{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.820723262374968,
  "eval_steps": 250,
  "global_step": 800,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0010259040779687098,
      "grad_norm": 9.798241135332411,
      "learning_rate": 3.3333333333333334e-08,
      "loss": 1.5793,
      "step": 1
    },
    {
      "epoch": 0.0020518081559374197,
      "grad_norm": 9.598717132793555,
      "learning_rate": 6.666666666666667e-08,
      "loss": 1.7229,
      "step": 2
    },
    {
      "epoch": 0.00307771223390613,
      "grad_norm": 10.260768981836485,
      "learning_rate": 1e-07,
      "loss": 1.6317,
      "step": 3
    },
    {
      "epoch": 0.004103616311874839,
      "grad_norm": 9.059955288981689,
      "learning_rate": 1.3333333333333334e-07,
      "loss": 1.6176,
      "step": 4
    },
    {
      "epoch": 0.005129520389843549,
      "grad_norm": 8.709116139461566,
      "learning_rate": 1.6666666666666665e-07,
      "loss": 1.5098,
      "step": 5
    },
    {
      "epoch": 0.00615542446781226,
      "grad_norm": 9.86995805736169,
      "learning_rate": 2e-07,
      "loss": 1.5554,
      "step": 6
    },
    {
      "epoch": 0.00718132854578097,
      "grad_norm": 9.269724208142758,
      "learning_rate": 2.3333333333333333e-07,
      "loss": 1.5685,
      "step": 7
    },
    {
      "epoch": 0.008207232623749679,
      "grad_norm": 9.788683578279787,
      "learning_rate": 2.6666666666666667e-07,
      "loss": 1.6332,
      "step": 8
    },
    {
      "epoch": 0.00923313670171839,
      "grad_norm": 9.329733080585035,
      "learning_rate": 3e-07,
      "loss": 1.5211,
      "step": 9
    },
    {
      "epoch": 0.010259040779687098,
      "grad_norm": 8.603455933686602,
      "learning_rate": 3.333333333333333e-07,
      "loss": 1.531,
      "step": 10
    },
    {
      "epoch": 0.011284944857655809,
      "grad_norm": 9.703453172271272,
      "learning_rate": 3.666666666666666e-07,
      "loss": 1.5336,
      "step": 11
    },
    {
      "epoch": 0.01231084893562452,
      "grad_norm": 8.366416586343489,
      "learning_rate": 4e-07,
      "loss": 1.5916,
      "step": 12
    },
    {
      "epoch": 0.013336753013593229,
      "grad_norm": 8.816992612572204,
      "learning_rate": 4.3333333333333335e-07,
      "loss": 1.6659,
      "step": 13
    },
    {
      "epoch": 0.01436265709156194,
      "grad_norm": 9.162032804849005,
      "learning_rate": 4.6666666666666666e-07,
      "loss": 1.5406,
      "step": 14
    },
    {
      "epoch": 0.015388561169530648,
      "grad_norm": 8.732450364954722,
      "learning_rate": 5e-07,
      "loss": 1.5083,
      "step": 15
    },
    {
      "epoch": 0.016414465247499357,
      "grad_norm": 8.622668586551894,
      "learning_rate": 5.333333333333333e-07,
      "loss": 1.5596,
      "step": 16
    },
    {
      "epoch": 0.01744036932546807,
      "grad_norm": 7.133974565680819,
      "learning_rate": 5.666666666666666e-07,
      "loss": 1.5639,
      "step": 17
    },
    {
      "epoch": 0.01846627340343678,
      "grad_norm": 6.991354736869125,
      "learning_rate": 6e-07,
      "loss": 1.4727,
      "step": 18
    },
    {
      "epoch": 0.019492177481405488,
      "grad_norm": 6.779705570699807,
      "learning_rate": 6.333333333333332e-07,
      "loss": 1.5018,
      "step": 19
    },
    {
      "epoch": 0.020518081559374197,
      "grad_norm": 6.980027846638078,
      "learning_rate": 6.666666666666666e-07,
      "loss": 1.5915,
      "step": 20
    },
    {
      "epoch": 0.02154398563734291,
      "grad_norm": 6.45230816051997,
      "learning_rate": 7e-07,
      "loss": 1.4633,
      "step": 21
    },
    {
      "epoch": 0.022569889715311618,
      "grad_norm": 5.935560467510461,
      "learning_rate": 7.333333333333332e-07,
      "loss": 1.519,
      "step": 22
    },
    {
      "epoch": 0.023595793793280327,
      "grad_norm": 5.412080062014559,
      "learning_rate": 7.666666666666667e-07,
      "loss": 1.5106,
      "step": 23
    },
    {
      "epoch": 0.02462169787124904,
      "grad_norm": 4.643500965809458,
      "learning_rate": 8e-07,
      "loss": 1.3979,
      "step": 24
    },
    {
      "epoch": 0.02564760194921775,
      "grad_norm": 4.425426929208595,
      "learning_rate": 8.333333333333333e-07,
      "loss": 1.3665,
      "step": 25
    },
    {
      "epoch": 0.026673506027186458,
      "grad_norm": 4.222303529390681,
      "learning_rate": 8.666666666666667e-07,
      "loss": 1.4494,
      "step": 26
    },
    {
      "epoch": 0.027699410105155167,
      "grad_norm": 4.369459716992631,
      "learning_rate": 9e-07,
      "loss": 1.419,
      "step": 27
    },
    {
      "epoch": 0.02872531418312388,
      "grad_norm": 4.039114576068017,
      "learning_rate": 9.333333333333333e-07,
      "loss": 1.4929,
      "step": 28
    },
    {
      "epoch": 0.029751218261092588,
      "grad_norm": 3.7836259253536135,
      "learning_rate": 9.666666666666666e-07,
      "loss": 1.3719,
      "step": 29
    },
    {
      "epoch": 0.030777122339061297,
      "grad_norm": 3.706293156036854,
      "learning_rate": 1e-06,
      "loss": 1.4779,
      "step": 30
    },
    {
      "epoch": 0.031803026417030006,
      "grad_norm": 3.460814827864659,
      "learning_rate": 9.999972311759113e-07,
      "loss": 1.3536,
      "step": 31
    },
    {
      "epoch": 0.032828930494998715,
      "grad_norm": 3.3017900915165086,
      "learning_rate": 9.99988924734311e-07,
      "loss": 1.3524,
      "step": 32
    },
    {
      "epoch": 0.03385483457296743,
      "grad_norm": 3.455224010733439,
      "learning_rate": 9.999750807671956e-07,
      "loss": 1.4297,
      "step": 33
    },
    {
      "epoch": 0.03488073865093614,
      "grad_norm": 3.63256540031341,
      "learning_rate": 9.999556994278908e-07,
      "loss": 1.4719,
      "step": 34
    },
    {
      "epoch": 0.03590664272890485,
      "grad_norm": 3.8514951366560823,
      "learning_rate": 9.999307809310508e-07,
      "loss": 1.3979,
      "step": 35
    },
    {
      "epoch": 0.03693254680687356,
      "grad_norm": 3.9586226136352396,
      "learning_rate": 9.999003255526553e-07,
      "loss": 1.3285,
      "step": 36
    },
    {
      "epoch": 0.03795845088484227,
      "grad_norm": 3.806464587016546,
      "learning_rate": 9.998643336300069e-07,
      "loss": 1.4278,
      "step": 37
    },
    {
      "epoch": 0.038984354962810976,
      "grad_norm": 3.153561780968849,
      "learning_rate": 9.998228055617262e-07,
      "loss": 1.4004,
      "step": 38
    },
    {
      "epoch": 0.040010259040779685,
      "grad_norm": 3.281790653235281,
      "learning_rate": 9.997757418077494e-07,
      "loss": 1.3469,
      "step": 39
    },
    {
      "epoch": 0.041036163118748394,
      "grad_norm": 3.1079807809050792,
      "learning_rate": 9.997231428893215e-07,
      "loss": 1.2929,
      "step": 40
    },
    {
      "epoch": 0.04206206719671711,
      "grad_norm": 2.940384594816459,
      "learning_rate": 9.99665009388991e-07,
      "loss": 1.2889,
      "step": 41
    },
    {
      "epoch": 0.04308797127468582,
      "grad_norm": 2.9910864286989036,
      "learning_rate": 9.996013419506033e-07,
      "loss": 1.3704,
      "step": 42
    },
    {
      "epoch": 0.04411387535265453,
      "grad_norm": 3.097652651590371,
      "learning_rate": 9.995321412792947e-07,
      "loss": 1.4157,
      "step": 43
    },
    {
      "epoch": 0.045139779430623236,
      "grad_norm": 2.9908105852033384,
      "learning_rate": 9.994574081414829e-07,
      "loss": 1.3665,
      "step": 44
    },
    {
      "epoch": 0.046165683508591945,
      "grad_norm": 3.1050321703200847,
      "learning_rate": 9.993771433648598e-07,
      "loss": 1.5106,
      "step": 45
    },
    {
      "epoch": 0.047191587586560654,
      "grad_norm": 3.0440366145712945,
      "learning_rate": 9.992913478383809e-07,
      "loss": 1.4094,
      "step": 46
    },
    {
      "epoch": 0.04821749166452936,
      "grad_norm": 2.6336951924027123,
      "learning_rate": 9.992000225122578e-07,
      "loss": 1.2391,
      "step": 47
    },
    {
      "epoch": 0.04924339574249808,
      "grad_norm": 2.9411301278665816,
      "learning_rate": 9.991031683979451e-07,
      "loss": 1.4004,
      "step": 48
    },
    {
      "epoch": 0.05026929982046679,
      "grad_norm": 2.588518890678191,
      "learning_rate": 9.990007865681312e-07,
      "loss": 1.4288,
      "step": 49
    },
    {
      "epoch": 0.0512952038984355,
      "grad_norm": 3.0977479141233384,
      "learning_rate": 9.98892878156725e-07,
      "loss": 1.2934,
      "step": 50
    },
    {
      "epoch": 0.052321107976404206,
      "grad_norm": 2.891254887655398,
      "learning_rate": 9.98779444358844e-07,
      "loss": 1.2993,
      "step": 51
    },
    {
      "epoch": 0.053347012054372915,
      "grad_norm": 2.8843613231387346,
      "learning_rate": 9.986604864308015e-07,
      "loss": 1.3691,
      "step": 52
    },
    {
      "epoch": 0.054372916132341624,
      "grad_norm": 3.058804698335043,
      "learning_rate": 9.985360056900914e-07,
      "loss": 1.3385,
      "step": 53
    },
    {
      "epoch": 0.05539882021031033,
      "grad_norm": 2.908824401209796,
      "learning_rate": 9.98406003515375e-07,
      "loss": 1.4562,
      "step": 54
    },
    {
      "epoch": 0.05642472428827905,
      "grad_norm": 3.5082197663137977,
      "learning_rate": 9.98270481346465e-07,
      "loss": 1.3692,
      "step": 55
    },
    {
      "epoch": 0.05745062836624776,
      "grad_norm": 2.94858359371842,
      "learning_rate": 9.981294406843093e-07,
      "loss": 1.3498,
      "step": 56
    },
    {
      "epoch": 0.05847653244421647,
      "grad_norm": 2.6795852716884374,
      "learning_rate": 9.979828830909754e-07,
      "loss": 1.3321,
      "step": 57
    },
    {
      "epoch": 0.059502436522185176,
      "grad_norm": 2.9978152584208053,
      "learning_rate": 9.978308101896316e-07,
      "loss": 1.3412,
      "step": 58
    },
    {
      "epoch": 0.060528340600153885,
      "grad_norm": 2.7552572011179497,
      "learning_rate": 9.97673223664531e-07,
      "loss": 1.4021,
      "step": 59
    },
    {
      "epoch": 0.061554244678122594,
      "grad_norm": 2.637599483243523,
      "learning_rate": 9.975101252609903e-07,
      "loss": 1.3292,
      "step": 60
    },
    {
      "epoch": 0.0625801487560913,
      "grad_norm": 2.938536066745437,
      "learning_rate": 9.973415167853734e-07,
      "loss": 1.2372,
      "step": 61
    },
    {
      "epoch": 0.06360605283406001,
      "grad_norm": 2.7530123525727466,
      "learning_rate": 9.971674001050686e-07,
      "loss": 1.3277,
      "step": 62
    },
    {
      "epoch": 0.06463195691202872,
      "grad_norm": 2.753702797656271,
      "learning_rate": 9.969877771484698e-07,
      "loss": 1.2756,
      "step": 63
    },
    {
      "epoch": 0.06565786098999743,
      "grad_norm": 2.707817535821633,
      "learning_rate": 9.968026499049549e-07,
      "loss": 1.2586,
      "step": 64
    },
    {
      "epoch": 0.06668376506796614,
      "grad_norm": 2.7234356724029567,
      "learning_rate": 9.966120204248625e-07,
      "loss": 1.3249,
      "step": 65
    },
    {
      "epoch": 0.06770966914593486,
      "grad_norm": 2.92461615737126,
      "learning_rate": 9.964158908194706e-07,
      "loss": 1.3265,
      "step": 66
    },
    {
      "epoch": 0.06873557322390357,
      "grad_norm": 2.593339469946713,
      "learning_rate": 9.962142632609732e-07,
      "loss": 1.3534,
      "step": 67
    },
    {
      "epoch": 0.06976147730187228,
      "grad_norm": 2.9860559274993466,
      "learning_rate": 9.960071399824547e-07,
      "loss": 1.3433,
      "step": 68
    },
    {
      "epoch": 0.07078738137984099,
      "grad_norm": 2.6875123035895796,
      "learning_rate": 9.957945232778672e-07,
      "loss": 1.3273,
      "step": 69
    },
    {
      "epoch": 0.0718132854578097,
      "grad_norm": 2.870657819908334,
      "learning_rate": 9.955764155020035e-07,
      "loss": 1.2982,
      "step": 70
    },
    {
      "epoch": 0.0728391895357784,
      "grad_norm": 3.0078112960966137,
      "learning_rate": 9.95352819070472e-07,
      "loss": 1.2967,
      "step": 71
    },
    {
      "epoch": 0.07386509361374712,
      "grad_norm": 2.6631173635617307,
      "learning_rate": 9.951237364596692e-07,
      "loss": 1.3039,
      "step": 72
    },
    {
      "epoch": 0.07489099769171582,
      "grad_norm": 2.903289343918584,
      "learning_rate": 9.94889170206753e-07,
      "loss": 1.2931,
      "step": 73
    },
    {
      "epoch": 0.07591690176968453,
      "grad_norm": 2.5803057970449594,
      "learning_rate": 9.946491229096141e-07,
      "loss": 1.3029,
      "step": 74
    },
    {
      "epoch": 0.07694280584765324,
      "grad_norm": 2.587911044453264,
      "learning_rate": 9.94403597226848e-07,
      "loss": 1.233,
      "step": 75
    },
    {
      "epoch": 0.07796870992562195,
      "grad_norm": 2.7832901812089923,
      "learning_rate": 9.941525958777235e-07,
      "loss": 1.2965,
      "step": 76
    },
    {
      "epoch": 0.07899461400359066,
      "grad_norm": 2.681414225480931,
      "learning_rate": 9.938961216421557e-07,
      "loss": 1.274,
      "step": 77
    },
    {
      "epoch": 0.08002051808155937,
      "grad_norm": 2.749129791384422,
      "learning_rate": 9.936341773606722e-07,
      "loss": 1.3564,
      "step": 78
    },
    {
      "epoch": 0.08104642215952808,
      "grad_norm": 2.615169662982429,
      "learning_rate": 9.93366765934384e-07,
      "loss": 1.3874,
      "step": 79
    },
    {
      "epoch": 0.08207232623749679,
      "grad_norm": 2.7336596773654627,
      "learning_rate": 9.930938903249516e-07,
      "loss": 1.2904,
      "step": 80
    },
    {
      "epoch": 0.08309823031546551,
      "grad_norm": 2.695518831245502,
      "learning_rate": 9.928155535545534e-07,
      "loss": 1.351,
      "step": 81
    },
    {
      "epoch": 0.08412413439343422,
      "grad_norm": 2.475635111783967,
      "learning_rate": 9.925317587058514e-07,
      "loss": 1.2317,
      "step": 82
    },
    {
      "epoch": 0.08515003847140293,
      "grad_norm": 2.8880852942092607,
      "learning_rate": 9.922425089219581e-07,
      "loss": 1.225,
      "step": 83
    },
    {
      "epoch": 0.08617594254937164,
      "grad_norm": 2.8478287390427353,
      "learning_rate": 9.919478074064001e-07,
      "loss": 1.261,
      "step": 84
    },
    {
      "epoch": 0.08720184662734035,
      "grad_norm": 2.637223722117266,
      "learning_rate": 9.91647657423084e-07,
      "loss": 1.3015,
      "step": 85
    },
    {
      "epoch": 0.08822775070530905,
      "grad_norm": 2.654960461830896,
      "learning_rate": 9.913420622962604e-07,
      "loss": 1.3029,
      "step": 86
    },
    {
      "epoch": 0.08925365478327776,
      "grad_norm": 3.075422859212528,
      "learning_rate": 9.910310254104854e-07,
      "loss": 1.2842,
      "step": 87
    },
    {
      "epoch": 0.09027955886124647,
      "grad_norm": 2.7924836932599635,
      "learning_rate": 9.907145502105846e-07,
      "loss": 1.2982,
      "step": 88
    },
    {
      "epoch": 0.09130546293921518,
      "grad_norm": 2.6508581949070016,
      "learning_rate": 9.90392640201615e-07,
      "loss": 1.3225,
      "step": 89
    },
    {
      "epoch": 0.09233136701718389,
      "grad_norm": 2.7887285958862904,
      "learning_rate": 9.900652989488253e-07,
      "loss": 1.3516,
      "step": 90
    },
    {
      "epoch": 0.0933572710951526,
      "grad_norm": 2.778310945122497,
      "learning_rate": 9.897325300776167e-07,
      "loss": 1.2409,
      "step": 91
    },
    {
      "epoch": 0.09438317517312131,
      "grad_norm": 2.771955143672296,
      "learning_rate": 9.893943372735032e-07,
      "loss": 1.3554,
      "step": 92
    },
    {
      "epoch": 0.09540907925109002,
      "grad_norm": 2.641330715037395,
      "learning_rate": 9.8905072428207e-07,
      "loss": 1.2767,
      "step": 93
    },
    {
      "epoch": 0.09643498332905873,
      "grad_norm": 2.7999670492066673,
      "learning_rate": 9.887016949089332e-07,
      "loss": 1.2957,
      "step": 94
    },
    {
      "epoch": 0.09746088740702745,
      "grad_norm": 2.9949645451613134,
      "learning_rate": 9.883472530196966e-07,
      "loss": 1.3359,
      "step": 95
    },
    {
      "epoch": 0.09848679148499616,
      "grad_norm": 2.497849004552429,
      "learning_rate": 9.879874025399087e-07,
      "loss": 1.3566,
      "step": 96
    },
    {
      "epoch": 0.09951269556296487,
      "grad_norm": 2.722740998468592,
      "learning_rate": 9.876221474550207e-07,
      "loss": 1.3512,
      "step": 97
    },
    {
      "epoch": 0.10053859964093358,
      "grad_norm": 2.953223838054628,
      "learning_rate": 9.872514918103405e-07,
      "loss": 1.3452,
      "step": 98
    },
    {
      "epoch": 0.10156450371890229,
      "grad_norm": 2.5063398128121417,
      "learning_rate": 9.868754397109895e-07,
      "loss": 1.3567,
      "step": 99
    },
    {
      "epoch": 0.102590407796871,
      "grad_norm": 2.5885598718043066,
      "learning_rate": 9.864939953218561e-07,
      "loss": 1.2673,
      "step": 100
    },
    {
      "epoch": 0.1036163118748397,
      "grad_norm": 2.620316954154948,
      "learning_rate": 9.8610716286755e-07,
      "loss": 1.3432,
      "step": 101
    },
    {
