{
  "best_metric": 10.349349021911621,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.8421052631578947,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004210526315789474,
      "grad_norm": 0.05298575758934021,
      "learning_rate": 1e-05,
      "loss": 10.3741,
      "step": 1
    },
    {
      "epoch": 0.004210526315789474,
      "eval_loss": 10.376794815063477,
      "eval_runtime": 0.8641,
      "eval_samples_per_second": 462.883,
      "eval_steps_per_second": 115.721,
      "step": 1
    },
    {
      "epoch": 0.008421052631578947,
      "grad_norm": 0.04920467734336853,
      "learning_rate": 2e-05,
      "loss": 10.3742,
      "step": 2
    },
    {
      "epoch": 0.01263157894736842,
      "grad_norm": 0.056816600263118744,
      "learning_rate": 3e-05,
      "loss": 10.376,
      "step": 3
    },
    {
      "epoch": 0.016842105263157894,
      "grad_norm": 0.05906955525279045,
      "learning_rate": 4e-05,
      "loss": 10.3744,
      "step": 4
    },
    {
      "epoch": 0.021052631578947368,
      "grad_norm": 0.06294029206037521,
      "learning_rate": 5e-05,
      "loss": 10.377,
      "step": 5
    },
    {
      "epoch": 0.02526315789473684,
      "grad_norm": 0.061749327927827835,
      "learning_rate": 6e-05,
      "loss": 10.3735,
      "step": 6
    },
    {
      "epoch": 0.029473684210526315,
      "grad_norm": 0.0676150694489479,
      "learning_rate": 7e-05,
      "loss": 10.3762,
      "step": 7
    },
    {
      "epoch": 0.03368421052631579,
      "grad_norm": 0.06195235997438431,
      "learning_rate": 8e-05,
      "loss": 10.3765,
      "step": 8
    },
    {
      "epoch": 0.037894736842105266,
      "grad_norm": 0.07176747918128967,
      "learning_rate": 9e-05,
      "loss": 10.3764,
      "step": 9
    },
    {
      "epoch": 0.042105263157894736,
      "grad_norm": 0.09072773158550262,
      "learning_rate": 0.0001,
      "loss": 10.3725,
      "step": 10
    },
    {
      "epoch": 0.04631578947368421,
      "grad_norm": 0.08036182820796967,
      "learning_rate": 9.999316524962345e-05,
      "loss": 10.3791,
      "step": 11
    },
    {
      "epoch": 0.05052631578947368,
      "grad_norm": 0.0946846678853035,
      "learning_rate": 9.997266286704631e-05,
      "loss": 10.3774,
      "step": 12
    },
    {
      "epoch": 0.05473684210526316,
      "grad_norm": 0.09597119688987732,
      "learning_rate": 9.993849845741524e-05,
      "loss": 10.3779,
      "step": 13
    },
    {
      "epoch": 0.05894736842105263,
      "grad_norm": 0.09982523322105408,
      "learning_rate": 9.989068136093873e-05,
      "loss": 10.3706,
      "step": 14
    },
    {
      "epoch": 0.06315789473684211,
      "grad_norm": 0.12407989054918289,
      "learning_rate": 9.98292246503335e-05,
      "loss": 10.3744,
      "step": 15
    },
    {
      "epoch": 0.06736842105263158,
      "grad_norm": 0.1279885321855545,
      "learning_rate": 9.975414512725057e-05,
      "loss": 10.373,
      "step": 16
    },
    {
      "epoch": 0.07157894736842105,
      "grad_norm": 0.13046935200691223,
      "learning_rate": 9.966546331768191e-05,
      "loss": 10.3728,
      "step": 17
    },
    {
      "epoch": 0.07578947368421053,
      "grad_norm": 0.10874418914318085,
      "learning_rate": 9.956320346634876e-05,
      "loss": 10.3726,
      "step": 18
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1167445108294487,
      "learning_rate": 9.944739353007344e-05,
      "loss": 10.372,
      "step": 19
    },
    {
      "epoch": 0.08421052631578947,
      "grad_norm": 0.15136179327964783,
      "learning_rate": 9.931806517013612e-05,
      "loss": 10.3683,
      "step": 20
    },
    {
      "epoch": 0.08842105263157894,
      "grad_norm": 0.1556064933538437,
      "learning_rate": 9.917525374361912e-05,
      "loss": 10.3703,
      "step": 21
    },
    {
      "epoch": 0.09263157894736843,
      "grad_norm": 0.15812093019485474,
      "learning_rate": 9.901899829374047e-05,
      "loss": 10.3702,
      "step": 22
    },
    {
      "epoch": 0.0968421052631579,
      "grad_norm": 0.16039413213729858,
      "learning_rate": 9.884934153917997e-05,
      "loss": 10.3712,
      "step": 23
    },
    {
      "epoch": 0.10105263157894737,
      "grad_norm": 0.23584923148155212,
      "learning_rate": 9.86663298624003e-05,
      "loss": 10.3647,
      "step": 24
    },
    {
      "epoch": 0.10526315789473684,
      "grad_norm": 0.21401426196098328,
