch": 0.5166164973029758,
      "grad_norm": 2.2970924377441406,
      "learning_rate": 8.694266532991133e-06,
      "loss": 0.9115,
      "step": 6393
    },
    {
      "epoch": 0.5166973070163033,
      "grad_norm": 2.5825273990631104,
      "learning_rate": 8.693825548758064e-06,
      "loss": 0.9039,
      "step": 6394
    },
    {
      "epoch": 0.5167781167296309,
      "grad_norm": 2.4262425899505615,
      "learning_rate": 8.693384501258653e-06,
      "loss": 1.1179,
      "step": 6395
    },
    {
      "epoch": 0.5168589264429584,
      "grad_norm": 2.5739622116088867,
      "learning_rate": 8.69294339050046e-06,
      "loss": 0.9481,
      "step": 6396
    },
    {
      "epoch": 0.516939736156286,
      "grad_norm": 2.987832546234131,
      "learning_rate": 8.692502216491034e-06,
      "loss": 0.968,
      "step": 6397
    },
    {
      "epoch": 0.5170205458696135,
      "grad_norm": 2.867089033126831,
      "learning_rate": 8.692060979237936e-06,
      "loss": 1.1234,
      "step": 6398
    },
    {
      "epoch": 0.5171013555829411,
      "grad_norm": 2.766071319580078,
      "learning_rate": 8.691619678748722e-06,
      "loss": 0.9993,
      "step": 6399
    },
    {
      "epoch": 0.5171821652962686,
      "grad_norm": 2.8523569107055664,
      "learning_rate": 8.69117831503095e-06,
      "loss": 0.9752,
      "step": 6400
    },
    {
      "epoch": 0.5172629750095962,
      "grad_norm": 2.809799909591675,
      "learning_rate": 8.69073688809218e-06,
      "loss": 0.9286,
      "step": 6401
    },
    {
      "epoch": 0.5173437847229237,
      "grad_norm": 2.233978748321533,
      "learning_rate": 8.69029539793997e-06,
      "loss": 0.9414,
      "step": 6402
    },
    {
      "epoch": 0.5174245944362512,
      "grad_norm": 2.607534646987915,
      "learning_rate": 8.689853844581886e-06,
      "loss": 1.0514,
      "step": 6403
    },
    {
      "epoch": 0.5175054041495788,
      "grad_norm": 2.491774559020996,
      "learning_rate": 8.689412228025487e-06,
      "loss": 1.0021,
      "step": 6404
    },
    {
      "epoch": 0.5175862138629064,
      "grad_norm": 2.586745262145996,
      "learning_rate": 8.688970548278339e-06,
      "loss": 1.0436,
      "step": 6405
    },
    {
      "epoch": 0.5176670235762338,
      "grad_norm": 2.635716676712036,
      "learning_rate": 8.688528805348008e-06,
      "loss": 0.9233,
      "step": 6406
    },
    {
      "epoch": 0.5177478332895614,
      "grad_norm": 2.7674312591552734,
      "learning_rate": 8.688086999242056e-06,
      "loss": 0.8556,
      "step": 6407
    },
    {
      "epoch": 0.517828643002889,
      "grad_norm": 2.457615375518799,
      "learning_rate": 8.687645129968054e-06,
      "loss": 0.9286,
      "step": 6408
    },
    {
      "epoch": 0.5179094527162165,
      "grad_norm": 3.614577054977417,
      "learning_rate": 8.687203197533567e-06,
      "loss": 0.9653,
      "step": 6409
    },
    {
      "epoch": 0.517990262429544,
      "grad_norm": 2.7701103687286377,
      "learning_rate": 8.686761201946168e-06,
      "loss": 1.0364,
      "step": 6410
    },
    {
      "epoch": 0.5180710721428716,
      "grad_norm": 2.433927297592163,
      "learning_rate": 8.686319143213424e-06,
      "loss": 0.9752,
      "step": 6411
    },
    {
      "epoch": 0.5181518818561991,
      "grad_norm": 2.5998244285583496,
      "learning_rate": 8.685877021342907e-06,
      "loss": 0.7826,
      "step": 6412
    },
    {
      "epoch": 0.5182326915695267,
      "grad_norm": 2.511528253555298,
      "learning_rate": 8.68543483634219e-06,
      "loss": 0.9163,
      "step": 6413
    },
    {
      "epoch": 0.5183135012828542,
      "grad_norm": 2.436471462249756,
      "learning_rate": 8.684992588218848e-06,
      "loss": 1.1039,
      "step": 6414
    },
    {
      "epoch": 0.5183943109961817,
      "grad_norm": 2.6319479942321777,
      "learning_rate": 8.684550276980453e-06,
      "loss": 1.0089,
      "step": 6415
    },
    {
      "epoch": 0.5184751207095093,
      "grad_norm": 2.991459369659424,
      "learning_rate": 8.684107902634581e-06,
      "loss": 0.939,
      "step": 6416
    },
    {
      "epoch": 0.5185559304228369,
      "grad_norm": 2.6421146392822266,
      "learning_rate": 8.683665465188811e-06,
      "loss": 0.9869,
      "step": 6417
    },
    {
      "epoch": 0.5186367401361643,
      "grad_norm": 2.276886224746704,
      "learning_rate": 8.683222964650721e-06,
      "loss": 0.8912,
      "step": 6418
    },
    {
      "epoch": 0.5187175498494919,
      "grad_norm": 2.6830523014068604,
      "learning_rate": 8.682780401027886e-06,
      "loss": 0.8998,
      "step": 6419
    },
    {
      "epoch": 0.5187983595628195,
      "grad_norm": 3.1952884197235107,
      "learning_rate": 8.68233777432789e-06,
      "loss": 1.0521,
      "step": 6420
    },
    {
      "epoch": 0.518879169276147,
      "grad_norm": 2.9305951595306396,
      "learning_rate": 8.681895084558314e-06,
      "loss": 0.9575,
      "step": 6421
    },
    {
      "epoch": 0.5189599789894745,
      "grad_norm": 2.691216230392456,
      "learning_rate": 8.681452331726737e-06,
      "loss": 0.8598,
      "step": 6422
    },
    {
      "epoch": 0.5190407887028021,
      "grad_norm": 2.364846706390381,
      "learning_rate": 8.681009515840744e-06,
      "loss": 1.0365,
      "step": 6423
    },
    {
      "epoch": 0.5191215984161296,
      "grad_norm": 2.915999174118042,
      "learning_rate": 8.680566636907922e-06,
      "loss": 0.8954,
      "step": 6424
    },
    {
      "epoch": 0.5192024081294572,
      "grad_norm": 2.4738638401031494,
      "learning_rate": 8.680123694935852e-06,
      "loss": 0.8674,
      "step": 6425
    },
    {
      "epoch": 0.5192832178427848,
      "grad_norm": 2.6491429805755615,
      "learning_rate": 8.679680689932123e-06,
      "loss": 0.9489,
      "step": 6426
    },
    {
      "epoch": 0.5193640275561122,
      "grad_norm": 2.5052988529205322,
      "learning_rate": 8.679237621904324e-06,
      "loss": 0.9889,
      "step": 6427
    },
    {
      "epoch": 0.5194448372694398,
      "grad_norm": 2.9090888500213623,
      "learning_rate": 8.678794490860039e-06,
      "loss": 1.0662,
      "step": 6428
    },
    {
      "epoch": 0.5195256469827674,
      "grad_norm": 2.6907501220703125,
      "learning_rate": 8.678351296806863e-06,
      "loss": 1.0126,
      "step": 6429
    },
    {
      "epoch": 0.5196064566960948,
      "grad_norm": 3.101341962814331,
      "learning_rate": 8.677908039752383e-06,
      "loss": 0.9364,
      "step": 6430
    },
    {
      "epoch": 0.5196872664094224,
      "grad_norm": 2.7768609523773193,
      "learning_rate": 8.677464719704194e-06,
      "loss": 0.8984,
      "step": 6431
    },
    {
      "epoch": 0.51976807612275,
      "grad_norm": 2.6042120456695557,
      "learning_rate": 8.677021336669887e-06,
      "loss": 0.9307,
      "step": 6432
    },
    {
      "epoch": 0.5198488858360775,
      "grad_norm": 3.216299295425415,
      "learning_rate": 8.676577890657056e-06,
      "loss": 1.0496,
      "step": 6433
    },
    {
      "epoch": 0.519929695549405,
      "grad_norm": 2.657881736755371,
      "learning_rate": 8.676134381673296e-06,
      "loss": 0.9144,
      "step": 6434
    },
    {
      "epoch": 0.5200105052627326,
      "grad_norm": 2.426283836364746,
      "learning_rate": 8.675690809726206e-06,
      "loss": 0.9773,
      "step": 6435
    },
    {
      "epoch": 0.5200913149760601,
      "grad_norm": 3.0907323360443115,
      "learning_rate": 8.67524717482338e-06,
      "loss": 0.9152,
      "step": 6436
    },
    {
      "epoch": 0.5201721246893877,
      "grad_norm": 2.756096839904785,
      "learning_rate": 8.674803476972418e-06,
      "loss": 1.0208,
      "step": 6437
    },
    {
      "epoch": 0.5202529344027153,
      "grad_norm": 2.7864134311676025,
      "learning_rate": 8.67435971618092e-06,
      "loss": 1.1348,
      "step": 6438
    },
    {
      "epoch": 0.5203337441160427,
      "grad_norm": 2.7034497261047363,
      "learning_rate": 8.673915892456484e-06,
      "loss": 0.9545,
      "step": 6439
    },
    {
      "epoch": 0.5204145538293703,
      "grad_norm": 2.788236618041992,
      "learning_rate": 8.673472005806715e-06,
      "loss": 0.9214,
      "step": 6440
    },
    {
      "epoch": 0.5204953635426979,
      "grad_norm": 2.6394190788269043,
      "learning_rate": 8.673028056239213e-06,
      "loss": 0.8938,
      "step": 6441
    },
    {
      "epoch": 0.5205761732560253,
      "grad_norm": 2.9849536418914795,
      "learning_rate": 8.672584043761583e-06,
      "loss": 1.0054,
      "step": 6442
    },
    {
      "epoch": 0.5206569829693529,
      "grad_norm": 2.5224430561065674,
      "learning_rate": 8.67213996838143e-06,
      "loss": 0.8606,
      "step": 6443
    },
    {
      "epoch": 0.5207377926826805,
      "grad_norm": 2.4942100048065186,
      "learning_rate": 8.67169583010636e-06,
      "loss": 0.9383,
      "step": 6444
    },
    {
      "epoch": 0.520818602396008,
      "grad_norm": 2.6450839042663574,
      "learning_rate": 8.67125162894398e-06,
      "loss": 1.0274,
      "step": 6445
    },
    {
      "epoch": 0.5208994121093355,
      "grad_norm": 2.9041450023651123,
      "learning_rate": 8.670807364901896e-06,
      "loss": 1.1383,
      "step": 6446
    },
    {
      "epoch": 0.5209802218226631,
      "grad_norm": 2.6695570945739746,
      "learning_rate": 8.67036303798772e-06,
      "loss": 0.9661,
      "step": 6447
    },
    {
      "epoch": 0.5210610315359906,
      "grad_norm": 2.626668691635132,
      "learning_rate": 8.669918648209062e-06,
      "loss": 0.9429,
      "step": 6448
    },
    {
      "epoch": 0.5211418412493182,
      "grad_norm": 3.0560965538024902,
      "learning_rate": 8.66947419557353e-06,
      "loss": 0.9078,
      "step": 6449
    },
    {
      "epoch": 0.5212226509626458,
      "grad_norm": 2.9653520584106445,
      "learning_rate": 8.66902968008874e-06,
      "loss": 1.0656,
      "step": 6450
    },
    {
      "epoch": 0.5213034606759732,
      "grad_norm": 2.8498342037200928,
      "learning_rate": 8.668585101762305e-06,
      "loss": 0.9314,
      "step": 6451
    },
    {
      "epoch": 0.5213842703893008,
      "grad_norm": 3.286355972290039,
      "learning_rate": 8.668140460601841e-06,
      "loss": 1.057,
      "step": 6452
    },
    {
      "epoch": 0.5214650801026284,
      "grad_norm": 2.399308204650879,
      "learning_rate": 8.66769575661496e-06,
      "loss": 0.9351,
      "step": 6453
    },
    {
      "epoch": 0.5215458898159558,
      "grad_norm": 2.535555362701416,
      "learning_rate": 8.667250989809279e-06,
      "loss": 0.9142,
      "step": 6454
    },
    {
      "epoch": 0.5216266995292834,
      "grad_norm": 2.7442376613616943,
      "learning_rate": 8.666806160192419e-06,
      "loss": 1.0837,
      "step": 6455
    },
    {
      "epoch": 0.521707509242611,
      "grad_norm": 2.702324151992798,
      "learning_rate": 8.666361267771994e-06,
      "loss": 1.1003,
      "step": 6456
    },
    {
      "epoch": 0.5217883189559385,
      "grad_norm": 2.9018096923828125,
      "learning_rate": 8.66591631255563e-06,
      "loss": 0.9353,
      "step": 6457
    },
    {
      "epoch": 0.521869128669266,
      "grad_norm": 2.298851728439331,
      "learning_rate": 8.665471294550943e-06,
      "loss": 1.1122,
      "step": 6458
    },
    {
      "epoch": 0.5219499383825936,
      "grad_norm": 2.289459705352783,
      "learning_rate": 8.665026213765558e-06,
      "loss": 0.929,
      "step": 6459
    },
    {
      "epoch": 0.5220307480959211,
      "grad_norm": 3.1866111755371094,
      "learning_rate": 8.664581070207098e-06,
      "loss": 0.9406,
      "step": 6460
    },
    {
      "epoch": 0.5221115578092487,
      "grad_norm": 2.3597025871276855,
      "learning_rate": 8.664135863883185e-06,
      "loss": 0.9293,
      "step": 6461
    },
    {
      "epoch": 0.5221923675225763,
      "grad_norm": 2.3700859546661377,
      "learning_rate": 8.663690594801446e-06,
      "loss": 0.8038,
      "step": 6462
    },
    {
      "epoch": 0.5222731772359037,
      "grad_norm": 2.5032904148101807,
      "learning_rate": 8.663245262969507e-06,
      "loss": 0.8848,
      "step": 6463
    },
    {
      "epoch": 0.5223539869492313,
      "grad_norm": 2.6346776485443115,
      "learning_rate": 8.662799868394995e-06,
      "loss": 1.0412,
      "step": 6464
    },
    {
      "epoch": 0.5224347966625589,
      "grad_norm": 2.7388508319854736,
      "learning_rate": 8.66235441108554e-06,
      "loss": 1.0035,
      "step": 6465
    },
    {
      "epoch": 0.5225156063758863,
      "grad_norm": 3.1407368183135986,
      "learning_rate": 8.66190889104877e-06,
      "loss": 1.0042,
      "step": 6466
    },
    {
      "epoch": 0.5225964160892139,
      "grad_norm": 2.942356824874878,
      "learning_rate": 8.661463308292317e-06,
      "loss": 0.9301,
      "step": 6467
    },
    {
      "epoch": 0.5226772258025415,
      "grad_norm": 2.982595443725586,
      "learning_rate": 8.661017662823812e-06,
      "loss": 0.9718,
      "step": 6468
    },
    {
      "epoch": 0.522758035515869,
      "grad_norm": 2.887547731399536,
      "learning_rate": 8.660571954650887e-06,
      "loss": 1.0011,
      "step": 6469
    },
    {
      "epoch": 0.5228388452291965,
      "grad_norm": 2.8909153938293457,
      "learning_rate": 8.660126183781179e-06,
      "loss": 0.9297,
      "step": 6470
    },
    {
      "epoch": 0.5229196549425241,
      "grad_norm": 2.794318914413452,
      "learning_rate": 8.65968035022232e-06,
      "loss": 1.0122,
      "step": 6471
    },
    {
      "epoch": 0.5230004646558516,
      "grad_norm": 2.5340795516967773,
      "learning_rate": 8.659234453981946e-06,
      "loss": 0.9236,
      "step": 6472
    },
    {
      "epoch": 0.5230812743691792,
      "grad_norm": 2.4756805896759033,
      "learning_rate": 8.658788495067696e-06,
      "loss": 0.9408,
      "step": 6473
    },
    {
      "epoch": 0.5231620840825068,
      "grad_norm": 2.262382984161377,
      "learning_rate": 8.658342473487207e-06,
      "loss": 1.0647,
      "step": 6474
    },
    {
      "epoch": 0.5232428937958342,
      "grad_norm": 2.6728365421295166,
      "learning_rate": 8.657896389248117e-06,
      "loss": 1.0638,
      "step": 6475
    },
    {
      "epoch": 0.5233237035091618,
      "grad_norm": 2.5527806282043457,
      "learning_rate": 8.657450242358069e-06,
      "loss": 1.0011,
      "step": 6476
    },
    {
      "epoch": 0.5234045132224894,
      "grad_norm": 2.3403241634368896,
      "learning_rate": 8.657004032824705e-06,
      "loss": 0.9129,
      "step": 6477
    },
    {
      "epoch": 0.5234853229358168,
      "grad_norm": 2.7129452228546143,
      "learning_rate": 8.656557760655663e-06,
      "loss": 0.976,
      "step": 6478
    },
    {
      "epoch": 0.5235661326491444,
      "grad_norm": 2.948230028152466,
      "learning_rate": 8.656111425858591e-06,
      "loss": 0.9973,
      "step": 6479
    },
    {
      "epoch": 0.523646942362472,
      "grad_norm": 2.6018083095550537,
      "learning_rate": 8.655665028441132e-06,
      "loss": 0.8752,
      "step": 6480
    },
    {
      "epoch": 0.5237277520757995,
      "grad_norm": 3.023979425430298,
      "learning_rate": 8.655218568410931e-06,
      "loss": 0.9006,
      "step": 6481
    },
    {
      "epoch": 0.523808561789127,
      "grad_norm": 2.693397045135498,
      "learning_rate": 8.654772045775636e-06,
      "loss": 0.9366,
      "step": 6482
    },
    {
      "epoch": 0.5238893715024546,
      "grad_norm": 2.627021074295044,
      "learning_rate": 8.654325460542894e-06,
      "loss": 1.012,
      "step": 6483
    },
    {
      "epoch": 0.5239701812157821,
      "grad_norm": 2.4792442321777344,
      "learning_rate": 8.653878812720356e-06,
      "loss": 0.8224,
      "step": 6484
    },
    {
      "epoch": 0.5240509909291097,
      "grad_norm": 2.6263723373413086,
      "learning_rate": 8.65343210231567e-06,
      "loss": 1.0071,
      "step": 6485
    },
    {
      "epoch": 0.5241318006424373,
      "grad_norm": 2.6514103412628174,
      "learning_rate": 8.652985329336485e-06,
      "loss": 0.8563,
      "step": 6486
    },
    {
      "epoch": 0.5242126103557647,
      "grad_norm": 2.460580348968506,
      "learning_rate": 8.652538493790457e-06,
      "loss": 0.9829,
      "step": 6487
    },
    {
      "epoch": 0.5242934200690923,
      "grad_norm": 2.533325433731079,
      "learning_rate": 8.652091595685238e-06,
      "loss": 0.9769,
      "step": 6488
    },
    {
      "epoch": 0.5243742297824199,
      "grad_norm": 2.8130195140838623,
      "learning_rate": 8.65164463502848e-06,
      "loss": 0.9436,
      "step": 6489
    },
    {
      "epoch": 0.5244550394957473,
      "grad_norm": 2.5389840602874756,
      "learning_rate": 8.651197611827842e-06,
      "loss": 0.9137,
      "step": 6490
    },
    {
      "epoch": 0.5245358492090749,
      "grad_norm": 2.9623239040374756,
      "learning_rate": 8.650750526090978e-06,
      "loss": 0.9418,
      "step": 6491
    },
    {
      "epoch": 0.5246166589224025,
      "grad_norm": 2.6795060634613037,
      "learning_rate": 8.650303377825549e-06,
      "loss": 1.053,
      "step": 6492
    },
    {
      "epoch": 0.52469746863573,
      "grad_norm": 2.6709554195404053,
      "learning_rate": 8.649856167039208e-06,
      "loss": 1.0574,
      "step": 6493
    },
    {
      "epoch": 0.5247782783490575,
      "grad_norm": 2.550464391708374,
      "learning_rate": 8.649408893739619e-06,
      "loss": 1.0703,
      "step": 6494
    },
    {
      "epoch": 0.5248590880623851,
      "grad_norm": 2.6652603149414062,
      "learning_rate": 8.648961557934439e-06,
      "loss": 0.8755,
      "step": 6495
    },
    {
      "epoch": 0.5249398977757126,
      "grad_norm": 3.513270378112793,
      "learning_rate": 8.648514159631333e-06,
      "loss": 0.9393,
      "step": 6496
    },
    {
      "epoch": 0.5250207074890402,
      "grad_norm": 3.112147808074951,
      "learning_rate": 8.648066698837965e-06,
      "loss": 1.0004,
      "step": 6497
    },
    {
      "epoch": 0.5251015172023678,
      "grad_norm": 2.3255598545074463,
      "learning_rate": 8.647619175561995e-06,
      "loss": 0.9419,
      "step": 6498
    },
    {
      "epoch": 0.5251823269156952,
      "grad_norm": 2.878896474838257,
      "learning_rate": 8.64717158981109e-06,
      "loss": 0.8765,
      "step": 6499
    },
    {
      "epoch": 0.5252631366290228,
      "grad_norm": 3.013002872467041,
      "learning_rate": 8.646723941592916e-06,
      "loss": 1.0088,
      "step": 6500
    },
    {
      "epoch": 0.5253439463423504,
      "grad_norm": 2.42220139503479,
      "learning_rate": 8.64627623091514e-06,
      "loss": 0.9786,
      "step": 6501
    },
    {
      "epoch": 0.5254247560556778,
      "grad_norm": 2.6195785999298096,
      "learning_rate": 8.64582845778543e-06,
      "loss": 0.8912,
      "step": 6502
    },
    {
      "epoch": 0.5255055657690054,
      "grad_norm": 2.3431684970855713,
      "learning_rate": 8.645380622211457e-06,
      "loss": 0.9507,
      "step": 6503
    },
    {
      "epoch": 0.525586375482333,
      "grad_norm": 2.744147777557373,
      "learning_rate": 8.644932724200888e-06,
      "loss": 0.93,
      "step": 6504
    },
    {
      "epoch": 0.5256671851956605,
      "grad_norm": 2.6927852630615234,
      "learning_rate": 8.644484763761397e-06,
      "loss": 0.9626,
      "step": 6505
    },
    {
      "epoch": 0.525747994908988,
      "grad_norm": 3.2699437141418457,
      "learning_rate": 8.644036740900657e-06,
      "loss": 1.0035,
      "step": 6506
    },
    {
      "epoch": 0.5258288046223156,
      "grad_norm": 3.2002522945404053,
      "learning_rate": 8.643588655626337e-06,
      "loss": 0.9647,
      "step": 6507
    },
    {
      "epoch": 0.5259096143356431,
      "grad_norm": 3.257383108139038,
      "learning_rate": 8.643140507946117e-06,
      "loss": 0.9159,
      "step": 6508
    },
    {
      "epoch": 0.5259904240489707,
      "grad_norm": 2.667975664138794,
      "learning_rate": 8.642692297867672e-06,
      "loss": 0.886,
      "step": 6509
    },
    {
      "epoch": 0.5260712337622983,
      "grad_norm": 2.6128931045532227,
      "learning_rate": 8.642244025398675e-06,
      "loss": 0.906,
      "step": 6510
    },
    {
      "epoch": 0.5261520434756257,
      "grad_norm": 3.0815443992614746,
      "learning_rate": 8.641795690546806e-06,
      "loss": 0.8116,
      "step": 6511
    },
    {
      "epoch": 0.5262328531889533,
      "grad_norm": 2.6134774684906006,
      "learning_rate": 8.641347293319746e-06,
      "loss": 0.9856,
      "step": 6512
    },
    {
      "epoch": 0.5263136629022809,
      "grad_norm": 2.595999240875244,
      "learning_rate": 8.640898833725172e-06,
      "loss": 0.9108,
      "step": 6513
    },
    {
      "epoch": 0.5263944726156083,
      "grad_norm": 2.8844268321990967,
      "learning_rate": 8.640450311770766e-06,
      "loss": 0.9831,
      "step": 6514
    },
    {
      "epoch": 0.5264752823289359,
      "grad_norm": 2.3618969917297363,
      "learning_rate": 8.64000172746421e-06,
      "loss": 0.9003,
      "step": 6515
    },
    {
      "epoch": 0.5265560920422635,
      "grad_norm": 2.6652464866638184,
      "learning_rate": 8.639553080813188e-06,
      "loss": 0.8747,
      "step": 6516
    },
    {
      "epoch": 0.526636901755591,
      "grad_norm": 2.7724783420562744,
      "learning_rate": 8.639104371825383e-06,
      "loss": 0.9955,
      "step": 6517
    },
    {
      "epoch": 0.5267177114689185,
      "grad_norm": 2.8445286750793457,
      "learning_rate": 8.638655600508481e-06,
      "loss": 1.0135,
      "step": 6518
    },
    {
      "epoch": 0.5267985211822461,
      "grad_norm": 2.5326647758483887,
      "learning_rate": 8.63820676687017e-06,
      "loss": 0.8841,
      "step": 6519
    },
    {
      "epoch": 0.5268793308955736,
      "grad_norm": 2.761439085006714,
      "learning_rate": 8.637757870918132e-06,
      "loss": 0.8969,
      "step": 6520
    },
    {
      "epoch": 0.5269601406089012,
      "grad_norm": 2.558090925216675,
      "learning_rate": 8.637308912660064e-06,
      "loss": 0.9825,
      "step": 6521
    },
    {
      "epoch": 0.5270409503222288,
      "grad_norm": 2.850323438644409,
      "learning_rate": 8.636859892103648e-06,
      "loss": 1.1012,
      "step": 6522
    },
    {
      "epoch": 0.5271217600355562,
      "grad_norm": 2.6106185913085938,
      "learning_rate": 8.636410809256577e-06,
      "loss": 0.8804,
      "step": 6523
    },
    {
      "epoch": 0.5272025697488838,
      "grad_norm": 2.455256938934326,
      "learning_rate": 8.635961664126543e-06,
      "loss": 1.0573,
      "step": 6524
    },
    {
      "epoch": 0.5272833794622114,
      "grad_norm": 2.6972479820251465,
      "learning_rate": 8.63551245672124e-06,
      "loss": 0.981,
      "step": 6525
    },
    {
      "epoch": 0.5273641891755388,
      "grad_norm": 2.7601194381713867,
      "learning_rate": 8.63506318704836e-06,
      "loss": 0.974,
      "step": 6526
    },
    {
      "epoch": 0.5274449988888664,
      "grad_norm": 2.5542311668395996,
      "learning_rate": 8.634613855115599e-06,
      "loss": 0.8569,
      "step": 6527
    },
    {
      "epoch": 0.527525808602194,
      "grad_norm": 2.483264446258545,
      "learning_rate": 8.634164460930653e-06,
      "loss": 0.8624,
      "step": 6528
    },
    {
      "epoch": 0.5276066183155216,
      "grad_norm": 2.668414831161499,
      "learning_rate": 8.633715004501219e-06,
      "loss": 0.9694,
      "step": 6529
    },
    {
      "epoch": 0.527687428028849,
      "grad_norm": 2.5893137454986572,
      "learning_rate": 8.633265485834993e-06,
      "loss": 0.9313,
      "step": 6530
    },
    {
      "epoch": 0.5277682377421766,
      "grad_norm": 3.162181854248047,
      "learning_rate": 8.632815904939678e-06,
      "loss": 0.9288,
      "step": 6531
    },
    {
      "epoch": 0.5278490474555042,
      "grad_norm": 2.2989354133605957,
      "learning_rate": 8.63236626182297e-06,
      "loss": 0.8628,
      "step": 6532
    },
    {
      "epoch": 0.5279298571688317,
      "grad_norm": 2.7202649116516113,
      "learning_rate": 8.631916556492576e-06,
      "loss": 1.0436,
      "step": 6533
    },
    {
      "epoch": 0.5280106668821593,
      "grad_norm": 2.672980785369873,
      "learning_rate": 8.631466788956191e-06,
      "loss": 0.9771,
      "step": 6534
    },
    {
      "epoch": 0.5280914765954868,
      "grad_norm": 2.866274833679199,
      "learning_rate": 8.631016959221526e-06,
      "loss": 1.0554,
      "step": 6535
    },
    {
      "epoch": 0.5281722863088143,
      "grad_norm": 2.7136855125427246,
      "learning_rate": 8.630567067296278e-06,
      "loss": 0.82,
      "step": 6536
    },
    {
      "epoch": 0.5282530960221419,
      "grad_norm": 3.0631158351898193,
      "learning_rate": 8.63011711318816e-06,
      "loss": 0.9494,
      "step": 6537
    },
    {
      "epoch": 0.5283339057354695,
      "grad_norm": 2.781109094619751,
      "learning_rate": 8.629667096904872e-06,
      "loss": 0.9363,
      "step": 6538
    },
    {
      "epoch": 0.5284147154487969,
      "grad_norm": 2.4523978233337402,
      "learning_rate": 8.629217018454126e-06,
      "loss": 0.9444,
      "step": 6539
    },
    {
      "epoch": 0.5284955251621245,
      "grad_norm": 2.6394524574279785,
      "learning_rate": 8.62876687784363e-06,
      "loss": 0.9864,
      "step": 6540
    },
    {
      "epoch": 0.5285763348754521,
      "grad_norm": 2.700120449066162,
      "learning_rate": 8.628316675081092e-06,
      "loss": 1.0109,
      "step": 6541
    },
    {
      "epoch": 0.5286571445887795,
      "grad_norm": 2.723179578781128,
      "learning_rate": 8.627866410174225e-06,
      "loss": 0.9629,
      "step": 6542
    },
    {
      "epoch": 0.5287379543021071,
      "grad_norm": 2.4862112998962402,
      "learning_rate": 8.62741608313074e-06,
      "loss": 0.9166,
      "step": 6543
    },
    {
      "epoch": 0.5288187640154347,
      "grad_norm": 2.704634666442871,
      "learning_rate": 8.62696569395835e-06,
      "loss": 0.8236,
      "step": 6544
    },
    {
      "epoch": 0.5288995737287622,
      "grad_norm": 2.9109954833984375,
      "learning_rate": 8.626515242664769e-06,
      "loss": 0.8608,
      "step": 6545
    },
    {
      "epoch": 0.5289803834420898,
      "grad_norm": 2.65484356880188,
      "learning_rate": 8.626064729257712e-06,
      "loss": 0.9483,
      "step": 6546
    },
    {
      "epoch": 0.5290611931554173,
      "grad_norm": 2.6971874237060547,
      "learning_rate": 8.625614153744897e-06,
      "loss": 0.9079,
      "step": 6547
    },
    {
      "epoch": 0.5291420028687448,
      "grad_norm": 2.643887758255005,
      "learning_rate": 8.62516351613404e-06,
      "loss": 1.1361,
      "step": 6548
    },
    {
      "epoch": 0.5292228125820724,
      "grad_norm": 2.857724905014038,
      "learning_rate": 8.624712816432857e-06,
      "loss": 1.0655,
      "step": 6549
    },
    {
      "epoch": 0.5293036222954,
      "grad_norm": 2.50374436378479,
      "learning_rate": 8.624262054649072e-06,
      "loss": 1.0183,
      "step": 6550
    },
    {
      "epoch": 0.5293844320087274,
      "grad_norm": 2.981233596801758,
      "learning_rate": 8.623811230790402e-06,
      "loss": 0.9639,
      "step": 6551
    },
    {
      "epoch": 0.529465241722055,
      "grad_norm": 2.4913814067840576,
      "learning_rate": 8.623360344864569e-06,
      "loss": 1.0296,
      "step": 6552
    },
    {
      "epoch": 0.5295460514353826,
      "grad_norm": 2.6574978828430176,
      "learning_rate": 8.622909396879298e-06,
      "loss": 0.9511,
      "step": 6553
    },
    {
      "epoch": 0.52962686114871,
      "grad_norm": 2.43039608001709,
      "learning_rate": 8.622458386842308e-06,
      "loss": 0.9564,
      "step": 6554
    },
    {
      "epoch": 0.5297076708620376,
      "grad_norm": 2.9565107822418213,
      "learning_rate": 8.62200731476133e-06,
      "loss": 0.978,
      "step": 6555
    },
    {
      "epoch": 0.5297884805753652,
      "grad_norm": 2.835583209991455,
      "learning_rate": 8.621556180644083e-06,
      "loss": 0.8722,
      "step": 6556
    },
    {
      "epoch": 0.5298692902886927,
      "grad_norm": 2.6030168533325195,
      "learning_rate": 8.621104984498299e-06,
      "loss": 0.9509,
      "step": 6557
    },
    {
      "epoch": 0.5299501000020203,
      "grad_norm": 2.6257119178771973,
      "learning_rate": 8.620653726331703e-06,
      "loss": 0.9394,
      "step": 6558
    },
    {
      "epoch": 0.5300309097153478,
      "grad_norm": 2.4450864791870117,
      "learning_rate": 8.620202406152027e-06,
      "loss": 0.9802,
      "step": 6559
    },
    {
      "epoch": 0.5301117194286753,
      "grad_norm": 2.6981029510498047,
      "learning_rate": 8.619751023966996e-06,
      "loss": 0.9408,
      "step": 6560
    },
    {
      "epoch": 0.5301925291420029,
      "grad_norm": 2.702737331390381,
      "learning_rate": 8.619299579784347e-06,
      "loss": 0.8579,
      "step": 6561
    },
    {
      "epoch": 0.5302733388553305,
      "grad_norm": 2.8359806537628174,
      "learning_rate": 8.618848073611807e-06,
      "loss": 1.0747,
      "step": 6562
    },
    {
      "epoch": 0.5303541485686579,
      "grad_norm": 2.9311721324920654,
      "learning_rate": 8.618396505457114e-06,
      "loss": 0.9953,
      "step": 6563
    },
    {
      "epoch": 0.5304349582819855,
      "grad_norm": 2.3743112087249756,
      "learning_rate": 8.617944875327998e-06,
      "loss": 0.9683,
      "step": 6564
    },
    {
      "epoch": 0.5305157679953131,
      "grad_norm": 2.845492124557495,
      "learning_rate": 8.617493183232198e-06,
      "loss": 0.9726,
      "step": 6565
    },
    {
      "epoch": 0.5305965777086405,
      "grad_norm": 2.371586799621582,
      "learning_rate": 8.617041429177447e-06,
      "loss": 0.9463,
      "step": 6566
    },
    {
      "epoch": 0.5306773874219681,
      "grad_norm": 2.7971203327178955,
      "learning_rate": 8.616589613171482e-06,
      "loss": 0.9046,
      "step": 6567
    },
    {
      "epoch": 0.5307581971352957,
      "grad_norm": 2.6262612342834473,
      "learning_rate": 8.616137735222047e-06,
      "loss": 1.0292,
      "step": 6568
    },
    {
      "epoch": 0.5308390068486232,
      "grad_norm": 2.373046875,
      "learning_rate": 8.615685795336877e-06,
      "loss": 0.9418,
      "step": 6569
    },
    {
      "epoch": 0.5309198165619508,
      "grad_norm": 2.4569997787475586,
      "learning_rate": 8.615233793523713e-06,
      "loss": 0.9875,
      "step": 6570
    },
    {
      "epoch": 0.5310006262752783,
      "grad_norm": 2.4921178817749023,
      "learning_rate": 8.614781729790298e-06,
      "loss": 0.8933,
      "step": 6571
    },
    {
      "epoch": 0.5310814359886058,
      "grad_norm": 2.863889694213867,
      "learning_rate": 8.614329604144373e-06,
      "loss": 0.9791,
      "step": 6572
    },
    {
      "epoch": 0.5311622457019334,
      "grad_norm": 2.38996958732605,
      "learning_rate": 8.613877416593686e-06,
      "loss": 0.9147,
      "step": 6573
    },
    {
      "epoch": 0.531243055415261,
      "grad_norm": 2.6225709915161133,
      "learning_rate": 8.613425167145977e-06,
      "loss": 0.9874,
      "step": 6574
    },
    {
      "epoch": 0.5313238651285884,
      "grad_norm": 2.7070398330688477,
      "learning_rate": 8.612972855808993e-06,
      "loss": 1.0344,
      "step": 6575
    },
    {
      "epoch": 0.531404674841916,
      "grad_norm": 2.5063893795013428,
      "learning_rate": 8.612520482590483e-06,
      "loss": 1.0081,
      "step": 6576
    },
    {
      "epoch": 0.5314854845552436,
      "grad_norm": 2.8511345386505127,
      "learning_rate": 8.612068047498195e-06,
      "loss": 0.9409,
      "step": 6577
    },
    {
      "epoch": 0.531566294268571,
      "grad_norm": 2.7853493690490723,
      "learning_rate": 8.611615550539874e-06,
      "loss": 0.8633,
      "step": 6578
    },
    {
      "epoch": 0.5316471039818986,
      "grad_norm": 2.6691391468048096,
      "learning_rate": 8.611162991723277e-06,
      "loss": 0.9804,
      "step": 6579
    },
    {
      "epoch": 0.5317279136952262,
      "grad_norm": 2.67984676361084,
      "learning_rate": 8.610710371056148e-06,
      "loss": 0.8895,
      "step": 6580
    },
    {
      "epoch": 0.5318087234085537,
      "grad_norm": 3.4826438426971436,
      "learning_rate": 8.610257688546244e-06,
      "loss": 1.0228,
      "step": 6581
    },
    {
      "epoch": 0.5318895331218813,
      "grad_norm": 2.6742348670959473,
      "learning_rate": 8.609804944201319e-06,
      "loss": 0.951,
      "step": 6582
    },
    {
      "epoch": 0.5319703428352088,
      "grad_norm": 2.6604034900665283,
      "learning_rate": 8.609352138029123e-06,
      "loss": 0.9672,
      "step": 6583
    },
    {
      "epoch": 0.5320511525485363,
      "grad_norm": 3.0618724822998047,
      "learning_rate": 8.608899270037414e-06,
      "loss": 0.9743,
      "step": 6584
    },
    {
      "epoch": 0.5321319622618639,
      "grad_norm": 2.5103518962860107,
      "learning_rate": 8.608446340233951e-06,
      "loss": 0.9295,
      "step": 6585
    },
    {
      "epoch": 0.5322127719751915,
      "grad_norm": 2.4761383533477783,
      "learning_rate": 8.607993348626486e-06,
      "loss": 1.021,
      "step": 6586
    },
    {
      "epoch": 0.5322935816885189,
      "grad_norm": 2.9981186389923096,
      "learning_rate": 8.607540295222784e-06,
      "loss": 0.9241,
      "step": 6587
    },
    {
      "epoch": 0.5323743914018465,
      "grad_norm": 2.822262763977051,
      "learning_rate": 8.6070871800306e-06,
      "loss": 0.865,
      "step": 6588
    },
    {
      "epoch": 0.5324552011151741,
      "grad_norm": 2.8145699501037598,
      "learning_rate": 8.606634003057697e-06,
      "loss": 0.9334,
      "step": 6589
    },
    {
      "epoch": 0.5325360108285015,
      "grad_norm": 2.835066318511963,
      "learning_rate": 8.606180764311836e-06,
      "loss": 0.9147,
      "step": 6590
    },
    {
      "epoch": 0.5326168205418291,
      "grad_norm": 2.5754809379577637,
      "learning_rate": 8.60572746380078e-06,
      "loss": 0.9673,
      "step": 6591
    },
    {
      "epoch": 0.5326976302551567,
      "grad_norm": 2.8904290199279785,
      "learning_rate": 8.60527410153229e-06,
      "loss": 0.842,
      "step": 6592
    },
    {
      "epoch": 0.5327784399684842,
      "grad_norm": 2.563721179962158,
      "learning_rate": 8.604820677514139e-06,
      "loss": 0.7603,
      "step": 6593
    },
    {
      "epoch": 0.5328592496818118,
      "grad_norm": 2.49072003364563,
      "learning_rate": 8.604367191754083e-06,
      "loss": 0.9352,
      "step": 6594
    },
    {
      "epoch": 0.5329400593951393,
      "grad_norm": 2.5579614639282227,
      "learning_rate": 8.603913644259898e-06,
      "loss": 1.1044,
      "step": 6595
    },
    {
      "epoch": 0.5330208691084668,
      "grad_norm": 2.7474677562713623,
      "learning_rate": 8.603460035039348e-06,
      "loss": 1.0202,
      "step": 6596
    },
    {
      "epoch": 0.5331016788217944,
      "grad_norm": 2.919316291809082,
      "learning_rate": 8.603006364100201e-06,
      "loss": 0.9191,
      "step": 6597
    },
    {
      "epoch": 0.533182488535122,
      "grad_norm": 2.361276865005493,
      "learning_rate": 8.60255263145023e-06,
      "loss": 1.0616,
      "step": 6598
    },
    {
      "epoch": 0.5332632982484494,
      "grad_norm": 2.8892202377319336,
      "learning_rate": 8.602098837097203e-06,
      "loss": 0.9934,
      "step": 6599
    },
    {
      "epoch": 0.533344107961777,
      "grad_norm": 2.831618309020996,
      "learning_rate": 8.601644981048897e-06,
      "loss": 0.968,
      "step": 6600
    },
    {
      "epoch": 0.5334249176751046,
      "grad_norm": 2.7197771072387695,
      "learning_rate": 8.601191063313084e-06,
      "loss": 1.0363,
      "step": 6601
    },
    {
      "epoch": 0.533505727388432,
      "grad_norm": 2.8328351974487305,
      "learning_rate": 8.600737083897534e-06,
      "loss": 0.9337,
      "step": 6602
    },
    {
      "epoch": 0.5335865371017596,
      "grad_norm": 2.942678928375244,
      "learning_rate": 8.60028304281003e-06,
      "loss": 0.9096,
      "step": 6603
    },
    {
      "epoch": 0.5336673468150872,
      "grad_norm": 2.599435806274414,
      "learning_rate": 8.599828940058343e-06,
      "loss": 0.9754,
      "step": 6604
    },
    {
      "epoch": 0.5337481565284147,
      "grad_norm": 2.5882532596588135,
      "learning_rate": 8.599374775650252e-06,
      "loss": 1.0987,
      "step": 6605
    },
    {
      "epoch": 0.5338289662417423,
      "grad_norm": 3.2290847301483154,
      "learning_rate": 8.598920549593536e-06,
      "loss": 0.8644,
      "step": 6606
    },
    {
      "epoch": 0.5339097759550698,
      "grad_norm": 2.351295232772827,
      "learning_rate": 8.598466261895976e-06,
      "loss": 0.9477,
      "step": 6607
    },
    {
      "epoch": 0.5339905856683973,
      "grad_norm": 2.7535810470581055,
      "learning_rate": 8.598011912565352e-06,
      "loss": 1.0735,
      "step": 6608
    },
    {
      "epoch": 0.5340713953817249,
      "grad_norm": 2.615550994873047,
      "learning_rate": 8.597557501609447e-06,
      "loss": 0.993,
      "step": 6609
    },
    {
      "epoch": 0.5341522050950525,
      "grad_norm": 2.746276617050171,
      "learning_rate": 8.59710302903604e-06,
      "loss": 0.919,
      "step": 6610
    },
    {
      "epoch": 0.5342330148083799,
      "grad_norm": 2.752744674682617,
      "learning_rate": 8.596648494852919e-06,
      "loss": 0.9607,
      "step": 6611
    },
    {
      "epoch": 0.5343138245217075,
      "grad_norm": 2.741203546524048,
      "learning_rate": 8.596193899067868e-06,
      "loss": 0.926,
      "step": 6612
    },
    {
      "epoch": 0.5343946342350351,
      "grad_norm": 2.859609365463257,
      "learning_rate": 8.595739241688673e-06,
      "loss": 0.887,
      "step": 6613
    },
    {
      "epoch": 0.5344754439483625,
      "grad_norm": 2.8038549423217773,
      "learning_rate": 8.59528452272312e-06,
      "loss": 0.8971,
      "step": 6614
    },
    {
      "epoch": 0.5345562536616901,
      "grad_norm": 3.0580995082855225,
      "learning_rate": 8.594829742179e-06,
      "loss": 1.0158,
      "step": 6615
    },
    {
      "epoch": 0.5346370633750177,
      "grad_norm": 2.649477243423462,
      "learning_rate": 8.5943749000641e-06,
      "loss": 1.0207,
      "step": 6616
    },
    {
      "epoch": 0.5347178730883452,
      "grad_norm": 3.593618869781494,
      "learning_rate": 8.593919996386212e-06,
      "loss": 0.9654,
      "step": 6617
    },
    {
      "epoch": 0.5347986828016728,
      "grad_norm": 2.7303850650787354,
      "learning_rate": 8.593465031153126e-06,
      "loss": 0.8854,
      "step": 6618
    },
    {
      "epoch": 0.5348794925150003,
      "grad_norm": 2.9342663288116455,
      "learning_rate": 8.593010004372634e-06,
      "loss": 0.8591,
      "step": 6619
    },
    {
      "epoch": 0.5349603022283278,
      "grad_norm": 3.488084554672241,
      "learning_rate": 8.592554916052531e-06,
      "loss": 0.9059,
      "step": 6620
    },
    {
      "epoch": 0.5350411119416554,
      "grad_norm": 2.514772653579712,
      "learning_rate": 8.592099766200613e-06,
      "loss": 0.9205,
      "step": 6621
    },
    {
      "epoch": 0.535121921654983,
      "grad_norm": 3.1306204795837402,
      "learning_rate": 8.59164455482467e-06,
      "loss": 0.9401,
      "step": 6622
    },
    {
      "epoch": 0.5352027313683104,
      "grad_norm": 2.5622668266296387,
      "learning_rate": 8.591189281932504e-06,
      "loss": 0.8657,
      "step": 6623
    },
    {
      "epoch": 0.535283541081638,
      "grad_norm": 2.785163402557373,
      "learning_rate": 8.590733947531911e-06,
      "loss": 0.9722,
      "step": 6624
    },
    {
      "epoch": 0.5353643507949656,
      "grad_norm": 2.698099374771118,
      "learning_rate": 8.590278551630691e-06,
      "loss": 1.0265,
      "step": 6625
    },
    {
      "epoch": 0.535445160508293,
      "grad_norm": 2.6029014587402344,
      "learning_rate": 8.589823094236642e-06,
      "loss": 0.9085,
      "step": 6626
    },
    {
      "epoch": 0.5355259702216206,
      "grad_norm": 3.1489837169647217,
      "learning_rate": 8.589367575357564e-06,
      "loss": 0.9826,
      "step": 6627
    },
    {
      "epoch": 0.5356067799349482,
      "grad_norm": 3.264782190322876,
      "learning_rate": 8.588911995001262e-06,
      "loss": 0.9107,
      "step": 6628
    },
    {
      "epoch": 0.5356875896482757,
      "grad_norm": 2.4482884407043457,
      "learning_rate": 8.588456353175536e-06,
      "loss": 1.0412,
      "step": 6629
    },
    {
      "epoch": 0.5357683993616033,
      "grad_norm": 2.366326093673706,
      "learning_rate": 8.588000649888194e-06,
      "loss": 0.926,
      "step": 6630
    },
    {
      "epoch": 0.5358492090749308,
      "grad_norm": 3.108952760696411,
      "learning_rate": 8.587544885147037e-06,
      "loss": 0.9872,
      "step": 6631
    },
    {
      "epoch": 0.5359300187882583,
      "grad_norm": 2.9690911769866943,
      "learning_rate": 8.587089058959872e-06,
      "loss": 0.9808,
      "step": 6632
    },
    {
      "epoch": 0.5360108285015859,
      "grad_norm": 2.4989426136016846,
      "learning_rate": 8.586633171334508e-06,
      "loss": 0.9387,
      "step": 6633
    },
    {
      "epoch": 0.5360916382149135,
      "grad_norm": 2.7757833003997803,
      "learning_rate": 8.586177222278753e-06,
      "loss": 0.9552,
      "step": 6634
    },
    {
      "epoch": 0.5361724479282409,
      "grad_norm": 2.7151308059692383,
      "learning_rate": 8.585721211800415e-06,
      "loss": 0.8695,
      "step": 6635
    },
    {
      "epoch": 0.5362532576415685,
      "grad_norm": 2.7193984985351562,
      "learning_rate": 8.585265139907303e-06,
      "loss": 1.0022,
      "step": 6636
    },
    {
      "epoch": 0.5363340673548961,
      "grad_norm": 2.4473953247070312,
      "learning_rate": 8.584809006607234e-06,
      "loss": 0.9212,
      "step": 6637
    },
    {
      "epoch": 0.5364148770682236,
      "grad_norm": 2.8431286811828613,
      "learning_rate": 8.584352811908015e-06,
      "loss": 0.9203,
      "step": 6638
    },
    {
      "epoch": 0.5364956867815511,
      "grad_norm": 3.207179546356201,
      "learning_rate": 8.58389655581746e-06,
      "loss": 0.9965,
      "step": 6639
    },
    {
      "epoch": 0.5365764964948787,
      "grad_norm": 2.723459005355835,
      "learning_rate": 8.583440238343385e-06,
      "loss": 0.9286,
      "step": 6640
    },
    {
      "epoch": 0.5366573062082062,
      "grad_norm": 2.376145601272583,
      "learning_rate": 8.582983859493607e-06,
      "loss": 0.9126,
      "step": 6641
    },
    {
      "epoch": 0.5367381159215338,
      "grad_norm": 2.7797765731811523,
      "learning_rate": 8.58252741927594e-06,
      "loss": 0.9176,
      "step": 6642
    },
    {
      "epoch": 0.5368189256348613,
      "grad_norm": 2.5153796672821045,
      "learning_rate": 8.582070917698204e-06,
      "loss": 0.9259,
      "step": 6643
    },
    {
      "epoch": 0.5368997353481888,
      "grad_norm": 2.734215021133423,
      "learning_rate": 8.581614354768218e-06,
      "loss": 0.9787,
      "step": 6644
    },
    {
      "epoch": 0.5369805450615164,
      "grad_norm": 2.379599094390869,
      "learning_rate": 8.581157730493798e-06,
      "loss": 0.983,
      "step": 6645
    },
    {
      "epoch": 0.537061354774844,
      "grad_norm": 2.7251689434051514,
      "learning_rate": 8.580701044882768e-06,
      "loss": 0.8952,
      "step": 6646
    },
    {
      "epoch": 0.5371421644881714,
      "grad_norm": 2.262371063232422,
      "learning_rate": 8.58024429794295e-06,
      "loss": 1.1348,
      "step": 6647
    },
    {
      "epoch": 0.537222974201499,
      "grad_norm": 2.9627561569213867,
      "learning_rate": 8.579787489682166e-06,
      "loss": 0.9859,
      "step": 6648
    },
    {
      "epoch": 0.5373037839148266,
      "grad_norm": 2.3098561763763428,
      "learning_rate": 8.57933062010824e-06,
      "loss": 1.0214,
      "step": 6649
    },
    {
      "epoch": 0.537384593628154,
      "grad_norm": 2.6136844158172607,
      "learning_rate": 8.578873689228996e-06,
      "loss": 1.0146,
      "step": 6650
    },
    {
      "epoch": 0.5374654033414816,
      "grad_norm": 2.760162353515625,
      "learning_rate": 8.578416697052263e-06,
      "loss": 0.9337,
      "step": 6651
    },
    {
      "epoch": 0.5375462130548092,
      "grad_norm": 2.3018577098846436,
      "learning_rate": 8.577959643585867e-06,
      "loss": 1.0456,
      "step": 6652
    },
    {
      "epoch": 0.5376270227681367,
      "grad_norm": 3.419112205505371,
      "learning_rate": 8.577502528837634e-06,
      "loss": 1.0884,
      "step": 6653
    },
    {
      "epoch": 0.5377078324814643,
      "grad_norm": 2.4860055446624756,
      "learning_rate": 8.577045352815397e-06,
      "loss": 1.0344,
      "step": 6654
    },
    {
      "epoch": 0.5377886421947918,
      "grad_norm": 2.4065463542938232,
      "learning_rate": 8.576588115526985e-06,
      "loss": 1.0467,
      "step": 6655
    },
    {
      "epoch": 0.5378694519081194,
      "grad_norm": 2.653153419494629,
      "learning_rate": 8.576130816980226e-06,
      "loss": 0.991,
      "step": 6656
    },
    {
      "epoch": 0.5379502616214469,
      "grad_norm": 2.1766774654388428,
      "learning_rate": 8.575673457182958e-06,
      "loss": 0.9505,
      "step": 6657
    },
    {
      "epoch": 0.5380310713347745,
      "grad_norm": 2.454543113708496,
      "learning_rate": 8.57521603614301e-06,
      "loss": 0.9752,
      "step": 6658
    },
    {
      "epoch": 0.538111881048102,
      "grad_norm": 3.144435405731201,
      "learning_rate": 8.57475855386822e-06,
      "loss": 0.926,
      "step": 6659
    },
    {
      "epoch": 0.5381926907614295,
      "grad_norm": 2.16461181640625,
      "learning_rate": 8.57430101036642e-06,
      "loss": 1.0754,
      "step": 6660
    },
    {
      "epoch": 0.5382735004747571,
      "grad_norm": 3.0113322734832764,
      "learning_rate": 8.57384340564545e-06,
      "loss": 1.0134,
      "step": 6661
    },
    {
      "epoch": 0.5383543101880847,
      "grad_norm": 2.7249655723571777,
      "learning_rate": 8.573385739713147e-06,
      "loss": 0.9509,
      "step": 6662
    },
    {
      "epoch": 0.5384351199014121,
      "grad_norm": 2.586225986480713,
      "learning_rate": 8.572928012577347e-06,
      "loss": 1.0151,
      "step": 6663
    },
    {
      "epoch": 0.5385159296147397,
      "grad_norm": 2.8383288383483887,
      "learning_rate": 8.572470224245892e-06,
      "loss": 0.9979,
      "step": 6664
    },
    {
      "epoch": 0.5385967393280673,
      "grad_norm": 2.82600998878479,
      "learning_rate": 8.572012374726623e-06,
      "loss": 0.9417,
      "step": 6665
    },
    {
      "epoch": 0.5386775490413948,
      "grad_norm": 2.528249740600586,
      "learning_rate": 8.571554464027381e-06,
      "loss": 0.9845,
      "step": 6666
    },
    {
      "epoch": 0.5387583587547223,
      "grad_norm": 3.2860031127929688,
      "learning_rate": 8.571096492156008e-06,
      "loss": 1.0213,
      "step": 6667
    },
    {
      "epoch": 0.5388391684680499,
      "grad_norm": 2.217886447906494,
      "learning_rate": 8.570638459120351e-06,
      "loss": 1.0046,
      "step": 6668
    },
    {
      "epoch": 0.5389199781813774,
      "grad_norm": 3.0639541149139404,
      "learning_rate": 8.570180364928252e-06,
      "loss": 0.9779,
      "step": 6669
    },
    {
      "epoch": 0.539000787894705,
      "grad_norm": 2.952514171600342,
      "learning_rate": 8.56972220958756e-06,
      "loss": 1.0254,
      "step": 6670
    },
    {
      "epoch": 0.5390815976080325,
      "grad_norm": 2.5589892864227295,
      "learning_rate": 8.569263993106118e-06,
      "loss": 1.0359,
      "step": 6671
    },
    {
      "epoch": 0.53916240732136,
      "grad_norm": 2.618180751800537,
      "learning_rate": 8.568805715491777e-06,
      "loss": 0.8866,
      "step": 6672
    },
    {
      "epoch": 0.5392432170346876,
      "grad_norm": 2.5594685077667236,
      "learning_rate": 8.568347376752387e-06,
      "loss": 0.9899,
      "step": 6673
    },
    {
      "epoch": 0.5393240267480152,
      "grad_norm": 2.4848663806915283,
      "learning_rate": 8.567888976895795e-06,
      "loss": 0.9149,
      "step": 6674
    },
    {
      "epoch": 0.5394048364613426,
      "grad_norm": 2.714613199234009,
      "learning_rate": 8.567430515929856e-06,
      "loss": 0.9139,
      "step": 6675
    },
    {
      "epoch": 0.5394856461746702,
      "grad_norm": 2.9260153770446777,
      "learning_rate": 8.566971993862419e-06,
      "loss": 0.9209,
      "step": 6676
    },
    {
      "epoch": 0.5395664558879978,
      "grad_norm": 2.8303847312927246,
      "learning_rate": 8.566513410701338e-06,
      "loss": 0.8885,
      "step": 6677
    },
    {
      "epoch": 0.5396472656013253,
      "grad_norm": 3.4838356971740723,
      "learning_rate": 8.566054766454471e-06,
      "loss": 0.9771,
      "step": 6678
    },
    {
      "epoch": 0.5397280753146528,
      "grad_norm": 2.4133317470550537,
      "learning_rate": 8.565596061129669e-06,
      "loss": 0.8819,
      "step": 6679
    },
    {
      "epoch": 0.5398088850279804,
      "grad_norm": 2.5783698558807373,
      "learning_rate": 8.56513729473479e-06,
      "loss": 0.9464,
      "step": 6680
    },
    {
      "epoch": 0.5398896947413079,
      "grad_norm": 2.557776689529419,
      "learning_rate": 8.564678467277693e-06,
      "loss": 0.9017,
      "step": 6681
    },
    {
      "epoch": 0.5399705044546355,
      "grad_norm": 2.467772960662842,
      "learning_rate": 8.564219578766236e-06,
      "loss": 1.0832,
      "step": 6682
    },
    {
      "epoch": 0.540051314167963,
      "grad_norm": 3.1684677600860596,
      "learning_rate": 8.563760629208277e-06,
      "loss": 0.9311,
      "step": 6683
    },
    {
      "epoch": 0.5401321238812905,
      "grad_norm": 3.209038019180298,
      "learning_rate": 8.563301618611678e-06,
      "loss": 0.8623,
      "step": 6684
    },
    {
      "epoch": 0.5402129335946181,
      "grad_norm": 2.0523617267608643,
      "learning_rate": 8.562842546984301e-06,
      "loss": 1.0347,
      "step": 6685
    },
    {
      "epoch": 0.5402937433079457,
      "grad_norm": 2.662071943283081,
      "learning_rate": 8.562383414334006e-06,
      "loss": 0.955,
      "step": 6686
    },
    {
      "epoch": 0.5403745530212731,
      "grad_norm": 2.708681106567383,
      "learning_rate": 8.561924220668663e-06,
      "loss": 0.968,
      "step": 6687
    },
    {
      "epoch": 0.5404553627346007,
      "grad_norm": 2.951434850692749,
      "learning_rate": 8.561464965996132e-06,
      "loss": 1.0313,
      "step": 6688
    },
    {
      "epoch": 0.5405361724479283,
      "grad_norm": 2.426435708999634,
      "learning_rate": 8.561005650324277e-06,
      "loss": 0.971,
      "step": 6689
    },
    {
      "epoch": 0.5406169821612558,
      "grad_norm": 2.5620830059051514,
      "learning_rate": 8.56054627366097e-06,
      "loss": 0.9397,
      "step": 6690
    },
    {
      "epoch": 0.5406977918745833,
      "grad_norm": 3.1869468688964844,
      "learning_rate": 8.560086836014078e-06,
      "loss": 0.9638,
      "step": 6691
    },
    {
      "epoch": 0.5407786015879109,
      "grad_norm": 2.8679094314575195,
      "learning_rate": 8.559627337391469e-06,
      "loss": 0.7878,
      "step": 6692
    },
    {
      "epoch": 0.5408594113012384,
      "grad_norm": 2.715805768966675,
      "learning_rate": 8.559167777801012e-06,
      "loss": 0.8415,
      "step": 6693
    },
    {
      "epoch": 0.540940221014566,
      "grad_norm": 2.559495687484741,
      "learning_rate": 8.55870815725058e-06,
      "loss": 0.8802,
      "step": 6694
    },
    {
      "epoch": 0.5410210307278935,
      "grad_norm": 2.8612630367279053,
      "learning_rate": 8.558248475748044e-06,
      "loss": 0.8723,
      "step": 6695
    },
    {
      "epoch": 0.541101840441221,
      "grad_norm": 2.882262945175171,
      "learning_rate": 8.557788733301278e-06,
      "loss": 0.9767,
      "step": 6696
    },
    {
      "epoch": 0.5411826501545486,
      "grad_norm": 2.5127177238464355,
      "learning_rate": 8.557328929918155e-06,
      "loss": 1.0426,
      "step": 6697
    },
    {
      "epoch": 0.5412634598678762,
      "grad_norm": 2.405726432800293,
      "learning_rate": 8.556869065606554e-06,
      "loss": 0.8544,
      "step": 6698
    },
    {
      "epoch": 0.5413442695812036,
      "grad_norm": 2.8429133892059326,
      "learning_rate": 8.556409140374347e-06,
      "loss": 1.0006,
      "step": 6699
    },
    {
      "epoch": 0.5414250792945312,
      "grad_norm": 3.339569091796875,
      "learning_rate": 8.555949154229412e-06,
      "loss": 0.9891,
      "step": 6700
    },
    {
      "epoch": 0.5415058890078588,
      "grad_norm": 2.438951015472412,
      "learning_rate": 8.55548910717963e-06,
      "loss": 0.9815,
      "step": 6701
    },
    {
      "epoch": 0.5415866987211863,
      "grad_norm": 2.578977108001709,
      "learning_rate": 8.555028999232879e-06,
      "loss": 1.0372,
      "step": 6702
    },
    {
      "epoch": 0.5416675084345138,
      "grad_norm": 2.533268451690674,
      "learning_rate": 8.554568830397039e-06,
      "loss": 0.8836,
      "step": 6703
    },
    {
      "epoch": 0.5417483181478414,
      "grad_norm": 2.6665029525756836,
      "learning_rate": 8.554108600679993e-06,
      "loss": 1.1245,
      "step": 6704
    },
    {
      "epoch": 0.5418291278611689,
      "grad_norm": 2.560865879058838,
      "learning_rate": 8.553648310089624e-06,
      "loss": 0.9025,
      "step": 6705
    },
    {
      "epoch": 0.5419099375744965,
      "grad_norm": 2.8905866146087646,
      "learning_rate": 8.553187958633813e-06,
      "loss": 0.9169,
      "step": 6706
    },
    {
      "epoch": 0.541990747287824,
      "grad_norm": 2.840790271759033,
      "learning_rate": 8.552727546320444e-06,
      "loss": 0.9517,
      "step": 6707
    },
    {
      "epoch": 0.5420715570011515,
      "grad_norm": 2.5757713317871094,
      "learning_rate": 8.552267073157408e-06,
      "loss": 1.091,
      "step": 6708
    },
    {
      "epoch": 0.5421523667144791,
      "grad_norm": 2.7778074741363525,
      "learning_rate": 8.55180653915259e-06,
      "loss": 0.7855,
      "step": 6709
    },
    {
      "epoch": 0.5422331764278067,
      "grad_norm": 2.9378228187561035,
      "learning_rate": 8.551345944313874e-06,
      "loss": 0.9325,
      "step": 6710
    },
    {
      "epoch": 0.5423139861411341,
      "grad_norm": 2.6456212997436523,
      "learning_rate": 8.55088528864915e-06,
      "loss": 1.0049,
      "step": 6711
    },
    {
      "epoch": 0.5423947958544617,
      "grad_norm": 2.3709723949432373,
      "learning_rate": 8.550424572166313e-06,
      "loss": 0.8832,
      "step": 6712
    },
    {
      "epoch": 0.5424756055677893,
      "grad_norm": 2.709120750427246,
      "learning_rate": 8.54996379487325e-06,
      "loss": 0.9989,
      "step": 6713
    },
    {
      "epoch": 0.5425564152811168,
      "grad_norm": 2.762544870376587,
      "learning_rate": 8.549502956777852e-06,
      "loss": 0.8828,
      "step": 6714
    },
    {
      "epoch": 0.5426372249944443,
      "grad_norm": 2.7496256828308105,
      "learning_rate": 8.549042057888013e-06,
      "loss": 1.0145,
      "step": 6715
    },
    {
      "epoch": 0.5427180347077719,
      "grad_norm": 2.6640915870666504,
      "learning_rate": 8.548581098211628e-06,
      "loss": 0.8594,
      "step": 6716
    },
    {
      "epoch": 0.5427988444210994,
      "grad_norm": 2.965928554534912,
      "learning_rate": 8.548120077756593e-06,
      "loss": 0.9964,
      "step": 6717
    },
    {
      "epoch": 0.542879654134427,
      "grad_norm": 2.6074132919311523,
      "learning_rate": 8.547658996530801e-06,
      "loss": 0.9174,
      "step": 6718
    },
    {
      "epoch": 0.5429604638477545,
      "grad_norm": 2.2748515605926514,
      "learning_rate": 8.547197854542154e-06,
      "loss": 1.0433,
      "step": 6719
    },
    {
      "epoch": 0.543041273561082,
      "grad_norm": 3.151670455932617,
      "learning_rate": 8.546736651798544e-06,
      "loss": 0.9248,
      "step": 6720
    },
    {
      "epoch": 0.5431220832744096,
      "grad_norm": 2.9950156211853027,
      "learning_rate": 8.546275388307874e-06,
      "loss": 0.8626,
      "step": 6721
    },
    {
      "epoch": 0.5432028929877372,
      "grad_norm": 2.853771686553955,
      "learning_rate": 8.545814064078045e-06,
      "loss": 0.9385,
      "step": 6722
    },
    {
      "epoch": 0.5432837027010646,
      "grad_norm": 2.752655506134033,
      "learning_rate": 8.545352679116956e-06,
      "loss": 0.9427,
      "step": 6723
    },
    {
      "epoch": 0.5433645124143922,
      "grad_norm": 2.7993664741516113,
      "learning_rate": 8.544891233432513e-06,
      "loss": 1.0047,
      "step": 6724
    },
    {
      "epoch": 0.5434453221277198,
      "grad_norm": 2.392876386642456,
      "learning_rate": 8.544429727032616e-06,
      "loss": 0.8839,
      "step": 6725
    },
    {
      "epoch": 0.5435261318410473,
      "grad_norm": 2.535710334777832,
      "learning_rate": 8.54396815992517e-06,
      "loss": 0.9808,
      "step": 6726
    },
    {
      "epoch": 0.5436069415543748,
      "grad_norm": 2.690338611602783,
      "learning_rate": 8.543506532118081e-06,
      "loss": 0.9425,
      "step": 6727
    },
    {
      "epoch": 0.5436877512677024,
      "grad_norm": 2.519347667694092,
      "learning_rate": 8.543044843619255e-06,
      "loss": 1.0035,
      "step": 6728
    },
    {
      "epoch": 0.5437685609810299,
      "grad_norm": 3.0266170501708984,
      "learning_rate": 8.542583094436603e-06,
      "loss": 1.0533,
      "step": 6729
    },
    {
      "epoch": 0.5438493706943575,
      "grad_norm": 2.4525606632232666,
      "learning_rate": 8.542121284578028e-06,
      "loss": 0.8922,
      "step": 6730
    },
    {
      "epoch": 0.543930180407685,
      "grad_norm": 2.5058019161224365,
      "learning_rate": 8.541659414051446e-06,
      "loss": 0.9306,
      "step": 6731
    },
    {
      "epoch": 0.5440109901210125,
      "grad_norm": 2.8877944946289062,
      "learning_rate": 8.541197482864763e-06,
      "loss": 0.8953,
      "step": 6732
    },
    {
      "epoch": 0.5440917998343401,
      "grad_norm": 3.1442575454711914,
      "learning_rate": 8.54073549102589e-06,
      "loss": 0.9758,
      "step": 6733
    },
    {
      "epoch": 0.5441726095476677,
      "grad_norm": 2.724261522293091,
      "learning_rate": 8.540273438542746e-06,
      "loss": 1.1241,
      "step": 6734
    },
    {
      "epoch": 0.5442534192609951,
      "grad_norm": 2.991394519805908,
      "learning_rate": 8.539811325423237e-06,
      "loss": 1.0696,
      "step": 6735
    },
    {
      "epoch": 0.5443342289743227,
      "grad_norm": 2.69022274017334,
      "learning_rate": 8.539349151675285e-06,
      "loss": 1.0104,
      "step": 6736
    },
    {
      "epoch": 0.5444150386876503,
      "grad_norm": 2.8307008743286133,
      "learning_rate": 8.5388869173068e-06,
      "loss": 1.0641,
      "step": 6737
    },
    {
      "epoch": 0.5444958484009778,
      "grad_norm": 2.664032459259033,
      "learning_rate": 8.538424622325705e-06,
      "loss": 0.9752,
      "step": 6738
    },
    {
      "epoch": 0.5445766581143053,
      "grad_norm": 2.362687110900879,
      "learning_rate": 8.537962266739912e-06,
      "loss": 0.9436,
      "step": 6739
    },
    {
      "epoch": 0.5446574678276329,
      "grad_norm": 2.7971317768096924,
      "learning_rate": 8.537499850557343e-06,
      "loss": 0.9524,
      "step": 6740
    },
    {
      "epoch": 0.5447382775409604,
      "grad_norm": 2.574246644973755,
      "learning_rate": 8.537037373785917e-06,
      "loss": 0.9919,
      "step": 6741
    },
    {
      "epoch": 0.544819087254288,
      "grad_norm": 2.4597177505493164,
      "learning_rate": 8.536574836433558e-06,
      "loss": 1.008,
      "step": 6742
    },
    {
      "epoch": 0.5448998969676155,
      "grad_norm": 2.4348719120025635,
      "learning_rate": 8.536112238508183e-06,
      "loss": 1.0923,
      "step": 6743
    },
    {
      "epoch": 0.544980706680943,
      "grad_norm": 2.532266855239868,
      "learning_rate": 8.53564958001772e-06,
      "loss": 0.9468,
      "step": 6744
    },
    {
      "epoch": 0.5450615163942706,
      "grad_norm": 2.626479387283325,
      "learning_rate": 8.535186860970093e-06,
      "loss": 0.9075,
      "step": 6745
    },
    {
      "epoch": 0.5451423261075982,
      "grad_norm": 2.474057674407959,
      "learning_rate": 8.534724081373224e-06,
      "loss": 0.9538,
      "step": 6746
    },
    {
      "epoch": 0.5452231358209256,
      "grad_norm": 2.630878210067749,
      "learning_rate": 8.53426124123504e-06,
      "loss": 0.8873,
      "step": 6747
    },
    {
      "epoch": 0.5453039455342532,
      "grad_norm": 2.385979413986206,
      "learning_rate": 8.53379834056347e-06,
      "loss": 0.9396,
      "step": 6748
    },
    {
      "epoch": 0.5453847552475808,
      "grad_norm": 2.3376524448394775,
      "learning_rate": 8.533335379366442e-06,
      "loss": 0.9887,
      "step": 6749
    },
    {
      "epoch": 0.5454655649609083,
      "grad_norm": 2.728322744369507,
      "learning_rate": 8.532872357651887e-06,
      "loss": 1.0434,
      "step": 6750
    },
    {
      "epoch": 0.5455463746742358,
      "grad_norm": 2.6892123222351074,
      "learning_rate": 8.532409275427731e-06,
      "loss": 0.8898,
      "step": 6751
    },
    {
      "epoch": 0.5456271843875634,
      "grad_norm": 2.754441022872925,
      "learning_rate": 8.531946132701909e-06,
      "loss": 0.886,
      "step": 6752
    },
    {
      "epoch": 0.5457079941008909,
      "grad_norm": 2.2685694694519043,
      "learning_rate": 8.531482929482352e-06,
      "loss": 1.0556,
      "step": 6753
    },
    {
      "epoch": 0.5457888038142185,
      "grad_norm": 2.628941059112549,
      "learning_rate": 8.531019665776994e-06,
      "loss": 0.9329,
      "step": 6754
    },
    {
      "epoch": 0.545869613527546,
      "grad_norm": 2.7105507850646973,
      "learning_rate": 8.53055634159377e-06,
      "loss": 0.9958,
      "step": 6755
    },
    {
      "epoch": 0.5459504232408735,
      "grad_norm": 3.347438097000122,
      "learning_rate": 8.530092956940617e-06,
      "loss": 0.9348,
      "step": 6756
    },
    {
      "epoch": 0.5460312329542011,
      "grad_norm": 2.9810469150543213,
      "learning_rate": 8.529629511825467e-06,
      "loss": 0.9408,
      "step": 6757
    },
    {
      "epoch": 0.5461120426675287,
      "grad_norm": 2.905885696411133,
      "learning_rate": 8.529166006256263e-06,
      "loss": 0.9919,
      "step": 6758
    },
    {
      "epoch": 0.5461928523808561,
      "grad_norm": 3.1253116130828857,
      "learning_rate": 8.52870244024094e-06,
      "loss": 0.9033,
      "step": 6759
    },
    {
      "epoch": 0.5462736620941837,
      "grad_norm": 2.615163803100586,
      "learning_rate": 8.52823881378744e-06,
      "loss": 0.8916,
      "step": 6760
    },
    {
      "epoch": 0.5463544718075113,
      "grad_norm": 2.942991018295288,
      "learning_rate": 8.527775126903704e-06,
      "loss": 0.949,
      "step": 6761
    },
    {
      "epoch": 0.5464352815208388,
      "grad_norm": 2.9247636795043945,
      "learning_rate": 8.527311379597672e-06,
      "loss": 0.9719,
      "step": 6762
    },
    {
      "epoch": 0.5465160912341663,
      "grad_norm": 2.6420843601226807,
      "learning_rate": 8.526847571877286e-06,
      "loss": 0.9688,
      "step": 6763
    },
    {
      "epoch": 0.5465969009474939,
      "grad_norm": 2.7086527347564697,
      "learning_rate": 8.526383703750494e-06,
      "loss": 0.998,
      "step": 6764
    },
    {
      "epoch": 0.5466777106608214,
      "grad_norm": 3.503931760787964,
      "learning_rate": 8.525919775225237e-06,
      "loss": 1.0477,
      "step": 6765
    },
    {
      "epoch": 0.546758520374149,
      "grad_norm": 2.868701457977295,
      "learning_rate": 8.525455786309464e-06,
      "loss": 0.9422,
      "step": 6766
    },
    {
      "epoch": 0.5468393300874765,
      "grad_norm": 2.823075532913208,
      "learning_rate": 8.524991737011119e-06,
      "loss": 0.8767,
      "step": 6767
    },
    {
      "epoch": 0.546920139800804,
      "grad_norm": 2.9327478408813477,
      "learning_rate": 8.524527627338152e-06,
      "loss": 0.9128,
      "step": 6768
    },
    {
      "epoch": 0.5470009495141316,
      "grad_norm": 2.5013482570648193,
      "learning_rate": 8.524063457298513e-06,
      "loss": 0.9008,
      "step": 6769
    },
    {
      "epoch": 0.5470817592274592,
      "grad_norm": 2.986111879348755,
      "learning_rate": 8.52359922690015e-06,
      "loss": 1.0234,
      "step": 6770
    },
    {
      "epoch": 0.5471625689407866,
      "grad_norm": 2.450460195541382,
      "learning_rate": 8.523134936151014e-06,
      "loss": 0.9352,
      "step": 6771
    },
    {
      "epoch": 0.5472433786541142,
      "grad_norm": 2.786874294281006,
      "learning_rate": 8.52267058505906e-06,
      "loss": 0.96,
      "step": 6772
    },
    {
      "epoch": 0.5473241883674418,
      "grad_norm": 2.892984390258789,
      "learning_rate": 8.522206173632238e-06,
      "loss": 0.9867,
      "step": 6773
    },
    {
      "epoch": 0.5474049980807693,
      "grad_norm": 2.5300848484039307,
      "learning_rate": 8.521741701878502e-06,
      "loss": 0.9617,
      "step": 6774
    },
    {
      "epoch": 0.5474858077940968,
      "grad_norm": 2.673757791519165,
      "learning_rate": 8.521277169805813e-06,
      "loss": 0.9111,
      "step": 6775
    },
    {
      "epoch": 0.5475666175074244,
      "grad_norm": 2.650216817855835,
      "learning_rate": 8.520812577422119e-06,
      "loss": 0.9323,
      "step": 6776
    },
    {
      "epoch": 0.5476474272207519,
      "grad_norm": 2.9507553577423096,
      "learning_rate": 8.520347924735387e-06,
      "loss": 0.891,
      "step": 6777
    },
    {
      "epoch": 0.5477282369340795,
      "grad_norm": 2.8094611167907715,
      "learning_rate": 8.519883211753566e-06,
      "loss": 1.0348,
      "step": 6778
    },
    {
      "epoch": 0.547809046647407,
      "grad_norm": 2.6251227855682373,
      "learning_rate": 8.51941843848462e-06,
      "loss": 0.9825,
      "step": 6779
    },
    {
      "epoch": 0.5478898563607345,
      "grad_norm": 2.7280211448669434,
      "learning_rate": 8.518953604936507e-06,
      "loss": 0.9915,
      "step": 6780
    },
    {
      "epoch": 0.5479706660740621,
      "grad_norm": 3.11441969871521,
      "learning_rate": 8.518488711117192e-06,
      "loss": 0.8633,
      "step": 6781
    },
    {
      "epoch": 0.5480514757873897,
      "grad_norm": 2.5203189849853516,
      "learning_rate": 8.518023757034636e-06,
      "loss": 1.0,
      "step": 6782
    },
    {
      "epoch": 0.5481322855007171,
      "grad_norm": 2.710209608078003,
      "learning_rate": 8.517558742696802e-06,
      "loss": 0.8671,
      "step": 6783
    },
    {
      "epoch": 0.5482130952140447,
      "grad_norm": 2.584124803543091,
      "learning_rate": 8.517093668111656e-06,
      "loss": 1.02,
      "step": 6784
    },
    {
      "epoch": 0.5482939049273723,
      "grad_norm": 2.77771258354187,
      "learning_rate": 8.516628533287162e-06,
      "loss": 1.0516,
      "step": 6785
    },
    {
      "epoch": 0.5483747146406999,
      "grad_norm": 2.7872426509857178,
      "learning_rate": 8.516163338231286e-06,
      "loss": 0.8756,
      "step": 6786
    },
    {
      "epoch": 0.5484555243540273,
      "grad_norm": 2.467712163925171,
      "learning_rate": 8.515698082951998e-06,
      "loss": 0.9397,
      "step": 6787
    },
    {
      "epoch": 0.5485363340673549,
      "grad_norm": 2.6244587898254395,
      "learning_rate": 8.515232767457265e-06,
      "loss": 1.0316,
      "step": 6788
    },
    {
      "epoch": 0.5486171437806825,
      "grad_norm": 2.871271848678589,
      "learning_rate": 8.514767391755057e-06,
      "loss": 0.994,
      "step": 6789
    },
    {
      "epoch": 0.54869795349401,
      "grad_norm": 2.5260796546936035,
      "learning_rate": 8.514301955853345e-06,
      "loss": 0.9915,
      "step": 6790
    },
    {
      "epoch": 0.5487787632073375,
      "grad_norm": 3.337914228439331,
      "learning_rate": 8.513836459760102e-06,
      "loss": 0.9211,
      "step": 6791
    },
    {
      "epoch": 0.5488595729206651,
      "grad_norm": 2.516460657119751,
      "learning_rate": 8.513370903483299e-06,
      "loss": 0.9487,
      "step": 6792
    },
    {
      "epoch": 0.5489403826339926,
      "grad_norm": 3.115722417831421,
      "learning_rate": 8.51290528703091e-06,
      "loss": 0.9131,
      "step": 6793
    },
    {
      "epoch": 0.5490211923473202,
      "grad_norm": 2.6946163177490234,
      "learning_rate": 8.51243961041091e-06,
      "loss": 0.9662,
      "step": 6794
    },
    {
      "epoch": 0.5491020020606477,
      "grad_norm": 2.703784704208374,
      "learning_rate": 8.511973873631278e-06,
      "loss": 0.9843,
      "step": 6795
    },
    {
      "epoch": 0.5491828117739752,
      "grad_norm": 2.922241449356079,
      "learning_rate": 8.511508076699985e-06,
      "loss": 0.9989,
      "step": 6796
    },
    {
      "epoch": 0.5492636214873028,
      "grad_norm": 3.141031265258789,
      "learning_rate": 8.511042219625015e-06,
      "loss": 1.0081,
      "step": 6797
    },
    {
      "epoch": 0.5493444312006304,
      "grad_norm": 2.9548773765563965,
      "learning_rate": 8.510576302414342e-06,
      "loss": 0.9744,
      "step": 6798
    },
    {
      "epoch": 0.5494252409139578,
      "grad_norm": 2.4163296222686768,
      "learning_rate": 8.510110325075948e-06,
      "loss": 1.0437,
      "step": 6799
    },
    {
      "epoch": 0.5495060506272854,
      "grad_norm": 2.843646764755249,
      "learning_rate": 8.509644287617817e-06,
      "loss": 1.1203,
      "step": 6800
    },
    {
      "epoch": 0.549586860340613,
      "grad_norm": 2.435694456100464,
      "learning_rate": 8.509178190047926e-06,
      "loss": 1.0005,
      "step": 6801
    },
    {
      "epoch": 0.5496676700539405,
      "grad_norm": 2.7450308799743652,
      "learning_rate": 8.508712032374263e-06,
      "loss": 0.9601,
      "step": 6802
    },
    {
      "epoch": 0.549748479767268,
      "grad_norm": 2.4280636310577393,
      "learning_rate": 8.50824581460481e-06,
      "loss": 0.8992,
      "step": 6803
    },
    {
      "epoch": 0.5498292894805956,
      "grad_norm": 2.5397446155548096,
      "learning_rate": 8.50777953674755e-06,
      "loss": 0.9561,
      "step": 6804
    },
    {
      "epoch": 0.5499100991939231,
      "grad_norm": 3.3508310317993164,
      "learning_rate": 8.50731319881047e-06,
      "loss": 1.0794,
      "step": 6805
    },
    {
      "epoch": 0.5499909089072507,
      "grad_norm": 2.627250909805298,
      "learning_rate": 8.50684680080156e-06,
      "loss": 0.977,
      "step": 6806
    },
    {
      "epoch": 0.5500717186205782,
      "grad_norm": 2.7339882850646973,
      "learning_rate": 8.506380342728808e-06,
      "loss": 0.8917,
      "step": 6807
    },
    {
      "epoch": 0.5501525283339057,
      "grad_norm": 2.551746368408203,
      "learning_rate": 8.505913824600201e-06,
      "loss": 1.0379,
      "step": 6808
    },
    {
      "epoch": 0.5502333380472333,
      "grad_norm": 2.5389673709869385,
      "learning_rate": 8.50544724642373e-06,
      "loss": 0.8873,
      "step": 6809
    },
    {
      "epoch": 0.5503141477605609,
      "grad_norm": 2.4108564853668213,
      "learning_rate": 8.504980608207386e-06,
      "loss": 0.9618,
      "step": 6810
    },
    {
      "epoch": 0.5503949574738883,
      "grad_norm": 2.981092691421509,
      "learning_rate": 8.504513909959164e-06,
      "loss": 0.8691,
      "step": 6811
    },
    {
      "epoch": 0.5504757671872159,
      "grad_norm": 2.546948194503784,
      "learning_rate": 8.504047151687054e-06,
      "loss": 0.9692,
      "step": 6812
    },
    {
      "epoch": 0.5505565769005435,
      "grad_norm": 2.4042325019836426,
      "learning_rate": 8.50358033339905e-06,
      "loss": 0.9395,
      "step": 6813
    },
    {
      "epoch": 0.550637386613871,
      "grad_norm": 2.7482199668884277,
      "learning_rate": 8.503113455103154e-06,
      "loss": 0.8634,
      "step": 6814
    },
    {
      "epoch": 0.5507181963271985,
      "grad_norm": 2.4952049255371094,
      "learning_rate": 8.502646516807354e-06,
      "loss": 0.9774,
      "step": 6815
    },
    {
      "epoch": 0.5507990060405261,
      "grad_norm": 2.6934478282928467,
      "learning_rate": 8.502179518519654e-06,
      "loss": 0.939,
      "step": 6816
    },
    {
      "epoch": 0.5508798157538536,
      "grad_norm": 3.1893129348754883,
      "learning_rate": 8.501712460248049e-06,
      "loss": 0.9484,
      "step": 6817
    },
    {
      "epoch": 0.5509606254671812,
      "grad_norm": 2.5171241760253906,
      "learning_rate": 8.501245342000539e-06,
      "loss": 0.9987,
      "step": 6818
    },
    {
      "epoch": 0.5510414351805087,
      "grad_norm": 2.6813125610351562,
      "learning_rate": 8.500778163785126e-06,
      "loss": 0.9635,
      "step": 6819
    },
    {
      "epoch": 0.5511222448938362,
      "grad_norm": 3.047844886779785,
      "learning_rate": 8.500310925609812e-06,
      "loss": 0.9055,
      "step": 6820
    },
    {
      "epoch": 0.5512030546071638,
      "grad_norm": 2.725375175476074,
      "learning_rate": 8.499843627482596e-06,
      "loss": 0.991,
      "step": 6821
    },
    {
      "epoch": 0.5512838643204914,
      "grad_norm": 2.3166511058807373,
      "learning_rate": 8.499376269411485e-06,
      "loss": 0.8638,
      "step": 6822
    },
    {
      "epoch": 0.5513646740338188,
      "grad_norm": 2.7540152072906494,
      "learning_rate": 8.498908851404484e-06,
      "loss": 1.0671,
      "step": 6823
    },
    {
      "epoch": 0.5514454837471464,
      "grad_norm": 2.56469464302063,
      "learning_rate": 8.498441373469597e-06,
      "loss": 0.9647,
      "step": 6824
    },
    {
      "epoch": 0.551526293460474,
      "grad_norm": 2.7377378940582275,
      "learning_rate": 8.497973835614831e-06,
      "loss": 1.0408,
      "step": 6825
    },
    {
      "epoch": 0.5516071031738015,
      "grad_norm": 2.9289684295654297,
      "learning_rate": 8.497506237848194e-06,
      "loss": 0.9382,
      "step": 6826
    },
    {
      "epoch": 0.551687912887129,
      "grad_norm": 2.7107300758361816,
      "learning_rate": 8.497038580177696e-06,
      "loss": 0.9402,
      "step": 6827
    },
    {
      "epoch": 0.5517687226004566,
      "grad_norm": 3.1685383319854736,
      "learning_rate": 8.496570862611346e-06,
      "loss": 1.0167,
      "step": 6828
    },
    {
      "epoch": 0.5518495323137841,
      "grad_norm": 2.8260293006896973,
      "learning_rate": 8.496103085157155e-06,
      "loss": 0.8844,
      "step": 6829
    },
    {
      "epoch": 0.5519303420271117,
      "grad_norm": 2.6349635124206543,
      "learning_rate": 8.495635247823132e-06,
      "loss": 1.0421,
      "step": 6830
    },
    {
      "epoch": 0.5520111517404392,
      "grad_norm": 2.400712728500366,
      "learning_rate": 8.495167350617296e-06,
      "loss": 1.0026,
      "step": 6831
    },
    {
      "epoch": 0.5520919614537667,
      "grad_norm": 2.3358864784240723,
      "learning_rate": 8.494699393547656e-06,
      "loss": 0.9533,
      "step": 6832
    },
    {
      "epoch": 0.5521727711670943,
      "grad_norm": 2.700319528579712,
      "learning_rate": 8.494231376622228e-06,
      "loss": 0.9356,
      "step": 6833
    },
    {
      "epoch": 0.5522535808804219,
      "grad_norm": 2.5020880699157715,
      "learning_rate": 8.49376329984903e-06,
      "loss": 0.8913,
      "step": 6834
    },
    {
      "epoch": 0.5523343905937493,
      "grad_norm": 2.6957719326019287,
      "learning_rate": 8.493295163236077e-06,
      "loss": 0.9682,
      "step": 6835
    },
    {
      "epoch": 0.5524152003070769,
      "grad_norm": 2.592831611633301,
      "learning_rate": 8.492826966791387e-06,
      "loss": 0.9135,
      "step": 6836
    },
    {
      "epoch": 0.5524960100204045,
      "grad_norm": 2.630263566970825,
      "learning_rate": 8.49235871052298e-06,
      "loss": 0.946,
      "step": 6837
    },
    {
      "epoch": 0.552576819733732,
      "grad_norm": 2.397728204727173,
      "learning_rate": 8.491890394438875e-06,
      "loss": 0.9447,
      "step": 6838
    },
    {
      "epoch": 0.5526576294470595,
      "grad_norm": 2.4945151805877686,
      "learning_rate": 8.491422018547096e-06,
      "loss": 1.0325,
      "step": 6839
    },
    {
      "epoch": 0.5527384391603871,
      "grad_norm": 2.8656532764434814,
      "learning_rate": 8.490953582855662e-06,
      "loss": 1.0034,
      "step": 6840
    },
    {
      "epoch": 0.5528192488737146,
      "grad_norm": 2.750896453857422,
      "learning_rate": 8.490485087372597e-06,
      "loss": 1.0123,
      "step": 6841
    },
    {
      "epoch": 0.5529000585870422,
      "grad_norm": 2.744513750076294,
      "learning_rate": 8.490016532105925e-06,
      "loss": 0.9346,
      "step": 6842
    },
    {
      "epoch": 0.5529808683003697,
      "grad_norm": 2.2509818077087402,
      "learning_rate": 8.489547917063672e-06,
      "loss": 0.9299,
      "step": 6843
    },
    {
      "epoch": 0.5530616780136972,
      "grad_norm": 3.089818239212036,
      "learning_rate": 8.489079242253863e-06,
      "loss": 1.0968,
      "step": 6844
    },
    {
      "epoch": 0.5531424877270248,
      "grad_norm": 3.1147658824920654,
      "learning_rate": 8.48861050768453e-06,
      "loss": 0.9114,
      "step": 6845
    },
    {
      "epoch": 0.5532232974403524,
      "grad_norm": 2.765091896057129,
      "learning_rate": 8.488141713363694e-06,
      "loss": 0.8719,
      "step": 6846
    },
    {
      "epoch": 0.5533041071536798,
      "grad_norm": 2.625201940536499,
      "learning_rate": 8.487672859299389e-06,
      "loss": 0.8287,
      "step": 6847
    },
    {
      "epoch": 0.5533849168670074,
      "grad_norm": 2.8303000926971436,
      "learning_rate": 8.487203945499644e-06,
      "loss": 0.9859,
      "step": 6848
    },
    {
      "epoch": 0.553465726580335,
      "grad_norm": 3.1322367191314697,
      "learning_rate": 8.486734971972488e-06,
      "loss": 0.9282,
      "step": 6849
    },
    {
      "epoch": 0.5535465362936625,
      "grad_norm": 2.78836989402771,
      "learning_rate": 8.48626593872596e-06,
      "loss": 0.8623,
      "step": 6850
    },
    {
      "epoch": 0.55362734600699,
      "grad_norm": 2.747438669204712,
      "learning_rate": 8.485796845768088e-06,
      "loss": 0.9469,
      "step": 6851
    },
    {
      "epoch": 0.5537081557203176,
      "grad_norm": 2.545588493347168,
      "learning_rate": 8.485327693106907e-06,
      "loss": 0.9366,
      "step": 6852
    },
    {
      "epoch": 0.5537889654336451,
      "grad_norm": 3.3030734062194824,
      "learning_rate": 8.484858480750453e-06,
      "loss": 0.9091,
      "step": 6853
    },
    {
      "epoch": 0.5538697751469727,
      "grad_norm": 2.8627467155456543,
      "learning_rate": 8.484389208706763e-06,
      "loss": 1.0392,
      "step": 6854
    },
    {
      "epoch": 0.5539505848603002,
      "grad_norm": 2.5307259559631348,
      "learning_rate": 8.483919876983875e-06,
      "loss": 1.0267,
      "step": 6855
    },
    {
      "epoch": 0.5540313945736277,
      "grad_norm": 2.709975242614746,
      "learning_rate": 8.483450485589825e-06,
      "loss": 1.0282,
      "step": 6856
    },
    {
      "epoch": 0.5541122042869553,
      "grad_norm": 3.306159019470215,
      "learning_rate": 8.482981034532655e-06,
      "loss": 0.9008,
      "step": 6857
    },
    {
      "epoch": 0.5541930140002829,
      "grad_norm": 2.525493621826172,
      "learning_rate": 8.482511523820407e-06,
      "loss": 1.053,
      "step": 6858
    },
    {
      "epoch": 0.5542738237136103,
      "grad_norm": 2.9141557216644287,
      "learning_rate": 8.482041953461117e-06,
      "loss": 0.9085,
      "step": 6859
    },
    {
      "epoch": 0.5543546334269379,
      "grad_norm": 2.58060359954834,
      "learning_rate": 8.481572323462831e-06,
      "loss": 0.9574,
      "step": 6860
    },
    {
      "epoch": 0.5544354431402655,
      "grad_norm": 2.615966558456421,
      "learning_rate": 8.481102633833595e-06,
      "loss": 0.8299,
      "step": 6861
    },
    {
      "epoch": 0.554516252853593,
      "grad_norm": 2.61551833152771,
      "learning_rate": 8.480632884581453e-06,
      "loss": 1.0211,
      "step": 6862
    },
    {
      "epoch": 0.5545970625669205,
      "grad_norm": 2.8956634998321533,
      "learning_rate": 8.480163075714445e-06,
      "loss": 0.8885,
      "step": 6863
    },
    {
      "epoch": 0.5546778722802481,
      "grad_norm": 2.880720615386963,
      "learning_rate": 8.479693207240624e-06,
      "loss": 1.1534,
      "step": 6864
    },
    {
      "epoch": 0.5547586819935756,
      "grad_norm": 2.5820982456207275,
      "learning_rate": 8.479223279168034e-06,
      "loss": 0.8149,
      "step": 6865
    },
    {
      "epoch": 0.5548394917069032,
      "grad_norm": 2.46889591217041,
      "learning_rate": 8.478753291504726e-06,
      "loss": 0.8981,
      "step": 6866
    },
    {
      "epoch": 0.5549203014202307,
      "grad_norm": 2.5251431465148926,
      "learning_rate": 8.478283244258749e-06,
      "loss": 1.0616,
      "step": 6867
    },
    {
      "epoch": 0.5550011111335582,
      "grad_norm": 2.763038396835327,
      "learning_rate": 8.47781313743815e-06,
      "loss": 0.9674,
      "step": 6868
    },
    {
      "epoch": 0.5550819208468858,
      "grad_norm": 2.5594913959503174,
      "learning_rate": 8.47734297105099e-06,
      "loss": 0.9419,
      "step": 6869
    },
    {
      "epoch": 0.5551627305602134,
      "grad_norm": 2.556610584259033,
      "learning_rate": 8.476872745105314e-06,
      "loss": 0.973,
      "step": 6870
    },
    {
      "epoch": 0.5552435402735408,
      "grad_norm": 3.166529417037964,
      "learning_rate": 8.476402459609177e-06,
      "loss": 0.9445,
      "step": 6871
    },
    {
      "epoch": 0.5553243499868684,
      "grad_norm": 2.978961229324341,
      "learning_rate": 8.475932114570636e-06,
      "loss": 0.9309,
      "step": 6872
    },
    {
      "epoch": 0.555405159700196,
      "grad_norm": 2.9318814277648926,
      "learning_rate": 8.475461709997747e-06,
      "loss": 0.8804,
      "step": 6873
    },
    {
      "epoch": 0.5554859694135235,
      "grad_norm": 2.61983323097229,
      "learning_rate": 8.474991245898564e-06,
      "loss": 1.0351,
      "step": 6874
    },
    {
      "epoch": 0.555566779126851,
      "grad_norm": 2.6989834308624268,
      "learning_rate": 8.474520722281148e-06,
      "loss": 0.9426,
      "step": 6875
    },
    {
      "epoch": 0.5556475888401786,
      "grad_norm": 2.703728675842285,
      "learning_rate": 8.474050139153556e-06,
      "loss": 0.9859,
      "step": 6876
    },
    {
      "epoch": 0.5557283985535061,
      "grad_norm": 2.4480912685394287,
      "learning_rate": 8.473579496523846e-06,
      "loss": 1.0633,
      "step": 6877
    },
    {
      "epoch": 0.5558092082668337,
      "grad_norm": 2.636629819869995,
      "learning_rate": 8.473108794400084e-06,
      "loss": 0.9045,
      "step": 6878
    },
    {
      "epoch": 0.5558900179801612,
      "grad_norm": 2.699205160140991,
      "learning_rate": 8.47263803279033e-06,
      "loss": 0.9455,
      "step": 6879
    },
    {
      "epoch": 0.5559708276934887,
      "grad_norm": 2.7170090675354004,
      "learning_rate": 8.472167211702646e-06,
      "loss": 1.059,
      "step": 6880
    },
    {
      "epoch": 0.5560516374068163,
      "grad_norm": 2.557217597961426,
      "learning_rate": 8.471696331145096e-06,
      "loss": 0.9927,
      "step": 6881
    },
    {
      "epoch": 0.5561324471201439,
      "grad_norm": 2.4627785682678223,
      "learning_rate": 8.471225391125745e-06,
      "loss": 0.9437,
      "step": 6882
    },
    {
      "epoch": 0.5562132568334713,
      "grad_norm": 2.7011351585388184,
      "learning_rate": 8.470754391652662e-06,
      "loss": 0.8459,
      "step": 6883
    },
    {
      "epoch": 0.5562940665467989,
      "grad_norm": 2.37245774269104,
      "learning_rate": 8.470283332733911e-06,
      "loss": 0.9837,
      "step": 6884
    },
    {
      "epoch": 0.5563748762601265,
      "grad_norm": 2.810795783996582,
      "learning_rate": 8.46981221437756e-06,
      "loss": 1.0007,
      "step": 6885
    },
    {
      "epoch": 0.556455685973454,
      "grad_norm": 2.83638596534729,
      "learning_rate": 8.46934103659168e-06,
      "loss": 1.0018,
      "step": 6886
    },
    {
      "epoch": 0.5565364956867815,
      "grad_norm": 2.2954795360565186,
      "learning_rate": 8.468869799384338e-06,
      "loss": 0.9581,
      "step": 6887
    },
    {
      "epoch": 0.5566173054001091,
      "grad_norm": 2.6287801265716553,
      "learning_rate": 8.46839850276361e-06,
      "loss": 0.9544,
      "step": 6888
    },
    {
      "epoch": 0.5566981151134366,
      "grad_norm": 2.4404377937316895,
      "learning_rate": 8.467927146737565e-06,
      "loss": 1.169,
      "step": 6889
    },
    {
      "epoch": 0.5567789248267642,
      "grad_norm": 2.6569740772247314,
      "learning_rate": 8.467455731314276e-06,
      "loss": 0.853,
      "step": 6890
    },
    {
      "epoch": 0.5568597345400917,
      "grad_norm": 2.9116122722625732,
      "learning_rate": 8.466984256501817e-06,
      "loss": 1.0032,
      "step": 6891
    },
    {
      "epoch": 0.5569405442534192,
      "grad_norm": 2.972496271133423,
      "learning_rate": 8.466512722308264e-06,
      "loss": 0.9627,
      "step": 6892
    },
    {
      "epoch": 0.5570213539667468,
      "grad_norm": 2.762817859649658,
      "learning_rate": 8.466041128741695e-06,
      "loss": 1.1623,
      "step": 6893
    },
    {
      "epoch": 0.5571021636800744,
      "grad_norm": 2.7529847621917725,
      "learning_rate": 8.465569475810185e-06,
      "loss": 0.9453,
      "step": 6894
    },
    {
      "epoch": 0.5571829733934018,
      "grad_norm": 2.5715386867523193,
      "learning_rate": 8.465097763521812e-06,
      "loss": 1.0209,
      "step": 6895
    },
    {
      "epoch": 0.5572637831067294,
      "grad_norm": 2.6773593425750732,
      "learning_rate": 8.464625991884658e-06,
      "loss": 0.8398,
      "step": 6896
    },
    {
      "epoch": 0.557344592820057,
      "grad_norm": 3.032032012939453,
      "learning_rate": 8.4641541609068e-06,
      "loss": 1.0124,
      "step": 6897
    },
    {
      "epoch": 0.5574254025333845,
      "grad_norm": 3.1945559978485107,
      "learning_rate": 8.463682270596322e-06,
      "loss": 0.958,
      "step": 6898
    },
    {
      "epoch": 0.557506212246712,
      "grad_norm": 2.645900011062622,
      "learning_rate": 8.463210320961303e-06,
      "loss": 0.8841,
      "step": 6899
    },
    {
      "epoch": 0.5575870219600396,
      "grad_norm": 2.7547786235809326,
      "learning_rate": 8.462738312009831e-06,
      "loss": 1.0065,
      "step": 6900
    },
    {
      "epoch": 0.5576678316733671,
      "grad_norm": 2.526700019836426,
      "learning_rate": 8.462266243749987e-06,
      "loss": 0.9441,
      "step": 6901
    },
    {
      "epoch": 0.5577486413866947,
      "grad_norm": 2.5337414741516113,
      "learning_rate": 8.461794116189857e-06,
      "loss": 1.057,
      "step": 6902
    },
    {
      "epoch": 0.5578294511000222,
      "grad_norm": 2.90421199798584,
      "learning_rate": 8.46132192933753e-06,
      "loss": 1.1456,
      "step": 6903
    },
    {
      "epoch": 0.5579102608133497,
      "grad_norm": 2.5645174980163574,
      "learning_rate": 8.46084968320109e-06,
      "loss": 1.0617,
      "step": 6904
    },
    {
      "epoch": 0.5579910705266773,
      "grad_norm": 2.4103856086730957,
      "learning_rate": 8.460377377788624e-06,
      "loss": 0.9526,
      "step": 6905
    },
    {
      "epoch": 0.5580718802400049,
      "grad_norm": 2.3386809825897217,
      "learning_rate": 8.459905013108225e-06,
      "loss": 0.9411,
      "step": 6906
    },
    {
      "epoch": 0.5581526899533323,
      "grad_norm": 2.7163736820220947,
      "learning_rate": 8.459432589167985e-06,
      "loss": 0.9477,
      "step": 6907
    },
    {
      "epoch": 0.5582334996666599,
      "grad_norm": 2.474764347076416,
      "learning_rate": 8.45896010597599e-06,
      "loss": 0.8838,
      "step": 6908
    },
    {
      "epoch": 0.5583143093799875,
      "grad_norm": 2.712538003921509,
      "learning_rate": 8.458487563540337e-06,
      "loss": 1.0101,
      "step": 6909
    },
    {
      "epoch": 0.558395119093315,
      "grad_norm": 2.8555169105529785,
      "learning_rate": 8.458014961869118e-06,
      "loss": 1.005,
      "step": 6910
    },
    {
      "epoch": 0.5584759288066425,
      "grad_norm": 2.7186081409454346,
      "learning_rate": 8.457542300970427e-06,
      "loss": 0.9449,
      "step": 6911
    },
    {
      "epoch": 0.5585567385199701,
      "grad_norm": 2.64795184135437,
      "learning_rate": 8.45706958085236e-06,
      "loss": 0.9931,
      "step": 6912
    },
    {
      "epoch": 0.5586375482332976,
      "grad_norm": 2.4156928062438965,
      "learning_rate": 8.456596801523014e-06,
      "loss": 1.0063,
      "step": 6913
    },
    {
      "epoch": 0.5587183579466252,
      "grad_norm": 2.5548274517059326,
      "learning_rate": 8.456123962990487e-06,
      "loss": 0.9713,
      "step": 6914
    },
    {
      "epoch": 0.5587991676599527,
      "grad_norm": 2.6821956634521484,
      "learning_rate": 8.455651065262876e-06,
      "loss": 0.8685,
      "step": 6915
    },
    {
      "epoch": 0.5588799773732803,
      "grad_norm": 2.311872959136963,
      "learning_rate": 8.45517810834828e-06,
      "loss": 1.0587,
      "step": 6916
    },
    {
      "epoch": 0.5589607870866078,
      "grad_norm": 3.150648355484009,
      "learning_rate": 8.454705092254803e-06,
      "loss": 0.9739,
      "step": 6917
    },
    {
      "epoch": 0.5590415967999354,
      "grad_norm": 2.1819918155670166,
      "learning_rate": 8.454232016990544e-06,
      "loss": 0.9802,
      "step": 6918
    },
    {
      "epoch": 0.559122406513263,
      "grad_norm": 2.6903281211853027,
      "learning_rate": 8.453758882563608e-06,
      "loss": 1.0517,
      "step": 6919
    },
    {
      "epoch": 0.5592032162265904,
      "grad_norm": 2.49411678314209,
      "learning_rate": 8.453285688982095e-06,
      "loss": 0.964,
      "step": 6920
    },
    {
      "epoch": 0.559284025939918,
      "grad_norm": 2.580589532852173,
      "learning_rate": 8.452812436254112e-06,
      "loss": 0.9212,
      "step": 6921
    },
    {
      "epoch": 0.5593648356532456,
      "grad_norm": 2.5534653663635254,
      "learning_rate": 8.452339124387763e-06,
      "loss": 1.0879,
      "step": 6922
    },
    {
      "epoch": 0.559445645366573,
      "grad_norm": 2.9323341846466064,
      "learning_rate": 8.451865753391158e-06,
      "loss": 1.0121,
      "step": 6923
    },
    {
      "epoch": 0.5595264550799006,
      "grad_norm": 2.8727517127990723,
      "learning_rate": 8.451392323272401e-06,
      "loss": 0.9481,
      "step": 6924
    },
    {
      "epoch": 0.5596072647932282,
      "grad_norm": 2.88751220703125,
      "learning_rate": 8.450918834039602e-06,
      "loss": 0.8822,
      "step": 6925
    },
    {
      "epoch": 0.5596880745065557,
      "grad_norm": 2.4941325187683105,
      "learning_rate": 8.450445285700875e-06,
      "loss": 0.9589,
      "step": 6926
    },
    {
      "epoch": 0.5597688842198832,
      "grad_norm": 2.9503421783447266,
      "learning_rate": 8.449971678264322e-06,
      "loss": 0.9996,
      "step": 6927
    },
    {
      "epoch": 0.5598496939332108,
      "grad_norm": 2.571575403213501,
      "learning_rate": 8.449498011738063e-06,
      "loss": 1.0112,
      "step": 6928
    },
    {
      "epoch": 0.5599305036465383,
      "grad_norm": 2.6412312984466553,
      "learning_rate": 8.449024286130207e-06,
      "loss": 0.8747,
      "step": 6929
    },
    {
      "epoch": 0.5600113133598659,
      "grad_norm": 2.6379234790802,
      "learning_rate": 8.448550501448867e-06,
      "loss": 0.9118,
      "step": 6930
    },
    {
      "epoch": 0.5600921230731934,
      "grad_norm": 2.6920595169067383,
      "learning_rate": 8.448076657702158e-06,
      "loss": 0.89,
      "step": 6931
    },
    {
      "epoch": 0.5601729327865209,
      "grad_norm": 2.924105644226074,
      "learning_rate": 8.4476027548982e-06,
      "loss": 0.9643,
      "step": 6932
    },
    {
      "epoch": 0.5602537424998485,
      "grad_norm": 2.5920870304107666,
      "learning_rate": 8.447128793045103e-06,
      "loss": 0.9073,
      "step": 6933
    },
    {
      "epoch": 0.5603345522131761,
      "grad_norm": 2.683720827102661,
      "learning_rate": 8.44665477215099e-06,
      "loss": 0.9764,
      "step": 6934
    },
    {
      "epoch": 0.5604153619265035,
      "grad_norm": 2.871969223022461,
      "learning_rate": 8.446180692223977e-06,
      "loss": 0.8684,
      "step": 6935
    },
    {
      "epoch": 0.5604961716398311,
      "grad_norm": 2.516979217529297,
      "learning_rate": 8.445706553272188e-06,
      "loss": 1.0522,
      "step": 6936
    },
    {
      "epoch": 0.5605769813531587,
      "grad_norm": 2.7543396949768066,
      "learning_rate": 8.445232355303739e-06,
      "loss": 0.976,
      "step": 6937
    },
    {
      "epoch": 0.5606577910664862,
      "grad_norm": 2.5632481575012207,
      "learning_rate": 8.444758098326753e-06,
      "loss": 0.9587,
      "step": 6938
    },
    {
      "epoch": 0.5607386007798137,
      "grad_norm": 2.6871962547302246,
      "learning_rate": 8.444283782349356e-06,
      "loss": 1.001,
      "step": 6939
    },
    {
      "epoch": 0.5608194104931413,
      "grad_norm": 2.6770501136779785,
      "learning_rate": 8.443809407379668e-06,
      "loss": 1.0675,
      "step": 6940
    },
    {
      "epoch": 0.5609002202064688,
      "grad_norm": 2.706904649734497,
      "learning_rate": 8.443334973425817e-06,
      "loss": 0.9417,
      "step": 6941
    },
    {
      "epoch": 0.5609810299197964,
      "grad_norm": 2.690647840499878,
      "learning_rate": 8.442860480495925e-06,
      "loss": 0.9333,
      "step": 6942
    },
    {
      "epoch": 0.561061839633124,
      "grad_norm": 3.1479380130767822,
      "learning_rate": 8.442385928598123e-06,
      "loss": 1.0568,
      "step": 6943
    },
    {
      "epoch": 0.5611426493464514,
      "grad_norm": 2.623107671737671,
      "learning_rate": 8.441911317740537e-06,
      "loss": 0.9015,
      "step": 6944
    },
    {
      "epoch": 0.561223459059779,
      "grad_norm": 2.5914926528930664,
      "learning_rate": 8.441436647931296e-06,
      "loss": 0.9363,
      "step": 6945
    },
    {
      "epoch": 0.5613042687731066,
      "grad_norm": 2.887040376663208,
      "learning_rate": 8.440961919178529e-06,
      "loss": 1.0202,
      "step": 6946
    },
    {
      "epoch": 0.561385078486434,
      "grad_norm": 2.8716883659362793,
      "learning_rate": 8.44048713149037e-06,
      "loss": 0.9894,
      "step": 6947
    },
    {
      "epoch": 0.5614658881997616,
      "grad_norm": 2.8763785362243652,
      "learning_rate": 8.44001228487495e-06,
      "loss": 1.0792,
      "step": 6948
    },
    {
      "epoch": 0.5615466979130892,
      "grad_norm": 2.7583351135253906,
      "learning_rate": 8.439537379340398e-06,
      "loss": 1.043,
      "step": 6949
    },
    {
      "epoch": 0.5616275076264167,
      "grad_norm": 2.1673460006713867,
      "learning_rate": 8.439062414894852e-06,
      "loss": 1.0191,
      "step": 6950
    },
    {
      "epoch": 0.5617083173397442,
      "grad_norm": 2.888838529586792,
      "learning_rate": 8.438587391546447e-06,
      "loss": 1.0027,
      "step": 6951
    },
    {
      "epoch": 0.5617891270530718,
      "grad_norm": 2.6036126613616943,
      "learning_rate": 8.438112309303318e-06,
      "loss": 0.9132,
      "step": 6952
    },
    {
      "epoch": 0.5618699367663993,
      "grad_norm": 2.374340295791626,
      "learning_rate": 8.4376371681736e-06,
      "loss": 0.9941,
      "step": 6953
    },
    {
      "epoch": 0.5619507464797269,
      "grad_norm": 2.4654552936553955,
      "learning_rate": 8.437161968165436e-06,
      "loss": 0.9955,
      "step": 6954
    },
    {
      "epoch": 0.5620315561930544,
      "grad_norm": 2.4640560150146484,
      "learning_rate": 8.43668670928696e-06,
      "loss": 0.9802,
      "step": 6955
    },
    {
      "epoch": 0.5621123659063819,
      "grad_norm": 2.866013526916504,
      "learning_rate": 8.436211391546315e-06,
      "loss": 0.8866,
      "step": 6956
    },
    {
      "epoch": 0.5621931756197095,
      "grad_norm": 3.00115704536438,
      "learning_rate": 8.43573601495164e-06,
      "loss": 0.8571,
      "step": 6957
    },
    {
      "epoch": 0.5622739853330371,
      "grad_norm": 2.5477354526519775,
      "learning_rate": 8.43526057951108e-06,
      "loss": 0.9537,
      "step": 6958
    },
    {
      "epoch": 0.5623547950463645,
      "grad_norm": 2.5641021728515625,
      "learning_rate": 8.434785085232777e-06,
      "loss": 0.9218,
      "step": 6959
    },
    {
      "epoch": 0.5624356047596921,
      "grad_norm": 2.412095069885254,
      "learning_rate": 8.434309532124872e-06,
      "loss": 1.0215,
      "step": 6960
    },
    {
      "epoch": 0.5625164144730197,
      "grad_norm": 2.4804725646972656,
      "learning_rate": 8.433833920195514e-06,
      "loss": 0.925,
      "step": 6961
    },
    {
      "epoch": 0.5625972241863472,
      "grad_norm": 2.8121204376220703,
      "learning_rate": 8.433358249452848e-06,
      "loss": 0.8633,
      "step": 6962
    },
    {
      "epoch": 0.5626780338996747,
      "grad_norm": 2.7459716796875,
      "learning_rate": 8.432882519905019e-06,
      "loss": 0.9291,
      "step": 6963
    },
    {
      "epoch": 0.5627588436130023,
      "grad_norm": 2.6513407230377197,
      "learning_rate": 8.432406731560178e-06,
      "loss": 0.9366,
      "step": 6964
    },
    {
      "epoch": 0.5628396533263298,
      "grad_norm": 2.5133769512176514,
      "learning_rate": 8.431930884426472e-06,
      "loss": 0.9824,
      "step": 6965
    },
    {
      "epoch": 0.5629204630396574,
      "grad_norm": 2.3399899005889893,
      "learning_rate": 8.431454978512052e-06,
      "loss": 1.0431,
      "step": 6966
    },
    {
      "epoch": 0.563001272752985,
      "grad_norm": 2.996201515197754,
      "learning_rate": 8.430979013825069e-06,
      "loss": 0.986,
      "step": 6967
    },
    {
      "epoch": 0.5630820824663124,
      "grad_norm": 2.232870578765869,
      "learning_rate": 8.430502990373677e-06,
      "loss": 1.0036,
      "step": 6968
    },
    {
      "epoch": 0.56316289217964,
      "grad_norm": 3.242262840270996,
      "learning_rate": 8.430026908166026e-06,
      "loss": 0.9543,
      "step": 6969
    },
    {
      "epoch": 0.5632437018929676,
      "grad_norm": 3.069676399230957,
      "learning_rate": 8.429550767210271e-06,
      "loss": 0.8638,
      "step": 6970
    },
    {
      "epoch": 0.563324511606295,
      "grad_norm": 2.892313003540039,
      "learning_rate": 8.42907456751457e-06,
      "loss": 0.9887,
      "step": 6971
    },
    {
      "epoch": 0.5634053213196226,
      "grad_norm": 2.5909368991851807,
      "learning_rate": 8.428598309087073e-06,
      "loss": 0.9523,
      "step": 6972
    },
    {
      "epoch": 0.5634861310329502,
      "grad_norm": 2.4261319637298584,
      "learning_rate": 8.428121991935945e-06,
      "loss": 0.9799,
      "step": 6973
    },
    {
      "epoch": 0.5635669407462777,
      "grad_norm": 2.2519493103027344,
      "learning_rate": 8.427645616069338e-06,
      "loss": 0.9108,
      "step": 6974
    },
    {
      "epoch": 0.5636477504596052,
      "grad_norm": 2.5064189434051514,
      "learning_rate": 8.427169181495413e-06,
      "loss": 0.9722,
      "step": 6975
    },
    {
      "epoch": 0.5637285601729328,
      "grad_norm": 2.6627674102783203,
      "learning_rate": 8.426692688222332e-06,
      "loss": 0.9325,
      "step": 6976
    },
    {
      "epoch": 0.5638093698862603,
      "grad_norm": 2.35725474357605,
      "learning_rate": 8.426216136258251e-06,
      "loss": 1.0678,
      "step": 6977
    },
    {
      "epoch": 0.5638901795995879,
      "grad_norm": 2.640176773071289,
      "learning_rate": 8.42573952561134e-06,
      "loss": 0.921,
      "step": 6978
    },
    {
      "epoch": 0.5639709893129154,
      "grad_norm": 2.5032410621643066,
      "learning_rate": 8.425262856289757e-06,
      "loss": 0.8822,
      "step": 6979
    },
    {
      "epoch": 0.5640517990262429,
      "grad_norm": 2.42099928855896,
      "learning_rate": 8.424786128301666e-06,
      "loss": 0.8589,
      "step": 6980
    },
    {
      "epoch": 0.5641326087395705,
      "grad_norm": 2.7244465351104736,
      "learning_rate": 8.424309341655235e-06,
      "loss": 0.9118,
      "step": 6981
    },
    {
      "epoch": 0.5642134184528981,
      "grad_norm": 2.6891651153564453,
      "learning_rate": 8.423832496358629e-06,
      "loss": 0.9213,
      "step": 6982
    },
    {
      "epoch": 0.5642942281662255,
      "grad_norm": 2.518742322921753,
      "learning_rate": 8.423355592420014e-06,
      "loss": 0.9701,
      "step": 6983
    },
    {
      "epoch": 0.5643750378795531,
      "grad_norm": 2.695110321044922,
      "learning_rate": 8.422878629847557e-06,
      "loss": 0.9443,
      "step": 6984
    },
    {
      "epoch": 0.5644558475928807,
      "grad_norm": 2.894137144088745,
      "learning_rate": 8.422401608649433e-06,
      "loss": 0.9273,
      "step": 6985
    },
    {
      "epoch": 0.5645366573062082,
      "grad_norm": 2.9529757499694824,
      "learning_rate": 8.421924528833806e-06,
      "loss": 0.9318,
      "step": 6986
    },
    {
      "epoch": 0.5646174670195357,
      "grad_norm": 2.6024718284606934,
      "learning_rate": 8.421447390408851e-06,
      "loss": 0.953,
      "step": 6987
    },
    {
      "epoch": 0.5646982767328633,
      "grad_norm": 2.8305442333221436,
      "learning_rate": 8.420970193382739e-06,
      "loss": 1.1961,
      "step": 6988
    },
    {
      "epoch": 0.5647790864461908,
      "grad_norm": 2.8597564697265625,
      "learning_rate": 8.420492937763642e-06,
      "loss": 0.9659,
      "step": 6989
    },
    {
      "epoch": 0.5648598961595184,
      "grad_norm": 2.536705732345581,
      "learning_rate": 8.420015623559737e-06,
      "loss": 0.9049,
      "step": 6990
    },
    {
      "epoch": 0.564940705872846,
      "grad_norm": 3.3660593032836914,
      "learning_rate": 8.419538250779197e-06,
      "loss": 0.9746,
      "step": 6991
    },
    {
      "epoch": 0.5650215155861734,
      "grad_norm": 3.0411932468414307,
      "learning_rate": 8.419060819430198e-06,
      "loss": 0.9653,
      "step": 6992
    },
    {
      "epoch": 0.565102325299501,
      "grad_norm": 2.857156276702881,
      "learning_rate": 8.41858332952092e-06,
      "loss": 0.9238,
      "step": 6993
    },
    {
      "epoch": 0.5651831350128286,
      "grad_norm": 2.5991122722625732,
      "learning_rate": 8.418105781059538e-06,
      "loss": 0.9415,
      "step": 6994
    },
    {
      "epoch": 0.565263944726156,
      "grad_norm": 2.8292133808135986,
      "learning_rate": 8.417628174054234e-06,
      "loss": 0.9808,
      "step": 6995
    },
    {
      "epoch": 0.5653447544394836,
      "grad_norm": 2.5704286098480225,
      "learning_rate": 8.417150508513187e-06,
      "loss": 0.9807,
      "step": 6996
    },
    {
      "epoch": 0.5654255641528112,
      "grad_norm": 3.0911080837249756,
      "learning_rate": 8.416672784444577e-06,
      "loss": 1.0539,
      "step": 6997
    },
    {
      "epoch": 0.5655063738661387,
      "grad_norm": 2.783277750015259,
      "learning_rate": 8.416195001856588e-06,
      "loss": 0.866,
      "step": 6998
    },
    {
      "epoch": 0.5655871835794662,
      "grad_norm": 2.933220624923706,
      "learning_rate": 8.415717160757403e-06,
      "loss": 0.8734,
      "step": 6999
    },
    {
      "epoch": 0.5656679932927938,
      "grad_norm": 2.308877468109131,
      "learning_rate": 8.415239261155206e-06,
      "loss": 0.9979,
      "step": 7000
    },
    {
      "epoch": 0.5656679932927938,
      "eval_loss": 0.8000102639198303,
      "eval_runtime": 816.7886,
      "eval_samples_per_second": 102.066,
      "eval_steps_per_second": 12.759,
      "step": 7000
    },
    {
      "epoch": 0.5657488030061213,
      "grad_norm": 2.592982769012451,
      "learning_rate": 8.414761303058183e-06,
      "loss": 1.0604,
      "step": 7001
    },
    {
      "epoch": 0.5658296127194489,
      "grad_norm": 2.4939510822296143,
      "learning_rate": 8.414283286474519e-06,
      "loss": 0.9632,
      "step": 7002
    },
    {
      "epoch": 0.5659104224327764,
      "grad_norm": 2.696521043777466,
      "learning_rate": 8.413805211412401e-06,
      "loss": 1.0048,
      "step": 7003
    },
    {
      "epoch": 0.5659912321461039,
      "grad_norm": 2.826878547668457,
      "learning_rate": 8.413327077880021e-06,
      "loss": 0.9808,
      "step": 7004
    },
    {
      "epoch": 0.5660720418594315,
      "grad_norm": 2.856379270553589,
      "learning_rate": 8.412848885885562e-06,
      "loss": 0.9076,
      "step": 7005
    },
    {
      "epoch": 0.5661528515727591,
      "grad_norm": 2.592928647994995,
      "learning_rate": 8.41237063543722e-06,
      "loss": 0.9142,
      "step": 7006
    },
    {
      "epoch": 0.5662336612860865,
      "grad_norm": 2.515559434890747,
      "learning_rate": 8.411892326543181e-06,
      "loss": 1.0286,
      "step": 7007
    },
    {
      "epoch": 0.5663144709994141,
      "grad_norm": 2.604215145111084,
      "learning_rate": 8.411413959211642e-06,
      "loss": 0.9673,
      "step": 7008
    },
    {
      "epoch": 0.5663952807127417,
      "grad_norm": 2.451754093170166,
      "learning_rate": 8.410935533450796e-06,
      "loss": 1.0003,
      "step": 7009
    },
    {
      "epoch": 0.5664760904260692,
      "grad_norm": 3.207737684249878,
      "learning_rate": 8.410457049268834e-06,
      "loss": 0.8736,
      "step": 7010
    },
    {
      "epoch": 0.5665569001393967,
      "grad_norm": 2.349968194961548,
      "learning_rate": 8.409978506673954e-06,
      "loss": 0.8771,
      "step": 7011
    },
    {
      "epoch": 0.5666377098527243,
      "grad_norm": 2.256417751312256,
      "learning_rate": 8.409499905674351e-06,
      "loss": 1.0773,
      "step": 7012
    },
    {
      "epoch": 0.5667185195660518,
      "grad_norm": 2.558096408843994,
      "learning_rate": 8.409021246278222e-06,
      "loss": 0.9073,
      "step": 7013
    },
    {
      "epoch": 0.5667993292793794,
      "grad_norm": 2.764721155166626,
      "learning_rate": 8.408542528493765e-06,
      "loss": 0.9293,
      "step": 7014
    },
    {
      "epoch": 0.566880138992707,
      "grad_norm": 2.7050933837890625,
      "learning_rate": 8.408063752329182e-06,
      "loss": 0.8146,
      "step": 7015
    },
    {
      "epoch": 0.5669609487060344,
      "grad_norm": 2.8810532093048096,
      "learning_rate": 8.407584917792672e-06,
      "loss": 0.9299,
      "step": 7016
    },
    {
      "epoch": 0.567041758419362,
      "grad_norm": 2.6436405181884766,
      "learning_rate": 8.407106024892436e-06,
      "loss": 0.9474,
      "step": 7017
    },
    {
      "epoch": 0.5671225681326896,
      "grad_norm": 2.8703296184539795,
      "learning_rate": 8.406627073636674e-06,
      "loss": 0.8982,
      "step": 7018
    },
    {
      "epoch": 0.567203377846017,
      "grad_norm": 2.899139642715454,
      "learning_rate": 8.406148064033592e-06,
      "loss": 1.0491,
      "step": 7019
    },
    {
      "epoch": 0.5672841875593446,
      "grad_norm": 2.6143975257873535,
      "learning_rate": 8.405668996091397e-06,
      "loss": 0.9591,
      "step": 7020
    },
    {
      "epoch": 0.5673649972726722,
      "grad_norm": 2.3370583057403564,
      "learning_rate": 8.405189869818286e-06,
      "loss": 1.0002,
      "step": 7021
    },
    {
      "epoch": 0.5674458069859997,
      "grad_norm": 2.314232349395752,
      "learning_rate": 8.404710685222473e-06,
      "loss": 0.9367,
      "step": 7022
    },
    {
      "epoch": 0.5675266166993272,
      "grad_norm": 3.1822986602783203,
      "learning_rate": 8.40423144231216e-06,
      "loss": 0.9731,
      "step": 7023
    },
    {
      "epoch": 0.5676074264126548,
      "grad_norm": 2.6218934059143066,
      "learning_rate": 8.40375214109556e-06,
      "loss": 0.9003,
      "step": 7024
    },
    {
      "epoch": 0.5676882361259823,
      "grad_norm": 2.5922951698303223,
      "learning_rate": 8.40327278158088e-06,
      "loss": 0.941,
      "step": 7025
    },
    {
      "epoch": 0.5677690458393099,
      "grad_norm": 2.763942241668701,
      "learning_rate": 8.402793363776329e-06,
      "loss": 1.0936,
      "step": 7026
    },
    {
      "epoch": 0.5678498555526375,
      "grad_norm": 2.2726926803588867,
      "learning_rate": 8.402313887690122e-06,
      "loss": 1.0425,
      "step": 7027
    },
    {
      "epoch": 0.5679306652659649,
      "grad_norm": 2.314948081970215,
      "learning_rate": 8.401834353330466e-06,
      "loss": 0.938,
      "step": 7028
    },
    {
      "epoch": 0.5680114749792925,
      "grad_norm": 2.565999984741211,
      "learning_rate": 8.401354760705578e-06,
      "loss": 1.0071,
      "step": 7029
    },
    {
      "epoch": 0.5680922846926201,
      "grad_norm": 2.7477312088012695,
      "learning_rate": 8.400875109823672e-06,
      "loss": 0.9649,
      "step": 7030
    },
    {
      "epoch": 0.5681730944059475,
      "grad_norm": 2.6854000091552734,
      "learning_rate": 8.400395400692962e-06,
      "loss": 0.9304,
      "step": 7031
    },
    {
      "epoch": 0.5682539041192751,
      "grad_norm": 2.1080284118652344,
      "learning_rate": 8.399915633321664e-06,
      "loss": 0.9809,
      "step": 7032
    },
    {
      "epoch": 0.5683347138326027,
      "grad_norm": 2.6586344242095947,
      "learning_rate": 8.399435807717998e-06,
      "loss": 1.17,
      "step": 7033
    },
    {
      "epoch": 0.5684155235459302,
      "grad_norm": 2.4868946075439453,
      "learning_rate": 8.39895592389018e-06,
      "loss": 0.8837,
      "step": 7034
    },
    {
      "epoch": 0.5684963332592577,
      "grad_norm": 2.6929616928100586,
      "learning_rate": 8.398475981846431e-06,
      "loss": 0.9189,
      "step": 7035
    },
    {
      "epoch": 0.5685771429725853,
      "grad_norm": 2.7626705169677734,
      "learning_rate": 8.397995981594966e-06,
      "loss": 0.9134,
      "step": 7036
    },
    {
      "epoch": 0.5686579526859128,
      "grad_norm": 2.5226573944091797,
      "learning_rate": 8.397515923144012e-06,
      "loss": 1.019,
      "step": 7037
    },
    {
      "epoch": 0.5687387623992404,
      "grad_norm": 3.266282558441162,
      "learning_rate": 8.397035806501792e-06,
      "loss": 0.9652,
      "step": 7038
    },
    {
      "epoch": 0.568819572112568,
      "grad_norm": 2.445091724395752,
      "learning_rate": 8.396555631676523e-06,
      "loss": 0.9725,
      "step": 7039
    },
    {
      "epoch": 0.5689003818258954,
      "grad_norm": 2.4900972843170166,
      "learning_rate": 8.396075398676435e-06,
      "loss": 0.9285,
      "step": 7040
    },
    {
      "epoch": 0.568981191539223,
      "grad_norm": 2.3779239654541016,
      "learning_rate": 8.395595107509751e-06,
      "loss": 0.9441,
      "step": 7041
    },
    {
      "epoch": 0.5690620012525506,
      "grad_norm": 2.8242993354797363,
      "learning_rate": 8.395114758184697e-06,
      "loss": 1.0378,
      "step": 7042
    },
    {
      "epoch": 0.569142810965878,
      "grad_norm": 2.4411308765411377,
      "learning_rate": 8.394634350709501e-06,
      "loss": 0.8461,
      "step": 7043
    },
    {
      "epoch": 0.5692236206792056,
      "grad_norm": 2.3372957706451416,
      "learning_rate": 8.39415388509239e-06,
      "loss": 0.9613,
      "step": 7044
    },
    {
      "epoch": 0.5693044303925332,
      "grad_norm": 2.3165688514709473,
      "learning_rate": 8.393673361341594e-06,
      "loss": 0.9957,
      "step": 7045
    },
    {
      "epoch": 0.5693852401058608,
      "grad_norm": 2.674180030822754,
      "learning_rate": 8.393192779465345e-06,
      "loss": 0.9776,
      "step": 7046
    },
    {
      "epoch": 0.5694660498191882,
      "grad_norm": 3.1991283893585205,
      "learning_rate": 8.39271213947187e-06,
      "loss": 0.8736,
      "step": 7047
    },
    {
      "epoch": 0.5695468595325158,
      "grad_norm": 2.841813802719116,
      "learning_rate": 8.392231441369405e-06,
      "loss": 0.9953,
      "step": 7048
    },
    {
      "epoch": 0.5696276692458434,
      "grad_norm": 3.1734557151794434,
      "learning_rate": 8.391750685166182e-06,
      "loss": 0.9715,
      "step": 7049
    },
    {
      "epoch": 0.5697084789591709,
      "grad_norm": 2.545691967010498,
      "learning_rate": 8.391269870870435e-06,
      "loss": 1.0736,
      "step": 7050
    },
    {
      "epoch": 0.5697892886724985,
      "grad_norm": 2.592190742492676,
      "learning_rate": 8.3907889984904e-06,
      "loss": 0.9419,
      "step": 7051
    },
    {
      "epoch": 0.569870098385826,
      "grad_norm": 2.8696320056915283,
      "learning_rate": 8.390308068034312e-06,
      "loss": 0.9299,
      "step": 7052
    },
    {
      "epoch": 0.5699509080991535,
      "grad_norm": 2.8169021606445312,
      "learning_rate": 8.389827079510406e-06,
      "loss": 1.1305,
      "step": 7053
    },
    {
      "epoch": 0.5700317178124811,
      "grad_norm": 2.3149781227111816,
      "learning_rate": 8.389346032926926e-06,
      "loss": 0.9445,
      "step": 7054
    },
    {
      "epoch": 0.5701125275258087,
      "grad_norm": 2.4093823432922363,
      "learning_rate": 8.388864928292106e-06,
      "loss": 1.1357,
      "step": 7055
    },
    {
      "epoch": 0.5701933372391361,
      "grad_norm": 2.9673547744750977,
      "learning_rate": 8.38838376561419e-06,
      "loss": 0.8642,
      "step": 7056
    },
    {
      "epoch": 0.5702741469524637,
      "grad_norm": 2.409287452697754,
      "learning_rate": 8.387902544901416e-06,
      "loss": 0.9395,
      "step": 7057
    },
    {
      "epoch": 0.5703549566657913,
      "grad_norm": 2.6744260787963867,
      "learning_rate": 8.387421266162027e-06,
      "loss": 1.0365,
      "step": 7058
    },
    {
      "epoch": 0.5704357663791187,
      "grad_norm": 2.8118560314178467,
      "learning_rate": 8.386939929404268e-06,
      "loss": 0.8687,
      "step": 7059
    },
    {
      "epoch": 0.5705165760924463,
      "grad_norm": 2.3100764751434326,
      "learning_rate": 8.386458534636382e-06,
      "loss": 1.105,
      "step": 7060
    },
    {
      "epoch": 0.5705973858057739,
      "grad_norm": 2.456096887588501,
      "learning_rate": 8.385977081866611e-06,
      "loss": 0.9708,
      "step": 7061
    },
    {
      "epoch": 0.5706781955191014,
      "grad_norm": 2.628122091293335,
      "learning_rate": 8.385495571103209e-06,
      "loss": 1.0057,
      "step": 7062
    },
    {
      "epoch": 0.570759005232429,
      "grad_norm": 3.010484457015991,
      "learning_rate": 8.385014002354415e-06,
      "loss": 0.9209,
      "step": 7063
    },
    {
      "epoch": 0.5708398149457565,
      "grad_norm": 3.124950647354126,
      "learning_rate": 8.384532375628478e-06,
      "loss": 0.9172,
      "step": 7064
    },
    {
      "epoch": 0.570920624659084,
      "grad_norm": 4.180485248565674,
      "learning_rate": 8.384050690933653e-06,
      "loss": 0.9709,
      "step": 7065
    },
    {
      "epoch": 0.5710014343724116,
      "grad_norm": 2.5992653369903564,
      "learning_rate": 8.383568948278185e-06,
      "loss": 1.1398,
      "step": 7066
    },
    {
      "epoch": 0.5710822440857392,
      "grad_norm": 2.78249454498291,
      "learning_rate": 8.383087147670325e-06,
      "loss": 0.9614,
      "step": 7067
    },
    {
      "epoch": 0.5711630537990666,
      "grad_norm": 2.8518197536468506,
      "learning_rate": 8.38260528911833e-06,
      "loss": 0.9481,
      "step": 7068
    },
    {
      "epoch": 0.5712438635123942,
      "grad_norm": 2.5273258686065674,
      "learning_rate": 8.382123372630448e-06,
      "loss": 0.8787,
      "step": 7069
    },
    {
      "epoch": 0.5713246732257218,
      "grad_norm": 2.849349021911621,
      "learning_rate": 8.381641398214935e-06,
      "loss": 0.9052,
      "step": 7070
    },
    {
      "epoch": 0.5714054829390492,
      "grad_norm": 2.7267680168151855,
      "learning_rate": 8.381159365880045e-06,
      "loss": 1.0531,
      "step": 7071
    },
    {
      "epoch": 0.5714862926523768,
      "grad_norm": 3.0048511028289795,
      "learning_rate": 8.380677275634035e-06,
      "loss": 1.0393,
      "step": 7072
    },
    {
      "epoch": 0.5715671023657044,
      "grad_norm": 2.633543014526367,
      "learning_rate": 8.380195127485161e-06,
      "loss": 1.0095,
      "step": 7073
    },
    {
      "epoch": 0.5716479120790319,
      "grad_norm": 2.8810977935791016,
      "learning_rate": 8.379712921441685e-06,
      "loss": 0.781,
      "step": 7074
    },
    {
      "epoch": 0.5717287217923595,
      "grad_norm": 2.575315475463867,
      "learning_rate": 8.379230657511861e-06,
      "loss": 1.0401,
      "step": 7075
    },
    {
      "epoch": 0.571809531505687,
      "grad_norm": 2.9347527027130127,
      "learning_rate": 8.378748335703953e-06,
      "loss": 0.9515,
      "step": 7076
    },
    {
      "epoch": 0.5718903412190145,
      "grad_norm": 2.5889923572540283,
      "learning_rate": 8.378265956026216e-06,
      "loss": 0.8905,
      "step": 7077
    },
    {
      "epoch": 0.5719711509323421,
      "grad_norm": 2.9868171215057373,
      "learning_rate": 8.377783518486919e-06,
      "loss": 0.9832,
      "step": 7078
    },
    {
      "epoch": 0.5720519606456697,
      "grad_norm": 2.6267364025115967,
      "learning_rate": 8.377301023094322e-06,
      "loss": 0.9181,
      "step": 7079
    },
    {
      "epoch": 0.5721327703589971,
      "grad_norm": 2.6298704147338867,
      "learning_rate": 8.376818469856687e-06,
      "loss": 0.9886,
      "step": 7080
    },
    {
      "epoch": 0.5722135800723247,
      "grad_norm": 2.3796985149383545,
      "learning_rate": 8.376335858782282e-06,
      "loss": 0.9637,
      "step": 7081
    },
    {
      "epoch": 0.5722943897856523,
      "grad_norm": 2.550248861312866,
      "learning_rate": 8.375853189879373e-06,
      "loss": 0.9606,
      "step": 7082
    },
    {
      "epoch": 0.5723751994989797,
      "grad_norm": 3.1329081058502197,
      "learning_rate": 8.375370463156225e-06,
      "loss": 1.0353,
      "step": 7083
    },
    {
      "epoch": 0.5724560092123073,
      "grad_norm": 2.2887074947357178,
      "learning_rate": 8.374887678621106e-06,
      "loss": 0.964,
      "step": 7084
    },
    {
      "epoch": 0.5725368189256349,
      "grad_norm": 2.3649778366088867,
      "learning_rate": 8.374404836282288e-06,
      "loss": 0.9174,
      "step": 7085
    },
    {
      "epoch": 0.5726176286389624,
      "grad_norm": 2.7177512645721436,
      "learning_rate": 8.373921936148037e-06,
      "loss": 0.9458,
      "step": 7086
    },
    {
      "epoch": 0.57269843835229,
      "grad_norm": 2.5574512481689453,
      "learning_rate": 8.373438978226627e-06,
      "loss": 1.0039,
      "step": 7087
    },
    {
      "epoch": 0.5727792480656175,
      "grad_norm": 2.699385166168213,
      "learning_rate": 8.372955962526326e-06,
      "loss": 0.9806,
      "step": 7088
    },
    {
      "epoch": 0.572860057778945,
      "grad_norm": 2.452263355255127,
      "learning_rate": 8.372472889055412e-06,
      "loss": 0.9901,
      "step": 7089
    },
    {
      "epoch": 0.5729408674922726,
      "grad_norm": 2.6215109825134277,
      "learning_rate": 8.371989757822154e-06,
      "loss": 0.934,
      "step": 7090
    },
    {
      "epoch": 0.5730216772056002,
      "grad_norm": 2.643709182739258,
      "learning_rate": 8.371506568834831e-06,
      "loss": 0.933,
      "step": 7091
    },
    {
      "epoch": 0.5731024869189276,
      "grad_norm": 3.367661952972412,
      "learning_rate": 8.371023322101716e-06,
      "loss": 1.0407,
      "step": 7092
    },
    {
      "epoch": 0.5731832966322552,
      "grad_norm": 2.4877586364746094,
      "learning_rate": 8.370540017631087e-06,
      "loss": 1.1873,
      "step": 7093
    },
    {
      "epoch": 0.5732641063455828,
      "grad_norm": 2.6113100051879883,
      "learning_rate": 8.370056655431224e-06,
      "loss": 0.9028,
      "step": 7094
    },
    {
      "epoch": 0.5733449160589102,
      "grad_norm": 2.850865364074707,
      "learning_rate": 8.3695732355104e-06,
      "loss": 0.9281,
      "step": 7095
    },
    {
      "epoch": 0.5734257257722378,
      "grad_norm": 2.618288993835449,
      "learning_rate": 8.369089757876901e-06,
      "loss": 0.846,
      "step": 7096
    },
    {
      "epoch": 0.5735065354855654,
      "grad_norm": 2.4725635051727295,
      "learning_rate": 8.368606222539004e-06,
      "loss": 1.0211,
      "step": 7097
    },
    {
      "epoch": 0.5735873451988929,
      "grad_norm": 2.657615900039673,
      "learning_rate": 8.368122629504994e-06,
      "loss": 0.9606,
      "step": 7098
    },
    {
      "epoch": 0.5736681549122205,
      "grad_norm": 2.788099527359009,
      "learning_rate": 8.367638978783149e-06,
      "loss": 1.0066,
      "step": 7099
    },
    {
      "epoch": 0.573748964625548,
      "grad_norm": 2.808821201324463,
      "learning_rate": 8.367155270381757e-06,
      "loss": 0.962,
      "step": 7100
    },
    {
      "epoch": 0.5738297743388755,
      "grad_norm": 2.753215789794922,
      "learning_rate": 8.3666715043091e-06,
      "loss": 0.8577,
      "step": 7101
    },
    {
      "epoch": 0.5739105840522031,
      "grad_norm": 2.542543649673462,
      "learning_rate": 8.366187680573466e-06,
      "loss": 1.0162,
      "step": 7102
    },
    {
      "epoch": 0.5739913937655307,
      "grad_norm": 2.808396816253662,
      "learning_rate": 8.365703799183141e-06,
      "loss": 1.0071,
      "step": 7103
    },
    {
      "epoch": 0.5740722034788581,
      "grad_norm": 2.6289422512054443,
      "learning_rate": 8.365219860146413e-06,
      "loss": 0.9115,
      "step": 7104
    },
    {
      "epoch": 0.5741530131921857,
      "grad_norm": 2.6151201725006104,
      "learning_rate": 8.364735863471569e-06,
      "loss": 0.9664,
      "step": 7105
    },
    {
      "epoch": 0.5742338229055133,
      "grad_norm": 2.8494882583618164,
      "learning_rate": 8.364251809166902e-06,
      "loss": 0.9145,
      "step": 7106
    },
    {
      "epoch": 0.5743146326188407,
      "grad_norm": 2.6571810245513916,
      "learning_rate": 8.363767697240697e-06,
      "loss": 0.8728,
      "step": 7107
    },
    {
      "epoch": 0.5743954423321683,
      "grad_norm": 2.619377851486206,
      "learning_rate": 8.363283527701252e-06,
      "loss": 0.8994,
      "step": 7108
    },
    {
      "epoch": 0.5744762520454959,
      "grad_norm": 2.767608642578125,
      "learning_rate": 8.362799300556856e-06,
      "loss": 0.8547,
      "step": 7109
    },
    {
      "epoch": 0.5745570617588234,
      "grad_norm": 2.7516744136810303,
      "learning_rate": 8.362315015815805e-06,
      "loss": 0.9478,
      "step": 7110
    },
    {
      "epoch": 0.574637871472151,
      "grad_norm": 2.5909171104431152,
      "learning_rate": 8.36183067348639e-06,
      "loss": 0.912,
      "step": 7111
    },
    {
      "epoch": 0.5747186811854785,
      "grad_norm": 2.6638970375061035,
      "learning_rate": 8.36134627357691e-06,
      "loss": 0.9576,
      "step": 7112
    },
    {
      "epoch": 0.574799490898806,
      "grad_norm": 2.5319557189941406,
      "learning_rate": 8.360861816095662e-06,
      "loss": 0.9856,
      "step": 7113
    },
    {
      "epoch": 0.5748803006121336,
      "grad_norm": 2.6411960124969482,
      "learning_rate": 8.36037730105094e-06,
      "loss": 0.8588,
      "step": 7114
    },
    {
      "epoch": 0.5749611103254612,
      "grad_norm": 2.6225197315216064,
      "learning_rate": 8.359892728451044e-06,
      "loss": 1.0094,
      "step": 7115
    },
    {
      "epoch": 0.5750419200387886,
      "grad_norm": 2.844329357147217,
      "learning_rate": 8.359408098304276e-06,
      "loss": 0.8998,
      "step": 7116
    },
    {
      "epoch": 0.5751227297521162,
      "grad_norm": 2.257103204727173,
      "learning_rate": 8.358923410618933e-06,
      "loss": 1.0439,
      "step": 7117
    },
    {
      "epoch": 0.5752035394654438,
      "grad_norm": 2.5816025733947754,
      "learning_rate": 8.358438665403318e-06,
      "loss": 0.9783,
      "step": 7118
    },
    {
      "epoch": 0.5752843491787712,
      "grad_norm": 2.5600202083587646,
      "learning_rate": 8.357953862665738e-06,
      "loss": 0.8909,
      "step": 7119
    },
    {
      "epoch": 0.5753651588920988,
      "grad_norm": 2.647162675857544,
      "learning_rate": 8.357469002414487e-06,
      "loss": 0.8291,
      "step": 7120
    },
    {
      "epoch": 0.5754459686054264,
      "grad_norm": 2.674480676651001,
      "learning_rate": 8.356984084657878e-06,
      "loss": 1.0018,
      "step": 7121
    },
    {
      "epoch": 0.5755267783187539,
      "grad_norm": 2.764030694961548,
      "learning_rate": 8.356499109404213e-06,
      "loss": 0.9144,
      "step": 7122
    },
    {
      "epoch": 0.5756075880320815,
      "grad_norm": 2.7310590744018555,
      "learning_rate": 8.356014076661797e-06,
      "loss": 0.9947,
      "step": 7123
    },
    {
      "epoch": 0.575688397745409,
      "grad_norm": 2.3526554107666016,
      "learning_rate": 8.35552898643894e-06,
      "loss": 0.9841,
      "step": 7124
    },
    {
      "epoch": 0.5757692074587365,
      "grad_norm": 2.3746774196624756,
      "learning_rate": 8.35504383874395e-06,
      "loss": 0.9476,
      "step": 7125
    },
    {
      "epoch": 0.5758500171720641,
      "grad_norm": 2.7105541229248047,
      "learning_rate": 8.354558633585135e-06,
      "loss": 0.9622,
      "step": 7126
    },
    {
      "epoch": 0.5759308268853917,
      "grad_norm": 2.2798879146575928,
      "learning_rate": 8.354073370970808e-06,
      "loss": 0.9252,
      "step": 7127
    },
    {
      "epoch": 0.5760116365987191,
      "grad_norm": 2.9079911708831787,
      "learning_rate": 8.353588050909278e-06,
      "loss": 0.9173,
      "step": 7128
    },
    {
      "epoch": 0.5760924463120467,
      "grad_norm": 2.9166505336761475,
      "learning_rate": 8.353102673408857e-06,
      "loss": 0.9159,
      "step": 7129
    },
    {
      "epoch": 0.5761732560253743,
      "grad_norm": 2.9057743549346924,
      "learning_rate": 8.35261723847786e-06,
      "loss": 0.9977,
      "step": 7130
    },
    {
      "epoch": 0.5762540657387017,
      "grad_norm": 2.689920425415039,
      "learning_rate": 8.352131746124602e-06,
      "loss": 0.9428,
      "step": 7131
    },
    {
      "epoch": 0.5763348754520293,
      "grad_norm": 2.1676008701324463,
      "learning_rate": 8.351646196357396e-06,
      "loss": 0.9955,
      "step": 7132
    },
    {
      "epoch": 0.5764156851653569,
      "grad_norm": 2.628889322280884,
      "learning_rate": 8.35116058918456e-06,
      "loss": 0.9724,
      "step": 7133
    },
    {
      "epoch": 0.5764964948786844,
      "grad_norm": 2.7892470359802246,
      "learning_rate": 8.350674924614411e-06,
      "loss": 0.8356,
      "step": 7134
    },
    {
      "epoch": 0.576577304592012,
      "grad_norm": 2.6595675945281982,
      "learning_rate": 8.350189202655265e-06,
      "loss": 0.8782,
      "step": 7135
    },
    {
      "epoch": 0.5766581143053395,
      "grad_norm": 3.222132682800293,
      "learning_rate": 8.349703423315446e-06,
      "loss": 1.0538,
      "step": 7136
    },
    {
      "epoch": 0.576738924018667,
      "grad_norm": 2.904111623764038,
      "learning_rate": 8.34921758660327e-06,
      "loss": 1.0349,
      "step": 7137
    },
    {
      "epoch": 0.5768197337319946,
      "grad_norm": 2.999847173690796,
      "learning_rate": 8.348731692527058e-06,
      "loss": 0.9379,
      "step": 7138
    },
    {
      "epoch": 0.5769005434453222,
      "grad_norm": 3.2096683979034424,
      "learning_rate": 8.348245741095139e-06,
      "loss": 0.9267,
      "step": 7139
    },
    {
      "epoch": 0.5769813531586496,
      "grad_norm": 2.9243977069854736,
      "learning_rate": 8.347759732315826e-06,
      "loss": 0.9898,
      "step": 7140
    },
    {
      "epoch": 0.5770621628719772,
      "grad_norm": 2.47684907913208,
      "learning_rate": 8.347273666197449e-06,
      "loss": 1.0357,
      "step": 7141
    },
    {
      "epoch": 0.5771429725853048,
      "grad_norm": 2.582578659057617,
      "learning_rate": 8.346787542748333e-06,
      "loss": 0.8993,
      "step": 7142
    },
    {
      "epoch": 0.5772237822986322,
      "grad_norm": 3.000941753387451,
      "learning_rate": 8.346301361976804e-06,
      "loss": 0.8562,
      "step": 7143
    },
    {
      "epoch": 0.5773045920119598,
      "grad_norm": 2.593186616897583,
      "learning_rate": 8.345815123891188e-06,
      "loss": 1.0636,
      "step": 7144
    },
    {
      "epoch": 0.5773854017252874,
      "grad_norm": 2.888679027557373,
      "learning_rate": 8.345328828499813e-06,
      "loss": 0.9278,
      "step": 7145
    },
    {
      "epoch": 0.5774662114386149,
      "grad_norm": 2.4658422470092773,
      "learning_rate": 8.34484247581101e-06,
      "loss": 0.9563,
      "step": 7146
    },
    {
      "epoch": 0.5775470211519425,
      "grad_norm": 3.0275471210479736,
      "learning_rate": 8.344356065833107e-06,
      "loss": 0.9294,
      "step": 7147
    },
    {
      "epoch": 0.57762783086527,
      "grad_norm": 2.5276520252227783,
      "learning_rate": 8.343869598574436e-06,
      "loss": 0.9542,
      "step": 7148
    },
    {
      "epoch": 0.5777086405785975,
      "grad_norm": 2.551970958709717,
      "learning_rate": 8.34338307404333e-06,
      "loss": 1.0405,
      "step": 7149
    },
    {
      "epoch": 0.5777894502919251,
      "grad_norm": 3.1526784896850586,
      "learning_rate": 8.34289649224812e-06,
      "loss": 0.9271,
      "step": 7150
    },
    {
      "epoch": 0.5778702600052527,
      "grad_norm": 2.6485044956207275,
      "learning_rate": 8.34240985319714e-06,
      "loss": 0.9898,
      "step": 7151
    },
    {
      "epoch": 0.5779510697185801,
      "grad_norm": 2.561781167984009,
      "learning_rate": 8.341923156898725e-06,
      "loss": 0.9021,
      "step": 7152
    },
    {
      "epoch": 0.5780318794319077,
      "grad_norm": 2.259307861328125,
      "learning_rate": 8.341436403361214e-06,
      "loss": 0.951,
      "step": 7153
    },
    {
      "epoch": 0.5781126891452353,
      "grad_norm": 2.5447819232940674,
      "learning_rate": 8.34094959259294e-06,
      "loss": 0.9541,
      "step": 7154
    },
    {
      "epoch": 0.5781934988585627,
      "grad_norm": 2.4141488075256348,
      "learning_rate": 8.340462724602243e-06,
      "loss": 0.9795,
      "step": 7155
    },
    {
      "epoch": 0.5782743085718903,
      "grad_norm": 3.415891647338867,
      "learning_rate": 8.339975799397462e-06,
      "loss": 1.0438,
      "step": 7156
    },
    {
      "epoch": 0.5783551182852179,
      "grad_norm": 2.540407657623291,
      "learning_rate": 8.339488816986934e-06,
      "loss": 1.0681,
      "step": 7157
    },
    {
      "epoch": 0.5784359279985454,
      "grad_norm": 2.658569812774658,
      "learning_rate": 8.339001777379004e-06,
      "loss": 0.9345,
      "step": 7158
    },
    {
      "epoch": 0.578516737711873,
      "grad_norm": 2.5978288650512695,
      "learning_rate": 8.338514680582011e-06,
      "loss": 0.9114,
      "step": 7159
    },
    {
      "epoch": 0.5785975474252005,
      "grad_norm": 2.3763232231140137,
      "learning_rate": 8.3380275266043e-06,
      "loss": 1.045,
      "step": 7160
    },
    {
      "epoch": 0.578678357138528,
      "grad_norm": 2.8415863513946533,
      "learning_rate": 8.337540315454213e-06,
      "loss": 0.9316,
      "step": 7161
    },
    {
      "epoch": 0.5787591668518556,
      "grad_norm": 2.7142231464385986,
      "learning_rate": 8.337053047140094e-06,
      "loss": 0.9099,
      "step": 7162
    },
    {
      "epoch": 0.5788399765651832,
      "grad_norm": 2.7461740970611572,
      "learning_rate": 8.33656572167029e-06,
      "loss": 1.0128,
      "step": 7163
    },
    {
      "epoch": 0.5789207862785106,
      "grad_norm": 3.922994375228882,
      "learning_rate": 8.33607833905315e-06,
      "loss": 1.022,
      "step": 7164
    },
    {
      "epoch": 0.5790015959918382,
      "grad_norm": 2.5308823585510254,
      "learning_rate": 8.335590899297018e-06,
      "loss": 0.9988,
      "step": 7165
    },
    {
      "epoch": 0.5790824057051658,
      "grad_norm": 2.4008872509002686,
      "learning_rate": 8.335103402410243e-06,
      "loss": 1.058,
      "step": 7166
    },
    {
      "epoch": 0.5791632154184932,
      "grad_norm": 2.4188337326049805,
      "learning_rate": 8.334615848401176e-06,
      "loss": 0.9581,
      "step": 7167
    },
    {
      "epoch": 0.5792440251318208,
      "grad_norm": 2.4827866554260254,
      "learning_rate": 8.334128237278168e-06,
      "loss": 0.924,
      "step": 7168
    },
    {
      "epoch": 0.5793248348451484,
      "grad_norm": 2.613064765930176,
      "learning_rate": 8.333640569049569e-06,
      "loss": 0.9715,
      "step": 7169
    },
    {
      "epoch": 0.5794056445584759,
      "grad_norm": 2.6098952293395996,
      "learning_rate": 8.333152843723732e-06,
      "loss": 0.9353,
      "step": 7170
    },
    {
      "epoch": 0.5794864542718035,
      "grad_norm": 3.0971457958221436,
      "learning_rate": 8.332665061309014e-06,
      "loss": 0.9005,
      "step": 7171
    },
    {
      "epoch": 0.579567263985131,
      "grad_norm": 2.992938756942749,
      "learning_rate": 8.332177221813765e-06,
      "loss": 1.087,
      "step": 7172
    },
    {
      "epoch": 0.5796480736984586,
      "grad_norm": 2.5264017581939697,
      "learning_rate": 8.331689325246339e-06,
      "loss": 0.9867,
      "step": 7173
    },
    {
      "epoch": 0.5797288834117861,
      "grad_norm": 2.314453363418579,
      "learning_rate": 8.3312013716151e-06,
      "loss": 0.9367,
      "step": 7174
    },
    {
      "epoch": 0.5798096931251137,
      "grad_norm": 2.806817054748535,
      "learning_rate": 8.330713360928398e-06,
      "loss": 0.9998,
      "step": 7175
    },
    {
      "epoch": 0.5798905028384412,
      "grad_norm": 2.609281063079834,
      "learning_rate": 8.330225293194595e-06,
      "loss": 1.1114,
      "step": 7176
    },
    {
      "epoch": 0.5799713125517687,
      "grad_norm": 2.8653433322906494,
      "learning_rate": 8.329737168422051e-06,
      "loss": 0.9069,
      "step": 7177
    },
    {
      "epoch": 0.5800521222650963,
      "grad_norm": 2.562836170196533,
      "learning_rate": 8.329248986619126e-06,
      "loss": 0.9759,
      "step": 7178
    },
    {
      "epoch": 0.5801329319784239,
      "grad_norm": 2.6038596630096436,
      "learning_rate": 8.328760747794179e-06,
      "loss": 0.842,
      "step": 7179
    },
    {
      "epoch": 0.5802137416917513,
      "grad_norm": 2.6815073490142822,
      "learning_rate": 8.328272451955574e-06,
      "loss": 0.9192,
      "step": 7180
    },
    {
      "epoch": 0.5802945514050789,
      "grad_norm": 2.736222267150879,
      "learning_rate": 8.327784099111676e-06,
      "loss": 1.022,
      "step": 7181
    },
    {
      "epoch": 0.5803753611184065,
      "grad_norm": 2.6508312225341797,
      "learning_rate": 8.327295689270847e-06,
      "loss": 1.0588,
      "step": 7182
    },
    {
      "epoch": 0.580456170831734,
      "grad_norm": 2.9490833282470703,
      "learning_rate": 8.326807222441454e-06,
      "loss": 0.8545,
      "step": 7183
    },
    {
      "epoch": 0.5805369805450615,
      "grad_norm": 2.8956761360168457,
      "learning_rate": 8.32631869863186e-06,
      "loss": 0.9728,
      "step": 7184
    },
    {
      "epoch": 0.5806177902583891,
      "grad_norm": 2.576817512512207,
      "learning_rate": 8.325830117850434e-06,
      "loss": 0.9309,
      "step": 7185
    },
    {
      "epoch": 0.5806985999717166,
      "grad_norm": 2.527831554412842,
      "learning_rate": 8.325341480105547e-06,
      "loss": 0.9836,
      "step": 7186
    },
    {
      "epoch": 0.5807794096850442,
      "grad_norm": 2.846911668777466,
      "learning_rate": 8.324852785405565e-06,
      "loss": 0.9205,
      "step": 7187
    },
    {
      "epoch": 0.5808602193983717,
      "grad_norm": 2.2964630126953125,
      "learning_rate": 8.32436403375886e-06,
      "loss": 1.0159,
      "step": 7188
    },
    {
      "epoch": 0.5809410291116992,
      "grad_norm": 2.5868728160858154,
      "learning_rate": 8.3238752251738e-06,
      "loss": 0.9503,
      "step": 7189
    },
    {
      "epoch": 0.5810218388250268,
      "grad_norm": 2.3481106758117676,
      "learning_rate": 8.32338635965876e-06,
      "loss": 0.8758,
      "step": 7190
    },
    {
      "epoch": 0.5811026485383544,
      "grad_norm": 3.0495853424072266,
      "learning_rate": 8.322897437222115e-06,
      "loss": 0.9098,
      "step": 7191
    },
    {
      "epoch": 0.5811834582516818,
      "grad_norm": 2.799420118331909,
      "learning_rate": 8.322408457872234e-06,
      "loss": 0.9523,
      "step": 7192
    },
    {
      "epoch": 0.5812642679650094,
      "grad_norm": 2.3985579013824463,
      "learning_rate": 8.321919421617495e-06,
      "loss": 0.9581,
      "step": 7193
    },
    {
      "epoch": 0.581345077678337,
      "grad_norm": 3.2880301475524902,
      "learning_rate": 8.321430328466273e-06,
      "loss": 0.9008,
      "step": 7194
    },
    {
      "epoch": 0.5814258873916645,
      "grad_norm": 2.5351295471191406,
      "learning_rate": 8.320941178426946e-06,
      "loss": 0.921,
      "step": 7195
    },
    {
      "epoch": 0.581506697104992,
      "grad_norm": 2.904498815536499,
      "learning_rate": 8.320451971507892e-06,
      "loss": 0.9153,
      "step": 7196
    },
    {
      "epoch": 0.5815875068183196,
      "grad_norm": 3.0620815753936768,
      "learning_rate": 8.319962707717489e-06,
      "loss": 1.0211,
      "step": 7197
    },
    {
      "epoch": 0.5816683165316471,
      "grad_norm": 2.6478400230407715,
      "learning_rate": 8.319473387064116e-06,
      "loss": 0.9977,
      "step": 7198
    },
    {
      "epoch": 0.5817491262449747,
      "grad_norm": 2.799684524536133,
      "learning_rate": 8.318984009556157e-06,
      "loss": 0.8999,
      "step": 7199
    },
    {
      "epoch": 0.5818299359583022,
      "grad_norm": 2.7371294498443604,
      "learning_rate": 8.31849457520199e-06,
      "loss": 0.9103,
      "step": 7200
    },
    {
      "epoch": 0.5819107456716297,
      "grad_norm": 2.4936740398406982,
      "learning_rate": 8.318005084010001e-06,
      "loss": 0.9903,
      "step": 7201
    },
    {
      "epoch": 0.5819915553849573,
      "grad_norm": 2.1372649669647217,
      "learning_rate": 8.317515535988574e-06,
      "loss": 0.9715,
      "step": 7202
    },
    {
      "epoch": 0.5820723650982849,
      "grad_norm": 2.432664155960083,
      "learning_rate": 8.31702593114609e-06,
      "loss": 1.106,
      "step": 7203
    },
    {
      "epoch": 0.5821531748116123,
      "grad_norm": 2.3609468936920166,
      "learning_rate": 8.31653626949094e-06,
      "loss": 1.1036,
      "step": 7204
    },
    {
      "epoch": 0.5822339845249399,
      "grad_norm": 2.9737699031829834,
      "learning_rate": 8.316046551031506e-06,
      "loss": 0.8853,
      "step": 7205
    },
    {
      "epoch": 0.5823147942382675,
      "grad_norm": 2.6651222705841064,
      "learning_rate": 8.315556775776179e-06,
      "loss": 0.9186,
      "step": 7206
    },
    {
      "epoch": 0.582395603951595,
      "grad_norm": 2.665828227996826,
      "learning_rate": 8.315066943733344e-06,
      "loss": 0.8584,
      "step": 7207
    },
    {
      "epoch": 0.5824764136649225,
      "grad_norm": 2.7233054637908936,
      "learning_rate": 8.314577054911395e-06,
      "loss": 0.987,
      "step": 7208
    },
    {
      "epoch": 0.5825572233782501,
      "grad_norm": 2.7811362743377686,
      "learning_rate": 8.31408710931872e-06,
      "loss": 1.026,
      "step": 7209
    },
    {
      "epoch": 0.5826380330915776,
      "grad_norm": 3.068962812423706,
      "learning_rate": 8.313597106963712e-06,
      "loss": 0.9771,
      "step": 7210
    },
    {
      "epoch": 0.5827188428049052,
      "grad_norm": 2.772899866104126,
      "learning_rate": 8.31310704785476e-06,
      "loss": 1.0116,
      "step": 7211
    },
    {
      "epoch": 0.5827996525182327,
      "grad_norm": 3.039334297180176,
      "learning_rate": 8.312616932000262e-06,
      "loss": 0.9946,
      "step": 7212
    },
    {
      "epoch": 0.5828804622315602,
      "grad_norm": 2.5728089809417725,
      "learning_rate": 8.312126759408613e-06,
      "loss": 0.9242,
      "step": 7213
    },
    {
      "epoch": 0.5829612719448878,
      "grad_norm": 2.7099645137786865,
      "learning_rate": 8.311636530088203e-06,
      "loss": 0.9262,
      "step": 7214
    },
    {
      "epoch": 0.5830420816582154,
      "grad_norm": 2.4517130851745605,
      "learning_rate": 8.311146244047433e-06,
      "loss": 0.9646,
      "step": 7215
    },
    {
      "epoch": 0.5831228913715428,
      "grad_norm": 2.8632915019989014,
      "learning_rate": 8.310655901294698e-06,
      "loss": 0.9943,
      "step": 7216
    },
    {
      "epoch": 0.5832037010848704,
      "grad_norm": 2.633495569229126,
      "learning_rate": 8.310165501838398e-06,
      "loss": 1.0139,
      "step": 7217
    },
    {
      "epoch": 0.583284510798198,
      "grad_norm": 2.336775302886963,
      "learning_rate": 8.309675045686932e-06,
      "loss": 0.937,
      "step": 7218
    },
    {
      "epoch": 0.5833653205115255,
      "grad_norm": 2.640777826309204,
      "learning_rate": 8.3091845328487e-06,
      "loss": 0.9503,
      "step": 7219
    },
    {
      "epoch": 0.583446130224853,
      "grad_norm": 2.6599156856536865,
      "learning_rate": 8.308693963332104e-06,
      "loss": 1.0876,
      "step": 7220
    },
    {
      "epoch": 0.5835269399381806,
      "grad_norm": 2.600266933441162,
      "learning_rate": 8.308203337145547e-06,
      "loss": 0.897,
      "step": 7221
    },
    {
