{
  "best_metric": 1.203765869140625,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 1.910828025477707,
  "eval_steps": 50,
  "global_step": 150,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.012738853503184714,
      "grad_norm": 0.730525016784668,
      "learning_rate": 1e-05,
      "loss": 1.4151,
      "step": 1
    },
    {
      "epoch": 0.012738853503184714,
      "eval_loss": 1.9231208562850952,
      "eval_runtime": 3.1054,
      "eval_samples_per_second": 42.506,
      "eval_steps_per_second": 10.627,
      "step": 1
    },
    {
      "epoch": 0.025477707006369428,
      "grad_norm": 0.9783929586410522,
      "learning_rate": 2e-05,
      "loss": 1.5352,
      "step": 2
    },
    {
      "epoch": 0.03821656050955414,
      "grad_norm": 1.061204195022583,
      "learning_rate": 3e-05,
      "loss": 1.7766,
      "step": 3
    },
    {
      "epoch": 0.050955414012738856,
      "grad_norm": 1.0660096406936646,
      "learning_rate": 4e-05,
      "loss": 1.7072,
      "step": 4
    },
    {
      "epoch": 0.06369426751592357,
      "grad_norm": 1.0237048864364624,
      "learning_rate": 5e-05,
      "loss": 1.7432,
      "step": 5
    },
    {
      "epoch": 0.07643312101910828,
      "grad_norm": 0.9316129088401794,
      "learning_rate": 6e-05,
      "loss": 1.7123,
      "step": 6
    },
    {
      "epoch": 0.08917197452229299,
      "grad_norm": 0.9872995615005493,
      "learning_rate": 7e-05,
      "loss": 1.7131,
      "step": 7
    },
    {
      "epoch": 0.10191082802547771,
      "grad_norm": 1.1821303367614746,
      "learning_rate": 8e-05,
      "loss": 1.7885,
      "step": 8
    },
    {
      "epoch": 0.11464968152866242,
      "grad_norm": 1.2145814895629883,
      "learning_rate": 9e-05,
      "loss": 1.7092,
      "step": 9
    },
    {
      "epoch": 0.12738853503184713,
      "grad_norm": 1.2002278566360474,
      "learning_rate": 0.0001,
      "loss": 1.7187,
      "step": 10
    },
    {
      "epoch": 0.14012738853503184,
      "grad_norm": 1.097898006439209,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.7364,
      "step": 11
    },
    {
      "epoch": 0.15286624203821655,
      "grad_norm": 1.2660833597183228,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.6647,
      "step": 12
    },
    {
      "epoch": 0.16560509554140126,
      "grad_norm": 1.2115747928619385,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.7278,
      "step": 13
    },
    {
      "epoch": 0.17834394904458598,
      "grad_norm": 1.304758071899414,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.7104,
      "step": 14
    },
    {
      "epoch": 0.1910828025477707,
      "grad_norm": 1.340864896774292,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.6743,
      "step": 15
    },
    {
      "epoch": 0.20382165605095542,
      "grad_norm": 1.41798734664917,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.6521,
      "step": 16
    },
    {
      "epoch": 0.21656050955414013,
      "grad_norm": 1.5173635482788086,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.5869,
      "step": 17
    },
    {
      "epoch": 0.22929936305732485,
      "grad_norm": 1.5008466243743896,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.6383,
      "step": 18
    },
    {
      "epoch": 0.24203821656050956,
      "grad_norm": 1.7662276029586792,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.5171,
      "step": 19
    },
    {
      "epoch": 0.25477707006369427,
      "grad_norm": 0.7256463766098022,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.2116,
      "step": 20
    },
    {
      "epoch": 0.267515923566879,
      "grad_norm": 0.7664700746536255,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.4651,
      "step": 21
    },
    {
      "epoch": 0.2802547770700637,
      "grad_norm": 0.667127788066864,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.4338,
      "step": 22
    },
    {
      "epoch": 0.2929936305732484,
      "grad_norm": 0.6543688774108887,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.5142,
      "step": 23
    },
    {
      "epoch": 0.3057324840764331,
      "grad_norm": 0.673316240310669,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.4425,
      "step": 24
    },
    {
      "epoch": 0.3184713375796178,
      "grad_norm": 0.7394607067108154,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.5683,
      "step": 25
    },
    {
      "epoch": 0.33121019108280253,
      "grad_norm": 0.7650450468063354,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.5173,
      "step": 26
    },
    {
      "epoch": 0.34394904458598724,
      "grad_norm": 0.8438877463340759,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.5401,
      "step": 27
    },
    {
      "epoch": 0.35668789808917195,
      "grad_norm": 0.8294534683227539,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.5294,
      "step": 28
    },
    {
      "epoch": 0.36942675159235666,
      "grad_norm": 0.8650810718536377,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.3832,
      "step": 29
    },
    {
      "epoch": 0.3821656050955414,