      "learning_rate": 9.847001329696653e-05,
      "loss": 10.3622,
      "step": 25
    },
    {
      "epoch": 0.10947368421052632,
      "grad_norm": 0.22885848581790924,
      "learning_rate": 9.826044551386744e-05,
      "loss": 10.3623,
      "step": 26
    },
    {
      "epoch": 0.11368421052631579,
      "grad_norm": 0.2253989279270172,
      "learning_rate": 9.803768380684242e-05,
      "loss": 10.3677,
      "step": 27
    },
    {
      "epoch": 0.11789473684210526,
      "grad_norm": 0.21368004381656647,
      "learning_rate": 9.780178907671789e-05,
      "loss": 10.3661,
      "step": 28
    },
    {
      "epoch": 0.12210526315789473,
      "grad_norm": 0.21518070995807648,
      "learning_rate": 9.755282581475769e-05,
      "loss": 10.3584,
      "step": 29
    },
    {
      "epoch": 0.12631578947368421,
      "grad_norm": 0.26163485646247864,
      "learning_rate": 9.729086208503174e-05,
      "loss": 10.3629,
      "step": 30
    },
    {
      "epoch": 0.13052631578947368,
      "grad_norm": 0.20369493961334229,
      "learning_rate": 9.701596950580806e-05,
      "loss": 10.3598,
      "step": 31
    },
    {
      "epoch": 0.13473684210526315,
      "grad_norm": 0.23807595670223236,
      "learning_rate": 9.672822322997305e-05,
      "loss": 10.3588,
      "step": 32
    },
    {
      "epoch": 0.13894736842105262,
      "grad_norm": 0.22704099118709564,
      "learning_rate": 9.642770192448536e-05,
      "loss": 10.3618,
      "step": 33
    },
    {
      "epoch": 0.1431578947368421,
      "grad_norm": 0.24535052478313446,
      "learning_rate": 9.611448774886924e-05,
      "loss": 10.3547,
      "step": 34
    },
    {
      "epoch": 0.14736842105263157,
      "grad_norm": 0.18469272553920746,
      "learning_rate": 9.578866633275288e-05,
      "loss": 10.36,
      "step": 35
    },
    {
      "epoch": 0.15157894736842106,
      "grad_norm": 0.23218542337417603,
      "learning_rate": 9.545032675245813e-05,
      "loss": 10.3571,
      "step": 36
    },
    {
      "epoch": 0.15578947368421053,
      "grad_norm": 0.20670539140701294,
      "learning_rate": 9.509956150664796e-05,
      "loss": 10.3537,
      "step": 37
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2033376842737198,
      "learning_rate": 9.473646649103818e-05,
      "loss": 10.3477,
      "step": 38
    },
    {
      "epoch": 0.16421052631578947,
      "grad_norm": 0.14542284607887268,
      "learning_rate": 9.43611409721806e-05,
      "loss": 10.3534,
      "step": 39
    },
    {
      "epoch": 0.16842105263157894,
      "grad_norm": 0.19221776723861694,
      "learning_rate": 9.397368756032445e-05,
      "loss": 10.3505,
      "step": 40
    },
    {
      "epoch": 0.1726315789473684,
      "grad_norm": 0.16828632354736328,
      "learning_rate": 9.357421218136386e-05,
      "loss": 10.3553,
      "step": 41
    },
    {
      "epoch": 0.17684210526315788,
      "grad_norm": 0.15972906351089478,
      "learning_rate": 9.316282404787871e-05,
      "loss": 10.3456,
      "step": 42
    },
    {
      "epoch": 0.18105263157894738,
      "grad_norm": 0.1627366542816162,
      "learning_rate": 9.273963562927695e-05,
      "loss": 10.3532,
      "step": 43
    },
    {
      "epoch": 0.18526315789473685,
      "grad_norm": 0.16887392103672028,
      "learning_rate": 9.230476262104677e-05,
      "loss": 10.3468,
      "step": 44
    },
    {
      "epoch": 0.18947368421052632,
      "grad_norm": 0.18476274609565735,
      "learning_rate": 9.185832391312644e-05,
      "loss": 10.3471,
      "step": 45
    },
    {
      "epoch": 0.1936842105263158,
      "grad_norm": 0.1950167566537857,
      "learning_rate": 9.140044155740101e-05,
      "loss": 10.3497,
      "step": 46
    },
    {
      "epoch": 0.19789473684210526,
      "grad_norm": 0.1808464229106903,
      "learning_rate": 9.093124073433463e-05,
      "loss": 10.351,
      "step": 47
    },
    {
      "epoch": 0.20210526315789473,
      "grad_norm": 0.19215361773967743,
      "learning_rate": 9.045084971874738e-05,
      "loss": 10.3543,
      "step": 48
    },
    {
      "epoch": 0.2063157894736842,
      "grad_norm": 0.16955988109111786,
      "learning_rate": 8.995939984474624e-05,
      "loss": 10.3558,
      "step": 49
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 0.19520843029022217,
      "learning_rate": 8.945702546981969e-05,
      "loss": 10.3524,
      "step": 50
    },
