{
  "best_metric": 0.28087058663368225,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.08219459570533237,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0004109729785266619,
      "grad_norm": 99.82452392578125,
      "learning_rate": 1e-05,
      "loss": 8.6848,
      "step": 1
    },
    {
      "epoch": 0.0004109729785266619,
      "eval_loss": 2.258553981781006,
      "eval_runtime": 299.2546,
      "eval_samples_per_second": 13.697,
      "eval_steps_per_second": 3.425,
      "step": 1
    },
    {
      "epoch": 0.0008219459570533238,
      "grad_norm": 102.91410064697266,
      "learning_rate": 2e-05,
      "loss": 8.7141,
      "step": 2
    },
    {
      "epoch": 0.0012329189355799856,
      "grad_norm": 88.38063049316406,
      "learning_rate": 3e-05,
      "loss": 7.6899,
      "step": 3
    },
    {
      "epoch": 0.0016438919141066476,
      "grad_norm": 65.69779968261719,
      "learning_rate": 4e-05,
      "loss": 6.2377,
      "step": 4
    },
    {
      "epoch": 0.0020548648926333093,
      "grad_norm": 38.535552978515625,
      "learning_rate": 5e-05,
      "loss": 4.8117,
      "step": 5
    },
    {
      "epoch": 0.0024658378711599712,
      "grad_norm": 49.42429733276367,
      "learning_rate": 6e-05,
      "loss": 3.4661,
      "step": 6
    },
    {
      "epoch": 0.002876810849686633,
      "grad_norm": 21.973173141479492,
      "learning_rate": 7e-05,
      "loss": 2.8121,
      "step": 7
    },
    {
      "epoch": 0.003287783828213295,
      "grad_norm": 11.168489456176758,
      "learning_rate": 8e-05,
      "loss": 2.3334,
      "step": 8
    },
    {
      "epoch": 0.003698756806739957,
      "grad_norm": 15.63986587524414,
      "learning_rate": 9e-05,
      "loss": 1.9635,
      "step": 9
    },
    {
      "epoch": 0.0041097297852666186,
      "grad_norm": 9.796710014343262,
      "learning_rate": 0.0001,
      "loss": 1.9314,
      "step": 10
    },
    {
      "epoch": 0.004520702763793281,
      "grad_norm": 6.944689750671387,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.7497,
      "step": 11
    },
    {
      "epoch": 0.0049316757423199424,
      "grad_norm": 4.380185127258301,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.4935,
      "step": 12
    },
    {
      "epoch": 0.005342648720846604,
      "grad_norm": 4.951937675476074,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.4793,
      "step": 13
    },
    {
      "epoch": 0.005753621699373266,
      "grad_norm": 126.91710662841797,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.5989,
      "step": 14
    },
    {
      "epoch": 0.006164594677899928,
      "grad_norm": 17.00688934326172,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.4243,
      "step": 15
    },
    {
      "epoch": 0.00657556765642659,
      "grad_norm": 4.84775972366333,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.4053,
      "step": 16
    },
    {
      "epoch": 0.006986540634953252,
      "grad_norm": 5.180408477783203,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.4272,
      "step": 17
    },
    {
      "epoch": 0.007397513613479914,
      "grad_norm": 3.69616961479187,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.531,
      "step": 18
    },
    {
      "epoch": 0.007808486592006576,
      "grad_norm": 3.7180888652801514,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.4695,
      "step": 19
    },
    {
      "epoch": 0.008219459570533237,
      "grad_norm": 4.4182305335998535,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.5831,
      "step": 20
    },
    {
      "epoch": 0.008630432549059899,
      "grad_norm": 3.2310924530029297,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.3927,
      "step": 21
    },
    {
      "epoch": 0.009041405527586562,
      "grad_norm": 3.0939457416534424,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.4012,
      "step": 22
    },
    {
      "epoch": 0.009452378506113223,
      "grad_norm": 3.1145191192626953,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.3784,
      "step": 23
    },
    {
      "epoch": 0.009863351484639885,
      "grad_norm": 3.215970993041992,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.3913,
      "step": 24
    },
    {
      "epoch": 0.010274324463166546,
      "grad_norm": 3.030649423599243,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.444,
      "step": 25
    },
    {
      "epoch": 0.010685297441693208,
      "grad_norm": 3.381289482116699,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.3824,
      "step": 26
    },
    {
      "epoch": 0.011096270420219871,
      "grad_norm": 3.311889886856079,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.4523,
      "step": 27
    },
    {
      "epoch": 0.011507243398746533,
      "grad_norm": 3.2556312084198,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.4612,
      "step": 28