      "epoch": 0.10464221595280841,
      "grad_norm": 2.7576659564287995,
      "learning_rate": 9.85714946632355e-07,
      "loss": 1.2954,
      "step": 102
    },
    {
      "epoch": 0.10566812003077712,
      "grad_norm": 2.6107378359741147,
      "learning_rate": 9.853173509601823e-07,
      "loss": 1.3089,
      "step": 103
    },
    {
      "epoch": 0.10669402410874583,
      "grad_norm": 2.72531292391567,
      "learning_rate": 9.84914380254522e-07,
      "loss": 1.2754,
      "step": 104
    },
    {
      "epoch": 0.10771992818671454,
      "grad_norm": 2.6204968346720197,
      "learning_rate": 9.845060389783937e-07,
      "loss": 1.311,
      "step": 105
    },
    {
      "epoch": 0.10874583226468325,
      "grad_norm": 2.403778961837896,
      "learning_rate": 9.840923316542983e-07,
      "loss": 1.2194,
      "step": 106
    },
    {
      "epoch": 0.10977173634265196,
      "grad_norm": 2.841225529794086,
      "learning_rate": 9.83673262864167e-07,
      "loss": 1.2714,
      "step": 107
    },
    {
      "epoch": 0.11079764042062067,
      "grad_norm": 2.680466409381784,
      "learning_rate": 9.832488372493108e-07,
      "loss": 1.3238,
      "step": 108
    },
    {
      "epoch": 0.11182354449858938,
      "grad_norm": 2.7899613949194393,
      "learning_rate": 9.82819059510369e-07,
      "loss": 1.3607,
      "step": 109
    },
    {
      "epoch": 0.1128494485765581,
      "grad_norm": 2.741199596943549,
      "learning_rate": 9.82383934407258e-07,
      "loss": 1.2373,
      "step": 110
    },
    {
      "epoch": 0.1138753526545268,
      "grad_norm": 2.680900338879034,
      "learning_rate": 9.819434667591166e-07,
      "loss": 1.332,
      "step": 111
    },
    {
      "epoch": 0.11490125673249552,
      "grad_norm": 2.6867928214379266,
      "learning_rate": 9.814976614442547e-07,
      "loss": 1.2636,
      "step": 112
    },
    {
      "epoch": 0.11592716081046422,
      "grad_norm": 2.530746336531516,
      "learning_rate": 9.810465234000987e-07,
      "loss": 1.2427,
      "step": 113
    },
    {
      "epoch": 0.11695306488843293,
      "grad_norm": 3.079117319825233,
      "learning_rate": 9.805900576231357e-07,
      "loss": 1.3352,
      "step": 114
    },
    {
      "epoch": 0.11797896896640164,
      "grad_norm": 2.705858226495669,
      "learning_rate": 9.801282691688595e-07,
      "loss": 1.2462,
      "step": 115
    },
    {
      "epoch": 0.11900487304437035,
      "grad_norm": 2.642585557515256,
      "learning_rate": 9.796611631517141e-07,
      "loss": 1.3102,
      "step": 116
    },
    {
      "epoch": 0.12003077712233906,
      "grad_norm": 2.591744882697605,
      "learning_rate": 9.791887447450374e-07,
      "loss": 1.3265,
      "step": 117
    },
    {
      "epoch": 0.12105668120030777,
      "grad_norm": 2.5766253217072137,
      "learning_rate": 9.787110191810026e-07,
      "loss": 1.3054,
      "step": 118
    },
    {
      "epoch": 0.12208258527827648,
      "grad_norm": 2.77284938502092,
      "learning_rate": 9.782279917505627e-07,
      "loss": 1.245,
      "step": 119
    },
    {
      "epoch": 0.12310848935624519,
      "grad_norm": 2.8563203008600837,
      "learning_rate": 9.77739667803389e-07,
      "loss": 1.3478,
      "step": 120
    },
    {
      "epoch": 0.1241343934342139,
      "grad_norm": 2.565482050398409,
      "learning_rate": 9.772460527478142e-07,
      "loss": 1.2775,
      "step": 121
    },
    {
      "epoch": 0.1251602975121826,
      "grad_norm": 2.7054539949137615,
      "learning_rate": 9.76747152050771e-07,
      "loss": 1.3187,
      "step": 122
    },
    {
      "epoch": 0.12618620159015131,
      "grad_norm": 2.5989084819509967,
      "learning_rate": 9.762429712377331e-07,
      "loss": 1.2934,
      "step": 123
    },
    {
      "epoch": 0.12721210566812002,
      "grad_norm": 2.510970489271258,
      "learning_rate": 9.75733515892652e-07,
      "loss": 1.2603,
      "step": 124
    },
    {
      "epoch": 0.12823800974608873,
      "grad_norm": 2.5595730950106925,
      "learning_rate": 9.752187916578967e-07,
      "loss": 1.2883,
      "step": 125
    },
    {
      "epoch": 0.12926391382405744,
      "grad_norm": 2.67442136336687,
      "learning_rate": 9.746988042341907e-07,
      "loss": 1.2817,
      "step": 126
    },
    {
      "epoch": 0.13028981790202615,
      "grad_norm": 2.708509570780934,
      "learning_rate": 9.741735593805486e-07,
      "loss": 1.2078,
      "step": 127
    },
    {
      "epoch": 0.13131572197999486,
      "grad_norm": 2.678010158167899,
      "learning_rate": 9.736430629142128e-07,
      "loss": 1.3796,
      "step": 128
    },
    {
      "epoch": 0.13234162605796357,
      "grad_norm": 2.844285265615473,
      "learning_rate": 9.731073207105896e-07,
      "loss": 1.3091,
      "step": 129
    },
    {
      "epoch": 0.13336753013593228,
      "grad_norm": 2.7814463008959334,
      "learning_rate": 9.725663387031816e-07,
      "loss": 1.3682,
      "step": 130
    },
    {
      "epoch": 0.13439343421390101,
      "grad_norm": 2.6513652068190403,
      "learning_rate": 9.720201228835256e-07,
      "loss": 1.2806,
      "step": 131
    },
    {
      "epoch": 0.13541933829186972,
      "grad_norm": 2.808046184764404,
      "learning_rate": 9.714686793011235e-07,
      "loss": 1.1529,
      "step": 132
    },
    {
      "epoch": 0.13644524236983843,
      "grad_norm": 2.6158983121591004,
      "learning_rate": 9.709120140633763e-07,
      "loss": 1.2301,
      "step": 133
    },
    {
      "epoch": 0.13747114644780714,
      "grad_norm": 2.6087976837956406,
      "learning_rate": 9.703501333355166e-07,
      "loss": 1.2512,
      "step": 134
    },
    {
      "epoch": 0.13849705052577585,
      "grad_norm": 2.3938681704241818,
      "learning_rate": 9.697830433405399e-07,
      "loss": 1.3158,
      "step": 135
    },
    {
      "epoch": 0.13952295460374456,
      "grad_norm": 2.5859416655464216,
      "learning_rate": 9.692107503591358e-07,
      "loss": 1.1996,
      "step": 136
    },
    {
      "epoch": 0.14054885868171327,
      "grad_norm": 2.814296800240032,
      "learning_rate": 9.68633260729619e-07,
      "loss": 1.3123,
      "step": 137
    },
    {
      "epoch": 0.14157476275968198,
      "grad_norm": 2.4891637404994125,
      "learning_rate": 9.680505808478581e-07,
      "loss": 1.2231,
      "step": 138
    },
    {
      "epoch": 0.14260066683765069,
      "grad_norm": 2.863453159988515,
      "learning_rate": 9.674627171672054e-07,
      "loss": 1.2959,
      "step": 139
    },
    {
      "epoch": 0.1436265709156194,
      "grad_norm": 2.538454344691091,
      "learning_rate": 9.668696761984254e-07,
      "loss": 1.2674,
      "step": 140
    },
    {
      "epoch": 0.1446524749935881,
      "grad_norm": 2.5004273929320395,
      "learning_rate": 9.662714645096229e-07,
      "loss": 1.2116,
      "step": 141
    },
    {
      "epoch": 0.1456783790715568,
      "grad_norm": 2.6371352778169657,
      "learning_rate": 9.656680887261692e-07,
      "loss": 1.2476,
      "step": 142
    },
    {
      "epoch": 0.14670428314952552,
      "grad_norm": 2.683569477616342,
      "learning_rate": 9.650595555306302e-07,
      "loss": 1.3046,
      "step": 143
    },
    {
      "epoch": 0.14773018722749423,
      "grad_norm": 2.912553498396746,
      "learning_rate": 9.644458716626911e-07,
      "loss": 1.2318,
      "step": 144
    },
    {
      "epoch": 0.14875609130546294,
      "grad_norm": 2.5818573358818266,
      "learning_rate": 9.63827043919083e-07,
      "loss": 1.3017,
      "step": 145
    },
    {
      "epoch": 0.14978199538343165,
      "grad_norm": 2.6673972834050543,
      "learning_rate": 9.63203079153506e-07,
      "loss": 1.2966,
      "step": 146
    },
    {
      "epoch": 0.15080789946140036,
      "grad_norm": 2.8672403287450523,
      "learning_rate": 9.625739842765556e-07,
      "loss": 1.3346,
      "step": 147
    },
    {
      "epoch": 0.15183380353936907,
      "grad_norm": 2.5970617764295496,
      "learning_rate": 9.619397662556433e-07,
      "loss": 1.2169,
      "step": 148
    },
    {
      "epoch": 0.15285970761733778,
      "grad_norm": 2.4979249789960556,
      "learning_rate": 9.61300432114922e-07,
      "loss": 1.3695,
      "step": 149
    },
    {
      "epoch": 0.15388561169530648,
      "grad_norm": 2.479601711066903,
      "learning_rate": 9.606559889352063e-07,
      "loss": 1.2008,
      "step": 150
    },
    {
      "epoch": 0.1549115157732752,
      "grad_norm": 2.5898796136386144,
      "learning_rate": 9.600064438538961e-07,
      "loss": 1.328,
      "step": 151
    },
    {
      "epoch": 0.1559374198512439,
      "grad_norm": 2.798734600781089,
      "learning_rate": 9.593518040648952e-07,
      "loss": 1.244,
      "step": 152
    },
    {
      "epoch": 0.1569633239292126,
      "grad_norm": 2.4882934421406837,
      "learning_rate": 9.586920768185333e-07,
      "loss": 1.2426,
      "step": 153
    },
    {
      "epoch": 0.15798922800718132,
      "grad_norm": 2.802602250049014,
      "learning_rate": 9.580272694214854e-07,
      "loss": 1.2628,
      "step": 154
    },
    {
      "epoch": 0.15901513208515003,
      "grad_norm": 2.930136632494125,
      "learning_rate": 9.573573892366903e-07,
      "loss": 1.2351,
      "step": 155
    },
    {
      "epoch": 0.16004103616311874,
      "grad_norm": 2.3313868113568446,
      "learning_rate": 9.566824436832695e-07,
      "loss": 1.303,
      "step": 156
    },
    {
      "epoch": 0.16106694024108745,
      "grad_norm": 2.616895617747635,
      "learning_rate": 9.56002440236445e-07,
      "loss": 1.3128,
      "step": 157
    },
    {
      "epoch": 0.16209284431905616,
      "grad_norm": 2.5932182189076647,
      "learning_rate": 9.553173864274566e-07,
      "loss": 1.3312,
      "step": 158
    },
    {
      "epoch": 0.16311874839702487,
      "grad_norm": 2.6881235109044184,
      "learning_rate": 9.54627289843478e-07,
      "loss": 1.2495,
      "step": 159
    },
    {
      "epoch": 0.16414465247499357,
      "grad_norm": 2.876970939352412,
      "learning_rate": 9.539321581275342e-07,
      "loss": 1.3992,
      "step": 160
    },
    {
      "epoch": 0.1651705565529623,
      "grad_norm": 2.762056056919623,
      "learning_rate": 9.532319989784139e-07,
      "loss": 1.3294,
      "step": 161
    },
    {
      "epoch": 0.16619646063093102,
      "grad_norm": 2.5107504330050636,
      "learning_rate": 9.525268201505878e-07,
      "loss": 1.331,
      "step": 162
    },
    {
      "epoch": 0.16722236470889973,
      "grad_norm": 2.7655428881842754,
      "learning_rate": 9.518166294541203e-07,
      "loss": 1.3166,
      "step": 163
    },
    {
      "epoch": 0.16824826878686844,
      "grad_norm": 2.5472143589800678,
      "learning_rate": 9.511014347545837e-07,
      "loss": 1.2435,
      "step": 164
    },
    {
      "epoch": 0.16927417286483715,
      "grad_norm": 2.9317558173367084,
      "learning_rate": 9.503812439729714e-07,
      "loss": 1.2583,
      "step": 165
    },
    {
      "epoch": 0.17030007694280586,
      "grad_norm": 2.723414624661715,
      "learning_rate": 9.496560650856096e-07,
      "loss": 1.2796,
      "step": 166
    },
    {
      "epoch": 0.17132598102077456,
      "grad_norm": 2.5426652630538418,
      "learning_rate": 9.489259061240695e-07,
      "loss": 1.2991,
      "step": 167
    },
    {
      "epoch": 0.17235188509874327,
      "grad_norm": 2.3609000901953716,
      "learning_rate": 9.481907751750779e-07,
      "loss": 1.2761,
      "step": 168
    },
    {
      "epoch": 0.17337778917671198,
      "grad_norm": 2.6737934941791885,
      "learning_rate": 9.474506803804279e-07,
      "loss": 1.2034,
      "step": 169
    },
    {
      "epoch": 0.1744036932546807,
      "grad_norm": 2.7034503852388494,
      "learning_rate": 9.467056299368887e-07,
      "loss": 1.2918,
      "step": 170
    },
    {
      "epoch": 0.1754295973326494,
      "grad_norm": 2.6495133759903857,
      "learning_rate": 9.459556320961151e-07,
      "loss": 1.268,
      "step": 171
    },
    {
      "epoch": 0.1764555014106181,
      "grad_norm": 2.6876955044079542,
      "learning_rate": 9.452006951645548e-07,
      "loss": 1.1678,
      "step": 172
    },
    {
      "epoch": 0.17748140548858682,
      "grad_norm": 2.5152653942206045,
      "learning_rate": 9.444408275033586e-07,
      "loss": 1.1947,
      "step": 173
    },
    {
      "epoch": 0.17850730956655553,
      "grad_norm": 2.802330037908763,
      "learning_rate": 9.436760375282857e-07,
      "loss": 1.3259,
      "step": 174
    },
    {
      "epoch": 0.17953321364452424,
      "grad_norm": 2.7422708968039116,
      "learning_rate": 9.429063337096119e-07,
      "loss": 1.2465,
      "step": 175
    },
    {
      "epoch": 0.18055911772249295,
      "grad_norm": 2.826066535856552,
      "learning_rate": 9.421317245720352e-07,
      "loss": 1.3284,
      "step": 176
    },
    {
      "epoch": 0.18158502180046165,
      "grad_norm": 2.466040916453339,
      "learning_rate": 9.41352218694581e-07,
      "loss": 1.2419,
      "step": 177
    },
    {
      "epoch": 0.18261092587843036,
      "grad_norm": 2.837491368634976,
      "learning_rate": 9.405678247105082e-07,
      "loss": 1.2956,
      "step": 178
    },
    {
      "epoch": 0.18363682995639907,
      "grad_norm": 2.704111429861623,
      "learning_rate": 9.397785513072126e-07,
      "loss": 1.2389,
      "step": 179
    },
    {
      "epoch": 0.18466273403436778,
      "grad_norm": 2.427829411975647,
      "learning_rate": 9.38984407226131e-07,
      "loss": 1.2553,
      "step": 180
    },
    {
      "epoch": 0.1856886381123365,
      "grad_norm": 2.6889270088014343,
      "learning_rate": 9.381854012626443e-07,
      "loss": 1.2249,
      "step": 181
    },
    {
      "epoch": 0.1867145421903052,
      "grad_norm": 3.0319367008791813,
      "learning_rate": 9.373815422659805e-07,
      "loss": 1.2405,
      "step": 182
    },
    {
      "epoch": 0.1877404462682739,
      "grad_norm": 2.7686852678293485,
      "learning_rate": 9.365728391391164e-07,
      "loss": 1.2309,
      "step": 183
    },
    {
      "epoch": 0.18876635034624262,
      "grad_norm": 2.6725011393667577,
      "learning_rate": 9.357593008386784e-07,
      "loss": 1.2986,
      "step": 184
    },
    {
      "epoch": 0.18979225442421133,
      "grad_norm": 2.4605351470645083,
      "learning_rate": 9.349409363748444e-07,
      "loss": 1.1675,
      "step": 185
    },
    {
      "epoch": 0.19081815850218004,
      "grad_norm": 2.592275913149399,
      "learning_rate": 9.341177548112436e-07,
      "loss": 1.3206,
      "step": 186
    },
    {
      "epoch": 0.19184406258014874,
      "grad_norm": 2.941294249687017,
      "learning_rate": 9.332897652648555e-07,
      "loss": 1.2446,
      "step": 187
    },
    {
      "epoch": 0.19286996665811745,
      "grad_norm": 2.741142925639391,
      "learning_rate": 9.324569769059096e-07,
      "loss": 1.3082,
      "step": 188
    },
    {
      "epoch": 0.19389587073608616,
      "grad_norm": 2.6064903500939747,
      "learning_rate": 9.316193989577843e-07,
      "loss": 1.2599,
      "step": 189
    },
    {
      "epoch": 0.1949217748140549,
      "grad_norm": 2.6477370525921966,
      "learning_rate": 9.30777040696903e-07,
      "loss": 1.2219,
      "step": 190
    },
    {
      "epoch": 0.1959476788920236,
      "grad_norm": 2.758046928790272,
      "learning_rate": 9.299299114526334e-07,
      "loss": 1.2352,
      "step": 191
    },
    {
      "epoch": 0.19697358296999232,
      "grad_norm": 2.6839436039460907,
      "learning_rate": 9.29078020607183e-07,
      "loss": 1.2706,
      "step": 192
    },
    {
      "epoch": 0.19799948704796103,
      "grad_norm": 2.4457645270891444,
      "learning_rate": 9.28221377595495e-07,
      "loss": 1.2831,
      "step": 193
    },
    {
      "epoch": 0.19902539112592973,
      "grad_norm": 2.547455908474382,
      "learning_rate": 9.273599919051452e-07,
      "loss": 1.3182,
      "step": 194
    },
    {
      "epoch": 0.20005129520389844,
      "grad_norm": 2.777434855703004,
      "learning_rate": 9.264938730762348e-07,
      "loss": 1.3255,
      "step": 195
    },
    {
      "epoch": 0.20107719928186715,
      "grad_norm": 2.615909597544814,
      "learning_rate": 9.256230307012869e-07,
      "loss": 1.1474,
      "step": 196
    },
    {
      "epoch": 0.20210310335983586,
      "grad_norm": 3.06791169964388,
      "learning_rate": 9.247474744251387e-07,
      "loss": 1.2645,
      "step": 197
    },
    {
      "epoch": 0.20312900743780457,
      "grad_norm": 2.6202584863601976,
      "learning_rate": 9.238672139448353e-07,
      "loss": 1.275,
      "step": 198
    },
    {
      "epoch": 0.20415491151577328,
      "grad_norm": 2.601162569382904,
      "learning_rate": 9.229822590095229e-07,
      "loss": 1.274,
      "step": 199
    },
    {
      "epoch": 0.205180815593742,
      "grad_norm": 2.6008093073665877,
      "learning_rate": 9.220926194203392e-07,
      "loss": 1.3114,
      "step": 200
    },
    {
      "epoch": 0.2062067196717107,
      "grad_norm": 2.7639300809591765,
      "learning_rate": 9.211983050303065e-07,
      "loss": 1.3243,
      "step": 201
    },
    {
      "epoch": 0.2072326237496794,
      "grad_norm": 2.414680199787791,
      "learning_rate": 9.202993257442216e-07,
