{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 0,
  "global_step": 337,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002967359050445104,
      "grad_norm": 1.5555781126022339,
      "learning_rate": 1e-05,
      "loss": 2.6685,
      "step": 1
    },
    {
      "epoch": 0.005934718100890208,
      "grad_norm": 1.5590428113937378,
      "learning_rate": 9.97032640949555e-06,
      "loss": 2.683,
      "step": 2
    },
    {
      "epoch": 0.008902077151335312,
      "grad_norm": 1.467544436454773,
      "learning_rate": 9.940652818991098e-06,
      "loss": 2.6761,
      "step": 3
    },
    {
      "epoch": 0.011869436201780416,
      "grad_norm": 1.413665771484375,
      "learning_rate": 9.910979228486648e-06,
      "loss": 2.6253,
      "step": 4
    },
    {
      "epoch": 0.01483679525222552,
      "grad_norm": 1.4075852632522583,
      "learning_rate": 9.881305637982197e-06,
      "loss": 2.632,
      "step": 5
    },
    {
      "epoch": 0.017804154302670624,
      "grad_norm": 1.396652102470398,
      "learning_rate": 9.851632047477747e-06,
      "loss": 2.6888,
      "step": 6
    },
    {
      "epoch": 0.020771513353115726,
      "grad_norm": 1.2711882591247559,
      "learning_rate": 9.821958456973294e-06,
      "loss": 2.63,
      "step": 7
    },
    {
      "epoch": 0.02373887240356083,
      "grad_norm": 1.1441361904144287,
      "learning_rate": 9.792284866468842e-06,
      "loss": 2.5566,
      "step": 8
    },
    {
      "epoch": 0.026706231454005934,
      "grad_norm": 1.1301014423370361,
      "learning_rate": 9.762611275964392e-06,
      "loss": 2.579,
      "step": 9
    },
    {
      "epoch": 0.02967359050445104,
      "grad_norm": 1.0782572031021118,
      "learning_rate": 9.732937685459941e-06,
      "loss": 2.5188,
      "step": 10
    },
    {
      "epoch": 0.032640949554896145,
      "grad_norm": 1.0123143196105957,
      "learning_rate": 9.70326409495549e-06,
      "loss": 2.5475,
      "step": 11
    },
    {
      "epoch": 0.03560830860534125,
      "grad_norm": 0.9147588610649109,
      "learning_rate": 9.673590504451039e-06,
      "loss": 2.4341,
      "step": 12
    },
    {
      "epoch": 0.03857566765578635,
      "grad_norm": 0.9908038973808289,
      "learning_rate": 9.643916913946588e-06,
      "loss": 2.4742,
      "step": 13
    },
    {
      "epoch": 0.04154302670623145,
      "grad_norm": 0.9412248134613037,
      "learning_rate": 9.614243323442138e-06,
      "loss": 2.404,
      "step": 14
    },
    {
      "epoch": 0.04451038575667656,
      "grad_norm": 0.9002366065979004,
      "learning_rate": 9.584569732937687e-06,
      "loss": 2.4078,
      "step": 15
    },
    {
      "epoch": 0.04747774480712166,
      "grad_norm": 0.9244178533554077,
      "learning_rate": 9.554896142433235e-06,
      "loss": 2.4475,
      "step": 16
    },
    {
      "epoch": 0.050445103857566766,
      "grad_norm": 0.9509070515632629,
      "learning_rate": 9.525222551928784e-06,
      "loss": 2.4401,
      "step": 17
    },
    {
      "epoch": 0.05341246290801187,
      "grad_norm": 0.9273353815078735,
      "learning_rate": 9.495548961424334e-06,
      "loss": 2.4526,
      "step": 18
    },
    {
      "epoch": 0.05637982195845697,
      "grad_norm": 0.8414579629898071,
      "learning_rate": 9.465875370919882e-06,
      "loss": 2.3536,
      "step": 19
    },
    {
      "epoch": 0.05934718100890208,
      "grad_norm": 0.8936970829963684,
      "learning_rate": 9.436201780415431e-06,
      "loss": 2.3934,
      "step": 20
    },
    {
      "epoch": 0.06231454005934718,
      "grad_norm": 0.8523764610290527,
      "learning_rate": 9.406528189910979e-06,
      "loss": 2.3833,
      "step": 21
    },
    {
      "epoch": 0.06528189910979229,
      "grad_norm": 0.7926989197731018,
      "learning_rate": 9.376854599406528e-06,
      "loss": 2.3302,
      "step": 22
    },
    {
      "epoch": 0.06824925816023739,
      "grad_norm": 0.79078209400177,
      "learning_rate": 9.347181008902078e-06,
      "loss": 2.3615,
      "step": 23
    },
    {
      "epoch": 0.0712166172106825,
      "grad_norm": 0.7669358849525452,
      "learning_rate": 9.317507418397626e-06,
      "loss": 2.2893,
      "step": 24
    },
    {
      "epoch": 0.07418397626112759,
      "grad_norm": 0.7894212007522583,
      "learning_rate": 9.287833827893175e-06,
      "loss": 2.2811,
      "step": 25
    },
    {
      "epoch": 0.0771513353115727,
      "grad_norm": 0.7898035645484924,
      "learning_rate": 9.258160237388725e-06,
      "loss": 2.2458,
      "step": 26
    },
    {
      "epoch": 0.08011869436201781,
      "grad_norm": 0.7668829560279846,
      "learning_rate": 9.228486646884274e-06,
      "loss": 2.3121,
      "step": 27
    },
    {
      "epoch": 0.0830860534124629,
      "grad_norm": 0.7801551222801208,
      "learning_rate": 9.198813056379822e-06,
      "loss": 2.261,
      "step": 28
    },
    {
      "epoch": 0.08605341246290801,
      "grad_norm": 0.6885457038879395,
      "learning_rate": 9.169139465875372e-06,
      "loss": 2.2047,
      "step": 29
    },
    {
      "epoch": 0.08902077151335312,
      "grad_norm": 0.6457021832466125,
      "learning_rate": 9.139465875370921e-06,
      "loss": 2.1508,
      "step": 30
    },
    {
      "epoch": 0.09198813056379822,
      "grad_norm": 0.6720436215400696,
      "learning_rate": 9.10979228486647e-06,
      "loss": 2.2211,
      "step": 31
    },
    {
      "epoch": 0.09495548961424333,
      "grad_norm": 0.684187114238739,
      "learning_rate": 9.080118694362018e-06,
      "loss": 2.1975,
      "step": 32
    },
    {
      "epoch": 0.09792284866468842,
      "grad_norm": 0.6878846287727356,
      "learning_rate": 9.050445103857568e-06,
      "loss": 2.1208,
      "step": 33
    },
    {
      "epoch": 0.10089020771513353,
      "grad_norm": 0.6933832764625549,
      "learning_rate": 9.020771513353116e-06,
      "loss": 2.1696,
      "step": 34
    },
    {
      "epoch": 0.10385756676557864,
      "grad_norm": 0.620203971862793,
      "learning_rate": 8.991097922848665e-06,
      "loss": 2.1174,
      "step": 35
    },
    {
      "epoch": 0.10682492581602374,
      "grad_norm": 0.6201160550117493,
      "learning_rate": 8.961424332344215e-06,
      "loss": 2.1494,
      "step": 36
    },
    {
      "epoch": 0.10979228486646884,
      "grad_norm": 0.6284067630767822,
      "learning_rate": 8.931750741839763e-06,
      "loss": 2.1778,
      "step": 37
    },
    {
      "epoch": 0.11275964391691394,
      "grad_norm": 0.5838620662689209,
      "learning_rate": 8.902077151335312e-06,
      "loss": 2.1085,
      "step": 38
    },
    {
      "epoch": 0.11572700296735905,
      "grad_norm": 0.601154088973999,
      "learning_rate": 8.872403560830862e-06,
      "loss": 2.1055,
      "step": 39
    },
    {
      "epoch": 0.11869436201780416,
      "grad_norm": 0.5567020177841187,
      "learning_rate": 8.842729970326411e-06,
      "loss": 2.0706,
      "step": 40
    },
    {
      "epoch": 0.12166172106824925,
      "grad_norm": 0.5745820999145508,
      "learning_rate": 8.813056379821959e-06,
      "loss": 2.0854,
      "step": 41
    },
    {
      "epoch": 0.12462908011869436,
      "grad_norm": 0.554350733757019,
      "learning_rate": 8.783382789317508e-06,
      "loss": 2.0453,
      "step": 42