    {
      "epoch": 0.21052631578947367,
      "eval_loss": 10.352336883544922,
      "eval_runtime": 0.8624,
      "eval_samples_per_second": 463.8,
      "eval_steps_per_second": 115.95,
      "step": 50
    },
    {
      "epoch": 0.21473684210526317,
      "grad_norm": 0.11945804953575134,
      "learning_rate": 8.894386393810563e-05,
      "loss": 10.3643,
      "step": 51
    },
    {
      "epoch": 0.21894736842105264,
      "grad_norm": 0.1887207329273224,
      "learning_rate": 8.842005554284296e-05,
      "loss": 10.3632,
      "step": 52
    },
    {
      "epoch": 0.2231578947368421,
      "grad_norm": 0.16580155491828918,
      "learning_rate": 8.788574348801675e-05,
      "loss": 10.361,
      "step": 53
    },
    {
      "epoch": 0.22736842105263158,
      "grad_norm": 0.1485416293144226,
      "learning_rate": 8.73410738492077e-05,
      "loss": 10.3597,
      "step": 54
    },
    {
      "epoch": 0.23157894736842105,
      "grad_norm": 0.10218663513660431,
      "learning_rate": 8.678619553365659e-05,
      "loss": 10.3542,
      "step": 55
    },
    {
      "epoch": 0.23578947368421052,
      "grad_norm": 0.11669065803289413,
      "learning_rate": 8.622126023955446e-05,
      "loss": 10.3574,
      "step": 56
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.09311048686504364,
      "learning_rate": 8.564642241456986e-05,
      "loss": 10.3521,
      "step": 57
    },
    {
      "epoch": 0.24421052631578946,
      "grad_norm": 0.13593854010105133,
      "learning_rate": 8.506183921362443e-05,
      "loss": 10.3575,
      "step": 58
    },
    {
      "epoch": 0.24842105263157896,
      "grad_norm": 0.14184492826461792,
      "learning_rate": 8.44676704559283e-05,
      "loss": 10.3584,
      "step": 59
    },
    {
      "epoch": 0.25263157894736843,
      "grad_norm": 0.10373972356319427,
      "learning_rate": 8.386407858128706e-05,
      "loss": 10.3501,
      "step": 60
    },
    {
      "epoch": 0.25684210526315787,
      "grad_norm": 0.0959099754691124,
      "learning_rate": 8.32512286056924e-05,
      "loss": 10.3537,
      "step": 61
    },
    {
      "epoch": 0.26105263157894737,
      "grad_norm": 0.09872995316982269,
      "learning_rate": 8.262928807620843e-05,
      "loss": 10.3557,
      "step": 62
    },
    {
      "epoch": 0.26526315789473687,
      "grad_norm": 0.09120391309261322,
      "learning_rate": 8.199842702516583e-05,
      "loss": 10.3473,
      "step": 63
    },
    {
      "epoch": 0.2694736842105263,
      "grad_norm": 0.08507146686315536,
      "learning_rate": 8.135881792367686e-05,
      "loss": 10.3475,
      "step": 64
    },
    {
      "epoch": 0.2736842105263158,
      "grad_norm": 0.09521874785423279,
      "learning_rate": 8.07106356344834e-05,
      "loss": 10.3499,
      "step": 65
    },
    {
      "epoch": 0.27789473684210525,
      "grad_norm": 0.09604507684707642,
      "learning_rate": 8.005405736415126e-05,
      "loss": 10.3504,
      "step": 66
    },
    {
      "epoch": 0.28210526315789475,
      "grad_norm": 0.07585348933935165,
      "learning_rate": 7.938926261462366e-05,
      "loss": 10.3476,
      "step": 67
    },
    {
      "epoch": 0.2863157894736842,
      "grad_norm": 0.08676667511463165,
      "learning_rate": 7.871643313414718e-05,
      "loss": 10.3478,
      "step": 68
    },
    {
      "epoch": 0.2905263157894737,
      "grad_norm": 0.08252411335706711,
      "learning_rate": 7.803575286758364e-05,
      "loss": 10.3481,
      "step": 69
    },
    {
      "epoch": 0.29473684210526313,
      "grad_norm": 0.06295716762542725,
      "learning_rate": 7.734740790612136e-05,
      "loss": 10.3467,
      "step": 70
    },
    {
      "epoch": 0.29894736842105263,
      "grad_norm": 0.06608200818300247,
      "learning_rate": 7.66515864363997e-05,
      "loss": 10.3474,
      "step": 71
    },
    {
      "epoch": 0.3031578947368421,
      "grad_norm": 0.06749081611633301,
      "learning_rate": 7.594847868906076e-05,
      "loss": 10.3457,
      "step": 72
    },
    {
      "epoch": 0.30736842105263157,
      "grad_norm": 0.050603706389665604,
      "learning_rate": 7.52382768867422e-05,
      "loss": 10.3482,
      "step": 73
    },
    {
      "epoch": 0.31157894736842107,
      "grad_norm": 0.07785230129957199,
      "learning_rate": 7.452117519152542e-05,
      "loss": 10.3466,
      "step": 74
    },
    {
      "epoch": 0.3157894736842105,