      "loss": 1.2389,
      "step": 202
    },
    {
      "epoch": 0.20825852782764812,
      "grad_norm": 2.6254293412284158,
      "learning_rate": 9.193956915185465e-07,
      "loss": 1.2058,
      "step": 203
    },
    {
      "epoch": 0.20928443190561682,
      "grad_norm": 2.448512350031076,
      "learning_rate": 9.184874123612981e-07,
      "loss": 1.2387,
      "step": 204
    },
    {
      "epoch": 0.21031033598358553,
      "grad_norm": 2.5690470438213406,
      "learning_rate": 9.175744983319373e-07,
      "loss": 1.26,
      "step": 205
    },
    {
      "epoch": 0.21133624006155424,
      "grad_norm": 2.754390048080123,
      "learning_rate": 9.166569595412574e-07,
      "loss": 1.1595,
      "step": 206
    },
    {
      "epoch": 0.21236214413952295,
      "grad_norm": 2.774729015636808,
      "learning_rate": 9.157348061512726e-07,
      "loss": 1.2775,
      "step": 207
    },
    {
      "epoch": 0.21338804821749166,
      "grad_norm": 2.7682686071100853,
      "learning_rate": 9.148080483751048e-07,
      "loss": 1.3079,
      "step": 208
    },
    {
      "epoch": 0.21441395229546037,
      "grad_norm": 2.782086823919593,
      "learning_rate": 9.138766964768711e-07,
      "loss": 1.2042,
      "step": 209
    },
    {
      "epoch": 0.21543985637342908,
      "grad_norm": 2.6465176185194674,
      "learning_rate": 9.129407607715696e-07,
      "loss": 1.3243,
      "step": 210
    },
    {
      "epoch": 0.2164657604513978,
      "grad_norm": 2.435336652939164,
      "learning_rate": 9.12000251624966e-07,
      "loss": 1.2992,
      "step": 211
    },
    {
      "epoch": 0.2174916645293665,
      "grad_norm": 2.598474192950846,
      "learning_rate": 9.110551794534775e-07,
      "loss": 1.2117,
      "step": 212
    },
    {
      "epoch": 0.2185175686073352,
      "grad_norm": 2.6249002298839756,
      "learning_rate": 9.101055547240586e-07,
      "loss": 1.305,
      "step": 213
    },
    {
      "epoch": 0.21954347268530391,
      "grad_norm": 2.82690930390355,
      "learning_rate": 9.091513879540844e-07,
      "loss": 1.3207,
      "step": 214
    },
    {
      "epoch": 0.22056937676327262,
      "grad_norm": 2.436101223347268,
      "learning_rate": 9.08192689711235e-07,
      "loss": 1.1575,
      "step": 215
    },
    {
      "epoch": 0.22159528084124133,
      "grad_norm": 2.3958033529310088,
      "learning_rate": 9.072294706133774e-07,
      "loss": 1.244,
      "step": 216
    },
    {
      "epoch": 0.22262118491921004,
      "grad_norm": 2.6527085410686078,
      "learning_rate": 9.062617413284485e-07,
      "loss": 1.172,
      "step": 217
    },
    {
      "epoch": 0.22364708899717875,
      "grad_norm": 2.648011195059752,
      "learning_rate": 9.052895125743369e-07,
      "loss": 1.3308,
      "step": 218
    },
    {
      "epoch": 0.2246729930751475,
      "grad_norm": 2.568920128652391,
      "learning_rate": 9.043127951187642e-07,
      "loss": 1.1837,
      "step": 219
    },
    {
      "epoch": 0.2256988971531162,
      "grad_norm": 2.5261344724437045,
      "learning_rate": 9.033315997791659e-07,
      "loss": 1.3071,
      "step": 220
    },
    {
      "epoch": 0.2267248012310849,
      "grad_norm": 2.482972261372125,
      "learning_rate": 9.023459374225708e-07,
      "loss": 1.2766,
      "step": 221
    },
    {
      "epoch": 0.2277507053090536,
      "grad_norm": 2.728013709536128,
      "learning_rate": 9.013558189654817e-07,
      "loss": 1.2879,
      "step": 222
    },
    {
      "epoch": 0.22877660938702232,
      "grad_norm": 2.509526243928691,
      "learning_rate": 9.003612553737543e-07,
      "loss": 1.2326,
      "step": 223
    },
    {
      "epoch": 0.22980251346499103,
      "grad_norm": 2.600184311388481,
      "learning_rate": 8.993622576624746e-07,
      "loss": 1.2468,
      "step": 224
    },
    {
      "epoch": 0.23082841754295974,
      "grad_norm": 2.996100810613903,
      "learning_rate": 8.983588368958387e-07,
      "loss": 1.2802,
      "step": 225
    },
    {
      "epoch": 0.23185432162092845,
      "grad_norm": 3.145733520640035,
      "learning_rate": 8.973510041870287e-07,
      "loss": 1.3561,
      "step": 226
    },
    {
      "epoch": 0.23288022569889716,
      "grad_norm": 2.677928117781213,
      "learning_rate": 8.963387706980907e-07,
      "loss": 1.2575,
      "step": 227
    },
    {
      "epoch": 0.23390612977686587,
      "grad_norm": 2.3983433304776542,
      "learning_rate": 8.953221476398105e-07,
      "loss": 1.3267,
      "step": 228
    },
    {
      "epoch": 0.23493203385483458,
      "grad_norm": 2.5880313443096146,
      "learning_rate": 8.943011462715897e-07,
      "loss": 1.3981,
      "step": 229
    },
    {
      "epoch": 0.23595793793280329,
      "grad_norm": 2.8392528250107882,
      "learning_rate": 8.932757779013213e-07,
      "loss": 1.2559,
      "step": 230
    },
    {
      "epoch": 0.236983842010772,
      "grad_norm": 2.5686226393466525,
      "learning_rate": 8.922460538852634e-07,
      "loss": 1.2202,
      "step": 231
    },
    {
      "epoch": 0.2380097460887407,
      "grad_norm": 2.509078780435471,
      "learning_rate": 8.912119856279149e-07,
      "loss": 1.1649,
      "step": 232
    },
    {
      "epoch": 0.2390356501667094,
      "grad_norm": 2.8313736339242257,
      "learning_rate": 8.901735845818884e-07,
      "loss": 1.303,
      "step": 233
    },
    {
      "epoch": 0.24006155424467812,
      "grad_norm": 2.63535412753487,
      "learning_rate": 8.891308622477829e-07,
      "loss": 1.2393,
      "step": 234
    },
    {
      "epoch": 0.24108745832264683,
      "grad_norm": 2.357470458300626,
      "learning_rate": 8.880838301740575e-07,
      "loss": 1.2319,
      "step": 235
    },
    {
      "epoch": 0.24211336240061554,
      "grad_norm": 2.633624233788043,
      "learning_rate": 8.870324999569024e-07,
      "loss": 1.3239,
      "step": 236
    },
    {
      "epoch": 0.24313926647858425,
      "grad_norm": 2.541409949570618,
      "learning_rate": 8.859768832401117e-07,
      "loss": 1.2875,
      "step": 237
    },
    {
      "epoch": 0.24416517055655296,
      "grad_norm": 2.7389024190179123,
      "learning_rate": 8.849169917149531e-07,
      "loss": 1.2948,
      "step": 238
    },
    {
      "epoch": 0.24519107463452167,
      "grad_norm": 2.7594743544624536,
      "learning_rate": 8.838528371200394e-07,
      "loss": 1.3199,
      "step": 239
    },
    {
      "epoch": 0.24621697871249038,
      "grad_norm": 2.553894247097559,
      "learning_rate": 8.827844312411982e-07,
      "loss": 1.3282,
      "step": 240
    },
    {
      "epoch": 0.24724288279045908,
      "grad_norm": 2.41692947947688,
      "learning_rate": 8.817117859113412e-07,
      "loss": 1.242,
      "step": 241
    },
    {
      "epoch": 0.2482687868684278,
      "grad_norm": 2.723064556658783,
      "learning_rate": 8.806349130103332e-07,
      "loss": 1.1648,
      "step": 242
    },
    {
      "epoch": 0.2492946909463965,
      "grad_norm": 2.643873563364823,
      "learning_rate": 8.795538244648609e-07,
      "loss": 1.2753,
      "step": 243
    },
    {
      "epoch": 0.2503205950243652,
      "grad_norm": 2.6856636444384385,
      "learning_rate": 8.784685322483003e-07,
      "loss": 1.2634,
      "step": 244
    },
    {
      "epoch": 0.2513464991023339,
      "grad_norm": 2.960535082783671,
      "learning_rate": 8.77379048380584e-07,
      "loss": 1.1963,
      "step": 245
    },
    {
      "epoch": 0.25237240318030263,
      "grad_norm": 3.2907999278170066,
      "learning_rate": 8.762853849280691e-07,
      "loss": 1.2236,
      "step": 246
    },
    {
      "epoch": 0.25339830725827134,
      "grad_norm": 2.706088121459989,
      "learning_rate": 8.751875540034025e-07,
      "loss": 1.2284,
      "step": 247
    },
    {
      "epoch": 0.25442421133624005,
      "grad_norm": 2.8342758112128656,
      "learning_rate": 8.740855677653867e-07,
      "loss": 1.3529,
      "step": 248
    },
    {
      "epoch": 0.25545011541420876,
      "grad_norm": 2.657409992434949,
      "learning_rate": 8.72979438418846e-07,
      "loss": 1.2265,
      "step": 249
    },
    {
      "epoch": 0.25647601949217746,
      "grad_norm": 2.733820536997817,
      "learning_rate": 8.718691782144907e-07,
      "loss": 1.3085,
      "step": 250
    },
    {
      "epoch": 0.25647601949217746,
      "eval_uground_MCTS_chains_SFT_val_loss": 1.2850571870803833,
      "eval_uground_MCTS_chains_SFT_val_runtime": 234.9849,
      "eval_uground_MCTS_chains_SFT_val_samples_per_second": 7.741,
      "eval_uground_MCTS_chains_SFT_val_steps_per_second": 0.97,
      "step": 250
    },
    {
      "epoch": 0.2575019235701462,
      "grad_norm": 2.6848225964360886,
      "learning_rate": 8.707547994487817e-07,
      "loss": 1.2179,
      "step": 251
    },
    {
      "epoch": 0.2585278276481149,
      "grad_norm": 2.6683129178340717,
      "learning_rate": 8.69636314463794e-07,
      "loss": 1.2185,
      "step": 252
    },
    {
      "epoch": 0.2595537317260836,
      "grad_norm": 2.7296939230032047,
      "learning_rate": 8.685137356470802e-07,
      "loss": 1.2974,
      "step": 253
    },
    {
      "epoch": 0.2605796358040523,
      "grad_norm": 2.5819217942943387,
      "learning_rate": 8.673870754315336e-07,
      "loss": 1.2832,
      "step": 254
    },
    {
      "epoch": 0.261605539882021,
      "grad_norm": 2.7286375807681202,
      "learning_rate": 8.662563462952498e-07,
      "loss": 1.1838,
      "step": 255
    },
    {
      "epoch": 0.2626314439599897,
      "grad_norm": 2.8389791464688825,
      "learning_rate": 8.651215607613891e-07,
      "loss": 1.2626,
      "step": 256
    },
    {
      "epoch": 0.26365734803795843,
      "grad_norm": 2.667042240854244,
      "learning_rate": 8.639827313980377e-07,
      "loss": 1.2649,
      "step": 257
    },
    {
      "epoch": 0.26468325211592714,
      "grad_norm": 2.6062798312002853,
      "learning_rate": 8.628398708180679e-07,
      "loss": 1.2032,
      "step": 258
    },
    {
      "epoch": 0.26570915619389585,
      "grad_norm": 2.8969301485628893,
      "learning_rate": 8.61692991679e-07,
      "loss": 1.2119,
      "step": 259
    },
    {
      "epoch": 0.26673506027186455,
      "grad_norm": 2.6804950419511164,
      "learning_rate": 8.605421066828598e-07,
      "loss": 1.2812,
      "step": 260
    },
    {
      "epoch": 0.26776096434983326,
      "grad_norm": 2.8172486497816807,
      "learning_rate": 8.593872285760399e-07,
      "loss": 1.2254,
      "step": 261
    },
    {
      "epoch": 0.26878686842780203,
      "grad_norm": 2.5633187420692307,
      "learning_rate": 8.582283701491575e-07,
      "loss": 1.2842,
      "step": 262
    },
    {
      "epoch": 0.26981277250577074,
      "grad_norm": 2.67558705929831,
      "learning_rate": 8.570655442369133e-07,
      "loss": 1.2636,
      "step": 263
    },
    {
      "epoch": 0.27083867658373945,
      "grad_norm": 2.8468850830803425,
      "learning_rate": 8.558987637179487e-07,
      "loss": 1.214,
      "step": 264
    },
    {
      "epoch": 0.27186458066170816,
      "grad_norm": 2.6831972490521965,
      "learning_rate": 8.547280415147037e-07,
      "loss": 1.2931,
      "step": 265
    },
    {
      "epoch": 0.27289048473967686,
      "grad_norm": 2.6462503105531896,
      "learning_rate": 8.535533905932737e-07,
      "loss": 1.2655,
      "step": 266
    },
    {
      "epoch": 0.2739163888176456,
      "grad_norm": 2.3586365587982967,
      "learning_rate": 8.523748239632659e-07,
      "loss": 1.2476,
      "step": 267
    },
    {
      "epoch": 0.2749422928956143,
      "grad_norm": 2.54829905356015,
      "learning_rate": 8.51192354677655e-07,
      "loss": 1.1603,
      "step": 268
    },
    {
      "epoch": 0.275968196973583,
      "grad_norm": 2.5266352307165705,
      "learning_rate": 8.500059958326384e-07,
      "loss": 1.2981,
      "step": 269
    },
    {
      "epoch": 0.2769941010515517,
      "grad_norm": 2.6922548930973167,
      "learning_rate": 8.488157605674924e-07,
      "loss": 1.2659,
      "step": 270
    },
    {
      "epoch": 0.2780200051295204,
      "grad_norm": 2.664526394721468,
      "learning_rate": 8.47621662064425e-07,
      "loss": 1.2074,
      "step": 271
    },
    {
      "epoch": 0.2790459092074891,
      "grad_norm": 2.6008818489590477,
      "learning_rate": 8.464237135484309e-07,
      "loss": 1.312,
      "step": 272
    },
    {
      "epoch": 0.2800718132854578,
      "grad_norm": 2.5513968093248733,
      "learning_rate": 8.452219282871451e-07,
      "loss": 1.2602,
      "step": 273
    },
    {
      "epoch": 0.28109771736342654,
      "grad_norm": 2.645895710122177,
      "learning_rate": 8.440163195906958e-07,
      "loss": 1.2199,
      "step": 274
    },
    {
      "epoch": 0.28212362144139524,
      "grad_norm": 2.555562047484234,
      "learning_rate": 8.42806900811556e-07,
      "loss": 1.2419,
      "step": 275
    },
    {
      "epoch": 0.28314952551936395,
      "grad_norm": 2.668319098287759,
      "learning_rate": 8.415936853443974e-07,
      "loss": 1.2866,
      "step": 276
    },
    {
      "epoch": 0.28417542959733266,
      "grad_norm": 2.5876812386597856,
      "learning_rate": 8.40376686625941e-07,
      "loss": 1.2495,
      "step": 277
    },
    {
      "epoch": 0.28520133367530137,
      "grad_norm": 2.761080694836167,
      "learning_rate": 8.391559181348081e-07,
      "loss": 1.3327,
      "step": 278
    },
    {
      "epoch": 0.2862272377532701,
      "grad_norm": 2.5148075285885323,
      "learning_rate": 8.379313933913714e-07,
      "loss": 1.2356,
      "step": 279
    },
    {
      "epoch": 0.2872531418312388,
      "grad_norm": 2.8048853074405913,
      "learning_rate": 8.367031259576056e-07,
      "loss": 1.3033,
      "step": 280
    },
    {
      "epoch": 0.2882790459092075,
      "grad_norm": 2.5855751773469,
      "learning_rate": 8.354711294369362e-07,
      "loss": 1.2116,
      "step": 281
    },
    {
      "epoch": 0.2893049499871762,
      "grad_norm": 2.6409683165441833,
      "learning_rate": 8.342354174740902e-07,
      "loss": 1.291,
      "step": 282
    },
    {
      "epoch": 0.2903308540651449,
      "grad_norm": 2.5609773198767583,
      "learning_rate": 8.329960037549433e-07,
      "loss": 1.1966,
      "step": 283
    },
    {
      "epoch": 0.2913567581431136,
      "grad_norm": 2.454732765461205,
      "learning_rate": 8.317529020063703e-07,
      "loss": 1.2519,
      "step": 284
    },
    {
      "epoch": 0.29238266222108233,
      "grad_norm": 2.5134417168975127,
      "learning_rate": 8.305061259960909e-07,
      "loss": 1.3021,
      "step": 285
    },
    {
      "epoch": 0.29340856629905104,
      "grad_norm": 2.7770733269324683,
      "learning_rate": 8.292556895325194e-07,
      "loss": 1.2724,
      "step": 286
    },
    {
      "epoch": 0.29443447037701975,
      "grad_norm": 2.7282701405098226,
      "learning_rate": 8.280016064646098e-07,
      "loss": 1.3078,
      "step": 287
    },
    {
      "epoch": 0.29546037445498846,
      "grad_norm": 2.9061694081822815,
      "learning_rate": 8.267438906817039e-07,
      "loss": 1.3072,
      "step": 288
    },
    {
      "epoch": 0.29648627853295717,
      "grad_norm": 2.789078860545301,
      "learning_rate": 8.25482556113377e-07,
      "loss": 1.2562,
      "step": 289
    },
    {
      "epoch": 0.2975121826109259,
      "grad_norm": 2.5716595982006147,
      "learning_rate": 8.242176167292826e-07,
      "loss": 1.242,
      "step": 290
    },
    {
      "epoch": 0.2985380866888946,
      "grad_norm": 2.639971746858166,
      "learning_rate": 8.229490865389998e-07,
      "loss": 1.2395,
      "step": 291
    },
    {
      "epoch": 0.2995639907668633,
      "grad_norm": 2.478159221268824,
      "learning_rate": 8.216769795918762e-07,
      "loss": 1.1481,
      "step": 292
    },
    {
      "epoch": 0.300589894844832,
      "grad_norm": 2.5242932402585376,
      "learning_rate": 8.204013099768732e-07,
      "loss": 1.2662,
      "step": 293
    },
    {
      "epoch": 0.3016157989228007,
      "grad_norm": 2.598195840538781,
      "learning_rate": 8.1912209182241e-07,
      "loss": 1.2947,
      "step": 294
    },
    {
      "epoch": 0.3026417030007694,
      "grad_norm": 2.7173961511536264,
      "learning_rate": 8.178393392962066e-07,
      "loss": 1.2896,
      "step": 295
    },
    {
      "epoch": 0.30366760707873813,
      "grad_norm": 2.7206884235512527,
      "learning_rate": 8.165530666051275e-07,
      "loss": 1.2786,
      "step": 296
    },
    {
      "epoch": 0.30469351115670684,
      "grad_norm": 2.7552783724459706,
      "learning_rate": 8.152632879950238e-07,
      "loss": 1.2365,
      "step": 297
    },
    {
      "epoch": 0.30571941523467555,
      "grad_norm": 2.6147443498413954,
      "learning_rate": 8.139700177505759e-07,
      "loss": 1.2179,
      "step": 298
    },
    {
      "epoch": 0.30674531931264426,
      "grad_norm": 2.606663274306208,
      "learning_rate": 8.126732701951351e-07,
      "loss": 1.3034,
      "step": 299
    },
    {
      "epoch": 0.30777122339061297,
      "grad_norm": 2.648199256292807,
      "learning_rate": 8.113730596905648e-07,
      "loss": 1.1252,
      "step": 300
    },
    {
      "epoch": 0.3087971274685817,
      "grad_norm": 2.4523122279421417,
      "learning_rate": 8.100694006370816e-07,