    },
    {
      "epoch": 0.12759643916913946,
      "grad_norm": 0.5471130609512329,
      "learning_rate": 8.753709198813058e-06,
      "loss": 2.0599,
      "step": 43
    },
    {
      "epoch": 0.13056379821958458,
      "grad_norm": 0.5599424242973328,
      "learning_rate": 8.724035608308606e-06,
      "loss": 2.0208,
      "step": 44
    },
    {
      "epoch": 0.13353115727002968,
      "grad_norm": 0.5747941732406616,
      "learning_rate": 8.694362017804155e-06,
      "loss": 2.0737,
      "step": 45
    },
    {
      "epoch": 0.13649851632047477,
      "grad_norm": 0.5319300889968872,
      "learning_rate": 8.664688427299705e-06,
      "loss": 2.0425,
      "step": 46
    },
    {
      "epoch": 0.1394658753709199,
      "grad_norm": 0.5038570761680603,
      "learning_rate": 8.635014836795252e-06,
      "loss": 2.0337,
      "step": 47
    },
    {
      "epoch": 0.142433234421365,
      "grad_norm": 0.5131641030311584,
      "learning_rate": 8.605341246290802e-06,
      "loss": 2.0335,
      "step": 48
    },
    {
      "epoch": 0.14540059347181009,
      "grad_norm": 0.5833173394203186,
      "learning_rate": 8.57566765578635e-06,
      "loss": 1.9506,
      "step": 49
    },
    {
      "epoch": 0.14836795252225518,
      "grad_norm": 0.516785204410553,
      "learning_rate": 8.5459940652819e-06,
      "loss": 1.9955,
      "step": 50
    },
    {
      "epoch": 0.1513353115727003,
      "grad_norm": 0.5017814636230469,
      "learning_rate": 8.516320474777449e-06,
      "loss": 1.9909,
      "step": 51
    },
    {
      "epoch": 0.1543026706231454,
      "grad_norm": 0.5120069980621338,
      "learning_rate": 8.486646884272998e-06,
      "loss": 2.0216,
      "step": 52
    },
    {
      "epoch": 0.1572700296735905,
      "grad_norm": 0.5009364485740662,
      "learning_rate": 8.456973293768546e-06,
      "loss": 2.006,
      "step": 53
    },
    {
      "epoch": 0.16023738872403562,
      "grad_norm": 0.5187188386917114,
      "learning_rate": 8.427299703264096e-06,
      "loss": 1.9846,
      "step": 54
    },
    {
      "epoch": 0.1632047477744807,
      "grad_norm": 0.4704589247703552,
      "learning_rate": 8.397626112759645e-06,
      "loss": 1.9522,
      "step": 55
    },
    {
      "epoch": 0.1661721068249258,
      "grad_norm": 0.5026980638504028,
      "learning_rate": 8.367952522255195e-06,
      "loss": 1.9088,
      "step": 56
    },
    {
      "epoch": 0.16913946587537093,
      "grad_norm": 0.47747883200645447,
      "learning_rate": 8.338278931750742e-06,
      "loss": 2.0106,
      "step": 57
    },
    {
      "epoch": 0.17210682492581603,
      "grad_norm": 0.46089866757392883,
      "learning_rate": 8.308605341246292e-06,
      "loss": 1.9678,
      "step": 58
    },
    {
      "epoch": 0.17507418397626112,
      "grad_norm": 0.5012717843055725,
      "learning_rate": 8.278931750741841e-06,
      "loss": 1.9628,
      "step": 59
    },
    {
      "epoch": 0.17804154302670624,
      "grad_norm": 0.5236649513244629,
      "learning_rate": 8.24925816023739e-06,
      "loss": 1.9642,
      "step": 60
    },
    {
      "epoch": 0.18100890207715134,
      "grad_norm": 0.48298126459121704,
      "learning_rate": 8.219584569732939e-06,
      "loss": 1.9857,
      "step": 61
    },
    {
      "epoch": 0.18397626112759644,
      "grad_norm": 0.46658381819725037,
      "learning_rate": 8.189910979228487e-06,
      "loss": 1.9117,
      "step": 62
    },
    {
      "epoch": 0.18694362017804153,
      "grad_norm": 0.44584569334983826,
      "learning_rate": 8.160237388724036e-06,
      "loss": 1.934,
      "step": 63
    },
    {
      "epoch": 0.18991097922848665,
      "grad_norm": 0.47703817486763,
      "learning_rate": 8.130563798219586e-06,
      "loss": 1.9219,
      "step": 64
    },
    {
      "epoch": 0.19287833827893175,
      "grad_norm": 0.828423798084259,
      "learning_rate": 8.100890207715133e-06,
      "loss": 1.9517,
      "step": 65
    },
    {
      "epoch": 0.19584569732937684,
      "grad_norm": 0.6144270300865173,
      "learning_rate": 8.071216617210683e-06,
      "loss": 1.9075,
      "step": 66
    },
    {
      "epoch": 0.19881305637982197,
      "grad_norm": 0.4625411629676819,
      "learning_rate": 8.041543026706232e-06,
      "loss": 1.8875,
      "step": 67
    },
    {
      "epoch": 0.20178041543026706,
      "grad_norm": 0.44810694456100464,
      "learning_rate": 8.011869436201782e-06,
      "loss": 1.9397,
      "step": 68
    },
    {
      "epoch": 0.20474777448071216,
      "grad_norm": 0.5223219990730286,
      "learning_rate": 7.98219584569733e-06,
      "loss": 1.8873,
      "step": 69
    },
    {
      "epoch": 0.20771513353115728,
      "grad_norm": 0.4946565330028534,
      "learning_rate": 7.95252225519288e-06,
      "loss": 1.8917,
      "step": 70
    },
    {
      "epoch": 0.21068249258160238,
      "grad_norm": 0.4554108679294586,
      "learning_rate": 7.922848664688429e-06,
      "loss": 1.903,
      "step": 71
    },
    {
      "epoch": 0.21364985163204747,
      "grad_norm": 0.4824586808681488,
      "learning_rate": 7.893175074183978e-06,
      "loss": 1.8526,
      "step": 72
    },
    {
      "epoch": 0.2166172106824926,
      "grad_norm": 0.46171435713768005,
      "learning_rate": 7.863501483679526e-06,
      "loss": 1.9143,
      "step": 73
    },
    {
      "epoch": 0.2195845697329377,
      "grad_norm": 0.4334982633590698,
      "learning_rate": 7.833827893175074e-06,
      "loss": 1.8644,
      "step": 74
    },
    {
      "epoch": 0.22255192878338279,
      "grad_norm": 0.4691421687602997,
      "learning_rate": 7.804154302670623e-06,
      "loss": 1.8397,
      "step": 75
    },
    {
      "epoch": 0.22551928783382788,
      "grad_norm": 0.47831249237060547,
      "learning_rate": 7.774480712166173e-06,
      "loss": 1.8724,
      "step": 76
    },
    {
      "epoch": 0.228486646884273,
      "grad_norm": 0.4349431097507477,
      "learning_rate": 7.744807121661722e-06,
      "loss": 1.86,
      "step": 77
    },
    {
      "epoch": 0.2314540059347181,
      "grad_norm": 0.4444494843482971,
      "learning_rate": 7.71513353115727e-06,
      "loss": 1.8684,
      "step": 78
    },
    {
      "epoch": 0.2344213649851632,
      "grad_norm": 0.42169326543807983,
      "learning_rate": 7.68545994065282e-06,
      "loss": 1.8684,
      "step": 79
    },
    {
      "epoch": 0.23738872403560832,
      "grad_norm": 0.45169129967689514,
      "learning_rate": 7.655786350148369e-06,
      "loss": 1.8451,
      "step": 80
    },
    {
      "epoch": 0.2403560830860534,
      "grad_norm": 0.5155013203620911,
      "learning_rate": 7.626112759643918e-06,
      "loss": 1.8165,
      "step": 81
    },
    {
      "epoch": 0.2433234421364985,
      "grad_norm": 0.6931116580963135,
      "learning_rate": 7.5964391691394664e-06,
      "loss": 1.885,
      "step": 82
    },
    {
      "epoch": 0.24629080118694363,
      "grad_norm": 0.4700143039226532,
      "learning_rate": 7.566765578635016e-06,
      "loss": 1.8673,
      "step": 83
    },
    {
      "epoch": 0.24925816023738873,
      "grad_norm": 0.43717002868652344,
      "learning_rate": 7.537091988130565e-06,
      "loss": 1.8186,
      "step": 84
    },
    {
      "epoch": 0.2522255192878338,
      "grad_norm": 0.4771670997142792,