      "grad_norm": 0.08538417518138885,
      "learning_rate": 7.379736965185368e-05,
      "loss": 10.3463,
      "step": 75
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.08417965471744537,
      "learning_rate": 7.30670581489344e-05,
      "loss": 10.3485,
      "step": 76
    },
    {
      "epoch": 0.32421052631578945,
      "grad_norm": 0.07258263230323792,
      "learning_rate": 7.233044034264034e-05,
      "loss": 10.3542,
      "step": 77
    },
    {
      "epoch": 0.32842105263157895,
      "grad_norm": 0.05613246187567711,
      "learning_rate": 7.158771761692464e-05,
      "loss": 10.348,
      "step": 78
    },
    {
      "epoch": 0.33263157894736844,
      "grad_norm": 0.05463242903351784,
      "learning_rate": 7.083909302476453e-05,
      "loss": 10.3419,
      "step": 79
    },
    {
      "epoch": 0.3368421052631579,
      "grad_norm": 0.06369648873806,
      "learning_rate": 7.008477123264848e-05,
      "loss": 10.3462,
      "step": 80
    },
    {
      "epoch": 0.3410526315789474,
      "grad_norm": 0.0644102618098259,
      "learning_rate": 6.932495846462261e-05,
      "loss": 10.3469,
      "step": 81
    },
    {
      "epoch": 0.3452631578947368,
      "grad_norm": 0.10663434118032455,
      "learning_rate": 6.855986244591104e-05,
      "loss": 10.3463,
      "step": 82
    },
    {
      "epoch": 0.3494736842105263,
      "grad_norm": 0.10768444091081619,
      "learning_rate": 6.778969234612584e-05,
      "loss": 10.3487,
      "step": 83
    },
    {
      "epoch": 0.35368421052631577,
      "grad_norm": 0.08826234191656113,
      "learning_rate": 6.701465872208216e-05,
      "loss": 10.347,
      "step": 84
    },
    {
      "epoch": 0.35789473684210527,
      "grad_norm": 0.07613211125135422,
      "learning_rate": 6.623497346023418e-05,
      "loss": 10.3469,
      "step": 85
    },
    {
      "epoch": 0.36210526315789476,
      "grad_norm": 0.10106461495161057,
      "learning_rate": 6.545084971874738e-05,
      "loss": 10.3506,
      "step": 86
    },
    {
      "epoch": 0.3663157894736842,
      "grad_norm": 0.09228483587503433,
      "learning_rate": 6.466250186922325e-05,
      "loss": 10.3473,
      "step": 87
    },
    {
      "epoch": 0.3705263157894737,
      "grad_norm": 0.0840250551700592,
      "learning_rate": 6.387014543809223e-05,
      "loss": 10.3474,
      "step": 88
    },
    {
      "epoch": 0.37473684210526315,
      "grad_norm": 0.13491995632648468,
      "learning_rate": 6.307399704769099e-05,
      "loss": 10.3494,
      "step": 89
    },
    {
      "epoch": 0.37894736842105264,
      "grad_norm": 0.10121510922908783,
      "learning_rate": 6.227427435703997e-05,
      "loss": 10.3502,
      "step": 90
    },
    {
      "epoch": 0.3831578947368421,
      "grad_norm": 0.09306132048368454,
      "learning_rate": 6.147119600233758e-05,
      "loss": 10.3435,
      "step": 91
    },
    {
      "epoch": 0.3873684210526316,
      "grad_norm": 0.10787343233823776,
      "learning_rate": 6.066498153718735e-05,
      "loss": 10.344,
      "step": 92
    },
    {
      "epoch": 0.391578947368421,
      "grad_norm": 0.0809551402926445,
      "learning_rate": 5.985585137257401e-05,
      "loss": 10.3518,
      "step": 93
    },
    {
      "epoch": 0.3957894736842105,
      "grad_norm": 0.11978474259376526,
      "learning_rate": 5.90440267166055e-05,
      "loss": 10.3501,
      "step": 94
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.11600308865308762,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 10.3523,
      "step": 95
    },
    {
      "epoch": 0.40421052631578946,
      "grad_norm": 0.12037135660648346,
      "learning_rate": 5.74131823855921e-05,
      "loss": 10.3489,
      "step": 96
    },
    {
      "epoch": 0.40842105263157896,
      "grad_norm": 0.10835761576890945,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 10.3477,
      "step": 97
    },
    {
      "epoch": 0.4126315789473684,
      "grad_norm": 0.14069664478302002,
      "learning_rate": 5.577423184847932e-05,
      "loss": 10.3474,
      "step": 98
    },
    {
      "epoch": 0.4168421052631579,
      "grad_norm": 0.12541352212429047,
      "learning_rate": 5.495227651252315e-05,
      "loss": 10.3497,
      "step": 99
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 0.21064390242099762,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 10.3525,
      "step": 100
    },
    {
      "epoch": 0.42105263157894735,