      "epoch": 0.5836077496515081,
      "grad_norm": 2.3513786792755127,
      "learning_rate": 8.307712654297428e-06,
      "loss": 0.934,
      "step": 7222
    },
    {
      "epoch": 0.5836885593648357,
      "grad_norm": 2.633467674255371,
      "learning_rate": 8.307221914796155e-06,
      "loss": 0.9819,
      "step": 7223
    },
    {
      "epoch": 0.5837693690781632,
      "grad_norm": 2.2107534408569336,
      "learning_rate": 8.306731118650135e-06,
      "loss": 1.0194,
      "step": 7224
    },
    {
      "epoch": 0.5838501787914907,
      "grad_norm": 2.882359504699707,
      "learning_rate": 8.306240265867768e-06,
      "loss": 0.9784,
      "step": 7225
    },
    {
      "epoch": 0.5839309885048183,
      "grad_norm": 2.542127847671509,
      "learning_rate": 8.305749356457468e-06,
      "loss": 0.8831,
      "step": 7226
    },
    {
      "epoch": 0.5840117982181459,
      "grad_norm": 2.408803939819336,
      "learning_rate": 8.305258390427638e-06,
      "loss": 0.9947,
      "step": 7227
    },
    {
      "epoch": 0.5840926079314733,
      "grad_norm": 2.930621385574341,
      "learning_rate": 8.30476736778669e-06,
      "loss": 0.9591,
      "step": 7228
    },
    {
      "epoch": 0.5841734176448009,
      "grad_norm": 3.039104461669922,
      "learning_rate": 8.304276288543031e-06,
      "loss": 0.9962,
      "step": 7229
    },
    {
      "epoch": 0.5842542273581285,
      "grad_norm": 2.72379469871521,
      "learning_rate": 8.303785152705076e-06,
      "loss": 0.8664,
      "step": 7230
    },
    {
      "epoch": 0.584335037071456,
      "grad_norm": 2.2838335037231445,
      "learning_rate": 8.303293960281233e-06,
      "loss": 1.021,
      "step": 7231
    },
    {
      "epoch": 0.5844158467847835,
      "grad_norm": 2.7396748065948486,
      "learning_rate": 8.302802711279917e-06,
      "loss": 0.9151,
      "step": 7232
    },
    {
      "epoch": 0.5844966564981111,
      "grad_norm": 2.399461269378662,
      "learning_rate": 8.302311405709542e-06,
      "loss": 0.9,
      "step": 7233
    },
    {
      "epoch": 0.5845774662114386,
      "grad_norm": 2.778099775314331,
      "learning_rate": 8.301820043578524e-06,
      "loss": 1.0041,
      "step": 7234
    },
    {
      "epoch": 0.5846582759247662,
      "grad_norm": 2.4411673545837402,
      "learning_rate": 8.301328624895277e-06,
      "loss": 1.0502,
      "step": 7235
    },
    {
      "epoch": 0.5847390856380937,
      "grad_norm": 3.5182604789733887,
      "learning_rate": 8.300837149668218e-06,
      "loss": 0.9082,
      "step": 7236
    },
    {
      "epoch": 0.5848198953514212,
      "grad_norm": 2.8782472610473633,
      "learning_rate": 8.300345617905763e-06,
      "loss": 0.9671,
      "step": 7237
    },
    {
      "epoch": 0.5849007050647488,
      "grad_norm": 2.3218936920166016,
      "learning_rate": 8.299854029616335e-06,
      "loss": 0.8876,
      "step": 7238
    },
    {
      "epoch": 0.5849815147780764,
      "grad_norm": 3.071110963821411,
      "learning_rate": 8.299362384808352e-06,
      "loss": 0.8901,
      "step": 7239
    },
    {
      "epoch": 0.5850623244914038,
      "grad_norm": 3.2636032104492188,
      "learning_rate": 8.298870683490232e-06,
      "loss": 1.0622,
      "step": 7240
    },
    {
      "epoch": 0.5851431342047314,
      "grad_norm": 2.4568161964416504,
      "learning_rate": 8.2983789256704e-06,
      "loss": 0.9507,
      "step": 7241
    },
    {
      "epoch": 0.585223943918059,
      "grad_norm": 2.903306722640991,
      "learning_rate": 8.297887111357279e-06,
      "loss": 0.9607,
      "step": 7242
    },
    {
      "epoch": 0.5853047536313865,
      "grad_norm": 2.5162227153778076,
      "learning_rate": 8.297395240559289e-06,
      "loss": 0.9627,
      "step": 7243
    },
    {
      "epoch": 0.585385563344714,
      "grad_norm": 2.6980717182159424,
      "learning_rate": 8.29690331328486e-06,
      "loss": 1.0173,
      "step": 7244
    },
    {
      "epoch": 0.5854663730580416,
      "grad_norm": 2.7115542888641357,
      "learning_rate": 8.29641132954241e-06,
      "loss": 0.9689,
      "step": 7245
    },
    {
      "epoch": 0.5855471827713691,
      "grad_norm": 2.9262375831604004,
      "learning_rate": 8.295919289340371e-06,
      "loss": 0.9344,
      "step": 7246
    },
    {
      "epoch": 0.5856279924846967,
      "grad_norm": 2.454115152359009,
      "learning_rate": 8.29542719268717e-06,
      "loss": 1.0366,
      "step": 7247
    },
    {
      "epoch": 0.5857088021980242,
      "grad_norm": 2.3883273601531982,
      "learning_rate": 8.294935039591235e-06,
      "loss": 0.9147,
      "step": 7248
    },
    {
      "epoch": 0.5857896119113517,
      "grad_norm": 2.728160858154297,
      "learning_rate": 8.294442830060993e-06,
      "loss": 0.9353,
      "step": 7249
    },
    {
      "epoch": 0.5858704216246793,
      "grad_norm": 2.707059621810913,
      "learning_rate": 8.293950564104878e-06,
      "loss": 0.9389,
      "step": 7250
    },
    {
      "epoch": 0.5859512313380069,
      "grad_norm": 2.561579942703247,
      "learning_rate": 8.293458241731319e-06,
      "loss": 0.9523,
      "step": 7251
    },
    {
      "epoch": 0.5860320410513343,
      "grad_norm": 2.8697996139526367,
      "learning_rate": 8.29296586294875e-06,
      "loss": 0.8341,
      "step": 7252
    },
    {
      "epoch": 0.5861128507646619,
      "grad_norm": 3.0197203159332275,
      "learning_rate": 8.292473427765603e-06,
      "loss": 0.9398,
      "step": 7253
    },
    {
      "epoch": 0.5861936604779895,
      "grad_norm": 2.7292654514312744,
      "learning_rate": 8.291980936190312e-06,
      "loss": 0.8719,
      "step": 7254
    },
    {
      "epoch": 0.586274470191317,
      "grad_norm": 3.2213222980499268,
      "learning_rate": 8.291488388231313e-06,
      "loss": 0.9678,
      "step": 7255
    },
    {
      "epoch": 0.5863552799046445,
      "grad_norm": 2.80500864982605,
      "learning_rate": 8.290995783897041e-06,
      "loss": 0.8987,
      "step": 7256
    },
    {
      "epoch": 0.5864360896179721,
      "grad_norm": 2.932123899459839,
      "learning_rate": 8.290503123195934e-06,
      "loss": 0.9552,
      "step": 7257
    },
    {
      "epoch": 0.5865168993312996,
      "grad_norm": 2.6169843673706055,
      "learning_rate": 8.290010406136433e-06,
      "loss": 0.9654,
      "step": 7258
    },
    {
      "epoch": 0.5865977090446272,
      "grad_norm": 2.683082103729248,
      "learning_rate": 8.289517632726972e-06,
      "loss": 0.942,
      "step": 7259
    },
    {
      "epoch": 0.5866785187579547,
      "grad_norm": 2.525327444076538,
      "learning_rate": 8.289024802975991e-06,
      "loss": 0.8518,
      "step": 7260
    },
    {
      "epoch": 0.5867593284712822,
      "grad_norm": 2.8130624294281006,
      "learning_rate": 8.288531916891936e-06,
      "loss": 0.9454,
      "step": 7261
    },
    {
      "epoch": 0.5868401381846098,
      "grad_norm": 2.7846462726593018,
      "learning_rate": 8.288038974483244e-06,
      "loss": 0.87,
      "step": 7262
    },
    {
      "epoch": 0.5869209478979374,
      "grad_norm": 2.4028542041778564,
      "learning_rate": 8.287545975758362e-06,
      "loss": 0.8436,
      "step": 7263
    },
    {
      "epoch": 0.5870017576112648,
      "grad_norm": 2.866311550140381,
      "learning_rate": 8.287052920725731e-06,
      "loss": 1.0084,
      "step": 7264
    },
    {
      "epoch": 0.5870825673245924,
      "grad_norm": 2.5690367221832275,
      "learning_rate": 8.286559809393796e-06,
      "loss": 0.9965,
      "step": 7265
    },
    {
      "epoch": 0.58716337703792,
      "grad_norm": 2.5451879501342773,
      "learning_rate": 8.286066641771005e-06,
      "loss": 0.9242,
      "step": 7266
    },
    {
      "epoch": 0.5872441867512475,
      "grad_norm": 2.705350399017334,
      "learning_rate": 8.285573417865802e-06,
      "loss": 0.8542,
      "step": 7267
    },
    {
      "epoch": 0.587324996464575,
      "grad_norm": 2.1438097953796387,
      "learning_rate": 8.285080137686637e-06,
      "loss": 1.1101,
      "step": 7268
    },
    {
      "epoch": 0.5874058061779026,
      "grad_norm": 2.670281410217285,
      "learning_rate": 8.284586801241957e-06,
      "loss": 0.8841,
      "step": 7269
    },
    {
      "epoch": 0.5874866158912301,
      "grad_norm": 2.326840877532959,
      "learning_rate": 8.284093408540213e-06,
      "loss": 0.9651,
      "step": 7270
    },
    {
      "epoch": 0.5875674256045577,
      "grad_norm": 2.614589214324951,
      "learning_rate": 8.283599959589854e-06,
      "loss": 0.8955,
      "step": 7271
    },
    {
      "epoch": 0.5876482353178852,
      "grad_norm": 2.9460232257843018,
      "learning_rate": 8.283106454399334e-06,
      "loss": 1.1043,
      "step": 7272
    },
    {
      "epoch": 0.5877290450312127,
      "grad_norm": 2.689347982406616,
      "learning_rate": 8.282612892977104e-06,
      "loss": 0.9888,
      "step": 7273
    },
    {
      "epoch": 0.5878098547445403,
      "grad_norm": 2.7261712551116943,
      "learning_rate": 8.282119275331617e-06,
      "loss": 1.0793,
      "step": 7274
    },
    {
      "epoch": 0.5878906644578679,
      "grad_norm": 2.2428059577941895,
      "learning_rate": 8.281625601471329e-06,
      "loss": 0.8454,
      "step": 7275
    },
    {
      "epoch": 0.5879714741711953,
      "grad_norm": 2.658480644226074,
      "learning_rate": 8.281131871404693e-06,
      "loss": 0.9489,
      "step": 7276
    },
    {
      "epoch": 0.5880522838845229,
      "grad_norm": 2.676419496536255,
      "learning_rate": 8.28063808514017e-06,
      "loss": 0.9293,
      "step": 7277
    },
    {
      "epoch": 0.5881330935978505,
      "grad_norm": 2.534071683883667,
      "learning_rate": 8.280144242686213e-06,
      "loss": 0.9243,
      "step": 7278
    },
    {
      "epoch": 0.588213903311178,
      "grad_norm": 2.4483423233032227,
      "learning_rate": 8.27965034405128e-06,
      "loss": 0.9228,
      "step": 7279
    },
    {
      "epoch": 0.5882947130245055,
      "grad_norm": 2.6890275478363037,
      "learning_rate": 8.279156389243835e-06,
      "loss": 0.9201,
      "step": 7280
    },
    {
      "epoch": 0.5883755227378331,
      "grad_norm": 2.8360040187835693,
      "learning_rate": 8.278662378272333e-06,
      "loss": 0.9064,
      "step": 7281
    },
    {
      "epoch": 0.5884563324511606,
      "grad_norm": 3.0665738582611084,
      "learning_rate": 8.278168311145238e-06,
      "loss": 0.8436,
      "step": 7282
    },
    {
      "epoch": 0.5885371421644882,
      "grad_norm": 2.62737774848938,
      "learning_rate": 8.277674187871012e-06,
      "loss": 0.8682,
      "step": 7283
    },
    {
      "epoch": 0.5886179518778157,
      "grad_norm": 3.6039865016937256,
      "learning_rate": 8.277180008458118e-06,
      "loss": 0.9289,
      "step": 7284
    },
    {
      "epoch": 0.5886987615911432,
      "grad_norm": 3.059206962585449,
      "learning_rate": 8.27668577291502e-06,
      "loss": 0.9672,
      "step": 7285
    },
    {
      "epoch": 0.5887795713044708,
      "grad_norm": 2.4514665603637695,
      "learning_rate": 8.276191481250183e-06,
      "loss": 0.9134,
      "step": 7286
    },
    {
      "epoch": 0.5888603810177984,
      "grad_norm": 2.8695521354675293,
      "learning_rate": 8.275697133472073e-06,
      "loss": 0.914,
      "step": 7287
    },
    {
      "epoch": 0.5889411907311258,
      "grad_norm": 2.4840245246887207,
      "learning_rate": 8.275202729589156e-06,
      "loss": 0.9021,
      "step": 7288
    },
    {
      "epoch": 0.5890220004444534,
      "grad_norm": 2.4578912258148193,
      "learning_rate": 8.274708269609902e-06,
      "loss": 0.972,
      "step": 7289
    },
    {
      "epoch": 0.589102810157781,
      "grad_norm": 2.822356939315796,
      "learning_rate": 8.274213753542778e-06,
      "loss": 0.9244,
      "step": 7290
    },
    {
      "epoch": 0.5891836198711085,
      "grad_norm": 2.9471700191497803,
      "learning_rate": 8.273719181396257e-06,
      "loss": 0.9063,
      "step": 7291
    },
    {
      "epoch": 0.589264429584436,
      "grad_norm": 2.3739912509918213,
      "learning_rate": 8.273224553178806e-06,
      "loss": 0.8829,
      "step": 7292
    },
    {
      "epoch": 0.5893452392977636,
      "grad_norm": 2.6234970092773438,
      "learning_rate": 8.272729868898897e-06,
      "loss": 1.0808,
      "step": 7293
    },
    {
      "epoch": 0.5894260490110911,
      "grad_norm": 3.0504112243652344,
      "learning_rate": 8.272235128565006e-06,
      "loss": 1.044,
      "step": 7294
    },
    {
      "epoch": 0.5895068587244187,
      "grad_norm": 3.024522542953491,
      "learning_rate": 8.271740332185605e-06,
      "loss": 0.9667,
      "step": 7295
    },
    {
      "epoch": 0.5895876684377462,
      "grad_norm": 3.0582995414733887,
      "learning_rate": 8.271245479769168e-06,
      "loss": 0.9668,
      "step": 7296
    },
    {
      "epoch": 0.5896684781510737,
      "grad_norm": 2.4406895637512207,
      "learning_rate": 8.27075057132417e-06,
      "loss": 0.8866,
      "step": 7297
    },
    {
      "epoch": 0.5897492878644013,
      "grad_norm": 2.813310384750366,
      "learning_rate": 8.27025560685909e-06,
      "loss": 0.8483,
      "step": 7298
    },
    {
      "epoch": 0.5898300975777289,
      "grad_norm": 2.3181216716766357,
      "learning_rate": 8.269760586382404e-06,
      "loss": 0.9512,
      "step": 7299
    },
    {
      "epoch": 0.5899109072910563,
      "grad_norm": 2.4975626468658447,
      "learning_rate": 8.26926550990259e-06,
      "loss": 1.0061,
      "step": 7300
    },
    {
      "epoch": 0.5899917170043839,
      "grad_norm": 2.848518133163452,
      "learning_rate": 8.268770377428131e-06,
      "loss": 1.0314,
      "step": 7301
    },
    {
      "epoch": 0.5900725267177115,
      "grad_norm": 2.6034278869628906,
      "learning_rate": 8.268275188967503e-06,
      "loss": 0.8208,
      "step": 7302
    },
    {
      "epoch": 0.5901533364310391,
      "grad_norm": 2.475684642791748,
      "learning_rate": 8.267779944529187e-06,
      "loss": 1.0302,
      "step": 7303
    },
    {
      "epoch": 0.5902341461443665,
      "grad_norm": 2.6110689640045166,
      "learning_rate": 8.267284644121669e-06,
      "loss": 0.915,
      "step": 7304
    },
    {
      "epoch": 0.5903149558576941,
      "grad_norm": 2.9698400497436523,
      "learning_rate": 8.266789287753432e-06,
      "loss": 0.8153,
      "step": 7305
    },
    {
      "epoch": 0.5903957655710217,
      "grad_norm": 2.4260551929473877,
      "learning_rate": 8.266293875432957e-06,
      "loss": 0.974,
      "step": 7306
    },
    {
      "epoch": 0.5904765752843492,
      "grad_norm": 2.4733686447143555,
      "learning_rate": 8.265798407168732e-06,
      "loss": 0.9055,
      "step": 7307
    },
    {
      "epoch": 0.5905573849976767,
      "grad_norm": 2.69268536567688,
      "learning_rate": 8.265302882969242e-06,
      "loss": 0.8344,
      "step": 7308
    },
    {
      "epoch": 0.5906381947110043,
      "grad_norm": 3.423067092895508,
      "learning_rate": 8.264807302842976e-06,
      "loss": 0.9971,
      "step": 7309
    },
    {
      "epoch": 0.5907190044243318,
      "grad_norm": 3.1059982776641846,
      "learning_rate": 8.264311666798419e-06,
      "loss": 0.9854,
      "step": 7310
    },
    {
      "epoch": 0.5907998141376594,
      "grad_norm": 2.9936695098876953,
      "learning_rate": 8.263815974844063e-06,
      "loss": 0.9392,
      "step": 7311
    },
    {
      "epoch": 0.5908806238509869,
      "grad_norm": 2.4910175800323486,
      "learning_rate": 8.263320226988395e-06,
      "loss": 0.9391,
      "step": 7312
    },
    {
      "epoch": 0.5909614335643144,
      "grad_norm": 2.7224578857421875,
      "learning_rate": 8.262824423239908e-06,
      "loss": 0.9075,
      "step": 7313
    },
    {
      "epoch": 0.591042243277642,
      "grad_norm": 2.775599718093872,
      "learning_rate": 8.262328563607094e-06,
      "loss": 1.0178,
      "step": 7314
    },
    {
      "epoch": 0.5911230529909696,
      "grad_norm": 2.9733424186706543,
      "learning_rate": 8.261832648098447e-06,
      "loss": 1.1278,
      "step": 7315
    },
    {
      "epoch": 0.591203862704297,
      "grad_norm": 2.845141649246216,
      "learning_rate": 8.261336676722457e-06,
      "loss": 0.9019,
      "step": 7316
    },
    {
      "epoch": 0.5912846724176246,
      "grad_norm": 2.9141879081726074,
      "learning_rate": 8.260840649487622e-06,
      "loss": 0.9286,
      "step": 7317
    },
    {
      "epoch": 0.5913654821309522,
      "grad_norm": 2.4800522327423096,
      "learning_rate": 8.260344566402436e-06,
      "loss": 0.9283,
      "step": 7318
    },
    {
      "epoch": 0.5914462918442797,
      "grad_norm": 3.3712503910064697,
      "learning_rate": 8.259848427475397e-06,
      "loss": 0.9504,
      "step": 7319
    },
    {
      "epoch": 0.5915271015576072,
      "grad_norm": 2.650286912918091,
      "learning_rate": 8.259352232715004e-06,
      "loss": 1.0402,
      "step": 7320
    },
    {
      "epoch": 0.5916079112709348,
      "grad_norm": 2.863093852996826,
      "learning_rate": 8.25885598212975e-06,
      "loss": 0.8921,
      "step": 7321
    },
    {
      "epoch": 0.5916887209842623,
      "grad_norm": 2.793614149093628,
      "learning_rate": 8.258359675728143e-06,
      "loss": 0.8471,
      "step": 7322
    },
    {
      "epoch": 0.5917695306975899,
      "grad_norm": 2.6360037326812744,
      "learning_rate": 8.257863313518676e-06,
      "loss": 0.9054,
      "step": 7323
    },
    {
      "epoch": 0.5918503404109174,
      "grad_norm": 2.5683095455169678,
      "learning_rate": 8.257366895509853e-06,
      "loss": 1.0138,
      "step": 7324
    },
    {
      "epoch": 0.5919311501242449,
      "grad_norm": 2.5132172107696533,
      "learning_rate": 8.25687042171018e-06,
      "loss": 0.9629,
      "step": 7325
    },
    {
      "epoch": 0.5920119598375725,
      "grad_norm": 3.5530898571014404,
      "learning_rate": 8.256373892128154e-06,
      "loss": 0.9259,
      "step": 7326
    },
    {
      "epoch": 0.5920927695509001,
      "grad_norm": 2.6273193359375,
      "learning_rate": 8.255877306772283e-06,
      "loss": 0.867,
      "step": 7327
    },
    {
      "epoch": 0.5921735792642275,
      "grad_norm": 3.0033140182495117,
      "learning_rate": 8.255380665651073e-06,
      "loss": 0.9118,
      "step": 7328
    },
    {
      "epoch": 0.5922543889775551,
      "grad_norm": 2.8479557037353516,
      "learning_rate": 8.254883968773028e-06,
      "loss": 0.9061,
      "step": 7329
    },
    {
      "epoch": 0.5923351986908827,
      "grad_norm": 2.35652756690979,
      "learning_rate": 8.254387216146658e-06,
      "loss": 1.0341,
      "step": 7330
    },
    {
      "epoch": 0.5924160084042102,
      "grad_norm": 2.7019267082214355,
      "learning_rate": 8.25389040778047e-06,
      "loss": 1.0876,
      "step": 7331
    },
    {
      "epoch": 0.5924968181175377,
      "grad_norm": 2.4694161415100098,
      "learning_rate": 8.25339354368297e-06,
      "loss": 1.1297,
      "step": 7332
    },
    {
      "epoch": 0.5925776278308653,
      "grad_norm": 2.877805471420288,
      "learning_rate": 8.252896623862674e-06,
      "loss": 0.9301,
      "step": 7333
    },
    {
      "epoch": 0.5926584375441928,
      "grad_norm": 2.8389344215393066,
      "learning_rate": 8.25239964832809e-06,
      "loss": 0.8912,
      "step": 7334
    },
    {
      "epoch": 0.5927392472575204,
      "grad_norm": 2.62739896774292,
      "learning_rate": 8.251902617087726e-06,
      "loss": 0.9648,
      "step": 7335
    },
    {
      "epoch": 0.5928200569708479,
      "grad_norm": 2.79373836517334,
      "learning_rate": 8.251405530150101e-06,
      "loss": 0.8026,
      "step": 7336
    },
    {
      "epoch": 0.5929008666841754,
      "grad_norm": 2.92045259475708,
      "learning_rate": 8.250908387523727e-06,
      "loss": 0.9487,
      "step": 7337
    },
    {
      "epoch": 0.592981676397503,
      "grad_norm": 2.3661844730377197,
      "learning_rate": 8.250411189217118e-06,
      "loss": 0.9737,
      "step": 7338
    },
    {
      "epoch": 0.5930624861108306,
      "grad_norm": 2.865668773651123,
      "learning_rate": 8.249913935238792e-06,
      "loss": 1.0048,
      "step": 7339
    },
    {
      "epoch": 0.593143295824158,
      "grad_norm": 2.6446855068206787,
      "learning_rate": 8.249416625597262e-06,
      "loss": 1.0937,
      "step": 7340
    },
    {
      "epoch": 0.5932241055374856,
      "grad_norm": 2.608536720275879,
      "learning_rate": 8.248919260301048e-06,
      "loss": 0.9387,
      "step": 7341
    },
    {
      "epoch": 0.5933049152508132,
      "grad_norm": 2.9983580112457275,
      "learning_rate": 8.248421839358669e-06,
      "loss": 0.8629,
      "step": 7342
    },
    {
      "epoch": 0.5933857249641407,
      "grad_norm": 2.7614946365356445,
      "learning_rate": 8.247924362778645e-06,
      "loss": 1.0247,
      "step": 7343
    },
    {
      "epoch": 0.5934665346774682,
      "grad_norm": 2.579474449157715,
      "learning_rate": 8.247426830569494e-06,
      "loss": 0.9084,
      "step": 7344
    },
    {
      "epoch": 0.5935473443907958,
      "grad_norm": 2.6016366481781006,
      "learning_rate": 8.24692924273974e-06,
      "loss": 1.019,
      "step": 7345
    },
    {
      "epoch": 0.5936281541041233,
      "grad_norm": 2.929500102996826,
      "learning_rate": 8.246431599297905e-06,
      "loss": 1.1036,
      "step": 7346
    },
    {
      "epoch": 0.5937089638174509,
      "grad_norm": 2.6554834842681885,
      "learning_rate": 8.245933900252514e-06,
      "loss": 0.9863,
      "step": 7347
    },
    {
      "epoch": 0.5937897735307784,
      "grad_norm": 3.1263253688812256,
      "learning_rate": 8.245436145612088e-06,
      "loss": 0.927,
      "step": 7348
    },
    {
      "epoch": 0.5938705832441059,
      "grad_norm": 2.7033302783966064,
      "learning_rate": 8.244938335385154e-06,
      "loss": 1.0174,
      "step": 7349
    },
    {
      "epoch": 0.5939513929574335,
      "grad_norm": 3.122936964035034,
      "learning_rate": 8.244440469580237e-06,
      "loss": 0.9281,
      "step": 7350
    },
    {
      "epoch": 0.5940322026707611,
      "grad_norm": 2.663856029510498,
      "learning_rate": 8.243942548205867e-06,
      "loss": 1.0489,
      "step": 7351
    },
    {
      "epoch": 0.5941130123840885,
      "grad_norm": 2.829989194869995,
      "learning_rate": 8.243444571270568e-06,
      "loss": 0.9527,
      "step": 7352
    },
    {
      "epoch": 0.5941938220974161,
      "grad_norm": 2.4378714561462402,
      "learning_rate": 8.242946538782875e-06,
      "loss": 0.8978,
      "step": 7353
    },
    {
      "epoch": 0.5942746318107437,
      "grad_norm": 2.8279662132263184,
      "learning_rate": 8.242448450751314e-06,
      "loss": 1.0525,
      "step": 7354
    },
    {
      "epoch": 0.5943554415240712,
      "grad_norm": 3.1409215927124023,
      "learning_rate": 8.241950307184416e-06,
      "loss": 0.982,
      "step": 7355
    },
    {
      "epoch": 0.5944362512373987,
      "grad_norm": 2.909543991088867,
      "learning_rate": 8.241452108090716e-06,
      "loss": 0.9226,
      "step": 7356
    },
    {
      "epoch": 0.5945170609507263,
      "grad_norm": 2.2005367279052734,
      "learning_rate": 8.240953853478742e-06,
      "loss": 0.9917,
      "step": 7357
    },
    {
      "epoch": 0.5945978706640538,
      "grad_norm": 2.7676758766174316,
      "learning_rate": 8.240455543357031e-06,
      "loss": 1.03,
      "step": 7358
    },
    {
      "epoch": 0.5946786803773814,
      "grad_norm": 2.4235360622406006,
      "learning_rate": 8.239957177734119e-06,
      "loss": 1.0517,
      "step": 7359
    },
    {
      "epoch": 0.5947594900907089,
      "grad_norm": 2.5609865188598633,
      "learning_rate": 8.23945875661854e-06,
      "loss": 0.9424,
      "step": 7360
    },
    {
      "epoch": 0.5948402998040364,
      "grad_norm": 2.680320978164673,
      "learning_rate": 8.238960280018832e-06,
      "loss": 0.9987,
      "step": 7361
    },
    {
      "epoch": 0.594921109517364,
      "grad_norm": 2.6540329456329346,
      "learning_rate": 8.23846174794353e-06,
      "loss": 0.959,
      "step": 7362
    },
    {
      "epoch": 0.5950019192306916,
      "grad_norm": 2.4815590381622314,
      "learning_rate": 8.237963160401176e-06,
      "loss": 0.9409,
      "step": 7363
    },
    {
      "epoch": 0.595082728944019,
      "grad_norm": 2.2560675144195557,
      "learning_rate": 8.237464517400308e-06,
      "loss": 0.9085,
      "step": 7364
    },
    {
      "epoch": 0.5951635386573466,
      "grad_norm": 2.394665479660034,
      "learning_rate": 8.236965818949467e-06,
      "loss": 0.9635,
      "step": 7365
    },
    {
      "epoch": 0.5952443483706742,
      "grad_norm": 2.595749855041504,
      "learning_rate": 8.236467065057193e-06,
      "loss": 0.9486,
      "step": 7366
    },
    {
      "epoch": 0.5953251580840017,
      "grad_norm": 2.5327694416046143,
      "learning_rate": 8.23596825573203e-06,
      "loss": 0.8764,
      "step": 7367
    },
    {
      "epoch": 0.5954059677973292,
      "grad_norm": 3.028306722640991,
      "learning_rate": 8.235469390982522e-06,
      "loss": 1.0335,
      "step": 7368
    },
    {
      "epoch": 0.5954867775106568,
      "grad_norm": 2.623086452484131,
      "learning_rate": 8.234970470817212e-06,
      "loss": 1.0315,
      "step": 7369
    },
    {
      "epoch": 0.5955675872239843,
      "grad_norm": 2.815917491912842,
      "learning_rate": 8.234471495244644e-06,
      "loss": 0.9318,
      "step": 7370
    },
    {
      "epoch": 0.5956483969373119,
      "grad_norm": 2.6206581592559814,
      "learning_rate": 8.23397246427337e-06,
      "loss": 0.9279,
      "step": 7371
    },
    {
      "epoch": 0.5957292066506394,
      "grad_norm": 2.9340102672576904,
      "learning_rate": 8.23347337791193e-06,
      "loss": 1.0592,
      "step": 7372
    },
    {
      "epoch": 0.5958100163639669,
      "grad_norm": 2.8901724815368652,
      "learning_rate": 8.232974236168875e-06,
      "loss": 1.0235,
      "step": 7373
    },
    {
      "epoch": 0.5958908260772945,
      "grad_norm": 2.838430166244507,
      "learning_rate": 8.232475039052755e-06,
      "loss": 1.0582,
      "step": 7374
    },
    {
      "epoch": 0.5959716357906221,
      "grad_norm": 2.7173655033111572,
      "learning_rate": 8.23197578657212e-06,
      "loss": 0.8986,
      "step": 7375
    },
    {
      "epoch": 0.5960524455039495,
      "grad_norm": 2.1216917037963867,
      "learning_rate": 8.23147647873552e-06,
      "loss": 1.0306,
      "step": 7376
    },
    {
      "epoch": 0.5961332552172771,
      "grad_norm": 3.172276735305786,
      "learning_rate": 8.230977115551508e-06,
      "loss": 0.9584,
      "step": 7377
    },
    {
      "epoch": 0.5962140649306047,
      "grad_norm": 2.7974860668182373,
      "learning_rate": 8.230477697028636e-06,
      "loss": 0.9677,
      "step": 7378
    },
    {
      "epoch": 0.5962948746439322,
      "grad_norm": 2.6010758876800537,
      "learning_rate": 8.229978223175459e-06,
      "loss": 1.0288,
      "step": 7379
    },
    {
      "epoch": 0.5963756843572597,
      "grad_norm": 3.3252294063568115,
      "learning_rate": 8.229478694000527e-06,
      "loss": 0.9599,
      "step": 7380
    },
    {
      "epoch": 0.5964564940705873,
      "grad_norm": 3.0079755783081055,
      "learning_rate": 8.228979109512405e-06,
      "loss": 0.8889,
      "step": 7381
    },
    {
      "epoch": 0.5965373037839148,
      "grad_norm": 3.624636650085449,
      "learning_rate": 8.228479469719641e-06,
      "loss": 1.0804,
      "step": 7382
    },
    {
      "epoch": 0.5966181134972424,
      "grad_norm": 2.2371068000793457,
      "learning_rate": 8.227979774630796e-06,
      "loss": 0.9555,
      "step": 7383
    },
    {
      "epoch": 0.5966989232105699,
      "grad_norm": 2.666361093521118,
      "learning_rate": 8.22748002425443e-06,
      "loss": 0.9783,
      "step": 7384
    },
    {
      "epoch": 0.5967797329238974,
      "grad_norm": 2.5665626525878906,
      "learning_rate": 8.2269802185991e-06,
      "loss": 0.9119,
      "step": 7385
    },
    {
      "epoch": 0.596860542637225,
      "grad_norm": 2.503783702850342,
      "learning_rate": 8.226480357673367e-06,
      "loss": 0.8624,
      "step": 7386
    },
    {
      "epoch": 0.5969413523505526,
      "grad_norm": 2.597759962081909,
      "learning_rate": 8.225980441485794e-06,
      "loss": 1.1148,
      "step": 7387
    },
    {
      "epoch": 0.59702216206388,
      "grad_norm": 2.4626972675323486,
      "learning_rate": 8.225480470044942e-06,
      "loss": 0.9465,
      "step": 7388
    },
    {
      "epoch": 0.5971029717772076,
      "grad_norm": 2.6293704509735107,
      "learning_rate": 8.224980443359374e-06,
      "loss": 0.9974,
      "step": 7389
    },
    {
      "epoch": 0.5971837814905352,
      "grad_norm": 2.6583855152130127,
      "learning_rate": 8.224480361437657e-06,
      "loss": 0.9461,
      "step": 7390
    },
    {
      "epoch": 0.5972645912038627,
      "grad_norm": 2.6637701988220215,
      "learning_rate": 8.223980224288351e-06,
      "loss": 0.9341,
      "step": 7391
    },
    {
      "epoch": 0.5973454009171902,
      "grad_norm": 2.804215431213379,
      "learning_rate": 8.223480031920029e-06,
      "loss": 1.0286,
      "step": 7392
    },
    {
      "epoch": 0.5974262106305178,
      "grad_norm": 2.570969581604004,
      "learning_rate": 8.22297978434125e-06,
      "loss": 0.9489,
      "step": 7393
    },
    {
      "epoch": 0.5975070203438453,
      "grad_norm": 2.333669900894165,
      "learning_rate": 8.222479481560588e-06,
      "loss": 0.9311,
      "step": 7394
    },
    {
      "epoch": 0.5975878300571729,
      "grad_norm": 2.5228443145751953,
      "learning_rate": 8.221979123586611e-06,
      "loss": 0.9284,
      "step": 7395
    },
    {
      "epoch": 0.5976686397705004,
      "grad_norm": 2.806471586227417,
      "learning_rate": 8.221478710427889e-06,
      "loss": 0.8775,
      "step": 7396
    },
    {
      "epoch": 0.5977494494838279,
      "grad_norm": 2.5036611557006836,
      "learning_rate": 8.22097824209299e-06,
      "loss": 0.9738,
      "step": 7397
    },
    {
      "epoch": 0.5978302591971555,
      "grad_norm": 2.396198034286499,
      "learning_rate": 8.220477718590486e-06,
      "loss": 0.9063,
      "step": 7398
    },
    {
      "epoch": 0.5979110689104831,
      "grad_norm": 3.1461191177368164,
      "learning_rate": 8.219977139928957e-06,
      "loss": 1.0042,
      "step": 7399
    },
    {
      "epoch": 0.5979918786238105,
      "grad_norm": 2.9181549549102783,
      "learning_rate": 8.219476506116968e-06,
      "loss": 0.9777,
      "step": 7400
    },
    {
      "epoch": 0.5980726883371381,
      "grad_norm": 3.023691415786743,
      "learning_rate": 8.218975817163098e-06,
      "loss": 0.9521,
      "step": 7401
    },
    {
      "epoch": 0.5981534980504657,
      "grad_norm": 2.5120723247528076,
      "learning_rate": 8.21847507307592e-06,
      "loss": 1.0271,
      "step": 7402
    },
    {
      "epoch": 0.5982343077637932,
      "grad_norm": 2.790689706802368,
      "learning_rate": 8.217974273864013e-06,
      "loss": 0.9678,
      "step": 7403
    },
    {
      "epoch": 0.5983151174771207,
      "grad_norm": 2.6812565326690674,
      "learning_rate": 8.217473419535956e-06,
      "loss": 1.0998,
      "step": 7404
    },
    {
      "epoch": 0.5983959271904483,
      "grad_norm": 2.502741813659668,
      "learning_rate": 8.216972510100322e-06,
      "loss": 0.9043,
      "step": 7405
    },
    {
      "epoch": 0.5984767369037758,
      "grad_norm": 2.2891488075256348,
      "learning_rate": 8.216471545565694e-06,
      "loss": 0.9692,
      "step": 7406
    },
    {
      "epoch": 0.5985575466171034,
      "grad_norm": 2.8040974140167236,
      "learning_rate": 8.215970525940653e-06,
      "loss": 0.9022,
      "step": 7407
    },
    {
      "epoch": 0.5986383563304309,
      "grad_norm": 2.492466688156128,
      "learning_rate": 8.215469451233778e-06,
      "loss": 0.8387,
      "step": 7408
    },
    {
      "epoch": 0.5987191660437584,
      "grad_norm": 2.9831202030181885,
      "learning_rate": 8.214968321453653e-06,
      "loss": 1.0022,
      "step": 7409
    },
    {
      "epoch": 0.598799975757086,
      "grad_norm": 2.6915781497955322,
      "learning_rate": 8.214467136608861e-06,
      "loss": 1.091,
      "step": 7410
    },
    {
      "epoch": 0.5988807854704136,
      "grad_norm": 2.6560022830963135,
      "learning_rate": 8.213965896707983e-06,
      "loss": 1.0015,
      "step": 7411
    },
    {
      "epoch": 0.598961595183741,
      "grad_norm": 2.7713708877563477,
      "learning_rate": 8.213464601759609e-06,
      "loss": 0.8648,
      "step": 7412
    },
    {
      "epoch": 0.5990424048970686,
      "grad_norm": 2.421834945678711,
      "learning_rate": 8.212963251772322e-06,
      "loss": 0.9333,
      "step": 7413
    },
    {
      "epoch": 0.5991232146103962,
      "grad_norm": 2.515977382659912,
      "learning_rate": 8.212461846754708e-06,
      "loss": 0.9433,
      "step": 7414
    },
    {
      "epoch": 0.5992040243237237,
      "grad_norm": 2.403157949447632,
      "learning_rate": 8.211960386715356e-06,
      "loss": 1.0099,
      "step": 7415
    },
    {
      "epoch": 0.5992848340370512,
      "grad_norm": 3.128310441970825,
      "learning_rate": 8.211458871662855e-06,
      "loss": 0.8759,
      "step": 7416
    },
    {
      "epoch": 0.5993656437503788,
      "grad_norm": 2.983551025390625,
      "learning_rate": 8.210957301605797e-06,
      "loss": 0.8309,
      "step": 7417
    },
    {
      "epoch": 0.5994464534637063,
      "grad_norm": 2.517723798751831,
      "learning_rate": 8.210455676552771e-06,
      "loss": 1.0831,
      "step": 7418
    },
    {
      "epoch": 0.5995272631770339,
      "grad_norm": 2.5016441345214844,
      "learning_rate": 8.209953996512366e-06,
      "loss": 0.9106,
      "step": 7419
    },
    {
      "epoch": 0.5996080728903614,
      "grad_norm": 2.536555290222168,
      "learning_rate": 8.209452261493178e-06,
      "loss": 0.9192,
      "step": 7420
    },
    {
      "epoch": 0.5996888826036889,
      "grad_norm": 2.7434983253479004,
      "learning_rate": 8.208950471503798e-06,
      "loss": 0.9539,
      "step": 7421
    },
    {
      "epoch": 0.5997696923170165,
      "grad_norm": 2.569958448410034,
      "learning_rate": 8.208448626552821e-06,
      "loss": 0.9704,
      "step": 7422
    },
    {
      "epoch": 0.5998505020303441,
      "grad_norm": 2.4909441471099854,
      "learning_rate": 8.207946726648846e-06,
      "loss": 0.9827,
      "step": 7423
    },
    {
      "epoch": 0.5999313117436715,
      "grad_norm": 3.1090803146362305,
      "learning_rate": 8.207444771800464e-06,
      "loss": 1.0315,
      "step": 7424
    },
    {
      "epoch": 0.6000121214569991,
      "grad_norm": 2.9979429244995117,
      "learning_rate": 8.206942762016275e-06,
      "loss": 0.8577,
      "step": 7425
    },
    {
      "epoch": 0.6000929311703267,
      "grad_norm": 2.3236005306243896,
      "learning_rate": 8.206440697304876e-06,
      "loss": 0.9198,
      "step": 7426
    },
    {
      "epoch": 0.6001737408836542,
      "grad_norm": 3.4106571674346924,
      "learning_rate": 8.205938577674869e-06,
      "loss": 0.8742,
      "step": 7427
    },
    {
      "epoch": 0.6002545505969817,
      "grad_norm": 2.469843864440918,
      "learning_rate": 8.20543640313485e-06,
      "loss": 1.1598,
      "step": 7428
    },
    {
      "epoch": 0.6003353603103093,
      "grad_norm": 2.6596758365631104,
      "learning_rate": 8.204934173693425e-06,
      "loss": 0.8582,
      "step": 7429
    },
    {
      "epoch": 0.6004161700236368,
      "grad_norm": 2.2257728576660156,
      "learning_rate": 8.20443188935919e-06,
      "loss": 0.9651,
      "step": 7430
    },
    {
      "epoch": 0.6004969797369644,
      "grad_norm": 2.8060038089752197,
      "learning_rate": 8.203929550140754e-06,
      "loss": 0.9679,
      "step": 7431
    },
    {
      "epoch": 0.6005777894502919,
      "grad_norm": 2.5629186630249023,
      "learning_rate": 8.203427156046715e-06,
      "loss": 0.9635,
      "step": 7432
    },
    {
      "epoch": 0.6006585991636195,
      "grad_norm": 2.5945253372192383,
      "learning_rate": 8.202924707085684e-06,
      "loss": 0.9805,
      "step": 7433
    },
    {
      "epoch": 0.600739408876947,
      "grad_norm": 2.6697072982788086,
      "learning_rate": 8.20242220326626e-06,
      "loss": 0.9685,
      "step": 7434
    },
    {
      "epoch": 0.6008202185902746,
      "grad_norm": 2.6331472396850586,
      "learning_rate": 8.201919644597056e-06,
      "loss": 0.9987,
      "step": 7435
    },
    {
      "epoch": 0.6009010283036021,
      "grad_norm": 2.4247093200683594,
      "learning_rate": 8.201417031086676e-06,
      "loss": 0.8182,
      "step": 7436
    },
    {
      "epoch": 0.6009818380169296,
      "grad_norm": 2.3713717460632324,
      "learning_rate": 8.20091436274373e-06,
      "loss": 1.0406,
      "step": 7437
    },
    {
      "epoch": 0.6010626477302572,
      "grad_norm": 2.7394397258758545,
      "learning_rate": 8.200411639576827e-06,
      "loss": 0.8715,
      "step": 7438
    },
    {
      "epoch": 0.6011434574435848,
      "grad_norm": 2.602278709411621,
      "learning_rate": 8.199908861594575e-06,
      "loss": 1.0104,
      "step": 7439
    },
    {
      "epoch": 0.6012242671569122,
      "grad_norm": 2.413912057876587,
      "learning_rate": 8.19940602880559e-06,
      "loss": 1.0092,
      "step": 7440
    },
    {
      "epoch": 0.6013050768702398,
      "grad_norm": 2.4183106422424316,
      "learning_rate": 8.19890314121848e-06,
      "loss": 0.867,
      "step": 7441
    },
    {
      "epoch": 0.6013858865835674,
      "grad_norm": 2.622666358947754,
      "learning_rate": 8.198400198841861e-06,
      "loss": 1.0506,
      "step": 7442
    },
    {
      "epoch": 0.6014666962968949,
      "grad_norm": 2.7496120929718018,
      "learning_rate": 8.197897201684347e-06,
      "loss": 0.9275,
      "step": 7443
    },
    {
      "epoch": 0.6015475060102224,
      "grad_norm": 2.7471072673797607,
      "learning_rate": 8.197394149754552e-06,
      "loss": 0.9175,
      "step": 7444
    },
    {
      "epoch": 0.60162831572355,
      "grad_norm": 2.3735315799713135,
      "learning_rate": 8.196891043061093e-06,
      "loss": 0.9111,
      "step": 7445
    },
    {
      "epoch": 0.6017091254368775,
      "grad_norm": 2.809560537338257,
      "learning_rate": 8.196387881612586e-06,
      "loss": 0.9707,
      "step": 7446
    },
    {
      "epoch": 0.6017899351502051,
      "grad_norm": 2.4325170516967773,
      "learning_rate": 8.19588466541765e-06,
      "loss": 0.9298,
      "step": 7447
    },
    {
      "epoch": 0.6018707448635326,
      "grad_norm": 2.7904233932495117,
      "learning_rate": 8.195381394484903e-06,
      "loss": 0.9868,
      "step": 7448
    },
    {
      "epoch": 0.6019515545768601,
      "grad_norm": 2.5866594314575195,
      "learning_rate": 8.194878068822967e-06,
      "loss": 0.9125,
      "step": 7449
    },
    {
      "epoch": 0.6020323642901877,
      "grad_norm": 2.3913865089416504,
      "learning_rate": 8.19437468844046e-06,
      "loss": 0.9177,
      "step": 7450
    },
    {
      "epoch": 0.6021131740035153,
      "grad_norm": 2.8396122455596924,
      "learning_rate": 8.193871253346005e-06,
      "loss": 1.0068,
      "step": 7451
    },
    {
      "epoch": 0.6021939837168427,
      "grad_norm": 2.1447858810424805,
      "learning_rate": 8.193367763548223e-06,
      "loss": 0.9674,
      "step": 7452
    },
    {
      "epoch": 0.6022747934301703,
      "grad_norm": 2.292912721633911,
      "learning_rate": 8.192864219055741e-06,
      "loss": 0.9786,
      "step": 7453
    },
    {
      "epoch": 0.6023556031434979,
      "grad_norm": 2.575404167175293,
      "learning_rate": 8.19236061987718e-06,
      "loss": 1.0059,
      "step": 7454
    },
    {
      "epoch": 0.6024364128568254,
      "grad_norm": 2.3621768951416016,
      "learning_rate": 8.191856966021166e-06,
      "loss": 0.8969,
      "step": 7455
    },
    {
      "epoch": 0.6025172225701529,
      "grad_norm": 2.828322410583496,
      "learning_rate": 8.191353257496328e-06,
      "loss": 0.9147,
      "step": 7456
    },
    {
      "epoch": 0.6025980322834805,
      "grad_norm": 2.3915247917175293,
      "learning_rate": 8.190849494311291e-06,
      "loss": 0.8263,
      "step": 7457
    },
    {
      "epoch": 0.602678841996808,
      "grad_norm": 2.792440891265869,
      "learning_rate": 8.190345676474684e-06,
      "loss": 0.9835,
      "step": 7458
    },
    {
      "epoch": 0.6027596517101356,
      "grad_norm": 2.724057912826538,
      "learning_rate": 8.189841803995135e-06,
      "loss": 0.9776,
      "step": 7459
    },
    {
      "epoch": 0.6028404614234631,
      "grad_norm": 2.665722608566284,
      "learning_rate": 8.189337876881276e-06,
      "loss": 0.9499,
      "step": 7460
    },
    {
      "epoch": 0.6029212711367906,
      "grad_norm": 2.611553192138672,
      "learning_rate": 8.188833895141737e-06,
      "loss": 0.9243,
      "step": 7461
    },
    {
      "epoch": 0.6030020808501182,
      "grad_norm": 2.957235813140869,
      "learning_rate": 8.188329858785152e-06,
      "loss": 1.12,
      "step": 7462
    },
    {
      "epoch": 0.6030828905634458,
      "grad_norm": 2.905423641204834,
      "learning_rate": 8.18782576782015e-06,
      "loss": 0.9059,
      "step": 7463
    },
    {
      "epoch": 0.6031637002767732,
      "grad_norm": 2.612804412841797,
      "learning_rate": 8.187321622255366e-06,
      "loss": 0.9177,
      "step": 7464
    },
    {
      "epoch": 0.6032445099901008,
      "grad_norm": 2.303976058959961,
      "learning_rate": 8.186817422099437e-06,
      "loss": 0.9688,
      "step": 7465
    },
    {
      "epoch": 0.6033253197034284,
      "grad_norm": 2.7772412300109863,
      "learning_rate": 8.186313167361e-06,
      "loss": 0.8792,
      "step": 7466
    },
    {
      "epoch": 0.6034061294167559,
      "grad_norm": 2.651742935180664,
      "learning_rate": 8.185808858048684e-06,
      "loss": 0.992,
      "step": 7467
    },
    {
      "epoch": 0.6034869391300834,
      "grad_norm": 2.477022886276245,
      "learning_rate": 8.185304494171136e-06,
      "loss": 1.0104,
      "step": 7468
    },
    {
      "epoch": 0.603567748843411,
      "grad_norm": 2.6424434185028076,
      "learning_rate": 8.18480007573699e-06,
      "loss": 1.0653,
      "step": 7469
    },
    {
      "epoch": 0.6036485585567385,
      "grad_norm": 2.523757219314575,
      "learning_rate": 8.184295602754883e-06,
      "loss": 1.0043,
      "step": 7470
    },
    {
      "epoch": 0.6037293682700661,
      "grad_norm": 2.7214486598968506,
      "learning_rate": 8.183791075233461e-06,
      "loss": 0.9176,
      "step": 7471
    },
    {
      "epoch": 0.6038101779833936,
      "grad_norm": 3.0870211124420166,
      "learning_rate": 8.183286493181361e-06,
      "loss": 0.8808,
      "step": 7472
    },
    {
      "epoch": 0.6038909876967211,
      "grad_norm": 2.231369972229004,
      "learning_rate": 8.182781856607229e-06,
      "loss": 0.9157,
      "step": 7473
    },
    {
      "epoch": 0.6039717974100487,
      "grad_norm": 2.3250484466552734,
      "learning_rate": 8.182277165519703e-06,
      "loss": 0.9548,
      "step": 7474
    },
    {
      "epoch": 0.6040526071233763,
      "grad_norm": 2.418064594268799,
      "learning_rate": 8.181772419927431e-06,
      "loss": 0.9099,
      "step": 7475
    },
    {
      "epoch": 0.6041334168367037,
      "grad_norm": 2.4413342475891113,
      "learning_rate": 8.18126761983906e-06,
      "loss": 1.0852,
      "step": 7476
    },
    {
      "epoch": 0.6042142265500313,
      "grad_norm": 2.696765422821045,
      "learning_rate": 8.18076276526323e-06,
      "loss": 0.9989,
      "step": 7477
    },
    {
      "epoch": 0.6042950362633589,
      "grad_norm": 2.770642042160034,
      "learning_rate": 8.18025785620859e-06,
      "loss": 1.0051,
      "step": 7478
    },
    {
      "epoch": 0.6043758459766864,
      "grad_norm": 3.536003589630127,
      "learning_rate": 8.179752892683793e-06,
      "loss": 0.9775,
      "step": 7479
    },
    {
      "epoch": 0.6044566556900139,
      "grad_norm": 2.349836587905884,
      "learning_rate": 8.179247874697482e-06,
      "loss": 0.9597,
      "step": 7480
    },
    {
      "epoch": 0.6045374654033415,
      "grad_norm": 2.5413570404052734,
      "learning_rate": 8.17874280225831e-06,
      "loss": 0.9222,
      "step": 7481
    },
    {
      "epoch": 0.604618275116669,
      "grad_norm": 2.447821855545044,
      "learning_rate": 8.178237675374924e-06,
      "loss": 0.9519,
      "step": 7482
    },
    {
      "epoch": 0.6046990848299966,
      "grad_norm": 2.8698275089263916,
      "learning_rate": 8.177732494055979e-06,
      "loss": 0.9258,
      "step": 7483
    },
    {
      "epoch": 0.6047798945433241,
      "grad_norm": 2.419726848602295,
      "learning_rate": 8.177227258310128e-06,
      "loss": 0.9944,
      "step": 7484
    },
    {
      "epoch": 0.6048607042566516,
      "grad_norm": 2.484703779220581,
      "learning_rate": 8.176721968146021e-06,
      "loss": 1.0323,
      "step": 7485
    },
    {
      "epoch": 0.6049415139699792,
      "grad_norm": 2.353822708129883,
      "learning_rate": 8.176216623572315e-06,
      "loss": 0.8977,
      "step": 7486
    },
    {
      "epoch": 0.6050223236833068,
      "grad_norm": 2.9588623046875,
      "learning_rate": 8.175711224597664e-06,
      "loss": 1.0361,
      "step": 7487
    },
    {
      "epoch": 0.6051031333966342,
      "grad_norm": 2.3673336505889893,
      "learning_rate": 8.175205771230725e-06,
      "loss": 0.9604,
      "step": 7488
    },
    {
      "epoch": 0.6051839431099618,
      "grad_norm": 2.559023380279541,
      "learning_rate": 8.174700263480156e-06,
      "loss": 1.001,
      "step": 7489
    },
    {
      "epoch": 0.6052647528232894,
      "grad_norm": 2.4415910243988037,
      "learning_rate": 8.174194701354614e-06,
      "loss": 0.9434,
      "step": 7490
    },
    {
      "epoch": 0.6053455625366169,
      "grad_norm": 3.00506329536438,
      "learning_rate": 8.173689084862758e-06,
      "loss": 1.0497,
      "step": 7491
    },
    {
      "epoch": 0.6054263722499444,
      "grad_norm": 2.8126955032348633,
      "learning_rate": 8.173183414013249e-06,
      "loss": 0.9418,
      "step": 7492
    },
    {
      "epoch": 0.605507181963272,
      "grad_norm": 2.812666416168213,
      "learning_rate": 8.172677688814746e-06,
      "loss": 0.983,
      "step": 7493
    },
    {
      "epoch": 0.6055879916765995,
      "grad_norm": 2.6819217205047607,
      "learning_rate": 8.172171909275912e-06,
      "loss": 0.9592,
      "step": 7494
    },
    {
      "epoch": 0.6056688013899271,
      "grad_norm": 3.085527181625366,
      "learning_rate": 8.171666075405411e-06,
      "loss": 0.9987,
      "step": 7495
    },
    {
      "epoch": 0.6057496111032546,
      "grad_norm": 2.5476555824279785,
      "learning_rate": 8.171160187211906e-06,
      "loss": 0.875,
      "step": 7496
    },
    {
      "epoch": 0.6058304208165821,
      "grad_norm": 2.542846441268921,
      "learning_rate": 8.17065424470406e-06,
      "loss": 0.9667,
      "step": 7497
    },
    {
      "epoch": 0.6059112305299097,
      "grad_norm": 2.782371759414673,
      "learning_rate": 8.170148247890541e-06,
      "loss": 0.9271,
      "step": 7498
    },
    {
      "epoch": 0.6059920402432373,
      "grad_norm": 2.778146743774414,
      "learning_rate": 8.169642196780015e-06,
      "loss": 0.9166,
      "step": 7499
    },
    {
      "epoch": 0.6060728499565647,
      "grad_norm": 2.539367914199829,
      "learning_rate": 8.169136091381146e-06,
      "loss": 0.891,
      "step": 7500
    },
    {
      "epoch": 0.6061536596698923,
      "grad_norm": 2.3302626609802246,
      "learning_rate": 8.168629931702606e-06,
      "loss": 0.9383,
      "step": 7501
    },
    {
      "epoch": 0.6062344693832199,
      "grad_norm": 2.7369325160980225,
      "learning_rate": 8.168123717753065e-06,
      "loss": 0.9104,
      "step": 7502
    },
    {
      "epoch": 0.6063152790965474,
      "grad_norm": 2.4281458854675293,
      "learning_rate": 8.16761744954119e-06,
      "loss": 0.8976,
      "step": 7503
    },
    {
      "epoch": 0.606396088809875,
      "grad_norm": 2.6938211917877197,
      "learning_rate": 8.167111127075656e-06,
      "loss": 1.1036,
      "step": 7504
    },
    {
      "epoch": 0.6064768985232025,
      "grad_norm": 3.0842552185058594,
      "learning_rate": 8.166604750365131e-06,
      "loss": 0.9602,
      "step": 7505
    },
    {
      "epoch": 0.60655770823653,
      "grad_norm": 2.772709608078003,
      "learning_rate": 8.16609831941829e-06,
      "loss": 1.1227,
      "step": 7506
    },
    {
      "epoch": 0.6066385179498576,
      "grad_norm": 3.11780047416687,
      "learning_rate": 8.165591834243807e-06,
      "loss": 0.9699,
      "step": 7507
    },
    {
      "epoch": 0.6067193276631851,
      "grad_norm": 2.5700080394744873,
      "learning_rate": 8.165085294850356e-06,
      "loss": 1.0149,
      "step": 7508
    },
    {
      "epoch": 0.6068001373765126,
      "grad_norm": 2.578562021255493,
      "learning_rate": 8.164578701246615e-06,
      "loss": 0.852,
      "step": 7509
    },
    {
      "epoch": 0.6068809470898402,
      "grad_norm": 2.594264507293701,
      "learning_rate": 8.164072053441259e-06,
      "loss": 0.9298,
      "step": 7510
    },
    {
      "epoch": 0.6069617568031678,
      "grad_norm": 2.7076611518859863,
      "learning_rate": 8.163565351442965e-06,
      "loss": 1.0175,
      "step": 7511
    },
    {
      "epoch": 0.6070425665164952,
      "grad_norm": 2.7060465812683105,
      "learning_rate": 8.163058595260413e-06,
      "loss": 0.8987,
      "step": 7512
    },
    {
      "epoch": 0.6071233762298228,
      "grad_norm": 2.303374767303467,
      "learning_rate": 8.162551784902284e-06,
      "loss": 0.9175,
      "step": 7513
    },
    {
      "epoch": 0.6072041859431504,
      "grad_norm": 2.9991860389709473,
      "learning_rate": 8.162044920377253e-06,
      "loss": 1.051,
      "step": 7514
    },
    {
      "epoch": 0.6072849956564779,
      "grad_norm": 2.8217544555664062,
      "learning_rate": 8.161538001694007e-06,
      "loss": 0.9991,
      "step": 7515
    },
    {
      "epoch": 0.6073658053698054,
      "grad_norm": 2.489773750305176,
      "learning_rate": 8.161031028861226e-06,
      "loss": 0.9147,
      "step": 7516
    },
    {
      "epoch": 0.607446615083133,
      "grad_norm": 3.427377462387085,
      "learning_rate": 8.160524001887592e-06,
      "loss": 0.9328,
      "step": 7517
    },
    {
      "epoch": 0.6075274247964605,
      "grad_norm": 2.4631266593933105,
      "learning_rate": 8.160016920781792e-06,
      "loss": 0.89,
      "step": 7518
    },
    {
      "epoch": 0.6076082345097881,
      "grad_norm": 2.5758299827575684,
      "learning_rate": 8.15950978555251e-06,
      "loss": 0.9989,
      "step": 7519
    },
    {
      "epoch": 0.6076890442231156,
      "grad_norm": 2.9490103721618652,
      "learning_rate": 8.15900259620843e-06,
      "loss": 0.9204,
      "step": 7520
    },
    {
      "epoch": 0.6077698539364431,
      "grad_norm": 2.7213451862335205,
      "learning_rate": 8.15849535275824e-06,
      "loss": 0.9274,
      "step": 7521
    },
    {
      "epoch": 0.6078506636497707,
      "grad_norm": 2.4654622077941895,
      "learning_rate": 8.15798805521063e-06,
      "loss": 0.9773,
      "step": 7522
    },
    {
      "epoch": 0.6079314733630983,
      "grad_norm": 2.4854447841644287,
      "learning_rate": 8.157480703574287e-06,
      "loss": 1.1008,
      "step": 7523
    },
    {
      "epoch": 0.6080122830764257,
      "grad_norm": 2.836308240890503,
      "learning_rate": 8.1569732978579e-06,
      "loss": 0.8587,
      "step": 7524
    },
    {
      "epoch": 0.6080930927897533,
      "grad_norm": 2.7810237407684326,
      "learning_rate": 8.156465838070161e-06,
      "loss": 0.9275,
      "step": 7525
    },
    {
      "epoch": 0.6081739025030809,
      "grad_norm": 2.991436004638672,
      "learning_rate": 8.155958324219761e-06,
      "loss": 1.0333,
      "step": 7526
    },
    {
      "epoch": 0.6082547122164084,
      "grad_norm": 2.6031649112701416,
      "learning_rate": 8.155450756315393e-06,
      "loss": 0.9937,
      "step": 7527
    },
    {
      "epoch": 0.608335521929736,
      "grad_norm": 2.783151865005493,
      "learning_rate": 8.15494313436575e-06,
      "loss": 0.9891,
      "step": 7528
    },
    {
      "epoch": 0.6084163316430635,
      "grad_norm": 2.942488193511963,
      "learning_rate": 8.154435458379527e-06,
      "loss": 1.0682,
      "step": 7529
    },
    {
      "epoch": 0.608497141356391,
      "grad_norm": 2.9991884231567383,
      "learning_rate": 8.153927728365416e-06,
      "loss": 0.8114,
      "step": 7530
    },
    {
      "epoch": 0.6085779510697186,
      "grad_norm": 2.4739584922790527,
      "learning_rate": 8.153419944332117e-06,
      "loss": 0.9758,
      "step": 7531
    },
    {
      "epoch": 0.6086587607830461,
      "grad_norm": 2.494847297668457,
      "learning_rate": 8.152912106288326e-06,
      "loss": 1.0421,
      "step": 7532
    },
    {
      "epoch": 0.6087395704963736,
      "grad_norm": 2.6952362060546875,
      "learning_rate": 8.152404214242741e-06,
      "loss": 0.9681,
      "step": 7533
    },
    {
      "epoch": 0.6088203802097012,
      "grad_norm": 2.7268807888031006,
      "learning_rate": 8.151896268204063e-06,
      "loss": 0.9916,
      "step": 7534
    },
    {
      "epoch": 0.6089011899230288,
      "grad_norm": 2.6562206745147705,
      "learning_rate": 8.151388268180987e-06,
      "loss": 0.8829,
      "step": 7535
    },
    {
      "epoch": 0.6089819996363562,
      "grad_norm": 2.884728193283081,
      "learning_rate": 8.150880214182217e-06,
      "loss": 0.8888,
      "step": 7536
    },
    {
      "epoch": 0.6090628093496838,
      "grad_norm": 3.1932179927825928,
      "learning_rate": 8.150372106216455e-06,
      "loss": 0.9161,
      "step": 7537
    },
    {
      "epoch": 0.6091436190630114,
      "grad_norm": 2.834439277648926,
      "learning_rate": 8.149863944292404e-06,
      "loss": 0.9637,
      "step": 7538
    },
    {
      "epoch": 0.6092244287763389,
      "grad_norm": 2.834709644317627,
      "learning_rate": 8.149355728418764e-06,
      "loss": 0.9427,
      "step": 7539
    },
    {
      "epoch": 0.6093052384896664,
      "grad_norm": 2.3518810272216797,
      "learning_rate": 8.148847458604245e-06,
      "loss": 1.0289,
      "step": 7540
    },
    {
      "epoch": 0.609386048202994,
      "grad_norm": 2.834364175796509,
      "learning_rate": 8.148339134857548e-06,
      "loss": 0.9526,
      "step": 7541
    },
    {
      "epoch": 0.6094668579163215,
      "grad_norm": 2.6153712272644043,
      "learning_rate": 8.14783075718738e-06,
      "loss": 0.9339,
      "step": 7542
    },
    {
      "epoch": 0.6095476676296491,
      "grad_norm": 2.8085949420928955,
      "learning_rate": 8.14732232560245e-06,
      "loss": 0.9856,
      "step": 7543
    },
    {
      "epoch": 0.6096284773429766,
      "grad_norm": 2.8711469173431396,
      "learning_rate": 8.146813840111467e-06,
      "loss": 0.9479,
      "step": 7544
    },
    {
      "epoch": 0.6097092870563041,
      "grad_norm": 2.610267400741577,
      "learning_rate": 8.146305300723138e-06,
      "loss": 0.982,
      "step": 7545
    },
    {
      "epoch": 0.6097900967696317,
      "grad_norm": 2.5013654232025146,
      "learning_rate": 8.145796707446173e-06,
      "loss": 1.023,
      "step": 7546
    },
    {
      "epoch": 0.6098709064829593,
      "grad_norm": 2.7156944274902344,
      "learning_rate": 8.145288060289283e-06,
      "loss": 0.9269,
      "step": 7547
    },
    {
      "epoch": 0.6099517161962867,
      "grad_norm": 2.9470441341400146,
      "learning_rate": 8.14477935926118e-06,
      "loss": 0.997,
      "step": 7548
    },
    {
      "epoch": 0.6100325259096143,
      "grad_norm": 2.524672031402588,
      "learning_rate": 8.14427060437058e-06,
      "loss": 0.9207,
      "step": 7549
    },
    {
      "epoch": 0.6101133356229419,
      "grad_norm": 2.64919114112854,
      "learning_rate": 8.143761795626192e-06,
      "loss": 0.8817,
      "step": 7550
    },
    {
      "epoch": 0.6101941453362694,
      "grad_norm": 2.4781274795532227,
      "learning_rate": 8.143252933036733e-06,
      "loss": 0.9171,
      "step": 7551
    },
    {
      "epoch": 0.610274955049597,
      "grad_norm": 2.727407932281494,
      "learning_rate": 8.142744016610919e-06,
      "loss": 0.82,
      "step": 7552
    },
    {
      "epoch": 0.6103557647629245,
      "grad_norm": 2.7073307037353516,
      "learning_rate": 8.142235046357465e-06,
      "loss": 0.8762,
      "step": 7553
    },
    {
      "epoch": 0.610436574476252,
      "grad_norm": 2.8427350521087646,
      "learning_rate": 8.141726022285091e-06,
      "loss": 1.0253,
      "step": 7554
    },
    {
      "epoch": 0.6105173841895796,
      "grad_norm": 2.542018413543701,
      "learning_rate": 8.141216944402513e-06,
      "loss": 1.0002,
      "step": 7555
    },
    {
      "epoch": 0.6105981939029071,
      "grad_norm": 2.543044090270996,
      "learning_rate": 8.140707812718449e-06,
      "loss": 0.8985,
      "step": 7556
    },
    {
      "epoch": 0.6106790036162346,
      "grad_norm": 2.4094595909118652,
      "learning_rate": 8.140198627241623e-06,
      "loss": 1.0356,
      "step": 7557
    },
    {
      "epoch": 0.6107598133295622,
      "grad_norm": 2.7881250381469727,
      "learning_rate": 8.139689387980754e-06,
      "loss": 0.9284,
      "step": 7558
    },
    {
      "epoch": 0.6108406230428898,
      "grad_norm": 2.783515691757202,
      "learning_rate": 8.139180094944564e-06,
      "loss": 0.9208,
      "step": 7559
    },
    {
      "epoch": 0.6109214327562172,
      "grad_norm": 2.75824236869812,
      "learning_rate": 8.138670748141775e-06,
      "loss": 0.9096,
      "step": 7560
    },
    {
      "epoch": 0.6110022424695448,
      "grad_norm": 2.610955238342285,
      "learning_rate": 8.138161347581113e-06,
      "loss": 0.8984,
      "step": 7561
    },
    {
      "epoch": 0.6110830521828724,
      "grad_norm": 2.691279172897339,
      "learning_rate": 8.137651893271303e-06,
      "loss": 0.974,
      "step": 7562
    },
    {
      "epoch": 0.6111638618962,
      "grad_norm": 2.675173759460449,
      "learning_rate": 8.137142385221069e-06,
      "loss": 0.9107,
      "step": 7563
    },
    {
      "epoch": 0.6112446716095274,
      "grad_norm": 2.454118013381958,
      "learning_rate": 8.136632823439136e-06,
      "loss": 0.9351,
      "step": 7564
    },
    {
      "epoch": 0.611325481322855,
      "grad_norm": 3.163738965988159,
      "learning_rate": 8.136123207934238e-06,
      "loss": 0.9331,
      "step": 7565
    },
    {
      "epoch": 0.6114062910361826,
      "grad_norm": 2.5796146392822266,
      "learning_rate": 8.135613538715097e-06,
      "loss": 0.8914,
      "step": 7566
    },
    {
      "epoch": 0.6114871007495101,
      "grad_norm": 2.610637903213501,
      "learning_rate": 8.135103815790445e-06,
      "loss": 0.9718,
      "step": 7567
    },
    {
      "epoch": 0.6115679104628376,
      "grad_norm": 2.7604098320007324,
      "learning_rate": 8.134594039169013e-06,
      "loss": 0.913,
      "step": 7568
    },
    {
      "epoch": 0.6116487201761652,
      "grad_norm": 2.9775872230529785,
      "learning_rate": 8.13408420885953e-06,
      "loss": 0.9978,
      "step": 7569
    },
    {
      "epoch": 0.6117295298894927,
      "grad_norm": 2.9809317588806152,
      "learning_rate": 8.13357432487073e-06,
      "loss": 0.9162,
      "step": 7570
    },
    {
      "epoch": 0.6118103396028203,
      "grad_norm": 2.4901397228240967,
      "learning_rate": 8.133064387211344e-06,
      "loss": 0.9872,
      "step": 7571
    },
    {
      "epoch": 0.6118911493161479,
      "grad_norm": 3.06878399848938,
      "learning_rate": 8.132554395890111e-06,
      "loss": 0.8949,
      "step": 7572
    },
    {
      "epoch": 0.6119719590294753,
      "grad_norm": 2.597285747528076,
      "learning_rate": 8.13204435091576e-06,
      "loss": 0.8572,
      "step": 7573
    },
    {
      "epoch": 0.6120527687428029,
      "grad_norm": 2.7203357219696045,
      "learning_rate": 8.13153425229703e-06,
      "loss": 0.8784,
      "step": 7574
    },
    {
      "epoch": 0.6121335784561305,
      "grad_norm": 2.711458206176758,
      "learning_rate": 8.131024100042658e-06,
      "loss": 0.9551,
      "step": 7575
    },
    {
      "epoch": 0.612214388169458,
      "grad_norm": 2.75632381439209,
      "learning_rate": 8.130513894161381e-06,
      "loss": 0.9443,
      "step": 7576
    },
    {
      "epoch": 0.6122951978827855,
      "grad_norm": 2.6390979290008545,
      "learning_rate": 8.130003634661936e-06,
      "loss": 0.9309,
      "step": 7577
    },
    {
      "epoch": 0.6123760075961131,
      "grad_norm": 2.926806926727295,
      "learning_rate": 8.129493321553063e-06,
      "loss": 0.9948,
      "step": 7578
    },
    {
      "epoch": 0.6124568173094406,
      "grad_norm": 2.803065061569214,
      "learning_rate": 8.128982954843504e-06,
      "loss": 0.9275,
      "step": 7579
    },
    {
      "epoch": 0.6125376270227681,
      "grad_norm": 2.6212358474731445,
      "learning_rate": 8.128472534542002e-06,
      "loss": 0.972,
      "step": 7580
    },
    {
      "epoch": 0.6126184367360957,
      "grad_norm": 2.381321668624878,
      "learning_rate": 8.127962060657295e-06,
      "loss": 0.8629,
      "step": 7581
    },
    {
      "epoch": 0.6126992464494232,
      "grad_norm": 2.4566802978515625,
      "learning_rate": 8.127451533198129e-06,
      "loss": 1.0621,
      "step": 7582
    },
    {
      "epoch": 0.6127800561627508,
      "grad_norm": 2.5291404724121094,
      "learning_rate": 8.126940952173247e-06,
      "loss": 1.0163,
      "step": 7583
    },
    {
      "epoch": 0.6128608658760784,
      "grad_norm": 2.574113368988037,
      "learning_rate": 8.126430317591395e-06,
      "loss": 0.9515,
      "step": 7584
    },
    {
      "epoch": 0.6129416755894058,
      "grad_norm": 2.8008768558502197,
      "learning_rate": 8.125919629461317e-06,
      "loss": 0.8503,
      "step": 7585
    },
    {
      "epoch": 0.6130224853027334,
      "grad_norm": 2.6720664501190186,
      "learning_rate": 8.125408887791761e-06,
      "loss": 1.0535,
      "step": 7586
    },
    {
      "epoch": 0.613103295016061,
      "grad_norm": 2.9697463512420654,
      "learning_rate": 8.124898092591474e-06,
      "loss": 1.0389,
      "step": 7587
    },
    {
      "epoch": 0.6131841047293884,
      "grad_norm": 2.585587978363037,
      "learning_rate": 8.124387243869208e-06,
      "loss": 0.8858,
      "step": 7588
    },
    {
      "epoch": 0.613264914442716,
      "grad_norm": 3.2030749320983887,
      "learning_rate": 8.123876341633707e-06,
      "loss": 0.9673,
      "step": 7589
    },
    {
      "epoch": 0.6133457241560436,
      "grad_norm": 2.4657204151153564,
      "learning_rate": 8.123365385893728e-06,
      "loss": 1.0361,
      "step": 7590
    },
    {
      "epoch": 0.6134265338693711,
      "grad_norm": 2.3138375282287598,
      "learning_rate": 8.122854376658019e-06,
      "loss": 1.0051,
      "step": 7591
    },
    {
      "epoch": 0.6135073435826986,
      "grad_norm": 2.5959856510162354,
      "learning_rate": 8.122343313935331e-06,
      "loss": 0.9098,
      "step": 7592
    },
    {
      "epoch": 0.6135881532960262,
      "grad_norm": 2.63698673248291,
      "learning_rate": 8.121832197734419e-06,
      "loss": 1.0095,
      "step": 7593
    },
    {
      "epoch": 0.6136689630093537,
      "grad_norm": 2.67195987701416,
      "learning_rate": 8.121321028064038e-06,
      "loss": 0.9975,
      "step": 7594
    },
    {
      "epoch": 0.6137497727226813,
      "grad_norm": 2.6742136478424072,
      "learning_rate": 8.120809804932938e-06,
      "loss": 0.9893,
      "step": 7595
    },
    {
      "epoch": 0.6138305824360089,
      "grad_norm": 2.8699092864990234,
      "learning_rate": 8.120298528349883e-06,
      "loss": 0.9978,
      "step": 7596
    },
    {
      "epoch": 0.6139113921493363,
      "grad_norm": 2.471646308898926,
      "learning_rate": 8.119787198323626e-06,
      "loss": 0.9892,
      "step": 7597
    },
    {
      "epoch": 0.6139922018626639,
      "grad_norm": 2.8393266201019287,
      "learning_rate": 8.119275814862924e-06,
      "loss": 1.0142,
      "step": 7598
    },
    {
      "epoch": 0.6140730115759915,
      "grad_norm": 2.6085870265960693,
      "learning_rate": 8.118764377976537e-06,
      "loss": 1.04,
      "step": 7599
    },
    {
      "epoch": 0.614153821289319,
      "grad_norm": 2.986100673675537,
      "learning_rate": 8.118252887673224e-06,
      "loss": 0.9623,
      "step": 7600
    },
    {
      "epoch": 0.6142346310026465,
      "grad_norm": 3.16513991355896,
      "learning_rate": 8.117741343961746e-06,
      "loss": 0.9224,
      "step": 7601
    },
    {
      "epoch": 0.6143154407159741,
      "grad_norm": 2.460918426513672,
      "learning_rate": 8.117229746850866e-06,
      "loss": 0.9181,
      "step": 7602
    },
    {
      "epoch": 0.6143962504293016,
      "grad_norm": 2.756314516067505,
      "learning_rate": 8.116718096349341e-06,
      "loss": 0.9514,
      "step": 7603
    },
    {
      "epoch": 0.6144770601426292,
      "grad_norm": 2.6732490062713623,
      "learning_rate": 8.116206392465942e-06,
      "loss": 1.0893,
      "step": 7604
    },
    {
      "epoch": 0.6145578698559567,
      "grad_norm": 2.8199009895324707,
      "learning_rate": 8.115694635209427e-06,
      "loss": 0.9249,
      "step": 7605
    },
    {
      "epoch": 0.6146386795692842,
      "grad_norm": 2.402858257293701,
      "learning_rate": 8.115182824588565e-06,
      "loss": 0.9422,
      "step": 7606
    },
    {
      "epoch": 0.6147194892826118,
      "grad_norm": 3.0272507667541504,
      "learning_rate": 8.11467096061212e-06,
      "loss": 0.9662,
      "step": 7607
    },
    {
      "epoch": 0.6148002989959394,
      "grad_norm": 3.0503029823303223,
      "learning_rate": 8.114159043288861e-06,
      "loss": 0.839,
      "step": 7608
    },
    {
      "epoch": 0.6148811087092668,
      "grad_norm": 2.922384023666382,
      "learning_rate": 8.113647072627553e-06,
      "loss": 0.8798,
      "step": 7609
    },
    {
      "epoch": 0.6149619184225944,
      "grad_norm": 2.7376458644866943,
      "learning_rate": 8.113135048636967e-06,
      "loss": 0.9187,
      "step": 7610
    },
    {
      "epoch": 0.615042728135922,
      "grad_norm": 3.5451629161834717,
      "learning_rate": 8.112622971325872e-06,
      "loss": 0.9327,
      "step": 7611
    },
    {
      "epoch": 0.6151235378492494,
      "grad_norm": 2.669274091720581,
      "learning_rate": 8.112110840703038e-06,
      "loss": 1.004,
      "step": 7612
    },
    {
      "epoch": 0.615204347562577,
      "grad_norm": 2.3365893363952637,
      "learning_rate": 8.11159865677724e-06,
      "loss": 1.1307,
      "step": 7613
    },
    {
      "epoch": 0.6152851572759046,
      "grad_norm": 2.4509685039520264,
      "learning_rate": 8.111086419557246e-06,
      "loss": 1.0756,
      "step": 7614
    },
    {
      "epoch": 0.6153659669892321,
      "grad_norm": 2.6364622116088867,
      "learning_rate": 8.110574129051831e-06,
      "loss": 1.0557,
      "step": 7615
    },
    {
      "epoch": 0.6154467767025597,
      "grad_norm": 2.7215840816497803,
      "learning_rate": 8.110061785269772e-06,
      "loss": 0.881,
      "step": 7616
    },
    {
      "epoch": 0.6155275864158872,
      "grad_norm": 2.5293588638305664,
      "learning_rate": 8.109549388219838e-06,
      "loss": 0.9913,
      "step": 7617
    },
    {
      "epoch": 0.6156083961292147,
      "grad_norm": 2.808417320251465,
      "learning_rate": 8.10903693791081e-06,
      "loss": 0.9183,
      "step": 7618
    },
    {
      "epoch": 0.6156892058425423,
      "grad_norm": 2.7782840728759766,
      "learning_rate": 8.108524434351466e-06,
      "loss": 0.9487,
      "step": 7619
    },
    {
      "epoch": 0.6157700155558699,
      "grad_norm": 2.7505362033843994,
      "learning_rate": 8.108011877550581e-06,
      "loss": 0.8463,
      "step": 7620
    },
    {
      "epoch": 0.6158508252691973,
      "grad_norm": 2.883913040161133,
      "learning_rate": 8.107499267516934e-06,
      "loss": 0.8409,
      "step": 7621
    },
    {
      "epoch": 0.6159316349825249,
      "grad_norm": 2.6276159286499023,
      "learning_rate": 8.106986604259307e-06,
      "loss": 0.947,
      "step": 7622
    },
    {
      "epoch": 0.6160124446958525,
      "grad_norm": 2.501807928085327,
      "learning_rate": 8.106473887786478e-06,
      "loss": 0.7892,
      "step": 7623
    },
    {
      "epoch": 0.61609325440918,
      "grad_norm": 2.743285655975342,
      "learning_rate": 8.10596111810723e-06,
      "loss": 1.0016,
      "step": 7624
    },
    {
      "epoch": 0.6161740641225075,
      "grad_norm": 3.7059848308563232,
      "learning_rate": 8.105448295230345e-06,
      "loss": 0.8897,
      "step": 7625
    },
    {
      "epoch": 0.6162548738358351,
      "grad_norm": 2.7780449390411377,
      "learning_rate": 8.104935419164608e-06,
      "loss": 0.9984,
      "step": 7626
    },
    {
      "epoch": 0.6163356835491626,
      "grad_norm": 2.756298542022705,
      "learning_rate": 8.104422489918802e-06,
      "loss": 1.0321,
      "step": 7627
    },
    {
      "epoch": 0.6164164932624902,
      "grad_norm": 2.452901601791382,
      "learning_rate": 8.10390950750171e-06,
      "loss": 0.9096,
      "step": 7628
    },
    {
      "epoch": 0.6164973029758177,
      "grad_norm": 2.7609567642211914,
      "learning_rate": 8.103396471922123e-06,
      "loss": 0.975,
      "step": 7629
    },
    {
      "epoch": 0.6165781126891452,
      "grad_norm": 2.735701560974121,
      "learning_rate": 8.102883383188825e-06,
      "loss": 0.9084,
      "step": 7630
    },
    {
      "epoch": 0.6166589224024728,
      "grad_norm": 2.292210817337036,
      "learning_rate": 8.102370241310605e-06,
      "loss": 0.9648,
      "step": 7631
    },
    {
      "epoch": 0.6167397321158004,
      "grad_norm": 3.1585936546325684,
      "learning_rate": 8.10185704629625e-06,
      "loss": 1.0051,
      "step": 7632
    },
    {
      "epoch": 0.6168205418291278,
      "grad_norm": 2.4181199073791504,
      "learning_rate": 8.101343798154551e-06,
      "loss": 1.0571,
      "step": 7633
    },
    {
      "epoch": 0.6169013515424554,
      "grad_norm": 2.7890429496765137,
      "learning_rate": 8.1008304968943e-06,
      "loss": 0.949,
      "step": 7634
    },
    {
      "epoch": 0.616982161255783,
      "grad_norm": 2.5355355739593506,
      "learning_rate": 8.100317142524287e-06,
      "loss": 0.9061,
      "step": 7635
    },
    {
      "epoch": 0.6170629709691104,
      "grad_norm": 2.5393078327178955,
      "learning_rate": 8.099803735053306e-06,
      "loss": 1.008,
      "step": 7636
    },
    {
      "epoch": 0.617143780682438,
      "grad_norm": 2.9275808334350586,
      "learning_rate": 8.099290274490149e-06,
      "loss": 0.9149,
      "step": 7637
    },
    {
      "epoch": 0.6172245903957656,
      "grad_norm": 2.668325901031494,
      "learning_rate": 8.09877676084361e-06,
      "loss": 0.9632,
      "step": 7638
    },
    {
      "epoch": 0.6173054001090931,
      "grad_norm": 2.934856414794922,
      "learning_rate": 8.098263194122486e-06,
      "loss": 0.9808,
      "step": 7639
    },
    {
      "epoch": 0.6173862098224207,
      "grad_norm": 3.2335379123687744,
      "learning_rate": 8.097749574335573e-06,
      "loss": 0.8622,
      "step": 7640
    },
    {
      "epoch": 0.6174670195357482,
      "grad_norm": 2.827376127243042,
      "learning_rate": 8.097235901491667e-06,
      "loss": 0.9455,
      "step": 7641
    },
    {
      "epoch": 0.6175478292490757,
      "grad_norm": 3.290057420730591,
      "learning_rate": 8.096722175599566e-06,
      "loss": 0.866,
      "step": 7642
    },
    {
      "epoch": 0.6176286389624033,
      "grad_norm": 2.7397851943969727,
      "learning_rate": 8.096208396668068e-06,
      "loss": 1.0122,
      "step": 7643
    },
    {
      "epoch": 0.6177094486757309,
      "grad_norm": 2.768148183822632,
      "learning_rate": 8.095694564705974e-06,
      "loss": 1.0163,
      "step": 7644
    },
    {
      "epoch": 0.6177902583890583,
      "grad_norm": 2.8673741817474365,
      "learning_rate": 8.095180679722085e-06,
      "loss": 0.9631,
      "step": 7645
    },
    {
      "epoch": 0.6178710681023859,
      "grad_norm": 2.8062868118286133,
      "learning_rate": 8.094666741725203e-06,
      "loss": 1.0469,
      "step": 7646
    },
    {
      "epoch": 0.6179518778157135,
      "grad_norm": 2.763274669647217,
      "learning_rate": 8.094152750724132e-06,
      "loss": 1.031,
      "step": 7647
    },
    {
      "epoch": 0.618032687529041,
      "grad_norm": 2.3404388427734375,
      "learning_rate": 8.09363870672767e-06,
      "loss": 0.9555,
      "step": 7648
    },
    {
      "epoch": 0.6181134972423685,
      "grad_norm": 2.8336215019226074,
      "learning_rate": 8.093124609744625e-06,
      "loss": 1.0455,
      "step": 7649
    },
    {
      "epoch": 0.6181943069556961,
      "grad_norm": 2.526571035385132,
      "learning_rate": 8.092610459783802e-06,
      "loss": 0.8752,
      "step": 7650
    },
    {
      "epoch": 0.6182751166690236,
      "grad_norm": 2.445449113845825,
      "learning_rate": 8.092096256854007e-06,
      "loss": 0.9826,
      "step": 7651
    },
    {
      "epoch": 0.6183559263823512,
      "grad_norm": 2.870954990386963,
      "learning_rate": 8.091582000964049e-06,
      "loss": 0.9998,
      "step": 7652
    },
    {
      "epoch": 0.6184367360956787,
      "grad_norm": 2.7198636531829834,
      "learning_rate": 8.091067692122731e-06,
      "loss": 0.9769,
      "step": 7653
    },
    {
      "epoch": 0.6185175458090062,
      "grad_norm": 2.578165292739868,
      "learning_rate": 8.090553330338867e-06,
      "loss": 0.8554,
      "step": 7654
    },
    {
      "epoch": 0.6185983555223338,
      "grad_norm": 2.680886745452881,
      "learning_rate": 8.090038915621263e-06,
      "loss": 0.9601,
      "step": 7655
    },
    {
      "epoch": 0.6186791652356614,
      "grad_norm": 2.909229278564453,
      "learning_rate": 8.089524447978734e-06,
      "loss": 0.8381,
      "step": 7656
    },
    {
      "epoch": 0.6187599749489888,
      "grad_norm": 2.6055145263671875,
      "learning_rate": 8.089009927420087e-06,
      "loss": 0.9023,
      "step": 7657
    },
    {
      "epoch": 0.6188407846623164,
      "grad_norm": 3.098064422607422,
      "learning_rate": 8.088495353954135e-06,
      "loss": 1.0305,
      "step": 7658
    },
    {
      "epoch": 0.618921594375644,
      "grad_norm": 2.6698648929595947,
      "learning_rate": 8.087980727589695e-06,
      "loss": 0.8798,
      "step": 7659
    },
    {
      "epoch": 0.6190024040889714,
      "grad_norm": 2.487757444381714,
      "learning_rate": 8.087466048335578e-06,
      "loss": 0.9364,
      "step": 7660
    },
    {
      "epoch": 0.619083213802299,
      "grad_norm": 2.4260802268981934,
      "learning_rate": 8.086951316200603e-06,
      "loss": 0.8854,
      "step": 7661
    },
    {
      "epoch": 0.6191640235156266,
      "grad_norm": 2.9222092628479004,
      "learning_rate": 8.08643653119358e-06,
      "loss": 1.0188,
      "step": 7662
    },
    {
      "epoch": 0.6192448332289541,
      "grad_norm": 2.7569892406463623,
      "learning_rate": 8.085921693323331e-06,
      "loss": 0.9232,
      "step": 7663
    },
    {
      "epoch": 0.6193256429422817,
      "grad_norm": 3.115506410598755,
      "learning_rate": 8.085406802598671e-06,
      "loss": 0.9975,
      "step": 7664
    },
    {
      "epoch": 0.6194064526556092,
      "grad_norm": 2.9450008869171143,
      "learning_rate": 8.084891859028423e-06,
      "loss": 0.9828,
      "step": 7665
    },
    {
      "epoch": 0.6194872623689367,
      "grad_norm": 2.8621959686279297,
      "learning_rate": 8.084376862621402e-06,
      "loss": 0.9365,
      "step": 7666
    },
    {
      "epoch": 0.6195680720822643,
      "grad_norm": 2.434705972671509,
      "learning_rate": 8.083861813386433e-06,
      "loss": 1.0411,
      "step": 7667
    },
    {
      "epoch": 0.6196488817955919,
      "grad_norm": 2.969108819961548,
      "learning_rate": 8.083346711332332e-06,
      "loss": 0.9242,
      "step": 7668
    },
    {
      "epoch": 0.6197296915089193,
      "grad_norm": 2.6051666736602783,
      "learning_rate": 8.082831556467927e-06,
      "loss": 0.9473,
      "step": 7669
    },
    {
      "epoch": 0.6198105012222469,
      "grad_norm": 3.0353684425354004,
      "learning_rate": 8.082316348802038e-06,
      "loss": 1.0033,
      "step": 7670
    },
    {
      "epoch": 0.6198913109355745,
      "grad_norm": 2.894141912460327,
      "learning_rate": 8.08180108834349e-06,
      "loss": 1.0795,
      "step": 7671
    },
    {
      "epoch": 0.619972120648902,
      "grad_norm": 2.572617769241333,
      "learning_rate": 8.081285775101111e-06,
      "loss": 0.9461,
      "step": 7672
    },
    {
      "epoch": 0.6200529303622295,
      "grad_norm": 3.017763376235962,
      "learning_rate": 8.080770409083722e-06,
      "loss": 1.0067,
      "step": 7673
    },
    {
      "epoch": 0.6201337400755571,
      "grad_norm": 2.480952262878418,
      "learning_rate": 8.080254990300153e-06,
      "loss": 0.9987,
      "step": 7674
    },
    {
      "epoch": 0.6202145497888846,
      "grad_norm": 2.371596574783325,
      "learning_rate": 8.079739518759232e-06,
      "loss": 0.9978,
      "step": 7675
    },
    {
      "epoch": 0.6202953595022122,
      "grad_norm": 2.7950267791748047,
      "learning_rate": 8.079223994469786e-06,
      "loss": 0.9195,
      "step": 7676
    },
    {
      "epoch": 0.6203761692155397,
      "grad_norm": 3.1806201934814453,
      "learning_rate": 8.078708417440647e-06,
      "loss": 0.9923,
      "step": 7677
    },
    {
      "epoch": 0.6204569789288672,
      "grad_norm": 3.063190460205078,
      "learning_rate": 8.078192787680645e-06,
      "loss": 0.9292,
      "step": 7678
    },
    {
      "epoch": 0.6205377886421948,
      "grad_norm": 2.8096413612365723,
      "learning_rate": 8.07767710519861e-06,
      "loss": 0.9318,
      "step": 7679
    },
    {
      "epoch": 0.6206185983555224,
      "grad_norm": 2.7971744537353516,
      "learning_rate": 8.077161370003376e-06,
      "loss": 0.8552,
      "step": 7680
    },
    {
      "epoch": 0.6206994080688498,
      "grad_norm": 2.738651752471924,
      "learning_rate": 8.076645582103775e-06,
      "loss": 0.8102,
      "step": 7681
    },
    {
      "epoch": 0.6207802177821774,
      "grad_norm": 2.5714449882507324,
      "learning_rate": 8.076129741508642e-06,
      "loss": 0.9529,
      "step": 7682
    },
    {
      "epoch": 0.620861027495505,
      "grad_norm": 2.2240352630615234,
      "learning_rate": 8.075613848226813e-06,
      "loss": 0.9133,
      "step": 7683
    },
    {
      "epoch": 0.6209418372088324,
      "grad_norm": 2.441371202468872,
      "learning_rate": 8.075097902267122e-06,
      "loss": 0.8994,
      "step": 7684
    },
    {
      "epoch": 0.62102264692216,
      "grad_norm": 2.7136309146881104,
      "learning_rate": 8.074581903638408e-06,
      "loss": 1.0391,
      "step": 7685
    },
    {
      "epoch": 0.6211034566354876,
      "grad_norm": 2.485348701477051,
      "learning_rate": 8.074065852349506e-06,
      "loss": 1.1127,
      "step": 7686
    },
    {
      "epoch": 0.6211842663488151,
      "grad_norm": 2.329624891281128,
      "learning_rate": 8.073549748409258e-06,
      "loss": 1.017,
      "step": 7687
    },
    {
      "epoch": 0.6212650760621427,
      "grad_norm": 3.4000937938690186,
      "learning_rate": 8.073033591826502e-06,
      "loss": 0.9897,
      "step": 7688
    },
    {
      "epoch": 0.6213458857754702,
      "grad_norm": 2.233670234680176,
      "learning_rate": 8.072517382610077e-06,
      "loss": 1.015,
      "step": 7689
    },
    {
      "epoch": 0.6214266954887978,
      "grad_norm": 2.7281813621520996,
      "learning_rate": 8.072001120768827e-06,
      "loss": 1.0299,
      "step": 7690
    },
    {
      "epoch": 0.6215075052021253,
      "grad_norm": 2.977562189102173,
      "learning_rate": 8.071484806311593e-06,
      "loss": 1.133,
      "step": 7691
    },
    {
      "epoch": 0.6215883149154529,
      "grad_norm": 3.0641415119171143,
      "learning_rate": 8.070968439247219e-06,
      "loss": 0.9024,
      "step": 7692
    },
    {
      "epoch": 0.6216691246287804,
      "grad_norm": 2.2583773136138916,
      "learning_rate": 8.070452019584549e-06,
      "loss": 0.9157,
      "step": 7693
    },
    {
      "epoch": 0.6217499343421079,
      "grad_norm": 2.4764106273651123,
      "learning_rate": 8.069935547332427e-06,
      "loss": 0.9392,
      "step": 7694
    },
    {
      "epoch": 0.6218307440554355,
      "grad_norm": 2.8662688732147217,
      "learning_rate": 8.0694190224997e-06,
      "loss": 1.0317,
      "step": 7695
    },
    {
      "epoch": 0.6219115537687631,
      "grad_norm": 3.5188772678375244,
      "learning_rate": 8.068902445095216e-06,
      "loss": 0.89,
      "step": 7696
    },
    {
      "epoch": 0.6219923634820905,
      "grad_norm": 2.7252142429351807,
      "learning_rate": 8.06838581512782e-06,
      "loss": 0.9942,
      "step": 7697
    },
    {
      "epoch": 0.6220731731954181,
      "grad_norm": 2.730548143386841,
      "learning_rate": 8.06786913260636e-06,
      "loss": 0.9666,
      "step": 7698
    },
    {
      "epoch": 0.6221539829087457,
      "grad_norm": 2.656900644302368,
      "learning_rate": 8.06735239753969e-06,
      "loss": 0.8996,
      "step": 7699
    },
    {
      "epoch": 0.6222347926220732,
      "grad_norm": 3.110222339630127,
      "learning_rate": 8.066835609936656e-06,
      "loss": 0.8834,
      "step": 7700
    },
    {
      "epoch": 0.6223156023354007,
      "grad_norm": 2.502372980117798,
      "learning_rate": 8.066318769806111e-06,
      "loss": 0.8727,
      "step": 7701
    },
    {
      "epoch": 0.6223964120487283,
      "grad_norm": 3.104184627532959,
      "learning_rate": 8.06580187715691e-06,
      "loss": 0.883,
      "step": 7702
    },
    {
      "epoch": 0.6224772217620558,
      "grad_norm": 2.5845611095428467,
      "learning_rate": 8.0652849319979e-06,
      "loss": 0.9605,
      "step": 7703
    },
    {
      "epoch": 0.6225580314753834,
      "grad_norm": 2.477426767349243,
      "learning_rate": 8.064767934337942e-06,
      "loss": 1.0792,
      "step": 7704
    },
    {
      "epoch": 0.6226388411887109,
      "grad_norm": 2.931091547012329,
      "learning_rate": 8.064250884185884e-06,
      "loss": 0.9479,
      "step": 7705
    },
    {
      "epoch": 0.6227196509020384,
      "grad_norm": 2.541217803955078,
      "learning_rate": 8.063733781550588e-06,
      "loss": 1.0394,
      "step": 7706
    },
    {
      "epoch": 0.622800460615366,
      "grad_norm": 3.0264840126037598,
      "learning_rate": 8.063216626440907e-06,
      "loss": 0.9937,
      "step": 7707
    },
    {
      "epoch": 0.6228812703286936,
      "grad_norm": 2.858987808227539,
      "learning_rate": 8.062699418865697e-06,
      "loss": 0.9599,
      "step": 7708
    },
    {
      "epoch": 0.622962080042021,
      "grad_norm": 2.7067854404449463,
      "learning_rate": 8.062182158833824e-06,
      "loss": 0.9244,
      "step": 7709
    },
    {
      "epoch": 0.6230428897553486,
      "grad_norm": 2.5970213413238525,
      "learning_rate": 8.061664846354138e-06,
      "loss": 0.895,
      "step": 7710
    },
    {
      "epoch": 0.6231236994686762,
      "grad_norm": 2.6465702056884766,
      "learning_rate": 8.061147481435507e-06,
      "loss": 0.9913,
      "step": 7711
    },
    {
      "epoch": 0.6232045091820037,
      "grad_norm": 2.559605598449707,
      "learning_rate": 8.060630064086788e-06,
      "loss": 1.0017,
      "step": 7712
    },
    {
      "epoch": 0.6232853188953312,
      "grad_norm": 2.321462869644165,
      "learning_rate": 8.060112594316843e-06,
      "loss": 0.9695,
      "step": 7713
    },
    {
      "epoch": 0.6233661286086588,
      "grad_norm": 2.7076878547668457,
      "learning_rate": 8.059595072134538e-06,
      "loss": 0.8661,
      "step": 7714
    },
    {
      "epoch": 0.6234469383219863,
      "grad_norm": 2.558384656906128,
      "learning_rate": 8.059077497548733e-06,
      "loss": 0.8615,
      "step": 7715
    },
    {
      "epoch": 0.6235277480353139,
      "grad_norm": 2.7877197265625,
      "learning_rate": 8.058559870568297e-06,
      "loss": 0.9967,
      "step": 7716
    },
    {
      "epoch": 0.6236085577486414,
      "grad_norm": 2.6813652515411377,
      "learning_rate": 8.058042191202094e-06,
      "loss": 1.0862,
      "step": 7717
    },
    {
      "epoch": 0.6236893674619689,
      "grad_norm": 2.652118444442749,
      "learning_rate": 8.057524459458988e-06,
      "loss": 0.9581,
      "step": 7718
    },
    {
      "epoch": 0.6237701771752965,
      "grad_norm": 3.2014918327331543,
      "learning_rate": 8.05700667534785e-06,
      "loss": 0.9084,
      "step": 7719
    },
    {
      "epoch": 0.6238509868886241,
      "grad_norm": 2.563494920730591,
      "learning_rate": 8.056488838877547e-06,
      "loss": 0.8831,
      "step": 7720
    },
    {
      "epoch": 0.6239317966019515,
      "grad_norm": 3.018420457839966,
      "learning_rate": 8.055970950056946e-06,
      "loss": 0.8915,
      "step": 7721
    },
    {
      "epoch": 0.6240126063152791,
      "grad_norm": 2.609955310821533,
      "learning_rate": 8.055453008894922e-06,
      "loss": 1.0439,
      "step": 7722
    },
    {
      "epoch": 0.6240934160286067,
      "grad_norm": 2.7135331630706787,
      "learning_rate": 8.054935015400345e-06,
      "loss": 0.809,
      "step": 7723
    },
    {
      "epoch": 0.6241742257419342,
      "grad_norm": 2.5502469539642334,
      "learning_rate": 8.054416969582085e-06,
      "loss": 1.0508,
      "step": 7724
    },
    {
      "epoch": 0.6242550354552617,
      "grad_norm": 2.7606074810028076,
      "learning_rate": 8.053898871449013e-06,
      "loss": 1.0064,
      "step": 7725
    },
    {
      "epoch": 0.6243358451685893,
      "grad_norm": 2.7798190116882324,
      "learning_rate": 8.053380721010007e-06,
      "loss": 0.9649,
      "step": 7726
    },
    {
      "epoch": 0.6244166548819168,
      "grad_norm": 2.289156436920166,
      "learning_rate": 8.052862518273939e-06,
      "loss": 0.9963,
      "step": 7727
    },
    {
      "epoch": 0.6244974645952444,
      "grad_norm": 2.4941487312316895,
      "learning_rate": 8.052344263249688e-06,
      "loss": 0.9801,
      "step": 7728
    },
    {
      "epoch": 0.6245782743085719,
      "grad_norm": 2.8087286949157715,
      "learning_rate": 8.051825955946124e-06,
      "loss": 1.0877,
      "step": 7729
    },
    {
      "epoch": 0.6246590840218994,
      "grad_norm": 2.871695041656494,
      "learning_rate": 8.051307596372132e-06,
      "loss": 1.0272,
      "step": 7730
    },
    {
      "epoch": 0.624739893735227,
      "grad_norm": 1.9633084535598755,
      "learning_rate": 8.050789184536584e-06,
      "loss": 1.0399,
      "step": 7731
    },
    {
      "epoch": 0.6248207034485546,
      "grad_norm": 2.7255969047546387,
      "learning_rate": 8.050270720448364e-06,
      "loss": 0.8947,
      "step": 7732
    },
    {
      "epoch": 0.624901513161882,
      "grad_norm": 2.166358232498169,
      "learning_rate": 8.049752204116349e-06,
      "loss": 0.9626,
      "step": 7733
    },
    {
      "epoch": 0.6249823228752096,
      "grad_norm": 2.5491621494293213,
      "learning_rate": 8.049233635549421e-06,
      "loss": 0.943,
      "step": 7734
    },
    {
      "epoch": 0.6250631325885372,
      "grad_norm": 2.7059831619262695,
      "learning_rate": 8.048715014756462e-06,
      "loss": 1.0776,
      "step": 7735
    },
    {
      "epoch": 0.6251439423018647,
      "grad_norm": 2.5231704711914062,
      "learning_rate": 8.048196341746353e-06,
      "loss": 0.9624,
      "step": 7736
    },
    {
      "epoch": 0.6252247520151922,
      "grad_norm": 2.5308783054351807,
      "learning_rate": 8.047677616527979e-06,
      "loss": 0.8851,
      "step": 7737
    },
    {
      "epoch": 0.6253055617285198,
      "grad_norm": 3.063410520553589,
      "learning_rate": 8.047158839110223e-06,
      "loss": 0.8693,
      "step": 7738
    },
    {
      "epoch": 0.6253863714418473,
      "grad_norm": 2.350961208343506,
      "learning_rate": 8.046640009501973e-06,
      "loss": 0.9015,
      "step": 7739
    },
    {
      "epoch": 0.6254671811551749,
      "grad_norm": 2.388249635696411,
      "learning_rate": 8.046121127712116e-06,
      "loss": 0.8678,
      "step": 7740
    },
    {
      "epoch": 0.6255479908685024,
      "grad_norm": 2.5745139122009277,
      "learning_rate": 8.045602193749536e-06,
      "loss": 0.9142,
      "step": 7741
    },
    {
      "epoch": 0.6256288005818299,
      "grad_norm": 2.595492362976074,
      "learning_rate": 8.045083207623122e-06,
      "loss": 0.9591,
      "step": 7742
    },
    {
      "epoch": 0.6257096102951575,
      "grad_norm": 2.4084506034851074,
      "learning_rate": 8.044564169341765e-06,
      "loss": 0.9244,
      "step": 7743
    },
    {
      "epoch": 0.6257904200084851,
      "grad_norm": 3.108823537826538,
      "learning_rate": 8.04404507891435e-06,
      "loss": 1.0732,
      "step": 7744
    },
    {
      "epoch": 0.6258712297218125,
      "grad_norm": 2.862159013748169,
      "learning_rate": 8.043525936349775e-06,
      "loss": 1.0014,
      "step": 7745
    },
    {
      "epoch": 0.6259520394351401,
      "grad_norm": 2.857966423034668,
      "learning_rate": 8.043006741656925e-06,
      "loss": 0.8613,
      "step": 7746
    },
    {
      "epoch": 0.6260328491484677,
      "grad_norm": 2.724604368209839,
      "learning_rate": 8.042487494844695e-06,
      "loss": 0.8828,
      "step": 7747
    },
    {
      "epoch": 0.6261136588617952,
      "grad_norm": 2.9525856971740723,
      "learning_rate": 8.041968195921981e-06,
      "loss": 0.9661,
      "step": 7748
    },
    {
      "epoch": 0.6261944685751227,
      "grad_norm": 2.5938563346862793,
      "learning_rate": 8.041448844897672e-06,
      "loss": 1.0016,
      "step": 7749
    },
    {
      "epoch": 0.6262752782884503,
      "grad_norm": 2.6016035079956055,
      "learning_rate": 8.04092944178067e-06,
      "loss": 0.8378,
      "step": 7750
    },
    {
      "epoch": 0.6263560880017778,
      "grad_norm": 2.6379079818725586,
      "learning_rate": 8.040409986579865e-06,
      "loss": 1.1437,
      "step": 7751
    },
    {
      "epoch": 0.6264368977151054,
      "grad_norm": 2.7775843143463135,
      "learning_rate": 8.039890479304156e-06,
      "loss": 0.9355,
      "step": 7752
    },
    {
      "epoch": 0.6265177074284329,
      "grad_norm": 2.626626968383789,
      "learning_rate": 8.039370919962443e-06,
      "loss": 0.9042,
      "step": 7753
    },
    {
      "epoch": 0.6265985171417604,
      "grad_norm": 2.5845067501068115,
      "learning_rate": 8.038851308563621e-06,
      "loss": 0.9599,
      "step": 7754
    },
    {
      "epoch": 0.626679326855088,
      "grad_norm": 2.277064085006714,
      "learning_rate": 8.038331645116593e-06,
      "loss": 0.9462,
      "step": 7755
    },
    {
      "epoch": 0.6267601365684156,
      "grad_norm": 2.5511014461517334,
      "learning_rate": 8.037811929630258e-06,
      "loss": 0.989,
      "step": 7756
    },
    {
      "epoch": 0.626840946281743,
      "grad_norm": 2.4857261180877686,
      "learning_rate": 8.037292162113519e-06,
      "loss": 0.989,
      "step": 7757
    },
    {
      "epoch": 0.6269217559950706,
      "grad_norm": 2.708094835281372,
      "learning_rate": 8.036772342575277e-06,
      "loss": 0.9264,
      "step": 7758
    },
    {
      "epoch": 0.6270025657083982,
      "grad_norm": 3.272844076156616,
      "learning_rate": 8.036252471024436e-06,
      "loss": 0.9398,
      "step": 7759
    },
    {
      "epoch": 0.6270833754217257,
      "grad_norm": 2.653989553451538,
      "learning_rate": 8.035732547469897e-06,
      "loss": 1.1018,
      "step": 7760
    },
    {
      "epoch": 0.6271641851350532,
      "grad_norm": 2.796238899230957,
      "learning_rate": 8.035212571920571e-06,
      "loss": 1.0137,
      "step": 7761
    },
    {
      "epoch": 0.6272449948483808,
      "grad_norm": 2.274282217025757,
      "learning_rate": 8.034692544385359e-06,
      "loss": 1.0252,
      "step": 7762
    },
    {
      "epoch": 0.6273258045617083,
      "grad_norm": 2.3375332355499268,
      "learning_rate": 8.034172464873169e-06,
      "loss": 1.0097,
      "step": 7763
    },
    {
      "epoch": 0.6274066142750359,
      "grad_norm": 3.058797597885132,
      "learning_rate": 8.03365233339291e-06,
      "loss": 0.9241,
      "step": 7764
    },
    {
      "epoch": 0.6274874239883634,
      "grad_norm": 2.5773017406463623,
      "learning_rate": 8.033132149953489e-06,
      "loss": 0.8924,
      "step": 7765
    },
    {
      "epoch": 0.6275682337016909,
      "grad_norm": 2.597038507461548,
      "learning_rate": 8.032611914563816e-06,
      "loss": 1.0032,
      "step": 7766
    },
    {
      "epoch": 0.6276490434150185,
      "grad_norm": 2.7467923164367676,
      "learning_rate": 8.032091627232803e-06,
      "loss": 1.0524,
      "step": 7767
    },
    {
      "epoch": 0.6277298531283461,
      "grad_norm": 2.6006813049316406,
      "learning_rate": 8.031571287969359e-06,
      "loss": 1.0723,
      "step": 7768
    },
    {
      "epoch": 0.6278106628416735,
      "grad_norm": 2.623356580734253,
      "learning_rate": 8.031050896782397e-06,
      "loss": 0.8168,
      "step": 7769
    },
    {
      "epoch": 0.6278914725550011,
      "grad_norm": 2.6809449195861816,
      "learning_rate": 8.030530453680832e-06,
      "loss": 0.9118,
      "step": 7770
    },
    {
      "epoch": 0.6279722822683287,
      "grad_norm": 3.0026824474334717,
      "learning_rate": 8.030009958673573e-06,
      "loss": 0.8335,
      "step": 7771
    },
    {
      "epoch": 0.6280530919816562,
      "grad_norm": 2.902270555496216,
      "learning_rate": 8.02948941176954e-06,
      "loss": 0.9533,
      "step": 7772
    },
    {
      "epoch": 0.6281339016949837,
      "grad_norm": 2.7863898277282715,
      "learning_rate": 8.028968812977645e-06,
      "loss": 0.9708,
      "step": 7773
    },
    {
      "epoch": 0.6282147114083113,
      "grad_norm": 3.016594171524048,
      "learning_rate": 8.028448162306807e-06,
      "loss": 1.0456,
      "step": 7774
    },
    {
      "epoch": 0.6282955211216388,
      "grad_norm": 2.954946994781494,
      "learning_rate": 8.027927459765944e-06,
      "loss": 1.0253,
      "step": 7775
    },
    {
      "epoch": 0.6283763308349664,
      "grad_norm": 2.2077760696411133,
      "learning_rate": 8.02740670536397e-06,
      "loss": 0.945,
      "step": 7776
    },
    {
      "epoch": 0.6284571405482939,
      "grad_norm": 2.8146488666534424,
      "learning_rate": 8.026885899109808e-06,
      "loss": 0.7922,
      "step": 7777
    },
    {
      "epoch": 0.6285379502616214,
      "grad_norm": 2.6830267906188965,
      "learning_rate": 8.02636504101238e-06,
      "loss": 1.0701,
      "step": 7778
    },
    {
      "epoch": 0.628618759974949,
      "grad_norm": 2.534834384918213,
      "learning_rate": 8.025844131080602e-06,
      "loss": 0.892,
      "step": 7779
    },
    {
      "epoch": 0.6286995696882766,
      "grad_norm": 3.6110360622406006,
      "learning_rate": 8.0253231693234e-06,
      "loss": 0.8395,
      "step": 7780
    },
    {
      "epoch": 0.628780379401604,
      "grad_norm": 2.882394790649414,
      "learning_rate": 8.024802155749696e-06,
      "loss": 0.9216,
      "step": 7781
    },
    {
      "epoch": 0.6288611891149316,
      "grad_norm": 2.889479398727417,
      "learning_rate": 8.02428109036841e-06,
      "loss": 0.9265,
      "step": 7782
    },
    {
      "epoch": 0.6289419988282592,
      "grad_norm": 2.5216856002807617,
      "learning_rate": 8.023759973188471e-06,
      "loss": 0.876,
      "step": 7783
    },
    {
      "epoch": 0.6290228085415867,
      "grad_norm": 2.7473416328430176,
      "learning_rate": 8.023238804218805e-06,
      "loss": 1.0997,
      "step": 7784
    },
    {
      "epoch": 0.6291036182549142,
      "grad_norm": 3.13901948928833,
      "learning_rate": 8.022717583468334e-06,
      "loss": 1.0614,
      "step": 7785
    },
    {
      "epoch": 0.6291844279682418,
      "grad_norm": 2.6193127632141113,
      "learning_rate": 8.022196310945988e-06,
      "loss": 1.0336,
      "step": 7786
    },
    {
      "epoch": 0.6292652376815693,
      "grad_norm": 3.199700355529785,
      "learning_rate": 8.021674986660696e-06,
      "loss": 0.9012,
      "step": 7787
    },
    {
      "epoch": 0.6293460473948969,
      "grad_norm": 2.636735200881958,
      "learning_rate": 8.021153610621385e-06,
      "loss": 0.939,
      "step": 7788
    },
    {
      "epoch": 0.6294268571082244,
      "grad_norm": 2.5263350009918213,
      "learning_rate": 8.020632182836986e-06,
      "loss": 0.845,
      "step": 7789
    },
    {
      "epoch": 0.6295076668215519,
      "grad_norm": 2.9626433849334717,
      "learning_rate": 8.02011070331643e-06,
      "loss": 0.899,
      "step": 7790
    },
    {
      "epoch": 0.6295884765348795,
      "grad_norm": 2.3330037593841553,
      "learning_rate": 8.019589172068646e-06,
      "loss": 0.9297,
      "step": 7791
    },
    {
      "epoch": 0.6296692862482071,
      "grad_norm": 2.7311487197875977,
      "learning_rate": 8.019067589102572e-06,
      "loss": 0.7996,
      "step": 7792
    },
    {
      "epoch": 0.6297500959615345,
      "grad_norm": 2.5752053260803223,
      "learning_rate": 8.018545954427138e-06,
      "loss": 0.8863,
      "step": 7793
    },
    {
      "epoch": 0.6298309056748621,
      "grad_norm": 2.6735050678253174,
      "learning_rate": 8.018024268051276e-06,
      "loss": 0.9915,
      "step": 7794
    },
    {
      "epoch": 0.6299117153881897,
      "grad_norm": 2.407761573791504,
      "learning_rate": 8.017502529983927e-06,
      "loss": 1.0075,
      "step": 7795
    },
    {
      "epoch": 0.6299925251015172,
      "grad_norm": 2.5616161823272705,
      "learning_rate": 8.016980740234022e-06,
      "loss": 0.8683,
      "step": 7796
    },
    {
      "epoch": 0.6300733348148447,
      "grad_norm": 2.460862874984741,
      "learning_rate": 8.0164588988105e-06,
      "loss": 0.9291,
      "step": 7797
    },
    {
      "epoch": 0.6301541445281723,
      "grad_norm": 2.6509780883789062,
      "learning_rate": 8.0159370057223e-06,
      "loss": 0.9686,
      "step": 7798
    },
    {
      "epoch": 0.6302349542414998,
      "grad_norm": 2.631164073944092,
      "learning_rate": 8.015415060978358e-06,
      "loss": 0.9321,
      "step": 7799
    },
    {
      "epoch": 0.6303157639548274,
      "grad_norm": 2.489335298538208,
      "learning_rate": 8.014893064587617e-06,
      "loss": 0.8782,
      "step": 7800
    },
    {
      "epoch": 0.6303965736681549,
      "grad_norm": 2.837106466293335,
      "learning_rate": 8.014371016559016e-06,
      "loss": 1.0091,
      "step": 7801
    },
    {
      "epoch": 0.6304773833814824,
      "grad_norm": 2.7319021224975586,
      "learning_rate": 8.013848916901494e-06,
      "loss": 0.9499,
      "step": 7802
    },
    {
      "epoch": 0.63055819309481,
      "grad_norm": 2.3950157165527344,
      "learning_rate": 8.013326765623999e-06,
      "loss": 0.9136,
      "step": 7803
    },
    {
      "epoch": 0.6306390028081376,
      "grad_norm": 2.824277400970459,
      "learning_rate": 8.01280456273547e-06,
      "loss": 0.8592,
      "step": 7804
    },
    {
      "epoch": 0.630719812521465,
      "grad_norm": 2.6503448486328125,
      "learning_rate": 8.01228230824485e-06,
      "loss": 0.9947,
      "step": 7805
    },
    {
      "epoch": 0.6308006222347926,
      "grad_norm": 2.688629388809204,
      "learning_rate": 8.011760002161087e-06,
      "loss": 0.9991,
      "step": 7806
    },
    {
      "epoch": 0.6308814319481202,
      "grad_norm": 2.8561294078826904,
      "learning_rate": 8.011237644493124e-06,
      "loss": 0.9442,
      "step": 7807
    },
    {
      "epoch": 0.6309622416614477,
      "grad_norm": 2.801499128341675,
      "learning_rate": 8.01071523524991e-06,
      "loss": 0.8982,
      "step": 7808
    },
    {
      "epoch": 0.6310430513747752,
      "grad_norm": 2.8403820991516113,
      "learning_rate": 8.010192774440394e-06,
      "loss": 0.9184,
      "step": 7809
    },
    {
      "epoch": 0.6311238610881028,
      "grad_norm": 2.9464499950408936,
      "learning_rate": 8.00967026207352e-06,
      "loss": 1.0028,
      "step": 7810
    },
    {
      "epoch": 0.6312046708014303,
      "grad_norm": 2.475886106491089,
      "learning_rate": 8.009147698158241e-06,
      "loss": 0.9276,
      "step": 7811
    },
    {
      "epoch": 0.6312854805147579,
      "grad_norm": 2.6976876258850098,
      "learning_rate": 8.008625082703507e-06,
      "loss": 0.8602,
      "step": 7812
    },
    {
      "epoch": 0.6313662902280854,
      "grad_norm": 2.429450511932373,
      "learning_rate": 8.008102415718269e-06,
      "loss": 0.9231,
      "step": 7813
    },
    {
      "epoch": 0.6314470999414129,
      "grad_norm": 2.8100454807281494,
      "learning_rate": 8.007579697211476e-06,
      "loss": 0.9139,
      "step": 7814
    },
    {
      "epoch": 0.6315279096547405,
      "grad_norm": 2.411275863647461,
      "learning_rate": 8.007056927192084e-06,
      "loss": 1.0832,
      "step": 7815
    },
    {
      "epoch": 0.6316087193680681,
      "grad_norm": 2.3729898929595947,
      "learning_rate": 8.006534105669046e-06,
      "loss": 0.9018,
      "step": 7816
    },
    {
      "epoch": 0.6316895290813955,
      "grad_norm": 3.1594552993774414,
      "learning_rate": 8.006011232651317e-06,
      "loss": 0.9886,
      "step": 7817
    },
    {
      "epoch": 0.6317703387947231,
      "grad_norm": 2.5499870777130127,
      "learning_rate": 8.005488308147852e-06,
      "loss": 0.937,
      "step": 7818
    },
    {
      "epoch": 0.6318511485080507,
      "grad_norm": 2.6816816329956055,
      "learning_rate": 8.004965332167609e-06,
      "loss": 1.0091,
      "step": 7819
    },
    {
      "epoch": 0.6319319582213783,
      "grad_norm": 2.468831777572632,
      "learning_rate": 8.004442304719541e-06,
      "loss": 1.0155,
      "step": 7820
    },
    {
      "epoch": 0.6320127679347057,
      "grad_norm": 2.800617218017578,
      "learning_rate": 8.003919225812612e-06,
      "loss": 0.9332,
      "step": 7821
    },
    {
      "epoch": 0.6320935776480333,
      "grad_norm": 2.770237684249878,
      "learning_rate": 8.003396095455778e-06,
      "loss": 0.9352,
      "step": 7822
    },
    {
      "epoch": 0.6321743873613609,
      "grad_norm": 2.297513723373413,
      "learning_rate": 8.002872913658e-06,
      "loss": 1.0963,
      "step": 7823
    },
    {
      "epoch": 0.6322551970746884,
      "grad_norm": 2.436514139175415,
      "learning_rate": 8.002349680428235e-06,
      "loss": 0.8675,
      "step": 7824
    },
    {
      "epoch": 0.6323360067880159,
      "grad_norm": 3.1877448558807373,
      "learning_rate": 8.00182639577545e-06,
      "loss": 0.9469,
      "step": 7825
    },
    {
      "epoch": 0.6324168165013435,
      "grad_norm": 2.697570323944092,
      "learning_rate": 8.001303059708605e-06,
      "loss": 0.8729,
      "step": 7826
    },
    {
      "epoch": 0.632497626214671,
      "grad_norm": 2.694343090057373,
      "learning_rate": 8.000779672236664e-06,
      "loss": 0.982,
      "step": 7827
    },
    {
      "epoch": 0.6325784359279986,
      "grad_norm": 2.614118814468384,
      "learning_rate": 8.000256233368592e-06,
      "loss": 0.8815,
      "step": 7828
    },
    {
      "epoch": 0.6326592456413261,
      "grad_norm": 2.5081512928009033,
      "learning_rate": 7.999732743113353e-06,
      "loss": 0.9012,
      "step": 7829
    },
    {
      "epoch": 0.6327400553546536,
      "grad_norm": 2.6738438606262207,
      "learning_rate": 7.999209201479913e-06,
      "loss": 0.9866,
      "step": 7830
    },
    {
      "epoch": 0.6328208650679812,
      "grad_norm": 2.5792617797851562,
      "learning_rate": 7.99868560847724e-06,
      "loss": 0.9247,
      "step": 7831
    },
    {
      "epoch": 0.6329016747813088,
      "grad_norm": 2.5665578842163086,
      "learning_rate": 7.9981619641143e-06,
      "loss": 0.904,
      "step": 7832
    },
    {
      "epoch": 0.6329824844946362,
      "grad_norm": 2.240414619445801,
      "learning_rate": 7.997638268400067e-06,
      "loss": 0.8802,
      "step": 7833
    },
    {
      "epoch": 0.6330632942079638,
      "grad_norm": 2.4527547359466553,
      "learning_rate": 7.997114521343505e-06,
      "loss": 1.0719,
      "step": 7834
    },
    {
      "epoch": 0.6331441039212914,
      "grad_norm": 2.39039945602417,
      "learning_rate": 7.996590722953586e-06,
      "loss": 1.0883,
      "step": 7835
    },
    {
      "epoch": 0.6332249136346189,
      "grad_norm": 2.4689130783081055,
      "learning_rate": 7.996066873239283e-06,
      "loss": 0.8954,
      "step": 7836
    },
    {
      "epoch": 0.6333057233479464,
      "grad_norm": 2.6633870601654053,
      "learning_rate": 7.995542972209567e-06,
      "loss": 0.9858,
      "step": 7837
    },
    {
      "epoch": 0.633386533061274,
      "grad_norm": 2.6453187465667725,
      "learning_rate": 7.995019019873411e-06,
      "loss": 1.068,
      "step": 7838
    },
    {
      "epoch": 0.6334673427746015,
      "grad_norm": 2.5491058826446533,
      "learning_rate": 7.994495016239789e-06,
      "loss": 1.013,
      "step": 7839
    },
    {
      "epoch": 0.6335481524879291,
      "grad_norm": 2.9839184284210205,
      "learning_rate": 7.993970961317678e-06,
      "loss": 1.1085,
      "step": 7840
    },
    {
      "epoch": 0.6336289622012566,
      "grad_norm": 2.376394510269165,
      "learning_rate": 7.99344685511605e-06,
      "loss": 0.9267,
      "step": 7841
    },
    {
      "epoch": 0.6337097719145841,
      "grad_norm": 2.6307690143585205,
      "learning_rate": 7.992922697643885e-06,
      "loss": 0.9711,
      "step": 7842
    },
    {
      "epoch": 0.6337905816279117,
      "grad_norm": 2.667869806289673,
      "learning_rate": 7.99239848891016e-06,
      "loss": 0.9759,
      "step": 7843
    },
    {
      "epoch": 0.6338713913412393,
      "grad_norm": 3.161916494369507,
      "learning_rate": 7.99187422892385e-06,
      "loss": 0.8621,
      "step": 7844
    },
    {
      "epoch": 0.6339522010545667,
      "grad_norm": 2.229846477508545,
      "learning_rate": 7.99134991769394e-06,
      "loss": 0.8275,
      "step": 7845
    },
    {
      "epoch": 0.6340330107678943,
      "grad_norm": 2.615090847015381,
      "learning_rate": 7.990825555229407e-06,
      "loss": 0.7687,
      "step": 7846
    },
    {
      "epoch": 0.6341138204812219,
      "grad_norm": 2.7549939155578613,
      "learning_rate": 7.99030114153923e-06,
      "loss": 0.9814,
      "step": 7847
    },
    {
      "epoch": 0.6341946301945494,
      "grad_norm": 2.917659282684326,
      "learning_rate": 7.989776676632395e-06,
      "loss": 0.8248,
      "step": 7848
    },
    {
      "epoch": 0.6342754399078769,
      "grad_norm": 2.4968743324279785,
      "learning_rate": 7.989252160517884e-06,
      "loss": 0.9364,
      "step": 7849
    },
    {
      "epoch": 0.6343562496212045,
      "grad_norm": 2.3302972316741943,
      "learning_rate": 7.988727593204679e-06,
      "loss": 0.9482,
      "step": 7850
    },
    {
      "epoch": 0.634437059334532,
      "grad_norm": 3.111441135406494,
      "learning_rate": 7.988202974701766e-06,
      "loss": 0.9921,
      "step": 7851
    },
    {
      "epoch": 0.6345178690478596,
      "grad_norm": 2.7978410720825195,
      "learning_rate": 7.987678305018128e-06,
      "loss": 0.9492,
      "step": 7852
    },
    {
      "epoch": 0.6345986787611871,
      "grad_norm": 2.696904420852661,
      "learning_rate": 7.987153584162754e-06,
      "loss": 1.0293,
      "step": 7853
    },
    {
      "epoch": 0.6346794884745146,
      "grad_norm": 2.556978225708008,
      "learning_rate": 7.986628812144632e-06,
      "loss": 1.0262,
      "step": 7854
    },
    {
      "epoch": 0.6347602981878422,
      "grad_norm": 2.380255699157715,
      "learning_rate": 7.986103988972746e-06,
      "loss": 0.8333,
      "step": 7855
    },
    {
      "epoch": 0.6348411079011698,
      "grad_norm": 2.783190965652466,
      "learning_rate": 7.985579114656089e-06,
      "loss": 0.9723,
      "step": 7856
    },
    {
      "epoch": 0.6349219176144972,
      "grad_norm": 2.297658920288086,
      "learning_rate": 7.985054189203648e-06,
      "loss": 0.9584,
      "step": 7857
    },
    {
      "epoch": 0.6350027273278248,
      "grad_norm": 2.73989200592041,
      "learning_rate": 7.984529212624417e-06,
      "loss": 1.0242,
      "step": 7858
    },
    {
      "epoch": 0.6350835370411524,
      "grad_norm": 2.856266498565674,
      "learning_rate": 7.984004184927383e-06,
      "loss": 0.9769,
      "step": 7859
    },
    {
      "epoch": 0.6351643467544799,
      "grad_norm": 2.6034038066864014,
      "learning_rate": 7.983479106121543e-06,
      "loss": 0.9549,
      "step": 7860
    },
    {
      "epoch": 0.6352451564678074,
      "grad_norm": 2.601545572280884,
      "learning_rate": 7.982953976215888e-06,
      "loss": 0.9885,
      "step": 7861
    },
    {
      "epoch": 0.635325966181135,
      "grad_norm": 2.3755569458007812,
      "learning_rate": 7.982428795219412e-06,
      "loss": 0.896,
      "step": 7862
    },
    {
      "epoch": 0.6354067758944625,
      "grad_norm": 2.7541861534118652,
      "learning_rate": 7.98190356314111e-06,
      "loss": 0.9988,
      "step": 7863
    },
    {
      "epoch": 0.6354875856077901,
      "grad_norm": 2.01902437210083,
      "learning_rate": 7.981378279989979e-06,
      "loss": 1.1493,
      "step": 7864
    },
    {
      "epoch": 0.6355683953211176,
      "grad_norm": 2.4613897800445557,
      "learning_rate": 7.980852945775017e-06,
      "loss": 0.9756,
      "step": 7865
    },
    {
      "epoch": 0.6356492050344451,
      "grad_norm": 2.7889516353607178,
      "learning_rate": 7.98032756050522e-06,
      "loss": 1.08,
      "step": 7866
    },
    {
      "epoch": 0.6357300147477727,
      "grad_norm": 2.5871918201446533,
      "learning_rate": 7.979802124189585e-06,
      "loss": 0.9198,
      "step": 7867
    },
    {
      "epoch": 0.6358108244611003,
      "grad_norm": 2.2707509994506836,
      "learning_rate": 7.979276636837115e-06,
      "loss": 0.9897,
      "step": 7868
    },
    {
      "epoch": 0.6358916341744277,
      "grad_norm": 2.6156702041625977,
      "learning_rate": 7.978751098456807e-06,
      "loss": 0.9799,
      "step": 7869
    },
    {
      "epoch": 0.6359724438877553,
      "grad_norm": 2.71252703666687,
      "learning_rate": 7.978225509057665e-06,
      "loss": 0.9228,
      "step": 7870
    },
    {
      "epoch": 0.6360532536010829,
      "grad_norm": 2.387237071990967,
      "learning_rate": 7.977699868648688e-06,
      "loss": 0.8877,
      "step": 7871
    },
    {
      "epoch": 0.6361340633144104,
      "grad_norm": 2.5913264751434326,
      "learning_rate": 7.977174177238882e-06,
      "loss": 0.8578,
      "step": 7872
    },
    {
      "epoch": 0.6362148730277379,
      "grad_norm": 3.011514902114868,
      "learning_rate": 7.976648434837249e-06,
      "loss": 0.8832,
      "step": 7873
    },
    {
      "epoch": 0.6362956827410655,
      "grad_norm": 2.521620750427246,
      "learning_rate": 7.976122641452796e-06,
      "loss": 0.9534,
      "step": 7874
    },
    {
      "epoch": 0.636376492454393,
      "grad_norm": 2.4802489280700684,
      "learning_rate": 7.975596797094525e-06,
      "loss": 1.0569,
      "step": 7875
    },
    {
      "epoch": 0.6364573021677206,
      "grad_norm": 2.9051873683929443,
      "learning_rate": 7.975070901771444e-06,
      "loss": 0.8106,
      "step": 7876
    },
    {
      "epoch": 0.6365381118810481,
      "grad_norm": 2.508762836456299,
      "learning_rate": 7.974544955492562e-06,
      "loss": 0.9217,
      "step": 7877
    },
    {
      "epoch": 0.6366189215943756,
      "grad_norm": 2.3968288898468018,
      "learning_rate": 7.974018958266885e-06,
      "loss": 0.9095,
      "step": 7878
    },
    {
      "epoch": 0.6366997313077032,
      "grad_norm": 2.5430245399475098,
      "learning_rate": 7.973492910103424e-06,
      "loss": 0.9602,
      "step": 7879
    },
    {
      "epoch": 0.6367805410210308,
      "grad_norm": 2.813250780105591,
      "learning_rate": 7.972966811011187e-06,
      "loss": 0.9238,
      "step": 7880
    },
    {
      "epoch": 0.6368613507343582,
      "grad_norm": 2.7060039043426514,
      "learning_rate": 7.972440660999185e-06,
      "loss": 0.8969,
      "step": 7881
    },
    {
      "epoch": 0.6369421604476858,
      "grad_norm": 2.887779474258423,
      "learning_rate": 7.971914460076434e-06,
      "loss": 0.8344,
      "step": 7882
    },
    {
      "epoch": 0.6370229701610134,
      "grad_norm": 2.6328446865081787,
      "learning_rate": 7.971388208251937e-06,
      "loss": 0.9091,
      "step": 7883
    },
    {
      "epoch": 0.6371037798743409,
      "grad_norm": 2.2077245712280273,
      "learning_rate": 7.970861905534718e-06,
      "loss": 0.9226,
      "step": 7884
    },
    {
      "epoch": 0.6371845895876684,
      "grad_norm": 2.7489259243011475,
      "learning_rate": 7.970335551933785e-06,
      "loss": 0.9441,
      "step": 7885
    },
    {
      "epoch": 0.637265399300996,
      "grad_norm": 2.5437917709350586,
      "learning_rate": 7.969809147458154e-06,
      "loss": 0.94,
      "step": 7886
    },
    {
      "epoch": 0.6373462090143235,
      "grad_norm": 2.82645320892334,
      "learning_rate": 7.969282692116844e-06,
      "loss": 1.0309,
      "step": 7887
    },
    {
      "epoch": 0.6374270187276511,
      "grad_norm": 2.450364589691162,
      "learning_rate": 7.968756185918869e-06,
      "loss": 0.9699,
      "step": 7888
    },
    {
      "epoch": 0.6375078284409786,
      "grad_norm": 2.5669620037078857,
      "learning_rate": 7.968229628873246e-06,
      "loss": 0.9245,
      "step": 7889
    },
    {
      "epoch": 0.6375886381543061,
      "grad_norm": 2.772475242614746,
      "learning_rate": 7.967703020988997e-06,
      "loss": 0.7552,
      "step": 7890
    },
    {
      "epoch": 0.6376694478676337,
      "grad_norm": 2.900303840637207,
      "learning_rate": 7.967176362275138e-06,
      "loss": 0.9327,
      "step": 7891
    },
    {
      "epoch": 0.6377502575809613,
      "grad_norm": 4.1801910400390625,
      "learning_rate": 7.96664965274069e-06,
      "loss": 1.0217,
      "step": 7892
    },
    {
      "epoch": 0.6378310672942887,
      "grad_norm": 2.6613759994506836,
      "learning_rate": 7.966122892394679e-06,
      "loss": 0.8932,
      "step": 7893
    },
    {
      "epoch": 0.6379118770076163,
      "grad_norm": 2.336002826690674,
      "learning_rate": 7.96559608124612e-06,
      "loss": 1.0525,
      "step": 7894
    },
    {
      "epoch": 0.6379926867209439,
      "grad_norm": 2.6573903560638428,
      "learning_rate": 7.965069219304043e-06,
      "loss": 0.9063,
      "step": 7895
    },
    {
      "epoch": 0.6380734964342714,
      "grad_norm": 2.5049591064453125,
      "learning_rate": 7.964542306577464e-06,
      "loss": 0.8362,
      "step": 7896
    },
    {
      "epoch": 0.6381543061475989,
      "grad_norm": 2.3977386951446533,
      "learning_rate": 7.964015343075416e-06,
      "loss": 1.0461,
      "step": 7897
    },
    {
      "epoch": 0.6382351158609265,
      "grad_norm": 3.4002206325531006,
      "learning_rate": 7.963488328806921e-06,
      "loss": 0.912,
      "step": 7898
    },
    {
      "epoch": 0.638315925574254,
      "grad_norm": 2.4830734729766846,
      "learning_rate": 7.962961263781003e-06,
      "loss": 1.0555,
      "step": 7899
    },
    {
      "epoch": 0.6383967352875816,
      "grad_norm": 2.5380730628967285,
      "learning_rate": 7.962434148006693e-06,
      "loss": 0.8836,
      "step": 7900
    },
    {
      "epoch": 0.6384775450009091,
      "grad_norm": 2.824876546859741,
      "learning_rate": 7.961906981493016e-06,
      "loss": 0.989,
      "step": 7901
    },
    {
      "epoch": 0.6385583547142366,
      "grad_norm": 2.6988911628723145,
      "learning_rate": 7.961379764249004e-06,
      "loss": 0.8882,
      "step": 7902
    },
    {
      "epoch": 0.6386391644275642,
      "grad_norm": 2.415241241455078,
      "learning_rate": 7.960852496283686e-06,
      "loss": 0.9554,
      "step": 7903
    },
    {
      "epoch": 0.6387199741408918,
      "grad_norm": 2.6575350761413574,
      "learning_rate": 7.960325177606093e-06,
      "loss": 0.9876,
      "step": 7904
    },
    {
      "epoch": 0.6388007838542192,
      "grad_norm": 2.8557636737823486,
      "learning_rate": 7.959797808225257e-06,
      "loss": 0.9652,
      "step": 7905
    },
    {
      "epoch": 0.6388815935675468,
      "grad_norm": 2.9141557216644287,
      "learning_rate": 7.959270388150209e-06,
      "loss": 1.0365,
      "step": 7906
    },
    {
      "epoch": 0.6389624032808744,
      "grad_norm": 2.2590649127960205,
      "learning_rate": 7.958742917389983e-06,
      "loss": 0.8244,
      "step": 7907
    },
    {
      "epoch": 0.6390432129942019,
      "grad_norm": 2.4103503227233887,
      "learning_rate": 7.958215395953614e-06,
      "loss": 0.9339,
      "step": 7908
    },
    {
      "epoch": 0.6391240227075294,
      "grad_norm": 2.4939024448394775,
      "learning_rate": 7.957687823850136e-06,
      "loss": 1.0206,
      "step": 7909
    },
    {
      "epoch": 0.639204832420857,
      "grad_norm": 2.592484474182129,
      "learning_rate": 7.957160201088588e-06,
      "loss": 1.1189,
      "step": 7910
    },
    {
      "epoch": 0.6392856421341845,
      "grad_norm": 2.637906789779663,
      "learning_rate": 7.956632527678003e-06,
      "loss": 1.0982,
      "step": 7911
    },
    {
      "epoch": 0.6393664518475121,
      "grad_norm": 2.261767864227295,
      "learning_rate": 7.95610480362742e-06,
      "loss": 0.9419,
      "step": 7912
    },
    {
      "epoch": 0.6394472615608396,
      "grad_norm": 2.8713479042053223,
      "learning_rate": 7.955577028945881e-06,
      "loss": 1.0715,
      "step": 7913
    },
    {
      "epoch": 0.6395280712741671,
      "grad_norm": 2.9200046062469482,
      "learning_rate": 7.955049203642421e-06,
      "loss": 0.9441,
      "step": 7914
    },
    {
      "epoch": 0.6396088809874947,
      "grad_norm": 2.5749030113220215,
      "learning_rate": 7.954521327726082e-06,
      "loss": 0.9433,
      "step": 7915
    },
    {
      "epoch": 0.6396896907008223,
      "grad_norm": 2.8914525508880615,
      "learning_rate": 7.953993401205907e-06,
      "loss": 0.9765,
      "step": 7916
    },
    {
      "epoch": 0.6397705004141497,
      "grad_norm": 2.4222776889801025,
      "learning_rate": 7.953465424090933e-06,
      "loss": 1.0381,
      "step": 7917
    },
    {
      "epoch": 0.6398513101274773,
      "grad_norm": 2.5520670413970947,
      "learning_rate": 7.95293739639021e-06,
      "loss": 1.0009,
      "step": 7918
    },
    {
      "epoch": 0.6399321198408049,
      "grad_norm": 3.170346736907959,
      "learning_rate": 7.952409318112778e-06,
      "loss": 0.9548,
      "step": 7919
    },
    {
      "epoch": 0.6400129295541324,
      "grad_norm": 2.340275526046753,
      "learning_rate": 7.951881189267681e-06,
      "loss": 0.9463,
      "step": 7920
    },
    {
      "epoch": 0.6400937392674599,
      "grad_norm": 2.8253910541534424,
      "learning_rate": 7.951353009863966e-06,
      "loss": 0.9926,
      "step": 7921
    },
    {
      "epoch": 0.6401745489807875,
      "grad_norm": 2.312244176864624,
      "learning_rate": 7.950824779910678e-06,
      "loss": 1.0231,
      "step": 7922
    },
    {
      "epoch": 0.640255358694115,
      "grad_norm": 2.3926475048065186,
      "learning_rate": 7.950296499416866e-06,
      "loss": 1.0799,
      "step": 7923
    },
    {
      "epoch": 0.6403361684074426,
      "grad_norm": 2.747842311859131,
      "learning_rate": 7.949768168391579e-06,
      "loss": 0.8584,
      "step": 7924
    },
    {
      "epoch": 0.6404169781207701,
      "grad_norm": 3.1573312282562256,
      "learning_rate": 7.94923978684386e-06,
      "loss": 0.8375,
      "step": 7925
    },
    {
      "epoch": 0.6404977878340976,
      "grad_norm": 2.4790127277374268,
      "learning_rate": 7.94871135478277e-06,
      "loss": 0.9673,
      "step": 7926
    },
    {
      "epoch": 0.6405785975474252,
      "grad_norm": 2.638072967529297,
      "learning_rate": 7.94818287221735e-06,
      "loss": 0.8478,
      "step": 7927
    },
    {
      "epoch": 0.6406594072607528,
      "grad_norm": 2.696831703186035,
      "learning_rate": 7.947654339156653e-06,
      "loss": 0.947,
      "step": 7928
    },
    {
      "epoch": 0.6407402169740802,
      "grad_norm": 3.0501718521118164,
      "learning_rate": 7.947125755609734e-06,
      "loss": 0.897,
      "step": 7929
    },
    {
      "epoch": 0.6408210266874078,
      "grad_norm": 2.7243447303771973,
      "learning_rate": 7.946597121585648e-06,
      "loss": 0.9673,
      "step": 7930
    },
    {
      "epoch": 0.6409018364007354,
      "grad_norm": 2.5512492656707764,
      "learning_rate": 7.946068437093445e-06,
      "loss": 0.8887,
      "step": 7931
    },
    {
      "epoch": 0.6409826461140629,
      "grad_norm": 2.7076799869537354,
      "learning_rate": 7.945539702142184e-06,
      "loss": 0.9501,
      "step": 7932
    },
    {
      "epoch": 0.6410634558273904,
      "grad_norm": 2.636019468307495,
      "learning_rate": 7.945010916740916e-06,
      "loss": 0.913,
      "step": 7933
    },
    {
      "epoch": 0.641144265540718,
      "grad_norm": 2.915334701538086,
      "learning_rate": 7.944482080898703e-06,
      "loss": 0.9734,
      "step": 7934
    },
    {
      "epoch": 0.6412250752540455,
      "grad_norm": 2.865138292312622,
      "learning_rate": 7.9439531946246e-06,
      "loss": 0.9657,
      "step": 7935
    },
    {
      "epoch": 0.6413058849673731,
      "grad_norm": 2.6116302013397217,
      "learning_rate": 7.943424257927667e-06,
      "loss": 0.8759,
      "step": 7936
    },
    {
      "epoch": 0.6413866946807006,
      "grad_norm": 2.7858128547668457,
      "learning_rate": 7.942895270816961e-06,
      "loss": 0.9374,
      "step": 7937
    },
    {
      "epoch": 0.6414675043940281,
      "grad_norm": 2.4870376586914062,
      "learning_rate": 7.942366233301545e-06,
      "loss": 0.9349,
      "step": 7938
    },
    {
      "epoch": 0.6415483141073557,
      "grad_norm": 2.4941608905792236,
      "learning_rate": 7.941837145390478e-06,
      "loss": 0.7869,
      "step": 7939
    },
    {
      "epoch": 0.6416291238206833,
      "grad_norm": 2.52121901512146,
      "learning_rate": 7.941308007092823e-06,
      "loss": 1.0242,
      "step": 7940
    },
    {
      "epoch": 0.6417099335340107,
      "grad_norm": 2.7866406440734863,
      "learning_rate": 7.940778818417643e-06,
      "loss": 0.9009,
      "step": 7941
    },
    {
      "epoch": 0.6417907432473383,
      "grad_norm": 2.4861245155334473,
      "learning_rate": 7.940249579374002e-06,
      "loss": 0.9445,
      "step": 7942
    },
    {
      "epoch": 0.6418715529606659,
      "grad_norm": 3.105573892593384,
      "learning_rate": 7.939720289970963e-06,
      "loss": 1.0505,
      "step": 7943
    },
    {
      "epoch": 0.6419523626739934,
      "grad_norm": 2.5582029819488525,
      "learning_rate": 7.939190950217592e-06,
      "loss": 0.9101,
      "step": 7944
    },
    {
      "epoch": 0.6420331723873209,
      "grad_norm": 3.139284610748291,
      "learning_rate": 7.938661560122958e-06,
      "loss": 1.0687,
      "step": 7945
    },
    {
      "epoch": 0.6421139821006485,
      "grad_norm": 2.891692638397217,
      "learning_rate": 7.938132119696125e-06,
      "loss": 0.8713,
      "step": 7946
    },
    {
      "epoch": 0.642194791813976,
      "grad_norm": 2.503300428390503,
      "learning_rate": 7.93760262894616e-06,
      "loss": 0.9078,
      "step": 7947
    },
    {
      "epoch": 0.6422756015273036,
      "grad_norm": 2.2113330364227295,
      "learning_rate": 7.937073087882137e-06,
      "loss": 0.9952,
      "step": 7948
    },
    {
      "epoch": 0.6423564112406311,
      "grad_norm": 2.411799907684326,
      "learning_rate": 7.936543496513121e-06,
      "loss": 0.9453,
      "step": 7949
    },
    {
      "epoch": 0.6424372209539587,
      "grad_norm": 2.4153037071228027,
      "learning_rate": 7.936013854848185e-06,
      "loss": 0.9972,
      "step": 7950
    },
    {
      "epoch": 0.6425180306672862,
      "grad_norm": 2.5781900882720947,
      "learning_rate": 7.9354841628964e-06,
      "loss": 0.9114,
      "step": 7951
    },
    {
      "epoch": 0.6425988403806138,
      "grad_norm": 2.9225873947143555,
      "learning_rate": 7.934954420666838e-06,
      "loss": 0.9453,
      "step": 7952
    },
    {
      "epoch": 0.6426796500939413,
      "grad_norm": 2.8506546020507812,
      "learning_rate": 7.93442462816857e-06,
      "loss": 0.9384,
      "step": 7953
    },
    {
      "epoch": 0.6427604598072688,
      "grad_norm": 2.8344874382019043,
      "learning_rate": 7.933894785410676e-06,
      "loss": 0.9396,
      "step": 7954
    },
    {
      "epoch": 0.6428412695205964,
      "grad_norm": 2.7729973793029785,
      "learning_rate": 7.933364892402227e-06,
      "loss": 1.0725,
      "step": 7955
    },
    {
      "epoch": 0.642922079233924,
      "grad_norm": 2.305830478668213,
      "learning_rate": 7.932834949152298e-06,
      "loss": 0.9691,
      "step": 7956
    },
    {
      "epoch": 0.6430028889472514,
      "grad_norm": 2.670337438583374,
      "learning_rate": 7.932304955669967e-06,
      "loss": 0.9467,
      "step": 7957
    },
    {
      "epoch": 0.643083698660579,
      "grad_norm": 3.4643208980560303,
      "learning_rate": 7.931774911964312e-06,
      "loss": 0.8375,
      "step": 7958
    },
    {
      "epoch": 0.6431645083739066,
      "grad_norm": 2.6386911869049072,
      "learning_rate": 7.93124481804441e-06,
      "loss": 0.9037,
      "step": 7959
    },
    {
      "epoch": 0.6432453180872341,
      "grad_norm": 2.4012837409973145,
      "learning_rate": 7.930714673919342e-06,
      "loss": 1.0759,
      "step": 7960
    },
    {
      "epoch": 0.6433261278005616,
      "grad_norm": 2.9881134033203125,
      "learning_rate": 7.930184479598185e-06,
      "loss": 0.9248,
      "step": 7961
    },
    {
      "epoch": 0.6434069375138892,
      "grad_norm": 2.3555521965026855,
      "learning_rate": 7.929654235090022e-06,
      "loss": 1.013,
      "step": 7962
    },
    {
      "epoch": 0.6434877472272167,
      "grad_norm": 2.7471446990966797,
      "learning_rate": 7.929123940403938e-06,
      "loss": 1.0243,
      "step": 7963
    },
    {
      "epoch": 0.6435685569405443,
      "grad_norm": 2.596599817276001,
      "learning_rate": 7.92859359554901e-06,
      "loss": 0.9914,
      "step": 7964
    },
    {
      "epoch": 0.6436493666538718,
      "grad_norm": 2.3659653663635254,
      "learning_rate": 7.928063200534323e-06,
      "loss": 0.9262,
      "step": 7965
    },
    {
      "epoch": 0.6437301763671993,
      "grad_norm": 2.971914052963257,
      "learning_rate": 7.927532755368965e-06,
      "loss": 1.1657,
      "step": 7966
    },
    {
      "epoch": 0.6438109860805269,
      "grad_norm": 2.639777421951294,
      "learning_rate": 7.927002260062018e-06,
      "loss": 0.8875,
      "step": 7967
    },
    {
      "epoch": 0.6438917957938545,
      "grad_norm": 2.5370595455169678,
      "learning_rate": 7.926471714622568e-06,
      "loss": 0.9521,
      "step": 7968
    },
    {
      "epoch": 0.6439726055071819,
      "grad_norm": 2.853273630142212,
      "learning_rate": 7.925941119059702e-06,
      "loss": 0.9387,
      "step": 7969
    },
    {
      "epoch": 0.6440534152205095,
      "grad_norm": 2.5535457134246826,
      "learning_rate": 7.92541047338251e-06,
      "loss": 0.9954,
      "step": 7970
    },
    {
      "epoch": 0.6441342249338371,
      "grad_norm": 3.0810725688934326,
      "learning_rate": 7.924879777600078e-06,
      "loss": 0.9183,
      "step": 7971
    },
    {
      "epoch": 0.6442150346471646,
      "grad_norm": 2.6284542083740234,
      "learning_rate": 7.9243490317215e-06,
      "loss": 0.9929,
      "step": 7972
    },
    {
      "epoch": 0.6442958443604921,
      "grad_norm": 2.6094870567321777,
      "learning_rate": 7.923818235755859e-06,
      "loss": 1.0417,
      "step": 7973
    },
    {
      "epoch": 0.6443766540738197,
      "grad_norm": 2.7049832344055176,
      "learning_rate": 7.923287389712251e-06,
      "loss": 0.9304,
      "step": 7974
    },
    {
      "epoch": 0.6444574637871472,
      "grad_norm": 2.2217698097229004,
      "learning_rate": 7.92275649359977e-06,
      "loss": 1.0143,
      "step": 7975
    },
    {
      "epoch": 0.6445382735004748,
      "grad_norm": 2.392277956008911,
      "learning_rate": 7.922225547427504e-06,
      "loss": 0.9486,
      "step": 7976
    },
    {
      "epoch": 0.6446190832138023,
      "grad_norm": 2.4179892539978027,
      "learning_rate": 7.921694551204552e-06,
      "loss": 1.0242,
      "step": 7977
    },
    {
      "epoch": 0.6446998929271298,
      "grad_norm": 2.6017119884490967,
      "learning_rate": 7.921163504940004e-06,
      "loss": 0.9248,
      "step": 7978
    },
    {
      "epoch": 0.6447807026404574,
      "grad_norm": 2.2930874824523926,
      "learning_rate": 7.920632408642959e-06,
      "loss": 0.9522,
      "step": 7979
    },
    {
      "epoch": 0.644861512353785,
      "grad_norm": 2.3802316188812256,
      "learning_rate": 7.92010126232251e-06,
      "loss": 1.0128,
      "step": 7980
    },
    {
      "epoch": 0.6449423220671124,
      "grad_norm": 2.891817331314087,
      "learning_rate": 7.919570065987757e-06,
      "loss": 1.0646,
      "step": 7981
    },
    {
      "epoch": 0.64502313178044,
      "grad_norm": 2.985229015350342,
      "learning_rate": 7.919038819647797e-06,
      "loss": 0.8542,
      "step": 7982
    },
    {
      "epoch": 0.6451039414937676,
      "grad_norm": 3.0287842750549316,
      "learning_rate": 7.918507523311732e-06,
      "loss": 0.9686,
      "step": 7983
    },
    {
      "epoch": 0.6451847512070951,
      "grad_norm": 2.559133529663086,
      "learning_rate": 7.917976176988656e-06,
      "loss": 0.9464,
      "step": 7984
    },
    {
      "epoch": 0.6452655609204226,
      "grad_norm": 3.291764736175537,
      "learning_rate": 7.917444780687674e-06,
      "loss": 1.015,
      "step": 7985
    },
    {
      "epoch": 0.6453463706337502,
      "grad_norm": 2.4056661128997803,
      "learning_rate": 7.916913334417887e-06,
      "loss": 1.0248,
      "step": 7986
    },
    {
      "epoch": 0.6454271803470777,
      "grad_norm": 2.574171543121338,
      "learning_rate": 7.916381838188396e-06,
      "loss": 0.9222,
      "step": 7987
    },
    {
      "epoch": 0.6455079900604053,
      "grad_norm": 2.3434150218963623,
      "learning_rate": 7.915850292008305e-06,
      "loss": 0.9589,
      "step": 7988
    },
    {
      "epoch": 0.6455887997737328,
      "grad_norm": 2.8135275840759277,
      "learning_rate": 7.915318695886717e-06,
      "loss": 0.8705,
      "step": 7989
    },
    {
      "epoch": 0.6456696094870603,
      "grad_norm": 2.5289838314056396,
      "learning_rate": 7.914787049832741e-06,
      "loss": 0.9562,
      "step": 7990
    },
    {
      "epoch": 0.6457504192003879,
      "grad_norm": 3.281461238861084,
      "learning_rate": 7.914255353855478e-06,
      "loss": 0.9741,
      "step": 7991
    },
    {
      "epoch": 0.6458312289137155,
      "grad_norm": 2.5670602321624756,
      "learning_rate": 7.913723607964037e-06,
      "loss": 0.8829,
      "step": 7992
    },
    {
      "epoch": 0.6459120386270429,
      "grad_norm": 3.479647159576416,
      "learning_rate": 7.913191812167524e-06,
      "loss": 0.8005,
      "step": 7993
    },
    {
      "epoch": 0.6459928483403705,
      "grad_norm": 2.5427286624908447,
      "learning_rate": 7.912659966475051e-06,
      "loss": 0.9571,
      "step": 7994
    },
    {
      "epoch": 0.6460736580536981,
      "grad_norm": 2.7731635570526123,
      "learning_rate": 7.91212807089572e-06,
      "loss": 1.0192,
      "step": 7995
    },
    {
      "epoch": 0.6461544677670256,
      "grad_norm": 2.5769543647766113,
      "learning_rate": 7.91159612543865e-06,
      "loss": 0.9193,
      "step": 7996
    },
    {
      "epoch": 0.6462352774803531,
      "grad_norm": 2.678894281387329,
      "learning_rate": 7.911064130112947e-06,
      "loss": 0.8745,
      "step": 7997
    },
    {
      "epoch": 0.6463160871936807,
      "grad_norm": 3.0910682678222656,
      "learning_rate": 7.910532084927724e-06,
      "loss": 0.8793,
      "step": 7998
    },
    {
      "epoch": 0.6463968969070082,
      "grad_norm": 2.6806740760803223,
      "learning_rate": 7.909999989892092e-06,
      "loss": 1.0389,
      "step": 7999
    },
    {
      "epoch": 0.6464777066203358,
      "grad_norm": 2.4299354553222656,
      "learning_rate": 7.909467845015167e-06,
      "loss": 1.0521,
      "step": 8000
    },
    {
      "epoch": 0.6464777066203358,
      "eval_loss": 0.7883428931236267,
      "eval_runtime": 813.9119,
      "eval_samples_per_second": 102.426,
      "eval_steps_per_second": 12.804,
      "step": 8000
    },
    {
      "epoch": 0.6465585163336633,
      "grad_norm": 2.5144739151000977,
      "learning_rate": 7.90893565030606e-06,
      "loss": 0.8928,
      "step": 8001
    },
    {
      "epoch": 0.6466393260469908,
      "grad_norm": 2.7249040603637695,
      "learning_rate": 7.90840340577389e-06,
      "loss": 0.9346,
      "step": 8002
    },
    {
      "epoch": 0.6467201357603184,
      "grad_norm": 2.625905990600586,
      "learning_rate": 7.90787111142777e-06,
      "loss": 0.8915,
      "step": 8003
    },
    {
      "epoch": 0.646800945473646,
      "grad_norm": 2.4242939949035645,
      "learning_rate": 7.90733876727682e-06,
      "loss": 1.0098,
      "step": 8004
    },
    {
      "epoch": 0.6468817551869734,
      "grad_norm": 2.6896331310272217,
      "learning_rate": 7.906806373330156e-06,
      "loss": 0.8493,
      "step": 8005
    },
    {
      "epoch": 0.646962564900301,
      "grad_norm": 2.746581792831421,
      "learning_rate": 7.906273929596895e-06,
      "loss": 0.8971,
      "step": 8006
    },
    {
      "epoch": 0.6470433746136286,
      "grad_norm": 2.4979236125946045,
      "learning_rate": 7.905741436086158e-06,
      "loss": 0.9366,
      "step": 8007
    },
    {
      "epoch": 0.6471241843269561,
      "grad_norm": 2.9039738178253174,
      "learning_rate": 7.905208892807069e-06,
      "loss": 1.057,
      "step": 8008
    },
    {
      "epoch": 0.6472049940402836,
      "grad_norm": 2.85227108001709,
      "learning_rate": 7.904676299768741e-06,
      "loss": 0.8735,
      "step": 8009
    },
    {
      "epoch": 0.6472858037536112,
      "grad_norm": 2.5249216556549072,
      "learning_rate": 7.904143656980303e-06,
      "loss": 0.9225,
      "step": 8010
    },
    {
      "epoch": 0.6473666134669387,
      "grad_norm": 3.0269346237182617,
      "learning_rate": 7.903610964450876e-06,
      "loss": 0.9081,
      "step": 8011
    },
    {
      "epoch": 0.6474474231802663,
      "grad_norm": 2.7412753105163574,
      "learning_rate": 7.903078222189582e-06,
      "loss": 1.0081,
      "step": 8012
    },
    {
      "epoch": 0.6475282328935938,
      "grad_norm": 3.5761208534240723,
      "learning_rate": 7.902545430205548e-06,
      "loss": 0.9045,
      "step": 8013
    },
    {
      "epoch": 0.6476090426069213,
      "grad_norm": 2.872438669204712,
      "learning_rate": 7.902012588507898e-06,
      "loss": 0.952,
      "step": 8014
    },
    {
      "epoch": 0.6476898523202489,
      "grad_norm": 3.0681557655334473,
      "learning_rate": 7.901479697105759e-06,
      "loss": 0.9553,
      "step": 8015
    },
    {
      "epoch": 0.6477706620335765,
      "grad_norm": 2.399759531021118,
      "learning_rate": 7.90094675600826e-06,
      "loss": 0.9147,
      "step": 8016
    },
    {
      "epoch": 0.6478514717469039,
      "grad_norm": 2.7116787433624268,
      "learning_rate": 7.900413765224522e-06,
      "loss": 0.9374,
      "step": 8017
    },
    {
      "epoch": 0.6479322814602315,
      "grad_norm": 2.808452844619751,
      "learning_rate": 7.899880724763681e-06,
      "loss": 0.9594,
      "step": 8018
    },
    {
      "epoch": 0.6480130911735591,
      "grad_norm": 2.55055570602417,
      "learning_rate": 7.899347634634864e-06,
      "loss": 0.906,
      "step": 8019
    },
    {
      "epoch": 0.6480939008868866,
      "grad_norm": 2.735206365585327,
      "learning_rate": 7.898814494847203e-06,
      "loss": 0.8791,
      "step": 8020
    },
    {
      "epoch": 0.6481747106002141,
      "grad_norm": 2.5543875694274902,
      "learning_rate": 7.898281305409828e-06,
      "loss": 0.8443,
      "step": 8021
    },
    {
      "epoch": 0.6482555203135417,
      "grad_norm": 2.7260234355926514,
      "learning_rate": 7.897748066331872e-06,
      "loss": 0.8795,
      "step": 8022
    },
    {
      "epoch": 0.6483363300268692,
      "grad_norm": 2.9717183113098145,
      "learning_rate": 7.897214777622466e-06,
      "loss": 0.9062,
      "step": 8023
    },
    {
      "epoch": 0.6484171397401968,
      "grad_norm": 2.705498456954956,
      "learning_rate": 7.896681439290746e-06,
      "loss": 0.9699,
      "step": 8024
    },
    {
      "epoch": 0.6484979494535243,
      "grad_norm": 2.6744043827056885,
      "learning_rate": 7.896148051345847e-06,
      "loss": 0.8502,
      "step": 8025
    },
    {
      "epoch": 0.6485787591668518,
      "grad_norm": 2.67856502532959,
      "learning_rate": 7.895614613796905e-06,
      "loss": 0.9153,
      "step": 8026
    },
    {
      "epoch": 0.6486595688801794,
      "grad_norm": 2.8819680213928223,
      "learning_rate": 7.895081126653055e-06,
      "loss": 0.8608,
      "step": 8027
    },
    {
      "epoch": 0.648740378593507,
      "grad_norm": 2.570169687271118,
      "learning_rate": 7.894547589923434e-06,
      "loss": 0.9676,
      "step": 8028
    },
    {
      "epoch": 0.6488211883068344,
      "grad_norm": 2.6497445106506348,
      "learning_rate": 7.89401400361718e-06,
      "loss": 0.9532,
      "step": 8029
    },
    {
      "epoch": 0.648901998020162,
      "grad_norm": 2.6665828227996826,
      "learning_rate": 7.893480367743435e-06,
      "loss": 0.8863,
      "step": 8030
    },
    {
      "epoch": 0.6489828077334896,
      "grad_norm": 2.9539785385131836,
      "learning_rate": 7.892946682311337e-06,
      "loss": 0.9181,
      "step": 8031
    },
    {
      "epoch": 0.6490636174468171,
      "grad_norm": 2.391578197479248,
      "learning_rate": 7.892412947330027e-06,
      "loss": 0.8551,
      "step": 8032
    },
    {
      "epoch": 0.6491444271601446,
      "grad_norm": 3.2689549922943115,
      "learning_rate": 7.891879162808647e-06,
      "loss": 0.8375,
      "step": 8033
    },
    {
      "epoch": 0.6492252368734722,
      "grad_norm": 2.7801010608673096,
      "learning_rate": 7.891345328756336e-06,
      "loss": 1.0153,
      "step": 8034
    },
    {
      "epoch": 0.6493060465867997,
      "grad_norm": 3.0100972652435303,
      "learning_rate": 7.890811445182242e-06,
      "loss": 0.9091,
      "step": 8035
    },
    {
      "epoch": 0.6493868563001273,
      "grad_norm": 2.7918031215667725,
      "learning_rate": 7.890277512095508e-06,
      "loss": 0.8684,
      "step": 8036
    },
    {
      "epoch": 0.6494676660134548,
      "grad_norm": 2.637303113937378,
      "learning_rate": 7.889743529505279e-06,
      "loss": 1.0033,
      "step": 8037
    },
    {
      "epoch": 0.6495484757267823,
      "grad_norm": 2.579993963241577,
      "learning_rate": 7.889209497420698e-06,
      "loss": 0.9412,
      "step": 8038
    },
    {
      "epoch": 0.6496292854401099,
      "grad_norm": 2.736687421798706,
      "learning_rate": 7.888675415850915e-06,
      "loss": 0.9336,
      "step": 8039
    },
    {
      "epoch": 0.6497100951534375,
      "grad_norm": 2.3833227157592773,
      "learning_rate": 7.888141284805076e-06,
      "loss": 0.9242,
      "step": 8040
    },
    {
      "epoch": 0.6497909048667649,
      "grad_norm": 3.161198616027832,
      "learning_rate": 7.887607104292329e-06,
      "loss": 0.9606,
      "step": 8041
    },
    {
      "epoch": 0.6498717145800925,
      "grad_norm": 2.431769847869873,
      "learning_rate": 7.887072874321824e-06,
      "loss": 0.959,
      "step": 8042
    },
    {
      "epoch": 0.6499525242934201,
      "grad_norm": 2.372891902923584,
      "learning_rate": 7.886538594902712e-06,
      "loss": 1.0232,
      "step": 8043
    },
    {
      "epoch": 0.6500333340067476,
      "grad_norm": 2.8621509075164795,
      "learning_rate": 7.886004266044143e-06,
      "loss": 0.8678,
      "step": 8044
    },
    {
      "epoch": 0.6501141437200751,
      "grad_norm": 2.5629258155822754,
      "learning_rate": 7.885469887755269e-06,
      "loss": 0.9922,
      "step": 8045
    },
    {
      "epoch": 0.6501949534334027,
      "grad_norm": 2.5273284912109375,
      "learning_rate": 7.88493546004524e-06,
      "loss": 0.8521,
      "step": 8046
    },
    {
      "epoch": 0.6502757631467302,
      "grad_norm": 2.6331684589385986,
      "learning_rate": 7.884400982923214e-06,
      "loss": 0.902,
      "step": 8047
    },
    {
      "epoch": 0.6503565728600578,
      "grad_norm": 2.39434552192688,
      "learning_rate": 7.883866456398341e-06,
      "loss": 1.0303,
      "step": 8048
    },
    {
      "epoch": 0.6504373825733853,
      "grad_norm": 2.408538818359375,
      "learning_rate": 7.88333188047978e-06,
      "loss": 0.9691,
      "step": 8049
    },
    {
      "epoch": 0.6505181922867128,
      "grad_norm": 2.6992383003234863,
      "learning_rate": 7.882797255176685e-06,