    },
    {
      "epoch": 0.011918216377273194,
      "grad_norm": 3.117842674255371,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.3742,
      "step": 29
    },
    {
      "epoch": 0.012329189355799856,
      "grad_norm": 2.9957573413848877,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.3595,
      "step": 30
    },
    {
      "epoch": 0.012740162334326517,
      "grad_norm": 2.4581284523010254,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.4175,
      "step": 31
    },
    {
      "epoch": 0.01315113531285318,
      "grad_norm": 2.6345133781433105,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.2999,
      "step": 32
    },
    {
      "epoch": 0.013562108291379842,
      "grad_norm": 2.2750868797302246,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.1897,
      "step": 33
    },
    {
      "epoch": 0.013973081269906503,
      "grad_norm": 3.5257420539855957,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.4714,
      "step": 34
    },
    {
      "epoch": 0.014384054248433165,
      "grad_norm": 2.380465269088745,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.2285,
      "step": 35
    },
    {
      "epoch": 0.014795027226959828,
      "grad_norm": 3.312361717224121,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.4777,
      "step": 36
    },
    {
      "epoch": 0.01520600020548649,
      "grad_norm": 2.7706005573272705,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.3272,
      "step": 37
    },
    {
      "epoch": 0.015616973184013151,
      "grad_norm": 2.307110071182251,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.1559,
      "step": 38
    },
    {
      "epoch": 0.016027946162539813,
      "grad_norm": 2.3895437717437744,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.4341,
      "step": 39
    },
    {
      "epoch": 0.016438919141066474,
      "grad_norm": 2.2602431774139404,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.1556,
      "step": 40
    },
    {
      "epoch": 0.016849892119593136,
      "grad_norm": 1.9967049360275269,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.1198,
      "step": 41
    },
    {
      "epoch": 0.017260865098119797,
      "grad_norm": 2.033595561981201,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.1691,
      "step": 42
    },
    {
      "epoch": 0.017671838076646462,
      "grad_norm": 2.3476226329803467,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.2033,
      "step": 43
    },
    {
      "epoch": 0.018082811055173124,
      "grad_norm": 2.4622278213500977,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.2166,
      "step": 44
    },
    {
      "epoch": 0.018493784033699785,
      "grad_norm": 1.9892240762710571,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.0709,
      "step": 45
    },
    {
      "epoch": 0.018904757012226447,
      "grad_norm": 1.8769431114196777,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.0773,
      "step": 46
    },
    {
      "epoch": 0.019315729990753108,
      "grad_norm": 1.9373986721038818,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.0586,
      "step": 47
    },
    {
      "epoch": 0.01972670296927977,
      "grad_norm": 2.0078210830688477,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.1058,
      "step": 48
    },
    {
      "epoch": 0.02013767594780643,
      "grad_norm": 2.487668514251709,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.002,
      "step": 49
    },
    {
      "epoch": 0.020548648926333093,
      "grad_norm": 2.062596321105957,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.0264,
      "step": 50
    },
    {
      "epoch": 0.020548648926333093,
      "eval_loss": 0.4309638440608978,
      "eval_runtime": 303.0316,
      "eval_samples_per_second": 13.527,
      "eval_steps_per_second": 3.382,
      "step": 50
    },
    {
      "epoch": 0.020959621904859754,
      "grad_norm": 14.787125587463379,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.6475,
      "step": 51
    },
    {
      "epoch": 0.021370594883386416,
      "grad_norm": 8.229430198669434,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.0006,
      "step": 52
    },
    {
      "epoch": 0.02178156786191308,
      "grad_norm": 6.33250093460083,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.5564,
      "step": 53
    },
    {
      "epoch": 0.022192540840439742,
      "grad_norm": 2.954143524169922,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.3861,
      "step": 54
    },
    {
      "epoch": 0.022603513818966404,
      "grad_norm": 32.080081939697266,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.7222,
      "step": 55
    },
    {
      "epoch": 0.023014486797493065,
      "grad_norm": 3.0146777629852295,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.4118,
      "step": 56
    },
    {
      "epoch": 0.023425459776019727,