      "loss": 1.2627,
      "step": 301
    },
    {
      "epoch": 0.3098230315465504,
      "grad_norm": 2.966417644370533,
      "learning_rate": 8.087623074730959e-07,
      "loss": 1.2693,
      "step": 302
    },
    {
      "epoch": 0.3108489356245191,
      "grad_norm": 2.7468880750816647,
      "learning_rate": 8.07451794675052e-07,
      "loss": 1.2192,
      "step": 303
    },
    {
      "epoch": 0.3118748397024878,
      "grad_norm": 2.7445599934088327,
      "learning_rate": 8.061378767572673e-07,
      "loss": 1.2482,
      "step": 304
    },
    {
      "epoch": 0.3129007437804565,
      "grad_norm": 2.377263362356011,
      "learning_rate": 8.048205682717724e-07,
      "loss": 1.22,
      "step": 305
    },
    {
      "epoch": 0.3139266478584252,
      "grad_norm": 2.682567930380336,
      "learning_rate": 8.034998838081489e-07,
      "loss": 1.3312,
      "step": 306
    },
    {
      "epoch": 0.31495255193639393,
      "grad_norm": 2.6542762724022717,
      "learning_rate": 8.021758379933686e-07,
      "loss": 1.2426,
      "step": 307
    },
    {
      "epoch": 0.31597845601436264,
      "grad_norm": 2.691345504635165,
      "learning_rate": 8.008484454916316e-07,
      "loss": 1.195,
      "step": 308
    },
    {
      "epoch": 0.31700436009233135,
      "grad_norm": 2.490433627351295,
      "learning_rate": 7.995177210042028e-07,
      "loss": 1.1885,
      "step": 309
    },
    {
      "epoch": 0.31803026417030006,
      "grad_norm": 2.6751579195821242,
      "learning_rate": 7.981836792692507e-07,
      "loss": 1.201,
      "step": 310
    },
    {
      "epoch": 0.31905616824826877,
      "grad_norm": 2.670875208828667,
      "learning_rate": 7.968463350616825e-07,
      "loss": 1.2289,
      "step": 311
    },
    {
      "epoch": 0.3200820723262375,
      "grad_norm": 2.745027291047488,
      "learning_rate": 7.955057031929819e-07,
      "loss": 1.2896,
      "step": 312
    },
    {
      "epoch": 0.3211079764042062,
      "grad_norm": 2.561763884212081,
      "learning_rate": 7.941617985110442e-07,
      "loss": 1.3243,
      "step": 313
    },
    {
      "epoch": 0.3221338804821749,
      "grad_norm": 2.6240454275926344,
      "learning_rate": 7.928146359000117e-07,
      "loss": 1.3075,
      "step": 314
    },
    {
      "epoch": 0.3231597845601436,
      "grad_norm": 2.6273454218750434,
      "learning_rate": 7.914642302801097e-07,
      "loss": 1.3616,
      "step": 315
    },
    {
      "epoch": 0.3241856886381123,
      "grad_norm": 2.563951932815581,
      "learning_rate": 7.901105966074806e-07,
      "loss": 1.2714,
      "step": 316
    },
    {
      "epoch": 0.325211592716081,
      "grad_norm": 2.5822875744818563,
      "learning_rate": 7.887537498740187e-07,
      "loss": 1.3283,
      "step": 317
    },
    {
      "epoch": 0.32623749679404973,
      "grad_norm": 2.6143885876617627,
      "learning_rate": 7.873937051072035e-07,
      "loss": 1.253,
      "step": 318
    },
    {
      "epoch": 0.32726340087201844,
      "grad_norm": 2.647031579928053,
      "learning_rate": 7.860304773699338e-07,
      "loss": 1.3734,
      "step": 319
    },
    {
      "epoch": 0.32828930494998715,
      "grad_norm": 2.6072419451698625,
      "learning_rate": 7.846640817603607e-07,
      "loss": 1.2476,
      "step": 320
    },
    {
      "epoch": 0.3293152090279559,
      "grad_norm": 2.6913716325413737,
      "learning_rate": 7.83294533411721e-07,
      "loss": 1.2436,
      "step": 321
    },
    {
      "epoch": 0.3303411131059246,
      "grad_norm": 2.5669118480954913,
      "learning_rate": 7.819218474921679e-07,
      "loss": 1.2327,
      "step": 322
    },
    {
      "epoch": 0.33136701718389333,
      "grad_norm": 2.4680065253742063,
      "learning_rate": 7.805460392046053e-07,
      "loss": 1.2094,
      "step": 323
    },
    {
      "epoch": 0.33239292126186204,
      "grad_norm": 2.791361144884489,
      "learning_rate": 7.791671237865174e-07,
      "loss": 1.2787,
      "step": 324
    },
    {
      "epoch": 0.33341882533983075,
      "grad_norm": 2.3530874149345355,
      "learning_rate": 7.777851165098011e-07,
      "loss": 1.2706,
      "step": 325
    },
    {
      "epoch": 0.33444472941779946,
      "grad_norm": 2.7864142928485442,
      "learning_rate": 7.764000326805966e-07,
      "loss": 1.2568,
      "step": 326
    },
    {
      "epoch": 0.33547063349576817,
      "grad_norm": 2.744024330823426,
      "learning_rate": 7.75011887639118e-07,
      "loss": 1.1573,
      "step": 327
    },
    {
      "epoch": 0.3364965375737369,
      "grad_norm": 2.4274620710022696,
      "learning_rate": 7.736206967594827e-07,
      "loss": 1.2552,
      "step": 328
    },
    {
      "epoch": 0.3375224416517056,
      "grad_norm": 2.433652617983981,
      "learning_rate": 7.722264754495421e-07,
      "loss": 1.1821,
      "step": 329
    },
    {
      "epoch": 0.3385483457296743,
      "grad_norm": 2.6471120476788963,
      "learning_rate": 7.708292391507105e-07,
      "loss": 1.2633,
      "step": 330
    },
    {
      "epoch": 0.339574249807643,
      "grad_norm": 2.4674613719486542,
      "learning_rate": 7.694290033377938e-07,
      "loss": 1.1524,
      "step": 331
    },
    {
      "epoch": 0.3406001538856117,
      "grad_norm": 2.7466941946417003,
      "learning_rate": 7.680257835188186e-07,
      "loss": 1.2265,
      "step": 332
    },
    {
      "epoch": 0.3416260579635804,
      "grad_norm": 2.5000486774159514,
      "learning_rate": 7.666195952348606e-07,
      "loss": 1.2611,
      "step": 333
    },
    {
      "epoch": 0.34265196204154913,
      "grad_norm": 2.567371463839335,
      "learning_rate": 7.652104540598712e-07,
      "loss": 1.2303,
      "step": 334
    },
    {
      "epoch": 0.34367786611951784,
      "grad_norm": 2.5729629715661266,
      "learning_rate": 7.63798375600507e-07,
      "loss": 1.2566,
      "step": 335
    },
    {
      "epoch": 0.34470377019748655,
      "grad_norm": 2.647727174139296,
      "learning_rate": 7.623833754959551e-07,
      "loss": 1.2118,
      "step": 336
    },
    {
      "epoch": 0.34572967427545526,
      "grad_norm": 2.8141439244256894,
      "learning_rate": 7.609654694177612e-07,
      "loss": 1.2393,
      "step": 337
    },
    {
      "epoch": 0.34675557835342397,
      "grad_norm": 2.5047983163137335,
      "learning_rate": 7.595446730696553e-07,
      "loss": 1.2848,
      "step": 338
    },
    {
      "epoch": 0.3477814824313927,
      "grad_norm": 2.8575885947114954,
      "learning_rate": 7.581210021873778e-07,
      "loss": 1.2739,
      "step": 339
    },
    {
      "epoch": 0.3488073865093614,
      "grad_norm": 2.5879263426142303,
      "learning_rate": 7.56694472538506e-07,
      "loss": 1.2399,
      "step": 340
    },
    {
      "epoch": 0.3498332905873301,
      "grad_norm": 2.634462986237627,
      "learning_rate": 7.552650999222783e-07,
      "loss": 1.1886,
      "step": 341
    },
    {
      "epoch": 0.3508591946652988,
      "grad_norm": 2.644196413288821,
      "learning_rate": 7.538329001694199e-07,
      "loss": 1.189,
      "step": 342
    },
    {
      "epoch": 0.3518850987432675,
      "grad_norm": 2.46112345657162,
      "learning_rate": 7.523978891419678e-07,
      "loss": 1.3348,
      "step": 343
    },
    {
      "epoch": 0.3529110028212362,
      "grad_norm": 2.749240231393701,
      "learning_rate": 7.509600827330942e-07,
      "loss": 1.3394,
      "step": 344
    },
    {
      "epoch": 0.35393690689920493,
      "grad_norm": 2.690306971315971,
      "learning_rate": 7.495194968669311e-07,
      "loss": 1.1576,
      "step": 345
    },
    {
      "epoch": 0.35496281097717364,
      "grad_norm": 2.539756877499731,
      "learning_rate": 7.480761474983943e-07,
      "loss": 1.2425,
      "step": 346
    },
    {
      "epoch": 0.35598871505514235,
      "grad_norm": 2.794783648579977,
      "learning_rate": 7.466300506130052e-07,
      "loss": 1.2182,
      "step": 347
    },
    {
      "epoch": 0.35701461913311106,
      "grad_norm": 2.824643687827716,
      "learning_rate": 7.451812222267157e-07,
      "loss": 1.218,
      "step": 348
    },
    {
      "epoch": 0.35804052321107976,
      "grad_norm": 2.667873083406682,
      "learning_rate": 7.437296783857296e-07,
      "loss": 1.1692,
      "step": 349
    },
    {
      "epoch": 0.3590664272890485,
      "grad_norm": 2.671390395878333,
      "learning_rate": 7.422754351663251e-07,
      "loss": 1.1613,
      "step": 350
    },
    {
      "epoch": 0.3600923313670172,
      "grad_norm": 2.8795003735438813,
      "learning_rate": 7.408185086746766e-07,
      "loss": 1.2302,
      "step": 351
    },
    {
      "epoch": 0.3611182354449859,
      "grad_norm": 2.6611306550789684,
      "learning_rate": 7.39358915046677e-07,
      "loss": 1.3027,
      "step": 352
    },
    {
      "epoch": 0.3621441395229546,
      "grad_norm": 2.6035638980444933,
      "learning_rate": 7.378966704477584e-07,
      "loss": 1.2342,
      "step": 353
    },
    {
      "epoch": 0.3631700436009233,
      "grad_norm": 2.4479505443923197,
      "learning_rate": 7.364317910727127e-07,
      "loss": 1.2259,
      "step": 354
    },
    {
      "epoch": 0.364195947678892,
      "grad_norm": 2.795523604698995,
      "learning_rate": 7.349642931455131e-07,
      "loss": 1.1781,
      "step": 355
    },
    {
      "epoch": 0.3652218517568607,
      "grad_norm": 2.5499819208366006,
      "learning_rate": 7.334941929191343e-07,
      "loss": 1.2789,
      "step": 356
    },
    {
      "epoch": 0.36624775583482944,
      "grad_norm": 2.7893689673549686,
      "learning_rate": 7.320215066753722e-07,
      "loss": 1.2962,
      "step": 357
    },
    {
      "epoch": 0.36727365991279814,
      "grad_norm": 2.644819217736545,
      "learning_rate": 7.305462507246629e-07,
      "loss": 1.2008,
      "step": 358
    },
    {
      "epoch": 0.36829956399076685,
      "grad_norm": 2.7769763671178795,
      "learning_rate": 7.290684414059034e-07,
      "loss": 1.2561,
      "step": 359
    },
    {
      "epoch": 0.36932546806873556,
      "grad_norm": 2.7516891619949715,
      "learning_rate": 7.2758809508627e-07,
      "loss": 1.1101,
      "step": 360
    },
    {
      "epoch": 0.37035137214670427,
      "grad_norm": 2.426378655817816,
      "learning_rate": 7.261052281610367e-07,
      "loss": 1.2136,
      "step": 361
    },
    {
      "epoch": 0.371377276224673,
      "grad_norm": 2.6154616984297414,
      "learning_rate": 7.246198570533944e-07,
      "loss": 1.2197,
      "step": 362
    },
    {
      "epoch": 0.3724031803026417,
      "grad_norm": 2.5360984469645467,
      "learning_rate": 7.231319982142679e-07,
      "loss": 1.1605,
      "step": 363
    },
    {
      "epoch": 0.3734290843806104,
      "grad_norm": 2.825045376287288,
      "learning_rate": 7.216416681221353e-07,
      "loss": 1.2416,
      "step": 364
    },
    {
      "epoch": 0.3744549884585791,
      "grad_norm": 2.5276250456119844,
      "learning_rate": 7.201488832828438e-07,
      "loss": 1.19,
      "step": 365
    },
    {
      "epoch": 0.3754808925365478,
      "grad_norm": 2.4928208511714627,
      "learning_rate": 7.186536602294278e-07,
      "loss": 1.2168,
      "step": 366
    },
    {
      "epoch": 0.3765067966145165,
      "grad_norm": 2.6143894885373102,
      "learning_rate": 7.171560155219256e-07,
      "loss": 1.2404,
      "step": 367
    },
    {
      "epoch": 0.37753270069248523,
      "grad_norm": 2.60017799652903,
      "learning_rate": 7.156559657471966e-07,
      "loss": 1.2826,
      "step": 368
    },
    {
      "epoch": 0.37855860477045394,
      "grad_norm": 2.6824928604628564,
      "learning_rate": 7.141535275187363e-07,
      "loss": 1.1865,
      "step": 369
    },
    {
      "epoch": 0.37958450884842265,
      "grad_norm": 2.603460949625,
      "learning_rate": 7.126487174764935e-07,
      "loss": 1.1988,
      "step": 370
    },
    {
      "epoch": 0.38061041292639136,
      "grad_norm": 2.679412906737572,
      "learning_rate": 7.11141552286685e-07,
      "loss": 1.2827,
      "step": 371
    },
    {
      "epoch": 0.38163631700436007,
      "grad_norm": 2.9611932859479055,
      "learning_rate": 7.096320486416124e-07,
      "loss": 1.2008,
      "step": 372
    },
    {
      "epoch": 0.3826622210823288,
      "grad_norm": 2.748214027279323,
      "learning_rate": 7.081202232594758e-07,
      "loss": 1.2062,
      "step": 373
    },
    {
      "epoch": 0.3836881251602975,
      "grad_norm": 2.610256542655334,
      "learning_rate": 7.06606092884189e-07,
      "loss": 1.2218,
      "step": 374
    },
    {
      "epoch": 0.3847140292382662,
      "grad_norm": 2.568752573794102,
      "learning_rate": 7.050896742851952e-07,
      "loss": 1.259,
      "step": 375
    },
    {
      "epoch": 0.3857399333162349,
      "grad_norm": 2.729994050097171,
      "learning_rate": 7.035709842572792e-07,
      "loss": 1.1736,
      "step": 376
    },
    {
      "epoch": 0.3867658373942036,
      "grad_norm": 2.865709205643247,
      "learning_rate": 7.020500396203837e-07,
      "loss": 1.2853,
      "step": 377
    },
    {
      "epoch": 0.3877917414721723,
      "grad_norm": 2.6383244403757065,
      "learning_rate": 7.005268572194207e-07,
      "loss": 1.2394,
      "step": 378
    },
    {
      "epoch": 0.3888176455501411,
      "grad_norm": 2.756577015435794,
      "learning_rate": 6.990014539240873e-07,
      "loss": 1.1993,
      "step": 379
    },
    {
      "epoch": 0.3898435496281098,
      "grad_norm": 2.5090775530327805,
      "learning_rate": 6.974738466286765e-07,
      "loss": 1.2351,
      "step": 380
    },
    {
      "epoch": 0.3908694537060785,
      "grad_norm": 2.7812691019728093,
      "learning_rate": 6.959440522518923e-07,
      "loss": 1.199,
      "step": 381
    },
    {
      "epoch": 0.3918953577840472,
      "grad_norm": 2.546674174824266,
      "learning_rate": 6.944120877366604e-07,
      "loss": 1.154,
      "step": 382
    },
    {
      "epoch": 0.3929212618620159,
      "grad_norm": 3.0015148003584557,
      "learning_rate": 6.928779700499419e-07,
      "loss": 1.2702,
      "step": 383
    },
    {
      "epoch": 0.39394716593998463,
      "grad_norm": 2.5008648231818627,
      "learning_rate": 6.913417161825449e-07,
      "loss": 1.2137,
      "step": 384
    },
    {
      "epoch": 0.39497307001795334,
      "grad_norm": 2.4588983698528772,
      "learning_rate": 6.898033431489361e-07,
      "loss": 1.1406,
      "step": 385
    },
    {
      "epoch": 0.39599897409592205,
      "grad_norm": 2.527134330865664,
      "learning_rate": 6.882628679870531e-07,
      "loss": 1.3066,
      "step": 386
    },
    {
      "epoch": 0.39702487817389076,
      "grad_norm": 2.6472686614664225,
      "learning_rate": 6.867203077581145e-07,
      "loss": 1.3113,
      "step": 387
    },
    {
      "epoch": 0.39805078225185947,
      "grad_norm": 2.764798950791151,
      "learning_rate": 6.851756795464323e-07,
      "loss": 1.2906,
      "step": 388
    },
    {
      "epoch": 0.3990766863298282,
      "grad_norm": 2.7405389624240653,
      "learning_rate": 6.836290004592213e-07,
      "loss": 1.2001,
      "step": 389
    },
    {
      "epoch": 0.4001025904077969,
      "grad_norm": 2.631861163892539,
      "learning_rate": 6.820802876264111e-07,
      "loss": 1.1641,
      "step": 390
    },
    {
      "epoch": 0.4011284944857656,
      "grad_norm": 2.6174482541499136,
      "learning_rate": 6.805295582004551e-07,
      "loss": 1.2187,
      "step": 391
    },
    {
      "epoch": 0.4021543985637343,
      "grad_norm": 2.852360371699908,
      "learning_rate": 6.789768293561413e-07,
      "loss": 1.308,
      "step": 392
    },
    {
      "epoch": 0.403180302641703,
      "grad_norm": 2.5465668197579743,
      "learning_rate": 6.774221182904017e-07,
      "loss": 1.2074,
      "step": 393
    },
    {
      "epoch": 0.4042062067196717,
      "grad_norm": 2.864528470906807,
      "learning_rate": 6.758654422221224e-07,
      "loss": 1.1637,
      "step": 394
    },
    {
      "epoch": 0.40523211079764043,
      "grad_norm": 2.8379197072181612,
      "learning_rate": 6.743068183919519e-07,
      "loss": 1.222,
      "step": 395
    },
    {
      "epoch": 0.40625801487560914,
      "grad_norm": 2.62449855028073,
      "learning_rate": 6.727462640621112e-07,
      "loss": 1.279,
      "step": 396
    },
    {
      "epoch": 0.40728391895357785,
      "grad_norm": 2.5226829698942397,
      "learning_rate": 6.711837965162019e-07,
      "loss": 1.2924,
      "step": 397
    },
    {
      "epoch": 0.40830982303154656,
      "grad_norm": 2.6680293582268715,
      "learning_rate": 6.69619433059015e-07,
      "loss": 1.203,
      "step": 398
    },
    {
      "epoch": 0.40933572710951527,
      "grad_norm": 2.6267175358172272,
      "learning_rate": 6.680531910163398e-07,
      "loss": 1.2044,
      "step": 399
    },
    {
      "epoch": 0.410361631187484,
      "grad_norm": 2.798539704369877,
      "learning_rate": 6.664850877347705e-07,
      "loss": 1.2283,
      "step": 400
    },
    {
      "epoch": 0.4113875352654527,
      "grad_norm": 2.706328663367393,
      "learning_rate": 6.649151405815161e-07,
      "loss": 1.2072,
      "step": 401
    },
    {
      "epoch": 0.4124134393434214,
      "grad_norm": 2.5865368220778735,