      "loss": 0.8808,
      "step": 8050
    },
    {
      "epoch": 0.6505990020000404,
      "grad_norm": 2.766754150390625,
      "learning_rate": 7.882262580498213e-06,
      "loss": 0.9663,
      "step": 8051
    },
    {
      "epoch": 0.650679811713368,
      "grad_norm": 2.825565814971924,
      "learning_rate": 7.881727856453522e-06,
      "loss": 0.911,
      "step": 8052
    },
    {
      "epoch": 0.6507606214266954,
      "grad_norm": 2.7521347999572754,
      "learning_rate": 7.881193083051768e-06,
      "loss": 0.9249,
      "step": 8053
    },
    {
      "epoch": 0.650841431140023,
      "grad_norm": 2.808980703353882,
      "learning_rate": 7.880658260302116e-06,
      "loss": 0.9069,
      "step": 8054
    },
    {
      "epoch": 0.6509222408533506,
      "grad_norm": 2.2507877349853516,
      "learning_rate": 7.880123388213722e-06,
      "loss": 0.895,
      "step": 8055
    },
    {
      "epoch": 0.6510030505666781,
      "grad_norm": 2.9410057067871094,
      "learning_rate": 7.879588466795746e-06,
      "loss": 0.969,
      "step": 8056
    },
    {
      "epoch": 0.6510838602800056,
      "grad_norm": 3.09829044342041,
      "learning_rate": 7.879053496057355e-06,
      "loss": 0.9526,
      "step": 8057
    },
    {
      "epoch": 0.6511646699933332,
      "grad_norm": 2.6134846210479736,
      "learning_rate": 7.878518476007707e-06,
      "loss": 1.0055,
      "step": 8058
    },
    {
      "epoch": 0.6512454797066607,
      "grad_norm": 2.373108386993408,
      "learning_rate": 7.877983406655968e-06,
      "loss": 0.8778,
      "step": 8059
    },
    {
      "epoch": 0.6513262894199883,
      "grad_norm": 2.6206023693084717,
      "learning_rate": 7.877448288011299e-06,
      "loss": 1.0567,
      "step": 8060
    },
    {
      "epoch": 0.6514070991333158,
      "grad_norm": 2.493279218673706,
      "learning_rate": 7.876913120082871e-06,
      "loss": 1.0763,
      "step": 8061
    },
    {
      "epoch": 0.6514879088466433,
      "grad_norm": 2.254305839538574,
      "learning_rate": 7.876377902879845e-06,
      "loss": 1.0401,
      "step": 8062
    },
    {
      "epoch": 0.6515687185599709,
      "grad_norm": 2.1235649585723877,
      "learning_rate": 7.875842636411391e-06,
      "loss": 1.0286,
      "step": 8063
    },
    {
      "epoch": 0.6516495282732985,
      "grad_norm": 3.1026573181152344,
      "learning_rate": 7.875307320686677e-06,
      "loss": 0.9726,
      "step": 8064
    },
    {
      "epoch": 0.6517303379866259,
      "grad_norm": 2.582230567932129,
      "learning_rate": 7.874771955714869e-06,
      "loss": 0.979,
      "step": 8065
    },
    {
      "epoch": 0.6518111476999535,
      "grad_norm": 2.8377366065979004,
      "learning_rate": 7.874236541505141e-06,
      "loss": 0.9474,
      "step": 8066
    },
    {
      "epoch": 0.6518919574132811,
      "grad_norm": 2.4523541927337646,
      "learning_rate": 7.873701078066656e-06,
      "loss": 0.9977,
      "step": 8067
    },
    {
      "epoch": 0.6519727671266086,
      "grad_norm": 2.3577163219451904,
      "learning_rate": 7.873165565408592e-06,
      "loss": 0.9095,
      "step": 8068
    },
    {
      "epoch": 0.6520535768399361,
      "grad_norm": 2.790398597717285,
      "learning_rate": 7.87263000354012e-06,
      "loss": 0.8775,
      "step": 8069
    },
    {
      "epoch": 0.6521343865532637,
      "grad_norm": 2.748030185699463,
      "learning_rate": 7.87209439247041e-06,
      "loss": 1.1089,
      "step": 8070
    },
    {
      "epoch": 0.6522151962665912,
      "grad_norm": 2.7407009601593018,
      "learning_rate": 7.871558732208637e-06,
      "loss": 0.8933,
      "step": 8071
    },
    {
      "epoch": 0.6522960059799188,
      "grad_norm": 2.4599592685699463,
      "learning_rate": 7.871023022763978e-06,
      "loss": 0.8096,
      "step": 8072
    },
    {
      "epoch": 0.6523768156932463,
      "grad_norm": 2.2446229457855225,
      "learning_rate": 7.870487264145605e-06,
      "loss": 0.9436,
      "step": 8073
    },
    {
      "epoch": 0.6524576254065738,
      "grad_norm": 2.4470279216766357,
      "learning_rate": 7.869951456362694e-06,
      "loss": 1.0174,
      "step": 8074
    },
    {
      "epoch": 0.6525384351199014,
      "grad_norm": 2.4033892154693604,
      "learning_rate": 7.869415599424428e-06,
      "loss": 1.0154,
      "step": 8075
    },
    {
      "epoch": 0.652619244833229,
      "grad_norm": 2.594316005706787,
      "learning_rate": 7.868879693339975e-06,
      "loss": 0.8648,
      "step": 8076
    },
    {
      "epoch": 0.6527000545465564,
      "grad_norm": 2.740948438644409,
      "learning_rate": 7.868343738118523e-06,
      "loss": 1.2255,
      "step": 8077
    },
    {
      "epoch": 0.652780864259884,
      "grad_norm": 2.3592514991760254,
      "learning_rate": 7.867807733769249e-06,
      "loss": 0.901,
      "step": 8078
    },
    {
      "epoch": 0.6528616739732116,
      "grad_norm": 2.428778886795044,
      "learning_rate": 7.86727168030133e-06,
      "loss": 0.9368,
      "step": 8079
    },
    {
      "epoch": 0.6529424836865392,
      "grad_norm": 2.8880114555358887,
      "learning_rate": 7.86673557772395e-06,
      "loss": 1.0381,
      "step": 8080
    },
    {
      "epoch": 0.6530232933998666,
      "grad_norm": 2.7028584480285645,
      "learning_rate": 7.866199426046292e-06,
      "loss": 0.942,
      "step": 8081
    },
    {
      "epoch": 0.6531041031131942,
      "grad_norm": 2.4617974758148193,
      "learning_rate": 7.865663225277537e-06,
      "loss": 0.9983,
      "step": 8082
    },
    {
      "epoch": 0.6531849128265218,
      "grad_norm": 2.67744517326355,
      "learning_rate": 7.86512697542687e-06,
      "loss": 0.9058,
      "step": 8083
    },
    {
      "epoch": 0.6532657225398493,
      "grad_norm": 2.6994595527648926,
      "learning_rate": 7.864590676503477e-06,
      "loss": 0.8427,
      "step": 8084
    },
    {
      "epoch": 0.6533465322531768,
      "grad_norm": 2.9935851097106934,
      "learning_rate": 7.864054328516539e-06,
      "loss": 1.068,
      "step": 8085
    },
    {
      "epoch": 0.6534273419665044,
      "grad_norm": 2.4924216270446777,
      "learning_rate": 7.863517931475247e-06,
      "loss": 0.8998,
      "step": 8086
    },
    {
      "epoch": 0.6535081516798319,
      "grad_norm": 2.3296256065368652,
      "learning_rate": 7.862981485388787e-06,
      "loss": 0.9844,
      "step": 8087
    },
    {
      "epoch": 0.6535889613931595,
      "grad_norm": 2.7485809326171875,
      "learning_rate": 7.862444990266346e-06,
      "loss": 0.899,
      "step": 8088
    },
    {
      "epoch": 0.653669771106487,
      "grad_norm": 3.0662386417388916,
      "learning_rate": 7.861908446117112e-06,
      "loss": 0.9192,
      "step": 8089
    },
    {
      "epoch": 0.6537505808198145,
      "grad_norm": 2.37210750579834,
      "learning_rate": 7.861371852950277e-06,
      "loss": 0.8889,
      "step": 8090
    },
    {
      "epoch": 0.6538313905331421,
      "grad_norm": 2.0970723628997803,
      "learning_rate": 7.860835210775032e-06,
      "loss": 1.0256,
      "step": 8091
    },
    {
      "epoch": 0.6539122002464697,
      "grad_norm": 2.4767050743103027,
      "learning_rate": 7.860298519600567e-06,
      "loss": 0.9691,
      "step": 8092
    },
    {
      "epoch": 0.6539930099597971,
      "grad_norm": 2.7629003524780273,
      "learning_rate": 7.859761779436073e-06,
      "loss": 0.9086,
      "step": 8093
    },
    {
      "epoch": 0.6540738196731247,
      "grad_norm": 2.7053186893463135,
      "learning_rate": 7.859224990290744e-06,
      "loss": 1.0455,
      "step": 8094
    },
    {
      "epoch": 0.6541546293864523,
      "grad_norm": 2.660536050796509,
      "learning_rate": 7.858688152173774e-06,
      "loss": 0.955,
      "step": 8095
    },
    {
      "epoch": 0.6542354390997798,
      "grad_norm": 2.680800437927246,
      "learning_rate": 7.858151265094358e-06,
      "loss": 0.888,
      "step": 8096
    },
    {
      "epoch": 0.6543162488131073,
      "grad_norm": 2.6603574752807617,
      "learning_rate": 7.857614329061694e-06,
      "loss": 1.0141,
      "step": 8097
    },
    {
      "epoch": 0.6543970585264349,
      "grad_norm": 2.6089746952056885,
      "learning_rate": 7.857077344084973e-06,
      "loss": 0.8461,
      "step": 8098
    },
    {
      "epoch": 0.6544778682397624,
      "grad_norm": 2.3982930183410645,
      "learning_rate": 7.856540310173397e-06,
      "loss": 0.9372,
      "step": 8099
    },
    {
      "epoch": 0.65455867795309,
      "grad_norm": 2.7135848999023438,
      "learning_rate": 7.856003227336163e-06,
      "loss": 0.9828,
      "step": 8100
    },
    {
      "epoch": 0.6546394876664176,
      "grad_norm": 2.6838266849517822,
      "learning_rate": 7.855466095582466e-06,
      "loss": 1.0033,
      "step": 8101
    },
    {
      "epoch": 0.654720297379745,
      "grad_norm": 3.121654510498047,
      "learning_rate": 7.854928914921511e-06,
      "loss": 0.8824,
      "step": 8102
    },
    {
      "epoch": 0.6548011070930726,
      "grad_norm": 2.817322254180908,
      "learning_rate": 7.854391685362497e-06,
      "loss": 1.0008,
      "step": 8103
    },
    {
      "epoch": 0.6548819168064002,
      "grad_norm": 2.29815673828125,
      "learning_rate": 7.853854406914625e-06,
      "loss": 0.842,
      "step": 8104
    },
    {
      "epoch": 0.6549627265197276,
      "grad_norm": 2.7330162525177,
      "learning_rate": 7.853317079587097e-06,
      "loss": 0.9539,
      "step": 8105
    },
    {
      "epoch": 0.6550435362330552,
      "grad_norm": 2.7535059452056885,
      "learning_rate": 7.852779703389117e-06,
      "loss": 0.7586,
      "step": 8106
    },
    {
      "epoch": 0.6551243459463828,
      "grad_norm": 2.806523084640503,
      "learning_rate": 7.852242278329887e-06,
      "loss": 1.0183,
      "step": 8107
    },
    {
      "epoch": 0.6552051556597103,
      "grad_norm": 3.1190836429595947,
      "learning_rate": 7.851704804418615e-06,
      "loss": 0.844,
      "step": 8108
    },
    {
      "epoch": 0.6552859653730378,
      "grad_norm": 2.6873362064361572,
      "learning_rate": 7.851167281664505e-06,
      "loss": 0.885,
      "step": 8109
    },
    {
      "epoch": 0.6553667750863654,
      "grad_norm": 2.3540689945220947,
      "learning_rate": 7.850629710076761e-06,
      "loss": 1.0014,
      "step": 8110
    },
    {
      "epoch": 0.6554475847996929,
      "grad_norm": 2.5897302627563477,
      "learning_rate": 7.850092089664596e-06,
      "loss": 1.1105,
      "step": 8111
    },
    {
      "epoch": 0.6555283945130205,
      "grad_norm": 2.765284299850464,
      "learning_rate": 7.849554420437212e-06,
      "loss": 0.9382,
      "step": 8112
    },
    {
      "epoch": 0.655609204226348,
      "grad_norm": 2.3324174880981445,
      "learning_rate": 7.849016702403822e-06,
      "loss": 0.9636,
      "step": 8113
    },
    {
      "epoch": 0.6556900139396755,
      "grad_norm": 2.6708390712738037,
      "learning_rate": 7.848478935573636e-06,
      "loss": 0.979,
      "step": 8114
    },
    {
      "epoch": 0.6557708236530031,
      "grad_norm": 2.6447081565856934,
      "learning_rate": 7.84794111995586e-06,
      "loss": 1.0572,
      "step": 8115
    },
    {
      "epoch": 0.6558516333663307,
      "grad_norm": 2.8259623050689697,
      "learning_rate": 7.847403255559712e-06,
      "loss": 0.9194,
      "step": 8116
    },
    {
      "epoch": 0.6559324430796581,
      "grad_norm": 2.4877662658691406,
      "learning_rate": 7.846865342394399e-06,
      "loss": 1.0061,
      "step": 8117
    },
    {
      "epoch": 0.6560132527929857,
      "grad_norm": 2.4083447456359863,
      "learning_rate": 7.846327380469136e-06,
      "loss": 1.0343,
      "step": 8118
    },
    {
      "epoch": 0.6560940625063133,
      "grad_norm": 2.6998558044433594,
      "learning_rate": 7.84578936979314e-06,
      "loss": 1.0667,
      "step": 8119
    },
    {
      "epoch": 0.6561748722196408,
      "grad_norm": 3.0838003158569336,
      "learning_rate": 7.845251310375622e-06,
      "loss": 0.9375,
      "step": 8120
    },
    {
      "epoch": 0.6562556819329683,
      "grad_norm": 2.4537813663482666,
      "learning_rate": 7.844713202225796e-06,
      "loss": 0.8858,
      "step": 8121
    },
    {
      "epoch": 0.6563364916462959,
      "grad_norm": 2.4450929164886475,
      "learning_rate": 7.844175045352883e-06,
      "loss": 1.0421,
      "step": 8122
    },
    {
      "epoch": 0.6564173013596234,
      "grad_norm": 3.038662910461426,
      "learning_rate": 7.843636839766098e-06,
      "loss": 0.9504,
      "step": 8123
    },
    {
      "epoch": 0.656498111072951,
      "grad_norm": 2.5889804363250732,
      "learning_rate": 7.843098585474661e-06,
      "loss": 0.9088,
      "step": 8124
    },
    {
      "epoch": 0.6565789207862786,
      "grad_norm": 3.5918288230895996,
      "learning_rate": 7.84256028248779e-06,
      "loss": 0.9749,
      "step": 8125
    },
    {
      "epoch": 0.656659730499606,
      "grad_norm": 2.735668659210205,
      "learning_rate": 7.842021930814704e-06,
      "loss": 1.0207,
      "step": 8126
    },
    {
      "epoch": 0.6567405402129336,
      "grad_norm": 2.8185863494873047,
      "learning_rate": 7.841483530464622e-06,
      "loss": 0.9252,
      "step": 8127
    },
    {
      "epoch": 0.6568213499262612,
      "grad_norm": 2.4732277393341064,
      "learning_rate": 7.840945081446771e-06,
      "loss": 0.8831,
      "step": 8128
    },
    {
      "epoch": 0.6569021596395886,
      "grad_norm": 2.7103240489959717,
      "learning_rate": 7.840406583770367e-06,
      "loss": 0.9958,
      "step": 8129
    },
    {
      "epoch": 0.6569829693529162,
      "grad_norm": 2.8222081661224365,
      "learning_rate": 7.839868037444638e-06,
      "loss": 0.897,
      "step": 8130
    },
    {
      "epoch": 0.6570637790662438,
      "grad_norm": 2.4207465648651123,
      "learning_rate": 7.839329442478808e-06,
      "loss": 1.0815,
      "step": 8131
    },
    {
      "epoch": 0.6571445887795713,
      "grad_norm": 2.7852301597595215,
      "learning_rate": 7.838790798882097e-06,
      "loss": 0.929,
      "step": 8132
    },
    {
      "epoch": 0.6572253984928988,
      "grad_norm": 3.249326705932617,
      "learning_rate": 7.838252106663735e-06,
      "loss": 0.9907,
      "step": 8133
    },
    {
      "epoch": 0.6573062082062264,
      "grad_norm": 3.042463779449463,
      "learning_rate": 7.837713365832946e-06,
      "loss": 0.8892,
      "step": 8134
    },
    {
      "epoch": 0.6573870179195539,
      "grad_norm": 2.8553247451782227,
      "learning_rate": 7.83717457639896e-06,
      "loss": 0.9131,
      "step": 8135
    },
    {
      "epoch": 0.6574678276328815,
      "grad_norm": 2.646902084350586,
      "learning_rate": 7.836635738371003e-06,
      "loss": 0.8496,
      "step": 8136
    },
    {
      "epoch": 0.657548637346209,
      "grad_norm": 2.9965171813964844,
      "learning_rate": 7.836096851758305e-06,
      "loss": 0.859,
      "step": 8137
    },
    {
      "epoch": 0.6576294470595365,
      "grad_norm": 2.6346452236175537,
      "learning_rate": 7.835557916570096e-06,
      "loss": 1.0676,
      "step": 8138
    },
    {
      "epoch": 0.6577102567728641,
      "grad_norm": 2.815836191177368,
      "learning_rate": 7.835018932815607e-06,
      "loss": 0.9813,
      "step": 8139
    },
    {
      "epoch": 0.6577910664861917,
      "grad_norm": 2.4747209548950195,
      "learning_rate": 7.834479900504066e-06,
      "loss": 0.9141,
      "step": 8140
    },
    {
      "epoch": 0.6578718761995191,
      "grad_norm": 3.1171391010284424,
      "learning_rate": 7.83394081964471e-06,
      "loss": 0.9211,
      "step": 8141
    },
    {
      "epoch": 0.6579526859128467,
      "grad_norm": 2.8758184909820557,
      "learning_rate": 7.83340169024677e-06,
      "loss": 1.0595,
      "step": 8142
    },
    {
      "epoch": 0.6580334956261743,
      "grad_norm": 2.6710498332977295,
      "learning_rate": 7.832862512319481e-06,
      "loss": 0.9622,
      "step": 8143
    },
    {
      "epoch": 0.6581143053395018,
      "grad_norm": 2.159796953201294,
      "learning_rate": 7.832323285872074e-06,
      "loss": 1.0378,
      "step": 8144
    },
    {
      "epoch": 0.6581951150528293,
      "grad_norm": 2.8067142963409424,
      "learning_rate": 7.83178401091379e-06,
      "loss": 0.9905,
      "step": 8145
    },
    {
      "epoch": 0.6582759247661569,
      "grad_norm": 2.759702205657959,
      "learning_rate": 7.831244687453864e-06,
      "loss": 0.9221,
      "step": 8146
    },
    {
      "epoch": 0.6583567344794844,
      "grad_norm": 3.2061405181884766,
      "learning_rate": 7.83070531550153e-06,
      "loss": 0.9053,
      "step": 8147
    },
    {
      "epoch": 0.658437544192812,
      "grad_norm": 2.638150215148926,
      "learning_rate": 7.83016589506603e-06,
      "loss": 0.8964,
      "step": 8148
    },
    {
      "epoch": 0.6585183539061396,
      "grad_norm": 3.284224510192871,
      "learning_rate": 7.829626426156602e-06,
      "loss": 1.04,
      "step": 8149
    },
    {
      "epoch": 0.658599163619467,
      "grad_norm": 2.38773512840271,
      "learning_rate": 7.829086908782485e-06,
      "loss": 0.9748,
      "step": 8150
    },
    {
      "epoch": 0.6586799733327946,
      "grad_norm": 2.14300274848938,
      "learning_rate": 7.828547342952919e-06,
      "loss": 0.9204,
      "step": 8151
    },
    {
      "epoch": 0.6587607830461222,
      "grad_norm": 2.5345678329467773,
      "learning_rate": 7.828007728677146e-06,
      "loss": 0.9704,
      "step": 8152
    },
    {
      "epoch": 0.6588415927594496,
      "grad_norm": 2.6654064655303955,
      "learning_rate": 7.827468065964412e-06,
      "loss": 0.9997,
      "step": 8153
    },
    {
      "epoch": 0.6589224024727772,
      "grad_norm": 2.908874273300171,
      "learning_rate": 7.826928354823954e-06,
      "loss": 0.9124,
      "step": 8154
    },
    {
      "epoch": 0.6590032121861048,
      "grad_norm": 3.229367733001709,
      "learning_rate": 7.82638859526502e-06,
      "loss": 0.9907,
      "step": 8155
    },
    {
      "epoch": 0.6590840218994323,
      "grad_norm": 2.381117820739746,
      "learning_rate": 7.825848787296853e-06,
      "loss": 0.9387,
      "step": 8156
    },
    {
      "epoch": 0.6591648316127598,
      "grad_norm": 2.5243749618530273,
      "learning_rate": 7.825308930928699e-06,
      "loss": 1.1287,
      "step": 8157
    },
    {
      "epoch": 0.6592456413260874,
      "grad_norm": 2.5855326652526855,
      "learning_rate": 7.824769026169807e-06,
      "loss": 1.01,
      "step": 8158
    },
    {
      "epoch": 0.6593264510394149,
      "grad_norm": 2.3746542930603027,
      "learning_rate": 7.824229073029419e-06,
      "loss": 1.0427,
      "step": 8159
    },
    {
      "epoch": 0.6594072607527425,
      "grad_norm": 2.572866439819336,
      "learning_rate": 7.823689071516787e-06,
      "loss": 1.0039,
      "step": 8160
    },
    {
      "epoch": 0.65948807046607,
      "grad_norm": 2.453367233276367,
      "learning_rate": 7.823149021641159e-06,
      "loss": 1.0211,
      "step": 8161
    },
    {
      "epoch": 0.6595688801793975,
      "grad_norm": 2.7661163806915283,
      "learning_rate": 7.822608923411786e-06,
      "loss": 1.0553,
      "step": 8162
    },
    {
      "epoch": 0.6596496898927251,
      "grad_norm": 2.6568291187286377,
      "learning_rate": 7.822068776837914e-06,
      "loss": 1.011,
      "step": 8163
    },
    {
      "epoch": 0.6597304996060527,
      "grad_norm": 2.8543918132781982,
      "learning_rate": 7.821528581928802e-06,
      "loss": 0.9701,
      "step": 8164
    },
    {
      "epoch": 0.6598113093193801,
      "grad_norm": 2.549863576889038,
      "learning_rate": 7.820988338693694e-06,
      "loss": 0.9714,
      "step": 8165
    },
    {
      "epoch": 0.6598921190327077,
      "grad_norm": 2.6213786602020264,
      "learning_rate": 7.82044804714185e-06,
      "loss": 0.9224,
      "step": 8166
    },
    {
      "epoch": 0.6599729287460353,
      "grad_norm": 2.7831413745880127,
      "learning_rate": 7.81990770728252e-06,
      "loss": 0.8969,
      "step": 8167
    },
    {
      "epoch": 0.6600537384593628,
      "grad_norm": 2.559598922729492,
      "learning_rate": 7.819367319124958e-06,
      "loss": 1.0287,
      "step": 8168
    },
    {
      "epoch": 0.6601345481726903,
      "grad_norm": 2.988743543624878,
      "learning_rate": 7.818826882678423e-06,
      "loss": 0.9466,
      "step": 8169
    },
    {
      "epoch": 0.6602153578860179,
      "grad_norm": 2.464355945587158,
      "learning_rate": 7.818286397952168e-06,
      "loss": 1.0238,
      "step": 8170
    },
    {
      "epoch": 0.6602961675993454,
      "grad_norm": 2.460078001022339,
      "learning_rate": 7.817745864955452e-06,
      "loss": 1.0904,
      "step": 8171
    },
    {
      "epoch": 0.660376977312673,
      "grad_norm": 2.3599300384521484,
      "learning_rate": 7.817205283697535e-06,
      "loss": 1.0396,
      "step": 8172
    },
    {
      "epoch": 0.6604577870260006,
      "grad_norm": 2.6572680473327637,
      "learning_rate": 7.816664654187673e-06,
      "loss": 0.9153,
      "step": 8173
    },
    {
      "epoch": 0.660538596739328,
      "grad_norm": 2.1598360538482666,
      "learning_rate": 7.816123976435125e-06,
      "loss": 1.0339,
      "step": 8174
    },
    {
      "epoch": 0.6606194064526556,
      "grad_norm": 3.005772590637207,
      "learning_rate": 7.815583250449152e-06,
      "loss": 0.9471,
      "step": 8175
    },
    {
      "epoch": 0.6607002161659832,
      "grad_norm": 2.7757091522216797,
      "learning_rate": 7.815042476239018e-06,
      "loss": 0.898,
      "step": 8176
    },
    {
      "epoch": 0.6607810258793106,
      "grad_norm": 2.637239456176758,
      "learning_rate": 7.814501653813984e-06,
      "loss": 0.9423,
      "step": 8177
    },
    {
      "epoch": 0.6608618355926382,
      "grad_norm": 2.377521514892578,
      "learning_rate": 7.81396078318331e-06,
      "loss": 0.9358,
      "step": 8178
    },
    {
      "epoch": 0.6609426453059658,
      "grad_norm": 2.9557089805603027,
      "learning_rate": 7.813419864356264e-06,
      "loss": 0.9624,
      "step": 8179
    },
    {
      "epoch": 0.6610234550192933,
      "grad_norm": 2.4435203075408936,
      "learning_rate": 7.812878897342107e-06,
      "loss": 0.8897,
      "step": 8180
    },
    {
      "epoch": 0.6611042647326208,
      "grad_norm": 2.9663431644439697,
      "learning_rate": 7.812337882150108e-06,
      "loss": 0.8366,
      "step": 8181
    },
    {
      "epoch": 0.6611850744459484,
      "grad_norm": 3.0248186588287354,
      "learning_rate": 7.81179681878953e-06,
      "loss": 0.9345,
      "step": 8182
    },
    {
      "epoch": 0.6612658841592759,
      "grad_norm": 3.1231157779693604,
      "learning_rate": 7.811255707269642e-06,
      "loss": 1.0285,
      "step": 8183
    },
    {
      "epoch": 0.6613466938726035,
      "grad_norm": 2.6495866775512695,
      "learning_rate": 7.810714547599714e-06,
      "loss": 0.8142,
      "step": 8184
    },
    {
      "epoch": 0.661427503585931,
      "grad_norm": 2.3191978931427,
      "learning_rate": 7.81017333978901e-06,
      "loss": 0.9625,
      "step": 8185
    },
    {
      "epoch": 0.6615083132992585,
      "grad_norm": 2.4325387477874756,
      "learning_rate": 7.8096320838468e-06,
      "loss": 0.8434,
      "step": 8186
    },
    {
      "epoch": 0.6615891230125861,
      "grad_norm": 2.579108238220215,
      "learning_rate": 7.80909077978236e-06,
      "loss": 0.9275,
      "step": 8187
    },
    {
      "epoch": 0.6616699327259137,
      "grad_norm": 2.8837814331054688,
      "learning_rate": 7.808549427604955e-06,
      "loss": 1.0955,
      "step": 8188
    },
    {
      "epoch": 0.6617507424392411,
      "grad_norm": 2.698129415512085,
      "learning_rate": 7.80800802732386e-06,
      "loss": 0.9433,
      "step": 8189
    },
    {
      "epoch": 0.6618315521525687,
      "grad_norm": 2.9319772720336914,
      "learning_rate": 7.807466578948349e-06,
      "loss": 1.0015,
      "step": 8190
    },
    {
      "epoch": 0.6619123618658963,
      "grad_norm": 2.947528123855591,
      "learning_rate": 7.806925082487694e-06,
      "loss": 1.0496,
      "step": 8191
    },
    {
      "epoch": 0.6619931715792238,
      "grad_norm": 2.6251354217529297,
      "learning_rate": 7.806383537951169e-06,
      "loss": 0.9827,
      "step": 8192
    },
    {
      "epoch": 0.6620739812925514,
      "grad_norm": 2.5413177013397217,
      "learning_rate": 7.805841945348049e-06,
      "loss": 0.9414,
      "step": 8193
    },
    {
      "epoch": 0.6621547910058789,
      "grad_norm": 2.597057580947876,
      "learning_rate": 7.805300304687614e-06,
      "loss": 0.8435,
      "step": 8194
    },
    {
      "epoch": 0.6622356007192064,
      "grad_norm": 2.618741989135742,
      "learning_rate": 7.804758615979136e-06,
      "loss": 0.9952,
      "step": 8195
    },
    {
      "epoch": 0.662316410432534,
      "grad_norm": 2.9278066158294678,
      "learning_rate": 7.804216879231894e-06,
      "loss": 0.8114,
      "step": 8196
    },
    {
      "epoch": 0.6623972201458616,
      "grad_norm": 2.789865255355835,
      "learning_rate": 7.803675094455171e-06,
      "loss": 0.9201,
      "step": 8197
    },
    {
      "epoch": 0.662478029859189,
      "grad_norm": 3.073566436767578,
      "learning_rate": 7.803133261658242e-06,
      "loss": 0.9657,
      "step": 8198
    },
    {
      "epoch": 0.6625588395725166,
      "grad_norm": 2.890922784805298,
      "learning_rate": 7.802591380850386e-06,
      "loss": 1.0548,
      "step": 8199
    },
    {
      "epoch": 0.6626396492858442,
      "grad_norm": 2.835726737976074,
      "learning_rate": 7.80204945204089e-06,
      "loss": 0.8772,
      "step": 8200
    },
    {
      "epoch": 0.6627204589991716,
      "grad_norm": 2.594926357269287,
      "learning_rate": 7.801507475239032e-06,
      "loss": 0.9912,
      "step": 8201
    },
    {
      "epoch": 0.6628012687124992,
      "grad_norm": 2.3858537673950195,
      "learning_rate": 7.800965450454095e-06,
      "loss": 0.9948,
      "step": 8202
    },
    {
      "epoch": 0.6628820784258268,
      "grad_norm": 2.5020763874053955,
      "learning_rate": 7.800423377695363e-06,
      "loss": 0.9253,
      "step": 8203
    },
    {
      "epoch": 0.6629628881391543,
      "grad_norm": 2.653742790222168,
      "learning_rate": 7.799881256972118e-06,
      "loss": 0.8953,
      "step": 8204
    },
    {
      "epoch": 0.6630436978524819,
      "grad_norm": 2.5069143772125244,
      "learning_rate": 7.799339088293649e-06,
      "loss": 0.8956,
      "step": 8205
    },
    {
      "epoch": 0.6631245075658094,
      "grad_norm": 3.1007113456726074,
      "learning_rate": 7.798796871669242e-06,
      "loss": 0.9522,
      "step": 8206
    },
    {
      "epoch": 0.663205317279137,
      "grad_norm": 2.8476850986480713,
      "learning_rate": 7.79825460710818e-06,
      "loss": 0.9397,
      "step": 8207
    },
    {
      "epoch": 0.6632861269924645,
      "grad_norm": 2.517516851425171,
      "learning_rate": 7.797712294619754e-06,
      "loss": 1.0802,
      "step": 8208
    },
    {
      "epoch": 0.663366936705792,
      "grad_norm": 2.7888779640197754,
      "learning_rate": 7.797169934213253e-06,
      "loss": 0.9823,
      "step": 8209
    },
    {
      "epoch": 0.6634477464191196,
      "grad_norm": 2.2836861610412598,
      "learning_rate": 7.796627525897964e-06,
      "loss": 0.9072,
      "step": 8210
    },
    {
      "epoch": 0.6635285561324471,
      "grad_norm": 2.598681688308716,
      "learning_rate": 7.796085069683178e-06,
      "loss": 1.1079,
      "step": 8211
    },
    {
      "epoch": 0.6636093658457747,
      "grad_norm": 2.5902888774871826,
      "learning_rate": 7.795542565578187e-06,
      "loss": 0.8208,
      "step": 8212
    },
    {
      "epoch": 0.6636901755591023,
      "grad_norm": 3.1470422744750977,
      "learning_rate": 7.79500001359228e-06,
      "loss": 0.9458,
      "step": 8213
    },
    {
      "epoch": 0.6637709852724297,
      "grad_norm": 2.673978805541992,
      "learning_rate": 7.794457413734753e-06,
      "loss": 0.9343,
      "step": 8214
    },
    {
      "epoch": 0.6638517949857573,
      "grad_norm": 2.3911330699920654,
      "learning_rate": 7.793914766014898e-06,
      "loss": 0.8567,
      "step": 8215
    },
    {
      "epoch": 0.6639326046990849,
      "grad_norm": 2.661830186843872,
      "learning_rate": 7.793372070442007e-06,
      "loss": 1.0665,
      "step": 8216
    },
    {
      "epoch": 0.6640134144124124,
      "grad_norm": 2.4099466800689697,
      "learning_rate": 7.792829327025379e-06,
      "loss": 0.9569,
      "step": 8217
    },
    {
      "epoch": 0.6640942241257399,
      "grad_norm": 2.650916337966919,
      "learning_rate": 7.792286535774307e-06,
      "loss": 0.9633,
      "step": 8218
    },
    {
      "epoch": 0.6641750338390675,
      "grad_norm": 2.642181873321533,
      "learning_rate": 7.79174369669809e-06,
      "loss": 0.9002,
      "step": 8219
    },
    {
      "epoch": 0.664255843552395,
      "grad_norm": 2.3400585651397705,
      "learning_rate": 7.791200809806025e-06,
      "loss": 0.8768,
      "step": 8220
    },
    {
      "epoch": 0.6643366532657226,
      "grad_norm": 2.772515296936035,
      "learning_rate": 7.790657875107408e-06,
      "loss": 1.0094,
      "step": 8221
    },
    {
      "epoch": 0.6644174629790501,
      "grad_norm": 2.719703435897827,
      "learning_rate": 7.79011489261154e-06,
      "loss": 1.098,
      "step": 8222
    },
    {
      "epoch": 0.6644982726923776,
      "grad_norm": 2.950763463973999,
      "learning_rate": 7.789571862327721e-06,
      "loss": 1.0463,
      "step": 8223
    },
    {
      "epoch": 0.6645790824057052,
      "grad_norm": 3.0480339527130127,
      "learning_rate": 7.78902878426525e-06,
      "loss": 0.9207,
      "step": 8224
    },
    {
      "epoch": 0.6646598921190328,
      "grad_norm": 2.881671190261841,
      "learning_rate": 7.788485658433434e-06,
      "loss": 0.955,
      "step": 8225
    },
    {
      "epoch": 0.6647407018323602,
      "grad_norm": 2.436127185821533,
      "learning_rate": 7.78794248484157e-06,
      "loss": 0.8773,
      "step": 8226
    },
    {
      "epoch": 0.6648215115456878,
      "grad_norm": 2.9438748359680176,
      "learning_rate": 7.787399263498961e-06,
      "loss": 0.8433,
      "step": 8227
    },
    {
      "epoch": 0.6649023212590154,
      "grad_norm": 2.633422374725342,
      "learning_rate": 7.786855994414915e-06,
      "loss": 1.0823,
      "step": 8228
    },
    {
      "epoch": 0.6649831309723429,
      "grad_norm": 2.4525909423828125,
      "learning_rate": 7.786312677598736e-06,
      "loss": 0.9231,
      "step": 8229
    },
    {
      "epoch": 0.6650639406856704,
      "grad_norm": 2.606020212173462,
      "learning_rate": 7.785769313059726e-06,
      "loss": 0.8937,
      "step": 8230
    },
    {
      "epoch": 0.665144750398998,
      "grad_norm": 2.752110242843628,
      "learning_rate": 7.785225900807194e-06,
      "loss": 0.9587,
      "step": 8231
    },
    {
      "epoch": 0.6652255601123255,
      "grad_norm": 2.667178153991699,
      "learning_rate": 7.78468244085045e-06,
      "loss": 0.9951,
      "step": 8232
    },
    {
      "epoch": 0.665306369825653,
      "grad_norm": 2.447206497192383,
      "learning_rate": 7.784138933198798e-06,
      "loss": 1.066,
      "step": 8233
    },
    {
      "epoch": 0.6653871795389806,
      "grad_norm": 2.6468465328216553,
      "learning_rate": 7.78359537786155e-06,
      "loss": 0.8543,
      "step": 8234
    },
    {
      "epoch": 0.6654679892523081,
      "grad_norm": 2.8490002155303955,
      "learning_rate": 7.783051774848011e-06,
      "loss": 1.0329,
      "step": 8235
    },
    {
      "epoch": 0.6655487989656357,
      "grad_norm": 3.0898759365081787,
      "learning_rate": 7.782508124167499e-06,
      "loss": 0.8809,
      "step": 8236
    },
    {
      "epoch": 0.6656296086789633,
      "grad_norm": 2.563471794128418,
      "learning_rate": 7.78196442582932e-06,
      "loss": 0.9191,
      "step": 8237
    },
    {
      "epoch": 0.6657104183922907,
      "grad_norm": 2.8497154712677,
      "learning_rate": 7.781420679842787e-06,
      "loss": 1.063,
      "step": 8238
    },
    {
      "epoch": 0.6657912281056183,
      "grad_norm": 2.5441057682037354,
      "learning_rate": 7.780876886217215e-06,
      "loss": 0.806,
      "step": 8239
    },
    {
      "epoch": 0.6658720378189459,
      "grad_norm": 2.5067288875579834,
      "learning_rate": 7.780333044961916e-06,
      "loss": 0.8253,
      "step": 8240
    },
    {
      "epoch": 0.6659528475322734,
      "grad_norm": 3.736954927444458,
      "learning_rate": 7.779789156086203e-06,
      "loss": 0.9403,
      "step": 8241
    },
    {
      "epoch": 0.6660336572456009,
      "grad_norm": 3.1867945194244385,
      "learning_rate": 7.779245219599397e-06,
      "loss": 0.9003,
      "step": 8242
    },
    {
      "epoch": 0.6661144669589285,
      "grad_norm": 2.6131770610809326,
      "learning_rate": 7.778701235510811e-06,
      "loss": 0.9347,
      "step": 8243
    },
    {
      "epoch": 0.666195276672256,
      "grad_norm": 2.5567126274108887,
      "learning_rate": 7.778157203829761e-06,
      "loss": 0.814,
      "step": 8244
    },
    {
      "epoch": 0.6662760863855836,
      "grad_norm": 2.6412107944488525,
      "learning_rate": 7.777613124565567e-06,
      "loss": 0.9403,
      "step": 8245
    },
    {
      "epoch": 0.6663568960989111,
      "grad_norm": 2.68034291267395,
      "learning_rate": 7.777068997727547e-06,
      "loss": 0.9767,
      "step": 8246
    },
    {
      "epoch": 0.6664377058122386,
      "grad_norm": 2.3616299629211426,
      "learning_rate": 7.77652482332502e-06,
      "loss": 1.0433,
      "step": 8247
    },
    {
      "epoch": 0.6665185155255662,
      "grad_norm": 2.2110915184020996,
      "learning_rate": 7.775980601367307e-06,
      "loss": 1.0506,
      "step": 8248
    },
    {
      "epoch": 0.6665993252388938,
      "grad_norm": 2.728952407836914,
      "learning_rate": 7.775436331863731e-06,
      "loss": 0.9137,
      "step": 8249
    },
    {
      "epoch": 0.6666801349522212,
      "grad_norm": 2.613842487335205,
      "learning_rate": 7.774892014823609e-06,
      "loss": 1.0901,
      "step": 8250
    },
    {
      "epoch": 0.6667609446655488,
      "grad_norm": 2.415443181991577,
      "learning_rate": 7.774347650256268e-06,
      "loss": 0.9084,
      "step": 8251
    },
    {
      "epoch": 0.6668417543788764,
      "grad_norm": 2.3626067638397217,
      "learning_rate": 7.773803238171031e-06,
      "loss": 0.9239,
      "step": 8252
    },
    {
      "epoch": 0.6669225640922039,
      "grad_norm": 2.2385334968566895,
      "learning_rate": 7.773258778577224e-06,
      "loss": 1.0494,
      "step": 8253
    },
    {
      "epoch": 0.6670033738055314,
      "grad_norm": 2.796081304550171,
      "learning_rate": 7.772714271484169e-06,
      "loss": 0.9133,
      "step": 8254
    },
    {
      "epoch": 0.667084183518859,
      "grad_norm": 2.7769646644592285,
      "learning_rate": 7.772169716901194e-06,
      "loss": 0.9149,
      "step": 8255
    },
    {
      "epoch": 0.6671649932321865,
      "grad_norm": 2.5033700466156006,
      "learning_rate": 7.771625114837625e-06,
      "loss": 1.0176,
      "step": 8256
    },
    {
      "epoch": 0.667245802945514,
      "grad_norm": 2.467130422592163,
      "learning_rate": 7.77108046530279e-06,
      "loss": 1.0605,
      "step": 8257
    },
    {
      "epoch": 0.6673266126588416,
      "grad_norm": 2.6402151584625244,
      "learning_rate": 7.770535768306019e-06,
      "loss": 0.9663,
      "step": 8258
    },
    {
      "epoch": 0.6674074223721691,
      "grad_norm": 2.4942843914031982,
      "learning_rate": 7.76999102385664e-06,
      "loss": 1.0421,
      "step": 8259
    },
    {
      "epoch": 0.6674882320854967,
      "grad_norm": 2.8600966930389404,
      "learning_rate": 7.769446231963982e-06,
      "loss": 0.9089,
      "step": 8260
    },
    {
      "epoch": 0.6675690417988243,
      "grad_norm": 2.88496470451355,
      "learning_rate": 7.768901392637378e-06,
      "loss": 0.9567,
      "step": 8261
    },
    {
      "epoch": 0.6676498515121517,
      "grad_norm": 2.750972032546997,
      "learning_rate": 7.768356505886158e-06,
      "loss": 0.9366,
      "step": 8262
    },
    {
      "epoch": 0.6677306612254793,
      "grad_norm": 3.0355007648468018,
      "learning_rate": 7.767811571719657e-06,
      "loss": 0.9799,
      "step": 8263
    },
    {
      "epoch": 0.6678114709388069,
      "grad_norm": 2.573878765106201,
      "learning_rate": 7.767266590147205e-06,
      "loss": 0.9333,
      "step": 8264
    },
    {
      "epoch": 0.6678922806521344,
      "grad_norm": 2.9451887607574463,
      "learning_rate": 7.76672156117814e-06,
      "loss": 0.948,
      "step": 8265
    },
    {
      "epoch": 0.6679730903654619,
      "grad_norm": 2.7100412845611572,
      "learning_rate": 7.766176484821794e-06,
      "loss": 0.9268,
      "step": 8266
    },
    {
      "epoch": 0.6680539000787895,
      "grad_norm": 2.3908088207244873,
      "learning_rate": 7.765631361087507e-06,
      "loss": 0.8855,
      "step": 8267
    },
    {
      "epoch": 0.668134709792117,
      "grad_norm": 2.676058053970337,
      "learning_rate": 7.765086189984609e-06,
      "loss": 0.9573,
      "step": 8268
    },
    {
      "epoch": 0.6682155195054446,
      "grad_norm": 2.635659694671631,
      "learning_rate": 7.764540971522443e-06,
      "loss": 1.0797,
      "step": 8269
    },
    {
      "epoch": 0.6682963292187721,
      "grad_norm": 2.4136404991149902,
      "learning_rate": 7.763995705710345e-06,
      "loss": 0.9517,
      "step": 8270
    },
    {
      "epoch": 0.6683771389320996,
      "grad_norm": 2.9774701595306396,
      "learning_rate": 7.763450392557656e-06,
      "loss": 0.9541,
      "step": 8271
    },
    {
      "epoch": 0.6684579486454272,
      "grad_norm": 2.8616385459899902,
      "learning_rate": 7.762905032073712e-06,
      "loss": 1.0143,
      "step": 8272
    },
    {
      "epoch": 0.6685387583587548,
      "grad_norm": 3.378654956817627,
      "learning_rate": 7.76235962426786e-06,
      "loss": 0.9294,
      "step": 8273
    },
    {
      "epoch": 0.6686195680720822,
      "grad_norm": 2.612964630126953,
      "learning_rate": 7.761814169149436e-06,
      "loss": 0.9196,
      "step": 8274
    },
    {
      "epoch": 0.6687003777854098,
      "grad_norm": 2.588641405105591,
      "learning_rate": 7.761268666727782e-06,
      "loss": 0.9922,
      "step": 8275
    },
    {
      "epoch": 0.6687811874987374,
      "grad_norm": 2.8573877811431885,
      "learning_rate": 7.760723117012245e-06,
      "loss": 0.8514,
      "step": 8276
    },
    {
      "epoch": 0.6688619972120649,
      "grad_norm": 2.6293280124664307,
      "learning_rate": 7.760177520012167e-06,
      "loss": 0.8541,
      "step": 8277
    },
    {
      "epoch": 0.6689428069253924,
      "grad_norm": 2.5145294666290283,
      "learning_rate": 7.759631875736892e-06,
      "loss": 0.854,
      "step": 8278
    },
    {
      "epoch": 0.66902361663872,
      "grad_norm": 2.9591727256774902,
      "learning_rate": 7.75908618419577e-06,
      "loss": 0.9361,
      "step": 8279
    },
    {
      "epoch": 0.6691044263520475,
      "grad_norm": 2.414062023162842,
      "learning_rate": 7.75854044539814e-06,
      "loss": 0.8483,
      "step": 8280
    },
    {
      "epoch": 0.669185236065375,
      "grad_norm": 2.8749969005584717,
      "learning_rate": 7.757994659353354e-06,
      "loss": 0.9572,
      "step": 8281
    },
    {
      "epoch": 0.6692660457787026,
      "grad_norm": 2.8766791820526123,
      "learning_rate": 7.757448826070761e-06,
      "loss": 1.0735,
      "step": 8282
    },
    {
      "epoch": 0.6693468554920301,
      "grad_norm": 2.4536221027374268,
      "learning_rate": 7.756902945559705e-06,
      "loss": 0.9159,
      "step": 8283
    },
    {
      "epoch": 0.6694276652053577,
      "grad_norm": 3.069279193878174,
      "learning_rate": 7.75635701782954e-06,
      "loss": 0.9608,
      "step": 8284
    },
    {
      "epoch": 0.6695084749186853,
      "grad_norm": 2.756054639816284,
      "learning_rate": 7.755811042889615e-06,
      "loss": 1.0801,
      "step": 8285
    },
    {
      "epoch": 0.6695892846320127,
      "grad_norm": 2.6374106407165527,
      "learning_rate": 7.755265020749281e-06,
      "loss": 1.0975,
      "step": 8286
    },
    {
      "epoch": 0.6696700943453403,
      "grad_norm": 2.7474169731140137,
      "learning_rate": 7.75471895141789e-06,
      "loss": 0.8723,
      "step": 8287
    },
    {
      "epoch": 0.6697509040586679,
      "grad_norm": 2.4243593215942383,
      "learning_rate": 7.754172834904797e-06,
      "loss": 0.9728,
      "step": 8288
    },
    {
      "epoch": 0.6698317137719954,
      "grad_norm": 2.5193612575531006,
      "learning_rate": 7.753626671219352e-06,
      "loss": 1.0481,
      "step": 8289
    },
    {
      "epoch": 0.6699125234853229,
      "grad_norm": 2.3451879024505615,
      "learning_rate": 7.753080460370912e-06,
      "loss": 1.0229,
      "step": 8290
    },
    {
      "epoch": 0.6699933331986505,
      "grad_norm": 4.024284362792969,
      "learning_rate": 7.75253420236883e-06,
      "loss": 0.7598,
      "step": 8291
    },
    {
      "epoch": 0.670074142911978,
      "grad_norm": 2.246476888656616,
      "learning_rate": 7.751987897222464e-06,
      "loss": 1.0429,
      "step": 8292
    },
    {
      "epoch": 0.6701549526253056,
      "grad_norm": 2.261608600616455,
      "learning_rate": 7.751441544941171e-06,
      "loss": 0.9731,
      "step": 8293
    },
    {
      "epoch": 0.6702357623386331,
      "grad_norm": 3.0265629291534424,
      "learning_rate": 7.750895145534308e-06,
      "loss": 1.0262,
      "step": 8294
    },
    {
      "epoch": 0.6703165720519606,
      "grad_norm": 2.9487316608428955,
      "learning_rate": 7.750348699011233e-06,
      "loss": 0.8359,
      "step": 8295
    },
    {
      "epoch": 0.6703973817652882,
      "grad_norm": 2.4979705810546875,
      "learning_rate": 7.749802205381307e-06,
      "loss": 0.9042,
      "step": 8296
    },
    {
      "epoch": 0.6704781914786158,
      "grad_norm": 2.4784200191497803,
      "learning_rate": 7.749255664653888e-06,
      "loss": 0.9499,
      "step": 8297
    },
    {
      "epoch": 0.6705590011919432,
      "grad_norm": 2.55683970451355,
      "learning_rate": 7.748709076838338e-06,
      "loss": 0.9525,
      "step": 8298
    },
    {
      "epoch": 0.6706398109052708,
      "grad_norm": 2.722597122192383,
      "learning_rate": 7.74816244194402e-06,
      "loss": 0.9261,
      "step": 8299
    },
    {
      "epoch": 0.6707206206185984,
      "grad_norm": 2.730271816253662,
      "learning_rate": 7.747615759980296e-06,
      "loss": 1.0247,
      "step": 8300
    },
    {
      "epoch": 0.6708014303319259,
      "grad_norm": 2.806570291519165,
      "learning_rate": 7.747069030956526e-06,
      "loss": 0.9996,
      "step": 8301
    },
    {
      "epoch": 0.6708822400452534,
      "grad_norm": 2.540257215499878,
      "learning_rate": 7.746522254882078e-06,
      "loss": 0.9455,
      "step": 8302
    },
    {
      "epoch": 0.670963049758581,
      "grad_norm": 2.505619525909424,
      "learning_rate": 7.745975431766317e-06,
      "loss": 0.9975,
      "step": 8303
    },
    {
      "epoch": 0.6710438594719085,
      "grad_norm": 3.088472366333008,
      "learning_rate": 7.745428561618606e-06,
      "loss": 0.9773,
      "step": 8304
    },
    {
      "epoch": 0.6711246691852361,
      "grad_norm": 2.639523506164551,
      "learning_rate": 7.744881644448315e-06,
      "loss": 0.9636,
      "step": 8305
    },
    {
      "epoch": 0.6712054788985636,
      "grad_norm": 3.1845033168792725,
      "learning_rate": 7.744334680264807e-06,
      "loss": 0.9452,
      "step": 8306
    },
    {
      "epoch": 0.6712862886118911,
      "grad_norm": 2.8969013690948486,
      "learning_rate": 7.743787669077454e-06,
      "loss": 1.0315,
      "step": 8307
    },
    {
      "epoch": 0.6713670983252187,
      "grad_norm": 2.420459270477295,
      "learning_rate": 7.743240610895623e-06,
      "loss": 1.0473,
      "step": 8308
    },
    {
      "epoch": 0.6714479080385463,
      "grad_norm": 2.644023895263672,
      "learning_rate": 7.742693505728684e-06,
      "loss": 0.795,
      "step": 8309
    },
    {
      "epoch": 0.6715287177518737,
      "grad_norm": 2.8094165325164795,
      "learning_rate": 7.74214635358601e-06,
      "loss": 1.0484,
      "step": 8310
    },
    {
      "epoch": 0.6716095274652013,
      "grad_norm": 2.2670786380767822,
      "learning_rate": 7.741599154476969e-06,
      "loss": 1.0099,
      "step": 8311
    },
    {
      "epoch": 0.6716903371785289,
      "grad_norm": 2.4911980628967285,
      "learning_rate": 7.741051908410935e-06,
      "loss": 0.905,
      "step": 8312
    },
    {
      "epoch": 0.6717711468918564,
      "grad_norm": 2.782815933227539,
      "learning_rate": 7.74050461539728e-06,
      "loss": 0.8601,
      "step": 8313
    },
    {
      "epoch": 0.6718519566051839,
      "grad_norm": 2.432807207107544,
      "learning_rate": 7.73995727544538e-06,
      "loss": 0.8356,
      "step": 8314
    },
    {
      "epoch": 0.6719327663185115,
      "grad_norm": 2.682298183441162,
      "learning_rate": 7.739409888564606e-06,
      "loss": 1.0302,
      "step": 8315
    },
    {
      "epoch": 0.672013576031839,
      "grad_norm": 2.6344592571258545,
      "learning_rate": 7.738862454764336e-06,
      "loss": 0.9202,
      "step": 8316
    },
    {
      "epoch": 0.6720943857451666,
      "grad_norm": 2.039740800857544,
      "learning_rate": 7.738314974053947e-06,
      "loss": 1.1516,
      "step": 8317
    },
    {
      "epoch": 0.6721751954584941,
      "grad_norm": 2.588088274002075,
      "learning_rate": 7.737767446442815e-06,
      "loss": 0.9694,
      "step": 8318
    },
    {
      "epoch": 0.6722560051718216,
      "grad_norm": 2.552767515182495,
      "learning_rate": 7.737219871940315e-06,
      "loss": 0.9992,
      "step": 8319
    },
    {
      "epoch": 0.6723368148851492,
      "grad_norm": 2.126105785369873,
      "learning_rate": 7.73667225055583e-06,
      "loss": 0.8482,
      "step": 8320
    },
    {
      "epoch": 0.6724176245984768,
      "grad_norm": 2.6231529712677,
      "learning_rate": 7.736124582298737e-06,
      "loss": 0.9491,
      "step": 8321
    },
    {
      "epoch": 0.6724984343118042,
      "grad_norm": 2.7928731441497803,
      "learning_rate": 7.735576867178417e-06,
      "loss": 0.9421,
      "step": 8322
    },
    {
      "epoch": 0.6725792440251318,
      "grad_norm": 3.0163440704345703,
      "learning_rate": 7.73502910520425e-06,
      "loss": 0.8872,
      "step": 8323
    },
    {
      "epoch": 0.6726600537384594,
      "grad_norm": 3.2160558700561523,
      "learning_rate": 7.73448129638562e-06,
      "loss": 0.998,
      "step": 8324
    },
    {
      "epoch": 0.6727408634517869,
      "grad_norm": 2.5027525424957275,
      "learning_rate": 7.73393344073191e-06,
      "loss": 0.8043,
      "step": 8325
    },
    {
      "epoch": 0.6728216731651144,
      "grad_norm": 2.5036063194274902,
      "learning_rate": 7.733385538252497e-06,
      "loss": 0.8813,
      "step": 8326
    },
    {
      "epoch": 0.672902482878442,
      "grad_norm": 3.024444818496704,
      "learning_rate": 7.732837588956775e-06,
      "loss": 0.8521,
      "step": 8327
    },
    {
      "epoch": 0.6729832925917695,
      "grad_norm": 2.3691868782043457,
      "learning_rate": 7.73228959285412e-06,
      "loss": 0.9211,
      "step": 8328
    },
    {
      "epoch": 0.6730641023050971,
      "grad_norm": 2.5722193717956543,
      "learning_rate": 7.731741549953927e-06,
      "loss": 0.9543,
      "step": 8329
    },
    {
      "epoch": 0.6731449120184246,
      "grad_norm": 3.1957387924194336,
      "learning_rate": 7.731193460265573e-06,
      "loss": 0.9296,
      "step": 8330
    },
    {
      "epoch": 0.6732257217317521,
      "grad_norm": 2.7510924339294434,
      "learning_rate": 7.730645323798451e-06,
      "loss": 0.9251,
      "step": 8331
    },
    {
      "epoch": 0.6733065314450797,
      "grad_norm": 2.670241594314575,
      "learning_rate": 7.730097140561949e-06,
      "loss": 0.8923,
      "step": 8332
    },
    {
      "epoch": 0.6733873411584073,
      "grad_norm": 2.8305563926696777,
      "learning_rate": 7.729548910565457e-06,
      "loss": 0.9963,
      "step": 8333
    },
    {
      "epoch": 0.6734681508717347,
      "grad_norm": 2.385990619659424,
      "learning_rate": 7.729000633818363e-06,
      "loss": 0.9195,
      "step": 8334
    },
    {
      "epoch": 0.6735489605850623,
      "grad_norm": 3.2661023139953613,
      "learning_rate": 7.728452310330055e-06,
      "loss": 0.9617,
      "step": 8335
    },
    {
      "epoch": 0.6736297702983899,
      "grad_norm": 2.773858070373535,
      "learning_rate": 7.72790394010993e-06,
      "loss": 0.8973,
      "step": 8336
    },
    {
      "epoch": 0.6737105800117175,
      "grad_norm": 2.914680242538452,
      "learning_rate": 7.727355523167378e-06,
      "loss": 0.9032,
      "step": 8337
    },
    {
      "epoch": 0.6737913897250449,
      "grad_norm": 2.5987532138824463,
      "learning_rate": 7.726807059511789e-06,
      "loss": 0.9761,
      "step": 8338
    },
    {
      "epoch": 0.6738721994383725,
      "grad_norm": 2.3974978923797607,
      "learning_rate": 7.72625854915256e-06,
      "loss": 0.9097,
      "step": 8339
    },
    {
      "epoch": 0.6739530091517001,
      "grad_norm": 2.736830472946167,
      "learning_rate": 7.725709992099089e-06,
      "loss": 1.0532,
      "step": 8340
    },
    {
      "epoch": 0.6740338188650276,
      "grad_norm": 2.79888653755188,
      "learning_rate": 7.725161388360764e-06,
      "loss": 1.0349,
      "step": 8341
    },
    {
      "epoch": 0.6741146285783551,
      "grad_norm": 2.7195255756378174,
      "learning_rate": 7.724612737946986e-06,
      "loss": 0.9932,
      "step": 8342
    },
    {
      "epoch": 0.6741954382916827,
      "grad_norm": 2.5283217430114746,
      "learning_rate": 7.724064040867152e-06,
      "loss": 0.9653,
      "step": 8343
    },
    {
      "epoch": 0.6742762480050102,
      "grad_norm": 2.5808815956115723,
      "learning_rate": 7.723515297130656e-06,
      "loss": 0.9035,
      "step": 8344
    },
    {
      "epoch": 0.6743570577183378,
      "grad_norm": 2.767984390258789,
      "learning_rate": 7.722966506746902e-06,
      "loss": 0.9965,
      "step": 8345
    },
    {
      "epoch": 0.6744378674316653,
      "grad_norm": 2.6212093830108643,
      "learning_rate": 7.722417669725288e-06,
      "loss": 0.9327,
      "step": 8346
    },
    {
      "epoch": 0.6745186771449928,
      "grad_norm": 2.387342929840088,
      "learning_rate": 7.721868786075212e-06,
      "loss": 0.9695,
      "step": 8347
    },
    {
      "epoch": 0.6745994868583204,
      "grad_norm": 2.599292516708374,
      "learning_rate": 7.721319855806076e-06,
      "loss": 1.0379,
      "step": 8348
    },
    {
      "epoch": 0.674680296571648,
      "grad_norm": 2.787990093231201,
      "learning_rate": 7.720770878927284e-06,
      "loss": 0.9756,
      "step": 8349
    },
    {
      "epoch": 0.6747611062849754,
      "grad_norm": 2.2481515407562256,
      "learning_rate": 7.720221855448233e-06,
      "loss": 0.9603,
      "step": 8350
    },
    {
      "epoch": 0.674841915998303,
      "grad_norm": 2.836638927459717,
      "learning_rate": 7.719672785378334e-06,
      "loss": 0.8768,
      "step": 8351
    },
    {
      "epoch": 0.6749227257116306,
      "grad_norm": 2.7812743186950684,
      "learning_rate": 7.719123668726986e-06,
      "loss": 1.0855,
      "step": 8352
    },
    {
      "epoch": 0.6750035354249581,
      "grad_norm": 3.177708387374878,
      "learning_rate": 7.718574505503596e-06,
      "loss": 0.9568,
      "step": 8353
    },
    {
      "epoch": 0.6750843451382856,
      "grad_norm": 2.8781800270080566,
      "learning_rate": 7.718025295717569e-06,
      "loss": 0.9938,
      "step": 8354
    },
    {
      "epoch": 0.6751651548516132,
      "grad_norm": 2.4317760467529297,
      "learning_rate": 7.71747603937831e-06,
      "loss": 0.9285,
      "step": 8355
    },
    {
      "epoch": 0.6752459645649407,
      "grad_norm": 2.6491196155548096,
      "learning_rate": 7.716926736495232e-06,
      "loss": 0.9991,
      "step": 8356
    },
    {
      "epoch": 0.6753267742782683,
      "grad_norm": 2.961341619491577,
      "learning_rate": 7.71637738707774e-06,
      "loss": 1.0286,
      "step": 8357
    },
    {
      "epoch": 0.6754075839915958,
      "grad_norm": 2.510984182357788,
      "learning_rate": 7.715827991135241e-06,
      "loss": 0.8888,
      "step": 8358
    },
    {
      "epoch": 0.6754883937049233,
      "grad_norm": 2.8395159244537354,
      "learning_rate": 7.715278548677145e-06,
      "loss": 0.9294,
      "step": 8359
    },
    {
      "epoch": 0.6755692034182509,
      "grad_norm": 2.6775362491607666,
      "learning_rate": 7.714729059712869e-06,
      "loss": 1.0098,
      "step": 8360
    },
    {
      "epoch": 0.6756500131315785,
      "grad_norm": 2.8229544162750244,
      "learning_rate": 7.714179524251814e-06,
      "loss": 0.8834,
      "step": 8361
    },
    {
      "epoch": 0.6757308228449059,
      "grad_norm": 2.85675048828125,
      "learning_rate": 7.713629942303403e-06,
      "loss": 0.7929,
      "step": 8362
    },
    {
      "epoch": 0.6758116325582335,
      "grad_norm": 2.634467124938965,
      "learning_rate": 7.71308031387704e-06,
      "loss": 0.9123,
      "step": 8363
    },
    {
      "epoch": 0.6758924422715611,
      "grad_norm": 2.678504467010498,
      "learning_rate": 7.712530638982143e-06,
      "loss": 0.907,
      "step": 8364
    },
    {
      "epoch": 0.6759732519848886,
      "grad_norm": 2.7769691944122314,
      "learning_rate": 7.711980917628128e-06,
      "loss": 0.8632,
      "step": 8365
    },
    {
      "epoch": 0.6760540616982161,
      "grad_norm": 2.5603690147399902,
      "learning_rate": 7.71143114982441e-06,
      "loss": 0.9156,
      "step": 8366
    },
    {
      "epoch": 0.6761348714115437,
      "grad_norm": 2.760061025619507,
      "learning_rate": 7.7108813355804e-06,
      "loss": 0.8569,
      "step": 8367
    },
    {
      "epoch": 0.6762156811248712,
      "grad_norm": 2.8086774349212646,
      "learning_rate": 7.710331474905522e-06,
      "loss": 0.9045,
      "step": 8368
    },
    {
      "epoch": 0.6762964908381988,
      "grad_norm": 2.4690489768981934,
      "learning_rate": 7.709781567809188e-06,
      "loss": 0.9131,
      "step": 8369
    },
    {
      "epoch": 0.6763773005515263,
      "grad_norm": 2.483729600906372,
      "learning_rate": 7.709231614300823e-06,
      "loss": 0.9583,
      "step": 8370
    },
    {
      "epoch": 0.6764581102648538,
      "grad_norm": 2.705441951751709,
      "learning_rate": 7.708681614389838e-06,
      "loss": 1.0647,
      "step": 8371
    },
    {
      "epoch": 0.6765389199781814,
      "grad_norm": 2.5196518898010254,
      "learning_rate": 7.708131568085661e-06,
      "loss": 0.8732,
      "step": 8372
    },
    {
      "epoch": 0.676619729691509,
      "grad_norm": 2.6411266326904297,
      "learning_rate": 7.70758147539771e-06,
      "loss": 0.8754,
      "step": 8373
    },
    {
      "epoch": 0.6767005394048364,
      "grad_norm": 2.3072762489318848,
      "learning_rate": 7.707031336335407e-06,
      "loss": 0.9353,
      "step": 8374
    },
    {
      "epoch": 0.676781349118164,
      "grad_norm": 2.587378978729248,
      "learning_rate": 7.706481150908172e-06,
      "loss": 0.8209,
      "step": 8375
    },
    {
      "epoch": 0.6768621588314916,
      "grad_norm": 2.600463628768921,
      "learning_rate": 7.70593091912543e-06,
      "loss": 0.9594,
      "step": 8376
    },
    {
      "epoch": 0.6769429685448191,
      "grad_norm": 2.5489156246185303,
      "learning_rate": 7.705380640996609e-06,
      "loss": 1.0448,
      "step": 8377
    },
    {
      "epoch": 0.6770237782581466,
      "grad_norm": 2.501147747039795,
      "learning_rate": 7.704830316531128e-06,
      "loss": 1.0695,
      "step": 8378
    },
    {
      "epoch": 0.6771045879714742,
      "grad_norm": 3.2265758514404297,
      "learning_rate": 7.704279945738416e-06,
      "loss": 0.9487,
      "step": 8379
    },
    {
      "epoch": 0.6771853976848017,
      "grad_norm": 2.878840684890747,
      "learning_rate": 7.703729528627899e-06,
      "loss": 0.9795,
      "step": 8380
    },
    {
      "epoch": 0.6772662073981293,
      "grad_norm": 2.609241485595703,
      "learning_rate": 7.703179065209003e-06,
      "loss": 1.0045,
      "step": 8381
    },
    {
      "epoch": 0.6773470171114568,
      "grad_norm": 2.75795841217041,
      "learning_rate": 7.702628555491159e-06,
      "loss": 0.9914,
      "step": 8382
    },
    {
      "epoch": 0.6774278268247843,
      "grad_norm": 2.7188820838928223,
      "learning_rate": 7.702077999483793e-06,
      "loss": 0.7822,
      "step": 8383
    },
    {
      "epoch": 0.6775086365381119,
      "grad_norm": 2.3923799991607666,
      "learning_rate": 7.701527397196336e-06,
      "loss": 0.8113,
      "step": 8384
    },
    {
      "epoch": 0.6775894462514395,
      "grad_norm": 2.942815065383911,
      "learning_rate": 7.700976748638218e-06,
      "loss": 1.0804,
      "step": 8385
    },
    {
      "epoch": 0.6776702559647669,
      "grad_norm": 3.0890114307403564,
      "learning_rate": 7.70042605381887e-06,
      "loss": 0.9667,
      "step": 8386
    },
    {
      "epoch": 0.6777510656780945,
      "grad_norm": 2.379671335220337,
      "learning_rate": 7.699875312747726e-06,
      "loss": 1.0517,
      "step": 8387
    },
    {
      "epoch": 0.6778318753914221,
      "grad_norm": 2.3722407817840576,
      "learning_rate": 7.699324525434217e-06,
      "loss": 0.9749,
      "step": 8388
    },
    {
      "epoch": 0.6779126851047496,
      "grad_norm": 2.7221665382385254,
      "learning_rate": 7.698773691887778e-06,
      "loss": 0.9211,
      "step": 8389
    },
    {
      "epoch": 0.6779934948180771,
      "grad_norm": 2.7167699337005615,
      "learning_rate": 7.698222812117843e-06,
      "loss": 0.9072,
      "step": 8390
    },
    {
      "epoch": 0.6780743045314047,
      "grad_norm": 2.2620158195495605,
      "learning_rate": 7.697671886133846e-06,
      "loss": 0.9907,
      "step": 8391
    },
    {
      "epoch": 0.6781551142447322,
      "grad_norm": 2.5893588066101074,
      "learning_rate": 7.697120913945224e-06,
      "loss": 0.9744,
      "step": 8392
    },
    {
      "epoch": 0.6782359239580598,
      "grad_norm": 2.9459714889526367,
      "learning_rate": 7.696569895561415e-06,
      "loss": 1.009,
      "step": 8393
    },
    {
      "epoch": 0.6783167336713873,
      "grad_norm": 3.053457260131836,
      "learning_rate": 7.696018830991852e-06,
      "loss": 0.8888,
      "step": 8394
    },
    {
      "epoch": 0.6783975433847148,
      "grad_norm": 2.751218557357788,
      "learning_rate": 7.69546772024598e-06,
      "loss": 1.0176,
      "step": 8395
    },
    {
      "epoch": 0.6784783530980424,
      "grad_norm": 2.663726329803467,
      "learning_rate": 7.694916563333234e-06,
      "loss": 0.968,
      "step": 8396
    },
    {
      "epoch": 0.67855916281137,
      "grad_norm": 2.3502562046051025,
      "learning_rate": 7.694365360263055e-06,
      "loss": 0.8048,
      "step": 8397
    },
    {
      "epoch": 0.6786399725246974,
      "grad_norm": 2.6676876544952393,
      "learning_rate": 7.693814111044885e-06,
      "loss": 0.8791,
      "step": 8398
    },
    {
      "epoch": 0.678720782238025,
      "grad_norm": 2.502436876296997,
      "learning_rate": 7.693262815688163e-06,
      "loss": 0.9708,
      "step": 8399
    },
    {
      "epoch": 0.6788015919513526,
      "grad_norm": 2.1671106815338135,
      "learning_rate": 7.692711474202334e-06,
      "loss": 0.8879,
      "step": 8400
    },
    {
      "epoch": 0.6788824016646801,
      "grad_norm": 3.063915491104126,
      "learning_rate": 7.692160086596838e-06,
      "loss": 1.0006,
      "step": 8401
    },
    {
      "epoch": 0.6789632113780076,
      "grad_norm": 2.5574464797973633,
      "learning_rate": 7.691608652881122e-06,
      "loss": 0.9383,
      "step": 8402
    },
    {
      "epoch": 0.6790440210913352,
      "grad_norm": 2.441051721572876,
      "learning_rate": 7.691057173064629e-06,
      "loss": 0.9807,
      "step": 8403
    },
    {
      "epoch": 0.6791248308046627,
      "grad_norm": 3.3872270584106445,
      "learning_rate": 7.690505647156806e-06,
      "loss": 0.9922,
      "step": 8404
    },
    {
      "epoch": 0.6792056405179903,
      "grad_norm": 2.5608327388763428,
      "learning_rate": 7.689954075167098e-06,
      "loss": 0.9392,
      "step": 8405
    },
    {
      "epoch": 0.6792864502313178,
      "grad_norm": 2.5460565090179443,
      "learning_rate": 7.689402457104954e-06,
      "loss": 1.1762,
      "step": 8406
    },
    {
      "epoch": 0.6793672599446453,
      "grad_norm": 2.524811267852783,
      "learning_rate": 7.688850792979816e-06,
      "loss": 0.8042,
      "step": 8407
    },
    {
      "epoch": 0.6794480696579729,
      "grad_norm": 2.9841954708099365,
      "learning_rate": 7.688299082801141e-06,
      "loss": 0.9428,
      "step": 8408
    },
    {
      "epoch": 0.6795288793713005,
      "grad_norm": 3.026340961456299,
      "learning_rate": 7.687747326578374e-06,
      "loss": 1.009,
      "step": 8409
    },
    {
      "epoch": 0.6796096890846279,
      "grad_norm": 2.856476068496704,
      "learning_rate": 7.687195524320965e-06,
      "loss": 0.9688,
      "step": 8410
    },
    {
      "epoch": 0.6796904987979555,
      "grad_norm": 2.6018548011779785,
      "learning_rate": 7.686643676038364e-06,
      "loss": 0.8525,
      "step": 8411
    },
    {
      "epoch": 0.6797713085112831,
      "grad_norm": 2.567927598953247,
      "learning_rate": 7.686091781740027e-06,
      "loss": 0.9956,
      "step": 8412
    },
    {
      "epoch": 0.6798521182246106,
      "grad_norm": 2.6431033611297607,
      "learning_rate": 7.685539841435406e-06,
      "loss": 0.9945,
      "step": 8413
    },
    {
      "epoch": 0.6799329279379381,
      "grad_norm": 2.5504331588745117,
      "learning_rate": 7.68498785513395e-06,
      "loss": 1.0153,
      "step": 8414
    },
    {
      "epoch": 0.6800137376512657,
      "grad_norm": 2.590113639831543,
      "learning_rate": 7.684435822845115e-06,
      "loss": 0.9778,
      "step": 8415
    },
    {
      "epoch": 0.6800945473645932,
      "grad_norm": 2.3256642818450928,
      "learning_rate": 7.683883744578359e-06,
      "loss": 1.0757,
      "step": 8416
    },
    {
      "epoch": 0.6801753570779208,
      "grad_norm": 3.3201260566711426,
      "learning_rate": 7.683331620343135e-06,
      "loss": 1.1739,
      "step": 8417
    },
    {
      "epoch": 0.6802561667912483,
      "grad_norm": 2.8348820209503174,
      "learning_rate": 7.6827794501489e-06,
      "loss": 0.9597,
      "step": 8418
    },
    {
      "epoch": 0.6803369765045758,
      "grad_norm": 2.7646586894989014,
      "learning_rate": 7.682227234005113e-06,
      "loss": 1.0271,
      "step": 8419
    },
    {
      "epoch": 0.6804177862179034,
      "grad_norm": 2.6002559661865234,
      "learning_rate": 7.681674971921227e-06,
      "loss": 0.9492,
      "step": 8420
    },
    {
      "epoch": 0.680498595931231,
      "grad_norm": 2.860055923461914,
      "learning_rate": 7.681122663906708e-06,
      "loss": 0.9503,
      "step": 8421
    },
    {
      "epoch": 0.6805794056445584,
      "grad_norm": 2.656940221786499,
      "learning_rate": 7.680570309971011e-06,
      "loss": 0.9084,
      "step": 8422
    },
    {
      "epoch": 0.680660215357886,
      "grad_norm": 2.8171634674072266,
      "learning_rate": 7.680017910123597e-06,
      "loss": 0.9681,
      "step": 8423
    },
    {
      "epoch": 0.6807410250712136,
      "grad_norm": 2.595752239227295,
      "learning_rate": 7.67946546437393e-06,
      "loss": 0.8406,
      "step": 8424
    },
    {
      "epoch": 0.6808218347845411,
      "grad_norm": 3.4196860790252686,
      "learning_rate": 7.67891297273147e-06,
      "loss": 0.9399,
      "step": 8425
    },
    {
      "epoch": 0.6809026444978686,
      "grad_norm": 2.6884870529174805,
      "learning_rate": 7.678360435205679e-06,
      "loss": 1.0277,
      "step": 8426
    },
    {
      "epoch": 0.6809834542111962,
      "grad_norm": 2.9558327198028564,
      "learning_rate": 7.677807851806022e-06,
      "loss": 0.9687,
      "step": 8427
    },
    {
      "epoch": 0.6810642639245237,
      "grad_norm": 2.5531744956970215,
      "learning_rate": 7.677255222541963e-06,
      "loss": 0.9916,
      "step": 8428
    },
    {
      "epoch": 0.6811450736378513,
      "grad_norm": 2.709683895111084,
      "learning_rate": 7.676702547422966e-06,
      "loss": 1.0118,
      "step": 8429
    },
    {
      "epoch": 0.6812258833511788,
      "grad_norm": 2.541829824447632,
      "learning_rate": 7.676149826458502e-06,
      "loss": 0.8386,
      "step": 8430
    },
    {
      "epoch": 0.6813066930645063,
      "grad_norm": 2.8499135971069336,
      "learning_rate": 7.675597059658031e-06,
      "loss": 0.979,
      "step": 8431
    },
    {
      "epoch": 0.6813875027778339,
      "grad_norm": 2.470608949661255,
      "learning_rate": 7.675044247031024e-06,
      "loss": 1.0212,
      "step": 8432
    },
    {
      "epoch": 0.6814683124911615,
      "grad_norm": 2.7002062797546387,
      "learning_rate": 7.67449138858695e-06,
      "loss": 1.0245,
      "step": 8433
    },
    {
      "epoch": 0.6815491222044889,
      "grad_norm": 2.56925892829895,
      "learning_rate": 7.673938484335276e-06,
      "loss": 1.0153,
      "step": 8434
    },
    {
      "epoch": 0.6816299319178165,
      "grad_norm": 2.4124858379364014,
      "learning_rate": 7.673385534285473e-06,
      "loss": 0.9645,
      "step": 8435
    },
    {
      "epoch": 0.6817107416311441,
      "grad_norm": 2.7462291717529297,
      "learning_rate": 7.672832538447011e-06,
      "loss": 0.8917,
      "step": 8436
    },
    {
      "epoch": 0.6817915513444716,
      "grad_norm": 2.918308973312378,
      "learning_rate": 7.672279496829364e-06,
      "loss": 1.0055,
      "step": 8437
    },
    {
      "epoch": 0.6818723610577991,
      "grad_norm": 2.783287763595581,
      "learning_rate": 7.671726409442002e-06,
      "loss": 0.9613,
      "step": 8438
    },
    {
      "epoch": 0.6819531707711267,
      "grad_norm": 2.619626522064209,
      "learning_rate": 7.671173276294397e-06,
      "loss": 1.051,
      "step": 8439
    },
    {
      "epoch": 0.6820339804844542,
      "grad_norm": 2.851304054260254,
      "learning_rate": 7.670620097396026e-06,
      "loss": 1.0064,
      "step": 8440
    },
    {
      "epoch": 0.6821147901977818,
      "grad_norm": 2.8876547813415527,
      "learning_rate": 7.670066872756362e-06,
      "loss": 0.953,
      "step": 8441
    },
    {
      "epoch": 0.6821955999111093,
      "grad_norm": 2.6084249019622803,
      "learning_rate": 7.669513602384879e-06,
      "loss": 0.9463,
      "step": 8442
    },
    {
      "epoch": 0.6822764096244368,
      "grad_norm": 2.5381696224212646,
      "learning_rate": 7.668960286291056e-06,
      "loss": 1.0155,
      "step": 8443
    },
    {
      "epoch": 0.6823572193377644,
      "grad_norm": 2.913601875305176,
      "learning_rate": 7.668406924484368e-06,
      "loss": 0.8306,
      "step": 8444
    },
    {
      "epoch": 0.682438029051092,
      "grad_norm": 2.6268184185028076,
      "learning_rate": 7.667853516974292e-06,
      "loss": 1.0605,
      "step": 8445
    },
    {
      "epoch": 0.6825188387644194,
      "grad_norm": 2.4881415367126465,
      "learning_rate": 7.66730006377031e-06,
      "loss": 1.0588,
      "step": 8446
    },
    {
      "epoch": 0.682599648477747,
      "grad_norm": 2.5792691707611084,
      "learning_rate": 7.6667465648819e-06,
      "loss": 0.9516,
      "step": 8447
    },
    {
      "epoch": 0.6826804581910746,
      "grad_norm": 2.578933000564575,
      "learning_rate": 7.666193020318537e-06,
      "loss": 0.9021,
      "step": 8448
    },
    {
      "epoch": 0.6827612679044021,
      "grad_norm": 2.4988861083984375,
      "learning_rate": 7.66563943008971e-06,
      "loss": 1.0085,
      "step": 8449
    },
    {
      "epoch": 0.6828420776177296,
      "grad_norm": 2.6755433082580566,
      "learning_rate": 7.665085794204896e-06,
      "loss": 1.0179,
      "step": 8450
    },
    {
      "epoch": 0.6829228873310572,
      "grad_norm": 2.3153629302978516,
      "learning_rate": 7.664532112673578e-06,
      "loss": 0.9813,
      "step": 8451
    },
    {
      "epoch": 0.6830036970443847,
      "grad_norm": 2.309171438217163,
      "learning_rate": 7.66397838550524e-06,
      "loss": 0.8567,
      "step": 8452
    },
    {
      "epoch": 0.6830845067577123,
      "grad_norm": 2.3666679859161377,
      "learning_rate": 7.663424612709364e-06,
      "loss": 0.9389,
      "step": 8453
    },
    {
      "epoch": 0.6831653164710398,
      "grad_norm": 2.8175573348999023,
      "learning_rate": 7.662870794295438e-06,
      "loss": 0.9105,
      "step": 8454
    },
    {
      "epoch": 0.6832461261843673,
      "grad_norm": 2.2377684116363525,
      "learning_rate": 7.662316930272945e-06,
      "loss": 0.9315,
      "step": 8455
    },
    {
      "epoch": 0.6833269358976949,
      "grad_norm": 2.5128984451293945,
      "learning_rate": 7.661763020651372e-06,
      "loss": 0.9953,
      "step": 8456
    },
    {
      "epoch": 0.6834077456110225,
      "grad_norm": 2.887550115585327,
      "learning_rate": 7.661209065440207e-06,
      "loss": 0.8619,
      "step": 8457
    },
    {
      "epoch": 0.6834885553243499,
      "grad_norm": 2.778005599975586,
      "learning_rate": 7.660655064648937e-06,
      "loss": 0.8838,
      "step": 8458
    },
    {
      "epoch": 0.6835693650376775,
      "grad_norm": 2.6972951889038086,
      "learning_rate": 7.660101018287053e-06,
      "loss": 0.9645,
      "step": 8459
    },
    {
      "epoch": 0.6836501747510051,
      "grad_norm": 2.66129207611084,
      "learning_rate": 7.659546926364038e-06,
      "loss": 0.8559,
      "step": 8460
    },
    {
      "epoch": 0.6837309844643326,
      "grad_norm": 2.4100239276885986,
      "learning_rate": 7.65899278888939e-06,
      "loss": 0.7838,
      "step": 8461
    },
    {
      "epoch": 0.6838117941776601,
      "grad_norm": 2.6647441387176514,
      "learning_rate": 7.658438605872596e-06,
      "loss": 0.9808,
      "step": 8462
    },
    {
      "epoch": 0.6838926038909877,
      "grad_norm": 2.9784247875213623,
      "learning_rate": 7.657884377323149e-06,
      "loss": 0.9254,
      "step": 8463
    },
    {
      "epoch": 0.6839734136043152,
      "grad_norm": 2.894587278366089,
      "learning_rate": 7.65733010325054e-06,
      "loss": 0.9022,
      "step": 8464
    },
    {
      "epoch": 0.6840542233176428,
      "grad_norm": 2.923002243041992,
      "learning_rate": 7.656775783664265e-06,
      "loss": 0.9115,
      "step": 8465
    },
    {
      "epoch": 0.6841350330309703,
      "grad_norm": 2.273637056350708,
      "learning_rate": 7.656221418573817e-06,
      "loss": 0.8948,
      "step": 8466
    },
    {
      "epoch": 0.6842158427442979,
      "grad_norm": 2.5338826179504395,
      "learning_rate": 7.65566700798869e-06,
      "loss": 1.0228,
      "step": 8467
    },
    {
      "epoch": 0.6842966524576254,
      "grad_norm": 2.8406641483306885,
      "learning_rate": 7.65511255191838e-06,
      "loss": 0.9498,
      "step": 8468
    },
    {
      "epoch": 0.684377462170953,
      "grad_norm": 2.497377872467041,
      "learning_rate": 7.654558050372385e-06,
      "loss": 0.9637,
      "step": 8469
    },
    {
      "epoch": 0.6844582718842805,
      "grad_norm": 2.7127885818481445,
      "learning_rate": 7.6540035033602e-06,
      "loss": 1.0735,
      "step": 8470
    },
    {
      "epoch": 0.684539081597608,
      "grad_norm": 2.4919934272766113,
      "learning_rate": 7.653448910891325e-06,
      "loss": 1.0278,
      "step": 8471
    },
    {
      "epoch": 0.6846198913109356,
      "grad_norm": 2.5153603553771973,
      "learning_rate": 7.652894272975257e-06,
      "loss": 0.9611,
      "step": 8472
    },
    {
      "epoch": 0.6847007010242632,
      "grad_norm": 2.4566097259521484,
      "learning_rate": 7.652339589621498e-06,
      "loss": 0.8935,
      "step": 8473
    },
    {
      "epoch": 0.6847815107375906,
      "grad_norm": 2.7795534133911133,
      "learning_rate": 7.651784860839547e-06,
      "loss": 0.9947,
      "step": 8474
    },
    {
      "epoch": 0.6848623204509182,
      "grad_norm": 3.1100776195526123,
      "learning_rate": 7.651230086638905e-06,
      "loss": 1.0381,
      "step": 8475
    },
    {
      "epoch": 0.6849431301642458,
      "grad_norm": 2.716601848602295,
      "learning_rate": 7.650675267029072e-06,
      "loss": 0.9526,
      "step": 8476
    },
    {
      "epoch": 0.6850239398775733,
      "grad_norm": 2.3293495178222656,
      "learning_rate": 7.650120402019556e-06,
      "loss": 0.9579,
      "step": 8477
    },
    {
      "epoch": 0.6851047495909008,
      "grad_norm": 2.3756096363067627,
      "learning_rate": 7.649565491619855e-06,
      "loss": 0.8161,
      "step": 8478
    },
    {
      "epoch": 0.6851855593042284,
      "grad_norm": 2.896602153778076,
      "learning_rate": 7.649010535839478e-06,
      "loss": 0.9511,
      "step": 8479
    },
    {
      "epoch": 0.6852663690175559,
      "grad_norm": 2.6385321617126465,
      "learning_rate": 7.648455534687927e-06,
      "loss": 0.9406,
      "step": 8480
    },
    {
      "epoch": 0.6853471787308835,
      "grad_norm": 2.6948325634002686,
      "learning_rate": 7.647900488174708e-06,
      "loss": 0.9515,
      "step": 8481
    },
    {
      "epoch": 0.685427988444211,
      "grad_norm": 2.516601800918579,
      "learning_rate": 7.647345396309328e-06,
      "loss": 0.8487,
      "step": 8482
    },
    {
      "epoch": 0.6855087981575385,
      "grad_norm": 2.4833476543426514,
      "learning_rate": 7.646790259101297e-06,
      "loss": 1.0024,
      "step": 8483
    },
    {
      "epoch": 0.6855896078708661,
      "grad_norm": 2.570033550262451,
      "learning_rate": 7.646235076560119e-06,
      "loss": 0.8984,
      "step": 8484
    },
    {
      "epoch": 0.6856704175841937,
      "grad_norm": 2.308443069458008,
      "learning_rate": 7.645679848695305e-06,
      "loss": 0.9218,
      "step": 8485
    },
    {
      "epoch": 0.6857512272975211,
      "grad_norm": 2.457383632659912,
      "learning_rate": 7.645124575516363e-06,
      "loss": 0.9703,
      "step": 8486
    },
    {
      "epoch": 0.6858320370108487,
      "grad_norm": 2.499898910522461,
      "learning_rate": 7.644569257032805e-06,
      "loss": 1.0285,
      "step": 8487
    },
    {
      "epoch": 0.6859128467241763,
      "grad_norm": 2.739859104156494,
      "learning_rate": 7.644013893254145e-06,
      "loss": 0.9798,
      "step": 8488
    },
    {
      "epoch": 0.6859936564375038,
      "grad_norm": 2.992964506149292,
      "learning_rate": 7.64345848418989e-06,
      "loss": 0.9043,
      "step": 8489
    },
    {
      "epoch": 0.6860744661508313,
      "grad_norm": 2.946100950241089,
      "learning_rate": 7.642903029849554e-06,
      "loss": 0.8497,
      "step": 8490
    },
    {
      "epoch": 0.6861552758641589,
      "grad_norm": 2.515207052230835,
      "learning_rate": 7.642347530242654e-06,
      "loss": 0.9575,
      "step": 8491
    },
    {
      "epoch": 0.6862360855774864,
      "grad_norm": 3.00736927986145,
      "learning_rate": 7.6417919853787e-06,
      "loss": 0.9283,
      "step": 8492
    },
    {
      "epoch": 0.686316895290814,
      "grad_norm": 2.454270601272583,
      "learning_rate": 7.64123639526721e-06,
      "loss": 1.0599,
      "step": 8493
    },
    {
      "epoch": 0.6863977050041415,
      "grad_norm": 2.7114779949188232,
      "learning_rate": 7.6406807599177e-06,
      "loss": 1.05,
      "step": 8494
    },
    {
      "epoch": 0.686478514717469,
      "grad_norm": 2.576533555984497,
      "learning_rate": 7.640125079339684e-06,
      "loss": 0.9726,
      "step": 8495
    },
    {
      "epoch": 0.6865593244307966,
      "grad_norm": 2.3898873329162598,
      "learning_rate": 7.639569353542683e-06,
      "loss": 0.8742,
      "step": 8496
    },
    {
      "epoch": 0.6866401341441242,
      "grad_norm": 2.7289464473724365,
      "learning_rate": 7.639013582536213e-06,
      "loss": 0.9219,
      "step": 8497
    },
    {
      "epoch": 0.6867209438574516,
      "grad_norm": 2.785449743270874,
      "learning_rate": 7.638457766329792e-06,
      "loss": 0.963,
      "step": 8498
    },
    {
      "epoch": 0.6868017535707792,
      "grad_norm": 2.609910249710083,
      "learning_rate": 7.637901904932943e-06,
      "loss": 0.8897,
      "step": 8499
    },
    {
      "epoch": 0.6868825632841068,
      "grad_norm": 2.7766265869140625,
      "learning_rate": 7.637345998355185e-06,
      "loss": 0.9813,
      "step": 8500
    },
    {
      "epoch": 0.6869633729974343,
      "grad_norm": 2.8442697525024414,
      "learning_rate": 7.636790046606037e-06,
      "loss": 1.0714,
      "step": 8501
    },
    {
      "epoch": 0.6870441827107618,
      "grad_norm": 2.5213828086853027,
      "learning_rate": 7.636234049695026e-06,
      "loss": 0.9575,
      "step": 8502
    },
    {
      "epoch": 0.6871249924240894,
      "grad_norm": 3.0026817321777344,
      "learning_rate": 7.63567800763167e-06,
      "loss": 0.8413,
      "step": 8503
    },
    {
      "epoch": 0.6872058021374169,
      "grad_norm": 2.7835092544555664,
      "learning_rate": 7.635121920425498e-06,
      "loss": 0.9992,
      "step": 8504
    },
    {
      "epoch": 0.6872866118507445,
      "grad_norm": 2.566882848739624,
      "learning_rate": 7.634565788086028e-06,
      "loss": 0.8768,
      "step": 8505
    },
    {
      "epoch": 0.687367421564072,
      "grad_norm": 2.6903584003448486,
      "learning_rate": 7.634009610622789e-06,
      "loss": 0.9405,
      "step": 8506
    },
    {
      "epoch": 0.6874482312773995,
      "grad_norm": 2.805124521255493,
      "learning_rate": 7.633453388045306e-06,
      "loss": 0.8344,
      "step": 8507
    },
    {
      "epoch": 0.6875290409907271,
      "grad_norm": 2.63955020904541,
      "learning_rate": 7.63289712036311e-06,
      "loss": 0.9395,
      "step": 8508
    },
    {
      "epoch": 0.6876098507040547,
      "grad_norm": 2.5578644275665283,
      "learning_rate": 7.63234080758572e-06,
      "loss": 0.8353,
      "step": 8509
    },
    {
      "epoch": 0.6876906604173821,
      "grad_norm": 2.5840113162994385,
      "learning_rate": 7.631784449722672e-06,
      "loss": 0.8967,
      "step": 8510
    },
    {
      "epoch": 0.6877714701307097,
      "grad_norm": 2.510427713394165,
      "learning_rate": 7.631228046783492e-06,
      "loss": 0.9796,
      "step": 8511
    },
    {
      "epoch": 0.6878522798440373,
      "grad_norm": 2.9449968338012695,
      "learning_rate": 7.63067159877771e-06,
      "loss": 0.8771,
      "step": 8512
    },
    {
      "epoch": 0.6879330895573648,
      "grad_norm": 2.2487404346466064,
      "learning_rate": 7.630115105714854e-06,
      "loss": 0.8831,
      "step": 8513
    },
    {
      "epoch": 0.6880138992706923,
      "grad_norm": 2.572007179260254,
      "learning_rate": 7.629558567604461e-06,
      "loss": 0.9316,
      "step": 8514
    },
    {
      "epoch": 0.6880947089840199,
      "grad_norm": 2.8729922771453857,
      "learning_rate": 7.629001984456059e-06,
      "loss": 0.8668,
      "step": 8515
    },
    {
      "epoch": 0.6881755186973474,
      "grad_norm": 2.8302195072174072,
      "learning_rate": 7.628445356279182e-06,
      "loss": 1.0108,
      "step": 8516
    },
    {
      "epoch": 0.688256328410675,
      "grad_norm": 2.6309752464294434,
      "learning_rate": 7.627888683083363e-06,
      "loss": 0.9727,
      "step": 8517
    },
    {
      "epoch": 0.6883371381240025,
      "grad_norm": 2.6336255073547363,
      "learning_rate": 7.6273319648781395e-06,
      "loss": 0.9099,
      "step": 8518
    },
    {
      "epoch": 0.68841794783733,
      "grad_norm": 2.4486746788024902,
      "learning_rate": 7.626775201673042e-06,
      "loss": 0.8963,
      "step": 8519
    },
    {
      "epoch": 0.6884987575506576,
      "grad_norm": 2.5037882328033447,
      "learning_rate": 7.6262183934776114e-06,
      "loss": 1.0884,
      "step": 8520
    },
    {
      "epoch": 0.6885795672639852,
      "grad_norm": 2.4955711364746094,
      "learning_rate": 7.62566154030138e-06,
      "loss": 0.8892,
      "step": 8521
    },
    {
      "epoch": 0.6886603769773126,
      "grad_norm": 2.296449661254883,
      "learning_rate": 7.625104642153889e-06,
      "loss": 0.9347,
      "step": 8522
    },
    {
      "epoch": 0.6887411866906402,
      "grad_norm": 2.760537624359131,
      "learning_rate": 7.624547699044673e-06,
      "loss": 1.0912,
      "step": 8523
    },
    {
      "epoch": 0.6888219964039678,
      "grad_norm": 2.649718761444092,
      "learning_rate": 7.623990710983275e-06,
      "loss": 0.8756,
      "step": 8524
    },
    {
      "epoch": 0.6889028061172953,
      "grad_norm": 2.7818546295166016,
      "learning_rate": 7.623433677979234e-06,
      "loss": 1.0735,
      "step": 8525
    },
    {
      "epoch": 0.6889836158306228,
      "grad_norm": 2.3486881256103516,
      "learning_rate": 7.622876600042088e-06,
      "loss": 0.9275,
      "step": 8526
    },
    {
      "epoch": 0.6890644255439504,
      "grad_norm": 2.675062656402588,
      "learning_rate": 7.622319477181381e-06,
      "loss": 0.9279,
      "step": 8527
    },
    {
      "epoch": 0.6891452352572779,
      "grad_norm": 2.6401591300964355,
      "learning_rate": 7.6217623094066554e-06,
      "loss": 1.0097,
      "step": 8528
    },
    {
      "epoch": 0.6892260449706055,
      "grad_norm": 2.5405540466308594,
      "learning_rate": 7.6212050967274495e-06,
      "loss": 0.7642,
      "step": 8529
    },
    {
      "epoch": 0.689306854683933,
      "grad_norm": 2.401169776916504,
      "learning_rate": 7.620647839153315e-06,
      "loss": 1.0061,
      "step": 8530
    },
    {
      "epoch": 0.6893876643972605,
      "grad_norm": 2.991809368133545,
      "learning_rate": 7.620090536693787e-06,
      "loss": 0.9315,
      "step": 8531
    },
    {
      "epoch": 0.6894684741105881,
      "grad_norm": 2.5979745388031006,
      "learning_rate": 7.61953318935842e-06,
      "loss": 0.9467,
      "step": 8532
    },
    {
      "epoch": 0.6895492838239157,
      "grad_norm": 2.3411121368408203,
      "learning_rate": 7.618975797156753e-06,
      "loss": 0.8983,
      "step": 8533
    },
    {
      "epoch": 0.6896300935372431,
      "grad_norm": 2.627819538116455,
      "learning_rate": 7.618418360098338e-06,
      "loss": 1.0355,
      "step": 8534
    },
    {
      "epoch": 0.6897109032505707,
      "grad_norm": 2.9491703510284424,
      "learning_rate": 7.617860878192718e-06,
      "loss": 0.9876,
      "step": 8535
    },
    {
      "epoch": 0.6897917129638983,
      "grad_norm": 2.737373113632202,
      "learning_rate": 7.617303351449444e-06,
      "loss": 0.942,
      "step": 8536
    },
    {
      "epoch": 0.6898725226772258,
      "grad_norm": 2.894930362701416,
      "learning_rate": 7.616745779878065e-06,
      "loss": 0.9467,
      "step": 8537
    },
    {
      "epoch": 0.6899533323905533,
      "grad_norm": 2.7718076705932617,
      "learning_rate": 7.616188163488132e-06,
      "loss": 0.8853,
      "step": 8538
    },
    {
      "epoch": 0.6900341421038809,
      "grad_norm": 2.4677348136901855,
      "learning_rate": 7.615630502289191e-06,
      "loss": 1.0362,
      "step": 8539
    },
    {
      "epoch": 0.6901149518172084,
      "grad_norm": 2.5967016220092773,
      "learning_rate": 7.615072796290797e-06,
      "loss": 1.0943,
      "step": 8540
    },
    {
      "epoch": 0.690195761530536,
      "grad_norm": 2.6121981143951416,
      "learning_rate": 7.614515045502502e-06,
      "loss": 0.8968,
      "step": 8541
    },
    {
      "epoch": 0.6902765712438635,
      "grad_norm": 2.5807816982269287,
      "learning_rate": 7.613957249933859e-06,
      "loss": 0.8829,
      "step": 8542
    },
    {
      "epoch": 0.690357380957191,
      "grad_norm": 2.292766809463501,
      "learning_rate": 7.6133994095944195e-06,
      "loss": 1.0097,
      "step": 8543
    },
    {
      "epoch": 0.6904381906705186,
      "grad_norm": 2.4342575073242188,
      "learning_rate": 7.612841524493741e-06,
      "loss": 0.9578,
      "step": 8544
    },
    {
      "epoch": 0.6905190003838462,
      "grad_norm": 2.439470052719116,
      "learning_rate": 7.612283594641376e-06,
      "loss": 0.951,
      "step": 8545
    },
    {
      "epoch": 0.6905998100971736,
      "grad_norm": 2.6422617435455322,
      "learning_rate": 7.611725620046884e-06,
      "loss": 0.9223,
      "step": 8546
    },
    {
      "epoch": 0.6906806198105012,
      "grad_norm": 2.6705915927886963,
      "learning_rate": 7.611167600719819e-06,
      "loss": 0.9802,
      "step": 8547
    },
    {
      "epoch": 0.6907614295238288,
      "grad_norm": 2.584794044494629,
      "learning_rate": 7.610609536669737e-06,
      "loss": 0.8957,
      "step": 8548
    },
    {
      "epoch": 0.6908422392371563,
      "grad_norm": 2.595289945602417,
      "learning_rate": 7.610051427906201e-06,
      "loss": 0.924,
      "step": 8549
    },
    {
      "epoch": 0.6909230489504838,
      "grad_norm": 2.8844175338745117,
      "learning_rate": 7.609493274438766e-06,
      "loss": 0.9296,
      "step": 8550
    },
    {
      "epoch": 0.6910038586638114,
      "grad_norm": 3.100165367126465,
      "learning_rate": 7.608935076276994e-06,
      "loss": 1.0573,
      "step": 8551
    },
    {
      "epoch": 0.6910846683771389,
      "grad_norm": 2.577888250350952,
      "learning_rate": 7.608376833430444e-06,
      "loss": 0.8948,
      "step": 8552
    },
    {
      "epoch": 0.6911654780904665,
      "grad_norm": 3.04179310798645,
      "learning_rate": 7.607818545908681e-06,
      "loss": 0.9858,
      "step": 8553
    },
    {
      "epoch": 0.691246287803794,
      "grad_norm": 2.8973894119262695,
      "learning_rate": 7.607260213721262e-06,
      "loss": 0.854,
      "step": 8554
    },
    {
      "epoch": 0.6913270975171215,
      "grad_norm": 3.1727302074432373,
      "learning_rate": 7.606701836877752e-06,
      "loss": 1.0172,
      "step": 8555
    },
    {
      "epoch": 0.6914079072304491,
      "grad_norm": 2.350161552429199,
      "learning_rate": 7.606143415387715e-06,
      "loss": 0.8157,
      "step": 8556
    },
    {
      "epoch": 0.6914887169437767,
      "grad_norm": 2.398963689804077,
      "learning_rate": 7.605584949260716e-06,
      "loss": 0.8972,
      "step": 8557
    },
    {
      "epoch": 0.6915695266571041,
      "grad_norm": 2.073056697845459,
      "learning_rate": 7.60502643850632e-06,
      "loss": 1.0482,
      "step": 8558
    },
    {
      "epoch": 0.6916503363704317,
      "grad_norm": 2.7007317543029785,
      "learning_rate": 7.6044678831340915e-06,
      "loss": 0.9308,
      "step": 8559
    },
    {
      "epoch": 0.6917311460837593,
      "grad_norm": 2.6965243816375732,
      "learning_rate": 7.603909283153598e-06,
      "loss": 1.0158,
      "step": 8560
    },
    {
      "epoch": 0.6918119557970868,
      "grad_norm": 2.4421262741088867,
      "learning_rate": 7.603350638574408e-06,
      "loss": 0.9155,
      "step": 8561
    },
    {
      "epoch": 0.6918927655104143,
      "grad_norm": 2.9302661418914795,
      "learning_rate": 7.602791949406088e-06,
      "loss": 0.8886,
      "step": 8562
    },
    {
      "epoch": 0.6919735752237419,
      "grad_norm": 2.614354372024536,
      "learning_rate": 7.602233215658209e-06,
      "loss": 1.0003,
      "step": 8563
    },
    {
      "epoch": 0.6920543849370694,
      "grad_norm": 2.6408298015594482,
      "learning_rate": 7.601674437340339e-06,
      "loss": 1.0349,
      "step": 8564
    },
    {
      "epoch": 0.692135194650397,
      "grad_norm": 2.468775987625122,
      "learning_rate": 7.601115614462049e-06,
      "loss": 1.0312,
      "step": 8565
    },
    {
      "epoch": 0.6922160043637245,
      "grad_norm": 2.7402100563049316,
      "learning_rate": 7.60055674703291e-06,
      "loss": 0.8751,
      "step": 8566
    },
    {
      "epoch": 0.692296814077052,
      "grad_norm": 2.747537136077881,
      "learning_rate": 7.599997835062496e-06,
      "loss": 0.9242,
      "step": 8567
    },
    {
      "epoch": 0.6923776237903796,
      "grad_norm": 2.6012680530548096,
      "learning_rate": 7.599438878560377e-06,
      "loss": 0.9521,
      "step": 8568
    },
    {
      "epoch": 0.6924584335037072,
      "grad_norm": 2.8128600120544434,
      "learning_rate": 7.598879877536129e-06,
      "loss": 1.0664,
      "step": 8569
    },
    {
      "epoch": 0.6925392432170346,
      "grad_norm": 2.7352778911590576,
      "learning_rate": 7.598320831999323e-06,
      "loss": 0.8944,
      "step": 8570
    },
    {
      "epoch": 0.6926200529303622,
      "grad_norm": 2.58370041847229,
      "learning_rate": 7.59776174195954e-06,
      "loss": 0.9682,
      "step": 8571
    },
    {
      "epoch": 0.6927008626436898,
      "grad_norm": 2.5282909870147705,
      "learning_rate": 7.597202607426349e-06,
      "loss": 0.8859,
      "step": 8572
    },
    {
      "epoch": 0.6927816723570173,
      "grad_norm": 3.129368543624878,
      "learning_rate": 7.59664342840933e-06,
      "loss": 1.035,
      "step": 8573
    },
    {
      "epoch": 0.6928624820703448,
      "grad_norm": 2.583153486251831,
      "learning_rate": 7.596084204918062e-06,
      "loss": 0.926,
      "step": 8574
    },
    {
      "epoch": 0.6929432917836724,
      "grad_norm": 2.4431116580963135,
      "learning_rate": 7.595524936962122e-06,
      "loss": 0.793,
      "step": 8575
    },
    {
      "epoch": 0.6930241014969999,
      "grad_norm": 2.709498643875122,
      "learning_rate": 7.594965624551085e-06,
      "loss": 0.9978,
      "step": 8576
    },
    {
      "epoch": 0.6931049112103275,
      "grad_norm": 2.3784282207489014,
      "learning_rate": 7.594406267694536e-06,
      "loss": 0.8461,
      "step": 8577
    },
    {
      "epoch": 0.693185720923655,
      "grad_norm": 2.5849270820617676,
      "learning_rate": 7.593846866402054e-06,
      "loss": 0.9594,
      "step": 8578
    },
    {
      "epoch": 0.6932665306369825,
      "grad_norm": 2.7068569660186768,
      "learning_rate": 7.593287420683219e-06,
      "loss": 0.8639,
      "step": 8579
    },
    {
      "epoch": 0.6933473403503101,
      "grad_norm": 2.623075246810913,
      "learning_rate": 7.592727930547613e-06,
      "loss": 1.1281,
      "step": 8580
    },
    {
      "epoch": 0.6934281500636377,
      "grad_norm": 2.4773852825164795,
      "learning_rate": 7.59216839600482e-06,
      "loss": 0.934,
      "step": 8581
    },
    {
      "epoch": 0.6935089597769651,
      "grad_norm": 2.8432528972625732,
      "learning_rate": 7.591608817064422e-06,
      "loss": 0.772,
      "step": 8582
    },
    {
      "epoch": 0.6935897694902927,
      "grad_norm": 2.484088659286499,
      "learning_rate": 7.5910491937360054e-06,
      "loss": 0.9708,
      "step": 8583
    },
    {
      "epoch": 0.6936705792036203,
      "grad_norm": 3.355008602142334,
      "learning_rate": 7.590489526029152e-06,
      "loss": 0.9861,
      "step": 8584
    },
    {
      "epoch": 0.6937513889169478,
      "grad_norm": 2.7875428199768066,
      "learning_rate": 7.589929813953452e-06,
      "loss": 0.8744,
      "step": 8585
    },
    {
      "epoch": 0.6938321986302753,
      "grad_norm": 2.3667221069335938,
      "learning_rate": 7.589370057518486e-06,
      "loss": 0.9079,
      "step": 8586
    },
    {
      "epoch": 0.6939130083436029,
      "grad_norm": 2.7356786727905273,
      "learning_rate": 7.588810256733847e-06,
      "loss": 0.8746,
      "step": 8587
    },
    {
      "epoch": 0.6939938180569304,
      "grad_norm": 2.7664761543273926,
      "learning_rate": 7.58825041160912e-06,
      "loss": 0.9542,
      "step": 8588
    },
    {
      "epoch": 0.694074627770258,
      "grad_norm": 2.645258665084839,
      "learning_rate": 7.587690522153894e-06,
      "loss": 0.9539,
      "step": 8589
    },
    {
      "epoch": 0.6941554374835855,
      "grad_norm": 2.487731695175171,
      "learning_rate": 7.587130588377758e-06,
      "loss": 0.9562,
      "step": 8590
    },
    {
      "epoch": 0.694236247196913,
      "grad_norm": 2.425704002380371,
      "learning_rate": 7.586570610290305e-06,
      "loss": 1.0059,
      "step": 8591
    },
    {
      "epoch": 0.6943170569102406,
      "grad_norm": 2.481106996536255,
      "learning_rate": 7.586010587901125e-06,
      "loss": 0.8474,
      "step": 8592
    },
    {
      "epoch": 0.6943978666235682,
      "grad_norm": 2.572812557220459,
      "learning_rate": 7.585450521219807e-06,
      "loss": 0.9305,
      "step": 8593
    },
    {
      "epoch": 0.6944786763368956,
      "grad_norm": 2.375962257385254,
      "learning_rate": 7.584890410255948e-06,
      "loss": 0.9888,
      "step": 8594
    },
    {
      "epoch": 0.6945594860502232,
      "grad_norm": 2.728224515914917,
      "learning_rate": 7.584330255019137e-06,
      "loss": 0.8738,
      "step": 8595
    },
    {
      "epoch": 0.6946402957635508,
      "grad_norm": 3.3338637351989746,
      "learning_rate": 7.583770055518971e-06,
      "loss": 0.9901,
      "step": 8596
    },
    {
      "epoch": 0.6947211054768784,
      "grad_norm": 2.8553481101989746,
      "learning_rate": 7.583209811765044e-06,
      "loss": 0.9182,
      "step": 8597
    },
    {
      "epoch": 0.6948019151902058,
      "grad_norm": 2.6169826984405518,
      "learning_rate": 7.582649523766952e-06,
      "loss": 0.9113,
      "step": 8598
    },
    {
      "epoch": 0.6948827249035334,
      "grad_norm": 3.1289563179016113,
      "learning_rate": 7.582089191534292e-06,
      "loss": 0.9261,
      "step": 8599
    },
    {
      "epoch": 0.694963534616861,
      "grad_norm": 2.7174198627471924,
      "learning_rate": 7.58152881507666e-06,
      "loss": 0.8609,
      "step": 8600
    },
    {
      "epoch": 0.6950443443301885,
      "grad_norm": 2.9810330867767334,
      "learning_rate": 7.580968394403653e-06,
      "loss": 0.948,
      "step": 8601
    },
    {
      "epoch": 0.695125154043516,
      "grad_norm": 2.527822256088257,
      "learning_rate": 7.580407929524871e-06,
      "loss": 0.9894,
      "step": 8602
    },
    {
      "epoch": 0.6952059637568436,
      "grad_norm": 3.117835521697998,
      "learning_rate": 7.579847420449913e-06,
      "loss": 1.0269,
      "step": 8603
    },
    {
      "epoch": 0.6952867734701711,
      "grad_norm": 2.935013771057129,
      "learning_rate": 7.5792868671883805e-06,
      "loss": 0.9498,
      "step": 8604
    },
    {
      "epoch": 0.6953675831834987,
      "grad_norm": 2.737457513809204,
      "learning_rate": 7.578726269749874e-06,
      "loss": 0.8966,
      "step": 8605
    },
    {
      "epoch": 0.6954483928968263,
      "grad_norm": 2.5293126106262207,
      "learning_rate": 7.578165628143993e-06,
      "loss": 0.9781,
      "step": 8606
    },
    {
      "epoch": 0.6955292026101537,
      "grad_norm": 2.5696821212768555,
      "learning_rate": 7.577604942380342e-06,
      "loss": 0.9699,
      "step": 8607
    },
    {
      "epoch": 0.6956100123234813,
      "grad_norm": 2.6878457069396973,
      "learning_rate": 7.577044212468523e-06,
      "loss": 1.0177,
      "step": 8608
    },
    {
      "epoch": 0.6956908220368089,
      "grad_norm": 2.777543306350708,
      "learning_rate": 7.576483438418142e-06,
      "loss": 0.9182,
      "step": 8609
    },
    {
      "epoch": 0.6957716317501363,
      "grad_norm": 2.3973729610443115,
      "learning_rate": 7.575922620238801e-06,
      "loss": 0.8788,
      "step": 8610
    },
    {
      "epoch": 0.6958524414634639,
      "grad_norm": 2.2486517429351807,
      "learning_rate": 7.575361757940107e-06,
      "loss": 1.0163,
      "step": 8611
    },
    {
      "epoch": 0.6959332511767915,
      "grad_norm": 2.6407740116119385,
      "learning_rate": 7.574800851531667e-06,
      "loss": 1.0108,
      "step": 8612
    },
    {
      "epoch": 0.696014060890119,
      "grad_norm": 2.4537734985351562,
      "learning_rate": 7.574239901023086e-06,
      "loss": 0.8857,
      "step": 8613
    },
    {
      "epoch": 0.6960948706034465,
      "grad_norm": 2.5676252841949463,
      "learning_rate": 7.573678906423973e-06,
      "loss": 0.9512,
      "step": 8614
    },
    {
      "epoch": 0.6961756803167741,
      "grad_norm": 2.4398884773254395,
      "learning_rate": 7.573117867743937e-06,
      "loss": 0.8377,
      "step": 8615
    },
    {
      "epoch": 0.6962564900301016,
      "grad_norm": 2.8415637016296387,
      "learning_rate": 7.572556784992586e-06,
      "loss": 0.9853,
      "step": 8616
    },
    {
      "epoch": 0.6963372997434292,
      "grad_norm": 2.5694844722747803,
      "learning_rate": 7.571995658179529e-06,
      "loss": 0.887,
      "step": 8617
    },
    {
      "epoch": 0.6964181094567568,
      "grad_norm": 2.49828839302063,
      "learning_rate": 7.57143448731438e-06,
      "loss": 1.0492,
      "step": 8618
    },
    {
      "epoch": 0.6964989191700842,
      "grad_norm": 3.0085999965667725,
      "learning_rate": 7.570873272406748e-06,
      "loss": 0.9792,
      "step": 8619
    },
    {
      "epoch": 0.6965797288834118,
      "grad_norm": 2.5902698040008545,
      "learning_rate": 7.570312013466248e-06,
      "loss": 0.9358,
      "step": 8620
    },
    {
      "epoch": 0.6966605385967394,
      "grad_norm": 2.436767578125,
      "learning_rate": 7.569750710502487e-06,
      "loss": 1.0047,
      "step": 8621
    },
    {
      "epoch": 0.6967413483100668,
      "grad_norm": 2.3210556507110596,
      "learning_rate": 7.569189363525086e-06,
      "loss": 1.0272,
      "step": 8622
    },
    {
      "epoch": 0.6968221580233944,
      "grad_norm": 2.89433217048645,
      "learning_rate": 7.568627972543654e-06,
      "loss": 0.8926,
      "step": 8623
    },
    {
      "epoch": 0.696902967736722,
      "grad_norm": 2.4591269493103027,
      "learning_rate": 7.568066537567811e-06,
      "loss": 0.9729,
      "step": 8624
    },
    {
      "epoch": 0.6969837774500495,
      "grad_norm": 2.656367778778076,
      "learning_rate": 7.567505058607169e-06,
      "loss": 0.9672,
      "step": 8625
    },
    {
      "epoch": 0.697064587163377,
      "grad_norm": 2.5143606662750244,
      "learning_rate": 7.566943535671346e-06,
      "loss": 1.0344,
      "step": 8626
    },
    {
      "epoch": 0.6971453968767046,
      "grad_norm": 2.8403632640838623,
      "learning_rate": 7.5663819687699605e-06,
      "loss": 0.9725,
      "step": 8627
    },
    {
      "epoch": 0.6972262065900321,
      "grad_norm": 3.015887975692749,
      "learning_rate": 7.565820357912631e-06,
      "loss": 1.0496,
      "step": 8628
    },
    {
      "epoch": 0.6973070163033597,
      "grad_norm": 2.7617697715759277,
      "learning_rate": 7.565258703108973e-06,
      "loss": 0.9673,
      "step": 8629
    },
    {
      "epoch": 0.6973878260166873,
      "grad_norm": 2.7930734157562256,
      "learning_rate": 7.564697004368613e-06,
      "loss": 1.0001,
      "step": 8630
    },
    {
      "epoch": 0.6974686357300147,
      "grad_norm": 2.5580267906188965,
      "learning_rate": 7.564135261701165e-06,
      "loss": 0.8846,
      "step": 8631
    },
    {
      "epoch": 0.6975494454433423,
      "grad_norm": 2.546013355255127,
      "learning_rate": 7.563573475116252e-06,
      "loss": 0.9289,
      "step": 8632
    },
    {
      "epoch": 0.6976302551566699,
      "grad_norm": 2.1583731174468994,
      "learning_rate": 7.563011644623499e-06,
      "loss": 0.9521,
      "step": 8633
    },
    {
      "epoch": 0.6977110648699973,
      "grad_norm": 2.7555742263793945,
      "learning_rate": 7.562449770232527e-06,
      "loss": 0.9649,
      "step": 8634
    },
    {
      "epoch": 0.6977918745833249,
      "grad_norm": 2.919497489929199,
      "learning_rate": 7.561887851952958e-06,
      "loss": 0.8866,
      "step": 8635
    },
    {
      "epoch": 0.6978726842966525,
      "grad_norm": 2.2876970767974854,
      "learning_rate": 7.561325889794417e-06,
      "loss": 0.9934,
      "step": 8636
    },
    {
      "epoch": 0.69795349400998,
      "grad_norm": 2.8355917930603027,
      "learning_rate": 7.560763883766531e-06,
      "loss": 0.8978,
      "step": 8637
    },
    {
      "epoch": 0.6980343037233075,
      "grad_norm": 3.190540075302124,
      "learning_rate": 7.560201833878924e-06,
      "loss": 0.8722,
      "step": 8638
    },
    {
      "epoch": 0.6981151134366351,
      "grad_norm": 3.1715128421783447,
      "learning_rate": 7.5596397401412234e-06,
      "loss": 0.873,
      "step": 8639
    },
    {
      "epoch": 0.6981959231499626,
      "grad_norm": 3.1202750205993652,
      "learning_rate": 7.559077602563057e-06,
      "loss": 0.9352,
      "step": 8640
    },
    {
      "epoch": 0.6982767328632902,
      "grad_norm": 2.6783177852630615,
      "learning_rate": 7.558515421154049e-06,
      "loss": 0.9136,
      "step": 8641
    },
    {
      "epoch": 0.6983575425766178,
      "grad_norm": 2.462207317352295,
      "learning_rate": 7.557953195923834e-06,
      "loss": 0.9122,
      "step": 8642
    },
    {
      "epoch": 0.6984383522899452,
      "grad_norm": 2.8287558555603027,
      "learning_rate": 7.557390926882037e-06,
      "loss": 0.898,
      "step": 8643
    },
    {
      "epoch": 0.6985191620032728,
      "grad_norm": 2.6783833503723145,
      "learning_rate": 7.556828614038292e-06,
      "loss": 0.9573,
      "step": 8644
    },
    {
      "epoch": 0.6985999717166004,
      "grad_norm": 2.578094720840454,
      "learning_rate": 7.556266257402226e-06,
      "loss": 0.9204,
      "step": 8645
    },
    {
      "epoch": 0.6986807814299278,
      "grad_norm": 2.8346030712127686,
      "learning_rate": 7.555703856983474e-06,
      "loss": 1.0098,
      "step": 8646
    },
    {
      "epoch": 0.6987615911432554,
      "grad_norm": 2.802305221557617,
      "learning_rate": 7.555141412791666e-06,
      "loss": 1.001,
      "step": 8647
    },
    {
      "epoch": 0.698842400856583,
      "grad_norm": 2.5147013664245605,
      "learning_rate": 7.554578924836441e-06,
      "loss": 0.8849,
      "step": 8648
    },
    {
      "epoch": 0.6989232105699105,
      "grad_norm": 2.50382137298584,
      "learning_rate": 7.554016393127425e-06,
      "loss": 1.0378,
      "step": 8649
    },
    {
      "epoch": 0.699004020283238,
      "grad_norm": 2.706132173538208,
      "learning_rate": 7.553453817674259e-06,
      "loss": 0.8929,
      "step": 8650
    },
    {
      "epoch": 0.6990848299965656,
      "grad_norm": 2.8096556663513184,
      "learning_rate": 7.552891198486575e-06,
      "loss": 0.8318,
      "step": 8651
    },
    {
      "epoch": 0.6991656397098931,
      "grad_norm": 2.7654807567596436,
      "learning_rate": 7.552328535574011e-06,
      "loss": 0.8296,
      "step": 8652
    },
    {
      "epoch": 0.6992464494232207,
      "grad_norm": 2.4102702140808105,
      "learning_rate": 7.551765828946202e-06,
      "loss": 0.9913,
      "step": 8653
    },
    {
      "epoch": 0.6993272591365483,
      "grad_norm": 2.799931049346924,
      "learning_rate": 7.5512030786127895e-06,
      "loss": 0.9169,
      "step": 8654
    },
    {
      "epoch": 0.6994080688498757,
      "grad_norm": 2.5439860820770264,
      "learning_rate": 7.55064028458341e-06,
      "loss": 0.9709,
      "step": 8655
    },
    {
      "epoch": 0.6994888785632033,
      "grad_norm": 2.6339478492736816,
      "learning_rate": 7.550077446867703e-06,
      "loss": 0.9449,
      "step": 8656
    },
    {
      "epoch": 0.6995696882765309,
      "grad_norm": 2.402557611465454,
      "learning_rate": 7.549514565475306e-06,
      "loss": 1.0367,
      "step": 8657
    },
    {
      "epoch": 0.6996504979898583,
      "grad_norm": 3.135340929031372,
      "learning_rate": 7.548951640415866e-06,
      "loss": 0.8973,
      "step": 8658
    },
    {
      "epoch": 0.6997313077031859,
      "grad_norm": 3.2426278591156006,
      "learning_rate": 7.548388671699019e-06,
      "loss": 0.9245,
      "step": 8659
    },
    {
      "epoch": 0.6998121174165135,
      "grad_norm": 2.369006872177124,
      "learning_rate": 7.547825659334408e-06,
      "loss": 0.931,
      "step": 8660
    },
    {
      "epoch": 0.699892927129841,
      "grad_norm": 2.7105281352996826,
      "learning_rate": 7.5472626033316775e-06,
      "loss": 0.9759,
      "step": 8661
    },
    {
      "epoch": 0.6999737368431685,
      "grad_norm": 2.8235816955566406,
      "learning_rate": 7.546699503700472e-06,
      "loss": 0.8962,
      "step": 8662
    },
    {
      "epoch": 0.7000545465564961,
      "grad_norm": 2.7356045246124268,
      "learning_rate": 7.546136360450434e-06,
      "loss": 0.974,
      "step": 8663
    },
    {
      "epoch": 0.7001353562698236,
      "grad_norm": 2.7572247982025146,
      "learning_rate": 7.54557317359121e-06,
      "loss": 0.8473,
      "step": 8664
    },
    {
      "epoch": 0.7002161659831512,
      "grad_norm": 2.7012922763824463,
      "learning_rate": 7.545009943132446e-06,
      "loss": 0.9708,
      "step": 8665
    },
    {
      "epoch": 0.7002969756964788,
      "grad_norm": 2.7031867504119873,
      "learning_rate": 7.544446669083788e-06,
      "loss": 0.962,
      "step": 8666
    },
    {
      "epoch": 0.7003777854098062,
      "grad_norm": 2.5701441764831543,
      "learning_rate": 7.543883351454884e-06,
      "loss": 1.0285,
      "step": 8667
    },
    {
      "epoch": 0.7004585951231338,
      "grad_norm": 3.391747236251831,
      "learning_rate": 7.543319990255382e-06,
      "loss": 0.9503,
      "step": 8668
    },
    {
      "epoch": 0.7005394048364614,
      "grad_norm": 2.4658329486846924,
      "learning_rate": 7.542756585494933e-06,
      "loss": 0.9535,
      "step": 8669
    },
    {
      "epoch": 0.7006202145497888,
      "grad_norm": 2.760211229324341,
      "learning_rate": 7.542193137183184e-06,
      "loss": 0.9103,
      "step": 8670
    },
    {
      "epoch": 0.7007010242631164,
      "grad_norm": 2.506631374359131,
      "learning_rate": 7.541629645329787e-06,
      "loss": 1.0329,
      "step": 8671
    },
    {
      "epoch": 0.700781833976444,
      "grad_norm": 2.7162492275238037,
      "learning_rate": 7.541066109944393e-06,
      "loss": 0.9238,
      "step": 8672
    },
    {
      "epoch": 0.7008626436897715,
      "grad_norm": 2.9053521156311035,
      "learning_rate": 7.540502531036653e-06,
      "loss": 1.0092,
      "step": 8673
    },
    {
      "epoch": 0.700943453403099,
      "grad_norm": 2.562232494354248,
      "learning_rate": 7.539938908616221e-06,
      "loss": 1.0301,
      "step": 8674
    },
    {
      "epoch": 0.7010242631164266,
      "grad_norm": 2.5382494926452637,
      "learning_rate": 7.53937524269275e-06,
      "loss": 0.907,
      "step": 8675
    },
    {
      "epoch": 0.7011050728297541,
      "grad_norm": 2.983947515487671,
      "learning_rate": 7.538811533275896e-06,
      "loss": 0.9538,
      "step": 8676
    },
    {
      "epoch": 0.7011858825430817,
      "grad_norm": 2.8068487644195557,
      "learning_rate": 7.5382477803753095e-06,
      "loss": 1.0288,
      "step": 8677
    },
    {
      "epoch": 0.7012666922564093,
      "grad_norm": 2.4507806301116943,
      "learning_rate": 7.537683984000651e-06,
      "loss": 0.9991,
      "step": 8678
    },
    {
      "epoch": 0.7013475019697367,
      "grad_norm": 2.6927502155303955,
      "learning_rate": 7.5371201441615745e-06,
      "loss": 0.9675,
      "step": 8679
    },
    {
      "epoch": 0.7014283116830643,
      "grad_norm": 2.58587384223938,
      "learning_rate": 7.53655626086774e-06,
      "loss": 0.9279,
      "step": 8680
    },
    {
      "epoch": 0.7015091213963919,
      "grad_norm": 2.3846824169158936,
      "learning_rate": 7.535992334128801e-06,
      "loss": 0.9453,
      "step": 8681
    },
    {
      "epoch": 0.7015899311097193,
      "grad_norm": 2.7196900844573975,
      "learning_rate": 7.535428363954418e-06,
      "loss": 0.8563,
      "step": 8682
    },
    {
      "epoch": 0.7016707408230469,
      "grad_norm": 2.4219930171966553,
      "learning_rate": 7.534864350354252e-06,
      "loss": 1.0132,
      "step": 8683
    },
    {
      "epoch": 0.7017515505363745,
      "grad_norm": 3.8839070796966553,
      "learning_rate": 7.534300293337962e-06,
      "loss": 0.8899,
      "step": 8684
    },
    {
      "epoch": 0.701832360249702,
      "grad_norm": 2.76360821723938,
      "learning_rate": 7.5337361929152085e-06,
      "loss": 1.0045,
      "step": 8685
    },
    {
      "epoch": 0.7019131699630295,
      "grad_norm": 2.3911209106445312,
      "learning_rate": 7.533172049095654e-06,
      "loss": 0.9208,
      "step": 8686
    },
    {
      "epoch": 0.7019939796763571,
      "grad_norm": 2.581125020980835,
      "learning_rate": 7.532607861888962e-06,
      "loss": 1.0219,
      "step": 8687
    },
    {
      "epoch": 0.7020747893896846,
      "grad_norm": 2.369431734085083,
      "learning_rate": 7.532043631304792e-06,
      "loss": 0.9991,
      "step": 8688
    },
    {
      "epoch": 0.7021555991030122,
      "grad_norm": 2.5366764068603516,
      "learning_rate": 7.531479357352812e-06,
      "loss": 0.8967,
      "step": 8689
    },
    {
      "epoch": 0.7022364088163398,
      "grad_norm": 2.5214669704437256,
      "learning_rate": 7.530915040042684e-06,
      "loss": 0.9072,
      "step": 8690
    },
    {
      "epoch": 0.7023172185296672,
      "grad_norm": 2.962538242340088,
      "learning_rate": 7.5303506793840755e-06,
      "loss": 0.9389,
      "step": 8691
    },
    {
      "epoch": 0.7023980282429948,
      "grad_norm": 2.8953888416290283,
      "learning_rate": 7.52978627538665e-06,
      "loss": 1.0155,
      "step": 8692
    },
    {
      "epoch": 0.7024788379563224,
      "grad_norm": 2.7045507431030273,
      "learning_rate": 7.529221828060076e-06,
      "loss": 1.0143,
      "step": 8693
    },
    {
      "epoch": 0.7025596476696498,
      "grad_norm": 2.3151791095733643,
      "learning_rate": 7.5286573374140205e-06,
      "loss": 0.977,
      "step": 8694
    },
    {
      "epoch": 0.7026404573829774,
      "grad_norm": 2.6774165630340576,
      "learning_rate": 7.528092803458154e-06,
      "loss": 0.8586,
      "step": 8695
    },
    {
      "epoch": 0.702721267096305,
      "grad_norm": 2.803908109664917,
      "learning_rate": 7.527528226202142e-06,
      "loss": 0.8626,
      "step": 8696
    },
    {
      "epoch": 0.7028020768096325,
      "grad_norm": 2.39029860496521,
      "learning_rate": 7.526963605655659e-06,
      "loss": 0.9367,
      "step": 8697
    },
    {
      "epoch": 0.70288288652296,
      "grad_norm": 2.5351829528808594,
      "learning_rate": 7.52639894182837e-06,
      "loss": 0.9001,
      "step": 8698
    },
    {
      "epoch": 0.7029636962362876,
      "grad_norm": 2.88838791847229,
      "learning_rate": 7.5258342347299504e-06,
      "loss": 0.9646,
      "step": 8699
    },
    {
      "epoch": 0.7030445059496151,
      "grad_norm": 2.9428274631500244,
      "learning_rate": 7.52526948437007e-06,
      "loss": 0.979,
      "step": 8700
    },
    {
      "epoch": 0.7031253156629427,
      "grad_norm": 2.27873158454895,
      "learning_rate": 7.524704690758405e-06,
      "loss": 0.8633,
      "step": 8701
    },
    {
      "epoch": 0.7032061253762703,
      "grad_norm": 2.4483258724212646,
      "learning_rate": 7.524139853904624e-06,
      "loss": 0.8733,
      "step": 8702
    },
    {
      "epoch": 0.7032869350895977,
      "grad_norm": 3.072887659072876,
      "learning_rate": 7.523574973818406e-06,
      "loss": 0.947,
      "step": 8703
    },
    {
      "epoch": 0.7033677448029253,
      "grad_norm": 2.6498184204101562,
      "learning_rate": 7.523010050509423e-06,
      "loss": 0.9445,
      "step": 8704
    },
    {
      "epoch": 0.7034485545162529,
      "grad_norm": 2.5125577449798584,
      "learning_rate": 7.522445083987353e-06,
      "loss": 0.8173,
      "step": 8705
    },
    {
      "epoch": 0.7035293642295803,
      "grad_norm": 2.689518690109253,
      "learning_rate": 7.521880074261869e-06,
      "loss": 0.9604,
      "step": 8706
    },
    {
      "epoch": 0.7036101739429079,
      "grad_norm": 3.4577524662017822,
      "learning_rate": 7.521315021342652e-06,
      "loss": 0.9367,
      "step": 8707
    },
    {
      "epoch": 0.7036909836562355,
      "grad_norm": 2.4654698371887207,
      "learning_rate": 7.520749925239378e-06,
      "loss": 0.9222,
      "step": 8708
    },
    {
      "epoch": 0.703771793369563,
      "grad_norm": 2.5475175380706787,
      "learning_rate": 7.520184785961727e-06,
      "loss": 0.9775,
      "step": 8709
    },
    {
      "epoch": 0.7038526030828905,
      "grad_norm": 2.6552414894104004,
      "learning_rate": 7.519619603519376e-06,
      "loss": 0.8956,
      "step": 8710
    },
    {
      "epoch": 0.7039334127962181,
      "grad_norm": 2.3695228099823,
      "learning_rate": 7.519054377922009e-06,
      "loss": 1.0462,
      "step": 8711
    },
    {
      "epoch": 0.7040142225095456,
      "grad_norm": 2.7073700428009033,
      "learning_rate": 7.518489109179304e-06,
      "loss": 0.912,
      "step": 8712
    },
    {
      "epoch": 0.7040950322228732,
      "grad_norm": 3.083273410797119,
      "learning_rate": 7.5179237973009435e-06,
      "loss": 0.991,
      "step": 8713
    },
    {
      "epoch": 0.7041758419362008,
      "grad_norm": 2.614108085632324,
      "learning_rate": 7.51735844229661e-06,
      "loss": 1.073,
      "step": 8714
    },
    {
      "epoch": 0.7042566516495282,
      "grad_norm": 2.8452632427215576,
      "learning_rate": 7.5167930441759875e-06,
      "loss": 0.9139,
      "step": 8715
    },
    {
      "epoch": 0.7043374613628558,
      "grad_norm": 2.1154541969299316,
      "learning_rate": 7.516227602948756e-06,
      "loss": 1.0226,
      "step": 8716
    },
    {
      "epoch": 0.7044182710761834,
      "grad_norm": 2.4039242267608643,
      "learning_rate": 7.515662118624607e-06,
      "loss": 0.9091,
      "step": 8717
    },
    {
      "epoch": 0.7044990807895108,
      "grad_norm": 2.7088048458099365,
      "learning_rate": 7.5150965912132205e-06,
      "loss": 0.9404,
      "step": 8718
    },
    {
      "epoch": 0.7045798905028384,
      "grad_norm": 2.4967024326324463,
      "learning_rate": 7.5145310207242836e-06,
      "loss": 0.9141,
      "step": 8719
    },
    {
      "epoch": 0.704660700216166,
      "grad_norm": 2.6987764835357666,
      "learning_rate": 7.513965407167485e-06,
      "loss": 0.9831,
      "step": 8720
    },
    {
      "epoch": 0.7047415099294935,
      "grad_norm": 2.955282688140869,
      "learning_rate": 7.51339975055251e-06,
      "loss": 1.0303,
      "step": 8721
    },
    {
      "epoch": 0.704822319642821,
      "grad_norm": 2.0802903175354004,
      "learning_rate": 7.512834050889048e-06,
      "loss": 0.8419,
      "step": 8722
    },
    {
      "epoch": 0.7049031293561486,
      "grad_norm": 2.7226569652557373,
      "learning_rate": 7.51226830818679e-06,
      "loss": 0.9608,
      "step": 8723
    },
    {
      "epoch": 0.7049839390694762,
      "grad_norm": 2.963134527206421,
      "learning_rate": 7.511702522455422e-06,
      "loss": 0.9277,
      "step": 8724
    },
    {
      "epoch": 0.7050647487828037,
      "grad_norm": 2.859212636947632,
      "learning_rate": 7.511136693704637e-06,
      "loss": 0.9165,
      "step": 8725
    },
    {
      "epoch": 0.7051455584961313,
      "grad_norm": 2.5749497413635254,
      "learning_rate": 7.510570821944126e-06,
      "loss": 0.947,
      "step": 8726
    },
    {
      "epoch": 0.7052263682094588,
      "grad_norm": 2.3279025554656982,
      "learning_rate": 7.510004907183581e-06,
      "loss": 0.919,
      "step": 8727
    },
    {
      "epoch": 0.7053071779227863,
      "grad_norm": 2.355889320373535,
      "learning_rate": 7.509438949432694e-06,
      "loss": 0.8945,
      "step": 8728
    },
    {
      "epoch": 0.7053879876361139,
      "grad_norm": 2.6222805976867676,
      "learning_rate": 7.50887294870116e-06,
      "loss": 0.9518,
      "step": 8729
    },
    {
      "epoch": 0.7054687973494415,
      "grad_norm": 3.2635293006896973,
      "learning_rate": 7.50830690499867e-06,
      "loss": 0.9052,
      "step": 8730
    },
    {
      "epoch": 0.7055496070627689,
      "grad_norm": 2.763084650039673,
      "learning_rate": 7.507740818334924e-06,
      "loss": 0.8967,
      "step": 8731
    },
    {
      "epoch": 0.7056304167760965,
      "grad_norm": 2.514324426651001,
      "learning_rate": 7.507174688719614e-06,
      "loss": 0.9678,
      "step": 8732
    },
    {
      "epoch": 0.7057112264894241,
      "grad_norm": 2.5376338958740234,
      "learning_rate": 7.506608516162437e-06,
      "loss": 0.9405,
      "step": 8733
    },
    {
      "epoch": 0.7057920362027515,
      "grad_norm": 2.724379301071167,
      "learning_rate": 7.50604230067309e-06,
      "loss": 0.8867,
      "step": 8734
    },
    {
      "epoch": 0.7058728459160791,
      "grad_norm": 2.402949094772339,
      "learning_rate": 7.505476042261271e-06,
      "loss": 0.844,
      "step": 8735
    },
    {
      "epoch": 0.7059536556294067,
      "grad_norm": 2.7474207878112793,
      "learning_rate": 7.504909740936681e-06,
      "loss": 1.0281,
      "step": 8736
    },
    {
      "epoch": 0.7060344653427342,
      "grad_norm": 2.48898983001709,
      "learning_rate": 7.504343396709017e-06,
      "loss": 0.9633,
      "step": 8737
    },
    {
      "epoch": 0.7061152750560618,
      "grad_norm": 2.1229522228240967,
      "learning_rate": 7.503777009587978e-06,
      "loss": 1.0316,
      "step": 8738
    },
    {
      "epoch": 0.7061960847693893,
      "grad_norm": 2.361743688583374,
      "learning_rate": 7.5032105795832685e-06,
      "loss": 0.9821,
      "step": 8739
    },
    {
      "epoch": 0.7062768944827168,
      "grad_norm": 2.911328077316284,
      "learning_rate": 7.502644106704586e-06,
      "loss": 1.1078,
      "step": 8740
    },
    {
      "epoch": 0.7063577041960444,
      "grad_norm": 2.2683932781219482,
      "learning_rate": 7.5020775909616365e-06,
      "loss": 0.9699,
      "step": 8741
    },
    {
      "epoch": 0.706438513909372,
      "grad_norm": 2.544445753097534,
      "learning_rate": 7.50151103236412e-06,
      "loss": 1.0466,
      "step": 8742
    },
    {
      "epoch": 0.7065193236226994,
      "grad_norm": 2.5798516273498535,
      "learning_rate": 7.500944430921743e-06,
      "loss": 0.9785,
      "step": 8743
    },
    {
      "epoch": 0.706600133336027,
      "grad_norm": 2.4686522483825684,
      "learning_rate": 7.500377786644207e-06,
      "loss": 0.9202,
      "step": 8744
    },
    {
      "epoch": 0.7066809430493546,
      "grad_norm": 2.7797043323516846,
      "learning_rate": 7.499811099541221e-06,
      "loss": 0.9207,
      "step": 8745
    },
    {
      "epoch": 0.706761752762682,
      "grad_norm": 2.6669442653656006,
      "learning_rate": 7.499244369622488e-06,
      "loss": 0.9433,
      "step": 8746
    },
    {
      "epoch": 0.7068425624760096,
      "grad_norm": 2.5092170238494873,
      "learning_rate": 7.4986775968977155e-06,
      "loss": 0.9155,
      "step": 8747
    },
    {
      "epoch": 0.7069233721893372,
      "grad_norm": 2.559154510498047,
      "learning_rate": 7.498110781376611e-06,
      "loss": 0.9418,
      "step": 8748
    },
    {
      "epoch": 0.7070041819026647,
      "grad_norm": 2.280831813812256,
      "learning_rate": 7.497543923068883e-06,
      "loss": 1.038,
      "step": 8749
    },
    {
      "epoch": 0.7070849916159923,
      "grad_norm": 2.3083012104034424,
      "learning_rate": 7.4969770219842395e-06,
      "loss": 1.0707,
      "step": 8750
    },
    {
      "epoch": 0.7071658013293198,
      "grad_norm": 3.4840195178985596,
      "learning_rate": 7.4964100781323915e-06,
      "loss": 1.0052,
      "step": 8751
    },
    {
      "epoch": 0.7072466110426473,
      "grad_norm": 3.099773406982422,
      "learning_rate": 7.495843091523049e-06,
      "loss": 1.0539,
      "step": 8752
    },
    {
      "epoch": 0.7073274207559749,
      "grad_norm": 2.438974142074585,
      "learning_rate": 7.495276062165922e-06,
      "loss": 1.1024,
      "step": 8753
    },
    {
      "epoch": 0.7074082304693025,
      "grad_norm": 2.3954732418060303,
      "learning_rate": 7.494708990070724e-06,
      "loss": 0.9245,
      "step": 8754
    },
    {
      "epoch": 0.7074890401826299,
      "grad_norm": 2.5976598262786865,
      "learning_rate": 7.494141875247165e-06,
      "loss": 0.9778,
      "step": 8755
    },
    {
      "epoch": 0.7075698498959575,
      "grad_norm": 2.151345729827881,
      "learning_rate": 7.493574717704964e-06,
      "loss": 0.9815,
      "step": 8756
    },
    {
      "epoch": 0.7076506596092851,
      "grad_norm": 2.6544687747955322,
      "learning_rate": 7.493007517453828e-06,
      "loss": 0.9092,
      "step": 8757
    },
    {
      "epoch": 0.7077314693226125,
      "grad_norm": 2.319226026535034,
      "learning_rate": 7.4924402745034745e-06,
      "loss": 0.9584,
      "step": 8758
    },
    {
      "epoch": 0.7078122790359401,
      "grad_norm": 2.8776092529296875,
      "learning_rate": 7.49187298886362e-06,
      "loss": 1.087,
      "step": 8759
    },
    {
      "epoch": 0.7078930887492677,
      "grad_norm": 2.3584213256835938,
      "learning_rate": 7.491305660543982e-06,
      "loss": 0.7846,
      "step": 8760
    },
    {
      "epoch": 0.7079738984625952,
      "grad_norm": 2.5362932682037354,
      "learning_rate": 7.490738289554273e-06,
      "loss": 0.8692,
      "step": 8761
    },
    {
      "epoch": 0.7080547081759228,
      "grad_norm": 2.2135183811187744,
      "learning_rate": 7.490170875904215e-06,
      "loss": 1.0255,
      "step": 8762
    },
    {
      "epoch": 0.7081355178892503,
      "grad_norm": 2.3536763191223145,
      "learning_rate": 7.489603419603524e-06,
      "loss": 0.8183,
      "step": 8763
    },
    {
      "epoch": 0.7082163276025778,
      "grad_norm": 2.256617784500122,
      "learning_rate": 7.48903592066192e-06,
      "loss": 0.9819,
      "step": 8764
    },
    {
      "epoch": 0.7082971373159054,
      "grad_norm": 2.7875657081604004,
      "learning_rate": 7.488468379089123e-06,
      "loss": 0.8544,
      "step": 8765
    },
    {
      "epoch": 0.708377947029233,
      "grad_norm": 3.078984498977661,
      "learning_rate": 7.487900794894853e-06,
      "loss": 0.9126,
      "step": 8766
    },
    {
      "epoch": 0.7084587567425604,
      "grad_norm": 2.362226963043213,
      "learning_rate": 7.487333168088832e-06,
      "loss": 0.8633,
      "step": 8767
    },
    {
      "epoch": 0.708539566455888,
      "grad_norm": 2.774829626083374,
      "learning_rate": 7.4867654986807824e-06,
      "loss": 0.9722,
      "step": 8768
    },
    {
      "epoch": 0.7086203761692156,
      "grad_norm": 2.991361379623413,
      "learning_rate": 7.486197786680425e-06,
      "loss": 0.9546,
      "step": 8769
    },
    {
      "epoch": 0.708701185882543,
      "grad_norm": 2.9520673751831055,
      "learning_rate": 7.485630032097486e-06,
      "loss": 0.9553,
      "step": 8770
    },
    {
      "epoch": 0.7087819955958706,
      "grad_norm": 3.1070826053619385,
      "learning_rate": 7.4850622349416894e-06,
      "loss": 1.0138,
      "step": 8771
    },
    {
      "epoch": 0.7088628053091982,
      "grad_norm": 2.724799633026123,
      "learning_rate": 7.484494395222758e-06,
      "loss": 0.947,
      "step": 8772
    },
    {
      "epoch": 0.7089436150225257,
      "grad_norm": 2.702982187271118,
      "learning_rate": 7.483926512950418e-06,
      "loss": 0.9498,
      "step": 8773
    },
    {
      "epoch": 0.7090244247358533,
      "grad_norm": 2.4756410121917725,
      "learning_rate": 7.483358588134398e-06,
      "loss": 0.9672,
      "step": 8774
    },
    {
      "epoch": 0.7091052344491808,
      "grad_norm": 2.4144980907440186,
      "learning_rate": 7.482790620784423e-06,
      "loss": 1.049,
      "step": 8775
    },
    {
      "epoch": 0.7091860441625083,
      "grad_norm": 2.344552516937256,
      "learning_rate": 7.482222610910222e-06,
      "loss": 0.8233,
      "step": 8776
    },
    {
      "epoch": 0.7092668538758359,
      "grad_norm": 2.753931760787964,
      "learning_rate": 7.481654558521523e-06,
      "loss": 0.9061,
      "step": 8777
    },
    {
      "epoch": 0.7093476635891635,
      "grad_norm": 2.382394552230835,
      "learning_rate": 7.481086463628057e-06,
      "loss": 0.9233,
      "step": 8778
    },
    {
      "epoch": 0.7094284733024909,
      "grad_norm": 2.821577787399292,
      "learning_rate": 7.480518326239552e-06,
      "loss": 1.034,
      "step": 8779
    },
    {
      "epoch": 0.7095092830158185,
      "grad_norm": 2.4641852378845215,
      "learning_rate": 7.47995014636574e-06,
      "loss": 0.9662,
      "step": 8780
    },
    {
      "epoch": 0.7095900927291461,
      "grad_norm": 6.780241966247559,
      "learning_rate": 7.479381924016351e-06,
      "loss": 0.9078,
      "step": 8781
    },
    {
      "epoch": 0.7096709024424736,
      "grad_norm": 2.5848770141601562,
      "learning_rate": 7.47881365920112e-06,
      "loss": 0.9667,
      "step": 8782
    },
    {
      "epoch": 0.7097517121558011,
      "grad_norm": 2.367079496383667,
      "learning_rate": 7.478245351929777e-06,
      "loss": 0.8423,
      "step": 8783
    },
    {
      "epoch": 0.7098325218691287,
      "grad_norm": 2.7461133003234863,
      "learning_rate": 7.4776770022120596e-06,
      "loss": 1.039,
      "step": 8784
    },
    {
      "epoch": 0.7099133315824562,
      "grad_norm": 2.403761148452759,
      "learning_rate": 7.477108610057699e-06,
      "loss": 0.8635,
      "step": 8785
    },
    {
      "epoch": 0.7099941412957838,
      "grad_norm": 2.949636936187744,
      "learning_rate": 7.47654017547643e-06,
      "loss": 0.9767,
      "step": 8786
    },
    {
      "epoch": 0.7100749510091113,
      "grad_norm": 2.945683479309082,
      "learning_rate": 7.4759716984779906e-06,
      "loss": 0.9774,
      "step": 8787
    },
    {
      "epoch": 0.7101557607224388,
      "grad_norm": 2.611204147338867,
      "learning_rate": 7.475403179072116e-06,
      "loss": 1.0105,
      "step": 8788
    },
    {
      "epoch": 0.7102365704357664,
      "grad_norm": 2.9422717094421387,
      "learning_rate": 7.474834617268545e-06,
      "loss": 0.9791,
      "step": 8789
    },
    {
      "epoch": 0.710317380149094,
      "grad_norm": 2.365835666656494,
      "learning_rate": 7.4742660130770165e-06,
      "loss": 0.9045,
      "step": 8790
    },
    {
      "epoch": 0.7103981898624214,
      "grad_norm": 2.431830644607544,
      "learning_rate": 7.473697366507264e-06,
      "loss": 0.971,
      "step": 8791
    },
    {
      "epoch": 0.710478999575749,
      "grad_norm": 2.978790521621704,
      "learning_rate": 7.4731286775690344e-06,
      "loss": 0.9703,
      "step": 8792
    },
    {
      "epoch": 0.7105598092890766,
      "grad_norm": 2.4299097061157227,
      "learning_rate": 7.472559946272063e-06,
      "loss": 1.082,
      "step": 8793
    },
    {
      "epoch": 0.710640619002404,
      "grad_norm": 2.591629981994629,
      "learning_rate": 7.4719911726260915e-06,
      "loss": 0.8756,
      "step": 8794
    },
    {
      "epoch": 0.7107214287157316,
      "grad_norm": 2.925849676132202,
      "learning_rate": 7.471422356640863e-06,
      "loss": 0.8851,
      "step": 8795
    },
    {
      "epoch": 0.7108022384290592,
      "grad_norm": 2.668818473815918,
      "learning_rate": 7.470853498326121e-06,
      "loss": 0.9577,
      "step": 8796
    },
    {
      "epoch": 0.7108830481423867,
      "grad_norm": 2.38317608833313,
      "learning_rate": 7.470284597691603e-06,
      "loss": 0.8331,
      "step": 8797
    },
    {
      "epoch": 0.7109638578557143,
      "grad_norm": 2.542328119277954,
      "learning_rate": 7.469715654747059e-06,
      "loss": 1.0084,
      "step": 8798
    },
    {
      "epoch": 0.7110446675690418,
      "grad_norm": 3.080044984817505,
      "learning_rate": 7.469146669502232e-06,
      "loss": 0.88,
      "step": 8799
    },
    {
      "epoch": 0.7111254772823693,
      "grad_norm": 2.602565050125122,
      "learning_rate": 7.468577641966866e-06,
      "loss": 1.0348,
      "step": 8800
    },
    {
      "epoch": 0.7112062869956969,
      "grad_norm": 2.9165420532226562,
      "learning_rate": 7.468008572150707e-06,
      "loss": 0.9912,
      "step": 8801
    },
    {
      "epoch": 0.7112870967090245,
      "grad_norm": 2.833052158355713,
      "learning_rate": 7.467439460063504e-06,
      "loss": 0.9313,
      "step": 8802
    },
    {
      "epoch": 0.7113679064223519,
      "grad_norm": 2.9368364810943604,
      "learning_rate": 7.466870305715002e-06,
      "loss": 0.9332,
      "step": 8803
    },
    {
      "epoch": 0.7114487161356795,
      "grad_norm": 3.3586065769195557,
      "learning_rate": 7.466301109114953e-06,
      "loss": 0.9256,
      "step": 8804
    },
    {
      "epoch": 0.7115295258490071,
      "grad_norm": 2.572722911834717,
      "learning_rate": 7.4657318702731e-06,
      "loss": 0.9497,
      "step": 8805
    },
    {
      "epoch": 0.7116103355623346,
      "grad_norm": 2.5348410606384277,
      "learning_rate": 7.465162589199197e-06,
      "loss": 0.9544,
      "step": 8806
    },
    {
      "epoch": 0.7116911452756621,
      "grad_norm": 3.143693208694458,
      "learning_rate": 7.464593265902995e-06,
      "loss": 0.9532,
      "step": 8807
    },
    {
      "epoch": 0.7117719549889897,
      "grad_norm": 2.5578513145446777,
      "learning_rate": 7.464023900394243e-06,
      "loss": 0.8862,
      "step": 8808
    },
    {
      "epoch": 0.7118527647023172,
      "grad_norm": 2.3635027408599854,
      "learning_rate": 7.463454492682693e-06,
      "loss": 0.9082,
      "step": 8809
    },
    {
      "epoch": 0.7119335744156448,
      "grad_norm": 2.573094606399536,
      "learning_rate": 7.462885042778097e-06,
      "loss": 0.9715,
      "step": 8810
    },
    {
      "epoch": 0.7120143841289723,
      "grad_norm": 2.5036468505859375,
      "learning_rate": 7.462315550690211e-06,
      "loss": 0.873,
      "step": 8811
    },
    {
      "epoch": 0.7120951938422998,
      "grad_norm": 2.525418281555176,
      "learning_rate": 7.461746016428787e-06,
      "loss": 0.9455,
      "step": 8812
    },
    {
      "epoch": 0.7121760035556274,
      "grad_norm": 2.4634768962860107,
      "learning_rate": 7.461176440003581e-06,
      "loss": 0.9364,
      "step": 8813
    },
    {
      "epoch": 0.712256813268955,
      "grad_norm": 2.6082589626312256,
      "learning_rate": 7.460606821424347e-06,
      "loss": 0.9037,
      "step": 8814
    },
    {
      "epoch": 0.7123376229822824,
      "grad_norm": 2.4430713653564453,
      "learning_rate": 7.460037160700842e-06,
      "loss": 0.8612,
      "step": 8815
    },
    {
      "epoch": 0.71241843269561,
      "grad_norm": 2.94598126411438,
      "learning_rate": 7.459467457842822e-06,
      "loss": 0.9126,
      "step": 8816
    },
    {
      "epoch": 0.7124992424089376,
      "grad_norm": 2.605842351913452,
      "learning_rate": 7.458897712860045e-06,
      "loss": 0.9074,
      "step": 8817
    },
    {
      "epoch": 0.712580052122265,
      "grad_norm": 2.8378100395202637,
      "learning_rate": 7.45832792576227e-06,
      "loss": 1.006,
      "step": 8818
    },
    {
      "epoch": 0.7126608618355926,
      "grad_norm": 2.8556532859802246,
      "learning_rate": 7.457758096559256e-06,
      "loss": 0.9265,
      "step": 8819
    },
    {
      "epoch": 0.7127416715489202,
      "grad_norm": 2.8925414085388184,
      "learning_rate": 7.457188225260763e-06,
      "loss": 0.9719,
      "step": 8820
    },
    {
      "epoch": 0.7128224812622477,
      "grad_norm": 3.1155261993408203,
      "learning_rate": 7.456618311876551e-06,
      "loss": 0.93,
      "step": 8821
    },
    {
      "epoch": 0.7129032909755753,
      "grad_norm": 2.681637763977051,
      "learning_rate": 7.456048356416381e-06,
      "loss": 0.8978,
      "step": 8822
    },
    {
      "epoch": 0.7129841006889028,
      "grad_norm": 2.8793821334838867,
      "learning_rate": 7.455478358890016e-06,
      "loss": 0.9453,
      "step": 8823
    },
    {
      "epoch": 0.7130649104022303,
      "grad_norm": 2.7457947731018066,
      "learning_rate": 7.454908319307218e-06,
      "loss": 0.8971,
      "step": 8824
    },
    {
      "epoch": 0.7131457201155579,
      "grad_norm": 2.4251136779785156,
      "learning_rate": 7.45433823767775e-06,
      "loss": 0.9628,
      "step": 8825
    },
    {
      "epoch": 0.7132265298288855,
      "grad_norm": 2.7069544792175293,
      "learning_rate": 7.453768114011377e-06,
      "loss": 0.9469,
      "step": 8826
    },
    {
      "epoch": 0.7133073395422129,
      "grad_norm": 3.0360372066497803,
      "learning_rate": 7.453197948317864e-06,
      "loss": 0.9199,
      "step": 8827
    },
    {
      "epoch": 0.7133881492555405,
      "grad_norm": 2.604696273803711,
      "learning_rate": 7.4526277406069735e-06,
      "loss": 0.8767,
      "step": 8828
    },
    {
      "epoch": 0.7134689589688681,
      "grad_norm": 2.3818135261535645,
      "learning_rate": 7.4520574908884765e-06,
      "loss": 0.9631,
      "step": 8829
    },
    {
      "epoch": 0.7135497686821956,
      "grad_norm": 2.156619071960449,
      "learning_rate": 7.451487199172136e-06,
      "loss": 1.0937,
      "step": 8830
    },
    {
      "epoch": 0.7136305783955231,
      "grad_norm": 2.309812307357788,
      "learning_rate": 7.450916865467725e-06,
      "loss": 1.0414,
      "step": 8831
    },
    {
      "epoch": 0.7137113881088507,
      "grad_norm": 3.0624377727508545,
      "learning_rate": 7.450346489785006e-06,
      "loss": 1.0111,
      "step": 8832
    },
    {
      "epoch": 0.7137921978221782,
      "grad_norm": 2.6573123931884766,
      "learning_rate": 7.44977607213375e-06,
      "loss": 1.0546,
      "step": 8833
    },
    {
      "epoch": 0.7138730075355058,
      "grad_norm": 2.9727978706359863,
      "learning_rate": 7.4492056125237275e-06,
      "loss": 0.9685,
      "step": 8834
    },
    {
      "epoch": 0.7139538172488333,
      "grad_norm": 2.5815351009368896,
      "learning_rate": 7.44863511096471e-06,
      "loss": 0.859,
      "step": 8835
    },
    {
      "epoch": 0.7140346269621608,
      "grad_norm": 2.6571624279022217,
      "learning_rate": 7.448064567466468e-06,
      "loss": 0.9839,
      "step": 8836
    },
    {
      "epoch": 0.7141154366754884,
      "grad_norm": 2.9274699687957764,
      "learning_rate": 7.447493982038774e-06,
      "loss": 0.9024,
      "step": 8837
    },
    {
      "epoch": 0.714196246388816,
      "grad_norm": 2.851759433746338,
      "learning_rate": 7.446923354691399e-06,
      "loss": 0.9148,
      "step": 8838
    },
    {
      "epoch": 0.7142770561021434,
      "grad_norm": 2.4229416847229004,
      "learning_rate": 7.446352685434117e-06,
      "loss": 0.8639,
      "step": 8839
    },
    {
      "epoch": 0.714357865815471,
      "grad_norm": 3.232449769973755,
      "learning_rate": 7.4457819742767045e-06,
      "loss": 0.9335,
      "step": 8840
    },
    {
      "epoch": 0.7144386755287986,
      "grad_norm": 2.5353729724884033,
      "learning_rate": 7.445211221228934e-06,
      "loss": 0.9065,
      "step": 8841
    },
    {
      "epoch": 0.714519485242126,
      "grad_norm": 2.435084581375122,
      "learning_rate": 7.444640426300581e-06,
      "loss": 0.9874,
      "step": 8842
    },
    {
      "epoch": 0.7146002949554536,
      "grad_norm": 2.8686530590057373,
      "learning_rate": 7.444069589501425e-06,
      "loss": 0.9891,
      "step": 8843
    },
    {
      "epoch": 0.7146811046687812,
      "grad_norm": 2.7953402996063232,
      "learning_rate": 7.443498710841238e-06,
      "loss": 1.0319,
      "step": 8844
    },
    {
      "epoch": 0.7147619143821087,
      "grad_norm": 3.346709728240967,
      "learning_rate": 7.442927790329804e-06,
      "loss": 0.829,
      "step": 8845
    },
    {
      "epoch": 0.7148427240954363,
      "grad_norm": 2.531511068344116,
      "learning_rate": 7.442356827976895e-06,
      "loss": 1.0826,
      "step": 8846
    },
    {
      "epoch": 0.7149235338087638,
      "grad_norm": 2.749894618988037,
      "learning_rate": 7.441785823792294e-06,
      "loss": 0.8799,
      "step": 8847
    },
    {
      "epoch": 0.7150043435220913,
      "grad_norm": 2.744140625,
      "learning_rate": 7.441214777785781e-06,
      "loss": 1.0979,
      "step": 8848
    },
    {
      "epoch": 0.7150851532354189,
      "grad_norm": 2.4637608528137207,
      "learning_rate": 7.440643689967135e-06,
      "loss": 0.904,
      "step": 8849
    },
    {
      "epoch": 0.7151659629487465,
      "grad_norm": 2.5459442138671875,
      "learning_rate": 7.440072560346139e-06,
      "loss": 0.8667,
      "step": 8850
    },
    {
      "epoch": 0.7152467726620739,
      "grad_norm": 2.253364086151123,
      "learning_rate": 7.439501388932574e-06,
      "loss": 0.8629,
      "step": 8851
    },
    {
      "epoch": 0.7153275823754015,
      "grad_norm": 2.4464023113250732,
      "learning_rate": 7.438930175736223e-06,
      "loss": 0.9178,
      "step": 8852
    },
    {
      "epoch": 0.7154083920887291,
      "grad_norm": 2.881850004196167,
      "learning_rate": 7.43835892076687e-06,
      "loss": 1.0262,
      "step": 8853
    },
    {
      "epoch": 0.7154892018020567,
      "grad_norm": 2.4822871685028076,
      "learning_rate": 7.437787624034297e-06,
      "loss": 0.9632,
      "step": 8854
    },
    {
      "epoch": 0.7155700115153841,
      "grad_norm": 2.4680581092834473,
      "learning_rate": 7.437216285548293e-06,
      "loss": 1.0259,
      "step": 8855
    },
    {
      "epoch": 0.7156508212287117,
      "grad_norm": 2.3201491832733154,
      "learning_rate": 7.436644905318639e-06,
      "loss": 0.9346,
      "step": 8856
    },
    {
      "epoch": 0.7157316309420393,
      "grad_norm": 2.338230848312378,
      "learning_rate": 7.4360734833551265e-06,
      "loss": 1.0319,
      "step": 8857
    },
    {
      "epoch": 0.7158124406553668,
      "grad_norm": 2.7920634746551514,
      "learning_rate": 7.435502019667537e-06,
      "loss": 0.9249,
      "step": 8858
    },
    {
      "epoch": 0.7158932503686943,
      "grad_norm": 2.5076868534088135,
      "learning_rate": 7.4349305142656635e-06,
      "loss": 0.9409,
      "step": 8859
    },
    {
      "epoch": 0.7159740600820219,
      "grad_norm": 2.9880881309509277,
      "learning_rate": 7.434358967159292e-06,
      "loss": 0.9233,
      "step": 8860
    },
    {
      "epoch": 0.7160548697953494,
      "grad_norm": 2.8383290767669678,
      "learning_rate": 7.433787378358211e-06,
      "loss": 0.9234,
      "step": 8861
    },
    {
      "epoch": 0.716135679508677,
      "grad_norm": 2.662801504135132,
      "learning_rate": 7.433215747872211e-06,
      "loss": 0.9522,
      "step": 8862
    },
    {
      "epoch": 0.7162164892220045,
      "grad_norm": 2.7637696266174316,
      "learning_rate": 7.432644075711084e-06,
      "loss": 1.0106,
      "step": 8863
    },
    {
      "epoch": 0.716297298935332,
      "grad_norm": 2.501694917678833,
      "learning_rate": 7.432072361884619e-06,
      "loss": 0.8882,
      "step": 8864
    },
    {
      "epoch": 0.7163781086486596,
      "grad_norm": 3.0861642360687256,
      "learning_rate": 7.4315006064026115e-06,
      "loss": 0.9129,
      "step": 8865
    },
    {
      "epoch": 0.7164589183619872,
      "grad_norm": 2.6039481163024902,
      "learning_rate": 7.430928809274851e-06,
      "loss": 0.8577,
      "step": 8866
    },
    {
      "epoch": 0.7165397280753146,
      "grad_norm": 2.7840166091918945,
      "learning_rate": 7.430356970511132e-06,
      "loss": 0.9922,
      "step": 8867
    },
    {
      "epoch": 0.7166205377886422,
      "grad_norm": 2.8131988048553467,
      "learning_rate": 7.429785090121249e-06,
      "loss": 0.8784,
      "step": 8868
    },
    {
      "epoch": 0.7167013475019698,
      "grad_norm": 2.765836000442505,
      "learning_rate": 7.429213168114997e-06,
      "loss": 0.8664,
      "step": 8869
    },
    {
      "epoch": 0.7167821572152973,
      "grad_norm": 3.1490299701690674,
      "learning_rate": 7.4286412045021706e-06,
      "loss": 0.9114,
      "step": 8870
    },
    {
      "epoch": 0.7168629669286248,
      "grad_norm": 2.8300869464874268,
      "learning_rate": 7.428069199292569e-06,
      "loss": 1.0112,
      "step": 8871
    },
    {
      "epoch": 0.7169437766419524,
      "grad_norm": 2.7464821338653564,
      "learning_rate": 7.427497152495986e-06,
      "loss": 1.0271,
      "step": 8872
    },
    {
      "epoch": 0.7170245863552799,
      "grad_norm": 2.8566503524780273,
      "learning_rate": 7.426925064122221e-06,
      "loss": 0.9684,
      "step": 8873
    },
    {
      "epoch": 0.7171053960686075,
      "grad_norm": 2.7995786666870117,
      "learning_rate": 7.426352934181072e-06,
      "loss": 0.9177,
      "step": 8874
    },
    {
      "epoch": 0.717186205781935,
      "grad_norm": 2.7830474376678467,
      "learning_rate": 7.42578076268234e-06,
      "loss": 0.9576,
      "step": 8875
    },
    {
      "epoch": 0.7172670154952625,
      "grad_norm": 2.440385103225708,
      "learning_rate": 7.42520854963582e-06,
      "loss": 0.9043,
      "step": 8876
    },
    {
      "epoch": 0.7173478252085901,
      "grad_norm": 2.4629225730895996,
      "learning_rate": 7.424636295051319e-06,
      "loss": 0.9895,
      "step": 8877
    },
    {
      "epoch": 0.7174286349219177,
      "grad_norm": 2.867295026779175,
      "learning_rate": 7.424063998938634e-06,
      "loss": 0.8997,
      "step": 8878
    },
    {
      "epoch": 0.7175094446352451,
      "grad_norm": 2.884681224822998,
      "learning_rate": 7.423491661307569e-06,
      "loss": 0.9008,
      "step": 8879
    },
    {
      "epoch": 0.7175902543485727,
