{
  "best_metric": 1.4807544946670532,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 2.030456852791878,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01015228426395939,
      "grad_norm": 0.7291321158409119,
      "learning_rate": 1e-05,
      "loss": 1.694,
      "step": 1
    },
    {
      "epoch": 0.01015228426395939,
      "eval_loss": 1.9204717874526978,
      "eval_runtime": 7.6104,
      "eval_samples_per_second": 21.812,
      "eval_steps_per_second": 5.519,
      "step": 1
    },
    {
      "epoch": 0.02030456852791878,
      "grad_norm": 0.80634605884552,
      "learning_rate": 2e-05,
      "loss": 2.0128,
      "step": 2
    },
    {
      "epoch": 0.030456852791878174,
      "grad_norm": 0.818202018737793,
      "learning_rate": 3e-05,
      "loss": 1.8612,
      "step": 3
    },
    {
      "epoch": 0.04060913705583756,
      "grad_norm": 0.8192562460899353,
      "learning_rate": 4e-05,
      "loss": 2.0149,
      "step": 4
    },
    {
      "epoch": 0.050761421319796954,
      "grad_norm": 0.8154361248016357,
      "learning_rate": 5e-05,
      "loss": 1.9988,
      "step": 5
    },
    {
      "epoch": 0.06091370558375635,
      "grad_norm": 0.8312411308288574,
      "learning_rate": 6e-05,
      "loss": 1.8842,
      "step": 6
    },
    {
      "epoch": 0.07106598984771574,
      "grad_norm": 0.8702840209007263,
      "learning_rate": 7e-05,
      "loss": 1.7678,
      "step": 7
    },
    {
      "epoch": 0.08121827411167512,
      "grad_norm": 0.8380058407783508,
      "learning_rate": 8e-05,
      "loss": 1.8245,
      "step": 8
    },
    {
      "epoch": 0.09137055837563451,
      "grad_norm": 0.8897891044616699,
      "learning_rate": 9e-05,
      "loss": 1.8679,
      "step": 9
    },
    {
      "epoch": 0.10152284263959391,
      "grad_norm": 0.7739996910095215,
      "learning_rate": 0.0001,
      "loss": 1.7507,
      "step": 10
    },
    {
      "epoch": 0.1116751269035533,
      "grad_norm": 0.8586166501045227,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.9417,
      "step": 11
    },
    {
      "epoch": 0.1218274111675127,
      "grad_norm": 0.8529003262519836,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.9287,
      "step": 12
    },
    {
      "epoch": 0.1319796954314721,
      "grad_norm": 0.8180898427963257,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.8649,
      "step": 13
    },
    {
      "epoch": 0.14213197969543148,
      "grad_norm": 0.8368061184883118,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.7576,
      "step": 14
    },
    {
      "epoch": 0.15228426395939088,
      "grad_norm": 0.750203549861908,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.801,
      "step": 15
    },
    {
      "epoch": 0.16243654822335024,
      "grad_norm": 0.9318957924842834,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.0455,
      "step": 16
    },
    {
      "epoch": 0.17258883248730963,
      "grad_norm": 0.8131672143936157,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.7534,
      "step": 17
    },
    {
      "epoch": 0.18274111675126903,
      "grad_norm": 0.7414889335632324,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.8609,
      "step": 18
    },
    {
      "epoch": 0.19289340101522842,
      "grad_norm": 0.724410891532898,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.6353,
      "step": 19
    },
    {
      "epoch": 0.20304568527918782,
      "grad_norm": 0.7799801230430603,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.8123,
      "step": 20
    },
    {
      "epoch": 0.2131979695431472,
      "grad_norm": 0.8134832382202148,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.7644,
      "step": 21
    },
    {
      "epoch": 0.2233502538071066,
      "grad_norm": 0.8975375294685364,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.6408,
      "step": 22
    },
    {
      "epoch": 0.233502538071066,
      "grad_norm": 1.0586955547332764,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.8749,
      "step": 23
    },
    {
      "epoch": 0.2436548223350254,
      "grad_norm": 1.335873007774353,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.8925,
      "step": 24
    },
    {
      "epoch": 0.25380710659898476,
      "grad_norm": 0.7456432580947876,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.6233,
      "step": 25
    },
    {
      "epoch": 0.2639593908629442,
      "grad_norm": 0.6994826793670654,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.8435,
      "step": 26
    },
    {
      "epoch": 0.27411167512690354,
      "grad_norm": 0.5480471253395081,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.6999,
      "step": 27
    },
    {
      "epoch": 0.28426395939086296,
      "grad_norm": 0.5642795562744141,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.7397,
      "step": 28
