{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 6981,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00014324595330181923,
      "grad_norm": 52.62998580932617,
      "learning_rate": 9.523809523809526e-07,
      "loss": 4.378,
      "step": 1
    },
    {
      "epoch": 0.00028649190660363845,
      "grad_norm": 73.16095733642578,
      "learning_rate": 1.9047619047619051e-06,
      "loss": 4.5535,
      "step": 2
    },
    {
      "epoch": 0.0004297378599054577,
      "grad_norm": 72.34996032714844,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 4.2944,
      "step": 3
    },
    {
      "epoch": 0.0005729838132072769,
      "grad_norm": 58.35626983642578,
      "learning_rate": 3.8095238095238102e-06,
      "loss": 4.7953,
      "step": 4
    },
    {
      "epoch": 0.0007162297665090961,
      "grad_norm": 87.64541625976562,
      "learning_rate": 4.7619047619047615e-06,
      "loss": 4.8454,
      "step": 5
    },
    {
      "epoch": 0.0008594757198109154,
      "grad_norm": 46.40412521362305,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 4.2617,
      "step": 6
    },
    {
      "epoch": 0.0010027216731127346,
      "grad_norm": 43.070899963378906,
      "learning_rate": 6.666666666666667e-06,
      "loss": 3.9821,
      "step": 7
    },
    {
      "epoch": 0.0011459676264145538,
      "grad_norm": 34.57389831542969,
      "learning_rate": 7.6190476190476205e-06,
      "loss": 4.1808,
      "step": 8
    },
    {
      "epoch": 0.001289213579716373,
      "grad_norm": 29.77797508239746,
      "learning_rate": 8.571428571428573e-06,
      "loss": 3.8951,
      "step": 9
    },
    {
      "epoch": 0.0014324595330181922,
      "grad_norm": 23.34331703186035,
      "learning_rate": 9.523809523809523e-06,
      "loss": 3.6618,
      "step": 10
    },
    {
      "epoch": 0.0015757054863200115,
      "grad_norm": 19.189594268798828,
      "learning_rate": 1.0476190476190477e-05,
      "loss": 3.4963,
      "step": 11
    },
    {
      "epoch": 0.0017189514396218307,
      "grad_norm": 26.155723571777344,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 3.6063,
      "step": 12
    },
    {
      "epoch": 0.00186219739292365,
      "grad_norm": 26.80401039123535,
      "learning_rate": 1.2380952380952381e-05,
      "loss": 3.2612,
      "step": 13
    },
    {
      "epoch": 0.0020054433462254693,
      "grad_norm": 31.526416778564453,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 3.5169,
      "step": 14
    },
    {
      "epoch": 0.0021486892995272885,
      "grad_norm": 12.625049591064453,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 2.8682,
      "step": 15
    },
    {
      "epoch": 0.0022919352528291076,
      "grad_norm": 8.687185287475586,
      "learning_rate": 1.5238095238095241e-05,
      "loss": 2.7834,
      "step": 16
    },
    {
      "epoch": 0.002435181206130927,
      "grad_norm": 6.890831470489502,
      "learning_rate": 1.6190476190476193e-05,
      "loss": 2.8426,
      "step": 17
    },
    {
      "epoch": 0.002578427159432746,
      "grad_norm": 5.462021827697754,
      "learning_rate": 1.7142857142857145e-05,
      "loss": 2.6669,
      "step": 18
    },
    {
      "epoch": 0.002721673112734565,
      "grad_norm": 5.529022216796875,
      "learning_rate": 1.8095238095238094e-05,
      "loss": 2.9437,
      "step": 19
    },
    {
      "epoch": 0.0028649190660363843,
      "grad_norm": 3.6151702404022217,
      "learning_rate": 1.9047619047619046e-05,
      "loss": 2.6288,
      "step": 20
    },
    {