      "grad_norm": 2.4002974033355713,
      "learning_rate": 7.422919282167926e-06,
      "loss": 1.0467,
      "step": 8880
    },
    {
      "epoch": 0.7176710640619003,
      "grad_norm": 2.7994320392608643,
      "learning_rate": 7.422346861529509e-06,
      "loss": 0.8691,
      "step": 8881
    },
    {
      "epoch": 0.7177518737752278,
      "grad_norm": 2.3295881748199463,
      "learning_rate": 7.421774399402122e-06,
      "loss": 1.0631,
      "step": 8882
    },
    {
      "epoch": 0.7178326834885553,
      "grad_norm": 2.559081554412842,
      "learning_rate": 7.421201895795569e-06,
      "loss": 0.9642,
      "step": 8883
    },
    {
      "epoch": 0.7179134932018829,
      "grad_norm": 2.916477680206299,
      "learning_rate": 7.420629350719656e-06,
      "loss": 0.9586,
      "step": 8884
    },
    {
      "epoch": 0.7179943029152104,
      "grad_norm": 2.9504342079162598,
      "learning_rate": 7.4200567641841905e-06,
      "loss": 0.9844,
      "step": 8885
    },
    {
      "epoch": 0.718075112628538,
      "grad_norm": 2.548313856124878,
      "learning_rate": 7.419484136198978e-06,
      "loss": 0.9824,
      "step": 8886
    },
    {
      "epoch": 0.7181559223418655,
      "grad_norm": 2.7055087089538574,
      "learning_rate": 7.418911466773827e-06,
      "loss": 0.8615,
      "step": 8887
    },
    {
      "epoch": 0.718236732055193,
      "grad_norm": 2.6341552734375,
      "learning_rate": 7.418338755918547e-06,
      "loss": 1.0053,
      "step": 8888
    },
    {
      "epoch": 0.7183175417685206,
      "grad_norm": 2.4791226387023926,
      "learning_rate": 7.417766003642945e-06,
      "loss": 1.2013,
      "step": 8889
    },
    {
      "epoch": 0.7183983514818482,
      "grad_norm": 2.622504711151123,
      "learning_rate": 7.417193209956832e-06,
      "loss": 0.928,
      "step": 8890
    },
    {
      "epoch": 0.7184791611951756,
      "grad_norm": 2.598707914352417,
      "learning_rate": 7.416620374870018e-06,
      "loss": 1.0093,
      "step": 8891
    },
    {
      "epoch": 0.7185599709085032,
      "grad_norm": 3.0351829528808594,
      "learning_rate": 7.416047498392316e-06,
      "loss": 0.8535,
      "step": 8892
    },
    {
      "epoch": 0.7186407806218308,
      "grad_norm": 2.9704062938690186,
      "learning_rate": 7.415474580533535e-06,
      "loss": 0.9562,
      "step": 8893
    },
    {
      "epoch": 0.7187215903351583,
      "grad_norm": 2.4888765811920166,
      "learning_rate": 7.414901621303492e-06,
      "loss": 1.039,
      "step": 8894
    },
    {
      "epoch": 0.7188024000484858,
      "grad_norm": 2.8852639198303223,
      "learning_rate": 7.414328620711994e-06,
      "loss": 0.8544,
      "step": 8895
    },
    {
      "epoch": 0.7188832097618134,
      "grad_norm": 2.756049871444702,
      "learning_rate": 7.413755578768863e-06,
      "loss": 0.9316,
      "step": 8896
    },
    {
      "epoch": 0.7189640194751409,
      "grad_norm": 2.4644150733947754,
      "learning_rate": 7.4131824954839075e-06,
      "loss": 1.041,
      "step": 8897
    },
    {
      "epoch": 0.7190448291884685,
      "grad_norm": 2.5486245155334473,
      "learning_rate": 7.4126093708669466e-06,
      "loss": 0.9131,
      "step": 8898
    },
    {
      "epoch": 0.719125638901796,
      "grad_norm": 2.486377716064453,
      "learning_rate": 7.412036204927794e-06,
      "loss": 0.9583,
      "step": 8899
    },
    {
      "epoch": 0.7192064486151235,
      "grad_norm": 2.431185245513916,
      "learning_rate": 7.411462997676269e-06,
      "loss": 1.0265,
      "step": 8900
    },
    {
      "epoch": 0.7192872583284511,
      "grad_norm": 2.976555109024048,
      "learning_rate": 7.4108897491221875e-06,
      "loss": 0.9008,
      "step": 8901
    },
    {
      "epoch": 0.7193680680417787,
      "grad_norm": 2.3858213424682617,
      "learning_rate": 7.410316459275369e-06,
      "loss": 0.8608,
      "step": 8902
    },
    {
      "epoch": 0.7194488777551061,
      "grad_norm": 2.9575653076171875,
      "learning_rate": 7.409743128145632e-06,
      "loss": 0.8934,
      "step": 8903
    },
    {
      "epoch": 0.7195296874684337,
      "grad_norm": 2.8733882904052734,
      "learning_rate": 7.409169755742797e-06,
      "loss": 0.9504,
      "step": 8904
    },
    {
      "epoch": 0.7196104971817613,
      "grad_norm": 3.163473129272461,
      "learning_rate": 7.408596342076684e-06,
      "loss": 0.95,
      "step": 8905
    },
    {
      "epoch": 0.7196913068950888,
      "grad_norm": 2.6298673152923584,
      "learning_rate": 7.4080228871571125e-06,
      "loss": 0.8962,
      "step": 8906
    },
    {
      "epoch": 0.7197721166084163,
      "grad_norm": 2.399029016494751,
      "learning_rate": 7.407449390993907e-06,
      "loss": 0.8262,
      "step": 8907
    },
    {
      "epoch": 0.7198529263217439,
      "grad_norm": 2.5172319412231445,
      "learning_rate": 7.40687585359689e-06,
      "loss": 1.0425,
      "step": 8908
    },
    {
      "epoch": 0.7199337360350714,
      "grad_norm": 2.794080972671509,
      "learning_rate": 7.406302274975883e-06,
      "loss": 0.9308,
      "step": 8909
    },
    {
      "epoch": 0.720014545748399,
      "grad_norm": 2.1614184379577637,
      "learning_rate": 7.405728655140711e-06,
      "loss": 0.9056,
      "step": 8910
    },
    {
      "epoch": 0.7200953554617265,
      "grad_norm": 2.6804192066192627,
      "learning_rate": 7.405154994101198e-06,
      "loss": 0.8006,
      "step": 8911
    },
    {
      "epoch": 0.720176165175054,
      "grad_norm": 2.838224411010742,
      "learning_rate": 7.404581291867172e-06,
      "loss": 0.9918,
      "step": 8912
    },
    {
      "epoch": 0.7202569748883816,
      "grad_norm": 2.758174419403076,
      "learning_rate": 7.404007548448455e-06,
      "loss": 0.9579,
      "step": 8913
    },
    {
      "epoch": 0.7203377846017092,
      "grad_norm": 2.366940975189209,
      "learning_rate": 7.403433763854878e-06,
      "loss": 1.0412,
      "step": 8914
    },
    {
      "epoch": 0.7204185943150366,
      "grad_norm": 2.428494930267334,
      "learning_rate": 7.402859938096265e-06,
      "loss": 1.0465,
      "step": 8915
    },
    {
      "epoch": 0.7204994040283642,
      "grad_norm": 2.558030843734741,
      "learning_rate": 7.402286071182449e-06,
      "loss": 0.9653,
      "step": 8916
    },
    {
      "epoch": 0.7205802137416918,
      "grad_norm": 2.6570165157318115,
      "learning_rate": 7.401712163123252e-06,
      "loss": 0.9815,
      "step": 8917
    },
    {
      "epoch": 0.7206610234550193,
      "grad_norm": 2.3573553562164307,
      "learning_rate": 7.4011382139285105e-06,
      "loss": 0.9072,
      "step": 8918
    },
    {
      "epoch": 0.7207418331683468,
      "grad_norm": 2.363840103149414,
      "learning_rate": 7.40056422360805e-06,
      "loss": 0.9098,
      "step": 8919
    },
    {
      "epoch": 0.7208226428816744,
      "grad_norm": 2.896048069000244,
      "learning_rate": 7.399990192171704e-06,
      "loss": 1.0659,
      "step": 8920
    },
    {
      "epoch": 0.7209034525950019,
      "grad_norm": 2.909825563430786,
      "learning_rate": 7.3994161196293035e-06,
      "loss": 0.9126,
      "step": 8921
    },
    {
      "epoch": 0.7209842623083295,
      "grad_norm": 2.5531294345855713,
      "learning_rate": 7.398842005990683e-06,
      "loss": 0.9924,
      "step": 8922
    },
    {
      "epoch": 0.721065072021657,
      "grad_norm": 2.868154764175415,
      "learning_rate": 7.398267851265671e-06,
      "loss": 0.8268,
      "step": 8923
    },
    {
      "epoch": 0.7211458817349845,
      "grad_norm": 3.251495838165283,
      "learning_rate": 7.397693655464106e-06,
      "loss": 1.0013,
      "step": 8924
    },
    {
      "epoch": 0.7212266914483121,
      "grad_norm": 2.6639227867126465,
      "learning_rate": 7.3971194185958206e-06,
      "loss": 1.063,
      "step": 8925
    },
    {
      "epoch": 0.7213075011616397,
      "grad_norm": 2.635226011276245,
      "learning_rate": 7.396545140670651e-06,
      "loss": 0.8988,
      "step": 8926
    },
    {
      "epoch": 0.7213883108749671,
      "grad_norm": 2.7212259769439697,
      "learning_rate": 7.395970821698433e-06,
      "loss": 0.8856,
      "step": 8927
    },
    {
      "epoch": 0.7214691205882947,
      "grad_norm": 2.4809279441833496,
      "learning_rate": 7.395396461689001e-06,
      "loss": 1.1287,
      "step": 8928
    },
    {
      "epoch": 0.7215499303016223,
      "grad_norm": 2.903891086578369,
      "learning_rate": 7.394822060652196e-06,
      "loss": 0.9249,
      "step": 8929
    },
    {
      "epoch": 0.7216307400149498,
      "grad_norm": 2.535151243209839,
      "learning_rate": 7.394247618597854e-06,
      "loss": 0.9742,
      "step": 8930
    },
    {
      "epoch": 0.7217115497282773,
      "grad_norm": 2.8602733612060547,
      "learning_rate": 7.393673135535812e-06,
      "loss": 1.0196,
      "step": 8931
    },
    {
      "epoch": 0.7217923594416049,
      "grad_norm": 2.4082558155059814,
      "learning_rate": 7.393098611475915e-06,
      "loss": 1.0145,
      "step": 8932
    },
    {
      "epoch": 0.7218731691549324,
      "grad_norm": 2.844003438949585,
      "learning_rate": 7.392524046427998e-06,
      "loss": 0.9257,
      "step": 8933
    },
    {
      "epoch": 0.72195397886826,
      "grad_norm": 2.696645736694336,
      "learning_rate": 7.3919494404019045e-06,
      "loss": 0.9545,
      "step": 8934
    },
    {
      "epoch": 0.7220347885815875,
      "grad_norm": 3.0856778621673584,
      "learning_rate": 7.391374793407475e-06,
      "loss": 1.0015,
      "step": 8935
    },
    {
      "epoch": 0.722115598294915,
      "grad_norm": 2.524808168411255,
      "learning_rate": 7.390800105454553e-06,
      "loss": 0.9554,
      "step": 8936
    },
    {
      "epoch": 0.7221964080082426,
      "grad_norm": 2.7753970623016357,
      "learning_rate": 7.390225376552981e-06,
      "loss": 0.9749,
      "step": 8937
    },
    {
      "epoch": 0.7222772177215702,
      "grad_norm": 2.7223031520843506,
      "learning_rate": 7.3896506067126015e-06,
      "loss": 1.0576,
      "step": 8938
    },
    {
      "epoch": 0.7223580274348976,
      "grad_norm": 2.9543721675872803,
      "learning_rate": 7.38907579594326e-06,
      "loss": 0.8495,
      "step": 8939
    },
    {
      "epoch": 0.7224388371482252,
      "grad_norm": 2.7078185081481934,
      "learning_rate": 7.3885009442548024e-06,
      "loss": 1.0531,
      "step": 8940
    },
    {
      "epoch": 0.7225196468615528,
      "grad_norm": 2.3410189151763916,
      "learning_rate": 7.387926051657074e-06,
      "loss": 0.9603,
      "step": 8941
    },
    {
      "epoch": 0.7226004565748803,
      "grad_norm": 2.4575016498565674,
      "learning_rate": 7.38735111815992e-06,
      "loss": 0.8717,
      "step": 8942
    },
    {
      "epoch": 0.7226812662882078,
      "grad_norm": 2.453535318374634,
      "learning_rate": 7.386776143773189e-06,
      "loss": 0.7974,
      "step": 8943
    },
    {
      "epoch": 0.7227620760015354,
      "grad_norm": 2.7958028316497803,
      "learning_rate": 7.386201128506728e-06,
      "loss": 0.8795,
      "step": 8944
    },
    {
      "epoch": 0.7228428857148629,
      "grad_norm": 3.515575647354126,
      "learning_rate": 7.385626072370387e-06,
      "loss": 0.9137,
      "step": 8945
    },
    {
      "epoch": 0.7229236954281905,
      "grad_norm": 2.450437307357788,
      "learning_rate": 7.385050975374014e-06,
      "loss": 0.9303,
      "step": 8946
    },
    {
      "epoch": 0.723004505141518,
      "grad_norm": 2.5931622982025146,
      "learning_rate": 7.384475837527461e-06,
      "loss": 0.8807,
      "step": 8947
    },
    {
      "epoch": 0.7230853148548455,
      "grad_norm": 2.7522566318511963,
      "learning_rate": 7.383900658840576e-06,
      "loss": 0.8882,
      "step": 8948
    },
    {
      "epoch": 0.7231661245681731,
      "grad_norm": 2.3719565868377686,
      "learning_rate": 7.383325439323212e-06,
      "loss": 0.874,
      "step": 8949
    },
    {
      "epoch": 0.7232469342815007,
      "grad_norm": 3.017683744430542,
      "learning_rate": 7.382750178985221e-06,
      "loss": 0.7991,
      "step": 8950
    },
    {
      "epoch": 0.7233277439948281,
      "grad_norm": 2.237924575805664,
      "learning_rate": 7.382174877836456e-06,
      "loss": 0.9674,
      "step": 8951
    },
    {
      "epoch": 0.7234085537081557,
      "grad_norm": 2.619729518890381,
      "learning_rate": 7.381599535886768e-06,
      "loss": 0.8787,
      "step": 8952
    },
    {
      "epoch": 0.7234893634214833,
      "grad_norm": 2.3338027000427246,
      "learning_rate": 7.381024153146016e-06,
      "loss": 0.8498,
      "step": 8953
    },
    {
      "epoch": 0.7235701731348108,
      "grad_norm": 2.7453088760375977,
      "learning_rate": 7.380448729624051e-06,
      "loss": 1.0517,
      "step": 8954
    },
    {
      "epoch": 0.7236509828481383,
      "grad_norm": 2.635988473892212,
      "learning_rate": 7.379873265330732e-06,
      "loss": 0.9731,
      "step": 8955
    },
    {
      "epoch": 0.7237317925614659,
      "grad_norm": 3.369051933288574,
      "learning_rate": 7.379297760275911e-06,
      "loss": 0.9114,
      "step": 8956
    },
    {
      "epoch": 0.7238126022747934,
      "grad_norm": 2.868903636932373,
      "learning_rate": 7.378722214469447e-06,
      "loss": 0.9999,
      "step": 8957
    },
    {
      "epoch": 0.723893411988121,
      "grad_norm": 2.838514804840088,
      "learning_rate": 7.378146627921199e-06,
      "loss": 0.9413,
      "step": 8958
    },
    {
      "epoch": 0.7239742217014485,
      "grad_norm": 3.0332632064819336,
      "learning_rate": 7.377571000641024e-06,
      "loss": 0.9165,
      "step": 8959
    },
    {
      "epoch": 0.724055031414776,
      "grad_norm": 2.7605502605438232,
      "learning_rate": 7.3769953326387825e-06,
      "loss": 0.9708,
      "step": 8960
    },
    {
      "epoch": 0.7241358411281036,
      "grad_norm": 2.6083874702453613,
      "learning_rate": 7.376419623924333e-06,
      "loss": 0.9836,
      "step": 8961
    },
    {
      "epoch": 0.7242166508414312,
      "grad_norm": 2.502711057662964,
      "learning_rate": 7.375843874507536e-06,
      "loss": 0.8841,
      "step": 8962
    },
    {
      "epoch": 0.7242974605547586,
      "grad_norm": 3.4860928058624268,
      "learning_rate": 7.375268084398253e-06,
      "loss": 0.9586,
      "step": 8963
    },
    {
      "epoch": 0.7243782702680862,
      "grad_norm": 2.41768217086792,
      "learning_rate": 7.374692253606346e-06,
      "loss": 0.9388,
      "step": 8964
    },
    {
      "epoch": 0.7244590799814138,
      "grad_norm": 2.621381998062134,
      "learning_rate": 7.374116382141679e-06,
      "loss": 0.9396,
      "step": 8965
    },
    {
      "epoch": 0.7245398896947413,
      "grad_norm": 1.9862269163131714,
      "learning_rate": 7.373540470014111e-06,
      "loss": 0.9621,
      "step": 8966
    },
    {
      "epoch": 0.7246206994080688,
      "grad_norm": 2.74418044090271,
      "learning_rate": 7.3729645172335095e-06,
      "loss": 0.8411,
      "step": 8967
    },
    {
      "epoch": 0.7247015091213964,
      "grad_norm": 2.856605291366577,
      "learning_rate": 7.372388523809739e-06,
      "loss": 0.804,
      "step": 8968
    },
    {
      "epoch": 0.7247823188347239,
      "grad_norm": 2.4016685485839844,
      "learning_rate": 7.371812489752665e-06,
      "loss": 0.8569,
      "step": 8969
    },
    {
      "epoch": 0.7248631285480515,
      "grad_norm": 2.506629228591919,
      "learning_rate": 7.371236415072153e-06,
      "loss": 1.0598,
      "step": 8970
    },
    {
      "epoch": 0.724943938261379,
      "grad_norm": 2.507174491882324,
      "learning_rate": 7.37066029977807e-06,
      "loss": 1.0069,
      "step": 8971
    },
    {
      "epoch": 0.7250247479747065,
      "grad_norm": 2.9702212810516357,
      "learning_rate": 7.370084143880282e-06,
      "loss": 0.9635,
      "step": 8972
    },
    {
      "epoch": 0.7251055576880341,
      "grad_norm": 3.0821893215179443,
      "learning_rate": 7.369507947388659e-06,
      "loss": 0.8754,
      "step": 8973
    },
    {
      "epoch": 0.7251863674013617,
      "grad_norm": 2.659006118774414,
      "learning_rate": 7.368931710313068e-06,
      "loss": 0.8889,
      "step": 8974
    },
    {
      "epoch": 0.7252671771146891,
      "grad_norm": 2.7876930236816406,
      "learning_rate": 7.368355432663382e-06,
      "loss": 0.936,
      "step": 8975
    },
    {
      "epoch": 0.7253479868280167,
      "grad_norm": 2.774790048599243,
      "learning_rate": 7.367779114449467e-06,
      "loss": 0.9198,
      "step": 8976
    },
    {
      "epoch": 0.7254287965413443,
      "grad_norm": 2.7417778968811035,
      "learning_rate": 7.367202755681198e-06,
      "loss": 0.7967,
      "step": 8977
    },
    {
      "epoch": 0.7255096062546718,
      "grad_norm": 2.626347303390503,
      "learning_rate": 7.366626356368443e-06,
      "loss": 0.9192,
      "step": 8978
    },
    {
      "epoch": 0.7255904159679993,
      "grad_norm": 2.4893431663513184,
      "learning_rate": 7.3660499165210765e-06,
      "loss": 0.9079,
      "step": 8979
    },
    {
      "epoch": 0.7256712256813269,
      "grad_norm": 2.3547041416168213,
      "learning_rate": 7.365473436148971e-06,
      "loss": 0.844,
      "step": 8980
    },
    {
      "epoch": 0.7257520353946544,
      "grad_norm": 2.3203346729278564,
      "learning_rate": 7.3648969152619995e-06,
      "loss": 0.9682,
      "step": 8981
    },
    {
      "epoch": 0.725832845107982,
      "grad_norm": 2.567094564437866,
      "learning_rate": 7.364320353870038e-06,
      "loss": 0.9329,
      "step": 8982
    },
    {
      "epoch": 0.7259136548213095,
      "grad_norm": 2.4649698734283447,
      "learning_rate": 7.36374375198296e-06,
      "loss": 0.9271,
      "step": 8983
    },
    {
      "epoch": 0.7259944645346371,
      "grad_norm": 2.674482583999634,
      "learning_rate": 7.363167109610641e-06,
      "loss": 0.9464,
      "step": 8984
    },
    {
      "epoch": 0.7260752742479646,
      "grad_norm": 2.594151020050049,
      "learning_rate": 7.362590426762961e-06,
      "loss": 0.8646,
      "step": 8985
    },
    {
      "epoch": 0.7261560839612922,
      "grad_norm": 2.744310140609741,
      "learning_rate": 7.362013703449794e-06,
      "loss": 0.9258,
      "step": 8986
    },
    {
      "epoch": 0.7262368936746197,
      "grad_norm": 2.228095293045044,
      "learning_rate": 7.3614369396810185e-06,
      "loss": 0.8247,
      "step": 8987
    },
    {
      "epoch": 0.7263177033879472,
      "grad_norm": 3.085430383682251,
      "learning_rate": 7.360860135466512e-06,
      "loss": 0.8957,
      "step": 8988
    },
    {
      "epoch": 0.7263985131012748,
      "grad_norm": 2.681002140045166,
      "learning_rate": 7.360283290816157e-06,
      "loss": 0.8328,
      "step": 8989
    },
    {
      "epoch": 0.7264793228146024,
      "grad_norm": 2.853421449661255,
      "learning_rate": 7.3597064057398285e-06,
      "loss": 0.9714,
      "step": 8990
    },
    {
      "epoch": 0.7265601325279298,
      "grad_norm": 2.6571035385131836,
      "learning_rate": 7.359129480247412e-06,
      "loss": 0.9846,
      "step": 8991
    },
    {
      "epoch": 0.7266409422412574,
      "grad_norm": 2.6871957778930664,
      "learning_rate": 7.358552514348787e-06,
      "loss": 0.966,
      "step": 8992
    },
    {
      "epoch": 0.726721751954585,
      "grad_norm": 2.384361982345581,
      "learning_rate": 7.357975508053834e-06,
      "loss": 0.9148,
      "step": 8993
    },
    {
      "epoch": 0.7268025616679125,
      "grad_norm": 2.662299633026123,
      "learning_rate": 7.357398461372438e-06,
      "loss": 1.0036,
      "step": 8994
    },
    {
      "epoch": 0.72688337138124,
      "grad_norm": 2.670008897781372,
      "learning_rate": 7.356821374314482e-06,
      "loss": 0.9544,
      "step": 8995
    },
    {
      "epoch": 0.7269641810945676,
      "grad_norm": 2.415121555328369,
      "learning_rate": 7.3562442468898485e-06,
      "loss": 0.9899,
      "step": 8996
    },
    {
      "epoch": 0.7270449908078951,
      "grad_norm": 2.4513497352600098,
      "learning_rate": 7.355667079108425e-06,
      "loss": 0.9313,
      "step": 8997
    },
    {
      "epoch": 0.7271258005212227,
      "grad_norm": 2.724459648132324,
      "learning_rate": 7.355089870980094e-06,
      "loss": 1.0078,
      "step": 8998
    },
    {
      "epoch": 0.7272066102345502,
      "grad_norm": 2.392666816711426,
      "learning_rate": 7.354512622514744e-06,
      "loss": 0.9195,
      "step": 8999
    },
    {
      "epoch": 0.7272874199478777,
      "grad_norm": 2.799349308013916,
      "learning_rate": 7.353935333722262e-06,
      "loss": 0.8417,
      "step": 9000
    },
    {
      "epoch": 0.7272874199478777,
      "eval_loss": 0.7830251455307007,
      "eval_runtime": 815.0654,
      "eval_samples_per_second": 102.281,
      "eval_steps_per_second": 12.785,
      "step": 9000
    },
    {
      "epoch": 0.7273682296612053,
      "grad_norm": 2.228584051132202,
      "learning_rate": 7.353358004612533e-06,
      "loss": 0.951,
      "step": 9001
    },
    {
      "epoch": 0.7274490393745329,
      "grad_norm": 2.6647536754608154,
      "learning_rate": 7.352780635195446e-06,
      "loss": 0.8601,
      "step": 9002
    },
    {
      "epoch": 0.7275298490878603,
      "grad_norm": 2.371415138244629,
      "learning_rate": 7.352203225480893e-06,
      "loss": 0.916,
      "step": 9003
    },
    {
      "epoch": 0.7276106588011879,
      "grad_norm": 2.6645593643188477,
      "learning_rate": 7.351625775478761e-06,
      "loss": 0.9452,
      "step": 9004
    },
    {
      "epoch": 0.7276914685145155,
      "grad_norm": 2.236506462097168,
      "learning_rate": 7.35104828519894e-06,
      "loss": 1.0534,
      "step": 9005
    },
    {
      "epoch": 0.727772278227843,
      "grad_norm": 2.4360313415527344,
      "learning_rate": 7.350470754651322e-06,
      "loss": 0.9478,
      "step": 9006
    },
    {
      "epoch": 0.7278530879411705,
      "grad_norm": 2.6017050743103027,
      "learning_rate": 7.3498931838458e-06,
      "loss": 0.8245,
      "step": 9007
    },
    {
      "epoch": 0.7279338976544981,
      "grad_norm": 2.8063132762908936,
      "learning_rate": 7.349315572792262e-06,
      "loss": 1.0436,
      "step": 9008
    },
    {
      "epoch": 0.7280147073678256,
      "grad_norm": 2.9765968322753906,
      "learning_rate": 7.348737921500606e-06,
      "loss": 0.9433,
      "step": 9009
    },
    {
      "epoch": 0.7280955170811532,
      "grad_norm": 2.7164366245269775,
      "learning_rate": 7.348160229980723e-06,
      "loss": 0.8589,
      "step": 9010
    },
    {
      "epoch": 0.7281763267944807,
      "grad_norm": 2.619481325149536,
      "learning_rate": 7.347582498242509e-06,
      "loss": 0.9097,
      "step": 9011
    },
    {
      "epoch": 0.7282571365078082,
      "grad_norm": 2.7392752170562744,
      "learning_rate": 7.347004726295857e-06,
      "loss": 0.9423,
      "step": 9012
    },
    {
      "epoch": 0.7283379462211358,
      "grad_norm": 2.6279306411743164,
      "learning_rate": 7.3464269141506665e-06,
      "loss": 0.9999,
      "step": 9013
    },
    {
      "epoch": 0.7284187559344634,
      "grad_norm": 2.804171085357666,
      "learning_rate": 7.3458490618168295e-06,
      "loss": 1.1039,
      "step": 9014
    },
    {
      "epoch": 0.7284995656477908,
      "grad_norm": 3.0502655506134033,
      "learning_rate": 7.345271169304246e-06,
      "loss": 0.8591,
      "step": 9015
    },
    {
      "epoch": 0.7285803753611184,
      "grad_norm": 2.96559739112854,
      "learning_rate": 7.3446932366228155e-06,
      "loss": 0.9103,
      "step": 9016
    },
    {
      "epoch": 0.728661185074446,
      "grad_norm": 2.4157369136810303,
      "learning_rate": 7.344115263782432e-06,
      "loss": 1.0023,
      "step": 9017
    },
    {
      "epoch": 0.7287419947877735,
      "grad_norm": 2.7634778022766113,
      "learning_rate": 7.343537250792998e-06,
      "loss": 0.9369,
      "step": 9018
    },
    {
      "epoch": 0.728822804501101,
      "grad_norm": 2.900740385055542,
      "learning_rate": 7.342959197664412e-06,
      "loss": 1.1396,
      "step": 9019
    },
    {
      "epoch": 0.7289036142144286,
      "grad_norm": 2.649463653564453,
      "learning_rate": 7.342381104406576e-06,
      "loss": 0.9305,
      "step": 9020
    },
    {
      "epoch": 0.7289844239277561,
      "grad_norm": 3.1890182495117188,
      "learning_rate": 7.34180297102939e-06,
      "loss": 0.9291,
      "step": 9021
    },
    {
      "epoch": 0.7290652336410837,
      "grad_norm": 2.8196585178375244,
      "learning_rate": 7.3412247975427586e-06,
      "loss": 0.8877,
      "step": 9022
    },
    {
      "epoch": 0.7291460433544112,
      "grad_norm": 2.358859062194824,
      "learning_rate": 7.340646583956582e-06,
      "loss": 0.9803,
      "step": 9023
    },
    {
      "epoch": 0.7292268530677387,
      "grad_norm": 2.893080472946167,
      "learning_rate": 7.340068330280764e-06,
      "loss": 0.9453,
      "step": 9024
    },
    {
      "epoch": 0.7293076627810663,
      "grad_norm": 2.7734615802764893,
      "learning_rate": 7.339490036525208e-06,
      "loss": 0.9046,
      "step": 9025
    },
    {
      "epoch": 0.7293884724943939,
      "grad_norm": 2.8442749977111816,
      "learning_rate": 7.338911702699822e-06,
      "loss": 1.0598,
      "step": 9026
    },
    {
      "epoch": 0.7294692822077213,
      "grad_norm": 2.563047170639038,
      "learning_rate": 7.338333328814507e-06,
      "loss": 0.9657,
      "step": 9027
    },
    {
      "epoch": 0.7295500919210489,
      "grad_norm": 2.7017898559570312,
      "learning_rate": 7.337754914879174e-06,
      "loss": 1.0422,
      "step": 9028
    },
    {
      "epoch": 0.7296309016343765,
      "grad_norm": 2.9704360961914062,
      "learning_rate": 7.3371764609037236e-06,
      "loss": 0.784,
      "step": 9029
    },
    {
      "epoch": 0.729711711347704,
      "grad_norm": 2.824162006378174,
      "learning_rate": 7.33659796689807e-06,
      "loss": 0.9321,
      "step": 9030
    },
    {
      "epoch": 0.7297925210610315,
      "grad_norm": 2.1505091190338135,
      "learning_rate": 7.336019432872117e-06,
      "loss": 0.9972,
      "step": 9031
    },
    {
      "epoch": 0.7298733307743591,
      "grad_norm": 2.6777167320251465,
      "learning_rate": 7.335440858835775e-06,
      "loss": 0.9419,
      "step": 9032
    },
    {
      "epoch": 0.7299541404876866,
      "grad_norm": 2.3954265117645264,
      "learning_rate": 7.334862244798953e-06,
      "loss": 1.0433,
      "step": 9033
    },
    {
      "epoch": 0.7300349502010142,
      "grad_norm": 2.8312015533447266,
      "learning_rate": 7.3342835907715625e-06,
      "loss": 0.9176,
      "step": 9034
    },
    {
      "epoch": 0.7301157599143417,
      "grad_norm": 3.6947519779205322,
      "learning_rate": 7.3337048967635135e-06,
      "loss": 0.8934,
      "step": 9035
    },
    {
      "epoch": 0.7301965696276692,
      "grad_norm": 2.8133158683776855,
      "learning_rate": 7.333126162784718e-06,
      "loss": 0.9151,
      "step": 9036
    },
    {
      "epoch": 0.7302773793409968,
      "grad_norm": 2.875257730484009,
      "learning_rate": 7.332547388845087e-06,
      "loss": 0.8651,
      "step": 9037
    },
    {
      "epoch": 0.7303581890543244,
      "grad_norm": 2.589542865753174,
      "learning_rate": 7.331968574954537e-06,
      "loss": 0.9141,
      "step": 9038
    },
    {
      "epoch": 0.7304389987676518,
      "grad_norm": 2.429004192352295,
      "learning_rate": 7.331389721122977e-06,
      "loss": 0.9255,
      "step": 9039
    },
    {
      "epoch": 0.7305198084809794,
      "grad_norm": 2.7485947608947754,
      "learning_rate": 7.330810827360324e-06,
      "loss": 0.9342,
      "step": 9040
    },
    {
      "epoch": 0.730600618194307,
      "grad_norm": 2.973147392272949,
      "learning_rate": 7.330231893676494e-06,
      "loss": 0.9602,
      "step": 9041
    },
    {
      "epoch": 0.7306814279076345,
      "grad_norm": 3.019298553466797,
      "learning_rate": 7.3296529200814005e-06,
      "loss": 1.0176,
      "step": 9042
    },
    {
      "epoch": 0.730762237620962,
      "grad_norm": 2.4859468936920166,
      "learning_rate": 7.32907390658496e-06,
      "loss": 0.9742,
      "step": 9043
    },
    {
      "epoch": 0.7308430473342896,
      "grad_norm": 2.8655483722686768,
      "learning_rate": 7.328494853197092e-06,
      "loss": 0.8597,
      "step": 9044
    },
    {
      "epoch": 0.7309238570476171,
      "grad_norm": 2.3558311462402344,
      "learning_rate": 7.327915759927713e-06,
      "loss": 0.9458,
      "step": 9045
    },
    {
      "epoch": 0.7310046667609447,
      "grad_norm": 2.892517328262329,
      "learning_rate": 7.327336626786739e-06,
      "loss": 0.8994,
      "step": 9046
    },
    {
      "epoch": 0.7310854764742722,
      "grad_norm": 2.5985705852508545,
      "learning_rate": 7.326757453784094e-06,
      "loss": 1.0047,
      "step": 9047
    },
    {
      "epoch": 0.7311662861875997,
      "grad_norm": 2.7899577617645264,
      "learning_rate": 7.326178240929693e-06,
      "loss": 0.9206,
      "step": 9048
    },
    {
      "epoch": 0.7312470959009273,
      "grad_norm": 2.733816146850586,
      "learning_rate": 7.32559898823346e-06,
      "loss": 0.8699,
      "step": 9049
    },
    {
      "epoch": 0.7313279056142549,
      "grad_norm": 2.8018996715545654,
      "learning_rate": 7.325019695705317e-06,
      "loss": 0.9548,
      "step": 9050
    },
    {
      "epoch": 0.7314087153275823,
      "grad_norm": 2.394652843475342,
      "learning_rate": 7.32444036335518e-06,
      "loss": 0.9311,
      "step": 9051
    },
    {
      "epoch": 0.7314895250409099,
      "grad_norm": 2.5981502532958984,
      "learning_rate": 7.323860991192978e-06,
      "loss": 0.8349,
      "step": 9052
    },
    {
      "epoch": 0.7315703347542375,
      "grad_norm": 2.5200812816619873,
      "learning_rate": 7.32328157922863e-06,
      "loss": 0.9657,
      "step": 9053
    },
    {
      "epoch": 0.731651144467565,
      "grad_norm": 2.7489216327667236,
      "learning_rate": 7.322702127472063e-06,
      "loss": 0.953,
      "step": 9054
    },
    {
      "epoch": 0.7317319541808925,
      "grad_norm": 2.780418872833252,
      "learning_rate": 7.3221226359332e-06,
      "loss": 0.8601,
      "step": 9055
    },
    {
      "epoch": 0.7318127638942201,
      "grad_norm": 3.3629696369171143,
      "learning_rate": 7.321543104621967e-06,
      "loss": 0.9465,
      "step": 9056
    },
    {
      "epoch": 0.7318935736075476,
      "grad_norm": 2.8562209606170654,
      "learning_rate": 7.3209635335482874e-06,
      "loss": 0.9981,
      "step": 9057
    },
    {
      "epoch": 0.7319743833208752,
      "grad_norm": 2.8071959018707275,
      "learning_rate": 7.3203839227220915e-06,
      "loss": 0.8345,
      "step": 9058
    },
    {
      "epoch": 0.7320551930342027,
      "grad_norm": 2.947610855102539,
      "learning_rate": 7.319804272153306e-06,
      "loss": 0.8278,
      "step": 9059
    },
    {
      "epoch": 0.7321360027475302,
      "grad_norm": 2.4837849140167236,
      "learning_rate": 7.319224581851857e-06,
      "loss": 0.9373,
      "step": 9060
    },
    {
      "epoch": 0.7322168124608578,
      "grad_norm": 2.406834602355957,
      "learning_rate": 7.318644851827674e-06,
      "loss": 0.8845,
      "step": 9061
    },
    {
      "epoch": 0.7322976221741854,
      "grad_norm": 2.5397162437438965,
      "learning_rate": 7.318065082090686e-06,
      "loss": 0.9404,
      "step": 9062
    },
    {
      "epoch": 0.7323784318875128,
      "grad_norm": 2.8372175693511963,
      "learning_rate": 7.317485272650825e-06,
      "loss": 0.9295,
      "step": 9063
    },
    {
      "epoch": 0.7324592416008404,
      "grad_norm": 2.714838981628418,
      "learning_rate": 7.31690542351802e-06,
      "loss": 0.9338,
      "step": 9064
    },
    {
      "epoch": 0.732540051314168,
      "grad_norm": 2.313499927520752,
      "learning_rate": 7.316325534702202e-06,
      "loss": 1.0098,
      "step": 9065
    },
    {
      "epoch": 0.7326208610274955,
      "grad_norm": 2.831343173980713,
      "learning_rate": 7.315745606213305e-06,
      "loss": 0.9202,
      "step": 9066
    },
    {
      "epoch": 0.732701670740823,
      "grad_norm": 2.8902063369750977,
      "learning_rate": 7.315165638061262e-06,
      "loss": 0.9094,
      "step": 9067
    },
    {
      "epoch": 0.7327824804541506,
      "grad_norm": 2.2195374965667725,
      "learning_rate": 7.314585630256001e-06,
      "loss": 0.9287,
      "step": 9068
    },
    {
      "epoch": 0.7328632901674781,
      "grad_norm": 2.9071133136749268,
      "learning_rate": 7.314005582807464e-06,
      "loss": 0.9714,
      "step": 9069
    },
    {
      "epoch": 0.7329440998808057,
      "grad_norm": 2.5509865283966064,
      "learning_rate": 7.31342549572558e-06,
      "loss": 0.9228,
      "step": 9070
    },
    {
      "epoch": 0.7330249095941332,
      "grad_norm": 2.6809580326080322,
      "learning_rate": 7.3128453690202875e-06,
      "loss": 1.0213,
      "step": 9071
    },
    {
      "epoch": 0.7331057193074607,
      "grad_norm": 2.3250210285186768,
      "learning_rate": 7.312265202701523e-06,
      "loss": 1.0098,
      "step": 9072
    },
    {
      "epoch": 0.7331865290207883,
      "grad_norm": 2.799264669418335,
      "learning_rate": 7.31168499677922e-06,
      "loss": 1.0227,
      "step": 9073
    },
    {
      "epoch": 0.7332673387341159,
      "grad_norm": 2.917973756790161,
      "learning_rate": 7.311104751263319e-06,
      "loss": 0.9799,
      "step": 9074
    },
    {
      "epoch": 0.7333481484474433,
      "grad_norm": 2.7183284759521484,
      "learning_rate": 7.310524466163758e-06,
      "loss": 0.97,
      "step": 9075
    },
    {
      "epoch": 0.7334289581607709,
      "grad_norm": 2.3030471801757812,
      "learning_rate": 7.309944141490474e-06,
      "loss": 0.9627,
      "step": 9076
    },
    {
      "epoch": 0.7335097678740985,
      "grad_norm": 2.5966598987579346,
      "learning_rate": 7.309363777253409e-06,
      "loss": 1.0361,
      "step": 9077
    },
    {
      "epoch": 0.733590577587426,
      "grad_norm": 2.7329165935516357,
      "learning_rate": 7.3087833734625e-06,
      "loss": 0.8653,
      "step": 9078
    },
    {
      "epoch": 0.7336713873007535,
      "grad_norm": 2.536996364593506,
      "learning_rate": 7.308202930127693e-06,
      "loss": 0.9458,
      "step": 9079
    },
    {
      "epoch": 0.7337521970140811,
      "grad_norm": 2.4987056255340576,
      "learning_rate": 7.307622447258925e-06,
      "loss": 1.0911,
      "step": 9080
    },
    {
      "epoch": 0.7338330067274086,
      "grad_norm": 2.4029359817504883,
      "learning_rate": 7.307041924866139e-06,
      "loss": 1.1116,
      "step": 9081
    },
    {
      "epoch": 0.7339138164407362,
      "grad_norm": 2.360539197921753,
      "learning_rate": 7.3064613629592806e-06,
      "loss": 0.9258,
      "step": 9082
    },
    {
      "epoch": 0.7339946261540637,
      "grad_norm": 2.5612308979034424,
      "learning_rate": 7.305880761548291e-06,
      "loss": 0.9422,
      "step": 9083
    },
    {
      "epoch": 0.7340754358673912,
      "grad_norm": 3.0456793308258057,
      "learning_rate": 7.305300120643114e-06,
      "loss": 0.8824,
      "step": 9084
    },
    {
      "epoch": 0.7341562455807188,
      "grad_norm": 2.5850014686584473,
      "learning_rate": 7.304719440253697e-06,
      "loss": 0.9751,
      "step": 9085
    },
    {
      "epoch": 0.7342370552940464,
      "grad_norm": 2.634831428527832,
      "learning_rate": 7.304138720389984e-06,
      "loss": 0.9048,
      "step": 9086
    },
    {
      "epoch": 0.7343178650073738,
      "grad_norm": 2.5613815784454346,
      "learning_rate": 7.3035579610619225e-06,
      "loss": 0.83,
      "step": 9087
    },
    {
      "epoch": 0.7343986747207014,
      "grad_norm": 3.141700506210327,
      "learning_rate": 7.302977162279457e-06,
      "loss": 0.9762,
      "step": 9088
    },
    {
      "epoch": 0.734479484434029,
      "grad_norm": 2.335719585418701,
      "learning_rate": 7.3023963240525385e-06,
      "loss": 0.9933,
      "step": 9089
    },
    {
      "epoch": 0.7345602941473565,
      "grad_norm": 2.5486817359924316,
      "learning_rate": 7.301815446391113e-06,
      "loss": 0.9592,
      "step": 9090
    },
    {
      "epoch": 0.734641103860684,
      "grad_norm": 2.6970295906066895,
      "learning_rate": 7.30123452930513e-06,
      "loss": 0.8509,
      "step": 9091
    },
    {
      "epoch": 0.7347219135740116,
      "grad_norm": 2.5547571182250977,
      "learning_rate": 7.300653572804539e-06,
      "loss": 0.8352,
      "step": 9092
    },
    {
      "epoch": 0.7348027232873391,
      "grad_norm": 2.8848555088043213,
      "learning_rate": 7.300072576899292e-06,
      "loss": 0.9291,
      "step": 9093
    },
    {
      "epoch": 0.7348835330006667,
      "grad_norm": 2.2670273780822754,
      "learning_rate": 7.299491541599338e-06,
      "loss": 1.0142,
      "step": 9094
    },
    {
      "epoch": 0.7349643427139942,
      "grad_norm": 2.5862274169921875,
      "learning_rate": 7.298910466914632e-06,
      "loss": 1.1024,
      "step": 9095
    },
    {
      "epoch": 0.7350451524273217,
      "grad_norm": 2.634575605392456,
      "learning_rate": 7.298329352855121e-06,
      "loss": 0.9297,
      "step": 9096
    },
    {
      "epoch": 0.7351259621406493,
      "grad_norm": 2.481804847717285,
      "learning_rate": 7.297748199430764e-06,
      "loss": 0.9552,
      "step": 9097
    },
    {
      "epoch": 0.7352067718539769,
      "grad_norm": 2.592578411102295,
      "learning_rate": 7.297167006651511e-06,
      "loss": 0.9029,
      "step": 9098
    },
    {
      "epoch": 0.7352875815673043,
      "grad_norm": 2.6386446952819824,
      "learning_rate": 7.296585774527316e-06,
      "loss": 1.0682,
      "step": 9099
    },
    {
      "epoch": 0.7353683912806319,
      "grad_norm": 2.4857497215270996,
      "learning_rate": 7.296004503068137e-06,
      "loss": 0.9442,
      "step": 9100
    },
    {
      "epoch": 0.7354492009939595,
      "grad_norm": 3.2034192085266113,
      "learning_rate": 7.295423192283928e-06,
      "loss": 0.9089,
      "step": 9101
    },
    {
      "epoch": 0.735530010707287,
      "grad_norm": 2.7047171592712402,
      "learning_rate": 7.294841842184645e-06,
      "loss": 1.0233,
      "step": 9102
    },
    {
      "epoch": 0.7356108204206145,
      "grad_norm": 2.326129913330078,
      "learning_rate": 7.294260452780248e-06,
      "loss": 1.0153,
      "step": 9103
    },
    {
      "epoch": 0.7356916301339421,
      "grad_norm": 2.405121326446533,
      "learning_rate": 7.293679024080689e-06,
      "loss": 1.045,
      "step": 9104
    },
    {
      "epoch": 0.7357724398472696,
      "grad_norm": 2.981825828552246,
      "learning_rate": 7.293097556095933e-06,
      "loss": 0.9242,
      "step": 9105
    },
    {
      "epoch": 0.7358532495605972,
      "grad_norm": 2.567793846130371,
      "learning_rate": 7.292516048835936e-06,
      "loss": 1.0347,
      "step": 9106
    },
    {
      "epoch": 0.7359340592739247,
      "grad_norm": 2.643747091293335,
      "learning_rate": 7.2919345023106566e-06,
      "loss": 1.0373,
      "step": 9107
    },
    {
      "epoch": 0.7360148689872522,
      "grad_norm": 2.5018110275268555,
      "learning_rate": 7.291352916530058e-06,
      "loss": 1.0742,
      "step": 9108
    },
    {
      "epoch": 0.7360956787005798,
      "grad_norm": 2.6694023609161377,
      "learning_rate": 7.2907712915041005e-06,
      "loss": 1.0641,
      "step": 9109
    },
    {
      "epoch": 0.7361764884139074,
      "grad_norm": 2.499962329864502,
      "learning_rate": 7.290189627242743e-06,
      "loss": 1.0189,
      "step": 9110
    },
    {
      "epoch": 0.7362572981272348,
      "grad_norm": 2.484152317047119,
      "learning_rate": 7.2896079237559546e-06,
      "loss": 0.9979,
      "step": 9111
    },
    {
      "epoch": 0.7363381078405624,
      "grad_norm": 2.7532966136932373,
      "learning_rate": 7.289026181053691e-06,
      "loss": 0.8251,
      "step": 9112
    },
    {
      "epoch": 0.73641891755389,
      "grad_norm": 2.6327338218688965,
      "learning_rate": 7.288444399145922e-06,
      "loss": 0.9772,
      "step": 9113
    },
    {
      "epoch": 0.7364997272672176,
      "grad_norm": 2.598501443862915,
      "learning_rate": 7.287862578042608e-06,
      "loss": 1.0511,
      "step": 9114
    },
    {
      "epoch": 0.736580536980545,
      "grad_norm": 2.483323335647583,
      "learning_rate": 7.287280717753716e-06,
      "loss": 0.8037,
      "step": 9115
    },
    {
      "epoch": 0.7366613466938726,
      "grad_norm": 2.7200582027435303,
      "learning_rate": 7.2866988182892116e-06,
      "loss": 1.0953,
      "step": 9116
    },
    {
      "epoch": 0.7367421564072002,
      "grad_norm": 2.8144047260284424,
      "learning_rate": 7.286116879659063e-06,
      "loss": 0.9657,
      "step": 9117
    },
    {
      "epoch": 0.7368229661205277,
      "grad_norm": 2.6776721477508545,
      "learning_rate": 7.2855349018732345e-06,
      "loss": 0.9313,
      "step": 9118
    },
    {
      "epoch": 0.7369037758338552,
      "grad_norm": 2.267498016357422,
      "learning_rate": 7.284952884941696e-06,
      "loss": 0.9559,
      "step": 9119
    },
    {
      "epoch": 0.7369845855471828,
      "grad_norm": 2.507622718811035,
      "learning_rate": 7.2843708288744155e-06,
      "loss": 0.9789,
      "step": 9120
    },
    {
      "epoch": 0.7370653952605103,
      "grad_norm": 3.1745150089263916,
      "learning_rate": 7.28378873368136e-06,
      "loss": 0.8794,
      "step": 9121
    },
    {
      "epoch": 0.7371462049738379,
      "grad_norm": 2.592698335647583,
      "learning_rate": 7.283206599372505e-06,
      "loss": 0.9868,
      "step": 9122
    },
    {
      "epoch": 0.7372270146871654,
      "grad_norm": 2.4423863887786865,
      "learning_rate": 7.282624425957816e-06,
      "loss": 0.9078,
      "step": 9123
    },
    {
      "epoch": 0.7373078244004929,
      "grad_norm": 2.5273303985595703,
      "learning_rate": 7.2820422134472635e-06,
      "loss": 0.9141,
      "step": 9124
    },
    {
      "epoch": 0.7373886341138205,
      "grad_norm": 2.8483357429504395,
      "learning_rate": 7.2814599618508255e-06,
      "loss": 0.8186,
      "step": 9125
    },
    {
      "epoch": 0.7374694438271481,
      "grad_norm": 2.963029384613037,
      "learning_rate": 7.280877671178468e-06,
      "loss": 0.8411,
      "step": 9126
    },
    {
      "epoch": 0.7375502535404755,
      "grad_norm": 2.508075714111328,
      "learning_rate": 7.280295341440168e-06,
      "loss": 0.7681,
      "step": 9127
    },
    {
      "epoch": 0.7376310632538031,
      "grad_norm": 2.446751832962036,
      "learning_rate": 7.279712972645898e-06,
      "loss": 0.9448,
      "step": 9128
    },
    {
      "epoch": 0.7377118729671307,
      "grad_norm": 2.707951545715332,
      "learning_rate": 7.279130564805633e-06,
      "loss": 0.905,
      "step": 9129
    },
    {
      "epoch": 0.7377926826804582,
      "grad_norm": 2.327008008956909,
      "learning_rate": 7.278548117929348e-06,
      "loss": 0.9702,
      "step": 9130
    },
    {
      "epoch": 0.7378734923937857,
      "grad_norm": 2.9833052158355713,
      "learning_rate": 7.27796563202702e-06,
      "loss": 1.0264,
      "step": 9131
    },
    {
      "epoch": 0.7379543021071133,
      "grad_norm": 2.0838663578033447,
      "learning_rate": 7.277383107108623e-06,
      "loss": 1.0132,
      "step": 9132
    },
    {
      "epoch": 0.7380351118204408,
      "grad_norm": 2.692185640335083,
      "learning_rate": 7.2768005431841385e-06,
      "loss": 0.9251,
      "step": 9133
    },
    {
      "epoch": 0.7381159215337684,
      "grad_norm": 2.429028272628784,
      "learning_rate": 7.27621794026354e-06,
      "loss": 0.9071,
      "step": 9134
    },
    {
      "epoch": 0.738196731247096,
      "grad_norm": 2.4352681636810303,
      "learning_rate": 7.2756352983568094e-06,
      "loss": 0.9305,
      "step": 9135
    },
    {
      "epoch": 0.7382775409604234,
      "grad_norm": 2.6816604137420654,
      "learning_rate": 7.275052617473923e-06,
      "loss": 0.9283,
      "step": 9136
    },
    {
      "epoch": 0.738358350673751,
      "grad_norm": 2.5244176387786865,
      "learning_rate": 7.274469897624863e-06,
      "loss": 0.9673,
      "step": 9137
    },
    {
      "epoch": 0.7384391603870786,
      "grad_norm": 2.489497423171997,
      "learning_rate": 7.273887138819608e-06,
      "loss": 1.0181,
      "step": 9138
    },
    {
      "epoch": 0.738519970100406,
      "grad_norm": 2.1473710536956787,
      "learning_rate": 7.273304341068143e-06,
      "loss": 1.0654,
      "step": 9139
    },
    {
      "epoch": 0.7386007798137336,
      "grad_norm": 3.047027349472046,
      "learning_rate": 7.272721504380446e-06,
      "loss": 0.9208,
      "step": 9140
    },
    {
      "epoch": 0.7386815895270612,
      "grad_norm": 2.316822052001953,
      "learning_rate": 7.272138628766501e-06,
      "loss": 0.9827,
      "step": 9141
    },
    {
      "epoch": 0.7387623992403887,
      "grad_norm": 3.2937958240509033,
      "learning_rate": 7.27155571423629e-06,
      "loss": 0.8491,
      "step": 9142
    },
    {
      "epoch": 0.7388432089537162,
      "grad_norm": 2.6288256645202637,
      "learning_rate": 7.2709727607998e-06,
      "loss": 0.9432,
      "step": 9143
    },
    {
      "epoch": 0.7389240186670438,
      "grad_norm": 2.5328307151794434,
      "learning_rate": 7.2703897684670125e-06,
      "loss": 0.9151,
      "step": 9144
    },
    {
      "epoch": 0.7390048283803713,
      "grad_norm": 2.8022024631500244,
      "learning_rate": 7.269806737247914e-06,
      "loss": 0.936,
      "step": 9145
    },
    {
      "epoch": 0.7390856380936989,
      "grad_norm": 2.4462738037109375,
      "learning_rate": 7.2692236671524915e-06,
      "loss": 1.0055,
      "step": 9146
    },
    {
      "epoch": 0.7391664478070264,
      "grad_norm": 2.5309243202209473,
      "learning_rate": 7.268640558190731e-06,
      "loss": 0.9735,
      "step": 9147
    },
    {
      "epoch": 0.7392472575203539,
      "grad_norm": 2.905632734298706,
      "learning_rate": 7.268057410372618e-06,
      "loss": 1.0658,
      "step": 9148
    },
    {
      "epoch": 0.7393280672336815,
      "grad_norm": 4.138669013977051,
      "learning_rate": 7.267474223708142e-06,
      "loss": 1.0168,
      "step": 9149
    },
    {
      "epoch": 0.7394088769470091,
      "grad_norm": 2.6521735191345215,
      "learning_rate": 7.266890998207291e-06,
      "loss": 0.9827,
      "step": 9150
    },
    {
      "epoch": 0.7394896866603365,
      "grad_norm": 2.6250736713409424,
      "learning_rate": 7.266307733880054e-06,
      "loss": 0.9865,
      "step": 9151
    },
    {
      "epoch": 0.7395704963736641,
      "grad_norm": 2.8561973571777344,
      "learning_rate": 7.265724430736423e-06,
      "loss": 1.0175,
      "step": 9152
    },
    {
      "epoch": 0.7396513060869917,
      "grad_norm": 2.865936040878296,
      "learning_rate": 7.265141088786385e-06,
      "loss": 0.9221,
      "step": 9153
    },
    {
      "epoch": 0.7397321158003192,
      "grad_norm": 2.615757703781128,
      "learning_rate": 7.264557708039935e-06,
      "loss": 0.8526,
      "step": 9154
    },
    {
      "epoch": 0.7398129255136467,
      "grad_norm": 2.5320639610290527,
      "learning_rate": 7.263974288507062e-06,
      "loss": 1.0548,
      "step": 9155
    },
    {
      "epoch": 0.7398937352269743,
      "grad_norm": 2.7111244201660156,
      "learning_rate": 7.263390830197761e-06,
      "loss": 0.8035,
      "step": 9156
    },
    {
      "epoch": 0.7399745449403018,
      "grad_norm": 2.625068426132202,
      "learning_rate": 7.262807333122024e-06,
      "loss": 0.8893,
      "step": 9157
    },
    {
      "epoch": 0.7400553546536294,
      "grad_norm": 3.314542531967163,
      "learning_rate": 7.262223797289843e-06,
      "loss": 0.8737,
      "step": 9158
    },
    {
      "epoch": 0.740136164366957,
      "grad_norm": 2.487816572189331,
      "learning_rate": 7.261640222711216e-06,
      "loss": 0.9601,
      "step": 9159
    },
    {
      "epoch": 0.7402169740802844,
      "grad_norm": 2.62138032913208,
      "learning_rate": 7.2610566093961356e-06,
      "loss": 1.0944,
      "step": 9160
    },
    {
      "epoch": 0.740297783793612,
      "grad_norm": 2.684164047241211,
      "learning_rate": 7.2604729573546e-06,
      "loss": 0.9524,
      "step": 9161
    },
    {
      "epoch": 0.7403785935069396,
      "grad_norm": 2.382800579071045,
      "learning_rate": 7.259889266596605e-06,
      "loss": 1.1115,
      "step": 9162
    },
    {
      "epoch": 0.740459403220267,
      "grad_norm": 3.1251096725463867,
      "learning_rate": 7.259305537132144e-06,
      "loss": 0.809,
      "step": 9163
    },
    {
      "epoch": 0.7405402129335946,
      "grad_norm": 2.339355707168579,
      "learning_rate": 7.258721768971222e-06,
      "loss": 0.9239,
      "step": 9164
    },
    {
      "epoch": 0.7406210226469222,
      "grad_norm": 2.5114083290100098,
      "learning_rate": 7.258137962123832e-06,
      "loss": 0.9411,
      "step": 9165
    },
    {
      "epoch": 0.7407018323602497,
      "grad_norm": 2.3658273220062256,
      "learning_rate": 7.257554116599975e-06,
      "loss": 1.0043,
      "step": 9166
    },
    {
      "epoch": 0.7407826420735772,
      "grad_norm": 2.6916441917419434,
      "learning_rate": 7.256970232409651e-06,
      "loss": 1.0074,
      "step": 9167
    },
    {
      "epoch": 0.7408634517869048,
      "grad_norm": 3.2465579509735107,
      "learning_rate": 7.256386309562862e-06,
      "loss": 0.9164,
      "step": 9168
    },
    {
      "epoch": 0.7409442615002323,
      "grad_norm": 2.397860288619995,
      "learning_rate": 7.255802348069604e-06,
      "loss": 0.9917,
      "step": 9169
    },
    {
      "epoch": 0.7410250712135599,
      "grad_norm": 2.5243847370147705,
      "learning_rate": 7.255218347939885e-06,
      "loss": 0.9158,
      "step": 9170
    },
    {
      "epoch": 0.7411058809268875,
      "grad_norm": 2.4917051792144775,
      "learning_rate": 7.2546343091837035e-06,
      "loss": 0.9196,
      "step": 9171
    },
    {
      "epoch": 0.7411866906402149,
      "grad_norm": 3.039865493774414,
      "learning_rate": 7.254050231811065e-06,
      "loss": 0.8602,
      "step": 9172
    },
    {
      "epoch": 0.7412675003535425,
      "grad_norm": 3.154069423675537,
      "learning_rate": 7.253466115831973e-06,
      "loss": 1.0926,
      "step": 9173
    },
    {
      "epoch": 0.7413483100668701,
      "grad_norm": 2.7796924114227295,
      "learning_rate": 7.2528819612564305e-06,
      "loss": 0.9239,
      "step": 9174
    },
    {
      "epoch": 0.7414291197801975,
      "grad_norm": 2.7398829460144043,
      "learning_rate": 7.252297768094443e-06,
      "loss": 1.0191,
      "step": 9175
    },
    {
      "epoch": 0.7415099294935251,
      "grad_norm": 2.4738523960113525,
      "learning_rate": 7.2517135363560185e-06,
      "loss": 0.8959,
      "step": 9176
    },
    {
      "epoch": 0.7415907392068527,
      "grad_norm": 2.7567601203918457,
      "learning_rate": 7.25112926605116e-06,
      "loss": 0.8983,
      "step": 9177
    },
    {
      "epoch": 0.7416715489201802,
      "grad_norm": 2.247624158859253,
      "learning_rate": 7.2505449571898775e-06,
      "loss": 0.824,
      "step": 9178
    },
    {
      "epoch": 0.7417523586335077,
      "grad_norm": 2.2220160961151123,
      "learning_rate": 7.249960609782179e-06,
      "loss": 1.0418,
      "step": 9179
    },
    {
      "epoch": 0.7418331683468353,
      "grad_norm": 2.7129971981048584,
      "learning_rate": 7.249376223838071e-06,
      "loss": 0.9933,
      "step": 9180
    },
    {
      "epoch": 0.7419139780601628,
      "grad_norm": 2.716585874557495,
      "learning_rate": 7.248791799367563e-06,
      "loss": 0.827,
      "step": 9181
    },
    {
      "epoch": 0.7419947877734904,
      "grad_norm": 2.348145008087158,
      "learning_rate": 7.248207336380666e-06,
      "loss": 0.9534,
      "step": 9182
    },
    {
      "epoch": 0.742075597486818,
      "grad_norm": 2.6000277996063232,
      "learning_rate": 7.247622834887388e-06,
      "loss": 0.8306,
      "step": 9183
    },
    {
      "epoch": 0.7421564072001454,
      "grad_norm": 2.988957166671753,
      "learning_rate": 7.2470382948977436e-06,
      "loss": 0.9401,
      "step": 9184
    },
    {
      "epoch": 0.742237216913473,
      "grad_norm": 2.2751994132995605,
      "learning_rate": 7.2464537164217405e-06,
      "loss": 0.8464,
      "step": 9185
    },
    {
      "epoch": 0.7423180266268006,
      "grad_norm": 2.605870246887207,
      "learning_rate": 7.245869099469396e-06,
      "loss": 0.9212,
      "step": 9186
    },
    {
      "epoch": 0.742398836340128,
      "grad_norm": 2.75929594039917,
      "learning_rate": 7.24528444405072e-06,
      "loss": 0.967,
      "step": 9187
    },
    {
      "epoch": 0.7424796460534556,
      "grad_norm": 2.563917875289917,
      "learning_rate": 7.244699750175726e-06,
      "loss": 0.7994,
      "step": 9188
    },
    {
      "epoch": 0.7425604557667832,
      "grad_norm": 2.3208367824554443,
      "learning_rate": 7.244115017854429e-06,
      "loss": 0.9049,
      "step": 9189
    },
    {
      "epoch": 0.7426412654801107,
      "grad_norm": 2.7053263187408447,
      "learning_rate": 7.243530247096845e-06,
      "loss": 0.9216,
      "step": 9190
    },
    {
      "epoch": 0.7427220751934382,
      "grad_norm": 2.4055285453796387,
      "learning_rate": 7.242945437912987e-06,
      "loss": 0.9529,
      "step": 9191
    },
    {
      "epoch": 0.7428028849067658,
      "grad_norm": 2.134317398071289,
      "learning_rate": 7.242360590312876e-06,
      "loss": 0.9731,
      "step": 9192
    },
    {
      "epoch": 0.7428836946200933,
      "grad_norm": 2.8379886150360107,
      "learning_rate": 7.241775704306525e-06,
      "loss": 0.9023,
      "step": 9193
    },
    {
      "epoch": 0.7429645043334209,
      "grad_norm": 2.562758684158325,
      "learning_rate": 7.241190779903953e-06,
      "loss": 0.9453,
      "step": 9194
    },
    {
      "epoch": 0.7430453140467485,
      "grad_norm": 2.912896156311035,
      "learning_rate": 7.240605817115179e-06,
      "loss": 0.9362,
      "step": 9195
    },
    {
      "epoch": 0.7431261237600759,
      "grad_norm": 3.105250358581543,
      "learning_rate": 7.240020815950222e-06,
      "loss": 0.8728,
      "step": 9196
    },
    {
      "epoch": 0.7432069334734035,
      "grad_norm": 2.6104366779327393,
      "learning_rate": 7.239435776419098e-06,
      "loss": 0.9926,
      "step": 9197
    },
    {
      "epoch": 0.7432877431867311,
      "grad_norm": 2.365262269973755,
      "learning_rate": 7.238850698531834e-06,
      "loss": 1.0324,
      "step": 9198
    },
    {
      "epoch": 0.7433685529000585,
      "grad_norm": 2.8031811714172363,
      "learning_rate": 7.238265582298445e-06,
      "loss": 0.8747,
      "step": 9199
    },
    {
      "epoch": 0.7434493626133861,
      "grad_norm": 2.52441668510437,
      "learning_rate": 7.237680427728956e-06,
      "loss": 0.9131,
      "step": 9200
    },
    {
      "epoch": 0.7435301723267137,
      "grad_norm": 3.2640960216522217,
      "learning_rate": 7.237095234833388e-06,
      "loss": 0.9751,
      "step": 9201
    },
    {
      "epoch": 0.7436109820400412,
      "grad_norm": 2.737494945526123,
      "learning_rate": 7.236510003621764e-06,
      "loss": 1.0418,
      "step": 9202
    },
    {
      "epoch": 0.7436917917533687,
      "grad_norm": 2.468668222427368,
      "learning_rate": 7.235924734104109e-06,
      "loss": 0.8864,
      "step": 9203
    },
    {
      "epoch": 0.7437726014666963,
      "grad_norm": 2.862825870513916,
      "learning_rate": 7.2353394262904456e-06,
      "loss": 0.9239,
      "step": 9204
    },
    {
      "epoch": 0.7438534111800238,
      "grad_norm": 2.7571980953216553,
      "learning_rate": 7.234754080190797e-06,
      "loss": 0.9473,
      "step": 9205
    },
    {
      "epoch": 0.7439342208933514,
      "grad_norm": 3.0559935569763184,
      "learning_rate": 7.234168695815194e-06,
      "loss": 0.9531,
      "step": 9206
    },
    {
      "epoch": 0.744015030606679,
      "grad_norm": 2.5840489864349365,
      "learning_rate": 7.233583273173658e-06,
      "loss": 0.9189,
      "step": 9207
    },
    {
      "epoch": 0.7440958403200064,
      "grad_norm": 3.0948474407196045,
      "learning_rate": 7.232997812276218e-06,
      "loss": 1.0294,
      "step": 9208
    },
    {
      "epoch": 0.744176650033334,
      "grad_norm": 2.4203271865844727,
      "learning_rate": 7.232412313132902e-06,
      "loss": 0.9397,
      "step": 9209
    },
    {
      "epoch": 0.7442574597466616,
      "grad_norm": 3.032336950302124,
      "learning_rate": 7.231826775753735e-06,
      "loss": 0.9868,
      "step": 9210
    },
    {
      "epoch": 0.744338269459989,
      "grad_norm": 3.3545117378234863,
      "learning_rate": 7.231241200148751e-06,
      "loss": 0.87,
      "step": 9211
    },
    {
      "epoch": 0.7444190791733166,
      "grad_norm": 2.4008750915527344,
      "learning_rate": 7.230655586327975e-06,
      "loss": 0.8474,
      "step": 9212
    },
    {
      "epoch": 0.7444998888866442,
      "grad_norm": 3.301022529602051,
      "learning_rate": 7.230069934301439e-06,
      "loss": 0.858,
      "step": 9213
    },
    {
      "epoch": 0.7445806985999717,
      "grad_norm": 2.817265033721924,
      "learning_rate": 7.2294842440791756e-06,
      "loss": 0.9113,
      "step": 9214
    },
    {
      "epoch": 0.7446615083132992,
      "grad_norm": 2.6447441577911377,
      "learning_rate": 7.228898515671214e-06,
      "loss": 1.0156,
      "step": 9215
    },
    {
      "epoch": 0.7447423180266268,
      "grad_norm": 2.373446226119995,
      "learning_rate": 7.228312749087585e-06,
      "loss": 0.8783,
      "step": 9216
    },
    {
      "epoch": 0.7448231277399543,
      "grad_norm": 2.9240570068359375,
      "learning_rate": 7.2277269443383225e-06,
      "loss": 1.0205,
      "step": 9217
    },
    {
      "epoch": 0.7449039374532819,
      "grad_norm": 2.993472099304199,
      "learning_rate": 7.227141101433463e-06,
      "loss": 0.8996,
      "step": 9218
    },
    {
      "epoch": 0.7449847471666095,
      "grad_norm": 2.8471803665161133,
      "learning_rate": 7.226555220383036e-06,
      "loss": 0.9852,
      "step": 9219
    },
    {
      "epoch": 0.7450655568799369,
      "grad_norm": 2.5014472007751465,
      "learning_rate": 7.225969301197079e-06,
      "loss": 1.0542,
      "step": 9220
    },
    {
      "epoch": 0.7451463665932645,
      "grad_norm": 2.8656868934631348,
      "learning_rate": 7.225383343885628e-06,
      "loss": 0.9652,
      "step": 9221
    },
    {
      "epoch": 0.7452271763065921,
      "grad_norm": 3.0316591262817383,
      "learning_rate": 7.224797348458714e-06,
      "loss": 0.924,
      "step": 9222
    },
    {
      "epoch": 0.7453079860199195,
      "grad_norm": 2.8609044551849365,
      "learning_rate": 7.224211314926382e-06,
      "loss": 0.9121,
      "step": 9223
    },
    {
      "epoch": 0.7453887957332471,
      "grad_norm": 2.761373519897461,
      "learning_rate": 7.223625243298662e-06,
      "loss": 0.933,
      "step": 9224
    },
    {
      "epoch": 0.7454696054465747,
      "grad_norm": 2.629418134689331,
      "learning_rate": 7.223039133585595e-06,
      "loss": 0.9997,
      "step": 9225
    },
    {
      "epoch": 0.7455504151599022,
      "grad_norm": 2.426096200942993,
      "learning_rate": 7.2224529857972205e-06,
      "loss": 0.8457,
      "step": 9226
    },
    {
      "epoch": 0.7456312248732297,
      "grad_norm": 3.3540234565734863,
      "learning_rate": 7.221866799943575e-06,
      "loss": 0.797,
      "step": 9227
    },
    {
      "epoch": 0.7457120345865573,
      "grad_norm": 2.6288182735443115,
      "learning_rate": 7.221280576034702e-06,
      "loss": 0.9058,
      "step": 9228
    },
    {
      "epoch": 0.7457928442998848,
      "grad_norm": 2.7626688480377197,
      "learning_rate": 7.22069431408064e-06,
      "loss": 0.9798,
      "step": 9229
    },
    {
      "epoch": 0.7458736540132124,
      "grad_norm": 3.0615322589874268,
      "learning_rate": 7.220108014091428e-06,
      "loss": 0.9302,
      "step": 9230
    },
    {
      "epoch": 0.74595446372654,
      "grad_norm": 2.7715580463409424,
      "learning_rate": 7.2195216760771125e-06,
      "loss": 1.0387,
      "step": 9231
    },
    {
      "epoch": 0.7460352734398674,
      "grad_norm": 2.9438533782958984,
      "learning_rate": 7.218935300047734e-06,
      "loss": 0.9236,
      "step": 9232
    },
    {
      "epoch": 0.746116083153195,
      "grad_norm": 2.639690637588501,
      "learning_rate": 7.218348886013335e-06,
      "loss": 1.0522,
      "step": 9233
    },
    {
      "epoch": 0.7461968928665226,
      "grad_norm": 2.5193910598754883,
      "learning_rate": 7.217762433983961e-06,
      "loss": 0.8731,
      "step": 9234
    },
    {
      "epoch": 0.74627770257985,
      "grad_norm": 3.033635139465332,
      "learning_rate": 7.217175943969655e-06,
      "loss": 0.931,
      "step": 9235
    },
    {
      "epoch": 0.7463585122931776,
      "grad_norm": 2.6699891090393066,
      "learning_rate": 7.216589415980462e-06,
      "loss": 0.9871,
      "step": 9236
    },
    {
      "epoch": 0.7464393220065052,
      "grad_norm": 2.3057162761688232,
      "learning_rate": 7.21600285002643e-06,
      "loss": 1.0662,
      "step": 9237
    },
    {
      "epoch": 0.7465201317198327,
      "grad_norm": 2.666623115539551,
      "learning_rate": 7.215416246117602e-06,
      "loss": 0.9298,
      "step": 9238
    },
    {
      "epoch": 0.7466009414331602,
      "grad_norm": 2.4717490673065186,
      "learning_rate": 7.21482960426403e-06,
      "loss": 0.8523,
      "step": 9239
    },
    {
      "epoch": 0.7466817511464878,
      "grad_norm": 3.2087392807006836,
      "learning_rate": 7.214242924475756e-06,
      "loss": 0.9559,
      "step": 9240
    },
    {
      "epoch": 0.7467625608598154,
      "grad_norm": 2.8614275455474854,
      "learning_rate": 7.2136562067628334e-06,
      "loss": 0.9902,
      "step": 9241
    },
    {
      "epoch": 0.7468433705731429,
      "grad_norm": 2.664296865463257,
      "learning_rate": 7.2130694511353074e-06,
      "loss": 0.8702,
      "step": 9242
    },
    {
      "epoch": 0.7469241802864705,
      "grad_norm": 2.6501624584198,
      "learning_rate": 7.2124826576032315e-06,
      "loss": 0.9428,
      "step": 9243
    },
    {
      "epoch": 0.747004989999798,
      "grad_norm": 2.4006431102752686,
      "learning_rate": 7.2118958261766515e-06,
      "loss": 0.8907,
      "step": 9244
    },
    {
      "epoch": 0.7470857997131255,
      "grad_norm": 2.529768705368042,
      "learning_rate": 7.211308956865623e-06,
      "loss": 1.0185,
      "step": 9245
    },
    {
      "epoch": 0.7471666094264531,
      "grad_norm": 2.6251213550567627,
      "learning_rate": 7.210722049680195e-06,
      "loss": 0.8266,
      "step": 9246
    },
    {
      "epoch": 0.7472474191397807,
      "grad_norm": 2.5997235774993896,
      "learning_rate": 7.2101351046304204e-06,
      "loss": 0.984,
      "step": 9247
    },
    {
      "epoch": 0.7473282288531081,
      "grad_norm": 2.433603525161743,
      "learning_rate": 7.209548121726351e-06,
      "loss": 0.8437,
      "step": 9248
    },
    {
      "epoch": 0.7474090385664357,
      "grad_norm": 2.428435802459717,
      "learning_rate": 7.208961100978043e-06,
      "loss": 0.8103,
      "step": 9249
    },
    {
      "epoch": 0.7474898482797633,
      "grad_norm": 2.557340383529663,
      "learning_rate": 7.208374042395547e-06,
      "loss": 0.9709,
      "step": 9250
    },
    {
      "epoch": 0.7475706579930907,
      "grad_norm": 2.5587244033813477,
      "learning_rate": 7.207786945988924e-06,
      "loss": 0.9366,
      "step": 9251
    },
    {
      "epoch": 0.7476514677064183,
      "grad_norm": 2.7646195888519287,
      "learning_rate": 7.207199811768222e-06,
      "loss": 1.086,
      "step": 9252
    },
    {
      "epoch": 0.7477322774197459,
      "grad_norm": 2.679720640182495,
      "learning_rate": 7.206612639743502e-06,
      "loss": 1.0583,
      "step": 9253
    },
    {
      "epoch": 0.7478130871330734,
      "grad_norm": 3.1740527153015137,
      "learning_rate": 7.20602542992482e-06,
      "loss": 1.0459,
      "step": 9254
    },
    {
      "epoch": 0.747893896846401,
      "grad_norm": 2.3224925994873047,
      "learning_rate": 7.205438182322233e-06,
      "loss": 0.9661,
      "step": 9255
    },
    {
      "epoch": 0.7479747065597285,
      "grad_norm": 2.633460760116577,
      "learning_rate": 7.2048508969457995e-06,
      "loss": 0.902,
      "step": 9256
    },
    {
      "epoch": 0.748055516273056,
      "grad_norm": 2.6579766273498535,
      "learning_rate": 7.204263573805579e-06,
      "loss": 0.8872,
      "step": 9257
    },
    {
      "epoch": 0.7481363259863836,
      "grad_norm": 2.569265604019165,
      "learning_rate": 7.2036762129116275e-06,
      "loss": 0.8392,
      "step": 9258
    },
    {
      "epoch": 0.7482171356997112,
      "grad_norm": 2.77009916305542,
      "learning_rate": 7.203088814274011e-06,
      "loss": 0.9,
      "step": 9259
    },
    {
      "epoch": 0.7482979454130386,
      "grad_norm": 2.763956069946289,
      "learning_rate": 7.202501377902784e-06,
      "loss": 0.8667,
      "step": 9260
    },
    {
      "epoch": 0.7483787551263662,
      "grad_norm": 2.542914390563965,
      "learning_rate": 7.201913903808011e-06,
      "loss": 0.9948,
      "step": 9261
    },
    {
      "epoch": 0.7484595648396938,
      "grad_norm": 2.641510248184204,
      "learning_rate": 7.201326391999754e-06,
      "loss": 0.9146,
      "step": 9262
    },
    {
      "epoch": 0.7485403745530212,
      "grad_norm": 2.4405808448791504,
      "learning_rate": 7.200738842488078e-06,
      "loss": 0.9546,
      "step": 9263
    },
    {
      "epoch": 0.7486211842663488,
      "grad_norm": 2.7633845806121826,
      "learning_rate": 7.20015125528304e-06,
      "loss": 0.8837,
      "step": 9264
    },
    {
      "epoch": 0.7487019939796764,
      "grad_norm": 2.6070621013641357,
      "learning_rate": 7.199563630394709e-06,
      "loss": 0.9535,
      "step": 9265
    },
    {
      "epoch": 0.7487828036930039,
      "grad_norm": 3.175579786300659,
      "learning_rate": 7.198975967833148e-06,
      "loss": 0.9229,
      "step": 9266
    },
    {
      "epoch": 0.7488636134063315,
      "grad_norm": 2.457697868347168,
      "learning_rate": 7.198388267608424e-06,
      "loss": 0.9232,
      "step": 9267
    },
    {
      "epoch": 0.748944423119659,
      "grad_norm": 3.052302360534668,
      "learning_rate": 7.1978005297305994e-06,
      "loss": 0.8864,
      "step": 9268
    },
    {
      "epoch": 0.7490252328329865,
      "grad_norm": 3.0551042556762695,
      "learning_rate": 7.197212754209744e-06,
      "loss": 0.8824,
      "step": 9269
    },
    {
      "epoch": 0.7491060425463141,
      "grad_norm": 3.2402610778808594,
      "learning_rate": 7.196624941055923e-06,
      "loss": 1.0016,
      "step": 9270
    },
    {
      "epoch": 0.7491868522596417,
      "grad_norm": 2.7520618438720703,
      "learning_rate": 7.196037090279206e-06,
      "loss": 0.8577,
      "step": 9271
    },
    {
      "epoch": 0.7492676619729691,
      "grad_norm": 2.7571287155151367,
      "learning_rate": 7.195449201889658e-06,
      "loss": 0.966,
      "step": 9272
    },
    {
      "epoch": 0.7493484716862967,
      "grad_norm": 2.440335273742676,
      "learning_rate": 7.194861275897352e-06,
      "loss": 0.9635,
      "step": 9273
    },
    {
      "epoch": 0.7494292813996243,
      "grad_norm": 2.439288854598999,
      "learning_rate": 7.194273312312357e-06,
      "loss": 0.9975,
      "step": 9274
    },
    {
      "epoch": 0.7495100911129517,
      "grad_norm": 3.4287447929382324,
      "learning_rate": 7.193685311144741e-06,
      "loss": 0.9167,
      "step": 9275
    },
    {
      "epoch": 0.7495909008262793,
      "grad_norm": 2.7884204387664795,
      "learning_rate": 7.193097272404578e-06,
      "loss": 0.9657,
      "step": 9276
    },
    {
      "epoch": 0.7496717105396069,
      "grad_norm": 2.3156442642211914,
      "learning_rate": 7.192509196101938e-06,
      "loss": 0.9158,
      "step": 9277
    },
    {
      "epoch": 0.7497525202529344,
      "grad_norm": 2.6749985218048096,
      "learning_rate": 7.191921082246893e-06,
      "loss": 0.9151,
      "step": 9278
    },
    {
      "epoch": 0.749833329966262,
      "grad_norm": 2.1590397357940674,
      "learning_rate": 7.191332930849517e-06,
      "loss": 1.0993,
      "step": 9279
    },
    {
      "epoch": 0.7499141396795895,
      "grad_norm": 2.533832550048828,
      "learning_rate": 7.190744741919884e-06,
      "loss": 0.8439,
      "step": 9280
    },
    {
      "epoch": 0.749994949392917,
      "grad_norm": 2.5818288326263428,
      "learning_rate": 7.190156515468069e-06,
      "loss": 1.0234,
      "step": 9281
    },
    {
      "epoch": 0.7500757591062446,
      "grad_norm": 2.3331825733184814,
      "learning_rate": 7.189568251504143e-06,
      "loss": 0.889,
      "step": 9282
    },
    {
      "epoch": 0.7501565688195722,
      "grad_norm": 2.4949986934661865,
      "learning_rate": 7.1889799500381855e-06,
      "loss": 0.9003,
      "step": 9283
    },
    {
      "epoch": 0.7502373785328996,
      "grad_norm": 2.7428500652313232,
      "learning_rate": 7.18839161108027e-06,
      "loss": 0.8927,
      "step": 9284
    },
    {
      "epoch": 0.7503181882462272,
      "grad_norm": 2.768780469894409,
      "learning_rate": 7.187803234640474e-06,
      "loss": 0.95,
      "step": 9285
    },
    {
      "epoch": 0.7503989979595548,
      "grad_norm": 2.724456787109375,
      "learning_rate": 7.187214820728877e-06,
      "loss": 0.9073,
      "step": 9286
    },
    {
      "epoch": 0.7504798076728822,
      "grad_norm": 2.688736915588379,
      "learning_rate": 7.186626369355555e-06,
      "loss": 0.8648,
      "step": 9287
    },
    {
      "epoch": 0.7505606173862098,
      "grad_norm": 2.1352972984313965,
      "learning_rate": 7.186037880530589e-06,
      "loss": 0.9212,
      "step": 9288
    },
    {
      "epoch": 0.7506414270995374,
      "grad_norm": 3.0642130374908447,
      "learning_rate": 7.185449354264055e-06,
      "loss": 0.914,
      "step": 9289
    },
    {
      "epoch": 0.7507222368128649,
      "grad_norm": 2.4261667728424072,
      "learning_rate": 7.184860790566035e-06,
      "loss": 1.0105,
      "step": 9290
    },
    {
      "epoch": 0.7508030465261925,
      "grad_norm": 2.8731417655944824,
      "learning_rate": 7.18427218944661e-06,
      "loss": 1.0317,
      "step": 9291
    },
    {
      "epoch": 0.75088385623952,
      "grad_norm": 2.8006937503814697,
      "learning_rate": 7.18368355091586e-06,
      "loss": 0.8769,
      "step": 9292
    },
    {
      "epoch": 0.7509646659528475,
      "grad_norm": 2.8680408000946045,
      "learning_rate": 7.183094874983868e-06,
      "loss": 0.9244,
      "step": 9293
    },
    {
      "epoch": 0.7510454756661751,
      "grad_norm": 2.4608442783355713,
      "learning_rate": 7.182506161660716e-06,
      "loss": 0.9152,
      "step": 9294
    },
    {
      "epoch": 0.7511262853795027,
      "grad_norm": 2.6141417026519775,
      "learning_rate": 7.181917410956489e-06,
      "loss": 0.9852,
      "step": 9295
    },
    {
      "epoch": 0.7512070950928301,
      "grad_norm": 2.662980556488037,
      "learning_rate": 7.181328622881269e-06,
      "loss": 0.9862,
      "step": 9296
    },
    {
      "epoch": 0.7512879048061577,
      "grad_norm": 3.090834379196167,
      "learning_rate": 7.1807397974451395e-06,
      "loss": 0.9018,
      "step": 9297
    },
    {
      "epoch": 0.7513687145194853,
      "grad_norm": 2.5306742191314697,
      "learning_rate": 7.18015093465819e-06,
      "loss": 0.9616,
      "step": 9298
    },
    {
      "epoch": 0.7514495242328127,
      "grad_norm": 3.2240328788757324,
      "learning_rate": 7.179562034530502e-06,
      "loss": 0.849,
      "step": 9299
    },
    {
      "epoch": 0.7515303339461403,
      "grad_norm": 3.143828868865967,
      "learning_rate": 7.1789730970721625e-06,
      "loss": 0.955,
      "step": 9300
    },
    {
      "epoch": 0.7516111436594679,
      "grad_norm": 2.6308634281158447,
      "learning_rate": 7.17838412229326e-06,
      "loss": 0.9277,
      "step": 9301
    },
    {
      "epoch": 0.7516919533727954,
      "grad_norm": 2.7284719944000244,
      "learning_rate": 7.177795110203884e-06,
      "loss": 0.9378,
      "step": 9302
    },
    {
      "epoch": 0.751772763086123,
      "grad_norm": 2.4026944637298584,
      "learning_rate": 7.177206060814117e-06,
      "loss": 0.952,
      "step": 9303
    },
    {
      "epoch": 0.7518535727994505,
      "grad_norm": 2.9648163318634033,
      "learning_rate": 7.176616974134054e-06,
      "loss": 0.8554,
      "step": 9304
    },
    {
      "epoch": 0.751934382512778,
      "grad_norm": 2.4647109508514404,
      "learning_rate": 7.176027850173781e-06,
      "loss": 0.817,
      "step": 9305
    },
    {
      "epoch": 0.7520151922261056,
      "grad_norm": 2.2996625900268555,
      "learning_rate": 7.17543868894339e-06,
      "loss": 0.8776,
      "step": 9306
    },
    {
      "epoch": 0.7520960019394332,
      "grad_norm": 2.7225048542022705,
      "learning_rate": 7.174849490452972e-06,
      "loss": 0.9062,
      "step": 9307
    },
    {
      "epoch": 0.7521768116527606,
      "grad_norm": 3.1516716480255127,
      "learning_rate": 7.174260254712617e-06,
      "loss": 0.9935,
      "step": 9308
    },
    {
      "epoch": 0.7522576213660882,
      "grad_norm": 2.550668478012085,
      "learning_rate": 7.173670981732419e-06,
      "loss": 1.1502,
      "step": 9309
    },
    {
      "epoch": 0.7523384310794158,
      "grad_norm": 2.4681434631347656,
      "learning_rate": 7.17308167152247e-06,
      "loss": 1.0161,
      "step": 9310
    },
    {
      "epoch": 0.7524192407927432,
      "grad_norm": 2.583268642425537,
      "learning_rate": 7.172492324092862e-06,
      "loss": 0.9408,
      "step": 9311
    },
    {
      "epoch": 0.7525000505060708,
      "grad_norm": 3.556342363357544,
      "learning_rate": 7.171902939453692e-06,
      "loss": 0.9428,
      "step": 9312
    },
    {
      "epoch": 0.7525808602193984,
      "grad_norm": 2.631986379623413,
      "learning_rate": 7.171313517615053e-06,
      "loss": 1.0738,
      "step": 9313
    },
    {
      "epoch": 0.7526616699327259,
      "grad_norm": 2.767961263656616,
      "learning_rate": 7.170724058587041e-06,
      "loss": 0.8905,
      "step": 9314
    },
    {
      "epoch": 0.7527424796460535,
      "grad_norm": 2.1314356327056885,
      "learning_rate": 7.17013456237975e-06,
      "loss": 0.9848,
      "step": 9315
    },
    {
      "epoch": 0.752823289359381,
      "grad_norm": 2.775268793106079,
      "learning_rate": 7.169545029003281e-06,
      "loss": 1.0272,
      "step": 9316
    },
    {
      "epoch": 0.7529040990727085,
      "grad_norm": 2.6422245502471924,
      "learning_rate": 7.168955458467726e-06,
      "loss": 0.8876,
      "step": 9317
    },
    {
      "epoch": 0.7529849087860361,
      "grad_norm": 2.1179001331329346,
      "learning_rate": 7.168365850783188e-06,
      "loss": 0.935,
      "step": 9318
    },
    {
      "epoch": 0.7530657184993637,
      "grad_norm": 2.697267532348633,
      "learning_rate": 7.167776205959761e-06,
      "loss": 0.8455,
      "step": 9319
    },
    {
      "epoch": 0.7531465282126911,
      "grad_norm": 2.6576733589172363,
      "learning_rate": 7.1671865240075475e-06,
      "loss": 0.9284,
      "step": 9320
    },
    {
      "epoch": 0.7532273379260187,
      "grad_norm": 3.2691657543182373,
      "learning_rate": 7.166596804936646e-06,
      "loss": 0.9172,
      "step": 9321
    },
    {
      "epoch": 0.7533081476393463,
      "grad_norm": 2.57712984085083,
      "learning_rate": 7.166007048757155e-06,
      "loss": 0.9596,
      "step": 9322
    },
    {
      "epoch": 0.7533889573526737,
      "grad_norm": 2.877537727355957,
      "learning_rate": 7.16541725547918e-06,
      "loss": 0.9322,
      "step": 9323
    },
    {
      "epoch": 0.7534697670660013,
      "grad_norm": 2.29923939704895,
      "learning_rate": 7.164827425112822e-06,
      "loss": 0.9925,
      "step": 9324
    },
    {
      "epoch": 0.7535505767793289,
      "grad_norm": 2.2775449752807617,
      "learning_rate": 7.164237557668177e-06,
      "loss": 0.9225,
      "step": 9325
    },
    {
      "epoch": 0.7536313864926564,
      "grad_norm": 2.4381635189056396,
      "learning_rate": 7.163647653155356e-06,
      "loss": 0.9624,
      "step": 9326
    },
    {
      "epoch": 0.753712196205984,
      "grad_norm": 3.1883251667022705,
      "learning_rate": 7.16305771158446e-06,
      "loss": 0.85,
      "step": 9327
    },
    {
      "epoch": 0.7537930059193115,
      "grad_norm": 2.524061918258667,
      "learning_rate": 7.162467732965592e-06,
      "loss": 0.955,
      "step": 9328
    },
    {
      "epoch": 0.753873815632639,
      "grad_norm": 2.958904981613159,
      "learning_rate": 7.161877717308857e-06,
      "loss": 1.0463,
      "step": 9329
    },
    {
      "epoch": 0.7539546253459666,
      "grad_norm": 3.006033420562744,
      "learning_rate": 7.161287664624364e-06,
      "loss": 1.0036,
      "step": 9330
    },
    {
      "epoch": 0.7540354350592942,
      "grad_norm": 2.6345412731170654,
      "learning_rate": 7.160697574922212e-06,
      "loss": 0.9288,
      "step": 9331
    },
    {
      "epoch": 0.7541162447726216,
      "grad_norm": 2.555610179901123,
      "learning_rate": 7.160107448212514e-06,
      "loss": 0.9351,
      "step": 9332
    },
    {
      "epoch": 0.7541970544859492,
      "grad_norm": 2.354729413986206,
      "learning_rate": 7.159517284505375e-06,
      "loss": 0.9694,
      "step": 9333
    },
    {
      "epoch": 0.7542778641992768,
      "grad_norm": 2.697465658187866,
      "learning_rate": 7.158927083810906e-06,
      "loss": 0.9136,
      "step": 9334
    },
    {
      "epoch": 0.7543586739126042,
      "grad_norm": 2.4910783767700195,
      "learning_rate": 7.158336846139212e-06,
      "loss": 0.9436,
      "step": 9335
    },
    {
      "epoch": 0.7544394836259318,
      "grad_norm": 2.5268726348876953,
      "learning_rate": 7.157746571500404e-06,
      "loss": 0.9455,
      "step": 9336
    },
    {
      "epoch": 0.7545202933392594,
      "grad_norm": 2.3225934505462646,
      "learning_rate": 7.157156259904592e-06,
      "loss": 0.9213,
      "step": 9337
    },
    {
      "epoch": 0.7546011030525869,
      "grad_norm": 2.3548941612243652,
      "learning_rate": 7.156565911361887e-06,
      "loss": 0.8115,
      "step": 9338
    },
    {
      "epoch": 0.7546819127659145,
      "grad_norm": 2.987687587738037,
      "learning_rate": 7.155975525882397e-06,
      "loss": 0.8953,
      "step": 9339
    },
    {
      "epoch": 0.754762722479242,
      "grad_norm": 2.4488444328308105,
      "learning_rate": 7.15538510347624e-06,
      "loss": 0.8765,
      "step": 9340
    },
    {
      "epoch": 0.7548435321925695,
      "grad_norm": 3.088615655899048,
      "learning_rate": 7.154794644153523e-06,
      "loss": 0.9218,
      "step": 9341
    },
    {
      "epoch": 0.7549243419058971,
      "grad_norm": 2.824294090270996,
      "learning_rate": 7.154204147924362e-06,
      "loss": 0.9115,
      "step": 9342
    },
    {
      "epoch": 0.7550051516192247,
      "grad_norm": 2.563656806945801,
      "learning_rate": 7.153613614798869e-06,
      "loss": 0.9283,
      "step": 9343
    },
    {
      "epoch": 0.7550859613325521,
      "grad_norm": 2.5003597736358643,
      "learning_rate": 7.15302304478716e-06,
      "loss": 0.9461,
      "step": 9344
    },
    {
      "epoch": 0.7551667710458797,
      "grad_norm": 4.22475004196167,
      "learning_rate": 7.152432437899349e-06,
      "loss": 0.9191,
      "step": 9345
    },
    {
      "epoch": 0.7552475807592073,
      "grad_norm": 2.362718105316162,
      "learning_rate": 7.151841794145553e-06,
      "loss": 0.983,
      "step": 9346
    },
    {
      "epoch": 0.7553283904725347,
      "grad_norm": 2.62459135055542,
      "learning_rate": 7.151251113535886e-06,
      "loss": 0.9091,
      "step": 9347
    },
    {
      "epoch": 0.7554092001858623,
      "grad_norm": 3.6033778190612793,
      "learning_rate": 7.150660396080469e-06,
      "loss": 0.9059,
      "step": 9348
    },
    {
      "epoch": 0.7554900098991899,
      "grad_norm": 2.774806261062622,
      "learning_rate": 7.150069641789414e-06,
      "loss": 0.8764,
      "step": 9349
    },
    {
      "epoch": 0.7555708196125174,
      "grad_norm": 2.9223997592926025,
      "learning_rate": 7.149478850672844e-06,
      "loss": 0.8836,
      "step": 9350
    },
    {
      "epoch": 0.755651629325845,
      "grad_norm": 2.658027410507202,
      "learning_rate": 7.148888022740875e-06,
      "loss": 0.9149,
      "step": 9351
    },
    {
      "epoch": 0.7557324390391725,
      "grad_norm": 2.5316708087921143,
      "learning_rate": 7.148297158003628e-06,
      "loss": 0.9406,
      "step": 9352
    },
    {
      "epoch": 0.7558132487525,
      "grad_norm": 2.776956558227539,
      "learning_rate": 7.147706256471222e-06,
      "loss": 1.0484,
      "step": 9353
    },
    {
      "epoch": 0.7558940584658276,
      "grad_norm": 2.474447250366211,
      "learning_rate": 7.147115318153778e-06,
      "loss": 0.8387,
      "step": 9354
    },
    {
      "epoch": 0.7559748681791552,
      "grad_norm": 2.3409523963928223,
      "learning_rate": 7.146524343061418e-06,
      "loss": 0.9392,
      "step": 9355
    },
    {
      "epoch": 0.7560556778924826,
      "grad_norm": 2.7103476524353027,
      "learning_rate": 7.145933331204264e-06,
      "loss": 0.9128,
      "step": 9356
    },
    {
      "epoch": 0.7561364876058102,
      "grad_norm": 3.336158037185669,
      "learning_rate": 7.145342282592438e-06,
      "loss": 0.921,
      "step": 9357
    },
    {
      "epoch": 0.7562172973191378,
      "grad_norm": 2.7014338970184326,
      "learning_rate": 7.144751197236063e-06,
      "loss": 1.0213,
      "step": 9358
    },
    {
      "epoch": 0.7562981070324652,
      "grad_norm": 2.6197993755340576,
      "learning_rate": 7.144160075145263e-06,
      "loss": 0.9492,
      "step": 9359
    },
    {
      "epoch": 0.7563789167457928,
      "grad_norm": 2.4035115242004395,
      "learning_rate": 7.143568916330163e-06,
      "loss": 0.9155,
      "step": 9360
    },
    {
      "epoch": 0.7564597264591204,
      "grad_norm": 2.1481218338012695,
      "learning_rate": 7.142977720800888e-06,
      "loss": 0.9322,
      "step": 9361
    },
    {
      "epoch": 0.7565405361724479,
      "grad_norm": 2.654609441757202,
      "learning_rate": 7.1423864885675634e-06,
      "loss": 1.0117,
      "step": 9362
    },
    {
      "epoch": 0.7566213458857755,
      "grad_norm": 2.4736640453338623,
      "learning_rate": 7.141795219640318e-06,
      "loss": 0.9867,
      "step": 9363
    },
    {
      "epoch": 0.756702155599103,
      "grad_norm": 2.554844379425049,
      "learning_rate": 7.141203914029273e-06,
      "loss": 0.759,
      "step": 9364
    },
    {
      "epoch": 0.7567829653124305,
      "grad_norm": 2.6223647594451904,
      "learning_rate": 7.140612571744562e-06,
      "loss": 0.8359,
      "step": 9365
    },
    {
      "epoch": 0.7568637750257581,
      "grad_norm": 2.4716081619262695,
      "learning_rate": 7.14002119279631e-06,
      "loss": 0.9236,
      "step": 9366
    },
    {
      "epoch": 0.7569445847390857,
      "grad_norm": 3.238743543624878,
      "learning_rate": 7.139429777194648e-06,
      "loss": 1.0816,
      "step": 9367
    },
    {
      "epoch": 0.7570253944524131,
      "grad_norm": 3.119372606277466,
      "learning_rate": 7.1388383249497025e-06,
      "loss": 0.9217,
      "step": 9368
    },
    {
      "epoch": 0.7571062041657407,
      "grad_norm": 2.5762522220611572,
      "learning_rate": 7.138246836071609e-06,
      "loss": 0.9669,
      "step": 9369
    },
    {
      "epoch": 0.7571870138790683,
      "grad_norm": 2.8306868076324463,
      "learning_rate": 7.13765531057049e-06,
      "loss": 0.8696,
      "step": 9370
    },
    {
      "epoch": 0.7572678235923959,
      "grad_norm": 2.553178310394287,
      "learning_rate": 7.1370637484564856e-06,
      "loss": 0.9538,
      "step": 9371
    },
    {
      "epoch": 0.7573486333057233,
      "grad_norm": 2.827378749847412,
      "learning_rate": 7.136472149739723e-06,
      "loss": 0.9093,
      "step": 9372
    },
    {
      "epoch": 0.7574294430190509,
      "grad_norm": 2.2313454151153564,
      "learning_rate": 7.135880514430334e-06,
      "loss": 1.0702,
      "step": 9373
    },
    {
      "epoch": 0.7575102527323785,
      "grad_norm": 2.437563419342041,
      "learning_rate": 7.1352888425384555e-06,
      "loss": 0.8496,
      "step": 9374
    },
    {
      "epoch": 0.757591062445706,
      "grad_norm": 2.5437726974487305,
      "learning_rate": 7.13469713407422e-06,
      "loss": 0.9566,
      "step": 9375
    },
    {
      "epoch": 0.7576718721590335,
      "grad_norm": 2.799818277359009,
      "learning_rate": 7.134105389047761e-06,
      "loss": 0.9318,
      "step": 9376
    },
    {
      "epoch": 0.7577526818723611,
      "grad_norm": 2.6814332008361816,
      "learning_rate": 7.133513607469214e-06,
      "loss": 0.8791,
      "step": 9377
    },
    {
      "epoch": 0.7578334915856886,
      "grad_norm": 2.529116153717041,
      "learning_rate": 7.132921789348714e-06,
      "loss": 1.0308,
      "step": 9378
    },
    {
      "epoch": 0.7579143012990162,
      "grad_norm": 2.8629038333892822,
      "learning_rate": 7.1323299346964015e-06,
      "loss": 0.9784,
      "step": 9379
    },
    {
      "epoch": 0.7579951110123437,
      "grad_norm": 2.7337749004364014,
      "learning_rate": 7.131738043522409e-06,
      "loss": 0.9443,
      "step": 9380
    },
    {
      "epoch": 0.7580759207256712,
      "grad_norm": 2.824218511581421,
      "learning_rate": 7.131146115836875e-06,
      "loss": 0.8851,
      "step": 9381
    },
    {
      "epoch": 0.7581567304389988,
      "grad_norm": 2.648939371109009,
      "learning_rate": 7.13055415164994e-06,
      "loss": 0.9399,
      "step": 9382
    },
    {
      "epoch": 0.7582375401523264,
      "grad_norm": 2.5566866397857666,
      "learning_rate": 7.129962150971741e-06,
      "loss": 0.94,
      "step": 9383
    },
    {
      "epoch": 0.7583183498656538,
      "grad_norm": 2.4530868530273438,
      "learning_rate": 7.1293701138124175e-06,
      "loss": 0.8466,
      "step": 9384
    },
    {
      "epoch": 0.7583991595789814,
      "grad_norm": 2.5966145992279053,
      "learning_rate": 7.1287780401821115e-06,
      "loss": 0.945,
      "step": 9385
    },
    {
      "epoch": 0.758479969292309,
      "grad_norm": 2.6472713947296143,
      "learning_rate": 7.1281859300909605e-06,
      "loss": 0.8775,
      "step": 9386
    },
    {
      "epoch": 0.7585607790056365,
      "grad_norm": 2.3527584075927734,
      "learning_rate": 7.12759378354911e-06,
      "loss": 1.0495,
      "step": 9387
    },
    {
      "epoch": 0.758641588718964,
      "grad_norm": 2.7990071773529053,
      "learning_rate": 7.1270016005666985e-06,
      "loss": 0.8712,
      "step": 9388
    },
    {
      "epoch": 0.7587223984322916,
      "grad_norm": 2.9060542583465576,
      "learning_rate": 7.1264093811538704e-06,
      "loss": 1.0565,
      "step": 9389
    },
    {
      "epoch": 0.7588032081456191,
      "grad_norm": 3.042029619216919,
      "learning_rate": 7.125817125320769e-06,
      "loss": 0.9588,
      "step": 9390
    },
    {
      "epoch": 0.7588840178589467,
      "grad_norm": 2.4240853786468506,
      "learning_rate": 7.125224833077537e-06,
      "loss": 1.0243,
      "step": 9391
    },
    {
      "epoch": 0.7589648275722742,
      "grad_norm": 2.659764528274536,
      "learning_rate": 7.124632504434321e-06,
      "loss": 0.8337,
      "step": 9392
    },
    {
      "epoch": 0.7590456372856017,
      "grad_norm": 2.4955406188964844,
      "learning_rate": 7.124040139401265e-06,
      "loss": 0.9165,
      "step": 9393
    },
    {
      "epoch": 0.7591264469989293,
      "grad_norm": 2.823073625564575,
      "learning_rate": 7.123447737988515e-06,
      "loss": 0.9566,
      "step": 9394
    },
    {
      "epoch": 0.7592072567122569,
      "grad_norm": 2.467170476913452,
      "learning_rate": 7.122855300206216e-06,
      "loss": 0.9647,
      "step": 9395
    },
    {
      "epoch": 0.7592880664255843,
      "grad_norm": 2.9635438919067383,
      "learning_rate": 7.122262826064518e-06,
      "loss": 0.9491,
      "step": 9396
    },
    {
      "epoch": 0.7593688761389119,
      "grad_norm": 2.6342523097991943,
      "learning_rate": 7.121670315573567e-06,
      "loss": 0.8406,
      "step": 9397
    },
    {
      "epoch": 0.7594496858522395,
      "grad_norm": 2.398297071456909,
      "learning_rate": 7.121077768743509e-06,
      "loss": 0.9417,
      "step": 9398
    },
    {
      "epoch": 0.759530495565567,
      "grad_norm": 2.964754581451416,
      "learning_rate": 7.1204851855844966e-06,
      "loss": 0.9138,
      "step": 9399
    },
    {
      "epoch": 0.7596113052788945,
      "grad_norm": 2.776533603668213,
      "learning_rate": 7.119892566106678e-06,
      "loss": 0.9521,
      "step": 9400
    },
    {
      "epoch": 0.7596921149922221,
      "grad_norm": 2.866835117340088,
      "learning_rate": 7.119299910320202e-06,
      "loss": 1.0627,
      "step": 9401
    },
    {
      "epoch": 0.7597729247055496,
      "grad_norm": 2.481116771697998,
      "learning_rate": 7.118707218235221e-06,
      "loss": 1.0436,
      "step": 9402
    },
    {
      "epoch": 0.7598537344188772,
      "grad_norm": 2.7597126960754395,
      "learning_rate": 7.118114489861886e-06,
      "loss": 1.0089,
      "step": 9403
    },
    {
      "epoch": 0.7599345441322047,
      "grad_norm": 2.7167396545410156,
      "learning_rate": 7.117521725210349e-06,
      "loss": 0.9041,
      "step": 9404
    },
    {
      "epoch": 0.7600153538455322,
      "grad_norm": 2.408841848373413,
      "learning_rate": 7.1169289242907634e-06,
      "loss": 0.9589,
      "step": 9405
    },
    {
      "epoch": 0.7600961635588598,
      "grad_norm": 2.7458863258361816,
      "learning_rate": 7.116336087113281e-06,
      "loss": 0.9494,
      "step": 9406
    },
    {
      "epoch": 0.7601769732721874,
      "grad_norm": 2.74415922164917,
      "learning_rate": 7.115743213688057e-06,
      "loss": 1.0392,
      "step": 9407
    },
    {
      "epoch": 0.7602577829855148,
      "grad_norm": 2.944753646850586,
      "learning_rate": 7.1151503040252435e-06,
      "loss": 0.8144,
      "step": 9408
    },
    {
      "epoch": 0.7603385926988424,
      "grad_norm": 2.4746129512786865,
      "learning_rate": 7.114557358134998e-06,
      "loss": 0.9022,
      "step": 9409
    },
    {
      "epoch": 0.76041940241217,
      "grad_norm": 2.4922099113464355,
      "learning_rate": 7.1139643760274756e-06,
      "loss": 0.9453,
      "step": 9410
    },
    {
      "epoch": 0.7605002121254975,
      "grad_norm": 2.807968854904175,
      "learning_rate": 7.113371357712833e-06,
      "loss": 0.9031,
      "step": 9411
    },
    {
      "epoch": 0.760581021838825,
      "grad_norm": 2.376249074935913,
      "learning_rate": 7.112778303201227e-06,
      "loss": 0.7973,
      "step": 9412
    },
    {
      "epoch": 0.7606618315521526,
      "grad_norm": 2.693059206008911,
      "learning_rate": 7.1121852125028144e-06,
      "loss": 1.0284,
      "step": 9413
    },
    {
      "epoch": 0.7607426412654801,
      "grad_norm": 2.5827879905700684,
      "learning_rate": 7.1115920856277545e-06,
      "loss": 0.8489,
      "step": 9414
    },
    {
      "epoch": 0.7608234509788077,
      "grad_norm": 2.589111089706421,
      "learning_rate": 7.1109989225862055e-06,
      "loss": 0.9226,
      "step": 9415
    },
    {
      "epoch": 0.7609042606921352,
      "grad_norm": 3.392749786376953,
      "learning_rate": 7.110405723388326e-06,
      "loss": 0.8925,
      "step": 9416
    },
    {
      "epoch": 0.7609850704054627,
      "grad_norm": 2.8906846046447754,
      "learning_rate": 7.1098124880442775e-06,
      "loss": 0.9414,
      "step": 9417
    },
    {
      "epoch": 0.7610658801187903,
      "grad_norm": 2.530301332473755,
      "learning_rate": 7.10921921656422e-06,
      "loss": 0.9049,
      "step": 9418
    },
    {
      "epoch": 0.7611466898321179,
      "grad_norm": 2.5978944301605225,
      "learning_rate": 7.1086259089583165e-06,
      "loss": 1.0372,
      "step": 9419
    },
    {
      "epoch": 0.7612274995454453,
      "grad_norm": 2.696030616760254,
      "learning_rate": 7.108032565236727e-06,
      "loss": 0.9258,
      "step": 9420
    },
    {
      "epoch": 0.7613083092587729,
      "grad_norm": 2.5111844539642334,
      "learning_rate": 7.107439185409613e-06,
      "loss": 0.9666,
      "step": 9421
    },
    {
      "epoch": 0.7613891189721005,
      "grad_norm": 2.845205783843994,
      "learning_rate": 7.106845769487142e-06,
      "loss": 0.8302,
      "step": 9422
    },
    {
      "epoch": 0.761469928685428,
      "grad_norm": 2.8490588665008545,
      "learning_rate": 7.106252317479473e-06,
      "loss": 0.8415,
      "step": 9423
    },
    {
      "epoch": 0.7615507383987555,
      "grad_norm": 2.5833804607391357,
      "learning_rate": 7.105658829396772e-06,
      "loss": 0.8523,
      "step": 9424
    },
    {
      "epoch": 0.7616315481120831,
      "grad_norm": 2.668302297592163,
      "learning_rate": 7.105065305249206e-06,
      "loss": 0.9104,
      "step": 9425
    },
    {
      "epoch": 0.7617123578254106,
      "grad_norm": 3.0312631130218506,
      "learning_rate": 7.104471745046937e-06,
      "loss": 0.9433,
      "step": 9426
    },
    {
      "epoch": 0.7617931675387382,
      "grad_norm": 2.984961986541748,
      "learning_rate": 7.103878148800134e-06,
      "loss": 1.0107,
      "step": 9427
    },
    {
      "epoch": 0.7618739772520657,
      "grad_norm": 2.511423349380493,
      "learning_rate": 7.103284516518966e-06,
      "loss": 0.8763,
      "step": 9428
    },
    {
      "epoch": 0.7619547869653932,
      "grad_norm": 2.8681230545043945,
      "learning_rate": 7.102690848213593e-06,
      "loss": 0.9905,
      "step": 9429
    },
    {
      "epoch": 0.7620355966787208,
      "grad_norm": 2.4794304370880127,
      "learning_rate": 7.102097143894191e-06,
      "loss": 0.973,
      "step": 9430
    },
    {
      "epoch": 0.7621164063920484,
      "grad_norm": 2.545872449874878,
      "learning_rate": 7.101503403570924e-06,
      "loss": 1.0162,
      "step": 9431
    },
    {
      "epoch": 0.7621972161053758,
      "grad_norm": 2.8261799812316895,
      "learning_rate": 7.1009096272539646e-06,
      "loss": 0.8995,
      "step": 9432
    },
    {
      "epoch": 0.7622780258187034,
      "grad_norm": 2.9498088359832764,
      "learning_rate": 7.10031581495348e-06,
      "loss": 1.0137,
      "step": 9433
    },
    {
      "epoch": 0.762358835532031,
      "grad_norm": 3.1932687759399414,
      "learning_rate": 7.099721966679642e-06,
      "loss": 0.863,
      "step": 9434
    },
    {
      "epoch": 0.7624396452453585,
      "grad_norm": 3.1584761142730713,
      "learning_rate": 7.099128082442621e-06,
      "loss": 0.8687,
      "step": 9435
    },
    {
      "epoch": 0.762520454958686,
      "grad_norm": 2.386131525039673,
      "learning_rate": 7.09853416225259e-06,
      "loss": 0.9076,
      "step": 9436
    },
    {
      "epoch": 0.7626012646720136,
      "grad_norm": 3.2778615951538086,
      "learning_rate": 7.09794020611972e-06,
      "loss": 0.9383,
      "step": 9437
    },
    {
      "epoch": 0.7626820743853411,
      "grad_norm": 2.6084465980529785,
      "learning_rate": 7.097346214054186e-06,
      "loss": 0.9834,
      "step": 9438
    },
    {
      "epoch": 0.7627628840986687,
      "grad_norm": 2.735183000564575,
      "learning_rate": 7.0967521860661604e-06,
      "loss": 0.8664,
      "step": 9439
    },
    {
      "epoch": 0.7628436938119962,
      "grad_norm": 2.414707660675049,
      "learning_rate": 7.096158122165816e-06,
      "loss": 1.1097,
      "step": 9440
    },
    {
      "epoch": 0.7629245035253237,
      "grad_norm": 3.164177417755127,
      "learning_rate": 7.09556402236333e-06,
      "loss": 1.0333,
      "step": 9441
    },
    {
      "epoch": 0.7630053132386513,
      "grad_norm": 2.6094696521759033,
      "learning_rate": 7.0949698866688774e-06,
      "loss": 0.9298,
      "step": 9442
    },
    {
      "epoch": 0.7630861229519789,
      "grad_norm": 2.682769536972046,
      "learning_rate": 7.094375715092635e-06,
      "loss": 0.9226,
      "step": 9443
    },
    {
      "epoch": 0.7631669326653063,
      "grad_norm": 2.2476346492767334,
      "learning_rate": 7.093781507644778e-06,
      "loss": 1.0783,
      "step": 9444
    },
    {
      "epoch": 0.7632477423786339,
      "grad_norm": 2.8686139583587646,
      "learning_rate": 7.093187264335484e-06,
      "loss": 0.9593,
      "step": 9445
    },
    {
      "epoch": 0.7633285520919615,
      "grad_norm": 2.404218912124634,
      "learning_rate": 7.092592985174932e-06,
      "loss": 0.9245,
      "step": 9446
    },
    {
      "epoch": 0.763409361805289,
      "grad_norm": 3.1972155570983887,
      "learning_rate": 7.091998670173299e-06,
      "loss": 0.9019,
      "step": 9447
    },
    {
      "epoch": 0.7634901715186165,
      "grad_norm": 2.677638053894043,
      "learning_rate": 7.091404319340765e-06,
      "loss": 0.9107,
      "step": 9448
    },
    {
      "epoch": 0.7635709812319441,
      "grad_norm": 2.568795919418335,
      "learning_rate": 7.09080993268751e-06,
      "loss": 1.0429,
      "step": 9449
    },
    {
      "epoch": 0.7636517909452716,
      "grad_norm": 2.8815431594848633,
      "learning_rate": 7.090215510223716e-06,
      "loss": 1.0192,
      "step": 9450
    },
    {
      "epoch": 0.7637326006585992,
      "grad_norm": 2.5321731567382812,
      "learning_rate": 7.089621051959559e-06,
      "loss": 0.977,
      "step": 9451
    },
    {
      "epoch": 0.7638134103719267,
      "grad_norm": 2.7652008533477783,
      "learning_rate": 7.089026557905227e-06,
      "loss": 0.9829,
      "step": 9452
    },
    {
      "epoch": 0.7638942200852542,
      "grad_norm": 2.2721855640411377,
      "learning_rate": 7.088432028070897e-06,
      "loss": 0.9826,
      "step": 9453
    },
    {
      "epoch": 0.7639750297985818,
      "grad_norm": 2.68190860748291,
      "learning_rate": 7.087837462466756e-06,
      "loss": 0.9249,
      "step": 9454
    },
    {
      "epoch": 0.7640558395119094,
      "grad_norm": 2.512723207473755,
      "learning_rate": 7.087242861102984e-06,
      "loss": 0.9233,
      "step": 9455
    },
    {
      "epoch": 0.7641366492252368,
      "grad_norm": 2.821420431137085,
      "learning_rate": 7.0866482239897675e-06,
      "loss": 0.8105,
      "step": 9456
    },
    {
      "epoch": 0.7642174589385644,
      "grad_norm": 2.5856151580810547,
      "learning_rate": 7.08605355113729e-06,
      "loss": 1.0468,
      "step": 9457
    },
    {
      "epoch": 0.764298268651892,
      "grad_norm": 2.2887649536132812,
      "learning_rate": 7.085458842555737e-06,
      "loss": 0.8821,
      "step": 9458
    },
    {
      "epoch": 0.7643790783652195,
      "grad_norm": 3.04512882232666,
      "learning_rate": 7.084864098255294e-06,
      "loss": 1.0228,
      "step": 9459
    },
    {
      "epoch": 0.764459888078547,
      "grad_norm": 3.017383337020874,
      "learning_rate": 7.0842693182461494e-06,
      "loss": 0.9401,
      "step": 9460
    },
    {
      "epoch": 0.7645406977918746,
      "grad_norm": 2.2113864421844482,
      "learning_rate": 7.083674502538489e-06,
      "loss": 1.0046,
      "step": 9461
    },
    {
      "epoch": 0.7646215075052021,
      "grad_norm": 2.4700751304626465,
      "learning_rate": 7.083079651142499e-06,
      "loss": 0.956,
      "step": 9462
    },
    {
      "epoch": 0.7647023172185297,
      "grad_norm": 3.114546537399292,
      "learning_rate": 7.082484764068371e-06,
      "loss": 1.096,
      "step": 9463
    },
    {
      "epoch": 0.7647831269318572,
      "grad_norm": 2.289763927459717,
      "learning_rate": 7.081889841326293e-06,
      "loss": 0.9989,
      "step": 9464
    },
    {
      "epoch": 0.7648639366451847,
      "grad_norm": 2.766525983810425,
      "learning_rate": 7.081294882926452e-06,
      "loss": 0.924,
      "step": 9465
    },
    {
      "epoch": 0.7649447463585123,
      "grad_norm": 2.4099643230438232,
      "learning_rate": 7.080699888879041e-06,
      "loss": 0.9167,
      "step": 9466
    },
    {
      "epoch": 0.7650255560718399,
      "grad_norm": 2.4988508224487305,
      "learning_rate": 7.08010485919425e-06,
      "loss": 0.9012,
      "step": 9467
    },
    {
      "epoch": 0.7651063657851673,
      "grad_norm": 2.955271005630493,
      "learning_rate": 7.0795097938822695e-06,
      "loss": 0.9786,
      "step": 9468
    },
    {
      "epoch": 0.7651871754984949,
      "grad_norm": 2.9606423377990723,
      "learning_rate": 7.078914692953294e-06,
      "loss": 0.9125,
      "step": 9469
    },
    {
      "epoch": 0.7652679852118225,
      "grad_norm": 2.853749990463257,
      "learning_rate": 7.078319556417513e-06,
      "loss": 0.9281,
      "step": 9470
    },
    {
      "epoch": 0.76534879492515,
      "grad_norm": 2.4236481189727783,
      "learning_rate": 7.077724384285123e-06,
      "loss": 1.0123,
      "step": 9471
    },
    {
      "epoch": 0.7654296046384775,
      "grad_norm": 2.6906888484954834,
      "learning_rate": 7.0771291765663156e-06,
      "loss": 0.9696,
      "step": 9472
    },
    {
      "epoch": 0.7655104143518051,
      "grad_norm": 2.526541233062744,
      "learning_rate": 7.076533933271284e-06,
      "loss": 0.9469,
      "step": 9473
    },
    {
      "epoch": 0.7655912240651326,
      "grad_norm": 2.3343493938446045,
      "learning_rate": 7.075938654410228e-06,
      "loss": 0.8985,
      "step": 9474
    },
    {
      "epoch": 0.7656720337784602,
      "grad_norm": 2.31614089012146,
      "learning_rate": 7.0753433399933406e-06,
      "loss": 0.8177,
      "step": 9475
    },
    {
      "epoch": 0.7657528434917877,
      "grad_norm": 3.08620285987854,
      "learning_rate": 7.074747990030816e-06,
      "loss": 0.8226,
      "step": 9476
    },
    {
      "epoch": 0.7658336532051152,
      "grad_norm": 2.4425439834594727,
      "learning_rate": 7.074152604532854e-06,
      "loss": 0.9616,
      "step": 9477
    },
    {
      "epoch": 0.7659144629184428,
      "grad_norm": 2.5771522521972656,
      "learning_rate": 7.073557183509651e-06,
      "loss": 1.143,
      "step": 9478
    },
    {
      "epoch": 0.7659952726317704,
      "grad_norm": 2.8968923091888428,
      "learning_rate": 7.072961726971405e-06,
      "loss": 0.8749,
      "step": 9479
    },
    {
      "epoch": 0.7660760823450978,
      "grad_norm": 2.923482894897461,
      "learning_rate": 7.072366234928316e-06,
      "loss": 0.9574,
      "step": 9480
    },
    {
      "epoch": 0.7661568920584254,
      "grad_norm": 2.593574285507202,
      "learning_rate": 7.071770707390582e-06,
      "loss": 0.8833,
      "step": 9481
    },
    {
      "epoch": 0.766237701771753,
      "grad_norm": 2.2582805156707764,
      "learning_rate": 7.071175144368403e-06,
      "loss": 1.1031,
      "step": 9482
    },
    {
      "epoch": 0.7663185114850805,
      "grad_norm": 3.0276057720184326,
      "learning_rate": 7.070579545871979e-06,
      "loss": 0.9027,
      "step": 9483
    },
    {
      "epoch": 0.766399321198408,
      "grad_norm": 2.48002290725708,
      "learning_rate": 7.069983911911513e-06,
      "loss": 1.0016,
      "step": 9484
    },
    {
      "epoch": 0.7664801309117356,
      "grad_norm": 2.7922701835632324,
      "learning_rate": 7.0693882424972074e-06,
      "loss": 1.0384,
      "step": 9485
    },
    {
      "epoch": 0.7665609406250631,
      "grad_norm": 2.288861036300659,
      "learning_rate": 7.068792537639261e-06,
      "loss": 0.9336,
      "step": 9486
    },
    {
      "epoch": 0.7666417503383907,
      "grad_norm": 2.3470566272735596,
      "learning_rate": 7.0681967973478795e-06,
      "loss": 0.9434,
      "step": 9487
    },
    {
      "epoch": 0.7667225600517182,
      "grad_norm": 2.6336758136749268,
      "learning_rate": 7.067601021633266e-06,
      "loss": 0.9795,
      "step": 9488
    },
    {
      "epoch": 0.7668033697650457,
      "grad_norm": 2.35019588470459,
      "learning_rate": 7.067005210505626e-06,
      "loss": 1.0797,
      "step": 9489
    },
    {
      "epoch": 0.7668841794783733,
      "grad_norm": 2.5723166465759277,
      "learning_rate": 7.066409363975161e-06,
      "loss": 1.0282,
      "step": 9490
    },
    {
      "epoch": 0.7669649891917009,
      "grad_norm": 2.7887842655181885,
      "learning_rate": 7.065813482052077e-06,
      "loss": 0.9215,
      "step": 9491
    },
    {
      "epoch": 0.7670457989050283,
      "grad_norm": 2.442120313644409,
      "learning_rate": 7.065217564746584e-06,
      "loss": 0.9704,
      "step": 9492
    },
    {
      "epoch": 0.7671266086183559,
      "grad_norm": 3.224236011505127,
      "learning_rate": 7.064621612068885e-06,
      "loss": 0.8509,
      "step": 9493
    },
    {
      "epoch": 0.7672074183316835,
      "grad_norm": 2.459911823272705,
      "learning_rate": 7.064025624029187e-06,
      "loss": 0.8022,
      "step": 9494
    },
    {
      "epoch": 0.767288228045011,
      "grad_norm": 2.5275039672851562,
      "learning_rate": 7.063429600637701e-06,
      "loss": 0.9088,
      "step": 9495
    },
    {
      "epoch": 0.7673690377583385,
      "grad_norm": 2.8273355960845947,
      "learning_rate": 7.062833541904631e-06,
      "loss": 0.9208,
      "step": 9496
    },
    {
      "epoch": 0.7674498474716661,
      "grad_norm": 2.6317386627197266,
      "learning_rate": 7.062237447840191e-06,
      "loss": 0.9965,
      "step": 9497
    },
    {
      "epoch": 0.7675306571849936,
      "grad_norm": 2.4512696266174316,
      "learning_rate": 7.061641318454586e-06,
      "loss": 1.0097,
      "step": 9498
    },
    {
      "epoch": 0.7676114668983212,
      "grad_norm": 4.055238723754883,
      "learning_rate": 7.0610451537580306e-06,
      "loss": 0.9901,
      "step": 9499
    },
    {
      "epoch": 0.7676922766116487,
      "grad_norm": 2.9475176334381104,
      "learning_rate": 7.060448953760732e-06,
      "loss": 1.043,
      "step": 9500
    },
    {
      "epoch": 0.7677730863249763,
      "grad_norm": 2.9297895431518555,
      "learning_rate": 7.059852718472904e-06,
      "loss": 0.9215,
      "step": 9501
    },
    {
      "epoch": 0.7678538960383038,
      "grad_norm": 2.8318982124328613,
      "learning_rate": 7.059256447904756e-06,
      "loss": 0.9256,
      "step": 9502
    },
    {
      "epoch": 0.7679347057516314,
      "grad_norm": 2.6794416904449463,
      "learning_rate": 7.058660142066506e-06,
      "loss": 0.8328,
      "step": 9503
    },
    {
      "epoch": 0.7680155154649589,
      "grad_norm": 3.0643105506896973,
      "learning_rate": 7.05806380096836e-06,
      "loss": 1.0295,
      "step": 9504
    },
    {
      "epoch": 0.7680963251782864,
      "grad_norm": 2.5476534366607666,
      "learning_rate": 7.057467424620539e-06,
      "loss": 0.9921,
      "step": 9505
    },
    {
      "epoch": 0.768177134891614,
      "grad_norm": 2.6490678787231445,
      "learning_rate": 7.056871013033252e-06,
      "loss": 0.941,
      "step": 9506
    },
    {
      "epoch": 0.7682579446049416,
      "grad_norm": 2.326160192489624,
      "learning_rate": 7.056274566216717e-06,
      "loss": 0.8238,
      "step": 9507
    },
    {
      "epoch": 0.768338754318269,
      "grad_norm": 2.27193546295166,
      "learning_rate": 7.055678084181148e-06,
      "loss": 0.9076,
      "step": 9508
    },
    {
      "epoch": 0.7684195640315966,
      "grad_norm": 3.777320623397827,
      "learning_rate": 7.055081566936763e-06,
      "loss": 0.9422,
      "step": 9509
    },
    {
      "epoch": 0.7685003737449242,
      "grad_norm": 2.8519272804260254,
      "learning_rate": 7.054485014493777e-06,
      "loss": 0.9746,
      "step": 9510
    },
    {
      "epoch": 0.7685811834582517,
      "grad_norm": 2.2314727306365967,
      "learning_rate": 7.053888426862412e-06,
      "loss": 0.9078,
      "step": 9511
    },
    {
      "epoch": 0.7686619931715792,
      "grad_norm": 2.3984363079071045,
      "learning_rate": 7.053291804052879e-06,
      "loss": 1.0029,
      "step": 9512
    },
    {
      "epoch": 0.7687428028849068,
      "grad_norm": 2.545302629470825,
      "learning_rate": 7.052695146075403e-06,
      "loss": 0.9006,
      "step": 9513
    },
    {
      "epoch": 0.7688236125982343,
      "grad_norm": 2.481466770172119,
      "learning_rate": 7.0520984529401995e-06,
      "loss": 0.907,
      "step": 9514
    },
    {
      "epoch": 0.7689044223115619,
      "grad_norm": 2.5026607513427734,
      "learning_rate": 7.05150172465749e-06,
      "loss": 0.9001,
      "step": 9515
    },
    {
      "epoch": 0.7689852320248894,
      "grad_norm": 2.6458675861358643,
      "learning_rate": 7.050904961237495e-06,
      "loss": 0.9075,
      "step": 9516
    },
    {
      "epoch": 0.7690660417382169,
      "grad_norm": 2.786869764328003,
      "learning_rate": 7.050308162690436e-06,
      "loss": 0.9283,
      "step": 9517
    },
    {
      "epoch": 0.7691468514515445,
      "grad_norm": 2.780898094177246,
      "learning_rate": 7.049711329026532e-06,
      "loss": 0.9822,
      "step": 9518
    },
    {
      "epoch": 0.7692276611648721,
      "grad_norm": 2.678616523742676,
      "learning_rate": 7.04911446025601e-06,
      "loss": 1.0265,
      "step": 9519
    },
    {
      "epoch": 0.7693084708781995,
      "grad_norm": 2.473202705383301,
      "learning_rate": 7.048517556389088e-06,
      "loss": 0.915,
      "step": 9520
    },
    {
      "epoch": 0.7693892805915271,
      "grad_norm": 2.3862879276275635,
      "learning_rate": 7.047920617435994e-06,
      "loss": 0.9869,
      "step": 9521
    },
    {
      "epoch": 0.7694700903048547,
      "grad_norm": 2.347320795059204,
      "learning_rate": 7.047323643406948e-06,
      "loss": 0.9873,
      "step": 9522
    },
    {
      "epoch": 0.7695509000181822,
      "grad_norm": 2.733719825744629,
      "learning_rate": 7.046726634312179e-06,
      "loss": 0.8149,
      "step": 9523
    },
    {
      "epoch": 0.7696317097315097,
      "grad_norm": 2.5755741596221924,
      "learning_rate": 7.046129590161908e-06,
      "loss": 1.0158,
      "step": 9524
    },
    {
      "epoch": 0.7697125194448373,
      "grad_norm": 2.6959922313690186,
      "learning_rate": 7.045532510966364e-06,
      "loss": 0.9994,
      "step": 9525
    },
    {
      "epoch": 0.7697933291581648,
      "grad_norm": 2.443272829055786,
      "learning_rate": 7.044935396735771e-06,
      "loss": 1.0158,
      "step": 9526
    },
    {
      "epoch": 0.7698741388714924,
      "grad_norm": 3.298548460006714,
      "learning_rate": 7.04433824748036e-06,
      "loss": 0.9775,
      "step": 9527
    },
    {
      "epoch": 0.7699549485848199,
      "grad_norm": 2.616126537322998,
      "learning_rate": 7.043741063210354e-06,
      "loss": 0.9429,
      "step": 9528
    },
    {
      "epoch": 0.7700357582981474,
      "grad_norm": 2.4621856212615967,
      "learning_rate": 7.043143843935985e-06,
      "loss": 1.0791,
      "step": 9529
    },
    {
      "epoch": 0.770116568011475,
      "grad_norm": 2.9813919067382812,
      "learning_rate": 7.042546589667481e-06,
      "loss": 1.0554,
      "step": 9530
    },
    {
      "epoch": 0.7701973777248026,
      "grad_norm": 2.506427764892578,
      "learning_rate": 7.0419493004150715e-06,
      "loss": 0.9289,
      "step": 9531
    },
    {
      "epoch": 0.77027818743813,
      "grad_norm": 2.6569604873657227,
      "learning_rate": 7.0413519761889835e-06,
      "loss": 0.8694,
      "step": 9532
    },
    {
      "epoch": 0.7703589971514576,
      "grad_norm": 2.4761314392089844,
      "learning_rate": 7.040754616999454e-06,
      "loss": 1.0014,
      "step": 9533
    },
    {
      "epoch": 0.7704398068647852,
      "grad_norm": 2.295379638671875,
      "learning_rate": 7.0401572228567094e-06,
      "loss": 0.8459,
      "step": 9534
    },
    {
      "epoch": 0.7705206165781127,
      "grad_norm": 2.436434030532837,
      "learning_rate": 7.039559793770983e-06,
      "loss": 0.9284,
      "step": 9535
    },
    {
      "epoch": 0.7706014262914402,
      "grad_norm": 2.2874650955200195,
      "learning_rate": 7.0389623297525065e-06,
      "loss": 0.8921,
      "step": 9536
    },
    {
      "epoch": 0.7706822360047678,
      "grad_norm": 2.787672519683838,
      "learning_rate": 7.038364830811516e-06,
      "loss": 0.8205,
      "step": 9537
    },
    {
      "epoch": 0.7707630457180953,
      "grad_norm": 2.6347997188568115,
      "learning_rate": 7.03776729695824e-06,
      "loss": 0.8769,
      "step": 9538
    },
    {
      "epoch": 0.7708438554314229,
      "grad_norm": 3.424485921859741,
      "learning_rate": 7.037169728202919e-06,
      "loss": 0.81,
      "step": 9539
    },
    {
      "epoch": 0.7709246651447504,
      "grad_norm": 2.503952741622925,
      "learning_rate": 7.036572124555783e-06,
      "loss": 0.7479,
      "step": 9540
    },
    {
      "epoch": 0.7710054748580779,
      "grad_norm": 2.5592153072357178,
      "learning_rate": 7.03597448602707e-06,
      "loss": 0.9672,
      "step": 9541
    },
    {
      "epoch": 0.7710862845714055,
      "grad_norm": 2.636793375015259,
      "learning_rate": 7.035376812627015e-06,
      "loss": 0.8467,
      "step": 9542
    },
    {
      "epoch": 0.7711670942847331,
      "grad_norm": 3.253643274307251,
      "learning_rate": 7.034779104365855e-06,
      "loss": 0.8573,
      "step": 9543
    },
    {
      "epoch": 0.7712479039980605,
      "grad_norm": 2.7215523719787598,
      "learning_rate": 7.034181361253829e-06,
      "loss": 0.9402,
      "step": 9544
    },
    {
      "epoch": 0.7713287137113881,
      "grad_norm": 2.5872609615325928,
      "learning_rate": 7.033583583301171e-06,
      "loss": 0.8611,
      "step": 9545
    },
    {
      "epoch": 0.7714095234247157,
      "grad_norm": 3.0744309425354004,
      "learning_rate": 7.032985770518123e-06,
      "loss": 0.9617,
      "step": 9546
    },
    {
      "epoch": 0.7714903331380432,
      "grad_norm": 2.5720481872558594,
      "learning_rate": 7.032387922914925e-06,
      "loss": 0.8442,
      "step": 9547
    },
    {
      "epoch": 0.7715711428513707,
      "grad_norm": 2.4857845306396484,
      "learning_rate": 7.031790040501812e-06,
      "loss": 0.9309,
      "step": 9548
    },
    {
      "epoch": 0.7716519525646983,
      "grad_norm": 2.5544092655181885,
      "learning_rate": 7.031192123289028e-06,
      "loss": 0.9437,
      "step": 9549
    },
    {
      "epoch": 0.7717327622780258,
      "grad_norm": 2.7516963481903076,
      "learning_rate": 7.030594171286813e-06,
      "loss": 0.8815,
      "step": 9550
    },
    {
      "epoch": 0.7718135719913534,
      "grad_norm": 2.5536606311798096,
      "learning_rate": 7.029996184505408e-06,
      "loss": 0.988,
      "step": 9551
    },
    {
      "epoch": 0.7718943817046809,
      "grad_norm": 2.81453800201416,
      "learning_rate": 7.029398162955054e-06,
      "loss": 0.9714,
      "step": 9552
    },
    {
      "epoch": 0.7719751914180084,
      "grad_norm": 2.854370355606079,
      "learning_rate": 7.028800106645996e-06,
      "loss": 1.0865,
      "step": 9553
    },
    {
      "epoch": 0.772056001131336,
      "grad_norm": 2.843309164047241,
      "learning_rate": 7.028202015588478e-06,
      "loss": 0.861,
      "step": 9554
    },
    {
      "epoch": 0.7721368108446636,
      "grad_norm": 2.750856876373291,
      "learning_rate": 7.02760388979274e-06,
      "loss": 0.9065,
      "step": 9555
    },
    {
      "epoch": 0.772217620557991,
      "grad_norm": 2.711421251296997,
      "learning_rate": 7.027005729269031e-06,
      "loss": 1.0317,
      "step": 9556
    },
    {
      "epoch": 0.7722984302713186,
      "grad_norm": 2.3563313484191895,
      "learning_rate": 7.026407534027592e-06,
      "loss": 0.9685,
      "step": 9557
    },
    {
      "epoch": 0.7723792399846462,
      "grad_norm": 2.8109371662139893,
      "learning_rate": 7.02580930407867e-06,
      "loss": 0.7691,
      "step": 9558
    },
    {
      "epoch": 0.7724600496979737,
      "grad_norm": 2.5159077644348145,
      "learning_rate": 7.025211039432512e-06,
      "loss": 0.9224,
      "step": 9559
    },
    {
      "epoch": 0.7725408594113012,
      "grad_norm": 2.9255714416503906,
      "learning_rate": 7.024612740099364e-06,
      "loss": 0.9872,
      "step": 9560
    },
    {
      "epoch": 0.7726216691246288,
      "grad_norm": 2.7818901538848877,
      "learning_rate": 7.024014406089475e-06,
      "loss": 0.9206,
      "step": 9561
    },
    {
      "epoch": 0.7727024788379563,
      "grad_norm": 2.6201705932617188,
      "learning_rate": 7.023416037413091e-06,
      "loss": 0.8499,
      "step": 9562
    },
    {
      "epoch": 0.7727832885512839,
      "grad_norm": 2.3721187114715576,
      "learning_rate": 7.022817634080461e-06,
      "loss": 1.0093,
      "step": 9563
    },
    {
      "epoch": 0.7728640982646114,
      "grad_norm": 2.4636049270629883,
      "learning_rate": 7.022219196101836e-06,
      "loss": 0.9829,
      "step": 9564
    },
    {
      "epoch": 0.7729449079779389,
      "grad_norm": 2.4145498275756836,
      "learning_rate": 7.021620723487464e-06,
      "loss": 0.8964,
      "step": 9565
    },
    {
      "epoch": 0.7730257176912665,
      "grad_norm": 2.788201332092285,
      "learning_rate": 7.021022216247595e-06,
      "loss": 0.8574,
      "step": 9566
    },
    {
      "epoch": 0.7731065274045941,
      "grad_norm": 2.318821430206299,
      "learning_rate": 7.02042367439248e-06,
      "loss": 0.8232,
      "step": 9567
    },
    {
      "epoch": 0.7731873371179215,
      "grad_norm": 2.2990260124206543,
      "learning_rate": 7.019825097932373e-06,
      "loss": 0.9651,
      "step": 9568
    },
    {
      "epoch": 0.7732681468312491,
      "grad_norm": 2.5955026149749756,
      "learning_rate": 7.019226486877525e-06,
      "loss": 0.904,
      "step": 9569
    },
    {
      "epoch": 0.7733489565445767,
      "grad_norm": 2.8680477142333984,
      "learning_rate": 7.018627841238188e-06,
      "loss": 0.9284,
      "step": 9570
    },
    {
      "epoch": 0.7734297662579042,
      "grad_norm": 2.5354461669921875,
      "learning_rate": 7.018029161024615e-06,
      "loss": 1.0128,
      "step": 9571
    },
    {
      "epoch": 0.7735105759712317,
      "grad_norm": 2.5947415828704834,
      "learning_rate": 7.017430446247062e-06,
      "loss": 0.8596,
      "step": 9572
    },
    {
      "epoch": 0.7735913856845593,
      "grad_norm": 2.6628518104553223,
      "learning_rate": 7.016831696915782e-06,
      "loss": 0.8106,
      "step": 9573
    },
    {
      "epoch": 0.7736721953978868,
      "grad_norm": 2.7208001613616943,
      "learning_rate": 7.016232913041029e-06,
      "loss": 0.9419,
      "step": 9574
    },
    {
      "epoch": 0.7737530051112144,
      "grad_norm": 2.77506947517395,
      "learning_rate": 7.01563409463306e-06,
      "loss": 0.9868,
      "step": 9575
    },
    {
      "epoch": 0.7738338148245419,
      "grad_norm": 2.494380235671997,
      "learning_rate": 7.015035241702133e-06,
      "loss": 0.852,
      "step": 9576
    },
    {
      "epoch": 0.7739146245378694,
      "grad_norm": 2.387253522872925,
      "learning_rate": 7.014436354258501e-06,
      "loss": 0.9764,
      "step": 9577
    },
    {
      "epoch": 0.773995434251197,
      "grad_norm": 2.615269899368286,
      "learning_rate": 7.013837432312427e-06,
      "loss": 0.8534,
      "step": 9578
    },
    {
      "epoch": 0.7740762439645246,
      "grad_norm": 3.229938268661499,
      "learning_rate": 7.013238475874163e-06,
      "loss": 0.8501,
      "step": 9579
    },
    {
      "epoch": 0.774157053677852,
      "grad_norm": 2.679093360900879,
      "learning_rate": 7.012639484953973e-06,
      "loss": 0.8724,
      "step": 9580
    },
    {
      "epoch": 0.7742378633911796,
      "grad_norm": 2.995333433151245,
      "learning_rate": 7.0120404595621125e-06,
      "loss": 0.9358,
      "step": 9581
    },
    {
      "epoch": 0.7743186731045072,
      "grad_norm": 3.551964521408081,
      "learning_rate": 7.011441399708842e-06,
      "loss": 0.9257,
      "step": 9582
    },
    {
      "epoch": 0.7743994828178347,
      "grad_norm": 2.7420544624328613,
      "learning_rate": 7.010842305404424e-06,
      "loss": 0.9684,
      "step": 9583
    },
    {
      "epoch": 0.7744802925311622,
      "grad_norm": 3.192138195037842,
      "learning_rate": 7.010243176659118e-06,
      "loss": 0.8429,
      "step": 9584
    },
    {
      "epoch": 0.7745611022444898,
      "grad_norm": 2.406846046447754,
      "learning_rate": 7.009644013483186e-06,
      "loss": 0.9449,
      "step": 9585
    },
    {
      "epoch": 0.7746419119578173,
      "grad_norm": 2.807209014892578,
      "learning_rate": 7.00904481588689e-06,
      "loss": 1.1005,
      "step": 9586
    },
    {
      "epoch": 0.7747227216711449,
      "grad_norm": 2.7669317722320557,
      "learning_rate": 7.008445583880492e-06,
      "loss": 1.0634,
      "step": 9587
    },
    {
      "epoch": 0.7748035313844724,
      "grad_norm": 2.5482051372528076,
      "learning_rate": 7.007846317474257e-06,
      "loss": 0.9123,
      "step": 9588
    },
    {
      "epoch": 0.7748843410977999,
      "grad_norm": 2.2731552124023438,
      "learning_rate": 7.007247016678448e-06,
      "loss": 0.9038,
      "step": 9589
    },
    {
      "epoch": 0.7749651508111275,
      "grad_norm": 2.675006628036499,
      "learning_rate": 7.006647681503331e-06,
      "loss": 0.9246,
      "step": 9590
    },
    {
      "epoch": 0.7750459605244551,
      "grad_norm": 2.4828829765319824,
      "learning_rate": 7.006048311959168e-06,
      "loss": 0.9194,
      "step": 9591
    },
    {
      "epoch": 0.7751267702377825,
      "grad_norm": 3.1538140773773193,
      "learning_rate": 7.0054489080562284e-06,
      "loss": 0.8672,
      "step": 9592
    },
    {
      "epoch": 0.7752075799511101,
      "grad_norm": 2.542527198791504,
      "learning_rate": 7.004849469804775e-06,
      "loss": 0.9722,
      "step": 9593
    },
    {
      "epoch": 0.7752883896644377,
      "grad_norm": 2.6315410137176514,
      "learning_rate": 7.004249997215079e-06,
      "loss": 0.9036,
      "step": 9594
    },
    {
      "epoch": 0.7753691993777652,
      "grad_norm": 2.555668830871582,
      "learning_rate": 7.0036504902974044e-06,
      "loss": 0.8578,
      "step": 9595
    },
    {
      "epoch": 0.7754500090910927,
      "grad_norm": 2.8489654064178467,
      "learning_rate": 7.00305094906202e-06,
      "loss": 1.0087,
      "step": 9596
    },
    {
      "epoch": 0.7755308188044203,
      "grad_norm": 2.558403968811035,
      "learning_rate": 7.0024513735191964e-06,
      "loss": 0.9447,
      "step": 9597
    },
    {
      "epoch": 0.7756116285177478,
      "grad_norm": 3.083716630935669,
      "learning_rate": 7.001851763679201e-06,
      "loss": 0.8327,
      "step": 9598
    },
    {
      "epoch": 0.7756924382310754,
      "grad_norm": 2.7847039699554443,
      "learning_rate": 7.0012521195523034e-06,
      "loss": 0.9479,
      "step": 9599
    },
    {
      "epoch": 0.7757732479444029,
      "grad_norm": 2.4878573417663574,
      "learning_rate": 7.000652441148777e-06,
      "loss": 0.8972,
      "step": 9600
    },
    {
      "epoch": 0.7758540576577304,
      "grad_norm": 2.6768605709075928,
      "learning_rate": 7.000052728478888e-06,
      "loss": 0.931,
      "step": 9601
    },
    {
      "epoch": 0.775934867371058,
      "grad_norm": 3.023346424102783,
      "learning_rate": 6.999452981552911e-06,
      "loss": 0.8828,
      "step": 9602
    },
    {
      "epoch": 0.7760156770843856,
      "grad_norm": 2.6194918155670166,
      "learning_rate": 6.998853200381118e-06,
      "loss": 1.0002,
      "step": 9603
    },
    {
      "epoch": 0.776096486797713,
      "grad_norm": 2.7090258598327637,
      "learning_rate": 6.998253384973784e-06,
      "loss": 0.9234,
      "step": 9604
    },
    {
      "epoch": 0.7761772965110406,
      "grad_norm": 2.317145824432373,
      "learning_rate": 6.997653535341177e-06,
      "loss": 0.8695,
      "step": 9605
    },
    {
      "epoch": 0.7762581062243682,
      "grad_norm": 2.125000238418579,
      "learning_rate": 6.997053651493576e-06,
      "loss": 0.9545,
      "step": 9606
    },
    {
      "epoch": 0.7763389159376957,
      "grad_norm": 2.7297613620758057,
      "learning_rate": 6.996453733441252e-06,
      "loss": 0.8297,
      "step": 9607
    },
    {
      "epoch": 0.7764197256510232,
      "grad_norm": 2.6772239208221436,
      "learning_rate": 6.995853781194484e-06,
      "loss": 1.0247,
      "step": 9608
    },
    {
      "epoch": 0.7765005353643508,
      "grad_norm": 2.860786199569702,
      "learning_rate": 6.995253794763545e-06,
      "loss": 0.997,
      "step": 9609
    },
    {
      "epoch": 0.7765813450776783,
      "grad_norm": 2.5108728408813477,
      "learning_rate": 6.994653774158711e-06,
      "loss": 0.981,
      "step": 9610
    },
    {
      "epoch": 0.7766621547910059,
      "grad_norm": 2.5882833003997803,
      "learning_rate": 6.9940537193902594e-06,
      "loss": 0.9718,
      "step": 9611
    },
    {
      "epoch": 0.7767429645043334,
      "grad_norm": 3.0769033432006836,
      "learning_rate": 6.993453630468468e-06,
      "loss": 0.9234,
      "step": 9612
    },
    {
      "epoch": 0.7768237742176609,
      "grad_norm": 2.6407887935638428,
      "learning_rate": 6.992853507403617e-06,
      "loss": 0.9675,
      "step": 9613
    },
    {
      "epoch": 0.7769045839309885,
      "grad_norm": 2.8648247718811035,
      "learning_rate": 6.992253350205982e-06,
      "loss": 0.9322,
      "step": 9614
    },
    {
      "epoch": 0.7769853936443161,
      "grad_norm": 2.362705707550049,
      "learning_rate": 6.991653158885842e-06,
      "loss": 0.9228,
      "step": 9615
    },
    {
      "epoch": 0.7770662033576435,
      "grad_norm": 2.714418411254883,
      "learning_rate": 6.99105293345348e-06,
      "loss": 0.9775,
      "step": 9616
    },
    {
      "epoch": 0.7771470130709711,
      "grad_norm": 2.5210421085357666,
      "learning_rate": 6.990452673919174e-06,
      "loss": 0.9148,
      "step": 9617
    },
    {
      "epoch": 0.7772278227842987,
      "grad_norm": 2.44423770904541,
      "learning_rate": 6.989852380293205e-06,
      "loss": 0.8282,
      "step": 9618
    },
    {
      "epoch": 0.7773086324976262,
      "grad_norm": 3.33247447013855,
      "learning_rate": 6.989252052585856e-06,
      "loss": 1.0303,
      "step": 9619
    },
    {
      "epoch": 0.7773894422109537,
      "grad_norm": 2.6625845432281494,
      "learning_rate": 6.988651690807407e-06,
      "loss": 0.9808,
      "step": 9620
    },
    {
      "epoch": 0.7774702519242813,
      "grad_norm": 2.823176145553589,
      "learning_rate": 6.988051294968142e-06,
      "loss": 1.0867,
      "step": 9621
    },
    {
      "epoch": 0.7775510616376088,
      "grad_norm": 2.1813924312591553,
      "learning_rate": 6.987450865078344e-06,
      "loss": 0.8795,
      "step": 9622
    },
    {
      "epoch": 0.7776318713509364,
      "grad_norm": 3.0216259956359863,
      "learning_rate": 6.986850401148299e-06,
      "loss": 0.9338,
      "step": 9623
    },
    {
      "epoch": 0.7777126810642639,
      "grad_norm": 2.6659300327301025,
      "learning_rate": 6.986249903188289e-06,
      "loss": 0.8721,
      "step": 9624
    },
    {
      "epoch": 0.7777934907775914,
      "grad_norm": 2.786710739135742,
      "learning_rate": 6.985649371208601e-06,
      "loss": 1.0178,
      "step": 9625
    },
    {
      "epoch": 0.777874300490919,
      "grad_norm": 2.4938833713531494,
      "learning_rate": 6.985048805219518e-06,
      "loss": 0.9823,
      "step": 9626
    },
    {
      "epoch": 0.7779551102042466,
      "grad_norm": 2.678549289703369,
      "learning_rate": 6.984448205231328e-06,
      "loss": 1.0505,
      "step": 9627
    },
    {
      "epoch": 0.778035919917574,
      "grad_norm": 2.584136724472046,
      "learning_rate": 6.983847571254317e-06,
      "loss": 0.9083,
      "step": 9628
    },
    {
      "epoch": 0.7781167296309016,
      "grad_norm": 2.886458158493042,
      "learning_rate": 6.983246903298775e-06,
      "loss": 0.8482,
      "step": 9629
    },
    {
      "epoch": 0.7781975393442292,
      "grad_norm": 2.893474578857422,
      "learning_rate": 6.982646201374985e-06,
      "loss": 0.8756,
      "step": 9630
    },
    {
      "epoch": 0.7782783490575568,
      "grad_norm": 2.426586389541626,
      "learning_rate": 6.982045465493241e-06,
      "loss": 0.9781,
      "step": 9631
    },
    {
      "epoch": 0.7783591587708842,
      "grad_norm": 2.4432895183563232,
      "learning_rate": 6.9814446956638305e-06,
      "loss": 0.8297,
      "step": 9632
    },
    {
      "epoch": 0.7784399684842118,
      "grad_norm": 2.637373924255371,
      "learning_rate": 6.98084389189704e-06,
      "loss": 0.9191,
      "step": 9633
    },
    {
      "epoch": 0.7785207781975394,
      "grad_norm": 2.6416330337524414,
      "learning_rate": 6.9802430542031645e-06,
      "loss": 1.075,
      "step": 9634
    },
    {
      "epoch": 0.7786015879108669,
      "grad_norm": 2.9617810249328613,
      "learning_rate": 6.979642182592491e-06,
      "loss": 0.8736,
      "step": 9635
    },
    {
      "epoch": 0.7786823976241944,
      "grad_norm": 2.736532688140869,
      "learning_rate": 6.979041277075313e-06,
      "loss": 0.9148,
      "step": 9636
    },
    {
      "epoch": 0.778763207337522,
      "grad_norm": 3.1691181659698486,
      "learning_rate": 6.978440337661923e-06,
      "loss": 1.0436,
      "step": 9637
    },
    {
      "epoch": 0.7788440170508495,
      "grad_norm": 2.742832660675049,
      "learning_rate": 6.977839364362612e-06,
      "loss": 0.9721,
      "step": 9638
    },
    {
      "epoch": 0.7789248267641771,
      "grad_norm": 2.6143288612365723,
      "learning_rate": 6.977238357187675e-06,
      "loss": 0.958,
      "step": 9639
    },
    {
      "epoch": 0.7790056364775046,
      "grad_norm": 3.0187857151031494,
      "learning_rate": 6.9766373161474054e-06,
      "loss": 0.8361,
      "step": 9640
    },
    {
      "epoch": 0.7790864461908321,
      "grad_norm": 2.5095255374908447,
      "learning_rate": 6.976036241252095e-06,
      "loss": 0.9714,
      "step": 9641
    },
    {
      "epoch": 0.7791672559041597,
      "grad_norm": 2.383462905883789,
      "learning_rate": 6.9754351325120426e-06,
      "loss": 0.9255,
      "step": 9642
    },
    {
      "epoch": 0.7792480656174873,
      "grad_norm": 3.18103289604187,
      "learning_rate": 6.974833989937543e-06,
      "loss": 0.7869,
      "step": 9643
    },
    {
      "epoch": 0.7793288753308147,
      "grad_norm": 2.3761940002441406,
      "learning_rate": 6.9742328135388896e-06,
      "loss": 0.9537,
      "step": 9644
    },
    {
      "epoch": 0.7794096850441423,
      "grad_norm": 2.1563880443573,
      "learning_rate": 6.973631603326382e-06,
      "loss": 1.1183,
      "step": 9645
    },
    {
      "epoch": 0.7794904947574699,
      "grad_norm": 2.8754329681396484,
      "learning_rate": 6.973030359310315e-06,
      "loss": 0.8996,
      "step": 9646
    },
    {
      "epoch": 0.7795713044707974,
      "grad_norm": 3.062407970428467,
      "learning_rate": 6.972429081500989e-06,
      "loss": 0.9424,
      "step": 9647
    },
    {
      "epoch": 0.779652114184125,
      "grad_norm": 3.4805405139923096,
      "learning_rate": 6.971827769908701e-06,
      "loss": 0.9288,
      "step": 9648
    },
    {
      "epoch": 0.7797329238974525,
      "grad_norm": 2.5562899112701416,
      "learning_rate": 6.971226424543749e-06,
      "loss": 0.882,
      "step": 9649
    },
    {
      "epoch": 0.77981373361078,
      "grad_norm": 2.508535146713257,
      "learning_rate": 6.970625045416435e-06,
      "loss": 0.8526,
      "step": 9650
    },
    {
      "epoch": 0.7798945433241076,
      "grad_norm": 2.536494016647339,
      "learning_rate": 6.97002363253706e-06,
      "loss": 0.9487,
      "step": 9651
    },
    {
      "epoch": 0.7799753530374351,
      "grad_norm": 2.586965322494507,
      "learning_rate": 6.96942218591592e-06,
      "loss": 0.9016,
      "step": 9652
    },
    {
      "epoch": 0.7800561627507626,
      "grad_norm": 2.5410730838775635,
      "learning_rate": 6.968820705563319e-06,
      "loss": 0.9202,
      "step": 9653
    },
    {
      "epoch": 0.7801369724640902,
      "grad_norm": 2.654521942138672,
      "learning_rate": 6.96821919148956e-06,
      "loss": 0.912,
      "step": 9654
    },
    {
      "epoch": 0.7802177821774178,
      "grad_norm": 2.298339366912842,
      "learning_rate": 6.967617643704945e-06,
      "loss": 0.9096,
      "step": 9655
    },
    {
      "epoch": 0.7802985918907452,
      "grad_norm": 2.3644843101501465,
      "learning_rate": 6.967016062219777e-06,
      "loss": 1.0322,
      "step": 9656
    },
    {
      "epoch": 0.7803794016040728,
      "grad_norm": 2.7085025310516357,
      "learning_rate": 6.966414447044359e-06,
      "loss": 0.9766,
      "step": 9657
    },
    {
      "epoch": 0.7804602113174004,
      "grad_norm": 2.3096635341644287,
      "learning_rate": 6.965812798188994e-06,
      "loss": 0.9163,
      "step": 9658
    },
    {
      "epoch": 0.7805410210307279,
      "grad_norm": 2.3417112827301025,
      "learning_rate": 6.96521111566399e-06,
      "loss": 0.9755,
      "step": 9659
    },
    {
      "epoch": 0.7806218307440554,
      "grad_norm": 2.7602450847625732,
      "learning_rate": 6.964609399479649e-06,
      "loss": 1.0441,
      "step": 9660
    },
    {
      "epoch": 0.780702640457383,
      "grad_norm": 2.400860548019409,
      "learning_rate": 6.964007649646281e-06,
      "loss": 0.8635,
      "step": 9661
    },
    {
      "epoch": 0.7807834501707105,
      "grad_norm": 2.5426175594329834,
      "learning_rate": 6.963405866174188e-06,
      "loss": 0.8881,
      "step": 9662
    },
    {
      "epoch": 0.7808642598840381,
      "grad_norm": 2.614553213119507,
      "learning_rate": 6.962804049073679e-06,
      "loss": 0.8525,
      "step": 9663
    },
    {
      "epoch": 0.7809450695973656,
      "grad_norm": 2.7390501499176025,
      "learning_rate": 6.962202198355062e-06,
      "loss": 1.0068,
      "step": 9664
    },
    {
      "epoch": 0.7810258793106931,
      "grad_norm": 2.543248176574707,
      "learning_rate": 6.961600314028647e-06,
      "loss": 0.9534,
      "step": 9665
    },
    {
      "epoch": 0.7811066890240207,
      "grad_norm": 3.6785783767700195,
      "learning_rate": 6.960998396104739e-06,
      "loss": 0.9088,
      "step": 9666
    },
    {
      "epoch": 0.7811874987373483,
      "grad_norm": 2.6666314601898193,
      "learning_rate": 6.960396444593651e-06,
      "loss": 0.977,
      "step": 9667
    },
    {
      "epoch": 0.7812683084506757,
      "grad_norm": 2.5132291316986084,
      "learning_rate": 6.959794459505691e-06,
      "loss": 0.8372,
      "step": 9668
    },
    {
      "epoch": 0.7813491181640033,
      "grad_norm": 2.5730397701263428,
      "learning_rate": 6.959192440851169e-06,
      "loss": 1.0269,
      "step": 9669
    },
    {
      "epoch": 0.7814299278773309,
      "grad_norm": 3.234506607055664,
      "learning_rate": 6.958590388640397e-06,
      "loss": 1.0491,
      "step": 9670
    },
    {
      "epoch": 0.7815107375906584,
      "grad_norm": 2.641125440597534,
      "learning_rate": 6.957988302883688e-06,
      "loss": 0.8949,
      "step": 9671
    },
    {
      "epoch": 0.781591547303986,
      "grad_norm": 2.720885753631592,
      "learning_rate": 6.957386183591351e-06,
      "loss": 1.0241,
      "step": 9672
    },
    {
      "epoch": 0.7816723570173135,
      "grad_norm": 2.790813446044922,
      "learning_rate": 6.9567840307737035e-06,
      "loss": 0.9351,
      "step": 9673
    },
    {
      "epoch": 0.781753166730641,
      "grad_norm": 2.3688058853149414,
      "learning_rate": 6.9561818444410545e-06,
      "loss": 0.868,
      "step": 9674
    },
    {
      "epoch": 0.7818339764439686,
      "grad_norm": 2.60577392578125,
      "learning_rate": 6.955579624603721e-06,
      "loss": 0.8864,
      "step": 9675
    },
    {
      "epoch": 0.7819147861572961,
      "grad_norm": 2.5541322231292725,
      "learning_rate": 6.954977371272016e-06,
      "loss": 0.9899,
      "step": 9676
    },
    {
      "epoch": 0.7819955958706236,
      "grad_norm": 2.799955129623413,
      "learning_rate": 6.954375084456254e-06,
      "loss": 0.9234,
      "step": 9677
    },
    {
      "epoch": 0.7820764055839512,
      "grad_norm": 2.387197971343994,
      "learning_rate": 6.953772764166753e-06,
      "loss": 1.006,
      "step": 9678
    },
    {
      "epoch": 0.7821572152972788,
      "grad_norm": 2.6300976276397705,
      "learning_rate": 6.953170410413828e-06,
      "loss": 0.8719,
      "step": 9679
    },
    {
      "epoch": 0.7822380250106062,
      "grad_norm": 2.34307861328125,
      "learning_rate": 6.952568023207795e-06,
      "loss": 0.9994,
      "step": 9680
    },
    {
      "epoch": 0.7823188347239338,
      "grad_norm": 2.5072860717773438,
      "learning_rate": 6.951965602558973e-06,
      "loss": 0.9497,
      "step": 9681
    },
    {
      "epoch": 0.7823996444372614,
      "grad_norm": 2.631420850753784,
      "learning_rate": 6.95136314847768e-06,
      "loss": 0.9123,
      "step": 9682
    },
    {
      "epoch": 0.7824804541505889,
      "grad_norm": 2.6064393520355225,
      "learning_rate": 6.950760660974233e-06,
      "loss": 1.0639,
      "step": 9683
    },
    {
      "epoch": 0.7825612638639164,
      "grad_norm": 2.7990331649780273,
      "learning_rate": 6.950158140058953e-06,
      "loss": 0.7694,
      "step": 9684
    },
    {
      "epoch": 0.782642073577244,
      "grad_norm": 2.9253673553466797,
      "learning_rate": 6.949555585742157e-06,
      "loss": 1.1346,
      "step": 9685
    },
    {
      "epoch": 0.7827228832905715,
      "grad_norm": 2.707361936569214,
      "learning_rate": 6.948952998034168e-06,
      "loss": 0.9165,
      "step": 9686
    },
    {
      "epoch": 0.7828036930038991,
      "grad_norm": 2.8470633029937744,
      "learning_rate": 6.948350376945307e-06,
      "loss": 0.941,
      "step": 9687
    },
    {
      "epoch": 0.7828845027172266,
      "grad_norm": 2.722989320755005,
      "learning_rate": 6.947747722485893e-06,
      "loss": 0.9912,
      "step": 9688
    },
    {
      "epoch": 0.7829653124305541,
      "grad_norm": 3.1429643630981445,
      "learning_rate": 6.947145034666253e-06,
      "loss": 0.9211,
      "step": 9689
    },
    {
      "epoch": 0.7830461221438817,
      "grad_norm": 2.6342954635620117,
      "learning_rate": 6.9465423134967035e-06,
      "loss": 0.9303,
      "step": 9690
    },
    {
      "epoch": 0.7831269318572093,
      "grad_norm": 2.8902268409729004,
      "learning_rate": 6.94593955898757e-06,
      "loss": 0.9141,
      "step": 9691
    },
    {
      "epoch": 0.7832077415705367,
      "grad_norm": 2.7754554748535156,
      "learning_rate": 6.945336771149177e-06,
      "loss": 0.8759,
      "step": 9692
    },
    {
      "epoch": 0.7832885512838643,
      "grad_norm": 2.7848026752471924,
      "learning_rate": 6.9447339499918485e-06,
      "loss": 0.9073,
      "step": 9693
    },
    {
      "epoch": 0.7833693609971919,
      "grad_norm": 2.5997676849365234,
      "learning_rate": 6.944131095525909e-06,
      "loss": 0.8533,
      "step": 9694
    },
    {
      "epoch": 0.7834501707105194,
      "grad_norm": 2.757568359375,
      "learning_rate": 6.943528207761684e-06,
      "loss": 0.9624,
      "step": 9695
    },
    {
      "epoch": 0.783530980423847,
      "grad_norm": 2.446953535079956,
      "learning_rate": 6.942925286709501e-06,
      "loss": 0.9607,
      "step": 9696
    },
    {
      "epoch": 0.7836117901371745,
      "grad_norm": 2.747488260269165,
      "learning_rate": 6.942322332379683e-06,
      "loss": 0.9538,
      "step": 9697
    },
    {
      "epoch": 0.783692599850502,
      "grad_norm": 3.0803427696228027,
      "learning_rate": 6.94171934478256e-06,
      "loss": 0.972,
      "step": 9698
    },
    {
      "epoch": 0.7837734095638296,
      "grad_norm": 2.4530720710754395,
      "learning_rate": 6.94111632392846e-06,
      "loss": 1.0084,
      "step": 9699
    },
    {
      "epoch": 0.7838542192771571,
      "grad_norm": 2.7214879989624023,
      "learning_rate": 6.94051326982771e-06,
      "loss": 1.0035,
      "step": 9700
    },
    {
      "epoch": 0.7839350289904846,
      "grad_norm": 2.7606163024902344,
      "learning_rate": 6.939910182490639e-06,
      "loss": 0.9441,
      "step": 9701
    },
    {
      "epoch": 0.7840158387038122,
      "grad_norm": 2.752997875213623,
      "learning_rate": 6.939307061927577e-06,
      "loss": 0.9443,
      "step": 9702
    },
    {
      "epoch": 0.7840966484171398,
      "grad_norm": 2.579437732696533,
      "learning_rate": 6.938703908148854e-06,
      "loss": 1.0311,
      "step": 9703
    },
    {
      "epoch": 0.7841774581304672,
      "grad_norm": 2.837007761001587,
      "learning_rate": 6.9381007211648e-06,
      "loss": 0.8975,
      "step": 9704
    },
    {
      "epoch": 0.7842582678437948,
      "grad_norm": 2.9079701900482178,
      "learning_rate": 6.937497500985746e-06,
      "loss": 0.9381,
      "step": 9705
    },
    {
      "epoch": 0.7843390775571224,
      "grad_norm": 2.8375697135925293,
      "learning_rate": 6.936894247622026e-06,
      "loss": 1.0136,
      "step": 9706
    },
    {
      "epoch": 0.7844198872704499,
      "grad_norm": 2.6654367446899414,
      "learning_rate": 6.936290961083968e-06,
      "loss": 0.9365,
      "step": 9707
    },
    {