      "learning_rate": 7.507418397626114e-06,
      "loss": 1.8373,
      "step": 85
    },
    {
      "epoch": 0.2551928783382789,
      "grad_norm": 0.4726479947566986,
      "learning_rate": 7.477744807121662e-06,
      "loss": 1.8542,
      "step": 86
    },
    {
      "epoch": 0.258160237388724,
      "grad_norm": 0.6090665459632874,
      "learning_rate": 7.4480712166172105e-06,
      "loss": 1.808,
      "step": 87
    },
    {
      "epoch": 0.26112759643916916,
      "grad_norm": 0.46400222182273865,
      "learning_rate": 7.41839762611276e-06,
      "loss": 1.8164,
      "step": 88
    },
    {
      "epoch": 0.26409495548961426,
      "grad_norm": 0.4562290906906128,
      "learning_rate": 7.388724035608309e-06,
      "loss": 1.8788,
      "step": 89
    },
    {
      "epoch": 0.26706231454005935,
      "grad_norm": 0.4658115804195404,
      "learning_rate": 7.359050445103858e-06,
      "loss": 1.817,
      "step": 90
    },
    {
      "epoch": 0.27002967359050445,
      "grad_norm": 0.4649907350540161,
      "learning_rate": 7.329376854599407e-06,
      "loss": 1.7972,
      "step": 91
    },
    {
      "epoch": 0.27299703264094954,
      "grad_norm": 0.434505432844162,
      "learning_rate": 7.299703264094956e-06,
      "loss": 1.7884,
      "step": 92
    },
    {
      "epoch": 0.27596439169139464,
      "grad_norm": 0.4366685450077057,
      "learning_rate": 7.270029673590505e-06,
      "loss": 1.7906,
      "step": 93
    },
    {
      "epoch": 0.2789317507418398,
      "grad_norm": 0.42524248361587524,
      "learning_rate": 7.2403560830860545e-06,
      "loss": 1.774,
      "step": 94
    },
    {
      "epoch": 0.2818991097922849,
      "grad_norm": 0.43574202060699463,
      "learning_rate": 7.210682492581603e-06,
      "loss": 1.8153,
      "step": 95
    },
    {
      "epoch": 0.28486646884273,
      "grad_norm": 0.511954128742218,
      "learning_rate": 7.181008902077153e-06,
      "loss": 1.7963,
      "step": 96
    },
    {
      "epoch": 0.2878338278931751,
      "grad_norm": 0.4813544750213623,
      "learning_rate": 7.151335311572701e-06,
      "loss": 1.8164,
      "step": 97
    },
    {
      "epoch": 0.29080118694362017,
      "grad_norm": 0.44782015681266785,
      "learning_rate": 7.12166172106825e-06,
      "loss": 1.8106,
      "step": 98
    },
    {
      "epoch": 0.29376854599406527,
      "grad_norm": 0.4343424141407013,
      "learning_rate": 7.091988130563799e-06,
      "loss": 1.7798,
      "step": 99
    },
    {
      "epoch": 0.29673590504451036,
      "grad_norm": 0.4410034418106079,
      "learning_rate": 7.062314540059347e-06,
      "loss": 1.8321,
      "step": 100
    },
    {
      "epoch": 0.2997032640949555,
      "grad_norm": 0.4282773733139038,
      "learning_rate": 7.032640949554897e-06,
      "loss": 1.7347,
      "step": 101
    },
    {
      "epoch": 0.3026706231454006,
      "grad_norm": 0.4454915225505829,
      "learning_rate": 7.0029673590504455e-06,
      "loss": 1.7774,
      "step": 102
    },
    {
      "epoch": 0.3056379821958457,
      "grad_norm": 0.5306017398834229,
      "learning_rate": 6.973293768545994e-06,
      "loss": 1.8005,
      "step": 103
    },
    {
      "epoch": 0.3086053412462908,
      "grad_norm": 0.7177593111991882,
      "learning_rate": 6.943620178041544e-06,
      "loss": 1.8215,
      "step": 104
    },
    {
      "epoch": 0.3115727002967359,
      "grad_norm": 0.41175395250320435,
      "learning_rate": 6.913946587537092e-06,
      "loss": 1.7983,
      "step": 105
    },
    {
      "epoch": 0.314540059347181,
      "grad_norm": 0.4323279857635498,
      "learning_rate": 6.884272997032642e-06,
      "loss": 1.7808,
      "step": 106
    },
    {
      "epoch": 0.31750741839762614,
      "grad_norm": 0.41738948225975037,
      "learning_rate": 6.85459940652819e-06,
      "loss": 1.7846,
      "step": 107
    },
    {
      "epoch": 0.32047477744807124,
      "grad_norm": 0.42175182700157166,
      "learning_rate": 6.82492581602374e-06,
      "loss": 1.7798,
      "step": 108
    },
    {
      "epoch": 0.32344213649851633,
      "grad_norm": 0.4261856973171234,
      "learning_rate": 6.795252225519289e-06,
      "loss": 1.7898,
      "step": 109
    },
    {
      "epoch": 0.3264094955489614,
      "grad_norm": 0.541006326675415,
      "learning_rate": 6.765578635014838e-06,
      "loss": 1.7475,
      "step": 110
    },
    {
      "epoch": 0.3293768545994065,
      "grad_norm": 0.5069206953048706,
      "learning_rate": 6.735905044510387e-06,
      "loss": 1.8048,
      "step": 111
    },
    {
      "epoch": 0.3323442136498516,
      "grad_norm": 0.4549408555030823,
      "learning_rate": 6.7062314540059345e-06,
      "loss": 1.7594,
      "step": 112
    },
    {
      "epoch": 0.3353115727002967,
      "grad_norm": 0.5172181129455566,
      "learning_rate": 6.676557863501484e-06,
      "loss": 1.7614,
      "step": 113
    },
    {
      "epoch": 0.33827893175074186,
      "grad_norm": 0.4310220181941986,
      "learning_rate": 6.646884272997033e-06,
      "loss": 1.7586,
      "step": 114
    },
    {
      "epoch": 0.34124629080118696,
      "grad_norm": 0.48071998357772827,
      "learning_rate": 6.617210682492582e-06,
      "loss": 1.7556,
      "step": 115
    },
    {
      "epoch": 0.34421364985163205,
      "grad_norm": 0.430143803358078,
      "learning_rate": 6.587537091988131e-06,
      "loss": 1.7271,
      "step": 116
    },
    {
      "epoch": 0.34718100890207715,
      "grad_norm": 0.45938631892204285,
      "learning_rate": 6.55786350148368e-06,
      "loss": 1.7229,
      "step": 117
    },
    {
      "epoch": 0.35014836795252224,
      "grad_norm": 0.4944155812263489,
      "learning_rate": 6.528189910979229e-06,
      "loss": 1.7565,
      "step": 118
    },
    {
      "epoch": 0.35311572700296734,
      "grad_norm": 0.4696439504623413,
      "learning_rate": 6.4985163204747785e-06,
      "loss": 1.734,
      "step": 119
    },
    {
      "epoch": 0.3560830860534125,
      "grad_norm": 0.46470603346824646,
      "learning_rate": 6.468842729970327e-06,
      "loss": 1.7337,
      "step": 120
    },
    {
      "epoch": 0.3590504451038576,
      "grad_norm": 0.5239853858947754,
      "learning_rate": 6.439169139465876e-06,
      "loss": 1.7614,
      "step": 121
    },
    {
      "epoch": 0.3620178041543027,
      "grad_norm": 0.44966644048690796,
      "learning_rate": 6.409495548961425e-06,
      "loss": 1.7274,
      "step": 122
    },
    {
      "epoch": 0.3649851632047478,
      "grad_norm": 0.4592658579349518,
      "learning_rate": 6.379821958456974e-06,
      "loss": 1.7262,
      "step": 123
    },
    {
      "epoch": 0.36795252225519287,
      "grad_norm": 0.4471616744995117,
      "learning_rate": 6.3501483679525235e-06,
      "loss": 1.7152,
      "step": 124
    },
    {
      "epoch": 0.37091988130563797,
      "grad_norm": 0.4259214997291565,
      "learning_rate": 6.320474777448071e-06,
      "loss": 1.6925,
      "step": 125
    },
    {
      "epoch": 0.37388724035608306,
      "grad_norm": 0.43465620279312134,
      "learning_rate": 6.29080118694362e-06,
      "loss": 1.7104,
      "step": 126
    },
    {
      "epoch": 0.3768545994065282,
      "grad_norm": 0.43034180998802185,
      "learning_rate": 6.2611275964391694e-06,