      "learning_rate": 6.633433669442064e-07,
      "loss": 1.2156,
      "step": 402
    },
    {
      "epoch": 0.4134393434213901,
      "grad_norm": 2.57901369585107,
      "learning_rate": 6.617697842307004e-07,
      "loss": 1.1958,
      "step": 403
    },
    {
      "epoch": 0.4144652474993588,
      "grad_norm": 2.7016388144398267,
      "learning_rate": 6.601944098688927e-07,
      "loss": 1.195,
      "step": 404
    },
    {
      "epoch": 0.4154911515773275,
      "grad_norm": 2.556707832936113,
      "learning_rate": 6.586172613065215e-07,
      "loss": 1.1804,
      "step": 405
    },
    {
      "epoch": 0.41651705565529623,
      "grad_norm": 2.636733912969758,
      "learning_rate": 6.570383560109745e-07,
      "loss": 1.2457,
      "step": 406
    },
    {
      "epoch": 0.41754295973326494,
      "grad_norm": 2.807001379119985,
      "learning_rate": 6.554577114690955e-07,
      "loss": 1.3283,
      "step": 407
    },
    {
      "epoch": 0.41856886381123365,
      "grad_norm": 2.5257169644640998,
      "learning_rate": 6.538753451869913e-07,
      "loss": 1.2884,
      "step": 408
    },
    {
      "epoch": 0.41959476788920236,
      "grad_norm": 2.5327516189035206,
      "learning_rate": 6.522912746898379e-07,
      "loss": 1.2235,
      "step": 409
    },
    {
      "epoch": 0.42062067196717107,
      "grad_norm": 2.90569340656041,
      "learning_rate": 6.507055175216849e-07,
      "loss": 1.1792,
      "step": 410
    },
    {
      "epoch": 0.4216465760451398,
      "grad_norm": 2.5114377824774317,
      "learning_rate": 6.491180912452631e-07,
      "loss": 1.3252,
      "step": 411
    },
    {
      "epoch": 0.4226724801231085,
      "grad_norm": 2.8201576436584963,
      "learning_rate": 6.475290134417891e-07,
      "loss": 1.2259,
      "step": 412
    },
    {
      "epoch": 0.4236983842010772,
      "grad_norm": 2.5655086880724065,
      "learning_rate": 6.459383017107703e-07,
      "loss": 1.2522,
      "step": 413
    },
    {
      "epoch": 0.4247242882790459,
      "grad_norm": 2.7241886166530165,
      "learning_rate": 6.443459736698105e-07,
      "loss": 1.1883,
      "step": 414
    },
    {
      "epoch": 0.4257501923570146,
      "grad_norm": 2.777112809092128,
      "learning_rate": 6.427520469544147e-07,
      "loss": 1.2589,
      "step": 415
    },
    {
      "epoch": 0.4267760964349833,
      "grad_norm": 2.5197028770911274,
      "learning_rate": 6.41156539217794e-07,
      "loss": 1.1807,
      "step": 416
    },
    {
      "epoch": 0.42780200051295203,
      "grad_norm": 2.7452576494954166,
      "learning_rate": 6.395594681306688e-07,
      "loss": 1.1969,
      "step": 417
    },
    {
      "epoch": 0.42882790459092074,
      "grad_norm": 2.419748401721261,
      "learning_rate": 6.379608513810753e-07,
      "loss": 1.2156,
      "step": 418
    },
    {
      "epoch": 0.42985380866888945,
      "grad_norm": 2.6007536361681347,
      "learning_rate": 6.363607066741672e-07,
      "loss": 1.1682,
      "step": 419
    },
    {
      "epoch": 0.43087971274685816,
      "grad_norm": 2.7690218693992614,
      "learning_rate": 6.347590517320217e-07,
      "loss": 1.2596,
      "step": 420
    },
    {
      "epoch": 0.43190561682482687,
      "grad_norm": 2.8021339088053887,
      "learning_rate": 6.331559042934418e-07,
      "loss": 1.2384,
      "step": 421
    },
    {
      "epoch": 0.4329315209027956,
      "grad_norm": 2.5407498593535345,
      "learning_rate": 6.315512821137606e-07,
      "loss": 1.1624,
      "step": 422
    },
    {
      "epoch": 0.4339574249807643,
      "grad_norm": 2.669181018356921,
      "learning_rate": 6.299452029646442e-07,
      "loss": 1.1831,
      "step": 423
    },
    {
      "epoch": 0.434983329058733,
      "grad_norm": 2.6902513257905567,
      "learning_rate": 6.28337684633895e-07,
      "loss": 1.1293,
      "step": 424
    },
    {
      "epoch": 0.4360092331367017,
      "grad_norm": 2.600350864738873,
      "learning_rate": 6.267287449252552e-07,
      "loss": 1.1606,
      "step": 425
    },
    {
      "epoch": 0.4370351372146704,
      "grad_norm": 2.646078066207826,
      "learning_rate": 6.251184016582088e-07,
      "loss": 1.2734,
      "step": 426
    },
    {
      "epoch": 0.4380610412926391,
      "grad_norm": 2.7517987447445904,
      "learning_rate": 6.235066726677845e-07,
      "loss": 1.2104,
      "step": 427
    },
    {
      "epoch": 0.43908694537060783,
      "grad_norm": 2.7451282679247764,
      "learning_rate": 6.218935758043586e-07,
      "loss": 1.1119,
      "step": 428
    },
    {
      "epoch": 0.44011284944857654,
      "grad_norm": 2.5657396333316105,
      "learning_rate": 6.202791289334571e-07,
      "loss": 1.2813,
      "step": 429
    },
    {
      "epoch": 0.44113875352654525,
      "grad_norm": 2.521858145927954,
      "learning_rate": 6.186633499355575e-07,
      "loss": 1.3072,
      "step": 430
    },
    {
      "epoch": 0.44216465760451396,
      "grad_norm": 2.7999275422555523,
      "learning_rate": 6.170462567058908e-07,
      "loss": 1.1443,
      "step": 431
    },
    {
      "epoch": 0.44319056168248266,
      "grad_norm": 2.8231518599718988,
      "learning_rate": 6.15427867154244e-07,
      "loss": 1.2586,
      "step": 432
    },
    {
      "epoch": 0.4442164657604514,
      "grad_norm": 2.6163010604612174,
      "learning_rate": 6.138081992047609e-07,
      "loss": 1.2694,
      "step": 433
    },
    {
      "epoch": 0.4452423698384201,
      "grad_norm": 2.4133746282742075,
      "learning_rate": 6.121872707957441e-07,
      "loss": 1.212,
      "step": 434
    },
    {
      "epoch": 0.4462682739163888,
      "grad_norm": 2.6353803046715405,
      "learning_rate": 6.105650998794559e-07,
      "loss": 1.2294,
      "step": 435
    },
    {
      "epoch": 0.4472941779943575,
      "grad_norm": 2.6224808581968744,
      "learning_rate": 6.089417044219201e-07,
      "loss": 1.2076,
      "step": 436
    },
    {
      "epoch": 0.44832008207232626,
      "grad_norm": 2.273912209414138,
      "learning_rate": 6.073171024027226e-07,
      "loss": 1.2212,
      "step": 437
    },
    {
      "epoch": 0.449345986150295,
      "grad_norm": 2.5146410294881605,
      "learning_rate": 6.056913118148121e-07,
      "loss": 1.2513,
      "step": 438
    },
    {
      "epoch": 0.4503718902282637,
      "grad_norm": 2.513841938933036,
      "learning_rate": 6.040643506643012e-07,
      "loss": 1.153,
      "step": 439
    },
    {
      "epoch": 0.4513977943062324,
      "grad_norm": 2.644906674789187,
      "learning_rate": 6.024362369702668e-07,
      "loss": 1.1479,
      "step": 440
    },
    {
      "epoch": 0.4524236983842011,
      "grad_norm": 2.7660556586821246,
      "learning_rate": 6.008069887645503e-07,
      "loss": 1.2301,
      "step": 441
    },
    {
      "epoch": 0.4534496024621698,
      "grad_norm": 2.4452998667055135,
      "learning_rate": 5.991766240915589e-07,
      "loss": 1.1561,
      "step": 442
    },
    {
      "epoch": 0.4544755065401385,
      "grad_norm": 2.5637907822415196,
      "learning_rate": 5.975451610080642e-07,
      "loss": 1.2151,
      "step": 443
    },
    {
      "epoch": 0.4555014106181072,
      "grad_norm": 2.634126808482155,
      "learning_rate": 5.959126175830033e-07,
      "loss": 1.2055,
      "step": 444
    },
    {
      "epoch": 0.45652731469607594,
      "grad_norm": 2.6995738192233323,
      "learning_rate": 5.942790118972786e-07,
      "loss": 1.1872,
      "step": 445
    },
    {
      "epoch": 0.45755321877404465,
      "grad_norm": 2.600607297977979,
      "learning_rate": 5.926443620435571e-07,
      "loss": 1.1858,
      "step": 446
    },
    {
      "epoch": 0.45857912285201335,
      "grad_norm": 2.7529683044640203,
      "learning_rate": 5.910086861260706e-07,
      "loss": 1.3244,
      "step": 447
    },
    {
      "epoch": 0.45960502692998206,
      "grad_norm": 2.4542539077385106,
      "learning_rate": 5.893720022604142e-07,
      "loss": 1.1912,
      "step": 448
    },
    {
      "epoch": 0.4606309310079508,
      "grad_norm": 2.7213636554229996,
      "learning_rate": 5.877343285733472e-07,
      "loss": 1.2338,
      "step": 449
    },
    {
      "epoch": 0.4616568350859195,
      "grad_norm": 2.5317246559946627,
      "learning_rate": 5.860956832025906e-07,
      "loss": 1.2244,
      "step": 450
    },
    {
      "epoch": 0.4626827391638882,
      "grad_norm": 2.7263289087726186,
      "learning_rate": 5.844560842966278e-07,
      "loss": 1.2377,
      "step": 451
    },
    {
      "epoch": 0.4637086432418569,
      "grad_norm": 2.5974508387893436,
      "learning_rate": 5.828155500145024e-07,
      "loss": 1.2219,
      "step": 452
    },
    {
      "epoch": 0.4647345473198256,
      "grad_norm": 2.492480904755007,
      "learning_rate": 5.811740985256179e-07,
      "loss": 1.2269,
      "step": 453
    },
    {
      "epoch": 0.4657604513977943,
      "grad_norm": 2.713260026567852,
      "learning_rate": 5.79531748009536e-07,
      "loss": 1.2114,
      "step": 454
    },
    {
      "epoch": 0.466786355475763,
      "grad_norm": 2.567056191750939,
      "learning_rate": 5.778885166557752e-07,
      "loss": 1.1719,
      "step": 455
    },
    {
      "epoch": 0.46781225955373174,
      "grad_norm": 2.2856765271337007,
      "learning_rate": 5.7624442266361e-07,
      "loss": 1.144,
      "step": 456
    },
    {
      "epoch": 0.46883816363170044,
      "grad_norm": 2.4580143502304552,
      "learning_rate": 5.745994842418683e-07,
      "loss": 1.2335,
      "step": 457
    },
    {
      "epoch": 0.46986406770966915,
      "grad_norm": 2.343512884988581,
      "learning_rate": 5.729537196087308e-07,
      "loss": 1.1682,
      "step": 458
    },
    {
      "epoch": 0.47088997178763786,
      "grad_norm": 2.5556051709696335,
      "learning_rate": 5.713071469915285e-07,
      "loss": 1.2655,
      "step": 459
    },
    {
      "epoch": 0.47191587586560657,
      "grad_norm": 2.4824036483855143,
      "learning_rate": 5.696597846265411e-07,
      "loss": 1.1478,
      "step": 460
    },
    {
      "epoch": 0.4729417799435753,
      "grad_norm": 2.6655990103099634,
      "learning_rate": 5.680116507587949e-07,
      "loss": 1.3385,
      "step": 461
    },
    {
      "epoch": 0.473967684021544,
      "grad_norm": 2.8244078669617014,
      "learning_rate": 5.663627636418609e-07,
      "loss": 1.19,
      "step": 462
    },
    {
      "epoch": 0.4749935880995127,
      "grad_norm": 2.4429409282936567,
      "learning_rate": 5.647131415376528e-07,
      "loss": 1.3615,
      "step": 463
    },
    {
      "epoch": 0.4760194921774814,
      "grad_norm": 2.9996249497285703,
      "learning_rate": 5.630628027162243e-07,
      "loss": 1.1419,
      "step": 464
    },
    {
      "epoch": 0.4770453962554501,
      "grad_norm": 2.938422500891204,
      "learning_rate": 5.614117654555666e-07,
      "loss": 1.232,
      "step": 465
    },
    {
      "epoch": 0.4780713003334188,
      "grad_norm": 2.7670713948237218,
      "learning_rate": 5.597600480414068e-07,
      "loss": 1.2346,
      "step": 466
    },
    {
      "epoch": 0.47909720441138753,
      "grad_norm": 2.794141488694239,
      "learning_rate": 5.58107668767005e-07,
      "loss": 1.2621,
      "step": 467
    },
    {
      "epoch": 0.48012310848935624,
      "grad_norm": 3.239639832399456,
      "learning_rate": 5.564546459329509e-07,
      "loss": 1.1259,
      "step": 468
    },
    {
      "epoch": 0.48114901256732495,
      "grad_norm": 2.557225226114097,
      "learning_rate": 5.548009978469626e-07,
      "loss": 1.2987,
      "step": 469
    },
    {
      "epoch": 0.48217491664529366,
      "grad_norm": 2.683440752301172,
      "learning_rate": 5.531467428236827e-07,
      "loss": 1.2546,
      "step": 470
    },
    {
      "epoch": 0.48320082072326237,
      "grad_norm": 2.508947166543078,
      "learning_rate": 5.514918991844758e-07,
      "loss": 1.2695,
      "step": 471
    },
    {
      "epoch": 0.4842267248012311,
      "grad_norm": 2.648393508091393,
      "learning_rate": 5.498364852572255e-07,
      "loss": 1.2868,
      "step": 472
    },
    {
      "epoch": 0.4852526288791998,
      "grad_norm": 2.302753325306599,
      "learning_rate": 5.481805193761315e-07,
      "loss": 1.1813,
      "step": 473
    },
    {
      "epoch": 0.4862785329571685,
      "grad_norm": 2.529547436028065,
      "learning_rate": 5.465240198815072e-07,
      "loss": 1.1878,
      "step": 474
    },
    {
      "epoch": 0.4873044370351372,
      "grad_norm": 2.4916972010131286,
      "learning_rate": 5.448670051195751e-07,
      "loss": 1.2886,
      "step": 475
    },
    {
      "epoch": 0.4883303411131059,
      "grad_norm": 2.5311295821586954,
      "learning_rate": 5.432094934422648e-07,
      "loss": 1.1612,
      "step": 476
    },
    {
      "epoch": 0.4893562451910746,
      "grad_norm": 2.5635956429429174,
      "learning_rate": 5.415515032070091e-07,
      "loss": 1.2791,
      "step": 477
    },
    {
      "epoch": 0.49038214926904333,
      "grad_norm": 2.500059103965476,
      "learning_rate": 5.398930527765415e-07,
      "loss": 1.1808,
      "step": 478
    },
    {
      "epoch": 0.49140805334701204,
      "grad_norm": 2.5521862908107154,
      "learning_rate": 5.38234160518692e-07,
      "loss": 1.2029,
      "step": 479
    },
    {
      "epoch": 0.49243395742498075,
      "grad_norm": 2.643620076098173,
      "learning_rate": 5.365748448061837e-07,
      "loss": 1.1923,
      "step": 480
    },
    {
      "epoch": 0.49345986150294946,
      "grad_norm": 2.5902182995299294,
      "learning_rate": 5.349151240164303e-07,
      "loss": 1.2226,
      "step": 481
    },
    {
      "epoch": 0.49448576558091817,
      "grad_norm": 2.768401067358843,
      "learning_rate": 5.332550165313312e-07,
      "loss": 1.174,
      "step": 482
    },
    {
      "epoch": 0.4955116696588869,
      "grad_norm": 2.557685695100021,
      "learning_rate": 5.315945407370686e-07,
      "loss": 1.1896,
      "step": 483
    },
    {
      "epoch": 0.4965375737368556,
      "grad_norm": 2.775440468842976,
      "learning_rate": 5.299337150239041e-07,
      "loss": 1.2212,
      "step": 484
    },
    {
      "epoch": 0.4975634778148243,
      "grad_norm": 2.9224945088224135,
      "learning_rate": 5.282725577859748e-07,
      "loss": 1.2949,
      "step": 485
    },
    {
      "epoch": 0.498589381892793,
      "grad_norm": 2.802770455656179,
      "learning_rate": 5.266110874210892e-07,
      "loss": 1.2073,
      "step": 486
    },
    {
      "epoch": 0.4996152859707617,
      "grad_norm": 2.824244076395358,
      "learning_rate": 5.249493223305244e-07,
      "loss": 1.1947,
      "step": 487
    },
    {
      "epoch": 0.5006411900487304,
      "grad_norm": 2.6923948351193507,
      "learning_rate": 5.232872809188208e-07,
      "loss": 1.2629,
      "step": 488
    },
    {
      "epoch": 0.5016670941266992,
      "grad_norm": 2.3968822015650386,
      "learning_rate": 5.216249815935797e-07,
      "loss": 1.2121,
      "step": 489
    },
    {
      "epoch": 0.5026929982046678,
      "grad_norm": 2.560325439840742,
      "learning_rate": 5.199624427652588e-07,
      "loss": 1.1595,
      "step": 490
    },
    {
      "epoch": 0.5037189022826366,
      "grad_norm": 2.506058680784885,
      "learning_rate": 5.182996828469683e-07,
      "loss": 1.3033,
      "step": 491
    },
    {
      "epoch": 0.5047448063606053,
      "grad_norm": 2.607500294914118,
      "learning_rate": 5.166367202542671e-07,
      "loss": 1.2084,
      "step": 492
    },
    {
      "epoch": 0.505770710438574,
      "grad_norm": 2.641008107839817,
      "learning_rate": 5.149735734049587e-07,
      "loss": 1.2329,
      "step": 493
    },
    {
      "epoch": 0.5067966145165427,
      "grad_norm": 2.541793580004398,
      "learning_rate": 5.133102607188874e-07,
      "loss": 1.2498,
      "step": 494
    },
    {
      "epoch": 0.5078225185945114,
      "grad_norm": 2.5646122609963187,
      "learning_rate": 5.11646800617734e-07,
      "loss": 1.1542,
      "step": 495
    },
    {
      "epoch": 0.5088484226724801,
      "grad_norm": 2.486587957501608,
      "learning_rate": 5.099832115248123e-07,
      "loss": 1.217,
      "step": 496
    },
    {
      "epoch": 0.5098743267504489,
      "grad_norm": 2.675452370610441,
      "learning_rate": 5.083195118648643e-07,
      "loss": 1.2943,
      "step": 497
    },
    {
      "epoch": 0.5109002308284175,
      "grad_norm": 2.5503918368054745,
      "learning_rate": 5.066557200638569e-07,
      "loss": 1.2284,
      "step": 498
    },
    {
      "epoch": 0.5119261349063863,
      "grad_norm": 2.6266723080797028,
      "learning_rate": 5.049918545487774e-07,
      "loss": 1.299,
      "step": 499
    },
    {
      "epoch": 0.5129520389843549,
      "grad_norm": 2.6465331754588055,
      "learning_rate": 5.033279337474294e-07,
      "loss": 1.252,
      "step": 500
    },
    {
      "epoch": 0.5129520389843549,
      "eval_uground_MCTS_chains_SFT_val_loss": 1.2743316888809204,
      "eval_uground_MCTS_chains_SFT_val_runtime": 211.067,
      "eval_uground_MCTS_chains_SFT_val_samples_per_second": 8.618,
      "eval_uground_MCTS_chains_SFT_val_steps_per_second": 1.08,
      "step": 500
    },
    {
      "epoch": 0.5139779430623237,
      "grad_norm": 3.0116752114032743,