      "epoch": 0.7845006969837774,
      "grad_norm": 3.0772650241851807,
      "learning_rate": 6.935687641381908e-06,
      "loss": 1.1799,
      "step": 9708
    },
    {
      "epoch": 0.784581506697105,
      "grad_norm": 2.334848165512085,
      "learning_rate": 6.935084288526179e-06,
      "loss": 0.9563,
      "step": 9709
    },
    {
      "epoch": 0.7846623164104325,
      "grad_norm": 2.4315085411071777,
      "learning_rate": 6.934480902527115e-06,
      "loss": 0.8677,
      "step": 9710
    },
    {
      "epoch": 0.7847431261237601,
      "grad_norm": 2.638908624649048,
      "learning_rate": 6.93387748339505e-06,
      "loss": 0.8608,
      "step": 9711
    },
    {
      "epoch": 0.7848239358370876,
      "grad_norm": 2.3801815509796143,
      "learning_rate": 6.933274031140319e-06,
      "loss": 0.9107,
      "step": 9712
    },
    {
      "epoch": 0.7849047455504151,
      "grad_norm": 2.940345525741577,
      "learning_rate": 6.932670545773259e-06,
      "loss": 0.8846,
      "step": 9713
    },
    {
      "epoch": 0.7849855552637427,
      "grad_norm": 3.1622703075408936,
      "learning_rate": 6.9320670273042034e-06,
      "loss": 0.8589,
      "step": 9714
    },
    {
      "epoch": 0.7850663649770703,
      "grad_norm": 2.639561891555786,
      "learning_rate": 6.931463475743492e-06,
      "loss": 0.9901,
      "step": 9715
    },
    {
      "epoch": 0.7851471746903977,
      "grad_norm": 2.6701161861419678,
      "learning_rate": 6.930859891101461e-06,
      "loss": 0.9628,
      "step": 9716
    },
    {
      "epoch": 0.7852279844037253,
      "grad_norm": 2.7424087524414062,
      "learning_rate": 6.930256273388448e-06,
      "loss": 0.9713,
      "step": 9717
    },
    {
      "epoch": 0.7853087941170529,
      "grad_norm": 2.560697555541992,
      "learning_rate": 6.929652622614793e-06,
      "loss": 0.857,
      "step": 9718
    },
    {
      "epoch": 0.7853896038303804,
      "grad_norm": 2.141119956970215,
      "learning_rate": 6.929048938790832e-06,
      "loss": 1.0336,
      "step": 9719
    },
    {
      "epoch": 0.785470413543708,
      "grad_norm": 2.396378517150879,
      "learning_rate": 6.928445221926909e-06,
      "loss": 0.9565,
      "step": 9720
    },
    {
      "epoch": 0.7855512232570355,
      "grad_norm": 2.7309021949768066,
      "learning_rate": 6.927841472033362e-06,
      "loss": 0.8041,
      "step": 9721
    },
    {
      "epoch": 0.785632032970363,
      "grad_norm": 3.081422805786133,
      "learning_rate": 6.9272376891205296e-06,
      "loss": 1.0131,
      "step": 9722
    },
    {
      "epoch": 0.7857128426836906,
      "grad_norm": 2.583186388015747,
      "learning_rate": 6.926633873198757e-06,
      "loss": 0.8351,
      "step": 9723
    },
    {
      "epoch": 0.7857936523970181,
      "grad_norm": 2.7975621223449707,
      "learning_rate": 6.926030024278384e-06,
      "loss": 0.8724,
      "step": 9724
    },
    {
      "epoch": 0.7858744621103456,
      "grad_norm": 2.965134620666504,
      "learning_rate": 6.925426142369752e-06,
      "loss": 0.877,
      "step": 9725
    },
    {
      "epoch": 0.7859552718236732,
      "grad_norm": 2.5071730613708496,
      "learning_rate": 6.924822227483208e-06,
      "loss": 0.9737,
      "step": 9726
    },
    {
      "epoch": 0.7860360815370008,
      "grad_norm": 2.840658664703369,
      "learning_rate": 6.92421827962909e-06,
      "loss": 0.9779,
      "step": 9727
    },
    {
      "epoch": 0.7861168912503282,
      "grad_norm": 3.170917272567749,
      "learning_rate": 6.923614298817747e-06,
      "loss": 0.9217,
      "step": 9728
    },
    {
      "epoch": 0.7861977009636558,
      "grad_norm": 2.484788179397583,
      "learning_rate": 6.923010285059521e-06,
      "loss": 1.0302,
      "step": 9729
    },
    {
      "epoch": 0.7862785106769834,
      "grad_norm": 2.6187918186187744,
      "learning_rate": 6.9224062383647595e-06,
      "loss": 1.0012,
      "step": 9730
    },
    {
      "epoch": 0.7863593203903109,
      "grad_norm": 3.17228627204895,
      "learning_rate": 6.921802158743807e-06,
      "loss": 0.9174,
      "step": 9731
    },
    {
      "epoch": 0.7864401301036384,
      "grad_norm": 2.9996063709259033,
      "learning_rate": 6.92119804620701e-06,
      "loss": 0.9287,
      "step": 9732
    },
    {
      "epoch": 0.786520939816966,
      "grad_norm": 3.090010166168213,
      "learning_rate": 6.920593900764714e-06,
      "loss": 0.9699,
      "step": 9733
    },
    {
      "epoch": 0.7866017495302935,
      "grad_norm": 2.606588125228882,
      "learning_rate": 6.91998972242727e-06,
      "loss": 0.9704,
      "step": 9734
    },
    {
      "epoch": 0.7866825592436211,
      "grad_norm": 2.7972769737243652,
      "learning_rate": 6.919385511205024e-06,
      "loss": 0.8863,
      "step": 9735
    },
    {
      "epoch": 0.7867633689569486,
      "grad_norm": 2.594951868057251,
      "learning_rate": 6.918781267108324e-06,
      "loss": 0.9085,
      "step": 9736
    },
    {
      "epoch": 0.7868441786702761,
      "grad_norm": 3.1590778827667236,
      "learning_rate": 6.918176990147522e-06,
      "loss": 0.8542,
      "step": 9737
    },
    {
      "epoch": 0.7869249883836037,
      "grad_norm": 2.806760311126709,
      "learning_rate": 6.917572680332965e-06,
      "loss": 0.911,
      "step": 9738
    },
    {
      "epoch": 0.7870057980969313,
      "grad_norm": 2.36431622505188,
      "learning_rate": 6.916968337675003e-06,
      "loss": 0.902,
      "step": 9739
    },
    {
      "epoch": 0.7870866078102587,
      "grad_norm": 2.8250107765197754,
      "learning_rate": 6.91636396218399e-06,
      "loss": 0.8955,
      "step": 9740
    },
    {
      "epoch": 0.7871674175235863,
      "grad_norm": 2.675884962081909,
      "learning_rate": 6.915759553870275e-06,
      "loss": 0.8415,
      "step": 9741
    },
    {
      "epoch": 0.7872482272369139,
      "grad_norm": 2.768265962600708,
      "learning_rate": 6.915155112744211e-06,
      "loss": 0.8819,
      "step": 9742
    },
    {
      "epoch": 0.7873290369502414,
      "grad_norm": 2.699328899383545,
      "learning_rate": 6.91455063881615e-06,
      "loss": 0.8715,
      "step": 9743
    },
    {
      "epoch": 0.787409846663569,
      "grad_norm": 2.888277292251587,
      "learning_rate": 6.913946132096447e-06,
      "loss": 0.9695,
      "step": 9744
    },
    {
      "epoch": 0.7874906563768965,
      "grad_norm": 2.8600375652313232,
      "learning_rate": 6.913341592595453e-06,
      "loss": 0.915,
      "step": 9745
    },
    {
      "epoch": 0.787571466090224,
      "grad_norm": 2.665877342224121,
      "learning_rate": 6.912737020323523e-06,
      "loss": 1.0342,
      "step": 9746
    },
    {
      "epoch": 0.7876522758035516,
      "grad_norm": 2.548896074295044,
      "learning_rate": 6.912132415291014e-06,
      "loss": 0.9424,
      "step": 9747
    },
    {
      "epoch": 0.7877330855168792,
      "grad_norm": 2.6893019676208496,
      "learning_rate": 6.91152777750828e-06,
      "loss": 0.7621,
      "step": 9748
    },
    {
      "epoch": 0.7878138952302066,
      "grad_norm": 2.669121026992798,
      "learning_rate": 6.910923106985678e-06,
      "loss": 0.8083,
      "step": 9749
    },
    {
      "epoch": 0.7878947049435342,
      "grad_norm": 2.5475196838378906,
      "learning_rate": 6.9103184037335615e-06,
      "loss": 0.8622,
      "step": 9750
    },
    {
      "epoch": 0.7879755146568618,
      "grad_norm": 2.509610652923584,
      "learning_rate": 6.90971366776229e-06,
      "loss": 0.8882,
      "step": 9751
    },
    {
      "epoch": 0.7880563243701892,
      "grad_norm": 3.3432157039642334,
      "learning_rate": 6.909108899082222e-06,
      "loss": 0.9586,
      "step": 9752
    },
    {
      "epoch": 0.7881371340835168,
      "grad_norm": 2.9079229831695557,
      "learning_rate": 6.908504097703713e-06,
      "loss": 0.8835,
      "step": 9753
    },
    {
      "epoch": 0.7882179437968444,
      "grad_norm": 2.9204115867614746,
      "learning_rate": 6.9078992636371246e-06,
      "loss": 1.0229,
      "step": 9754
    },
    {
      "epoch": 0.7882987535101719,
      "grad_norm": 2.441419839859009,
      "learning_rate": 6.907294396892815e-06,
      "loss": 0.8887,
      "step": 9755
    },
    {
      "epoch": 0.7883795632234994,
      "grad_norm": 3.024362087249756,
      "learning_rate": 6.9066894974811425e-06,
      "loss": 0.9457,
      "step": 9756
    },
    {
      "epoch": 0.788460372936827,
      "grad_norm": 2.8817200660705566,
      "learning_rate": 6.906084565412471e-06,
      "loss": 0.8425,
      "step": 9757
    },
    {
      "epoch": 0.7885411826501546,
      "grad_norm": 2.3224284648895264,
      "learning_rate": 6.905479600697158e-06,
      "loss": 0.9562,
      "step": 9758
    },
    {
      "epoch": 0.7886219923634821,
      "grad_norm": 2.697815179824829,
      "learning_rate": 6.9048746033455675e-06,
      "loss": 0.9303,
      "step": 9759
    },
    {
      "epoch": 0.7887028020768097,
      "grad_norm": 2.7465643882751465,
      "learning_rate": 6.904269573368061e-06,
      "loss": 0.9734,
      "step": 9760
    },
    {
      "epoch": 0.7887836117901372,
      "grad_norm": 3.1014504432678223,
      "learning_rate": 6.903664510775e-06,
      "loss": 0.9998,
      "step": 9761
    },
    {
      "epoch": 0.7888644215034647,
      "grad_norm": 2.543888807296753,
      "learning_rate": 6.90305941557675e-06,
      "loss": 0.8978,
      "step": 9762
    },
    {
      "epoch": 0.7889452312167923,
      "grad_norm": 3.0319859981536865,
      "learning_rate": 6.9024542877836735e-06,
      "loss": 0.9864,
      "step": 9763
    },
    {
      "epoch": 0.7890260409301199,
      "grad_norm": 3.4651505947113037,
      "learning_rate": 6.9018491274061325e-06,
      "loss": 1.0205,
      "step": 9764
    },
    {
      "epoch": 0.7891068506434473,
      "grad_norm": 2.6538500785827637,
      "learning_rate": 6.901243934454498e-06,
      "loss": 0.9339,
      "step": 9765
    },
    {
      "epoch": 0.7891876603567749,
      "grad_norm": 2.7711353302001953,
      "learning_rate": 6.9006387089391315e-06,
      "loss": 0.9507,
      "step": 9766
    },
    {
      "epoch": 0.7892684700701025,
      "grad_norm": 2.5237932205200195,
      "learning_rate": 6.900033450870398e-06,
      "loss": 0.9447,
      "step": 9767
    },
    {
      "epoch": 0.78934927978343,
      "grad_norm": 2.416130304336548,
      "learning_rate": 6.899428160258665e-06,
      "loss": 0.8435,
      "step": 9768
    },
    {
      "epoch": 0.7894300894967575,
      "grad_norm": 2.563319444656372,
      "learning_rate": 6.8988228371143025e-06,
      "loss": 0.8626,
      "step": 9769
    },
    {
      "epoch": 0.7895108992100851,
      "grad_norm": 2.9576029777526855,
      "learning_rate": 6.898217481447675e-06,
      "loss": 0.8971,
      "step": 9770
    },
    {
      "epoch": 0.7895917089234126,
      "grad_norm": 2.78456974029541,
      "learning_rate": 6.897612093269153e-06,
      "loss": 0.9585,
      "step": 9771
    },
    {
      "epoch": 0.7896725186367402,
      "grad_norm": 2.344691038131714,
      "learning_rate": 6.897006672589102e-06,
      "loss": 1.0789,
      "step": 9772
    },
    {
      "epoch": 0.7897533283500677,
      "grad_norm": 2.1248655319213867,
      "learning_rate": 6.896401219417896e-06,
      "loss": 0.9265,
      "step": 9773
    },
    {
      "epoch": 0.7898341380633952,
      "grad_norm": 2.875714063644409,
      "learning_rate": 6.8957957337659e-06,
      "loss": 0.9438,
      "step": 9774
    },
    {
      "epoch": 0.7899149477767228,
      "grad_norm": 3.0175061225891113,
      "learning_rate": 6.895190215643488e-06,
      "loss": 0.9157,
      "step": 9775
    },
    {
      "epoch": 0.7899957574900504,
      "grad_norm": 2.4655911922454834,
      "learning_rate": 6.89458466506103e-06,
      "loss": 1.0023,
      "step": 9776
    },
    {
      "epoch": 0.7900765672033778,
      "grad_norm": 2.8080813884735107,
      "learning_rate": 6.893979082028899e-06,
      "loss": 0.8888,
      "step": 9777
    },
    {
      "epoch": 0.7901573769167054,
      "grad_norm": 2.922720432281494,
      "learning_rate": 6.893373466557464e-06,
      "loss": 0.8943,
      "step": 9778
    },
    {
      "epoch": 0.790238186630033,
      "grad_norm": 2.578737258911133,
      "learning_rate": 6.892767818657101e-06,
      "loss": 0.9522,
      "step": 9779
    },
    {
      "epoch": 0.7903189963433604,
      "grad_norm": 2.356459379196167,
      "learning_rate": 6.892162138338181e-06,
      "loss": 0.8621,
      "step": 9780
    },
    {
      "epoch": 0.790399806056688,
      "grad_norm": 2.190770149230957,
      "learning_rate": 6.891556425611079e-06,
      "loss": 1.0324,
      "step": 9781
    },
    {
      "epoch": 0.7904806157700156,
      "grad_norm": 3.4735827445983887,
      "learning_rate": 6.890950680486169e-06,
      "loss": 0.8301,
      "step": 9782
    },
    {
      "epoch": 0.7905614254833431,
      "grad_norm": 3.018254041671753,
      "learning_rate": 6.8903449029738265e-06,
      "loss": 0.8631,
      "step": 9783
    },
    {
      "epoch": 0.7906422351966707,
      "grad_norm": 3.063438892364502,
      "learning_rate": 6.889739093084427e-06,
      "loss": 0.889,
      "step": 9784
    },
    {
      "epoch": 0.7907230449099982,
      "grad_norm": 3.142925262451172,
      "learning_rate": 6.889133250828346e-06,
      "loss": 0.7975,
      "step": 9785
    },
    {
      "epoch": 0.7908038546233257,
      "grad_norm": 2.408189058303833,
      "learning_rate": 6.888527376215959e-06,
      "loss": 0.8847,
      "step": 9786
    },
    {
      "epoch": 0.7908846643366533,
      "grad_norm": 2.9360907077789307,
      "learning_rate": 6.887921469257647e-06,
      "loss": 0.8699,
      "step": 9787
    },
    {
      "epoch": 0.7909654740499809,
      "grad_norm": 2.627872943878174,
      "learning_rate": 6.887315529963784e-06,
      "loss": 0.9903,
      "step": 9788
    },
    {
      "epoch": 0.7910462837633083,
      "grad_norm": 2.3716421127319336,
      "learning_rate": 6.886709558344748e-06,
      "loss": 0.8687,
      "step": 9789
    },
    {
      "epoch": 0.7911270934766359,
      "grad_norm": 2.794773578643799,
      "learning_rate": 6.886103554410921e-06,
      "loss": 1.019,
      "step": 9790
    },
    {
      "epoch": 0.7912079031899635,
      "grad_norm": 2.796485662460327,
      "learning_rate": 6.885497518172681e-06,
      "loss": 1.0813,
      "step": 9791
    },
    {
      "epoch": 0.791288712903291,
      "grad_norm": 2.507392168045044,
      "learning_rate": 6.884891449640407e-06,
      "loss": 0.9772,
      "step": 9792
    },
    {
      "epoch": 0.7913695226166185,
      "grad_norm": 2.811211109161377,
      "learning_rate": 6.88428534882448e-06,
      "loss": 0.898,
      "step": 9793
    },
    {
      "epoch": 0.7914503323299461,
      "grad_norm": 3.025217294692993,
      "learning_rate": 6.883679215735282e-06,
      "loss": 0.9232,
      "step": 9794
    },
    {
      "epoch": 0.7915311420432736,
      "grad_norm": 2.4453773498535156,
      "learning_rate": 6.883073050383193e-06,
      "loss": 0.9294,
      "step": 9795
    },
    {
      "epoch": 0.7916119517566012,
      "grad_norm": 2.9261538982391357,
      "learning_rate": 6.8824668527785954e-06,
      "loss": 0.9779,
      "step": 9796
    },
    {
      "epoch": 0.7916927614699287,
      "grad_norm": 2.2668650150299072,
      "learning_rate": 6.881860622931873e-06,
      "loss": 1.049,
      "step": 9797
    },
    {
      "epoch": 0.7917735711832562,
      "grad_norm": 2.7034614086151123,
      "learning_rate": 6.881254360853409e-06,
      "loss": 0.9288,
      "step": 9798
    },
    {
      "epoch": 0.7918543808965838,
      "grad_norm": 2.669090986251831,
      "learning_rate": 6.880648066553588e-06,
      "loss": 0.8742,
      "step": 9799
    },
    {
      "epoch": 0.7919351906099114,
      "grad_norm": 3.4627444744110107,
      "learning_rate": 6.88004174004279e-06,
      "loss": 0.8534,
      "step": 9800
    },
    {
      "epoch": 0.7920160003232388,
      "grad_norm": 2.5556793212890625,
      "learning_rate": 6.879435381331405e-06,
      "loss": 0.8797,
      "step": 9801
    },
    {
      "epoch": 0.7920968100365664,
      "grad_norm": 2.601912498474121,
      "learning_rate": 6.8788289904298155e-06,
      "loss": 0.8918,
      "step": 9802
    },
    {
      "epoch": 0.792177619749894,
      "grad_norm": 2.544543981552124,
      "learning_rate": 6.878222567348409e-06,
      "loss": 0.8968,
      "step": 9803
    },
    {
      "epoch": 0.7922584294632214,
      "grad_norm": 3.0917084217071533,
      "learning_rate": 6.8776161120975714e-06,
      "loss": 0.9293,
      "step": 9804
    },
    {
      "epoch": 0.792339239176549,
      "grad_norm": 2.291499376296997,
      "learning_rate": 6.877009624687691e-06,
      "loss": 0.8988,
      "step": 9805
    },
    {
      "epoch": 0.7924200488898766,
      "grad_norm": 2.528583526611328,
      "learning_rate": 6.8764031051291535e-06,
      "loss": 0.95,
      "step": 9806
    },
    {
      "epoch": 0.7925008586032041,
      "grad_norm": 2.412081241607666,
      "learning_rate": 6.875796553432349e-06,
      "loss": 0.9062,
      "step": 9807
    },
    {
      "epoch": 0.7925816683165317,
      "grad_norm": 2.9253103733062744,
      "learning_rate": 6.875189969607664e-06,
      "loss": 0.9501,
      "step": 9808
    },
    {
      "epoch": 0.7926624780298592,
      "grad_norm": 2.699721574783325,
      "learning_rate": 6.8745833536654895e-06,
      "loss": 1.0665,
      "step": 9809
    },
    {
      "epoch": 0.7927432877431867,
      "grad_norm": 2.3653342723846436,
      "learning_rate": 6.873976705616215e-06,
      "loss": 0.9201,
      "step": 9810
    },
    {
      "epoch": 0.7928240974565143,
      "grad_norm": 2.5907254219055176,
      "learning_rate": 6.873370025470232e-06,
      "loss": 0.955,
      "step": 9811
    },
    {
      "epoch": 0.7929049071698419,
      "grad_norm": 2.5645127296447754,
      "learning_rate": 6.872763313237929e-06,
      "loss": 0.8921,
      "step": 9812
    },
    {
      "epoch": 0.7929857168831693,
      "grad_norm": 2.475836992263794,
      "learning_rate": 6.8721565689297e-06,
      "loss": 0.9178,
      "step": 9813
    },
    {
      "epoch": 0.7930665265964969,
      "grad_norm": 2.7445790767669678,
      "learning_rate": 6.871549792555935e-06,
      "loss": 0.9757,
      "step": 9814
    },
    {
      "epoch": 0.7931473363098245,
      "grad_norm": 2.585686445236206,
      "learning_rate": 6.870942984127029e-06,
      "loss": 0.8806,
      "step": 9815
    },
    {
      "epoch": 0.793228146023152,
      "grad_norm": 2.713639974594116,
      "learning_rate": 6.870336143653372e-06,
      "loss": 0.9453,
      "step": 9816
    },
    {
      "epoch": 0.7933089557364795,
      "grad_norm": 2.8230977058410645,
      "learning_rate": 6.86972927114536e-06,
      "loss": 0.9789,
      "step": 9817
    },
    {
      "epoch": 0.7933897654498071,
      "grad_norm": 3.0862629413604736,
      "learning_rate": 6.869122366613387e-06,
      "loss": 0.9532,
      "step": 9818
    },
    {
      "epoch": 0.7934705751631346,
      "grad_norm": 2.866936445236206,
      "learning_rate": 6.868515430067848e-06,
      "loss": 0.9657,
      "step": 9819
    },
    {
      "epoch": 0.7935513848764622,
      "grad_norm": 2.8744053840637207,
      "learning_rate": 6.867908461519138e-06,
      "loss": 0.9024,
      "step": 9820
    },
    {
      "epoch": 0.7936321945897897,
      "grad_norm": 2.412216901779175,
      "learning_rate": 6.867301460977652e-06,
      "loss": 0.9955,
      "step": 9821
    },
    {
      "epoch": 0.7937130043031172,
      "grad_norm": 2.5999553203582764,
      "learning_rate": 6.86669442845379e-06,
      "loss": 1.0087,
      "step": 9822
    },
    {
      "epoch": 0.7937938140164448,
      "grad_norm": 2.983273506164551,
      "learning_rate": 6.866087363957943e-06,
      "loss": 0.9415,
      "step": 9823
    },
    {
      "epoch": 0.7938746237297724,
      "grad_norm": 2.738760232925415,
      "learning_rate": 6.865480267500514e-06,
      "loss": 0.9421,
      "step": 9824
    },
    {
      "epoch": 0.7939554334430998,
      "grad_norm": 2.6678097248077393,
      "learning_rate": 6.864873139091897e-06,
      "loss": 0.9528,
      "step": 9825
    },
    {
      "epoch": 0.7940362431564274,
      "grad_norm": 2.8385188579559326,
      "learning_rate": 6.864265978742494e-06,
      "loss": 0.9275,
      "step": 9826
    },
    {
      "epoch": 0.794117052869755,
      "grad_norm": 2.6046605110168457,
      "learning_rate": 6.863658786462702e-06,
      "loss": 0.9594,
      "step": 9827
    },
    {
      "epoch": 0.7941978625830824,
      "grad_norm": 2.5831477642059326,
      "learning_rate": 6.863051562262922e-06,
      "loss": 0.9989,
      "step": 9828
    },
    {
      "epoch": 0.79427867229641,
      "grad_norm": 2.5561978816986084,
      "learning_rate": 6.862444306153555e-06,
      "loss": 0.8679,
      "step": 9829
    },
    {
      "epoch": 0.7943594820097376,
      "grad_norm": 3.213993549346924,
      "learning_rate": 6.861837018145e-06,
      "loss": 0.9804,
      "step": 9830
    },
    {
      "epoch": 0.7944402917230651,
      "grad_norm": 2.7921385765075684,
      "learning_rate": 6.861229698247658e-06,
      "loss": 0.8718,
      "step": 9831
    },
    {
      "epoch": 0.7945211014363927,
      "grad_norm": 2.426623582839966,
      "learning_rate": 6.860622346471933e-06,
      "loss": 0.873,
      "step": 9832
    },
    {
      "epoch": 0.7946019111497202,
      "grad_norm": 2.5576171875,
      "learning_rate": 6.8600149628282265e-06,
      "loss": 0.837,
      "step": 9833
    },
    {
      "epoch": 0.7946827208630477,
      "grad_norm": 2.269954204559326,
      "learning_rate": 6.859407547326941e-06,
      "loss": 0.96,
      "step": 9834
    },
    {
      "epoch": 0.7947635305763753,
      "grad_norm": 2.6326749324798584,
      "learning_rate": 6.85880009997848e-06,
      "loss": 0.9024,
      "step": 9835
    },
    {
      "epoch": 0.7948443402897029,
      "grad_norm": 2.807593822479248,
      "learning_rate": 6.85819262079325e-06,
      "loss": 0.9605,
      "step": 9836
    },
    {
      "epoch": 0.7949251500030303,
      "grad_norm": 2.447377920150757,
      "learning_rate": 6.857585109781652e-06,
      "loss": 0.935,
      "step": 9837
    },
    {
      "epoch": 0.7950059597163579,
      "grad_norm": 2.6915740966796875,
      "learning_rate": 6.856977566954095e-06,
      "loss": 0.886,
      "step": 9838
    },
    {
      "epoch": 0.7950867694296855,
      "grad_norm": 2.9169421195983887,
      "learning_rate": 6.8563699923209794e-06,
      "loss": 0.941,
      "step": 9839
    },
    {
      "epoch": 0.795167579143013,
      "grad_norm": 2.1841347217559814,
      "learning_rate": 6.855762385892718e-06,
      "loss": 0.9259,
      "step": 9840
    },
    {
      "epoch": 0.7952483888563405,
      "grad_norm": 2.310147762298584,
      "learning_rate": 6.855154747679713e-06,
      "loss": 0.8719,
      "step": 9841
    },
    {
      "epoch": 0.7953291985696681,
      "grad_norm": 2.3946635723114014,
      "learning_rate": 6.854547077692374e-06,
      "loss": 0.9772,
      "step": 9842
    },
    {
      "epoch": 0.7954100082829956,
      "grad_norm": 2.599512815475464,
      "learning_rate": 6.853939375941108e-06,
      "loss": 0.9794,
      "step": 9843
    },
    {
      "epoch": 0.7954908179963232,
      "grad_norm": 2.9427289962768555,
      "learning_rate": 6.853331642436325e-06,
      "loss": 0.8381,
      "step": 9844
    },
    {
      "epoch": 0.7955716277096507,
      "grad_norm": 2.641378164291382,
      "learning_rate": 6.85272387718843e-06,
      "loss": 0.9253,
      "step": 9845
    },
    {
      "epoch": 0.7956524374229782,
      "grad_norm": 2.788924217224121,
      "learning_rate": 6.852116080207837e-06,
      "loss": 1.0463,
      "step": 9846
    },
    {
      "epoch": 0.7957332471363058,
      "grad_norm": 3.3630847930908203,
      "learning_rate": 6.8515082515049535e-06,
      "loss": 0.9495,
      "step": 9847
    },
    {
      "epoch": 0.7958140568496334,
      "grad_norm": 2.465693712234497,
      "learning_rate": 6.850900391090191e-06,
      "loss": 0.8036,
      "step": 9848
    },
    {
      "epoch": 0.7958948665629608,
      "grad_norm": 2.459573745727539,
      "learning_rate": 6.850292498973962e-06,
      "loss": 0.956,
      "step": 9849
    },
    {
      "epoch": 0.7959756762762884,
      "grad_norm": 2.578242778778076,
      "learning_rate": 6.849684575166676e-06,
      "loss": 0.966,
      "step": 9850
    },
    {
      "epoch": 0.796056485989616,
      "grad_norm": 2.8725247383117676,
      "learning_rate": 6.849076619678745e-06,
      "loss": 0.9189,
      "step": 9851
    },
    {
      "epoch": 0.7961372957029434,
      "grad_norm": 2.562953472137451,
      "learning_rate": 6.848468632520585e-06,
      "loss": 0.9005,
      "step": 9852
    },
    {
      "epoch": 0.796218105416271,
      "grad_norm": 2.785662889480591,
      "learning_rate": 6.847860613702605e-06,
      "loss": 1.0629,
      "step": 9853
    },
    {
      "epoch": 0.7962989151295986,
      "grad_norm": 2.6662254333496094,
      "learning_rate": 6.847252563235224e-06,
      "loss": 0.9215,
      "step": 9854
    },
    {
      "epoch": 0.7963797248429261,
      "grad_norm": 2.75342059135437,
      "learning_rate": 6.846644481128852e-06,
      "loss": 1.0127,
      "step": 9855
    },
    {
      "epoch": 0.7964605345562537,
      "grad_norm": 2.9609148502349854,
      "learning_rate": 6.8460363673939055e-06,
      "loss": 0.9866,
      "step": 9856
    },
    {
      "epoch": 0.7965413442695812,
      "grad_norm": 3.680875539779663,
      "learning_rate": 6.8454282220408005e-06,
      "loss": 0.9998,
      "step": 9857
    },
    {
      "epoch": 0.7966221539829087,
      "grad_norm": 2.6819679737091064,
      "learning_rate": 6.844820045079954e-06,
      "loss": 1.005,
      "step": 9858
    },
    {
      "epoch": 0.7967029636962363,
      "grad_norm": 2.4925944805145264,
      "learning_rate": 6.844211836521779e-06,
      "loss": 0.9933,
      "step": 9859
    },
    {
      "epoch": 0.7967837734095639,
      "grad_norm": 2.279554843902588,
      "learning_rate": 6.843603596376697e-06,
      "loss": 0.8336,
      "step": 9860
    },
    {
      "epoch": 0.7968645831228913,
      "grad_norm": 2.840665817260742,
      "learning_rate": 6.842995324655123e-06,
      "loss": 0.8977,
      "step": 9861
    },
    {
      "epoch": 0.7969453928362189,
      "grad_norm": 2.678210496902466,
      "learning_rate": 6.842387021367476e-06,
      "loss": 0.8747,
      "step": 9862
    },
    {
      "epoch": 0.7970262025495465,
      "grad_norm": 2.3635852336883545,
      "learning_rate": 6.841778686524174e-06,
      "loss": 1.137,
      "step": 9863
    },
    {
      "epoch": 0.797107012262874,
      "grad_norm": 2.741825819015503,
      "learning_rate": 6.8411703201356385e-06,
      "loss": 1.0271,
      "step": 9864
    },
    {
      "epoch": 0.7971878219762015,
      "grad_norm": 2.799586772918701,
      "learning_rate": 6.840561922212285e-06,
      "loss": 1.0293,
      "step": 9865
    },
    {
      "epoch": 0.7972686316895291,
      "grad_norm": 2.8495535850524902,
      "learning_rate": 6.8399534927645396e-06,
      "loss": 0.9574,
      "step": 9866
    },
    {
      "epoch": 0.7973494414028566,
      "grad_norm": 2.7940566539764404,
      "learning_rate": 6.839345031802819e-06,
      "loss": 0.9242,
      "step": 9867
    },
    {
      "epoch": 0.7974302511161842,
      "grad_norm": 2.7938547134399414,
      "learning_rate": 6.838736539337547e-06,
      "loss": 0.9976,
      "step": 9868
    },
    {
      "epoch": 0.7975110608295117,
      "grad_norm": 2.463744640350342,
      "learning_rate": 6.838128015379144e-06,
      "loss": 0.999,
      "step": 9869
    },
    {
      "epoch": 0.7975918705428392,
      "grad_norm": 2.3955326080322266,
      "learning_rate": 6.837519459938034e-06,
      "loss": 1.0032,
      "step": 9870
    },
    {
      "epoch": 0.7976726802561668,
      "grad_norm": 2.0644900798797607,
      "learning_rate": 6.836910873024637e-06,
      "loss": 0.9661,
      "step": 9871
    },
    {
      "epoch": 0.7977534899694944,
      "grad_norm": 3.0256295204162598,
      "learning_rate": 6.836302254649382e-06,
      "loss": 0.9018,
      "step": 9872
    },
    {
      "epoch": 0.7978342996828218,
      "grad_norm": 2.729816198348999,
      "learning_rate": 6.835693604822687e-06,
      "loss": 0.8376,
      "step": 9873
    },
    {
      "epoch": 0.7979151093961494,
      "grad_norm": 2.375227451324463,
      "learning_rate": 6.835084923554982e-06,
      "loss": 1.035,
      "step": 9874
    },
    {
      "epoch": 0.797995919109477,
      "grad_norm": 2.579192876815796,
      "learning_rate": 6.834476210856689e-06,
      "loss": 1.0178,
      "step": 9875
    },
    {
      "epoch": 0.7980767288228044,
      "grad_norm": 2.5040221214294434,
      "learning_rate": 6.833867466738235e-06,
      "loss": 1.0459,
      "step": 9876
    },
    {
      "epoch": 0.798157538536132,
      "grad_norm": 2.7196125984191895,
      "learning_rate": 6.833258691210048e-06,
      "loss": 1.0056,
      "step": 9877
    },
    {
      "epoch": 0.7982383482494596,
      "grad_norm": 2.4624030590057373,
      "learning_rate": 6.83264988428255e-06,
      "loss": 0.9431,
      "step": 9878
    },
    {
      "epoch": 0.7983191579627871,
      "grad_norm": 2.6063098907470703,
      "learning_rate": 6.832041045966174e-06,
      "loss": 0.8946,
      "step": 9879
    },
    {
      "epoch": 0.7983999676761147,
      "grad_norm": 2.7179946899414062,
      "learning_rate": 6.831432176271345e-06,
      "loss": 0.9484,
      "step": 9880
    },
    {
      "epoch": 0.7984807773894422,
      "grad_norm": 2.3139700889587402,
      "learning_rate": 6.830823275208489e-06,
      "loss": 0.856,
      "step": 9881
    },
    {
      "epoch": 0.7985615871027697,
      "grad_norm": 2.4322259426116943,
      "learning_rate": 6.8302143427880405e-06,
      "loss": 0.9994,
      "step": 9882
    },
    {
      "epoch": 0.7986423968160973,
      "grad_norm": 2.5613677501678467,
      "learning_rate": 6.829605379020425e-06,
      "loss": 0.8508,
      "step": 9883
    },
    {
      "epoch": 0.7987232065294249,
      "grad_norm": 2.6813414096832275,
      "learning_rate": 6.828996383916076e-06,
      "loss": 0.904,
      "step": 9884
    },
    {
      "epoch": 0.7988040162427523,
      "grad_norm": 2.4728004932403564,
      "learning_rate": 6.82838735748542e-06,
      "loss": 0.9978,
      "step": 9885
    },
    {
      "epoch": 0.7988848259560799,
      "grad_norm": 2.4219539165496826,
      "learning_rate": 6.827778299738891e-06,
      "loss": 0.9248,
      "step": 9886
    },
    {
      "epoch": 0.7989656356694075,
      "grad_norm": 2.4130876064300537,
      "learning_rate": 6.8271692106869195e-06,
      "loss": 0.8378,
      "step": 9887
    },
    {
      "epoch": 0.7990464453827351,
      "grad_norm": 2.6504530906677246,
      "learning_rate": 6.826560090339939e-06,
      "loss": 0.8864,
      "step": 9888
    },
    {
      "epoch": 0.7991272550960625,
      "grad_norm": 2.934321403503418,
      "learning_rate": 6.825950938708381e-06,
      "loss": 1.0809,
      "step": 9889
    },
    {
      "epoch": 0.7992080648093901,
      "grad_norm": 2.4860029220581055,
      "learning_rate": 6.825341755802679e-06,
      "loss": 0.94,
      "step": 9890
    },
    {
      "epoch": 0.7992888745227177,
      "grad_norm": 2.6035618782043457,
      "learning_rate": 6.8247325416332675e-06,
      "loss": 0.9673,
      "step": 9891
    },
    {
      "epoch": 0.7993696842360452,
      "grad_norm": 2.690329074859619,
      "learning_rate": 6.824123296210579e-06,
      "loss": 0.9245,
      "step": 9892
    },
    {
      "epoch": 0.7994504939493727,
      "grad_norm": 2.507843255996704,
      "learning_rate": 6.823514019545052e-06,
      "loss": 0.9987,
      "step": 9893
    },
    {
      "epoch": 0.7995313036627003,
      "grad_norm": 2.463536024093628,
      "learning_rate": 6.822904711647118e-06,
      "loss": 1.1466,
      "step": 9894
    },
    {
      "epoch": 0.7996121133760278,
      "grad_norm": 2.800693988800049,
      "learning_rate": 6.822295372527216e-06,
      "loss": 0.8854,
      "step": 9895
    },
    {
      "epoch": 0.7996929230893554,
      "grad_norm": 2.625532865524292,
      "learning_rate": 6.82168600219578e-06,
      "loss": 0.8947,
      "step": 9896
    },
    {
      "epoch": 0.7997737328026829,
      "grad_norm": 2.919964075088501,
      "learning_rate": 6.82107660066325e-06,
      "loss": 0.9991,
      "step": 9897
    },
    {
      "epoch": 0.7998545425160104,
      "grad_norm": 2.7522058486938477,
      "learning_rate": 6.82046716794006e-06,
      "loss": 0.9677,
      "step": 9898
    },
    {
      "epoch": 0.799935352229338,
      "grad_norm": 2.916342258453369,
      "learning_rate": 6.819857704036652e-06,
      "loss": 0.9301,
      "step": 9899
    },
    {
      "epoch": 0.8000161619426656,
      "grad_norm": 2.8729188442230225,
      "learning_rate": 6.819248208963461e-06,
      "loss": 0.9065,
      "step": 9900
    },
    {
      "epoch": 0.800096971655993,
      "grad_norm": 2.786853551864624,
      "learning_rate": 6.818638682730929e-06,
      "loss": 0.9544,
      "step": 9901
    },
    {
      "epoch": 0.8001777813693206,
      "grad_norm": 2.7986268997192383,
      "learning_rate": 6.818029125349494e-06,
      "loss": 0.9148,
      "step": 9902
    },
    {
      "epoch": 0.8002585910826482,
      "grad_norm": 2.4643256664276123,
      "learning_rate": 6.817419536829597e-06,
      "loss": 1.0353,
      "step": 9903
    },
    {
      "epoch": 0.8003394007959757,
      "grad_norm": 2.592578411102295,
      "learning_rate": 6.816809917181677e-06,
      "loss": 0.9834,
      "step": 9904
    },
    {
      "epoch": 0.8004202105093032,
      "grad_norm": 2.952615261077881,
      "learning_rate": 6.8162002664161794e-06,
      "loss": 0.9166,
      "step": 9905
    },
    {
      "epoch": 0.8005010202226308,
      "grad_norm": 3.0346031188964844,
      "learning_rate": 6.815590584543542e-06,
      "loss": 0.9002,
      "step": 9906
    },
    {
      "epoch": 0.8005818299359583,
      "grad_norm": 2.2955641746520996,
      "learning_rate": 6.81498087157421e-06,
      "loss": 0.8635,
      "step": 9907
    },
    {
      "epoch": 0.8006626396492859,
      "grad_norm": 2.3769054412841797,
      "learning_rate": 6.814371127518624e-06,
      "loss": 0.9364,
      "step": 9908
    },
    {
      "epoch": 0.8007434493626134,
      "grad_norm": 2.8720016479492188,
      "learning_rate": 6.813761352387229e-06,
      "loss": 0.9804,
      "step": 9909
    },
    {
      "epoch": 0.8008242590759409,
      "grad_norm": 2.956225872039795,
      "learning_rate": 6.8131515461904685e-06,
      "loss": 0.9734,
      "step": 9910
    },
    {
      "epoch": 0.8009050687892685,
      "grad_norm": 2.920043468475342,
      "learning_rate": 6.812541708938787e-06,
      "loss": 1.0127,
      "step": 9911
    },
    {
      "epoch": 0.8009858785025961,
      "grad_norm": 2.2908806800842285,
      "learning_rate": 6.811931840642628e-06,
      "loss": 1.006,
      "step": 9912
    },
    {
      "epoch": 0.8010666882159235,
      "grad_norm": 2.7652716636657715,
      "learning_rate": 6.811321941312441e-06,
      "loss": 0.9693,
      "step": 9913
    },
    {
      "epoch": 0.8011474979292511,
      "grad_norm": 2.4583494663238525,
      "learning_rate": 6.8107120109586685e-06,
      "loss": 0.911,
      "step": 9914
    },
    {
      "epoch": 0.8012283076425787,
      "grad_norm": 2.4979212284088135,
      "learning_rate": 6.810102049591759e-06,
      "loss": 1.0718,
      "step": 9915
    },
    {
      "epoch": 0.8013091173559062,
      "grad_norm": 2.608715295791626,
      "learning_rate": 6.809492057222158e-06,
      "loss": 0.9271,
      "step": 9916
    },
    {
      "epoch": 0.8013899270692337,
      "grad_norm": 2.429481267929077,
      "learning_rate": 6.808882033860316e-06,
      "loss": 0.9787,
      "step": 9917
    },
    {
      "epoch": 0.8014707367825613,
      "grad_norm": 2.6573383808135986,
      "learning_rate": 6.808271979516677e-06,
      "loss": 1.0015,
      "step": 9918
    },
    {
      "epoch": 0.8015515464958888,
      "grad_norm": 2.55149245262146,
      "learning_rate": 6.807661894201695e-06,
      "loss": 0.9501,
      "step": 9919
    },
    {
      "epoch": 0.8016323562092164,
      "grad_norm": 2.7874369621276855,
      "learning_rate": 6.807051777925812e-06,
      "loss": 0.8918,
      "step": 9920
    },
    {
      "epoch": 0.8017131659225439,
      "grad_norm": 3.394378662109375,
      "learning_rate": 6.806441630699488e-06,
      "loss": 0.9506,
      "step": 9921
    },
    {
      "epoch": 0.8017939756358714,
      "grad_norm": 2.581477642059326,
      "learning_rate": 6.805831452533165e-06,
      "loss": 0.9971,
      "step": 9922
    },
    {
      "epoch": 0.801874785349199,
      "grad_norm": 2.5167365074157715,
      "learning_rate": 6.805221243437297e-06,
      "loss": 0.8361,
      "step": 9923
    },
    {
      "epoch": 0.8019555950625266,
      "grad_norm": 3.6850173473358154,
      "learning_rate": 6.804611003422333e-06,
      "loss": 1.0683,
      "step": 9924
    },
    {
      "epoch": 0.802036404775854,
      "grad_norm": 2.8270466327667236,
      "learning_rate": 6.80400073249873e-06,
      "loss": 0.9289,
      "step": 9925
    },
    {
      "epoch": 0.8021172144891816,
      "grad_norm": 2.3623886108398438,
      "learning_rate": 6.803390430676935e-06,
      "loss": 0.9429,
      "step": 9926
    },
    {
      "epoch": 0.8021980242025092,
      "grad_norm": 2.547598361968994,
      "learning_rate": 6.802780097967405e-06,
      "loss": 0.9129,
      "step": 9927
    },
    {
      "epoch": 0.8022788339158367,
      "grad_norm": 2.7918617725372314,
      "learning_rate": 6.802169734380592e-06,
      "loss": 0.9014,
      "step": 9928
    },
    {
      "epoch": 0.8023596436291642,
      "grad_norm": 2.6616594791412354,
      "learning_rate": 6.801559339926948e-06,
      "loss": 0.9323,
      "step": 9929
    },
    {
      "epoch": 0.8024404533424918,
      "grad_norm": 2.622589111328125,
      "learning_rate": 6.800948914616932e-06,
      "loss": 0.942,
      "step": 9930
    },
    {
      "epoch": 0.8025212630558193,
      "grad_norm": 2.566311836242676,
      "learning_rate": 6.8003384584609954e-06,
      "loss": 0.9182,
      "step": 9931
    },
    {
      "epoch": 0.8026020727691469,
      "grad_norm": 3.059619188308716,
      "learning_rate": 6.7997279714695945e-06,
      "loss": 0.9075,
      "step": 9932
    },
    {
      "epoch": 0.8026828824824744,
      "grad_norm": 2.622683048248291,
      "learning_rate": 6.799117453653188e-06,
      "loss": 0.9596,
      "step": 9933
    },
    {
      "epoch": 0.8027636921958019,
      "grad_norm": 2.312102794647217,
      "learning_rate": 6.798506905022229e-06,
      "loss": 0.9342,
      "step": 9934
    },
    {
      "epoch": 0.8028445019091295,
      "grad_norm": 2.541771411895752,
      "learning_rate": 6.7978963255871775e-06,
      "loss": 1.0073,
      "step": 9935
    },
    {
      "epoch": 0.8029253116224571,
      "grad_norm": 2.8717281818389893,
      "learning_rate": 6.797285715358491e-06,
      "loss": 0.9759,
      "step": 9936
    },
    {
      "epoch": 0.8030061213357845,
      "grad_norm": 2.7421064376831055,
      "learning_rate": 6.796675074346625e-06,
      "loss": 0.8878,
      "step": 9937
    },
    {
      "epoch": 0.8030869310491121,
      "grad_norm": 2.3162598609924316,
      "learning_rate": 6.7960644025620405e-06,
      "loss": 1.0536,
      "step": 9938
    },
    {
      "epoch": 0.8031677407624397,
      "grad_norm": 2.7054810523986816,
      "learning_rate": 6.795453700015198e-06,
      "loss": 1.0363,
      "step": 9939
    },
    {
      "epoch": 0.8032485504757672,
      "grad_norm": 2.5268211364746094,
      "learning_rate": 6.794842966716554e-06,
      "loss": 0.8561,
      "step": 9940
    },
    {
      "epoch": 0.8033293601890947,
      "grad_norm": 2.5131895542144775,
      "learning_rate": 6.7942322026765725e-06,
      "loss": 0.8584,
      "step": 9941
    },
    {
      "epoch": 0.8034101699024223,
      "grad_norm": 2.794346332550049,
      "learning_rate": 6.793621407905713e-06,
      "loss": 0.9088,
      "step": 9942
    },
    {
      "epoch": 0.8034909796157498,
      "grad_norm": 2.537999153137207,
      "learning_rate": 6.793010582414437e-06,
      "loss": 1.0139,
      "step": 9943
    },
    {
      "epoch": 0.8035717893290774,
      "grad_norm": 2.5075199604034424,
      "learning_rate": 6.792399726213205e-06,
      "loss": 0.9189,
      "step": 9944
    },
    {
      "epoch": 0.8036525990424049,
      "grad_norm": 2.7178080081939697,
      "learning_rate": 6.791788839312481e-06,
      "loss": 0.8522,
      "step": 9945
    },
    {
      "epoch": 0.8037334087557324,
      "grad_norm": 2.673673391342163,
      "learning_rate": 6.791177921722727e-06,
      "loss": 0.9387,
      "step": 9946
    },
    {
      "epoch": 0.80381421846906,
      "grad_norm": 2.8458967208862305,
      "learning_rate": 6.790566973454409e-06,
      "loss": 0.89,
      "step": 9947
    },
    {
      "epoch": 0.8038950281823876,
      "grad_norm": 2.802276611328125,
      "learning_rate": 6.789955994517987e-06,
      "loss": 1.0193,
      "step": 9948
    },
    {
      "epoch": 0.803975837895715,
      "grad_norm": 2.669790029525757,
      "learning_rate": 6.789344984923931e-06,
      "loss": 0.9449,
      "step": 9949
    },
    {
      "epoch": 0.8040566476090426,
      "grad_norm": 2.5872020721435547,
      "learning_rate": 6.788733944682702e-06,
      "loss": 0.956,
      "step": 9950
    },
    {
      "epoch": 0.8041374573223702,
      "grad_norm": 2.780992269515991,
      "learning_rate": 6.788122873804766e-06,
      "loss": 0.948,
      "step": 9951
    },
    {
      "epoch": 0.8042182670356977,
      "grad_norm": 2.6253044605255127,
      "learning_rate": 6.787511772300589e-06,
      "loss": 0.9695,
      "step": 9952
    },
    {
      "epoch": 0.8042990767490252,
      "grad_norm": 3.0298964977264404,
      "learning_rate": 6.78690064018064e-06,
      "loss": 0.8866,
      "step": 9953
    },
    {
      "epoch": 0.8043798864623528,
      "grad_norm": 2.735344409942627,
      "learning_rate": 6.786289477455385e-06,
      "loss": 0.8452,
      "step": 9954
    },
    {
      "epoch": 0.8044606961756803,
      "grad_norm": 2.6249313354492188,
      "learning_rate": 6.785678284135291e-06,
      "loss": 0.8771,
      "step": 9955
    },
    {
      "epoch": 0.8045415058890079,
      "grad_norm": 2.6650261878967285,
      "learning_rate": 6.7850670602308275e-06,
      "loss": 1.0703,
      "step": 9956
    },
    {
      "epoch": 0.8046223156023354,
      "grad_norm": 2.552743434906006,
      "learning_rate": 6.784455805752462e-06,
      "loss": 0.9084,
      "step": 9957
    },
    {
      "epoch": 0.8047031253156629,
      "grad_norm": 2.681318521499634,
      "learning_rate": 6.783844520710664e-06,
      "loss": 1.043,
      "step": 9958
    },
    {
      "epoch": 0.8047839350289905,
      "grad_norm": 2.7370269298553467,
      "learning_rate": 6.783233205115904e-06,
      "loss": 0.9245,
      "step": 9959
    },
    {
      "epoch": 0.8048647447423181,
      "grad_norm": 2.591618299484253,
      "learning_rate": 6.782621858978653e-06,
      "loss": 0.7968,
      "step": 9960
    },
    {
      "epoch": 0.8049455544556455,
      "grad_norm": 2.88142728805542,
      "learning_rate": 6.78201048230938e-06,
      "loss": 0.9588,
      "step": 9961
    },
    {
      "epoch": 0.8050263641689731,
      "grad_norm": 2.8691976070404053,
      "learning_rate": 6.7813990751185585e-06,
      "loss": 0.9014,
      "step": 9962
    },
    {
      "epoch": 0.8051071738823007,
      "grad_norm": 2.421128988265991,
      "learning_rate": 6.780787637416659e-06,
      "loss": 0.9896,
      "step": 9963
    },
    {
      "epoch": 0.8051879835956282,
      "grad_norm": 2.5189719200134277,
      "learning_rate": 6.780176169214155e-06,
      "loss": 0.9405,
      "step": 9964
    },
    {
      "epoch": 0.8052687933089557,
      "grad_norm": 2.573498249053955,
      "learning_rate": 6.7795646705215176e-06,
      "loss": 0.8544,
      "step": 9965
    },
    {
      "epoch": 0.8053496030222833,
      "grad_norm": 2.145843267440796,
      "learning_rate": 6.778953141349222e-06,
      "loss": 0.9849,
      "step": 9966
    },
    {
      "epoch": 0.8054304127356108,
      "grad_norm": 2.535832166671753,
      "learning_rate": 6.778341581707742e-06,
      "loss": 0.9573,
      "step": 9967
    },
    {
      "epoch": 0.8055112224489384,
      "grad_norm": 3.017761468887329,
      "learning_rate": 6.777729991607551e-06,
      "loss": 0.9537,
      "step": 9968
    },
    {
      "epoch": 0.8055920321622659,
      "grad_norm": 2.8418562412261963,
      "learning_rate": 6.777118371059126e-06,
      "loss": 0.9567,
      "step": 9969
    },
    {
      "epoch": 0.8056728418755934,
      "grad_norm": 2.78224778175354,
      "learning_rate": 6.7765067200729415e-06,
      "loss": 0.9891,
      "step": 9970
    },
    {
      "epoch": 0.805753651588921,
      "grad_norm": 2.618445634841919,
      "learning_rate": 6.7758950386594725e-06,
      "loss": 0.8203,
      "step": 9971
    },
    {
      "epoch": 0.8058344613022486,
      "grad_norm": 2.5080177783966064,
      "learning_rate": 6.775283326829199e-06,
      "loss": 0.7917,
      "step": 9972
    },
    {
      "epoch": 0.805915271015576,
      "grad_norm": 2.6570470333099365,
      "learning_rate": 6.7746715845925935e-06,
      "loss": 0.9759,
      "step": 9973
    },
    {
      "epoch": 0.8059960807289036,
      "grad_norm": 2.826026201248169,
      "learning_rate": 6.7740598119601365e-06,
      "loss": 0.8451,
      "step": 9974
    },
    {
      "epoch": 0.8060768904422312,
      "grad_norm": 2.806075096130371,
      "learning_rate": 6.773448008942307e-06,
      "loss": 0.9144,
      "step": 9975
    },
    {
      "epoch": 0.8061577001555587,
      "grad_norm": 2.6879494190216064,
      "learning_rate": 6.772836175549582e-06,
      "loss": 0.8722,
      "step": 9976
    },
    {
      "epoch": 0.8062385098688862,
      "grad_norm": 2.495375633239746,
      "learning_rate": 6.77222431179244e-06,
      "loss": 0.9778,
      "step": 9977
    },
    {
      "epoch": 0.8063193195822138,
      "grad_norm": 2.7937846183776855,
      "learning_rate": 6.7716124176813645e-06,
      "loss": 0.8248,
      "step": 9978
    },
    {
      "epoch": 0.8064001292955413,
      "grad_norm": 2.7677974700927734,
      "learning_rate": 6.771000493226831e-06,
      "loss": 0.9429,
      "step": 9979
    },
    {
      "epoch": 0.8064809390088689,
      "grad_norm": 2.934624195098877,
      "learning_rate": 6.770388538439324e-06,
      "loss": 0.8823,
      "step": 9980
    },
    {
      "epoch": 0.8065617487221964,
      "grad_norm": 3.1143798828125,
      "learning_rate": 6.769776553329322e-06,
      "loss": 0.9946,
      "step": 9981
    },
    {
      "epoch": 0.8066425584355239,
      "grad_norm": 2.7064270973205566,
      "learning_rate": 6.76916453790731e-06,
      "loss": 0.9639,
      "step": 9982
    },
    {
      "epoch": 0.8067233681488515,
      "grad_norm": 2.8486764430999756,
      "learning_rate": 6.768552492183768e-06,
      "loss": 0.9309,
      "step": 9983
    },
    {
      "epoch": 0.8068041778621791,
      "grad_norm": 2.335663080215454,
      "learning_rate": 6.767940416169179e-06,
      "loss": 0.9013,
      "step": 9984
    },
    {
      "epoch": 0.8068849875755065,
      "grad_norm": 2.4963512420654297,
      "learning_rate": 6.767328309874026e-06,
      "loss": 1.0221,
      "step": 9985
    },
    {
      "epoch": 0.8069657972888341,
      "grad_norm": 2.660006523132324,
      "learning_rate": 6.766716173308795e-06,
      "loss": 0.9881,
      "step": 9986
    },
    {
      "epoch": 0.8070466070021617,
      "grad_norm": 2.5680716037750244,
      "learning_rate": 6.766104006483968e-06,
      "loss": 0.9788,
      "step": 9987
    },
    {
      "epoch": 0.8071274167154892,
      "grad_norm": 3.0067737102508545,
      "learning_rate": 6.765491809410032e-06,
      "loss": 1.0068,
      "step": 9988
    },
    {
      "epoch": 0.8072082264288167,
      "grad_norm": 2.735895872116089,
      "learning_rate": 6.764879582097472e-06,
      "loss": 1.0031,
      "step": 9989
    },
    {
      "epoch": 0.8072890361421443,
      "grad_norm": 2.764328718185425,
      "learning_rate": 6.764267324556773e-06,
      "loss": 0.9728,
      "step": 9990
    },
    {
      "epoch": 0.8073698458554718,
      "grad_norm": 3.4726850986480713,
      "learning_rate": 6.763655036798421e-06,
      "loss": 1.14,
      "step": 9991
    },
    {
      "epoch": 0.8074506555687994,
      "grad_norm": 3.1520864963531494,
      "learning_rate": 6.763042718832907e-06,
      "loss": 0.8831,
      "step": 9992
    },
    {
      "epoch": 0.8075314652821269,
      "grad_norm": 2.4253652095794678,
      "learning_rate": 6.762430370670712e-06,
      "loss": 0.9169,
      "step": 9993
    },
    {
      "epoch": 0.8076122749954544,
      "grad_norm": 2.8476674556732178,
      "learning_rate": 6.761817992322329e-06,
      "loss": 0.9232,
      "step": 9994
    },
    {
      "epoch": 0.807693084708782,
      "grad_norm": 2.554034471511841,
      "learning_rate": 6.761205583798246e-06,
      "loss": 0.8709,
      "step": 9995
    },
    {
      "epoch": 0.8077738944221096,
      "grad_norm": 2.48921537399292,
      "learning_rate": 6.76059314510895e-06,
      "loss": 0.9303,
      "step": 9996
    },
    {
      "epoch": 0.807854704135437,
      "grad_norm": 2.599687099456787,
      "learning_rate": 6.759980676264932e-06,
      "loss": 0.9226,
      "step": 9997
    },
    {
      "epoch": 0.8079355138487646,
      "grad_norm": 2.7335262298583984,
      "learning_rate": 6.759368177276684e-06,
      "loss": 0.9733,
      "step": 9998
    },
    {
      "epoch": 0.8080163235620922,
      "grad_norm": 2.438054323196411,
      "learning_rate": 6.758755648154692e-06,
      "loss": 0.7867,
      "step": 9999
    },
    {
      "epoch": 0.8080971332754197,
      "grad_norm": 2.3297009468078613,
      "learning_rate": 6.758143088909453e-06,
      "loss": 0.9875,
      "step": 10000
    },
    {
      "epoch": 0.8080971332754197,
      "eval_loss": 0.7729527354240417,
      "eval_runtime": 815.2614,
      "eval_samples_per_second": 102.257,
      "eval_steps_per_second": 12.782,
      "step": 10000
    },
    {
      "epoch": 0.8081779429887472,
      "grad_norm": 2.6086413860321045,
      "learning_rate": 6.757530499551451e-06,
      "loss": 1.0694,
      "step": 10001
    },
    {
      "epoch": 0.8082587527020748,
      "grad_norm": 2.852970600128174,
      "learning_rate": 6.756917880091186e-06,
      "loss": 0.909,
      "step": 10002
    },
    {
      "epoch": 0.8083395624154023,
      "grad_norm": 2.8168246746063232,
      "learning_rate": 6.756305230539146e-06,
      "loss": 1.0034,
      "step": 10003
    },
    {
      "epoch": 0.8084203721287299,
      "grad_norm": 2.691760778427124,
      "learning_rate": 6.755692550905826e-06,
      "loss": 0.847,
      "step": 10004
    },
    {
      "epoch": 0.8085011818420574,
      "grad_norm": 2.7063395977020264,
      "learning_rate": 6.755079841201719e-06,
      "loss": 1.0821,
      "step": 10005
    },
    {
      "epoch": 0.8085819915553849,
      "grad_norm": 2.810978412628174,
      "learning_rate": 6.754467101437321e-06,
      "loss": 0.9827,
      "step": 10006
    },
    {
      "epoch": 0.8086628012687125,
      "grad_norm": 2.860426664352417,
      "learning_rate": 6.753854331623122e-06,
      "loss": 0.9415,
      "step": 10007
    },
    {
      "epoch": 0.8087436109820401,
      "grad_norm": 2.674544334411621,
      "learning_rate": 6.7532415317696234e-06,
      "loss": 1.0137,
      "step": 10008
    },
    {
      "epoch": 0.8088244206953675,
      "grad_norm": 2.3481979370117188,
      "learning_rate": 6.752628701887317e-06,
      "loss": 0.9346,
      "step": 10009
    },
    {
      "epoch": 0.8089052304086951,
      "grad_norm": 3.004422903060913,
      "learning_rate": 6.7520158419867e-06,
      "loss": 0.8694,
      "step": 10010
    },
    {
      "epoch": 0.8089860401220227,
      "grad_norm": 3.0733017921447754,
      "learning_rate": 6.75140295207827e-06,
      "loss": 0.9039,
      "step": 10011
    },
    {
      "epoch": 0.8090668498353502,
      "grad_norm": 2.641711473464966,
      "learning_rate": 6.750790032172523e-06,
      "loss": 1.105,
      "step": 10012
    },
    {
      "epoch": 0.8091476595486777,
      "grad_norm": 3.112874984741211,
      "learning_rate": 6.750177082279959e-06,
      "loss": 0.863,
      "step": 10013
    },
    {
      "epoch": 0.8092284692620053,
      "grad_norm": 2.5871925354003906,
      "learning_rate": 6.749564102411074e-06,
      "loss": 1.0219,
      "step": 10014
    },
    {
      "epoch": 0.8093092789753328,
      "grad_norm": 2.4820566177368164,
      "learning_rate": 6.748951092576367e-06,
      "loss": 0.9423,
      "step": 10015
    },
    {
      "epoch": 0.8093900886886604,
      "grad_norm": 2.7655489444732666,
      "learning_rate": 6.7483380527863394e-06,
      "loss": 1.0334,
      "step": 10016
    },
    {
      "epoch": 0.8094708984019879,
      "grad_norm": 2.9179372787475586,
      "learning_rate": 6.74772498305149e-06,
      "loss": 0.8532,
      "step": 10017
    },
    {
      "epoch": 0.8095517081153155,
      "grad_norm": 2.9236321449279785,
      "learning_rate": 6.747111883382318e-06,
      "loss": 0.8832,
      "step": 10018
    },
    {
      "epoch": 0.809632517828643,
      "grad_norm": 2.989229202270508,
      "learning_rate": 6.746498753789327e-06,
      "loss": 0.9418,
      "step": 10019
    },
    {
      "epoch": 0.8097133275419706,
      "grad_norm": 2.2503018379211426,
      "learning_rate": 6.745885594283016e-06,
      "loss": 1.0469,
      "step": 10020
    },
    {
      "epoch": 0.8097941372552981,
      "grad_norm": 2.6143321990966797,
      "learning_rate": 6.745272404873887e-06,
      "loss": 0.9602,
      "step": 10021
    },
    {
      "epoch": 0.8098749469686256,
      "grad_norm": 2.563796043395996,
      "learning_rate": 6.744659185572444e-06,
      "loss": 0.9779,
      "step": 10022
    },
    {
      "epoch": 0.8099557566819532,
      "grad_norm": 2.7808890342712402,
      "learning_rate": 6.744045936389191e-06,
      "loss": 0.9327,
      "step": 10023
    },
    {
      "epoch": 0.8100365663952808,
      "grad_norm": 3.1725735664367676,
      "learning_rate": 6.743432657334628e-06,
      "loss": 1.0153,
      "step": 10024
    },
    {
      "epoch": 0.8101173761086082,
      "grad_norm": 2.564241409301758,
      "learning_rate": 6.7428193484192605e-06,
      "loss": 0.8836,
      "step": 10025
    },
    {
      "epoch": 0.8101981858219358,
      "grad_norm": 2.9634478092193604,
      "learning_rate": 6.742206009653593e-06,
      "loss": 0.9728,
      "step": 10026
    },
    {
      "epoch": 0.8102789955352634,
      "grad_norm": 2.711183547973633,
      "learning_rate": 6.741592641048132e-06,
      "loss": 1.021,
      "step": 10027
    },
    {
      "epoch": 0.8103598052485909,
      "grad_norm": 2.8670310974121094,
      "learning_rate": 6.74097924261338e-06,
      "loss": 0.9941,
      "step": 10028
    },
    {
      "epoch": 0.8104406149619184,
      "grad_norm": 2.6566109657287598,
      "learning_rate": 6.7403658143598464e-06,
      "loss": 0.9375,
      "step": 10029
    },
    {
      "epoch": 0.810521424675246,
      "grad_norm": 3.2555360794067383,
      "learning_rate": 6.739752356298035e-06,
      "loss": 0.9621,
      "step": 10030
    },
    {
      "epoch": 0.8106022343885735,
      "grad_norm": 2.604140281677246,
      "learning_rate": 6.739138868438456e-06,
      "loss": 0.9054,
      "step": 10031
    },
    {
      "epoch": 0.8106830441019011,
      "grad_norm": 2.6726272106170654,
      "learning_rate": 6.738525350791614e-06,
      "loss": 0.9611,
      "step": 10032
    },
    {
      "epoch": 0.8107638538152286,
      "grad_norm": 3.002016544342041,
      "learning_rate": 6.737911803368017e-06,
      "loss": 0.9564,
      "step": 10033
    },
    {
      "epoch": 0.8108446635285561,
      "grad_norm": 2.8780055046081543,
      "learning_rate": 6.737298226178175e-06,
      "loss": 0.989,
      "step": 10034
    },
    {
      "epoch": 0.8109254732418837,
      "grad_norm": 2.6711130142211914,
      "learning_rate": 6.736684619232597e-06,
      "loss": 0.9191,
      "step": 10035
    },
    {
      "epoch": 0.8110062829552113,
      "grad_norm": 2.675698757171631,
      "learning_rate": 6.7360709825417925e-06,
      "loss": 0.9342,
      "step": 10036
    },
    {
      "epoch": 0.8110870926685387,
      "grad_norm": 2.771097421646118,
      "learning_rate": 6.735457316116273e-06,
      "loss": 0.9834,
      "step": 10037
    },
    {
      "epoch": 0.8111679023818663,
      "grad_norm": 2.6776483058929443,
      "learning_rate": 6.7348436199665445e-06,
      "loss": 0.9516,
      "step": 10038
    },
    {
      "epoch": 0.8112487120951939,
      "grad_norm": 2.557114601135254,
      "learning_rate": 6.734229894103124e-06,
      "loss": 0.77,
      "step": 10039
    },
    {
      "epoch": 0.8113295218085214,
      "grad_norm": 2.6638333797454834,
      "learning_rate": 6.733616138536519e-06,
      "loss": 0.8616,
      "step": 10040
    },
    {
      "epoch": 0.8114103315218489,
      "grad_norm": 2.5837926864624023,
      "learning_rate": 6.733002353277243e-06,
      "loss": 0.9252,
      "step": 10041
    },
    {
      "epoch": 0.8114911412351765,
      "grad_norm": 2.385305643081665,
      "learning_rate": 6.73238853833581e-06,
      "loss": 0.8884,
      "step": 10042
    },
    {
      "epoch": 0.811571950948504,
      "grad_norm": 2.212360382080078,
      "learning_rate": 6.73177469372273e-06,
      "loss": 0.8848,
      "step": 10043
    },
    {
      "epoch": 0.8116527606618316,
      "grad_norm": 2.545518398284912,
      "learning_rate": 6.73116081944852e-06,
      "loss": 0.8787,
      "step": 10044
    },
    {
      "epoch": 0.8117335703751591,
      "grad_norm": 2.412827730178833,
      "learning_rate": 6.730546915523693e-06,
      "loss": 0.9444,
      "step": 10045
    },
    {
      "epoch": 0.8118143800884866,
      "grad_norm": 3.18371844291687,
      "learning_rate": 6.7299329819587615e-06,
      "loss": 0.9335,
      "step": 10046
    },
    {
      "epoch": 0.8118951898018142,
      "grad_norm": 2.795923948287964,
      "learning_rate": 6.729319018764244e-06,
      "loss": 1.0072,
      "step": 10047
    },
    {
      "epoch": 0.8119759995151418,
      "grad_norm": 2.6364145278930664,
      "learning_rate": 6.728705025950656e-06,
      "loss": 0.9055,
      "step": 10048
    },
    {
      "epoch": 0.8120568092284692,
      "grad_norm": 3.0788910388946533,
      "learning_rate": 6.728091003528511e-06,
      "loss": 0.9018,
      "step": 10049
    },
    {
      "epoch": 0.8121376189417968,
      "grad_norm": 2.7288432121276855,
      "learning_rate": 6.727476951508327e-06,
      "loss": 1.0376,
      "step": 10050
    },
    {
      "epoch": 0.8122184286551244,
      "grad_norm": 2.7107584476470947,
      "learning_rate": 6.726862869900624e-06,
      "loss": 0.9126,
      "step": 10051
    },
    {
      "epoch": 0.8122992383684519,
      "grad_norm": 2.771477222442627,
      "learning_rate": 6.726248758715914e-06,
      "loss": 1.0442,
      "step": 10052
    },
    {
      "epoch": 0.8123800480817794,
      "grad_norm": 2.47438383102417,
      "learning_rate": 6.725634617964721e-06,
      "loss": 0.7962,
      "step": 10053
    },
    {
      "epoch": 0.812460857795107,
      "grad_norm": 2.296757698059082,
      "learning_rate": 6.7250204476575594e-06,
      "loss": 0.8564,
      "step": 10054
    },
    {
      "epoch": 0.8125416675084345,
      "grad_norm": 2.618964433670044,
      "learning_rate": 6.724406247804952e-06,
      "loss": 0.9034,
      "step": 10055
    },
    {
      "epoch": 0.8126224772217621,
      "grad_norm": 2.6588761806488037,
      "learning_rate": 6.7237920184174165e-06,
      "loss": 0.8751,
      "step": 10056
    },
    {
      "epoch": 0.8127032869350896,
      "grad_norm": 2.5719103813171387,
      "learning_rate": 6.723177759505473e-06,
      "loss": 1.0358,
      "step": 10057
    },
    {
      "epoch": 0.8127840966484171,
      "grad_norm": 2.8363654613494873,
      "learning_rate": 6.722563471079643e-06,
      "loss": 1.1362,
      "step": 10058
    },
    {
      "epoch": 0.8128649063617447,
      "grad_norm": 2.587702989578247,
      "learning_rate": 6.721949153150449e-06,
      "loss": 0.997,
      "step": 10059
    },
    {
      "epoch": 0.8129457160750723,
      "grad_norm": 2.430743932723999,
      "learning_rate": 6.721334805728409e-06,
      "loss": 1.0193,
      "step": 10060
    },
    {
      "epoch": 0.8130265257883997,
      "grad_norm": 2.8744118213653564,
      "learning_rate": 6.72072042882405e-06,
      "loss": 0.969,
      "step": 10061
    },
    {
      "epoch": 0.8131073355017273,
      "grad_norm": 2.4647622108459473,
      "learning_rate": 6.720106022447891e-06,
      "loss": 0.8981,
      "step": 10062
    },
    {
      "epoch": 0.8131881452150549,
      "grad_norm": 2.7981672286987305,
      "learning_rate": 6.719491586610457e-06,
      "loss": 0.9021,
      "step": 10063
    },
    {
      "epoch": 0.8132689549283824,
      "grad_norm": 2.9667656421661377,
      "learning_rate": 6.718877121322271e-06,
      "loss": 0.9479,
      "step": 10064
    },
    {
      "epoch": 0.8133497646417099,
      "grad_norm": 2.5634799003601074,
      "learning_rate": 6.718262626593861e-06,
      "loss": 1.0027,
      "step": 10065
    },
    {
      "epoch": 0.8134305743550375,
      "grad_norm": 3.1184725761413574,
      "learning_rate": 6.717648102435745e-06,
      "loss": 0.8169,
      "step": 10066
    },
    {
      "epoch": 0.813511384068365,
      "grad_norm": 2.696802854537964,
      "learning_rate": 6.717033548858455e-06,
      "loss": 0.9675,
      "step": 10067
    },
    {
      "epoch": 0.8135921937816926,
      "grad_norm": 2.2710602283477783,
      "learning_rate": 6.716418965872513e-06,
      "loss": 1.0478,
      "step": 10068
    },
    {
      "epoch": 0.8136730034950201,
      "grad_norm": 3.26822829246521,
      "learning_rate": 6.715804353488445e-06,
      "loss": 0.878,
      "step": 10069
    },
    {
      "epoch": 0.8137538132083476,
      "grad_norm": 2.9366466999053955,
      "learning_rate": 6.71518971171678e-06,
      "loss": 0.865,
      "step": 10070
    },
    {
      "epoch": 0.8138346229216752,
      "grad_norm": 2.584158420562744,
      "learning_rate": 6.714575040568044e-06,
      "loss": 0.9311,
      "step": 10071
    },
    {
      "epoch": 0.8139154326350028,
      "grad_norm": 2.484003782272339,
      "learning_rate": 6.713960340052765e-06,
      "loss": 0.9868,
      "step": 10072
    },
    {
      "epoch": 0.8139962423483302,
      "grad_norm": 2.606750249862671,
      "learning_rate": 6.713345610181474e-06,
      "loss": 0.8955,
      "step": 10073
    },
    {
      "epoch": 0.8140770520616578,
      "grad_norm": 2.787959575653076,
      "learning_rate": 6.7127308509646935e-06,
      "loss": 0.9094,
      "step": 10074
    },
    {
      "epoch": 0.8141578617749854,
      "grad_norm": 2.514554023742676,
      "learning_rate": 6.71211606241296e-06,
      "loss": 0.9302,
      "step": 10075
    },
    {
      "epoch": 0.8142386714883129,
      "grad_norm": 2.673313617706299,
      "learning_rate": 6.7115012445367985e-06,
      "loss": 1.0401,
      "step": 10076
    },
    {
      "epoch": 0.8143194812016404,
      "grad_norm": 2.5632903575897217,
      "learning_rate": 6.710886397346741e-06,
      "loss": 0.9086,
      "step": 10077
    },
    {
      "epoch": 0.814400290914968,
      "grad_norm": 2.2498035430908203,
      "learning_rate": 6.710271520853319e-06,
      "loss": 0.9825,
      "step": 10078
    },
    {
      "epoch": 0.8144811006282955,
      "grad_norm": 2.883054733276367,
      "learning_rate": 6.709656615067063e-06,
      "loss": 0.8546,
      "step": 10079
    },
    {
      "epoch": 0.8145619103416231,
      "grad_norm": 2.5208911895751953,
      "learning_rate": 6.709041679998505e-06,
      "loss": 0.8928,
      "step": 10080
    },
    {
      "epoch": 0.8146427200549506,
      "grad_norm": 2.398552417755127,
      "learning_rate": 6.708426715658177e-06,
      "loss": 0.8561,
      "step": 10081
    },
    {
      "epoch": 0.8147235297682781,
      "grad_norm": 2.866697072982788,
      "learning_rate": 6.707811722056612e-06,
      "loss": 0.9007,
      "step": 10082
    },
    {
      "epoch": 0.8148043394816057,
      "grad_norm": 2.902155876159668,
      "learning_rate": 6.707196699204345e-06,
      "loss": 0.9017,
      "step": 10083
    },
    {
      "epoch": 0.8148851491949333,
      "grad_norm": 2.83586049079895,
      "learning_rate": 6.706581647111907e-06,
      "loss": 1.0539,
      "step": 10084
    },
    {
      "epoch": 0.8149659589082607,
      "grad_norm": 2.565206527709961,
      "learning_rate": 6.705966565789834e-06,
      "loss": 0.7901,
      "step": 10085
    },
    {
      "epoch": 0.8150467686215883,
      "grad_norm": 2.6432831287384033,
      "learning_rate": 6.705351455248661e-06,
      "loss": 0.9973,
      "step": 10086
    },
    {
      "epoch": 0.8151275783349159,
      "grad_norm": 2.851337194442749,
      "learning_rate": 6.704736315498922e-06,
      "loss": 0.9275,
      "step": 10087
    },
    {
      "epoch": 0.8152083880482434,
      "grad_norm": 2.192448139190674,
      "learning_rate": 6.704121146551154e-06,
      "loss": 0.8284,
      "step": 10088
    },
    {
      "epoch": 0.8152891977615709,
      "grad_norm": 2.867723226547241,
      "learning_rate": 6.703505948415896e-06,
      "loss": 0.9557,
      "step": 10089
    },
    {
      "epoch": 0.8153700074748985,
      "grad_norm": 2.345224142074585,
      "learning_rate": 6.7028907211036806e-06,
      "loss": 0.9346,
      "step": 10090
    },
    {
      "epoch": 0.815450817188226,
      "grad_norm": 2.7858355045318604,
      "learning_rate": 6.702275464625045e-06,
      "loss": 0.969,
      "step": 10091
    },
    {
      "epoch": 0.8155316269015536,
      "grad_norm": 3.0980935096740723,
      "learning_rate": 6.701660178990531e-06,
      "loss": 0.878,
      "step": 10092
    },
    {
      "epoch": 0.8156124366148811,
      "grad_norm": 3.0792758464813232,
      "learning_rate": 6.701044864210673e-06,
      "loss": 0.8974,
      "step": 10093
    },
    {
      "epoch": 0.8156932463282086,
      "grad_norm": 3.3279123306274414,
      "learning_rate": 6.700429520296012e-06,
      "loss": 0.9739,
      "step": 10094
    },
    {
      "epoch": 0.8157740560415362,
      "grad_norm": 2.926795721054077,
      "learning_rate": 6.699814147257088e-06,
      "loss": 0.9833,
      "step": 10095
    },
    {
      "epoch": 0.8158548657548638,
      "grad_norm": 2.509044647216797,
      "learning_rate": 6.6991987451044385e-06,
      "loss": 1.0915,
      "step": 10096
    },
    {
      "epoch": 0.8159356754681912,
      "grad_norm": 2.4170308113098145,
      "learning_rate": 6.6985833138486055e-06,
      "loss": 0.9636,
      "step": 10097
    },
    {
      "epoch": 0.8160164851815188,
      "grad_norm": 3.246962308883667,
      "learning_rate": 6.697967853500132e-06,
      "loss": 0.8535,
      "step": 10098
    },
    {
      "epoch": 0.8160972948948464,
      "grad_norm": 2.6362879276275635,
      "learning_rate": 6.697352364069553e-06,
      "loss": 0.9736,
      "step": 10099
    },
    {
      "epoch": 0.8161781046081739,
      "grad_norm": 2.961191415786743,
      "learning_rate": 6.696736845567417e-06,
      "loss": 0.8553,
      "step": 10100
    },
    {
      "epoch": 0.8162589143215014,
      "grad_norm": 2.4501521587371826,
      "learning_rate": 6.6961212980042615e-06,
      "loss": 0.978,
      "step": 10101
    },
    {
      "epoch": 0.816339724034829,
      "grad_norm": 2.997964859008789,
      "learning_rate": 6.695505721390632e-06,
      "loss": 0.8973,
      "step": 10102
    },
    {
      "epoch": 0.8164205337481565,
      "grad_norm": 3.016570806503296,
      "learning_rate": 6.694890115737072e-06,
      "loss": 0.975,
      "step": 10103
    },
    {
      "epoch": 0.8165013434614841,
      "grad_norm": 2.6183083057403564,
      "learning_rate": 6.694274481054125e-06,
      "loss": 0.948,
      "step": 10104
    },
    {
      "epoch": 0.8165821531748116,
      "grad_norm": 2.6524343490600586,
      "learning_rate": 6.6936588173523335e-06,
      "loss": 1.0718,
      "step": 10105
    },
    {
      "epoch": 0.8166629628881391,
      "grad_norm": 2.4056267738342285,
      "learning_rate": 6.693043124642244e-06,
      "loss": 0.9673,
      "step": 10106
    },
    {
      "epoch": 0.8167437726014667,
      "grad_norm": 3.0743188858032227,
      "learning_rate": 6.6924274029344024e-06,
      "loss": 0.8724,
      "step": 10107
    },
    {
      "epoch": 0.8168245823147943,
      "grad_norm": 2.5183846950531006,
      "learning_rate": 6.691811652239352e-06,
      "loss": 1.002,
      "step": 10108
    },
    {
      "epoch": 0.8169053920281217,
      "grad_norm": 2.6143288612365723,
      "learning_rate": 6.691195872567643e-06,
      "loss": 0.9921,
      "step": 10109
    },
    {
      "epoch": 0.8169862017414493,
      "grad_norm": 2.5690085887908936,
      "learning_rate": 6.690580063929819e-06,
      "loss": 0.8967,
      "step": 10110
    },
    {
      "epoch": 0.8170670114547769,
      "grad_norm": 3.043970823287964,
      "learning_rate": 6.6899642263364296e-06,
      "loss": 0.9835,
      "step": 10111
    },
    {
      "epoch": 0.8171478211681044,
      "grad_norm": 2.7340104579925537,
      "learning_rate": 6.6893483597980205e-06,
      "loss": 0.9379,
      "step": 10112
    },
    {
      "epoch": 0.8172286308814319,
      "grad_norm": 2.3324716091156006,
      "learning_rate": 6.68873246432514e-06,
      "loss": 0.9084,
      "step": 10113
    },
    {
      "epoch": 0.8173094405947595,
      "grad_norm": 2.45967173576355,
      "learning_rate": 6.68811653992834e-06,
      "loss": 0.9476,
      "step": 10114
    },
    {
      "epoch": 0.817390250308087,
      "grad_norm": 3.017951011657715,
      "learning_rate": 6.6875005866181665e-06,
      "loss": 0.8792,
      "step": 10115
    },
    {
      "epoch": 0.8174710600214146,
      "grad_norm": 2.4142134189605713,
      "learning_rate": 6.68688460440517e-06,
      "loss": 1.0397,
      "step": 10116
    },
    {
      "epoch": 0.8175518697347421,
      "grad_norm": 2.2612359523773193,
      "learning_rate": 6.686268593299902e-06,
      "loss": 0.916,
      "step": 10117
    },
    {
      "epoch": 0.8176326794480696,
      "grad_norm": 2.9187474250793457,
      "learning_rate": 6.685652553312912e-06,
      "loss": 0.9164,
      "step": 10118
    },
    {
      "epoch": 0.8177134891613972,
      "grad_norm": 2.5997235774993896,
      "learning_rate": 6.685036484454751e-06,
      "loss": 0.8463,
      "step": 10119
    },
    {
      "epoch": 0.8177942988747248,
      "grad_norm": 2.774573802947998,
      "learning_rate": 6.684420386735973e-06,
      "loss": 1.1914,
      "step": 10120
    },
    {
      "epoch": 0.8178751085880522,
      "grad_norm": 2.281858444213867,
      "learning_rate": 6.683804260167128e-06,
      "loss": 1.1155,
      "step": 10121
    },
    {
      "epoch": 0.8179559183013798,
      "grad_norm": 2.6453025341033936,
      "learning_rate": 6.683188104758771e-06,
      "loss": 0.8863,
      "step": 10122
    },
    {
      "epoch": 0.8180367280147074,
      "grad_norm": 2.6469886302948,
      "learning_rate": 6.682571920521452e-06,
      "loss": 1.0179,
      "step": 10123
    },
    {
      "epoch": 0.8181175377280349,
      "grad_norm": 2.680061101913452,
      "learning_rate": 6.6819557074657285e-06,
      "loss": 0.8888,
      "step": 10124
    },
    {
      "epoch": 0.8181983474413624,
      "grad_norm": 2.871833324432373,
      "learning_rate": 6.681339465602152e-06,
      "loss": 0.854,
      "step": 10125
    },
    {
      "epoch": 0.81827915715469,
      "grad_norm": 2.630047082901001,
      "learning_rate": 6.6807231949412775e-06,
      "loss": 0.898,
      "step": 10126
    },
    {
      "epoch": 0.8183599668680175,
      "grad_norm": 2.623267412185669,
      "learning_rate": 6.680106895493661e-06,
      "loss": 0.7925,
      "step": 10127
    },
    {
      "epoch": 0.8184407765813451,
      "grad_norm": 2.749006509780884,
      "learning_rate": 6.67949056726986e-06,
      "loss": 1.0612,
      "step": 10128
    },
    {
      "epoch": 0.8185215862946726,
      "grad_norm": 2.7983627319335938,
      "learning_rate": 6.678874210280426e-06,
      "loss": 1.1179,
      "step": 10129
    },
    {
      "epoch": 0.8186023960080001,
      "grad_norm": 2.661078929901123,
      "learning_rate": 6.67825782453592e-06,
      "loss": 0.8077,
      "step": 10130
    },
    {
      "epoch": 0.8186832057213277,
      "grad_norm": 2.6801116466522217,
      "learning_rate": 6.677641410046896e-06,
      "loss": 0.8544,
      "step": 10131
    },
    {
      "epoch": 0.8187640154346553,
      "grad_norm": 2.988719940185547,
      "learning_rate": 6.6770249668239165e-06,
      "loss": 0.9211,
      "step": 10132
    },
    {
      "epoch": 0.8188448251479827,
      "grad_norm": 2.723395824432373,
      "learning_rate": 6.676408494877534e-06,
      "loss": 0.8983,
      "step": 10133
    },
    {
      "epoch": 0.8189256348613103,
      "grad_norm": 2.9994168281555176,
      "learning_rate": 6.675791994218311e-06,
      "loss": 0.9042,
      "step": 10134
    },
    {
      "epoch": 0.8190064445746379,
      "grad_norm": 2.401463031768799,
      "learning_rate": 6.675175464856806e-06,
      "loss": 0.9688,
      "step": 10135
    },
    {
      "epoch": 0.8190872542879654,
      "grad_norm": 2.3743197917938232,
      "learning_rate": 6.674558906803576e-06,
      "loss": 0.8271,
      "step": 10136
    },
    {
      "epoch": 0.8191680640012929,
      "grad_norm": 2.615125894546509,
      "learning_rate": 6.673942320069185e-06,
      "loss": 0.9315,
      "step": 10137
    },
    {
      "epoch": 0.8192488737146205,
      "grad_norm": 2.834791660308838,
      "learning_rate": 6.673325704664191e-06,
      "loss": 1.0527,
      "step": 10138
    },
    {
      "epoch": 0.819329683427948,
      "grad_norm": 2.7235167026519775,
      "learning_rate": 6.672709060599156e-06,
      "loss": 0.8809,
      "step": 10139
    },
    {
      "epoch": 0.8194104931412756,
      "grad_norm": 2.63031005859375,
      "learning_rate": 6.672092387884643e-06,
      "loss": 0.9727,
      "step": 10140
    },
    {
      "epoch": 0.8194913028546031,
      "grad_norm": 2.686319589614868,
      "learning_rate": 6.671475686531211e-06,
      "loss": 0.9677,
      "step": 10141
    },
    {
      "epoch": 0.8195721125679306,
      "grad_norm": 3.041576623916626,
      "learning_rate": 6.670858956549427e-06,
      "loss": 0.9037,
      "step": 10142
    },
    {
      "epoch": 0.8196529222812582,
      "grad_norm": 3.731450319290161,
      "learning_rate": 6.670242197949849e-06,
      "loss": 1.0614,
      "step": 10143
    },
    {
      "epoch": 0.8197337319945858,
      "grad_norm": 2.628826856613159,
      "learning_rate": 6.669625410743044e-06,
      "loss": 0.9932,
      "step": 10144
    },
    {
      "epoch": 0.8198145417079132,
      "grad_norm": 2.3889551162719727,
      "learning_rate": 6.669008594939574e-06,
      "loss": 0.9255,
      "step": 10145
    },
    {
      "epoch": 0.8198953514212408,
      "grad_norm": 2.936560869216919,
      "learning_rate": 6.668391750550006e-06,
      "loss": 0.8694,
      "step": 10146
    },
    {
      "epoch": 0.8199761611345684,
      "grad_norm": 2.8249928951263428,
      "learning_rate": 6.6677748775849035e-06,
      "loss": 0.9415,
      "step": 10147
    },
    {
      "epoch": 0.820056970847896,
      "grad_norm": 2.964205265045166,
      "learning_rate": 6.667157976054835e-06,
      "loss": 0.9028,
      "step": 10148
    },
    {
      "epoch": 0.8201377805612234,
      "grad_norm": 2.5520401000976562,
      "learning_rate": 6.66654104597036e-06,
      "loss": 0.8618,
      "step": 10149
    },
    {
      "epoch": 0.820218590274551,
      "grad_norm": 3.233186721801758,
      "learning_rate": 6.66592408734205e-06,
      "loss": 0.8467,
      "step": 10150
    },
    {
      "epoch": 0.8202993999878786,
      "grad_norm": 2.4776675701141357,
      "learning_rate": 6.665307100180472e-06,
      "loss": 0.9657,
      "step": 10151
    },
    {
      "epoch": 0.8203802097012061,
      "grad_norm": 2.51839280128479,
      "learning_rate": 6.66469008449619e-06,
      "loss": 0.9224,
      "step": 10152
    },
    {
      "epoch": 0.8204610194145336,
      "grad_norm": 2.646523952484131,
      "learning_rate": 6.664073040299777e-06,
      "loss": 0.9803,
      "step": 10153
    },
    {
      "epoch": 0.8205418291278612,
      "grad_norm": 2.538114547729492,
      "learning_rate": 6.663455967601797e-06,
      "loss": 0.9243,
      "step": 10154
    },
    {
      "epoch": 0.8206226388411887,
      "grad_norm": 2.5881614685058594,
      "learning_rate": 6.662838866412822e-06,
      "loss": 1.0662,
      "step": 10155
    },
    {
      "epoch": 0.8207034485545163,
      "grad_norm": 2.826141595840454,
      "learning_rate": 6.662221736743422e-06,
      "loss": 0.925,
      "step": 10156
    },
    {
      "epoch": 0.8207842582678438,
      "grad_norm": 2.7673892974853516,
      "learning_rate": 6.6616045786041625e-06,
      "loss": 0.9167,
      "step": 10157
    },
    {
      "epoch": 0.8208650679811713,
      "grad_norm": 2.562490701675415,
      "learning_rate": 6.660987392005618e-06,
      "loss": 0.8849,
      "step": 10158
    },
    {
      "epoch": 0.8209458776944989,
      "grad_norm": 3.130794048309326,
      "learning_rate": 6.660370176958358e-06,
      "loss": 0.9633,
      "step": 10159
    },
    {
      "epoch": 0.8210266874078265,
      "grad_norm": 2.4005680084228516,
      "learning_rate": 6.659752933472954e-06,
      "loss": 0.9505,
      "step": 10160
    },
    {
      "epoch": 0.8211074971211539,
      "grad_norm": 2.835517644882202,
      "learning_rate": 6.659135661559977e-06,
      "loss": 0.9847,
      "step": 10161
    },
    {
      "epoch": 0.8211883068344815,
      "grad_norm": 2.4224259853363037,
      "learning_rate": 6.658518361230002e-06,
      "loss": 0.8529,
      "step": 10162
    },
    {
      "epoch": 0.8212691165478091,
      "grad_norm": 3.1083216667175293,
      "learning_rate": 6.6579010324936e-06,
      "loss": 0.8878,
      "step": 10163
    },
    {
      "epoch": 0.8213499262611366,
      "grad_norm": 3.0418636798858643,
      "learning_rate": 6.6572836753613425e-06,
      "loss": 0.9649,
      "step": 10164
    },
    {
      "epoch": 0.8214307359744641,
      "grad_norm": 2.8070740699768066,
      "learning_rate": 6.656666289843808e-06,
      "loss": 0.9534,
      "step": 10165
    },
    {
      "epoch": 0.8215115456877917,
      "grad_norm": 2.583364248275757,
      "learning_rate": 6.656048875951566e-06,
      "loss": 0.7991,
      "step": 10166
    },
    {
      "epoch": 0.8215923554011192,
      "grad_norm": 2.580359697341919,
      "learning_rate": 6.655431433695195e-06,
      "loss": 0.8787,
      "step": 10167
    },
    {
      "epoch": 0.8216731651144468,
      "grad_norm": 2.6281025409698486,
      "learning_rate": 6.654813963085268e-06,
      "loss": 0.9272,
      "step": 10168
    },
    {
      "epoch": 0.8217539748277743,
      "grad_norm": 2.742807149887085,
      "learning_rate": 6.654196464132362e-06,
      "loss": 0.9912,
      "step": 10169
    },
    {
      "epoch": 0.8218347845411018,
      "grad_norm": 2.8565542697906494,
      "learning_rate": 6.653578936847052e-06,
      "loss": 0.8122,
      "step": 10170
    },
    {
      "epoch": 0.8219155942544294,
      "grad_norm": 2.4775171279907227,
      "learning_rate": 6.652961381239919e-06,
      "loss": 0.9722,
      "step": 10171
    },
    {
      "epoch": 0.821996403967757,
      "grad_norm": 2.6974034309387207,
      "learning_rate": 6.652343797321532e-06,
      "loss": 0.7939,
      "step": 10172
    },
    {
      "epoch": 0.8220772136810844,
      "grad_norm": 2.4402425289154053,
      "learning_rate": 6.651726185102477e-06,
      "loss": 1.0766,
      "step": 10173
    },
    {
      "epoch": 0.822158023394412,
      "grad_norm": 3.0757877826690674,
      "learning_rate": 6.651108544593327e-06,
      "loss": 0.9869,
      "step": 10174
    },
    {
      "epoch": 0.8222388331077396,
      "grad_norm": 2.5132808685302734,
      "learning_rate": 6.6504908758046645e-06,
      "loss": 1.0028,
      "step": 10175
    },
    {
      "epoch": 0.8223196428210671,
      "grad_norm": 2.898428201675415,
      "learning_rate": 6.649873178747065e-06,
      "loss": 0.881,
      "step": 10176
    },
    {
      "epoch": 0.8224004525343946,
      "grad_norm": 2.6547205448150635,
      "learning_rate": 6.649255453431112e-06,
      "loss": 0.8828,
      "step": 10177
    },
    {
      "epoch": 0.8224812622477222,
      "grad_norm": 2.9147605895996094,
      "learning_rate": 6.648637699867379e-06,
      "loss": 0.9364,
      "step": 10178
    },
    {
      "epoch": 0.8225620719610497,
      "grad_norm": 2.318533182144165,
      "learning_rate": 6.648019918066456e-06,
      "loss": 0.9524,
      "step": 10179
    },
    {
      "epoch": 0.8226428816743773,
      "grad_norm": 2.617670774459839,
      "learning_rate": 6.647402108038916e-06,
      "loss": 0.8285,
      "step": 10180
    },
    {
      "epoch": 0.8227236913877048,
      "grad_norm": 2.6547958850860596,
      "learning_rate": 6.646784269795347e-06,
      "loss": 0.9326,
      "step": 10181
    },
    {
      "epoch": 0.8228045011010323,
      "grad_norm": 2.443333148956299,
      "learning_rate": 6.646166403346326e-06,
      "loss": 0.9704,
      "step": 10182
    },
    {
      "epoch": 0.8228853108143599,
      "grad_norm": 2.5647428035736084,
      "learning_rate": 6.645548508702436e-06,
      "loss": 0.8155,
      "step": 10183
    },
    {
      "epoch": 0.8229661205276875,
      "grad_norm": 3.127439498901367,
      "learning_rate": 6.644930585874263e-06,
      "loss": 0.9464,
      "step": 10184
    },
    {
      "epoch": 0.8230469302410149,
      "grad_norm": 3.0389275550842285,
      "learning_rate": 6.6443126348723905e-06,
      "loss": 0.8466,
      "step": 10185
    },
    {
      "epoch": 0.8231277399543425,
      "grad_norm": 2.54085636138916,
      "learning_rate": 6.643694655707399e-06,
      "loss": 0.8709,
      "step": 10186
    },
    {
      "epoch": 0.8232085496676701,
      "grad_norm": 2.587632656097412,
      "learning_rate": 6.6430766483898765e-06,
      "loss": 0.9208,
      "step": 10187
    },
    {
      "epoch": 0.8232893593809976,
      "grad_norm": 2.5194945335388184,
      "learning_rate": 6.642458612930406e-06,
      "loss": 0.9098,
      "step": 10188
    },
    {
      "epoch": 0.8233701690943251,
      "grad_norm": 2.77231764793396,
      "learning_rate": 6.641840549339573e-06,
      "loss": 0.9631,
      "step": 10189
    },
    {
      "epoch": 0.8234509788076527,
      "grad_norm": 2.6095011234283447,
      "learning_rate": 6.641222457627964e-06,
      "loss": 0.964,
      "step": 10190
    },
    {
      "epoch": 0.8235317885209802,
      "grad_norm": 2.629227638244629,
      "learning_rate": 6.6406043378061665e-06,
      "loss": 1.0529,
      "step": 10191
    },
    {
      "epoch": 0.8236125982343078,
      "grad_norm": 2.665407180786133,
      "learning_rate": 6.6399861898847654e-06,
      "loss": 0.891,
      "step": 10192
    },
    {
      "epoch": 0.8236934079476353,
      "grad_norm": 2.96156644821167,
      "learning_rate": 6.63936801387435e-06,
      "loss": 0.9784,
      "step": 10193
    },
    {
      "epoch": 0.8237742176609628,
      "grad_norm": 2.704313278198242,
      "learning_rate": 6.638749809785504e-06,
      "loss": 0.9028,
      "step": 10194
    },
    {
      "epoch": 0.8238550273742904,
      "grad_norm": 2.567556858062744,
      "learning_rate": 6.6381315776288225e-06,
      "loss": 0.8058,
      "step": 10195
    },
    {
      "epoch": 0.823935837087618,
      "grad_norm": 2.753659248352051,
      "learning_rate": 6.63751331741489e-06,
      "loss": 0.8175,
      "step": 10196
    },
    {
      "epoch": 0.8240166468009454,
      "grad_norm": 2.827409505844116,
      "learning_rate": 6.636895029154295e-06,
      "loss": 0.994,
      "step": 10197
    },
    {
      "epoch": 0.824097456514273,
      "grad_norm": 2.9777119159698486,
      "learning_rate": 6.63627671285763e-06,
      "loss": 0.8782,
      "step": 10198
    },
    {
      "epoch": 0.8241782662276006,
      "grad_norm": 2.711076259613037,
      "learning_rate": 6.6356583685354845e-06,
      "loss": 0.9108,
      "step": 10199
    },
    {
      "epoch": 0.8242590759409281,
      "grad_norm": 2.158092975616455,
      "learning_rate": 6.635039996198447e-06,
      "loss": 0.9218,
      "step": 10200
    },
    {
      "epoch": 0.8243398856542556,
      "grad_norm": 2.5535473823547363,
      "learning_rate": 6.634421595857113e-06,
      "loss": 0.9166,
      "step": 10201
    },
    {
      "epoch": 0.8244206953675832,
      "grad_norm": 2.590787887573242,
      "learning_rate": 6.63380316752207e-06,
      "loss": 0.8717,
      "step": 10202
    },
    {
      "epoch": 0.8245015050809107,
      "grad_norm": 3.331803798675537,
      "learning_rate": 6.633184711203912e-06,
      "loss": 0.9595,
      "step": 10203
    },
    {
      "epoch": 0.8245823147942383,
      "grad_norm": 2.619899272918701,
      "learning_rate": 6.632566226913232e-06,
      "loss": 0.9667,
      "step": 10204
    },
    {
      "epoch": 0.8246631245075658,
      "grad_norm": 2.8658032417297363,
      "learning_rate": 6.631947714660622e-06,
      "loss": 0.8818,
      "step": 10205
    },
    {
      "epoch": 0.8247439342208933,
      "grad_norm": 2.6257688999176025,
      "learning_rate": 6.6313291744566775e-06,
      "loss": 0.9268,
      "step": 10206
    },
    {
      "epoch": 0.8248247439342209,
      "grad_norm": 2.7508773803710938,
      "learning_rate": 6.630710606311992e-06,
      "loss": 0.9832,
      "step": 10207
    },
    {
      "epoch": 0.8249055536475485,
      "grad_norm": 2.3672330379486084,
      "learning_rate": 6.630092010237158e-06,
      "loss": 0.8869,
      "step": 10208
    },
    {
      "epoch": 0.8249863633608759,
      "grad_norm": 2.609768867492676,
      "learning_rate": 6.629473386242773e-06,
      "loss": 0.9656,
      "step": 10209
    },
    {
      "epoch": 0.8250671730742035,
      "grad_norm": 2.485883951187134,
      "learning_rate": 6.628854734339432e-06,
      "loss": 1.0547,
      "step": 10210
    },
    {
      "epoch": 0.8251479827875311,
      "grad_norm": 2.6045002937316895,
      "learning_rate": 6.62823605453773e-06,
      "loss": 0.8622,
      "step": 10211
    },
    {
      "epoch": 0.8252287925008586,
      "grad_norm": 2.529799699783325,
      "learning_rate": 6.627617346848265e-06,
      "loss": 0.8746,
      "step": 10212
    },
    {
      "epoch": 0.8253096022141861,
      "grad_norm": 2.90537166595459,
      "learning_rate": 6.626998611281633e-06,
      "loss": 0.9279,
      "step": 10213
    },
    {
      "epoch": 0.8253904119275137,
      "grad_norm": 2.5929040908813477,
      "learning_rate": 6.626379847848431e-06,
      "loss": 0.8579,
      "step": 10214
    },
    {
      "epoch": 0.8254712216408412,
      "grad_norm": 2.4857747554779053,
      "learning_rate": 6.625761056559259e-06,
      "loss": 1.0123,
      "step": 10215
    },
    {
      "epoch": 0.8255520313541688,
      "grad_norm": 3.461653470993042,
      "learning_rate": 6.625142237424712e-06,
      "loss": 0.9556,
      "step": 10216
    },
    {
      "epoch": 0.8256328410674963,
      "grad_norm": 2.7623372077941895,
      "learning_rate": 6.624523390455392e-06,
      "loss": 0.8098,
      "step": 10217
    },
    {
      "epoch": 0.8257136507808238,
      "grad_norm": 2.3995823860168457,
      "learning_rate": 6.623904515661897e-06,
      "loss": 0.8664,
      "step": 10218
    },
    {
      "epoch": 0.8257944604941514,
      "grad_norm": 2.5108940601348877,
      "learning_rate": 6.623285613054826e-06,
      "loss": 0.9601,
      "step": 10219
    },
    {
      "epoch": 0.825875270207479,
      "grad_norm": 2.5409669876098633,
      "learning_rate": 6.622666682644782e-06,
      "loss": 0.9741,
      "step": 10220
    },
    {
      "epoch": 0.8259560799208064,
      "grad_norm": 3.002204418182373,
      "learning_rate": 6.622047724442363e-06,
      "loss": 0.9147,
      "step": 10221
    },
    {
      "epoch": 0.826036889634134,
      "grad_norm": 2.9399683475494385,
      "learning_rate": 6.621428738458171e-06,
      "loss": 0.8211,
      "step": 10222
    },
    {
      "epoch": 0.8261176993474616,
      "grad_norm": 2.6760101318359375,
      "learning_rate": 6.620809724702811e-06,
      "loss": 0.9768,
      "step": 10223
    },
    {
      "epoch": 0.8261985090607891,
      "grad_norm": 2.6501832008361816,
      "learning_rate": 6.62019068318688e-06,
      "loss": 0.8796,
      "step": 10224
    },
    {
      "epoch": 0.8262793187741166,
      "grad_norm": 2.7179949283599854,
      "learning_rate": 6.6195716139209835e-06,
      "loss": 0.9877,
      "step": 10225
    },
    {
      "epoch": 0.8263601284874442,
      "grad_norm": 2.8635172843933105,
      "learning_rate": 6.618952516915723e-06,
      "loss": 0.8617,
      "step": 10226
    },
    {
      "epoch": 0.8264409382007717,
      "grad_norm": 2.3890533447265625,
      "learning_rate": 6.618333392181705e-06,
      "loss": 0.8804,
      "step": 10227
    },
    {
      "epoch": 0.8265217479140993,
      "grad_norm": 2.7411887645721436,
      "learning_rate": 6.6177142397295315e-06,
      "loss": 1.0229,
      "step": 10228
    },
    {
      "epoch": 0.8266025576274268,
      "grad_norm": 3.016167163848877,
      "learning_rate": 6.617095059569807e-06,
      "loss": 1.0175,
      "step": 10229
    },
    {
      "epoch": 0.8266833673407543,
      "grad_norm": 3.2278783321380615,
      "learning_rate": 6.616475851713139e-06,
      "loss": 0.8996,
      "step": 10230
    },
    {
      "epoch": 0.8267641770540819,
      "grad_norm": 2.789527177810669,
      "learning_rate": 6.615856616170129e-06,
      "loss": 0.9442,
      "step": 10231
    },
    {
      "epoch": 0.8268449867674095,
      "grad_norm": 2.6740267276763916,
      "learning_rate": 6.6152373529513855e-06,
      "loss": 0.9406,
      "step": 10232
    },
    {
      "epoch": 0.8269257964807369,
      "grad_norm": 2.271397590637207,
      "learning_rate": 6.614618062067515e-06,
      "loss": 0.9842,
      "step": 10233
    },
    {
      "epoch": 0.8270066061940645,
      "grad_norm": 2.798659563064575,
      "learning_rate": 6.6139987435291244e-06,
      "loss": 0.9679,
      "step": 10234
    },
    {
      "epoch": 0.8270874159073921,
      "grad_norm": 2.850368022918701,
      "learning_rate": 6.613379397346821e-06,
      "loss": 1.0395,
      "step": 10235
    },
    {
      "epoch": 0.8271682256207196,
      "grad_norm": 2.6009531021118164,
      "learning_rate": 6.612760023531212e-06,
      "loss": 0.8984,
      "step": 10236
    },
    {
      "epoch": 0.8272490353340471,
      "grad_norm": 3.158233165740967,
      "learning_rate": 6.612140622092906e-06,
      "loss": 0.8778,
      "step": 10237
    },
    {
      "epoch": 0.8273298450473747,
      "grad_norm": 2.669201374053955,
      "learning_rate": 6.611521193042514e-06,
      "loss": 0.9173,
      "step": 10238
    },
    {
      "epoch": 0.8274106547607022,
      "grad_norm": 2.4236388206481934,
      "learning_rate": 6.6109017363906415e-06,
      "loss": 1.0802,
      "step": 10239
    },
    {
      "epoch": 0.8274914644740298,
      "grad_norm": 2.54530668258667,
      "learning_rate": 6.610282252147903e-06,
      "loss": 0.9442,
      "step": 10240
    },
    {
      "epoch": 0.8275722741873573,
      "grad_norm": 2.8342511653900146,
      "learning_rate": 6.6096627403249036e-06,
      "loss": 0.9262,
      "step": 10241
    },
    {
      "epoch": 0.8276530839006848,
      "grad_norm": 2.8013253211975098,
      "learning_rate": 6.609043200932257e-06,
      "loss": 0.8559,
      "step": 10242
    },
    {
      "epoch": 0.8277338936140124,
      "grad_norm": 3.2630040645599365,
      "learning_rate": 6.608423633980574e-06,
      "loss": 0.8682,
      "step": 10243
    },
    {
      "epoch": 0.82781470332734,
      "grad_norm": 2.5720272064208984,
      "learning_rate": 6.607804039480468e-06,
      "loss": 0.7945,
      "step": 10244
    },
    {
      "epoch": 0.8278955130406674,
      "grad_norm": 2.8171470165252686,
      "learning_rate": 6.607184417442547e-06,
      "loss": 0.9817,
      "step": 10245
    },
    {
      "epoch": 0.827976322753995,
      "grad_norm": 2.729215621948242,
      "learning_rate": 6.606564767877428e-06,
      "loss": 0.9457,
      "step": 10246
    },
    {
      "epoch": 0.8280571324673226,
      "grad_norm": 2.5352158546447754,
      "learning_rate": 6.60594509079572e-06,
      "loss": 1.0682,
      "step": 10247
    },
    {
      "epoch": 0.8281379421806501,
      "grad_norm": 2.6266767978668213,
      "learning_rate": 6.605325386208041e-06,
      "loss": 0.9063,
      "step": 10248
    },
    {
      "epoch": 0.8282187518939776,
      "grad_norm": 3.0243353843688965,
      "learning_rate": 6.604705654125001e-06,
      "loss": 0.9711,
      "step": 10249
    },
    {
      "epoch": 0.8282995616073052,
      "grad_norm": 2.0301334857940674,
      "learning_rate": 6.604085894557217e-06,
      "loss": 0.9394,
      "step": 10250
    },
    {
      "epoch": 0.8283803713206327,
      "grad_norm": 2.6448678970336914,
      "learning_rate": 6.603466107515304e-06,
      "loss": 1.0479,
      "step": 10251
    },
    {
      "epoch": 0.8284611810339603,
      "grad_norm": 2.66023325920105,
      "learning_rate": 6.602846293009877e-06,
      "loss": 0.9664,
      "step": 10252
    },
    {
      "epoch": 0.8285419907472878,
      "grad_norm": 2.9944941997528076,
      "learning_rate": 6.60222645105155e-06,
      "loss": 0.9087,
      "step": 10253
    },
    {
      "epoch": 0.8286228004606153,
      "grad_norm": 2.3767995834350586,
      "learning_rate": 6.601606581650942e-06,
      "loss": 0.8815,
      "step": 10254
    },
    {
      "epoch": 0.8287036101739429,
      "grad_norm": 2.890199899673462,
      "learning_rate": 6.600986684818669e-06,
      "loss": 0.9807,
      "step": 10255
    },
    {
      "epoch": 0.8287844198872705,
      "grad_norm": 2.7976033687591553,
      "learning_rate": 6.600366760565349e-06,
      "loss": 0.9007,
      "step": 10256
    },
    {
      "epoch": 0.8288652296005979,
      "grad_norm": 2.4705512523651123,
      "learning_rate": 6.599746808901598e-06,
      "loss": 1.0106,
      "step": 10257
    },
    {
      "epoch": 0.8289460393139255,
      "grad_norm": 2.668663263320923,
      "learning_rate": 6.5991268298380365e-06,
      "loss": 0.9168,
      "step": 10258
    },
    {
      "epoch": 0.8290268490272531,
      "grad_norm": 3.10408616065979,
      "learning_rate": 6.59850682338528e-06,
      "loss": 0.8787,
      "step": 10259
    },
    {
      "epoch": 0.8291076587405806,
      "grad_norm": 3.365107536315918,
      "learning_rate": 6.597886789553952e-06,
      "loss": 1.0074,
      "step": 10260
    },
    {
      "epoch": 0.8291884684539081,
      "grad_norm": 2.9234774112701416,
      "learning_rate": 6.597266728354669e-06,
      "loss": 1.01,
      "step": 10261
    },
    {
      "epoch": 0.8292692781672357,
      "grad_norm": 2.556942939758301,
      "learning_rate": 6.596646639798053e-06,
      "loss": 0.8714,
      "step": 10262
    },
    {
      "epoch": 0.8293500878805632,
      "grad_norm": 2.7647294998168945,
      "learning_rate": 6.596026523894723e-06,
      "loss": 1.0166,
      "step": 10263
    },
    {
      "epoch": 0.8294308975938908,
      "grad_norm": 3.5150368213653564,
      "learning_rate": 6.595406380655301e-06,
      "loss": 0.9164,
      "step": 10264
    },
    {
      "epoch": 0.8295117073072183,
      "grad_norm": 3.1027796268463135,
      "learning_rate": 6.5947862100904094e-06,
      "loss": 0.9209,
      "step": 10265
    },
    {
      "epoch": 0.8295925170205458,
      "grad_norm": 3.0092806816101074,
      "learning_rate": 6.59416601221067e-06,
      "loss": 0.9362,
      "step": 10266
    },
    {
      "epoch": 0.8296733267338734,
      "grad_norm": 2.6388099193573,
      "learning_rate": 6.593545787026702e-06,
      "loss": 1.0916,
      "step": 10267
    },
    {
      "epoch": 0.829754136447201,
      "grad_norm": 2.347038984298706,
      "learning_rate": 6.592925534549133e-06,
      "loss": 1.0426,
      "step": 10268
    },
    {
      "epoch": 0.8298349461605284,
      "grad_norm": 2.541348457336426,
      "learning_rate": 6.592305254788584e-06,
      "loss": 0.9192,
      "step": 10269
    },
    {
      "epoch": 0.829915755873856,
      "grad_norm": 2.2369649410247803,
      "learning_rate": 6.591684947755678e-06,
      "loss": 0.8332,
      "step": 10270
    },
    {
      "epoch": 0.8299965655871836,
      "grad_norm": 2.574091911315918,
      "learning_rate": 6.591064613461042e-06,
      "loss": 0.9477,
      "step": 10271
    },
    {
      "epoch": 0.8300773753005111,
      "grad_norm": 2.4311954975128174,
      "learning_rate": 6.5904442519153e-06,
      "loss": 0.9134,
      "step": 10272
    },
    {
      "epoch": 0.8301581850138386,
      "grad_norm": 2.4001121520996094,
      "learning_rate": 6.589823863129074e-06,
      "loss": 0.9378,
      "step": 10273
    },
    {
      "epoch": 0.8302389947271662,
      "grad_norm": 2.173936128616333,
      "learning_rate": 6.589203447112997e-06,
      "loss": 1.0063,
      "step": 10274
    },
    {
      "epoch": 0.8303198044404938,
      "grad_norm": 2.3609704971313477,
      "learning_rate": 6.588583003877686e-06,
      "loss": 0.9422,
      "step": 10275
    },
    {
      "epoch": 0.8304006141538213,
      "grad_norm": 2.4362683296203613,
      "learning_rate": 6.587962533433776e-06,
      "loss": 0.9251,
      "step": 10276
    },
    {
      "epoch": 0.8304814238671488,
      "grad_norm": 2.4698076248168945,
      "learning_rate": 6.587342035791889e-06,
      "loss": 1.0722,
      "step": 10277
    },
    {
      "epoch": 0.8305622335804764,
      "grad_norm": 3.0186030864715576,
      "learning_rate": 6.586721510962655e-06,
      "loss": 1.0061,
      "step": 10278
    },
    {
      "epoch": 0.8306430432938039,
      "grad_norm": 3.0380935668945312,
      "learning_rate": 6.5861009589567015e-06,
      "loss": 0.9737,
      "step": 10279
    },
    {
      "epoch": 0.8307238530071315,
      "grad_norm": 2.6261894702911377,
      "learning_rate": 6.5854803797846566e-06,
      "loss": 0.9573,
      "step": 10280
    },
    {
      "epoch": 0.830804662720459,
      "grad_norm": 2.7082245349884033,
      "learning_rate": 6.5848597734571495e-06,
      "loss": 0.9382,
      "step": 10281
    },
    {
      "epoch": 0.8308854724337865,
      "grad_norm": 2.721151113510132,
      "learning_rate": 6.584239139984811e-06,
      "loss": 0.88,
      "step": 10282
    },
    {
      "epoch": 0.8309662821471141,
      "grad_norm": 2.8823740482330322,
      "learning_rate": 6.5836184793782686e-06,
      "loss": 1.0344,
      "step": 10283
    },
    {
      "epoch": 0.8310470918604417,
      "grad_norm": 2.827423095703125,
      "learning_rate": 6.582997791648154e-06,
      "loss": 0.9251,
      "step": 10284
    },
    {
      "epoch": 0.8311279015737691,
      "grad_norm": 2.6694536209106445,
      "learning_rate": 6.582377076805099e-06,
      "loss": 0.9052,
      "step": 10285
    },
    {
      "epoch": 0.8312087112870967,
      "grad_norm": 2.617753267288208,
      "learning_rate": 6.581756334859734e-06,
      "loss": 0.8906,
      "step": 10286
    },
    {
      "epoch": 0.8312895210004243,
      "grad_norm": 2.254328489303589,
      "learning_rate": 6.58113556582269e-06,
      "loss": 0.9846,
      "step": 10287
    },
    {
      "epoch": 0.8313703307137518,
      "grad_norm": 2.3716840744018555,
      "learning_rate": 6.5805147697046e-06,
      "loss": 0.9119,
      "step": 10288
    },
    {
      "epoch": 0.8314511404270793,
      "grad_norm": 3.1163132190704346,
      "learning_rate": 6.579893946516098e-06,
      "loss": 0.8303,
      "step": 10289
    },
    {
      "epoch": 0.8315319501404069,
      "grad_norm": 2.55526065826416,
      "learning_rate": 6.579273096267818e-06,
      "loss": 0.8802,
      "step": 10290
    },
    {
      "epoch": 0.8316127598537344,
      "grad_norm": 2.5996673107147217,
      "learning_rate": 6.578652218970389e-06,
      "loss": 1.0192,
      "step": 10291
    },
    {
      "epoch": 0.831693569567062,
      "grad_norm": 2.965198040008545,
      "learning_rate": 6.578031314634447e-06,
      "loss": 0.9942,
      "step": 10292
    },
    {
      "epoch": 0.8317743792803896,
      "grad_norm": 2.2330403327941895,
      "learning_rate": 6.57741038327063e-06,
      "loss": 0.8312,
      "step": 10293
    },
    {
      "epoch": 0.831855188993717,
      "grad_norm": 2.1520514488220215,
      "learning_rate": 6.57678942488957e-06,
      "loss": 1.1504,
      "step": 10294
    },
    {
      "epoch": 0.8319359987070446,
      "grad_norm": 2.4626641273498535,
      "learning_rate": 6.576168439501902e-06,
      "loss": 0.9926,
      "step": 10295
    },
    {
      "epoch": 0.8320168084203722,
      "grad_norm": 2.5862386226654053,
      "learning_rate": 6.5755474271182655e-06,
      "loss": 0.8666,
      "step": 10296
    },
    {
      "epoch": 0.8320976181336996,
      "grad_norm": 2.8516321182250977,
      "learning_rate": 6.5749263877492934e-06,
      "loss": 0.8282,
      "step": 10297
    },
    {
      "epoch": 0.8321784278470272,
      "grad_norm": 2.7109827995300293,
      "learning_rate": 6.574305321405622e-06,
      "loss": 1.0934,
      "step": 10298
    },
    {
      "epoch": 0.8322592375603548,
      "grad_norm": 2.948587656021118,
      "learning_rate": 6.573684228097893e-06,
      "loss": 0.9018,
      "step": 10299
    },
    {
      "epoch": 0.8323400472736823,
      "grad_norm": 2.943542003631592,
      "learning_rate": 6.5730631078367406e-06,
      "loss": 0.9913,
      "step": 10300
    },
    {
      "epoch": 0.8324208569870098,
      "grad_norm": 2.990828514099121,
      "learning_rate": 6.572441960632803e-06,
      "loss": 0.8745,
      "step": 10301
    },
    {
      "epoch": 0.8325016667003374,
      "grad_norm": 2.6374523639678955,
      "learning_rate": 6.571820786496721e-06,
      "loss": 0.9704,
      "step": 10302
    },
    {
      "epoch": 0.8325824764136649,
      "grad_norm": 2.488494396209717,
      "learning_rate": 6.571199585439133e-06,
      "loss": 0.9956,
      "step": 10303
    },
    {
      "epoch": 0.8326632861269925,
      "grad_norm": 2.843135118484497,
      "learning_rate": 6.570578357470678e-06,
      "loss": 0.8777,
      "step": 10304
    },
    {
      "epoch": 0.83274409584032,
      "grad_norm": 2.8891029357910156,
      "learning_rate": 6.569957102601999e-06,
      "loss": 0.9228,
      "step": 10305
    },
    {
      "epoch": 0.8328249055536475,
      "grad_norm": 2.464301347732544,
      "learning_rate": 6.569335820843732e-06,
      "loss": 0.97,
      "step": 10306
    },
    {
      "epoch": 0.8329057152669751,
      "grad_norm": 2.515451431274414,
      "learning_rate": 6.568714512206522e-06,
      "loss": 0.9194,
      "step": 10307
    },
    {
      "epoch": 0.8329865249803027,
      "grad_norm": 2.717715263366699,
      "learning_rate": 6.568093176701008e-06,
      "loss": 0.9511,
      "step": 10308
    },
    {
      "epoch": 0.8330673346936301,
      "grad_norm": 3.003147840499878,
      "learning_rate": 6.567471814337834e-06,
      "loss": 0.898,
      "step": 10309
    },
    {
      "epoch": 0.8331481444069577,
      "grad_norm": 3.3430235385894775,
      "learning_rate": 6.566850425127639e-06,
      "loss": 0.9178,
      "step": 10310
    },
    {
      "epoch": 0.8332289541202853,
      "grad_norm": 2.6437325477600098,
      "learning_rate": 6.566229009081071e-06,
      "loss": 0.9032,
      "step": 10311
    },
    {
      "epoch": 0.8333097638336128,
      "grad_norm": 2.8529152870178223,
      "learning_rate": 6.565607566208768e-06,
      "loss": 0.8734,
      "step": 10312
    },
    {
      "epoch": 0.8333905735469403,
      "grad_norm": 2.3621718883514404,
      "learning_rate": 6.564986096521379e-06,
      "loss": 1.0013,
      "step": 10313
    },
    {
      "epoch": 0.8334713832602679,
      "grad_norm": 2.4563350677490234,
      "learning_rate": 6.5643646000295425e-06,
      "loss": 0.9843,
      "step": 10314
    },
    {
      "epoch": 0.8335521929735954,
      "grad_norm": 2.558286428451538,
      "learning_rate": 6.5637430767439096e-06,
      "loss": 1.0255,
      "step": 10315
    },
    {
      "epoch": 0.833633002686923,
      "grad_norm": 2.2724075317382812,
      "learning_rate": 6.563121526675121e-06,
      "loss": 0.9457,
      "step": 10316
    },
    {
      "epoch": 0.8337138124002506,
      "grad_norm": 2.591700315475464,
      "learning_rate": 6.5624999498338234e-06,
      "loss": 1.0062,
      "step": 10317
    },
    {
      "epoch": 0.833794622113578,
      "grad_norm": 2.2724504470825195,
      "learning_rate": 6.561878346230664e-06,
      "loss": 0.8854,
      "step": 10318
    },
    {
      "epoch": 0.8338754318269056,
      "grad_norm": 2.7128658294677734,
      "learning_rate": 6.5612567158762894e-06,
      "loss": 0.9872,
      "step": 10319
    },
    {
      "epoch": 0.8339562415402332,
      "grad_norm": 2.780306816101074,
      "learning_rate": 6.560635058781342e-06,
      "loss": 0.8845,
      "step": 10320
    },
    {
      "epoch": 0.8340370512535606,
      "grad_norm": 2.6698758602142334,
      "learning_rate": 6.5600133749564775e-06,
      "loss": 1.0092,
      "step": 10321
    },
    {
      "epoch": 0.8341178609668882,
      "grad_norm": 2.7282967567443848,
      "learning_rate": 6.559391664412338e-06,
      "loss": 0.9888,
      "step": 10322
    },
    {
      "epoch": 0.8341986706802158,
      "grad_norm": 3.131373167037964,
      "learning_rate": 6.558769927159573e-06,
      "loss": 1.0423,
      "step": 10323
    },
    {
      "epoch": 0.8342794803935433,
      "grad_norm": 2.3342626094818115,
      "learning_rate": 6.558148163208832e-06,
      "loss": 0.9927,
      "step": 10324
    },
    {
      "epoch": 0.8343602901068708,
      "grad_norm": 2.7721657752990723,
      "learning_rate": 6.557526372570765e-06,
      "loss": 0.9143,
      "step": 10325
    },
    {
      "epoch": 0.8344410998201984,
      "grad_norm": 2.354970693588257,
      "learning_rate": 6.556904555256019e-06,
      "loss": 1.0307,
      "step": 10326
    },
    {
      "epoch": 0.8345219095335259,
      "grad_norm": 2.6982693672180176,
      "learning_rate": 6.556282711275247e-06,
      "loss": 0.8372,
      "step": 10327
    },
    {
      "epoch": 0.8346027192468535,
      "grad_norm": 2.881446123123169,
      "learning_rate": 6.555660840639097e-06,
      "loss": 0.9558,
      "step": 10328
    },
    {
      "epoch": 0.834683528960181,
      "grad_norm": 2.7588305473327637,
      "learning_rate": 6.555038943358225e-06,
      "loss": 0.88,
      "step": 10329
    },
    {
      "epoch": 0.8347643386735085,
      "grad_norm": 2.5668423175811768,
      "learning_rate": 6.554417019443278e-06,
      "loss": 0.9132,
      "step": 10330
    },
    {
      "epoch": 0.8348451483868361,
      "grad_norm": 2.724865198135376,
      "learning_rate": 6.553795068904909e-06,
      "loss": 0.9082,
      "step": 10331
    },
    {
      "epoch": 0.8349259581001637,
      "grad_norm": 2.9270431995391846,
      "learning_rate": 6.553173091753771e-06,
      "loss": 0.9653,
      "step": 10332
    },
    {
      "epoch": 0.8350067678134911,
      "grad_norm": 2.477363109588623,
      "learning_rate": 6.552551088000519e-06,
      "loss": 0.9978,
      "step": 10333
    },
    {
      "epoch": 0.8350875775268187,
      "grad_norm": 2.592115879058838,
      "learning_rate": 6.551929057655802e-06,
      "loss": 1.0549,
      "step": 10334
    },
    {
      "epoch": 0.8351683872401463,
      "grad_norm": 3.223541498184204,
      "learning_rate": 6.551307000730278e-06,
      "loss": 0.9469,
      "step": 10335
    },
    {
      "epoch": 0.8352491969534738,
      "grad_norm": 2.4590396881103516,
      "learning_rate": 6.5506849172346e-06,
      "loss": 0.9423,
      "step": 10336
    },
    {
      "epoch": 0.8353300066668014,
      "grad_norm": 2.496253252029419,
      "learning_rate": 6.5500628071794215e-06,
      "loss": 0.8054,
      "step": 10337
    },
    {
      "epoch": 0.8354108163801289,
      "grad_norm": 2.501291513442993,
      "learning_rate": 6.549440670575399e-06,
      "loss": 0.9644,
      "step": 10338
    },
    {
      "epoch": 0.8354916260934564,
      "grad_norm": 2.6153807640075684,
      "learning_rate": 6.548818507433189e-06,
      "loss": 0.8983,
      "step": 10339
    },
    {
      "epoch": 0.835572435806784,
      "grad_norm": 2.555190086364746,
      "learning_rate": 6.548196317763445e-06,
      "loss": 0.8627,
      "step": 10340
    },
    {
      "epoch": 0.8356532455201116,
      "grad_norm": 2.7666139602661133,
      "learning_rate": 6.547574101576826e-06,
      "loss": 0.9579,
      "step": 10341
    },
    {
      "epoch": 0.835734055233439,
      "grad_norm": 2.5289762020111084,
      "learning_rate": 6.546951858883986e-06,
      "loss": 0.9392,
      "step": 10342
    },
    {
      "epoch": 0.8358148649467666,
      "grad_norm": 2.3865489959716797,
      "learning_rate": 6.546329589695588e-06,
      "loss": 0.933,
      "step": 10343
    },
    {
      "epoch": 0.8358956746600942,
      "grad_norm": 2.9827826023101807,
      "learning_rate": 6.545707294022286e-06,
      "loss": 0.8892,
      "step": 10344
    },
    {
      "epoch": 0.8359764843734216,
      "grad_norm": 2.7289986610412598,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.9109,
      "step": 10345
    },
    {
      "epoch": 0.8360572940867492,
      "grad_norm": 2.163463592529297,
      "learning_rate": 6.5444626232636045e-06,
      "loss": 0.9232,
      "step": 10346
    },
    {
      "epoch": 0.8361381038000768,
      "grad_norm": 2.7812235355377197,
      "learning_rate": 6.543840248199546e-06,
      "loss": 0.8579,
      "step": 10347
    },
    {
      "epoch": 0.8362189135134043,
      "grad_norm": 2.5764145851135254,
      "learning_rate": 6.543217846693217e-06,
      "loss": 0.9469,
      "step": 10348
    },
    {
      "epoch": 0.8362997232267319,
      "grad_norm": 2.4816665649414062,
      "learning_rate": 6.542595418755286e-06,
      "loss": 1.0085,
      "step": 10349
    },
    {
      "epoch": 0.8363805329400594,
      "grad_norm": 2.3393003940582275,
      "learning_rate": 6.5419729643964055e-06,
      "loss": 0.954,
      "step": 10350
    },
    {
      "epoch": 0.8364613426533869,
      "grad_norm": 2.690126895904541,
      "learning_rate": 6.541350483627242e-06,
      "loss": 0.9522,
      "step": 10351
    },
    {
      "epoch": 0.8365421523667145,
      "grad_norm": 2.5432019233703613,
      "learning_rate": 6.5407279764584555e-06,
      "loss": 0.9274,
      "step": 10352
    },
    {
      "epoch": 0.836622962080042,
      "grad_norm": 2.5258004665374756,
      "learning_rate": 6.540105442900707e-06,
      "loss": 0.9676,
      "step": 10353
    },
    {
      "epoch": 0.8367037717933695,
      "grad_norm": 2.782893180847168,
      "learning_rate": 6.539482882964661e-06,
      "loss": 0.8797,
      "step": 10354
    },
    {
      "epoch": 0.8367845815066971,
      "grad_norm": 2.819946765899658,
      "learning_rate": 6.538860296660978e-06,
      "loss": 0.8726,
      "step": 10355
    },
    {
      "epoch": 0.8368653912200247,
      "grad_norm": 2.5403528213500977,
      "learning_rate": 6.538237684000324e-06,
      "loss": 0.8837,
      "step": 10356
    },
    {
      "epoch": 0.8369462009333521,
      "grad_norm": 2.641281843185425,
      "learning_rate": 6.537615044993362e-06,
      "loss": 0.8482,
      "step": 10357
    },
    {
      "epoch": 0.8370270106466797,
      "grad_norm": 2.8591344356536865,
      "learning_rate": 6.536992379650755e-06,
      "loss": 0.8728,
      "step": 10358
    },
    {
      "epoch": 0.8371078203600073,
      "grad_norm": 2.840751886367798,
      "learning_rate": 6.5363696879831686e-06,
      "loss": 0.9887,
      "step": 10359
    },
    {
      "epoch": 0.8371886300733348,
      "grad_norm": 2.555953025817871,
      "learning_rate": 6.535746970001268e-06,
      "loss": 0.8691,
      "step": 10360
    },
    {
      "epoch": 0.8372694397866624,
      "grad_norm": 2.655471086502075,
      "learning_rate": 6.535124225715719e-06,
      "loss": 0.9269,
      "step": 10361
    },
    {
      "epoch": 0.8373502494999899,
      "grad_norm": 2.682887077331543,
      "learning_rate": 6.534501455137188e-06,
      "loss": 0.852,
      "step": 10362
    },
    {
      "epoch": 0.8374310592133174,
      "grad_norm": 2.834235668182373,
      "learning_rate": 6.53387865827634e-06,
      "loss": 0.8463,
      "step": 10363
    },
    {
      "epoch": 0.837511868926645,
      "grad_norm": 2.5176448822021484,
      "learning_rate": 6.5332558351438454e-06,
      "loss": 0.9175,
      "step": 10364
    },
    {
      "epoch": 0.8375926786399726,
      "grad_norm": 2.8520047664642334,
      "learning_rate": 6.532632985750369e-06,
      "loss": 0.9253,
      "step": 10365
    },
    {
      "epoch": 0.8376734883533,
      "grad_norm": 2.656947135925293,
      "learning_rate": 6.5320101101065795e-06,
      "loss": 0.9316,
      "step": 10366
    },
    {
      "epoch": 0.8377542980666276,
      "grad_norm": 2.4157190322875977,
      "learning_rate": 6.531387208223143e-06,
      "loss": 0.9568,
      "step": 10367
    },
    {
      "epoch": 0.8378351077799552,
      "grad_norm": 2.4857492446899414,
      "learning_rate": 6.530764280110732e-06,
      "loss": 1.0323,
      "step": 10368
    },
    {
      "epoch": 0.8379159174932826,
      "grad_norm": 2.6493284702301025,
      "learning_rate": 6.5301413257800126e-06,
      "loss": 0.9611,
      "step": 10369
    },
    {
      "epoch": 0.8379967272066102,
      "grad_norm": 2.4020230770111084,
      "learning_rate": 6.5295183452416575e-06,
      "loss": 1.0994,
      "step": 10370
    },
    {
      "epoch": 0.8380775369199378,
      "grad_norm": 2.5846850872039795,
      "learning_rate": 6.528895338506334e-06,
      "loss": 0.8125,
      "step": 10371
    },
    {
      "epoch": 0.8381583466332653,
      "grad_norm": 2.467902421951294,
      "learning_rate": 6.528272305584717e-06,
      "loss": 0.9869,
      "step": 10372
    },
    {
      "epoch": 0.8382391563465929,
      "grad_norm": 2.4679744243621826,
      "learning_rate": 6.527649246487471e-06,
      "loss": 0.8842,
      "step": 10373
    },
    {
      "epoch": 0.8383199660599204,
      "grad_norm": 2.61643648147583,
      "learning_rate": 6.5270261612252725e-06,
      "loss": 1.0172,
      "step": 10374
    },
    {
      "epoch": 0.8384007757732479,
      "grad_norm": 2.2779440879821777,
      "learning_rate": 6.526403049808791e-06,
      "loss": 0.9067,
      "step": 10375
    },
    {
      "epoch": 0.8384815854865755,
      "grad_norm": 2.7108347415924072,
      "learning_rate": 6.525779912248702e-06,
      "loss": 0.8728,
      "step": 10376
    },
    {
      "epoch": 0.838562395199903,
      "grad_norm": 2.429582118988037,
      "learning_rate": 6.525156748555674e-06,
      "loss": 0.9884,
      "step": 10377
    },
    {
      "epoch": 0.8386432049132305,
      "grad_norm": 3.2600700855255127,
      "learning_rate": 6.524533558740385e-06,
      "loss": 0.9382,
      "step": 10378
    },
    {
      "epoch": 0.8387240146265581,
      "grad_norm": 3.188776731491089,
      "learning_rate": 6.523910342813504e-06,
      "loss": 1.016,
      "step": 10379
    },
    {
      "epoch": 0.8388048243398857,
      "grad_norm": 3.175647258758545,
      "learning_rate": 6.523287100785709e-06,
      "loss": 0.9298,
      "step": 10380
    },
    {
      "epoch": 0.8388856340532131,
      "grad_norm": 2.506345510482788,
      "learning_rate": 6.522663832667672e-06,
      "loss": 0.9197,
      "step": 10381
    },
    {
      "epoch": 0.8389664437665407,
      "grad_norm": 2.5089378356933594,
      "learning_rate": 6.52204053847007e-06,
      "loss": 0.865,
      "step": 10382
    },
    {
      "epoch": 0.8390472534798683,
      "grad_norm": 2.816622495651245,
      "learning_rate": 6.521417218203579e-06,
      "loss": 0.9613,
      "step": 10383
    },
    {
      "epoch": 0.8391280631931958,
      "grad_norm": 2.5885701179504395,
      "learning_rate": 6.520793871878871e-06,
      "loss": 0.9186,
      "step": 10384
    },
    {
      "epoch": 0.8392088729065234,
      "grad_norm": 2.666088581085205,
      "learning_rate": 6.520170499506626e-06,
      "loss": 0.937,
      "step": 10385
    },
    {
      "epoch": 0.8392896826198509,
      "grad_norm": 2.1622211933135986,
      "learning_rate": 6.519547101097522e-06,
      "loss": 0.9675,
      "step": 10386
    },
    {
      "epoch": 0.8393704923331784,
      "grad_norm": 2.9082515239715576,
      "learning_rate": 6.518923676662231e-06,
      "loss": 0.9938,
      "step": 10387
    },
    {
      "epoch": 0.839451302046506,
      "grad_norm": 2.3681159019470215,
      "learning_rate": 6.518300226211437e-06,
      "loss": 1.0046,
      "step": 10388
    },
    {
      "epoch": 0.8395321117598336,
      "grad_norm": 2.5531108379364014,
      "learning_rate": 6.517676749755813e-06,
      "loss": 1.1372,
      "step": 10389
    },
    {
      "epoch": 0.839612921473161,
      "grad_norm": 3.0269222259521484,
      "learning_rate": 6.5170532473060425e-06,
      "loss": 0.9001,
      "step": 10390
    },
    {
      "epoch": 0.8396937311864886,
      "grad_norm": 2.360135555267334,
      "learning_rate": 6.5164297188728e-06,
      "loss": 0.9343,
      "step": 10391
    },
    {
      "epoch": 0.8397745408998162,
      "grad_norm": 2.1920666694641113,
      "learning_rate": 6.515806164466768e-06,
      "loss": 0.9894,
      "step": 10392
    },
    {
      "epoch": 0.8398553506131436,
      "grad_norm": 2.4082422256469727,
      "learning_rate": 6.515182584098624e-06,
      "loss": 0.8902,
      "step": 10393
    },
    {
      "epoch": 0.8399361603264712,
      "grad_norm": 2.6963584423065186,
      "learning_rate": 6.514558977779052e-06,
      "loss": 0.9596,
      "step": 10394
    },
    {
      "epoch": 0.8400169700397988,
      "grad_norm": 3.0328590869903564,
      "learning_rate": 6.513935345518731e-06,
      "loss": 1.0362,
      "step": 10395
    },
    {
      "epoch": 0.8400977797531263,
      "grad_norm": 2.9155263900756836,
      "learning_rate": 6.51331168732834e-06,
      "loss": 0.9551,
      "step": 10396
    },
    {
      "epoch": 0.8401785894664539,
      "grad_norm": 2.918823719024658,
      "learning_rate": 6.5126880032185634e-06,
      "loss": 0.973,
      "step": 10397
    },
    {
      "epoch": 0.8402593991797814,
      "grad_norm": 2.338228940963745,
      "learning_rate": 6.512064293200084e-06,
      "loss": 0.8797,
      "step": 10398
    },
    {
      "epoch": 0.8403402088931089,
      "grad_norm": 2.837373971939087,
      "learning_rate": 6.511440557283584e-06,
      "loss": 0.86,
      "step": 10399
    },
    {
      "epoch": 0.8404210186064365,
      "grad_norm": 2.5574188232421875,
      "learning_rate": 6.5108167954797455e-06,
      "loss": 0.8753,
      "step": 10400
    },
    {
      "epoch": 0.840501828319764,
      "grad_norm": 2.618107795715332,
      "learning_rate": 6.510193007799251e-06,
      "loss": 0.8964,
      "step": 10401
    },
    {
      "epoch": 0.8405826380330915,
      "grad_norm": 2.5248045921325684,
      "learning_rate": 6.509569194252787e-06,
      "loss": 0.9029,
      "step": 10402
    },
    {
      "epoch": 0.8406634477464191,
      "grad_norm": 2.905775547027588,
      "learning_rate": 6.508945354851037e-06,
      "loss": 0.9027,
      "step": 10403
    },
    {
      "epoch": 0.8407442574597467,
      "grad_norm": 2.5878195762634277,
      "learning_rate": 6.508321489604685e-06,
      "loss": 1.0004,
      "step": 10404
    },
    {
      "epoch": 0.8408250671730743,
      "grad_norm": 3.027595043182373,
      "learning_rate": 6.507697598524417e-06,
      "loss": 0.9041,
      "step": 10405
    },
    {
      "epoch": 0.8409058768864017,
      "grad_norm": 2.4799745082855225,
      "learning_rate": 6.5070736816209205e-06,
      "loss": 0.9806,
      "step": 10406
    },
    {
      "epoch": 0.8409866865997293,
      "grad_norm": 2.5228214263916016,
      "learning_rate": 6.5064497389048775e-06,
      "loss": 0.9409,
      "step": 10407
    },
    {
      "epoch": 0.8410674963130569,
      "grad_norm": 2.364022731781006,
      "learning_rate": 6.5058257703869786e-06,
      "loss": 0.9317,
      "step": 10408
    },
    {
      "epoch": 0.8411483060263844,
      "grad_norm": 3.2667176723480225,
      "learning_rate": 6.50520177607791e-06,
      "loss": 0.8653,
      "step": 10409
    },
    {
      "epoch": 0.8412291157397119,
      "grad_norm": 2.6843714714050293,
      "learning_rate": 6.504577755988357e-06,
      "loss": 0.9667,
      "step": 10410
    },
    {
      "epoch": 0.8413099254530395,
      "grad_norm": 2.670832395553589,
      "learning_rate": 6.50395371012901e-06,
      "loss": 0.9697,
      "step": 10411
    },
    {
      "epoch": 0.841390735166367,
      "grad_norm": 2.891305923461914,
      "learning_rate": 6.503329638510556e-06,
      "loss": 0.9298,
      "step": 10412
    },
    {
      "epoch": 0.8414715448796946,
      "grad_norm": 2.956672191619873,
      "learning_rate": 6.502705541143685e-06,
      "loss": 1.0445,
      "step": 10413
    },
    {
      "epoch": 0.8415523545930221,
      "grad_norm": 2.436795234680176,
      "learning_rate": 6.502081418039086e-06,
      "loss": 0.9082,
      "step": 10414
    },
    {
      "epoch": 0.8416331643063496,
      "grad_norm": 3.1604461669921875,
      "learning_rate": 6.501457269207446e-06,
      "loss": 0.8802,
      "step": 10415
    },
    {
      "epoch": 0.8417139740196772,
      "grad_norm": 2.384186267852783,
      "learning_rate": 6.500833094659461e-06,
      "loss": 0.9525,
      "step": 10416
    },
    {
      "epoch": 0.8417947837330048,
      "grad_norm": 2.745936155319214,
      "learning_rate": 6.500208894405817e-06,
      "loss": 0.8545,
      "step": 10417
    },
    {
      "epoch": 0.8418755934463322,
      "grad_norm": 2.5840983390808105,
      "learning_rate": 6.499584668457205e-06,
      "loss": 0.9579,
      "step": 10418
    },
    {
      "epoch": 0.8419564031596598,
      "grad_norm": 2.8174636363983154,
      "learning_rate": 6.498960416824319e-06,
      "loss": 0.9557,
      "step": 10419
    },
    {
      "epoch": 0.8420372128729874,
      "grad_norm": 2.354670763015747,
      "learning_rate": 6.498336139517849e-06,
      "loss": 0.8537,
      "step": 10420
    },
    {
      "epoch": 0.8421180225863149,
      "grad_norm": 2.612330675125122,
      "learning_rate": 6.497711836548488e-06,
      "loss": 0.9151,
      "step": 10421
    },
    {
      "epoch": 0.8421988322996424,
      "grad_norm": 2.763094663619995,
      "learning_rate": 6.497087507926929e-06,
      "loss": 0.9803,
      "step": 10422
    },
    {
      "epoch": 0.84227964201297,
      "grad_norm": 2.8730318546295166,
      "learning_rate": 6.4964631536638655e-06,
      "loss": 1.0824,
      "step": 10423
    },
    {
      "epoch": 0.8423604517262975,
      "grad_norm": 2.7079296112060547,
      "learning_rate": 6.4958387737699916e-06,
      "loss": 1.0005,
      "step": 10424
    },
    {
      "epoch": 0.842441261439625,
      "grad_norm": 3.3459935188293457,
      "learning_rate": 6.495214368256e-06,
      "loss": 0.8695,
      "step": 10425
    },
    {
      "epoch": 0.8425220711529526,
      "grad_norm": 2.3343212604522705,
      "learning_rate": 6.494589937132585e-06,
      "loss": 0.9076,
      "step": 10426
    },
    {
      "epoch": 0.8426028808662801,
      "grad_norm": 2.8797125816345215,
      "learning_rate": 6.493965480410443e-06,
      "loss": 0.8911,
      "step": 10427
    },
    {
      "epoch": 0.8426836905796077,
      "grad_norm": 2.4214296340942383,
      "learning_rate": 6.493340998100268e-06,
      "loss": 0.9846,
      "step": 10428
    },
    {
      "epoch": 0.8427645002929353,
      "grad_norm": 2.4357798099517822,
      "learning_rate": 6.492716490212757e-06,
      "loss": 0.8807,
      "step": 10429
    },
    {
      "epoch": 0.8428453100062627,
      "grad_norm": 2.704138994216919,
      "learning_rate": 6.492091956758606e-06,
      "loss": 0.9051,
      "step": 10430
    },
    {
      "epoch": 0.8429261197195903,
      "grad_norm": 2.5518760681152344,
      "learning_rate": 6.491467397748514e-06,
      "loss": 1.0004,
      "step": 10431
    },
    {
      "epoch": 0.8430069294329179,
      "grad_norm": 2.386793613433838,
      "learning_rate": 6.490842813193174e-06,
      "loss": 0.9185,
      "step": 10432
    },
    {
      "epoch": 0.8430877391462454,
      "grad_norm": 2.8249313831329346,
      "learning_rate": 6.4902182031032866e-06,
      "loss": 1.0037,
      "step": 10433
    },
    {
      "epoch": 0.8431685488595729,
      "grad_norm": 2.5971086025238037,
      "learning_rate": 6.489593567489548e-06,
      "loss": 0.8821,
      "step": 10434
    },
    {
      "epoch": 0.8432493585729005,
      "grad_norm": 2.56463360786438,
      "learning_rate": 6.4889689063626585e-06,
      "loss": 0.9605,
      "step": 10435
    },
    {
      "epoch": 0.843330168286228,
      "grad_norm": 2.573432207107544,
      "learning_rate": 6.488344219733316e-06,
      "loss": 0.903,
      "step": 10436
    },
    {
      "epoch": 0.8434109779995556,
      "grad_norm": 2.7142174243927,
      "learning_rate": 6.487719507612219e-06,
      "loss": 1.094,
      "step": 10437
    },
    {
      "epoch": 0.8434917877128831,
      "grad_norm": 2.6429712772369385,
      "learning_rate": 6.487094770010069e-06,
      "loss": 0.9953,
      "step": 10438
    },
    {
      "epoch": 0.8435725974262106,
      "grad_norm": 2.763674020767212,
      "learning_rate": 6.486470006937567e-06,
      "loss": 0.8782,
      "step": 10439
    },
    {
      "epoch": 0.8436534071395382,
      "grad_norm": 2.7453370094299316,
      "learning_rate": 6.48584521840541e-06,
      "loss": 0.9211,
      "step": 10440
    },
    {
      "epoch": 0.8437342168528658,
      "grad_norm": 2.8645360469818115,
      "learning_rate": 6.485220404424304e-06,
      "loss": 0.882,
      "step": 10441
    },
    {
      "epoch": 0.8438150265661932,
      "grad_norm": 2.6044774055480957,
      "learning_rate": 6.4845955650049454e-06,
      "loss": 0.8915,
      "step": 10442
    },
    {
      "epoch": 0.8438958362795208,
      "grad_norm": 2.4029619693756104,
      "learning_rate": 6.4839707001580395e-06,
      "loss": 0.9417,
      "step": 10443
    },
    {
      "epoch": 0.8439766459928484,
      "grad_norm": 2.8236749172210693,
      "learning_rate": 6.483345809894289e-06,
      "loss": 0.9543,
      "step": 10444
    },
    {
      "epoch": 0.8440574557061759,
      "grad_norm": 2.9325554370880127,
      "learning_rate": 6.482720894224397e-06,
      "loss": 0.9072,
      "step": 10445
    },
    {
      "epoch": 0.8441382654195034,
      "grad_norm": 2.2445032596588135,
      "learning_rate": 6.482095953159062e-06,
      "loss": 0.8364,
      "step": 10446
    },
    {
      "epoch": 0.844219075132831,
      "grad_norm": 2.74062442779541,
      "learning_rate": 6.481470986708994e-06,
      "loss": 0.9539,
      "step": 10447
    },
    {
      "epoch": 0.8442998848461585,
      "grad_norm": 2.362377405166626,
      "learning_rate": 6.480845994884893e-06,
      "loss": 0.9643,
      "step": 10448
    },
    {
      "epoch": 0.8443806945594861,
      "grad_norm": 2.9559617042541504,
      "learning_rate": 6.480220977697467e-06,
      "loss": 0.8588,
      "step": 10449
    },
    {
      "epoch": 0.8444615042728136,
      "grad_norm": 2.6655261516571045,
      "learning_rate": 6.479595935157417e-06,
      "loss": 0.9298,
      "step": 10450
    },
    {
      "epoch": 0.8445423139861411,
      "grad_norm": 2.5437235832214355,
      "learning_rate": 6.478970867275451e-06,
      "loss": 0.9388,
      "step": 10451
    },
    {
      "epoch": 0.8446231236994687,
      "grad_norm": 2.9839928150177,
      "learning_rate": 6.478345774062276e-06,
      "loss": 0.9739,
      "step": 10452
    },
    {
      "epoch": 0.8447039334127963,
      "grad_norm": 2.721081495285034,
      "learning_rate": 6.477720655528597e-06,
      "loss": 0.9002,
      "step": 10453
    },
    {
      "epoch": 0.8447847431261237,
      "grad_norm": 2.4781289100646973,
      "learning_rate": 6.477095511685117e-06,
      "loss": 1.0952,
      "step": 10454
    },
    {
      "epoch": 0.8448655528394513,
      "grad_norm": 2.4446706771850586,
      "learning_rate": 6.476470342542552e-06,
      "loss": 1.0701,
      "step": 10455
    },
    {
      "epoch": 0.8449463625527789,
      "grad_norm": 2.812384605407715,
      "learning_rate": 6.4758451481116014e-06,
      "loss": 1.0214,
      "step": 10456
    },
    {
      "epoch": 0.8450271722661064,
      "grad_norm": 3.1847965717315674,
      "learning_rate": 6.475219928402976e-06,
      "loss": 0.9729,
      "step": 10457
    },
    {
      "epoch": 0.8451079819794339,
      "grad_norm": 2.3405656814575195,
      "learning_rate": 6.474594683427385e-06,
      "loss": 1.065,
      "step": 10458
    },
    {
      "epoch": 0.8451887916927615,
      "grad_norm": 2.5131845474243164,
      "learning_rate": 6.4739694131955385e-06,
      "loss": 0.9468,
      "step": 10459
    },
    {
      "epoch": 0.845269601406089,
      "grad_norm": 2.65091609954834,
      "learning_rate": 6.4733441177181435e-06,
      "loss": 0.9437,
      "step": 10460
    },
    {
      "epoch": 0.8453504111194166,
      "grad_norm": 2.3615477085113525,
      "learning_rate": 6.47271879700591e-06,
      "loss": 0.8991,
      "step": 10461
    },
    {
      "epoch": 0.8454312208327441,
      "grad_norm": 2.9846079349517822,
      "learning_rate": 6.47209345106955e-06,
      "loss": 0.9782,
      "step": 10462
    },
    {
      "epoch": 0.8455120305460716,
      "grad_norm": 2.397481918334961,
      "learning_rate": 6.4714680799197725e-06,
      "loss": 0.9648,
      "step": 10463
    },
    {
      "epoch": 0.8455928402593992,
      "grad_norm": 3.003866672515869,
      "learning_rate": 6.47084268356729e-06,
      "loss": 0.961,
      "step": 10464
    },
    {
      "epoch": 0.8456736499727268,
      "grad_norm": 2.726534843444824,
      "learning_rate": 6.470217262022812e-06,
      "loss": 0.9583,
      "step": 10465
    },
    {
      "epoch": 0.8457544596860542,
      "grad_norm": 2.547591209411621,
      "learning_rate": 6.469591815297051e-06,
      "loss": 0.9565,
      "step": 10466
    },
    {
      "epoch": 0.8458352693993818,
      "grad_norm": 2.7158493995666504,
      "learning_rate": 6.4689663434007235e-06,
      "loss": 0.8202,
      "step": 10467
    },
    {
      "epoch": 0.8459160791127094,
      "grad_norm": 2.6833715438842773,
      "learning_rate": 6.468340846344536e-06,
      "loss": 0.8625,
      "step": 10468
    },
    {
      "epoch": 0.8459968888260369,
      "grad_norm": 2.635164499282837,
      "learning_rate": 6.4677153241392065e-06,
      "loss": 1.0071,
      "step": 10469
    },
    {
      "epoch": 0.8460776985393644,
      "grad_norm": 2.7561097145080566,
      "learning_rate": 6.467089776795446e-06,
      "loss": 0.8652,
      "step": 10470
    },
    {
      "epoch": 0.846158508252692,
      "grad_norm": 2.521637201309204,
      "learning_rate": 6.466464204323969e-06,
      "loss": 1.0617,
      "step": 10471
    },
    {
      "epoch": 0.8462393179660195,
      "grad_norm": 3.266026020050049,
      "learning_rate": 6.46583860673549e-06,
      "loss": 0.9337,
      "step": 10472
    },
    {
      "epoch": 0.8463201276793471,
      "grad_norm": 2.7632336616516113,
      "learning_rate": 6.465212984040727e-06,
      "loss": 0.9889,
      "step": 10473
    },
    {
      "epoch": 0.8464009373926746,
      "grad_norm": 2.851417064666748,
      "learning_rate": 6.464587336250389e-06,
      "loss": 1.0364,
      "step": 10474
    },
    {
      "epoch": 0.8464817471060021,
      "grad_norm": 2.3812038898468018,
      "learning_rate": 6.4639616633752e-06,
      "loss": 0.9583,
      "step": 10475
    },
    {
      "epoch": 0.8465625568193297,
      "grad_norm": 2.4048590660095215,
      "learning_rate": 6.463335965425871e-06,
      "loss": 1.0242,
      "step": 10476
    },
    {
      "epoch": 0.8466433665326573,
      "grad_norm": 2.4567837715148926,
      "learning_rate": 6.462710242413118e-06,
      "loss": 0.8045,
      "step": 10477
    },
    {
      "epoch": 0.8467241762459847,
      "grad_norm": 2.8058969974517822,
      "learning_rate": 6.4620844943476615e-06,
      "loss": 1.0603,
      "step": 10478
    },
    {
      "epoch": 0.8468049859593123,
      "grad_norm": 2.433835983276367,
      "learning_rate": 6.461458721240217e-06,
      "loss": 0.8385,
      "step": 10479
    },
    {
      "epoch": 0.8468857956726399,
      "grad_norm": 2.4225337505340576,
      "learning_rate": 6.460832923101502e-06,
      "loss": 0.9962,
      "step": 10480
    },
    {
      "epoch": 0.8469666053859674,
      "grad_norm": 3.16461181640625,
      "learning_rate": 6.460207099942237e-06,
      "loss": 0.8905,
      "step": 10481
    },
    {
      "epoch": 0.8470474150992949,
      "grad_norm": 3.0563158988952637,
      "learning_rate": 6.459581251773139e-06,
      "loss": 0.8266,
      "step": 10482
    },
    {
      "epoch": 0.8471282248126225,
      "grad_norm": 2.891674757003784,
      "learning_rate": 6.458955378604929e-06,
      "loss": 0.9402,
      "step": 10483
    },
    {
      "epoch": 0.84720903452595,
      "grad_norm": 2.4817230701446533,
      "learning_rate": 6.458329480448324e-06,
      "loss": 0.9309,
      "step": 10484
    },
    {
      "epoch": 0.8472898442392776,
      "grad_norm": 2.5383822917938232,
      "learning_rate": 6.457703557314048e-06,
      "loss": 0.937,
      "step": 10485
    },
    {
      "epoch": 0.8473706539526051,
      "grad_norm": 2.653319835662842,
      "learning_rate": 6.457077609212817e-06,
      "loss": 0.9991,
      "step": 10486
    },
    {
      "epoch": 0.8474514636659326,
      "grad_norm": 2.8143310546875,
      "learning_rate": 6.456451636155355e-06,
      "loss": 0.878,
      "step": 10487
    },
    {
      "epoch": 0.8475322733792602,
      "grad_norm": 2.7485921382904053,
      "learning_rate": 6.455825638152383e-06,
      "loss": 1.0788,
      "step": 10488
    },
    {
      "epoch": 0.8476130830925878,
      "grad_norm": 2.448368787765503,
      "learning_rate": 6.455199615214623e-06,
      "loss": 0.9665,
      "step": 10489
    },
    {
      "epoch": 0.8476938928059152,
      "grad_norm": 2.6947338581085205,
      "learning_rate": 6.454573567352797e-06,
      "loss": 0.8635,
      "step": 10490
    },
    {
      "epoch": 0.8477747025192428,
      "grad_norm": 2.86772084236145,
      "learning_rate": 6.453947494577627e-06,
      "loss": 0.8238,
      "step": 10491
    },
    {
      "epoch": 0.8478555122325704,
      "grad_norm": 3.0618245601654053,
      "learning_rate": 6.453321396899837e-06,
      "loss": 0.9415,
      "step": 10492
    },
    {
      "epoch": 0.8479363219458979,
      "grad_norm": 2.782890558242798,
      "learning_rate": 6.452695274330149e-06,
      "loss": 0.9395,
      "step": 10493
    },
    {
      "epoch": 0.8480171316592254,
      "grad_norm": 2.7398335933685303,
      "learning_rate": 6.452069126879289e-06,
      "loss": 0.8563,
      "step": 10494
    },
    {
      "epoch": 0.848097941372553,
      "grad_norm": 2.7994000911712646,
      "learning_rate": 6.451442954557981e-06,
      "loss": 1.0517,
      "step": 10495
    },
    {
      "epoch": 0.8481787510858805,
      "grad_norm": 2.945470094680786,
      "learning_rate": 6.450816757376949e-06,
      "loss": 0.8899,
      "step": 10496
    },
    {
      "epoch": 0.8482595607992081,
      "grad_norm": 2.6887476444244385,
      "learning_rate": 6.450190535346918e-06,
      "loss": 0.9508,
      "step": 10497
    },
    {
      "epoch": 0.8483403705125356,
      "grad_norm": 2.4166359901428223,
      "learning_rate": 6.449564288478616e-06,
      "loss": 0.9646,
      "step": 10498
    },
    {
      "epoch": 0.8484211802258631,
      "grad_norm": 2.211223602294922,
      "learning_rate": 6.448938016782766e-06,
      "loss": 0.9367,
      "step": 10499
    },
    {
      "epoch": 0.8485019899391907,
      "grad_norm": 2.7613167762756348,
      "learning_rate": 6.448311720270096e-06,
      "loss": 0.8813,
      "step": 10500
    },
    {
      "epoch": 0.8485827996525183,
      "grad_norm": 2.7050299644470215,
      "learning_rate": 6.447685398951333e-06,
      "loss": 0.936,
      "step": 10501
    },
    {
      "epoch": 0.8486636093658457,
      "grad_norm": 2.3954224586486816,
      "learning_rate": 6.4470590528372054e-06,
      "loss": 0.7174,
      "step": 10502
    },
    {
      "epoch": 0.8487444190791733,
      "grad_norm": 2.9367053508758545,
      "learning_rate": 6.446432681938439e-06,
      "loss": 0.8673,
      "step": 10503
    },
    {
      "epoch": 0.8488252287925009,
      "grad_norm": 2.6032369136810303,
      "learning_rate": 6.445806286265764e-06,
      "loss": 0.9011,
      "step": 10504
    },
    {
      "epoch": 0.8489060385058284,
      "grad_norm": 2.7029058933258057,
      "learning_rate": 6.445179865829905e-06,
      "loss": 0.8865,
      "step": 10505
    },
    {
      "epoch": 0.8489868482191559,
      "grad_norm": 2.5612471103668213,
      "learning_rate": 6.444553420641597e-06,
      "loss": 0.9992,
      "step": 10506
    },
    {
      "epoch": 0.8490676579324835,
      "grad_norm": 2.1173813343048096,
      "learning_rate": 6.443926950711564e-06,
      "loss": 0.9415,
      "step": 10507
    },
    {
      "epoch": 0.849148467645811,
      "grad_norm": 2.9662208557128906,
      "learning_rate": 6.4433004560505405e-06,
      "loss": 0.9405,
      "step": 10508
    },
    {
      "epoch": 0.8492292773591386,
      "grad_norm": 2.5690298080444336,
      "learning_rate": 6.442673936669255e-06,
      "loss": 0.886,
      "step": 10509
    },
    {
      "epoch": 0.8493100870724661,
      "grad_norm": 2.6806228160858154,
      "learning_rate": 6.4420473925784365e-06,
      "loss": 0.8541,
      "step": 10510
    },
    {
      "epoch": 0.8493908967857936,
      "grad_norm": 2.445883274078369,
      "learning_rate": 6.441420823788819e-06,
      "loss": 0.8864,
      "step": 10511
    },
    {
      "epoch": 0.8494717064991212,
      "grad_norm": 2.7138330936431885,
      "learning_rate": 6.440794230311133e-06,
      "loss": 0.9217,
      "step": 10512
    },
    {
      "epoch": 0.8495525162124488,
      "grad_norm": 2.6860594749450684,
      "learning_rate": 6.440167612156109e-06,
      "loss": 0.9295,
      "step": 10513
    },
    {
      "epoch": 0.8496333259257762,
      "grad_norm": 2.641309976577759,
      "learning_rate": 6.439540969334481e-06,
      "loss": 1.03,
      "step": 10514
    },
    {
      "epoch": 0.8497141356391038,
      "grad_norm": 2.674391031265259,
      "learning_rate": 6.4389143018569834e-06,
      "loss": 0.9904,
      "step": 10515
    },
    {
      "epoch": 0.8497949453524314,
      "grad_norm": 2.4982192516326904,
      "learning_rate": 6.438287609734346e-06,
      "loss": 1.0411,
      "step": 10516
    },
    {
      "epoch": 0.8498757550657589,
      "grad_norm": 2.8474903106689453,
      "learning_rate": 6.437660892977305e-06,
      "loss": 0.7843,
      "step": 10517
    },
    {
      "epoch": 0.8499565647790864,
      "grad_norm": 2.4621262550354004,
      "learning_rate": 6.437034151596595e-06,
      "loss": 0.8982,
      "step": 10518
    },
    {
      "epoch": 0.850037374492414,
      "grad_norm": 2.777421236038208,
      "learning_rate": 6.436407385602948e-06,
      "loss": 0.9847,
      "step": 10519
    },
    {
      "epoch": 0.8501181842057415,
      "grad_norm": 2.378206729888916,
      "learning_rate": 6.435780595007102e-06,
      "loss": 0.9996,
      "step": 10520
    },
    {
      "epoch": 0.8501989939190691,
      "grad_norm": 2.539928436279297,
      "learning_rate": 6.435153779819788e-06,
      "loss": 0.978,
      "step": 10521
    },
    {
      "epoch": 0.8502798036323966,
      "grad_norm": 2.253937244415283,
      "learning_rate": 6.4345269400517485e-06,
      "loss": 0.9378,
      "step": 10522
    },
    {
      "epoch": 0.8503606133457241,
      "grad_norm": 2.5191359519958496,
      "learning_rate": 6.433900075713714e-06,
      "loss": 0.8788,
      "step": 10523
    },
    {
      "epoch": 0.8504414230590517,
      "grad_norm": 2.7018587589263916,
      "learning_rate": 6.4332731868164235e-06,
      "loss": 0.8876,
      "step": 10524
    },
    {
      "epoch": 0.8505222327723793,
      "grad_norm": 2.474862813949585,
      "learning_rate": 6.432646273370613e-06,
      "loss": 0.939,
      "step": 10525
    },
    {
      "epoch": 0.8506030424857067,
      "grad_norm": 2.328767776489258,
      "learning_rate": 6.432019335387023e-06,
      "loss": 0.9886,
      "step": 10526
    },
    {
      "epoch": 0.8506838521990343,
      "grad_norm": 2.6595728397369385,
      "learning_rate": 6.431392372876386e-06,
      "loss": 0.9237,
      "step": 10527
    },
    {
      "epoch": 0.8507646619123619,
      "grad_norm": 2.5876286029815674,
      "learning_rate": 6.430765385849447e-06,
      "loss": 1.087,
      "step": 10528
    },
    {
      "epoch": 0.8508454716256894,
      "grad_norm": 2.5601816177368164,
      "learning_rate": 6.430138374316939e-06,
      "loss": 1.0598,
      "step": 10529
    },
    {
      "epoch": 0.8509262813390169,
      "grad_norm": 2.8048312664031982,
      "learning_rate": 6.429511338289604e-06,
      "loss": 0.9395,
      "step": 10530
    },
    {
      "epoch": 0.8510070910523445,
      "grad_norm": 2.326653480529785,
      "learning_rate": 6.428884277778183e-06,
      "loss": 0.9201,
      "step": 10531
    },
    {
      "epoch": 0.851087900765672,
      "grad_norm": 2.5945656299591064,
      "learning_rate": 6.428257192793411e-06,
      "loss": 0.8322,