      "loss": 1.7448,
      "step": 127
    },
    {
      "epoch": 0.3798219584569733,
      "grad_norm": 0.4406508207321167,
      "learning_rate": 6.231454005934718e-06,
      "loss": 1.7299,
      "step": 128
    },
    {
      "epoch": 0.3827893175074184,
      "grad_norm": 0.4358867406845093,
      "learning_rate": 6.201780415430268e-06,
      "loss": 1.7094,
      "step": 129
    },
    {
      "epoch": 0.3857566765578635,
      "grad_norm": 0.46925172209739685,
      "learning_rate": 6.172106824925816e-06,
      "loss": 1.7138,
      "step": 130
    },
    {
      "epoch": 0.3887240356083086,
      "grad_norm": 0.45740288496017456,
      "learning_rate": 6.142433234421366e-06,
      "loss": 1.7344,
      "step": 131
    },
    {
      "epoch": 0.3916913946587537,
      "grad_norm": 0.4303436577320099,
      "learning_rate": 6.112759643916914e-06,
      "loss": 1.7016,
      "step": 132
    },
    {
      "epoch": 0.39465875370919884,
      "grad_norm": 0.4608151316642761,
      "learning_rate": 6.083086053412464e-06,
      "loss": 1.7213,
      "step": 133
    },
    {
      "epoch": 0.39762611275964393,
      "grad_norm": 0.4390476942062378,
      "learning_rate": 6.0534124629080126e-06,
      "loss": 1.7341,
      "step": 134
    },
    {
      "epoch": 0.40059347181008903,
      "grad_norm": 0.44573843479156494,
      "learning_rate": 6.023738872403562e-06,
      "loss": 1.7235,
      "step": 135
    },
    {
      "epoch": 0.4035608308605341,
      "grad_norm": 0.45029526948928833,
      "learning_rate": 5.994065281899111e-06,
      "loss": 1.7148,
      "step": 136
    },
    {
      "epoch": 0.4065281899109792,
      "grad_norm": 0.5462357401847839,
      "learning_rate": 5.964391691394659e-06,
      "loss": 1.6683,
      "step": 137
    },
    {
      "epoch": 0.4094955489614243,
      "grad_norm": 0.5839945673942566,
      "learning_rate": 5.934718100890208e-06,
      "loss": 1.6809,
      "step": 138
    },
    {
      "epoch": 0.4124629080118694,
      "grad_norm": 0.5675076246261597,
      "learning_rate": 5.905044510385757e-06,
      "loss": 1.6973,
      "step": 139
    },
    {
      "epoch": 0.41543026706231456,
      "grad_norm": 0.45286524295806885,
      "learning_rate": 5.875370919881306e-06,
      "loss": 1.7043,
      "step": 140
    },
    {
      "epoch": 0.41839762611275966,
      "grad_norm": 0.43389052152633667,
      "learning_rate": 5.845697329376855e-06,
      "loss": 1.7251,
      "step": 141
    },
    {
      "epoch": 0.42136498516320475,
      "grad_norm": 0.4973659813404083,
      "learning_rate": 5.8160237388724035e-06,
      "loss": 1.7092,
      "step": 142
    },
    {
      "epoch": 0.42433234421364985,
      "grad_norm": 0.5145835876464844,
      "learning_rate": 5.786350148367953e-06,
      "loss": 1.7155,
      "step": 143
    },
    {
      "epoch": 0.42729970326409494,
      "grad_norm": 0.42716243863105774,
      "learning_rate": 5.756676557863502e-06,
      "loss": 1.705,
      "step": 144
    },
    {
      "epoch": 0.43026706231454004,
      "grad_norm": 0.6541968584060669,
      "learning_rate": 5.727002967359051e-06,
      "loss": 1.6899,
      "step": 145
    },
    {
      "epoch": 0.4332344213649852,
      "grad_norm": 0.4450303912162781,
      "learning_rate": 5.6973293768546e-06,
      "loss": 1.7204,
      "step": 146
    },
    {
      "epoch": 0.4362017804154303,
      "grad_norm": 0.5097622871398926,
      "learning_rate": 5.667655786350149e-06,
      "loss": 1.7106,
      "step": 147
    },
    {
      "epoch": 0.4391691394658754,
      "grad_norm": 0.4638642370700836,
      "learning_rate": 5.637982195845698e-06,
      "loss": 1.6704,
      "step": 148
    },
    {
      "epoch": 0.4421364985163205,
      "grad_norm": 0.631533682346344,
      "learning_rate": 5.6083086053412475e-06,
      "loss": 1.6867,
      "step": 149
    },
    {
      "epoch": 0.44510385756676557,
      "grad_norm": 0.43542999029159546,
      "learning_rate": 5.578635014836796e-06,
      "loss": 1.6724,
      "step": 150
    },
    {
      "epoch": 0.44807121661721067,
      "grad_norm": 0.4347829818725586,
      "learning_rate": 5.548961424332344e-06,
      "loss": 1.7161,
      "step": 151
    },
    {
      "epoch": 0.45103857566765576,
      "grad_norm": 0.42149391770362854,
      "learning_rate": 5.5192878338278934e-06,
      "loss": 1.6849,
      "step": 152
    },
    {
      "epoch": 0.4540059347181009,
      "grad_norm": 0.43702206015586853,
      "learning_rate": 5.489614243323442e-06,
      "loss": 1.675,
      "step": 153
    },
    {
      "epoch": 0.456973293768546,
      "grad_norm": 0.43425336480140686,
      "learning_rate": 5.459940652818992e-06,
      "loss": 1.699,
      "step": 154
    },
    {
      "epoch": 0.4599406528189911,
      "grad_norm": 0.4301832616329193,
      "learning_rate": 5.43026706231454e-06,
      "loss": 1.6709,
      "step": 155
    },
    {
      "epoch": 0.4629080118694362,
      "grad_norm": 0.4581262767314911,
      "learning_rate": 5.40059347181009e-06,
      "loss": 1.6596,
      "step": 156
    },
    {
      "epoch": 0.4658753709198813,
      "grad_norm": 0.4849711060523987,
      "learning_rate": 5.370919881305638e-06,
      "loss": 1.6597,
      "step": 157
    },
    {
      "epoch": 0.4688427299703264,
      "grad_norm": 0.5430138111114502,
      "learning_rate": 5.341246290801188e-06,
      "loss": 1.6597,
      "step": 158
    },
    {
      "epoch": 0.47181008902077154,
      "grad_norm": 0.4438075125217438,
      "learning_rate": 5.3115727002967366e-06,
      "loss": 1.6681,
      "step": 159
    },
    {
      "epoch": 0.47477744807121663,
      "grad_norm": 0.4786677956581116,
      "learning_rate": 5.281899109792285e-06,
      "loss": 1.6831,
      "step": 160
    },
    {
      "epoch": 0.47774480712166173,
      "grad_norm": 0.4415838122367859,
      "learning_rate": 5.252225519287835e-06,
      "loss": 1.6713,
      "step": 161
    },
    {
      "epoch": 0.4807121661721068,
      "grad_norm": 0.7021270394325256,
      "learning_rate": 5.222551928783383e-06,
      "loss": 1.6688,
      "step": 162
    },
    {
      "epoch": 0.4836795252225519,
      "grad_norm": 0.4373079538345337,
      "learning_rate": 5.192878338278933e-06,
      "loss": 1.6945,
      "step": 163
    },
    {
      "epoch": 0.486646884272997,
      "grad_norm": 0.4287460148334503,
      "learning_rate": 5.163204747774481e-06,
      "loss": 1.666,
      "step": 164
    },
    {
      "epoch": 0.4896142433234421,
      "grad_norm": 0.5587551593780518,
      "learning_rate": 5.133531157270029e-06,
      "loss": 1.6851,
      "step": 165
    },
    {
      "epoch": 0.49258160237388726,
      "grad_norm": 0.46015986800193787,
      "learning_rate": 5.103857566765579e-06,
      "loss": 1.6493,
      "step": 166
    },
    {
      "epoch": 0.49554896142433236,
      "grad_norm": 0.46552950143814087,
      "learning_rate": 5.0741839762611275e-06,
      "loss": 1.6801,
      "step": 167
    },
    {
      "epoch": 0.49851632047477745,
      "grad_norm": 0.5727176666259766,
      "learning_rate": 5.044510385756677e-06,
      "loss": 1.6788,
      "step": 168
    },
    {
      "epoch": 0.5014836795252225,
      "grad_norm": 0.44388559460639954,
      "learning_rate": 5.014836795252226e-06,
      "loss": 1.6576,