      "eval_loss": 10.3496675491333,
      "eval_runtime": 0.8601,
      "eval_samples_per_second": 465.083,
      "eval_steps_per_second": 116.271,
      "step": 100
    },
    {
      "epoch": 0.42526315789473684,
      "grad_norm": 0.13670872151851654,
      "learning_rate": 5.330452921628497e-05,
      "loss": 10.3592,
      "step": 101
    },
    {
      "epoch": 0.42947368421052634,
      "grad_norm": 0.1379602998495102,
      "learning_rate": 5.247918773366112e-05,
      "loss": 10.3587,
      "step": 102
    },
    {
      "epoch": 0.4336842105263158,
      "grad_norm": 0.1390412151813507,
      "learning_rate": 5.165316846586541e-05,
      "loss": 10.3573,
      "step": 103
    },
    {
      "epoch": 0.4378947368421053,
      "grad_norm": 0.10719850659370422,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 10.3533,
      "step": 104
    },
    {
      "epoch": 0.4421052631578947,
      "grad_norm": 0.09825416654348373,
      "learning_rate": 5e-05,
      "loss": 10.351,
      "step": 105
    },
    {
      "epoch": 0.4463157894736842,
      "grad_norm": 0.0736796036362648,
      "learning_rate": 4.917330276168208e-05,
      "loss": 10.3541,
      "step": 106
    },
    {
      "epoch": 0.45052631578947366,
      "grad_norm": 0.07018779963254929,
      "learning_rate": 4.834683153413459e-05,
      "loss": 10.3507,
      "step": 107
    },
    {
      "epoch": 0.45473684210526316,
      "grad_norm": 0.08102770894765854,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 10.3565,
      "step": 108
    },
    {
      "epoch": 0.4589473684210526,
      "grad_norm": 0.07456878572702408,
      "learning_rate": 4.669547078371504e-05,
      "loss": 10.3462,
      "step": 109
    },
    {
      "epoch": 0.4631578947368421,
      "grad_norm": 0.05995337292551994,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 10.3489,
      "step": 110
    },
    {
      "epoch": 0.4673684210526316,
      "grad_norm": 0.05509920045733452,
      "learning_rate": 4.504772348747687e-05,
      "loss": 10.3445,
      "step": 111
    },
    {
      "epoch": 0.47157894736842104,
      "grad_norm": 0.06844080239534378,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 10.3496,
      "step": 112
    },
    {
      "epoch": 0.47578947368421054,
      "grad_norm": 0.09006761759519577,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 10.3412,
      "step": 113
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.06060045212507248,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 10.3486,
      "step": 114
    },
    {
      "epoch": 0.4842105263157895,
      "grad_norm": 0.07553546875715256,
      "learning_rate": 4.17702704859633e-05,
      "loss": 10.3458,
      "step": 115
    },
    {
      "epoch": 0.4884210526315789,
      "grad_norm": 0.06478956341743469,
      "learning_rate": 4.095597328339452e-05,
      "loss": 10.3494,
      "step": 116
    },
    {
      "epoch": 0.4926315789473684,
      "grad_norm": 0.051422376185655594,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 10.3481,
      "step": 117
    },
    {
      "epoch": 0.4968421052631579,
      "grad_norm": 0.08252200484275818,
      "learning_rate": 3.933501846281267e-05,
      "loss": 10.347,
      "step": 118
    },
    {
      "epoch": 0.5010526315789474,
      "grad_norm": 0.07404341548681259,
      "learning_rate": 3.852880399766243e-05,
      "loss": 10.3483,
      "step": 119
    },
    {
      "epoch": 0.5052631578947369,
      "grad_norm": 0.07694444060325623,
      "learning_rate": 3.772572564296005e-05,
      "loss": 10.3494,
      "step": 120
    },
    {
      "epoch": 0.5094736842105263,
      "grad_norm": 0.06970033049583435,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 10.3497,
      "step": 121
    },
    {
      "epoch": 0.5136842105263157,
      "grad_norm": 0.0691736564040184,
      "learning_rate": 3.612985456190778e-05,
      "loss": 10.3498,
      "step": 122
    },
    {
      "epoch": 0.5178947368421053,
      "grad_norm": 0.06187010928988457,
      "learning_rate": 3.533749813077677e-05,
      "loss": 10.3469,
      "step": 123
    },
    {
      "epoch": 0.5221052631578947,
      "grad_norm": 0.06911417841911316,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 10.3485,
      "step": 124
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 0.05010535940527916,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 10.3464,
      "step": 125
    },