      "step": 10532
    },
    {
      "epoch": 0.8511687104789996,
      "grad_norm": 2.785053253173828,
      "learning_rate": 6.427630083346033e-06,
      "loss": 0.9954,
      "step": 10533
    },
    {
      "epoch": 0.8512495201923271,
      "grad_norm": 2.6776323318481445,
      "learning_rate": 6.4270029494467904e-06,
      "loss": 0.8333,
      "step": 10534
    },
    {
      "epoch": 0.8513303299056547,
      "grad_norm": 2.6399784088134766,
      "learning_rate": 6.4263757911064195e-06,
      "loss": 0.9619,
      "step": 10535
    },
    {
      "epoch": 0.8514111396189822,
      "grad_norm": 2.997673511505127,
      "learning_rate": 6.425748608335668e-06,
      "loss": 0.984,
      "step": 10536
    },
    {
      "epoch": 0.8514919493323098,
      "grad_norm": 2.540785551071167,
      "learning_rate": 6.4251214011452735e-06,
      "loss": 0.8854,
      "step": 10537
    },
    {
      "epoch": 0.8515727590456373,
      "grad_norm": 2.7109200954437256,
      "learning_rate": 6.424494169545981e-06,
      "loss": 0.9228,
      "step": 10538
    },
    {
      "epoch": 0.8516535687589648,
      "grad_norm": 2.867309093475342,
      "learning_rate": 6.423866913548532e-06,
      "loss": 0.9037,
      "step": 10539
    },
    {
      "epoch": 0.8517343784722924,
      "grad_norm": 2.5105175971984863,
      "learning_rate": 6.423239633163673e-06,
      "loss": 0.8214,
      "step": 10540
    },
    {
      "epoch": 0.85181518818562,
      "grad_norm": 2.6527299880981445,
      "learning_rate": 6.4226123284021416e-06,
      "loss": 0.9401,
      "step": 10541
    },
    {
      "epoch": 0.8518959978989474,
      "grad_norm": 2.8117775917053223,
      "learning_rate": 6.4219849992746885e-06,
      "loss": 0.9948,
      "step": 10542
    },
    {
      "epoch": 0.851976807612275,
      "grad_norm": 2.792485475540161,
      "learning_rate": 6.421357645792054e-06,
      "loss": 0.981,
      "step": 10543
    },
    {
      "epoch": 0.8520576173256026,
      "grad_norm": 2.7991905212402344,
      "learning_rate": 6.4207302679649865e-06,
      "loss": 1.0453,
      "step": 10544
    },
    {
      "epoch": 0.8521384270389301,
      "grad_norm": 3.1634392738342285,
      "learning_rate": 6.420102865804228e-06,
      "loss": 0.9755,
      "step": 10545
    },
    {
      "epoch": 0.8522192367522576,
      "grad_norm": 2.360008716583252,
      "learning_rate": 6.419475439320527e-06,
      "loss": 0.9726,
      "step": 10546
    },
    {
      "epoch": 0.8523000464655852,
      "grad_norm": 2.892042875289917,
      "learning_rate": 6.418847988524629e-06,
      "loss": 0.9303,
      "step": 10547
    },
    {
      "epoch": 0.8523808561789127,
      "grad_norm": 2.1792070865631104,
      "learning_rate": 6.418220513427282e-06,
      "loss": 0.893,
      "step": 10548
    },
    {
      "epoch": 0.8524616658922403,
      "grad_norm": 2.467843770980835,
      "learning_rate": 6.417593014039229e-06,
      "loss": 0.9939,
      "step": 10549
    },
    {
      "epoch": 0.8525424756055678,
      "grad_norm": 2.352982997894287,
      "learning_rate": 6.416965490371223e-06,
      "loss": 0.8349,
      "step": 10550
    },
    {
      "epoch": 0.8526232853188953,
      "grad_norm": 2.815352439880371,
      "learning_rate": 6.4163379424340075e-06,
      "loss": 0.9593,
      "step": 10551
    },
    {
      "epoch": 0.8527040950322229,
      "grad_norm": 2.481914758682251,
      "learning_rate": 6.415710370238334e-06,
      "loss": 0.9265,
      "step": 10552
    },
    {
      "epoch": 0.8527849047455505,
      "grad_norm": 2.6599695682525635,
      "learning_rate": 6.41508277379495e-06,
      "loss": 1.0223,
      "step": 10553
    },
    {
      "epoch": 0.8528657144588779,
      "grad_norm": 2.587871789932251,
      "learning_rate": 6.414455153114604e-06,
      "loss": 1.0665,
      "step": 10554
    },
    {
      "epoch": 0.8529465241722055,
      "grad_norm": 2.924708604812622,
      "learning_rate": 6.413827508208046e-06,
      "loss": 0.8857,
      "step": 10555
    },
    {
      "epoch": 0.8530273338855331,
      "grad_norm": 2.524275064468384,
      "learning_rate": 6.413199839086029e-06,
      "loss": 0.9162,
      "step": 10556
    },
    {
      "epoch": 0.8531081435988606,
      "grad_norm": 3.098365306854248,
      "learning_rate": 6.4125721457592984e-06,
      "loss": 0.8304,
      "step": 10557
    },
    {
      "epoch": 0.8531889533121881,
      "grad_norm": 2.604714870452881,
      "learning_rate": 6.411944428238608e-06,
      "loss": 0.9379,
      "step": 10558
    },
    {
      "epoch": 0.8532697630255157,
      "grad_norm": 2.5910441875457764,
      "learning_rate": 6.411316686534709e-06,
      "loss": 1.085,
      "step": 10559
    },
    {
      "epoch": 0.8533505727388432,
      "grad_norm": 2.3200278282165527,
      "learning_rate": 6.410688920658352e-06,
      "loss": 0.9693,
      "step": 10560
    },
    {
      "epoch": 0.8534313824521708,
      "grad_norm": 2.490990161895752,
      "learning_rate": 6.410061130620291e-06,
      "loss": 0.9117,
      "step": 10561
    },
    {
      "epoch": 0.8535121921654983,
      "grad_norm": 2.962435245513916,
      "learning_rate": 6.409433316431276e-06,
      "loss": 0.8826,
      "step": 10562
    },
    {
      "epoch": 0.8535930018788258,
      "grad_norm": 2.776380777359009,
      "learning_rate": 6.4088054781020625e-06,
      "loss": 0.8119,
      "step": 10563
    },
    {
      "epoch": 0.8536738115921534,
      "grad_norm": 2.5014638900756836,
      "learning_rate": 6.4081776156434025e-06,
      "loss": 0.969,
      "step": 10564
    },
    {
      "epoch": 0.853754621305481,
      "grad_norm": 2.2608299255371094,
      "learning_rate": 6.40754972906605e-06,
      "loss": 0.9605,
      "step": 10565
    },
    {
      "epoch": 0.8538354310188084,
      "grad_norm": 2.999452590942383,
      "learning_rate": 6.4069218183807605e-06,
      "loss": 0.9088,
      "step": 10566
    },
    {
      "epoch": 0.853916240732136,
      "grad_norm": 2.4732658863067627,
      "learning_rate": 6.406293883598285e-06,
      "loss": 0.8978,
      "step": 10567
    },
    {
      "epoch": 0.8539970504454636,
      "grad_norm": 2.768444776535034,
      "learning_rate": 6.405665924729382e-06,
      "loss": 1.0506,
      "step": 10568
    },
    {
      "epoch": 0.8540778601587911,
      "grad_norm": 2.745413064956665,
      "learning_rate": 6.405037941784805e-06,
      "loss": 0.8907,
      "step": 10569
    },
    {
      "epoch": 0.8541586698721186,
      "grad_norm": 2.9249978065490723,
      "learning_rate": 6.404409934775311e-06,
      "loss": 0.9404,
      "step": 10570
    },
    {
      "epoch": 0.8542394795854462,
      "grad_norm": 3.0861124992370605,
      "learning_rate": 6.4037819037116564e-06,
      "loss": 1.026,
      "step": 10571
    },
    {
      "epoch": 0.8543202892987737,
      "grad_norm": 2.9700753688812256,
      "learning_rate": 6.4031538486045954e-06,
      "loss": 0.9146,
      "step": 10572
    },
    {
      "epoch": 0.8544010990121013,
      "grad_norm": 2.7021214962005615,
      "learning_rate": 6.402525769464889e-06,
      "loss": 0.8618,
      "step": 10573
    },
    {
      "epoch": 0.8544819087254288,
      "grad_norm": 2.956753730773926,
      "learning_rate": 6.401897666303291e-06,
      "loss": 1.0302,
      "step": 10574
    },
    {
      "epoch": 0.8545627184387563,
      "grad_norm": 3.2103488445281982,
      "learning_rate": 6.401269539130562e-06,
      "loss": 0.929,
      "step": 10575
    },
    {
      "epoch": 0.8546435281520839,
      "grad_norm": 2.6895854473114014,
      "learning_rate": 6.4006413879574594e-06,
      "loss": 0.9415,
      "step": 10576
    },
    {
      "epoch": 0.8547243378654115,
      "grad_norm": 2.826874256134033,
      "learning_rate": 6.400013212794741e-06,
      "loss": 0.9348,
      "step": 10577
    },
    {
      "epoch": 0.8548051475787389,
      "grad_norm": 2.940903902053833,
      "learning_rate": 6.399385013653166e-06,
      "loss": 0.8599,
      "step": 10578
    },
    {
      "epoch": 0.8548859572920665,
      "grad_norm": 2.674924373626709,
      "learning_rate": 6.398756790543498e-06,
      "loss": 0.9335,
      "step": 10579
    },
    {
      "epoch": 0.8549667670053941,
      "grad_norm": 2.5626094341278076,
      "learning_rate": 6.39812854347649e-06,
      "loss": 0.8634,
      "step": 10580
    },
    {
      "epoch": 0.8550475767187216,
      "grad_norm": 2.4545671939849854,
      "learning_rate": 6.397500272462906e-06,
      "loss": 0.8704,
      "step": 10581
    },
    {
      "epoch": 0.8551283864320491,
      "grad_norm": 3.2266173362731934,
      "learning_rate": 6.396871977513508e-06,
      "loss": 0.9394,
      "step": 10582
    },
    {
      "epoch": 0.8552091961453767,
      "grad_norm": 2.3609201908111572,
      "learning_rate": 6.396243658639056e-06,
      "loss": 0.92,
      "step": 10583
    },
    {
      "epoch": 0.8552900058587042,
      "grad_norm": 2.5651919841766357,
      "learning_rate": 6.395615315850311e-06,
      "loss": 1.0117,
      "step": 10584
    },
    {
      "epoch": 0.8553708155720318,
      "grad_norm": 2.9391074180603027,
      "learning_rate": 6.394986949158037e-06,
      "loss": 0.9646,
      "step": 10585
    },
    {
      "epoch": 0.8554516252853593,
      "grad_norm": 2.545461654663086,
      "learning_rate": 6.394358558572991e-06,
      "loss": 0.9538,
      "step": 10586
    },
    {
      "epoch": 0.8555324349986868,
      "grad_norm": 2.345226287841797,
      "learning_rate": 6.393730144105943e-06,
      "loss": 0.9813,
      "step": 10587
    },
    {
      "epoch": 0.8556132447120144,
      "grad_norm": 2.9139716625213623,
      "learning_rate": 6.39310170576765e-06,
      "loss": 0.7992,
      "step": 10588
    },
    {
      "epoch": 0.855694054425342,
      "grad_norm": 2.6127424240112305,
      "learning_rate": 6.3924732435688815e-06,
      "loss": 0.9688,
      "step": 10589
    },
    {
      "epoch": 0.8557748641386694,
      "grad_norm": 2.747901201248169,
      "learning_rate": 6.3918447575203975e-06,
      "loss": 0.925,
      "step": 10590
    },
    {
      "epoch": 0.855855673851997,
      "grad_norm": 2.4753477573394775,
      "learning_rate": 6.391216247632963e-06,
      "loss": 0.8328,
      "step": 10591
    },
    {
      "epoch": 0.8559364835653246,
      "grad_norm": 2.527433156967163,
      "learning_rate": 6.390587713917344e-06,
      "loss": 0.9202,
      "step": 10592
    },
    {
      "epoch": 0.8560172932786521,
      "grad_norm": 3.302743673324585,
      "learning_rate": 6.389959156384307e-06,
      "loss": 0.8891,
      "step": 10593
    },
    {
      "epoch": 0.8560981029919796,
      "grad_norm": 2.222712993621826,
      "learning_rate": 6.389330575044612e-06,
      "loss": 0.9005,
      "step": 10594
    },
    {
      "epoch": 0.8561789127053072,
      "grad_norm": 2.5155506134033203,
      "learning_rate": 6.388701969909032e-06,
      "loss": 1.0071,
      "step": 10595
    },
    {
      "epoch": 0.8562597224186347,
      "grad_norm": 2.667337417602539,
      "learning_rate": 6.388073340988329e-06,
      "loss": 0.9755,
      "step": 10596
    },
    {
      "epoch": 0.8563405321319623,
      "grad_norm": 2.794900894165039,
      "learning_rate": 6.387444688293272e-06,
      "loss": 0.9474,
      "step": 10597
    },
    {
      "epoch": 0.8564213418452898,
      "grad_norm": 2.404810905456543,
      "learning_rate": 6.386816011834627e-06,
      "loss": 0.9183,
      "step": 10598
    },
    {
      "epoch": 0.8565021515586173,
      "grad_norm": 2.6825006008148193,
      "learning_rate": 6.386187311623162e-06,
      "loss": 0.9499,
      "step": 10599
    },
    {
      "epoch": 0.8565829612719449,
      "grad_norm": 2.584521532058716,
      "learning_rate": 6.385558587669646e-06,
      "loss": 0.9438,
      "step": 10600
    },
    {
      "epoch": 0.8566637709852725,
      "grad_norm": 2.64414644241333,
      "learning_rate": 6.384929839984847e-06,
      "loss": 0.9696,
      "step": 10601
    },
    {
      "epoch": 0.8567445806985999,
      "grad_norm": 2.6944527626037598,
      "learning_rate": 6.384301068579532e-06,
      "loss": 0.8774,
      "step": 10602
    },
    {
      "epoch": 0.8568253904119275,
      "grad_norm": 2.4189651012420654,
      "learning_rate": 6.3836722734644765e-06,
      "loss": 1.0631,
      "step": 10603
    },
    {
      "epoch": 0.8569062001252551,
      "grad_norm": 2.6345582008361816,
      "learning_rate": 6.3830434546504425e-06,
      "loss": 0.9469,
      "step": 10604
    },
    {
      "epoch": 0.8569870098385826,
      "grad_norm": 2.7669363021850586,
      "learning_rate": 6.3824146121482055e-06,
      "loss": 0.9544,
      "step": 10605
    },
    {
      "epoch": 0.8570678195519101,
      "grad_norm": 2.6450963020324707,
      "learning_rate": 6.381785745968533e-06,
      "loss": 1.1208,
      "step": 10606
    },
    {
      "epoch": 0.8571486292652377,
      "grad_norm": 2.425590991973877,
      "learning_rate": 6.381156856122199e-06,
      "loss": 0.9461,
      "step": 10607
    },
    {
      "epoch": 0.8572294389785652,
      "grad_norm": 2.8030965328216553,
      "learning_rate": 6.380527942619971e-06,
      "loss": 0.9005,
      "step": 10608
    },
    {
      "epoch": 0.8573102486918928,
      "grad_norm": 2.6762855052948,
      "learning_rate": 6.379899005472624e-06,
      "loss": 0.978,
      "step": 10609
    },
    {
      "epoch": 0.8573910584052203,
      "grad_norm": 2.991990566253662,
      "learning_rate": 6.379270044690928e-06,
      "loss": 0.9275,
      "step": 10610
    },
    {
      "epoch": 0.8574718681185478,
      "grad_norm": 2.6101531982421875,
      "learning_rate": 6.378641060285657e-06,
      "loss": 0.9789,
      "step": 10611
    },
    {
      "epoch": 0.8575526778318754,
      "grad_norm": 2.340989828109741,
      "learning_rate": 6.3780120522675835e-06,
      "loss": 0.8883,
      "step": 10612
    },
    {
      "epoch": 0.857633487545203,
      "grad_norm": 3.0196053981781006,
      "learning_rate": 6.377383020647483e-06,
      "loss": 0.8825,
      "step": 10613
    },
    {
      "epoch": 0.8577142972585304,
      "grad_norm": 2.589050054550171,
      "learning_rate": 6.376753965436124e-06,
      "loss": 0.9256,
      "step": 10614
    },
    {
      "epoch": 0.857795106971858,
      "grad_norm": 2.5810017585754395,
      "learning_rate": 6.376124886644286e-06,
      "loss": 0.9272,
      "step": 10615
    },
    {
      "epoch": 0.8578759166851856,
      "grad_norm": 2.5934975147247314,
      "learning_rate": 6.375495784282741e-06,
      "loss": 1.0079,
      "step": 10616
    },
    {
      "epoch": 0.8579567263985131,
      "grad_norm": 2.4420084953308105,
      "learning_rate": 6.3748666583622655e-06,
      "loss": 0.9891,
      "step": 10617
    },
    {
      "epoch": 0.8580375361118406,
      "grad_norm": 2.4460995197296143,
      "learning_rate": 6.3742375088936325e-06,
      "loss": 0.9663,
      "step": 10618
    },
    {
      "epoch": 0.8581183458251682,
      "grad_norm": 2.510251760482788,
      "learning_rate": 6.37360833588762e-06,
      "loss": 0.903,
      "step": 10619
    },
    {
      "epoch": 0.8581991555384957,
      "grad_norm": 2.613154888153076,
      "learning_rate": 6.372979139355003e-06,
      "loss": 0.8368,
      "step": 10620
    },
    {
      "epoch": 0.8582799652518233,
      "grad_norm": 2.815469980239868,
      "learning_rate": 6.372349919306559e-06,
      "loss": 0.9071,
      "step": 10621
    },
    {
      "epoch": 0.8583607749651508,
      "grad_norm": 3.0432651042938232,
      "learning_rate": 6.371720675753065e-06,
      "loss": 0.909,
      "step": 10622
    },
    {
      "epoch": 0.8584415846784783,
      "grad_norm": 3.003157138824463,
      "learning_rate": 6.371091408705299e-06,
      "loss": 0.9256,
      "step": 10623
    },
    {
      "epoch": 0.8585223943918059,
      "grad_norm": 2.535301923751831,
      "learning_rate": 6.370462118174037e-06,
      "loss": 1.048,
      "step": 10624
    },
    {
      "epoch": 0.8586032041051335,
      "grad_norm": 2.4820950031280518,
      "learning_rate": 6.3698328041700575e-06,
      "loss": 1.0474,
      "step": 10625
    },
    {
      "epoch": 0.8586840138184609,
      "grad_norm": 3.022475242614746,
      "learning_rate": 6.369203466704141e-06,
      "loss": 0.943,
      "step": 10626
    },
    {
      "epoch": 0.8587648235317885,
      "grad_norm": 2.51517915725708,
      "learning_rate": 6.368574105787065e-06,
      "loss": 0.868,
      "step": 10627
    },
    {
      "epoch": 0.8588456332451161,
      "grad_norm": 2.576625347137451,
      "learning_rate": 6.367944721429608e-06,
      "loss": 0.9348,
      "step": 10628
    },
    {
      "epoch": 0.8589264429584436,
      "grad_norm": 2.760350227355957,
      "learning_rate": 6.3673153136425515e-06,
      "loss": 0.9838,
      "step": 10629
    },
    {
      "epoch": 0.8590072526717711,
      "grad_norm": 2.6769909858703613,
      "learning_rate": 6.366685882436676e-06,
      "loss": 0.8802,
      "step": 10630
    },
    {
      "epoch": 0.8590880623850987,
      "grad_norm": 2.7748525142669678,
      "learning_rate": 6.366056427822761e-06,
      "loss": 0.832,
      "step": 10631
    },
    {
      "epoch": 0.8591688720984262,
      "grad_norm": 3.260263681411743,
      "learning_rate": 6.365426949811589e-06,
      "loss": 1.0036,
      "step": 10632
    },
    {
      "epoch": 0.8592496818117538,
      "grad_norm": 2.397263526916504,
      "learning_rate": 6.3647974484139404e-06,
      "loss": 0.8667,
      "step": 10633
    },
    {
      "epoch": 0.8593304915250813,
      "grad_norm": 2.635941743850708,
      "learning_rate": 6.364167923640596e-06,
      "loss": 0.9143,
      "step": 10634
    },
    {
      "epoch": 0.8594113012384088,
      "grad_norm": 2.3416852951049805,
      "learning_rate": 6.363538375502341e-06,
      "loss": 0.9969,
      "step": 10635
    },
    {
      "epoch": 0.8594921109517364,
      "grad_norm": 2.5194973945617676,
      "learning_rate": 6.362908804009954e-06,
      "loss": 0.7951,
      "step": 10636
    },
    {
      "epoch": 0.859572920665064,
      "grad_norm": 2.572483777999878,
      "learning_rate": 6.3622792091742215e-06,
      "loss": 0.937,
      "step": 10637
    },
    {
      "epoch": 0.8596537303783914,
      "grad_norm": 2.6123387813568115,
      "learning_rate": 6.361649591005926e-06,
      "loss": 1.0149,
      "step": 10638
    },
    {
      "epoch": 0.859734540091719,
      "grad_norm": 2.845313787460327,
      "learning_rate": 6.36101994951585e-06,
      "loss": 0.8273,
      "step": 10639
    },
    {
      "epoch": 0.8598153498050466,
      "grad_norm": 2.2015128135681152,
      "learning_rate": 6.360390284714781e-06,
      "loss": 0.9506,
      "step": 10640
    },
    {
      "epoch": 0.8598961595183741,
      "grad_norm": 2.50612473487854,
      "learning_rate": 6.359760596613499e-06,
      "loss": 0.8745,
      "step": 10641
    },
    {
      "epoch": 0.8599769692317016,
      "grad_norm": 2.620586395263672,
      "learning_rate": 6.359130885222792e-06,
      "loss": 0.8983,
      "step": 10642
    },
    {
      "epoch": 0.8600577789450292,
      "grad_norm": 2.8248183727264404,
      "learning_rate": 6.358501150553444e-06,
      "loss": 0.9756,
      "step": 10643
    },
    {
      "epoch": 0.8601385886583567,
      "grad_norm": 2.6578304767608643,
      "learning_rate": 6.357871392616244e-06,
      "loss": 0.9275,
      "step": 10644
    },
    {
      "epoch": 0.8602193983716843,
      "grad_norm": 2.0792160034179688,
      "learning_rate": 6.357241611421975e-06,
      "loss": 0.9592,
      "step": 10645
    },
    {
      "epoch": 0.8603002080850118,
      "grad_norm": 2.506110429763794,
      "learning_rate": 6.356611806981425e-06,
      "loss": 0.7845,
      "step": 10646
    },
    {
      "epoch": 0.8603810177983393,
      "grad_norm": 2.1438324451446533,
      "learning_rate": 6.355981979305379e-06,
      "loss": 1.0082,
      "step": 10647
    },
    {
      "epoch": 0.8604618275116669,
      "grad_norm": 2.546485185623169,
      "learning_rate": 6.355352128404629e-06,
      "loss": 1.0375,
      "step": 10648
    },
    {
      "epoch": 0.8605426372249945,
      "grad_norm": 2.4790470600128174,
      "learning_rate": 6.354722254289958e-06,
      "loss": 0.8797,
      "step": 10649
    },
    {
      "epoch": 0.8606234469383219,
      "grad_norm": 2.379716157913208,
      "learning_rate": 6.3540923569721555e-06,
      "loss": 0.9556,
      "step": 10650
    },
    {
      "epoch": 0.8607042566516495,
      "grad_norm": 2.492772340774536,
      "learning_rate": 6.3534624364620115e-06,
      "loss": 1.019,
      "step": 10651
    },
    {
      "epoch": 0.8607850663649771,
      "grad_norm": 2.5661613941192627,
      "learning_rate": 6.352832492770316e-06,
      "loss": 0.9823,
      "step": 10652
    },
    {
      "epoch": 0.8608658760783046,
      "grad_norm": 2.65501070022583,
      "learning_rate": 6.352202525907854e-06,
      "loss": 0.8606,
      "step": 10653
    },
    {
      "epoch": 0.8609466857916321,
      "grad_norm": 2.75138783454895,
      "learning_rate": 6.35157253588542e-06,
      "loss": 0.8956,
      "step": 10654
    },
    {
      "epoch": 0.8610274955049597,
      "grad_norm": 2.963960886001587,
      "learning_rate": 6.350942522713799e-06,
      "loss": 1.0333,
      "step": 10655
    },
    {
      "epoch": 0.8611083052182872,
      "grad_norm": 2.7790400981903076,
      "learning_rate": 6.350312486403789e-06,
      "loss": 0.9164,
      "step": 10656
    },
    {
      "epoch": 0.8611891149316148,
      "grad_norm": 2.602954626083374,
      "learning_rate": 6.349682426966175e-06,
      "loss": 0.7521,
      "step": 10657
    },
    {
      "epoch": 0.8612699246449423,
      "grad_norm": 2.724419116973877,
      "learning_rate": 6.349052344411749e-06,
      "loss": 1.0504,
      "step": 10658
    },
    {
      "epoch": 0.8613507343582698,
      "grad_norm": 2.1897552013397217,
      "learning_rate": 6.348422238751306e-06,
      "loss": 1.0841,
      "step": 10659
    },
    {
      "epoch": 0.8614315440715974,
      "grad_norm": 2.358163595199585,
      "learning_rate": 6.347792109995636e-06,
      "loss": 1.062,
      "step": 10660
    },
    {
      "epoch": 0.861512353784925,
      "grad_norm": 2.888578176498413,
      "learning_rate": 6.34716195815553e-06,
      "loss": 0.9526,
      "step": 10661
    },
    {
      "epoch": 0.8615931634982524,
      "grad_norm": 2.2651138305664062,
      "learning_rate": 6.346531783241786e-06,
      "loss": 0.8792,
      "step": 10662
    },
    {
      "epoch": 0.86167397321158,
      "grad_norm": 2.759732246398926,
      "learning_rate": 6.3459015852651914e-06,
      "loss": 0.7242,
      "step": 10663
    },
    {
      "epoch": 0.8617547829249076,
      "grad_norm": 2.562382221221924,
      "learning_rate": 6.345271364236545e-06,
      "loss": 0.9803,
      "step": 10664
    },
    {
      "epoch": 0.8618355926382352,
      "grad_norm": 2.7765278816223145,
      "learning_rate": 6.344641120166637e-06,
      "loss": 0.9276,
      "step": 10665
    },
    {
      "epoch": 0.8619164023515626,
      "grad_norm": 2.386147975921631,
      "learning_rate": 6.344010853066265e-06,
      "loss": 0.9007,
      "step": 10666
    },
    {
      "epoch": 0.8619972120648902,
      "grad_norm": 2.420410394668579,
      "learning_rate": 6.343380562946221e-06,
      "loss": 0.8955,
      "step": 10667
    },
    {
      "epoch": 0.8620780217782178,
      "grad_norm": 2.726506471633911,
      "learning_rate": 6.342750249817303e-06,
      "loss": 0.8754,
      "step": 10668
    },
    {
      "epoch": 0.8621588314915453,
      "grad_norm": 2.5447473526000977,
      "learning_rate": 6.342119913690306e-06,
      "loss": 0.8546,
      "step": 10669
    },
    {
      "epoch": 0.8622396412048728,
      "grad_norm": 2.5017807483673096,
      "learning_rate": 6.341489554576026e-06,
      "loss": 0.8573,
      "step": 10670
    },
    {
      "epoch": 0.8623204509182004,
      "grad_norm": 2.4028189182281494,
      "learning_rate": 6.340859172485259e-06,
      "loss": 0.8024,
      "step": 10671
    },
    {
      "epoch": 0.8624012606315279,
      "grad_norm": 2.5131757259368896,
      "learning_rate": 6.3402287674288025e-06,
      "loss": 0.9333,
      "step": 10672
    },
    {
      "epoch": 0.8624820703448555,
      "grad_norm": 2.429248809814453,
      "learning_rate": 6.339598339417452e-06,
      "loss": 0.9257,
      "step": 10673
    },
    {
      "epoch": 0.862562880058183,
      "grad_norm": 2.963897228240967,
      "learning_rate": 6.33896788846201e-06,
      "loss": 0.8952,
      "step": 10674
    },
    {
      "epoch": 0.8626436897715105,
      "grad_norm": 2.5240468978881836,
      "learning_rate": 6.338337414573269e-06,
      "loss": 0.8979,
      "step": 10675
    },
    {
      "epoch": 0.8627244994848381,
      "grad_norm": 2.6063954830169678,
      "learning_rate": 6.337706917762032e-06,
      "loss": 0.9515,
      "step": 10676
    },
    {
      "epoch": 0.8628053091981657,
      "grad_norm": 2.316326379776001,
      "learning_rate": 6.337076398039095e-06,
      "loss": 0.9365,
      "step": 10677
    },
    {
      "epoch": 0.8628861189114931,
      "grad_norm": 2.583449125289917,
      "learning_rate": 6.336445855415257e-06,
      "loss": 0.7777,
      "step": 10678
    },
    {
      "epoch": 0.8629669286248207,
      "grad_norm": 2.994060516357422,
      "learning_rate": 6.33581528990132e-06,
      "loss": 0.9925,
      "step": 10679
    },
    {
      "epoch": 0.8630477383381483,
      "grad_norm": 2.4650230407714844,
      "learning_rate": 6.335184701508084e-06,
      "loss": 0.959,
      "step": 10680
    },
    {
      "epoch": 0.8631285480514758,
      "grad_norm": 2.5867812633514404,
      "learning_rate": 6.3345540902463456e-06,
      "loss": 0.8716,
      "step": 10681
    },
    {
      "epoch": 0.8632093577648033,
      "grad_norm": 2.8306491374969482,
      "learning_rate": 6.333923456126912e-06,
      "loss": 0.9326,
      "step": 10682
    },
    {
      "epoch": 0.8632901674781309,
      "grad_norm": 2.8601202964782715,
      "learning_rate": 6.333292799160578e-06,
      "loss": 0.9888,
      "step": 10683
    },
    {
      "epoch": 0.8633709771914584,
      "grad_norm": 2.715182065963745,
      "learning_rate": 6.332662119358149e-06,
      "loss": 0.8933,
      "step": 10684
    },
    {
      "epoch": 0.863451786904786,
      "grad_norm": 2.2428197860717773,
      "learning_rate": 6.332031416730426e-06,
      "loss": 0.9219,
      "step": 10685
    },
    {
      "epoch": 0.8635325966181135,
      "grad_norm": 2.675428628921509,
      "learning_rate": 6.331400691288212e-06,
      "loss": 0.9913,
      "step": 10686
    },
    {
      "epoch": 0.863613406331441,
      "grad_norm": 2.7280433177948,
      "learning_rate": 6.33076994304231e-06,
      "loss": 1.0188,
      "step": 10687
    },
    {
      "epoch": 0.8636942160447686,
      "grad_norm": 2.759230136871338,
      "learning_rate": 6.330139172003521e-06,
      "loss": 0.8432,
      "step": 10688
    },
    {
      "epoch": 0.8637750257580962,
      "grad_norm": 2.314729928970337,
      "learning_rate": 6.329508378182651e-06,
      "loss": 0.9584,
      "step": 10689
    },
    {
      "epoch": 0.8638558354714236,
      "grad_norm": 2.8474714756011963,
      "learning_rate": 6.328877561590503e-06,
      "loss": 0.9268,
      "step": 10690
    },
    {
      "epoch": 0.8639366451847512,
      "grad_norm": 2.595289707183838,
      "learning_rate": 6.328246722237882e-06,
      "loss": 0.9734,
      "step": 10691
    },
    {
      "epoch": 0.8640174548980788,
      "grad_norm": 2.816744089126587,
      "learning_rate": 6.327615860135592e-06,
      "loss": 1.0148,
      "step": 10692
    },
    {
      "epoch": 0.8640982646114063,
      "grad_norm": 3.141812801361084,
      "learning_rate": 6.326984975294438e-06,
      "loss": 1.0503,
      "step": 10693
    },
    {
      "epoch": 0.8641790743247338,
      "grad_norm": 2.314528226852417,
      "learning_rate": 6.326354067725226e-06,
      "loss": 0.9588,
      "step": 10694
    },
    {
      "epoch": 0.8642598840380614,
      "grad_norm": 2.6816062927246094,
      "learning_rate": 6.325723137438762e-06,
      "loss": 0.8807,
      "step": 10695
    },
    {
      "epoch": 0.8643406937513889,
      "grad_norm": 2.7583794593811035,
      "learning_rate": 6.325092184445852e-06,
      "loss": 0.9194,
      "step": 10696
    },
    {
      "epoch": 0.8644215034647165,
      "grad_norm": 2.718410015106201,
      "learning_rate": 6.3244612087573034e-06,
      "loss": 0.9402,
      "step": 10697
    },
    {
      "epoch": 0.864502313178044,
      "grad_norm": 2.365786552429199,
      "learning_rate": 6.323830210383922e-06,
      "loss": 0.8833,
      "step": 10698
    },
    {
      "epoch": 0.8645831228913715,
      "grad_norm": 2.589773416519165,
      "learning_rate": 6.323199189336517e-06,
      "loss": 0.9362,
      "step": 10699
    },
    {
      "epoch": 0.8646639326046991,
      "grad_norm": 2.79266619682312,
      "learning_rate": 6.322568145625896e-06,
      "loss": 0.953,
      "step": 10700
    },
    {
      "epoch": 0.8647447423180267,
      "grad_norm": 2.7299811840057373,
      "learning_rate": 6.321937079262866e-06,
      "loss": 0.8029,
      "step": 10701
    },
    {
      "epoch": 0.8648255520313541,
      "grad_norm": 2.4152779579162598,
      "learning_rate": 6.321305990258235e-06,
      "loss": 1.0027,
      "step": 10702
    },
    {
      "epoch": 0.8649063617446817,
      "grad_norm": 3.3846495151519775,
      "learning_rate": 6.320674878622815e-06,
      "loss": 0.8574,
      "step": 10703
    },
    {
      "epoch": 0.8649871714580093,
      "grad_norm": 2.5887768268585205,
      "learning_rate": 6.320043744367412e-06,
      "loss": 0.9724,
      "step": 10704
    },
    {
      "epoch": 0.8650679811713368,
      "grad_norm": 2.4225897789001465,
      "learning_rate": 6.31941258750284e-06,
      "loss": 0.8693,
      "step": 10705
    },
    {
      "epoch": 0.8651487908846643,
      "grad_norm": 2.640592575073242,
      "learning_rate": 6.318781408039904e-06,
      "loss": 1.0469,
      "step": 10706
    },
    {
      "epoch": 0.8652296005979919,
      "grad_norm": 2.973048210144043,
      "learning_rate": 6.31815020598942e-06,
      "loss": 1.0066,
      "step": 10707
    },
    {
      "epoch": 0.8653104103113194,
      "grad_norm": 2.6845743656158447,
      "learning_rate": 6.317518981362194e-06,
      "loss": 0.9655,
      "step": 10708
    },
    {
      "epoch": 0.865391220024647,
      "grad_norm": 2.4086756706237793,
      "learning_rate": 6.31688773416904e-06,
      "loss": 0.9585,
      "step": 10709
    },
    {
      "epoch": 0.8654720297379745,
      "grad_norm": 2.196768283843994,
      "learning_rate": 6.31625646442077e-06,
      "loss": 0.9601,
      "step": 10710
    },
    {
      "epoch": 0.865552839451302,
      "grad_norm": 3.2099034786224365,
      "learning_rate": 6.315625172128195e-06,
      "loss": 0.9698,
      "step": 10711
    },
    {
      "epoch": 0.8656336491646296,
      "grad_norm": 2.5552661418914795,
      "learning_rate": 6.314993857302129e-06,
      "loss": 1.1014,
      "step": 10712
    },
    {
      "epoch": 0.8657144588779572,
      "grad_norm": 3.1415045261383057,
      "learning_rate": 6.314362519953384e-06,
      "loss": 0.9447,
      "step": 10713
    },
    {
      "epoch": 0.8657952685912846,
      "grad_norm": 2.824753522872925,
      "learning_rate": 6.313731160092771e-06,
      "loss": 0.9919,
      "step": 10714
    },
    {
      "epoch": 0.8658760783046122,
      "grad_norm": 2.9019217491149902,
      "learning_rate": 6.313099777731109e-06,
      "loss": 0.9108,
      "step": 10715
    },
    {
      "epoch": 0.8659568880179398,
      "grad_norm": 2.5668866634368896,
      "learning_rate": 6.312468372879207e-06,
      "loss": 0.9482,
      "step": 10716
    },
    {
      "epoch": 0.8660376977312673,
      "grad_norm": 2.8544135093688965,
      "learning_rate": 6.311836945547882e-06,
      "loss": 1.0704,
      "step": 10717
    },
    {
      "epoch": 0.8661185074445948,
      "grad_norm": 2.715951442718506,
      "learning_rate": 6.311205495747947e-06,
      "loss": 0.9377,
      "step": 10718
    },
    {
      "epoch": 0.8661993171579224,
      "grad_norm": 2.9735267162323,
      "learning_rate": 6.310574023490222e-06,
      "loss": 0.8002,
      "step": 10719
    },
    {
      "epoch": 0.8662801268712499,
      "grad_norm": 2.4859063625335693,
      "learning_rate": 6.309942528785515e-06,
      "loss": 0.9082,
      "step": 10720
    },
    {
      "epoch": 0.8663609365845775,
      "grad_norm": 2.8067026138305664,
      "learning_rate": 6.309311011644649e-06,
      "loss": 0.9845,
      "step": 10721
    },
    {
      "epoch": 0.866441746297905,
      "grad_norm": 2.8802614212036133,
      "learning_rate": 6.3086794720784364e-06,
      "loss": 1.081,
      "step": 10722
    },
    {
      "epoch": 0.8665225560112325,
      "grad_norm": 2.5529253482818604,
      "learning_rate": 6.308047910097694e-06,
      "loss": 0.8574,
      "step": 10723
    },
    {
      "epoch": 0.8666033657245601,
      "grad_norm": 2.744807481765747,
      "learning_rate": 6.3074163257132405e-06,
      "loss": 0.8429,
      "step": 10724
    },
    {
      "epoch": 0.8666841754378877,
      "grad_norm": 2.576052188873291,
      "learning_rate": 6.306784718935892e-06,
      "loss": 0.9237,
      "step": 10725
    },
    {
      "epoch": 0.8667649851512151,
      "grad_norm": 2.367292881011963,
      "learning_rate": 6.306153089776468e-06,
      "loss": 0.9179,
      "step": 10726
    },
    {
      "epoch": 0.8668457948645427,
      "grad_norm": 3.018024444580078,
      "learning_rate": 6.305521438245788e-06,
      "loss": 0.9679,
      "step": 10727
    },
    {
      "epoch": 0.8669266045778703,
      "grad_norm": 2.719452142715454,
      "learning_rate": 6.304889764354665e-06,
      "loss": 0.782,
      "step": 10728
    },
    {
      "epoch": 0.8670074142911978,
      "grad_norm": 2.57442045211792,
      "learning_rate": 6.304258068113924e-06,
      "loss": 0.9226,
      "step": 10729
    },
    {
      "epoch": 0.8670882240045253,
      "grad_norm": 2.4634058475494385,
      "learning_rate": 6.303626349534382e-06,
      "loss": 0.86,
      "step": 10730
    },
    {
      "epoch": 0.8671690337178529,
      "grad_norm": 2.393235921859741,
      "learning_rate": 6.302994608626859e-06,
      "loss": 0.9199,
      "step": 10731
    },
    {
      "epoch": 0.8672498434311804,
      "grad_norm": 2.4220023155212402,
      "learning_rate": 6.3023628454021744e-06,
      "loss": 1.0465,
      "step": 10732
    },
    {
      "epoch": 0.867330653144508,
      "grad_norm": 3.016916036605835,
      "learning_rate": 6.301731059871151e-06,
      "loss": 0.8538,
      "step": 10733
    },
    {
      "epoch": 0.8674114628578355,
      "grad_norm": 2.7756950855255127,
      "learning_rate": 6.3010992520446055e-06,
      "loss": 0.8465,
      "step": 10734
    },
    {
      "epoch": 0.867492272571163,
      "grad_norm": 2.587876319885254,
      "learning_rate": 6.300467421933365e-06,
      "loss": 0.8998,
      "step": 10735
    },
    {
      "epoch": 0.8675730822844906,
      "grad_norm": 2.5709636211395264,
      "learning_rate": 6.299835569548247e-06,
      "loss": 0.8722,
      "step": 10736
    },
    {
      "epoch": 0.8676538919978182,
      "grad_norm": 3.024399757385254,
      "learning_rate": 6.299203694900076e-06,
      "loss": 0.9851,
      "step": 10737
    },
    {
      "epoch": 0.8677347017111456,
      "grad_norm": 2.966047763824463,
      "learning_rate": 6.298571797999672e-06,
      "loss": 0.8631,
      "step": 10738
    },
    {
      "epoch": 0.8678155114244732,
      "grad_norm": 2.9161622524261475,
      "learning_rate": 6.297939878857859e-06,
      "loss": 0.846,
      "step": 10739
    },
    {
      "epoch": 0.8678963211378008,
      "grad_norm": 2.444450616836548,
      "learning_rate": 6.297307937485462e-06,
      "loss": 0.8635,
      "step": 10740
    },
    {
      "epoch": 0.8679771308511283,
      "grad_norm": 2.7127749919891357,
      "learning_rate": 6.296675973893304e-06,
      "loss": 0.8632,
      "step": 10741
    },
    {
      "epoch": 0.8680579405644558,
      "grad_norm": 2.566418409347534,
      "learning_rate": 6.296043988092205e-06,
      "loss": 0.8537,
      "step": 10742
    },
    {
      "epoch": 0.8681387502777834,
      "grad_norm": 2.485527753829956,
      "learning_rate": 6.2954119800929955e-06,
      "loss": 0.9469,
      "step": 10743
    },
    {
      "epoch": 0.8682195599911109,
      "grad_norm": 2.256040573120117,
      "learning_rate": 6.294779949906496e-06,
      "loss": 0.969,
      "step": 10744
    },
    {
      "epoch": 0.8683003697044385,
      "grad_norm": 2.8519866466522217,
      "learning_rate": 6.2941478975435334e-06,
      "loss": 1.0519,
      "step": 10745
    },
    {
      "epoch": 0.868381179417766,
      "grad_norm": 3.0991079807281494,
      "learning_rate": 6.293515823014933e-06,
      "loss": 1.1046,
      "step": 10746
    },
    {
      "epoch": 0.8684619891310935,
      "grad_norm": 2.535867691040039,
      "learning_rate": 6.29288372633152e-06,
      "loss": 0.8546,
      "step": 10747
    },
    {
      "epoch": 0.8685427988444211,
      "grad_norm": 2.7065234184265137,
      "learning_rate": 6.29225160750412e-06,
      "loss": 0.8375,
      "step": 10748
    },
    {
      "epoch": 0.8686236085577487,
      "grad_norm": 2.5868163108825684,
      "learning_rate": 6.291619466543564e-06,
      "loss": 0.8276,
      "step": 10749
    },
    {
      "epoch": 0.8687044182710761,
      "grad_norm": 2.6736459732055664,
      "learning_rate": 6.290987303460674e-06,
      "loss": 0.9697,
      "step": 10750
    },
    {
      "epoch": 0.8687852279844037,
      "grad_norm": 2.3866159915924072,
      "learning_rate": 6.29035511826628e-06,
      "loss": 0.9777,
      "step": 10751
    },
    {
      "epoch": 0.8688660376977313,
      "grad_norm": 2.549373149871826,
      "learning_rate": 6.289722910971208e-06,
      "loss": 0.7991,
      "step": 10752
    },
    {
      "epoch": 0.8689468474110588,
      "grad_norm": 2.6806037425994873,
      "learning_rate": 6.289090681586289e-06,
      "loss": 0.9878,
      "step": 10753
    },
    {
      "epoch": 0.8690276571243863,
      "grad_norm": 3.136216640472412,
      "learning_rate": 6.28845843012235e-06,
      "loss": 0.8673,
      "step": 10754
    },
    {
      "epoch": 0.8691084668377139,
      "grad_norm": 2.7405874729156494,
      "learning_rate": 6.287826156590219e-06,
      "loss": 0.9072,
      "step": 10755
    },
    {
      "epoch": 0.8691892765510414,
      "grad_norm": 3.0692496299743652,
      "learning_rate": 6.287193861000727e-06,
      "loss": 0.8264,
      "step": 10756
    },
    {
      "epoch": 0.869270086264369,
      "grad_norm": 2.371486186981201,
      "learning_rate": 6.286561543364703e-06,
      "loss": 0.9797,
      "step": 10757
    },
    {
      "epoch": 0.8693508959776965,
      "grad_norm": 2.7062017917633057,
      "learning_rate": 6.285929203692977e-06,
      "loss": 0.9089,
      "step": 10758
    },
    {
      "epoch": 0.869431705691024,
      "grad_norm": 2.3986001014709473,
      "learning_rate": 6.285296841996378e-06,
      "loss": 0.9209,
      "step": 10759
    },
    {
      "epoch": 0.8695125154043516,
      "grad_norm": 2.299267530441284,
      "learning_rate": 6.2846644582857396e-06,
      "loss": 0.9217,
      "step": 10760
    },
    {
      "epoch": 0.8695933251176792,
      "grad_norm": 2.544961929321289,
      "learning_rate": 6.284032052571891e-06,
      "loss": 1.0939,
      "step": 10761
    },
    {
      "epoch": 0.8696741348310066,
      "grad_norm": 2.4752156734466553,
      "learning_rate": 6.283399624865666e-06,
      "loss": 0.9844,
      "step": 10762
    },
    {
      "epoch": 0.8697549445443342,
      "grad_norm": 2.646388530731201,
      "learning_rate": 6.282767175177893e-06,
      "loss": 0.8684,
      "step": 10763
    },
    {
      "epoch": 0.8698357542576618,
      "grad_norm": 3.183969259262085,
      "learning_rate": 6.282134703519408e-06,
      "loss": 0.8831,
      "step": 10764
    },
    {
      "epoch": 0.8699165639709893,
      "grad_norm": 2.540184497833252,
      "learning_rate": 6.281502209901043e-06,
      "loss": 0.8108,
      "step": 10765
    },
    {
      "epoch": 0.8699973736843168,
      "grad_norm": 2.5926268100738525,
      "learning_rate": 6.280869694333629e-06,
      "loss": 0.9253,
      "step": 10766
    },
    {
      "epoch": 0.8700781833976444,
      "grad_norm": 2.694242000579834,
      "learning_rate": 6.280237156827999e-06,
      "loss": 0.8878,
      "step": 10767
    },
    {
      "epoch": 0.8701589931109719,
      "grad_norm": 2.509291410446167,
      "learning_rate": 6.279604597394991e-06,
      "loss": 0.8167,
      "step": 10768
    },
    {
      "epoch": 0.8702398028242995,
      "grad_norm": 2.694880962371826,
      "learning_rate": 6.278972016045436e-06,
      "loss": 0.9564,
      "step": 10769
    },
    {
      "epoch": 0.870320612537627,
      "grad_norm": 2.6930549144744873,
      "learning_rate": 6.278339412790169e-06,
      "loss": 1.0387,
      "step": 10770
    },
    {
      "epoch": 0.8704014222509545,
      "grad_norm": 2.729775905609131,
      "learning_rate": 6.277706787640025e-06,
      "loss": 0.7416,
      "step": 10771
    },
    {
      "epoch": 0.8704822319642821,
      "grad_norm": 2.8160128593444824,
      "learning_rate": 6.277074140605842e-06,
      "loss": 1.0056,
      "step": 10772
    },
    {
      "epoch": 0.8705630416776097,
      "grad_norm": 2.5572054386138916,
      "learning_rate": 6.2764414716984514e-06,
      "loss": 0.9884,
      "step": 10773
    },
    {
      "epoch": 0.8706438513909371,
      "grad_norm": 2.755920886993408,
      "learning_rate": 6.275808780928691e-06,
      "loss": 1.0284,
      "step": 10774
    },
    {
      "epoch": 0.8707246611042647,
      "grad_norm": 2.579279661178589,
      "learning_rate": 6.275176068307399e-06,
      "loss": 0.8889,
      "step": 10775
    },
    {
      "epoch": 0.8708054708175923,
      "grad_norm": 2.5251824855804443,
      "learning_rate": 6.27454333384541e-06,
      "loss": 0.9342,
      "step": 10776
    },
    {
      "epoch": 0.8708862805309198,
      "grad_norm": 2.084838628768921,
      "learning_rate": 6.273910577553561e-06,
      "loss": 0.9613,
      "step": 10777
    },
    {
      "epoch": 0.8709670902442473,
      "grad_norm": 2.7771565914154053,
      "learning_rate": 6.273277799442692e-06,
      "loss": 1.0266,
      "step": 10778
    },
    {
      "epoch": 0.8710478999575749,
      "grad_norm": 2.3363239765167236,
      "learning_rate": 6.272644999523639e-06,
      "loss": 0.8887,
      "step": 10779
    },
    {
      "epoch": 0.8711287096709024,
      "grad_norm": 2.9889473915100098,
      "learning_rate": 6.272012177807243e-06,
      "loss": 1.011,
      "step": 10780
    },
    {
      "epoch": 0.87120951938423,
      "grad_norm": 2.5180137157440186,
      "learning_rate": 6.271379334304337e-06,
      "loss": 1.037,
      "step": 10781
    },
    {
      "epoch": 0.8712903290975575,
      "grad_norm": 2.5055792331695557,
      "learning_rate": 6.270746469025767e-06,
      "loss": 0.8885,
      "step": 10782
    },
    {
      "epoch": 0.871371138810885,
      "grad_norm": 2.5165748596191406,
      "learning_rate": 6.270113581982368e-06,
      "loss": 0.9617,
      "step": 10783
    },
    {
      "epoch": 0.8714519485242126,
      "grad_norm": 2.2549095153808594,
      "learning_rate": 6.26948067318498e-06,
      "loss": 0.9547,
      "step": 10784
    },
    {
      "epoch": 0.8715327582375402,
      "grad_norm": 2.5069286823272705,
      "learning_rate": 6.268847742644445e-06,
      "loss": 0.8673,
      "step": 10785
    },
    {
      "epoch": 0.8716135679508676,
      "grad_norm": 2.8266592025756836,
      "learning_rate": 6.2682147903716036e-06,
      "loss": 0.8074,
      "step": 10786
    },
    {
      "epoch": 0.8716943776641952,
      "grad_norm": 2.5912058353424072,
      "learning_rate": 6.267581816377294e-06,
      "loss": 0.9483,
      "step": 10787
    },
    {
      "epoch": 0.8717751873775228,
      "grad_norm": 2.6459462642669678,
      "learning_rate": 6.266948820672362e-06,
      "loss": 1.0598,
      "step": 10788
    },
    {
      "epoch": 0.8718559970908503,
      "grad_norm": 2.4300882816314697,
      "learning_rate": 6.266315803267645e-06,
      "loss": 0.9263,
      "step": 10789
    },
    {
      "epoch": 0.8719368068041778,
      "grad_norm": 2.8088088035583496,
      "learning_rate": 6.265682764173987e-06,
      "loss": 0.8896,
      "step": 10790
    },
    {
      "epoch": 0.8720176165175054,
      "grad_norm": 3.032182455062866,
      "learning_rate": 6.26504970340223e-06,
      "loss": 0.9081,
      "step": 10791
    },
    {
      "epoch": 0.872098426230833,
      "grad_norm": 2.6282832622528076,
      "learning_rate": 6.264416620963217e-06,
      "loss": 0.9339,
      "step": 10792
    },
    {
      "epoch": 0.8721792359441605,
      "grad_norm": 2.961758613586426,
      "learning_rate": 6.263783516867792e-06,
      "loss": 0.9128,
      "step": 10793
    },
    {
      "epoch": 0.872260045657488,
      "grad_norm": 2.9102795124053955,
      "learning_rate": 6.263150391126799e-06,
      "loss": 1.1049,
      "step": 10794
    },
    {
      "epoch": 0.8723408553708156,
      "grad_norm": 2.5261833667755127,
      "learning_rate": 6.262517243751078e-06,
      "loss": 0.913,
      "step": 10795
    },
    {
      "epoch": 0.8724216650841431,
      "grad_norm": 2.563030958175659,
      "learning_rate": 6.261884074751479e-06,
      "loss": 0.9188,
      "step": 10796
    },
    {
      "epoch": 0.8725024747974707,
      "grad_norm": 2.705185651779175,
      "learning_rate": 6.261250884138841e-06,
      "loss": 0.9966,
      "step": 10797
    },
    {
      "epoch": 0.8725832845107983,
      "grad_norm": 2.666682481765747,
      "learning_rate": 6.260617671924012e-06,
      "loss": 0.7857,
      "step": 10798
    },
    {
      "epoch": 0.8726640942241257,
      "grad_norm": 2.5954337120056152,
      "learning_rate": 6.259984438117837e-06,
      "loss": 0.8968,
      "step": 10799
    },
    {
      "epoch": 0.8727449039374533,
      "grad_norm": 2.9078593254089355,
      "learning_rate": 6.259351182731164e-06,
      "loss": 0.9886,
      "step": 10800
    },
    {
      "epoch": 0.8728257136507809,
      "grad_norm": 2.364166498184204,
      "learning_rate": 6.258717905774835e-06,
      "loss": 0.8949,
      "step": 10801
    },
    {
      "epoch": 0.8729065233641083,
      "grad_norm": 2.4891860485076904,
      "learning_rate": 6.2580846072597e-06,
      "loss": 0.9635,
      "step": 10802
    },
    {
      "epoch": 0.8729873330774359,
      "grad_norm": 2.585794687271118,
      "learning_rate": 6.257451287196603e-06,
      "loss": 0.993,
      "step": 10803
    },
    {
      "epoch": 0.8730681427907635,
      "grad_norm": 2.3488502502441406,
      "learning_rate": 6.256817945596393e-06,
      "loss": 0.8275,
      "step": 10804
    },
    {
      "epoch": 0.873148952504091,
      "grad_norm": 2.637300491333008,
      "learning_rate": 6.2561845824699175e-06,
      "loss": 0.943,
      "step": 10805
    },
    {
      "epoch": 0.8732297622174185,
      "grad_norm": 2.2985503673553467,
      "learning_rate": 6.255551197828024e-06,
      "loss": 0.8112,
      "step": 10806
    },
    {
      "epoch": 0.8733105719307461,
      "grad_norm": 3.118016481399536,
      "learning_rate": 6.254917791681561e-06,
      "loss": 0.8169,
      "step": 10807
    },
    {
      "epoch": 0.8733913816440736,
      "grad_norm": 2.5043869018554688,
      "learning_rate": 6.2542843640413784e-06,
      "loss": 0.9848,
      "step": 10808
    },
    {
      "epoch": 0.8734721913574012,
      "grad_norm": 2.850107192993164,
      "learning_rate": 6.253650914918322e-06,
      "loss": 0.911,
      "step": 10809
    },
    {
      "epoch": 0.8735530010707288,
      "grad_norm": 2.5064802169799805,
      "learning_rate": 6.253017444323246e-06,
      "loss": 0.9335,
      "step": 10810
    },
    {
      "epoch": 0.8736338107840562,
      "grad_norm": 3.565614700317383,
      "learning_rate": 6.252383952266996e-06,
      "loss": 1.0437,
      "step": 10811
    },
    {
      "epoch": 0.8737146204973838,
      "grad_norm": 2.5139176845550537,
      "learning_rate": 6.251750438760425e-06,
      "loss": 0.8993,
      "step": 10812
    },
    {
      "epoch": 0.8737954302107114,
      "grad_norm": 2.882888078689575,
      "learning_rate": 6.25111690381438e-06,
      "loss": 0.9241,
      "step": 10813
    },
    {
      "epoch": 0.8738762399240388,
      "grad_norm": 2.9219565391540527,
      "learning_rate": 6.250483347439717e-06,
      "loss": 0.9589,
      "step": 10814
    },
    {
      "epoch": 0.8739570496373664,
      "grad_norm": 2.3526394367218018,
      "learning_rate": 6.249849769647283e-06,
      "loss": 0.8644,
      "step": 10815
    },
    {
      "epoch": 0.874037859350694,
      "grad_norm": 2.953166961669922,
      "learning_rate": 6.249216170447934e-06,
      "loss": 0.9873,
      "step": 10816
    },
    {
      "epoch": 0.8741186690640215,
      "grad_norm": 2.7625043392181396,
      "learning_rate": 6.248582549852517e-06,
      "loss": 0.8488,
      "step": 10817
    },
    {
      "epoch": 0.874199478777349,
      "grad_norm": 2.5680489540100098,
      "learning_rate": 6.247948907871888e-06,
      "loss": 0.8417,
      "step": 10818
    },
    {
      "epoch": 0.8742802884906766,
      "grad_norm": 2.6536850929260254,
      "learning_rate": 6.247315244516897e-06,
      "loss": 0.8816,
      "step": 10819
    },
    {
      "epoch": 0.8743610982040041,
      "grad_norm": 3.0687880516052246,
      "learning_rate": 6.2466815597984e-06,
      "loss": 0.9783,
      "step": 10820
    },
    {
      "epoch": 0.8744419079173317,
      "grad_norm": 2.711142063140869,
      "learning_rate": 6.246047853727249e-06,
      "loss": 0.8074,
      "step": 10821
    },
    {
      "epoch": 0.8745227176306593,
      "grad_norm": 2.937854766845703,
      "learning_rate": 6.245414126314297e-06,
      "loss": 0.8972,
      "step": 10822
    },
    {
      "epoch": 0.8746035273439867,
      "grad_norm": 2.8597121238708496,
      "learning_rate": 6.2447803775704e-06,
      "loss": 0.9454,
      "step": 10823
    },
    {
      "epoch": 0.8746843370573143,
      "grad_norm": 2.685940742492676,
      "learning_rate": 6.244146607506412e-06,
      "loss": 0.9485,
      "step": 10824
    },
    {
      "epoch": 0.8747651467706419,
      "grad_norm": 2.585789680480957,
      "learning_rate": 6.243512816133188e-06,
      "loss": 0.9593,
      "step": 10825
    },
    {
      "epoch": 0.8748459564839693,
      "grad_norm": 2.7482876777648926,
      "learning_rate": 6.2428790034615814e-06,
      "loss": 0.9468,
      "step": 10826
    },
    {
      "epoch": 0.8749267661972969,
      "grad_norm": 2.5974881649017334,
      "learning_rate": 6.2422451695024515e-06,
      "loss": 0.8735,
      "step": 10827
    },
    {
      "epoch": 0.8750075759106245,
      "grad_norm": 2.5537526607513428,
      "learning_rate": 6.241611314266651e-06,
      "loss": 0.9705,
      "step": 10828
    },
    {
      "epoch": 0.875088385623952,
      "grad_norm": 2.82592511177063,
      "learning_rate": 6.240977437765039e-06,
      "loss": 1.0177,
      "step": 10829
    },
    {
      "epoch": 0.8751691953372795,
      "grad_norm": 2.7693116664886475,
      "learning_rate": 6.24034354000847e-06,
      "loss": 0.8391,
      "step": 10830
    },
    {
      "epoch": 0.8752500050506071,
      "grad_norm": 2.3902838230133057,
      "learning_rate": 6.2397096210078035e-06,
      "loss": 0.8922,
      "step": 10831
    },
    {
      "epoch": 0.8753308147639346,
      "grad_norm": 2.4729909896850586,
      "learning_rate": 6.239075680773895e-06,
      "loss": 1.0435,
      "step": 10832
    },
    {
      "epoch": 0.8754116244772622,
      "grad_norm": 2.984990358352661,
      "learning_rate": 6.238441719317603e-06,
      "loss": 0.9861,
      "step": 10833
    },
    {
      "epoch": 0.8754924341905898,
      "grad_norm": 3.044724225997925,
      "learning_rate": 6.237807736649784e-06,
      "loss": 0.8173,
      "step": 10834
    },
    {
      "epoch": 0.8755732439039172,
      "grad_norm": 2.6723055839538574,
      "learning_rate": 6.237173732781301e-06,
      "loss": 1.0277,
      "step": 10835
    },
    {
      "epoch": 0.8756540536172448,
      "grad_norm": 2.329822063446045,
      "learning_rate": 6.236539707723008e-06,
      "loss": 0.942,
      "step": 10836
    },
    {
      "epoch": 0.8757348633305724,
      "grad_norm": 2.3134264945983887,
      "learning_rate": 6.235905661485768e-06,
      "loss": 0.9642,
      "step": 10837
    },
    {
      "epoch": 0.8758156730438998,
      "grad_norm": 2.7943315505981445,
      "learning_rate": 6.23527159408044e-06,
      "loss": 0.9402,
      "step": 10838
    },
    {
      "epoch": 0.8758964827572274,
      "grad_norm": 2.4493045806884766,
      "learning_rate": 6.234637505517883e-06,
      "loss": 1.0155,
      "step": 10839
    },
    {
      "epoch": 0.875977292470555,
      "grad_norm": 2.632335901260376,
      "learning_rate": 6.234003395808956e-06,
      "loss": 0.9214,
      "step": 10840
    },
    {
      "epoch": 0.8760581021838825,
      "grad_norm": 2.8725292682647705,
      "learning_rate": 6.233369264964525e-06,
      "loss": 0.8162,
      "step": 10841
    },
    {
      "epoch": 0.87613891189721,
      "grad_norm": 2.914325714111328,
      "learning_rate": 6.232735112995445e-06,
      "loss": 0.9773,
      "step": 10842
    },
    {
      "epoch": 0.8762197216105376,
      "grad_norm": 2.4684174060821533,
      "learning_rate": 6.232100939912581e-06,
      "loss": 0.8336,
      "step": 10843
    },
    {
      "epoch": 0.8763005313238651,
      "grad_norm": 2.9864718914031982,
      "learning_rate": 6.2314667457267944e-06,
      "loss": 0.8743,
      "step": 10844
    },
    {
      "epoch": 0.8763813410371927,
      "grad_norm": 2.4806735515594482,
      "learning_rate": 6.230832530448947e-06,
      "loss": 0.9626,
      "step": 10845
    },
    {
      "epoch": 0.8764621507505203,
      "grad_norm": 3.071540355682373,
      "learning_rate": 6.230198294089901e-06,
      "loss": 0.8701,
      "step": 10846
    },
    {
      "epoch": 0.8765429604638477,
      "grad_norm": 2.8915750980377197,
      "learning_rate": 6.2295640366605205e-06,
      "loss": 0.8878,
      "step": 10847
    },
    {
      "epoch": 0.8766237701771753,
      "grad_norm": 2.8225057125091553,
      "learning_rate": 6.228929758171667e-06,
      "loss": 0.8599,
      "step": 10848
    },
    {
      "epoch": 0.8767045798905029,
      "grad_norm": 3.210167646408081,
      "learning_rate": 6.228295458634206e-06,
      "loss": 0.8908,
      "step": 10849
    },
    {
      "epoch": 0.8767853896038303,
      "grad_norm": 2.648663282394409,
      "learning_rate": 6.227661138059e-06,
      "loss": 0.8926,
      "step": 10850
    },
    {
      "epoch": 0.8768661993171579,
      "grad_norm": 2.393606185913086,
      "learning_rate": 6.227026796456915e-06,
      "loss": 0.9686,
      "step": 10851
    },
    {
      "epoch": 0.8769470090304855,
      "grad_norm": 2.7983009815216064,
      "learning_rate": 6.226392433838815e-06,
      "loss": 0.9171,
      "step": 10852
    },
    {
      "epoch": 0.877027818743813,
      "grad_norm": 2.3201189041137695,
      "learning_rate": 6.2257580502155655e-06,
      "loss": 0.9764,
      "step": 10853
    },
    {
      "epoch": 0.8771086284571405,
      "grad_norm": 2.386796712875366,
      "learning_rate": 6.22512364559803e-06,
      "loss": 0.8711,
      "step": 10854
    },
    {
      "epoch": 0.8771894381704681,
      "grad_norm": 2.350531816482544,
      "learning_rate": 6.2244892199970775e-06,
      "loss": 0.909,
      "step": 10855
    },
    {
      "epoch": 0.8772702478837956,
      "grad_norm": 2.6143767833709717,
      "learning_rate": 6.223854773423571e-06,
      "loss": 0.9039,
      "step": 10856
    },
    {
      "epoch": 0.8773510575971232,
      "grad_norm": 2.9482271671295166,
      "learning_rate": 6.223220305888379e-06,
      "loss": 0.9026,
      "step": 10857
    },
    {
      "epoch": 0.8774318673104508,
      "grad_norm": 2.7969493865966797,
      "learning_rate": 6.222585817402368e-06,
      "loss": 0.8364,
      "step": 10858
    },
    {
      "epoch": 0.8775126770237782,
      "grad_norm": 2.87231707572937,
      "learning_rate": 6.221951307976405e-06,
      "loss": 0.8836,
      "step": 10859
    },
    {
      "epoch": 0.8775934867371058,
      "grad_norm": 2.7605488300323486,
      "learning_rate": 6.221316777621358e-06,
      "loss": 0.894,
      "step": 10860
    },
    {
      "epoch": 0.8776742964504334,
      "grad_norm": 2.188354015350342,
      "learning_rate": 6.220682226348096e-06,
      "loss": 0.8875,
      "step": 10861
    },
    {
      "epoch": 0.8777551061637608,
      "grad_norm": 2.5554821491241455,
      "learning_rate": 6.220047654167484e-06,
      "loss": 0.7976,
      "step": 10862
    },
    {
      "epoch": 0.8778359158770884,
      "grad_norm": 2.6469528675079346,
      "learning_rate": 6.2194130610903945e-06,
      "loss": 0.8315,
      "step": 10863
    },
    {
      "epoch": 0.877916725590416,
      "grad_norm": 2.3564810752868652,
      "learning_rate": 6.218778447127694e-06,
      "loss": 0.7794,
      "step": 10864
    },
    {
      "epoch": 0.8779975353037435,
      "grad_norm": 2.6810898780822754,
      "learning_rate": 6.2181438122902535e-06,
      "loss": 0.9194,
      "step": 10865
    },
    {
      "epoch": 0.878078345017071,
      "grad_norm": 3.0478122234344482,
      "learning_rate": 6.217509156588941e-06,
      "loss": 0.937,
      "step": 10866
    },
    {
      "epoch": 0.8781591547303986,
      "grad_norm": 2.703763008117676,
      "learning_rate": 6.2168744800346296e-06,
      "loss": 0.913,
      "step": 10867
    },
    {
      "epoch": 0.8782399644437261,
      "grad_norm": 2.9125471115112305,
      "learning_rate": 6.216239782638185e-06,
      "loss": 0.9359,
      "step": 10868
    },
    {
      "epoch": 0.8783207741570537,
      "grad_norm": 2.6811916828155518,
      "learning_rate": 6.215605064410484e-06,
      "loss": 1.0597,
      "step": 10869
    },
    {
      "epoch": 0.8784015838703813,
      "grad_norm": 3.099846363067627,
      "learning_rate": 6.2149703253623925e-06,
      "loss": 0.9312,
      "step": 10870
    },
    {
      "epoch": 0.8784823935837087,
      "grad_norm": 2.476093053817749,
      "learning_rate": 6.214335565504785e-06,
      "loss": 0.9466,
      "step": 10871
    },
    {
      "epoch": 0.8785632032970363,
      "grad_norm": 2.362313985824585,
      "learning_rate": 6.213700784848532e-06,
      "loss": 0.8724,
      "step": 10872
    },
    {
      "epoch": 0.8786440130103639,
      "grad_norm": 2.5924103260040283,
      "learning_rate": 6.213065983404507e-06,
      "loss": 0.8915,
      "step": 10873
    },
    {
      "epoch": 0.8787248227236913,
      "grad_norm": 2.9155962467193604,
      "learning_rate": 6.2124311611835816e-06,
      "loss": 0.927,
      "step": 10874
    },
    {
      "epoch": 0.8788056324370189,
      "grad_norm": 2.7962844371795654,
      "learning_rate": 6.211796318196631e-06,
      "loss": 0.9504,
      "step": 10875
    },
    {
      "epoch": 0.8788864421503465,
      "grad_norm": 2.7248804569244385,
      "learning_rate": 6.211161454454524e-06,
      "loss": 0.9766,
      "step": 10876
    },
    {
      "epoch": 0.878967251863674,
      "grad_norm": 2.827620267868042,
      "learning_rate": 6.210526569968139e-06,
      "loss": 0.9193,
      "step": 10877
    },
    {
      "epoch": 0.8790480615770015,
      "grad_norm": 2.3883299827575684,
      "learning_rate": 6.2098916647483476e-06,
      "loss": 0.7755,
      "step": 10878
    },
    {
      "epoch": 0.8791288712903291,
      "grad_norm": 2.251025915145874,
      "learning_rate": 6.209256738806024e-06,
      "loss": 0.9077,
      "step": 10879
    },
    {
      "epoch": 0.8792096810036566,
      "grad_norm": 2.837599992752075,
      "learning_rate": 6.208621792152045e-06,
      "loss": 0.8625,
      "step": 10880
    },
    {
      "epoch": 0.8792904907169842,
      "grad_norm": 2.9547653198242188,
      "learning_rate": 6.207986824797284e-06,
      "loss": 0.9137,
      "step": 10881
    },
    {
      "epoch": 0.8793713004303118,
      "grad_norm": 2.667105197906494,
      "learning_rate": 6.207351836752615e-06,
      "loss": 0.9531,
      "step": 10882
    },
    {
      "epoch": 0.8794521101436392,
      "grad_norm": 2.4493191242218018,
      "learning_rate": 6.206716828028918e-06,
      "loss": 1.0634,
      "step": 10883
    },
    {
      "epoch": 0.8795329198569668,
      "grad_norm": 2.635343313217163,
      "learning_rate": 6.206081798637066e-06,
      "loss": 0.9985,
      "step": 10884
    },
    {
      "epoch": 0.8796137295702944,
      "grad_norm": 2.2418689727783203,
      "learning_rate": 6.205446748587935e-06,
      "loss": 0.98,
      "step": 10885
    },
    {
      "epoch": 0.8796945392836218,
      "grad_norm": 2.3565685749053955,
      "learning_rate": 6.204811677892405e-06,
      "loss": 0.98,
      "step": 10886
    },
    {
      "epoch": 0.8797753489969494,
      "grad_norm": 2.800968885421753,
      "learning_rate": 6.20417658656135e-06,
      "loss": 0.9563,
      "step": 10887
    },
    {
      "epoch": 0.879856158710277,
      "grad_norm": 2.515580177307129,
      "learning_rate": 6.20354147460565e-06,
      "loss": 0.9428,
      "step": 10888
    },
    {
      "epoch": 0.8799369684236045,
      "grad_norm": 2.726541042327881,
      "learning_rate": 6.2029063420361826e-06,
      "loss": 0.7699,
      "step": 10889
    },
    {
      "epoch": 0.880017778136932,
      "grad_norm": 2.676410436630249,
      "learning_rate": 6.202271188863823e-06,
      "loss": 0.9635,
      "step": 10890
    },
    {
      "epoch": 0.8800985878502596,
      "grad_norm": 2.9437105655670166,
      "learning_rate": 6.201636015099455e-06,
      "loss": 1.0041,
      "step": 10891
    },
    {
      "epoch": 0.8801793975635871,
      "grad_norm": 2.6276378631591797,
      "learning_rate": 6.201000820753953e-06,
      "loss": 0.9229,
      "step": 10892
    },
    {
      "epoch": 0.8802602072769147,
      "grad_norm": 2.8323397636413574,
      "learning_rate": 6.200365605838199e-06,
      "loss": 0.9107,
      "step": 10893
    },
    {
      "epoch": 0.8803410169902423,
      "grad_norm": 2.2161734104156494,
      "learning_rate": 6.199730370363072e-06,
      "loss": 0.9561,
      "step": 10894
    },
    {
      "epoch": 0.8804218267035697,
      "grad_norm": 2.293673515319824,
      "learning_rate": 6.199095114339452e-06,
      "loss": 0.8399,
      "step": 10895
    },
    {
      "epoch": 0.8805026364168973,
      "grad_norm": 2.8733720779418945,
      "learning_rate": 6.198459837778219e-06,
      "loss": 0.9495,
      "step": 10896
    },
    {
      "epoch": 0.8805834461302249,
      "grad_norm": 2.476972818374634,
      "learning_rate": 6.197824540690254e-06,
      "loss": 0.9946,
      "step": 10897
    },
    {
      "epoch": 0.8806642558435523,
      "grad_norm": 2.2801151275634766,
      "learning_rate": 6.19718922308644e-06,
      "loss": 0.7916,
      "step": 10898
    },
    {
      "epoch": 0.8807450655568799,
      "grad_norm": 2.419496774673462,
      "learning_rate": 6.196553884977654e-06,
      "loss": 0.8658,
      "step": 10899
    },
    {
      "epoch": 0.8808258752702075,
      "grad_norm": 2.6121389865875244,
      "learning_rate": 6.195918526374782e-06,
      "loss": 0.9492,
      "step": 10900
    },
    {
      "epoch": 0.880906684983535,
      "grad_norm": 2.437551498413086,
      "learning_rate": 6.195283147288704e-06,
      "loss": 0.9233,
      "step": 10901
    },
    {
      "epoch": 0.8809874946968625,
      "grad_norm": 2.4147794246673584,
      "learning_rate": 6.194647747730305e-06,
      "loss": 1.1507,
      "step": 10902
    },
    {
      "epoch": 0.8810683044101901,
      "grad_norm": 2.5784285068511963,
      "learning_rate": 6.194012327710464e-06,
      "loss": 0.8709,
      "step": 10903
    },
    {
      "epoch": 0.8811491141235176,
      "grad_norm": 2.5563831329345703,
      "learning_rate": 6.1933768872400665e-06,
      "loss": 1.0097,
      "step": 10904
    },
    {
      "epoch": 0.8812299238368452,
      "grad_norm": 2.711738348007202,
      "learning_rate": 6.1927414263299966e-06,
      "loss": 0.8831,
      "step": 10905
    },
    {
      "epoch": 0.8813107335501728,
      "grad_norm": 2.670205593109131,
      "learning_rate": 6.192105944991138e-06,
      "loss": 0.913,
      "step": 10906
    },
    {
      "epoch": 0.8813915432635002,
      "grad_norm": 2.5287325382232666,
      "learning_rate": 6.191470443234373e-06,
      "loss": 0.9675,
      "step": 10907
    },
    {
      "epoch": 0.8814723529768278,
      "grad_norm": 2.4057791233062744,
      "learning_rate": 6.19083492107059e-06,
      "loss": 0.9638,
      "step": 10908
    },
    {
      "epoch": 0.8815531626901554,
      "grad_norm": 2.631802797317505,
      "learning_rate": 6.19019937851067e-06,
      "loss": 0.9026,
      "step": 10909
    },
    {
      "epoch": 0.8816339724034828,
      "grad_norm": 2.7692408561706543,
      "learning_rate": 6.189563815565499e-06,
      "loss": 0.8508,
      "step": 10910
    },
    {
      "epoch": 0.8817147821168104,
      "grad_norm": 2.1514453887939453,
      "learning_rate": 6.188928232245966e-06,
      "loss": 0.8769,
      "step": 10911
    },
    {
      "epoch": 0.881795591830138,
      "grad_norm": 2.7200186252593994,
      "learning_rate": 6.188292628562953e-06,
      "loss": 0.7712,
      "step": 10912
    },
    {
      "epoch": 0.8818764015434655,
      "grad_norm": 3.0090413093566895,
      "learning_rate": 6.187657004527348e-06,
      "loss": 1.01,
      "step": 10913
    },
    {
      "epoch": 0.881957211256793,
      "grad_norm": 2.3555073738098145,
      "learning_rate": 6.187021360150038e-06,
      "loss": 0.9096,
      "step": 10914
    },
    {
      "epoch": 0.8820380209701206,
      "grad_norm": 2.7792162895202637,
      "learning_rate": 6.186385695441909e-06,
      "loss": 0.999,
      "step": 10915
    },
    {
      "epoch": 0.8821188306834481,
      "grad_norm": 2.5428154468536377,
      "learning_rate": 6.18575001041385e-06,
      "loss": 1.0028,
      "step": 10916
    },
    {
      "epoch": 0.8821996403967757,
      "grad_norm": 2.696187973022461,
      "learning_rate": 6.185114305076748e-06,
      "loss": 1.0562,
      "step": 10917
    },
    {
      "epoch": 0.8822804501101033,
      "grad_norm": 2.7940385341644287,
      "learning_rate": 6.184478579441491e-06,
      "loss": 0.861,
      "step": 10918
    },
    {
      "epoch": 0.8823612598234307,
      "grad_norm": 2.8222694396972656,
      "learning_rate": 6.183842833518966e-06,
      "loss": 0.9965,
      "step": 10919
    },
    {
      "epoch": 0.8824420695367583,
      "grad_norm": 2.582070827484131,
      "learning_rate": 6.183207067320065e-06,
      "loss": 0.8817,
      "step": 10920
    },
    {
      "epoch": 0.8825228792500859,
      "grad_norm": 2.641030788421631,
      "learning_rate": 6.1825712808556734e-06,
      "loss": 0.9073,
      "step": 10921
    },
    {
      "epoch": 0.8826036889634135,
      "grad_norm": 2.370635747909546,
      "learning_rate": 6.1819354741366855e-06,
      "loss": 1.0039,
      "step": 10922
    },
    {
      "epoch": 0.8826844986767409,
      "grad_norm": 2.588547945022583,
      "learning_rate": 6.181299647173987e-06,
      "loss": 0.967,
      "step": 10923
    },
    {
      "epoch": 0.8827653083900685,
      "grad_norm": 2.910203456878662,
      "learning_rate": 6.180663799978469e-06,
      "loss": 0.9606,
      "step": 10924
    },
    {
      "epoch": 0.8828461181033961,
      "grad_norm": 2.415900945663452,
      "learning_rate": 6.180027932561022e-06,
      "loss": 0.9355,
      "step": 10925
    },
    {
      "epoch": 0.8829269278167236,
      "grad_norm": 2.5070905685424805,
      "learning_rate": 6.179392044932539e-06,
      "loss": 0.9732,
      "step": 10926
    },
    {
      "epoch": 0.8830077375300511,
      "grad_norm": 3.0649960041046143,
      "learning_rate": 6.178756137103908e-06,
      "loss": 0.9669,
      "step": 10927
    },
    {
      "epoch": 0.8830885472433787,
      "grad_norm": 2.8635520935058594,
      "learning_rate": 6.178120209086024e-06,
      "loss": 0.9476,
      "step": 10928
    },
    {
      "epoch": 0.8831693569567062,
      "grad_norm": 2.382916212081909,
      "learning_rate": 6.177484260889775e-06,
      "loss": 0.9309,
      "step": 10929
    },
    {
      "epoch": 0.8832501666700338,
      "grad_norm": 2.9815895557403564,
      "learning_rate": 6.176848292526057e-06,
      "loss": 0.9731,
      "step": 10930
    },
    {
      "epoch": 0.8833309763833613,
      "grad_norm": 3.019881248474121,
      "learning_rate": 6.176212304005759e-06,
      "loss": 1.0128,
      "step": 10931
    },
    {
      "epoch": 0.8834117860966888,
      "grad_norm": 2.5681111812591553,
      "learning_rate": 6.175576295339776e-06,
      "loss": 0.8593,
      "step": 10932
    },
    {
      "epoch": 0.8834925958100164,
      "grad_norm": 2.969653606414795,
      "learning_rate": 6.174940266539003e-06,
      "loss": 0.905,
      "step": 10933
    },
    {
      "epoch": 0.883573405523344,
      "grad_norm": 2.4689438343048096,
      "learning_rate": 6.174304217614331e-06,
      "loss": 0.8833,
      "step": 10934
    },
    {
      "epoch": 0.8836542152366714,
      "grad_norm": 3.0760090351104736,
      "learning_rate": 6.173668148576652e-06,
      "loss": 0.945,
      "step": 10935
    },
    {
      "epoch": 0.883735024949999,
      "grad_norm": 2.8236286640167236,
      "learning_rate": 6.173032059436868e-06,
      "loss": 0.9239,
      "step": 10936
    },
    {
      "epoch": 0.8838158346633266,
      "grad_norm": 2.3985254764556885,
      "learning_rate": 6.172395950205865e-06,
      "loss": 0.9703,
      "step": 10937
    },
    {
      "epoch": 0.883896644376654,
      "grad_norm": 2.001253128051758,
      "learning_rate": 6.171759820894544e-06,
      "loss": 0.9547,
      "step": 10938
    },
    {
      "epoch": 0.8839774540899816,
      "grad_norm": 2.5760116577148438,
      "learning_rate": 6.171123671513797e-06,
      "loss": 0.8207,
      "step": 10939
    },
    {
      "epoch": 0.8840582638033092,
      "grad_norm": 2.603823184967041,
      "learning_rate": 6.170487502074521e-06,
      "loss": 1.0051,
      "step": 10940
    },
    {
      "epoch": 0.8841390735166367,
      "grad_norm": 2.527780771255493,
      "learning_rate": 6.169851312587612e-06,
      "loss": 0.9405,
      "step": 10941
    },
    {
      "epoch": 0.8842198832299643,
      "grad_norm": 2.5697104930877686,
      "learning_rate": 6.169215103063967e-06,
      "loss": 0.862,
      "step": 10942
    },
    {
      "epoch": 0.8843006929432918,
      "grad_norm": 2.6602511405944824,
      "learning_rate": 6.168578873514481e-06,
      "loss": 0.9605,
      "step": 10943
    },
    {
      "epoch": 0.8843815026566193,
      "grad_norm": 2.4392359256744385,
      "learning_rate": 6.1679426239500525e-06,
      "loss": 0.8319,
      "step": 10944
    },
    {
      "epoch": 0.8844623123699469,
      "grad_norm": 2.3442440032958984,
      "learning_rate": 6.167306354381579e-06,
      "loss": 0.9502,
      "step": 10945
    },
    {
      "epoch": 0.8845431220832745,
      "grad_norm": 2.7411835193634033,
      "learning_rate": 6.166670064819957e-06,
      "loss": 0.9613,
      "step": 10946
    },
    {
      "epoch": 0.8846239317966019,
      "grad_norm": 2.5848867893218994,
      "learning_rate": 6.166033755276087e-06,
      "loss": 0.9501,
      "step": 10947
    },
    {
      "epoch": 0.8847047415099295,
      "grad_norm": 2.587322235107422,
      "learning_rate": 6.165397425760865e-06,
      "loss": 0.9307,
      "step": 10948
    },
    {
      "epoch": 0.8847855512232571,
      "grad_norm": 2.7896323204040527,
      "learning_rate": 6.16476107628519e-06,
      "loss": 0.8878,
      "step": 10949
    },
    {
      "epoch": 0.8848663609365846,
      "grad_norm": 2.699521541595459,
      "learning_rate": 6.164124706859962e-06,
      "loss": 0.8462,
      "step": 10950
    },
    {
      "epoch": 0.8849471706499121,
      "grad_norm": 2.4634876251220703,
      "learning_rate": 6.16348831749608e-06,
      "loss": 0.81,
      "step": 10951
    },
    {
      "epoch": 0.8850279803632397,
      "grad_norm": 2.4452576637268066,
      "learning_rate": 6.162851908204446e-06,
      "loss": 0.9694,
      "step": 10952
    },
    {
      "epoch": 0.8851087900765672,
      "grad_norm": 2.78956937789917,
      "learning_rate": 6.162215478995956e-06,
      "loss": 0.8283,
      "step": 10953
    },
    {
      "epoch": 0.8851895997898948,
      "grad_norm": 2.6653823852539062,
      "learning_rate": 6.161579029881514e-06,
      "loss": 1.0106,
      "step": 10954
    },
    {
      "epoch": 0.8852704095032223,
      "grad_norm": 3.050967216491699,
      "learning_rate": 6.160942560872019e-06,
      "loss": 0.984,
      "step": 10955
    },
    {
      "epoch": 0.8853512192165498,
      "grad_norm": 2.719074249267578,
      "learning_rate": 6.160306071978374e-06,
      "loss": 1.0023,
      "step": 10956
    },
    {
      "epoch": 0.8854320289298774,
      "grad_norm": 2.5805485248565674,
      "learning_rate": 6.1596695632114765e-06,
      "loss": 0.9016,
      "step": 10957
    },
    {
      "epoch": 0.885512838643205,
      "grad_norm": 2.601243734359741,
      "learning_rate": 6.159033034582234e-06,
      "loss": 0.9219,
      "step": 10958
    },
    {
      "epoch": 0.8855936483565324,
      "grad_norm": 2.6648449897766113,
      "learning_rate": 6.158396486101545e-06,
      "loss": 0.9181,
      "step": 10959
    },
    {
      "epoch": 0.88567445806986,
      "grad_norm": 2.6205074787139893,
      "learning_rate": 6.157759917780313e-06,
      "loss": 0.912,
      "step": 10960
    },
    {
      "epoch": 0.8857552677831876,
      "grad_norm": 2.8496949672698975,
      "learning_rate": 6.1571233296294405e-06,
      "loss": 0.8803,
      "step": 10961
    },
    {
      "epoch": 0.885836077496515,
      "grad_norm": 3.2378621101379395,
      "learning_rate": 6.156486721659831e-06,
      "loss": 0.9554,
      "step": 10962
    },
    {
      "epoch": 0.8859168872098426,
      "grad_norm": 2.833557367324829,
      "learning_rate": 6.155850093882388e-06,
      "loss": 0.8796,
      "step": 10963
    },
    {
      "epoch": 0.8859976969231702,
      "grad_norm": 2.7678937911987305,
      "learning_rate": 6.155213446308016e-06,
      "loss": 0.9024,
      "step": 10964
    },
    {
      "epoch": 0.8860785066364977,
      "grad_norm": 2.7520134449005127,
      "learning_rate": 6.1545767789476195e-06,
      "loss": 0.8869,
      "step": 10965
    },
    {
      "epoch": 0.8861593163498253,
      "grad_norm": 2.827279567718506,
      "learning_rate": 6.153940091812101e-06,
      "loss": 0.9128,
      "step": 10966
    },
    {
      "epoch": 0.8862401260631528,
      "grad_norm": 2.4025306701660156,
      "learning_rate": 6.153303384912367e-06,
      "loss": 0.911,
      "step": 10967
    },
    {
      "epoch": 0.8863209357764803,
      "grad_norm": 2.7287189960479736,
      "learning_rate": 6.152666658259323e-06,
      "loss": 0.9942,
      "step": 10968
    },
    {
      "epoch": 0.8864017454898079,
      "grad_norm": 2.780747413635254,
      "learning_rate": 6.152029911863872e-06,
      "loss": 1.056,
      "step": 10969
    },
    {
      "epoch": 0.8864825552031355,
      "grad_norm": 3.0128583908081055,
      "learning_rate": 6.151393145736925e-06,
      "loss": 1.0137,
      "step": 10970
    },
    {
      "epoch": 0.8865633649164629,
      "grad_norm": 2.7319211959838867,
      "learning_rate": 6.150756359889382e-06,
      "loss": 0.8488,
      "step": 10971
    },
    {
      "epoch": 0.8866441746297905,
      "grad_norm": 2.742210865020752,
      "learning_rate": 6.150119554332156e-06,
      "loss": 0.8695,
      "step": 10972
    },
    {
      "epoch": 0.8867249843431181,
      "grad_norm": 2.666001319885254,
      "learning_rate": 6.1494827290761505e-06,
      "loss": 0.8589,
      "step": 10973
    },
    {
      "epoch": 0.8868057940564456,
      "grad_norm": 2.4835453033447266,
      "learning_rate": 6.1488458841322704e-06,
      "loss": 0.8999,
      "step": 10974
    },
    {
      "epoch": 0.8868866037697731,
      "grad_norm": 2.4682981967926025,
      "learning_rate": 6.14820901951143e-06,
      "loss": 0.8755,
      "step": 10975
    },
    {
      "epoch": 0.8869674134831007,
      "grad_norm": 2.6600914001464844,
      "learning_rate": 6.14757213522453e-06,
      "loss": 0.9856,
      "step": 10976
    },
    {
      "epoch": 0.8870482231964282,
      "grad_norm": 2.5532233715057373,
      "learning_rate": 6.146935231282484e-06,
      "loss": 1.0455,
      "step": 10977
    },
    {
      "epoch": 0.8871290329097558,
      "grad_norm": 2.6507294178009033,
      "learning_rate": 6.1462983076961965e-06,
      "loss": 0.9244,
      "step": 10978
    },
    {
      "epoch": 0.8872098426230833,
      "grad_norm": 2.4792628288269043,
      "learning_rate": 6.145661364476582e-06,
      "loss": 0.7741,
      "step": 10979
    },
    {
      "epoch": 0.8872906523364108,
      "grad_norm": 2.6015565395355225,
      "learning_rate": 6.145024401634543e-06,
      "loss": 0.9,
      "step": 10980
    },
    {
      "epoch": 0.8873714620497384,
      "grad_norm": 2.4832382202148438,
      "learning_rate": 6.144387419180994e-06,
      "loss": 0.9151,
      "step": 10981
    },
    {
      "epoch": 0.887452271763066,
      "grad_norm": 2.9805192947387695,
      "learning_rate": 6.143750417126843e-06,
      "loss": 1.0015,
      "step": 10982
    },
    {
      "epoch": 0.8875330814763934,
      "grad_norm": 2.3034491539001465,
      "learning_rate": 6.143113395483001e-06,
      "loss": 0.8822,
      "step": 10983
    },
    {
      "epoch": 0.887613891189721,
      "grad_norm": 2.5549511909484863,
      "learning_rate": 6.142476354260378e-06,
      "loss": 1.0219,
      "step": 10984
    },
    {
      "epoch": 0.8876947009030486,
      "grad_norm": 2.888711929321289,
      "learning_rate": 6.141839293469887e-06,
      "loss": 0.8712,
      "step": 10985
    },
    {
      "epoch": 0.887775510616376,
      "grad_norm": 2.5501506328582764,
      "learning_rate": 6.141202213122436e-06,
      "loss": 1.0281,
      "step": 10986
    },
    {
      "epoch": 0.8878563203297036,
      "grad_norm": 2.496525764465332,
      "learning_rate": 6.14056511322894e-06,
      "loss": 0.8497,
      "step": 10987
    },
    {
      "epoch": 0.8879371300430312,
      "grad_norm": 2.4370250701904297,
      "learning_rate": 6.139927993800308e-06,
      "loss": 0.8917,
      "step": 10988
    },
    {
      "epoch": 0.8880179397563587,
      "grad_norm": 2.651669502258301,
      "learning_rate": 6.139290854847455e-06,
      "loss": 0.9555,
      "step": 10989
    },
    {
      "epoch": 0.8880987494696863,
      "grad_norm": 2.6617887020111084,
      "learning_rate": 6.138653696381292e-06,
      "loss": 0.9536,
      "step": 10990
    },
    {
      "epoch": 0.8881795591830138,
      "grad_norm": 2.939648151397705,
      "learning_rate": 6.1380165184127315e-06,
      "loss": 0.9768,
      "step": 10991
    },
    {
      "epoch": 0.8882603688963413,
      "grad_norm": 2.7883365154266357,
      "learning_rate": 6.137379320952688e-06,
      "loss": 0.9624,
      "step": 10992
    },
    {
      "epoch": 0.8883411786096689,
      "grad_norm": 2.86812686920166,
      "learning_rate": 6.1367421040120765e-06,
      "loss": 0.9721,
      "step": 10993
    },
    {
      "epoch": 0.8884219883229965,
      "grad_norm": 2.7602572441101074,
      "learning_rate": 6.136104867601806e-06,
      "loss": 0.8592,
      "step": 10994
    },
    {
      "epoch": 0.8885027980363239,
      "grad_norm": 3.019890785217285,
      "learning_rate": 6.135467611732798e-06,
      "loss": 0.9147,
      "step": 10995
    },
    {
      "epoch": 0.8885836077496515,
      "grad_norm": 2.5440359115600586,
      "learning_rate": 6.13483033641596e-06,
      "loss": 0.8378,
      "step": 10996
    },
    {
      "epoch": 0.8886644174629791,
      "grad_norm": 2.7020857334136963,
      "learning_rate": 6.134193041662213e-06,
      "loss": 1.0581,
      "step": 10997
    },
    {
      "epoch": 0.8887452271763066,
      "grad_norm": 2.3741939067840576,
      "learning_rate": 6.133555727482468e-06,
      "loss": 0.9238,
      "step": 10998
    },
    {
      "epoch": 0.8888260368896341,
      "grad_norm": 3.2570548057556152,
      "learning_rate": 6.132918393887643e-06,
      "loss": 0.9342,
      "step": 10999
    },
    {
      "epoch": 0.8889068466029617,
      "grad_norm": 2.0186386108398438,
      "learning_rate": 6.132281040888653e-06,
      "loss": 1.014,
      "step": 11000
    },
    {
      "epoch": 0.8889068466029617,
      "eval_loss": 0.7640535831451416,
      "eval_runtime": 815.0076,
      "eval_samples_per_second": 102.289,
      "eval_steps_per_second": 12.786,
      "step": 11000
    },
    {
      "epoch": 0.8889876563162892,
      "grad_norm": 2.4831643104553223,
      "learning_rate": 6.131643668496417e-06,
      "loss": 0.9275,
      "step": 11001
    },
    {
      "epoch": 0.8890684660296168,
      "grad_norm": 2.905310869216919,
      "learning_rate": 6.131006276721845e-06,
      "loss": 0.8717,
      "step": 11002
    },
    {
      "epoch": 0.8891492757429443,
      "grad_norm": 2.3836982250213623,
      "learning_rate": 6.130368865575861e-06,
      "loss": 0.9481,
      "step": 11003
    },
    {
      "epoch": 0.8892300854562718,
      "grad_norm": 2.923142910003662,
      "learning_rate": 6.129731435069379e-06,
      "loss": 1.0165,
      "step": 11004
    },
    {
      "epoch": 0.8893108951695994,
      "grad_norm": 2.755363941192627,
      "learning_rate": 6.129093985213317e-06,
      "loss": 1.0159,
      "step": 11005
    },
    {
      "epoch": 0.889391704882927,
      "grad_norm": 2.3933568000793457,
      "learning_rate": 6.1284565160185925e-06,
      "loss": 0.868,
      "step": 11006
    },
    {
      "epoch": 0.8894725145962544,
      "grad_norm": 2.53499698638916,
      "learning_rate": 6.127819027496127e-06,
      "loss": 0.89,
      "step": 11007
    },
    {
      "epoch": 0.889553324309582,
      "grad_norm": 2.4543845653533936,
      "learning_rate": 6.127181519656834e-06,
      "loss": 1.0205,
      "step": 11008
    },
    {
      "epoch": 0.8896341340229096,
      "grad_norm": 2.6794471740722656,
      "learning_rate": 6.126543992511638e-06,
      "loss": 0.8723,
      "step": 11009
    },
    {
      "epoch": 0.889714943736237,
      "grad_norm": 2.9267797470092773,
      "learning_rate": 6.1259064460714514e-06,
      "loss": 0.938,
      "step": 11010
    },
    {
      "epoch": 0.8897957534495646,
      "grad_norm": 2.8477814197540283,
      "learning_rate": 6.1252688803472016e-06,
      "loss": 0.8794,
      "step": 11011
    },
    {
      "epoch": 0.8898765631628922,
      "grad_norm": 2.8415892124176025,
      "learning_rate": 6.124631295349803e-06,
      "loss": 0.9823,
      "step": 11012
    },
    {
      "epoch": 0.8899573728762197,
      "grad_norm": 2.753502607345581,
      "learning_rate": 6.123993691090178e-06,
      "loss": 1.0443,
      "step": 11013
    },
    {
      "epoch": 0.8900381825895473,
      "grad_norm": 2.7093663215637207,
      "learning_rate": 6.1233560675792465e-06,
      "loss": 0.881,
      "step": 11014
    },
    {
      "epoch": 0.8901189923028748,
      "grad_norm": 2.673405170440674,
      "learning_rate": 6.122718424827931e-06,
      "loss": 0.9464,
      "step": 11015
    },
    {
      "epoch": 0.8901998020162023,
      "grad_norm": 2.669354200363159,
      "learning_rate": 6.122080762847151e-06,
      "loss": 0.9228,
      "step": 11016
    },
    {
      "epoch": 0.8902806117295299,
      "grad_norm": 2.303164005279541,
      "learning_rate": 6.12144308164783e-06,
      "loss": 0.938,
      "step": 11017
    },
    {
      "epoch": 0.8903614214428575,
      "grad_norm": 2.6203126907348633,
      "learning_rate": 6.120805381240888e-06,
      "loss": 0.7984,
      "step": 11018
    },
    {
      "epoch": 0.8904422311561849,
      "grad_norm": 2.3148365020751953,
      "learning_rate": 6.120167661637247e-06,
      "loss": 0.8208,
      "step": 11019
    },
    {
      "epoch": 0.8905230408695125,
      "grad_norm": 2.8378329277038574,
      "learning_rate": 6.119529922847832e-06,
      "loss": 0.9023,
      "step": 11020
    },
    {
      "epoch": 0.8906038505828401,
      "grad_norm": 2.271371841430664,
      "learning_rate": 6.1188921648835646e-06,
      "loss": 0.8225,
      "step": 11021
    },
    {
      "epoch": 0.8906846602961676,
      "grad_norm": 2.5016963481903076,
      "learning_rate": 6.118254387755367e-06,
      "loss": 0.8901,
      "step": 11022
    },
    {
      "epoch": 0.8907654700094951,
      "grad_norm": 2.6315226554870605,
      "learning_rate": 6.117616591474166e-06,
      "loss": 0.988,
      "step": 11023
    },
    {
      "epoch": 0.8908462797228227,
      "grad_norm": 3.0709304809570312,
      "learning_rate": 6.1169787760508806e-06,
      "loss": 1.0308,
      "step": 11024
    },
    {
      "epoch": 0.8909270894361502,
      "grad_norm": 2.647787094116211,
      "learning_rate": 6.11634094149644e-06,
      "loss": 0.9285,
      "step": 11025
    },
    {
      "epoch": 0.8910078991494778,
      "grad_norm": 2.7675271034240723,
      "learning_rate": 6.115703087821765e-06,
      "loss": 0.9306,
      "step": 11026
    },
    {
      "epoch": 0.8910887088628053,
      "grad_norm": 2.588698148727417,
      "learning_rate": 6.115065215037782e-06,
      "loss": 0.8785,
      "step": 11027
    },
    {
      "epoch": 0.8911695185761328,
      "grad_norm": 2.85332989692688,
      "learning_rate": 6.1144273231554165e-06,
      "loss": 0.8812,
      "step": 11028
    },
    {
      "epoch": 0.8912503282894604,
      "grad_norm": 2.28715181350708,
      "learning_rate": 6.113789412185594e-06,
      "loss": 1.0643,
      "step": 11029
    },
    {
      "epoch": 0.891331138002788,
      "grad_norm": 2.4027044773101807,
      "learning_rate": 6.113151482139241e-06,
      "loss": 0.9467,
      "step": 11030
    },
    {
      "epoch": 0.8914119477161154,
      "grad_norm": 3.2616589069366455,
      "learning_rate": 6.112513533027282e-06,
      "loss": 0.9655,
      "step": 11031
    },
    {
      "epoch": 0.891492757429443,
      "grad_norm": 2.323657512664795,
      "learning_rate": 6.111875564860645e-06,
      "loss": 0.9994,
      "step": 11032
    },
    {
      "epoch": 0.8915735671427706,
      "grad_norm": 2.5008647441864014,
      "learning_rate": 6.1112375776502565e-06,
      "loss": 0.911,
      "step": 11033
    },
    {
      "epoch": 0.891654376856098,
      "grad_norm": 2.446787118911743,
      "learning_rate": 6.110599571407043e-06,
      "loss": 0.9701,
      "step": 11034
    },
    {
      "epoch": 0.8917351865694256,
      "grad_norm": 2.3228187561035156,
      "learning_rate": 6.109961546141932e-06,
      "loss": 0.9283,
      "step": 11035
    },
    {
      "epoch": 0.8918159962827532,
      "grad_norm": 2.728739023208618,
      "learning_rate": 6.109323501865853e-06,
      "loss": 0.8999,
      "step": 11036
    },
    {
      "epoch": 0.8918968059960807,
      "grad_norm": 2.7874202728271484,
      "learning_rate": 6.108685438589732e-06,
      "loss": 0.8816,
      "step": 11037
    },
    {
      "epoch": 0.8919776157094083,
      "grad_norm": 2.433786392211914,
      "learning_rate": 6.108047356324498e-06,
      "loss": 0.9881,
      "step": 11038
    },
    {
      "epoch": 0.8920584254227358,
      "grad_norm": 2.3185479640960693,
      "learning_rate": 6.107409255081082e-06,
      "loss": 0.8745,
      "step": 11039
    },
    {
      "epoch": 0.8921392351360633,
      "grad_norm": 2.803328514099121,
      "learning_rate": 6.1067711348704104e-06,
      "loss": 0.9704,
      "step": 11040
    },
    {
      "epoch": 0.8922200448493909,
      "grad_norm": 2.8452160358428955,
      "learning_rate": 6.1061329957034145e-06,
      "loss": 0.893,
      "step": 11041
    },
    {
      "epoch": 0.8923008545627185,
      "grad_norm": 2.792412042617798,
      "learning_rate": 6.105494837591023e-06,
      "loss": 0.9528,
      "step": 11042
    },
    {
      "epoch": 0.8923816642760459,
      "grad_norm": 2.8253371715545654,
      "learning_rate": 6.104856660544165e-06,
      "loss": 0.8227,
      "step": 11043
    },
    {
      "epoch": 0.8924624739893735,
      "grad_norm": 3.3019015789031982,
      "learning_rate": 6.104218464573773e-06,
      "loss": 0.9624,
      "step": 11044
    },
    {
      "epoch": 0.8925432837027011,
      "grad_norm": 2.3970048427581787,
      "learning_rate": 6.103580249690777e-06,
      "loss": 0.9066,
      "step": 11045
    },
    {
      "epoch": 0.8926240934160286,
      "grad_norm": 2.600074529647827,
      "learning_rate": 6.102942015906109e-06,
      "loss": 1.0343,
      "step": 11046
    },
    {
      "epoch": 0.8927049031293561,
      "grad_norm": 2.8846118450164795,
      "learning_rate": 6.102303763230698e-06,
      "loss": 0.8015,
      "step": 11047
    },
    {
      "epoch": 0.8927857128426837,
      "grad_norm": 3.0641016960144043,
      "learning_rate": 6.101665491675479e-06,
      "loss": 0.8428,
      "step": 11048
    },
    {
      "epoch": 0.8928665225560112,
      "grad_norm": 2.2472636699676514,
      "learning_rate": 6.101027201251381e-06,
      "loss": 1.0,
      "step": 11049
    },
    {
      "epoch": 0.8929473322693388,
      "grad_norm": 2.562082052230835,
      "learning_rate": 6.100388891969337e-06,
      "loss": 1.0599,
      "step": 11050
    },
    {
      "epoch": 0.8930281419826663,
      "grad_norm": 2.611246109008789,
      "learning_rate": 6.099750563840282e-06,
      "loss": 0.9197,
      "step": 11051
    },
    {
      "epoch": 0.8931089516959939,
      "grad_norm": 2.5023462772369385,
      "learning_rate": 6.099112216875147e-06,
      "loss": 0.8714,
      "step": 11052
    },
    {
      "epoch": 0.8931897614093214,
      "grad_norm": 2.7679619789123535,
      "learning_rate": 6.098473851084865e-06,
      "loss": 0.8747,
      "step": 11053
    },
    {
      "epoch": 0.893270571122649,
      "grad_norm": 2.6410112380981445,
      "learning_rate": 6.097835466480372e-06,
      "loss": 0.9315,
      "step": 11054
    },
    {
      "epoch": 0.8933513808359765,
      "grad_norm": 2.6186962127685547,
      "learning_rate": 6.097197063072598e-06,
      "loss": 0.9532,
      "step": 11055
    },
    {
      "epoch": 0.893432190549304,
      "grad_norm": 2.960371494293213,
      "learning_rate": 6.096558640872482e-06,
      "loss": 0.945,
      "step": 11056
    },
    {
      "epoch": 0.8935130002626316,
      "grad_norm": 2.620387077331543,
      "learning_rate": 6.0959201998909555e-06,
      "loss": 1.0442,
      "step": 11057
    },
    {
      "epoch": 0.8935938099759592,
      "grad_norm": 2.879689931869507,
      "learning_rate": 6.095281740138954e-06,
      "loss": 1.0425,
      "step": 11058
    },
    {
      "epoch": 0.8936746196892866,
      "grad_norm": 2.5625710487365723,
      "learning_rate": 6.094643261627413e-06,
      "loss": 0.8893,
      "step": 11059
    },
    {
      "epoch": 0.8937554294026142,
      "grad_norm": 2.65523099899292,
      "learning_rate": 6.09400476436727e-06,
      "loss": 0.901,
      "step": 11060
    },
    {
      "epoch": 0.8938362391159418,
      "grad_norm": 2.5786983966827393,
      "learning_rate": 6.093366248369456e-06,
      "loss": 0.9458,
      "step": 11061
    },
    {
      "epoch": 0.8939170488292693,
      "grad_norm": 3.3469996452331543,
      "learning_rate": 6.092727713644912e-06,
      "loss": 0.9274,
      "step": 11062
    },
    {
      "epoch": 0.8939978585425968,
      "grad_norm": 2.6339290142059326,
      "learning_rate": 6.0920891602045705e-06,
      "loss": 0.9025,
      "step": 11063
    },
    {
      "epoch": 0.8940786682559244,
      "grad_norm": 2.3962159156799316,
      "learning_rate": 6.091450588059374e-06,
      "loss": 0.9841,
      "step": 11064
    },
    {
      "epoch": 0.8941594779692519,
      "grad_norm": 2.6802737712860107,
      "learning_rate": 6.090811997220256e-06,
      "loss": 0.9009,
      "step": 11065
    },
    {
      "epoch": 0.8942402876825795,
      "grad_norm": 2.751603126525879,
      "learning_rate": 6.090173387698154e-06,
      "loss": 0.9618,
      "step": 11066
    },
    {
      "epoch": 0.894321097395907,
      "grad_norm": 2.7731077671051025,
      "learning_rate": 6.089534759504005e-06,
      "loss": 0.7689,
      "step": 11067
    },
    {
      "epoch": 0.8944019071092345,
      "grad_norm": 2.53085994720459,
      "learning_rate": 6.088896112648751e-06,
      "loss": 0.9838,
      "step": 11068
    },
    {
      "epoch": 0.8944827168225621,
      "grad_norm": 2.406844139099121,
      "learning_rate": 6.0882574471433266e-06,
      "loss": 0.9046,
      "step": 11069
    },
    {
      "epoch": 0.8945635265358897,
      "grad_norm": 2.21785831451416,
      "learning_rate": 6.087618762998673e-06,
      "loss": 0.6916,
      "step": 11070
    },
    {
      "epoch": 0.8946443362492171,
      "grad_norm": 2.4833412170410156,
      "learning_rate": 6.086980060225727e-06,
      "loss": 0.9391,
      "step": 11071
    },
    {
      "epoch": 0.8947251459625447,
      "grad_norm": 2.9493179321289062,
      "learning_rate": 6.08634133883543e-06,
      "loss": 1.0017,
      "step": 11072
    },
    {
      "epoch": 0.8948059556758723,
      "grad_norm": 2.618835687637329,
      "learning_rate": 6.0857025988387205e-06,
      "loss": 0.9473,
      "step": 11073
    },
    {
      "epoch": 0.8948867653891998,
      "grad_norm": 2.991755485534668,
      "learning_rate": 6.085063840246541e-06,
      "loss": 1.0111,
      "step": 11074
    },
    {
      "epoch": 0.8949675751025273,
      "grad_norm": 2.582691192626953,
      "learning_rate": 6.084425063069827e-06,
      "loss": 0.8572,
      "step": 11075
    },
    {
      "epoch": 0.8950483848158549,
      "grad_norm": 2.7862629890441895,
      "learning_rate": 6.083786267319526e-06,
      "loss": 0.9168,
      "step": 11076
    },
    {
      "epoch": 0.8951291945291824,
      "grad_norm": 3.048041343688965,
      "learning_rate": 6.083147453006572e-06,
      "loss": 0.8811,
      "step": 11077
    },
    {
      "epoch": 0.89521000424251,
      "grad_norm": 2.6280767917633057,
      "learning_rate": 6.082508620141911e-06,
      "loss": 0.7797,
      "step": 11078
    },
    {
      "epoch": 0.8952908139558375,
      "grad_norm": 2.5700151920318604,
      "learning_rate": 6.081869768736485e-06,
      "loss": 0.9484,
      "step": 11079
    },
    {
      "epoch": 0.895371623669165,
      "grad_norm": 2.745976686477661,
      "learning_rate": 6.0812308988012326e-06,
      "loss": 0.9675,
      "step": 11080
    },
    {
      "epoch": 0.8954524333824926,
      "grad_norm": 2.3650968074798584,
      "learning_rate": 6.0805920103470975e-06,
      "loss": 0.8606,
      "step": 11081
    },
    {
      "epoch": 0.8955332430958202,
      "grad_norm": 2.681852340698242,
      "learning_rate": 6.079953103385025e-06,
      "loss": 0.8773,
      "step": 11082
    },
    {
      "epoch": 0.8956140528091476,
      "grad_norm": 2.5203349590301514,
      "learning_rate": 6.079314177925952e-06,
      "loss": 0.9918,
      "step": 11083
    },
    {
      "epoch": 0.8956948625224752,
      "grad_norm": 2.5127477645874023,
      "learning_rate": 6.078675233980829e-06,
      "loss": 1.0656,
      "step": 11084
    },
    {
      "epoch": 0.8957756722358028,
      "grad_norm": 2.8647875785827637,
      "learning_rate": 6.078036271560594e-06,
      "loss": 0.9295,
      "step": 11085
    },
    {
      "epoch": 0.8958564819491303,
      "grad_norm": 2.4162609577178955,
      "learning_rate": 6.077397290676194e-06,
      "loss": 0.9008,
      "step": 11086
    },
    {
      "epoch": 0.8959372916624578,
      "grad_norm": 2.674287796020508,
      "learning_rate": 6.076758291338571e-06,
      "loss": 0.9015,
      "step": 11087
    },
    {
      "epoch": 0.8960181013757854,
      "grad_norm": 3.113100290298462,
      "learning_rate": 6.076119273558672e-06,
      "loss": 0.9221,
      "step": 11088
    },
    {
      "epoch": 0.8960989110891129,
      "grad_norm": 2.8635690212249756,
      "learning_rate": 6.07548023734744e-06,
      "loss": 1.0285,
      "step": 11089
    },
    {
      "epoch": 0.8961797208024405,
      "grad_norm": 2.6006035804748535,
      "learning_rate": 6.07484118271582e-06,
      "loss": 0.9718,
      "step": 11090
    },
    {
      "epoch": 0.896260530515768,
      "grad_norm": 3.4972732067108154,
      "learning_rate": 6.0742021096747584e-06,
      "loss": 0.8588,
      "step": 11091
    },
    {
      "epoch": 0.8963413402290955,
      "grad_norm": 2.4558448791503906,
      "learning_rate": 6.073563018235202e-06,
      "loss": 0.8262,
      "step": 11092
    },
    {
      "epoch": 0.8964221499424231,
      "grad_norm": 2.393813371658325,
      "learning_rate": 6.0729239084080935e-06,
      "loss": 1.0198,
      "step": 11093
    },
    {
      "epoch": 0.8965029596557507,
      "grad_norm": 2.940751314163208,
      "learning_rate": 6.072284780204383e-06,
      "loss": 0.8167,
      "step": 11094
    },
    {
      "epoch": 0.8965837693690781,
      "grad_norm": 2.8081979751586914,
      "learning_rate": 6.071645633635015e-06,
      "loss": 0.9558,
      "step": 11095
    },
    {
      "epoch": 0.8966645790824057,
      "grad_norm": 3.488110303878784,
      "learning_rate": 6.071006468710936e-06,
      "loss": 0.9361,
      "step": 11096
    },
    {
      "epoch": 0.8967453887957333,
      "grad_norm": 2.4365170001983643,
      "learning_rate": 6.070367285443096e-06,
      "loss": 0.8705,
      "step": 11097
    },
    {
      "epoch": 0.8968261985090608,
      "grad_norm": 2.6625611782073975,
      "learning_rate": 6.069728083842441e-06,
      "loss": 1.0411,
      "step": 11098
    },
    {
      "epoch": 0.8969070082223883,
      "grad_norm": 2.440619707107544,
      "learning_rate": 6.069088863919919e-06,
      "loss": 0.9265,
      "step": 11099
    },
    {
      "epoch": 0.8969878179357159,
      "grad_norm": 2.6538734436035156,
      "learning_rate": 6.06844962568648e-06,
      "loss": 1.0621,
      "step": 11100
    },
    {
      "epoch": 0.8970686276490434,
      "grad_norm": 2.297936201095581,
      "learning_rate": 6.0678103691530686e-06,
      "loss": 0.8572,
      "step": 11101
    },
    {
      "epoch": 0.897149437362371,
      "grad_norm": 2.6770808696746826,
      "learning_rate": 6.067171094330637e-06,
      "loss": 0.9088,
      "step": 11102
    },
    {
      "epoch": 0.8972302470756985,
      "grad_norm": 2.618973970413208,
      "learning_rate": 6.0665318012301345e-06,
      "loss": 0.9641,
      "step": 11103
    },
    {
      "epoch": 0.897311056789026,
      "grad_norm": 2.47927188873291,
      "learning_rate": 6.065892489862509e-06,
      "loss": 0.9679,
      "step": 11104
    },
    {
      "epoch": 0.8973918665023536,
      "grad_norm": 2.793241500854492,
      "learning_rate": 6.065253160238712e-06,
      "loss": 0.9245,
      "step": 11105
    },
    {
      "epoch": 0.8974726762156812,
      "grad_norm": 2.7728309631347656,
      "learning_rate": 6.064613812369692e-06,
      "loss": 0.9167,
      "step": 11106
    },
    {
      "epoch": 0.8975534859290086,
      "grad_norm": 2.337017774581909,
      "learning_rate": 6.063974446266402e-06,
      "loss": 0.9126,
      "step": 11107
    },
    {
      "epoch": 0.8976342956423362,
      "grad_norm": 2.579535722732544,
      "learning_rate": 6.063335061939789e-06,
      "loss": 0.9078,
      "step": 11108
    },
    {
      "epoch": 0.8977151053556638,
      "grad_norm": 2.4813663959503174,
      "learning_rate": 6.062695659400807e-06,
      "loss": 0.9675,
      "step": 11109
    },
    {
      "epoch": 0.8977959150689913,
      "grad_norm": 2.7406249046325684,
      "learning_rate": 6.062056238660408e-06,
      "loss": 0.9199,
      "step": 11110
    },
    {
      "epoch": 0.8978767247823188,
      "grad_norm": 2.772345542907715,
      "learning_rate": 6.061416799729541e-06,
      "loss": 0.9108,
      "step": 11111
    },
    {
      "epoch": 0.8979575344956464,
      "grad_norm": 2.8032472133636475,
      "learning_rate": 6.060777342619159e-06,
      "loss": 0.9564,
      "step": 11112
    },
    {
      "epoch": 0.8980383442089739,
      "grad_norm": 2.816871404647827,
      "learning_rate": 6.060137867340217e-06,
      "loss": 0.818,
      "step": 11113
    },
    {
      "epoch": 0.8981191539223015,
      "grad_norm": 2.663496494293213,
      "learning_rate": 6.0594983739036614e-06,
      "loss": 0.9222,
      "step": 11114
    },
    {
      "epoch": 0.898199963635629,
      "grad_norm": 2.336209774017334,
      "learning_rate": 6.0588588623204535e-06,
      "loss": 0.8684,
      "step": 11115
    },
    {
      "epoch": 0.8982807733489565,
      "grad_norm": 2.6434648036956787,
      "learning_rate": 6.05821933260154e-06,
      "loss": 0.9634,
      "step": 11116
    },
    {
      "epoch": 0.8983615830622841,
      "grad_norm": 2.485868215560913,
      "learning_rate": 6.057579784757879e-06,
      "loss": 0.9125,
      "step": 11117
    },
    {
      "epoch": 0.8984423927756117,
      "grad_norm": 2.5955581665039062,
      "learning_rate": 6.05694021880042e-06,
      "loss": 0.9674,
      "step": 11118
    },
    {
      "epoch": 0.8985232024889391,
      "grad_norm": 3.049374580383301,
      "learning_rate": 6.056300634740121e-06,
      "loss": 0.8956,
      "step": 11119
    },
    {
      "epoch": 0.8986040122022667,
      "grad_norm": 2.6581337451934814,
      "learning_rate": 6.055661032587934e-06,
      "loss": 0.8908,
      "step": 11120
    },
    {
      "epoch": 0.8986848219155943,
      "grad_norm": 2.7783305644989014,
      "learning_rate": 6.055021412354817e-06,
      "loss": 0.9802,
      "step": 11121
    },
    {
      "epoch": 0.8987656316289218,
      "grad_norm": 2.3142786026000977,
      "learning_rate": 6.054381774051721e-06,
      "loss": 1.0919,
      "step": 11122
    },
    {
      "epoch": 0.8988464413422493,
      "grad_norm": 2.6475307941436768,
      "learning_rate": 6.053742117689604e-06,
      "loss": 1.0031,
      "step": 11123
    },
    {
      "epoch": 0.8989272510555769,
      "grad_norm": 3.1276907920837402,
      "learning_rate": 6.053102443279422e-06,
      "loss": 1.0043,
      "step": 11124
    },
    {
      "epoch": 0.8990080607689044,
      "grad_norm": 2.736304759979248,
      "learning_rate": 6.052462750832128e-06,
      "loss": 0.9356,
      "step": 11125
    },
    {
      "epoch": 0.899088870482232,
      "grad_norm": 2.513861894607544,
      "learning_rate": 6.0518230403586845e-06,
      "loss": 0.8544,
      "step": 11126
    },
    {
      "epoch": 0.8991696801955595,
      "grad_norm": 2.8509857654571533,
      "learning_rate": 6.0511833118700425e-06,
      "loss": 1.0235,
      "step": 11127
    },
    {
      "epoch": 0.899250489908887,
      "grad_norm": 2.5635993480682373,
      "learning_rate": 6.05054356537716e-06,
      "loss": 0.916,
      "step": 11128
    },
    {
      "epoch": 0.8993312996222146,
      "grad_norm": 2.838994264602661,
      "learning_rate": 6.049903800890997e-06,
      "loss": 0.8859,
      "step": 11129
    },
    {
      "epoch": 0.8994121093355422,
      "grad_norm": 2.704026699066162,
      "learning_rate": 6.04926401842251e-06,
      "loss": 0.9902,
      "step": 11130
    },
    {
      "epoch": 0.8994929190488696,
      "grad_norm": 2.9368770122528076,
      "learning_rate": 6.048624217982655e-06,
      "loss": 0.9803,
      "step": 11131
    },
    {
      "epoch": 0.8995737287621972,
      "grad_norm": 2.286797046661377,
      "learning_rate": 6.04798439958239e-06,
      "loss": 0.8546,
      "step": 11132
    },
    {
      "epoch": 0.8996545384755248,
      "grad_norm": 2.619764804840088,
      "learning_rate": 6.047344563232678e-06,
      "loss": 1.0141,
      "step": 11133
    },
    {
      "epoch": 0.8997353481888523,
      "grad_norm": 2.3920695781707764,
      "learning_rate": 6.046704708944473e-06,
      "loss": 0.8981,
      "step": 11134
    },
    {
      "epoch": 0.8998161579021798,
      "grad_norm": 2.8092894554138184,
      "learning_rate": 6.046064836728739e-06,
      "loss": 0.9764,
      "step": 11135
    },
    {
      "epoch": 0.8998969676155074,
      "grad_norm": 2.5988481044769287,
      "learning_rate": 6.04542494659643e-06,
      "loss": 0.9469,
      "step": 11136
    },
    {
      "epoch": 0.8999777773288349,
      "grad_norm": 2.972332000732422,
      "learning_rate": 6.044785038558509e-06,
      "loss": 0.8987,
      "step": 11137
    },
    {
      "epoch": 0.9000585870421625,
      "grad_norm": 2.5728893280029297,
      "learning_rate": 6.044145112625937e-06,
      "loss": 1.062,
      "step": 11138
    },
    {
      "epoch": 0.90013939675549,
      "grad_norm": 2.64799165725708,
      "learning_rate": 6.04350516880967e-06,
      "loss": 0.8855,
      "step": 11139
    },
    {
      "epoch": 0.9002202064688175,
      "grad_norm": 2.9226863384246826,
      "learning_rate": 6.042865207120674e-06,
      "loss": 0.8502,
      "step": 11140
    },
    {
      "epoch": 0.9003010161821451,
      "grad_norm": 2.4827048778533936,
      "learning_rate": 6.042225227569908e-06,
      "loss": 0.9605,
      "step": 11141
    },
    {
      "epoch": 0.9003818258954727,
      "grad_norm": 2.5582778453826904,
      "learning_rate": 6.04158523016833e-06,
      "loss": 0.947,
      "step": 11142
    },
    {
      "epoch": 0.9004626356088001,
      "grad_norm": 2.6642837524414062,
      "learning_rate": 6.040945214926906e-06,
      "loss": 0.8842,
      "step": 11143
    },
    {
      "epoch": 0.9005434453221277,
      "grad_norm": 2.6373517513275146,
      "learning_rate": 6.040305181856597e-06,
      "loss": 0.9237,
      "step": 11144
    },
    {
      "epoch": 0.9006242550354553,
      "grad_norm": 2.9530553817749023,
      "learning_rate": 6.039665130968365e-06,
      "loss": 0.93,
      "step": 11145
    },
    {
      "epoch": 0.9007050647487828,
      "grad_norm": 2.6635923385620117,
      "learning_rate": 6.039025062273171e-06,
      "loss": 0.843,
      "step": 11146
    },
    {
      "epoch": 0.9007858744621103,
      "grad_norm": 2.7025904655456543,
      "learning_rate": 6.038384975781979e-06,
      "loss": 0.8554,
      "step": 11147
    },
    {
      "epoch": 0.9008666841754379,
      "grad_norm": 3.593562126159668,
      "learning_rate": 6.037744871505753e-06,
      "loss": 0.9221,
      "step": 11148
    },
    {
      "epoch": 0.9009474938887654,
      "grad_norm": 2.8435449600219727,
      "learning_rate": 6.037104749455457e-06,
      "loss": 0.8964,
      "step": 11149
    },
    {
      "epoch": 0.901028303602093,
      "grad_norm": 2.2871532440185547,
      "learning_rate": 6.03646460964205e-06,
      "loss": 1.0125,
      "step": 11150
    },
    {
      "epoch": 0.9011091133154205,
      "grad_norm": 3.5378856658935547,
      "learning_rate": 6.0358244520765015e-06,
      "loss": 0.8169,
      "step": 11151
    },
    {
      "epoch": 0.901189923028748,
      "grad_norm": 3.1749320030212402,
      "learning_rate": 6.035184276769775e-06,
      "loss": 0.9427,
      "step": 11152
    },
    {
      "epoch": 0.9012707327420756,
      "grad_norm": 2.827220916748047,
      "learning_rate": 6.034544083732832e-06,
      "loss": 0.8491,
      "step": 11153
    },
    {
      "epoch": 0.9013515424554032,
      "grad_norm": 3.2153000831604004,
      "learning_rate": 6.033903872976639e-06,
      "loss": 0.8981,
      "step": 11154
    },
    {
      "epoch": 0.9014323521687306,
      "grad_norm": 2.887916088104248,
      "learning_rate": 6.033263644512162e-06,
      "loss": 1.0583,
      "step": 11155
    },
    {
      "epoch": 0.9015131618820582,
      "grad_norm": 2.70062255859375,
      "learning_rate": 6.032623398350367e-06,
      "loss": 0.9759,
      "step": 11156
    },
    {
      "epoch": 0.9015939715953858,
      "grad_norm": 2.5867743492126465,
      "learning_rate": 6.031983134502219e-06,
      "loss": 0.8996,
      "step": 11157
    },
    {
      "epoch": 0.9016747813087133,
      "grad_norm": 2.6162171363830566,
      "learning_rate": 6.031342852978684e-06,
      "loss": 0.8708,
      "step": 11158
    },
    {
      "epoch": 0.9017555910220408,
      "grad_norm": 2.669729709625244,
      "learning_rate": 6.030702553790729e-06,
      "loss": 0.7751,
      "step": 11159
    },
    {
      "epoch": 0.9018364007353684,
      "grad_norm": 2.3106298446655273,
      "learning_rate": 6.030062236949321e-06,
      "loss": 0.8672,
      "step": 11160
    },
    {
      "epoch": 0.9019172104486959,
      "grad_norm": 2.2521443367004395,
      "learning_rate": 6.029421902465425e-06,
      "loss": 0.8974,
      "step": 11161
    },
    {
      "epoch": 0.9019980201620235,
      "grad_norm": 2.7022483348846436,
      "learning_rate": 6.028781550350011e-06,
      "loss": 0.9219,
      "step": 11162
    },
    {
      "epoch": 0.902078829875351,
      "grad_norm": 3.1653780937194824,
      "learning_rate": 6.028141180614047e-06,
      "loss": 0.9184,
      "step": 11163
    },
    {
      "epoch": 0.9021596395886785,
      "grad_norm": 2.475628614425659,
      "learning_rate": 6.0275007932684986e-06,
      "loss": 0.9396,
      "step": 11164
    },
    {
      "epoch": 0.9022404493020061,
      "grad_norm": 3.138230562210083,
      "learning_rate": 6.026860388324336e-06,
      "loss": 1.0098,
      "step": 11165
    },
    {
      "epoch": 0.9023212590153337,
      "grad_norm": 2.9436910152435303,
      "learning_rate": 6.026219965792526e-06,
      "loss": 1.0456,
      "step": 11166
    },
    {
      "epoch": 0.9024020687286611,
      "grad_norm": 2.330030918121338,
      "learning_rate": 6.025579525684039e-06,
      "loss": 1.0196,
      "step": 11167
    },
    {
      "epoch": 0.9024828784419887,
      "grad_norm": 2.46738600730896,
      "learning_rate": 6.024939068009843e-06,
      "loss": 0.9496,
      "step": 11168
    },
    {
      "epoch": 0.9025636881553163,
      "grad_norm": 2.793869733810425,
      "learning_rate": 6.0242985927809085e-06,
      "loss": 0.8757,
      "step": 11169
    },
    {
      "epoch": 0.9026444978686438,
      "grad_norm": 3.0422780513763428,
      "learning_rate": 6.023658100008205e-06,
      "loss": 1.0815,
      "step": 11170
    },
    {
      "epoch": 0.9027253075819713,
      "grad_norm": 2.429168701171875,
      "learning_rate": 6.0230175897027035e-06,
      "loss": 0.9945,
      "step": 11171
    },
    {
      "epoch": 0.9028061172952989,
      "grad_norm": 2.726804256439209,
      "learning_rate": 6.022377061875372e-06,
      "loss": 0.9619,
      "step": 11172
    },
    {
      "epoch": 0.9028869270086264,
      "grad_norm": 2.9235785007476807,
      "learning_rate": 6.021736516537183e-06,
      "loss": 1.0218,
      "step": 11173
    },
    {
      "epoch": 0.902967736721954,
      "grad_norm": 2.6464831829071045,
      "learning_rate": 6.0210959536991095e-06,
      "loss": 0.9229,
      "step": 11174
    },
    {
      "epoch": 0.9030485464352815,
      "grad_norm": 2.5794496536254883,
      "learning_rate": 6.0204553733721185e-06,
      "loss": 0.8573,
      "step": 11175
    },
    {
      "epoch": 0.903129356148609,
      "grad_norm": 2.67161226272583,
      "learning_rate": 6.019814775567183e-06,
      "loss": 0.95,
      "step": 11176
    },
    {
      "epoch": 0.9032101658619366,
      "grad_norm": 2.5570461750030518,
      "learning_rate": 6.019174160295277e-06,
      "loss": 0.8372,
      "step": 11177
    },
    {
      "epoch": 0.9032909755752642,
      "grad_norm": 2.561553716659546,
      "learning_rate": 6.018533527567369e-06,
      "loss": 0.9202,
      "step": 11178
    },
    {
      "epoch": 0.9033717852885916,
      "grad_norm": 2.3138413429260254,
      "learning_rate": 6.017892877394435e-06,
      "loss": 0.8676,
      "step": 11179
    },
    {
      "epoch": 0.9034525950019192,
      "grad_norm": 2.2205593585968018,
      "learning_rate": 6.017252209787449e-06,
      "loss": 0.8491,
      "step": 11180
    },
    {
      "epoch": 0.9035334047152468,
      "grad_norm": 2.1689064502716064,
      "learning_rate": 6.016611524757378e-06,
      "loss": 1.0517,
      "step": 11181
    },
    {
      "epoch": 0.9036142144285744,
      "grad_norm": 2.705734968185425,
      "learning_rate": 6.015970822315201e-06,
      "loss": 0.9101,
      "step": 11182
    },
    {
      "epoch": 0.9036950241419018,
      "grad_norm": 3.0479984283447266,
      "learning_rate": 6.015330102471889e-06,
      "loss": 0.993,
      "step": 11183
    },
    {
      "epoch": 0.9037758338552294,
      "grad_norm": 3.1683247089385986,
      "learning_rate": 6.0146893652384165e-06,
      "loss": 0.8702,
      "step": 11184
    },
    {
      "epoch": 0.903856643568557,
      "grad_norm": 2.922452688217163,
      "learning_rate": 6.014048610625758e-06,
      "loss": 0.8839,
      "step": 11185
    },
    {
      "epoch": 0.9039374532818845,
      "grad_norm": 2.803436040878296,
      "learning_rate": 6.013407838644888e-06,
      "loss": 0.9096,
      "step": 11186
    },
    {
      "epoch": 0.904018262995212,
      "grad_norm": 2.449906587600708,
      "learning_rate": 6.012767049306781e-06,
      "loss": 0.9873,
      "step": 11187
    },
    {
      "epoch": 0.9040990727085396,
      "grad_norm": 2.6783127784729004,
      "learning_rate": 6.012126242622414e-06,
      "loss": 0.9777,
      "step": 11188
    },
    {
      "epoch": 0.9041798824218671,
      "grad_norm": 2.1947245597839355,
      "learning_rate": 6.011485418602758e-06,
      "loss": 0.992,
      "step": 11189
    },
    {
      "epoch": 0.9042606921351947,
      "grad_norm": 2.6547415256500244,
      "learning_rate": 6.010844577258795e-06,
      "loss": 0.9082,
      "step": 11190
    },
    {
      "epoch": 0.9043415018485222,
      "grad_norm": 3.241971969604492,
      "learning_rate": 6.010203718601497e-06,
      "loss": 0.9937,
      "step": 11191
    },
    {
      "epoch": 0.9044223115618497,
      "grad_norm": 2.841972827911377,
      "learning_rate": 6.009562842641841e-06,
      "loss": 0.95,
      "step": 11192
    },
    {
      "epoch": 0.9045031212751773,
      "grad_norm": 2.440422534942627,
      "learning_rate": 6.008921949390805e-06,
      "loss": 1.0262,
      "step": 11193
    },
    {
      "epoch": 0.9045839309885049,
      "grad_norm": 2.7049107551574707,
      "learning_rate": 6.008281038859364e-06,
      "loss": 0.9716,
      "step": 11194
    },
    {
      "epoch": 0.9046647407018323,
      "grad_norm": 2.8333685398101807,
      "learning_rate": 6.007640111058495e-06,
      "loss": 0.9343,
      "step": 11195
    },
    {
      "epoch": 0.9047455504151599,
      "grad_norm": 3.6632275581359863,
      "learning_rate": 6.006999165999179e-06,
      "loss": 0.9483,
      "step": 11196
    },
    {
      "epoch": 0.9048263601284875,
      "grad_norm": 3.449188709259033,
      "learning_rate": 6.006358203692389e-06,
      "loss": 0.8426,
      "step": 11197
    },
    {
      "epoch": 0.904907169841815,
      "grad_norm": 2.5142784118652344,
      "learning_rate": 6.005717224149108e-06,
      "loss": 0.894,
      "step": 11198
    },
    {
      "epoch": 0.9049879795551425,
      "grad_norm": 2.63183331489563,
      "learning_rate": 6.005076227380311e-06,
      "loss": 0.8921,
      "step": 11199
    },
    {
      "epoch": 0.9050687892684701,
      "grad_norm": 2.4130797386169434,
      "learning_rate": 6.0044352133969774e-06,
      "loss": 0.9531,
      "step": 11200
    },
    {
      "epoch": 0.9051495989817976,
      "grad_norm": 2.420818567276001,
      "learning_rate": 6.0037941822100865e-06,
      "loss": 0.8832,
      "step": 11201
    },
    {
      "epoch": 0.9052304086951252,
      "grad_norm": 2.771743059158325,
      "learning_rate": 6.00315313383062e-06,
      "loss": 0.8849,
      "step": 11202
    },
    {
      "epoch": 0.9053112184084527,
      "grad_norm": 2.915541172027588,
      "learning_rate": 6.002512068269553e-06,
      "loss": 0.8743,
      "step": 11203
    },
    {
      "epoch": 0.9053920281217802,
      "grad_norm": 2.7193355560302734,
      "learning_rate": 6.001870985537869e-06,
      "loss": 1.0206,
      "step": 11204
    },
    {
      "epoch": 0.9054728378351078,
      "grad_norm": 2.4656622409820557,
      "learning_rate": 6.0012298856465455e-06,
      "loss": 0.8858,
      "step": 11205
    },
    {
      "epoch": 0.9055536475484354,
      "grad_norm": 2.5466904640197754,
      "learning_rate": 6.000588768606566e-06,
      "loss": 0.9497,
      "step": 11206
    },
    {
      "epoch": 0.9056344572617628,
      "grad_norm": 2.4730398654937744,
      "learning_rate": 5.999947634428908e-06,
      "loss": 1.0108,
      "step": 11207
    },
    {
      "epoch": 0.9057152669750904,
      "grad_norm": 2.6922435760498047,
      "learning_rate": 5.999306483124557e-06,
      "loss": 0.9906,
      "step": 11208
    },
    {
      "epoch": 0.905796076688418,
      "grad_norm": 2.407792091369629,
      "learning_rate": 5.998665314704488e-06,
      "loss": 0.9333,
      "step": 11209
    },
    {
      "epoch": 0.9058768864017455,
      "grad_norm": 2.6304855346679688,
      "learning_rate": 5.99802412917969e-06,
      "loss": 0.9595,
      "step": 11210
    },
    {
      "epoch": 0.905957696115073,
      "grad_norm": 2.8021180629730225,
      "learning_rate": 5.9973829265611395e-06,
      "loss": 0.9729,
      "step": 11211
    },
    {
      "epoch": 0.9060385058284006,
      "grad_norm": 2.408714532852173,
      "learning_rate": 5.9967417068598214e-06,
      "loss": 0.8815,
      "step": 11212
    },
    {
      "epoch": 0.9061193155417281,
      "grad_norm": 2.826852798461914,
      "learning_rate": 5.996100470086716e-06,
      "loss": 0.968,
      "step": 11213
    },
    {
      "epoch": 0.9062001252550557,
      "grad_norm": 2.453939199447632,
      "learning_rate": 5.995459216252808e-06,
      "loss": 0.8878,
      "step": 11214
    },
    {
      "epoch": 0.9062809349683832,
      "grad_norm": 2.2264769077301025,
      "learning_rate": 5.99481794536908e-06,
      "loss": 1.1432,
      "step": 11215
    },
    {
      "epoch": 0.9063617446817107,
      "grad_norm": 2.6396334171295166,
      "learning_rate": 5.994176657446517e-06,
      "loss": 0.8677,
      "step": 11216
    },
    {
      "epoch": 0.9064425543950383,
      "grad_norm": 2.892902135848999,
      "learning_rate": 5.993535352496099e-06,
      "loss": 0.8049,
      "step": 11217
    },
    {
      "epoch": 0.9065233641083659,
      "grad_norm": 2.7703239917755127,
      "learning_rate": 5.992894030528815e-06,
      "loss": 0.9139,
      "step": 11218
    },
    {
      "epoch": 0.9066041738216933,
      "grad_norm": 2.794983148574829,
      "learning_rate": 5.992252691555644e-06,
      "loss": 0.9116,
      "step": 11219
    },
    {
      "epoch": 0.9066849835350209,
      "grad_norm": 3.235292911529541,
      "learning_rate": 5.9916113355875746e-06,
      "loss": 0.9596,
      "step": 11220
    },
    {
      "epoch": 0.9067657932483485,
      "grad_norm": 2.62172532081604,
      "learning_rate": 5.99096996263559e-06,
      "loss": 0.9795,
      "step": 11221
    },
    {
      "epoch": 0.906846602961676,
      "grad_norm": 2.896568775177002,
      "learning_rate": 5.990328572710675e-06,
      "loss": 0.9722,
      "step": 11222
    },
    {
      "epoch": 0.9069274126750035,
      "grad_norm": 2.6957719326019287,
      "learning_rate": 5.989687165823816e-06,
      "loss": 0.8538,
      "step": 11223
    },
    {
      "epoch": 0.9070082223883311,
      "grad_norm": 2.6205925941467285,
      "learning_rate": 5.989045741985999e-06,
      "loss": 0.9752,
      "step": 11224
    },
    {
      "epoch": 0.9070890321016586,
      "grad_norm": 2.3968451023101807,
      "learning_rate": 5.9884043012082095e-06,
      "loss": 1.0031,
      "step": 11225
    },
    {
      "epoch": 0.9071698418149862,
      "grad_norm": 2.6111690998077393,
      "learning_rate": 5.987762843501432e-06,
      "loss": 0.9345,
      "step": 11226
    },
    {
      "epoch": 0.9072506515283137,
      "grad_norm": 2.851182222366333,
      "learning_rate": 5.987121368876657e-06,
      "loss": 0.9331,
      "step": 11227
    },
    {
      "epoch": 0.9073314612416412,
      "grad_norm": 2.383997678756714,
      "learning_rate": 5.9864798773448686e-06,
      "loss": 0.9757,
      "step": 11228
    },
    {
      "epoch": 0.9074122709549688,
      "grad_norm": 2.7834765911102295,
      "learning_rate": 5.985838368917054e-06,
      "loss": 1.1084,
      "step": 11229
    },
    {
      "epoch": 0.9074930806682964,
      "grad_norm": 2.503272533416748,
      "learning_rate": 5.985196843604202e-06,
      "loss": 0.8915,
      "step": 11230
    },
    {
      "epoch": 0.9075738903816238,
      "grad_norm": 2.509212017059326,
      "learning_rate": 5.984555301417301e-06,
      "loss": 1.0119,
      "step": 11231
    },
    {
      "epoch": 0.9076547000949514,
      "grad_norm": 2.4349122047424316,
      "learning_rate": 5.983913742367337e-06,
      "loss": 0.8884,
      "step": 11232
    },
    {
      "epoch": 0.907735509808279,
      "grad_norm": 2.9278783798217773,
      "learning_rate": 5.983272166465299e-06,
      "loss": 0.9587,
      "step": 11233
    },
    {
      "epoch": 0.9078163195216065,
      "grad_norm": 2.4104089736938477,
      "learning_rate": 5.982630573722176e-06,
      "loss": 0.9787,
      "step": 11234
    },
    {
      "epoch": 0.907897129234934,
      "grad_norm": 2.575124502182007,
      "learning_rate": 5.981988964148957e-06,
      "loss": 0.9259,
      "step": 11235
    },
    {
      "epoch": 0.9079779389482616,
      "grad_norm": 2.7539727687835693,
      "learning_rate": 5.981347337756631e-06,
      "loss": 0.9839,
      "step": 11236
    },
    {
      "epoch": 0.9080587486615891,
      "grad_norm": 2.7652857303619385,
      "learning_rate": 5.980705694556187e-06,
      "loss": 1.0116,
      "step": 11237
    },
    {
      "epoch": 0.9081395583749167,
      "grad_norm": 3.1519527435302734,
      "learning_rate": 5.980064034558616e-06,
      "loss": 0.8814,
      "step": 11238
    },
    {
      "epoch": 0.9082203680882442,
      "grad_norm": 2.9400460720062256,
      "learning_rate": 5.9794223577749085e-06,
      "loss": 0.8787,
      "step": 11239
    },
    {
      "epoch": 0.9083011778015717,
      "grad_norm": 2.1733686923980713,
      "learning_rate": 5.978780664216051e-06,
      "loss": 1.0534,
      "step": 11240
    },
    {
      "epoch": 0.9083819875148993,
      "grad_norm": 2.6477341651916504,
      "learning_rate": 5.97813895389304e-06,
      "loss": 0.8846,
      "step": 11241
    },
    {
      "epoch": 0.9084627972282269,
      "grad_norm": 2.5101115703582764,
      "learning_rate": 5.977497226816863e-06,
      "loss": 0.9324,
      "step": 11242
    },
    {
      "epoch": 0.9085436069415543,
      "grad_norm": 2.8830316066741943,
      "learning_rate": 5.976855482998512e-06,
      "loss": 0.9997,
      "step": 11243
    },
    {
      "epoch": 0.9086244166548819,
      "grad_norm": 2.2825920581817627,
      "learning_rate": 5.976213722448976e-06,
      "loss": 0.8071,
      "step": 11244
    },
    {
      "epoch": 0.9087052263682095,
      "grad_norm": 2.728428840637207,
      "learning_rate": 5.97557194517925e-06,
      "loss": 0.9208,
      "step": 11245
    },
    {
      "epoch": 0.908786036081537,
      "grad_norm": 2.7809622287750244,
      "learning_rate": 5.974930151200327e-06,
      "loss": 0.9794,
      "step": 11246
    },
    {
      "epoch": 0.9088668457948645,
      "grad_norm": 2.585909605026245,
      "learning_rate": 5.974288340523196e-06,
      "loss": 0.9535,
      "step": 11247
    },
    {
      "epoch": 0.9089476555081921,
      "grad_norm": 2.725247859954834,
      "learning_rate": 5.97364651315885e-06,
      "loss": 0.9258,
      "step": 11248
    },
    {
      "epoch": 0.9090284652215196,
      "grad_norm": 2.8630080223083496,
      "learning_rate": 5.973004669118287e-06,
      "loss": 0.8904,
      "step": 11249
    },
    {
      "epoch": 0.9091092749348472,
      "grad_norm": 2.7674100399017334,
      "learning_rate": 5.972362808412493e-06,
      "loss": 0.8673,
      "step": 11250
    },
    {
      "epoch": 0.9091900846481747,
      "grad_norm": 2.413344383239746,
      "learning_rate": 5.971720931052466e-06,
      "loss": 0.8533,
      "step": 11251
    },
    {
      "epoch": 0.9092708943615022,
      "grad_norm": 2.512573480606079,
      "learning_rate": 5.971079037049199e-06,
      "loss": 0.9182,
      "step": 11252
    },
    {
      "epoch": 0.9093517040748298,
      "grad_norm": 2.508984327316284,
      "learning_rate": 5.9704371264136865e-06,
      "loss": 0.9513,
      "step": 11253
    },
    {
      "epoch": 0.9094325137881574,
      "grad_norm": 2.908663749694824,
      "learning_rate": 5.9697951991569205e-06,
      "loss": 0.9486,
      "step": 11254
    },
    {
      "epoch": 0.9095133235014848,
      "grad_norm": 3.1995677947998047,
      "learning_rate": 5.969153255289899e-06,
      "loss": 0.924,
      "step": 11255
    },
    {
      "epoch": 0.9095941332148124,
      "grad_norm": 2.529698133468628,
      "learning_rate": 5.9685112948236135e-06,
      "loss": 0.9569,
      "step": 11256
    },
    {
      "epoch": 0.90967494292814,
      "grad_norm": 2.919224739074707,
      "learning_rate": 5.967869317769062e-06,
      "loss": 0.9777,
      "step": 11257
    },
    {
      "epoch": 0.9097557526414675,
      "grad_norm": 2.4557557106018066,
      "learning_rate": 5.96722732413724e-06,
      "loss": 0.9035,
      "step": 11258
    },
    {
      "epoch": 0.909836562354795,
      "grad_norm": 2.8355343341827393,
      "learning_rate": 5.966585313939142e-06,
      "loss": 1.1082,
      "step": 11259
    },
    {
      "epoch": 0.9099173720681226,
      "grad_norm": 2.2926454544067383,
      "learning_rate": 5.965943287185764e-06,
      "loss": 0.8098,
      "step": 11260
    },
    {
      "epoch": 0.9099981817814501,
      "grad_norm": 2.4217917919158936,
      "learning_rate": 5.965301243888105e-06,
      "loss": 0.985,
      "step": 11261
    },
    {
      "epoch": 0.9100789914947777,
      "grad_norm": 2.621795654296875,
      "learning_rate": 5.964659184057157e-06,
      "loss": 1.0147,
      "step": 11262
    },
    {
      "epoch": 0.9101598012081052,
      "grad_norm": 2.48290753364563,
      "learning_rate": 5.964017107703921e-06,
      "loss": 1.0239,
      "step": 11263
    },
    {
      "epoch": 0.9102406109214327,
      "grad_norm": 2.5411014556884766,
      "learning_rate": 5.9633750148393924e-06,
      "loss": 0.8103,
      "step": 11264
    },
    {
      "epoch": 0.9103214206347603,
      "grad_norm": 2.88592267036438,
      "learning_rate": 5.962732905474569e-06,
      "loss": 0.966,
      "step": 11265
    },
    {
      "epoch": 0.9104022303480879,
      "grad_norm": 3.0505177974700928,
      "learning_rate": 5.962090779620447e-06,
      "loss": 0.8293,
      "step": 11266
    },
    {
      "epoch": 0.9104830400614153,
      "grad_norm": 2.4384031295776367,
      "learning_rate": 5.961448637288029e-06,
      "loss": 0.9133,
      "step": 11267
    },
    {
      "epoch": 0.9105638497747429,
      "grad_norm": 2.8146374225616455,
      "learning_rate": 5.960806478488308e-06,
      "loss": 0.8945,
      "step": 11268
    },
    {
      "epoch": 0.9106446594880705,
      "grad_norm": 2.360950231552124,
      "learning_rate": 5.960164303232287e-06,
      "loss": 0.9087,
      "step": 11269
    },
    {
      "epoch": 0.910725469201398,
      "grad_norm": 2.6275992393493652,
      "learning_rate": 5.95952211153096e-06,
      "loss": 0.9402,
      "step": 11270
    },
    {
      "epoch": 0.9108062789147255,
      "grad_norm": 2.512693166732788,
      "learning_rate": 5.958879903395333e-06,
      "loss": 0.9689,
      "step": 11271
    },
    {
      "epoch": 0.9108870886280531,
      "grad_norm": 2.736257314682007,
      "learning_rate": 5.958237678836399e-06,
      "loss": 0.8769,
      "step": 11272
    },
    {
      "epoch": 0.9109678983413806,
      "grad_norm": 2.865622043609619,
      "learning_rate": 5.957595437865161e-06,
      "loss": 0.8717,
      "step": 11273
    },
    {
      "epoch": 0.9110487080547082,
      "grad_norm": 2.982100486755371,
      "learning_rate": 5.956953180492618e-06,
      "loss": 0.8607,
      "step": 11274
    },
    {
      "epoch": 0.9111295177680357,
      "grad_norm": 2.5475263595581055,
      "learning_rate": 5.956310906729773e-06,
      "loss": 0.8857,
      "step": 11275
    },
    {
      "epoch": 0.9112103274813632,
      "grad_norm": 2.3581409454345703,
      "learning_rate": 5.955668616587622e-06,
      "loss": 1.0149,
      "step": 11276
    },
    {
      "epoch": 0.9112911371946908,
      "grad_norm": 2.7486884593963623,
      "learning_rate": 5.95502631007717e-06,
      "loss": 0.9528,
      "step": 11277
    },
    {
      "epoch": 0.9113719469080184,
      "grad_norm": 2.519463062286377,
      "learning_rate": 5.954383987209416e-06,
      "loss": 1.1178,
      "step": 11278
    },
    {
      "epoch": 0.9114527566213458,
      "grad_norm": 2.4370343685150146,
      "learning_rate": 5.953741647995361e-06,
      "loss": 0.8749,
      "step": 11279
    },
    {
      "epoch": 0.9115335663346734,
      "grad_norm": 2.405346155166626,
      "learning_rate": 5.953099292446007e-06,
      "loss": 0.8948,
      "step": 11280
    },
    {
      "epoch": 0.911614376048001,
      "grad_norm": 2.662438154220581,
      "learning_rate": 5.952456920572358e-06,
      "loss": 0.8326,
      "step": 11281
    },
    {
      "epoch": 0.9116951857613285,
      "grad_norm": 3.1241025924682617,
      "learning_rate": 5.9518145323854145e-06,
      "loss": 0.8969,
      "step": 11282
    },
    {
      "epoch": 0.911775995474656,
      "grad_norm": 3.142529249191284,
      "learning_rate": 5.951172127896181e-06,
      "loss": 0.9985,
      "step": 11283
    },
    {
      "epoch": 0.9118568051879836,
      "grad_norm": 2.650031566619873,
      "learning_rate": 5.950529707115655e-06,
      "loss": 0.8894,
      "step": 11284
    },
    {
      "epoch": 0.9119376149013111,
      "grad_norm": 3.0189099311828613,
      "learning_rate": 5.949887270054848e-06,
      "loss": 1.0169,
      "step": 11285
    },
    {
      "epoch": 0.9120184246146387,
      "grad_norm": 2.3782217502593994,
      "learning_rate": 5.949244816724757e-06,
      "loss": 0.8812,
      "step": 11286
    },
    {
      "epoch": 0.9120992343279662,
      "grad_norm": 2.6692121028900146,
      "learning_rate": 5.948602347136386e-06,
      "loss": 0.874,
      "step": 11287
    },
    {
      "epoch": 0.9121800440412937,
      "grad_norm": 2.5997726917266846,
      "learning_rate": 5.947959861300742e-06,
      "loss": 0.8085,
      "step": 11288
    },
    {
      "epoch": 0.9122608537546213,
      "grad_norm": 2.5190415382385254,
      "learning_rate": 5.947317359228828e-06,
      "loss": 0.9172,
      "step": 11289
    },
    {
      "epoch": 0.9123416634679489,
      "grad_norm": 3.0348873138427734,
      "learning_rate": 5.946674840931647e-06,
      "loss": 0.9727,
      "step": 11290
    },
    {
      "epoch": 0.9124224731812763,
      "grad_norm": 2.6810731887817383,
      "learning_rate": 5.946032306420205e-06,
      "loss": 0.8782,
      "step": 11291
    },
    {
      "epoch": 0.9125032828946039,
      "grad_norm": 2.9404735565185547,
      "learning_rate": 5.945389755705508e-06,
      "loss": 0.9064,
      "step": 11292
    },
    {
      "epoch": 0.9125840926079315,
      "grad_norm": 2.676630735397339,
      "learning_rate": 5.9447471887985595e-06,
      "loss": 1.0119,
      "step": 11293
    },
    {
      "epoch": 0.912664902321259,
      "grad_norm": 2.9339253902435303,
      "learning_rate": 5.944104605710367e-06,
      "loss": 0.8553,
      "step": 11294
    },
    {
      "epoch": 0.9127457120345865,
      "grad_norm": 3.127005100250244,
      "learning_rate": 5.943462006451934e-06,
      "loss": 0.8845,
      "step": 11295
    },
    {
      "epoch": 0.9128265217479141,
      "grad_norm": 2.6678290367126465,
      "learning_rate": 5.942819391034269e-06,
      "loss": 0.9765,
      "step": 11296
    },
    {
      "epoch": 0.9129073314612416,
      "grad_norm": 2.41629695892334,
      "learning_rate": 5.942176759468378e-06,
      "loss": 0.9757,
      "step": 11297
    },
    {
      "epoch": 0.9129881411745692,
      "grad_norm": 2.2940430641174316,
      "learning_rate": 5.941534111765267e-06,
      "loss": 1.0025,
      "step": 11298
    },
    {
      "epoch": 0.9130689508878967,
      "grad_norm": 2.623856782913208,
      "learning_rate": 5.940891447935944e-06,
      "loss": 0.9508,
      "step": 11299
    },
    {
      "epoch": 0.9131497606012242,
      "grad_norm": 2.4364163875579834,
      "learning_rate": 5.940248767991415e-06,
      "loss": 1.0775,
      "step": 11300
    },
    {
      "epoch": 0.9132305703145518,
      "grad_norm": 2.95477032661438,
      "learning_rate": 5.939606071942687e-06,
      "loss": 0.8752,
      "step": 11301
    },
    {
      "epoch": 0.9133113800278794,
      "grad_norm": 2.120548963546753,
      "learning_rate": 5.938963359800771e-06,
      "loss": 0.852,
      "step": 11302
    },
    {
      "epoch": 0.9133921897412068,
      "grad_norm": 2.6929080486297607,
      "learning_rate": 5.938320631576671e-06,
      "loss": 0.9214,
      "step": 11303
    },
    {
      "epoch": 0.9134729994545344,
      "grad_norm": 2.3618075847625732,
      "learning_rate": 5.937677887281399e-06,
      "loss": 0.9602,
      "step": 11304
    },
    {
      "epoch": 0.913553809167862,
      "grad_norm": 2.6843745708465576,
      "learning_rate": 5.937035126925962e-06,
      "loss": 0.9806,
      "step": 11305
    },
    {
      "epoch": 0.9136346188811895,
      "grad_norm": 2.4290525913238525,
      "learning_rate": 5.936392350521369e-06,
      "loss": 0.8757,
      "step": 11306
    },
    {
      "epoch": 0.913715428594517,
      "grad_norm": 2.4888224601745605,
      "learning_rate": 5.9357495580786286e-06,
      "loss": 0.9865,
      "step": 11307
    },
    {
      "epoch": 0.9137962383078446,
      "grad_norm": 2.39178729057312,
      "learning_rate": 5.935106749608752e-06,
      "loss": 0.9695,
      "step": 11308
    },
    {
      "epoch": 0.9138770480211722,
      "grad_norm": 2.755352258682251,
      "learning_rate": 5.934463925122749e-06,
      "loss": 0.9641,
      "step": 11309
    },
    {
      "epoch": 0.9139578577344997,
      "grad_norm": 2.814725875854492,
      "learning_rate": 5.933821084631625e-06,
      "loss": 0.9578,
      "step": 11310
    },
    {
      "epoch": 0.9140386674478272,
      "grad_norm": 2.35859751701355,
      "learning_rate": 5.9331782281463965e-06,
      "loss": 0.8058,
      "step": 11311
    },
    {
      "epoch": 0.9141194771611548,
      "grad_norm": 2.6462783813476562,
      "learning_rate": 5.932535355678072e-06,
      "loss": 0.8071,
      "step": 11312
    },
    {
      "epoch": 0.9142002868744823,
      "grad_norm": 2.5416271686553955,
      "learning_rate": 5.93189246723766e-06,
      "loss": 1.0333,
      "step": 11313
    },
    {
      "epoch": 0.9142810965878099,
      "grad_norm": 2.3862597942352295,
      "learning_rate": 5.931249562836176e-06,
      "loss": 0.9642,
      "step": 11314
    },
    {
      "epoch": 0.9143619063011375,
      "grad_norm": 2.9840433597564697,
      "learning_rate": 5.9306066424846254e-06,
      "loss": 0.8035,
      "step": 11315
    },
    {
      "epoch": 0.9144427160144649,
      "grad_norm": 2.784557342529297,
      "learning_rate": 5.929963706194027e-06,
      "loss": 0.8911,
      "step": 11316
    },
    {
      "epoch": 0.9145235257277925,
      "grad_norm": 2.5682101249694824,
      "learning_rate": 5.929320753975388e-06,
      "loss": 0.9524,
      "step": 11317
    },
    {
      "epoch": 0.9146043354411201,
      "grad_norm": 2.7034237384796143,
      "learning_rate": 5.9286777858397195e-06,
      "loss": 0.9586,
      "step": 11318
    },
    {
      "epoch": 0.9146851451544475,
      "grad_norm": 2.3170664310455322,
      "learning_rate": 5.928034801798037e-06,
      "loss": 0.8701,
      "step": 11319
    },
    {
      "epoch": 0.9147659548677751,
      "grad_norm": 2.2595629692077637,
      "learning_rate": 5.927391801861355e-06,
      "loss": 0.9048,
      "step": 11320
    },
    {
      "epoch": 0.9148467645811027,
      "grad_norm": 2.599877119064331,
      "learning_rate": 5.926748786040681e-06,
      "loss": 0.9804,
      "step": 11321
    },
    {
      "epoch": 0.9149275742944302,
      "grad_norm": 2.9440958499908447,
      "learning_rate": 5.926105754347034e-06,
      "loss": 0.8105,
      "step": 11322
    },
    {
      "epoch": 0.9150083840077577,
      "grad_norm": 2.644850730895996,
      "learning_rate": 5.925462706791421e-06,
      "loss": 1.0119,
      "step": 11323
    },
    {
      "epoch": 0.9150891937210853,
      "grad_norm": 3.3473072052001953,
      "learning_rate": 5.9248196433848645e-06,
      "loss": 0.9653,
      "step": 11324
    },
    {
      "epoch": 0.9151700034344128,
      "grad_norm": 2.77569842338562,
      "learning_rate": 5.924176564138372e-06,
      "loss": 0.8518,
      "step": 11325
    },
    {
      "epoch": 0.9152508131477404,
      "grad_norm": 2.5223371982574463,
      "learning_rate": 5.9235334690629586e-06,
      "loss": 0.9585,
      "step": 11326
    },
    {
      "epoch": 0.915331622861068,
      "grad_norm": 2.3707287311553955,
      "learning_rate": 5.922890358169642e-06,
      "loss": 0.8922,
      "step": 11327
    },
    {
      "epoch": 0.9154124325743954,
      "grad_norm": 3.036858320236206,
      "learning_rate": 5.922247231469435e-06,
      "loss": 0.945,
      "step": 11328
    },
    {
      "epoch": 0.915493242287723,
      "grad_norm": 2.6230826377868652,
      "learning_rate": 5.921604088973352e-06,
      "loss": 0.8826,
      "step": 11329
    },
    {
      "epoch": 0.9155740520010506,
      "grad_norm": 2.5088369846343994,
      "learning_rate": 5.9209609306924106e-06,
      "loss": 0.9406,
      "step": 11330
    },
    {
      "epoch": 0.915654861714378,
      "grad_norm": 2.4708962440490723,
      "learning_rate": 5.9203177566376256e-06,
      "loss": 0.8927,
      "step": 11331
    },
    {
      "epoch": 0.9157356714277056,
      "grad_norm": 2.6827518939971924,
      "learning_rate": 5.919674566820013e-06,
      "loss": 1.0397,
      "step": 11332
    },
    {
      "epoch": 0.9158164811410332,
      "grad_norm": 2.4582741260528564,
      "learning_rate": 5.919031361250588e-06,
      "loss": 0.8563,
      "step": 11333
    },
    {
      "epoch": 0.9158972908543607,
      "grad_norm": 2.6187996864318848,
      "learning_rate": 5.9183881399403694e-06,
      "loss": 0.8182,
      "step": 11334
    },
    {
      "epoch": 0.9159781005676882,
      "grad_norm": 2.827922821044922,
      "learning_rate": 5.917744902900372e-06,
      "loss": 0.9741,
      "step": 11335
    },
    {
      "epoch": 0.9160589102810158,
      "grad_norm": 3.082225799560547,
      "learning_rate": 5.917101650141616e-06,
      "loss": 1.0021,
      "step": 11336
    },
    {
      "epoch": 0.9161397199943433,
      "grad_norm": 2.4827330112457275,
      "learning_rate": 5.916458381675113e-06,
      "loss": 0.9795,
      "step": 11337
    },
    {
      "epoch": 0.9162205297076709,
      "grad_norm": 3.0935449600219727,
      "learning_rate": 5.915815097511887e-06,
      "loss": 0.8948,
      "step": 11338
    },
    {
      "epoch": 0.9163013394209985,
      "grad_norm": 2.7609193325042725,
      "learning_rate": 5.915171797662952e-06,
      "loss": 1.0349,
      "step": 11339
    },
    {
      "epoch": 0.9163821491343259,
      "grad_norm": 2.6350550651550293,
      "learning_rate": 5.914528482139328e-06,
      "loss": 0.973,
      "step": 11340
    },
    {
      "epoch": 0.9164629588476535,
      "grad_norm": 2.7748446464538574,
      "learning_rate": 5.913885150952032e-06,
      "loss": 0.879,
      "step": 11341
    },
    {
      "epoch": 0.9165437685609811,
      "grad_norm": 2.544610023498535,
      "learning_rate": 5.9132418041120845e-06,
      "loss": 0.8538,
      "step": 11342
    },
    {
      "epoch": 0.9166245782743085,
      "grad_norm": 2.6684980392456055,
      "learning_rate": 5.912598441630501e-06,
      "loss": 0.893,
      "step": 11343
    },
    {
      "epoch": 0.9167053879876361,
      "grad_norm": 2.9629220962524414,
      "learning_rate": 5.911955063518307e-06,
      "loss": 0.8841,
      "step": 11344
    },
    {
      "epoch": 0.9167861977009637,
      "grad_norm": 2.4688479900360107,
      "learning_rate": 5.9113116697865145e-06,
      "loss": 0.9434,
      "step": 11345
    },
    {
      "epoch": 0.9168670074142912,
      "grad_norm": 2.7888643741607666,
      "learning_rate": 5.910668260446149e-06,
      "loss": 0.8742,
      "step": 11346
    },
    {
      "epoch": 0.9169478171276187,
      "grad_norm": 2.280099630355835,
      "learning_rate": 5.910024835508228e-06,
      "loss": 0.8853,
      "step": 11347
    },
    {
      "epoch": 0.9170286268409463,
      "grad_norm": 2.539198160171509,
      "learning_rate": 5.9093813949837735e-06,
      "loss": 1.0329,
      "step": 11348
    },
    {
      "epoch": 0.9171094365542738,
      "grad_norm": 2.62003493309021,
      "learning_rate": 5.908737938883803e-06,
      "loss": 0.9036,
      "step": 11349
    },
    {
      "epoch": 0.9171902462676014,
      "grad_norm": 2.6804158687591553,
      "learning_rate": 5.908094467219341e-06,
      "loss": 0.9506,
      "step": 11350
    },
    {
      "epoch": 0.917271055980929,
      "grad_norm": 2.666212558746338,
      "learning_rate": 5.907450980001405e-06,
      "loss": 0.9333,
      "step": 11351
    },
    {
      "epoch": 0.9173518656942564,
      "grad_norm": 3.5210061073303223,
      "learning_rate": 5.90680747724102e-06,
      "loss": 0.8667,
      "step": 11352
    },
    {
      "epoch": 0.917432675407584,
      "grad_norm": 2.6402230262756348,
      "learning_rate": 5.906163958949205e-06,
      "loss": 0.8378,
      "step": 11353
    },
    {
      "epoch": 0.9175134851209116,
      "grad_norm": 3.05657696723938,
      "learning_rate": 5.905520425136983e-06,
      "loss": 0.9953,
      "step": 11354
    },
    {
      "epoch": 0.917594294834239,
      "grad_norm": 2.8205697536468506,
      "learning_rate": 5.904876875815376e-06,