      "epoch": 0.0030081650193382035,
      "grad_norm": 5.00957727432251,
      "learning_rate": 2e-05,
      "loss": 2.8314,
      "step": 21
    },
    {
      "epoch": 0.003151410972640023,
      "grad_norm": 7.0872721672058105,
      "learning_rate": 2.0952380952380954e-05,
      "loss": 3.1902,
      "step": 22
    },
    {
      "epoch": 0.0032946569259418423,
      "grad_norm": 4.451307773590088,
      "learning_rate": 2.1904761904761906e-05,
      "loss": 2.7511,
      "step": 23
    },
    {
      "epoch": 0.0034379028792436614,
      "grad_norm": 4.2892961502075195,
      "learning_rate": 2.2857142857142858e-05,
      "loss": 2.5161,
      "step": 24
    },
    {
      "epoch": 0.0035811488325454806,
      "grad_norm": 5.240790367126465,
      "learning_rate": 2.380952380952381e-05,
      "loss": 2.6864,
      "step": 25
    },
    {
      "epoch": 0.0037243947858473,
      "grad_norm": 4.106722354888916,
      "learning_rate": 2.4761904761904762e-05,
      "loss": 2.8796,
      "step": 26
    },
    {
      "epoch": 0.003867640739149119,
      "grad_norm": 3.803208827972412,
      "learning_rate": 2.5714285714285714e-05,
      "loss": 2.4555,
      "step": 27
    },
    {
      "epoch": 0.004010886692450939,
      "grad_norm": 3.8796379566192627,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 2.4737,
      "step": 28
    },
    {
      "epoch": 0.004154132645752757,
      "grad_norm": 2.8030145168304443,
      "learning_rate": 2.7619047619047622e-05,
      "loss": 2.2508,
      "step": 29
    },
    {
      "epoch": 0.004297378599054577,
      "grad_norm": 2.5193729400634766,
      "learning_rate": 2.857142857142857e-05,
      "loss": 2.431,
      "step": 30
    },
    {
      "epoch": 0.004440624552356396,
      "grad_norm": 9.163617134094238,
      "learning_rate": 2.9523809523809526e-05,
      "loss": 2.4488,
      "step": 31
    },
    {
      "epoch": 0.004583870505658215,
      "grad_norm": 2.1598594188690186,
      "learning_rate": 3.0476190476190482e-05,
      "loss": 2.4681,
      "step": 32
    },
    {
      "epoch": 0.004727116458960034,
      "grad_norm": 1.586427927017212,
      "learning_rate": 3.142857142857143e-05,
      "loss": 2.4252,
      "step": 33
    },
    {
      "epoch": 0.004870362412261854,
      "grad_norm": 1.845141887664795,
      "learning_rate": 3.2380952380952386e-05,
      "loss": 2.5847,
      "step": 34
    },
    {
      "epoch": 0.005013608365563673,
      "grad_norm": 2.1396284103393555,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 2.4139,
      "step": 35
    },
    {
      "epoch": 0.005156854318865492,
      "grad_norm": 1.5211702585220337,
      "learning_rate": 3.428571428571429e-05,
      "loss": 2.3666,
      "step": 36
    },
    {
      "epoch": 0.0053001002721673116,
      "grad_norm": 1.7743985652923584,
      "learning_rate": 3.523809523809524e-05,
      "loss": 2.3948,
      "step": 37
    },
    {
      "epoch": 0.00544334622546913,
      "grad_norm": 1.732538104057312,
      "learning_rate": 3.619047619047619e-05,
      "loss": 2.2876,
      "step": 38
    },
    {
      "epoch": 0.00558659217877095,
      "grad_norm": 1.6113958358764648,
      "learning_rate": 3.7142857142857143e-05,
      "loss": 2.4636,
      "step": 39
    },
    {
      "epoch": 0.005729838132072769,
      "grad_norm": 1.706359624862671,
      "learning_rate": 3.809523809523809e-05,
      "loss": 2.3282,
      "step": 40
    },
    {
      "epoch": 0.005873084085374588,
      "grad_norm": 1.6610397100448608,