      "grad_norm": 0.9150643348693848,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.5288,
      "step": 30
    },
    {
      "epoch": 0.39490445859872614,
      "grad_norm": 0.9153487682342529,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.4732,
      "step": 31
    },
    {
      "epoch": 0.40764331210191085,
      "grad_norm": 1.045650839805603,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.4168,
      "step": 32
    },
    {
      "epoch": 0.42038216560509556,
      "grad_norm": 1.134645700454712,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.5262,
      "step": 33
    },
    {
      "epoch": 0.43312101910828027,
      "grad_norm": 1.2910881042480469,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.5426,
      "step": 34
    },
    {
      "epoch": 0.445859872611465,
      "grad_norm": 1.3072689771652222,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.4739,
      "step": 35
    },
    {
      "epoch": 0.4585987261146497,
      "grad_norm": 1.3429385423660278,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.4015,
      "step": 36
    },
    {
      "epoch": 0.4713375796178344,
      "grad_norm": 1.5992823839187622,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.4074,
      "step": 37
    },
    {
      "epoch": 0.4840764331210191,
      "grad_norm": 2.0143518447875977,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.5051,
      "step": 38
    },
    {
      "epoch": 0.4968152866242038,
      "grad_norm": 0.510002851486206,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.1671,
      "step": 39
    },
    {
      "epoch": 0.5095541401273885,
      "grad_norm": 0.6488459706306458,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.4333,
      "step": 40
    },
    {
      "epoch": 0.5222929936305732,
      "grad_norm": 0.6607312560081482,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.4202,
      "step": 41
    },
    {
      "epoch": 0.535031847133758,
      "grad_norm": 0.6778765916824341,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.4926,
      "step": 42
    },
    {
      "epoch": 0.5477707006369427,
      "grad_norm": 0.6211956143379211,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.3819,
      "step": 43
    },
    {
      "epoch": 0.5605095541401274,
      "grad_norm": 0.6790802478790283,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.3139,
      "step": 44
    },
    {
      "epoch": 0.5732484076433121,
      "grad_norm": 0.7075676321983337,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.4416,
      "step": 45
    },
    {
      "epoch": 0.5859872611464968,
      "grad_norm": 0.7339006662368774,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.3549,
      "step": 46
    },
    {
      "epoch": 0.5987261146496815,
      "grad_norm": 0.7455741167068481,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.3055,
      "step": 47
    },
    {
      "epoch": 0.6114649681528662,
      "grad_norm": 0.8291700482368469,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.3461,
      "step": 48
    },
    {
      "epoch": 0.6242038216560509,
      "grad_norm": 0.9561677575111389,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.3612,
      "step": 49
    },
    {
      "epoch": 0.6369426751592356,
      "grad_norm": 0.9051001667976379,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.4297,
      "step": 50
    },
    {
      "epoch": 0.6369426751592356,
      "eval_loss": 1.3635605573654175,
      "eval_runtime": 3.143,
      "eval_samples_per_second": 41.998,
      "eval_steps_per_second": 10.499,
      "step": 50
    },
    {
      "epoch": 0.6496815286624203,
      "grad_norm": 1.0236021280288696,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.4105,
      "step": 51
    },
    {
      "epoch": 0.6624203821656051,
      "grad_norm": 1.0871549844741821,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.4374,
      "step": 52
    },
    {
      "epoch": 0.6751592356687898,
      "grad_norm": 1.3056539297103882,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.5019,
      "step": 53
    },
    {
      "epoch": 0.6878980891719745,
      "grad_norm": 1.073501706123352,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.2992,
      "step": 54
    },
    {
      "epoch": 0.7006369426751592,
      "grad_norm": 1.3480092287063599,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.5648,
      "step": 55
    },
    {
      "epoch": 0.7133757961783439,
      "grad_norm": 1.4414317607879639,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.3422,
      "step": 56
    },
    {
      "epoch": 0.7261146496815286,
      "grad_norm": 1.5923312902450562,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.405,
      "step": 57
    },
    {
      "epoch": 0.7388535031847133,
      "grad_norm": 0.44930320978164673,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.2227,
      "step": 58
    },
    {
      "epoch": 0.7515923566878981,
      "grad_norm": 0.5464533567428589,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.3616,
      "step": 59
    },
    {
      "epoch": 0.7643312101910829,