      "grad_norm": 3.2158637046813965,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.5024,
      "step": 57
    },
    {
      "epoch": 0.02383643275454639,
      "grad_norm": 4.174438953399658,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.3529,
      "step": 58
    },
    {
      "epoch": 0.02424740573307305,
      "grad_norm": 4.474514961242676,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.2999,
      "step": 59
    },
    {
      "epoch": 0.02465837871159971,
      "grad_norm": 2.1648683547973633,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.4036,
      "step": 60
    },
    {
      "epoch": 0.025069351690126373,
      "grad_norm": 2.116624355316162,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.3803,
      "step": 61
    },
    {
      "epoch": 0.025480324668653034,
      "grad_norm": 2.377899408340454,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.3239,
      "step": 62
    },
    {
      "epoch": 0.0258912976471797,
      "grad_norm": 5.777559757232666,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.2052,
      "step": 63
    },
    {
      "epoch": 0.02630227062570636,
      "grad_norm": 2.0995659828186035,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.1951,
      "step": 64
    },
    {
      "epoch": 0.026713243604233022,
      "grad_norm": 2.492811918258667,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.4303,
      "step": 65
    },
    {
      "epoch": 0.027124216582759684,
      "grad_norm": 2.533114194869995,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.3062,
      "step": 66
    },
    {
      "epoch": 0.027535189561286345,
      "grad_norm": 2.158507823944092,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.1784,
      "step": 67
    },
    {
      "epoch": 0.027946162539813007,
      "grad_norm": 2.376920223236084,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.2497,
      "step": 68
    },
    {
      "epoch": 0.02835713551833967,
      "grad_norm": 2.451343059539795,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.3026,
      "step": 69
    },
    {
      "epoch": 0.02876810849686633,
      "grad_norm": 2.025463819503784,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.2421,
      "step": 70
    },
    {
      "epoch": 0.02917908147539299,
      "grad_norm": 2.333592653274536,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.24,
      "step": 71
    },
    {
      "epoch": 0.029590054453919656,
      "grad_norm": 2.4560556411743164,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.4416,
      "step": 72
    },
    {
      "epoch": 0.030001027432446318,
      "grad_norm": 2.0353782176971436,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.2523,
      "step": 73
    },
    {
      "epoch": 0.03041200041097298,
      "grad_norm": 2.174056053161621,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.2772,
      "step": 74
    },
    {
      "epoch": 0.03082297338949964,
      "grad_norm": 1.9931261539459229,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.1877,
      "step": 75
    },
    {
      "epoch": 0.031233946368026302,
      "grad_norm": 2.3647842407226562,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.2875,
      "step": 76
    },
    {
      "epoch": 0.03164491934655297,
      "grad_norm": 1.9131511449813843,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.1767,
      "step": 77
    },
    {
      "epoch": 0.032055892325079625,
      "grad_norm": 2.5212457180023193,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.5935,
      "step": 78
    },
    {
      "epoch": 0.03246686530360629,
      "grad_norm": 2.2391068935394287,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.3051,
      "step": 79
    },
    {
      "epoch": 0.03287783828213295,
      "grad_norm": 1.9801801443099976,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.3152,
      "step": 80
    },
    {
      "epoch": 0.03328881126065961,
      "grad_norm": 1.7763584852218628,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.31,
      "step": 81
    },
    {
      "epoch": 0.03369978423918627,
      "grad_norm": 1.9331237077713013,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.2905,
      "step": 82
    },
    {
      "epoch": 0.034110757217712936,
      "grad_norm": 2.0055196285247803,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.2186,
      "step": 83
    },
    {
      "epoch": 0.034521730196239594,
      "grad_norm": 2.0965957641601562,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.2158,
      "step": 84
    },
    {
      "epoch": 0.03493270317476626,
      "grad_norm": 2.1323089599609375,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.261,
      "step": 85
    },
    {
      "epoch": 0.035343676153292924,
      "grad_norm": 1.7872354984283447,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.2137,
      "step": 86
    },