    {
      "epoch": 0.5305263157894737,
      "grad_norm": 0.06975674629211426,
      "learning_rate": 3.298534127791785e-05,
      "loss": 10.345,
      "step": 126
    },
    {
      "epoch": 0.5347368421052632,
      "grad_norm": 0.07023079693317413,
      "learning_rate": 3.221030765387417e-05,
      "loss": 10.3479,
      "step": 127
    },
    {
      "epoch": 0.5389473684210526,
      "grad_norm": 0.04967440664768219,
      "learning_rate": 3.144013755408895e-05,
      "loss": 10.3449,
      "step": 128
    },
    {
      "epoch": 0.5431578947368421,
      "grad_norm": 0.06098603829741478,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 10.3492,
      "step": 129
    },
    {
      "epoch": 0.5473684210526316,
      "grad_norm": 0.05341643467545509,
      "learning_rate": 2.991522876735154e-05,
      "loss": 10.3481,
      "step": 130
    },
    {
      "epoch": 0.5515789473684211,
      "grad_norm": 0.06596950441598892,
      "learning_rate": 2.916090697523549e-05,
      "loss": 10.3515,
      "step": 131
    },
    {
      "epoch": 0.5557894736842105,
      "grad_norm": 0.051306821405887604,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 10.3453,
      "step": 132
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.08631924539804459,
      "learning_rate": 2.766955965735968e-05,
      "loss": 10.3483,
      "step": 133
    },
    {
      "epoch": 0.5642105263157895,
      "grad_norm": 0.0771772637963295,
      "learning_rate": 2.693294185106562e-05,
      "loss": 10.3476,
      "step": 134
    },
    {
      "epoch": 0.5684210526315789,
      "grad_norm": 0.0852055475115776,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 10.3461,
      "step": 135
    },
    {
      "epoch": 0.5726315789473684,
      "grad_norm": 0.05603896453976631,
      "learning_rate": 2.547882480847461e-05,
      "loss": 10.3477,
      "step": 136
    },
    {
      "epoch": 0.5768421052631579,
      "grad_norm": 0.06591824442148209,
      "learning_rate": 2.476172311325783e-05,
      "loss": 10.3462,
      "step": 137
    },
    {
      "epoch": 0.5810526315789474,
      "grad_norm": 0.09183630347251892,
      "learning_rate": 2.405152131093926e-05,
      "loss": 10.3453,
      "step": 138
    },
    {
      "epoch": 0.5852631578947368,
      "grad_norm": 0.09669166058301926,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 10.3482,
      "step": 139
    },
    {
      "epoch": 0.5894736842105263,
      "grad_norm": 0.09249909967184067,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 10.3458,
      "step": 140
    },
    {
      "epoch": 0.5936842105263158,
      "grad_norm": 0.06754100322723389,
      "learning_rate": 2.196424713241637e-05,
      "loss": 10.3494,
      "step": 141
    },
    {
      "epoch": 0.5978947368421053,
      "grad_norm": 0.10845322906970978,
      "learning_rate": 2.128356686585282e-05,
      "loss": 10.3437,
      "step": 142
    },
    {
      "epoch": 0.6021052631578947,
      "grad_norm": 0.13180933892726898,
      "learning_rate": 2.061073738537635e-05,
      "loss": 10.3518,
      "step": 143
    },
    {
      "epoch": 0.6063157894736843,
      "grad_norm": 0.09026509523391724,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 10.3428,
      "step": 144
    },
    {
      "epoch": 0.6105263157894737,
      "grad_norm": 0.10887766629457474,
      "learning_rate": 1.928936436551661e-05,
      "loss": 10.3444,
      "step": 145
    },
    {
      "epoch": 0.6147368421052631,
      "grad_norm": 0.10717064887285233,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 10.3466,
      "step": 146
    },
    {
      "epoch": 0.6189473684210526,
      "grad_norm": 0.09877806156873703,
      "learning_rate": 1.800157297483417e-05,
      "loss": 10.3468,
      "step": 147
    },
    {
      "epoch": 0.6231578947368421,
      "grad_norm": 0.12459009885787964,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 10.3432,
      "step": 148
    },
    {
      "epoch": 0.6273684210526316,
      "grad_norm": 0.11336024105548859,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 10.3513,
      "step": 149
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 0.2063036561012268,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 10.3476,
      "step": 150
    },
    {
      "epoch": 0.631578947368421,
      "eval_loss": 10.349417686462402,
      "eval_runtime": 0.8643,
      "eval_samples_per_second": 462.795,