      "learning_rate": 3.904761904761905e-05,
      "loss": 2.4035,
      "step": 41
    },
    {
      "epoch": 0.006016330038676407,
      "grad_norm": 1.9866093397140503,
      "learning_rate": 4e-05,
      "loss": 2.1891,
      "step": 42
    },
    {
      "epoch": 0.006159575991978227,
      "grad_norm": 1.3556532859802246,
      "learning_rate": 4.095238095238095e-05,
      "loss": 2.4585,
      "step": 43
    },
    {
      "epoch": 0.006302821945280046,
      "grad_norm": 1.4916921854019165,
      "learning_rate": 4.190476190476191e-05,
      "loss": 2.3047,
      "step": 44
    },
    {
      "epoch": 0.006446067898581865,
      "grad_norm": 1.7516685724258423,
      "learning_rate": 4.2857142857142856e-05,
      "loss": 2.3223,
      "step": 45
    },
    {
      "epoch": 0.0065893138518836845,
      "grad_norm": 1.7386003732681274,
      "learning_rate": 4.380952380952381e-05,
      "loss": 2.2594,
      "step": 46
    },
    {
      "epoch": 0.006732559805185503,
      "grad_norm": 1.6626988649368286,
      "learning_rate": 4.476190476190477e-05,
      "loss": 2.1952,
      "step": 47
    },
    {
      "epoch": 0.006875805758487323,
      "grad_norm": 1.5003330707550049,
      "learning_rate": 4.5714285714285716e-05,
      "loss": 2.2867,
      "step": 48
    },
    {
      "epoch": 0.007019051711789142,
      "grad_norm": 1.6588246822357178,
      "learning_rate": 4.666666666666667e-05,
      "loss": 2.1556,
      "step": 49
    },
    {
      "epoch": 0.007162297665090961,
      "grad_norm": 1.3544559478759766,
      "learning_rate": 4.761904761904762e-05,
      "loss": 2.1987,
      "step": 50
    },
    {
      "epoch": 0.00730554361839278,
      "grad_norm": 1.466995358467102,
      "learning_rate": 4.8571428571428576e-05,
      "loss": 2.1098,
      "step": 51
    },
    {
      "epoch": 0.0074487895716946,
      "grad_norm": 1.5698860883712769,
      "learning_rate": 4.9523809523809525e-05,
      "loss": 2.4818,
      "step": 52
    },
    {
      "epoch": 0.007592035524996419,
      "grad_norm": 1.5944514274597168,
      "learning_rate": 5.047619047619048e-05,
      "loss": 2.21,
      "step": 53
    },
    {
      "epoch": 0.007735281478298238,
      "grad_norm": 1.328236699104309,
      "learning_rate": 5.142857142857143e-05,
      "loss": 2.0475,
      "step": 54
    },
    {
      "epoch": 0.007878527431600058,
      "grad_norm": 1.5962390899658203,
      "learning_rate": 5.2380952380952384e-05,
      "loss": 2.2377,
      "step": 55
    },
    {
      "epoch": 0.008021773384901877,
      "grad_norm": 1.5493494272232056,
      "learning_rate": 5.333333333333333e-05,
      "loss": 2.2419,
      "step": 56
    },
    {
      "epoch": 0.008165019338203695,
      "grad_norm": 1.5521552562713623,
      "learning_rate": 5.428571428571428e-05,
      "loss": 2.0745,
      "step": 57
    },
    {
      "epoch": 0.008308265291505515,
      "grad_norm": 1.398070216178894,
      "learning_rate": 5.5238095238095244e-05,
      "loss": 2.2003,
      "step": 58
    },
    {
      "epoch": 0.008451511244807334,
      "grad_norm": 1.4636003971099854,
      "learning_rate": 5.619047619047619e-05,
      "loss": 2.2911,
      "step": 59
    },
    {
      "epoch": 0.008594757198109154,
      "grad_norm": 1.1128668785095215,
      "learning_rate": 5.714285714285714e-05,
      "loss": 2.0765,
      "step": 60
    },
    {
      "epoch": 0.008738003151410973,
      "grad_norm": 1.301720380783081,
      "learning_rate": 5.8095238095238104e-05,
      "loss": 2.1891,
      "step": 61
    },
    {