    {
      "epoch": 0.03575464913181958,
      "grad_norm": 2.0680534839630127,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.1599,
      "step": 87
    },
    {
      "epoch": 0.03616562211034625,
      "grad_norm": 1.9625688791275024,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.1587,
      "step": 88
    },
    {
      "epoch": 0.036576595088872906,
      "grad_norm": 1.9819324016571045,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.22,
      "step": 89
    },
    {
      "epoch": 0.03698756806739957,
      "grad_norm": 1.8013478517532349,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.1443,
      "step": 90
    },
    {
      "epoch": 0.03739854104592623,
      "grad_norm": 1.606767177581787,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.0728,
      "step": 91
    },
    {
      "epoch": 0.037809514024452893,
      "grad_norm": 2.007246255874634,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.2271,
      "step": 92
    },
    {
      "epoch": 0.03822048700297955,
      "grad_norm": 1.763731598854065,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.1322,
      "step": 93
    },
    {
      "epoch": 0.038631459981506217,
      "grad_norm": 1.7357094287872314,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.0478,
      "step": 94
    },
    {
      "epoch": 0.039042432960032875,
      "grad_norm": 1.4845335483551025,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.9269,
      "step": 95
    },
    {
      "epoch": 0.03945340593855954,
      "grad_norm": 1.636317253112793,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.1051,
      "step": 96
    },
    {
      "epoch": 0.039864378917086205,
      "grad_norm": 1.710037112236023,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.0736,
      "step": 97
    },
    {
      "epoch": 0.04027535189561286,
      "grad_norm": 1.9065608978271484,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.0447,
      "step": 98
    },
    {
      "epoch": 0.04068632487413953,
      "grad_norm": 1.6300371885299683,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.9586,
      "step": 99
    },
    {
      "epoch": 0.041097297852666186,
      "grad_norm": 1.8696982860565186,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.9802,
      "step": 100
    },
    {
      "epoch": 0.041097297852666186,
      "eval_loss": 0.35236656665802,
      "eval_runtime": 302.2005,
      "eval_samples_per_second": 13.564,
      "eval_steps_per_second": 3.392,
      "step": 100
    },
    {
      "epoch": 0.04150827083119285,
      "grad_norm": 6.189826965332031,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.7071,
      "step": 101
    },
    {
      "epoch": 0.04191924380971951,
      "grad_norm": 4.489702224731445,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.5446,
      "step": 102
    },
    {
      "epoch": 0.042330216788246174,
      "grad_norm": 2.949436902999878,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.4995,
      "step": 103
    },
    {
      "epoch": 0.04274118976677283,
      "grad_norm": 2.4982078075408936,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.3625,
      "step": 104
    },
    {
      "epoch": 0.0431521627452995,
      "grad_norm": 1.9499236345291138,
      "learning_rate": 5e-05,
      "loss": 1.2754,
      "step": 105
    },
    {
      "epoch": 0.04356313572382616,
      "grad_norm": 2.0557985305786133,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.2812,
      "step": 106
    },
    {
      "epoch": 0.04397410870235282,
      "grad_norm": 2.3099803924560547,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.4246,
      "step": 107
    },
    {
      "epoch": 0.044385081680879485,
      "grad_norm": 2.25260591506958,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.2934,
      "step": 108
    },
    {
      "epoch": 0.04479605465940614,
      "grad_norm": 2.0996761322021484,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.3142,
      "step": 109
    },
    {
      "epoch": 0.04520702763793281,
      "grad_norm": 2.28292179107666,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.3963,
      "step": 110
    },
    {
      "epoch": 0.045618000616459466,
      "grad_norm": 1.9222716093063354,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.2815,
      "step": 111
    },
    {
      "epoch": 0.04602897359498613,
      "grad_norm": 2.6586039066314697,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.2013,
      "step": 112
    },
    {
      "epoch": 0.04643994657351279,
      "grad_norm": 1.9669302701950073,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.2238,
      "step": 113
    },
    {
      "epoch": 0.046850919552039454,
      "grad_norm": 2.147646903991699,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.175,
      "step": 114
    },
    {
      "epoch": 0.04726189253056612,