      "step": 169
    },
    {
      "epoch": 0.5044510385756676,
      "grad_norm": 0.4283711910247803,
      "learning_rate": 4.985163204747775e-06,
      "loss": 1.6614,
      "step": 170
    },
    {
      "epoch": 0.5074183976261127,
      "grad_norm": 0.5084763169288635,
      "learning_rate": 4.955489614243324e-06,
      "loss": 1.6682,
      "step": 171
    },
    {
      "epoch": 0.5103857566765578,
      "grad_norm": 0.5325101613998413,
      "learning_rate": 4.925816023738873e-06,
      "loss": 1.6571,
      "step": 172
    },
    {
      "epoch": 0.5133531157270029,
      "grad_norm": 0.44875359535217285,
      "learning_rate": 4.896142433234421e-06,
      "loss": 1.6634,
      "step": 173
    },
    {
      "epoch": 0.516320474777448,
      "grad_norm": 0.526423990726471,
      "learning_rate": 4.866468842729971e-06,
      "loss": 1.6602,
      "step": 174
    },
    {
      "epoch": 0.5192878338278932,
      "grad_norm": 0.45827770233154297,
      "learning_rate": 4.836795252225519e-06,
      "loss": 1.649,
      "step": 175
    },
    {
      "epoch": 0.5222551928783383,
      "grad_norm": 0.6596139073371887,
      "learning_rate": 4.807121661721069e-06,
      "loss": 1.6968,
      "step": 176
    },
    {
      "epoch": 0.5252225519287834,
      "grad_norm": 0.4827237129211426,
      "learning_rate": 4.7774480712166174e-06,
      "loss": 1.6739,
      "step": 177
    },
    {
      "epoch": 0.5281899109792285,
      "grad_norm": 0.4733069837093353,
      "learning_rate": 4.747774480712167e-06,
      "loss": 1.6612,
      "step": 178
    },
    {
      "epoch": 0.5311572700296736,
      "grad_norm": 0.486103355884552,
      "learning_rate": 4.718100890207716e-06,
      "loss": 1.6303,
      "step": 179
    },
    {
      "epoch": 0.5341246290801187,
      "grad_norm": 0.5725518465042114,
      "learning_rate": 4.688427299703264e-06,
      "loss": 1.7068,
      "step": 180
    },
    {
      "epoch": 0.5370919881305638,
      "grad_norm": 0.4758003354072571,
      "learning_rate": 4.658753709198813e-06,
      "loss": 1.6436,
      "step": 181
    },
    {
      "epoch": 0.5400593471810089,
      "grad_norm": 0.45277777314186096,
      "learning_rate": 4.629080118694362e-06,
      "loss": 1.6318,
      "step": 182
    },
    {
      "epoch": 0.543026706231454,
      "grad_norm": 0.7258154153823853,
      "learning_rate": 4.599406528189911e-06,
      "loss": 1.6428,
      "step": 183
    },
    {
      "epoch": 0.5459940652818991,
      "grad_norm": 0.5648652911186218,
      "learning_rate": 4.5697329376854606e-06,
      "loss": 1.6261,
      "step": 184
    },
    {
      "epoch": 0.5489614243323442,
      "grad_norm": 0.45321282744407654,
      "learning_rate": 4.540059347181009e-06,
      "loss": 1.6354,
      "step": 185
    },
    {
      "epoch": 0.5519287833827893,
      "grad_norm": 0.5343035459518433,
      "learning_rate": 4.510385756676558e-06,
      "loss": 1.6547,
      "step": 186
    },
    {
      "epoch": 0.5548961424332344,
      "grad_norm": 0.5379279255867004,
      "learning_rate": 4.480712166172107e-06,
      "loss": 1.6224,
      "step": 187
    },
    {
      "epoch": 0.5578635014836796,
      "grad_norm": 0.4859595000743866,
      "learning_rate": 4.451038575667656e-06,
      "loss": 1.6717,
      "step": 188
    },
    {
      "epoch": 0.5608308605341247,
      "grad_norm": 0.6033043265342712,
      "learning_rate": 4.4213649851632055e-06,
      "loss": 1.6319,
      "step": 189
    },
    {
      "epoch": 0.5637982195845698,
      "grad_norm": 0.46907198429107666,
      "learning_rate": 4.391691394658754e-06,
      "loss": 1.6432,
      "step": 190
    },
    {
      "epoch": 0.5667655786350149,
      "grad_norm": 0.47217652201652527,
      "learning_rate": 4.362017804154303e-06,
      "loss": 1.667,
      "step": 191
    },
    {
      "epoch": 0.56973293768546,
      "grad_norm": 0.48675742745399475,
      "learning_rate": 4.332344213649852e-06,
      "loss": 1.6309,
      "step": 192
    },
    {
      "epoch": 0.5727002967359051,
      "grad_norm": 0.777526319026947,
      "learning_rate": 4.302670623145401e-06,
      "loss": 1.6237,
      "step": 193
    },
    {
      "epoch": 0.5756676557863502,
      "grad_norm": 0.4463180899620056,
      "learning_rate": 4.27299703264095e-06,
      "loss": 1.6016,
      "step": 194
    },
    {
      "epoch": 0.5786350148367952,
      "grad_norm": 0.5856812596321106,
      "learning_rate": 4.243323442136499e-06,
      "loss": 1.6713,
      "step": 195
    },
    {
      "epoch": 0.5816023738872403,
      "grad_norm": 0.4704248905181885,
      "learning_rate": 4.213649851632048e-06,
      "loss": 1.6342,
      "step": 196
    },
    {
      "epoch": 0.5845697329376854,
      "grad_norm": 0.49075794219970703,
      "learning_rate": 4.183976261127597e-06,
      "loss": 1.6421,
      "step": 197
    },
    {
      "epoch": 0.5875370919881305,
      "grad_norm": 0.5137872099876404,
      "learning_rate": 4.154302670623146e-06,
      "loss": 1.6177,
      "step": 198
    },
    {
      "epoch": 0.5905044510385756,
      "grad_norm": 0.4413808286190033,
      "learning_rate": 4.124629080118695e-06,
      "loss": 1.6336,
      "step": 199
    },
    {
      "epoch": 0.5934718100890207,
      "grad_norm": 0.6047279834747314,
      "learning_rate": 4.094955489614243e-06,
      "loss": 1.6224,
      "step": 200
    },
    {
      "epoch": 0.5964391691394659,
      "grad_norm": 0.7798176407814026,
      "learning_rate": 4.065281899109793e-06,
      "loss": 1.626,
      "step": 201
    },
    {
      "epoch": 0.599406528189911,
      "grad_norm": 0.560275137424469,
      "learning_rate": 4.0356083086053414e-06,
      "loss": 1.6624,
      "step": 202
    },
    {
      "epoch": 0.6023738872403561,
      "grad_norm": 0.47962644696235657,
      "learning_rate": 4.005934718100891e-06,
      "loss": 1.6302,
      "step": 203
    },
    {
      "epoch": 0.6053412462908012,
      "grad_norm": 0.7026333212852478,
      "learning_rate": 3.97626112759644e-06,
      "loss": 1.6245,
      "step": 204
    },
    {
      "epoch": 0.6083086053412463,
      "grad_norm": 0.5532521605491638,
      "learning_rate": 3.946587537091989e-06,
      "loss": 1.6518,
      "step": 205
    },
    {
      "epoch": 0.6112759643916914,
      "grad_norm": 0.5071268677711487,
      "learning_rate": 3.916913946587537e-06,
      "loss": 1.6106,
      "step": 206
    },
    {
      "epoch": 0.6142433234421365,
      "grad_norm": 0.6073765158653259,
      "learning_rate": 3.887240356083086e-06,
      "loss": 1.616,
      "step": 207
    },
    {
      "epoch": 0.6172106824925816,
      "grad_norm": 0.6259579062461853,
      "learning_rate": 3.857566765578635e-06,
      "loss": 1.6174,
      "step": 208
    },
    {
      "epoch": 0.6201780415430267,
      "grad_norm": 0.45534420013427734,
      "learning_rate": 3.8278931750741846e-06,
      "loss": 1.6267,
      "step": 209
    },
    {
      "epoch": 0.6231454005934718,
      "grad_norm": 0.43893375992774963,
      "learning_rate": 3.7982195845697332e-06,
      "loss": 1.6348,
      "step": 210
    },
    {
      "epoch": 0.6261127596439169,
      "grad_norm": 0.4532437026500702,
      "learning_rate": 3.7685459940652823e-06,
      "loss": 1.6111,
      "step": 211
    },
    {
      "epoch": 0.629080118694362,