      "learning_rate": 5.016639760882288e-07,
      "loss": 1.2094,
      "step": 501
    },
    {
      "epoch": 0.5150038471402923,
      "grad_norm": 2.5726062219813244,
      "learning_rate": 5e-07,
      "loss": 1.2787,
      "step": 502
    },
    {
      "epoch": 0.5160297512182611,
      "grad_norm": 2.5559171539739585,
      "learning_rate": 4.98336023911771e-07,
      "loss": 1.1275,
      "step": 503
    },
    {
      "epoch": 0.5170556552962298,
      "grad_norm": 2.652830973325525,
      "learning_rate": 4.966720662525707e-07,
      "loss": 1.3168,
      "step": 504
    },
    {
      "epoch": 0.5180815593741985,
      "grad_norm": 2.700794383275604,
      "learning_rate": 4.950081454512225e-07,
      "loss": 1.299,
      "step": 505
    },
    {
      "epoch": 0.5191074634521672,
      "grad_norm": 2.784294521735747,
      "learning_rate": 4.933442799361431e-07,
      "loss": 1.2955,
      "step": 506
    },
    {
      "epoch": 0.520133367530136,
      "grad_norm": 2.4208374454069337,
      "learning_rate": 4.916804881351357e-07,
      "loss": 1.1661,
      "step": 507
    },
    {
      "epoch": 0.5211592716081046,
      "grad_norm": 2.569371008392117,
      "learning_rate": 4.900167884751877e-07,
      "loss": 1.2028,
      "step": 508
    },
    {
      "epoch": 0.5221851756860734,
      "grad_norm": 2.7873211845718027,
      "learning_rate": 4.883531993822659e-07,
      "loss": 1.2652,
      "step": 509
    },
    {
      "epoch": 0.523211079764042,
      "grad_norm": 2.6113158053359333,
      "learning_rate": 4.866897392811126e-07,
      "loss": 1.3185,
      "step": 510
    },
    {
      "epoch": 0.5242369838420108,
      "grad_norm": 2.3767956668686385,
      "learning_rate": 4.850264265950413e-07,
      "loss": 1.2051,
      "step": 511
    },
    {
      "epoch": 0.5252628879199794,
      "grad_norm": 2.884877451969795,
      "learning_rate": 4.833632797457331e-07,
      "loss": 1.1469,
      "step": 512
    },
    {
      "epoch": 0.5262887919979482,
      "grad_norm": 2.571309281671942,
      "learning_rate": 4.817003171530317e-07,
      "loss": 1.2533,
      "step": 513
    },
    {
      "epoch": 0.5273146960759169,
      "grad_norm": 2.8580251389569633,
      "learning_rate": 4.800375572347413e-07,
      "loss": 1.2175,
      "step": 514
    },
    {
      "epoch": 0.5283406001538856,
      "grad_norm": 2.657024380838802,
      "learning_rate": 4.783750184064203e-07,
      "loss": 1.2324,
      "step": 515
    },
    {
      "epoch": 0.5293665042318543,
      "grad_norm": 2.4907546697657095,
      "learning_rate": 4.767127190811793e-07,
      "loss": 1.1992,
      "step": 516
    },
    {
      "epoch": 0.530392408309823,
      "grad_norm": 2.782096492852977,
      "learning_rate": 4.750506776694757e-07,
      "loss": 1.2924,
      "step": 517
    },
    {
      "epoch": 0.5314183123877917,
      "grad_norm": 2.427383520783994,
      "learning_rate": 4.7338891257891076e-07,
      "loss": 1.1579,
      "step": 518
    },
    {
      "epoch": 0.5324442164657605,
      "grad_norm": 2.5770073682499484,
      "learning_rate": 4.717274422140252e-07,
      "loss": 1.2246,
      "step": 519
    },
    {
      "epoch": 0.5334701205437291,
      "grad_norm": 2.4572503599210576,
      "learning_rate": 4.7006628497609604e-07,
      "loss": 1.2358,
      "step": 520
    },
    {
      "epoch": 0.5344960246216979,
      "grad_norm": 2.6448246950333254,
      "learning_rate": 4.6840545926293146e-07,
      "loss": 1.2461,
      "step": 521
    },
    {
      "epoch": 0.5355219286996665,
      "grad_norm": 2.5891258589659247,
      "learning_rate": 4.6674498346866887e-07,
      "loss": 1.1114,
      "step": 522
    },
    {
      "epoch": 0.5365478327776353,
      "grad_norm": 2.6559153198303678,
      "learning_rate": 4.650848759835697e-07,
      "loss": 1.2646,
      "step": 523
    },
    {
      "epoch": 0.5375737368556041,
      "grad_norm": 2.673133305478356,
      "learning_rate": 4.634251551938161e-07,
      "loss": 1.2122,
      "step": 524
    },
    {
      "epoch": 0.5385996409335727,
      "grad_norm": 2.68538453093342,
      "learning_rate": 4.6176583948130803e-07,
      "loss": 1.2235,
      "step": 525
    },
    {
      "epoch": 0.5396255450115415,
      "grad_norm": 2.440546575305456,
      "learning_rate": 4.601069472234584e-07,
      "loss": 1.2358,
      "step": 526
    },
    {
      "epoch": 0.5406514490895101,
      "grad_norm": 2.6044073884926924,
      "learning_rate": 4.584484967929909e-07,
      "loss": 1.2677,
      "step": 527
    },
    {
      "epoch": 0.5416773531674789,
      "grad_norm": 2.5476463792153305,
      "learning_rate": 4.5679050655773534e-07,
      "loss": 1.1461,
      "step": 528
    },
    {
      "epoch": 0.5427032572454475,
      "grad_norm": 2.436273544492887,
      "learning_rate": 4.5513299488042487e-07,
      "loss": 1.2116,
      "step": 529
    },
    {
      "epoch": 0.5437291613234163,
      "grad_norm": 2.5753873910087,
      "learning_rate": 4.5347598011849275e-07,
      "loss": 1.1925,
      "step": 530
    },
    {
      "epoch": 0.544755065401385,
      "grad_norm": 2.522315717983984,
      "learning_rate": 4.5181948062386846e-07,
      "loss": 1.2488,
      "step": 531
    },
    {
      "epoch": 0.5457809694793537,
      "grad_norm": 2.4561592505381777,
      "learning_rate": 4.501635147427745e-07,
      "loss": 1.1357,
      "step": 532
    },
    {
      "epoch": 0.5468068735573224,
      "grad_norm": 2.6167475341297375,
      "learning_rate": 4.485081008155243e-07,
      "loss": 1.1628,
      "step": 533
    },
    {
      "epoch": 0.5478327776352911,
      "grad_norm": 2.318388883687291,
      "learning_rate": 4.4685325717631734e-07,
      "loss": 1.1911,
      "step": 534
    },
    {
      "epoch": 0.5488586817132598,
      "grad_norm": 2.551623209073283,
      "learning_rate": 4.4519900215303733e-07,
      "loss": 1.2293,
      "step": 535
    },
    {
      "epoch": 0.5498845857912286,
      "grad_norm": 3.0020726476405875,
      "learning_rate": 4.4354535406704907e-07,
      "loss": 1.1972,
      "step": 536
    },
    {
      "epoch": 0.5509104898691972,
      "grad_norm": 2.4079569221490393,
      "learning_rate": 4.418923312329952e-07,
      "loss": 1.126,
      "step": 537
    },
    {
      "epoch": 0.551936393947166,
      "grad_norm": 2.6031632771818893,
      "learning_rate": 4.4023995195859313e-07,
      "loss": 1.2331,
      "step": 538
    },
    {
      "epoch": 0.5529622980251346,
      "grad_norm": 2.6258547925667353,
      "learning_rate": 4.385882345444335e-07,
      "loss": 1.2573,
      "step": 539
    },
    {
      "epoch": 0.5539882021031034,
      "grad_norm": 2.510150905831169,
      "learning_rate": 4.369371972837757e-07,
      "loss": 1.26,
      "step": 540
    },
    {
      "epoch": 0.555014106181072,
      "grad_norm": 2.7544197615747406,
      "learning_rate": 4.352868584623471e-07,
      "loss": 1.2229,
      "step": 541
    },
    {
      "epoch": 0.5560400102590408,
      "grad_norm": 2.6505858678051863,
      "learning_rate": 4.33637236358139e-07,
      "loss": 1.1452,
      "step": 542
    },
    {
      "epoch": 0.5570659143370095,
      "grad_norm": 2.61741589849154,
      "learning_rate": 4.319883492412051e-07,
      "loss": 1.2923,
      "step": 543
    },
    {
      "epoch": 0.5580918184149782,
      "grad_norm": 2.7258410404480182,
      "learning_rate": 4.30340215373459e-07,
      "loss": 1.1754,
      "step": 544
    },
    {
      "epoch": 0.5591177224929469,
      "grad_norm": 2.551873559715973,
      "learning_rate": 4.286928530084714e-07,
      "loss": 1.2079,
      "step": 545
    },
    {
      "epoch": 0.5601436265709157,
      "grad_norm": 2.7405100803593885,
      "learning_rate": 4.2704628039126914e-07,
      "loss": 1.1972,
      "step": 546
    },
    {
      "epoch": 0.5611695306488843,
      "grad_norm": 2.8909985376686684,
      "learning_rate": 4.2540051575813165e-07,
      "loss": 1.2332,
      "step": 547
    },
    {
      "epoch": 0.5621954347268531,
      "grad_norm": 2.78196270770764,
      "learning_rate": 4.2375557733639006e-07,
      "loss": 1.197,
      "step": 548
    },
    {
      "epoch": 0.5632213388048217,
      "grad_norm": 2.5697422125069784,
      "learning_rate": 4.2211148334422465e-07,
      "loss": 1.2079,
      "step": 549
    },
    {
      "epoch": 0.5642472428827905,
      "grad_norm": 2.6354651408048966,
      "learning_rate": 4.20468251990464e-07,
      "loss": 1.2577,
      "step": 550
    },
    {
      "epoch": 0.5652731469607591,
      "grad_norm": 2.5724042590969938,
      "learning_rate": 4.1882590147438197e-07,
      "loss": 1.2442,
      "step": 551
    },
    {
      "epoch": 0.5662990510387279,
      "grad_norm": 2.830434169158481,
      "learning_rate": 4.1718444998549756e-07,
      "loss": 1.2195,
      "step": 552
    },
    {
      "epoch": 0.5673249551166966,
      "grad_norm": 2.663741290125766,
      "learning_rate": 4.1554391570337223e-07,
      "loss": 1.3125,
      "step": 553
    },
    {
      "epoch": 0.5683508591946653,
      "grad_norm": 2.5970031739016846,
      "learning_rate": 4.1390431679740953e-07,
      "loss": 1.1242,
      "step": 554
    },
    {
      "epoch": 0.569376763272634,
      "grad_norm": 2.768091392231377,
      "learning_rate": 4.1226567142665287e-07,
      "loss": 1.2086,
      "step": 555
    },
    {
      "epoch": 0.5704026673506027,
      "grad_norm": 2.6640670169831173,
      "learning_rate": 4.106279977395858e-07,
      "loss": 1.2559,
      "step": 556
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 2.53235756296073,
      "learning_rate": 4.0899131387392943e-07,
      "loss": 1.1748,
      "step": 557
    },
    {
      "epoch": 0.5724544755065402,
      "grad_norm": 2.682901944624015,
      "learning_rate": 4.073556379564429e-07,
      "loss": 1.1939,
      "step": 558
    },
    {
      "epoch": 0.5734803795845088,
      "grad_norm": 2.5999213234263534,
      "learning_rate": 4.057209881027214e-07,
      "loss": 1.1405,
      "step": 559
    },
    {
      "epoch": 0.5745062836624776,
      "grad_norm": 2.8359536761627755,
      "learning_rate": 4.0408738241699685e-07,
      "loss": 1.2553,
      "step": 560
    },
    {
      "epoch": 0.5755321877404462,
      "grad_norm": 2.478862534452602,
      "learning_rate": 4.0245483899193586e-07,
      "loss": 1.1883,
      "step": 561
    },
    {
      "epoch": 0.576558091818415,
      "grad_norm": 2.9624404297867297,
      "learning_rate": 4.00823375908441e-07,
      "loss": 1.3241,
      "step": 562
    },
    {
      "epoch": 0.5775839958963837,
      "grad_norm": 2.7214306290037467,
      "learning_rate": 3.991930112354496e-07,
      "loss": 1.1486,
      "step": 563
    },
    {
      "epoch": 0.5786098999743524,
      "grad_norm": 2.5096893205743283,
      "learning_rate": 3.9756376302973325e-07,
      "loss": 1.2576,
      "step": 564
    },
    {
      "epoch": 0.5796358040523211,
      "grad_norm": 2.571283901595518,
      "learning_rate": 3.959356493356989e-07,
      "loss": 1.1346,
      "step": 565
    },
    {
      "epoch": 0.5806617081302898,
      "grad_norm": 2.644842706151376,
      "learning_rate": 3.943086881851878e-07,
      "loss": 1.1977,
      "step": 566
    },
    {
      "epoch": 0.5816876122082585,
      "grad_norm": 2.807481417987691,
      "learning_rate": 3.9268289759727736e-07,
      "loss": 1.2563,
      "step": 567
    },
    {
      "epoch": 0.5827135162862273,
      "grad_norm": 2.864431606411388,
      "learning_rate": 3.9105829557807973e-07,
      "loss": 1.2385,
      "step": 568
    },
    {
      "epoch": 0.5837394203641959,
      "grad_norm": 2.7536651973511312,
      "learning_rate": 3.894349001205441e-07,
      "loss": 1.175,
      "step": 569
    },
    {
      "epoch": 0.5847653244421647,
      "grad_norm": 2.557792248642305,
      "learning_rate": 3.87812729204256e-07,
      "loss": 1.263,
      "step": 570
    },
    {
      "epoch": 0.5857912285201333,
      "grad_norm": 2.6951777375825947,
      "learning_rate": 3.861918007952393e-07,
      "loss": 1.1879,
      "step": 571
    },
    {
      "epoch": 0.5868171325981021,
      "grad_norm": 2.5039948130909595,
      "learning_rate": 3.84572132845756e-07,
      "loss": 1.1872,
      "step": 572
    },
    {
      "epoch": 0.5878430366760707,
      "grad_norm": 2.744912722045592,
      "learning_rate": 3.8295374329410926e-07,
      "loss": 1.0991,
      "step": 573
    },
    {
      "epoch": 0.5888689407540395,
      "grad_norm": 2.7544422679030127,
      "learning_rate": 3.8133665006444255e-07,
      "loss": 1.2738,
      "step": 574
    },
    {
      "epoch": 0.5898948448320082,
      "grad_norm": 2.696877329145256,
      "learning_rate": 3.7972087106654296e-07,
      "loss": 1.1795,
      "step": 575
    },
    {
      "epoch": 0.5909207489099769,
      "grad_norm": 2.6792688172902386,
      "learning_rate": 3.781064241956414e-07,
      "loss": 1.2181,
      "step": 576
    },
    {
      "epoch": 0.5919466529879456,
      "grad_norm": 2.7459246071356636,
      "learning_rate": 3.7649332733221575e-07,
      "loss": 1.2205,
      "step": 577
    },
    {
      "epoch": 0.5929725570659143,
      "grad_norm": 2.5648768805482374,
      "learning_rate": 3.7488159834179135e-07,
      "loss": 1.1505,
      "step": 578
    },
    {
      "epoch": 0.593998461143883,
      "grad_norm": 2.430751747310971,
      "learning_rate": 3.732712550747449e-07,
      "loss": 1.1854,
      "step": 579
    },
    {
      "epoch": 0.5950243652218518,
      "grad_norm": 2.705026052502069,
      "learning_rate": 3.716623153661049e-07,
      "loss": 1.2227,
      "step": 580
    },
    {
      "epoch": 0.5960502692998204,
      "grad_norm": 2.7017035978712545,
      "learning_rate": 3.700547970353558e-07,
      "loss": 1.1729,
      "step": 581
    },
    {
      "epoch": 0.5970761733777892,
      "grad_norm": 2.5019196552045972,
      "learning_rate": 3.6844871788623945e-07,
      "loss": 1.1364,
      "step": 582
    },
    {
      "epoch": 0.5981020774557579,
      "grad_norm": 2.8882315979569886,
      "learning_rate": 3.6684409570655804e-07,
      "loss": 1.1923,
      "step": 583
    },
    {
      "epoch": 0.5991279815337266,
      "grad_norm": 2.6337103485212334,
      "learning_rate": 3.652409482679783e-07,
      "loss": 1.2562,
      "step": 584
    },
    {
      "epoch": 0.6001538856116954,
      "grad_norm": 2.8149166619473287,
      "learning_rate": 3.6363929332583263e-07,
      "loss": 1.2216,
      "step": 585
    },
    {
      "epoch": 0.601179789689664,
      "grad_norm": 2.5198337744735,
      "learning_rate": 3.6203914861892476e-07,
      "loss": 1.2763,
      "step": 586
    },
    {
      "epoch": 0.6022056937676328,
      "grad_norm": 2.5771491568943645,
      "learning_rate": 3.604405318693311e-07,
      "loss": 1.2214,
      "step": 587
    },
    {
      "epoch": 0.6032315978456014,
      "grad_norm": 2.6091632070488426,
      "learning_rate": 3.588434607822061e-07,
      "loss": 1.1978,
      "step": 588
    },
    {
      "epoch": 0.6042575019235702,
      "grad_norm": 2.5250674602170537,
      "learning_rate": 3.5724795304558513e-07,
      "loss": 1.1383,
      "step": 589
    },
    {
      "epoch": 0.6052834060015388,
      "grad_norm": 2.818153148170644,
      "learning_rate": 3.5565402633018957e-07,
      "loss": 1.2158,
      "step": 590
    },
    {
      "epoch": 0.6063093100795076,
      "grad_norm": 2.4250679915698643,
      "learning_rate": 3.540616982892297e-07,
      "loss": 1.2316,
      "step": 591
    },
    {
      "epoch": 0.6073352141574763,
      "grad_norm": 2.6768142400804544,
      "learning_rate": 3.5247098655821103e-07,
      "loss": 1.2056,
      "step": 592
    },
    {
      "epoch": 0.608361118235445,
      "grad_norm": 2.6144828178712514,
      "learning_rate": 3.5088190875473686e-07,
      "loss": 1.2977,
      "step": 593
    },
    {
      "epoch": 0.6093870223134137,
      "grad_norm": 2.4216338604546146,
      "learning_rate": 3.4929448247831514e-07,
      "loss": 1.2196,
      "step": 594
    },
    {
      "epoch": 0.6104129263913824,
      "grad_norm": 2.4463464458523814,
      "learning_rate": 3.4770872531016215e-07,
      "loss": 1.2126,
      "step": 595
    },
    {
      "epoch": 0.6114388304693511,
      "grad_norm": 2.746300393158143,
      "learning_rate": 3.4612465481300867e-07,
      "loss": 1.2965,
      "step": 596
    },
    {
      "epoch": 0.6124647345473199,
      "grad_norm": 2.550377036020753,
      "learning_rate": 3.445422885309045e-07,
      "loss": 1.255,
      "step": 597
    },
    {
      "epoch": 0.6134906386252885,
      "grad_norm": 2.5581073815113515,
      "learning_rate": 3.429616439890257e-07,
      "loss": 1.2221,
      "step": 598
    },
    {
      "epoch": 0.6145165427032573,
      "grad_norm": 2.6287106761798764,
      "learning_rate": 3.4138273869347846e-07,
      "loss": 1.1759,
      "step": 599
    },
    {
      "epoch": 0.6155424467812259,
      "grad_norm": 2.7064703602194236,
      "learning_rate": 3.398055901311073e-07,
      "loss": 1.2003,
      "step": 600
    },
    {
      "epoch": 0.6165683508591947,
      "grad_norm": 2.670960915963785,
      "learning_rate": 3.3823021576929964e-07,
      "loss": 1.1922,
      "step": 601
    },
    {
      "epoch": 0.6175942549371634,
      "grad_norm": 2.5857309156591604,