    },
    {
      "epoch": 0.29441624365482233,
      "grad_norm": 0.5542154312133789,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.8022,
      "step": 29
    },
    {
      "epoch": 0.30456852791878175,
      "grad_norm": 0.5566151738166809,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.8316,
      "step": 30
    },
    {
      "epoch": 0.3147208121827411,
      "grad_norm": 0.5346680879592896,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.8187,
      "step": 31
    },
    {
      "epoch": 0.3248730964467005,
      "grad_norm": 0.511399507522583,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.5272,
      "step": 32
    },
    {
      "epoch": 0.3350253807106599,
      "grad_norm": 0.5856063961982727,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.7773,
      "step": 33
    },
    {
      "epoch": 0.34517766497461927,
      "grad_norm": 0.545522928237915,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.787,
      "step": 34
    },
    {
      "epoch": 0.3553299492385787,
      "grad_norm": 0.5445789694786072,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.8507,
      "step": 35
    },
    {
      "epoch": 0.36548223350253806,
      "grad_norm": 0.5248554348945618,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.6546,
      "step": 36
    },
    {
      "epoch": 0.3756345177664975,
      "grad_norm": 0.5966745018959045,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.7359,
      "step": 37
    },
    {
      "epoch": 0.38578680203045684,
      "grad_norm": 0.5989910364151001,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.7916,
      "step": 38
    },
    {
      "epoch": 0.39593908629441626,
      "grad_norm": 0.6065903902053833,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.6755,
      "step": 39
    },
    {
      "epoch": 0.40609137055837563,
      "grad_norm": 0.6157575249671936,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.5601,
      "step": 40
    },
    {
      "epoch": 0.41624365482233505,
      "grad_norm": 0.6427657604217529,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.6133,
      "step": 41
    },
    {
      "epoch": 0.4263959390862944,
      "grad_norm": 0.6492039561271667,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.7375,
      "step": 42
    },
    {
      "epoch": 0.4365482233502538,
      "grad_norm": 0.7545812726020813,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.7681,
      "step": 43
    },
    {
      "epoch": 0.4467005076142132,
      "grad_norm": 0.7499914765357971,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.733,
      "step": 44
    },
    {
      "epoch": 0.45685279187817257,
      "grad_norm": 0.772696852684021,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.7454,
      "step": 45
    },
    {
      "epoch": 0.467005076142132,
      "grad_norm": 0.7788810133934021,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.5849,
      "step": 46
    },
    {
      "epoch": 0.47715736040609136,
      "grad_norm": 0.8338989019393921,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.5808,
      "step": 47
    },
    {
      "epoch": 0.4873096446700508,
      "grad_norm": 1.129300594329834,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.6567,
      "step": 48
    },
    {
      "epoch": 0.49746192893401014,
      "grad_norm": 0.4577413499355316,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.6247,
      "step": 49
    },
    {
      "epoch": 0.5076142131979695,
      "grad_norm": 0.5409057140350342,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.7515,
      "step": 50
    },
    {
      "epoch": 0.5076142131979695,
      "eval_loss": 1.6396291255950928,
      "eval_runtime": 7.6659,
      "eval_samples_per_second": 21.654,
      "eval_steps_per_second": 5.479,
      "step": 50
    },
    {
      "epoch": 0.5177664974619289,
      "grad_norm": 0.5438342690467834,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.6741,
      "step": 51
    },
    {
      "epoch": 0.5279187817258884,
      "grad_norm": 0.480343222618103,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.6306,
      "step": 52
    },
    {
      "epoch": 0.5380710659898477,
      "grad_norm": 0.4987044036388397,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.6827,
      "step": 53
    },
    {
      "epoch": 0.5482233502538071,
      "grad_norm": 0.5006952881813049,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.7046,
      "step": 54
    },
    {
      "epoch": 0.5583756345177665,
      "grad_norm": 0.5399014949798584,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.6806,
      "step": 55
    },
    {
      "epoch": 0.5685279187817259,
      "grad_norm": 0.5678799152374268,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.7275,
      "step": 56
    },
    {
      "epoch": 0.5786802030456852,