      "eval_steps_per_second": 115.699,
      "step": 150
    },
    {
      "epoch": 0.6357894736842106,
      "grad_norm": 0.10644425451755524,
      "learning_rate": 1.553232954407171e-05,
      "loss": 10.3562,
      "step": 151
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.134618878364563,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 10.3563,
      "step": 152
    },
    {
      "epoch": 0.6442105263157895,
      "grad_norm": 0.12683887779712677,
      "learning_rate": 1.435357758543015e-05,
      "loss": 10.3576,
      "step": 153
    },
    {
      "epoch": 0.6484210526315789,
      "grad_norm": 0.11942125856876373,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 10.3577,
      "step": 154
    },
    {
      "epoch": 0.6526315789473685,
      "grad_norm": 0.10792864114046097,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 10.3562,
      "step": 155
    },
    {
      "epoch": 0.6568421052631579,
      "grad_norm": 0.0912218764424324,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 10.3523,
      "step": 156
    },
    {
      "epoch": 0.6610526315789473,
      "grad_norm": 0.07179052382707596,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 10.3536,
      "step": 157
    },
    {
      "epoch": 0.6652631578947369,
      "grad_norm": 0.05848036706447601,
      "learning_rate": 1.157994445715706e-05,
      "loss": 10.3545,
      "step": 158
    },
    {
      "epoch": 0.6694736842105263,
      "grad_norm": 0.06571789085865021,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 10.3551,
      "step": 159
    },
    {
      "epoch": 0.6736842105263158,
      "grad_norm": 0.07704383879899979,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 10.3506,
      "step": 160
    },
    {
      "epoch": 0.6778947368421052,
      "grad_norm": 0.07392904907464981,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 10.3503,
      "step": 161
    },
    {
      "epoch": 0.6821052631578948,
      "grad_norm": 0.049188822507858276,
      "learning_rate": 9.549150281252633e-06,
      "loss": 10.3499,
      "step": 162
    },
    {
      "epoch": 0.6863157894736842,
      "grad_norm": 0.05214214697480202,
      "learning_rate": 9.068759265665384e-06,
      "loss": 10.3485,
      "step": 163
    },
    {
      "epoch": 0.6905263157894737,
      "grad_norm": 0.04309199005365372,
      "learning_rate": 8.599558442598998e-06,
      "loss": 10.344,
      "step": 164
    },
    {
      "epoch": 0.6947368421052632,
      "grad_norm": 0.05458575859665871,
      "learning_rate": 8.141676086873572e-06,
      "loss": 10.3499,
      "step": 165
    },
    {
      "epoch": 0.6989473684210527,
      "grad_norm": 0.06369191408157349,
      "learning_rate": 7.695237378953223e-06,
      "loss": 10.348,
      "step": 166
    },
    {
      "epoch": 0.7031578947368421,
      "grad_norm": 0.06323923170566559,
      "learning_rate": 7.260364370723044e-06,
      "loss": 10.3523,
      "step": 167
    },
    {
      "epoch": 0.7073684210526315,
      "grad_norm": 0.0629107877612114,
      "learning_rate": 6.837175952121306e-06,
      "loss": 10.3502,
      "step": 168
    },
    {
      "epoch": 0.7115789473684211,
      "grad_norm": 0.0670945793390274,
      "learning_rate": 6.425787818636131e-06,
      "loss": 10.346,
      "step": 169
    },
    {
      "epoch": 0.7157894736842105,
      "grad_norm": 0.09251123666763306,
      "learning_rate": 6.026312439675552e-06,
      "loss": 10.3509,
      "step": 170
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.0668419674038887,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 10.3455,
      "step": 171
    },
    {
      "epoch": 0.7242105263157895,
      "grad_norm": 0.07016400247812271,
      "learning_rate": 5.263533508961827e-06,
      "loss": 10.3458,
      "step": 172
    },
    {
      "epoch": 0.728421052631579,
      "grad_norm": 0.05713383108377457,
      "learning_rate": 4.900438493352055e-06,
      "loss": 10.3478,
      "step": 173
    },
    {
      "epoch": 0.7326315789473684,
      "grad_norm": 0.0667177215218544,
      "learning_rate": 4.549673247541875e-06,
      "loss": 10.3444,
      "step": 174
    },
    {
      "epoch": 0.7368421052631579,
      "grad_norm": 0.08438938856124878,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 10.344,
      "step": 175
    },
    {
      "epoch": 0.7410526315789474,
      "grad_norm": 0.07812978327274323,
      "learning_rate": 3.885512251130763e-06,