      "learning_rate": 3.3665663305579344e-07,
      "loss": 1.2312,
      "step": 602
    },
    {
      "epoch": 0.6186201590151321,
      "grad_norm": 2.6327880988339194,
      "learning_rate": 3.3508485941848383e-07,
      "loss": 1.1737,
      "step": 603
    },
    {
      "epoch": 0.6196460630931008,
      "grad_norm": 2.6453978254038133,
      "learning_rate": 3.335149122652293e-07,
      "loss": 1.1843,
      "step": 604
    },
    {
      "epoch": 0.6206719671710695,
      "grad_norm": 2.5723820192230598,
      "learning_rate": 3.3194680898366023e-07,
      "loss": 1.2656,
      "step": 605
    },
    {
      "epoch": 0.6216978712490382,
      "grad_norm": 2.5714268383189047,
      "learning_rate": 3.303805669409848e-07,
      "loss": 1.1766,
      "step": 606
    },
    {
      "epoch": 0.622723775327007,
      "grad_norm": 2.545114874785041,
      "learning_rate": 3.288162034837981e-07,
      "loss": 1.2778,
      "step": 607
    },
    {
      "epoch": 0.6237496794049756,
      "grad_norm": 2.3647835710703435,
      "learning_rate": 3.272537359378887e-07,
      "loss": 1.2307,
      "step": 608
    },
    {
      "epoch": 0.6247755834829444,
      "grad_norm": 2.971636171732171,
      "learning_rate": 3.2569318160804805e-07,
      "loss": 1.163,
      "step": 609
    },
    {
      "epoch": 0.625801487560913,
      "grad_norm": 2.4541682678346795,
      "learning_rate": 3.2413455777787746e-07,
      "loss": 1.216,
      "step": 610
    },
    {
      "epoch": 0.6268273916388818,
      "grad_norm": 2.3227177006306423,
      "learning_rate": 3.225778817095982e-07,
      "loss": 1.1738,
      "step": 611
    },
    {
      "epoch": 0.6278532957168504,
      "grad_norm": 2.5261842954294402,
      "learning_rate": 3.2102317064385876e-07,
      "loss": 1.2311,
      "step": 612
    },
    {
      "epoch": 0.6288791997948192,
      "grad_norm": 2.397049474115395,
      "learning_rate": 3.19470441799545e-07,
      "loss": 1.1192,
      "step": 613
    },
    {
      "epoch": 0.6299051038727879,
      "grad_norm": 2.593266135696937,
      "learning_rate": 3.179197123735889e-07,
      "loss": 1.2214,
      "step": 614
    },
    {
      "epoch": 0.6309310079507566,
      "grad_norm": 2.590251241187722,
      "learning_rate": 3.1637099954077875e-07,
      "loss": 1.1793,
      "step": 615
    },
    {
      "epoch": 0.6319569120287253,
      "grad_norm": 2.771110361934025,
      "learning_rate": 3.148243204535677e-07,
      "loss": 1.1597,
      "step": 616
    },
    {
      "epoch": 0.632982816106694,
      "grad_norm": 2.506372299796566,
      "learning_rate": 3.1327969224188546e-07,
      "loss": 1.2679,
      "step": 617
    },
    {
      "epoch": 0.6340087201846627,
      "grad_norm": 2.6050212777549,
      "learning_rate": 3.117371320129469e-07,
      "loss": 1.2335,
      "step": 618
    },
    {
      "epoch": 0.6350346242626315,
      "grad_norm": 2.6519984662810825,
      "learning_rate": 3.101966568510639e-07,
      "loss": 1.207,
      "step": 619
    },
    {
      "epoch": 0.6360605283406001,
      "grad_norm": 2.4414939383947947,
      "learning_rate": 3.086582838174551e-07,
      "loss": 1.1293,
      "step": 620
    },
    {
      "epoch": 0.6370864324185689,
      "grad_norm": 2.4498656223793933,
      "learning_rate": 3.0712202995005806e-07,
      "loss": 1.1637,
      "step": 621
    },
    {
      "epoch": 0.6381123364965375,
      "grad_norm": 2.4980445735258368,
      "learning_rate": 3.055879122633397e-07,
      "loss": 1.2441,
      "step": 622
    },
    {
      "epoch": 0.6391382405745063,
      "grad_norm": 2.770776748857224,
      "learning_rate": 3.040559477481077e-07,
      "loss": 1.2255,
      "step": 623
    },
    {
      "epoch": 0.640164144652475,
      "grad_norm": 2.722646125854624,
      "learning_rate": 3.025261533713235e-07,
      "loss": 1.164,
      "step": 624
    },
    {
      "epoch": 0.6411900487304437,
      "grad_norm": 2.6111255488966827,
      "learning_rate": 3.009985460759127e-07,
      "loss": 1.155,
      "step": 625
    },
    {
      "epoch": 0.6422159528084124,
      "grad_norm": 2.774808082342546,
      "learning_rate": 2.994731427805792e-07,
      "loss": 1.1843,
      "step": 626
    },
    {
      "epoch": 0.6432418568863811,
      "grad_norm": 2.729046492958925,
      "learning_rate": 2.979499603796163e-07,
      "loss": 1.1878,
      "step": 627
    },
    {
      "epoch": 0.6442677609643498,
      "grad_norm": 2.3088017207024105,
      "learning_rate": 2.964290157427207e-07,
      "loss": 1.107,
      "step": 628
    },
    {
      "epoch": 0.6452936650423186,
      "grad_norm": 2.769640462107753,
      "learning_rate": 2.9491032571480486e-07,
      "loss": 1.3354,
      "step": 629
    },
    {
      "epoch": 0.6463195691202872,
      "grad_norm": 2.6586300568678087,
      "learning_rate": 2.9339390711581105e-07,
      "loss": 1.2318,
      "step": 630
    },
    {
      "epoch": 0.647345473198256,
      "grad_norm": 2.8192282998223304,
      "learning_rate": 2.9187977674052424e-07,
      "loss": 1.2764,
      "step": 631
    },
    {
      "epoch": 0.6483713772762246,
      "grad_norm": 2.8997868168822056,
      "learning_rate": 2.9036795135838764e-07,
      "loss": 1.2166,
      "step": 632
    },
    {
      "epoch": 0.6493972813541934,
      "grad_norm": 2.708571116756967,
      "learning_rate": 2.88858447713315e-07,
      "loss": 1.2067,
      "step": 633
    },
    {
      "epoch": 0.650423185432162,
      "grad_norm": 2.620361247323107,
      "learning_rate": 2.8735128252350674e-07,
      "loss": 1.1406,
      "step": 634
    },
    {
      "epoch": 0.6514490895101308,
      "grad_norm": 2.532649091877776,
      "learning_rate": 2.858464724812638e-07,
      "loss": 1.2767,
      "step": 635
    },
    {
      "epoch": 0.6524749935880995,
      "grad_norm": 2.654953427048768,
      "learning_rate": 2.843440342528035e-07,
      "loss": 1.1596,
      "step": 636
    },
    {
      "epoch": 0.6535008976660682,
      "grad_norm": 2.639809330039523,
      "learning_rate": 2.8284398447807435e-07,
      "loss": 1.1958,
      "step": 637
    },
    {
      "epoch": 0.6545268017440369,
      "grad_norm": 2.539389618176952,
      "learning_rate": 2.813463397705723e-07,
      "loss": 1.2032,
      "step": 638
    },
    {
      "epoch": 0.6555527058220056,
      "grad_norm": 2.427603600782808,
      "learning_rate": 2.798511167171562e-07,
      "loss": 1.2111,
      "step": 639
    },
    {
      "epoch": 0.6565786098999743,
      "grad_norm": 2.522000223511891,
      "learning_rate": 2.783583318778646e-07,
      "loss": 1.1634,
      "step": 640
    },
    {
      "epoch": 0.6576045139779431,
      "grad_norm": 2.5130918357831455,
      "learning_rate": 2.768680017857319e-07,
      "loss": 1.2287,
      "step": 641
    },
    {
      "epoch": 0.6586304180559118,
      "grad_norm": 2.7221180506008285,
      "learning_rate": 2.753801429466056e-07,
      "loss": 1.2837,
      "step": 642
    },
    {
      "epoch": 0.6596563221338805,
      "grad_norm": 2.7920485812373403,
      "learning_rate": 2.7389477183896313e-07,
      "loss": 1.0884,
      "step": 643
    },
    {
      "epoch": 0.6606822262118492,
      "grad_norm": 2.4167742978657145,
      "learning_rate": 2.7241190491372987e-07,
      "loss": 1.1931,
      "step": 644
    },
    {
      "epoch": 0.6617081302898179,
      "grad_norm": 2.6876746385015586,
      "learning_rate": 2.7093155859409667e-07,
      "loss": 1.263,
      "step": 645
    },
    {
      "epoch": 0.6627340343677867,
      "grad_norm": 2.544082336181996,
      "learning_rate": 2.6945374927533697e-07,
      "loss": 1.1986,
      "step": 646
    },
    {
      "epoch": 0.6637599384457553,
      "grad_norm": 2.5769749930127177,
      "learning_rate": 2.6797849332462785e-07,
      "loss": 1.3002,
      "step": 647
    },
    {
      "epoch": 0.6647858425237241,
      "grad_norm": 2.765429290049575,
      "learning_rate": 2.665058070808654e-07,
      "loss": 1.293,
      "step": 648
    },
    {
      "epoch": 0.6658117466016927,
      "grad_norm": 2.7225782297173216,
      "learning_rate": 2.650357068544869e-07,
      "loss": 1.22,
      "step": 649
    },
    {
      "epoch": 0.6668376506796615,
      "grad_norm": 2.4495687580818433,
      "learning_rate": 2.635682089272875e-07,
      "loss": 1.1719,
      "step": 650
    },
    {
      "epoch": 0.6678635547576302,
      "grad_norm": 2.515217933274136,
      "learning_rate": 2.621033295522417e-07,
      "loss": 1.3224,
      "step": 651
    },
    {
      "epoch": 0.6688894588355989,
      "grad_norm": 2.6424207426144815,
      "learning_rate": 2.6064108495332293e-07,
      "loss": 1.1448,
      "step": 652
    },
    {
      "epoch": 0.6699153629135676,
      "grad_norm": 2.5858491438102056,
      "learning_rate": 2.5918149132532336e-07,
      "loss": 1.1662,
      "step": 653
    },
    {
      "epoch": 0.6709412669915363,
      "grad_norm": 2.6381268455957594,
      "learning_rate": 2.5772456483367497e-07,
      "loss": 1.1797,
      "step": 654
    },
    {
      "epoch": 0.671967171069505,
      "grad_norm": 2.6228781987684364,
      "learning_rate": 2.5627032161427036e-07,
      "loss": 1.2355,
      "step": 655
    },
    {
      "epoch": 0.6729930751474738,
      "grad_norm": 2.615456043233998,
      "learning_rate": 2.5481877777328424e-07,
      "loss": 1.2654,
      "step": 656
    },
    {
      "epoch": 0.6740189792254424,
      "grad_norm": 2.416561917193722,
      "learning_rate": 2.5336994938699503e-07,
      "loss": 1.12,
      "step": 657
    },
    {
      "epoch": 0.6750448833034112,
      "grad_norm": 2.642465270297251,
      "learning_rate": 2.5192385250160586e-07,
      "loss": 1.1921,
      "step": 658
    },
    {
      "epoch": 0.6760707873813798,
      "grad_norm": 2.7480834455937866,
      "learning_rate": 2.50480503133069e-07,
      "loss": 1.241,
      "step": 659
    },
    {
      "epoch": 0.6770966914593486,
      "grad_norm": 2.709971968793763,
      "learning_rate": 2.4903991726690583e-07,
      "loss": 1.1892,
      "step": 660
    },
    {
      "epoch": 0.6781225955373172,
      "grad_norm": 2.5423357822987125,
      "learning_rate": 2.4760211085803214e-07,
      "loss": 1.061,
      "step": 661
    },
    {
      "epoch": 0.679148499615286,
      "grad_norm": 2.5471809431180947,
      "learning_rate": 2.461670998305801e-07,
      "loss": 1.1406,
      "step": 662
    },
    {
      "epoch": 0.6801744036932547,
      "grad_norm": 2.2614394749490865,
      "learning_rate": 2.447349000777216e-07,
      "loss": 1.1914,
      "step": 663
    },
    {
      "epoch": 0.6812003077712234,
      "grad_norm": 2.5975786485688745,
      "learning_rate": 2.4330552746149404e-07,
      "loss": 1.1927,
      "step": 664
    },
    {
      "epoch": 0.6822262118491921,
      "grad_norm": 2.5643004029139362,
      "learning_rate": 2.418789978126219e-07,
      "loss": 1.2564,
      "step": 665
    },
    {
      "epoch": 0.6832521159271608,
      "grad_norm": 2.5694519450290922,
      "learning_rate": 2.4045532693034474e-07,
      "loss": 1.287,
      "step": 666
    },
    {
      "epoch": 0.6842780200051295,
      "grad_norm": 2.583217401294905,
      "learning_rate": 2.3903453058223876e-07,
      "loss": 1.2259,
      "step": 667
    },
    {
      "epoch": 0.6853039240830983,
      "grad_norm": 2.7713992830326264,
      "learning_rate": 2.3761662450404492e-07,
      "loss": 1.2404,
      "step": 668
    },
    {
      "epoch": 0.6863298281610669,
      "grad_norm": 2.683261081042423,
      "learning_rate": 2.3620162439949303e-07,
      "loss": 1.2424,
      "step": 669
    },
    {
      "epoch": 0.6873557322390357,
      "grad_norm": 2.6813466710719274,
      "learning_rate": 2.347895459401288e-07,
      "loss": 1.1871,
      "step": 670
    },
    {
      "epoch": 0.6883816363170043,
      "grad_norm": 2.4980969369442985,
      "learning_rate": 2.3338040476513947e-07,
      "loss": 1.196,
      "step": 671
    },
    {
      "epoch": 0.6894075403949731,
      "grad_norm": 2.6573309673650116,
      "learning_rate": 2.319742164811813e-07,
      "loss": 1.221,
      "step": 672
    },
    {
      "epoch": 0.6904334444729417,
      "grad_norm": 2.7446852008084224,
      "learning_rate": 2.305709966622062e-07,
      "loss": 1.2488,
      "step": 673
    },
    {
      "epoch": 0.6914593485509105,
      "grad_norm": 2.5674080914108384,
      "learning_rate": 2.2917076084928948e-07,
      "loss": 1.2597,
      "step": 674
    },
    {
      "epoch": 0.6924852526288792,
      "grad_norm": 2.811651191362524,
      "learning_rate": 2.2777352455045784e-07,
      "loss": 1.2413,
      "step": 675
    },
    {
      "epoch": 0.6935111567068479,
      "grad_norm": 2.572728259688939,
      "learning_rate": 2.2637930324051747e-07,
      "loss": 1.2547,
      "step": 676
    },
    {
      "epoch": 0.6945370607848166,
      "grad_norm": 2.700506373166308,
      "learning_rate": 2.2498811236088195e-07,
      "loss": 1.1831,
      "step": 677
    },
    {
      "epoch": 0.6955629648627853,
      "grad_norm": 2.5836554498566864,
      "learning_rate": 2.2359996731940345e-07,
      "loss": 1.1322,
      "step": 678
    },
    {
      "epoch": 0.696588868940754,
      "grad_norm": 2.4909146663935204,
      "learning_rate": 2.2221488349019902e-07,
      "loss": 1.1805,
      "step": 679
    },
    {
      "epoch": 0.6976147730187228,
      "grad_norm": 2.380457276679437,
      "learning_rate": 2.2083287621348256e-07,
      "loss": 1.1997,
      "step": 680
    },
    {
      "epoch": 0.6986406770966914,
      "grad_norm": 2.78505809062204,
      "learning_rate": 2.194539607953948e-07,
      "loss": 1.2253,
      "step": 681
    },
    {
      "epoch": 0.6996665811746602,
      "grad_norm": 2.5665582250329257,
      "learning_rate": 2.180781525078319e-07,
      "loss": 1.1259,
      "step": 682
    },
    {
      "epoch": 0.7006924852526288,
      "grad_norm": 2.555086105402453,
      "learning_rate": 2.167054665882791e-07,
      "loss": 1.1908,
      "step": 683
    },
    {
      "epoch": 0.7017183893305976,
      "grad_norm": 2.6561979962463154,
      "learning_rate": 2.1533591823963926e-07,
      "loss": 1.1986,
      "step": 684
    },
    {
      "epoch": 0.7027442934085663,
      "grad_norm": 2.527387762301175,
      "learning_rate": 2.139695226300663e-07,
      "loss": 1.2669,
      "step": 685
    },
    {
      "epoch": 0.703770197486535,
      "grad_norm": 2.7391799488802575,
      "learning_rate": 2.1260629489279657e-07,
      "loss": 1.1581,
      "step": 686
    },
    {
      "epoch": 0.7047961015645037,
      "grad_norm": 2.4323409695512495,
      "learning_rate": 2.1124625012598135e-07,
      "loss": 1.2396,
      "step": 687
    },
    {
      "epoch": 0.7058220056424724,
      "grad_norm": 2.599314144274195,
      "learning_rate": 2.0988940339251937e-07,
      "loss": 1.1927,
      "step": 688
    },
    {
      "epoch": 0.7068479097204411,
      "grad_norm": 2.7206436512698797,
      "learning_rate": 2.0853576971989035e-07,
      "loss": 1.1842,
      "step": 689
    },
    {
      "epoch": 0.7078738137984099,
      "grad_norm": 2.438206997849408,
      "learning_rate": 2.0718536409998833e-07,
      "loss": 1.1859,
      "step": 690
    },
    {
      "epoch": 0.7088997178763785,
      "grad_norm": 2.425979251612811,
      "learning_rate": 2.058382014889558e-07,
      "loss": 1.2472,
      "step": 691
    },
    {
      "epoch": 0.7099256219543473,
      "grad_norm": 3.1059613200905507,
      "learning_rate": 2.0449429680701797e-07,
      "loss": 1.1009,
      "step": 692
    },
    {
      "epoch": 0.7109515260323159,
      "grad_norm": 2.549687762265441,
      "learning_rate": 2.0315366493831755e-07,
      "loss": 1.2296,
      "step": 693
    },
    {
      "epoch": 0.7119774301102847,
      "grad_norm": 2.5370967790110126,
      "learning_rate": 2.0181632073074923e-07,
      "loss": 1.2098,
      "step": 694
    },
    {
      "epoch": 0.7130033341882533,
      "grad_norm": 2.4536439877324883,
      "learning_rate": 2.0048227899579728e-07,
      "loss": 1.1183,
      "step": 695
    },
    {
      "epoch": 0.7140292382662221,
      "grad_norm": 2.55316768712919,
      "learning_rate": 1.991515545083684e-07,
      "loss": 1.1191,
      "step": 696
    },
    {
      "epoch": 0.7150551423441908,
      "grad_norm": 2.4788318107618124,
      "learning_rate": 1.978241620066315e-07,
      "loss": 1.1502,
      "step": 697
    },
    {
      "epoch": 0.7160810464221595,
      "grad_norm": 2.6045581314247253,
      "learning_rate": 1.9650011619185126e-07,
      "loss": 1.1872,
      "step": 698
    },
    {
      "epoch": 0.7171069505001283,
      "grad_norm": 2.6627567620404107,
      "learning_rate": 1.9517943172822753e-07,
      "loss": 1.2355,
      "step": 699
    },
    {
      "epoch": 0.718132854578097,
      "grad_norm": 2.730504463070005,
      "learning_rate": 1.938621232427327e-07,
      "loss": 1.2131,
      "step": 700
    },
    {
      "epoch": 0.7191587586560657,
      "grad_norm": 2.4953047796323498,
      "learning_rate": 1.9254820532494787e-07,
      "loss": 1.2416,
      "step": 701
    },
    {
      "epoch": 0.7201846627340344,
      "grad_norm": 2.5640586712114986,
      "learning_rate": 1.9123769252690407e-07,
      "loss": 1.2348,
      "step": 702
    },
    {
      "epoch": 0.7212105668120031,