      "epoch": 0.008881249104712791,
      "grad_norm": 1.5715515613555908,
      "learning_rate": 5.904761904761905e-05,
      "loss": 2.237,
      "step": 62
    },
    {
      "epoch": 0.009024495058014611,
      "grad_norm": 1.0565502643585205,
      "learning_rate": 6e-05,
      "loss": 2.0478,
      "step": 63
    },
    {
      "epoch": 0.00916774101131643,
      "grad_norm": 1.0669859647750854,
      "learning_rate": 6.0952380952380964e-05,
      "loss": 2.2018,
      "step": 64
    },
    {
      "epoch": 0.00931098696461825,
      "grad_norm": 1.3056137561798096,
      "learning_rate": 6.19047619047619e-05,
      "loss": 2.048,
      "step": 65
    },
    {
      "epoch": 0.009454232917920068,
      "grad_norm": 1.6640822887420654,
      "learning_rate": 6.285714285714286e-05,
      "loss": 2.1539,
      "step": 66
    },
    {
      "epoch": 0.009597478871221888,
      "grad_norm": 1.364961862564087,
      "learning_rate": 6.38095238095238e-05,
      "loss": 2.1301,
      "step": 67
    },
    {
      "epoch": 0.009740724824523707,
      "grad_norm": 1.196677327156067,
      "learning_rate": 6.476190476190477e-05,
      "loss": 2.062,
      "step": 68
    },
    {
      "epoch": 0.009883970777825527,
      "grad_norm": 1.2314097881317139,
      "learning_rate": 6.571428571428571e-05,
      "loss": 2.1485,
      "step": 69
    },
    {
      "epoch": 0.010027216731127346,
      "grad_norm": 0.9541298747062683,
      "learning_rate": 6.666666666666667e-05,
      "loss": 2.0575,
      "step": 70
    },
    {
      "epoch": 0.010170462684429164,
      "grad_norm": 1.3738226890563965,
      "learning_rate": 6.761904761904763e-05,
      "loss": 2.2395,
      "step": 71
    },
    {
      "epoch": 0.010313708637730984,
      "grad_norm": 1.2460862398147583,
      "learning_rate": 6.857142857142858e-05,
      "loss": 2.3738,
      "step": 72
    },
    {
      "epoch": 0.010456954591032804,
      "grad_norm": 1.5750643014907837,
      "learning_rate": 6.952380952380952e-05,
      "loss": 2.1665,
      "step": 73
    },
    {
      "epoch": 0.010600200544334623,
      "grad_norm": 1.3629311323165894,
      "learning_rate": 7.047619047619048e-05,
      "loss": 2.0315,
      "step": 74
    },
    {
      "epoch": 0.010743446497636441,
      "grad_norm": 1.502973198890686,
      "learning_rate": 7.142857142857143e-05,
      "loss": 2.0981,
      "step": 75
    },
    {
      "epoch": 0.01088669245093826,
      "grad_norm": 1.2985895872116089,
      "learning_rate": 7.238095238095238e-05,
      "loss": 2.0413,
      "step": 76
    },
    {
      "epoch": 0.01102993840424008,
      "grad_norm": 1.2793923616409302,
      "learning_rate": 7.333333333333333e-05,
      "loss": 2.0376,
      "step": 77
    },
    {
      "epoch": 0.0111731843575419,
      "grad_norm": 1.0606118440628052,
      "learning_rate": 7.428571428571429e-05,
      "loss": 2.1592,
      "step": 78
    },
    {
      "epoch": 0.01131643031084372,
      "grad_norm": 1.101787805557251,
      "learning_rate": 7.523809523809524e-05,
      "loss": 2.2685,
      "step": 79
    },
    {
      "epoch": 0.011459676264145537,
      "grad_norm": 1.2614856958389282,
      "learning_rate": 7.619047619047618e-05,
      "loss": 2.1567,
      "step": 80
    },
    {
      "epoch": 0.011602922217447357,
      "grad_norm": 1.119423270225525,
      "learning_rate": 7.714285714285715e-05,
      "loss": 2.2293,
      "step": 81
    },
    {
      "epoch": 0.011746168170749176,
      "grad_norm": 1.317784070968628,