      "loss": 10.3506,
      "step": 176
    },
    {
      "epoch": 0.7452631578947368,
      "grad_norm": 0.09120804071426392,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 10.3491,
      "step": 177
    },
    {
      "epoch": 0.7494736842105263,
      "grad_norm": 0.0724586546421051,
      "learning_rate": 3.271776770026963e-06,
      "loss": 10.3476,
      "step": 178
    },
    {
      "epoch": 0.7536842105263157,
      "grad_norm": 0.0684080496430397,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 10.3475,
      "step": 179
    },
    {
      "epoch": 0.7578947368421053,
      "grad_norm": 0.08776247501373291,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 10.3442,
      "step": 180
    },
    {
      "epoch": 0.7621052631578947,
      "grad_norm": 0.07232668250799179,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 10.3483,
      "step": 181
    },
    {
      "epoch": 0.7663157894736842,
      "grad_norm": 0.07222525775432587,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 10.3428,
      "step": 182
    },
    {
      "epoch": 0.7705263157894737,
      "grad_norm": 0.0601932555437088,
      "learning_rate": 1.962316193157593e-06,
      "loss": 10.3476,
      "step": 183
    },
    {
      "epoch": 0.7747368421052632,
      "grad_norm": 0.05897534638643265,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 10.3463,
      "step": 184
    },
    {
      "epoch": 0.7789473684210526,
      "grad_norm": 0.06230738013982773,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 10.3471,
      "step": 185
    },
    {
      "epoch": 0.783157894736842,
      "grad_norm": 0.07036654651165009,
      "learning_rate": 1.333670137599713e-06,
      "loss": 10.3465,
      "step": 186
    },
    {
      "epoch": 0.7873684210526316,
      "grad_norm": 0.07231266796588898,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 10.3447,
      "step": 187
    },
    {
      "epoch": 0.791578947368421,
      "grad_norm": 0.08135058730840683,
      "learning_rate": 9.810017062595322e-07,
      "loss": 10.3465,
      "step": 188
    },
    {
      "epoch": 0.7957894736842105,
      "grad_norm": 0.1063043475151062,
      "learning_rate": 8.247462563808817e-07,
      "loss": 10.3428,
      "step": 189
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.10244079679250717,
      "learning_rate": 6.819348298638839e-07,
      "loss": 10.3484,
      "step": 190
    },
    {
      "epoch": 0.8042105263157895,
      "grad_norm": 0.09512383490800858,
      "learning_rate": 5.526064699265753e-07,
      "loss": 10.3469,
      "step": 191
    },
    {
      "epoch": 0.8084210526315789,
      "grad_norm": 0.0843498706817627,
      "learning_rate": 4.367965336512403e-07,
      "loss": 10.3512,
      "step": 192
    },
    {
      "epoch": 0.8126315789473684,
      "grad_norm": 0.09366335719823837,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 10.3462,
      "step": 193
    },
    {
      "epoch": 0.8168421052631579,
      "grad_norm": 0.1242184117436409,
      "learning_rate": 2.458548727494292e-07,
      "loss": 10.3507,
      "step": 194
    },
    {
      "epoch": 0.8210526315789474,
      "grad_norm": 0.09218117594718933,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 10.3461,
      "step": 195
    },
    {
      "epoch": 0.8252631578947368,
      "grad_norm": 0.12053912878036499,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 10.3445,
      "step": 196
    },
    {
      "epoch": 0.8294736842105264,
      "grad_norm": 0.09178328514099121,
      "learning_rate": 6.150154258476315e-08,
      "loss": 10.3436,
      "step": 197
    },
    {
      "epoch": 0.8336842105263158,
      "grad_norm": 0.11465375125408173,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 10.3437,
      "step": 198
    },
    {
      "epoch": 0.8378947368421052,
      "grad_norm": 0.1588142216205597,
      "learning_rate": 6.834750376549792e-09,
      "loss": 10.3467,
      "step": 199
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 0.17497660219669342,
      "learning_rate": 0.0,
      "loss": 10.3441,
      "step": 200
    },
    {
      "epoch": 0.8421052631578947,
      "eval_loss": 10.349349021911621,
      "eval_runtime": 0.8695,
      "eval_samples_per_second": 460.055,
      "eval_steps_per_second": 115.014,
      "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": 22310132121600.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}