      "grad_norm": 0.46695148944854736,
      "learning_rate": 3.738872403560831e-06,
      "loss": 1.6472,
      "step": 212
    },
    {
      "epoch": 0.6320474777448071,
      "grad_norm": 0.5121713280677795,
      "learning_rate": 3.70919881305638e-06,
      "loss": 1.61,
      "step": 213
    },
    {
      "epoch": 0.6350148367952523,
      "grad_norm": 0.5956472158432007,
      "learning_rate": 3.679525222551929e-06,
      "loss": 1.6268,
      "step": 214
    },
    {
      "epoch": 0.6379821958456974,
      "grad_norm": 0.4677529036998749,
      "learning_rate": 3.649851632047478e-06,
      "loss": 1.6208,
      "step": 215
    },
    {
      "epoch": 0.6409495548961425,
      "grad_norm": 0.8256700038909912,
      "learning_rate": 3.6201780415430273e-06,
      "loss": 1.6537,
      "step": 216
    },
    {
      "epoch": 0.6439169139465876,
      "grad_norm": 0.4774104356765747,
      "learning_rate": 3.5905044510385763e-06,
      "loss": 1.6777,
      "step": 217
    },
    {
      "epoch": 0.6468842729970327,
      "grad_norm": 0.45575234293937683,
      "learning_rate": 3.560830860534125e-06,
      "loss": 1.6277,
      "step": 218
    },
    {
      "epoch": 0.6498516320474778,
      "grad_norm": 0.4724850654602051,
      "learning_rate": 3.5311572700296736e-06,
      "loss": 1.6176,
      "step": 219
    },
    {
      "epoch": 0.6528189910979229,
      "grad_norm": 0.502692699432373,
      "learning_rate": 3.5014836795252227e-06,
      "loss": 1.639,
      "step": 220
    },
    {
      "epoch": 0.655786350148368,
      "grad_norm": 0.5757880806922913,
      "learning_rate": 3.471810089020772e-06,
      "loss": 1.6592,
      "step": 221
    },
    {
      "epoch": 0.658753709198813,
      "grad_norm": 0.44157612323760986,
      "learning_rate": 3.442136498516321e-06,
      "loss": 1.6274,
      "step": 222
    },
    {
      "epoch": 0.6617210682492581,
      "grad_norm": 0.41496533155441284,
      "learning_rate": 3.41246290801187e-06,
      "loss": 1.6375,
      "step": 223
    },
    {
      "epoch": 0.6646884272997032,
      "grad_norm": 0.46265709400177,
      "learning_rate": 3.382789317507419e-06,
      "loss": 1.6031,
      "step": 224
    },
    {
      "epoch": 0.6676557863501483,
      "grad_norm": 0.46054109930992126,
      "learning_rate": 3.3531157270029673e-06,
      "loss": 1.6139,
      "step": 225
    },
    {
      "epoch": 0.6706231454005934,
      "grad_norm": 0.5033138990402222,
      "learning_rate": 3.3234421364985163e-06,
      "loss": 1.6115,
      "step": 226
    },
    {
      "epoch": 0.6735905044510386,
      "grad_norm": 0.44652995467185974,
      "learning_rate": 3.2937685459940654e-06,
      "loss": 1.6074,
      "step": 227
    },
    {
      "epoch": 0.6765578635014837,
      "grad_norm": 0.46622881293296814,
      "learning_rate": 3.2640949554896145e-06,
      "loss": 1.6329,
      "step": 228
    },
    {
      "epoch": 0.6795252225519288,
      "grad_norm": 0.4660681486129761,
      "learning_rate": 3.2344213649851636e-06,
      "loss": 1.6188,
      "step": 229
    },
    {
      "epoch": 0.6824925816023739,
      "grad_norm": 0.47356855869293213,
      "learning_rate": 3.2047477744807127e-06,
      "loss": 1.6059,
      "step": 230
    },
    {
      "epoch": 0.685459940652819,
      "grad_norm": 0.4421982169151306,
      "learning_rate": 3.1750741839762617e-06,
      "loss": 1.62,
      "step": 231
    },
    {
      "epoch": 0.6884272997032641,
      "grad_norm": 0.4574885964393616,
      "learning_rate": 3.14540059347181e-06,
      "loss": 1.6552,
      "step": 232
    },
    {
      "epoch": 0.6913946587537092,
      "grad_norm": 0.46729278564453125,
      "learning_rate": 3.115727002967359e-06,
      "loss": 1.5724,
      "step": 233
    },
    {
      "epoch": 0.6943620178041543,
      "grad_norm": 0.48040372133255005,
      "learning_rate": 3.086053412462908e-06,
      "loss": 1.5932,
      "step": 234
    },
    {
      "epoch": 0.6973293768545994,
      "grad_norm": 0.4446007311344147,
      "learning_rate": 3.056379821958457e-06,
      "loss": 1.6108,
      "step": 235
    },
    {
      "epoch": 0.7002967359050445,
      "grad_norm": 0.6134002804756165,
      "learning_rate": 3.0267062314540063e-06,
      "loss": 1.5658,
      "step": 236
    },
    {
      "epoch": 0.7032640949554896,
      "grad_norm": 0.5650774240493774,
      "learning_rate": 2.9970326409495554e-06,
      "loss": 1.6165,
      "step": 237
    },
    {
      "epoch": 0.7062314540059347,
      "grad_norm": 0.5392011404037476,
      "learning_rate": 2.967359050445104e-06,
      "loss": 1.616,
      "step": 238
    },
    {
      "epoch": 0.7091988130563798,
      "grad_norm": 0.5072977542877197,
      "learning_rate": 2.937685459940653e-06,
      "loss": 1.6487,
      "step": 239
    },
    {
      "epoch": 0.712166172106825,
      "grad_norm": 0.4656478762626648,
      "learning_rate": 2.9080118694362018e-06,
      "loss": 1.606,
      "step": 240
    },
    {
      "epoch": 0.7151335311572701,
      "grad_norm": 0.46908023953437805,
      "learning_rate": 2.878338278931751e-06,
      "loss": 1.6197,
      "step": 241
    },
    {
      "epoch": 0.7181008902077152,
      "grad_norm": 0.46259936690330505,
      "learning_rate": 2.8486646884273e-06,
      "loss": 1.6262,
      "step": 242
    },
    {
      "epoch": 0.7210682492581603,
      "grad_norm": 0.47601285576820374,
      "learning_rate": 2.818991097922849e-06,
      "loss": 1.6429,
      "step": 243
    },
    {
      "epoch": 0.7240356083086054,
      "grad_norm": 0.4617674648761749,
      "learning_rate": 2.789317507418398e-06,
      "loss": 1.6269,
      "step": 244
    },
    {
      "epoch": 0.7270029673590505,
      "grad_norm": 0.751435399055481,
      "learning_rate": 2.7596439169139467e-06,
      "loss": 1.62,
      "step": 245
    },
    {
      "epoch": 0.7299703264094956,
      "grad_norm": 0.6891562938690186,
      "learning_rate": 2.729970326409496e-06,
      "loss": 1.5813,
      "step": 246
    },
    {
      "epoch": 0.7329376854599406,
      "grad_norm": 0.46907439827919006,
      "learning_rate": 2.700296735905045e-06,
      "loss": 1.6297,
      "step": 247
    },
    {
      "epoch": 0.7359050445103857,
      "grad_norm": 0.6780754327774048,
      "learning_rate": 2.670623145400594e-06,
      "loss": 1.6329,
      "step": 248
    },
    {
      "epoch": 0.7388724035608308,
      "grad_norm": 0.452311635017395,
      "learning_rate": 2.6409495548961426e-06,
      "loss": 1.6091,
      "step": 249
    },
    {
      "epoch": 0.7418397626112759,
      "grad_norm": 0.46394577622413635,
      "learning_rate": 2.6112759643916917e-06,
      "loss": 1.5904,
      "step": 250
    },
    {
      "epoch": 0.744807121661721,
      "grad_norm": 0.4922093152999878,
      "learning_rate": 2.5816023738872403e-06,
      "loss": 1.5897,
      "step": 251
    },
    {
      "epoch": 0.7477744807121661,
      "grad_norm": 0.7256433367729187,
      "learning_rate": 2.5519287833827894e-06,
      "loss": 1.5871,
      "step": 252
    },
    {
      "epoch": 0.7507418397626113,
      "grad_norm": 0.512948215007782,
      "learning_rate": 2.5222551928783385e-06,
      "loss": 1.589,
      "step": 253
    },
    {
      "epoch": 0.7537091988130564,
      "grad_norm": 0.5184987187385559,