      "grad_norm": 0.5736947059631348,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.8071,
      "step": 57
    },
    {
      "epoch": 0.5888324873096447,
      "grad_norm": 0.5423876047134399,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.6724,
      "step": 58
    },
    {
      "epoch": 0.5989847715736041,
      "grad_norm": 0.6047144532203674,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.7872,
      "step": 59
    },
    {
      "epoch": 0.6091370558375635,
      "grad_norm": 0.5819637179374695,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.7783,
      "step": 60
    },
    {
      "epoch": 0.6192893401015228,
      "grad_norm": 0.6512382626533508,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.6646,
      "step": 61
    },
    {
      "epoch": 0.6294416243654822,
      "grad_norm": 0.6138646602630615,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.7131,
      "step": 62
    },
    {
      "epoch": 0.6395939086294417,
      "grad_norm": 0.6422168016433716,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.6838,
      "step": 63
    },
    {
      "epoch": 0.649746192893401,
      "grad_norm": 0.6840230226516724,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.7402,
      "step": 64
    },
    {
      "epoch": 0.6598984771573604,
      "grad_norm": 0.6549532413482666,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.6616,
      "step": 65
    },
    {
      "epoch": 0.6700507614213198,
      "grad_norm": 0.742430567741394,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.6679,
      "step": 66
    },
    {
      "epoch": 0.6802030456852792,
      "grad_norm": 0.7295668721199036,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.6294,
      "step": 67
    },
    {
      "epoch": 0.6903553299492385,
      "grad_norm": 0.8265715837478638,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.8901,
      "step": 68
    },
    {
      "epoch": 0.700507614213198,
      "grad_norm": 0.7593281269073486,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.5688,
      "step": 69
    },
    {
      "epoch": 0.7106598984771574,
      "grad_norm": 0.8952975869178772,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.7755,
      "step": 70
    },
    {
      "epoch": 0.7208121827411168,
      "grad_norm": 1.0053350925445557,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.5932,
      "step": 71
    },
    {
      "epoch": 0.7309644670050761,
      "grad_norm": 0.9705185294151306,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.5989,
      "step": 72
    },
    {
      "epoch": 0.7411167512690355,
      "grad_norm": 0.4911041557788849,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.708,
      "step": 73
    },
    {
      "epoch": 0.751269035532995,
      "grad_norm": 0.5251993536949158,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.8025,
      "step": 74
    },
    {
      "epoch": 0.7614213197969543,
      "grad_norm": 0.5147970914840698,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.5758,
      "step": 75
    },
    {
      "epoch": 0.7715736040609137,
      "grad_norm": 0.5723939538002014,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.7717,
      "step": 76
    },
    {
      "epoch": 0.7817258883248731,
      "grad_norm": 0.5487803816795349,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.6682,
      "step": 77
    },
    {
      "epoch": 0.7918781725888325,
      "grad_norm": 0.5720694065093994,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.7009,
      "step": 78
    },
    {
      "epoch": 0.8020304568527918,
      "grad_norm": 0.5712814927101135,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.6973,
      "step": 79
    },
    {
      "epoch": 0.8121827411167513,
      "grad_norm": 0.5587448477745056,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.5849,
      "step": 80
    },
    {
      "epoch": 0.8223350253807107,
      "grad_norm": 0.6082424521446228,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.7704,
      "step": 81
    },
    {
      "epoch": 0.8324873096446701,
      "grad_norm": 0.6438080072402954,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.7454,
      "step": 82
    },
    {
      "epoch": 0.8426395939086294,
      "grad_norm": 0.6118745803833008,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.5767,
      "step": 83
    },
    {
      "epoch": 0.8527918781725888,
      "grad_norm": 0.6005804538726807,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.5711,
      "step": 84
    },
    {
      "epoch": 0.8629441624365483,
      "grad_norm": 0.6292843222618103,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.5878,
      "step": 85
    },
    {
      "epoch": 0.8730964467005076,
      "grad_norm": 0.6168345808982849,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.4827,
      "step": 86
    },
    {