      "grad_norm": 1.997481107711792,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.306,
      "step": 115
    },
    {
      "epoch": 0.04767286550909278,
      "grad_norm": 1.8799959421157837,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.2445,
      "step": 116
    },
    {
      "epoch": 0.04808383848761944,
      "grad_norm": 1.9030442237854004,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.2654,
      "step": 117
    },
    {
      "epoch": 0.0484948114661461,
      "grad_norm": 1.9143301248550415,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.3011,
      "step": 118
    },
    {
      "epoch": 0.048905784444672765,
      "grad_norm": 2.298109531402588,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.3096,
      "step": 119
    },
    {
      "epoch": 0.04931675742319942,
      "grad_norm": 1.7382055521011353,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.124,
      "step": 120
    },
    {
      "epoch": 0.04972773040172609,
      "grad_norm": 1.9285649061203003,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.3284,
      "step": 121
    },
    {
      "epoch": 0.050138703380252746,
      "grad_norm": 2.0173795223236084,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.3295,
      "step": 122
    },
    {
      "epoch": 0.05054967635877941,
      "grad_norm": 1.6439995765686035,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.2016,
      "step": 123
    },
    {
      "epoch": 0.05096064933730607,
      "grad_norm": 1.826642632484436,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.1813,
      "step": 124
    },
    {
      "epoch": 0.051371622315832734,
      "grad_norm": 1.9849226474761963,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.3747,
      "step": 125
    },
    {
      "epoch": 0.0517825952943594,
      "grad_norm": 1.99949049949646,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.2911,
      "step": 126
    },
    {
      "epoch": 0.05219356827288606,
      "grad_norm": 2.221797227859497,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.2424,
      "step": 127
    },
    {
      "epoch": 0.05260454125141272,
      "grad_norm": 2.1024458408355713,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.2834,
      "step": 128
    },
    {
      "epoch": 0.05301551422993938,
      "grad_norm": 1.8467835187911987,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.2616,
      "step": 129
    },
    {
      "epoch": 0.053426487208466045,
      "grad_norm": 1.8827816247940063,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.2407,
      "step": 130
    },
    {
      "epoch": 0.0538374601869927,
      "grad_norm": 1.7796611785888672,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.0954,
      "step": 131
    },
    {
      "epoch": 0.05424843316551937,
      "grad_norm": 1.6813939809799194,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.1754,
      "step": 132
    },
    {
      "epoch": 0.054659406144046026,
      "grad_norm": 1.9201902151107788,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.2356,
      "step": 133
    },
    {
      "epoch": 0.05507037912257269,
      "grad_norm": 1.535643219947815,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.0475,
      "step": 134
    },
    {
      "epoch": 0.055481352101099356,
      "grad_norm": 1.7866325378417969,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.2014,
      "step": 135
    },
    {
      "epoch": 0.055892325079626014,
      "grad_norm": 1.5998573303222656,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.1075,
      "step": 136
    },
    {
      "epoch": 0.05630329805815268,
      "grad_norm": 1.8415613174438477,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.1282,
      "step": 137
    },
    {
      "epoch": 0.05671427103667934,
      "grad_norm": 1.7491577863693237,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.129,
      "step": 138
    },
    {
      "epoch": 0.057125244015206,
      "grad_norm": 1.6540237665176392,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.0292,
      "step": 139
    },
    {
      "epoch": 0.05753621699373266,
      "grad_norm": 1.4938299655914307,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.0299,
      "step": 140
    },
    {
      "epoch": 0.057947189972259325,
      "grad_norm": 1.718437671661377,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.0044,
      "step": 141
    },
    {
      "epoch": 0.05835816295078598,
      "grad_norm": 1.7997852563858032,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.1825,
      "step": 142
    },
    {
      "epoch": 0.05876913592931265,
      "grad_norm": 1.6285719871520996,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.0148,
      "step": 143
    },
    {
      "epoch": 0.05918010890783931,
      "grad_norm": 1.731074571609497,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.0091,
      "step": 144