      "grad_norm": 0.5873711705207825,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.3376,
      "step": 60
    },
    {
      "epoch": 0.7770700636942676,
      "grad_norm": 0.6632224917411804,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.442,
      "step": 61
    },
    {
      "epoch": 0.7898089171974523,
      "grad_norm": 0.6500707864761353,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.3928,
      "step": 62
    },
    {
      "epoch": 0.802547770700637,
      "grad_norm": 0.6918773055076599,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.315,
      "step": 63
    },
    {
      "epoch": 0.8152866242038217,
      "grad_norm": 0.7319614291191101,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.4207,
      "step": 64
    },
    {
      "epoch": 0.8280254777070064,
      "grad_norm": 0.7440667748451233,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.3226,
      "step": 65
    },
    {
      "epoch": 0.8407643312101911,
      "grad_norm": 0.7762590050697327,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.3935,
      "step": 66
    },
    {
      "epoch": 0.8535031847133758,
      "grad_norm": 0.8076745867729187,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.3657,
      "step": 67
    },
    {
      "epoch": 0.8662420382165605,
      "grad_norm": 0.8834313154220581,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.3637,
      "step": 68
    },
    {
      "epoch": 0.8789808917197452,
      "grad_norm": 0.956062376499176,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.4393,
      "step": 69
    },
    {
      "epoch": 0.89171974522293,
      "grad_norm": 1.0407336950302124,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.3919,
      "step": 70
    },
    {
      "epoch": 0.9044585987261147,
      "grad_norm": 1.08828866481781,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.3064,
      "step": 71
    },
    {
      "epoch": 0.9171974522292994,
      "grad_norm": 1.1810286045074463,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.345,
      "step": 72
    },
    {
      "epoch": 0.9299363057324841,
      "grad_norm": 1.1818805932998657,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.4282,
      "step": 73
    },
    {
      "epoch": 0.9426751592356688,
      "grad_norm": 1.2071267366409302,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.2402,
      "step": 74
    },
    {
      "epoch": 0.9554140127388535,
      "grad_norm": 1.3365556001663208,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.2618,
      "step": 75
    },
    {
      "epoch": 0.9681528662420382,
      "grad_norm": 1.5835633277893066,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.1645,
      "step": 76
    },
    {
      "epoch": 0.9808917197452229,
      "grad_norm": 0.6107880473136902,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.3169,
      "step": 77
    },
    {
      "epoch": 0.9936305732484076,
      "grad_norm": 0.9823195934295654,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.3774,
      "step": 78
    },
    {
      "epoch": 1.0063694267515924,
      "grad_norm": 2.4235544204711914,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.0893,
      "step": 79
    },
    {
      "epoch": 1.019108280254777,
      "grad_norm": 0.4601570665836334,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.1624,
      "step": 80
    },
    {
      "epoch": 1.0318471337579618,
      "grad_norm": 0.5115529298782349,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.3283,
      "step": 81
    },
    {
      "epoch": 1.0445859872611465,
      "grad_norm": 0.5322657227516174,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.2238,
      "step": 82
    },
    {
      "epoch": 1.0573248407643312,
      "grad_norm": 0.5813700556755066,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.3318,
      "step": 83
    },
    {
      "epoch": 1.070063694267516,
      "grad_norm": 0.5919923782348633,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.1778,
      "step": 84
    },
    {
      "epoch": 1.0828025477707006,
      "grad_norm": 0.6315931081771851,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.2893,
      "step": 85
    },
    {
      "epoch": 1.0955414012738853,
      "grad_norm": 0.6535232663154602,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.2274,
      "step": 86
    },
    {
      "epoch": 1.10828025477707,
      "grad_norm": 0.7134358286857605,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.2859,
      "step": 87
    },
    {
      "epoch": 1.1210191082802548,
      "grad_norm": 0.6813247799873352,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.1289,
      "step": 88
    },
    {
      "epoch": 1.1337579617834395,
      "grad_norm": 0.7537347674369812,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.1257,
      "step": 89
    },
    {
      "epoch": 1.1464968152866242,
      "grad_norm": 0.7934715151786804,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.0256,
      "step": 90
    },
    {
      "epoch": 1.1592356687898089,
      "grad_norm": 0.9412715435028076,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.1815,
      "step": 91