      "learning_rate": 7.80952380952381e-05,
      "loss": 2.2105,
      "step": 82
    },
    {
      "epoch": 0.011889414124050996,
      "grad_norm": 1.2064156532287598,
      "learning_rate": 7.904761904761905e-05,
      "loss": 2.0107,
      "step": 83
    },
    {
      "epoch": 0.012032660077352814,
      "grad_norm": 1.4291565418243408,
      "learning_rate": 8e-05,
      "loss": 2.2087,
      "step": 84
    },
    {
      "epoch": 0.012175906030654634,
      "grad_norm": 1.3053336143493652,
      "learning_rate": 8.095238095238096e-05,
      "loss": 2.1808,
      "step": 85
    },
    {
      "epoch": 0.012319151983956453,
      "grad_norm": 1.3130074739456177,
      "learning_rate": 8.19047619047619e-05,
      "loss": 2.3055,
      "step": 86
    },
    {
      "epoch": 0.012462397937258273,
      "grad_norm": 1.3293081521987915,
      "learning_rate": 8.285714285714287e-05,
      "loss": 2.1222,
      "step": 87
    },
    {
      "epoch": 0.012605643890560092,
      "grad_norm": 1.3411742448806763,
      "learning_rate": 8.380952380952382e-05,
      "loss": 2.2056,
      "step": 88
    },
    {
      "epoch": 0.01274888984386191,
      "grad_norm": 1.3213112354278564,
      "learning_rate": 8.476190476190477e-05,
      "loss": 2.0332,
      "step": 89
    },
    {
      "epoch": 0.01289213579716373,
      "grad_norm": 1.166063666343689,
      "learning_rate": 8.571428571428571e-05,
      "loss": 2.1501,
      "step": 90
    },
    {
      "epoch": 0.01303538175046555,
      "grad_norm": 1.291751742362976,
      "learning_rate": 8.666666666666667e-05,
      "loss": 2.1473,
      "step": 91
    },
    {
      "epoch": 0.013178627703767369,
      "grad_norm": 1.4382574558258057,
      "learning_rate": 8.761904761904762e-05,
      "loss": 1.9233,
      "step": 92
    },
    {
      "epoch": 0.013321873657069187,
      "grad_norm": 1.0422803163528442,
      "learning_rate": 8.857142857142857e-05,
      "loss": 2.0612,
      "step": 93
    },
    {
      "epoch": 0.013465119610371007,
      "grad_norm": 1.269909143447876,
      "learning_rate": 8.952380952380953e-05,
      "loss": 2.2321,
      "step": 94
    },
    {
      "epoch": 0.013608365563672826,
      "grad_norm": 1.3586599826812744,
      "learning_rate": 9.047619047619048e-05,
      "loss": 1.9371,
      "step": 95
    },
    {
      "epoch": 0.013751611516974646,
      "grad_norm": 1.4789193868637085,
      "learning_rate": 9.142857142857143e-05,
      "loss": 1.9251,
      "step": 96
    },
    {
      "epoch": 0.013894857470276465,
      "grad_norm": 1.35099458694458,
      "learning_rate": 9.238095238095239e-05,
      "loss": 2.2233,
      "step": 97
    },
    {
      "epoch": 0.014038103423578283,
      "grad_norm": 1.0202860832214355,
      "learning_rate": 9.333333333333334e-05,
      "loss": 2.1776,
      "step": 98
    },
    {
      "epoch": 0.014181349376880103,
      "grad_norm": 1.2178406715393066,
      "learning_rate": 9.428571428571429e-05,
      "loss": 2.0692,
      "step": 99
    },
    {
      "epoch": 0.014324595330181922,
      "grad_norm": 1.655634880065918,
      "learning_rate": 9.523809523809524e-05,
      "loss": 2.1217,
      "step": 100
    },
    {
      "epoch": 0.014467841283483742,
      "grad_norm": 1.3825751543045044,
      "learning_rate": 9.61904761904762e-05,
      "loss": 2.0644,
      "step": 101
    },
    {
      "epoch": 0.01461108723678556,
      "grad_norm": 1.1038153171539307,
      "learning_rate": 9.714285714285715e-05,
      "loss": 2.0095,
      "step": 102
    },
    {