    },
    {
      "epoch": 0.05959108188636597,
      "grad_norm": 1.5902398824691772,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.9629,
      "step": 145
    },
    {
      "epoch": 0.060002054864892636,
      "grad_norm": 1.4852415323257446,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.9152,
      "step": 146
    },
    {
      "epoch": 0.060413027843419294,
      "grad_norm": 1.5001593828201294,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.9508,
      "step": 147
    },
    {
      "epoch": 0.06082400082194596,
      "grad_norm": 1.4033710956573486,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.8395,
      "step": 148
    },
    {
      "epoch": 0.06123497380047262,
      "grad_norm": 1.8549343347549438,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.952,
      "step": 149
    },
    {
      "epoch": 0.06164594677899928,
      "grad_norm": 1.366256594657898,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.8303,
      "step": 150
    },
    {
      "epoch": 0.06164594677899928,
      "eval_loss": 0.3162507712841034,
      "eval_runtime": 302.1878,
      "eval_samples_per_second": 13.564,
      "eval_steps_per_second": 3.392,
      "step": 150
    },
    {
      "epoch": 0.06205691975752594,
      "grad_norm": 4.3530497550964355,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.554,
      "step": 151
    },
    {
      "epoch": 0.062467892736052605,
      "grad_norm": 3.813164234161377,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.413,
      "step": 152
    },
    {
      "epoch": 0.06287886571457926,
      "grad_norm": 3.1085216999053955,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.401,
      "step": 153
    },
    {
      "epoch": 0.06328983869310593,
      "grad_norm": 2.7696313858032227,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.3471,
      "step": 154
    },
    {
      "epoch": 0.06370081167163259,
      "grad_norm": 3.7587575912475586,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.4271,
      "step": 155
    },
    {
      "epoch": 0.06411178465015925,
      "grad_norm": 2.270883560180664,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.2572,
      "step": 156
    },
    {
      "epoch": 0.06452275762868591,
      "grad_norm": 2.6226305961608887,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.4167,
      "step": 157
    },
    {
      "epoch": 0.06493373060721258,
      "grad_norm": 2.6284401416778564,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.4415,
      "step": 158
    },
    {
      "epoch": 0.06534470358573924,
      "grad_norm": 2.1877174377441406,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.2546,
      "step": 159
    },
    {
      "epoch": 0.0657556765642659,
      "grad_norm": 1.9456294775009155,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.2911,
      "step": 160
    },
    {
      "epoch": 0.06616664954279255,
      "grad_norm": 2.1140735149383545,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.2687,
      "step": 161
    },
    {
      "epoch": 0.06657762252131923,
      "grad_norm": 1.6488206386566162,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.1436,
      "step": 162
    },
    {
      "epoch": 0.06698859549984588,
      "grad_norm": 1.4730253219604492,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.0965,
      "step": 163
    },
    {
      "epoch": 0.06739956847837254,
      "grad_norm": 1.7242522239685059,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.0815,
      "step": 164
    },
    {
      "epoch": 0.06781054145689921,
      "grad_norm": 1.9750295877456665,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.2384,
      "step": 165
    },
    {
      "epoch": 0.06822151443542587,
      "grad_norm": 1.8953503370285034,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.2582,
      "step": 166
    },
    {
      "epoch": 0.06863248741395253,
      "grad_norm": 1.625638484954834,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.0583,
      "step": 167
    },
    {
      "epoch": 0.06904346039247919,
      "grad_norm": 1.67536461353302,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.169,
      "step": 168
    },
    {
      "epoch": 0.06945443337100586,
      "grad_norm": 1.9972889423370361,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.2651,
      "step": 169
    },
    {
      "epoch": 0.06986540634953252,
      "grad_norm": 1.823053240776062,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.224,
      "step": 170
    },
    {
      "epoch": 0.07027637932805918,
      "grad_norm": 1.693700909614563,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.1543,
      "step": 171
    },
    {
      "epoch": 0.07068735230658585,
      "grad_norm": 1.7095065116882324,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.1754,
      "step": 172
    },
    {
      "epoch": 0.0710983252851125,