      "epoch": 0.883248730964467,
      "grad_norm": 0.6699876189231873,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.6753,
      "step": 87
    },
    {
      "epoch": 0.8934010152284264,
      "grad_norm": 0.7486642003059387,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.5591,
      "step": 88
    },
    {
      "epoch": 0.9035532994923858,
      "grad_norm": 0.716096818447113,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.5256,
      "step": 89
    },
    {
      "epoch": 0.9137055837563451,
      "grad_norm": 0.6980071663856506,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.6128,
      "step": 90
    },
    {
      "epoch": 0.9238578680203046,
      "grad_norm": 0.776879072189331,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.5798,
      "step": 91
    },
    {
      "epoch": 0.934010152284264,
      "grad_norm": 0.8649686574935913,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.6465,
      "step": 92
    },
    {
      "epoch": 0.9441624365482234,
      "grad_norm": 0.8168870210647583,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.7388,
      "step": 93
    },
    {
      "epoch": 0.9543147208121827,
      "grad_norm": 0.836371123790741,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.5316,
      "step": 94
    },
    {
      "epoch": 0.9644670050761421,
      "grad_norm": 0.975714385509491,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.6221,
      "step": 95
    },
    {
      "epoch": 0.9746192893401016,
      "grad_norm": 1.0654964447021484,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.6136,
      "step": 96
    },
    {
      "epoch": 0.9847715736040609,
      "grad_norm": 0.6085041165351868,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.6456,
      "step": 97
    },
    {
      "epoch": 0.9949238578680203,
      "grad_norm": 0.7240885496139526,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.5469,
      "step": 98
    },
    {
      "epoch": 1.0050761421319796,
      "grad_norm": 1.5985159873962402,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.3111,
      "step": 99
    },
    {
      "epoch": 1.015228426395939,
      "grad_norm": 0.5017366409301758,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.5656,
      "step": 100
    },
    {
      "epoch": 1.015228426395939,
      "eval_loss": 1.5486623048782349,
      "eval_runtime": 7.6734,
      "eval_samples_per_second": 21.633,
      "eval_steps_per_second": 5.473,
      "step": 100
    },
    {
      "epoch": 1.0253807106598984,
      "grad_norm": 0.5553914308547974,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.5885,
      "step": 101
    },
    {
      "epoch": 1.0355329949238579,
      "grad_norm": 0.5437448620796204,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.52,
      "step": 102
    },
    {
      "epoch": 1.0456852791878173,
      "grad_norm": 0.5706773996353149,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.5359,
      "step": 103
    },
    {
      "epoch": 1.0558375634517767,
      "grad_norm": 0.5832688808441162,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.5736,
      "step": 104
    },
    {
      "epoch": 1.0659898477157361,
      "grad_norm": 0.584896445274353,
      "learning_rate": 5e-05,
      "loss": 1.3852,
      "step": 105
    },
    {
      "epoch": 1.0761421319796955,
      "grad_norm": 0.6052612066268921,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.4096,
      "step": 106
    },
    {
      "epoch": 1.0862944162436547,
      "grad_norm": 0.6644023060798645,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.5471,
      "step": 107
    },
    {
      "epoch": 1.0964467005076142,
      "grad_norm": 0.7001641392707825,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.5418,
      "step": 108
    },
    {
      "epoch": 1.1065989847715736,
      "grad_norm": 0.6971859931945801,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.3906,
      "step": 109
    },
    {
      "epoch": 1.116751269035533,
      "grad_norm": 0.7651525735855103,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.4862,
      "step": 110
    },
    {
      "epoch": 1.1269035532994924,
      "grad_norm": 0.8193091750144958,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.551,
      "step": 111
    },
    {
      "epoch": 1.1370558375634519,
      "grad_norm": 0.8337458372116089,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.3594,
      "step": 112
    },
    {
      "epoch": 1.1472081218274113,
      "grad_norm": 0.821225106716156,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.2333,
      "step": 113
    },
    {
      "epoch": 1.1573604060913705,
      "grad_norm": 0.8680731058120728,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.3171,
      "step": 114
    },
    {
      "epoch": 1.16751269035533,