      "epoch": 0.01475433319008738,
      "grad_norm": 1.4572913646697998,
      "learning_rate": 9.80952380952381e-05,
      "loss": 2.1681,
      "step": 103
    },
    {
      "epoch": 0.0148975791433892,
      "grad_norm": 1.0186891555786133,
      "learning_rate": 9.904761904761905e-05,
      "loss": 1.9391,
      "step": 104
    },
    {
      "epoch": 0.015040825096691019,
      "grad_norm": 1.3062372207641602,
      "learning_rate": 0.0001,
      "loss": 2.1501,
      "step": 105
    },
    {
      "epoch": 0.015184071049992838,
      "grad_norm": 1.5498818159103394,
      "learning_rate": 0.00010095238095238096,
      "loss": 2.1133,
      "step": 106
    },
    {
      "epoch": 0.015327317003294656,
      "grad_norm": 1.3852585554122925,
      "learning_rate": 0.0001019047619047619,
      "loss": 2.0824,
      "step": 107
    },
    {
      "epoch": 0.015470562956596476,
      "grad_norm": 1.3924949169158936,
      "learning_rate": 0.00010285714285714286,
      "loss": 2.1442,
      "step": 108
    },
    {
      "epoch": 0.015613808909898295,
      "grad_norm": 1.6720349788665771,
      "learning_rate": 0.00010380952380952383,
      "loss": 2.2242,
      "step": 109
    },
    {
      "epoch": 0.015757054863200115,
      "grad_norm": 1.2025669813156128,
      "learning_rate": 0.00010476190476190477,
      "loss": 2.2154,
      "step": 110
    },
    {
      "epoch": 0.015900300816501935,
      "grad_norm": 1.3543643951416016,
      "learning_rate": 0.00010571428571428572,
      "loss": 1.9969,
      "step": 111
    },
    {
      "epoch": 0.016043546769803754,
      "grad_norm": 1.2689183950424194,
      "learning_rate": 0.00010666666666666667,
      "loss": 2.0138,
      "step": 112
    },
    {
      "epoch": 0.016186792723105574,
      "grad_norm": 1.564605951309204,
      "learning_rate": 0.00010761904761904762,
      "loss": 2.1687,
      "step": 113
    },
    {
      "epoch": 0.01633003867640739,
      "grad_norm": 1.6459993124008179,
      "learning_rate": 0.00010857142857142856,
      "loss": 1.9199,
      "step": 114
    },
    {
      "epoch": 0.01647328462970921,
      "grad_norm": 1.2635573148727417,
      "learning_rate": 0.00010952380952380953,
      "loss": 2.0291,
      "step": 115
    },
    {
      "epoch": 0.01661653058301103,
      "grad_norm": 1.8486382961273193,
      "learning_rate": 0.00011047619047619049,
      "loss": 2.0796,
      "step": 116
    },
    {
      "epoch": 0.01675977653631285,
      "grad_norm": 1.4162074327468872,
      "learning_rate": 0.00011142857142857144,
      "loss": 2.0421,
      "step": 117
    },
    {
      "epoch": 0.01690302248961467,
      "grad_norm": 1.2740123271942139,
      "learning_rate": 0.00011238095238095239,
      "loss": 2.0934,
      "step": 118
    },
    {
      "epoch": 0.017046268442916488,
      "grad_norm": 1.0773913860321045,
      "learning_rate": 0.00011333333333333334,
      "loss": 2.2551,
      "step": 119
    },
    {
      "epoch": 0.017189514396218308,
      "grad_norm": 1.8074616193771362,
      "learning_rate": 0.00011428571428571428,
      "loss": 2.178,
      "step": 120
    },
    {
      "epoch": 0.017332760349520127,
      "grad_norm": 1.403294324874878,
      "learning_rate": 0.00011523809523809524,
      "loss": 1.9464,
      "step": 121
    },
    {
      "epoch": 0.017476006302821947,
      "grad_norm": 1.5251725912094116,
      "learning_rate": 0.00011619047619047621,
      "loss": 2.0521,
      "step": 122
    },
    {
      "epoch": 0.017619252256123763,
      "grad_norm": 1.192023754119873,