      "grad_norm": 1.7292845249176025,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.1931,
      "step": 173
    },
    {
      "epoch": 0.07150929826363916,
      "grad_norm": 1.8461198806762695,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.1184,
      "step": 174
    },
    {
      "epoch": 0.07192027124216582,
      "grad_norm": 1.8154035806655884,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.1492,
      "step": 175
    },
    {
      "epoch": 0.0723312442206925,
      "grad_norm": 1.8658778667449951,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.2567,
      "step": 176
    },
    {
      "epoch": 0.07274221719921915,
      "grad_norm": 1.8199588060379028,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.1761,
      "step": 177
    },
    {
      "epoch": 0.07315319017774581,
      "grad_norm": 1.7595168352127075,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.1521,
      "step": 178
    },
    {
      "epoch": 0.07356416315627247,
      "grad_norm": 1.9610886573791504,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.203,
      "step": 179
    },
    {
      "epoch": 0.07397513613479914,
      "grad_norm": 1.7113889455795288,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.1194,
      "step": 180
    },
    {
      "epoch": 0.0743861091133258,
      "grad_norm": 1.8753197193145752,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.2144,
      "step": 181
    },
    {
      "epoch": 0.07479708209185246,
      "grad_norm": 2.0304226875305176,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.2055,
      "step": 182
    },
    {
      "epoch": 0.07520805507037913,
      "grad_norm": 1.924513816833496,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.2379,
      "step": 183
    },
    {
      "epoch": 0.07561902804890579,
      "grad_norm": 1.7012286186218262,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.1809,
      "step": 184
    },
    {
      "epoch": 0.07603000102743245,
      "grad_norm": 1.708716630935669,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.1141,
      "step": 185
    },
    {
      "epoch": 0.0764409740059591,
      "grad_norm": 1.7929389476776123,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.1881,
      "step": 186
    },
    {
      "epoch": 0.07685194698448577,
      "grad_norm": 1.637284755706787,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.0349,
      "step": 187
    },
    {
      "epoch": 0.07726291996301243,
      "grad_norm": 1.953851580619812,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.1291,
      "step": 188
    },
    {
      "epoch": 0.07767389294153909,
      "grad_norm": 1.9498717784881592,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.147,
      "step": 189
    },
    {
      "epoch": 0.07808486592006575,
      "grad_norm": 1.534037470817566,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.967,
      "step": 190
    },
    {
      "epoch": 0.07849583889859242,
      "grad_norm": 1.8027256727218628,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.0618,
      "step": 191
    },
    {
      "epoch": 0.07890681187711908,
      "grad_norm": 1.7773703336715698,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.093,
      "step": 192
    },
    {
      "epoch": 0.07931778485564574,
      "grad_norm": 1.7668685913085938,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.0807,
      "step": 193
    },
    {
      "epoch": 0.07972875783417241,
      "grad_norm": 1.692906141281128,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.0268,
      "step": 194
    },
    {
      "epoch": 0.08013973081269907,
      "grad_norm": 1.9769437313079834,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.0075,
      "step": 195
    },
    {
      "epoch": 0.08055070379122573,
      "grad_norm": 1.9087471961975098,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.0192,
      "step": 196
    },
    {
      "epoch": 0.08096167676975238,
      "grad_norm": 1.750394344329834,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.0551,
      "step": 197
    },
    {
      "epoch": 0.08137264974827906,
      "grad_norm": 1.772232174873352,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.0097,
      "step": 198
    },
    {
      "epoch": 0.08178362272680571,
      "grad_norm": 1.878016471862793,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.0983,
      "step": 199
    },
    {
      "epoch": 0.08219459570533237,
      "grad_norm": 2.1359243392944336,
      "learning_rate": 0.0,
      "loss": 1.0053,
      "step": 200
    },
    {
      "epoch": 0.08219459570533237,
      "eval_loss": 0.28087058663368225,
      "eval_runtime": 302.6335,
      "eval_samples_per_second": 13.544,
      "eval_steps_per_second": 3.387,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.861999239200768e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}