      "grad_norm": 0.8808449506759644,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.3503,
      "step": 115
    },
    {
      "epoch": 1.1776649746192893,
      "grad_norm": 0.9254382848739624,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.4595,
      "step": 116
    },
    {
      "epoch": 1.1878172588832487,
      "grad_norm": 1.2700564861297607,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.2803,
      "step": 117
    },
    {
      "epoch": 1.1979695431472082,
      "grad_norm": 0.9289053082466125,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.2169,
      "step": 118
    },
    {
      "epoch": 1.2081218274111676,
      "grad_norm": 1.0314911603927612,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.3335,
      "step": 119
    },
    {
      "epoch": 1.218274111675127,
      "grad_norm": 1.0336929559707642,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.221,
      "step": 120
    },
    {
      "epoch": 1.2284263959390862,
      "grad_norm": 1.1959046125411987,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.1413,
      "step": 121
    },
    {
      "epoch": 1.2385786802030456,
      "grad_norm": 1.5569109916687012,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.1702,
      "step": 122
    },
    {
      "epoch": 1.248730964467005,
      "grad_norm": 1.3973491191864014,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.3704,
      "step": 123
    },
    {
      "epoch": 1.2588832487309645,
      "grad_norm": 1.7734311819076538,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.5551,
      "step": 124
    },
    {
      "epoch": 1.2690355329949239,
      "grad_norm": 1.2938789129257202,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.6164,
      "step": 125
    },
    {
      "epoch": 1.2791878172588833,
      "grad_norm": 0.9389830231666565,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.4727,
      "step": 126
    },
    {
      "epoch": 1.2893401015228427,
      "grad_norm": 0.8181756734848022,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.517,
      "step": 127
    },
    {
      "epoch": 1.299492385786802,
      "grad_norm": 0.7993819713592529,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.5517,
      "step": 128
    },
    {
      "epoch": 1.3096446700507614,
      "grad_norm": 0.7154996395111084,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.4461,
      "step": 129
    },
    {
      "epoch": 1.3197969543147208,
      "grad_norm": 0.7990055680274963,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.5857,
      "step": 130
    },
    {
      "epoch": 1.3299492385786802,
      "grad_norm": 0.7646247148513794,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.419,
      "step": 131
    },
    {
      "epoch": 1.3401015228426396,
      "grad_norm": 0.7962736487388611,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.4192,
      "step": 132
    },
    {
      "epoch": 1.350253807106599,
      "grad_norm": 0.8432332873344421,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.5356,
      "step": 133
    },
    {
      "epoch": 1.3604060913705585,
      "grad_norm": 0.7631216049194336,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.3423,
      "step": 134
    },
    {
      "epoch": 1.3705583756345177,
      "grad_norm": 0.8354220986366272,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.3963,
      "step": 135
    },
    {
      "epoch": 1.380710659898477,
      "grad_norm": 0.8573411107063293,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.461,
      "step": 136
    },
    {
      "epoch": 1.3908629441624365,
      "grad_norm": 0.8249084949493408,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.4064,
      "step": 137
    },
    {
      "epoch": 1.401015228426396,
      "grad_norm": 0.906618595123291,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.3833,
      "step": 138
    },
    {
      "epoch": 1.4111675126903553,
      "grad_norm": 0.9852612018585205,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.3802,
      "step": 139
    },
    {
      "epoch": 1.4213197969543148,
      "grad_norm": 0.9798620939254761,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.2791,
      "step": 140
    },
    {
      "epoch": 1.4314720812182742,
      "grad_norm": 0.9310597777366638,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.2705,
      "step": 141
    },
    {
      "epoch": 1.4416243654822334,
      "grad_norm": 0.9294157028198242,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.1918,
      "step": 142
    },
    {
      "epoch": 1.451776649746193,
      "grad_norm": 1.1194959878921509,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.3822,
      "step": 143
    },
    {
      "epoch": 1.4619289340101522,
      "grad_norm": 1.119828701019287,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.1602,
      "step": 144
    },
    {
      "epoch": 1.4720812182741116,