      "loss": 0.9638,
      "step": 11355
    },
    {
      "epoch": 0.9176751045475666,
      "grad_norm": 2.873640775680542,
      "learning_rate": 5.904233310995409e-06,
      "loss": 0.8999,
      "step": 11356
    },
    {
      "epoch": 0.9177559142608942,
      "grad_norm": 2.755690813064575,
      "learning_rate": 5.9035897306880986e-06,
      "loss": 0.8995,
      "step": 11357
    },
    {
      "epoch": 0.9178367239742217,
      "grad_norm": 2.4933114051818848,
      "learning_rate": 5.902946134904473e-06,
      "loss": 0.846,
      "step": 11358
    },
    {
      "epoch": 0.9179175336875492,
      "grad_norm": 2.5199472904205322,
      "learning_rate": 5.9023025236555545e-06,
      "loss": 0.7741,
      "step": 11359
    },
    {
      "epoch": 0.9179983434008768,
      "grad_norm": 2.5322792530059814,
      "learning_rate": 5.901658896952365e-06,
      "loss": 0.787,
      "step": 11360
    },
    {
      "epoch": 0.9180791531142043,
      "grad_norm": 2.953186273574829,
      "learning_rate": 5.901015254805929e-06,
      "loss": 0.9253,
      "step": 11361
    },
    {
      "epoch": 0.9181599628275319,
      "grad_norm": 2.3440465927124023,
      "learning_rate": 5.900371597227271e-06,
      "loss": 0.9632,
      "step": 11362
    },
    {
      "epoch": 0.9182407725408595,
      "grad_norm": 2.270451307296753,
      "learning_rate": 5.899727924227416e-06,
      "loss": 1.0529,
      "step": 11363
    },
    {
      "epoch": 0.9183215822541869,
      "grad_norm": 2.456655502319336,
      "learning_rate": 5.899084235817387e-06,
      "loss": 0.9195,
      "step": 11364
    },
    {
      "epoch": 0.9184023919675145,
      "grad_norm": 2.6576099395751953,
      "learning_rate": 5.89844053200821e-06,
      "loss": 0.9569,
      "step": 11365
    },
    {
      "epoch": 0.9184832016808421,
      "grad_norm": 3.1898820400238037,
      "learning_rate": 5.89779681281091e-06,
      "loss": 0.8924,
      "step": 11366
    },
    {
      "epoch": 0.9185640113941695,
      "grad_norm": 2.303389310836792,
      "learning_rate": 5.8971530782365105e-06,
      "loss": 0.9054,
      "step": 11367
    },
    {
      "epoch": 0.9186448211074971,
      "grad_norm": 2.348397731781006,
      "learning_rate": 5.896509328296038e-06,
      "loss": 0.8843,
      "step": 11368
    },
    {
      "epoch": 0.9187256308208247,
      "grad_norm": 2.2865302562713623,
      "learning_rate": 5.895865563000521e-06,
      "loss": 0.8138,
      "step": 11369
    },
    {
      "epoch": 0.9188064405341522,
      "grad_norm": 2.529402256011963,
      "learning_rate": 5.895221782360983e-06,
      "loss": 0.8935,
      "step": 11370
    },
    {
      "epoch": 0.9188872502474797,
      "grad_norm": 2.8028602600097656,
      "learning_rate": 5.89457798638845e-06,
      "loss": 1.0034,
      "step": 11371
    },
    {
      "epoch": 0.9189680599608073,
      "grad_norm": 2.6138877868652344,
      "learning_rate": 5.893934175093951e-06,
      "loss": 0.944,
      "step": 11372
    },
    {
      "epoch": 0.9190488696741348,
      "grad_norm": 2.643620729446411,
      "learning_rate": 5.893290348488512e-06,
      "loss": 0.7615,
      "step": 11373
    },
    {
      "epoch": 0.9191296793874624,
      "grad_norm": 2.6800639629364014,
      "learning_rate": 5.892646506583158e-06,
      "loss": 0.9227,
      "step": 11374
    },
    {
      "epoch": 0.91921048910079,
      "grad_norm": 2.541599988937378,
      "learning_rate": 5.89200264938892e-06,
      "loss": 0.9378,
      "step": 11375
    },
    {
      "epoch": 0.9192912988141174,
      "grad_norm": 2.6753969192504883,
      "learning_rate": 5.891358776916822e-06,
      "loss": 0.9243,
      "step": 11376
    },
    {
      "epoch": 0.919372108527445,
      "grad_norm": 2.499925374984741,
      "learning_rate": 5.890714889177895e-06,
      "loss": 0.9163,
      "step": 11377
    },
    {
      "epoch": 0.9194529182407726,
      "grad_norm": 2.6245625019073486,
      "learning_rate": 5.890070986183168e-06,
      "loss": 1.0037,
      "step": 11378
    },
    {
      "epoch": 0.9195337279541,
      "grad_norm": 2.6513335704803467,
      "learning_rate": 5.889427067943665e-06,
      "loss": 0.8895,
      "step": 11379
    },
    {
      "epoch": 0.9196145376674276,
      "grad_norm": 2.487545967102051,
      "learning_rate": 5.8887831344704195e-06,
      "loss": 0.8848,
      "step": 11380
    },
    {
      "epoch": 0.9196953473807552,
      "grad_norm": 2.648772954940796,
      "learning_rate": 5.888139185774459e-06,
      "loss": 0.7962,
      "step": 11381
    },
    {
      "epoch": 0.9197761570940827,
      "grad_norm": 2.3208978176116943,
      "learning_rate": 5.887495221866811e-06,
      "loss": 1.0391,
      "step": 11382
    },
    {
      "epoch": 0.9198569668074102,
      "grad_norm": 2.3278167247772217,
      "learning_rate": 5.8868512427585064e-06,
      "loss": 0.8511,
      "step": 11383
    },
    {
      "epoch": 0.9199377765207378,
      "grad_norm": 2.843883514404297,
      "learning_rate": 5.886207248460575e-06,
      "loss": 0.9639,
      "step": 11384
    },
    {
      "epoch": 0.9200185862340653,
      "grad_norm": 2.954684257507324,
      "learning_rate": 5.885563238984046e-06,
      "loss": 0.8697,
      "step": 11385
    },
    {
      "epoch": 0.9200993959473929,
      "grad_norm": 3.011845827102661,
      "learning_rate": 5.884919214339952e-06,
      "loss": 0.9803,
      "step": 11386
    },
    {
      "epoch": 0.9201802056607205,
      "grad_norm": 2.6969895362854004,
      "learning_rate": 5.884275174539324e-06,
      "loss": 0.934,
      "step": 11387
    },
    {
      "epoch": 0.9202610153740479,
      "grad_norm": 2.4887447357177734,
      "learning_rate": 5.883631119593187e-06,
      "loss": 0.9138,
      "step": 11388
    },
    {
      "epoch": 0.9203418250873755,
      "grad_norm": 2.7332000732421875,
      "learning_rate": 5.88298704951258e-06,
      "loss": 0.8979,
      "step": 11389
    },
    {
      "epoch": 0.9204226348007031,
      "grad_norm": 2.726731061935425,
      "learning_rate": 5.8823429643085275e-06,
      "loss": 0.8644,
      "step": 11390
    },
    {
      "epoch": 0.9205034445140305,
      "grad_norm": 2.412320852279663,
      "learning_rate": 5.881698863992067e-06,
      "loss": 0.9154,
      "step": 11391
    },
    {
      "epoch": 0.9205842542273581,
      "grad_norm": 2.885359048843384,
      "learning_rate": 5.881054748574226e-06,
      "loss": 1.0249,
      "step": 11392
    },
    {
      "epoch": 0.9206650639406857,
      "grad_norm": 2.8027565479278564,
      "learning_rate": 5.880410618066038e-06,
      "loss": 0.8415,
      "step": 11393
    },
    {
      "epoch": 0.9207458736540132,
      "grad_norm": 3.0334484577178955,
      "learning_rate": 5.879766472478535e-06,
      "loss": 0.919,
      "step": 11394
    },
    {
      "epoch": 0.9208266833673407,
      "grad_norm": 2.489262342453003,
      "learning_rate": 5.879122311822752e-06,
      "loss": 0.9561,
      "step": 11395
    },
    {
      "epoch": 0.9209074930806683,
      "grad_norm": 2.3767006397247314,
      "learning_rate": 5.878478136109719e-06,
      "loss": 1.0179,
      "step": 11396
    },
    {
      "epoch": 0.9209883027939958,
      "grad_norm": 2.608154058456421,
      "learning_rate": 5.877833945350471e-06,
      "loss": 0.9476,
      "step": 11397
    },
    {
      "epoch": 0.9210691125073234,
      "grad_norm": 2.684385061264038,
      "learning_rate": 5.87718973955604e-06,
      "loss": 0.9262,
      "step": 11398
    },
    {
      "epoch": 0.921149922220651,
      "grad_norm": 2.503622055053711,
      "learning_rate": 5.876545518737462e-06,
      "loss": 0.9795,
      "step": 11399
    },
    {
      "epoch": 0.9212307319339784,
      "grad_norm": 2.7895336151123047,
      "learning_rate": 5.875901282905768e-06,
      "loss": 0.8742,
      "step": 11400
    },
    {
      "epoch": 0.921311541647306,
      "grad_norm": 2.5915136337280273,
      "learning_rate": 5.875257032071995e-06,
      "loss": 0.8651,
      "step": 11401
    },
    {
      "epoch": 0.9213923513606336,
      "grad_norm": 2.8383326530456543,
      "learning_rate": 5.874612766247174e-06,
      "loss": 0.9806,
      "step": 11402
    },
    {
      "epoch": 0.921473161073961,
      "grad_norm": 2.4563441276550293,
      "learning_rate": 5.873968485442345e-06,
      "loss": 0.7725,
      "step": 11403
    },
    {
      "epoch": 0.9215539707872886,
      "grad_norm": 2.784400463104248,
      "learning_rate": 5.8733241896685366e-06,
      "loss": 1.0246,
      "step": 11404
    },
    {
      "epoch": 0.9216347805006162,
      "grad_norm": 2.898120880126953,
      "learning_rate": 5.8726798789367886e-06,
      "loss": 0.9011,
      "step": 11405
    },
    {
      "epoch": 0.9217155902139437,
      "grad_norm": 3.285613775253296,
      "learning_rate": 5.872035553258136e-06,
      "loss": 0.9468,
      "step": 11406
    },
    {
      "epoch": 0.9217963999272712,
      "grad_norm": 2.813210964202881,
      "learning_rate": 5.871391212643614e-06,
      "loss": 0.8887,
      "step": 11407
    },
    {
      "epoch": 0.9218772096405988,
      "grad_norm": 2.130511999130249,
      "learning_rate": 5.870746857104256e-06,
      "loss": 0.954,
      "step": 11408
    },
    {
      "epoch": 0.9219580193539263,
      "grad_norm": 2.8262295722961426,
      "learning_rate": 5.870102486651102e-06,
      "loss": 0.8483,
      "step": 11409
    },
    {
      "epoch": 0.9220388290672539,
      "grad_norm": 2.5086472034454346,
      "learning_rate": 5.8694581012951866e-06,
      "loss": 0.9501,
      "step": 11410
    },
    {
      "epoch": 0.9221196387805815,
      "grad_norm": 3.1171772480010986,
      "learning_rate": 5.868813701047549e-06,
      "loss": 0.8756,
      "step": 11411
    },
    {
      "epoch": 0.9222004484939089,
      "grad_norm": 2.7673299312591553,
      "learning_rate": 5.868169285919222e-06,
      "loss": 1.1057,
      "step": 11412
    },
    {
      "epoch": 0.9222812582072365,
      "grad_norm": 2.42545747756958,
      "learning_rate": 5.867524855921246e-06,
      "loss": 0.9058,
      "step": 11413
    },
    {
      "epoch": 0.9223620679205641,
      "grad_norm": 2.694911479949951,
      "learning_rate": 5.866880411064657e-06,
      "loss": 0.9866,
      "step": 11414
    },
    {
      "epoch": 0.9224428776338915,
      "grad_norm": 2.726897716522217,
      "learning_rate": 5.866235951360495e-06,
      "loss": 0.9312,
      "step": 11415
    },
    {
      "epoch": 0.9225236873472191,
      "grad_norm": 2.3903236389160156,
      "learning_rate": 5.865591476819794e-06,
      "loss": 1.0821,
      "step": 11416
    },
    {
      "epoch": 0.9226044970605467,
      "grad_norm": 2.8196592330932617,
      "learning_rate": 5.864946987453598e-06,
      "loss": 0.8985,
      "step": 11417
    },
    {
      "epoch": 0.9226853067738742,
      "grad_norm": 2.540738344192505,
      "learning_rate": 5.864302483272939e-06,
      "loss": 0.8386,
      "step": 11418
    },
    {
      "epoch": 0.9227661164872017,
      "grad_norm": 2.7647039890289307,
      "learning_rate": 5.863657964288863e-06,
      "loss": 1.019,
      "step": 11419
    },
    {
      "epoch": 0.9228469262005293,
      "grad_norm": 2.655989170074463,
      "learning_rate": 5.8630134305124035e-06,
      "loss": 0.8882,
      "step": 11420
    },
    {
      "epoch": 0.9229277359138568,
      "grad_norm": 2.9405517578125,
      "learning_rate": 5.8623688819546e-06,
      "loss": 1.0154,
      "step": 11421
    },
    {
      "epoch": 0.9230085456271844,
      "grad_norm": 3.0355875492095947,
      "learning_rate": 5.861724318626495e-06,
      "loss": 0.997,
      "step": 11422
    },
    {
      "epoch": 0.923089355340512,
      "grad_norm": 2.91152024269104,
      "learning_rate": 5.861079740539128e-06,
      "loss": 0.9131,
      "step": 11423
    },
    {
      "epoch": 0.9231701650538394,
      "grad_norm": 2.527146339416504,
      "learning_rate": 5.860435147703536e-06,
      "loss": 1.0204,
      "step": 11424
    },
    {
      "epoch": 0.923250974767167,
      "grad_norm": 3.000993251800537,
      "learning_rate": 5.8597905401307634e-06,
      "loss": 0.9303,
      "step": 11425
    },
    {
      "epoch": 0.9233317844804946,
      "grad_norm": 2.335820198059082,
      "learning_rate": 5.8591459178318465e-06,
      "loss": 0.8627,
      "step": 11426
    },
    {
      "epoch": 0.923412594193822,
      "grad_norm": 2.655054807662964,
      "learning_rate": 5.85850128081783e-06,
      "loss": 0.8509,
      "step": 11427
    },
    {
      "epoch": 0.9234934039071496,
      "grad_norm": 2.9699034690856934,
      "learning_rate": 5.857856629099752e-06,
      "loss": 0.9142,
      "step": 11428
    },
    {
      "epoch": 0.9235742136204772,
      "grad_norm": 2.5918145179748535,
      "learning_rate": 5.857211962688656e-06,
      "loss": 1.0522,
      "step": 11429
    },
    {
      "epoch": 0.9236550233338047,
      "grad_norm": 2.4602110385894775,
      "learning_rate": 5.856567281595582e-06,
      "loss": 0.8648,
      "step": 11430
    },
    {
      "epoch": 0.9237358330471322,
      "grad_norm": 2.568098783493042,
      "learning_rate": 5.855922585831573e-06,
      "loss": 0.8071,
      "step": 11431
    },
    {
      "epoch": 0.9238166427604598,
      "grad_norm": 2.309749126434326,
      "learning_rate": 5.85527787540767e-06,
      "loss": 0.9676,
      "step": 11432
    },
    {
      "epoch": 0.9238974524737873,
      "grad_norm": 2.800462245941162,
      "learning_rate": 5.8546331503349185e-06,
      "loss": 0.8561,
      "step": 11433
    },
    {
      "epoch": 0.9239782621871149,
      "grad_norm": 2.488215446472168,
      "learning_rate": 5.853988410624356e-06,
      "loss": 0.8526,
      "step": 11434
    },
    {
      "epoch": 0.9240590719004425,
      "grad_norm": 3.203184127807617,
      "learning_rate": 5.853343656287029e-06,
      "loss": 0.892,
      "step": 11435
    },
    {
      "epoch": 0.9241398816137699,
      "grad_norm": 2.5497629642486572,
      "learning_rate": 5.852698887333979e-06,
      "loss": 0.942,
      "step": 11436
    },
    {
      "epoch": 0.9242206913270975,
      "grad_norm": 2.2810022830963135,
      "learning_rate": 5.85205410377625e-06,
      "loss": 0.9922,
      "step": 11437
    },
    {
      "epoch": 0.9243015010404251,
      "grad_norm": 3.3598952293395996,
      "learning_rate": 5.851409305624886e-06,
      "loss": 0.9115,
      "step": 11438
    },
    {
      "epoch": 0.9243823107537527,
      "grad_norm": 2.994471788406372,
      "learning_rate": 5.850764492890929e-06,
      "loss": 0.9589,
      "step": 11439
    },
    {
      "epoch": 0.9244631204670801,
      "grad_norm": 2.88105845451355,
      "learning_rate": 5.850119665585427e-06,
      "loss": 0.9863,
      "step": 11440
    },
    {
      "epoch": 0.9245439301804077,
      "grad_norm": 2.943375825881958,
      "learning_rate": 5.8494748237194184e-06,
      "loss": 0.8862,
      "step": 11441
    },
    {
      "epoch": 0.9246247398937353,
      "grad_norm": 2.7711851596832275,
      "learning_rate": 5.8488299673039525e-06,
      "loss": 0.8328,
      "step": 11442
    },
    {
      "epoch": 0.9247055496070627,
      "grad_norm": 2.6126210689544678,
      "learning_rate": 5.848185096350073e-06,
      "loss": 1.066,
      "step": 11443
    },
    {
      "epoch": 0.9247863593203903,
      "grad_norm": 2.4791460037231445,
      "learning_rate": 5.847540210868825e-06,
      "loss": 1.0345,
      "step": 11444
    },
    {
      "epoch": 0.9248671690337179,
      "grad_norm": 2.5218231678009033,
      "learning_rate": 5.846895310871252e-06,
      "loss": 0.9926,
      "step": 11445
    },
    {
      "epoch": 0.9249479787470454,
      "grad_norm": 2.2506418228149414,
      "learning_rate": 5.846250396368403e-06,
      "loss": 0.9138,
      "step": 11446
    },
    {
      "epoch": 0.925028788460373,
      "grad_norm": 2.818552017211914,
      "learning_rate": 5.8456054673713215e-06,
      "loss": 0.9325,
      "step": 11447
    },
    {
      "epoch": 0.9251095981737005,
      "grad_norm": 2.802095413208008,
      "learning_rate": 5.844960523891054e-06,
      "loss": 0.8851,
      "step": 11448
    },
    {
      "epoch": 0.925190407887028,
      "grad_norm": 2.4871411323547363,
      "learning_rate": 5.844315565938645e-06,
      "loss": 0.8751,
      "step": 11449
    },
    {
      "epoch": 0.9252712176003556,
      "grad_norm": 2.527529001235962,
      "learning_rate": 5.843670593525146e-06,
      "loss": 0.9122,
      "step": 11450
    },
    {
      "epoch": 0.9253520273136832,
      "grad_norm": 2.765439987182617,
      "learning_rate": 5.8430256066616e-06,
      "loss": 0.9113,
      "step": 11451
    },
    {
      "epoch": 0.9254328370270106,
      "grad_norm": 2.457642078399658,
      "learning_rate": 5.842380605359054e-06,
      "loss": 0.9469,
      "step": 11452
    },
    {
      "epoch": 0.9255136467403382,
      "grad_norm": 2.3127601146698,
      "learning_rate": 5.841735589628556e-06,
      "loss": 0.8463,
      "step": 11453
    },
    {
      "epoch": 0.9255944564536658,
      "grad_norm": 2.3116958141326904,
      "learning_rate": 5.841090559481155e-06,
      "loss": 1.0376,
      "step": 11454
    },
    {
      "epoch": 0.9256752661669932,
      "grad_norm": 2.455389976501465,
      "learning_rate": 5.840445514927896e-06,
      "loss": 1.0043,
      "step": 11455
    },
    {
      "epoch": 0.9257560758803208,
      "grad_norm": 2.7778866291046143,
      "learning_rate": 5.839800455979829e-06,
      "loss": 1.0258,
      "step": 11456
    },
    {
      "epoch": 0.9258368855936484,
      "grad_norm": 2.987760305404663,
      "learning_rate": 5.839155382648003e-06,
      "loss": 0.9582,
      "step": 11457
    },
    {
      "epoch": 0.9259176953069759,
      "grad_norm": 2.9499425888061523,
      "learning_rate": 5.838510294943465e-06,
      "loss": 0.858,
      "step": 11458
    },
    {
      "epoch": 0.9259985050203035,
      "grad_norm": 2.903751850128174,
      "learning_rate": 5.837865192877263e-06,
      "loss": 0.815,
      "step": 11459
    },
    {
      "epoch": 0.926079314733631,
      "grad_norm": 2.858501434326172,
      "learning_rate": 5.837220076460449e-06,
      "loss": 0.831,
      "step": 11460
    },
    {
      "epoch": 0.9261601244469585,
      "grad_norm": 2.4036664962768555,
      "learning_rate": 5.83657494570407e-06,
      "loss": 0.8755,
      "step": 11461
    },
    {
      "epoch": 0.9262409341602861,
      "grad_norm": 2.3154263496398926,
      "learning_rate": 5.835929800619177e-06,
      "loss": 0.9799,
      "step": 11462
    },
    {
      "epoch": 0.9263217438736137,
      "grad_norm": 2.4877054691314697,
      "learning_rate": 5.835284641216816e-06,
      "loss": 0.8289,
      "step": 11463
    },
    {
      "epoch": 0.9264025535869411,
      "grad_norm": 2.2633888721466064,
      "learning_rate": 5.8346394675080446e-06,
      "loss": 1.1154,
      "step": 11464
    },
    {
      "epoch": 0.9264833633002687,
      "grad_norm": 2.742274045944214,
      "learning_rate": 5.8339942795039055e-06,
      "loss": 0.9021,
      "step": 11465
    },
    {
      "epoch": 0.9265641730135963,
      "grad_norm": 2.681795120239258,
      "learning_rate": 5.833349077215452e-06,
      "loss": 0.9081,
      "step": 11466
    },
    {
      "epoch": 0.9266449827269237,
      "grad_norm": 2.5320637226104736,
      "learning_rate": 5.832703860653736e-06,
      "loss": 0.9118,
      "step": 11467
    },
    {
      "epoch": 0.9267257924402513,
      "grad_norm": 2.741290330886841,
      "learning_rate": 5.832058629829808e-06,
      "loss": 0.9423,
      "step": 11468
    },
    {
      "epoch": 0.9268066021535789,
      "grad_norm": 2.9602246284484863,
      "learning_rate": 5.831413384754716e-06,
      "loss": 0.903,
      "step": 11469
    },
    {
      "epoch": 0.9268874118669064,
      "grad_norm": 2.332145929336548,
      "learning_rate": 5.8307681254395165e-06,
      "loss": 0.9625,
      "step": 11470
    },
    {
      "epoch": 0.926968221580234,
      "grad_norm": 2.47112774848938,
      "learning_rate": 5.830122851895259e-06,
      "loss": 0.9085,
      "step": 11471
    },
    {
      "epoch": 0.9270490312935615,
      "grad_norm": 2.4056050777435303,
      "learning_rate": 5.8294775641329945e-06,
      "loss": 0.9285,
      "step": 11472
    },
    {
      "epoch": 0.927129841006889,
      "grad_norm": 2.653259754180908,
      "learning_rate": 5.8288322621637776e-06,
      "loss": 0.9261,
      "step": 11473
    },
    {
      "epoch": 0.9272106507202166,
      "grad_norm": 2.3907012939453125,
      "learning_rate": 5.8281869459986585e-06,
      "loss": 0.9271,
      "step": 11474
    },
    {
      "epoch": 0.9272914604335442,
      "grad_norm": 3.4400665760040283,
      "learning_rate": 5.82754161564869e-06,
      "loss": 0.964,
      "step": 11475
    },
    {
      "epoch": 0.9273722701468716,
      "grad_norm": 3.1091995239257812,
      "learning_rate": 5.826896271124928e-06,
      "loss": 0.8983,
      "step": 11476
    },
    {
      "epoch": 0.9274530798601992,
      "grad_norm": 2.6850452423095703,
      "learning_rate": 5.826250912438421e-06,
      "loss": 0.9606,
      "step": 11477
    },
    {
      "epoch": 0.9275338895735268,
      "grad_norm": 2.686000108718872,
      "learning_rate": 5.8256055396002275e-06,
      "loss": 0.9391,
      "step": 11478
    },
    {
      "epoch": 0.9276146992868542,
      "grad_norm": 2.9149484634399414,
      "learning_rate": 5.824960152621397e-06,
      "loss": 0.9082,
      "step": 11479
    },
    {
      "epoch": 0.9276955090001818,
      "grad_norm": 2.213421583175659,
      "learning_rate": 5.824314751512985e-06,
      "loss": 0.9635,
      "step": 11480
    },
    {
      "epoch": 0.9277763187135094,
      "grad_norm": 2.699484348297119,
      "learning_rate": 5.823669336286046e-06,
      "loss": 0.9009,
      "step": 11481
    },
    {
      "epoch": 0.9278571284268369,
      "grad_norm": 2.49141526222229,
      "learning_rate": 5.823023906951636e-06,
      "loss": 0.9613,
      "step": 11482
    },
    {
      "epoch": 0.9279379381401645,
      "grad_norm": 2.307377815246582,
      "learning_rate": 5.822378463520805e-06,
      "loss": 0.8796,
      "step": 11483
    },
    {
      "epoch": 0.928018747853492,
      "grad_norm": 2.557276487350464,
      "learning_rate": 5.821733006004613e-06,
      "loss": 0.9419,
      "step": 11484
    },
    {
      "epoch": 0.9280995575668195,
      "grad_norm": 2.382101058959961,
      "learning_rate": 5.821087534414112e-06,
      "loss": 0.9555,
      "step": 11485
    },
    {
      "epoch": 0.9281803672801471,
      "grad_norm": 2.5004258155822754,
      "learning_rate": 5.820442048760357e-06,
      "loss": 0.9311,
      "step": 11486
    },
    {
      "epoch": 0.9282611769934747,
      "grad_norm": 3.6078615188598633,
      "learning_rate": 5.8197965490544064e-06,
      "loss": 0.9319,
      "step": 11487
    },
    {
      "epoch": 0.9283419867068021,
      "grad_norm": 2.3218090534210205,
      "learning_rate": 5.819151035307314e-06,
      "loss": 0.9341,
      "step": 11488
    },
    {
      "epoch": 0.9284227964201297,
      "grad_norm": 2.539309501647949,
      "learning_rate": 5.818505507530137e-06,
      "loss": 0.933,
      "step": 11489
    },
    {
      "epoch": 0.9285036061334573,
      "grad_norm": 2.954179048538208,
      "learning_rate": 5.8178599657339305e-06,
      "loss": 0.9342,
      "step": 11490
    },
    {
      "epoch": 0.9285844158467847,
      "grad_norm": 2.8786184787750244,
      "learning_rate": 5.817214409929751e-06,
      "loss": 0.8229,
      "step": 11491
    },
    {
      "epoch": 0.9286652255601123,
      "grad_norm": 2.192934036254883,
      "learning_rate": 5.816568840128658e-06,
      "loss": 1.0281,
      "step": 11492
    },
    {
      "epoch": 0.9287460352734399,
      "grad_norm": 2.5164315700531006,
      "learning_rate": 5.815923256341704e-06,
      "loss": 0.9558,
      "step": 11493
    },
    {
      "epoch": 0.9288268449867674,
      "grad_norm": 2.596343755722046,
      "learning_rate": 5.815277658579951e-06,
      "loss": 0.9164,
      "step": 11494
    },
    {
      "epoch": 0.928907654700095,
      "grad_norm": 3.173933982849121,
      "learning_rate": 5.8146320468544536e-06,
      "loss": 1.0063,
      "step": 11495
    },
    {
      "epoch": 0.9289884644134225,
      "grad_norm": 2.6160390377044678,
      "learning_rate": 5.81398642117627e-06,
      "loss": 0.8698,
      "step": 11496
    },
    {
      "epoch": 0.92906927412675,
      "grad_norm": 2.4642016887664795,
      "learning_rate": 5.8133407815564595e-06,
      "loss": 0.8666,
      "step": 11497
    },
    {
      "epoch": 0.9291500838400776,
      "grad_norm": 2.741631507873535,
      "learning_rate": 5.812695128006079e-06,
      "loss": 0.9123,
      "step": 11498
    },
    {
      "epoch": 0.9292308935534052,
      "grad_norm": 2.4259939193725586,
      "learning_rate": 5.812049460536187e-06,
      "loss": 1.0552,
      "step": 11499
    },
    {
      "epoch": 0.9293117032667326,
      "grad_norm": 2.5590362548828125,
      "learning_rate": 5.811403779157844e-06,
      "loss": 0.973,
      "step": 11500
    },
    {
      "epoch": 0.9293925129800602,
      "grad_norm": 2.7256057262420654,
      "learning_rate": 5.810758083882107e-06,
      "loss": 1.0253,
      "step": 11501
    },
    {
      "epoch": 0.9294733226933878,
      "grad_norm": 2.9486336708068848,
      "learning_rate": 5.810112374720034e-06,
      "loss": 0.9112,
      "step": 11502
    },
    {
      "epoch": 0.9295541324067152,
      "grad_norm": 2.5152158737182617,
      "learning_rate": 5.809466651682688e-06,
      "loss": 0.9632,
      "step": 11503
    },
    {
      "epoch": 0.9296349421200428,
      "grad_norm": 2.4745943546295166,
      "learning_rate": 5.808820914781127e-06,
      "loss": 0.8528,
      "step": 11504
    },
    {
      "epoch": 0.9297157518333704,
      "grad_norm": 2.594898223876953,
      "learning_rate": 5.80817516402641e-06,
      "loss": 0.9564,
      "step": 11505
    },
    {
      "epoch": 0.9297965615466979,
      "grad_norm": 2.7474186420440674,
      "learning_rate": 5.807529399429599e-06,
      "loss": 0.9599,
      "step": 11506
    },
    {
      "epoch": 0.9298773712600255,
      "grad_norm": 2.5169644355773926,
      "learning_rate": 5.806883621001754e-06,
      "loss": 0.9271,
      "step": 11507
    },
    {
      "epoch": 0.929958180973353,
      "grad_norm": 2.7150564193725586,
      "learning_rate": 5.806237828753935e-06,
      "loss": 0.8374,
      "step": 11508
    },
    {
      "epoch": 0.9300389906866805,
      "grad_norm": 2.2712574005126953,
      "learning_rate": 5.8055920226972005e-06,
      "loss": 0.9027,
      "step": 11509
    },
    {
      "epoch": 0.9301198004000081,
      "grad_norm": 2.483797550201416,
      "learning_rate": 5.804946202842616e-06,
      "loss": 0.8502,
      "step": 11510
    },
    {
      "epoch": 0.9302006101133357,
      "grad_norm": 2.591059446334839,
      "learning_rate": 5.80430036920124e-06,
      "loss": 0.9023,
      "step": 11511
    },
    {
      "epoch": 0.9302814198266631,
      "grad_norm": 2.3627212047576904,
      "learning_rate": 5.803654521784135e-06,
      "loss": 0.9751,
      "step": 11512
    },
    {
      "epoch": 0.9303622295399907,
      "grad_norm": 2.7235360145568848,
      "learning_rate": 5.803008660602364e-06,
      "loss": 0.9483,
      "step": 11513
    },
    {
      "epoch": 0.9304430392533183,
      "grad_norm": 2.6170332431793213,
      "learning_rate": 5.802362785666987e-06,
      "loss": 0.9834,
      "step": 11514
    },
    {
      "epoch": 0.9305238489666458,
      "grad_norm": 2.68587327003479,
      "learning_rate": 5.801716896989068e-06,
      "loss": 0.9299,
      "step": 11515
    },
    {
      "epoch": 0.9306046586799733,
      "grad_norm": 2.193279266357422,
      "learning_rate": 5.801070994579668e-06,
      "loss": 0.8976,
      "step": 11516
    },
    {
      "epoch": 0.9306854683933009,
      "grad_norm": 2.9214203357696533,
      "learning_rate": 5.800425078449849e-06,
      "loss": 0.8915,
      "step": 11517
    },
    {
      "epoch": 0.9307662781066284,
      "grad_norm": 2.671398639678955,
      "learning_rate": 5.799779148610677e-06,
      "loss": 0.8175,
      "step": 11518
    },
    {
      "epoch": 0.930847087819956,
      "grad_norm": 2.8005127906799316,
      "learning_rate": 5.799133205073213e-06,
      "loss": 0.8693,
      "step": 11519
    },
    {
      "epoch": 0.9309278975332835,
      "grad_norm": 3.3705873489379883,
      "learning_rate": 5.798487247848521e-06,
      "loss": 0.8651,
      "step": 11520
    },
    {
      "epoch": 0.931008707246611,
      "grad_norm": 2.396902322769165,
      "learning_rate": 5.7978412769476656e-06,
      "loss": 0.8154,
      "step": 11521
    },
    {
      "epoch": 0.9310895169599386,
      "grad_norm": 2.690016984939575,
      "learning_rate": 5.797195292381707e-06,
      "loss": 0.9181,
      "step": 11522
    },
    {
      "epoch": 0.9311703266732662,
      "grad_norm": 2.621349573135376,
      "learning_rate": 5.796549294161716e-06,
      "loss": 1.0243,
      "step": 11523
    },
    {
      "epoch": 0.9312511363865936,
      "grad_norm": 2.9115445613861084,
      "learning_rate": 5.795903282298752e-06,
      "loss": 1.0949,
      "step": 11524
    },
    {
      "epoch": 0.9313319460999212,
      "grad_norm": 2.867117404937744,
      "learning_rate": 5.79525725680388e-06,
      "loss": 0.915,
      "step": 11525
    },
    {
      "epoch": 0.9314127558132488,
      "grad_norm": 2.8154852390289307,
      "learning_rate": 5.794611217688167e-06,
      "loss": 0.9369,
      "step": 11526
    },
    {
      "epoch": 0.9314935655265763,
      "grad_norm": 2.3597569465637207,
      "learning_rate": 5.793965164962675e-06,
      "loss": 1.0562,
      "step": 11527
    },
    {
      "epoch": 0.9315743752399038,
      "grad_norm": 3.2262609004974365,
      "learning_rate": 5.793319098638471e-06,
      "loss": 0.8527,
      "step": 11528
    },
    {
      "epoch": 0.9316551849532314,
      "grad_norm": 2.5392205715179443,
      "learning_rate": 5.792673018726624e-06,
      "loss": 0.8943,
      "step": 11529
    },
    {
      "epoch": 0.9317359946665589,
      "grad_norm": 2.468940496444702,
      "learning_rate": 5.792026925238192e-06,
      "loss": 0.899,
      "step": 11530
    },
    {
      "epoch": 0.9318168043798865,
      "grad_norm": 2.5027191638946533,
      "learning_rate": 5.791380818184248e-06,
      "loss": 1.014,
      "step": 11531
    },
    {
      "epoch": 0.931897614093214,
      "grad_norm": 2.5542562007904053,
      "learning_rate": 5.790734697575855e-06,
      "loss": 0.8202,
      "step": 11532
    },
    {
      "epoch": 0.9319784238065415,
      "grad_norm": 2.9731252193450928,
      "learning_rate": 5.790088563424081e-06,
      "loss": 1.0461,
      "step": 11533
    },
    {
      "epoch": 0.9320592335198691,
      "grad_norm": 2.508049726486206,
      "learning_rate": 5.789442415739991e-06,
      "loss": 0.9014,
      "step": 11534
    },
    {
      "epoch": 0.9321400432331967,
      "grad_norm": 2.503194570541382,
      "learning_rate": 5.7887962545346545e-06,
      "loss": 0.8512,
      "step": 11535
    },
    {
      "epoch": 0.9322208529465241,
      "grad_norm": 2.6812469959259033,
      "learning_rate": 5.788150079819135e-06,
      "loss": 0.894,
      "step": 11536
    },
    {
      "epoch": 0.9323016626598517,
      "grad_norm": 3.24379301071167,
      "learning_rate": 5.7875038916045044e-06,
      "loss": 0.9462,
      "step": 11537
    },
    {
      "epoch": 0.9323824723731793,
      "grad_norm": 2.8694512844085693,
      "learning_rate": 5.7868576899018256e-06,
      "loss": 0.8687,
      "step": 11538
    },
    {
      "epoch": 0.9324632820865068,
      "grad_norm": 2.6820590496063232,
      "learning_rate": 5.786211474722171e-06,
      "loss": 0.8236,
      "step": 11539
    },
    {
      "epoch": 0.9325440917998343,
      "grad_norm": 2.6474108695983887,
      "learning_rate": 5.785565246076605e-06,
      "loss": 0.992,
      "step": 11540
    },
    {
      "epoch": 0.9326249015131619,
      "grad_norm": 2.641235589981079,
      "learning_rate": 5.7849190039761986e-06,
      "loss": 0.9407,
      "step": 11541
    },
    {
      "epoch": 0.9327057112264894,
      "grad_norm": 2.4772539138793945,
      "learning_rate": 5.784272748432019e-06,
      "loss": 0.9484,
      "step": 11542
    },
    {
      "epoch": 0.932786520939817,
      "grad_norm": 2.6564748287200928,
      "learning_rate": 5.7836264794551345e-06,
      "loss": 0.8055,
      "step": 11543
    },
    {
      "epoch": 0.9328673306531445,
      "grad_norm": 2.9654014110565186,
      "learning_rate": 5.782980197056614e-06,
      "loss": 0.8799,
      "step": 11544
    },
    {
      "epoch": 0.932948140366472,
      "grad_norm": 2.4450013637542725,
      "learning_rate": 5.78233390124753e-06,
      "loss": 1.0247,
      "step": 11545
    },
    {
      "epoch": 0.9330289500797996,
      "grad_norm": 2.6594128608703613,
      "learning_rate": 5.781687592038949e-06,
      "loss": 0.9457,
      "step": 11546
    },
    {
      "epoch": 0.9331097597931272,
      "grad_norm": 2.8365838527679443,
      "learning_rate": 5.78104126944194e-06,
      "loss": 0.9219,
      "step": 11547
    },
    {
      "epoch": 0.9331905695064546,
      "grad_norm": 3.188791036605835,
      "learning_rate": 5.780394933467576e-06,
      "loss": 0.9533,
      "step": 11548
    },
    {
      "epoch": 0.9332713792197822,
      "grad_norm": 2.2971277236938477,
      "learning_rate": 5.779748584126926e-06,
      "loss": 0.9989,
      "step": 11549
    },
    {
      "epoch": 0.9333521889331098,
      "grad_norm": 3.2613959312438965,
      "learning_rate": 5.779102221431057e-06,
      "loss": 0.9592,
      "step": 11550
    },
    {
      "epoch": 0.9334329986464373,
      "grad_norm": 2.406313180923462,
      "learning_rate": 5.778455845391047e-06,
      "loss": 0.9709,
      "step": 11551
    },
    {
      "epoch": 0.9335138083597648,
      "grad_norm": 2.714195966720581,
      "learning_rate": 5.777809456017958e-06,
      "loss": 0.945,
      "step": 11552
    },
    {
      "epoch": 0.9335946180730924,
      "grad_norm": 2.3232004642486572,
      "learning_rate": 5.777163053322869e-06,
      "loss": 0.884,
      "step": 11553
    },
    {
      "epoch": 0.9336754277864199,
      "grad_norm": 2.48897123336792,
      "learning_rate": 5.776516637316844e-06,
      "loss": 0.9891,
      "step": 11554
    },
    {
      "epoch": 0.9337562374997475,
      "grad_norm": 2.561631202697754,
      "learning_rate": 5.775870208010962e-06,
      "loss": 1.0345,
      "step": 11555
    },
    {
      "epoch": 0.933837047213075,
      "grad_norm": 3.3465235233306885,
      "learning_rate": 5.775223765416289e-06,
      "loss": 0.9146,
      "step": 11556
    },
    {
      "epoch": 0.9339178569264025,
      "grad_norm": 2.860692024230957,
      "learning_rate": 5.7745773095439005e-06,
      "loss": 0.9705,
      "step": 11557
    },
    {
      "epoch": 0.9339986666397301,
      "grad_norm": 2.452435255050659,
      "learning_rate": 5.7739308404048665e-06,
      "loss": 0.9616,
      "step": 11558
    },
    {
      "epoch": 0.9340794763530577,
      "grad_norm": 2.4608728885650635,
      "learning_rate": 5.7732843580102615e-06,
      "loss": 1.0741,
      "step": 11559
    },
    {
      "epoch": 0.9341602860663851,
      "grad_norm": 2.8395230770111084,
      "learning_rate": 5.772637862371156e-06,
      "loss": 0.9014,
      "step": 11560
    },
    {
      "epoch": 0.9342410957797127,
      "grad_norm": 2.319927930831909,
      "learning_rate": 5.771991353498624e-06,
      "loss": 0.9134,
      "step": 11561
    },
    {
      "epoch": 0.9343219054930403,
      "grad_norm": 2.8315300941467285,
      "learning_rate": 5.771344831403739e-06,
      "loss": 0.8373,
      "step": 11562
    },
    {
      "epoch": 0.9344027152063678,
      "grad_norm": 2.3799524307250977,
      "learning_rate": 5.770698296097573e-06,
      "loss": 0.9199,
      "step": 11563
    },
    {
      "epoch": 0.9344835249196953,
      "grad_norm": 2.1802234649658203,
      "learning_rate": 5.770051747591202e-06,
      "loss": 0.9585,
      "step": 11564
    },
    {
      "epoch": 0.9345643346330229,
      "grad_norm": 2.759897232055664,
      "learning_rate": 5.769405185895699e-06,
      "loss": 0.9132,
      "step": 11565
    },
    {
      "epoch": 0.9346451443463504,
      "grad_norm": 2.5097436904907227,
      "learning_rate": 5.768758611022136e-06,
      "loss": 0.9136,
      "step": 11566
    },
    {
      "epoch": 0.934725954059678,
      "grad_norm": 2.9719035625457764,
      "learning_rate": 5.768112022981589e-06,
      "loss": 0.9139,
      "step": 11567
    },
    {
      "epoch": 0.9348067637730055,
      "grad_norm": 2.778306245803833,
      "learning_rate": 5.767465421785131e-06,
      "loss": 1.0107,
      "step": 11568
    },
    {
      "epoch": 0.9348875734863331,
      "grad_norm": 2.8643646240234375,
      "learning_rate": 5.766818807443839e-06,
      "loss": 1.0238,
      "step": 11569
    },
    {
      "epoch": 0.9349683831996606,
      "grad_norm": 2.584343671798706,
      "learning_rate": 5.766172179968788e-06,
      "loss": 1.028,
      "step": 11570
    },
    {
      "epoch": 0.9350491929129882,
      "grad_norm": 2.721311092376709,
      "learning_rate": 5.765525539371049e-06,
      "loss": 0.8965,
      "step": 11571
    },
    {
      "epoch": 0.9351300026263157,
      "grad_norm": 2.5223119258880615,
      "learning_rate": 5.764878885661703e-06,
      "loss": 1.0345,
      "step": 11572
    },
    {
      "epoch": 0.9352108123396432,
      "grad_norm": 2.4973721504211426,
      "learning_rate": 5.764232218851822e-06,
      "loss": 0.8728,
      "step": 11573
    },
    {
      "epoch": 0.9352916220529708,
      "grad_norm": 2.6296730041503906,
      "learning_rate": 5.763585538952485e-06,
      "loss": 0.8364,
      "step": 11574
    },
    {
      "epoch": 0.9353724317662984,
      "grad_norm": 2.2104923725128174,
      "learning_rate": 5.7629388459747635e-06,
      "loss": 1.0195,
      "step": 11575
    },
    {
      "epoch": 0.9354532414796258,
      "grad_norm": 2.313533306121826,
      "learning_rate": 5.7622921399297375e-06,
      "loss": 0.9308,
      "step": 11576
    },
    {
      "epoch": 0.9355340511929534,
      "grad_norm": 2.5132343769073486,
      "learning_rate": 5.761645420828481e-06,
      "loss": 0.8205,
      "step": 11577
    },
    {
      "epoch": 0.935614860906281,
      "grad_norm": 2.088826894760132,
      "learning_rate": 5.760998688682073e-06,
      "loss": 1.0048,
      "step": 11578
    },
    {
      "epoch": 0.9356956706196085,
      "grad_norm": 2.9883954524993896,
      "learning_rate": 5.76035194350159e-06,
      "loss": 0.9718,
      "step": 11579
    },
    {
      "epoch": 0.935776480332936,
      "grad_norm": 3.0133285522460938,
      "learning_rate": 5.759705185298109e-06,
      "loss": 0.8568,
      "step": 11580
    },
    {
      "epoch": 0.9358572900462636,
      "grad_norm": 2.6837682723999023,
      "learning_rate": 5.759058414082704e-06,
      "loss": 1.0335,
      "step": 11581
    },
    {
      "epoch": 0.9359380997595911,
      "grad_norm": 2.656585693359375,
      "learning_rate": 5.758411629866459e-06,
      "loss": 0.862,
      "step": 11582
    },
    {
      "epoch": 0.9360189094729187,
      "grad_norm": 2.485264778137207,
      "learning_rate": 5.7577648326604465e-06,
      "loss": 0.8882,
      "step": 11583
    },
    {
      "epoch": 0.9360997191862462,
      "grad_norm": 2.6140716075897217,
      "learning_rate": 5.757118022475749e-06,
      "loss": 0.9033,
      "step": 11584
    },
    {
      "epoch": 0.9361805288995737,
      "grad_norm": 2.6041059494018555,
      "learning_rate": 5.756471199323441e-06,
      "loss": 0.8124,
      "step": 11585
    },
    {
      "epoch": 0.9362613386129013,
      "grad_norm": 2.636671304702759,
      "learning_rate": 5.755824363214603e-06,
      "loss": 0.9513,
      "step": 11586
    },
    {
      "epoch": 0.9363421483262289,
      "grad_norm": 2.6526453495025635,
      "learning_rate": 5.755177514160312e-06,
      "loss": 0.9581,
      "step": 11587
    },
    {
      "epoch": 0.9364229580395563,
      "grad_norm": 3.1483845710754395,
      "learning_rate": 5.754530652171651e-06,
      "loss": 0.8236,
      "step": 11588
    },
    {
      "epoch": 0.9365037677528839,
      "grad_norm": 2.5279927253723145,
      "learning_rate": 5.753883777259693e-06,
      "loss": 0.9465,
      "step": 11589
    },
    {
      "epoch": 0.9365845774662115,
      "grad_norm": 2.7167091369628906,
      "learning_rate": 5.753236889435523e-06,
      "loss": 1.0296,
      "step": 11590
    },
    {
      "epoch": 0.936665387179539,
      "grad_norm": 3.3632664680480957,
      "learning_rate": 5.752589988710216e-06,
      "loss": 0.8465,
      "step": 11591
    },
    {
      "epoch": 0.9367461968928665,
      "grad_norm": 2.8106067180633545,
      "learning_rate": 5.751943075094857e-06,
      "loss": 1.0021,
      "step": 11592
    },
    {
      "epoch": 0.9368270066061941,
      "grad_norm": 2.6745994091033936,
      "learning_rate": 5.751296148600521e-06,
      "loss": 0.9987,
      "step": 11593
    },
    {
      "epoch": 0.9369078163195216,
      "grad_norm": 2.5426883697509766,
      "learning_rate": 5.750649209238294e-06,
      "loss": 1.0077,
      "step": 11594
    },
    {
      "epoch": 0.9369886260328492,
      "grad_norm": 2.624891996383667,
      "learning_rate": 5.75000225701925e-06,
      "loss": 1.0869,
      "step": 11595
    },
    {
      "epoch": 0.9370694357461767,
      "grad_norm": 2.8439857959747314,
      "learning_rate": 5.749355291954475e-06,
      "loss": 0.958,
      "step": 11596
    },
    {
      "epoch": 0.9371502454595042,
      "grad_norm": 2.8124709129333496,
      "learning_rate": 5.7487083140550444e-06,
      "loss": 0.8651,
      "step": 11597
    },
    {
      "epoch": 0.9372310551728318,
      "grad_norm": 2.540600538253784,
      "learning_rate": 5.748061323332047e-06,
      "loss": 0.8403,
      "step": 11598
    },
    {
      "epoch": 0.9373118648861594,
      "grad_norm": 2.6409599781036377,
      "learning_rate": 5.7474143197965584e-06,
      "loss": 0.9303,
      "step": 11599
    },
    {
      "epoch": 0.9373926745994868,
      "grad_norm": 2.486414909362793,
      "learning_rate": 5.7467673034596605e-06,
      "loss": 0.8395,
      "step": 11600
    },
    {
      "epoch": 0.9374734843128144,
      "grad_norm": 2.6569576263427734,
      "learning_rate": 5.746120274332439e-06,
      "loss": 0.9399,
      "step": 11601
    },
    {
      "epoch": 0.937554294026142,
      "grad_norm": 2.4254419803619385,
      "learning_rate": 5.745473232425972e-06,
      "loss": 0.8859,
      "step": 11602
    },
    {
      "epoch": 0.9376351037394695,
      "grad_norm": 2.6574559211730957,
      "learning_rate": 5.744826177751341e-06,
      "loss": 0.9295,
      "step": 11603
    },
    {
      "epoch": 0.937715913452797,
      "grad_norm": 2.792635679244995,
      "learning_rate": 5.7441791103196345e-06,
      "loss": 0.8113,
      "step": 11604
    },
    {
      "epoch": 0.9377967231661246,
      "grad_norm": 2.413358449935913,
      "learning_rate": 5.743532030141929e-06,
      "loss": 0.7997,
      "step": 11605
    },
    {
      "epoch": 0.9378775328794521,
      "grad_norm": 2.2785565853118896,
      "learning_rate": 5.742884937229312e-06,
      "loss": 1.0717,
      "step": 11606
    },
    {
      "epoch": 0.9379583425927797,
      "grad_norm": 2.803117275238037,
      "learning_rate": 5.742237831592862e-06,
      "loss": 0.8737,
      "step": 11607
    },
    {
      "epoch": 0.9380391523061072,
      "grad_norm": 2.532715082168579,
      "learning_rate": 5.741590713243666e-06,
      "loss": 0.9361,
      "step": 11608
    },
    {
      "epoch": 0.9381199620194347,
      "grad_norm": 2.657824754714966,
      "learning_rate": 5.740943582192806e-06,
      "loss": 0.9157,
      "step": 11609
    },
    {
      "epoch": 0.9382007717327623,
      "grad_norm": 2.352769613265991,
      "learning_rate": 5.740296438451367e-06,
      "loss": 0.9595,
      "step": 11610
    },
    {
      "epoch": 0.9382815814460899,
      "grad_norm": 2.897944688796997,
      "learning_rate": 5.73964928203043e-06,
      "loss": 0.9243,
      "step": 11611
    },
    {
      "epoch": 0.9383623911594173,
      "grad_norm": 2.463521957397461,
      "learning_rate": 5.739002112941085e-06,
      "loss": 0.8229,
      "step": 11612
    },
    {
      "epoch": 0.9384432008727449,
      "grad_norm": 2.732213020324707,
      "learning_rate": 5.738354931194411e-06,
      "loss": 0.8576,
      "step": 11613
    },
    {
      "epoch": 0.9385240105860725,
      "grad_norm": 2.7223618030548096,
      "learning_rate": 5.737707736801494e-06,
      "loss": 0.8618,
      "step": 11614
    },
    {
      "epoch": 0.9386048202994,
      "grad_norm": 2.5126633644104004,
      "learning_rate": 5.73706052977342e-06,
      "loss": 1.0201,
      "step": 11615
    },
    {
      "epoch": 0.9386856300127275,
      "grad_norm": 2.850412607192993,
      "learning_rate": 5.736413310121274e-06,
      "loss": 0.9298,
      "step": 11616
    },
    {
      "epoch": 0.9387664397260551,
      "grad_norm": 2.631685495376587,
      "learning_rate": 5.73576607785614e-06,
      "loss": 0.963,
      "step": 11617
    },
    {
      "epoch": 0.9388472494393826,
      "grad_norm": 2.328568458557129,
      "learning_rate": 5.735118832989105e-06,
      "loss": 0.9466,
      "step": 11618
    },
    {
      "epoch": 0.9389280591527102,
      "grad_norm": 2.7688541412353516,
      "learning_rate": 5.734471575531253e-06,
      "loss": 0.8964,
      "step": 11619
    },
    {
      "epoch": 0.9390088688660377,
      "grad_norm": 2.0908613204956055,
      "learning_rate": 5.733824305493672e-06,
      "loss": 0.962,
      "step": 11620
    },
    {
      "epoch": 0.9390896785793652,
      "grad_norm": 2.739166259765625,
      "learning_rate": 5.733177022887447e-06,
      "loss": 0.9175,
      "step": 11621
    },
    {
      "epoch": 0.9391704882926928,
      "grad_norm": 2.4673783779144287,
      "learning_rate": 5.732529727723665e-06,
      "loss": 0.9727,
      "step": 11622
    },
    {
      "epoch": 0.9392512980060204,
      "grad_norm": 2.764119863510132,
      "learning_rate": 5.731882420013411e-06,
      "loss": 0.8901,
      "step": 11623
    },
    {
      "epoch": 0.9393321077193478,
      "grad_norm": 2.6008598804473877,
      "learning_rate": 5.731235099767776e-06,
      "loss": 0.8897,
      "step": 11624
    },
    {
      "epoch": 0.9394129174326754,
      "grad_norm": 2.988536834716797,
      "learning_rate": 5.73058776699784e-06,
      "loss": 0.7958,
      "step": 11625
    },
    {
      "epoch": 0.939493727146003,
      "grad_norm": 2.476825714111328,
      "learning_rate": 5.729940421714698e-06,
      "loss": 0.9187,
      "step": 11626
    },
    {
      "epoch": 0.9395745368593305,
      "grad_norm": 3.226663112640381,
      "learning_rate": 5.729293063929432e-06,
      "loss": 1.0076,
      "step": 11627
    },
    {
      "epoch": 0.939655346572658,
      "grad_norm": 2.6984446048736572,
      "learning_rate": 5.728645693653132e-06,
      "loss": 0.9631,
      "step": 11628
    },
    {
      "epoch": 0.9397361562859856,
      "grad_norm": 2.6082351207733154,
      "learning_rate": 5.727998310896885e-06,
      "loss": 0.9045,
      "step": 11629
    },
    {
      "epoch": 0.9398169659993131,
      "grad_norm": 2.7023630142211914,
      "learning_rate": 5.72735091567178e-06,
      "loss": 0.8335,
      "step": 11630
    },
    {
      "epoch": 0.9398977757126407,
      "grad_norm": 2.7454357147216797,
      "learning_rate": 5.726703507988904e-06,
      "loss": 0.8524,
      "step": 11631
    },
    {
      "epoch": 0.9399785854259682,
      "grad_norm": 2.3165290355682373,
      "learning_rate": 5.7260560878593486e-06,
      "loss": 1.0407,
      "step": 11632
    },
    {
      "epoch": 0.9400593951392957,
      "grad_norm": 3.1288325786590576,
      "learning_rate": 5.725408655294199e-06,
      "loss": 0.8654,
      "step": 11633
    },
    {
      "epoch": 0.9401402048526233,
      "grad_norm": 2.6203765869140625,
      "learning_rate": 5.724761210304544e-06,
      "loss": 0.9561,
      "step": 11634
    },
    {
      "epoch": 0.9402210145659509,
      "grad_norm": 2.4022812843322754,
      "learning_rate": 5.724113752901476e-06,
      "loss": 0.8731,
      "step": 11635
    },
    {
      "epoch": 0.9403018242792783,
      "grad_norm": 2.7591936588287354,
      "learning_rate": 5.723466283096082e-06,
      "loss": 0.9516,
      "step": 11636
    },
    {
      "epoch": 0.9403826339926059,
      "grad_norm": 2.518235445022583,
      "learning_rate": 5.7228188008994525e-06,
      "loss": 0.9368,
      "step": 11637
    },
    {
      "epoch": 0.9404634437059335,
      "grad_norm": 2.76204252243042,
      "learning_rate": 5.722171306322677e-06,
      "loss": 1.0127,
      "step": 11638
    },
    {
      "epoch": 0.940544253419261,
      "grad_norm": 2.6521084308624268,
      "learning_rate": 5.721523799376845e-06,
      "loss": 0.8873,
      "step": 11639
    },
    {
      "epoch": 0.9406250631325885,
      "grad_norm": 2.2734222412109375,
      "learning_rate": 5.720876280073047e-06,
      "loss": 1.0376,
      "step": 11640
    },
    {
      "epoch": 0.9407058728459161,
      "grad_norm": 3.156776189804077,
      "learning_rate": 5.720228748422376e-06,
      "loss": 0.912,
      "step": 11641
    },
    {
      "epoch": 0.9407866825592436,
      "grad_norm": 2.487765312194824,
      "learning_rate": 5.719581204435919e-06,
      "loss": 0.9032,
      "step": 11642
    },
    {
      "epoch": 0.9408674922725712,
      "grad_norm": 2.592525005340576,
      "learning_rate": 5.7189336481247685e-06,
      "loss": 0.9763,
      "step": 11643
    },
    {
      "epoch": 0.9409483019858987,
      "grad_norm": 2.4446444511413574,
      "learning_rate": 5.718286079500015e-06,
      "loss": 1.0842,
      "step": 11644
    },
    {
      "epoch": 0.9410291116992262,
      "grad_norm": 2.3044979572296143,
      "learning_rate": 5.717638498572751e-06,
      "loss": 0.9632,
      "step": 11645
    },
    {
      "epoch": 0.9411099214125538,
      "grad_norm": 2.7817041873931885,
      "learning_rate": 5.7169909053540666e-06,
      "loss": 0.867,
      "step": 11646
    },
    {
      "epoch": 0.9411907311258814,
      "grad_norm": 2.5680007934570312,
      "learning_rate": 5.716343299855054e-06,
      "loss": 0.8192,
      "step": 11647
    },
    {
      "epoch": 0.9412715408392088,
      "grad_norm": 2.3593533039093018,
      "learning_rate": 5.7156956820868035e-06,
      "loss": 1.0092,
      "step": 11648
    },
    {
      "epoch": 0.9413523505525364,
      "grad_norm": 2.8043808937072754,
      "learning_rate": 5.715048052060413e-06,
      "loss": 0.9201,
      "step": 11649
    },
    {
      "epoch": 0.941433160265864,
      "grad_norm": 2.6572163105010986,
      "learning_rate": 5.7144004097869664e-06,
      "loss": 0.8527,
      "step": 11650
    },
    {
      "epoch": 0.9415139699791915,
      "grad_norm": 2.7471683025360107,
      "learning_rate": 5.713752755277564e-06,
      "loss": 0.9413,
      "step": 11651
    },
    {
      "epoch": 0.941594779692519,
      "grad_norm": 2.9516263008117676,
      "learning_rate": 5.713105088543294e-06,
      "loss": 0.9228,
      "step": 11652
    },
    {
      "epoch": 0.9416755894058466,
      "grad_norm": 2.697279214859009,
      "learning_rate": 5.712457409595249e-06,
      "loss": 0.9134,
      "step": 11653
    },
    {
      "epoch": 0.9417563991191741,
      "grad_norm": 2.719963788986206,
      "learning_rate": 5.711809718444525e-06,
      "loss": 0.9863,
      "step": 11654
    },
    {
      "epoch": 0.9418372088325017,
      "grad_norm": 2.6032958030700684,
      "learning_rate": 5.711162015102216e-06,
      "loss": 0.8067,
      "step": 11655
    },
    {
      "epoch": 0.9419180185458292,
      "grad_norm": 2.5723934173583984,
      "learning_rate": 5.7105142995794104e-06,
      "loss": 0.9703,
      "step": 11656
    },
    {
      "epoch": 0.9419988282591567,
      "grad_norm": 2.735833168029785,
      "learning_rate": 5.709866571887208e-06,
      "loss": 0.9125,
      "step": 11657
    },
    {
      "epoch": 0.9420796379724843,
      "grad_norm": 2.497631549835205,
      "learning_rate": 5.7092188320366994e-06,
      "loss": 0.9172,
      "step": 11658
    },
    {
      "epoch": 0.9421604476858119,
      "grad_norm": 2.514263153076172,
      "learning_rate": 5.70857108003898e-06,
      "loss": 1.0135,
      "step": 11659
    },
    {
      "epoch": 0.9422412573991393,
      "grad_norm": 2.6322433948516846,
      "learning_rate": 5.707923315905142e-06,
      "loss": 0.8427,
      "step": 11660
    },
    {
      "epoch": 0.9423220671124669,
      "grad_norm": 3.2455856800079346,
      "learning_rate": 5.707275539646284e-06,
      "loss": 0.8514,
      "step": 11661
    },
    {
      "epoch": 0.9424028768257945,
      "grad_norm": 2.391359806060791,
      "learning_rate": 5.706627751273496e-06,
      "loss": 0.973,
      "step": 11662
    },
    {
      "epoch": 0.942483686539122,
      "grad_norm": 2.312131643295288,
      "learning_rate": 5.705979950797878e-06,
      "loss": 0.8773,
      "step": 11663
    },
    {
      "epoch": 0.9425644962524495,
      "grad_norm": 2.034846782684326,
      "learning_rate": 5.7053321382305214e-06,
      "loss": 0.942,
      "step": 11664
    },
    {
      "epoch": 0.9426453059657771,
      "grad_norm": 2.5560832023620605,
      "learning_rate": 5.704684313582526e-06,
      "loss": 0.927,
      "step": 11665
    },
    {
      "epoch": 0.9427261156791046,
      "grad_norm": 2.4851796627044678,
      "learning_rate": 5.704036476864982e-06,
      "loss": 1.0638,
      "step": 11666
    },
    {
      "epoch": 0.9428069253924322,
      "grad_norm": 3.1105778217315674,
      "learning_rate": 5.7033886280889894e-06,
      "loss": 0.8441,
      "step": 11667
    },
    {
      "epoch": 0.9428877351057597,
      "grad_norm": 2.5052027702331543,
      "learning_rate": 5.702740767265643e-06,
      "loss": 0.9546,
      "step": 11668
    },
    {
      "epoch": 0.9429685448190872,
      "grad_norm": 2.8185160160064697,
      "learning_rate": 5.7020928944060395e-06,
      "loss": 1.0313,
      "step": 11669
    },
    {
      "epoch": 0.9430493545324148,
      "grad_norm": 2.9099888801574707,
      "learning_rate": 5.701445009521273e-06,
      "loss": 1.024,
      "step": 11670
    },
    {
      "epoch": 0.9431301642457424,
      "grad_norm": 2.643374443054199,
      "learning_rate": 5.700797112622445e-06,
      "loss": 1.0331,
      "step": 11671
    },
    {
      "epoch": 0.9432109739590698,
      "grad_norm": 2.4487547874450684,
      "learning_rate": 5.700149203720648e-06,
      "loss": 0.9712,
      "step": 11672
    },
    {
      "epoch": 0.9432917836723974,
      "grad_norm": 2.403853416442871,
      "learning_rate": 5.699501282826979e-06,
      "loss": 1.0223,
      "step": 11673
    },
    {
      "epoch": 0.943372593385725,
      "grad_norm": 2.1139395236968994,
      "learning_rate": 5.698853349952539e-06,
      "loss": 1.0247,
      "step": 11674
    },
    {
      "epoch": 0.9434534030990525,
      "grad_norm": 3.068643808364868,
      "learning_rate": 5.6982054051084235e-06,
      "loss": 0.8979,
      "step": 11675
    },
    {
      "epoch": 0.94353421281238,
      "grad_norm": 2.852083206176758,
      "learning_rate": 5.697557448305729e-06,
      "loss": 0.9458,
      "step": 11676
    },
    {
      "epoch": 0.9436150225257076,
      "grad_norm": 2.364699125289917,
      "learning_rate": 5.696909479555557e-06,
      "loss": 0.8893,
      "step": 11677
    },
    {
      "epoch": 0.9436958322390351,
      "grad_norm": 2.350175142288208,
      "learning_rate": 5.696261498869e-06,
      "loss": 1.0053,
      "step": 11678
    },
    {
      "epoch": 0.9437766419523627,
      "grad_norm": 2.7102997303009033,
      "learning_rate": 5.695613506257162e-06,
      "loss": 0.8596,
      "step": 11679
    },
    {
      "epoch": 0.9438574516656902,
      "grad_norm": 2.601069927215576,
      "learning_rate": 5.69496550173114e-06,
      "loss": 0.8036,
      "step": 11680
    },
    {
      "epoch": 0.9439382613790177,
      "grad_norm": 2.622908115386963,
      "learning_rate": 5.69431748530203e-06,
      "loss": 0.9233,
      "step": 11681
    },
    {
      "epoch": 0.9440190710923453,
      "grad_norm": 2.724900960922241,
      "learning_rate": 5.693669456980935e-06,
      "loss": 1.043,
      "step": 11682
    },
    {
      "epoch": 0.9440998808056729,
      "grad_norm": 2.661947727203369,
      "learning_rate": 5.693021416778951e-06,
      "loss": 1.0079,
      "step": 11683
    },
    {
      "epoch": 0.9441806905190003,
      "grad_norm": 2.5352721214294434,
      "learning_rate": 5.692373364707178e-06,
      "loss": 0.8936,
      "step": 11684
    },
    {
      "epoch": 0.9442615002323279,
      "grad_norm": 2.724100351333618,
      "learning_rate": 5.691725300776717e-06,
      "loss": 0.9612,
      "step": 11685
    },
    {
      "epoch": 0.9443423099456555,
      "grad_norm": 3.0862293243408203,
      "learning_rate": 5.691077224998667e-06,
      "loss": 0.9231,
      "step": 11686
    },
    {
      "epoch": 0.944423119658983,
      "grad_norm": 3.1719329357147217,
      "learning_rate": 5.690429137384127e-06,
      "loss": 0.964,
      "step": 11687
    },
    {
      "epoch": 0.9445039293723105,
      "grad_norm": 2.7404732704162598,
      "learning_rate": 5.689781037944198e-06,
      "loss": 0.9465,
      "step": 11688
    },
    {
      "epoch": 0.9445847390856381,
      "grad_norm": 3.070930004119873,
      "learning_rate": 5.689132926689982e-06,
      "loss": 0.8778,
      "step": 11689
    },
    {
      "epoch": 0.9446655487989656,
      "grad_norm": 2.8368983268737793,
      "learning_rate": 5.688484803632574e-06,
      "loss": 1.0098,
      "step": 11690
    },
    {
      "epoch": 0.9447463585122932,
      "grad_norm": 2.6157639026641846,
      "learning_rate": 5.687836668783083e-06,
      "loss": 0.9648,
      "step": 11691
    },
    {
      "epoch": 0.9448271682256207,
      "grad_norm": 2.602119207382202,
      "learning_rate": 5.687188522152603e-06,
      "loss": 0.8824,
      "step": 11692
    },
    {
      "epoch": 0.9449079779389482,
      "grad_norm": 2.6795992851257324,
      "learning_rate": 5.686540363752241e-06,
      "loss": 0.9899,
      "step": 11693
    },
    {
      "epoch": 0.9449887876522758,
      "grad_norm": 2.348895788192749,
      "learning_rate": 5.685892193593093e-06,
      "loss": 0.8316,
      "step": 11694
    },
    {
      "epoch": 0.9450695973656034,
      "grad_norm": 3.1809682846069336,
      "learning_rate": 5.685244011686264e-06,
      "loss": 0.9253,
      "step": 11695
    },
    {
      "epoch": 0.9451504070789308,
      "grad_norm": 2.47281813621521,
      "learning_rate": 5.684595818042854e-06,
      "loss": 1.0117,
      "step": 11696
    },
    {
      "epoch": 0.9452312167922584,
      "grad_norm": 2.3931689262390137,
      "learning_rate": 5.683947612673966e-06,
      "loss": 1.0186,
      "step": 11697
    },
    {
      "epoch": 0.945312026505586,
      "grad_norm": 2.7105016708374023,
      "learning_rate": 5.683299395590701e-06,
      "loss": 0.8382,
      "step": 11698
    },
    {
      "epoch": 0.9453928362189136,
      "grad_norm": 3.033129930496216,
      "learning_rate": 5.682651166804165e-06,
      "loss": 0.8572,
      "step": 11699
    },
    {
      "epoch": 0.945473645932241,
      "grad_norm": 3.0416135787963867,
      "learning_rate": 5.682002926325456e-06,
      "loss": 0.8278,
      "step": 11700
    },
    {
      "epoch": 0.9455544556455686,
      "grad_norm": 2.209350109100342,
      "learning_rate": 5.681354674165678e-06,
      "loss": 0.862,
      "step": 11701
    },
    {
      "epoch": 0.9456352653588962,
      "grad_norm": 2.855954885482788,
      "learning_rate": 5.680706410335936e-06,
      "loss": 0.9371,
      "step": 11702
    },
    {
      "epoch": 0.9457160750722237,
      "grad_norm": 2.812636375427246,
      "learning_rate": 5.680058134847332e-06,
      "loss": 0.8857,
      "step": 11703
    },
    {
      "epoch": 0.9457968847855512,
      "grad_norm": 3.201815128326416,
      "learning_rate": 5.679409847710968e-06,
      "loss": 0.935,
      "step": 11704
    },
    {
      "epoch": 0.9458776944988788,
      "grad_norm": 2.952446699142456,
      "learning_rate": 5.67876154893795e-06,
      "loss": 0.9729,
      "step": 11705
    },
    {
      "epoch": 0.9459585042122063,
      "grad_norm": 2.510310173034668,
      "learning_rate": 5.67811323853938e-06,
      "loss": 0.8319,
      "step": 11706
    },
    {
      "epoch": 0.9460393139255339,
      "grad_norm": 2.7408740520477295,
      "learning_rate": 5.677464916526363e-06,
      "loss": 0.9462,
      "step": 11707
    },
    {
      "epoch": 0.9461201236388614,
      "grad_norm": 3.0367515087127686,
      "learning_rate": 5.676816582910004e-06,
      "loss": 0.9367,
      "step": 11708
    },
    {
      "epoch": 0.9462009333521889,
      "grad_norm": 2.750922918319702,
      "learning_rate": 5.676168237701405e-06,
      "loss": 0.9806,
      "step": 11709
    },
    {
      "epoch": 0.9462817430655165,
      "grad_norm": 2.858391284942627,
      "learning_rate": 5.675519880911673e-06,
      "loss": 0.8769,
      "step": 11710
    },
    {
      "epoch": 0.9463625527788441,
      "grad_norm": 2.731144666671753,
      "learning_rate": 5.67487151255191e-06,
      "loss": 0.9623,
      "step": 11711
    },
    {
      "epoch": 0.9464433624921715,
      "grad_norm": 2.8071606159210205,
      "learning_rate": 5.674223132633224e-06,
      "loss": 0.9204,
      "step": 11712
    },
    {
      "epoch": 0.9465241722054991,
      "grad_norm": 2.459951400756836,
      "learning_rate": 5.673574741166719e-06,
      "loss": 0.9877,
      "step": 11713
    },
    {
      "epoch": 0.9466049819188267,
      "grad_norm": 2.5169243812561035,
      "learning_rate": 5.6729263381635e-06,
      "loss": 0.9025,
      "step": 11714
    },
    {
      "epoch": 0.9466857916321542,
      "grad_norm": 2.9062554836273193,
      "learning_rate": 5.672277923634671e-06,
      "loss": 0.8722,
      "step": 11715
    },
    {
      "epoch": 0.9467666013454817,
      "grad_norm": 3.1349399089813232,
      "learning_rate": 5.671629497591343e-06,
      "loss": 0.9472,
      "step": 11716
    },
    {
      "epoch": 0.9468474110588093,
      "grad_norm": 2.732954740524292,
      "learning_rate": 5.6709810600446165e-06,
      "loss": 0.889,
      "step": 11717
    },
    {
      "epoch": 0.9469282207721368,
      "grad_norm": 2.821624279022217,
      "learning_rate": 5.6703326110056e-06,
      "loss": 0.8655,
      "step": 11718
    },
    {
      "epoch": 0.9470090304854644,
      "grad_norm": 2.6683051586151123,
      "learning_rate": 5.6696841504853994e-06,
      "loss": 0.97,
      "step": 11719
    },
    {
      "epoch": 0.9470898401987919,
      "grad_norm": 2.8609542846679688,
      "learning_rate": 5.6690356784951216e-06,
      "loss": 0.8594,
      "step": 11720
    },
    {
      "epoch": 0.9471706499121194,
      "grad_norm": 2.2158589363098145,
      "learning_rate": 5.668387195045874e-06,
      "loss": 0.9758,
      "step": 11721
    },
    {
      "epoch": 0.947251459625447,
      "grad_norm": 2.7605020999908447,
      "learning_rate": 5.667738700148763e-06,
      "loss": 1.0052,
      "step": 11722
    },
    {
      "epoch": 0.9473322693387746,
      "grad_norm": 2.9931535720825195,
      "learning_rate": 5.667090193814894e-06,
      "loss": 0.9068,
      "step": 11723
    },
    {
      "epoch": 0.947413079052102,
      "grad_norm": 2.7512810230255127,
      "learning_rate": 5.666441676055378e-06,
      "loss": 0.9482,
      "step": 11724
    },
    {
      "epoch": 0.9474938887654296,
      "grad_norm": 2.234410047531128,
      "learning_rate": 5.665793146881319e-06,
      "loss": 1.0454,
      "step": 11725
    },
    {
      "epoch": 0.9475746984787572,
      "grad_norm": 2.208270788192749,
      "learning_rate": 5.665144606303826e-06,
      "loss": 0.7288,
      "step": 11726
    },
    {
      "epoch": 0.9476555081920847,
      "grad_norm": 2.417513370513916,
      "learning_rate": 5.664496054334008e-06,
      "loss": 0.9563,
      "step": 11727
    },
    {
      "epoch": 0.9477363179054122,
      "grad_norm": 2.571308135986328,
      "learning_rate": 5.663847490982973e-06,
      "loss": 0.9953,
      "step": 11728
    },
    {
      "epoch": 0.9478171276187398,
      "grad_norm": 2.855367660522461,
      "learning_rate": 5.6631989162618265e-06,
      "loss": 0.9072,
      "step": 11729
    },
    {
      "epoch": 0.9478979373320673,
      "grad_norm": 2.699425458908081,
      "learning_rate": 5.662550330181681e-06,
      "loss": 0.9318,
      "step": 11730
    },
    {
      "epoch": 0.9479787470453949,
      "grad_norm": 2.2659034729003906,
      "learning_rate": 5.6619017327536415e-06,
      "loss": 0.9078,
      "step": 11731
    },
    {
      "epoch": 0.9480595567587224,
      "grad_norm": 2.4166064262390137,
      "learning_rate": 5.661253123988821e-06,
      "loss": 0.8629,
      "step": 11732
    },
    {
      "epoch": 0.9481403664720499,
      "grad_norm": 2.674800157546997,
      "learning_rate": 5.660604503898325e-06,
      "loss": 0.9452,
      "step": 11733
    },
    {
      "epoch": 0.9482211761853775,
      "grad_norm": 2.9452109336853027,
      "learning_rate": 5.659955872493265e-06,
      "loss": 0.8996,
      "step": 11734
    },
    {
      "epoch": 0.9483019858987051,
      "grad_norm": 2.279548406600952,
      "learning_rate": 5.659307229784748e-06,
      "loss": 0.8853,
      "step": 11735
    },
    {
      "epoch": 0.9483827956120325,
      "grad_norm": 2.522388219833374,
      "learning_rate": 5.658658575783888e-06,
      "loss": 0.9592,
      "step": 11736
    },
    {
      "epoch": 0.9484636053253601,
      "grad_norm": 2.332998514175415,
      "learning_rate": 5.6580099105017895e-06,
      "loss": 0.9054,
      "step": 11737
    },
    {
      "epoch": 0.9485444150386877,
      "grad_norm": 2.6834816932678223,
      "learning_rate": 5.657361233949568e-06,
      "loss": 0.9384,
      "step": 11738
    },
    {
      "epoch": 0.9486252247520152,
      "grad_norm": 2.5026962757110596,
      "learning_rate": 5.65671254613833e-06,
      "loss": 1.0743,
      "step": 11739
    },
    {
      "epoch": 0.9487060344653427,
      "grad_norm": 2.5357184410095215,
      "learning_rate": 5.656063847079186e-06,
      "loss": 0.9328,
      "step": 11740
    },
    {
      "epoch": 0.9487868441786703,
      "grad_norm": 2.4060933589935303,
      "learning_rate": 5.655415136783249e-06,
      "loss": 0.9872,
      "step": 11741
    },
    {
      "epoch": 0.9488676538919978,
      "grad_norm": 2.616105794906616,
      "learning_rate": 5.6547664152616284e-06,
      "loss": 1.0001,
      "step": 11742
    },
    {
      "epoch": 0.9489484636053254,
      "grad_norm": 2.723905563354492,
      "learning_rate": 5.654117682525434e-06,
      "loss": 0.8787,
      "step": 11743
    },
    {
      "epoch": 0.9490292733186529,
      "grad_norm": 3.021120548248291,
      "learning_rate": 5.65346893858578e-06,
      "loss": 0.9593,
      "step": 11744
    },
    {
      "epoch": 0.9491100830319804,
      "grad_norm": 2.3811423778533936,
      "learning_rate": 5.6528201834537746e-06,
      "loss": 0.9215,
      "step": 11745
    },
    {
      "epoch": 0.949190892745308,
      "grad_norm": 2.196046829223633,
      "learning_rate": 5.652171417140533e-06,
      "loss": 0.888,
      "step": 11746
    },
    {
      "epoch": 0.9492717024586356,
      "grad_norm": 2.4733779430389404,
      "learning_rate": 5.651522639657164e-06,
      "loss": 0.8295,
      "step": 11747
    },
    {
      "epoch": 0.949352512171963,
      "grad_norm": 2.6456706523895264,
      "learning_rate": 5.650873851014781e-06,
      "loss": 0.9543,
      "step": 11748
    },
    {
      "epoch": 0.9494333218852906,
      "grad_norm": 2.7628049850463867,
      "learning_rate": 5.650225051224496e-06,
      "loss": 0.8961,
      "step": 11749
    },
    {
      "epoch": 0.9495141315986182,
      "grad_norm": 2.874279260635376,
      "learning_rate": 5.6495762402974215e-06,
      "loss": 0.9824,
      "step": 11750
    },
    {
      "epoch": 0.9495949413119457,
      "grad_norm": 2.812365770339966,
      "learning_rate": 5.648927418244668e-06,
      "loss": 0.9168,
      "step": 11751
    },
    {
      "epoch": 0.9496757510252732,
      "grad_norm": 2.616420030593872,
      "learning_rate": 5.648278585077352e-06,
      "loss": 0.9293,
      "step": 11752
    },
    {
      "epoch": 0.9497565607386008,
      "grad_norm": 3.3047990798950195,
      "learning_rate": 5.6476297408065836e-06,
      "loss": 0.8761,
      "step": 11753
    },
    {
      "epoch": 0.9498373704519283,
      "grad_norm": 2.7111291885375977,
      "learning_rate": 5.646980885443478e-06,
      "loss": 0.8583,
      "step": 11754
    },
    {
      "epoch": 0.9499181801652559,
      "grad_norm": 2.859605312347412,
      "learning_rate": 5.646332018999145e-06,
      "loss": 0.8666,
      "step": 11755
    },
    {
      "epoch": 0.9499989898785834,
      "grad_norm": 2.3949639797210693,
      "learning_rate": 5.645683141484703e-06,
      "loss": 0.844,
      "step": 11756
    },
    {
      "epoch": 0.9500797995919109,
      "grad_norm": 2.8391880989074707,
      "learning_rate": 5.645034252911262e-06,
      "loss": 0.9814,
      "step": 11757
    },
    {
      "epoch": 0.9501606093052385,
      "grad_norm": 2.696854829788208,
      "learning_rate": 5.644385353289939e-06,
      "loss": 1.0628,
      "step": 11758
    },
    {
      "epoch": 0.9502414190185661,
      "grad_norm": 2.8728668689727783,
      "learning_rate": 5.643736442631842e-06,
      "loss": 0.8773,
      "step": 11759
    },
    {
      "epoch": 0.9503222287318935,
      "grad_norm": 2.1819210052490234,
      "learning_rate": 5.643087520948093e-06,
      "loss": 0.941,
      "step": 11760
    },
    {
      "epoch": 0.9504030384452211,
      "grad_norm": 3.0423099994659424,
      "learning_rate": 5.642438588249802e-06,
      "loss": 0.9751,
      "step": 11761
    },
    {
      "epoch": 0.9504838481585487,
      "grad_norm": 2.6552722454071045,
      "learning_rate": 5.6417896445480846e-06,
      "loss": 0.8125,
      "step": 11762
    },
    {
      "epoch": 0.9505646578718762,
      "grad_norm": 2.639998435974121,
      "learning_rate": 5.6411406898540555e-06,
      "loss": 0.8741,
      "step": 11763
    },
    {
      "epoch": 0.9506454675852037,
      "grad_norm": 2.564969062805176,
      "learning_rate": 5.6404917241788295e-06,
      "loss": 0.9727,
      "step": 11764
    },
    {
      "epoch": 0.9507262772985313,
      "grad_norm": 2.4525957107543945,
      "learning_rate": 5.6398427475335214e-06,
      "loss": 0.9602,
      "step": 11765
    },
    {
      "epoch": 0.9508070870118588,
      "grad_norm": 2.4959049224853516,
      "learning_rate": 5.63919375992925e-06,
      "loss": 0.8862,
      "step": 11766
    },
    {
      "epoch": 0.9508878967251864,
      "grad_norm": 2.6759307384490967,
      "learning_rate": 5.638544761377127e-06,
      "loss": 0.9905,
      "step": 11767
    },
    {
      "epoch": 0.9509687064385139,
      "grad_norm": 2.340460777282715,
      "learning_rate": 5.637895751888269e-06,
      "loss": 0.9592,
      "step": 11768
    },
    {
      "epoch": 0.9510495161518414,
      "grad_norm": 2.380915641784668,
      "learning_rate": 5.637246731473792e-06,
      "loss": 0.9191,
      "step": 11769
    },
    {
      "epoch": 0.951130325865169,
      "grad_norm": 2.7404839992523193,
      "learning_rate": 5.636597700144814e-06,
      "loss": 0.9549,
      "step": 11770
    },
    {
      "epoch": 0.9512111355784966,
      "grad_norm": 2.454563856124878,
      "learning_rate": 5.635948657912449e-06,
      "loss": 0.8845,
      "step": 11771
    },
    {
      "epoch": 0.951291945291824,
      "grad_norm": 2.698922634124756,
      "learning_rate": 5.635299604787815e-06,
      "loss": 0.9333,
      "step": 11772
    },
    {
      "epoch": 0.9513727550051516,
      "grad_norm": 2.2386131286621094,
      "learning_rate": 5.634650540782028e-06,
      "loss": 0.8995,
      "step": 11773
    },
    {
      "epoch": 0.9514535647184792,
      "grad_norm": 2.602815628051758,
      "learning_rate": 5.6340014659062044e-06,
      "loss": 0.9921,
      "step": 11774
    },
    {
      "epoch": 0.9515343744318067,
      "grad_norm": 2.311607837677002,
      "learning_rate": 5.633352380171464e-06,
      "loss": 1.0103,
      "step": 11775
    },
    {
      "epoch": 0.9516151841451342,
      "grad_norm": 2.5385825634002686,
      "learning_rate": 5.6327032835889204e-06,
      "loss": 0.8515,
      "step": 11776
    },
    {
      "epoch": 0.9516959938584618,
      "grad_norm": 3.2339069843292236,
      "learning_rate": 5.6320541761696925e-06,
      "loss": 0.9709,
      "step": 11777
    },
    {
      "epoch": 0.9517768035717893,
      "grad_norm": 2.827669143676758,
      "learning_rate": 5.631405057924899e-06,
      "loss": 0.8341,
      "step": 11778
    },
    {
      "epoch": 0.9518576132851169,
      "grad_norm": 2.6024420261383057,
      "learning_rate": 5.630755928865656e-06,
      "loss": 0.9085,
      "step": 11779
    },
    {
      "epoch": 0.9519384229984444,
      "grad_norm": 2.5606136322021484,
      "learning_rate": 5.630106789003083e-06,
      "loss": 0.8564,
      "step": 11780
    },
    {
      "epoch": 0.9520192327117719,
      "grad_norm": 2.8526065349578857,
      "learning_rate": 5.629457638348299e-06,
      "loss": 0.9309,
      "step": 11781
    },
    {
      "epoch": 0.9521000424250995,
      "grad_norm": 2.884697675704956,
      "learning_rate": 5.628808476912417e-06,
      "loss": 0.9203,
      "step": 11782
    },
    {
      "epoch": 0.9521808521384271,
      "grad_norm": 2.9439496994018555,
      "learning_rate": 5.628159304706564e-06,
      "loss": 0.9687,
      "step": 11783
    },
    {
      "epoch": 0.9522616618517545,
      "grad_norm": 2.4576220512390137,
      "learning_rate": 5.627510121741852e-06,
      "loss": 0.9139,
      "step": 11784
    },
    {
      "epoch": 0.9523424715650821,
      "grad_norm": 3.0397865772247314,
      "learning_rate": 5.626860928029403e-06,
      "loss": 0.8678,
      "step": 11785
    },
    {
      "epoch": 0.9524232812784097,
      "grad_norm": 2.5104920864105225,
      "learning_rate": 5.626211723580335e-06,
      "loss": 0.8331,
      "step": 11786
    },
    {
      "epoch": 0.9525040909917372,
      "grad_norm": 2.589230537414551,
      "learning_rate": 5.6255625084057685e-06,
      "loss": 0.8887,
      "step": 11787
    },
    {
      "epoch": 0.9525849007050647,
      "grad_norm": 2.8371191024780273,
      "learning_rate": 5.624913282516822e-06,
      "loss": 1.0051,
      "step": 11788
    },
    {
      "epoch": 0.9526657104183923,
      "grad_norm": 3.225444793701172,
      "learning_rate": 5.624264045924616e-06,
      "loss": 0.9531,
      "step": 11789
    },
    {
      "epoch": 0.9527465201317198,
      "grad_norm": 2.4302711486816406,
      "learning_rate": 5.623614798640267e-06,
      "loss": 0.9652,
      "step": 11790
    },
    {
      "epoch": 0.9528273298450474,
      "grad_norm": 2.6602742671966553,
      "learning_rate": 5.622965540674901e-06,
      "loss": 0.9664,
      "step": 11791
    },
    {
      "epoch": 0.952908139558375,
      "grad_norm": 2.643425941467285,
      "learning_rate": 5.622316272039633e-06,
      "loss": 0.9502,
      "step": 11792
    },
    {
      "epoch": 0.9529889492717024,
      "grad_norm": 2.4053399562835693,
      "learning_rate": 5.621666992745586e-06,
      "loss": 0.9174,
      "step": 11793
    },
    {
      "epoch": 0.95306975898503,
      "grad_norm": 2.7074623107910156,
      "learning_rate": 5.621017702803879e-06,
      "loss": 0.8457,
      "step": 11794
    },
    {
      "epoch": 0.9531505686983576,
      "grad_norm": 2.365626573562622,
      "learning_rate": 5.620368402225637e-06,
      "loss": 0.9687,
      "step": 11795
    },
    {
      "epoch": 0.953231378411685,
      "grad_norm": 2.4549739360809326,
      "learning_rate": 5.619719091021973e-06,
      "loss": 0.9489,
      "step": 11796
    },
    {
      "epoch": 0.9533121881250126,
      "grad_norm": 3.2549610137939453,
      "learning_rate": 5.619069769204017e-06,
      "loss": 1.0024,
      "step": 11797
    },
    {
      "epoch": 0.9533929978383402,
      "grad_norm": 2.302922248840332,
      "learning_rate": 5.618420436782886e-06,
      "loss": 0.9862,
      "step": 11798
    },
    {
      "epoch": 0.9534738075516677,
      "grad_norm": 2.9284064769744873,
      "learning_rate": 5.6177710937696996e-06,
      "loss": 0.987,
      "step": 11799
    },
    {
      "epoch": 0.9535546172649952,
      "grad_norm": 2.7569353580474854,
      "learning_rate": 5.617121740175582e-06,
      "loss": 0.9758,
      "step": 11800
    },
    {
      "epoch": 0.9536354269783228,
      "grad_norm": 2.929086446762085,
      "learning_rate": 5.616472376011654e-06,
      "loss": 0.9325,
      "step": 11801
    },
    {
      "epoch": 0.9537162366916503,
      "grad_norm": 2.611757278442383,
      "learning_rate": 5.61582300128904e-06,
      "loss": 0.8772,
      "step": 11802
    },
    {
      "epoch": 0.9537970464049779,
      "grad_norm": 2.6273276805877686,
      "learning_rate": 5.615173616018861e-06,
      "loss": 1.0233,
      "step": 11803
    },
    {
      "epoch": 0.9538778561183054,
      "grad_norm": 2.3987762928009033,
      "learning_rate": 5.614524220212236e-06,
      "loss": 0.829,
      "step": 11804
    },
    {
      "epoch": 0.9539586658316329,
      "grad_norm": 2.251645088195801,
      "learning_rate": 5.613874813880293e-06,
      "loss": 0.9795,
      "step": 11805
    },
    {
      "epoch": 0.9540394755449605,
      "grad_norm": 2.7683653831481934,
      "learning_rate": 5.613225397034152e-06,
      "loss": 0.8576,
      "step": 11806
    },
    {
      "epoch": 0.9541202852582881,
      "grad_norm": 2.711724281311035,
      "learning_rate": 5.612575969684936e-06,
      "loss": 0.9559,
      "step": 11807
    },
    {
      "epoch": 0.9542010949716155,
      "grad_norm": 2.6666228771209717,
      "learning_rate": 5.611926531843768e-06,
      "loss": 0.8438,
      "step": 11808
    },
    {
      "epoch": 0.9542819046849431,
      "grad_norm": 3.5984926223754883,
      "learning_rate": 5.611277083521772e-06,
      "loss": 1.0368,
      "step": 11809
    },
    {
      "epoch": 0.9543627143982707,
      "grad_norm": 2.522277593612671,
      "learning_rate": 5.610627624730071e-06,
      "loss": 1.0333,
      "step": 11810
    },
    {
      "epoch": 0.9544435241115982,
      "grad_norm": 2.439326763153076,
      "learning_rate": 5.609978155479789e-06,
      "loss": 0.9332,
      "step": 11811
    },
    {
      "epoch": 0.9545243338249257,
      "grad_norm": 2.7858619689941406,
      "learning_rate": 5.60932867578205e-06,
      "loss": 0.9524,
      "step": 11812
    },
    {
      "epoch": 0.9546051435382533,
      "grad_norm": 2.5886895656585693,
      "learning_rate": 5.608679185647976e-06,
      "loss": 0.9871,
      "step": 11813
    },
    {
      "epoch": 0.9546859532515808,
      "grad_norm": 2.249063014984131,
      "learning_rate": 5.608029685088694e-06,
      "loss": 0.9825,
      "step": 11814
    },
    {
      "epoch": 0.9547667629649084,
      "grad_norm": 2.3512027263641357,
      "learning_rate": 5.607380174115328e-06,
      "loss": 0.8383,
      "step": 11815
    },
    {
      "epoch": 0.954847572678236,
      "grad_norm": 2.7193336486816406,
      "learning_rate": 5.6067306527390005e-06,
      "loss": 1.0348,
      "step": 11816
    },
    {
      "epoch": 0.9549283823915634,
      "grad_norm": 2.9844541549682617,
      "learning_rate": 5.606081120970838e-06,
      "loss": 0.8671,
      "step": 11817
    },
    {
      "epoch": 0.955009192104891,
      "grad_norm": 2.8504726886749268,
      "learning_rate": 5.605431578821965e-06,
      "loss": 0.9323,
      "step": 11818
    },
    {
      "epoch": 0.9550900018182186,
      "grad_norm": 2.616696834564209,
      "learning_rate": 5.604782026303508e-06,
      "loss": 0.9002,
      "step": 11819
    },
    {
      "epoch": 0.955170811531546,
      "grad_norm": 3.192596197128296,
      "learning_rate": 5.6041324634265895e-06,
      "loss": 0.8543,
      "step": 11820
    },
    {
      "epoch": 0.9552516212448736,
      "grad_norm": 2.398898124694824,
      "learning_rate": 5.603482890202335e-06,
      "loss": 1.0712,
      "step": 11821
    },
    {
      "epoch": 0.9553324309582012,
      "grad_norm": 2.7207298278808594,
      "learning_rate": 5.6028333066418725e-06,
      "loss": 0.9207,
      "step": 11822
    },
    {
      "epoch": 0.9554132406715287,
      "grad_norm": 2.7692246437072754,
      "learning_rate": 5.602183712756328e-06,
      "loss": 0.9189,
      "step": 11823
    },
    {
      "epoch": 0.9554940503848562,
      "grad_norm": 2.6773533821105957,
      "learning_rate": 5.601534108556824e-06,
      "loss": 1.0012,
      "step": 11824
    },
    {
      "epoch": 0.9555748600981838,
      "grad_norm": 2.6970131397247314,
      "learning_rate": 5.600884494054491e-06,
      "loss": 0.9982,
      "step": 11825
    },
    {
      "epoch": 0.9556556698115114,
      "grad_norm": 2.2516486644744873,
      "learning_rate": 5.600234869260451e-06,
      "loss": 0.9003,
      "step": 11826
    },
    {
      "epoch": 0.9557364795248389,
      "grad_norm": 2.576528310775757,
      "learning_rate": 5.599585234185836e-06,
      "loss": 0.9896,
      "step": 11827
    },
    {
      "epoch": 0.9558172892381664,
      "grad_norm": 2.8807623386383057,
      "learning_rate": 5.598935588841768e-06,
      "loss": 0.885,
      "step": 11828
    },
    {
      "epoch": 0.955898098951494,
      "grad_norm": 2.4815165996551514,
      "learning_rate": 5.598285933239373e-06,
      "loss": 0.8403,
      "step": 11829
    },
    {
      "epoch": 0.9559789086648215,
      "grad_norm": 2.2415409088134766,
      "learning_rate": 5.5976362673897825e-06,
      "loss": 0.863,
      "step": 11830
    },
    {
      "epoch": 0.9560597183781491,
      "grad_norm": 2.512897491455078,
      "learning_rate": 5.59698659130412e-06,
      "loss": 0.9813,
      "step": 11831
    },
    {
      "epoch": 0.9561405280914766,
      "grad_norm": 2.87442684173584,
      "learning_rate": 5.596336904993516e-06,
      "loss": 0.9838,
      "step": 11832
    },
    {
      "epoch": 0.9562213378048041,
      "grad_norm": 2.697709321975708,
      "learning_rate": 5.595687208469096e-06,
      "loss": 0.9258,
      "step": 11833
    },
    {
      "epoch": 0.9563021475181317,
      "grad_norm": 2.5676422119140625,
      "learning_rate": 5.5950375017419875e-06,
      "loss": 0.9948,
      "step": 11834
    },
    {
      "epoch": 0.9563829572314593,
      "grad_norm": 2.5715384483337402,
      "learning_rate": 5.5943877848233185e-06,
      "loss": 0.911,
      "step": 11835
    },
    {
      "epoch": 0.9564637669447867,
      "grad_norm": 2.583868980407715,
      "learning_rate": 5.59373805772422e-06,
      "loss": 0.9617,
      "step": 11836
    },
    {
      "epoch": 0.9565445766581143,
      "grad_norm": 2.7021234035491943,
      "learning_rate": 5.593088320455815e-06,
      "loss": 0.9524,
      "step": 11837
    },
    {
      "epoch": 0.9566253863714419,
      "grad_norm": 2.6744143962860107,
      "learning_rate": 5.592438573029236e-06,
      "loss": 0.8343,
      "step": 11838
    },
    {
      "epoch": 0.9567061960847694,
      "grad_norm": 2.4851043224334717,
      "learning_rate": 5.591788815455611e-06,
      "loss": 0.9552,
      "step": 11839
    },
    {
      "epoch": 0.956787005798097,
      "grad_norm": 2.3759748935699463,
      "learning_rate": 5.591139047746068e-06,
      "loss": 0.9101,
      "step": 11840
    },
    {
      "epoch": 0.9568678155114245,
      "grad_norm": 2.605360746383667,
      "learning_rate": 5.590489269911738e-06,
      "loss": 0.8584,
      "step": 11841
    },
    {
      "epoch": 0.956948625224752,
      "grad_norm": 2.6257309913635254,
      "learning_rate": 5.589839481963745e-06,
      "loss": 0.923,
      "step": 11842
    },
    {
      "epoch": 0.9570294349380796,
      "grad_norm": 2.8049874305725098,
      "learning_rate": 5.589189683913224e-06,
      "loss": 0.9867,
      "step": 11843
    },
    {
      "epoch": 0.9571102446514071,
      "grad_norm": 2.438661575317383,
      "learning_rate": 5.588539875771301e-06,
      "loss": 1.0174,
      "step": 11844
    },
    {
      "epoch": 0.9571910543647346,
      "grad_norm": 2.449471950531006,
      "learning_rate": 5.587890057549108e-06,
      "loss": 0.9168,
      "step": 11845
    },
    {
      "epoch": 0.9572718640780622,
      "grad_norm": 2.5123777389526367,
      "learning_rate": 5.587240229257773e-06,
      "loss": 0.99,
      "step": 11846
    },
    {
      "epoch": 0.9573526737913898,
      "grad_norm": 2.5206658840179443,
      "learning_rate": 5.586590390908426e-06,
      "loss": 0.8916,
      "step": 11847
    },
    {
      "epoch": 0.9574334835047172,
      "grad_norm": 2.9969146251678467,
      "learning_rate": 5.585940542512199e-06,
      "loss": 0.9834,
      "step": 11848
    },
    {
      "epoch": 0.9575142932180448,
      "grad_norm": 2.512291193008423,
      "learning_rate": 5.585290684080219e-06,
      "loss": 0.7877,
      "step": 11849
    },
    {
      "epoch": 0.9575951029313724,
      "grad_norm": 2.7943925857543945,
      "learning_rate": 5.584640815623621e-06,
      "loss": 0.968,
      "step": 11850
    },
    {
      "epoch": 0.9576759126446999,
      "grad_norm": 2.564974308013916,
      "learning_rate": 5.583990937153533e-06,
      "loss": 0.8463,
      "step": 11851
    },
    {
      "epoch": 0.9577567223580274,
      "grad_norm": 2.8840105533599854,
      "learning_rate": 5.583341048681085e-06,
      "loss": 0.908,
      "step": 11852
    },
    {
      "epoch": 0.957837532071355,
      "grad_norm": 2.7245535850524902,
      "learning_rate": 5.582691150217408e-06,
      "loss": 0.9197,
      "step": 11853
    },
    {
      "epoch": 0.9579183417846825,
      "grad_norm": 2.4554336071014404,
      "learning_rate": 5.582041241773637e-06,
      "loss": 0.9541,
      "step": 11854
    },
    {
      "epoch": 0.9579991514980101,
      "grad_norm": 2.4145753383636475,
      "learning_rate": 5.5813913233609e-06,
      "loss": 1.1204,
      "step": 11855
    },
    {
      "epoch": 0.9580799612113376,
      "grad_norm": 2.252882719039917,
      "learning_rate": 5.580741394990329e-06,
      "loss": 0.9932,
      "step": 11856
    },
    {
      "epoch": 0.9581607709246651,
      "grad_norm": 2.604893445968628,
      "learning_rate": 5.580091456673055e-06,
      "loss": 0.9917,
      "step": 11857
    },
    {
      "epoch": 0.9582415806379927,
      "grad_norm": 3.008949041366577,
      "learning_rate": 5.579441508420213e-06,
      "loss": 0.9565,
      "step": 11858
    },
    {
      "epoch": 0.9583223903513203,
      "grad_norm": 2.3516547679901123,
      "learning_rate": 5.5787915502429315e-06,
      "loss": 0.899,
      "step": 11859
    },
    {
      "epoch": 0.9584032000646477,
      "grad_norm": 2.8255181312561035,
      "learning_rate": 5.578141582152344e-06,
      "loss": 0.8812,
      "step": 11860
    },
    {
      "epoch": 0.9584840097779753,
      "grad_norm": 2.4160497188568115,
      "learning_rate": 5.577491604159583e-06,
      "loss": 0.9587,
      "step": 11861
    },
    {
      "epoch": 0.9585648194913029,
      "grad_norm": 2.656268358230591,
      "learning_rate": 5.576841616275782e-06,
      "loss": 0.8639,
      "step": 11862
    },
    {
      "epoch": 0.9586456292046304,
      "grad_norm": 2.7525899410247803,
      "learning_rate": 5.576191618512071e-06,
      "loss": 1.0354,
      "step": 11863
    },
    {
      "epoch": 0.958726438917958,
      "grad_norm": 2.9054365158081055,
      "learning_rate": 5.575541610879587e-06,
      "loss": 0.9397,
      "step": 11864
    },
    {
      "epoch": 0.9588072486312855,
      "grad_norm": 2.5695607662200928,
      "learning_rate": 5.57489159338946e-06,
      "loss": 0.8876,
      "step": 11865
    },
    {
      "epoch": 0.958888058344613,
      "grad_norm": 2.642263174057007,
      "learning_rate": 5.574241566052824e-06,
      "loss": 0.9237,
      "step": 11866
    },
    {
      "epoch": 0.9589688680579406,
      "grad_norm": 2.4560844898223877,
      "learning_rate": 5.573591528880812e-06,
      "loss": 0.8213,
      "step": 11867
    },
    {
      "epoch": 0.9590496777712681,
      "grad_norm": 2.4272077083587646,
      "learning_rate": 5.572941481884557e-06,
      "loss": 0.9842,
      "step": 11868
    },
    {
      "epoch": 0.9591304874845956,
      "grad_norm": 2.425368547439575,
      "learning_rate": 5.572291425075195e-06,
      "loss": 0.7658,
      "step": 11869
    },
    {
      "epoch": 0.9592112971979232,
      "grad_norm": 2.9005725383758545,
      "learning_rate": 5.5716413584638594e-06,
      "loss": 0.888,
      "step": 11870
    },
    {
      "epoch": 0.9592921069112508,
      "grad_norm": 2.5874617099761963,
      "learning_rate": 5.570991282061681e-06,
      "loss": 0.9456,
      "step": 11871
    },
    {
      "epoch": 0.9593729166245782,
      "grad_norm": 2.476445198059082,
      "learning_rate": 5.570341195879799e-06,
      "loss": 0.9491,
      "step": 11872
    },
    {
      "epoch": 0.9594537263379058,
      "grad_norm": 2.938046455383301,
      "learning_rate": 5.5696910999293444e-06,
      "loss": 1.0074,
      "step": 11873
    },
    {
      "epoch": 0.9595345360512334,
      "grad_norm": 2.6625962257385254,
      "learning_rate": 5.569040994221453e-06,
      "loss": 1.0006,
      "step": 11874
    },
    {
      "epoch": 0.9596153457645609,
      "grad_norm": 2.6659927368164062,
      "learning_rate": 5.568390878767258e-06,
      "loss": 0.9031,
      "step": 11875
    },
    {
      "epoch": 0.9596961554778884,
      "grad_norm": 2.371853828430176,
      "learning_rate": 5.567740753577898e-06,
      "loss": 0.9263,
      "step": 11876
    },
    {
      "epoch": 0.959776965191216,
      "grad_norm": 2.314671516418457,
      "learning_rate": 5.567090618664503e-06,
      "loss": 0.9129,
      "step": 11877
    },
    {
      "epoch": 0.9598577749045435,
      "grad_norm": 2.497958183288574,
      "learning_rate": 5.566440474038213e-06,
      "loss": 0.8058,
      "step": 11878
    },
    {
      "epoch": 0.9599385846178711,
      "grad_norm": 2.8782222270965576,
      "learning_rate": 5.56579031971016e-06,
      "loss": 0.9175,
      "step": 11879
    },
    {
      "epoch": 0.9600193943311986,
      "grad_norm": 3.011770009994507,
      "learning_rate": 5.565140155691482e-06,
      "loss": 0.9486,
      "step": 11880
    },
    {
      "epoch": 0.9601002040445261,
      "grad_norm": 2.551955223083496,
      "learning_rate": 5.564489981993313e-06,
      "loss": 0.8927,
      "step": 11881
    },
    {
      "epoch": 0.9601810137578537,
      "grad_norm": 2.612999200820923,
      "learning_rate": 5.563839798626789e-06,
      "loss": 0.9371,
      "step": 11882
    },
    {
      "epoch": 0.9602618234711813,
      "grad_norm": 3.1354992389678955,
      "learning_rate": 5.5631896056030475e-06,
      "loss": 0.976,
      "step": 11883
    },
    {
      "epoch": 0.9603426331845087,
      "grad_norm": 2.590576171875,
      "learning_rate": 5.562539402933225e-06,
      "loss": 0.8114,
      "step": 11884
    },
    {
      "epoch": 0.9604234428978363,
      "grad_norm": 2.5892324447631836,
      "learning_rate": 5.561889190628454e-06,
      "loss": 0.9457,
      "step": 11885
    },
    {
      "epoch": 0.9605042526111639,
      "grad_norm": 2.3162331581115723,
      "learning_rate": 5.5612389686998755e-06,
      "loss": 0.9203,
      "step": 11886
    },
    {
      "epoch": 0.9605850623244914,
      "grad_norm": 2.729736566543579,
      "learning_rate": 5.5605887371586254e-06,
      "loss": 0.8252,
      "step": 11887
    },
    {
      "epoch": 0.960665872037819,
      "grad_norm": 2.5039331912994385,
      "learning_rate": 5.559938496015838e-06,
      "loss": 0.8686,
      "step": 11888
    },
    {
      "epoch": 0.9607466817511465,
      "grad_norm": 2.8261559009552,
      "learning_rate": 5.559288245282652e-06,
      "loss": 0.828,
      "step": 11889
    },
    {
      "epoch": 0.960827491464474,
      "grad_norm": 2.9055511951446533,
      "learning_rate": 5.558637984970208e-06,
      "loss": 0.9344,
      "step": 11890
    },
    {
      "epoch": 0.9609083011778016,
      "grad_norm": 2.3357317447662354,
      "learning_rate": 5.557987715089637e-06,
      "loss": 1.0102,
      "step": 11891
    },
    {
      "epoch": 0.9609891108911292,
      "grad_norm": 2.7615742683410645,
      "learning_rate": 5.55733743565208e-06,
      "loss": 0.9038,
      "step": 11892
    },
    {
      "epoch": 0.9610699206044566,
      "grad_norm": 2.6449334621429443,
      "learning_rate": 5.556687146668675e-06,
      "loss": 0.9393,
      "step": 11893
    },
    {
      "epoch": 0.9611507303177842,
      "grad_norm": 2.7535130977630615,
      "learning_rate": 5.55603684815056e-06,
      "loss": 1.0233,
      "step": 11894
    },
    {
      "epoch": 0.9612315400311118,
      "grad_norm": 2.990295886993408,
      "learning_rate": 5.555386540108872e-06,
      "loss": 1.0242,
      "step": 11895
    },
    {
      "epoch": 0.9613123497444392,
      "grad_norm": 2.797285318374634,
      "learning_rate": 5.55473622255475e-06,
      "loss": 0.9128,
      "step": 11896
    },
    {
      "epoch": 0.9613931594577668,
      "grad_norm": 2.8205981254577637,
      "learning_rate": 5.55408589549933e-06,
      "loss": 0.8859,
      "step": 11897
    },
    {
      "epoch": 0.9614739691710944,
      "grad_norm": 2.5844244956970215,
      "learning_rate": 5.5534355589537545e-06,
      "loss": 0.8984,
      "step": 11898
    },
    {
      "epoch": 0.9615547788844219,
      "grad_norm": 2.822474241256714,
      "learning_rate": 5.552785212929159e-06,
      "loss": 0.9482,
      "step": 11899
    },
    {
      "epoch": 0.9616355885977494,
      "grad_norm": 2.5194263458251953,
      "learning_rate": 5.5521348574366864e-06,
      "loss": 0.912,
      "step": 11900
    },
    {
      "epoch": 0.961716398311077,
      "grad_norm": 3.069188117980957,
      "learning_rate": 5.55148449248747e-06,
      "loss": 0.96,
      "step": 11901
    },
    {
      "epoch": 0.9617972080244045,
      "grad_norm": 2.8065030574798584,
      "learning_rate": 5.5508341180926524e-06,
      "loss": 1.038,
      "step": 11902
    },
    {
      "epoch": 0.9618780177377321,
      "grad_norm": 2.3951709270477295,
      "learning_rate": 5.5501837342633725e-06,
      "loss": 0.9176,
      "step": 11903
    },
    {
      "epoch": 0.9619588274510597,
      "grad_norm": 3.390561103820801,
      "learning_rate": 5.54953334101077e-06,
      "loss": 0.8808,
      "step": 11904
    },
    {
      "epoch": 0.9620396371643871,
      "grad_norm": 2.726771593093872,
      "learning_rate": 5.548882938345984e-06,
      "loss": 0.9092,
      "step": 11905
    },
    {
      "epoch": 0.9621204468777147,
      "grad_norm": 2.804781436920166,
      "learning_rate": 5.5482325262801554e-06,
      "loss": 0.8905,
      "step": 11906
    },
    {
      "epoch": 0.9622012565910423,
      "grad_norm": 2.394566535949707,
      "learning_rate": 5.547582104824423e-06,
      "loss": 0.8691,
      "step": 11907
    },
    {
      "epoch": 0.9622820663043697,
      "grad_norm": 2.655407428741455,
      "learning_rate": 5.546931673989927e-06,
      "loss": 0.9035,
      "step": 11908
    },
    {
      "epoch": 0.9623628760176973,
      "grad_norm": 2.811229944229126,
      "learning_rate": 5.546281233787809e-06,
      "loss": 0.9811,
      "step": 11909
    },
    {
      "epoch": 0.9624436857310249,
      "grad_norm": 2.610511064529419,
      "learning_rate": 5.545630784229208e-06,
      "loss": 0.9878,
      "step": 11910
    },
    {
      "epoch": 0.9625244954443524,
      "grad_norm": 2.0195400714874268,
      "learning_rate": 5.544980325325264e-06,
      "loss": 0.9336,
      "step": 11911
    },
    {
      "epoch": 0.96260530515768,
      "grad_norm": 3.0447850227355957,
      "learning_rate": 5.5443298570871205e-06,
      "loss": 0.952,
      "step": 11912
    },
    {
      "epoch": 0.9626861148710075,
      "grad_norm": 2.9219303131103516,
      "learning_rate": 5.543679379525917e-06,
      "loss": 0.9372,
      "step": 11913
    },
    {
      "epoch": 0.962766924584335,
      "grad_norm": 2.240722894668579,
      "learning_rate": 5.543028892652794e-06,
      "loss": 0.8187,
      "step": 11914
    },
    {
      "epoch": 0.9628477342976626,
      "grad_norm": 2.5248091220855713,
      "learning_rate": 5.542378396478894e-06,
      "loss": 0.8664,
      "step": 11915
    },
    {
      "epoch": 0.9629285440109902,
      "grad_norm": 2.4241583347320557,
      "learning_rate": 5.541727891015357e-06,
      "loss": 1.0349,
      "step": 11916
    },
    {
      "epoch": 0.9630093537243176,
      "grad_norm": 3.2690250873565674,
      "learning_rate": 5.541077376273327e-06,
      "loss": 0.8942,
      "step": 11917
    },
    {
      "epoch": 0.9630901634376452,
      "grad_norm": 2.689641237258911,
      "learning_rate": 5.540426852263943e-06,
      "loss": 1.0096,
      "step": 11918
    },
    {
      "epoch": 0.9631709731509728,
      "grad_norm": 2.767505168914795,
      "learning_rate": 5.5397763189983475e-06,
      "loss": 0.8507,
      "step": 11919
    },
    {
      "epoch": 0.9632517828643002,
      "grad_norm": 3.236069440841675,
      "learning_rate": 5.539125776487684e-06,
      "loss": 1.0064,
      "step": 11920
    },
    {
      "epoch": 0.9633325925776278,
      "grad_norm": 2.6238200664520264,
      "learning_rate": 5.538475224743094e-06,
      "loss": 1.0667,
      "step": 11921
    },
    {
      "epoch": 0.9634134022909554,
      "grad_norm": 3.1788158416748047,
      "learning_rate": 5.537824663775719e-06,
      "loss": 0.9107,
      "step": 11922
    },
    {
      "epoch": 0.9634942120042829,
      "grad_norm": 2.5138230323791504,
      "learning_rate": 5.5371740935967026e-06,
      "loss": 0.9272,
      "step": 11923
    },
    {
      "epoch": 0.9635750217176104,
      "grad_norm": 3.8556137084960938,
      "learning_rate": 5.536523514217186e-06,
      "loss": 0.8909,
      "step": 11924
    },
    {
      "epoch": 0.963655831430938,
      "grad_norm": 2.61544132232666,
      "learning_rate": 5.5358729256483145e-06,
      "loss": 0.935,
      "step": 11925
    },
    {
      "epoch": 0.9637366411442655,
      "grad_norm": 2.8928191661834717,
      "learning_rate": 5.53522232790123e-06,
      "loss": 0.9328,
      "step": 11926
    },
    {
      "epoch": 0.9638174508575931,
      "grad_norm": 2.9073996543884277,
      "learning_rate": 5.534571720987076e-06,
      "loss": 0.8416,
      "step": 11927
    },
    {
      "epoch": 0.9638982605709207,
      "grad_norm": 3.052318572998047,
      "learning_rate": 5.533921104916993e-06,
      "loss": 0.8459,
      "step": 11928
    },
    {
      "epoch": 0.9639790702842481,
      "grad_norm": 2.600609302520752,
      "learning_rate": 5.5332704797021295e-06,
      "loss": 0.8499,
      "step": 11929
    },
    {
      "epoch": 0.9640598799975757,
      "grad_norm": 2.682948589324951,
      "learning_rate": 5.532619845353624e-06,
      "loss": 0.8424,
      "step": 11930
    },
    {
      "epoch": 0.9641406897109033,
      "grad_norm": 2.9137091636657715,
      "learning_rate": 5.531969201882625e-06,
      "loss": 0.9018,
      "step": 11931
    },
    {
      "epoch": 0.9642214994242307,
      "grad_norm": 2.423874616622925,
      "learning_rate": 5.531318549300273e-06,
      "loss": 1.011,
      "step": 11932
    },
    {
      "epoch": 0.9643023091375583,
      "grad_norm": 2.705209255218506,
      "learning_rate": 5.5306678876177135e-06,
      "loss": 0.941,
      "step": 11933
    },
    {
      "epoch": 0.9643831188508859,
      "grad_norm": 2.503817319869995,
      "learning_rate": 5.530017216846091e-06,
      "loss": 0.9454,
      "step": 11934
    },
    {
      "epoch": 0.9644639285642134,
      "grad_norm": 2.6077113151550293,
      "learning_rate": 5.529366536996549e-06,
      "loss": 0.9686,
      "step": 11935
    },
    {
      "epoch": 0.964544738277541,
      "grad_norm": 2.6398589611053467,
      "learning_rate": 5.528715848080233e-06,
      "loss": 0.9402,
      "step": 11936
    },
    {
      "epoch": 0.9646255479908685,
      "grad_norm": 2.7943220138549805,
      "learning_rate": 5.5280651501082886e-06,
      "loss": 0.8676,
      "step": 11937
    },
    {
      "epoch": 0.964706357704196,
      "grad_norm": 3.4839236736297607,
      "learning_rate": 5.527414443091856e-06,
      "loss": 0.9526,
      "step": 11938
    },
    {
      "epoch": 0.9647871674175236,
      "grad_norm": 2.680088996887207,
      "learning_rate": 5.526763727042087e-06,
      "loss": 0.8524,
      "step": 11939
    },
    {
      "epoch": 0.9648679771308512,
      "grad_norm": 2.4965760707855225,
      "learning_rate": 5.5261130019701225e-06,
      "loss": 0.9418,
      "step": 11940
    },
    {
      "epoch": 0.9649487868441786,
      "grad_norm": 2.5393991470336914,
      "learning_rate": 5.525462267887108e-06,
      "loss": 0.8356,
      "step": 11941
    },
    {
      "epoch": 0.9650295965575062,
      "grad_norm": 2.610919952392578,
      "learning_rate": 5.524811524804191e-06,
      "loss": 0.9368,
      "step": 11942
    },
    {
      "epoch": 0.9651104062708338,
      "grad_norm": 2.57777738571167,
      "learning_rate": 5.524160772732517e-06,
      "loss": 0.9092,
      "step": 11943
    },
    {
      "epoch": 0.9651912159841612,
      "grad_norm": 2.3386118412017822,
      "learning_rate": 5.5235100116832275e-06,
      "loss": 0.8654,
      "step": 11944
    },
    {
      "epoch": 0.9652720256974888,
      "grad_norm": 2.5760498046875,
      "learning_rate": 5.522859241667475e-06,
      "loss": 0.916,
      "step": 11945
    },
    {
      "epoch": 0.9653528354108164,
      "grad_norm": 2.5684726238250732,
      "learning_rate": 5.5222084626964e-06,
      "loss": 0.8797,
      "step": 11946
    },
    {
      "epoch": 0.9654336451241439,
      "grad_norm": 2.869614601135254,
      "learning_rate": 5.521557674781153e-06,
      "loss": 0.967,
      "step": 11947
    },
    {
      "epoch": 0.9655144548374714,
      "grad_norm": 2.1509933471679688,
      "learning_rate": 5.520906877932877e-06,
      "loss": 0.9737,
      "step": 11948
    },
    {
      "epoch": 0.965595264550799,
      "grad_norm": 2.528146982192993,
      "learning_rate": 5.520256072162722e-06,
      "loss": 0.8562,
      "step": 11949
    },
    {
      "epoch": 0.9656760742641265,
      "grad_norm": 2.6308462619781494,
      "learning_rate": 5.519605257481832e-06,
      "loss": 0.994,
      "step": 11950
    },
    {
      "epoch": 0.9657568839774541,
      "grad_norm": 3.2644970417022705,
      "learning_rate": 5.518954433901356e-06,
      "loss": 1.0149,
      "step": 11951
    },
    {
      "epoch": 0.9658376936907817,
      "grad_norm": 3.344829559326172,
      "learning_rate": 5.518303601432437e-06,
      "loss": 1.0481,
      "step": 11952
    },
    {
      "epoch": 0.9659185034041091,
      "grad_norm": 2.8689024448394775,
      "learning_rate": 5.517652760086227e-06,
      "loss": 1.0843,
      "step": 11953
    },
    {
      "epoch": 0.9659993131174367,
      "grad_norm": 2.743041753768921,
      "learning_rate": 5.517001909873871e-06,
      "loss": 0.8881,
      "step": 11954
    },
    {
      "epoch": 0.9660801228307643,
      "grad_norm": 2.755493640899658,
      "learning_rate": 5.516351050806518e-06,
      "loss": 0.9139,
      "step": 11955
    },
    {
      "epoch": 0.9661609325440919,
      "grad_norm": 2.8872082233428955,
      "learning_rate": 5.515700182895314e-06,
      "loss": 0.848,
      "step": 11956
    },
    {
      "epoch": 0.9662417422574193,
      "grad_norm": 2.6378605365753174,
      "learning_rate": 5.515049306151408e-06,
      "loss": 0.8225,
      "step": 11957
    },
    {
      "epoch": 0.9663225519707469,
      "grad_norm": 2.765263557434082,
      "learning_rate": 5.514398420585945e-06,
      "loss": 0.8984,
      "step": 11958
    },
    {
      "epoch": 0.9664033616840745,
      "grad_norm": 2.951899290084839,
      "learning_rate": 5.513747526210077e-06,
      "loss": 0.942,
      "step": 11959
    },
    {
      "epoch": 0.966484171397402,
      "grad_norm": 2.599083423614502,
      "learning_rate": 5.51309662303495e-06,
      "loss": 0.8274,
      "step": 11960
    },
    {
      "epoch": 0.9665649811107295,
      "grad_norm": 2.4001996517181396,
      "learning_rate": 5.512445711071714e-06,
      "loss": 0.7794,
      "step": 11961
    },
    {
      "epoch": 0.9666457908240571,
      "grad_norm": 2.3266730308532715,
      "learning_rate": 5.511794790331516e-06,
      "loss": 1.0175,
      "step": 11962
    },
    {
      "epoch": 0.9667266005373846,
      "grad_norm": 3.0140881538391113,
      "learning_rate": 5.511143860825506e-06,
      "loss": 1.0362,
      "step": 11963
    },
    {
      "epoch": 0.9668074102507122,
      "grad_norm": 2.9782848358154297,
      "learning_rate": 5.510492922564832e-06,
      "loss": 0.9554,
      "step": 11964
    },
    {
      "epoch": 0.9668882199640397,
      "grad_norm": 3.11130690574646,
      "learning_rate": 5.509841975560644e-06,
      "loss": 0.8305,
      "step": 11965
    },
    {
      "epoch": 0.9669690296773672,
      "grad_norm": 2.929426670074463,
      "learning_rate": 5.509191019824087e-06,
      "loss": 0.8546,
      "step": 11966
    },
    {
      "epoch": 0.9670498393906948,
      "grad_norm": 2.5395426750183105,
      "learning_rate": 5.508540055366317e-06,
      "loss": 0.8987,
      "step": 11967
    },
    {
      "epoch": 0.9671306491040224,
      "grad_norm": 3.2608258724212646,
      "learning_rate": 5.5078890821984795e-06,
      "loss": 0.9975,
      "step": 11968
    },
    {
      "epoch": 0.9672114588173498,
      "grad_norm": 2.535095691680908,
      "learning_rate": 5.5072381003317245e-06,
      "loss": 0.8209,
      "step": 11969
    },
    {
      "epoch": 0.9672922685306774,
      "grad_norm": 2.618473768234253,
      "learning_rate": 5.5065871097772015e-06,
      "loss": 0.9664,
      "step": 11970
    },
    {
      "epoch": 0.967373078244005,
      "grad_norm": 2.96295428276062,
      "learning_rate": 5.505936110546061e-06,
      "loss": 0.9803,
      "step": 11971
    },
    {
      "epoch": 0.9674538879573324,
      "grad_norm": 2.8674564361572266,
      "learning_rate": 5.505285102649452e-06,
      "loss": 0.8752,
      "step": 11972
    },
    {
      "epoch": 0.96753469767066,
      "grad_norm": 3.0477328300476074,
      "learning_rate": 5.504634086098527e-06,
      "loss": 0.8243,
      "step": 11973
    },
    {
      "epoch": 0.9676155073839876,
      "grad_norm": 3.2491588592529297,
      "learning_rate": 5.503983060904436e-06,
      "loss": 0.9981,
      "step": 11974
    },
    {
      "epoch": 0.9676963170973151,
      "grad_norm": 2.4708495140075684,
      "learning_rate": 5.5033320270783265e-06,
      "loss": 0.8138,
      "step": 11975
    },
    {
      "epoch": 0.9677771268106427,
      "grad_norm": 2.916604518890381,
      "learning_rate": 5.502680984631351e-06,
      "loss": 0.8819,
      "step": 11976
    },
    {
      "epoch": 0.9678579365239702,
      "grad_norm": 2.530425786972046,
      "learning_rate": 5.502029933574662e-06,
      "loss": 0.8206,
      "step": 11977
    },
    {
      "epoch": 0.9679387462372977,
      "grad_norm": 2.459057331085205,
      "learning_rate": 5.501378873919407e-06,
      "loss": 0.8387,
      "step": 11978
    },
    {
      "epoch": 0.9680195559506253,
      "grad_norm": 2.8737850189208984,
      "learning_rate": 5.50072780567674e-06,
      "loss": 0.8975,
      "step": 11979
    },
    {
      "epoch": 0.9681003656639529,
      "grad_norm": 2.7501003742218018,
      "learning_rate": 5.50007672885781e-06,
      "loss": 0.951,
      "step": 11980
    },
    {
      "epoch": 0.9681811753772803,
      "grad_norm": 2.6252670288085938,
      "learning_rate": 5.499425643473771e-06,
      "loss": 0.9224,
      "step": 11981
    },
    {
      "epoch": 0.9682619850906079,
      "grad_norm": 2.9512546062469482,
      "learning_rate": 5.498774549535773e-06,
      "loss": 0.8445,
      "step": 11982
    },
    {
      "epoch": 0.9683427948039355,
      "grad_norm": 2.7344777584075928,
      "learning_rate": 5.498123447054966e-06,
      "loss": 0.7579,
      "step": 11983
    },
    {
      "epoch": 0.968423604517263,
      "grad_norm": 2.848266124725342,
      "learning_rate": 5.4974723360425066e-06,
      "loss": 0.8675,
      "step": 11984
    },
    {
      "epoch": 0.9685044142305905,
      "grad_norm": 2.897792100906372,
      "learning_rate": 5.4968212165095415e-06,
      "loss": 0.9332,
      "step": 11985
    },
    {
      "epoch": 0.9685852239439181,
      "grad_norm": 2.544478416442871,
      "learning_rate": 5.496170088467225e-06,
      "loss": 0.9673,
      "step": 11986
    },
    {
      "epoch": 0.9686660336572456,
      "grad_norm": 2.631855010986328,
      "learning_rate": 5.495518951926709e-06,
      "loss": 0.87,
      "step": 11987
    },
    {
      "epoch": 0.9687468433705732,
      "grad_norm": 2.7205095291137695,
      "learning_rate": 5.494867806899149e-06,
      "loss": 0.9345,
      "step": 11988
    },
    {
      "epoch": 0.9688276530839007,
      "grad_norm": 2.4885149002075195,
      "learning_rate": 5.494216653395691e-06,
      "loss": 0.8209,
      "step": 11989
    },
    {
      "epoch": 0.9689084627972282,
      "grad_norm": 2.4030749797821045,
      "learning_rate": 5.493565491427495e-06,
      "loss": 1.019,
      "step": 11990
    },
    {
      "epoch": 0.9689892725105558,
      "grad_norm": 3.011697292327881,
      "learning_rate": 5.492914321005707e-06,
      "loss": 0.8068,
      "step": 11991
    },
    {
      "epoch": 0.9690700822238834,
      "grad_norm": 2.883098602294922,
      "learning_rate": 5.492263142141486e-06,
      "loss": 0.9623,
      "step": 11992
    },
    {
      "epoch": 0.9691508919372108,
      "grad_norm": 2.875462293624878,
      "learning_rate": 5.491611954845981e-06,
      "loss": 0.9659,
      "step": 11993
    },
    {
      "epoch": 0.9692317016505384,
      "grad_norm": 2.7993433475494385,
      "learning_rate": 5.4909607591303474e-06,
      "loss": 0.9546,
      "step": 11994
    },
    {
      "epoch": 0.969312511363866,
      "grad_norm": 2.930129289627075,
      "learning_rate": 5.490309555005738e-06,
      "loss": 0.9378,
      "step": 11995
    },
    {
      "epoch": 0.9693933210771934,
      "grad_norm": 2.551013231277466,
      "learning_rate": 5.489658342483306e-06,
      "loss": 0.9169,
      "step": 11996
    },
    {
      "epoch": 0.969474130790521,
      "grad_norm": 2.400601863861084,
      "learning_rate": 5.489007121574205e-06,
      "loss": 0.9753,
      "step": 11997
    },
    {
      "epoch": 0.9695549405038486,
      "grad_norm": 2.214895248413086,
      "learning_rate": 5.488355892289591e-06,
      "loss": 1.0205,
      "step": 11998
    },
    {
      "epoch": 0.9696357502171761,
      "grad_norm": 2.7257273197174072,
      "learning_rate": 5.4877046546406146e-06,
      "loss": 0.8608,
      "step": 11999
    },
    {
      "epoch": 0.9697165599305037,
      "grad_norm": 2.630215883255005,
      "learning_rate": 5.487053408638431e-06,
      "loss": 0.9609,
      "step": 12000
    },
    {
      "epoch": 0.9697165599305037,
      "eval_loss": 0.7585760354995728,
      "eval_runtime": 815.1901,
      "eval_samples_per_second": 102.266,
      "eval_steps_per_second": 12.784,
      "step": 12000
    },
    {
      "epoch": 0.9697973696438312,
      "grad_norm": 3.021000385284424,
      "learning_rate": 5.486402154294196e-06,
      "loss": 0.8275,
      "step": 12001
    },
    {
      "epoch": 0.9698781793571587,
      "grad_norm": 2.782231092453003,
      "learning_rate": 5.485750891619064e-06,
      "loss": 0.8195,
      "step": 12002
    },
    {
      "epoch": 0.9699589890704863,
      "grad_norm": 2.2484138011932373,
      "learning_rate": 5.4850996206241855e-06,
      "loss": 0.9467,
      "step": 12003
    },
    {
      "epoch": 0.9700397987838139,
      "grad_norm": 2.9472787380218506,
      "learning_rate": 5.484448341320722e-06,
      "loss": 1.0583,
      "step": 12004
    },
    {
      "epoch": 0.9701206084971413,
      "grad_norm": 3.0565943717956543,
      "learning_rate": 5.4837970537198214e-06,
      "loss": 0.8226,
      "step": 12005
    },
    {
      "epoch": 0.9702014182104689,
      "grad_norm": 2.46749210357666,
      "learning_rate": 5.483145757832645e-06,
      "loss": 0.8567,
      "step": 12006
    },
    {
      "epoch": 0.9702822279237965,
      "grad_norm": 2.705247402191162,
      "learning_rate": 5.4824944536703424e-06,
      "loss": 0.8806,
      "step": 12007
    },
    {
      "epoch": 0.970363037637124,
      "grad_norm": 2.4201948642730713,
      "learning_rate": 5.481843141244073e-06,
      "loss": 0.8836,
      "step": 12008
    },
    {
      "epoch": 0.9704438473504515,
      "grad_norm": 2.4932138919830322,
      "learning_rate": 5.48119182056499e-06,
      "loss": 0.9435,
      "step": 12009
    },
    {
      "epoch": 0.9705246570637791,
      "grad_norm": 2.6153910160064697,
      "learning_rate": 5.480540491644251e-06,
      "loss": 0.7817,
      "step": 12010
    },
    {
      "epoch": 0.9706054667771066,
      "grad_norm": 2.635456085205078,
      "learning_rate": 5.479889154493008e-06,
      "loss": 0.9235,
      "step": 12011
    },
    {
      "epoch": 0.9706862764904342,
      "grad_norm": 2.6182751655578613,
      "learning_rate": 5.479237809122421e-06,
      "loss": 0.906,
      "step": 12012
    },
    {
      "epoch": 0.9707670862037617,
      "grad_norm": 2.5122549533843994,
      "learning_rate": 5.478586455543642e-06,
      "loss": 0.9398,
      "step": 12013
    },
    {
      "epoch": 0.9708478959170892,
      "grad_norm": 2.6415774822235107,
      "learning_rate": 5.47793509376783e-06,
      "loss": 0.9097,
      "step": 12014
    },
    {
      "epoch": 0.9709287056304168,
      "grad_norm": 2.48339581489563,
      "learning_rate": 5.47728372380614e-06,
      "loss": 0.9476,
      "step": 12015
    },
    {
      "epoch": 0.9710095153437444,
      "grad_norm": 2.688344955444336,
      "learning_rate": 5.476632345669731e-06,
      "loss": 0.9421,
      "step": 12016
    },
    {
      "epoch": 0.9710903250570718,
      "grad_norm": 2.872114896774292,
      "learning_rate": 5.475980959369754e-06,
      "loss": 1.0038,
      "step": 12017
    },
    {
      "epoch": 0.9711711347703994,
      "grad_norm": 2.2378416061401367,
      "learning_rate": 5.475329564917372e-06,
      "loss": 0.8921,
      "step": 12018
    },
    {
      "epoch": 0.971251944483727,
      "grad_norm": 2.508561849594116,
      "learning_rate": 5.4746781623237365e-06,
      "loss": 0.9155,
      "step": 12019
    },
    {
      "epoch": 0.9713327541970544,
      "grad_norm": 2.6900382041931152,
      "learning_rate": 5.474026751600009e-06,
      "loss": 0.9107,
      "step": 12020
    },
    {
      "epoch": 0.971413563910382,
      "grad_norm": 2.8969435691833496,
      "learning_rate": 5.473375332757344e-06,
      "loss": 0.7556,
      "step": 12021
    },
    {
      "epoch": 0.9714943736237096,
      "grad_norm": 2.584876775741577,
      "learning_rate": 5.472723905806898e-06,
      "loss": 0.8506,
      "step": 12022
    },
    {
      "epoch": 0.9715751833370371,
      "grad_norm": 2.85139799118042,
      "learning_rate": 5.472072470759829e-06,
      "loss": 0.8943,
      "step": 12023
    },
    {
      "epoch": 0.9716559930503647,
      "grad_norm": 2.6479482650756836,
      "learning_rate": 5.471421027627297e-06,
      "loss": 0.9549,
      "step": 12024
    },
    {
      "epoch": 0.9717368027636922,
      "grad_norm": 2.532615900039673,
      "learning_rate": 5.470769576420456e-06,
      "loss": 0.879,
      "step": 12025
    },
    {
      "epoch": 0.9718176124770197,
      "grad_norm": 3.0018045902252197,
      "learning_rate": 5.470118117150467e-06,
      "loss": 0.8712,
      "step": 12026
    },
    {
      "epoch": 0.9718984221903473,
      "grad_norm": 2.863424777984619,
      "learning_rate": 5.469466649828487e-06,
      "loss": 0.8958,
      "step": 12027
    },
    {
      "epoch": 0.9719792319036749,
      "grad_norm": 2.624499797821045,
      "learning_rate": 5.468815174465672e-06,
      "loss": 0.9215,
      "step": 12028
    },
    {
      "epoch": 0.9720600416170023,
      "grad_norm": 2.3520593643188477,
      "learning_rate": 5.46816369107318e-06,
      "loss": 0.7808,
      "step": 12029
    },
    {
      "epoch": 0.9721408513303299,
      "grad_norm": 3.0101473331451416,
      "learning_rate": 5.4675121996621736e-06,
      "loss": 0.8992,
      "step": 12030
    },
    {
      "epoch": 0.9722216610436575,
      "grad_norm": 2.4136831760406494,
      "learning_rate": 5.466860700243808e-06,
      "loss": 0.9802,
      "step": 12031
    },
    {
      "epoch": 0.972302470756985,
      "grad_norm": 2.4630825519561768,
      "learning_rate": 5.466209192829244e-06,
      "loss": 0.9886,
      "step": 12032
    },
    {
      "epoch": 0.9723832804703125,
      "grad_norm": 2.6207082271575928,
      "learning_rate": 5.465557677429636e-06,
      "loss": 0.9481,
      "step": 12033
    },
    {
      "epoch": 0.9724640901836401,
      "grad_norm": 2.466914415359497,
      "learning_rate": 5.464906154056147e-06,
      "loss": 0.9045,
      "step": 12034
    },
    {
      "epoch": 0.9725448998969676,
      "grad_norm": 2.6892282962799072,
      "learning_rate": 5.464254622719936e-06,
      "loss": 0.9514,
      "step": 12035
    },
    {
      "epoch": 0.9726257096102952,
      "grad_norm": 2.581954002380371,
      "learning_rate": 5.4636030834321595e-06,
      "loss": 0.8804,
      "step": 12036
    },
    {
      "epoch": 0.9727065193236227,
      "grad_norm": 2.4629170894622803,
      "learning_rate": 5.462951536203979e-06,
      "loss": 0.9844,
      "step": 12037
    },
    {
      "epoch": 0.9727873290369502,
      "grad_norm": 2.5060341358184814,
      "learning_rate": 5.462299981046553e-06,
      "loss": 0.9414,
      "step": 12038
    },
    {
      "epoch": 0.9728681387502778,
      "grad_norm": 2.9171016216278076,
      "learning_rate": 5.46164841797104e-06,
      "loss": 0.9069,
      "step": 12039
    },
    {
      "epoch": 0.9729489484636054,
      "grad_norm": 2.539320468902588,
      "learning_rate": 5.460996846988602e-06,
      "loss": 0.8887,
      "step": 12040
    },
    {
      "epoch": 0.9730297581769328,
      "grad_norm": 2.4009783267974854,
      "learning_rate": 5.460345268110399e-06,
      "loss": 0.8742,
      "step": 12041
    },
    {
      "epoch": 0.9731105678902604,
      "grad_norm": 2.4767725467681885,
      "learning_rate": 5.459693681347588e-06,
      "loss": 0.9974,
      "step": 12042
    },
    {
      "epoch": 0.973191377603588,
      "grad_norm": 3.1554806232452393,
      "learning_rate": 5.4590420867113325e-06,
      "loss": 0.9001,
      "step": 12043
    },
    {
      "epoch": 0.9732721873169154,
      "grad_norm": 2.9938976764678955,
      "learning_rate": 5.45839048421279e-06,
      "loss": 0.9279,
      "step": 12044
    },
    {
      "epoch": 0.973352997030243,
      "grad_norm": 2.3939712047576904,
      "learning_rate": 5.457738873863122e-06,
      "loss": 0.9318,
      "step": 12045
    },
    {
      "epoch": 0.9734338067435706,
      "grad_norm": 2.897372007369995,
      "learning_rate": 5.45708725567349e-06,
      "loss": 0.9627,
      "step": 12046
    },
    {
      "epoch": 0.9735146164568981,
      "grad_norm": 2.5381247997283936,
      "learning_rate": 5.4564356296550524e-06,
      "loss": 0.8693,
      "step": 12047
    },
    {
      "epoch": 0.9735954261702257,
      "grad_norm": 2.783339262008667,
      "learning_rate": 5.455783995818972e-06,
      "loss": 0.963,
      "step": 12048
    },
    {
      "epoch": 0.9736762358835532,
      "grad_norm": 2.798495054244995,
      "learning_rate": 5.45513235417641e-06,
      "loss": 0.9404,
      "step": 12049
    },
    {
      "epoch": 0.9737570455968807,
      "grad_norm": 2.6823842525482178,
      "learning_rate": 5.454480704738524e-06,
      "loss": 0.9302,
      "step": 12050
    },
    {
      "epoch": 0.9738378553102083,
      "grad_norm": 2.952755928039551,
      "learning_rate": 5.45382904751648e-06,
      "loss": 0.9424,
      "step": 12051
    },
    {
      "epoch": 0.9739186650235359,
      "grad_norm": 2.4954073429107666,
      "learning_rate": 5.453177382521436e-06,
      "loss": 1.122,
      "step": 12052
    },
    {
      "epoch": 0.9739994747368633,
      "grad_norm": 2.240203857421875,
      "learning_rate": 5.452525709764555e-06,
      "loss": 1.0148,
      "step": 12053
    },
    {
      "epoch": 0.9740802844501909,
      "grad_norm": 2.3791136741638184,
      "learning_rate": 5.451874029256997e-06,
      "loss": 0.9732,
      "step": 12054
    },
    {
      "epoch": 0.9741610941635185,
      "grad_norm": 2.5452420711517334,
      "learning_rate": 5.4512223410099264e-06,
      "loss": 0.8804,
      "step": 12055
    },
    {
      "epoch": 0.974241903876846,
      "grad_norm": 2.996633291244507,
      "learning_rate": 5.450570645034501e-06,
      "loss": 0.9005,
      "step": 12056
    },
    {
      "epoch": 0.9743227135901735,
      "grad_norm": 2.428724527359009,
      "learning_rate": 5.449918941341887e-06,
      "loss": 0.9323,
      "step": 12057
    },
    {
      "epoch": 0.9744035233035011,
      "grad_norm": 2.5964088439941406,
      "learning_rate": 5.4492672299432424e-06,
      "loss": 0.974,
      "step": 12058
    },
    {
      "epoch": 0.9744843330168286,
      "grad_norm": 3.083604097366333,
      "learning_rate": 5.448615510849735e-06,
      "loss": 0.9759,
      "step": 12059
    },
    {
      "epoch": 0.9745651427301562,
      "grad_norm": 2.5858798027038574,
      "learning_rate": 5.447963784072522e-06,
      "loss": 0.9801,
      "step": 12060
    },
    {
      "epoch": 0.9746459524434837,
      "grad_norm": 2.552480697631836,
      "learning_rate": 5.447312049622767e-06,
      "loss": 0.8827,
      "step": 12061
    },
    {
      "epoch": 0.9747267621568112,
      "grad_norm": 2.3750808238983154,
      "learning_rate": 5.446660307511635e-06,
      "loss": 0.8882,
      "step": 12062
    },
    {
      "epoch": 0.9748075718701388,
      "grad_norm": 2.459052085876465,
      "learning_rate": 5.446008557750288e-06,
      "loss": 0.9051,
      "step": 12063
    },
    {
      "epoch": 0.9748883815834664,
      "grad_norm": 2.8203341960906982,
      "learning_rate": 5.445356800349886e-06,
      "loss": 0.8869,
      "step": 12064
    },
    {
      "epoch": 0.9749691912967938,
      "grad_norm": 2.5629642009735107,
      "learning_rate": 5.444705035321596e-06,
      "loss": 0.929,
      "step": 12065
    },
    {
      "epoch": 0.9750500010101214,
      "grad_norm": 2.6861367225646973,
      "learning_rate": 5.444053262676577e-06,
      "loss": 0.8866,
      "step": 12066
    },
    {
      "epoch": 0.975130810723449,
      "grad_norm": 2.8241360187530518,
      "learning_rate": 5.443401482425996e-06,
      "loss": 0.9322,
      "step": 12067
    },
    {
      "epoch": 0.9752116204367764,
      "grad_norm": 2.5909547805786133,
      "learning_rate": 5.442749694581015e-06,
      "loss": 0.8902,
      "step": 12068
    },
    {
      "epoch": 0.975292430150104,
      "grad_norm": 2.752270221710205,
      "learning_rate": 5.442097899152798e-06,
      "loss": 0.8437,
      "step": 12069
    },
    {
      "epoch": 0.9753732398634316,
      "grad_norm": 2.348259687423706,
      "learning_rate": 5.441446096152507e-06,
      "loss": 0.9196,
      "step": 12070
    },
    {
      "epoch": 0.9754540495767591,
      "grad_norm": 2.414008855819702,
      "learning_rate": 5.440794285591309e-06,
      "loss": 1.0726,
      "step": 12071
    },
    {
      "epoch": 0.9755348592900867,
      "grad_norm": 3.0321905612945557,
      "learning_rate": 5.440142467480364e-06,
      "loss": 0.9362,
      "step": 12072
    },
    {
      "epoch": 0.9756156690034142,
      "grad_norm": 2.7788331508636475,
      "learning_rate": 5.43949064183084e-06,
      "loss": 0.9472,
      "step": 12073
    },
    {
      "epoch": 0.9756964787167417,
      "grad_norm": 2.6454648971557617,
      "learning_rate": 5.438838808653899e-06,
      "loss": 0.8909,
      "step": 12074
    },
    {
      "epoch": 0.9757772884300693,
      "grad_norm": 2.9046497344970703,
      "learning_rate": 5.4381869679607045e-06,
      "loss": 0.8763,
      "step": 12075
    },
    {
      "epoch": 0.9758580981433969,
      "grad_norm": 2.7738723754882812,
      "learning_rate": 5.437535119762423e-06,
      "loss": 0.9295,
      "step": 12076
    },
    {
      "epoch": 0.9759389078567243,
      "grad_norm": 2.4629805088043213,
      "learning_rate": 5.436883264070219e-06,
      "loss": 0.9496,
      "step": 12077
    },
    {
      "epoch": 0.9760197175700519,
      "grad_norm": 2.4428911209106445,
      "learning_rate": 5.436231400895255e-06,
      "loss": 0.9442,
      "step": 12078
    },
    {
      "epoch": 0.9761005272833795,
      "grad_norm": 2.0524792671203613,
      "learning_rate": 5.435579530248698e-06,
      "loss": 1.0176,
      "step": 12079
    },
    {
      "epoch": 0.976181336996707,
      "grad_norm": 2.963853597640991,
      "learning_rate": 5.434927652141711e-06,
      "loss": 0.9538,
      "step": 12080
    },
    {
      "epoch": 0.9762621467100345,
      "grad_norm": 2.7061660289764404,
      "learning_rate": 5.434275766585462e-06,
      "loss": 0.9298,
      "step": 12081
    },
    {
      "epoch": 0.9763429564233621,
      "grad_norm": 2.973245620727539,
      "learning_rate": 5.433623873591114e-06,
      "loss": 0.9745,
      "step": 12082
    },
    {
      "epoch": 0.9764237661366896,
      "grad_norm": 2.5026097297668457,
      "learning_rate": 5.432971973169834e-06,
      "loss": 0.8883,
      "step": 12083
    },
    {
      "epoch": 0.9765045758500172,
      "grad_norm": 2.575617790222168,
      "learning_rate": 5.432320065332785e-06,
      "loss": 0.85,
      "step": 12084
    },
    {
      "epoch": 0.9765853855633447,
      "grad_norm": 3.1300740242004395,
      "learning_rate": 5.431668150091135e-06,
      "loss": 0.9331,
      "step": 12085
    },
    {
      "epoch": 0.9766661952766723,
      "grad_norm": 2.5608370304107666,
      "learning_rate": 5.431016227456047e-06,
      "loss": 0.865,
      "step": 12086
    },
    {
      "epoch": 0.9767470049899998,
      "grad_norm": 2.486452102661133,
      "learning_rate": 5.43036429743869e-06,
      "loss": 1.0273,
      "step": 12087
    },
    {
      "epoch": 0.9768278147033274,
      "grad_norm": 2.638866662979126,
      "learning_rate": 5.429712360050229e-06,
      "loss": 0.9897,
      "step": 12088
    },
    {
      "epoch": 0.9769086244166549,
      "grad_norm": 2.524085760116577,
      "learning_rate": 5.429060415301829e-06,
      "loss": 0.9344,
      "step": 12089
    },
    {
      "epoch": 0.9769894341299824,
      "grad_norm": 2.0070433616638184,
      "learning_rate": 5.428408463204657e-06,
      "loss": 1.0262,
      "step": 12090
    },
    {
      "epoch": 0.97707024384331,
      "grad_norm": 2.591197967529297,
      "learning_rate": 5.427756503769881e-06,
      "loss": 0.8865,
      "step": 12091
    },
    {
      "epoch": 0.9771510535566376,
      "grad_norm": 2.7331416606903076,
      "learning_rate": 5.4271045370086625e-06,
      "loss": 0.9492,
      "step": 12092
    },
    {
      "epoch": 0.977231863269965,
      "grad_norm": 2.887516498565674,
      "learning_rate": 5.426452562932175e-06,
      "loss": 0.9706,
      "step": 12093
    },
    {
      "epoch": 0.9773126729832926,
      "grad_norm": 2.3011393547058105,
      "learning_rate": 5.42580058155158e-06,
      "loss": 0.8829,
      "step": 12094
    },
    {
      "epoch": 0.9773934826966202,
      "grad_norm": 2.5624382495880127,
      "learning_rate": 5.425148592878047e-06,
      "loss": 0.9892,
      "step": 12095
    },
    {
      "epoch": 0.9774742924099477,
      "grad_norm": 2.3409554958343506,
      "learning_rate": 5.424496596922742e-06,
      "loss": 0.9166,
      "step": 12096
    },
    {
      "epoch": 0.9775551021232752,
      "grad_norm": 2.6583638191223145,
      "learning_rate": 5.423844593696831e-06,
      "loss": 0.8867,
      "step": 12097
    },
    {
      "epoch": 0.9776359118366028,
      "grad_norm": 2.8415184020996094,
      "learning_rate": 5.423192583211483e-06,
      "loss": 0.9739,
      "step": 12098
    },
    {
      "epoch": 0.9777167215499303,
      "grad_norm": 2.552694320678711,
      "learning_rate": 5.422540565477865e-06,
      "loss": 0.9481,
      "step": 12099
    },
    {
      "epoch": 0.9777975312632579,
      "grad_norm": 3.0070302486419678,
      "learning_rate": 5.421888540507144e-06,
      "loss": 0.9894,
      "step": 12100
    },
    {
      "epoch": 0.9778783409765854,
      "grad_norm": 2.724087953567505,
      "learning_rate": 5.421236508310489e-06,
      "loss": 0.8188,
      "step": 12101
    },
    {
      "epoch": 0.9779591506899129,
      "grad_norm": 2.6836183071136475,
      "learning_rate": 5.420584468899066e-06,
      "loss": 0.8178,
      "step": 12102
    },
    {
      "epoch": 0.9780399604032405,
      "grad_norm": 2.4756271839141846,
      "learning_rate": 5.419932422284044e-06,
      "loss": 0.9692,
      "step": 12103
    },
    {
      "epoch": 0.9781207701165681,
      "grad_norm": 2.5955512523651123,
      "learning_rate": 5.41928036847659e-06,
      "loss": 0.8955,
      "step": 12104
    },
    {
      "epoch": 0.9782015798298955,
      "grad_norm": 2.2906126976013184,
      "learning_rate": 5.418628307487872e-06,
      "loss": 0.9172,
      "step": 12105
    },
    {
      "epoch": 0.9782823895432231,
      "grad_norm": 2.5296623706817627,
      "learning_rate": 5.41797623932906e-06,
      "loss": 0.9272,
      "step": 12106
    },
    {
      "epoch": 0.9783631992565507,
      "grad_norm": 2.603579044342041,
      "learning_rate": 5.41732416401132e-06,
      "loss": 0.9143,
      "step": 12107
    },
    {
      "epoch": 0.9784440089698782,
      "grad_norm": 2.800060272216797,
      "learning_rate": 5.416672081545824e-06,
      "loss": 0.9438,
      "step": 12108
    },
    {
      "epoch": 0.9785248186832057,
      "grad_norm": 2.8698110580444336,
      "learning_rate": 5.4160199919437375e-06,
      "loss": 0.8999,
      "step": 12109
    },
    {
      "epoch": 0.9786056283965333,
      "grad_norm": 2.505112409591675,
      "learning_rate": 5.415367895216229e-06,
      "loss": 0.94,
      "step": 12110
    },
    {
      "epoch": 0.9786864381098608,
      "grad_norm": 3.3058252334594727,
      "learning_rate": 5.414715791374469e-06,
      "loss": 0.936,
      "step": 12111
    },
    {
      "epoch": 0.9787672478231884,
      "grad_norm": 2.4477574825286865,
      "learning_rate": 5.414063680429625e-06,
      "loss": 0.9525,
      "step": 12112
    },
    {
      "epoch": 0.9788480575365159,
      "grad_norm": 2.613408327102661,
      "learning_rate": 5.413411562392868e-06,
      "loss": 0.9188,
      "step": 12113
    },
    {
      "epoch": 0.9789288672498434,
      "grad_norm": 2.8560428619384766,
      "learning_rate": 5.412759437275366e-06,
      "loss": 0.9135,
      "step": 12114
    },
    {
      "epoch": 0.979009676963171,
      "grad_norm": 2.201775312423706,
      "learning_rate": 5.412107305088289e-06,
      "loss": 0.9843,
      "step": 12115
    },
    {
      "epoch": 0.9790904866764986,
      "grad_norm": 2.7511961460113525,
      "learning_rate": 5.411455165842806e-06,
      "loss": 0.9251,
      "step": 12116
    },
    {
      "epoch": 0.979171296389826,
      "grad_norm": 2.503464698791504,
      "learning_rate": 5.410803019550086e-06,
      "loss": 0.9065,
      "step": 12117
    },
    {
      "epoch": 0.9792521061031536,
      "grad_norm": 2.3945083618164062,
      "learning_rate": 5.4101508662213e-06,
      "loss": 0.9408,
      "step": 12118
    },
    {
      "epoch": 0.9793329158164812,
      "grad_norm": 2.640806198120117,
      "learning_rate": 5.409498705867616e-06,
      "loss": 0.8195,
      "step": 12119
    },
    {
      "epoch": 0.9794137255298087,
      "grad_norm": 2.9981961250305176,
      "learning_rate": 5.408846538500205e-06,
      "loss": 0.9757,
      "step": 12120
    },
    {
      "epoch": 0.9794945352431362,
      "grad_norm": 2.7424895763397217,
      "learning_rate": 5.408194364130238e-06,
      "loss": 0.8409,
      "step": 12121
    },
    {
      "epoch": 0.9795753449564638,
      "grad_norm": 2.4994630813598633,
      "learning_rate": 5.407542182768884e-06,
      "loss": 0.9793,
      "step": 12122
    },
    {
      "epoch": 0.9796561546697913,
      "grad_norm": 2.414841890335083,
      "learning_rate": 5.406889994427313e-06,
      "loss": 0.8574,
      "step": 12123
    },
    {
      "epoch": 0.9797369643831189,
      "grad_norm": 2.671175479888916,
      "learning_rate": 5.406237799116696e-06,
      "loss": 0.8671,
      "step": 12124
    },
    {
      "epoch": 0.9798177740964464,
      "grad_norm": 2.3028671741485596,
      "learning_rate": 5.405585596848204e-06,
      "loss": 0.8778,
      "step": 12125
    },
    {
      "epoch": 0.9798985838097739,
      "grad_norm": 2.698812961578369,
      "learning_rate": 5.404933387633007e-06,
      "loss": 0.8468,
      "step": 12126
    },
    {
      "epoch": 0.9799793935231015,
      "grad_norm": 2.5293102264404297,
      "learning_rate": 5.404281171482275e-06,
      "loss": 0.8957,
      "step": 12127
    },
    {
      "epoch": 0.9800602032364291,
      "grad_norm": 2.2571027278900146,
      "learning_rate": 5.40362894840718e-06,
      "loss": 0.8962,
      "step": 12128
    },
    {
      "epoch": 0.9801410129497565,
      "grad_norm": 2.41410493850708,
      "learning_rate": 5.402976718418893e-06,
      "loss": 1.0511,
      "step": 12129
    },
    {
      "epoch": 0.9802218226630841,
      "grad_norm": 2.474343776702881,
      "learning_rate": 5.402324481528587e-06,
      "loss": 0.9717,
      "step": 12130
    },
    {
      "epoch": 0.9803026323764117,
      "grad_norm": 2.4942386150360107,
      "learning_rate": 5.401672237747428e-06,
      "loss": 0.9959,
      "step": 12131
    },
    {
      "epoch": 0.9803834420897392,
      "grad_norm": 2.3961400985717773,
      "learning_rate": 5.401019987086593e-06,
      "loss": 0.9601,
      "step": 12132
    },
    {
      "epoch": 0.9804642518030667,
      "grad_norm": 2.85259747505188,
      "learning_rate": 5.4003677295572496e-06,
      "loss": 0.871,
      "step": 12133
    },
    {
      "epoch": 0.9805450615163943,
      "grad_norm": 2.4561750888824463,
      "learning_rate": 5.399715465170571e-06,
      "loss": 0.9385,
      "step": 12134
    },
    {
      "epoch": 0.9806258712297218,
      "grad_norm": 2.253120183944702,
      "learning_rate": 5.399063193937729e-06,
      "loss": 0.9056,
      "step": 12135
    },
    {
      "epoch": 0.9807066809430494,
      "grad_norm": 2.6282289028167725,
      "learning_rate": 5.398410915869896e-06,
      "loss": 0.9145,
      "step": 12136
    },
    {
      "epoch": 0.9807874906563769,
      "grad_norm": 2.1766889095306396,
      "learning_rate": 5.397758630978241e-06,
      "loss": 0.9398,
      "step": 12137
    },
    {
      "epoch": 0.9808683003697044,
      "grad_norm": 2.755300760269165,
      "learning_rate": 5.397106339273941e-06,
      "loss": 0.9192,
      "step": 12138
    },
    {
      "epoch": 0.980949110083032,
      "grad_norm": 2.4856979846954346,
      "learning_rate": 5.396454040768164e-06,
      "loss": 0.8801,
      "step": 12139
    },
    {
      "epoch": 0.9810299197963596,
      "grad_norm": 2.632225513458252,
      "learning_rate": 5.395801735472084e-06,
      "loss": 0.9607,
      "step": 12140
    },
    {
      "epoch": 0.981110729509687,
      "grad_norm": 3.21994686126709,
      "learning_rate": 5.395149423396872e-06,
      "loss": 0.9186,
      "step": 12141
    },
    {
      "epoch": 0.9811915392230146,
      "grad_norm": 2.641205072402954,
      "learning_rate": 5.394497104553702e-06,
      "loss": 0.8499,
      "step": 12142
    },
    {
      "epoch": 0.9812723489363422,
      "grad_norm": 2.4946446418762207,
      "learning_rate": 5.393844778953748e-06,
      "loss": 1.0907,
      "step": 12143
    },
    {
      "epoch": 0.9813531586496697,
      "grad_norm": 2.7536115646362305,
      "learning_rate": 5.39319244660818e-06,
      "loss": 0.9752,
      "step": 12144
    },
    {
      "epoch": 0.9814339683629972,
      "grad_norm": 2.757713794708252,
      "learning_rate": 5.39254010752817e-06,
      "loss": 0.8335,
      "step": 12145
    },
    {
      "epoch": 0.9815147780763248,
      "grad_norm": 2.7080740928649902,
      "learning_rate": 5.391887761724897e-06,
      "loss": 0.9056,
      "step": 12146
    },
    {
      "epoch": 0.9815955877896523,
      "grad_norm": 2.723893165588379,
      "learning_rate": 5.3912354092095265e-06,
      "loss": 0.9127,
      "step": 12147
    },
    {
      "epoch": 0.9816763975029799,
      "grad_norm": 2.4144506454467773,
      "learning_rate": 5.390583049993236e-06,
      "loss": 0.9482,
      "step": 12148
    },
    {
      "epoch": 0.9817572072163074,
      "grad_norm": 2.5960702896118164,
      "learning_rate": 5.389930684087199e-06,
      "loss": 0.8589,
      "step": 12149
    },
    {
      "epoch": 0.9818380169296349,
      "grad_norm": 2.481485605239868,
      "learning_rate": 5.3892783115025895e-06,
      "loss": 0.8986,
      "step": 12150
    },
    {
      "epoch": 0.9819188266429625,
      "grad_norm": 2.679650068283081,
      "learning_rate": 5.388625932250576e-06,
      "loss": 0.8275,
      "step": 12151
    },
    {
      "epoch": 0.9819996363562901,
      "grad_norm": 2.44803786277771,
      "learning_rate": 5.38797354634234e-06,
      "loss": 0.9115,
      "step": 12152
    },
    {
      "epoch": 0.9820804460696175,
      "grad_norm": 2.4601192474365234,
      "learning_rate": 5.387321153789047e-06,
      "loss": 0.8299,
      "step": 12153
    },
    {
      "epoch": 0.9821612557829451,
      "grad_norm": 2.5130109786987305,
      "learning_rate": 5.386668754601878e-06,
      "loss": 0.9465,
      "step": 12154
    },
    {
      "epoch": 0.9822420654962727,
      "grad_norm": 2.3025498390197754,
      "learning_rate": 5.386016348792004e-06,
      "loss": 0.9691,
      "step": 12155
    },
    {
      "epoch": 0.9823228752096002,
      "grad_norm": 2.6452362537384033,
      "learning_rate": 5.385363936370598e-06,
      "loss": 0.9182,
      "step": 12156
    },
    {
      "epoch": 0.9824036849229277,
      "grad_norm": 3.0965425968170166,
      "learning_rate": 5.384711517348837e-06,
      "loss": 0.996,
      "step": 12157
    },
    {
      "epoch": 0.9824844946362553,
      "grad_norm": 2.5991649627685547,
      "learning_rate": 5.384059091737892e-06,
      "loss": 0.8321,
      "step": 12158
    },
    {
      "epoch": 0.9825653043495828,
      "grad_norm": 2.765760898590088,
      "learning_rate": 5.38340665954894e-06,
      "loss": 0.9406,
      "step": 12159
    },
    {
      "epoch": 0.9826461140629104,
      "grad_norm": 2.2090837955474854,
      "learning_rate": 5.382754220793156e-06,
      "loss": 0.9335,
      "step": 12160
    },
    {
      "epoch": 0.9827269237762379,
      "grad_norm": 2.5422046184539795,
      "learning_rate": 5.382101775481712e-06,
      "loss": 1.0029,
      "step": 12161
    },
    {
      "epoch": 0.9828077334895654,
      "grad_norm": 2.9563913345336914,
      "learning_rate": 5.3814493236257855e-06,
      "loss": 0.9966,
      "step": 12162
    },
    {
      "epoch": 0.982888543202893,
      "grad_norm": 2.5533783435821533,
      "learning_rate": 5.380796865236549e-06,
      "loss": 0.9375,
      "step": 12163
    },
    {
      "epoch": 0.9829693529162206,
      "grad_norm": 2.479666233062744,
      "learning_rate": 5.3801444003251815e-06,
      "loss": 1.0035,
      "step": 12164
    },
    {
      "epoch": 0.983050162629548,
      "grad_norm": 2.4869494438171387,
      "learning_rate": 5.3794919289028535e-06,
      "loss": 0.9296,
      "step": 12165
    },
    {
      "epoch": 0.9831309723428756,
      "grad_norm": 2.7353363037109375,
      "learning_rate": 5.378839450980744e-06,
      "loss": 0.9772,
      "step": 12166
    },
    {
      "epoch": 0.9832117820562032,
      "grad_norm": 2.5732433795928955,
      "learning_rate": 5.3781869665700235e-06,
      "loss": 0.9728,
      "step": 12167
    },
    {
      "epoch": 0.9832925917695307,
      "grad_norm": 2.4385268688201904,
      "learning_rate": 5.377534475681875e-06,
      "loss": 0.7798,
      "step": 12168
    },
    {
      "epoch": 0.9833734014828582,
      "grad_norm": 2.479956865310669,
      "learning_rate": 5.376881978327467e-06,
      "loss": 0.9799,
      "step": 12169
    },
    {
      "epoch": 0.9834542111961858,
      "grad_norm": 2.8344802856445312,
      "learning_rate": 5.376229474517979e-06,
      "loss": 0.9356,
      "step": 12170
    },
    {
      "epoch": 0.9835350209095133,
      "grad_norm": 2.779484987258911,
      "learning_rate": 5.375576964264585e-06,
      "loss": 0.9546,
      "step": 12171
    },
    {
      "epoch": 0.9836158306228409,
      "grad_norm": 2.6164891719818115,
      "learning_rate": 5.374924447578462e-06,
      "loss": 0.8925,
      "step": 12172
    },
    {
      "epoch": 0.9836966403361684,
      "grad_norm": 2.648625135421753,
      "learning_rate": 5.374271924470787e-06,
      "loss": 0.9378,
      "step": 12173
    },
    {
      "epoch": 0.9837774500494959,
      "grad_norm": 2.5899453163146973,
      "learning_rate": 5.373619394952734e-06,
      "loss": 0.9926,
      "step": 12174
    },
    {
      "epoch": 0.9838582597628235,
      "grad_norm": 2.894723653793335,
      "learning_rate": 5.372966859035481e-06,
      "loss": 0.9491,
      "step": 12175
    },
    {
      "epoch": 0.9839390694761511,
      "grad_norm": 2.368407726287842,
      "learning_rate": 5.372314316730203e-06,
      "loss": 0.8911,
      "step": 12176
    },
    {
      "epoch": 0.9840198791894785,
      "grad_norm": 2.3118772506713867,
      "learning_rate": 5.371661768048077e-06,
      "loss": 0.7905,
      "step": 12177
    },
    {
      "epoch": 0.9841006889028061,
      "grad_norm": 2.4478707313537598,
      "learning_rate": 5.371009213000279e-06,