      "grad_norm": 2.301017245817361,
      "learning_rate": 1.8993059936291845e-07,
      "loss": 1.1246,
      "step": 703
    },
    {
      "epoch": 0.7222364708899718,
      "grad_norm": 2.560508237614933,
      "learning_rate": 1.8862694030943528e-07,
      "loss": 1.1593,
      "step": 704
    },
    {
      "epoch": 0.7232623749679405,
      "grad_norm": 2.5384108530663667,
      "learning_rate": 1.873267298048649e-07,
      "loss": 1.177,
      "step": 705
    },
    {
      "epoch": 0.7242882790459092,
      "grad_norm": 2.6017567089602784,
      "learning_rate": 1.8602998224942406e-07,
      "loss": 1.1913,
      "step": 706
    },
    {
      "epoch": 0.725314183123878,
      "grad_norm": 2.563942704168396,
      "learning_rate": 1.847367120049762e-07,
      "loss": 1.1599,
      "step": 707
    },
    {
      "epoch": 0.7263400872018466,
      "grad_norm": 2.477017573966492,
      "learning_rate": 1.834469333948725e-07,
      "loss": 1.3285,
      "step": 708
    },
    {
      "epoch": 0.7273659912798154,
      "grad_norm": 2.606062790037432,
      "learning_rate": 1.8216066070379332e-07,
      "loss": 1.1228,
      "step": 709
    },
    {
      "epoch": 0.728391895357784,
      "grad_norm": 2.627021153673398,
      "learning_rate": 1.808779081775901e-07,
      "loss": 1.254,
      "step": 710
    },
    {
      "epoch": 0.7294177994357528,
      "grad_norm": 2.732112582668254,
      "learning_rate": 1.7959869002312665e-07,
      "loss": 1.1785,
      "step": 711
    },
    {
      "epoch": 0.7304437035137215,
      "grad_norm": 2.7648989482089794,
      "learning_rate": 1.7832302040812392e-07,
      "loss": 1.2254,
      "step": 712
    },
    {
      "epoch": 0.7314696075916902,
      "grad_norm": 2.5265344506070226,
      "learning_rate": 1.7705091346100016e-07,
      "loss": 1.248,
      "step": 713
    },
    {
      "epoch": 0.7324955116696589,
      "grad_norm": 2.487884119195698,
      "learning_rate": 1.757823832707175e-07,
      "loss": 1.2019,
      "step": 714
    },
    {
      "epoch": 0.7335214157476276,
      "grad_norm": 2.7066983532459763,
      "learning_rate": 1.7451744388662326e-07,
      "loss": 1.1771,
      "step": 715
    },
    {
      "epoch": 0.7345473198255963,
      "grad_norm": 2.496085937161839,
      "learning_rate": 1.7325610931829616e-07,
      "loss": 1.1365,
      "step": 716
    },
    {
      "epoch": 0.735573223903565,
      "grad_norm": 2.5310750605865966,
      "learning_rate": 1.719983935353903e-07,
      "loss": 1.2424,
      "step": 717
    },
    {
      "epoch": 0.7365991279815337,
      "grad_norm": 2.6323596452399727,
      "learning_rate": 1.7074431046748074e-07,
      "loss": 1.2113,
      "step": 718
    },
    {
      "epoch": 0.7376250320595025,
      "grad_norm": 2.773768814333001,
      "learning_rate": 1.6949387400390912e-07,
      "loss": 1.2481,
      "step": 719
    },
    {
      "epoch": 0.7386509361374711,
      "grad_norm": 2.6027832225629637,
      "learning_rate": 1.682470979936298e-07,
      "loss": 1.1914,
      "step": 720
    },
    {
      "epoch": 0.7396768402154399,
      "grad_norm": 2.5140857014311084,
      "learning_rate": 1.6700399624505663e-07,
      "loss": 1.1656,
      "step": 721
    },
    {
      "epoch": 0.7407027442934085,
      "grad_norm": 2.6602123194953835,
      "learning_rate": 1.6576458252590986e-07,
      "loss": 1.2504,
      "step": 722
    },
    {
      "epoch": 0.7417286483713773,
      "grad_norm": 2.481197593104395,
      "learning_rate": 1.6452887056306376e-07,
      "loss": 1.1759,
      "step": 723
    },
    {
      "epoch": 0.742754552449346,
      "grad_norm": 2.6801836727829804,
      "learning_rate": 1.6329687404239445e-07,
      "loss": 1.3156,
      "step": 724
    },
    {
      "epoch": 0.7437804565273147,
      "grad_norm": 2.5124931482279105,
      "learning_rate": 1.6206860660862858e-07,
      "loss": 1.2196,
      "step": 725
    },
    {
      "epoch": 0.7448063606052834,
      "grad_norm": 2.664235589191919,
      "learning_rate": 1.6084408186519194e-07,
      "loss": 1.1395,
      "step": 726
    },
    {
      "epoch": 0.7458322646832521,
      "grad_norm": 2.327075273677125,
      "learning_rate": 1.5962331337405914e-07,
      "loss": 1.2351,
      "step": 727
    },
    {
      "epoch": 0.7468581687612208,
      "grad_norm": 2.412427852412854,
      "learning_rate": 1.584063146556025e-07,
      "loss": 1.2697,
      "step": 728
    },
    {
      "epoch": 0.7478840728391896,
      "grad_norm": 2.5543949822741987,
      "learning_rate": 1.5719309918844414e-07,
      "loss": 1.23,
      "step": 729
    },
    {
      "epoch": 0.7489099769171582,
      "grad_norm": 4.7369349678689066,
      "learning_rate": 1.5598368040930427e-07,
      "loss": 1.1512,
      "step": 730
    },
    {
      "epoch": 0.749935880995127,
      "grad_norm": 2.6739445766939745,
      "learning_rate": 1.5477807171285489e-07,
      "loss": 1.2274,
      "step": 731
    },
    {
      "epoch": 0.7509617850730956,
      "grad_norm": 2.8907340492672535,
      "learning_rate": 1.5357628645156918e-07,
      "loss": 1.1342,
      "step": 732
    },
    {
      "epoch": 0.7519876891510644,
      "grad_norm": 2.725694829583633,
      "learning_rate": 1.5237833793557515e-07,
      "loss": 1.2117,
      "step": 733
    },
    {
      "epoch": 0.753013593229033,
      "grad_norm": 2.4983300901122916,
      "learning_rate": 1.5118423943250768e-07,
      "loss": 1.1886,
      "step": 734
    },
    {
      "epoch": 0.7540394973070018,
      "grad_norm": 2.97591503230041,
      "learning_rate": 1.499940041673616e-07,
      "loss": 1.1669,
      "step": 735
    },
    {
      "epoch": 0.7550654013849705,
      "grad_norm": 2.4033448682516942,
      "learning_rate": 1.4880764532234514e-07,
      "loss": 1.0508,
      "step": 736
    },
    {
      "epoch": 0.7560913054629392,
      "grad_norm": 2.508957495407041,
      "learning_rate": 1.4762517603673408e-07,
      "loss": 1.1536,
      "step": 737
    },
    {
      "epoch": 0.7571172095409079,
      "grad_norm": 2.782391609960307,
      "learning_rate": 1.4644660940672627e-07,
      "loss": 1.1624,
      "step": 738
    },
    {
      "epoch": 0.7581431136188767,
      "grad_norm": 2.43794343363202,
      "learning_rate": 1.4527195848529634e-07,
      "loss": 1.2295,
      "step": 739
    },
    {
      "epoch": 0.7591690176968453,
      "grad_norm": 2.566694597309234,
      "learning_rate": 1.4410123628205134e-07,
      "loss": 1.1343,
      "step": 740
    },
    {
      "epoch": 0.7601949217748141,
      "grad_norm": 2.6362502212431576,
      "learning_rate": 1.4293445576308673e-07,
      "loss": 1.2362,
      "step": 741
    },
    {
      "epoch": 0.7612208258527827,
      "grad_norm": 2.4532291147501653,
      "learning_rate": 1.417716298508424e-07,
      "loss": 1.1928,
      "step": 742
    },
    {
      "epoch": 0.7622467299307515,
      "grad_norm": 2.595969156532276,
      "learning_rate": 1.4061277142396006e-07,
      "loss": 1.2755,
      "step": 743
    },
    {
      "epoch": 0.7632726340087201,
      "grad_norm": 2.517605616531437,
      "learning_rate": 1.3945789331714013e-07,
      "loss": 1.1298,
      "step": 744
    },
    {
      "epoch": 0.7642985380866889,
      "grad_norm": 2.5082145361741315,
      "learning_rate": 1.3830700832099994e-07,
      "loss": 1.1012,
      "step": 745
    },
    {
      "epoch": 0.7653244421646576,
      "grad_norm": 2.5628568172056925,
      "learning_rate": 1.3716012918193205e-07,
      "loss": 1.2271,
      "step": 746
    },
    {
      "epoch": 0.7663503462426263,
      "grad_norm": 2.4325845713405387,
      "learning_rate": 1.3601726860196227e-07,
      "loss": 1.1876,
      "step": 747
    },
    {
      "epoch": 0.767376250320595,
      "grad_norm": 2.507778203499113,
      "learning_rate": 1.3487843923861098e-07,
      "loss": 1.1841,
      "step": 748
    },
    {
      "epoch": 0.7684021543985637,
      "grad_norm": 2.492682324587268,
      "learning_rate": 1.337436537047501e-07,
      "loss": 1.2382,
      "step": 749
    },
    {
      "epoch": 0.7694280584765324,
      "grad_norm": 2.695111243135506,
      "learning_rate": 1.3261292456846646e-07,
      "loss": 1.192,
      "step": 750
    },
    {
      "epoch": 0.7694280584765324,
      "eval_uground_MCTS_chains_SFT_val_loss": 1.2723054885864258,
      "eval_uground_MCTS_chains_SFT_val_runtime": 210.8326,
      "eval_uground_MCTS_chains_SFT_val_samples_per_second": 8.628,
      "eval_uground_MCTS_chains_SFT_val_steps_per_second": 1.081,
      "step": 750
    },
    {
      "epoch": 0.7704539625545012,
      "grad_norm": 2.4499636226982187,
      "learning_rate": 1.3148626435291977e-07,
      "loss": 1.2008,
      "step": 751
    },
    {
      "epoch": 0.7714798666324698,
      "grad_norm": 2.749941205607747,
      "learning_rate": 1.30363685536206e-07,
      "loss": 1.1632,
      "step": 752
    },
    {
      "epoch": 0.7725057707104386,
      "grad_norm": 2.6515417658669325,
      "learning_rate": 1.2924520055121834e-07,
      "loss": 1.2075,
      "step": 753
    },
    {
      "epoch": 0.7735316747884072,
      "grad_norm": 2.668556743002617,
      "learning_rate": 1.2813082178550928e-07,
      "loss": 1.2162,
      "step": 754
    },
    {
      "epoch": 0.774557578866376,
      "grad_norm": 2.393774949444245,
      "learning_rate": 1.2702056158115405e-07,
      "loss": 1.1836,
      "step": 755
    },
    {
      "epoch": 0.7755834829443446,
      "grad_norm": 2.383685066946397,
      "learning_rate": 1.2591443223461333e-07,
      "loss": 1.2551,
      "step": 756
    },
    {
      "epoch": 0.7766093870223134,
      "grad_norm": 2.5950121153810164,
      "learning_rate": 1.2481244599659752e-07,
      "loss": 1.1199,
      "step": 757
    },
    {
      "epoch": 0.7776352911002822,
      "grad_norm": 2.683432845893599,
      "learning_rate": 1.2371461507193075e-07,
      "loss": 1.2948,
      "step": 758
    },
    {
      "epoch": 0.7786611951782508,
      "grad_norm": 2.5444949929602236,
      "learning_rate": 1.2262095161941588e-07,
      "loss": 1.1559,
      "step": 759
    },
    {
      "epoch": 0.7796870992562196,
      "grad_norm": 2.825484071880231,
      "learning_rate": 1.215314677516997e-07,
      "loss": 1.1641,
      "step": 760
    },
    {
      "epoch": 0.7807130033341882,
      "grad_norm": 2.7865772894604017,
      "learning_rate": 1.2044617553513897e-07,
      "loss": 1.1873,
      "step": 761
    },
    {
      "epoch": 0.781738907412157,
      "grad_norm": 2.5228443815683224,
      "learning_rate": 1.1936508698966663e-07,
      "loss": 1.2681,
      "step": 762
    },
    {
      "epoch": 0.7827648114901257,
      "grad_norm": 2.8999815031168636,
      "learning_rate": 1.1828821408865886e-07,
      "loss": 1.2122,
      "step": 763
    },
    {
      "epoch": 0.7837907155680944,
      "grad_norm": 2.5689794885209563,
      "learning_rate": 1.1721556875880167e-07,
      "loss": 1.2239,
      "step": 764
    },
    {
      "epoch": 0.7848166196460631,
      "grad_norm": 2.507318965085582,
      "learning_rate": 1.1614716287996063e-07,
      "loss": 1.1625,
      "step": 765
    },
    {
      "epoch": 0.7858425237240318,
      "grad_norm": 2.650879555302289,
      "learning_rate": 1.150830082850468e-07,
      "loss": 1.2091,
      "step": 766
    },
    {
      "epoch": 0.7868684278020005,
      "grad_norm": 2.644278727926541,
      "learning_rate": 1.1402311675988834e-07,
      "loss": 1.273,
      "step": 767
    },
    {
      "epoch": 0.7878943318799693,
      "grad_norm": 2.780296372251445,
      "learning_rate": 1.1296750004309757e-07,
      "loss": 1.236,
      "step": 768
    },
    {
      "epoch": 0.7889202359579379,
      "grad_norm": 2.5188724688574213,
      "learning_rate": 1.1191616982594259e-07,
      "loss": 1.1625,
      "step": 769
    },
    {
      "epoch": 0.7899461400359067,
      "grad_norm": 2.523287196036484,
      "learning_rate": 1.1086913775221706e-07,
      "loss": 1.1488,
      "step": 770
    },
    {
      "epoch": 0.7909720441138753,
      "grad_norm": 2.6829517454733716,
      "learning_rate": 1.0982641541811161e-07,
      "loss": 1.2307,
      "step": 771
    },
    {
      "epoch": 0.7919979481918441,
      "grad_norm": 2.6511117351470697,
      "learning_rate": 1.0878801437208496e-07,
      "loss": 1.229,
      "step": 772
    },
    {
      "epoch": 0.7930238522698128,
      "grad_norm": 2.9646303582070157,
      "learning_rate": 1.077539461147366e-07,
      "loss": 1.1874,
      "step": 773
    },
    {
      "epoch": 0.7940497563477815,
      "grad_norm": 2.7202486187139097,
      "learning_rate": 1.0672422209867876e-07,
      "loss": 1.154,
      "step": 774
    },
    {
      "epoch": 0.7950756604257502,
      "grad_norm": 2.5478407063833424,
      "learning_rate": 1.056988537284103e-07,
      "loss": 1.1646,
      "step": 775
    },
    {
      "epoch": 0.7961015645037189,
      "grad_norm": 2.6329035656681943,
      "learning_rate": 1.0467785236018944e-07,
      "loss": 1.284,
      "step": 776
    },
    {
      "epoch": 0.7971274685816876,
      "grad_norm": 2.581763231489404,
      "learning_rate": 1.0366122930190934e-07,
      "loss": 1.1138,
      "step": 777
    },
    {
      "epoch": 0.7981533726596564,
      "grad_norm": 2.509180799267851,
      "learning_rate": 1.026489958129712e-07,
      "loss": 1.2524,
      "step": 778
    },
    {
      "epoch": 0.799179276737625,
      "grad_norm": 2.277115810698049,
      "learning_rate": 1.0164116310416127e-07,
      "loss": 1.2171,
      "step": 779
    },
    {
      "epoch": 0.8002051808155938,
      "grad_norm": 2.5306432670776386,
      "learning_rate": 1.0063774233752542e-07,
      "loss": 1.2668,
      "step": 780
    },
    {
      "epoch": 0.8012310848935624,
      "grad_norm": 2.7565443335969966,
      "learning_rate": 9.963874462624566e-08,
      "loss": 1.2028,
      "step": 781
    },
    {
      "epoch": 0.8022569889715312,
      "grad_norm": 2.717640654108823,
      "learning_rate": 9.864418103451827e-08,
      "loss": 1.1879,
      "step": 782
    },
    {
      "epoch": 0.8032828930494998,
      "grad_norm": 2.428371028068662,
      "learning_rate": 9.765406257742915e-08,
      "loss": 1.2411,
      "step": 783
    },
    {
      "epoch": 0.8043087971274686,
      "grad_norm": 2.7349523704191516,
      "learning_rate": 9.666840022083422e-08,
      "loss": 1.1647,
      "step": 784
    },
    {
      "epoch": 0.8053347012054373,
      "grad_norm": 2.5706570443073224,
      "learning_rate": 9.568720488123577e-08,
      "loss": 1.2041,
      "step": 785
    },
    {
      "epoch": 0.806360605283406,
      "grad_norm": 2.5312663092030974,
      "learning_rate": 9.471048742566312e-08,
      "loss": 1.2376,
      "step": 786
    },
    {
      "epoch": 0.8073865093613747,
      "grad_norm": 2.567303045700278,
      "learning_rate": 9.373825867155155e-08,
      "loss": 1.2416,
      "step": 787
    },
    {
      "epoch": 0.8084124134393434,
      "grad_norm": 2.5607337841796842,
      "learning_rate": 9.27705293866226e-08,
      "loss": 1.1564,
      "step": 788
    },
    {
      "epoch": 0.8094383175173121,
      "grad_norm": 2.455235264261172,
      "learning_rate": 9.180731028876492e-08,
      "loss": 1.1916,
      "step": 789
    },
    {
      "epoch": 0.8104642215952809,
      "grad_norm": 2.3169514617874594,
      "learning_rate": 9.084861204591549e-08,
      "loss": 1.2042,
      "step": 790
    },
    {
      "epoch": 0.8114901256732495,
      "grad_norm": 2.4016974955821593,
      "learning_rate": 8.989444527594143e-08,
      "loss": 1.0985,
      "step": 791
    },
    {
      "epoch": 0.8125160297512183,
      "grad_norm": 2.682553790748834,
      "learning_rate": 8.894482054652247e-08,
      "loss": 1.1976,
      "step": 792
    },
    {
      "epoch": 0.8135419338291869,
      "grad_norm": 2.5902122938763843,
      "learning_rate": 8.7999748375034e-08,
      "loss": 1.2057,
      "step": 793
    },
    {
      "epoch": 0.8145678379071557,
      "grad_norm": 2.4236373303760272,
      "learning_rate": 8.705923922843039e-08,
      "loss": 1.2162,
      "step": 794
    },
    {
      "epoch": 0.8155937419851244,
      "grad_norm": 2.453178416086071,
      "learning_rate": 8.612330352312891e-08,
      "loss": 1.1699,
      "step": 795
    },
    {
      "epoch": 0.8166196460630931,
      "grad_norm": 2.332974640258587,
      "learning_rate": 8.519195162489528e-08,
      "loss": 1.1845,
      "step": 796
    },
    {
      "epoch": 0.8176455501410618,
      "grad_norm": 2.7398187199253767,
      "learning_rate": 8.426519384872732e-08,
      "loss": 1.2276,
      "step": 797
    },
    {
      "epoch": 0.8186714542190305,
      "grad_norm": 2.799522945659722,
      "learning_rate": 8.334304045874246e-08,
      "loss": 1.1882,
      "step": 798
    },
    {
      "epoch": 0.8196973582969992,
      "grad_norm": 2.597041968467517,
      "learning_rate": 8.242550166806272e-08,
      "loss": 1.21,
      "step": 799
    },
    {
      "epoch": 0.820723262374968,
      "grad_norm": 2.922217189437273,
      "learning_rate": 8.151258763870177e-08,
      "loss": 1.2113,
      "step": 800
    }
  ],
  "logging_steps": 1,
  "max_steps": 974,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 247351370907648.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}