      "grad_norm": 1.1303843259811401,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.1408,
      "step": 145
    },
    {
      "epoch": 1.482233502538071,
      "grad_norm": 1.4049675464630127,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.2422,
      "step": 146
    },
    {
      "epoch": 1.4923857868020305,
      "grad_norm": 0.9562484622001648,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.3199,
      "step": 147
    },
    {
      "epoch": 1.50253807106599,
      "grad_norm": 0.7429844737052917,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.5144,
      "step": 148
    },
    {
      "epoch": 1.512690355329949,
      "grad_norm": 0.8964042663574219,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.6035,
      "step": 149
    },
    {
      "epoch": 1.5228426395939088,
      "grad_norm": 0.9342640042304993,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.6336,
      "step": 150
    },
    {
      "epoch": 1.5228426395939088,
      "eval_loss": 1.5060570240020752,
      "eval_runtime": 7.6664,
      "eval_samples_per_second": 21.653,
      "eval_steps_per_second": 5.478,
      "step": 150
    },
    {
      "epoch": 1.532994923857868,
      "grad_norm": 0.9249311089515686,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.5501,
      "step": 151
    },
    {
      "epoch": 1.5431472081218274,
      "grad_norm": 0.8187703490257263,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.4895,
      "step": 152
    },
    {
      "epoch": 1.5532994923857868,
      "grad_norm": 0.9012784957885742,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.5277,
      "step": 153
    },
    {
      "epoch": 1.5634517766497462,
      "grad_norm": 0.7938726544380188,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.2788,
      "step": 154
    },
    {
      "epoch": 1.5736040609137056,
      "grad_norm": 0.8638392090797424,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.3972,
      "step": 155
    },
    {
      "epoch": 1.5837563451776648,
      "grad_norm": 0.8909347057342529,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.5049,
      "step": 156
    },
    {
      "epoch": 1.5939086294416245,
      "grad_norm": 0.8673722743988037,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.4228,
      "step": 157
    },
    {
      "epoch": 1.6040609137055837,
      "grad_norm": 0.8636196851730347,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.3808,
      "step": 158
    },
    {
      "epoch": 1.614213197969543,
      "grad_norm": 0.9390663504600525,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.4431,
      "step": 159
    },
    {
      "epoch": 1.6243654822335025,
      "grad_norm": 0.9971596598625183,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.4337,
      "step": 160
    },
    {
      "epoch": 1.634517766497462,
      "grad_norm": 0.961604118347168,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.2047,
      "step": 161
    },
    {
      "epoch": 1.6446700507614214,
      "grad_norm": 1.011372447013855,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.3067,
      "step": 162
    },
    {
      "epoch": 1.6548223350253806,
      "grad_norm": 1.0928382873535156,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.4216,
      "step": 163
    },
    {
      "epoch": 1.6649746192893402,
      "grad_norm": 1.0573028326034546,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.4101,
      "step": 164
    },
    {
      "epoch": 1.6751269035532994,
      "grad_norm": 0.9755818247795105,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.198,
      "step": 165
    },
    {
      "epoch": 1.6852791878172588,
      "grad_norm": 1.1219879388809204,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.3912,
      "step": 166
    },
    {
      "epoch": 1.6954314720812182,
      "grad_norm": 1.1401968002319336,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.2936,
      "step": 167
    },
    {
      "epoch": 1.7055837563451777,
      "grad_norm": 1.2365508079528809,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.1554,
      "step": 168
    },
    {
      "epoch": 1.715736040609137,
      "grad_norm": 1.2871695756912231,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.1339,
      "step": 169
    },
    {
      "epoch": 1.7258883248730963,
      "grad_norm": 1.515007495880127,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.1042,
      "step": 170
    },
    {
      "epoch": 1.736040609137056,
      "grad_norm": 1.1687333583831787,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.2151,
      "step": 171
    },
    {
      "epoch": 1.7461928934010151,
      "grad_norm": 0.7266783118247986,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.5054,
      "step": 172
    },
    {
      "epoch": 1.7563451776649748,