      "loss": 0.9669,
      "step": 12178
    },
    {
      "epoch": 0.9841814986161337,
      "grad_norm": 2.403913736343384,
      "learning_rate": 5.3703566515979865e-06,
      "loss": 0.9257,
      "step": 12179
    },
    {
      "epoch": 0.9842623083294612,
      "grad_norm": 2.834071159362793,
      "learning_rate": 5.369704083852376e-06,
      "loss": 0.9239,
      "step": 12180
    },
    {
      "epoch": 0.9843431180427887,
      "grad_norm": 2.5608348846435547,
      "learning_rate": 5.369051509774625e-06,
      "loss": 0.8549,
      "step": 12181
    },
    {
      "epoch": 0.9844239277561163,
      "grad_norm": 2.570713996887207,
      "learning_rate": 5.368398929375911e-06,
      "loss": 0.9413,
      "step": 12182
    },
    {
      "epoch": 0.9845047374694438,
      "grad_norm": 2.7243850231170654,
      "learning_rate": 5.36774634266741e-06,
      "loss": 1.0564,
      "step": 12183
    },
    {
      "epoch": 0.9845855471827714,
      "grad_norm": 2.613048553466797,
      "learning_rate": 5.367093749660299e-06,
      "loss": 0.9697,
      "step": 12184
    },
    {
      "epoch": 0.9846663568960989,
      "grad_norm": 2.798172950744629,
      "learning_rate": 5.366441150365755e-06,
      "loss": 0.8594,
      "step": 12185
    },
    {
      "epoch": 0.9847471666094264,
      "grad_norm": 2.6238584518432617,
      "learning_rate": 5.365788544794958e-06,
      "loss": 0.9041,
      "step": 12186
    },
    {
      "epoch": 0.984827976322754,
      "grad_norm": 2.3103373050689697,
      "learning_rate": 5.365135932959083e-06,
      "loss": 0.9151,
      "step": 12187
    },
    {
      "epoch": 0.9849087860360816,
      "grad_norm": 2.1761138439178467,
      "learning_rate": 5.364483314869308e-06,
      "loss": 0.9919,
      "step": 12188
    },
    {
      "epoch": 0.984989595749409,
      "grad_norm": 2.738905906677246,
      "learning_rate": 5.363830690536812e-06,
      "loss": 0.9131,
      "step": 12189
    },
    {
      "epoch": 0.9850704054627366,
      "grad_norm": 2.7202906608581543,
      "learning_rate": 5.3631780599727715e-06,
      "loss": 0.8734,
      "step": 12190
    },
    {
      "epoch": 0.9851512151760642,
      "grad_norm": 2.608964204788208,
      "learning_rate": 5.362525423188366e-06,
      "loss": 1.0752,
      "step": 12191
    },
    {
      "epoch": 0.9852320248893917,
      "grad_norm": 2.9981770515441895,
      "learning_rate": 5.361872780194772e-06,
      "loss": 0.7842,
      "step": 12192
    },
    {
      "epoch": 0.9853128346027192,
      "grad_norm": 2.777306318283081,
      "learning_rate": 5.361220131003169e-06,
      "loss": 0.8846,
      "step": 12193
    },
    {
      "epoch": 0.9853936443160468,
      "grad_norm": 3.3271985054016113,
      "learning_rate": 5.360567475624734e-06,
      "loss": 0.7995,
      "step": 12194
    },
    {
      "epoch": 0.9854744540293743,
      "grad_norm": 2.412200450897217,
      "learning_rate": 5.359914814070646e-06,
      "loss": 0.821,
      "step": 12195
    },
    {
      "epoch": 0.9855552637427019,
      "grad_norm": 2.6865506172180176,
      "learning_rate": 5.359262146352085e-06,
      "loss": 1.0232,
      "step": 12196
    },
    {
      "epoch": 0.9856360734560294,
      "grad_norm": 2.4174892902374268,
      "learning_rate": 5.358609472480227e-06,
      "loss": 0.8605,
      "step": 12197
    },
    {
      "epoch": 0.9857168831693569,
      "grad_norm": 2.217020034790039,
      "learning_rate": 5.357956792466252e-06,
      "loss": 0.9156,
      "step": 12198
    },
    {
      "epoch": 0.9857976928826845,
      "grad_norm": 2.7196741104125977,
      "learning_rate": 5.35730410632134e-06,
      "loss": 0.762,
      "step": 12199
    },
    {
      "epoch": 0.9858785025960121,
      "grad_norm": 2.831913471221924,
      "learning_rate": 5.356651414056669e-06,
      "loss": 0.9218,
      "step": 12200
    },
    {
      "epoch": 0.9859593123093395,
      "grad_norm": 2.774486780166626,
      "learning_rate": 5.355998715683417e-06,
      "loss": 1.0029,
      "step": 12201
    },
    {
      "epoch": 0.9860401220226671,
      "grad_norm": 2.3571219444274902,
      "learning_rate": 5.355346011212764e-06,
      "loss": 1.0688,
      "step": 12202
    },
    {
      "epoch": 0.9861209317359947,
      "grad_norm": 2.823669910430908,
      "learning_rate": 5.354693300655891e-06,
      "loss": 0.9271,
      "step": 12203
    },
    {
      "epoch": 0.9862017414493222,
      "grad_norm": 2.444176197052002,
      "learning_rate": 5.3540405840239725e-06,
      "loss": 0.9965,
      "step": 12204
    },
    {
      "epoch": 0.9862825511626497,
      "grad_norm": 2.6155660152435303,
      "learning_rate": 5.353387861328194e-06,
      "loss": 0.9131,
      "step": 12205
    },
    {
      "epoch": 0.9863633608759773,
      "grad_norm": 2.318523645401001,
      "learning_rate": 5.352735132579732e-06,
      "loss": 1.0011,
      "step": 12206
    },
    {
      "epoch": 0.9864441705893048,
      "grad_norm": 2.41597843170166,
      "learning_rate": 5.352082397789764e-06,
      "loss": 0.934,
      "step": 12207
    },
    {
      "epoch": 0.9865249803026324,
      "grad_norm": 2.3817427158355713,
      "learning_rate": 5.351429656969473e-06,
      "loss": 0.9334,
      "step": 12208
    },
    {
      "epoch": 0.9866057900159599,
      "grad_norm": 3.1035239696502686,
      "learning_rate": 5.350776910130039e-06,
      "loss": 0.9043,
      "step": 12209
    },
    {
      "epoch": 0.9866865997292874,
      "grad_norm": 2.416182518005371,
      "learning_rate": 5.35012415728264e-06,
      "loss": 0.8499,
      "step": 12210
    },
    {
      "epoch": 0.986767409442615,
      "grad_norm": 2.56245493888855,
      "learning_rate": 5.34947139843846e-06,
      "loss": 0.8969,
      "step": 12211
    },
    {
      "epoch": 0.9868482191559426,
      "grad_norm": 2.5546376705169678,
      "learning_rate": 5.348818633608671e-06,
      "loss": 0.9497,
      "step": 12212
    },
    {
      "epoch": 0.98692902886927,
      "grad_norm": 2.3924479484558105,
      "learning_rate": 5.348165862804463e-06,
      "loss": 0.9101,
      "step": 12213
    },
    {
      "epoch": 0.9870098385825976,
      "grad_norm": 2.4456634521484375,
      "learning_rate": 5.3475130860370106e-06,
      "loss": 0.8853,
      "step": 12214
    },
    {
      "epoch": 0.9870906482959252,
      "grad_norm": 2.8276612758636475,
      "learning_rate": 5.346860303317495e-06,
      "loss": 0.8544,
      "step": 12215
    },
    {
      "epoch": 0.9871714580092528,
      "grad_norm": 2.8824312686920166,
      "learning_rate": 5.346207514657098e-06,
      "loss": 0.8829,
      "step": 12216
    },
    {
      "epoch": 0.9872522677225802,
      "grad_norm": 2.500629425048828,
      "learning_rate": 5.345554720067e-06,
      "loss": 1.0615,
      "step": 12217
    },
    {
      "epoch": 0.9873330774359078,
      "grad_norm": 2.476496696472168,
      "learning_rate": 5.3449019195583795e-06,
      "loss": 0.8171,
      "step": 12218
    },
    {
      "epoch": 0.9874138871492354,
      "grad_norm": 2.551996946334839,
      "learning_rate": 5.344249113142422e-06,
      "loss": 0.848,
      "step": 12219
    },
    {
      "epoch": 0.9874946968625629,
      "grad_norm": 2.895494222640991,
      "learning_rate": 5.3435963008303046e-06,
      "loss": 0.9491,
      "step": 12220
    },
    {
      "epoch": 0.9875755065758904,
      "grad_norm": 2.8388452529907227,
      "learning_rate": 5.342943482633211e-06,
      "loss": 0.8374,
      "step": 12221
    },
    {
      "epoch": 0.987656316289218,
      "grad_norm": 2.273324966430664,
      "learning_rate": 5.34229065856232e-06,
      "loss": 0.9174,
      "step": 12222
    },
    {
      "epoch": 0.9877371260025455,
      "grad_norm": 2.447150945663452,
      "learning_rate": 5.341637828628814e-06,
      "loss": 0.9445,
      "step": 12223
    },
    {
      "epoch": 0.9878179357158731,
      "grad_norm": 2.7661643028259277,
      "learning_rate": 5.340984992843874e-06,
      "loss": 0.9384,
      "step": 12224
    },
    {
      "epoch": 0.9878987454292006,
      "grad_norm": 2.543273448944092,
      "learning_rate": 5.340332151218684e-06,
      "loss": 0.837,
      "step": 12225
    },
    {
      "epoch": 0.9879795551425281,
      "grad_norm": 2.4272313117980957,
      "learning_rate": 5.339679303764421e-06,
      "loss": 0.9039,
      "step": 12226
    },
    {
      "epoch": 0.9880603648558557,
      "grad_norm": 2.3481764793395996,
      "learning_rate": 5.339026450492272e-06,
      "loss": 0.9525,
      "step": 12227
    },
    {
      "epoch": 0.9881411745691833,
      "grad_norm": 2.6108486652374268,
      "learning_rate": 5.338373591413414e-06,
      "loss": 0.9075,
      "step": 12228
    },
    {
      "epoch": 0.9882219842825107,
      "grad_norm": 3.0879838466644287,
      "learning_rate": 5.337720726539032e-06,
      "loss": 0.8332,
      "step": 12229
    },
    {
      "epoch": 0.9883027939958383,
      "grad_norm": 2.506908416748047,
      "learning_rate": 5.337067855880305e-06,
      "loss": 0.965,
      "step": 12230
    },
    {
      "epoch": 0.9883836037091659,
      "grad_norm": 2.6116840839385986,
      "learning_rate": 5.33641497944842e-06,
      "loss": 0.885,
      "step": 12231
    },
    {
      "epoch": 0.9884644134224934,
      "grad_norm": 2.8873798847198486,
      "learning_rate": 5.335762097254554e-06,
      "loss": 0.853,
      "step": 12232
    },
    {
      "epoch": 0.9885452231358209,
      "grad_norm": 2.668135166168213,
      "learning_rate": 5.3351092093098944e-06,
      "loss": 0.8436,
      "step": 12233
    },
    {
      "epoch": 0.9886260328491485,
      "grad_norm": 3.144533634185791,
      "learning_rate": 5.334456315625618e-06,
      "loss": 0.9255,
      "step": 12234
    },
    {
      "epoch": 0.988706842562476,
      "grad_norm": 2.8562543392181396,
      "learning_rate": 5.333803416212911e-06,
      "loss": 0.985,
      "step": 12235
    },
    {
      "epoch": 0.9887876522758036,
      "grad_norm": 2.622060537338257,
      "learning_rate": 5.333150511082955e-06,
      "loss": 0.9229,
      "step": 12236
    },
    {
      "epoch": 0.9888684619891311,
      "grad_norm": 2.6421568393707275,
      "learning_rate": 5.332497600246933e-06,
      "loss": 0.7765,
      "step": 12237
    },
    {
      "epoch": 0.9889492717024586,
      "grad_norm": 2.7658445835113525,
      "learning_rate": 5.331844683716027e-06,
      "loss": 0.8947,
      "step": 12238
    },
    {
      "epoch": 0.9890300814157862,
      "grad_norm": 2.8542468547821045,
      "learning_rate": 5.331191761501421e-06,
      "loss": 0.8648,
      "step": 12239
    },
    {
      "epoch": 0.9891108911291138,
      "grad_norm": 2.5092406272888184,
      "learning_rate": 5.330538833614297e-06,
      "loss": 1.0049,
      "step": 12240
    },
    {
      "epoch": 0.9891917008424412,
      "grad_norm": 2.5922608375549316,
      "learning_rate": 5.3298859000658395e-06,
      "loss": 1.0946,
      "step": 12241
    },
    {
      "epoch": 0.9892725105557688,
      "grad_norm": 2.412569046020508,
      "learning_rate": 5.329232960867231e-06,
      "loss": 1.0575,
      "step": 12242
    },
    {
      "epoch": 0.9893533202690964,
      "grad_norm": 2.937439203262329,
      "learning_rate": 5.328580016029653e-06,
      "loss": 0.9753,
      "step": 12243
    },
    {
      "epoch": 0.9894341299824239,
      "grad_norm": 2.7120580673217773,
      "learning_rate": 5.327927065564291e-06,
      "loss": 0.9585,
      "step": 12244
    },
    {
      "epoch": 0.9895149396957514,
      "grad_norm": 2.4741392135620117,
      "learning_rate": 5.3272741094823275e-06,
      "loss": 0.9583,
      "step": 12245
    },
    {
      "epoch": 0.989595749409079,
      "grad_norm": 2.476374387741089,
      "learning_rate": 5.326621147794946e-06,
      "loss": 0.9511,
      "step": 12246
    },
    {
      "epoch": 0.9896765591224065,
      "grad_norm": 2.838733196258545,
      "learning_rate": 5.325968180513331e-06,
      "loss": 0.8448,
      "step": 12247
    },
    {
      "epoch": 0.9897573688357341,
      "grad_norm": 2.329197883605957,
      "learning_rate": 5.325315207648667e-06,
      "loss": 0.9221,
      "step": 12248
    },
    {
      "epoch": 0.9898381785490616,
      "grad_norm": 2.675673246383667,
      "learning_rate": 5.3246622292121344e-06,
      "loss": 0.8106,
      "step": 12249
    },
    {
      "epoch": 0.9899189882623891,
      "grad_norm": 2.3492445945739746,
      "learning_rate": 5.324009245214922e-06,
      "loss": 0.9264,
      "step": 12250
    },
    {
      "epoch": 0.9899997979757167,
      "grad_norm": 2.4474477767944336,
      "learning_rate": 5.32335625566821e-06,
      "loss": 0.9301,
      "step": 12251
    },
    {
      "epoch": 0.9900806076890443,
      "grad_norm": 2.737623691558838,
      "learning_rate": 5.322703260583183e-06,
      "loss": 0.8766,
      "step": 12252
    },
    {
      "epoch": 0.9901614174023717,
      "grad_norm": 2.7616465091705322,
      "learning_rate": 5.322050259971027e-06,
      "loss": 0.9792,
      "step": 12253
    },
    {
      "epoch": 0.9902422271156993,
      "grad_norm": 3.4910755157470703,
      "learning_rate": 5.321397253842924e-06,
      "loss": 0.9654,
      "step": 12254
    },
    {
      "epoch": 0.9903230368290269,
      "grad_norm": 2.699347734451294,
      "learning_rate": 5.320744242210061e-06,
      "loss": 0.9645,
      "step": 12255
    },
    {
      "epoch": 0.9904038465423544,
      "grad_norm": 2.415865182876587,
      "learning_rate": 5.320091225083622e-06,
      "loss": 0.9715,
      "step": 12256
    },
    {
      "epoch": 0.9904846562556819,
      "grad_norm": 2.8747668266296387,
      "learning_rate": 5.319438202474788e-06,
      "loss": 0.9351,
      "step": 12257
    },
    {
      "epoch": 0.9905654659690095,
      "grad_norm": 2.717130184173584,
      "learning_rate": 5.318785174394751e-06,
      "loss": 0.8969,
      "step": 12258
    },
    {
      "epoch": 0.990646275682337,
      "grad_norm": 3.111077308654785,
      "learning_rate": 5.3181321408546885e-06,
      "loss": 0.8883,
      "step": 12259
    },
    {
      "epoch": 0.9907270853956646,
      "grad_norm": 3.141089677810669,
      "learning_rate": 5.317479101865788e-06,
      "loss": 1.0838,
      "step": 12260
    },
    {
      "epoch": 0.9908078951089921,
      "grad_norm": 2.755082607269287,
      "learning_rate": 5.316826057439236e-06,
      "loss": 0.9619,
      "step": 12261
    },
    {
      "epoch": 0.9908887048223196,
      "grad_norm": 2.669637441635132,
      "learning_rate": 5.316173007586215e-06,
      "loss": 1.0097,
      "step": 12262
    },
    {
      "epoch": 0.9909695145356472,
      "grad_norm": 2.529109477996826,
      "learning_rate": 5.315519952317912e-06,
      "loss": 0.9657,
      "step": 12263
    },
    {
      "epoch": 0.9910503242489748,
      "grad_norm": 2.3322739601135254,
      "learning_rate": 5.314866891645514e-06,
      "loss": 0.885,
      "step": 12264
    },
    {
      "epoch": 0.9911311339623022,
      "grad_norm": 2.8430070877075195,
      "learning_rate": 5.314213825580201e-06,
      "loss": 0.892,
      "step": 12265
    },
    {
      "epoch": 0.9912119436756298,
      "grad_norm": 2.8258538246154785,
      "learning_rate": 5.3135607541331646e-06,
      "loss": 0.9957,
      "step": 12266
    },
    {
      "epoch": 0.9912927533889574,
      "grad_norm": 2.687260627746582,
      "learning_rate": 5.312907677315585e-06,
      "loss": 0.9395,
      "step": 12267
    },
    {
      "epoch": 0.9913735631022849,
      "grad_norm": 2.6090567111968994,
      "learning_rate": 5.3122545951386505e-06,
      "loss": 0.9182,
      "step": 12268
    },
    {
      "epoch": 0.9914543728156124,
      "grad_norm": 2.947476387023926,
      "learning_rate": 5.311601507613547e-06,
      "loss": 1.029,
      "step": 12269
    },
    {
      "epoch": 0.99153518252894,
      "grad_norm": 2.5828793048858643,
      "learning_rate": 5.310948414751461e-06,
      "loss": 0.959,
      "step": 12270
    },
    {
      "epoch": 0.9916159922422675,
      "grad_norm": 2.347775459289551,
      "learning_rate": 5.310295316563575e-06,
      "loss": 0.9352,
      "step": 12271
    },
    {
      "epoch": 0.9916968019555951,
      "grad_norm": 2.6231443881988525,
      "learning_rate": 5.309642213061079e-06,
      "loss": 0.8537,
      "step": 12272
    },
    {
      "epoch": 0.9917776116689226,
      "grad_norm": 2.8878562450408936,
      "learning_rate": 5.308989104255157e-06,
      "loss": 0.9144,
      "step": 12273
    },
    {
      "epoch": 0.9918584213822501,
      "grad_norm": 2.5320656299591064,
      "learning_rate": 5.308335990156994e-06,
      "loss": 1.0086,
      "step": 12274
    },
    {
      "epoch": 0.9919392310955777,
      "grad_norm": 2.5271451473236084,
      "learning_rate": 5.3076828707777795e-06,
      "loss": 0.8926,
      "step": 12275
    },
    {
      "epoch": 0.9920200408089053,
      "grad_norm": 2.6236629486083984,
      "learning_rate": 5.307029746128697e-06,
      "loss": 0.8536,
      "step": 12276
    },
    {
      "epoch": 0.9921008505222327,
      "grad_norm": 2.6657965183258057,
      "learning_rate": 5.3063766162209354e-06,
      "loss": 0.8521,
      "step": 12277
    },
    {
      "epoch": 0.9921816602355603,
      "grad_norm": 2.7572550773620605,
      "learning_rate": 5.305723481065679e-06,
      "loss": 0.725,
      "step": 12278
    },
    {
      "epoch": 0.9922624699488879,
      "grad_norm": 2.477506160736084,
      "learning_rate": 5.305070340674114e-06,
      "loss": 0.9167,
      "step": 12279
    },
    {
      "epoch": 0.9923432796622154,
      "grad_norm": 2.4824118614196777,
      "learning_rate": 5.304417195057432e-06,
      "loss": 1.0134,
      "step": 12280
    },
    {
      "epoch": 0.9924240893755429,
      "grad_norm": 2.8695285320281982,
      "learning_rate": 5.303764044226814e-06,
      "loss": 0.9027,
      "step": 12281
    },
    {
      "epoch": 0.9925048990888705,
      "grad_norm": 2.9349610805511475,
      "learning_rate": 5.303110888193449e-06,
      "loss": 0.9602,
      "step": 12282
    },
    {
      "epoch": 0.992585708802198,
      "grad_norm": 3.3354389667510986,
      "learning_rate": 5.302457726968525e-06,
      "loss": 0.9154,
      "step": 12283
    },
    {
      "epoch": 0.9926665185155256,
      "grad_norm": 2.677093982696533,
      "learning_rate": 5.301804560563229e-06,
      "loss": 0.8953,
      "step": 12284
    },
    {
      "epoch": 0.9927473282288531,
      "grad_norm": 2.967055082321167,
      "learning_rate": 5.3011513889887445e-06,
      "loss": 0.9596,
      "step": 12285
    },
    {
      "epoch": 0.9928281379421806,
      "grad_norm": 2.8245794773101807,
      "learning_rate": 5.300498212256266e-06,
      "loss": 0.9689,
      "step": 12286
    },
    {
      "epoch": 0.9929089476555082,
      "grad_norm": 3.164083480834961,
      "learning_rate": 5.2998450303769734e-06,
      "loss": 0.9935,
      "step": 12287
    },
    {
      "epoch": 0.9929897573688358,
      "grad_norm": 2.2264647483825684,
      "learning_rate": 5.299191843362057e-06,
      "loss": 1.109,
      "step": 12288
    },
    {
      "epoch": 0.9930705670821632,
      "grad_norm": 2.7262375354766846,
      "learning_rate": 5.298538651222705e-06,
      "loss": 0.9294,
      "step": 12289
    },
    {
      "epoch": 0.9931513767954908,
      "grad_norm": 2.357172966003418,
      "learning_rate": 5.297885453970106e-06,
      "loss": 0.9459,
      "step": 12290
    },
    {
      "epoch": 0.9932321865088184,
      "grad_norm": 2.671567916870117,
      "learning_rate": 5.297232251615445e-06,
      "loss": 0.8271,
      "step": 12291
    },
    {
      "epoch": 0.9933129962221459,
      "grad_norm": 3.148672580718994,
      "learning_rate": 5.296579044169913e-06,
      "loss": 0.8556,
      "step": 12292
    },
    {
      "epoch": 0.9933938059354734,
      "grad_norm": 3.001763105392456,
      "learning_rate": 5.2959258316446935e-06,
      "loss": 0.9584,
      "step": 12293
    },
    {
      "epoch": 0.993474615648801,
      "grad_norm": 2.171727418899536,
      "learning_rate": 5.2952726140509794e-06,
      "loss": 0.8663,
      "step": 12294
    },
    {
      "epoch": 0.9935554253621285,
      "grad_norm": 2.8575220108032227,
      "learning_rate": 5.294619391399954e-06,
      "loss": 0.979,
      "step": 12295
    },
    {
      "epoch": 0.9936362350754561,
      "grad_norm": 2.449154853820801,
      "learning_rate": 5.2939661637028085e-06,
      "loss": 0.9035,
      "step": 12296
    },
    {
      "epoch": 0.9937170447887836,
      "grad_norm": 3.02980899810791,
      "learning_rate": 5.29331293097073e-06,
      "loss": 0.7961,
      "step": 12297
    },
    {
      "epoch": 0.9937978545021111,
      "grad_norm": 3.2935004234313965,
      "learning_rate": 5.292659693214908e-06,
      "loss": 0.9608,
      "step": 12298
    },
    {
      "epoch": 0.9938786642154387,
      "grad_norm": 2.5653650760650635,
      "learning_rate": 5.292006450446529e-06,
      "loss": 0.9354,
      "step": 12299
    },
    {
      "epoch": 0.9939594739287663,
      "grad_norm": 3.002340316772461,
      "learning_rate": 5.2913532026767845e-06,
      "loss": 0.9007,
      "step": 12300
    },
    {
      "epoch": 0.9940402836420937,
      "grad_norm": 2.487698793411255,
      "learning_rate": 5.290699949916859e-06,
      "loss": 0.943,
      "step": 12301
    },
    {
      "epoch": 0.9941210933554213,
      "grad_norm": 2.4758803844451904,
      "learning_rate": 5.2900466921779436e-06,
      "loss": 0.9927,
      "step": 12302
    },
    {
      "epoch": 0.9942019030687489,
      "grad_norm": 2.691206455230713,
      "learning_rate": 5.289393429471227e-06,
      "loss": 0.9391,
      "step": 12303
    },
    {
      "epoch": 0.9942827127820764,
      "grad_norm": 2.8046562671661377,
      "learning_rate": 5.288740161807897e-06,
      "loss": 0.9743,
      "step": 12304
    },
    {
      "epoch": 0.9943635224954039,
      "grad_norm": 2.924703598022461,
      "learning_rate": 5.288086889199143e-06,
      "loss": 0.9712,
      "step": 12305
    },
    {
      "epoch": 0.9944443322087315,
      "grad_norm": 2.697329521179199,
      "learning_rate": 5.2874336116561545e-06,
      "loss": 0.8344,
      "step": 12306
    },
    {
      "epoch": 0.994525141922059,
      "grad_norm": 2.530019760131836,
      "learning_rate": 5.28678032919012e-06,
      "loss": 0.8061,
      "step": 12307
    },
    {
      "epoch": 0.9946059516353866,
      "grad_norm": 2.3434691429138184,
      "learning_rate": 5.28612704181223e-06,
      "loss": 0.9062,
      "step": 12308
    },
    {
      "epoch": 0.9946867613487141,
      "grad_norm": 2.569121837615967,
      "learning_rate": 5.285473749533671e-06,
      "loss": 0.8273,
      "step": 12309
    },
    {
      "epoch": 0.9947675710620416,
      "grad_norm": 2.8427610397338867,
      "learning_rate": 5.284820452365635e-06,
      "loss": 0.834,
      "step": 12310
    },
    {
      "epoch": 0.9948483807753692,
      "grad_norm": 2.7131824493408203,
      "learning_rate": 5.28416715031931e-06,
      "loss": 0.7951,
      "step": 12311
    },
    {
      "epoch": 0.9949291904886968,
      "grad_norm": 2.529956817626953,
      "learning_rate": 5.283513843405886e-06,
      "loss": 0.8515,
      "step": 12312
    },
    {
      "epoch": 0.9950100002020242,
      "grad_norm": 2.633648157119751,
      "learning_rate": 5.282860531636552e-06,
      "loss": 1.0068,
      "step": 12313
    },
    {
      "epoch": 0.9950908099153518,
      "grad_norm": 2.5329723358154297,
      "learning_rate": 5.282207215022499e-06,
      "loss": 0.9443,
      "step": 12314
    },
    {
      "epoch": 0.9951716196286794,
      "grad_norm": 2.7503702640533447,
      "learning_rate": 5.281553893574916e-06,
      "loss": 0.9101,
      "step": 12315
    },
    {
      "epoch": 0.9952524293420069,
      "grad_norm": 2.7381479740142822,
      "learning_rate": 5.280900567304989e-06,
      "loss": 0.8936,
      "step": 12316
    },
    {
      "epoch": 0.9953332390553344,
      "grad_norm": 2.5200910568237305,
      "learning_rate": 5.280247236223916e-06,
      "loss": 0.9186,
      "step": 12317
    },
    {
      "epoch": 0.995414048768662,
      "grad_norm": 2.981550931930542,
      "learning_rate": 5.279593900342881e-06,
      "loss": 0.8855,
      "step": 12318
    },
    {
      "epoch": 0.9954948584819895,
      "grad_norm": 2.4293437004089355,
      "learning_rate": 5.278940559673075e-06,
      "loss": 0.964,
      "step": 12319
    },
    {
      "epoch": 0.9955756681953171,
      "grad_norm": 2.6228461265563965,
      "learning_rate": 5.278287214225689e-06,
      "loss": 0.8185,
      "step": 12320
    },
    {
      "epoch": 0.9956564779086446,
      "grad_norm": 2.576329469680786,
      "learning_rate": 5.277633864011913e-06,
      "loss": 0.8389,
      "step": 12321
    },
    {
      "epoch": 0.9957372876219721,
      "grad_norm": 2.8689534664154053,
      "learning_rate": 5.276980509042937e-06,
      "loss": 0.9261,
      "step": 12322
    },
    {
      "epoch": 0.9958180973352997,
      "grad_norm": 2.744128942489624,
      "learning_rate": 5.276327149329953e-06,
      "loss": 0.8265,
      "step": 12323
    },
    {
      "epoch": 0.9958989070486273,
      "grad_norm": 2.3744585514068604,
      "learning_rate": 5.275673784884147e-06,
      "loss": 0.8113,
      "step": 12324
    },
    {
      "epoch": 0.9959797167619547,
      "grad_norm": 2.8023927211761475,
      "learning_rate": 5.275020415716717e-06,
      "loss": 0.8206,
      "step": 12325
    },
    {
      "epoch": 0.9960605264752823,
      "grad_norm": 2.3721160888671875,
      "learning_rate": 5.274367041838847e-06,
      "loss": 0.8316,
      "step": 12326
    },
    {
      "epoch": 0.9961413361886099,
      "grad_norm": 2.561278820037842,
      "learning_rate": 5.27371366326173e-06,
      "loss": 1.0216,
      "step": 12327
    },
    {
      "epoch": 0.9962221459019374,
      "grad_norm": 2.513955593109131,
      "learning_rate": 5.273060279996557e-06,
      "loss": 0.8201,
      "step": 12328
    },
    {
      "epoch": 0.9963029556152649,
      "grad_norm": 2.638197898864746,
      "learning_rate": 5.27240689205452e-06,
      "loss": 0.9466,
      "step": 12329
    },
    {
      "epoch": 0.9963837653285925,
      "grad_norm": 2.419740676879883,
      "learning_rate": 5.2717534994468066e-06,
      "loss": 0.945,
      "step": 12330
    },
    {
      "epoch": 0.99646457504192,
      "grad_norm": 2.927049160003662,
      "learning_rate": 5.271100102184612e-06,
      "loss": 0.9586,
      "step": 12331
    },
    {
      "epoch": 0.9965453847552476,
      "grad_norm": 2.7897491455078125,
      "learning_rate": 5.270446700279124e-06,
      "loss": 1.0664,
      "step": 12332
    },
    {
      "epoch": 0.9966261944685751,
      "grad_norm": 3.367159366607666,
      "learning_rate": 5.269793293741535e-06,
      "loss": 1.0409,
      "step": 12333
    },
    {
      "epoch": 0.9967070041819026,
      "grad_norm": 2.744915008544922,
      "learning_rate": 5.269139882583038e-06,
      "loss": 0.874,
      "step": 12334
    },
    {
      "epoch": 0.9967878138952302,
      "grad_norm": 2.3484771251678467,
      "learning_rate": 5.2684864668148206e-06,
      "loss": 0.8897,
      "step": 12335
    },
    {
      "epoch": 0.9968686236085578,
      "grad_norm": 2.1473448276519775,
      "learning_rate": 5.2678330464480775e-06,
      "loss": 1.0396,
      "step": 12336
    },
    {
      "epoch": 0.9969494333218852,
      "grad_norm": 3.1765217781066895,
      "learning_rate": 5.267179621494e-06,
      "loss": 0.8927,
      "step": 12337
    },
    {
      "epoch": 0.9970302430352128,
      "grad_norm": 2.411862850189209,
      "learning_rate": 5.266526191963777e-06,
      "loss": 0.8184,
      "step": 12338
    },
    {
      "epoch": 0.9971110527485404,
      "grad_norm": 2.7513530254364014,
      "learning_rate": 5.265872757868603e-06,
      "loss": 0.926,
      "step": 12339
    },
    {
      "epoch": 0.9971918624618679,
      "grad_norm": 3.0000483989715576,
      "learning_rate": 5.265219319219669e-06,
      "loss": 0.8717,
      "step": 12340
    },
    {
      "epoch": 0.9972726721751954,
      "grad_norm": 2.771310329437256,
      "learning_rate": 5.264565876028166e-06,
      "loss": 0.9364,
      "step": 12341
    },
    {
      "epoch": 0.997353481888523,
      "grad_norm": 2.723525047302246,
      "learning_rate": 5.263912428305285e-06,
      "loss": 0.9053,
      "step": 12342
    },
    {
      "epoch": 0.9974342916018506,
      "grad_norm": 2.8177742958068848,
      "learning_rate": 5.263258976062223e-06,
      "loss": 0.9837,
      "step": 12343
    },
    {
      "epoch": 0.9975151013151781,
      "grad_norm": 2.651573657989502,
      "learning_rate": 5.2626055193101644e-06,
      "loss": 1.0256,
      "step": 12344
    },
    {
      "epoch": 0.9975959110285056,
      "grad_norm": 2.6109039783477783,
      "learning_rate": 5.261952058060309e-06,
      "loss": 0.8424,
      "step": 12345
    },
    {
      "epoch": 0.9976767207418332,
      "grad_norm": 2.802781105041504,
      "learning_rate": 5.261298592323843e-06,
      "loss": 0.794,
      "step": 12346
    },
    {
      "epoch": 0.9977575304551607,
      "grad_norm": 3.1382737159729004,
      "learning_rate": 5.260645122111963e-06,
      "loss": 0.8761,
      "step": 12347
    },
    {
      "epoch": 0.9978383401684883,
      "grad_norm": 2.3803482055664062,
      "learning_rate": 5.259991647435858e-06,
      "loss": 0.9459,
      "step": 12348
    },
    {
      "epoch": 0.9979191498818158,
      "grad_norm": 2.5168089866638184,
      "learning_rate": 5.259338168306723e-06,
      "loss": 0.8965,
      "step": 12349
    },
    {
      "epoch": 0.9979999595951433,
      "grad_norm": 2.7386581897735596,
      "learning_rate": 5.258684684735749e-06,
      "loss": 0.8913,
      "step": 12350
    },
    {
      "epoch": 0.9980807693084709,
      "grad_norm": 2.4279236793518066,
      "learning_rate": 5.258031196734131e-06,
      "loss": 0.9048,
      "step": 12351
    },
    {
      "epoch": 0.9981615790217985,
      "grad_norm": 2.523301362991333,
      "learning_rate": 5.257377704313056e-06,
      "loss": 0.8991,
      "step": 12352
    },
    {
      "epoch": 0.9982423887351259,
      "grad_norm": 2.689197063446045,
      "learning_rate": 5.256724207483723e-06,
      "loss": 0.8349,
      "step": 12353
    },
    {
      "epoch": 0.9983231984484535,
      "grad_norm": 2.5127463340759277,
      "learning_rate": 5.2560707062573225e-06,
      "loss": 0.8636,
      "step": 12354
    },
    {
      "epoch": 0.9984040081617811,
      "grad_norm": 2.3001163005828857,
      "learning_rate": 5.255417200645046e-06,
      "loss": 0.9016,
      "step": 12355
    },
    {
      "epoch": 0.9984848178751086,
      "grad_norm": 3.1291275024414062,
      "learning_rate": 5.254763690658089e-06,
      "loss": 0.9736,
      "step": 12356
    },
    {
      "epoch": 0.9985656275884361,
      "grad_norm": 3.4742558002471924,
      "learning_rate": 5.254110176307643e-06,
      "loss": 0.9567,
      "step": 12357
    },
    {
      "epoch": 0.9986464373017637,
      "grad_norm": 2.3015217781066895,
      "learning_rate": 5.2534566576049005e-06,
      "loss": 0.9369,
      "step": 12358
    },
    {
      "epoch": 0.9987272470150912,
      "grad_norm": 2.615185499191284,
      "learning_rate": 5.252803134561057e-06,
      "loss": 0.8202,
      "step": 12359
    },
    {
      "epoch": 0.9988080567284188,
      "grad_norm": 3.7705647945404053,
      "learning_rate": 5.252149607187302e-06,
      "loss": 0.9719,
      "step": 12360
    },
    {
      "epoch": 0.9988888664417463,
      "grad_norm": 2.679399013519287,
      "learning_rate": 5.251496075494834e-06,
      "loss": 0.8473,
      "step": 12361
    },
    {
      "epoch": 0.9989696761550738,
      "grad_norm": 2.5937654972076416,
      "learning_rate": 5.250842539494843e-06,
      "loss": 0.9509,
      "step": 12362
    },
    {
      "epoch": 0.9990504858684014,
      "grad_norm": 2.3290038108825684,
      "learning_rate": 5.250188999198522e-06,
      "loss": 0.901,
      "step": 12363
    },
    {
      "epoch": 0.999131295581729,
      "grad_norm": 2.8928582668304443,
      "learning_rate": 5.249535454617067e-06,
      "loss": 0.8926,
      "step": 12364
    },
    {
      "epoch": 0.9992121052950564,
      "grad_norm": 2.4332470893859863,
      "learning_rate": 5.248881905761671e-06,
      "loss": 0.8268,
      "step": 12365
    },
    {
      "epoch": 0.999292915008384,
      "grad_norm": 2.907656669616699,
      "learning_rate": 5.248228352643525e-06,
      "loss": 0.9369,
      "step": 12366
    },
    {
      "epoch": 0.9993737247217116,
      "grad_norm": 2.733362913131714,
      "learning_rate": 5.247574795273827e-06,
      "loss": 0.8138,
      "step": 12367
    },
    {
      "epoch": 0.9994545344350391,
      "grad_norm": 2.6702473163604736,
      "learning_rate": 5.246921233663768e-06,
      "loss": 0.9376,
      "step": 12368
    },
    {
      "epoch": 0.9995353441483666,
      "grad_norm": 2.54905366897583,
      "learning_rate": 5.2462676678245415e-06,
      "loss": 0.9463,
      "step": 12369
    },
    {
      "epoch": 0.9996161538616942,
      "grad_norm": 2.7659668922424316,
      "learning_rate": 5.245614097767343e-06,
      "loss": 0.9509,
      "step": 12370
    },
    {
      "epoch": 0.9996969635750217,
      "grad_norm": 2.5379488468170166,
      "learning_rate": 5.244960523503368e-06,
      "loss": 0.903,
      "step": 12371
    },
    {
      "epoch": 0.9997777732883493,
      "grad_norm": 2.2945473194122314,
      "learning_rate": 5.244306945043807e-06,
      "loss": 1.0499,
      "step": 12372
    },
    {
      "epoch": 0.9998585830016768,
      "grad_norm": 3.2162699699401855,
      "learning_rate": 5.2436533623998575e-06,
      "loss": 0.9818,
      "step": 12373
    },
    {
      "epoch": 0.9999393927150043,
      "grad_norm": 3.6801867485046387,
      "learning_rate": 5.242999775582711e-06,
      "loss": 0.9017,
      "step": 12374
    },
    {
      "epoch": 1.0000202024283318,
      "grad_norm": 2.544635772705078,
      "learning_rate": 5.2423461846035665e-06,
      "loss": 0.9105,
      "step": 12375
    },
    {
      "epoch": 1.0001010121416594,
      "grad_norm": 2.5052084922790527,
      "learning_rate": 5.241692589473613e-06,
      "loss": 0.8389,
      "step": 12376
    },
    {
      "epoch": 1.000181821854987,
      "grad_norm": 2.650294542312622,
      "learning_rate": 5.241038990204047e-06,
      "loss": 0.8178,
      "step": 12377
    },
    {
      "epoch": 1.0002626315683145,
      "grad_norm": 2.218303680419922,
      "learning_rate": 5.240385386806064e-06,
      "loss": 0.9017,
      "step": 12378
    },
    {
      "epoch": 1.000343441281642,
      "grad_norm": 2.2635653018951416,
      "learning_rate": 5.239731779290858e-06,
      "loss": 0.8809,
      "step": 12379
    },
    {
      "epoch": 1.0004242509949697,
      "grad_norm": 2.671069622039795,
      "learning_rate": 5.239078167669622e-06,
      "loss": 0.8148,
      "step": 12380
    },
    {
      "epoch": 1.000505060708297,
      "grad_norm": 2.1969265937805176,
      "learning_rate": 5.2384245519535545e-06,
      "loss": 0.7317,
      "step": 12381
    },
    {
      "epoch": 1.0005858704216246,
      "grad_norm": 2.548107385635376,
      "learning_rate": 5.237770932153849e-06,
      "loss": 0.8617,
      "step": 12382
    },
    {
      "epoch": 1.0006666801349522,
      "grad_norm": 2.4351718425750732,
      "learning_rate": 5.2371173082816985e-06,
      "loss": 0.7876,
      "step": 12383
    },
    {
      "epoch": 1.0007474898482798,
      "grad_norm": 2.3911964893341064,
      "learning_rate": 5.2364636803483e-06,
      "loss": 0.7921,
      "step": 12384
    },
    {
      "epoch": 1.0008282995616073,
      "grad_norm": 2.0781474113464355,
      "learning_rate": 5.2358100483648475e-06,
      "loss": 0.8085,
      "step": 12385
    },
    {
      "epoch": 1.000909109274935,
      "grad_norm": 2.5982773303985596,
      "learning_rate": 5.235156412342537e-06,
      "loss": 0.7222,
      "step": 12386
    },
    {
      "epoch": 1.0009899189882623,
      "grad_norm": 2.546323776245117,
      "learning_rate": 5.234502772292563e-06,
      "loss": 0.7774,
      "step": 12387
    },
    {
      "epoch": 1.0010707287015899,
      "grad_norm": 2.678215503692627,
      "learning_rate": 5.233849128226121e-06,
      "loss": 0.7791,
      "step": 12388
    },
    {
      "epoch": 1.0011515384149174,
      "grad_norm": 2.8298981189727783,
      "learning_rate": 5.233195480154406e-06,
      "loss": 0.7541,
      "step": 12389
    },
    {
      "epoch": 1.001232348128245,
      "grad_norm": 2.4765350818634033,
      "learning_rate": 5.232541828088616e-06,
      "loss": 0.7497,
      "step": 12390
    },
    {
      "epoch": 1.0013131578415726,
      "grad_norm": 2.929046869277954,
      "learning_rate": 5.231888172039941e-06,
      "loss": 0.862,
      "step": 12391
    },
    {
      "epoch": 1.0013939675549002,
      "grad_norm": 2.8621582984924316,
      "learning_rate": 5.231234512019583e-06,
      "loss": 0.9023,
      "step": 12392
    },
    {
      "epoch": 1.0014747772682275,
      "grad_norm": 2.355471611022949,
      "learning_rate": 5.230580848038732e-06,
      "loss": 0.9258,
      "step": 12393
    },
    {
      "epoch": 1.0015555869815551,
      "grad_norm": 2.8851842880249023,
      "learning_rate": 5.2299271801085875e-06,
      "loss": 0.8527,
      "step": 12394
    },
    {
      "epoch": 1.0016363966948827,
      "grad_norm": 2.6690969467163086,
      "learning_rate": 5.229273508240343e-06,
      "loss": 0.8402,
      "step": 12395
    },
    {
      "epoch": 1.0017172064082103,
      "grad_norm": 2.405996322631836,
      "learning_rate": 5.2286198324451964e-06,
      "loss": 0.7549,
      "step": 12396
    },
    {
      "epoch": 1.0017980161215378,
      "grad_norm": 2.793443441390991,
      "learning_rate": 5.227966152734341e-06,
      "loss": 0.7942,
      "step": 12397
    },
    {
      "epoch": 1.0018788258348654,
      "grad_norm": 2.437676429748535,
      "learning_rate": 5.227312469118976e-06,
      "loss": 0.7605,
      "step": 12398
    },
    {
      "epoch": 1.0019596355481928,
      "grad_norm": 3.5769991874694824,
      "learning_rate": 5.226658781610293e-06,
      "loss": 0.872,
      "step": 12399
    },
    {
      "epoch": 1.0020404452615204,
      "grad_norm": 2.2302238941192627,
      "learning_rate": 5.226005090219493e-06,
      "loss": 0.9423,
      "step": 12400
    },
    {
      "epoch": 1.002121254974848,
      "grad_norm": 2.1968255043029785,
      "learning_rate": 5.22535139495777e-06,
      "loss": 0.8641,
      "step": 12401
    },
    {
      "epoch": 1.0022020646881755,
      "grad_norm": 2.4747560024261475,
      "learning_rate": 5.22469769583632e-06,
      "loss": 0.8181,
      "step": 12402
    },
    {
      "epoch": 1.002282874401503,
      "grad_norm": 3.3412742614746094,
      "learning_rate": 5.2240439928663375e-06,
      "loss": 0.7322,
      "step": 12403
    },
    {
      "epoch": 1.0023636841148307,
      "grad_norm": 2.458338737487793,
      "learning_rate": 5.223390286059023e-06,
      "loss": 0.8743,
      "step": 12404
    },
    {
      "epoch": 1.002444493828158,
      "grad_norm": 2.582960844039917,
      "learning_rate": 5.22273657542557e-06,
      "loss": 0.823,
      "step": 12405
    },
    {
      "epoch": 1.0025253035414856,
      "grad_norm": 2.654578685760498,
      "learning_rate": 5.222082860977176e-06,
      "loss": 0.8594,
      "step": 12406
    },
    {
      "epoch": 1.0026061132548132,
      "grad_norm": 2.927464246749878,
      "learning_rate": 5.221429142725036e-06,
      "loss": 0.8869,
      "step": 12407
    },
    {
      "epoch": 1.0026869229681408,
      "grad_norm": 2.0888020992279053,
      "learning_rate": 5.220775420680348e-06,
      "loss": 0.7466,
      "step": 12408
    },
    {
      "epoch": 1.0027677326814683,
      "grad_norm": 2.579902172088623,
      "learning_rate": 5.22012169485431e-06,
      "loss": 0.8496,
      "step": 12409
    },
    {
      "epoch": 1.002848542394796,
      "grad_norm": 3.1141583919525146,
      "learning_rate": 5.219467965258117e-06,
      "loss": 0.8589,
      "step": 12410
    },
    {
      "epoch": 1.0029293521081235,
      "grad_norm": 2.3609068393707275,
      "learning_rate": 5.218814231902965e-06,
      "loss": 0.8282,
      "step": 12411
    },
    {
      "epoch": 1.0030101618214509,
      "grad_norm": 3.273946762084961,
      "learning_rate": 5.2181604948000534e-06,
      "loss": 0.8372,
      "step": 12412
    },
    {
      "epoch": 1.0030909715347784,
      "grad_norm": 2.5162534713745117,
      "learning_rate": 5.217506753960575e-06,
      "loss": 0.67,
      "step": 12413
    },
    {
      "epoch": 1.003171781248106,
      "grad_norm": 2.6449105739593506,
      "learning_rate": 5.216853009395732e-06,
      "loss": 0.8688,
      "step": 12414
    },
    {
      "epoch": 1.0032525909614336,
      "grad_norm": 2.5569684505462646,
      "learning_rate": 5.216199261116719e-06,
      "loss": 0.7325,
      "step": 12415
    },
    {
      "epoch": 1.0033334006747612,
      "grad_norm": 2.832686424255371,
      "learning_rate": 5.215545509134732e-06,
      "loss": 0.9348,
      "step": 12416
    },
    {
      "epoch": 1.0034142103880888,
      "grad_norm": 2.564300775527954,
      "learning_rate": 5.21489175346097e-06,
      "loss": 0.8142,
      "step": 12417
    },
    {
      "epoch": 1.0034950201014161,
      "grad_norm": 2.928838014602661,
      "learning_rate": 5.21423799410663e-06,
      "loss": 0.6983,
      "step": 12418
    },
    {
      "epoch": 1.0035758298147437,
      "grad_norm": 2.172785758972168,
      "learning_rate": 5.213584231082908e-06,
      "loss": 0.8403,
      "step": 12419
    },
    {
      "epoch": 1.0036566395280713,
      "grad_norm": 2.230769395828247,
      "learning_rate": 5.212930464401002e-06,
      "loss": 0.7288,
      "step": 12420
    },
    {
      "epoch": 1.0037374492413988,
      "grad_norm": 2.494647264480591,
      "learning_rate": 5.212276694072112e-06,
      "loss": 0.9566,
      "step": 12421
    },
    {
      "epoch": 1.0038182589547264,
      "grad_norm": 2.367222785949707,
      "learning_rate": 5.211622920107431e-06,
      "loss": 0.944,
      "step": 12422
    },
    {
      "epoch": 1.003899068668054,
      "grad_norm": 2.6778035163879395,
      "learning_rate": 5.210969142518159e-06,
      "loss": 0.7903,
      "step": 12423
    },
    {
      "epoch": 1.0039798783813814,
      "grad_norm": 2.5383713245391846,
      "learning_rate": 5.210315361315494e-06,
      "loss": 0.7225,
      "step": 12424
    },
    {
      "epoch": 1.004060688094709,
      "grad_norm": 2.898833751678467,
      "learning_rate": 5.209661576510633e-06,
      "loss": 0.791,
      "step": 12425
    },
    {
      "epoch": 1.0041414978080365,
      "grad_norm": 3.2661795616149902,
      "learning_rate": 5.209007788114775e-06,
      "loss": 0.8662,
      "step": 12426
    },
    {
      "epoch": 1.004222307521364,
      "grad_norm": 3.6123569011688232,
      "learning_rate": 5.208353996139115e-06,
      "loss": 0.814,
      "step": 12427
    },
    {
      "epoch": 1.0043031172346917,
      "grad_norm": 2.734471082687378,
      "learning_rate": 5.207700200594854e-06,
      "loss": 0.8,
      "step": 12428
    },
    {
      "epoch": 1.0043839269480193,
      "grad_norm": 2.50447416305542,
      "learning_rate": 5.207046401493188e-06,
      "loss": 0.6917,
      "step": 12429
    },
    {
      "epoch": 1.0044647366613466,
      "grad_norm": 2.889979362487793,
      "learning_rate": 5.2063925988453155e-06,
      "loss": 0.7898,
      "step": 12430
    },
    {
      "epoch": 1.0045455463746742,
      "grad_norm": 2.468834638595581,
      "learning_rate": 5.205738792662435e-06,
      "loss": 0.8287,
      "step": 12431
    },
    {
      "epoch": 1.0046263560880018,
      "grad_norm": 2.7482471466064453,
      "learning_rate": 5.205084982955745e-06,
      "loss": 0.8683,
      "step": 12432
    },
    {
      "epoch": 1.0047071658013293,
      "grad_norm": 2.8472559452056885,
      "learning_rate": 5.2044311697364405e-06,
      "loss": 0.7325,
      "step": 12433
    },
    {
      "epoch": 1.004787975514657,
      "grad_norm": 2.7432994842529297,
      "learning_rate": 5.203777353015725e-06,
      "loss": 0.7982,
      "step": 12434
    },
    {
      "epoch": 1.0048687852279845,
      "grad_norm": 2.612081527709961,
      "learning_rate": 5.203123532804793e-06,
      "loss": 0.7473,
      "step": 12435
    },
    {
      "epoch": 1.0049495949413119,
      "grad_norm": 2.707618236541748,
      "learning_rate": 5.202469709114842e-06,
      "loss": 0.7284,
      "step": 12436
    },
    {
      "epoch": 1.0050304046546394,
      "grad_norm": 2.7868945598602295,
      "learning_rate": 5.201815881957074e-06,
      "loss": 0.8548,
      "step": 12437
    },
    {
      "epoch": 1.005111214367967,
      "grad_norm": 2.8199827671051025,
      "learning_rate": 5.201162051342687e-06,
      "loss": 0.8966,
      "step": 12438
    },
    {
      "epoch": 1.0051920240812946,
      "grad_norm": 3.1978096961975098,
      "learning_rate": 5.200508217282876e-06,
      "loss": 0.7633,
      "step": 12439
    },
    {
      "epoch": 1.0052728337946222,
      "grad_norm": 2.757664680480957,
      "learning_rate": 5.199854379788843e-06,
      "loss": 0.837,
      "step": 12440
    },
    {
      "epoch": 1.0053536435079498,
      "grad_norm": 3.1558547019958496,
      "learning_rate": 5.199200538871786e-06,
      "loss": 0.7882,
      "step": 12441
    },
    {
      "epoch": 1.0054344532212771,
      "grad_norm": 2.5289220809936523,
      "learning_rate": 5.198546694542903e-06,
      "loss": 0.919,
      "step": 12442
    },
    {
      "epoch": 1.0055152629346047,
      "grad_norm": 2.881176233291626,
      "learning_rate": 5.197892846813393e-06,
      "loss": 0.8632,
      "step": 12443
    },
    {
      "epoch": 1.0055960726479323,
      "grad_norm": 2.6454885005950928,
      "learning_rate": 5.1972389956944544e-06,
      "loss": 0.9492,
      "step": 12444
    },
    {
      "epoch": 1.0056768823612598,
      "grad_norm": 2.912611484527588,
      "learning_rate": 5.196585141197288e-06,
      "loss": 0.9273,
      "step": 12445
    },
    {
      "epoch": 1.0057576920745874,
      "grad_norm": 2.206540584564209,
      "learning_rate": 5.1959312833330895e-06,
      "loss": 0.8335,
      "step": 12446
    },
    {
      "epoch": 1.005838501787915,
      "grad_norm": 2.196087121963501,
      "learning_rate": 5.195277422113062e-06,
      "loss": 0.7331,
      "step": 12447
    },
    {
      "epoch": 1.0059193115012424,
      "grad_norm": 3.0699610710144043,
      "learning_rate": 5.1946235575484005e-06,
      "loss": 0.8245,
      "step": 12448
    },
    {
      "epoch": 1.00600012121457,
      "grad_norm": 2.666667938232422,
      "learning_rate": 5.193969689650308e-06,
      "loss": 0.7672,
      "step": 12449
    },
    {
      "epoch": 1.0060809309278975,
      "grad_norm": 2.46816086769104,
      "learning_rate": 5.19331581842998e-06,
      "loss": 0.7748,
      "step": 12450
    },
    {
      "epoch": 1.006161740641225,
      "grad_norm": 2.4281744956970215,
      "learning_rate": 5.192661943898618e-06,
      "loss": 0.8805,
      "step": 12451
    },
    {
      "epoch": 1.0062425503545527,
      "grad_norm": 2.9114654064178467,
      "learning_rate": 5.192008066067421e-06,
      "loss": 0.8672,
      "step": 12452
    },
    {
      "epoch": 1.0063233600678803,
      "grad_norm": 2.379807233810425,
      "learning_rate": 5.1913541849475866e-06,
      "loss": 0.8261,
      "step": 12453
    },
    {
      "epoch": 1.0064041697812076,
      "grad_norm": 2.6848325729370117,
      "learning_rate": 5.190700300550317e-06,
      "loss": 0.9017,
      "step": 12454
    },
    {
      "epoch": 1.0064849794945352,
      "grad_norm": 2.5436956882476807,
      "learning_rate": 5.19004641288681e-06,
      "loss": 0.7208,
      "step": 12455
    },
    {
      "epoch": 1.0065657892078628,
      "grad_norm": 2.371771812438965,
      "learning_rate": 5.189392521968266e-06,
      "loss": 0.8036,
      "step": 12456
    },
    {
      "epoch": 1.0066465989211903,
      "grad_norm": 2.7062172889709473,
      "learning_rate": 5.188738627805884e-06,
      "loss": 0.9021,
      "step": 12457
    },
    {
      "epoch": 1.006727408634518,
      "grad_norm": 2.474684476852417,
      "learning_rate": 5.1880847304108625e-06,
      "loss": 0.8564,
      "step": 12458
    },
    {
      "epoch": 1.0068082183478455,
      "grad_norm": 3.052793502807617,
      "learning_rate": 5.187430829794405e-06,
      "loss": 0.7722,
      "step": 12459
    },
    {
      "epoch": 1.0068890280611729,
      "grad_norm": 2.415121078491211,
      "learning_rate": 5.186776925967706e-06,
      "loss": 0.8317,
      "step": 12460
    },
    {
      "epoch": 1.0069698377745004,
      "grad_norm": 2.5590007305145264,
      "learning_rate": 5.186123018941967e-06,
      "loss": 0.7778,
      "step": 12461
    },
    {
      "epoch": 1.007050647487828,
      "grad_norm": 2.7981984615325928,
      "learning_rate": 5.1854691087283915e-06,
      "loss": 0.78,
      "step": 12462
    },
    {
      "epoch": 1.0071314572011556,
      "grad_norm": 2.3507823944091797,
      "learning_rate": 5.184815195338176e-06,
      "loss": 0.9592,
      "step": 12463
    },
    {
      "epoch": 1.0072122669144832,
      "grad_norm": 2.577120065689087,
      "learning_rate": 5.18416127878252e-06,
      "loss": 0.9041,
      "step": 12464
    },
    {
      "epoch": 1.0072930766278108,
      "grad_norm": 2.737476110458374,
      "learning_rate": 5.183507359072626e-06,
      "loss": 0.9182,
      "step": 12465
    },
    {
      "epoch": 1.0073738863411381,
      "grad_norm": 2.4286694526672363,
      "learning_rate": 5.1828534362196924e-06,
      "loss": 0.8065,
      "step": 12466
    },
    {
      "epoch": 1.0074546960544657,
      "grad_norm": 2.646874189376831,
      "learning_rate": 5.182199510234919e-06,
      "loss": 0.9631,
      "step": 12467
    },
    {
      "epoch": 1.0075355057677933,
      "grad_norm": 2.3519954681396484,
      "learning_rate": 5.181545581129507e-06,
      "loss": 0.7719,
      "step": 12468
    },
    {
      "epoch": 1.0076163154811208,
      "grad_norm": 2.4610695838928223,
      "learning_rate": 5.180891648914656e-06,
      "loss": 0.8303,
      "step": 12469
    },
    {
      "epoch": 1.0076971251944484,
      "grad_norm": 3.065570116043091,
      "learning_rate": 5.180237713601566e-06,
      "loss": 0.7641,
      "step": 12470
    },
    {
      "epoch": 1.007777934907776,
      "grad_norm": 2.477292537689209,
      "learning_rate": 5.17958377520144e-06,
      "loss": 0.9094,
      "step": 12471
    },
    {
      "epoch": 1.0078587446211034,
      "grad_norm": 2.3511126041412354,
      "learning_rate": 5.178929833725473e-06,
      "loss": 0.8413,
      "step": 12472
    },
    {
      "epoch": 1.007939554334431,
      "grad_norm": 2.6266164779663086,
      "learning_rate": 5.178275889184872e-06,
      "loss": 0.8049,
      "step": 12473
    },
    {
      "epoch": 1.0080203640477585,
      "grad_norm": 2.6902618408203125,
      "learning_rate": 5.177621941590833e-06,
      "loss": 0.8439,
      "step": 12474
    },
    {
      "epoch": 1.008101173761086,
      "grad_norm": 2.522594690322876,
      "learning_rate": 5.176967990954557e-06,
      "loss": 0.8213,
      "step": 12475
    },
    {
      "epoch": 1.0081819834744137,
      "grad_norm": 3.043595790863037,
      "learning_rate": 5.176314037287246e-06,
      "loss": 0.7734,
      "step": 12476
    },
    {
      "epoch": 1.0082627931877413,
      "grad_norm": 2.538238286972046,
      "learning_rate": 5.1756600806001e-06,
      "loss": 0.8054,
      "step": 12477
    },
    {
      "epoch": 1.0083436029010686,
      "grad_norm": 2.71919846534729,
      "learning_rate": 5.175006120904319e-06,
      "loss": 0.895,
      "step": 12478
    },
    {
      "epoch": 1.0084244126143962,
      "grad_norm": 2.378711700439453,
      "learning_rate": 5.1743521582111054e-06,
      "loss": 0.8879,
      "step": 12479
    },
    {
      "epoch": 1.0085052223277238,
      "grad_norm": 2.6883904933929443,
      "learning_rate": 5.1736981925316575e-06,
      "loss": 0.8416,
      "step": 12480
    },
    {
      "epoch": 1.0085860320410514,
      "grad_norm": 2.489894390106201,
      "learning_rate": 5.173044223877181e-06,
      "loss": 0.8498,
      "step": 12481
    },
    {
      "epoch": 1.008666841754379,
      "grad_norm": 2.707709312438965,
      "learning_rate": 5.172390252258871e-06,
      "loss": 0.7953,
      "step": 12482
    },
    {
      "epoch": 1.0087476514677065,
      "grad_norm": 2.8330583572387695,
      "learning_rate": 5.171736277687931e-06,
      "loss": 0.7782,
      "step": 12483
    },
    {
      "epoch": 1.0088284611810339,
      "grad_norm": 2.5243594646453857,
      "learning_rate": 5.171082300175562e-06,
      "loss": 0.8177,
      "step": 12484
    },
    {
      "epoch": 1.0089092708943614,
      "grad_norm": 2.657561779022217,
      "learning_rate": 5.170428319732966e-06,
      "loss": 0.8734,
      "step": 12485
    },
    {
      "epoch": 1.008990080607689,
      "grad_norm": 2.4913158416748047,
      "learning_rate": 5.169774336371342e-06,
      "loss": 0.9185,
      "step": 12486
    },
    {
      "epoch": 1.0090708903210166,
      "grad_norm": 2.6465888023376465,
      "learning_rate": 5.1691203501018935e-06,
      "loss": 0.8314,
      "step": 12487
    },
    {
      "epoch": 1.0091517000343442,
      "grad_norm": 2.3169476985931396,
      "learning_rate": 5.1684663609358195e-06,
      "loss": 0.9192,
      "step": 12488
    },
    {
      "epoch": 1.0092325097476718,
      "grad_norm": 2.53898024559021,
      "learning_rate": 5.167812368884323e-06,
      "loss": 0.8389,
      "step": 12489
    },
    {
      "epoch": 1.0093133194609991,
      "grad_norm": 2.3562941551208496,
      "learning_rate": 5.167158373958605e-06,
      "loss": 0.7824,
      "step": 12490
    },
    {
      "epoch": 1.0093941291743267,
      "grad_norm": 3.0374176502227783,
      "learning_rate": 5.166504376169867e-06,
      "loss": 0.6628,
      "step": 12491
    },
    {
      "epoch": 1.0094749388876543,
      "grad_norm": 2.5810141563415527,
      "learning_rate": 5.1658503755293075e-06,
      "loss": 0.7645,
      "step": 12492
    },
    {
      "epoch": 1.0095557486009819,
      "grad_norm": 2.4141604900360107,
      "learning_rate": 5.165196372048133e-06,
      "loss": 0.7173,
      "step": 12493
    },
    {
      "epoch": 1.0096365583143094,
      "grad_norm": 2.61149001121521,
      "learning_rate": 5.164542365737539e-06,
      "loss": 0.7786,
      "step": 12494
    },
    {
      "epoch": 1.009717368027637,
      "grad_norm": 2.6056108474731445,
      "learning_rate": 5.1638883566087324e-06,
      "loss": 0.883,
      "step": 12495
    },
    {
      "epoch": 1.0097981777409644,
      "grad_norm": 3.0438790321350098,
      "learning_rate": 5.1632343446729135e-06,
      "loss": 0.6845,
      "step": 12496
    },
    {
      "epoch": 1.009878987454292,
      "grad_norm": 3.2974979877471924,
      "learning_rate": 5.162580329941283e-06,
      "loss": 0.9291,
      "step": 12497
    },
    {
      "epoch": 1.0099597971676195,
      "grad_norm": 2.6835813522338867,
      "learning_rate": 5.161926312425042e-06,
      "loss": 0.7465,
      "step": 12498
    },
    {
      "epoch": 1.010040606880947,
      "grad_norm": 2.5462892055511475,
      "learning_rate": 5.161272292135394e-06,
      "loss": 0.7749,
      "step": 12499
    },
    {
      "epoch": 1.0101214165942747,
      "grad_norm": 2.438927412033081,
      "learning_rate": 5.160618269083538e-06,
      "loss": 0.867,
      "step": 12500
    },
    {
      "epoch": 1.0102022263076023,
      "grad_norm": 2.727975368499756,
      "learning_rate": 5.159964243280681e-06,
      "loss": 0.7904,
      "step": 12501
    },
    {
      "epoch": 1.0102830360209296,
      "grad_norm": 2.4510722160339355,
      "learning_rate": 5.159310214738019e-06,
      "loss": 0.8352,
      "step": 12502
    },
    {
      "epoch": 1.0103638457342572,
      "grad_norm": 2.4921584129333496,
      "learning_rate": 5.158656183466757e-06,
      "loss": 0.8254,
      "step": 12503
    },
    {
      "epoch": 1.0104446554475848,
      "grad_norm": 2.5787413120269775,
      "learning_rate": 5.158002149478096e-06,
      "loss": 0.6811,
      "step": 12504
    },
    {
      "epoch": 1.0105254651609124,
      "grad_norm": 2.7388923168182373,
      "learning_rate": 5.157348112783239e-06,
      "loss": 0.7902,
      "step": 12505
    },
    {
      "epoch": 1.01060627487424,
      "grad_norm": 2.5851552486419678,
      "learning_rate": 5.156694073393388e-06,
      "loss": 0.9329,
      "step": 12506
    },
    {
      "epoch": 1.0106870845875675,
      "grad_norm": 2.8699493408203125,
      "learning_rate": 5.156040031319744e-06,
      "loss": 0.8148,
      "step": 12507
    },
    {
      "epoch": 1.0107678943008949,
      "grad_norm": 3.3402721881866455,
      "learning_rate": 5.155385986573511e-06,
      "loss": 0.7249,
      "step": 12508
    },
    {
      "epoch": 1.0108487040142224,
      "grad_norm": 2.719470977783203,
      "learning_rate": 5.154731939165891e-06,
      "loss": 0.8114,
      "step": 12509
    },
    {
      "epoch": 1.01092951372755,
      "grad_norm": 2.856743574142456,
      "learning_rate": 5.154077889108085e-06,
      "loss": 0.7428,
      "step": 12510
    },
    {
      "epoch": 1.0110103234408776,
      "grad_norm": 2.2968685626983643,
      "learning_rate": 5.153423836411293e-06,
      "loss": 0.8198,
      "step": 12511
    },
    {
      "epoch": 1.0110911331542052,
      "grad_norm": 3.1482110023498535,
      "learning_rate": 5.152769781086723e-06,
      "loss": 0.8385,
      "step": 12512
    },
    {
      "epoch": 1.0111719428675328,
      "grad_norm": 2.8562870025634766,
      "learning_rate": 5.152115723145572e-06,
      "loss": 0.8137,
      "step": 12513
    },
    {
      "epoch": 1.0112527525808601,
      "grad_norm": 2.7162911891937256,
      "learning_rate": 5.151461662599047e-06,
      "loss": 0.7455,
      "step": 12514
    },
    {
      "epoch": 1.0113335622941877,
      "grad_norm": 2.4655370712280273,
      "learning_rate": 5.1508075994583465e-06,
      "loss": 0.7117,
      "step": 12515
    },
    {
      "epoch": 1.0114143720075153,
      "grad_norm": 2.6548609733581543,
      "learning_rate": 5.150153533734677e-06,
      "loss": 0.8141,
      "step": 12516
    },
    {
      "epoch": 1.0114951817208429,
      "grad_norm": 2.7891759872436523,
      "learning_rate": 5.149499465439237e-06,
      "loss": 0.8306,
      "step": 12517
    },
    {
      "epoch": 1.0115759914341704,
      "grad_norm": 3.0253827571868896,
      "learning_rate": 5.148845394583233e-06,
      "loss": 0.7892,
      "step": 12518
    },
    {
      "epoch": 1.011656801147498,
      "grad_norm": 2.6911189556121826,
      "learning_rate": 5.148191321177864e-06,
      "loss": 0.8156,
      "step": 12519
    },
    {
      "epoch": 1.0117376108608254,
      "grad_norm": 2.3544700145721436,
      "learning_rate": 5.147537245234334e-06,
      "loss": 0.9618,
      "step": 12520
    },
    {
      "epoch": 1.011818420574153,
      "grad_norm": 2.5659430027008057,
      "learning_rate": 5.1468831667638475e-06,
      "loss": 0.7331,
      "step": 12521
    },
    {
      "epoch": 1.0118992302874805,
      "grad_norm": 2.667371988296509,
      "learning_rate": 5.146229085777605e-06,
      "loss": 0.789,
      "step": 12522
    },
    {
      "epoch": 1.011980040000808,
      "grad_norm": 2.415018320083618,
      "learning_rate": 5.145575002286811e-06,
      "loss": 0.8749,
      "step": 12523
    },
    {
      "epoch": 1.0120608497141357,
      "grad_norm": 2.4200429916381836,
      "learning_rate": 5.144920916302669e-06,
      "loss": 0.7911,
      "step": 12524
    },
    {
      "epoch": 1.0121416594274633,
      "grad_norm": 2.622950553894043,
      "learning_rate": 5.144266827836378e-06,
      "loss": 0.8284,
      "step": 12525
    },
    {
      "epoch": 1.0122224691407906,
      "grad_norm": 2.663954019546509,
      "learning_rate": 5.143612736899145e-06,
      "loss": 0.9321,
      "step": 12526
    },
    {
      "epoch": 1.0123032788541182,
      "grad_norm": 2.183441638946533,
      "learning_rate": 5.142958643502172e-06,
      "loss": 0.7527,
      "step": 12527
    },
    {
      "epoch": 1.0123840885674458,
      "grad_norm": 2.886779546737671,
      "learning_rate": 5.14230454765666e-06,
      "loss": 1.0143,
      "step": 12528
    },
    {
      "epoch": 1.0124648982807734,
      "grad_norm": 2.8194220066070557,
      "learning_rate": 5.141650449373815e-06,
      "loss": 0.8163,
      "step": 12529
    },
    {
      "epoch": 1.012545707994101,
      "grad_norm": 2.681802988052368,
      "learning_rate": 5.14099634866484e-06,
      "loss": 0.8449,
      "step": 12530
    },
    {
      "epoch": 1.0126265177074285,
      "grad_norm": 2.327749013900757,
      "learning_rate": 5.1403422455409334e-06,
      "loss": 0.8555,
      "step": 12531
    },
    {
      "epoch": 1.0127073274207559,
      "grad_norm": 2.747070074081421,
      "learning_rate": 5.139688140013305e-06,
      "loss": 0.8523,
      "step": 12532
    },
    {
      "epoch": 1.0127881371340834,
      "grad_norm": 2.4575886726379395,
      "learning_rate": 5.139034032093153e-06,
      "loss": 0.7795,
      "step": 12533
    },
    {
      "epoch": 1.012868946847411,
      "grad_norm": 2.783709764480591,
      "learning_rate": 5.138379921791684e-06,
      "loss": 0.8766,
      "step": 12534
    },
    {
      "epoch": 1.0129497565607386,
      "grad_norm": 2.537069320678711,
      "learning_rate": 5.1377258091201e-06,
      "loss": 0.7238,
      "step": 12535
    },
    {
      "epoch": 1.0130305662740662,
      "grad_norm": 2.3739278316497803,
      "learning_rate": 5.137071694089604e-06,
      "loss": 0.9084,
      "step": 12536
    },
    {
      "epoch": 1.0131113759873938,
      "grad_norm": 3.3975095748901367,
      "learning_rate": 5.1364175767114e-06,
      "loss": 0.8118,
      "step": 12537
    },
    {
      "epoch": 1.0131921857007211,
      "grad_norm": 2.5251986980438232,
      "learning_rate": 5.135763456996692e-06,
      "loss": 0.8556,
      "step": 12538
    },
    {
      "epoch": 1.0132729954140487,
      "grad_norm": 2.407038927078247,
      "learning_rate": 5.135109334956682e-06,
      "loss": 0.8468,
      "step": 12539
    },
    {
      "epoch": 1.0133538051273763,
      "grad_norm": 2.478182315826416,
      "learning_rate": 5.134455210602575e-06,
      "loss": 0.7428,
      "step": 12540
    },
    {
      "epoch": 1.0134346148407039,
      "grad_norm": 2.464784860610962,
      "learning_rate": 5.133801083945573e-06,
      "loss": 0.6878,
      "step": 12541
    },
    {
      "epoch": 1.0135154245540314,
      "grad_norm": 2.6182079315185547,
      "learning_rate": 5.1331469549968814e-06,
      "loss": 0.8197,
      "step": 12542
    },
    {
      "epoch": 1.013596234267359,
      "grad_norm": 2.762474298477173,
      "learning_rate": 5.132492823767702e-06,
      "loss": 0.7491,
      "step": 12543
    },
    {
      "epoch": 1.0136770439806866,
      "grad_norm": 2.3340630531311035,
      "learning_rate": 5.131838690269242e-06,
      "loss": 0.8649,
      "step": 12544
    },
    {
      "epoch": 1.013757853694014,
      "grad_norm": 2.8064169883728027,
      "learning_rate": 5.1311845545126995e-06,
      "loss": 0.9643,
      "step": 12545
    },
    {
      "epoch": 1.0138386634073415,
      "grad_norm": 2.4351413249969482,
      "learning_rate": 5.130530416509283e-06,
      "loss": 0.8348,
      "step": 12546
    },
    {
      "epoch": 1.013919473120669,
      "grad_norm": 2.338689088821411,
      "learning_rate": 5.129876276270195e-06,
      "loss": 0.8594,
      "step": 12547
    },
    {
      "epoch": 1.0140002828339967,
      "grad_norm": 2.895043134689331,
      "learning_rate": 5.129222133806638e-06,
      "loss": 0.7953,
      "step": 12548
    },
    {
      "epoch": 1.0140810925473243,
      "grad_norm": 2.9791154861450195,
      "learning_rate": 5.128567989129816e-06,
      "loss": 0.8836,
      "step": 12549
    },
    {
      "epoch": 1.0141619022606518,
      "grad_norm": 2.254404306411743,
      "learning_rate": 5.127913842250936e-06,
      "loss": 0.804,
      "step": 12550
    },
    {
      "epoch": 1.0142427119739792,
      "grad_norm": 2.5991082191467285,
      "learning_rate": 5.127259693181199e-06,
      "loss": 0.8925,
      "step": 12551
    },
    {
      "epoch": 1.0143235216873068,
      "grad_norm": 2.7920894622802734,
      "learning_rate": 5.126605541931811e-06,
      "loss": 0.9347,
      "step": 12552
    },
    {
      "epoch": 1.0144043314006344,
      "grad_norm": 2.6438891887664795,
      "learning_rate": 5.125951388513972e-06,
      "loss": 0.6857,
      "step": 12553
    },
    {
      "epoch": 1.014485141113962,
      "grad_norm": 2.456252336502075,
      "learning_rate": 5.125297232938892e-06,
      "loss": 0.8561,
      "step": 12554
    },
    {
      "epoch": 1.0145659508272895,
      "grad_norm": 2.6659305095672607,
      "learning_rate": 5.124643075217771e-06,
      "loss": 0.8925,
      "step": 12555
    },
    {
      "epoch": 1.014646760540617,
      "grad_norm": 2.682535409927368,
      "learning_rate": 5.123988915361814e-06,
      "loss": 0.7246,
      "step": 12556
    },
    {
      "epoch": 1.0147275702539444,
      "grad_norm": 2.886742353439331,
      "learning_rate": 5.123334753382224e-06,
      "loss": 0.6959,
      "step": 12557
    },
    {
      "epoch": 1.014808379967272,
      "grad_norm": 2.5031943321228027,
      "learning_rate": 5.1226805892902095e-06,
      "loss": 0.7845,
      "step": 12558
    },
    {
      "epoch": 1.0148891896805996,
      "grad_norm": 2.4899203777313232,
      "learning_rate": 5.122026423096968e-06,
      "loss": 0.7403,
      "step": 12559
    },
    {
      "epoch": 1.0149699993939272,
      "grad_norm": 2.523195505142212,
      "learning_rate": 5.121372254813712e-06,
      "loss": 0.803,
      "step": 12560
    },
    {
      "epoch": 1.0150508091072548,
      "grad_norm": 3.197504758834839,
      "learning_rate": 5.120718084451639e-06,
      "loss": 0.8498,
      "step": 12561
    },
    {
      "epoch": 1.0151316188205823,
      "grad_norm": 2.8150250911712646,
      "learning_rate": 5.120063912021957e-06,
      "loss": 0.8401,
      "step": 12562
    },
    {
      "epoch": 1.0152124285339097,
      "grad_norm": 2.3248839378356934,
      "learning_rate": 5.1194097375358675e-06,
      "loss": 0.8333,
      "step": 12563
    },
    {
      "epoch": 1.0152932382472373,
      "grad_norm": 2.4993953704833984,
      "learning_rate": 5.118755561004577e-06,
      "loss": 0.8824,
      "step": 12564
    },
    {
      "epoch": 1.0153740479605649,
      "grad_norm": 2.494349718093872,
      "learning_rate": 5.11810138243929e-06,
      "loss": 0.794,
      "step": 12565
    },
    {
      "epoch": 1.0154548576738924,
      "grad_norm": 2.4131264686584473,
      "learning_rate": 5.117447201851212e-06,
      "loss": 0.8869,
      "step": 12566
    },
    {
      "epoch": 1.01553566738722,
      "grad_norm": 2.6513562202453613,
      "learning_rate": 5.1167930192515434e-06,
      "loss": 0.8108,
      "step": 12567
    },
    {
      "epoch": 1.0156164771005476,
      "grad_norm": 2.5764501094818115,
      "learning_rate": 5.116138834651494e-06,
      "loss": 0.8413,
      "step": 12568
    },
    {
      "epoch": 1.015697286813875,
      "grad_norm": 2.739586114883423,
      "learning_rate": 5.115484648062265e-06,
      "loss": 0.9202,
      "step": 12569
    },
    {
      "epoch": 1.0157780965272025,
      "grad_norm": 2.8270676136016846,
      "learning_rate": 5.114830459495063e-06,
      "loss": 0.8101,
      "step": 12570
    },
    {
      "epoch": 1.01585890624053,
      "grad_norm": 2.359265089035034,
      "learning_rate": 5.114176268961089e-06,
      "loss": 0.8041,
      "step": 12571
    },
    {
      "epoch": 1.0159397159538577,
      "grad_norm": 2.790734052658081,
      "learning_rate": 5.113522076471553e-06,
      "loss": 0.8016,
      "step": 12572
    },
    {
      "epoch": 1.0160205256671853,
      "grad_norm": 2.2723278999328613,
      "learning_rate": 5.1128678820376565e-06,
      "loss": 0.9098,
      "step": 12573
    },
    {
      "epoch": 1.0161013353805128,
      "grad_norm": 2.446974515914917,
      "learning_rate": 5.112213685670604e-06,
      "loss": 0.7743,
      "step": 12574
    },
    {
      "epoch": 1.0161821450938402,
      "grad_norm": 3.0320420265197754,
      "learning_rate": 5.111559487381603e-06,
      "loss": 0.8606,
      "step": 12575
    },
    {
      "epoch": 1.0162629548071678,
      "grad_norm": 2.7096023559570312,
      "learning_rate": 5.110905287181855e-06,
      "loss": 0.8023,
      "step": 12576
    },
    {
      "epoch": 1.0163437645204954,
      "grad_norm": 2.3316524028778076,
      "learning_rate": 5.110251085082567e-06,
      "loss": 0.7733,
      "step": 12577
    },
    {
      "epoch": 1.016424574233823,
      "grad_norm": 2.485217332839966,
      "learning_rate": 5.109596881094942e-06,
      "loss": 0.8495,
      "step": 12578
    },
    {
      "epoch": 1.0165053839471505,
      "grad_norm": 2.262326955795288,
      "learning_rate": 5.108942675230188e-06,
      "loss": 0.8006,
      "step": 12579
    },
    {
      "epoch": 1.016586193660478,
      "grad_norm": 2.484311580657959,
      "learning_rate": 5.1082884674995085e-06,
      "loss": 0.8535,
      "step": 12580
    },
    {
      "epoch": 1.0166670033738054,
      "grad_norm": 2.561978578567505,
      "learning_rate": 5.107634257914107e-06,
      "loss": 0.8623,
      "step": 12581
    },
    {
      "epoch": 1.016747813087133,
      "grad_norm": 2.7206268310546875,
      "learning_rate": 5.106980046485189e-06,
      "loss": 0.7643,
      "step": 12582
    },
    {
      "epoch": 1.0168286228004606,
      "grad_norm": 2.300199031829834,
      "learning_rate": 5.106325833223963e-06,
      "loss": 0.8361,
      "step": 12583
    },
    {
      "epoch": 1.0169094325137882,
      "grad_norm": 2.913027763366699,
      "learning_rate": 5.10567161814163e-06,
      "loss": 0.9844,
      "step": 12584
    },
    {
      "epoch": 1.0169902422271158,
      "grad_norm": 2.632624387741089,
      "learning_rate": 5.105017401249397e-06,
      "loss": 0.8527,
      "step": 12585
    },
    {
      "epoch": 1.0170710519404433,
      "grad_norm": 2.1495048999786377,
      "learning_rate": 5.104363182558467e-06,
      "loss": 0.7796,
      "step": 12586
    },
    {
      "epoch": 1.0171518616537707,
      "grad_norm": 2.6786608695983887,
      "learning_rate": 5.103708962080048e-06,
      "loss": 0.7279,
      "step": 12587
    },
    {
      "epoch": 1.0172326713670983,
      "grad_norm": 2.464973211288452,
      "learning_rate": 5.103054739825345e-06,
      "loss": 0.8749,
      "step": 12588
    },
    {
      "epoch": 1.0173134810804259,
      "grad_norm": 2.6223528385162354,
      "learning_rate": 5.102400515805561e-06,
      "loss": 0.8116,
      "step": 12589
    },
    {
      "epoch": 1.0173942907937534,
      "grad_norm": 2.671349048614502,
      "learning_rate": 5.101746290031903e-06,
      "loss": 0.8056,
      "step": 12590
    },
    {
      "epoch": 1.017475100507081,
      "grad_norm": 2.7348575592041016,
      "learning_rate": 5.101092062515578e-06,
      "loss": 0.8513,
      "step": 12591
    },
    {
      "epoch": 1.0175559102204086,
      "grad_norm": 3.5210304260253906,
      "learning_rate": 5.100437833267788e-06,
      "loss": 0.8594,
      "step": 12592
    },
    {
      "epoch": 1.017636719933736,
      "grad_norm": 2.3753433227539062,
      "learning_rate": 5.099783602299739e-06,
      "loss": 0.8299,
      "step": 12593
    },
    {
      "epoch": 1.0177175296470635,
      "grad_norm": 2.5451722145080566,
      "learning_rate": 5.099129369622639e-06,
      "loss": 0.844,
      "step": 12594
    },
    {
      "epoch": 1.017798339360391,
      "grad_norm": 3.6367604732513428,
      "learning_rate": 5.098475135247689e-06,
      "loss": 0.8029,
      "step": 12595
    },
    {
      "epoch": 1.0178791490737187,
      "grad_norm": 2.472900867462158,
      "learning_rate": 5.097820899186098e-06,
      "loss": 0.8026,
      "step": 12596
    },
    {
      "epoch": 1.0179599587870463,
      "grad_norm": 3.27156138420105,
      "learning_rate": 5.0971666614490725e-06,
      "loss": 0.9626,
      "step": 12597
    },
    {
      "epoch": 1.0180407685003738,
      "grad_norm": 2.752840518951416,
      "learning_rate": 5.096512422047812e-06,
      "loss": 0.8839,
      "step": 12598
    },
    {
      "epoch": 1.0181215782137012,
      "grad_norm": 2.6791982650756836,
      "learning_rate": 5.095858180993529e-06,
      "loss": 0.7915,
      "step": 12599
    },
    {
      "epoch": 1.0182023879270288,
      "grad_norm": 2.264230489730835,
      "learning_rate": 5.095203938297426e-06,
      "loss": 0.8246,
      "step": 12600
    },
    {
      "epoch": 1.0182831976403564,
      "grad_norm": 2.516456127166748,
      "learning_rate": 5.094549693970707e-06,
      "loss": 0.727,
      "step": 12601
    },
    {
      "epoch": 1.018364007353684,
      "grad_norm": 2.3891682624816895,
      "learning_rate": 5.093895448024581e-06,
      "loss": 0.8913,
      "step": 12602
    },
    {
      "epoch": 1.0184448170670115,
      "grad_norm": 2.3722238540649414,
      "learning_rate": 5.093241200470252e-06,
      "loss": 0.766,
      "step": 12603
    },
    {
      "epoch": 1.018525626780339,
      "grad_norm": 2.553745746612549,
      "learning_rate": 5.092586951318924e-06,
      "loss": 0.7955,
      "step": 12604
    },
    {
      "epoch": 1.0186064364936664,
      "grad_norm": 2.5744190216064453,
      "learning_rate": 5.0919327005818065e-06,
      "loss": 0.7883,
      "step": 12605
    },
    {
      "epoch": 1.018687246206994,
      "grad_norm": 3.2034945487976074,
      "learning_rate": 5.0912784482701015e-06,
      "loss": 0.8694,
      "step": 12606
    },
    {
      "epoch": 1.0187680559203216,
      "grad_norm": 2.4338269233703613,
      "learning_rate": 5.090624194395018e-06,
      "loss": 0.7803,
      "step": 12607
    },
    {
      "epoch": 1.0188488656336492,
      "grad_norm": 2.874695062637329,
      "learning_rate": 5.089969938967759e-06,
      "loss": 0.8685,
      "step": 12608
    },
    {
      "epoch": 1.0189296753469768,
      "grad_norm": 2.651587963104248,
      "learning_rate": 5.089315681999531e-06,
      "loss": 0.8423,
      "step": 12609
    },
    {
      "epoch": 1.0190104850603043,
      "grad_norm": 2.773283004760742,
      "learning_rate": 5.088661423501542e-06,
      "loss": 0.858,
      "step": 12610
    },
    {
      "epoch": 1.0190912947736317,
      "grad_norm": 2.34829044342041,
      "learning_rate": 5.088007163484997e-06,
      "loss": 0.8556,
      "step": 12611
    },
    {
      "epoch": 1.0191721044869593,
      "grad_norm": 3.038344383239746,
      "learning_rate": 5.087352901961098e-06,
      "loss": 0.8405,
      "step": 12612
    },
    {
      "epoch": 1.0192529142002869,
      "grad_norm": 2.6653900146484375,
      "learning_rate": 5.0866986389410564e-06,
      "loss": 0.8067,
      "step": 12613
    },
    {
      "epoch": 1.0193337239136144,
      "grad_norm": 2.984839677810669,
      "learning_rate": 5.086044374436076e-06,
      "loss": 0.8352,
      "step": 12614
    },
    {
      "epoch": 1.019414533626942,
      "grad_norm": 2.7549664974212646,
      "learning_rate": 5.085390108457362e-06,
      "loss": 0.7722,
      "step": 12615
    },
    {
      "epoch": 1.0194953433402696,
      "grad_norm": 2.6267752647399902,
      "learning_rate": 5.08473584101612e-06,
      "loss": 0.8012,
      "step": 12616
    },
    {
      "epoch": 1.019576153053597,
      "grad_norm": 2.9457740783691406,
      "learning_rate": 5.084081572123558e-06,
      "loss": 0.8584,
      "step": 12617
    },
    {
      "epoch": 1.0196569627669245,
      "grad_norm": 2.533068895339966,
      "learning_rate": 5.083427301790881e-06,
      "loss": 0.8131,
      "step": 12618
    },
    {
      "epoch": 1.019737772480252,
      "grad_norm": 2.8118538856506348,
      "learning_rate": 5.082773030029297e-06,
      "loss": 0.7693,
      "step": 12619
    },
    {
      "epoch": 1.0198185821935797,
      "grad_norm": 2.4821224212646484,
      "learning_rate": 5.082118756850007e-06,
      "loss": 0.9049,
      "step": 12620
    },
    {
      "epoch": 1.0198993919069073,
      "grad_norm": 2.4987287521362305,
      "learning_rate": 5.081464482264223e-06,
      "loss": 0.8304,
      "step": 12621
    },
    {
      "epoch": 1.0199802016202348,
      "grad_norm": 2.7845113277435303,
      "learning_rate": 5.080810206283147e-06,
      "loss": 0.791,
      "step": 12622
    },
    {
      "epoch": 1.0200610113335622,
      "grad_norm": 2.3762574195861816,
      "learning_rate": 5.080155928917986e-06,
      "loss": 0.8095,
      "step": 12623
    },
    {
      "epoch": 1.0201418210468898,
      "grad_norm": 2.5248916149139404,
      "learning_rate": 5.079501650179948e-06,
      "loss": 0.7802,
      "step": 12624
    },
    {
      "epoch": 1.0202226307602174,
      "grad_norm": 2.6407883167266846,
      "learning_rate": 5.078847370080239e-06,
      "loss": 0.7451,
      "step": 12625
    },
    {
      "epoch": 1.020303440473545,
      "grad_norm": 2.595813512802124,
      "learning_rate": 5.078193088630062e-06,
      "loss": 0.8572,
      "step": 12626
    },
    {
      "epoch": 1.0203842501868725,
      "grad_norm": 2.273280382156372,
      "learning_rate": 5.077538805840629e-06,
      "loss": 0.8425,
      "step": 12627
    },
    {
      "epoch": 1.0204650599002,
      "grad_norm": 2.8976593017578125,
      "learning_rate": 5.07688452172314e-06,
      "loss": 0.824,
      "step": 12628
    },
    {
      "epoch": 1.0205458696135274,
      "grad_norm": 2.5143322944641113,
      "learning_rate": 5.076230236288805e-06,
      "loss": 0.8648,
      "step": 12629
    },
    {
      "epoch": 1.020626679326855,
      "grad_norm": 2.59073805809021,
      "learning_rate": 5.07557594954883e-06,
      "loss": 0.779,
      "step": 12630
    },
    {
      "epoch": 1.0207074890401826,
      "grad_norm": 2.656137704849243,
      "learning_rate": 5.07492166151442e-06,
      "loss": 0.8114,
      "step": 12631
    },
    {
      "epoch": 1.0207882987535102,
      "grad_norm": 2.878208875656128,
      "learning_rate": 5.074267372196784e-06,
      "loss": 0.8133,
      "step": 12632
    },
    {
      "epoch": 1.0208691084668378,
      "grad_norm": 2.3239173889160156,
      "learning_rate": 5.0736130816071265e-06,
      "loss": 0.8897,
      "step": 12633
    },
    {
      "epoch": 1.0209499181801653,
      "grad_norm": 2.847738027572632,
      "learning_rate": 5.072958789756653e-06,
      "loss": 0.8624,
      "step": 12634
    },
    {
      "epoch": 1.0210307278934927,
      "grad_norm": 2.5505475997924805,
      "learning_rate": 5.0723044966565716e-06,
      "loss": 0.8422,
      "step": 12635
    },
    {
      "epoch": 1.0211115376068203,
      "grad_norm": 2.650987148284912,
      "learning_rate": 5.071650202318088e-06,
      "loss": 0.791,
      "step": 12636
    },
    {
      "epoch": 1.0211923473201479,
      "grad_norm": 3.117178440093994,
      "learning_rate": 5.070995906752409e-06,
      "loss": 0.8437,
      "step": 12637
    },
    {
      "epoch": 1.0212731570334754,
      "grad_norm": 2.452789068222046,
      "learning_rate": 5.0703416099707404e-06,
      "loss": 0.8355,
      "step": 12638
    },
    {
      "epoch": 1.021353966746803,
      "grad_norm": 2.923870801925659,
      "learning_rate": 5.0696873119842906e-06,
      "loss": 0.833,
      "step": 12639
    },
    {
      "epoch": 1.0214347764601306,
      "grad_norm": 2.6646676063537598,
      "learning_rate": 5.0690330128042645e-06,
      "loss": 0.7995,
      "step": 12640
    },
    {
      "epoch": 1.021515586173458,
      "grad_norm": 2.2564642429351807,
      "learning_rate": 5.0683787124418695e-06,
      "loss": 0.775,
      "step": 12641
    },
    {
      "epoch": 1.0215963958867855,
      "grad_norm": 2.532127618789673,
      "learning_rate": 5.06772441090831e-06,
      "loss": 0.7502,
      "step": 12642
    },
    {
      "epoch": 1.021677205600113,
      "grad_norm": 2.6183149814605713,
      "learning_rate": 5.067070108214795e-06,
      "loss": 0.7897,
      "step": 12643
    },
    {
      "epoch": 1.0217580153134407,
      "grad_norm": 2.713432550430298,
      "learning_rate": 5.066415804372532e-06,
      "loss": 0.8426,
      "step": 12644
    },
    {
      "epoch": 1.0218388250267683,
      "grad_norm": 2.9958345890045166,
      "learning_rate": 5.065761499392725e-06,
      "loss": 0.7845,
      "step": 12645
    },
    {
      "epoch": 1.0219196347400958,
      "grad_norm": 3.305631637573242,
      "learning_rate": 5.065107193286581e-06,
      "loss": 0.8995,
      "step": 12646
    },
    {
      "epoch": 1.0220004444534232,
      "grad_norm": 2.428516149520874,
      "learning_rate": 5.0644528860653096e-06,
      "loss": 0.8722,
      "step": 12647
    },
    {
      "epoch": 1.0220812541667508,
      "grad_norm": 2.7107222080230713,
      "learning_rate": 5.0637985777401145e-06,
      "loss": 0.8177,
      "step": 12648
    },
    {
      "epoch": 1.0221620638800784,
      "grad_norm": 2.5419907569885254,
      "learning_rate": 5.063144268322203e-06,
      "loss": 0.7565,
      "step": 12649
    },
    {
      "epoch": 1.022242873593406,
      "grad_norm": 2.2726142406463623,
      "learning_rate": 5.0624899578227825e-06,
      "loss": 0.7221,
      "step": 12650
    },
    {
      "epoch": 1.0223236833067335,
      "grad_norm": 2.45625901222229,
      "learning_rate": 5.061835646253059e-06,
      "loss": 0.7756,
      "step": 12651
    },
    {
      "epoch": 1.022404493020061,
      "grad_norm": 3.0126771926879883,
      "learning_rate": 5.061181333624241e-06,
      "loss": 0.8195,
      "step": 12652
    },
    {
      "epoch": 1.0224853027333884,
      "grad_norm": 2.5299274921417236,
      "learning_rate": 5.060527019947533e-06,
      "loss": 0.8244,
      "step": 12653
    },
    {
      "epoch": 1.022566112446716,
      "grad_norm": 2.408522367477417,
      "learning_rate": 5.059872705234144e-06,
      "loss": 0.8256,
      "step": 12654
    },
    {
      "epoch": 1.0226469221600436,
      "grad_norm": 2.281392812728882,
      "learning_rate": 5.059218389495277e-06,
      "loss": 0.8795,
      "step": 12655
    },
    {
      "epoch": 1.0227277318733712,
      "grad_norm": 3.077888011932373,
      "learning_rate": 5.058564072742145e-06,
      "loss": 0.773,
      "step": 12656
    },
    {
      "epoch": 1.0228085415866988,
      "grad_norm": 3.169032096862793,
      "learning_rate": 5.057909754985948e-06,
      "loss": 0.7304,
      "step": 12657
    },
    {
      "epoch": 1.0228893513000263,
      "grad_norm": 2.776294469833374,
      "learning_rate": 5.057255436237899e-06,
      "loss": 0.8589,
      "step": 12658
    },
    {
      "epoch": 1.0229701610133537,
      "grad_norm": 2.7166430950164795,
      "learning_rate": 5.0566011165092e-06,
      "loss": 0.7925,
      "step": 12659
    },
    {
      "epoch": 1.0230509707266813,
      "grad_norm": 3.373649835586548,
      "learning_rate": 5.0559467958110635e-06,
      "loss": 0.7587,
      "step": 12660
    },
    {
      "epoch": 1.0231317804400089,
      "grad_norm": 2.545762538909912,
      "learning_rate": 5.05529247415469e-06,
      "loss": 0.799,
      "step": 12661
    },
    {
      "epoch": 1.0232125901533364,
      "grad_norm": 2.73091721534729,
      "learning_rate": 5.0546381515512896e-06,
      "loss": 0.8279,
      "step": 12662
    },
    {
      "epoch": 1.023293399866664,
      "grad_norm": 2.65697979927063,
      "learning_rate": 5.0539838280120715e-06,
      "loss": 0.8238,
      "step": 12663
    },
    {
      "epoch": 1.0233742095799916,
      "grad_norm": 2.65800142288208,
      "learning_rate": 5.053329503548239e-06,
      "loss": 0.785,
      "step": 12664
    },
    {
      "epoch": 1.0234550192933192,
      "grad_norm": 3.661139965057373,
      "learning_rate": 5.052675178170999e-06,
      "loss": 0.8133,
      "step": 12665
    },
    {
      "epoch": 1.0235358290066465,
      "grad_norm": 2.6694295406341553,
      "learning_rate": 5.052020851891563e-06,
      "loss": 0.8975,
      "step": 12666
    },
    {
      "epoch": 1.023616638719974,
      "grad_norm": 2.9427649974823,
      "learning_rate": 5.051366524721133e-06,
      "loss": 0.8424,
      "step": 12667
    },
    {
      "epoch": 1.0236974484333017,
      "grad_norm": 2.6894493103027344,
      "learning_rate": 5.050712196670918e-06,
      "loss": 0.8914,
      "step": 12668
    },
    {
      "epoch": 1.0237782581466293,
      "grad_norm": 2.7558834552764893,
      "learning_rate": 5.050057867752126e-06,
      "loss": 0.8858,
      "step": 12669
    },
    {
      "epoch": 1.0238590678599568,
      "grad_norm": 3.1157639026641846,
      "learning_rate": 5.049403537975964e-06,
      "loss": 0.817,
      "step": 12670
    },
    {
      "epoch": 1.0239398775732842,
      "grad_norm": 2.6721572875976562,
      "learning_rate": 5.048749207353636e-06,
      "loss": 0.8365,
      "step": 12671
    },
    {
      "epoch": 1.0240206872866118,
      "grad_norm": 2.8743832111358643,
      "learning_rate": 5.048094875896354e-06,
      "loss": 0.8592,
      "step": 12672
    },
    {
      "epoch": 1.0241014969999394,
      "grad_norm": 2.6651179790496826,
      "learning_rate": 5.047440543615321e-06,
      "loss": 0.702,
      "step": 12673
    },
    {
      "epoch": 1.024182306713267,
      "grad_norm": 2.6311092376708984,
      "learning_rate": 5.0467862105217455e-06,
      "loss": 0.7006,
      "step": 12674
    },
    {
      "epoch": 1.0242631164265945,
      "grad_norm": 2.699047803878784,
      "learning_rate": 5.0461318766268364e-06,
      "loss": 0.8941,
      "step": 12675
    },
    {
      "epoch": 1.024343926139922,
      "grad_norm": 2.647075891494751,
      "learning_rate": 5.045477541941798e-06,
      "loss": 0.726,
      "step": 12676
    },
    {
      "epoch": 1.0244247358532497,
      "grad_norm": 2.4319651126861572,
      "learning_rate": 5.044823206477839e-06,
      "loss": 0.9222,
      "step": 12677
    },
    {
      "epoch": 1.024505545566577,
      "grad_norm": 2.899913787841797,
      "learning_rate": 5.044168870246166e-06,
      "loss": 0.9301,
      "step": 12678
    },
    {
      "epoch": 1.0245863552799046,
      "grad_norm": 2.800304412841797,
      "learning_rate": 5.043514533257987e-06,
      "loss": 0.7484,
      "step": 12679
    },
    {
      "epoch": 1.0246671649932322,
      "grad_norm": 2.4168550968170166,
      "learning_rate": 5.042860195524509e-06,
      "loss": 0.8403,
      "step": 12680
    },
    {
      "epoch": 1.0247479747065598,
      "grad_norm": 2.4322919845581055,
      "learning_rate": 5.04220585705694e-06,
      "loss": 0.8399,
      "step": 12681
    },
    {
      "epoch": 1.0248287844198873,
      "grad_norm": 2.780128002166748,
      "learning_rate": 5.0415515178664846e-06,
      "loss": 0.7998,
      "step": 12682
    },
    {
      "epoch": 1.024909594133215,
      "grad_norm": 2.586167573928833,
      "learning_rate": 5.040897177964353e-06,
      "loss": 0.7803,
      "step": 12683
    },
    {
      "epoch": 1.0249904038465423,
      "grad_norm": 3.1910691261291504,
      "learning_rate": 5.040242837361751e-06,
      "loss": 0.7963,
      "step": 12684
    },
    {
      "epoch": 1.0250712135598699,
      "grad_norm": 2.624464511871338,
      "learning_rate": 5.039588496069883e-06,
      "loss": 0.8317,
      "step": 12685
    },
    {
      "epoch": 1.0251520232731974,
      "grad_norm": 2.69215989112854,
      "learning_rate": 5.038934154099964e-06,
      "loss": 0.7799,
      "step": 12686
    },
    {
      "epoch": 1.025232832986525,
      "grad_norm": 2.8588006496429443,
      "learning_rate": 5.038279811463193e-06,
      "loss": 0.9083,
      "step": 12687
    },
    {
      "epoch": 1.0253136426998526,
      "grad_norm": 2.506113290786743,
      "learning_rate": 5.037625468170783e-06,
      "loss": 0.8346,
      "step": 12688
    },
    {
      "epoch": 1.0253944524131802,
      "grad_norm": 2.533895492553711,
      "learning_rate": 5.0369711242339396e-06,
      "loss": 0.8112,
      "step": 12689
    },
    {
      "epoch": 1.0254752621265075,
      "grad_norm": 2.662245512008667,
      "learning_rate": 5.036316779663869e-06,
      "loss": 0.8136,
      "step": 12690
    },
    {
      "epoch": 1.025556071839835,
      "grad_norm": 2.8152735233306885,
      "learning_rate": 5.0356624344717785e-06,
      "loss": 0.9144,
      "step": 12691
    },
    {
      "epoch": 1.0256368815531627,
      "grad_norm": 2.935899496078491,
      "learning_rate": 5.035008088668879e-06,
      "loss": 0.8417,
      "step": 12692
    },
    {
      "epoch": 1.0257176912664903,
      "grad_norm": 2.5224225521087646,
      "learning_rate": 5.034353742266372e-06,
      "loss": 0.8094,
      "step": 12693
    },
    {
      "epoch": 1.0257985009798178,
      "grad_norm": 2.627077341079712,
      "learning_rate": 5.033699395275471e-06,
      "loss": 0.8952,
      "step": 12694
    },
    {
      "epoch": 1.0258793106931454,
      "grad_norm": 2.5839152336120605,
      "learning_rate": 5.033045047707379e-06,
      "loss": 0.9106,
      "step": 12695
    },
    {
      "epoch": 1.0259601204064728,
      "grad_norm": 2.659386157989502,
      "learning_rate": 5.0323906995733055e-06,
      "loss": 0.8824,
      "step": 12696
    },
    {
      "epoch": 1.0260409301198004,
      "grad_norm": 2.8633780479431152,
      "learning_rate": 5.031736350884456e-06,
      "loss": 0.9488,
      "step": 12697
    },
    {
      "epoch": 1.026121739833128,
      "grad_norm": 2.4078078269958496,
      "learning_rate": 5.03108200165204e-06,
      "loss": 0.8426,
      "step": 12698
    },
    {
      "epoch": 1.0262025495464555,
      "grad_norm": 2.6873888969421387,
      "learning_rate": 5.030427651887264e-06,
      "loss": 0.7887,
      "step": 12699
    },
    {
      "epoch": 1.026283359259783,
      "grad_norm": 3.359510898590088,
      "learning_rate": 5.029773301601338e-06,
      "loss": 0.7892,
      "step": 12700
    },
    {
      "epoch": 1.0263641689731107,
      "grad_norm": 2.856140375137329,
      "learning_rate": 5.0291189508054624e-06,
      "loss": 0.8906,
      "step": 12701
    },
    {
      "epoch": 1.026444978686438,
      "grad_norm": 2.6707277297973633,
      "learning_rate": 5.028464599510853e-06,
      "loss": 0.8341,
      "step": 12702
    },
    {
      "epoch": 1.0265257883997656,
      "grad_norm": 2.730442523956299,
      "learning_rate": 5.027810247728712e-06,
      "loss": 0.7867,
      "step": 12703
    },
    {
      "epoch": 1.0266065981130932,
      "grad_norm": 2.8404700756073,
      "learning_rate": 5.027155895470248e-06,
      "loss": 0.8352,
      "step": 12704
    },
    {
      "epoch": 1.0266874078264208,
      "grad_norm": 2.6206679344177246,
      "learning_rate": 5.0265015427466705e-06,
      "loss": 0.8447,
      "step": 12705
    },
    {
      "epoch": 1.0267682175397483,
      "grad_norm": 2.6233479976654053,
      "learning_rate": 5.025847189569183e-06,
      "loss": 0.8776,
      "step": 12706
    },
    {
      "epoch": 1.026849027253076,
      "grad_norm": 2.4091241359710693,
      "learning_rate": 5.025192835948996e-06,
      "loss": 0.8689,
      "step": 12707
    },
    {
      "epoch": 1.0269298369664033,
      "grad_norm": 2.6468141078948975,
      "learning_rate": 5.024538481897319e-06,
      "loss": 0.8165,
      "step": 12708
    },
    {
      "epoch": 1.0270106466797309,
      "grad_norm": 2.6091692447662354,
      "learning_rate": 5.0238841274253545e-06,
      "loss": 0.7286,
      "step": 12709
    },
    {
      "epoch": 1.0270914563930584,
      "grad_norm": 2.8415677547454834,
      "learning_rate": 5.023229772544313e-06,
      "loss": 0.9361,
      "step": 12710
    },
    {
      "epoch": 1.027172266106386,
      "grad_norm": 2.6208531856536865,
      "learning_rate": 5.022575417265402e-06,
      "loss": 0.7295,
      "step": 12711
    },
    {
      "epoch": 1.0272530758197136,
      "grad_norm": 2.9239227771759033,
      "learning_rate": 5.0219210615998274e-06,
      "loss": 0.8331,
      "step": 12712
    },
    {
      "epoch": 1.0273338855330412,
      "grad_norm": 2.4374606609344482,
      "learning_rate": 5.021266705558797e-06,
      "loss": 0.7484,
      "step": 12713
    },
    {
      "epoch": 1.0274146952463685,
      "grad_norm": 2.185107946395874,
      "learning_rate": 5.020612349153521e-06,
      "loss": 0.8039,
      "step": 12714
    },
    {
      "epoch": 1.027495504959696,
      "grad_norm": 2.581763744354248,
      "learning_rate": 5.019957992395204e-06,
      "loss": 0.7838,
      "step": 12715
    },
    {
      "epoch": 1.0275763146730237,
      "grad_norm": 2.4704651832580566,
      "learning_rate": 5.0193036352950556e-06,
      "loss": 0.8881,
      "step": 12716
    },
    {
      "epoch": 1.0276571243863513,
      "grad_norm": 2.8464388847351074,
      "learning_rate": 5.0186492778642815e-06,
      "loss": 0.807,
      "step": 12717
    },
    {
      "epoch": 1.0277379340996788,
      "grad_norm": 2.9068238735198975,
      "learning_rate": 5.0179949201140905e-06,
      "loss": 0.8626,
      "step": 12718
    },
    {
      "epoch": 1.0278187438130064,
      "grad_norm": 2.8587801456451416,
      "learning_rate": 5.0173405620556905e-06,
      "loss": 0.892,
      "step": 12719
    },
    {
      "epoch": 1.0278995535263338,
      "grad_norm": 2.780148983001709,
      "learning_rate": 5.016686203700288e-06,
      "loss": 0.851,
      "step": 12720
    },
    {
      "epoch": 1.0279803632396614,
      "grad_norm": 2.440889596939087,
      "learning_rate": 5.016031845059089e-06,
      "loss": 0.8264,
      "step": 12721
    },
    {
      "epoch": 1.028061172952989,
      "grad_norm": 2.604332447052002,
      "learning_rate": 5.015377486143305e-06,
      "loss": 0.8601,
      "step": 12722
    },
    {
      "epoch": 1.0281419826663165,
      "grad_norm": 2.7778687477111816,
      "learning_rate": 5.014723126964143e-06,
      "loss": 0.7743,
      "step": 12723
    },
    {
      "epoch": 1.028222792379644,
      "grad_norm": 2.172650098800659,
      "learning_rate": 5.014068767532806e-06,
      "loss": 0.8466,
      "step": 12724
    },
    {
      "epoch": 1.0283036020929717,
      "grad_norm": 2.3984363079071045,
      "learning_rate": 5.013414407860507e-06,
      "loss": 0.8429,
      "step": 12725
    },
    {
      "epoch": 1.028384411806299,
      "grad_norm": 2.6857895851135254,
      "learning_rate": 5.01276004795845e-06,
      "loss": 0.8363,
      "step": 12726
    },
    {
      "epoch": 1.0284652215196266,
      "grad_norm": 2.552767276763916,
      "learning_rate": 5.0121056878378475e-06,
      "loss": 0.7717,
      "step": 12727
    },
    {
      "epoch": 1.0285460312329542,
      "grad_norm": 2.4570114612579346,
      "learning_rate": 5.011451327509901e-06,
      "loss": 0.9366,
      "step": 12728
    },
    {
      "epoch": 1.0286268409462818,
      "grad_norm": 2.271482467651367,
      "learning_rate": 5.010796966985822e-06,
      "loss": 0.8003,
      "step": 12729
    },
    {
      "epoch": 1.0287076506596093,
      "grad_norm": 2.92912220954895,
      "learning_rate": 5.010142606276816e-06,
      "loss": 0.864,
      "step": 12730
    },
    {
      "epoch": 1.028788460372937,
      "grad_norm": 2.7757370471954346,
      "learning_rate": 5.009488245394092e-06,
      "loss": 0.7798,
      "step": 12731
    },
    {
      "epoch": 1.0288692700862643,
      "grad_norm": 2.60307240486145,
      "learning_rate": 5.008833884348856e-06,
      "loss": 0.7594,
      "step": 12732
    },
    {
      "epoch": 1.0289500797995919,
      "grad_norm": 2.7053911685943604,
      "learning_rate": 5.008179523152319e-06,
      "loss": 0.8843,
      "step": 12733
    },
    {
      "epoch": 1.0290308895129194,
      "grad_norm": 2.5982367992401123,
      "learning_rate": 5.007525161815685e-06,
      "loss": 0.8155,
      "step": 12734
    },
    {
      "epoch": 1.029111699226247,
      "grad_norm": 2.5147147178649902,
      "learning_rate": 5.006870800350163e-06,
      "loss": 0.8066,
      "step": 12735
    },
    {
      "epoch": 1.0291925089395746,
      "grad_norm": 2.2974839210510254,
      "learning_rate": 5.0062164387669605e-06,
      "loss": 0.8844,
      "step": 12736
    },
    {
      "epoch": 1.0292733186529022,
      "grad_norm": 2.545964479446411,
      "learning_rate": 5.005562077077287e-06,
      "loss": 0.8175,
      "step": 12737
    },
    {
      "epoch": 1.0293541283662295,
      "grad_norm": 2.3717331886291504,
      "learning_rate": 5.004907715292346e-06,
      "loss": 0.7335,
      "step": 12738
    },
    {
      "epoch": 1.029434938079557,
      "grad_norm": 2.3097434043884277,
      "learning_rate": 5.004253353423351e-06,
      "loss": 0.8945,
      "step": 12739
    },
    {
      "epoch": 1.0295157477928847,
      "grad_norm": 2.6539955139160156,
      "learning_rate": 5.003598991481503e-06,
      "loss": 0.8038,
      "step": 12740
    },
    {
      "epoch": 1.0295965575062123,
      "grad_norm": 2.718130350112915,
      "learning_rate": 5.0029446294780146e-06,
      "loss": 0.813,
      "step": 12741
    },
    {
      "epoch": 1.0296773672195398,
      "grad_norm": 2.3481106758117676,
      "learning_rate": 5.0022902674240915e-06,
      "loss": 0.7984,
      "step": 12742
    },
    {
      "epoch": 1.0297581769328674,
      "grad_norm": 2.7545816898345947,
      "learning_rate": 5.0016359053309415e-06,
      "loss": 0.8325,
      "step": 12743
    },
    {
      "epoch": 1.0298389866461948,
      "grad_norm": 2.639677047729492,
      "learning_rate": 5.000981543209773e-06,
      "loss": 0.914,
      "step": 12744
    },
    {
      "epoch": 1.0299197963595224,
      "grad_norm": 2.8748130798339844,
      "learning_rate": 5.000327181071793e-06,
      "loss": 0.7105,
      "step": 12745
    },
    {
      "epoch": 1.03000060607285,
      "grad_norm": 2.184514284133911,
      "learning_rate": 4.9996728189282075e-06,
      "loss": 0.8635,
      "step": 12746
    },
    {
      "epoch": 1.0300814157861775,
      "grad_norm": 3.0235283374786377,
      "learning_rate": 4.9990184567902275e-06,
      "loss": 0.8444,
      "step": 12747
    },
    {
      "epoch": 1.030162225499505,
      "grad_norm": 2.3344459533691406,
      "learning_rate": 4.998364094669059e-06,
      "loss": 0.8783,
      "step": 12748
    },
    {
      "epoch": 1.0302430352128327,
      "grad_norm": 2.846672296524048,
      "learning_rate": 4.997709732575909e-06,
      "loss": 0.8181,
      "step": 12749
    },
    {
      "epoch": 1.03032384492616,
      "grad_norm": 2.2562077045440674,
      "learning_rate": 4.997055370521985e-06,
      "loss": 0.8856,
      "step": 12750
    },
    {
      "epoch": 1.0304046546394876,
      "grad_norm": 2.255314826965332,
      "learning_rate": 4.996401008518499e-06,
      "loss": 0.6853,
      "step": 12751
    },
    {
      "epoch": 1.0304854643528152,
      "grad_norm": 3.0052249431610107,
      "learning_rate": 4.995746646576651e-06,
      "loss": 0.6795,
      "step": 12752
    },
    {
      "epoch": 1.0305662740661428,
      "grad_norm": 3.18621826171875,
      "learning_rate": 4.995092284707654e-06,
      "loss": 0.7444,
      "step": 12753
    },
    {
      "epoch": 1.0306470837794703,
      "grad_norm": 3.149110794067383,
      "learning_rate": 4.9944379229227155e-06,
      "loss": 0.7594,
      "step": 12754
    },
    {
      "epoch": 1.030727893492798,
      "grad_norm": 2.6996846199035645,
      "learning_rate": 4.993783561233041e-06,
      "loss": 0.7711,
      "step": 12755
    },
    {
      "epoch": 1.0308087032061253,
      "grad_norm": 2.3134026527404785,
      "learning_rate": 4.993129199649838e-06,
      "loss": 0.8269,
      "step": 12756
    },
    {
      "epoch": 1.0308895129194529,
      "grad_norm": 2.3625435829162598,
      "learning_rate": 4.992474838184318e-06,
      "loss": 0.8723,
      "step": 12757
    },
    {
      "epoch": 1.0309703226327804,
      "grad_norm": 2.7127509117126465,
      "learning_rate": 4.991820476847683e-06,
      "loss": 0.8032,
      "step": 12758
    },
    {
      "epoch": 1.031051132346108,
      "grad_norm": 3.0470070838928223,
      "learning_rate": 4.9911661156511445e-06,
      "loss": 0.887,
      "step": 12759
    },
    {
      "epoch": 1.0311319420594356,
      "grad_norm": 2.3634185791015625,
      "learning_rate": 4.990511754605911e-06,
      "loss": 0.7546,
      "step": 12760
    },
    {
      "epoch": 1.0312127517727632,
      "grad_norm": 2.869631767272949,
      "learning_rate": 4.989857393723187e-06,
      "loss": 0.8403,
      "step": 12761
    },
    {
      "epoch": 1.0312935614860905,
      "grad_norm": 2.5108728408813477,
      "learning_rate": 4.989203033014179e-06,
      "loss": 0.7941,
      "step": 12762
    },
    {
      "epoch": 1.031374371199418,
      "grad_norm": 2.6491148471832275,
      "learning_rate": 4.9885486724901e-06,
      "loss": 0.811,
      "step": 12763
    },
    {
      "epoch": 1.0314551809127457,
      "grad_norm": 2.439948797225952,
      "learning_rate": 4.987894312162154e-06,
      "loss": 0.8451,
      "step": 12764
    },
    {
      "epoch": 1.0315359906260733,
      "grad_norm": 2.49680495262146,
      "learning_rate": 4.987239952041549e-06,
      "loss": 0.8132,
      "step": 12765
    },
    {
      "epoch": 1.0316168003394008,
      "grad_norm": 2.9315829277038574,
      "learning_rate": 4.9865855921394945e-06,
      "loss": 0.7809,
      "step": 12766
    },
    {
      "epoch": 1.0316976100527284,
      "grad_norm": 2.4637374877929688,
      "learning_rate": 4.985931232467195e-06,
      "loss": 0.8845,
      "step": 12767
    },
    {
      "epoch": 1.0317784197660558,
      "grad_norm": 2.5532619953155518,
      "learning_rate": 4.985276873035859e-06,
      "loss": 0.8286,
      "step": 12768
    },
    {
      "epoch": 1.0318592294793834,
      "grad_norm": 2.7578811645507812,
      "learning_rate": 4.984622513856696e-06,
      "loss": 0.8114,
      "step": 12769
    },
    {
      "epoch": 1.031940039192711,
      "grad_norm": 2.925610065460205,
      "learning_rate": 4.983968154940912e-06,
      "loss": 0.7349,
      "step": 12770
    },
    {
      "epoch": 1.0320208489060385,
      "grad_norm": 2.2802493572235107,
      "learning_rate": 4.983313796299714e-06,
      "loss": 0.8446,
      "step": 12771
    },
    {
      "epoch": 1.032101658619366,
      "grad_norm": 2.626570224761963,
      "learning_rate": 4.982659437944313e-06,
      "loss": 0.8845,
      "step": 12772
    },
    {
      "epoch": 1.0321824683326937,
      "grad_norm": 2.2684454917907715,
      "learning_rate": 4.98200507988591e-06,
      "loss": 0.8328,
      "step": 12773
    },
    {
      "epoch": 1.032263278046021,
      "grad_norm": 2.4540622234344482,
      "learning_rate": 4.981350722135719e-06,
      "loss": 0.815,
      "step": 12774
    },
    {
      "epoch": 1.0323440877593486,
      "grad_norm": 2.575643301010132,
      "learning_rate": 4.980696364704945e-06,
      "loss": 0.8865,
      "step": 12775
    },
    {
      "epoch": 1.0324248974726762,
      "grad_norm": 2.8141043186187744,
      "learning_rate": 4.980042007604797e-06,
      "loss": 0.8462,
      "step": 12776
    },
    {
      "epoch": 1.0325057071860038,
      "grad_norm": 2.6783711910247803,
      "learning_rate": 4.979387650846481e-06,
      "loss": 0.7909,
      "step": 12777
    },
    {
      "epoch": 1.0325865168993313,
      "grad_norm": 2.9930763244628906,
      "learning_rate": 4.978733294441203e-06,
      "loss": 0.8268,
      "step": 12778
    },
    {
      "epoch": 1.032667326612659,
      "grad_norm": 2.710965633392334,
      "learning_rate": 4.978078938400174e-06,
      "loss": 0.7757,
      "step": 12779
    },
    {
      "epoch": 1.0327481363259863,
      "grad_norm": 3.494272232055664,
      "learning_rate": 4.9774245827346e-06,
      "loss": 0.8253,
      "step": 12780
    },
    {
      "epoch": 1.0328289460393139,
      "grad_norm": 2.375830888748169,
      "learning_rate": 4.9767702274556885e-06,
      "loss": 0.899,
      "step": 12781
    },
    {
      "epoch": 1.0329097557526414,
      "grad_norm": 2.467590093612671,
      "learning_rate": 4.976115872574648e-06,
      "loss": 0.8084,
      "step": 12782
    },
    {
      "epoch": 1.032990565465969,
      "grad_norm": 2.582261800765991,
      "learning_rate": 4.975461518102682e-06,
      "loss": 0.8396,
      "step": 12783
    },
    {
      "epoch": 1.0330713751792966,
      "grad_norm": 2.7683613300323486,
      "learning_rate": 4.974807164051003e-06,
      "loss": 0.9251,
      "step": 12784
    },
    {
      "epoch": 1.0331521848926242,
      "grad_norm": 3.1756012439727783,
      "learning_rate": 4.974152810430818e-06,
      "loss": 0.8297,
      "step": 12785
    },
    {
      "epoch": 1.0332329946059517,
      "grad_norm": 2.457958459854126,
      "learning_rate": 4.973498457253332e-06,
      "loss": 0.9918,
      "step": 12786
    },
    {
      "epoch": 1.033313804319279,
      "grad_norm": 2.488844871520996,
      "learning_rate": 4.972844104529753e-06,
      "loss": 0.7827,
      "step": 12787
    },
    {
      "epoch": 1.0333946140326067,
      "grad_norm": 2.1935784816741943,
      "learning_rate": 4.972189752271291e-06,
      "loss": 0.7634,
      "step": 12788
    },
    {
      "epoch": 1.0334754237459343,
      "grad_norm": 2.20938777923584,
      "learning_rate": 4.971535400489148e-06,
      "loss": 0.8776,
      "step": 12789
    },
    {
      "epoch": 1.0335562334592618,
      "grad_norm": 2.3523964881896973,
      "learning_rate": 4.970881049194538e-06,
      "loss": 0.7805,
      "step": 12790
    },
    {
      "epoch": 1.0336370431725894,
      "grad_norm": 2.588494300842285,
      "learning_rate": 4.970226698398666e-06,
      "loss": 0.8473,
      "step": 12791
    },
    {
      "epoch": 1.0337178528859168,
      "grad_norm": 2.520622730255127,
      "learning_rate": 4.9695723481127374e-06,
      "loss": 0.7837,
      "step": 12792
    },
    {
      "epoch": 1.0337986625992444,
      "grad_norm": 2.6883108615875244,
      "learning_rate": 4.968917998347961e-06,
      "loss": 0.8244,
      "step": 12793
    },
    {
      "epoch": 1.033879472312572,
      "grad_norm": 2.437628746032715,
      "learning_rate": 4.968263649115546e-06,
      "loss": 0.7811,
      "step": 12794
    },
    {
      "epoch": 1.0339602820258995,
      "grad_norm": 2.5353872776031494,
      "learning_rate": 4.967609300426697e-06,
      "loss": 0.8087,
      "step": 12795
    },
    {
      "epoch": 1.034041091739227,
      "grad_norm": 2.777150869369507,
      "learning_rate": 4.966954952292623e-06,
      "loss": 0.7576,
      "step": 12796
    },
    {
      "epoch": 1.0341219014525547,
      "grad_norm": 3.088149309158325,
      "learning_rate": 4.966300604724532e-06,
      "loss": 0.7606,
      "step": 12797
    },
    {
      "epoch": 1.0342027111658822,
      "grad_norm": 2.8645825386047363,
      "learning_rate": 4.965646257733629e-06,
      "loss": 0.8096,
      "step": 12798
    },
    {
      "epoch": 1.0342835208792096,
      "grad_norm": 2.754312038421631,
      "learning_rate": 4.964991911331122e-06,
      "loss": 0.7841,
      "step": 12799
    },
    {
      "epoch": 1.0343643305925372,
      "grad_norm": 2.337752342224121,
      "learning_rate": 4.9643375655282214e-06,
      "loss": 0.8053,
      "step": 12800
    },
    {
      "epoch": 1.0344451403058648,
      "grad_norm": 2.6112146377563477,
      "learning_rate": 4.963683220336133e-06,
      "loss": 0.83,
      "step": 12801
    },
    {
      "epoch": 1.0345259500191923,
      "grad_norm": 2.616032600402832,
      "learning_rate": 4.963028875766062e-06,
      "loss": 0.7731,
      "step": 12802
    },
    {
      "epoch": 1.03460675973252,
      "grad_norm": 2.3240854740142822,
      "learning_rate": 4.9623745318292175e-06,
      "loss": 0.9083,
      "step": 12803
    },
    {
      "epoch": 1.0346875694458473,
      "grad_norm": 2.695591926574707,
      "learning_rate": 4.961720188536808e-06,
      "loss": 0.8537,
      "step": 12804
    },
    {
      "epoch": 1.0347683791591749,
      "grad_norm": 2.967604398727417,
      "learning_rate": 4.961065845900038e-06,
      "loss": 0.8413,
      "step": 12805
    },
    {
      "epoch": 1.0348491888725024,
      "grad_norm": 2.7116377353668213,
      "learning_rate": 4.960411503930117e-06,
      "loss": 0.7494,
      "step": 12806
    },
    {
      "epoch": 1.03492999858583,
      "grad_norm": 3.1584997177124023,
      "learning_rate": 4.959757162638253e-06,
      "loss": 0.7371,
      "step": 12807
    },
    {
      "epoch": 1.0350108082991576,
      "grad_norm": 2.4165217876434326,
      "learning_rate": 4.95910282203565e-06,
      "loss": 0.7482,
      "step": 12808
    },
    {
      "epoch": 1.0350916180124852,
      "grad_norm": 2.883049488067627,
      "learning_rate": 4.958448482133516e-06,
      "loss": 0.8778,
      "step": 12809
    },
    {
      "epoch": 1.0351724277258127,
      "grad_norm": 2.670912981033325,
      "learning_rate": 4.957794142943063e-06,
      "loss": 0.8581,
      "step": 12810
    },
    {
      "epoch": 1.03525323743914,
      "grad_norm": 2.74064302444458,
      "learning_rate": 4.957139804475492e-06,
      "loss": 0.843,
      "step": 12811
    },
    {
      "epoch": 1.0353340471524677,
      "grad_norm": 2.5970287322998047,
      "learning_rate": 4.956485466742014e-06,
      "loss": 0.7506,
      "step": 12812
    },
    {
      "epoch": 1.0354148568657953,
      "grad_norm": 2.4259068965911865,
      "learning_rate": 4.955831129753835e-06,
      "loss": 0.9008,
      "step": 12813
    },
    {
      "epoch": 1.0354956665791228,
      "grad_norm": 2.599001407623291,
      "learning_rate": 4.955176793522163e-06,
      "loss": 0.7608,
      "step": 12814
    },
    {
      "epoch": 1.0355764762924504,
      "grad_norm": 2.6436216831207275,
      "learning_rate": 4.954522458058203e-06,
      "loss": 0.8819,
      "step": 12815
    },
    {
      "epoch": 1.035657286005778,
      "grad_norm": 2.8188560009002686,
      "learning_rate": 4.953868123373167e-06,
      "loss": 0.6734,
      "step": 12816
    },
    {
      "epoch": 1.0357380957191054,
      "grad_norm": 3.691556692123413,
      "learning_rate": 4.953213789478255e-06,
      "loss": 0.7709,
      "step": 12817
    },
    {
      "epoch": 1.035818905432433,
      "grad_norm": 2.5950491428375244,
      "learning_rate": 4.95255945638468e-06,
      "loss": 0.8796,
      "step": 12818
    },
    {
      "epoch": 1.0358997151457605,
      "grad_norm": 2.524399995803833,
      "learning_rate": 4.951905124103648e-06,
      "loss": 0.8444,
      "step": 12819
    },
    {
      "epoch": 1.035980524859088,
      "grad_norm": 2.501587390899658,
      "learning_rate": 4.951250792646365e-06,
      "loss": 0.947,
      "step": 12820
    },
    {
      "epoch": 1.0360613345724157,
      "grad_norm": 2.551919937133789,
      "learning_rate": 4.950596462024037e-06,
      "loss": 0.8316,
      "step": 12821
    },
    {
      "epoch": 1.0361421442857432,
      "grad_norm": 2.523668050765991,
      "learning_rate": 4.9499421322478755e-06,
      "loss": 0.7983,
      "step": 12822
    },
    {
      "epoch": 1.0362229539990706,
      "grad_norm": 3.2756011486053467,
      "learning_rate": 4.9492878033290826e-06,
      "loss": 0.8154,
      "step": 12823
    },
    {
      "epoch": 1.0363037637123982,
      "grad_norm": 2.44270396232605,
      "learning_rate": 4.948633475278869e-06,
      "loss": 0.8198,
      "step": 12824
    },
    {
      "epoch": 1.0363845734257258,
      "grad_norm": 2.660534381866455,
      "learning_rate": 4.94797914810844e-06,
      "loss": 0.7953,
      "step": 12825
    },
    {
      "epoch": 1.0364653831390533,
      "grad_norm": 2.6788415908813477,
      "learning_rate": 4.947324821829002e-06,
      "loss": 0.6867,
      "step": 12826
    },
    {
      "epoch": 1.036546192852381,
      "grad_norm": 2.509154796600342,
      "learning_rate": 4.946670496451762e-06,
      "loss": 0.7959,
      "step": 12827
    },
    {
      "epoch": 1.0366270025657085,
      "grad_norm": 2.484999895095825,
      "learning_rate": 4.946016171987929e-06,
      "loss": 0.837,
      "step": 12828
    },
    {
      "epoch": 1.0367078122790359,
      "grad_norm": 2.3741610050201416,
      "learning_rate": 4.945361848448711e-06,
      "loss": 0.8531,
      "step": 12829
    },
    {
      "epoch": 1.0367886219923634,
      "grad_norm": 2.699448585510254,
      "learning_rate": 4.944707525845311e-06,
      "loss": 0.9241,
      "step": 12830
    },
    {
      "epoch": 1.036869431705691,
      "grad_norm": 2.320544958114624,
      "learning_rate": 4.944053204188937e-06,
      "loss": 0.868,
      "step": 12831
    },
    {
      "epoch": 1.0369502414190186,
      "grad_norm": 2.888483762741089,
      "learning_rate": 4.943398883490801e-06,
      "loss": 0.8137,
      "step": 12832
    },
    {
      "epoch": 1.0370310511323462,
      "grad_norm": 2.6746714115142822,
      "learning_rate": 4.942744563762102e-06,
      "loss": 0.8417,
      "step": 12833
    },
    {
      "epoch": 1.0371118608456737,
      "grad_norm": 2.575489044189453,
      "learning_rate": 4.942090245014052e-06,
      "loss": 0.8192,
      "step": 12834
    },
    {
      "epoch": 1.037192670559001,
      "grad_norm": 2.713181972503662,
      "learning_rate": 4.941435927257857e-06,
      "loss": 0.8827,
      "step": 12835
    },
    {
      "epoch": 1.0372734802723287,
      "grad_norm": 2.8507280349731445,
      "learning_rate": 4.940781610504724e-06,
      "loss": 0.9301,
      "step": 12836
    },
    {
      "epoch": 1.0373542899856563,
      "grad_norm": 2.50119686126709,
      "learning_rate": 4.940127294765857e-06,
      "loss": 0.9033,
      "step": 12837
    },
    {
      "epoch": 1.0374350996989838,
      "grad_norm": 2.282935857772827,
      "learning_rate": 4.939472980052468e-06,
      "loss": 0.8336,
      "step": 12838
    },
    {
      "epoch": 1.0375159094123114,
      "grad_norm": 2.280317544937134,
      "learning_rate": 4.9388186663757605e-06,
      "loss": 0.8113,
      "step": 12839
    },
    {
      "epoch": 1.037596719125639,
      "grad_norm": 2.6849544048309326,
      "learning_rate": 4.938164353746942e-06,
      "loss": 0.8208,
      "step": 12840
    },
    {
      "epoch": 1.0376775288389664,
      "grad_norm": 2.924696683883667,
      "learning_rate": 4.937510042177219e-06,
      "loss": 0.6863,
      "step": 12841
    },
    {
      "epoch": 1.037758338552294,
      "grad_norm": 2.3179073333740234,
      "learning_rate": 4.936855731677799e-06,
      "loss": 0.7647,
      "step": 12842
    },
    {
      "epoch": 1.0378391482656215,
      "grad_norm": 2.255683422088623,
      "learning_rate": 4.936201422259886e-06,
      "loss": 0.8274,
      "step": 12843
    },
    {
      "epoch": 1.037919957978949,
      "grad_norm": 2.973609209060669,
      "learning_rate": 4.935547113934692e-06,
      "loss": 0.8438,
      "step": 12844
    },
    {
      "epoch": 1.0380007676922767,
      "grad_norm": 2.8407390117645264,
      "learning_rate": 4.93489280671342e-06,
      "loss": 0.8347,
      "step": 12845
    },
    {
      "epoch": 1.0380815774056042,
      "grad_norm": 2.5327024459838867,
      "learning_rate": 4.9342385006072765e-06,
      "loss": 0.7966,
      "step": 12846
    },
    {
      "epoch": 1.0381623871189316,
      "grad_norm": 3.217491626739502,
      "learning_rate": 4.93358419562747e-06,
      "loss": 0.8251,
      "step": 12847
    },
    {
      "epoch": 1.0382431968322592,
      "grad_norm": 2.3137214183807373,
      "learning_rate": 4.932929891785205e-06,
      "loss": 0.848,
      "step": 12848
    },
    {
      "epoch": 1.0383240065455868,
      "grad_norm": 2.7785751819610596,
      "learning_rate": 4.932275589091691e-06,
      "loss": 0.7204,
      "step": 12849
    },
    {
      "epoch": 1.0384048162589143,
      "grad_norm": 2.8221638202667236,
      "learning_rate": 4.931621287558134e-06,
      "loss": 0.7979,
      "step": 12850
    },
    {
      "epoch": 1.038485625972242,
      "grad_norm": 2.9343366622924805,
      "learning_rate": 4.930966987195738e-06,
      "loss": 0.7742,
      "step": 12851
    },
    {
      "epoch": 1.0385664356855695,
      "grad_norm": 2.5015809535980225,
      "learning_rate": 4.930312688015711e-06,
      "loss": 0.8307,
      "step": 12852
    },
    {
      "epoch": 1.0386472453988969,
      "grad_norm": 3.0302183628082275,
      "learning_rate": 4.929658390029262e-06,
      "loss": 0.8124,
      "step": 12853
    },
    {
      "epoch": 1.0387280551122244,
      "grad_norm": 3.35300874710083,
      "learning_rate": 4.929004093247593e-06,
      "loss": 0.7601,
      "step": 12854
    },
    {
      "epoch": 1.038808864825552,
      "grad_norm": 3.1182520389556885,
      "learning_rate": 4.928349797681914e-06,
      "loss": 0.7175,
      "step": 12855
    },
    {
      "epoch": 1.0388896745388796,
      "grad_norm": 2.6093690395355225,
      "learning_rate": 4.92769550334343e-06,
      "loss": 0.8113,
      "step": 12856
    },
    {
      "epoch": 1.0389704842522072,
      "grad_norm": 2.6275503635406494,
      "learning_rate": 4.92704121024335e-06,
      "loss": 0.8617,
      "step": 12857
    },
    {
      "epoch": 1.0390512939655348,
      "grad_norm": 2.651615619659424,
      "learning_rate": 4.926386918392875e-06,
      "loss": 0.7714,
      "step": 12858
    },
    {
      "epoch": 1.039132103678862,
      "grad_norm": 2.784255027770996,
      "learning_rate": 4.925732627803217e-06,
      "loss": 0.7314,
      "step": 12859
    },
    {
      "epoch": 1.0392129133921897,
      "grad_norm": 2.509617805480957,
      "learning_rate": 4.925078338485581e-06,
      "loss": 0.8124,
      "step": 12860
    },
    {
      "epoch": 1.0392937231055173,
      "grad_norm": 3.1349759101867676,
      "learning_rate": 4.924424050451172e-06,
      "loss": 0.773,
      "step": 12861
    },
    {
      "epoch": 1.0393745328188448,
      "grad_norm": 2.5523223876953125,
      "learning_rate": 4.923769763711196e-06,
      "loss": 0.7478,
      "step": 12862
    },
    {
      "epoch": 1.0394553425321724,
      "grad_norm": 2.086899995803833,
      "learning_rate": 4.923115478276863e-06,
      "loss": 0.8137,
      "step": 12863
    },
    {
      "epoch": 1.0395361522455,
      "grad_norm": 2.5579464435577393,
      "learning_rate": 4.922461194159373e-06,
      "loss": 0.7144,
      "step": 12864
    },
    {
      "epoch": 1.0396169619588274,
      "grad_norm": 2.9754111766815186,
      "learning_rate": 4.921806911369938e-06,
      "loss": 0.8312,
      "step": 12865
    },
    {
      "epoch": 1.039697771672155,
      "grad_norm": 2.61491060256958,
      "learning_rate": 4.921152629919763e-06,
      "loss": 0.8016,
      "step": 12866
    },
    {
      "epoch": 1.0397785813854825,
      "grad_norm": 2.2346484661102295,
      "learning_rate": 4.920498349820053e-06,
      "loss": 0.8497,
      "step": 12867
    },
    {
      "epoch": 1.03985939109881,
      "grad_norm": 2.6187667846679688,
      "learning_rate": 4.919844071082014e-06,
      "loss": 0.841,
      "step": 12868
    },
    {
      "epoch": 1.0399402008121377,
      "grad_norm": 2.577465534210205,
      "learning_rate": 4.919189793716857e-06,
      "loss": 0.874,
      "step": 12869
    },
    {
      "epoch": 1.0400210105254653,
      "grad_norm": 2.6797478199005127,
      "learning_rate": 4.91853551773578e-06,
      "loss": 0.8406,
      "step": 12870
    },
    {
      "epoch": 1.0401018202387926,
      "grad_norm": 2.720130681991577,
      "learning_rate": 4.917881243149994e-06,
      "loss": 0.7676,
      "step": 12871
    },
    {
      "epoch": 1.0401826299521202,
      "grad_norm": 2.9457404613494873,
      "learning_rate": 4.9172269699707065e-06,
      "loss": 0.8268,
      "step": 12872
    },
    {
      "epoch": 1.0402634396654478,
      "grad_norm": 3.3807499408721924,
      "learning_rate": 4.916572698209121e-06,
      "loss": 0.8266,
      "step": 12873
    },
    {
      "epoch": 1.0403442493787753,
      "grad_norm": 2.8594462871551514,
      "learning_rate": 4.915918427876442e-06,
      "loss": 0.8403,
      "step": 12874
    },
    {
      "epoch": 1.040425059092103,
      "grad_norm": 2.592507839202881,
      "learning_rate": 4.915264158983882e-06,
      "loss": 0.8523,
      "step": 12875
    },
    {
      "epoch": 1.0405058688054305,
      "grad_norm": 2.3449959754943848,
      "learning_rate": 4.91460989154264e-06,
      "loss": 0.8336,
      "step": 12876
    },
    {
      "epoch": 1.0405866785187579,
      "grad_norm": 2.2409980297088623,
      "learning_rate": 4.913955625563925e-06,
      "loss": 0.8671,
      "step": 12877
    },
    {
      "epoch": 1.0406674882320854,
      "grad_norm": 2.4352972507476807,
      "learning_rate": 4.913301361058945e-06,
      "loss": 0.8769,
      "step": 12878
    },
    {
      "epoch": 1.040748297945413,
      "grad_norm": 2.753619909286499,
      "learning_rate": 4.912647098038903e-06,
      "loss": 0.7833,
      "step": 12879
    },
    {
      "epoch": 1.0408291076587406,
      "grad_norm": 2.6027324199676514,
      "learning_rate": 4.911992836515004e-06,
      "loss": 0.8694,
      "step": 12880
    },
    {
      "epoch": 1.0409099173720682,
      "grad_norm": 2.600783586502075,
      "learning_rate": 4.91133857649846e-06,
      "loss": 0.7395,
      "step": 12881
    },
    {
      "epoch": 1.0409907270853958,
      "grad_norm": 2.7033865451812744,
      "learning_rate": 4.91068431800047e-06,
      "loss": 0.7953,
      "step": 12882
    },
    {
      "epoch": 1.041071536798723,
      "grad_norm": 2.8208181858062744,
      "learning_rate": 4.910030061032242e-06,
      "loss": 0.8189,
      "step": 12883
    },
    {
      "epoch": 1.0411523465120507,
      "grad_norm": 2.617600917816162,
      "learning_rate": 4.9093758056049824e-06,
      "loss": 0.844,
      "step": 12884
    },
    {
      "epoch": 1.0412331562253783,
      "grad_norm": 2.503328800201416,
      "learning_rate": 4.9087215517299e-06,
      "loss": 0.8647,
      "step": 12885
    },
    {
      "epoch": 1.0413139659387058,
      "grad_norm": 2.663407325744629,
      "learning_rate": 4.908067299418194e-06,
      "loss": 0.7519,
      "step": 12886
    },
    {
      "epoch": 1.0413947756520334,
      "grad_norm": 2.375281572341919,
      "learning_rate": 4.907413048681076e-06,
      "loss": 0.8333,
      "step": 12887
    },
    {
      "epoch": 1.041475585365361,
      "grad_norm": 2.865656614303589,
      "learning_rate": 4.90675879952975e-06,
      "loss": 0.8715,
      "step": 12888
    },
    {
      "epoch": 1.0415563950786884,
      "grad_norm": 2.389338970184326,
      "learning_rate": 4.906104551975421e-06,
      "loss": 0.8152,
      "step": 12889
    },
    {
      "epoch": 1.041637204792016,
      "grad_norm": 2.327669620513916,
      "learning_rate": 4.905450306029293e-06,
      "loss": 0.8006,
      "step": 12890
    },
    {
      "epoch": 1.0417180145053435,
      "grad_norm": 2.764125347137451,
      "learning_rate": 4.904796061702577e-06,
      "loss": 0.8706,
      "step": 12891
    },
    {
      "epoch": 1.041798824218671,
      "grad_norm": 2.936788320541382,
      "learning_rate": 4.904141819006472e-06,
      "loss": 0.8794,
      "step": 12892
    },
    {
      "epoch": 1.0418796339319987,
      "grad_norm": 2.675888776779175,
      "learning_rate": 4.903487577952188e-06,
      "loss": 0.8843,
      "step": 12893
    },
    {
      "epoch": 1.0419604436453263,
      "grad_norm": 2.5857391357421875,
      "learning_rate": 4.902833338550931e-06,
      "loss": 0.8065,
      "step": 12894
    },
    {
      "epoch": 1.0420412533586536,
      "grad_norm": 2.7993967533111572,
      "learning_rate": 4.902179100813903e-06,
      "loss": 0.8547,
      "step": 12895
    },
    {
      "epoch": 1.0421220630719812,
      "grad_norm": 2.5567092895507812,
      "learning_rate": 4.901524864752311e-06,
      "loss": 0.7886,
      "step": 12896
    },
    {
      "epoch": 1.0422028727853088,
      "grad_norm": 2.6120681762695312,
      "learning_rate": 4.900870630377364e-06,
      "loss": 0.8189,
      "step": 12897
    },
    {
      "epoch": 1.0422836824986363,
      "grad_norm": 2.7176361083984375,
      "learning_rate": 4.900216397700262e-06,
      "loss": 0.8435,
      "step": 12898
    },
    {
      "epoch": 1.042364492211964,
      "grad_norm": 2.7765636444091797,
      "learning_rate": 4.899562166732213e-06,
      "loss": 0.8413,
      "step": 12899
    },
    {
      "epoch": 1.0424453019252915,
      "grad_norm": 2.5095574855804443,
      "learning_rate": 4.898907937484424e-06,
      "loss": 0.7782,
      "step": 12900
    },
    {
      "epoch": 1.0425261116386189,
      "grad_norm": 2.6688530445098877,
      "learning_rate": 4.8982537099680975e-06,
      "loss": 0.8314,
      "step": 12901
    },
    {
      "epoch": 1.0426069213519464,
      "grad_norm": 2.7962560653686523,
      "learning_rate": 4.897599484194439e-06,
      "loss": 0.8385,
      "step": 12902
    },
    {
      "epoch": 1.042687731065274,
      "grad_norm": 2.8190219402313232,
      "learning_rate": 4.8969452601746564e-06,
      "loss": 0.8413,
      "step": 12903
    },
    {
      "epoch": 1.0427685407786016,
      "grad_norm": 3.0162580013275146,
      "learning_rate": 4.896291037919953e-06,
      "loss": 0.7438,
      "step": 12904
    },
    {
      "epoch": 1.0428493504919292,
      "grad_norm": 2.2201485633850098,
      "learning_rate": 4.895636817441534e-06,
      "loss": 0.8379,
      "step": 12905
    },
    {
      "epoch": 1.0429301602052568,
      "grad_norm": 2.630707025527954,
      "learning_rate": 4.894982598750606e-06,
      "loss": 0.7655,
      "step": 12906
    },
    {
      "epoch": 1.043010969918584,
      "grad_norm": 2.468411922454834,
      "learning_rate": 4.8943283818583716e-06,
      "loss": 0.8834,
      "step": 12907
    },
    {
      "epoch": 1.0430917796319117,
      "grad_norm": 2.922224521636963,
      "learning_rate": 4.893674166776039e-06,
      "loss": 0.7395,
      "step": 12908
    },
    {
      "epoch": 1.0431725893452393,
      "grad_norm": 2.400918960571289,
      "learning_rate": 4.893019953514811e-06,
      "loss": 0.82,
      "step": 12909
    },
    {
      "epoch": 1.0432533990585668,
      "grad_norm": 2.39370059967041,
      "learning_rate": 4.892365742085895e-06,
      "loss": 0.7025,
      "step": 12910
    },
    {
      "epoch": 1.0433342087718944,
      "grad_norm": 2.4781503677368164,
      "learning_rate": 4.891711532500494e-06,
      "loss": 0.8592,
      "step": 12911
    },
    {
      "epoch": 1.043415018485222,
      "grad_norm": 2.745285749435425,
      "learning_rate": 4.891057324769813e-06,
      "loss": 0.88,
      "step": 12912
    },
    {
      "epoch": 1.0434958281985494,
      "grad_norm": 2.4261934757232666,
      "learning_rate": 4.890403118905058e-06,
      "loss": 0.7642,
      "step": 12913
    },
    {
      "epoch": 1.043576637911877,
      "grad_norm": 2.5502376556396484,
      "learning_rate": 4.889748914917435e-06,
      "loss": 0.7788,
      "step": 12914
    },
    {
      "epoch": 1.0436574476252045,
      "grad_norm": 3.543602705001831,
      "learning_rate": 4.889094712818146e-06,
      "loss": 0.841,
      "step": 12915
    },
    {
      "epoch": 1.043738257338532,
      "grad_norm": 2.63369083404541,
      "learning_rate": 4.8884405126184e-06,
      "loss": 0.8548,
      "step": 12916
    },
    {
      "epoch": 1.0438190670518597,
      "grad_norm": 2.512787342071533,
      "learning_rate": 4.887786314329398e-06,
      "loss": 0.8282,
      "step": 12917
    },
    {
      "epoch": 1.0438998767651873,
      "grad_norm": 2.8143913745880127,
      "learning_rate": 4.887132117962344e-06,
      "loss": 0.8347,
      "step": 12918
    },
    {
      "epoch": 1.0439806864785148,
      "grad_norm": 2.6553726196289062,
      "learning_rate": 4.886477923528449e-06,
      "loss": 0.7903,
      "step": 12919
    },
    {
      "epoch": 1.0440614961918422,
      "grad_norm": 2.450526714324951,
      "learning_rate": 4.8858237310389115e-06,
      "loss": 0.9034,
      "step": 12920
    },
    {
      "epoch": 1.0441423059051698,
      "grad_norm": 2.8087244033813477,
      "learning_rate": 4.8851695405049395e-06,
      "loss": 0.8621,
      "step": 12921
    },
    {
      "epoch": 1.0442231156184973,
      "grad_norm": 2.3381245136260986,
      "learning_rate": 4.884515351937737e-06,
      "loss": 0.8188,
      "step": 12922
    },
    {
      "epoch": 1.044303925331825,
      "grad_norm": 3.1490652561187744,
      "learning_rate": 4.883861165348507e-06,
      "loss": 0.7268,
      "step": 12923
    },
    {
      "epoch": 1.0443847350451525,
      "grad_norm": 2.547015428543091,
      "learning_rate": 4.8832069807484565e-06,
      "loss": 0.7608,
      "step": 12924
    },
    {
      "epoch": 1.0444655447584799,
      "grad_norm": 3.16630482673645,
      "learning_rate": 4.88255279814879e-06,
      "loss": 0.9203,
      "step": 12925
    },
    {
      "epoch": 1.0445463544718074,
      "grad_norm": 2.810124397277832,
      "learning_rate": 4.8818986175607114e-06,
      "loss": 0.7702,
      "step": 12926
    },
    {
      "epoch": 1.044627164185135,
      "grad_norm": 2.728691339492798,
      "learning_rate": 4.881244438995424e-06,
      "loss": 0.9028,
      "step": 12927
    },
    {
      "epoch": 1.0447079738984626,
      "grad_norm": 2.4419636726379395,
      "learning_rate": 4.880590262464136e-06,
      "loss": 0.7741,
      "step": 12928
    },
    {
      "epoch": 1.0447887836117902,
      "grad_norm": 2.686403274536133,
      "learning_rate": 4.879936087978046e-06,
      "loss": 0.786,
      "step": 12929
    },
    {
      "epoch": 1.0448695933251178,
      "grad_norm": 2.2678635120391846,
      "learning_rate": 4.879281915548363e-06,
      "loss": 0.8956,
      "step": 12930
    },
    {
      "epoch": 1.0449504030384453,
      "grad_norm": 2.7791199684143066,
      "learning_rate": 4.878627745186291e-06,
      "loss": 0.8339,
      "step": 12931
    },
    {
      "epoch": 1.0450312127517727,
      "grad_norm": 2.401492118835449,
      "learning_rate": 4.877973576903032e-06,
      "loss": 0.9785,
      "step": 12932
    },
    {
      "epoch": 1.0451120224651003,
      "grad_norm": 2.7468011379241943,
      "learning_rate": 4.877319410709792e-06,
      "loss": 0.8165,
      "step": 12933
    },
    {
      "epoch": 1.0451928321784278,
      "grad_norm": 2.9621524810791016,
      "learning_rate": 4.876665246617778e-06,
      "loss": 0.7445,
      "step": 12934
    },
    {
      "epoch": 1.0452736418917554,
      "grad_norm": 2.692807912826538,
      "learning_rate": 4.876011084638188e-06,
      "loss": 0.83,
      "step": 12935
    },
    {
      "epoch": 1.045354451605083,
      "grad_norm": 2.5371296405792236,
      "learning_rate": 4.875356924782231e-06,
      "loss": 0.8248,
      "step": 12936
    },
    {
      "epoch": 1.0454352613184106,
      "grad_norm": 2.3327078819274902,
      "learning_rate": 4.874702767061109e-06,
      "loss": 0.8322,
      "step": 12937
    },
    {
      "epoch": 1.045516071031738,
      "grad_norm": 2.231863260269165,
      "learning_rate": 4.874048611486029e-06,
      "loss": 0.75,
      "step": 12938
    },
    {
      "epoch": 1.0455968807450655,
      "grad_norm": 2.6811535358428955,
      "learning_rate": 4.873394458068191e-06,
      "loss": 0.8114,
      "step": 12939
    },
    {
      "epoch": 1.045677690458393,
      "grad_norm": 2.4019546508789062,
      "learning_rate": 4.872740306818801e-06,
      "loss": 0.8005,
      "step": 12940
    },
    {
      "epoch": 1.0457585001717207,
      "grad_norm": 2.4113852977752686,
      "learning_rate": 4.872086157749065e-06,
      "loss": 0.9376,
      "step": 12941
    },
    {
      "epoch": 1.0458393098850483,
      "grad_norm": 2.4300756454467773,
      "learning_rate": 4.871432010870184e-06,
      "loss": 0.7849,
      "step": 12942
    },
    {
      "epoch": 1.0459201195983758,
      "grad_norm": 2.8537158966064453,
      "learning_rate": 4.870777866193362e-06,
      "loss": 0.8639,
      "step": 12943
    },
    {
      "epoch": 1.0460009293117032,
      "grad_norm": 2.6500296592712402,
      "learning_rate": 4.870123723729809e-06,
      "loss": 0.8609,
      "step": 12944
    },
    {
      "epoch": 1.0460817390250308,
      "grad_norm": 2.716092824935913,
      "learning_rate": 4.869469583490718e-06,
      "loss": 0.7456,
      "step": 12945
    },
    {
      "epoch": 1.0461625487383583,
      "grad_norm": 2.320584774017334,
      "learning_rate": 4.868815445487301e-06,
      "loss": 0.9307,
      "step": 12946
    },
    {
      "epoch": 1.046243358451686,
      "grad_norm": 2.245326042175293,
      "learning_rate": 4.868161309730761e-06,
      "loss": 0.8419,
      "step": 12947
    },
    {
      "epoch": 1.0463241681650135,
      "grad_norm": 2.4302146434783936,
      "learning_rate": 4.867507176232299e-06,
      "loss": 0.8272,
      "step": 12948
    },
    {
      "epoch": 1.046404977878341,
      "grad_norm": 2.7676596641540527,
      "learning_rate": 4.866853045003119e-06,
      "loss": 0.8039,
      "step": 12949
    },
    {
      "epoch": 1.0464857875916684,
      "grad_norm": 2.416679859161377,
      "learning_rate": 4.8661989160544295e-06,
      "loss": 0.7932,
      "step": 12950
    },
    {
      "epoch": 1.046566597304996,
      "grad_norm": 2.6678290367126465,
      "learning_rate": 4.865544789397427e-06,
      "loss": 0.8741,
      "step": 12951
    },
    {
      "epoch": 1.0466474070183236,
      "grad_norm": 2.4413371086120605,
      "learning_rate": 4.864890665043319e-06,
      "loss": 0.8358,
      "step": 12952
    },
    {
      "epoch": 1.0467282167316512,
      "grad_norm": 2.3341283798217773,
      "learning_rate": 4.86423654300331e-06,
      "loss": 0.7614,
      "step": 12953
    },
    {
      "epoch": 1.0468090264449788,
      "grad_norm": 3.1425116062164307,
      "learning_rate": 4.863582423288602e-06,
      "loss": 0.8203,
      "step": 12954
    },
    {
      "epoch": 1.0468898361583063,
      "grad_norm": 2.804582118988037,
      "learning_rate": 4.862928305910396e-06,
      "loss": 0.8434,
      "step": 12955
    },
    {
      "epoch": 1.0469706458716337,
      "grad_norm": 2.468287467956543,
      "learning_rate": 4.862274190879903e-06,
      "loss": 0.8474,
      "step": 12956
    },
    {
      "epoch": 1.0470514555849613,
      "grad_norm": 2.8406238555908203,
      "learning_rate": 4.861620078208317e-06,
      "loss": 0.7911,
      "step": 12957
    },
    {
      "epoch": 1.0471322652982888,
      "grad_norm": 2.4930367469787598,
      "learning_rate": 4.8609659679068475e-06,
      "loss": 0.9259,
      "step": 12958
    },
    {
      "epoch": 1.0472130750116164,
      "grad_norm": 2.488539218902588,
      "learning_rate": 4.8603118599866975e-06,
      "loss": 0.8723,
      "step": 12959
    },
    {
      "epoch": 1.047293884724944,
      "grad_norm": 2.456427574157715,
      "learning_rate": 4.859657754459068e-06,
      "loss": 0.8189,
      "step": 12960
    },
    {
      "epoch": 1.0473746944382716,
      "grad_norm": 2.627305269241333,
      "learning_rate": 4.859003651335163e-06,
      "loss": 0.7841,
      "step": 12961
    },
    {
      "epoch": 1.047455504151599,
      "grad_norm": 2.6415815353393555,
      "learning_rate": 4.858349550626187e-06,
      "loss": 0.7202,
      "step": 12962
    },
    {
      "epoch": 1.0475363138649265,
      "grad_norm": 2.7195494174957275,
      "learning_rate": 4.8576954523433415e-06,
      "loss": 0.7869,
      "step": 12963
    },
    {
      "epoch": 1.047617123578254,
      "grad_norm": 2.7807328701019287,
      "learning_rate": 4.85704135649783e-06,
      "loss": 0.7887,
      "step": 12964
    },
    {
      "epoch": 1.0476979332915817,
      "grad_norm": 2.895347833633423,
      "learning_rate": 4.856387263100854e-06,
      "loss": 0.822,
      "step": 12965
    },
    {
      "epoch": 1.0477787430049093,
      "grad_norm": 2.4659616947174072,
      "learning_rate": 4.8557331721636235e-06,
      "loss": 0.9038,
      "step": 12966
    },
    {
      "epoch": 1.0478595527182368,
      "grad_norm": 2.6420671939849854,
      "learning_rate": 4.8550790836973325e-06,
      "loss": 0.8497,
      "step": 12967
    },
    {
      "epoch": 1.0479403624315642,
      "grad_norm": 2.4824156761169434,
      "learning_rate": 4.8544249977131895e-06,
      "loss": 0.9595,
      "step": 12968
    },
    {
      "epoch": 1.0480211721448918,
      "grad_norm": 2.7139363288879395,
      "learning_rate": 4.853770914222396e-06,
      "loss": 0.912,
      "step": 12969
    },
    {
      "epoch": 1.0481019818582193,
      "grad_norm": 2.9319398403167725,
      "learning_rate": 4.853116833236154e-06,
      "loss": 0.8654,
      "step": 12970
    },
    {
      "epoch": 1.048182791571547,
      "grad_norm": 2.4399850368499756,
      "learning_rate": 4.852462754765665e-06,
      "loss": 0.8604,
      "step": 12971
    },
    {
      "epoch": 1.0482636012848745,
      "grad_norm": 2.647242307662964,
      "learning_rate": 4.851808678822137e-06,
      "loss": 0.716,
      "step": 12972
    },
    {
      "epoch": 1.048344410998202,
      "grad_norm": 2.568131923675537,
      "learning_rate": 4.851154605416769e-06,
      "loss": 0.7549,
      "step": 12973
    },
    {
      "epoch": 1.0484252207115294,
      "grad_norm": 2.3960366249084473,
      "learning_rate": 4.850500534560764e-06,
      "loss": 0.8559,
      "step": 12974
    },
    {
      "epoch": 1.048506030424857,
      "grad_norm": 2.702261447906494,
      "learning_rate": 4.849846466265326e-06,
      "loss": 0.9374,
      "step": 12975
    },
    {
      "epoch": 1.0485868401381846,
      "grad_norm": 2.4157943725585938,
      "learning_rate": 4.849192400541654e-06,
      "loss": 0.8004,
      "step": 12976
    },
    {
      "epoch": 1.0486676498515122,
      "grad_norm": 2.5824832916259766,
      "learning_rate": 4.848538337400954e-06,
      "loss": 0.9134,
      "step": 12977
    },
    {
      "epoch": 1.0487484595648398,
      "grad_norm": 2.6210951805114746,
      "learning_rate": 4.847884276854429e-06,
      "loss": 0.787,
      "step": 12978
    },
    {
      "epoch": 1.0488292692781673,
      "grad_norm": 2.176785945892334,
      "learning_rate": 4.8472302189132795e-06,
      "loss": 0.943,
      "step": 12979
    },
    {
      "epoch": 1.0489100789914947,
      "grad_norm": 3.3187334537506104,
      "learning_rate": 4.846576163588708e-06,
      "loss": 0.8931,
      "step": 12980
    },
    {
      "epoch": 1.0489908887048223,
      "grad_norm": 2.90089750289917,
      "learning_rate": 4.8459221108919185e-06,
      "loss": 0.7279,
      "step": 12981
    },
    {
      "epoch": 1.0490716984181498,
      "grad_norm": 2.589691162109375,
      "learning_rate": 4.84526806083411e-06,
      "loss": 0.8896,
      "step": 12982
    },
    {
      "epoch": 1.0491525081314774,
      "grad_norm": 2.450986385345459,
      "learning_rate": 4.844614013426489e-06,
      "loss": 0.8449,
      "step": 12983
    },
    {
      "epoch": 1.049233317844805,
      "grad_norm": 2.4798226356506348,
      "learning_rate": 4.8439599686802566e-06,
      "loss": 0.9056,
      "step": 12984
    },
    {
      "epoch": 1.0493141275581326,
      "grad_norm": 2.879004716873169,
      "learning_rate": 4.843305926606613e-06,
      "loss": 0.8751,
      "step": 12985
    },
    {
      "epoch": 1.04939493727146,
      "grad_norm": 2.4587759971618652,
      "learning_rate": 4.8426518872167615e-06,
      "loss": 0.9262,
      "step": 12986
    },
    {
      "epoch": 1.0494757469847875,
      "grad_norm": 2.6293201446533203,
      "learning_rate": 4.841997850521906e-06,
      "loss": 0.8178,
      "step": 12987
    },
    {
      "epoch": 1.049556556698115,
      "grad_norm": 2.8607261180877686,
      "learning_rate": 4.841343816533245e-06,
      "loss": 0.7812,
      "step": 12988
    },
    {
      "epoch": 1.0496373664114427,
      "grad_norm": 2.734184741973877,
      "learning_rate": 4.840689785261983e-06,
      "loss": 0.8692,
      "step": 12989
    },
    {
      "epoch": 1.0497181761247703,
      "grad_norm": 2.922410011291504,
      "learning_rate": 4.840035756719322e-06,
      "loss": 0.7434,
      "step": 12990
    },
    {
      "epoch": 1.0497989858380978,
      "grad_norm": 2.6176400184631348,
      "learning_rate": 4.839381730916463e-06,
      "loss": 0.8148,
      "step": 12991
    },
    {
      "epoch": 1.0498797955514252,
      "grad_norm": 2.8842031955718994,
      "learning_rate": 4.838727707864607e-06,
      "loss": 0.7583,
      "step": 12992
    },
    {
      "epoch": 1.0499606052647528,
      "grad_norm": 2.6824910640716553,
      "learning_rate": 4.838073687574959e-06,
      "loss": 0.8783,
      "step": 12993
    },
    {
      "epoch": 1.0500414149780803,
      "grad_norm": 2.2635958194732666,
      "learning_rate": 4.837419670058719e-06,
      "loss": 0.8861,
      "step": 12994
    },
    {
      "epoch": 1.050122224691408,
      "grad_norm": 2.542126417160034,
      "learning_rate": 4.836765655327088e-06,
      "loss": 0.8662,
      "step": 12995
    },
    {
      "epoch": 1.0502030344047355,
      "grad_norm": 2.507847309112549,
      "learning_rate": 4.836111643391268e-06,
      "loss": 0.8216,
      "step": 12996
    },
    {
      "epoch": 1.050283844118063,
      "grad_norm": 2.5112709999084473,
      "learning_rate": 4.8354576342624624e-06,
      "loss": 0.8836,
      "step": 12997
    },
    {
      "epoch": 1.0503646538313904,
      "grad_norm": 2.362645149230957,
      "learning_rate": 4.834803627951869e-06,
      "loss": 0.8943,
      "step": 12998
    },
    {
      "epoch": 1.050445463544718,
      "grad_norm": 2.359710931777954,
      "learning_rate": 4.834149624470693e-06,
      "loss": 0.8865,
      "step": 12999
    },
    {
      "epoch": 1.0505262732580456,
      "grad_norm": 4.162075042724609,
      "learning_rate": 4.833495623830136e-06,
      "loss": 0.7883,
      "step": 13000
    },
    {
      "epoch": 1.0505262732580456,
      "eval_loss": 0.754727840423584,
      "eval_runtime": 818.5065,
      "eval_samples_per_second": 101.851,
      "eval_steps_per_second": 12.732,
      "step": 13000
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 24748,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.1025855761417765e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}