    },
    {
      "epoch": 1.1719745222929936,
      "grad_norm": 0.9789512753486633,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.101,
      "step": 92
    },
    {
      "epoch": 1.1847133757961783,
      "grad_norm": 1.0922623872756958,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.9928,
      "step": 93
    },
    {
      "epoch": 1.197452229299363,
      "grad_norm": 1.2500115633010864,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.1307,
      "step": 94
    },
    {
      "epoch": 1.2101910828025477,
      "grad_norm": 1.223982572555542,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.0143,
      "step": 95
    },
    {
      "epoch": 1.2229299363057324,
      "grad_norm": 1.468043565750122,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.2461,
      "step": 96
    },
    {
      "epoch": 1.2356687898089171,
      "grad_norm": 1.6290546655654907,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.051,
      "step": 97
    },
    {
      "epoch": 1.2484076433121019,
      "grad_norm": 1.1483049392700195,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.0254,
      "step": 98
    },
    {
      "epoch": 1.2611464968152866,
      "grad_norm": 0.7350722551345825,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.2225,
      "step": 99
    },
    {
      "epoch": 1.2738853503184713,
      "grad_norm": 0.7307099103927612,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.2505,
      "step": 100
    },
    {
      "epoch": 1.2738853503184713,
      "eval_loss": 1.2614315748214722,
      "eval_runtime": 3.1356,
      "eval_samples_per_second": 42.097,
      "eval_steps_per_second": 10.524,
      "step": 100
    },
    {
      "epoch": 1.286624203821656,
      "grad_norm": 0.7599360942840576,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.2487,
      "step": 101
    },
    {
      "epoch": 1.2993630573248407,
      "grad_norm": 0.7184345126152039,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.2761,
      "step": 102
    },
    {
      "epoch": 1.3121019108280254,
      "grad_norm": 0.7206115126609802,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.2801,
      "step": 103
    },
    {
      "epoch": 1.3248407643312101,
      "grad_norm": 0.7325707674026489,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.292,
      "step": 104
    },
    {
      "epoch": 1.3375796178343948,
      "grad_norm": 0.7475815415382385,
      "learning_rate": 5e-05,
      "loss": 1.2098,
      "step": 105
    },
    {
      "epoch": 1.3503184713375795,
      "grad_norm": 0.7716426849365234,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.251,
      "step": 106
    },
    {
      "epoch": 1.3630573248407643,
      "grad_norm": 0.81484055519104,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.163,
      "step": 107
    },
    {
      "epoch": 1.3757961783439492,
      "grad_norm": 0.836017370223999,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.2662,
      "step": 108
    },
    {
      "epoch": 1.388535031847134,
      "grad_norm": 0.9110499024391174,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.2193,
      "step": 109
    },
    {
      "epoch": 1.4012738853503186,
      "grad_norm": 0.9469859600067139,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.1439,
      "step": 110
    },
    {
      "epoch": 1.4140127388535033,
      "grad_norm": 1.011849284172058,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.1469,
      "step": 111
    },
    {
      "epoch": 1.426751592356688,
      "grad_norm": 1.071759581565857,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.1344,
      "step": 112
    },
    {
      "epoch": 1.4394904458598727,
      "grad_norm": 1.0705921649932861,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.0314,
      "step": 113
    },
    {
      "epoch": 1.4522292993630574,
      "grad_norm": 1.132436752319336,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.9669,
      "step": 114
    },
    {
      "epoch": 1.4649681528662422,
      "grad_norm": 1.3897563219070435,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.0581,
      "step": 115
    },
    {
      "epoch": 1.4777070063694269,
      "grad_norm": 1.6292494535446167,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.0959,
      "step": 116
    },
    {
      "epoch": 1.4904458598726116,
      "grad_norm": 1.1249570846557617,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.921,
      "step": 117
    },
    {
      "epoch": 1.5031847133757963,
      "grad_norm": 0.6542145609855652,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.2543,
      "step": 118
    },
    {
      "epoch": 1.515923566878981,
      "grad_norm": 0.7548701167106628,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.2312,
      "step": 119
    },
    {
      "epoch": 1.5286624203821657,
      "grad_norm": 0.7549881935119629,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.1718,
      "step": 120
    },
    {
      "epoch": 1.5414012738853504,
      "grad_norm": 0.7595965266227722,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.241,
      "step": 121
    },
    {