      "learning_rate": 2.4925816023738876e-06,
      "loss": 1.5808,
      "step": 254
    },
    {
      "epoch": 0.7566765578635015,
      "grad_norm": 0.5400562882423401,
      "learning_rate": 2.4629080118694367e-06,
      "loss": 1.604,
      "step": 255
    },
    {
      "epoch": 0.7596439169139466,
      "grad_norm": 0.49138668179512024,
      "learning_rate": 2.4332344213649853e-06,
      "loss": 1.6144,
      "step": 256
    },
    {
      "epoch": 0.7626112759643917,
      "grad_norm": 0.5963372588157654,
      "learning_rate": 2.4035608308605344e-06,
      "loss": 1.67,
      "step": 257
    },
    {
      "epoch": 0.7655786350148368,
      "grad_norm": 0.4776369631290436,
      "learning_rate": 2.3738872403560835e-06,
      "loss": 1.5471,
      "step": 258
    },
    {
      "epoch": 0.7685459940652819,
      "grad_norm": 0.46721333265304565,
      "learning_rate": 2.344213649851632e-06,
      "loss": 1.585,
      "step": 259
    },
    {
      "epoch": 0.771513353115727,
      "grad_norm": 0.5722032189369202,
      "learning_rate": 2.314540059347181e-06,
      "loss": 1.6047,
      "step": 260
    },
    {
      "epoch": 0.7744807121661721,
      "grad_norm": 0.4744715392589569,
      "learning_rate": 2.2848664688427303e-06,
      "loss": 1.575,
      "step": 261
    },
    {
      "epoch": 0.7774480712166172,
      "grad_norm": 0.593866229057312,
      "learning_rate": 2.255192878338279e-06,
      "loss": 1.6163,
      "step": 262
    },
    {
      "epoch": 0.7804154302670623,
      "grad_norm": 0.4665597379207611,
      "learning_rate": 2.225519287833828e-06,
      "loss": 1.5885,
      "step": 263
    },
    {
      "epoch": 0.7833827893175074,
      "grad_norm": 0.4636242389678955,
      "learning_rate": 2.195845697329377e-06,
      "loss": 1.5775,
      "step": 264
    },
    {
      "epoch": 0.7863501483679525,
      "grad_norm": 0.48517903685569763,
      "learning_rate": 2.166172106824926e-06,
      "loss": 1.5902,
      "step": 265
    },
    {
      "epoch": 0.7893175074183977,
      "grad_norm": 0.45555201172828674,
      "learning_rate": 2.136498516320475e-06,
      "loss": 1.58,
      "step": 266
    },
    {
      "epoch": 0.7922848664688428,
      "grad_norm": 0.45862534642219543,
      "learning_rate": 2.106824925816024e-06,
      "loss": 1.5554,
      "step": 267
    },
    {
      "epoch": 0.7952522255192879,
      "grad_norm": 0.4659798741340637,
      "learning_rate": 2.077151335311573e-06,
      "loss": 1.5825,
      "step": 268
    },
    {
      "epoch": 0.798219584569733,
      "grad_norm": 0.5027198195457458,
      "learning_rate": 2.0474777448071216e-06,
      "loss": 1.5782,
      "step": 269
    },
    {
      "epoch": 0.8011869436201781,
      "grad_norm": 0.6607580184936523,
      "learning_rate": 2.0178041543026707e-06,
      "loss": 1.5949,
      "step": 270
    },
    {
      "epoch": 0.8041543026706232,
      "grad_norm": 0.5816552042961121,
      "learning_rate": 1.98813056379822e-06,
      "loss": 1.6091,
      "step": 271
    },
    {
      "epoch": 0.8071216617210683,
      "grad_norm": 0.48888540267944336,
      "learning_rate": 1.9584569732937684e-06,
      "loss": 1.5578,
      "step": 272
    },
    {
      "epoch": 0.8100890207715133,
      "grad_norm": 0.4814195930957794,
      "learning_rate": 1.9287833827893175e-06,
      "loss": 1.6032,
      "step": 273
    },
    {
      "epoch": 0.8130563798219584,
      "grad_norm": 0.7086522579193115,
      "learning_rate": 1.8991097922848666e-06,
      "loss": 1.5919,
      "step": 274
    },
    {
      "epoch": 0.8160237388724035,
      "grad_norm": 0.4650281071662903,
      "learning_rate": 1.8694362017804155e-06,
      "loss": 1.583,
      "step": 275
    },
    {
      "epoch": 0.8189910979228486,
      "grad_norm": 0.4690192937850952,
      "learning_rate": 1.8397626112759646e-06,
      "loss": 1.5362,
      "step": 276
    },
    {
      "epoch": 0.8219584569732937,
      "grad_norm": 0.8297029137611389,
      "learning_rate": 1.8100890207715136e-06,
      "loss": 1.5959,
      "step": 277
    },
    {
      "epoch": 0.8249258160237388,
      "grad_norm": 0.518366813659668,
      "learning_rate": 1.7804154302670625e-06,
      "loss": 1.5935,
      "step": 278
    },
    {
      "epoch": 0.827893175074184,
      "grad_norm": 0.5858137011528015,
      "learning_rate": 1.7507418397626114e-06,
      "loss": 1.5556,
      "step": 279
    },
    {
      "epoch": 0.8308605341246291,
      "grad_norm": 0.5444852709770203,
      "learning_rate": 1.7210682492581604e-06,
      "loss": 1.5771,
      "step": 280
    },
    {
      "epoch": 0.8338278931750742,
      "grad_norm": 0.729849636554718,
      "learning_rate": 1.6913946587537095e-06,
      "loss": 1.5895,
      "step": 281
    },
    {
      "epoch": 0.8367952522255193,
      "grad_norm": 0.5181935429573059,
      "learning_rate": 1.6617210682492582e-06,
      "loss": 1.6054,
      "step": 282
    },
    {
      "epoch": 0.8397626112759644,
      "grad_norm": 0.48367366194725037,
      "learning_rate": 1.6320474777448073e-06,
      "loss": 1.617,
      "step": 283
    },
    {
      "epoch": 0.8427299703264095,
      "grad_norm": 0.46795350313186646,
      "learning_rate": 1.6023738872403563e-06,
      "loss": 1.5845,
      "step": 284
    },
    {
      "epoch": 0.8456973293768546,
      "grad_norm": 0.534070611000061,
      "learning_rate": 1.572700296735905e-06,
      "loss": 1.6164,
      "step": 285
    },
    {
      "epoch": 0.8486646884272997,
      "grad_norm": 0.6131719946861267,
      "learning_rate": 1.543026706231454e-06,
      "loss": 1.568,
      "step": 286
    },
    {
      "epoch": 0.8516320474777448,
      "grad_norm": 0.5441489815711975,
      "learning_rate": 1.5133531157270031e-06,
      "loss": 1.6022,
      "step": 287
    },
    {
      "epoch": 0.8545994065281899,
      "grad_norm": 0.6075683832168579,
      "learning_rate": 1.483679525222552e-06,
      "loss": 1.5795,
      "step": 288
    },
    {
      "epoch": 0.857566765578635,
      "grad_norm": 0.4614003300666809,
      "learning_rate": 1.4540059347181009e-06,
      "loss": 1.5886,
      "step": 289
    },
    {
      "epoch": 0.8605341246290801,
      "grad_norm": 0.45874059200286865,
      "learning_rate": 1.42433234421365e-06,
      "loss": 1.5873,
      "step": 290
    },
    {
      "epoch": 0.8635014836795252,
      "grad_norm": 0.46380406618118286,
      "learning_rate": 1.394658753709199e-06,
      "loss": 1.5793,
      "step": 291
    },
    {
      "epoch": 0.8664688427299704,
      "grad_norm": 0.4983331561088562,
      "learning_rate": 1.364985163204748e-06,
      "loss": 1.6048,
      "step": 292
    },
    {
      "epoch": 0.8694362017804155,
      "grad_norm": 0.5502035617828369,
      "learning_rate": 1.335311572700297e-06,
      "loss": 1.5721,
      "step": 293
    },
    {
      "epoch": 0.8724035608308606,
      "grad_norm": 0.4903833270072937,
      "learning_rate": 1.3056379821958458e-06,
      "loss": 1.6266,
      "step": 294
    },
    {
      "epoch": 0.8753709198813057,
      "grad_norm": 0.7122423648834229,
      "learning_rate": 1.2759643916913947e-06,
      "loss": 1.5608,
      "step": 295
    },
    {
      "epoch": 0.8783382789317508,
      "grad_norm": 0.47747164964675903,
      "learning_rate": 1.2462908011869438e-06,