      "grad_norm": 0.7849690318107605,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.5274,
      "step": 173
    },
    {
      "epoch": 1.766497461928934,
      "grad_norm": 0.7627509236335754,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.4538,
      "step": 174
    },
    {
      "epoch": 1.7766497461928934,
      "grad_norm": 0.823388397693634,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.482,
      "step": 175
    },
    {
      "epoch": 1.7868020304568528,
      "grad_norm": 0.8710838556289673,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.5931,
      "step": 176
    },
    {
      "epoch": 1.796954314720812,
      "grad_norm": 0.9284811615943909,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.5351,
      "step": 177
    },
    {
      "epoch": 1.8071065989847717,
      "grad_norm": 0.8276659846305847,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.301,
      "step": 178
    },
    {
      "epoch": 1.8172588832487309,
      "grad_norm": 0.9095329642295837,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.4175,
      "step": 179
    },
    {
      "epoch": 1.8274111675126905,
      "grad_norm": 0.9896294474601746,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.6145,
      "step": 180
    },
    {
      "epoch": 1.8375634517766497,
      "grad_norm": 1.0880430936813354,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.4819,
      "step": 181
    },
    {
      "epoch": 1.8477157360406091,
      "grad_norm": 0.9309003949165344,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.3353,
      "step": 182
    },
    {
      "epoch": 1.8578680203045685,
      "grad_norm": 0.9569535255432129,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.3366,
      "step": 183
    },
    {
      "epoch": 1.868020304568528,
      "grad_norm": 0.9854280352592468,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.4042,
      "step": 184
    },
    {
      "epoch": 1.8781725888324874,
      "grad_norm": 1.068068027496338,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.3141,
      "step": 185
    },
    {
      "epoch": 1.8883248730964466,
      "grad_norm": 1.064318060874939,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.3181,
      "step": 186
    },
    {
      "epoch": 1.8984771573604062,
      "grad_norm": 1.1086726188659668,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.2969,
      "step": 187
    },
    {
      "epoch": 1.9086294416243654,
      "grad_norm": 1.1145756244659424,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.1048,
      "step": 188
    },
    {
      "epoch": 1.9187817258883249,
      "grad_norm": 1.107241153717041,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.2283,
      "step": 189
    },
    {
      "epoch": 1.9289340101522843,
      "grad_norm": 1.1122037172317505,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.2167,
      "step": 190
    },
    {
      "epoch": 1.9390862944162437,
      "grad_norm": 1.319201111793518,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.2276,
      "step": 191
    },
    {
      "epoch": 1.9492385786802031,
      "grad_norm": 1.2451070547103882,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.1836,
      "step": 192
    },
    {
      "epoch": 1.9593908629441623,
      "grad_norm": 1.3413528203964233,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.0294,
      "step": 193
    },
    {
      "epoch": 1.969543147208122,
      "grad_norm": 1.5231115818023682,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.11,
      "step": 194
    },
    {
      "epoch": 1.9796954314720812,
      "grad_norm": 1.0309418439865112,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.2832,
      "step": 195
    },
    {
      "epoch": 1.9898477157360406,
      "grad_norm": 0.8608314394950867,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.2385,
      "step": 196
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.0627083778381348,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.5563,
      "step": 197
    },
    {
      "epoch": 2.010152284263959,
      "grad_norm": 0.5498896837234497,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.3595,
      "step": 198
    },
    {
      "epoch": 2.020304568527919,
      "grad_norm": 0.7041447758674622,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.6647,
      "step": 199
    },
    {
      "epoch": 2.030456852791878,
      "grad_norm": 0.7375603318214417,
      "learning_rate": 0.0,
      "loss": 1.3941,
      "step": 200
    },
    {
      "epoch": 2.030456852791878,
      "eval_loss": 1.4807544946670532,
      "eval_runtime": 7.6594,
      "eval_samples_per_second": 21.673,
      "eval_steps_per_second": 5.483,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "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": 8.427209730136474e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}