      "epoch": 1.5541401273885351,
      "grad_norm": 0.7441968321800232,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.0611,
      "step": 122
    },
    {
      "epoch": 1.5668789808917198,
      "grad_norm": 0.7548097968101501,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.1194,
      "step": 123
    },
    {
      "epoch": 1.5796178343949046,
      "grad_norm": 0.8410933017730713,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.2075,
      "step": 124
    },
    {
      "epoch": 1.5923566878980893,
      "grad_norm": 0.8075399398803711,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.1333,
      "step": 125
    },
    {
      "epoch": 1.605095541401274,
      "grad_norm": 0.8704642653465271,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.158,
      "step": 126
    },
    {
      "epoch": 1.6178343949044587,
      "grad_norm": 0.9110318422317505,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.211,
      "step": 127
    },
    {
      "epoch": 1.6305732484076434,
      "grad_norm": 0.8437381982803345,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.9478,
      "step": 128
    },
    {
      "epoch": 1.643312101910828,
      "grad_norm": 0.95970618724823,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.1144,
      "step": 129
    },
    {
      "epoch": 1.6560509554140128,
      "grad_norm": 0.9990697503089905,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.136,
      "step": 130
    },
    {
      "epoch": 1.6687898089171975,
      "grad_norm": 1.11715567111969,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.2196,
      "step": 131
    },
    {
      "epoch": 1.6815286624203822,
      "grad_norm": 1.0742250680923462,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.0805,
      "step": 132
    },
    {
      "epoch": 1.694267515923567,
      "grad_norm": 1.1880393028259277,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.1496,
      "step": 133
    },
    {
      "epoch": 1.7070063694267517,
      "grad_norm": 1.320546269416809,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.0415,
      "step": 134
    },
    {
      "epoch": 1.7197452229299364,
      "grad_norm": 1.5339378118515015,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.9827,
      "step": 135
    },
    {
      "epoch": 1.732484076433121,
      "grad_norm": 1.0027467012405396,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.8373,
      "step": 136
    },
    {
      "epoch": 1.7452229299363058,
      "grad_norm": 0.4806833267211914,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.231,
      "step": 137
    },
    {
      "epoch": 1.7579617834394905,
      "grad_norm": 0.5429365038871765,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.0698,
      "step": 138
    },
    {
      "epoch": 1.7707006369426752,
      "grad_norm": 0.6483641266822815,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.1061,
      "step": 139
    },
    {
      "epoch": 1.78343949044586,
      "grad_norm": 0.6727161407470703,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.2198,
      "step": 140
    },
    {
      "epoch": 1.7961783439490446,
      "grad_norm": 0.7277352213859558,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.2109,
      "step": 141
    },
    {
      "epoch": 1.8089171974522293,
      "grad_norm": 0.7775723934173584,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.2011,
      "step": 142
    },
    {
      "epoch": 1.821656050955414,
      "grad_norm": 0.7981808185577393,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.0889,
      "step": 143
    },
    {
      "epoch": 1.8343949044585988,
      "grad_norm": 0.9179716110229492,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.0686,
      "step": 144
    },
    {
      "epoch": 1.8471337579617835,
      "grad_norm": 0.9146676659584045,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.1423,
      "step": 145
    },
    {
      "epoch": 1.8598726114649682,
      "grad_norm": 0.9447951316833496,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.1823,
      "step": 146
    },
    {
      "epoch": 1.872611464968153,
      "grad_norm": 0.9554328918457031,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.1351,
      "step": 147
    },
    {
      "epoch": 1.8853503184713376,
      "grad_norm": 1.0493661165237427,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.1145,
      "step": 148
    },
    {
      "epoch": 1.8980891719745223,
      "grad_norm": 1.0875847339630127,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.1727,
      "step": 149
    },
    {
      "epoch": 1.910828025477707,
      "grad_norm": 1.1676983833312988,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.1159,
      "step": 150
    },
    {
      "epoch": 1.910828025477707,
      "eval_loss": 1.203765869140625,
      "eval_runtime": 3.072,
      "eval_samples_per_second": 42.969,
      "eval_steps_per_second": 10.742,
      "step": 150
    }
  ],
  "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": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.124263930298368e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}