      "loss": 1.5856,
      "step": 296
    },
    {
      "epoch": 0.8813056379821959,
      "grad_norm": 0.499977171421051,
      "learning_rate": 1.2166172106824927e-06,
      "loss": 1.555,
      "step": 297
    },
    {
      "epoch": 0.884272997032641,
      "grad_norm": 0.4627906382083893,
      "learning_rate": 1.1869436201780417e-06,
      "loss": 1.5967,
      "step": 298
    },
    {
      "epoch": 0.887240356083086,
      "grad_norm": 0.4661864936351776,
      "learning_rate": 1.1572700296735906e-06,
      "loss": 1.5802,
      "step": 299
    },
    {
      "epoch": 0.8902077151335311,
      "grad_norm": 0.5619197487831116,
      "learning_rate": 1.1275964391691395e-06,
      "loss": 1.5766,
      "step": 300
    },
    {
      "epoch": 0.8931750741839762,
      "grad_norm": 0.5081275105476379,
      "learning_rate": 1.0979228486646885e-06,
      "loss": 1.6063,
      "step": 301
    },
    {
      "epoch": 0.8961424332344213,
      "grad_norm": 0.45451945066452026,
      "learning_rate": 1.0682492581602374e-06,
      "loss": 1.5594,
      "step": 302
    },
    {
      "epoch": 0.8991097922848664,
      "grad_norm": 0.44967764616012573,
      "learning_rate": 1.0385756676557865e-06,
      "loss": 1.557,
      "step": 303
    },
    {
      "epoch": 0.9020771513353115,
      "grad_norm": 0.4656503200531006,
      "learning_rate": 1.0089020771513354e-06,
      "loss": 1.5759,
      "step": 304
    },
    {
      "epoch": 0.9050445103857567,
      "grad_norm": 0.46309301257133484,
      "learning_rate": 9.792284866468842e-07,
      "loss": 1.6054,
      "step": 305
    },
    {
      "epoch": 0.9080118694362018,
      "grad_norm": 0.45573991537094116,
      "learning_rate": 9.495548961424333e-07,
      "loss": 1.5956,
      "step": 306
    },
    {
      "epoch": 0.9109792284866469,
      "grad_norm": 0.706634521484375,
      "learning_rate": 9.198813056379823e-07,
      "loss": 1.5274,
      "step": 307
    },
    {
      "epoch": 0.913946587537092,
      "grad_norm": 0.4630567133426666,
      "learning_rate": 8.902077151335312e-07,
      "loss": 1.6051,
      "step": 308
    },
    {
      "epoch": 0.9169139465875371,
      "grad_norm": 0.8506336808204651,
      "learning_rate": 8.605341246290802e-07,
      "loss": 1.5698,
      "step": 309
    },
    {
      "epoch": 0.9198813056379822,
      "grad_norm": 0.48109981417655945,
      "learning_rate": 8.308605341246291e-07,
      "loss": 1.5675,
      "step": 310
    },
    {
      "epoch": 0.9228486646884273,
      "grad_norm": 0.4885670840740204,
      "learning_rate": 8.011869436201782e-07,
      "loss": 1.5523,
      "step": 311
    },
    {
      "epoch": 0.9258160237388724,
      "grad_norm": 0.48182421922683716,
      "learning_rate": 7.71513353115727e-07,
      "loss": 1.5701,
      "step": 312
    },
    {
      "epoch": 0.9287833827893175,
      "grad_norm": 0.49333518743515015,
      "learning_rate": 7.41839762611276e-07,
      "loss": 1.5864,
      "step": 313
    },
    {
      "epoch": 0.9317507418397626,
      "grad_norm": 0.6127849221229553,
      "learning_rate": 7.12166172106825e-07,
      "loss": 1.6149,
      "step": 314
    },
    {
      "epoch": 0.9347181008902077,
      "grad_norm": 0.7807979583740234,
      "learning_rate": 6.82492581602374e-07,
      "loss": 1.6104,
      "step": 315
    },
    {
      "epoch": 0.9376854599406528,
      "grad_norm": 0.45212405920028687,
      "learning_rate": 6.528189910979229e-07,
      "loss": 1.5692,
      "step": 316
    },
    {
      "epoch": 0.9406528189910979,
      "grad_norm": 0.47415590286254883,
      "learning_rate": 6.231454005934719e-07,
      "loss": 1.5733,
      "step": 317
    },
    {
      "epoch": 0.9436201780415431,
      "grad_norm": 0.4578646719455719,
      "learning_rate": 5.934718100890209e-07,
      "loss": 1.5963,
      "step": 318
    },
    {
      "epoch": 0.9465875370919882,
      "grad_norm": 0.5850412249565125,
      "learning_rate": 5.637982195845697e-07,
      "loss": 1.5719,
      "step": 319
    },
    {
      "epoch": 0.9495548961424333,
      "grad_norm": 0.4761084020137787,
      "learning_rate": 5.341246290801187e-07,
      "loss": 1.5647,
      "step": 320
    },
    {
      "epoch": 0.9525222551928784,
      "grad_norm": 0.5709928870201111,
      "learning_rate": 5.044510385756677e-07,
      "loss": 1.5546,
      "step": 321
    },
    {
      "epoch": 0.9554896142433235,
      "grad_norm": 0.48186010122299194,
      "learning_rate": 4.7477744807121665e-07,
      "loss": 1.5733,
      "step": 322
    },
    {
      "epoch": 0.9584569732937686,
      "grad_norm": 0.45225703716278076,
      "learning_rate": 4.451038575667656e-07,
      "loss": 1.5724,
      "step": 323
    },
    {
      "epoch": 0.9614243323442137,
      "grad_norm": 0.46753621101379395,
      "learning_rate": 4.1543026706231454e-07,
      "loss": 1.6256,
      "step": 324
    },
    {
      "epoch": 0.9643916913946587,
      "grad_norm": 0.48810645937919617,
      "learning_rate": 3.857566765578635e-07,
      "loss": 1.6059,
      "step": 325
    },
    {
      "epoch": 0.9673590504451038,
      "grad_norm": 0.7288530468940735,
      "learning_rate": 3.560830860534125e-07,
      "loss": 1.5805,
      "step": 326
    },
    {
      "epoch": 0.9703264094955489,
      "grad_norm": 0.47065672278404236,
      "learning_rate": 3.2640949554896146e-07,
      "loss": 1.5776,
      "step": 327
    },
    {
      "epoch": 0.973293768545994,
      "grad_norm": 0.461771696805954,
      "learning_rate": 2.9673590504451043e-07,
      "loss": 1.5667,
      "step": 328
    },
    {
      "epoch": 0.9762611275964391,
      "grad_norm": 0.45015615224838257,
      "learning_rate": 2.6706231454005935e-07,
      "loss": 1.5761,
      "step": 329
    },
    {
      "epoch": 0.9792284866468842,
      "grad_norm": 0.49132898449897766,
      "learning_rate": 2.3738872403560833e-07,
      "loss": 1.5619,
      "step": 330
    },
    {
      "epoch": 0.9821958456973294,
      "grad_norm": 0.4527791142463684,
      "learning_rate": 2.0771513353115727e-07,
      "loss": 1.5594,
      "step": 331
    },
    {
      "epoch": 0.9851632047477745,
      "grad_norm": 0.4398275315761566,
      "learning_rate": 1.7804154302670624e-07,
      "loss": 1.5944,
      "step": 332
    },
    {
      "epoch": 0.9881305637982196,
      "grad_norm": 0.4522647261619568,
      "learning_rate": 1.4836795252225522e-07,
      "loss": 1.6167,
      "step": 333
    },
    {
      "epoch": 0.9910979228486647,
      "grad_norm": 0.452622652053833,
      "learning_rate": 1.1869436201780416e-07,
      "loss": 1.5632,
      "step": 334
    },
    {
      "epoch": 0.9940652818991098,
      "grad_norm": 0.778292715549469,
      "learning_rate": 8.902077151335312e-08,
      "loss": 1.6169,
      "step": 335
    },
    {
      "epoch": 0.9970326409495549,
      "grad_norm": 0.47834548354148865,
      "learning_rate": 5.934718100890208e-08,
      "loss": 1.6213,
      "step": 336
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.45398491621017456,
      "learning_rate": 2.967359050445104e-08,
      "loss": 1.5757,
      "step": 337
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 337,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.5385960380366848e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}