      "learning_rate": 0.00011714285714285715,
      "loss": 2.089,
      "step": 123
    },
    {
      "epoch": 0.017762498209425583,
      "grad_norm": 1.5174809694290161,
      "learning_rate": 0.0001180952380952381,
      "loss": 2.0408,
      "step": 124
    },
    {
      "epoch": 0.017905744162727402,
      "grad_norm": 1.5903838872909546,
      "learning_rate": 0.00011904761904761905,
      "loss": 2.0947,
      "step": 125
    },
    {
      "epoch": 0.018048990116029222,
      "grad_norm": 1.828599452972412,
      "learning_rate": 0.00012,
      "loss": 2.1239,
      "step": 126
    },
    {
      "epoch": 0.01819223606933104,
      "grad_norm": 1.2842992544174194,
      "learning_rate": 0.00012095238095238095,
      "loss": 2.0186,
      "step": 127
    },
    {
      "epoch": 0.01833548202263286,
      "grad_norm": 1.359697699546814,
      "learning_rate": 0.00012190476190476193,
      "loss": 1.9914,
      "step": 128
    },
    {
      "epoch": 0.01847872797593468,
      "grad_norm": 1.45980703830719,
      "learning_rate": 0.00012285714285714287,
      "loss": 1.977,
      "step": 129
    },
    {
      "epoch": 0.0186219739292365,
      "grad_norm": 1.476110816001892,
      "learning_rate": 0.0001238095238095238,
      "loss": 1.9878,
      "step": 130
    },
    {
      "epoch": 0.01876521988253832,
      "grad_norm": 1.417382836341858,
      "learning_rate": 0.00012476190476190478,
      "loss": 1.9036,
      "step": 131
    },
    {
      "epoch": 0.018908465835840136,
      "grad_norm": 1.361007809638977,
      "learning_rate": 0.00012571428571428572,
      "loss": 2.0625,
      "step": 132
    },
    {
      "epoch": 0.019051711789141956,
      "grad_norm": 1.3317992687225342,
      "learning_rate": 0.00012666666666666666,
      "loss": 2.093,
      "step": 133
    },
    {
      "epoch": 0.019194957742443775,
      "grad_norm": 1.46342933177948,
      "learning_rate": 0.0001276190476190476,
      "loss": 2.0367,
      "step": 134
    },
    {
      "epoch": 0.019338203695745595,
      "grad_norm": 1.2505223751068115,
      "learning_rate": 0.00012857142857142858,
      "loss": 2.2136,
      "step": 135
    },
    {
      "epoch": 0.019481449649047414,
      "grad_norm": 1.5655434131622314,
      "learning_rate": 0.00012952380952380954,
      "loss": 2.0605,
      "step": 136
    },
    {
      "epoch": 0.019624695602349234,
      "grad_norm": 1.0419528484344482,
      "learning_rate": 0.0001304761904761905,
      "loss": 1.9831,
      "step": 137
    },
    {
      "epoch": 0.019767941555651054,
      "grad_norm": 1.260705828666687,
      "learning_rate": 0.00013142857142857143,
      "loss": 1.9628,
      "step": 138
    },
    {
      "epoch": 0.019911187508952873,
      "grad_norm": 1.676471471786499,
      "learning_rate": 0.00013238095238095237,
      "loss": 2.1263,
      "step": 139
    },
    {
      "epoch": 0.020054433462254693,
      "grad_norm": 1.489518642425537,
      "learning_rate": 0.00013333333333333334,
      "loss": 2.1856,
      "step": 140
    },
    {
      "epoch": 0.02019767941555651,
      "grad_norm": 1.6098390817642212,
      "learning_rate": 0.00013428571428571428,
      "loss": 2.071,
      "step": 141
    },
    {
      "epoch": 0.02034092536885833,
      "grad_norm": 1.4824872016906738,
      "learning_rate": 0.00013523809523809525,
      "loss": 1.8687,
      "step": 142
    },
    {
      "epoch": 0.020484171322160148,
      "grad_norm": 1.339866280555725,
      "learning_rate": 0.0001361904761904762,
      "loss": 2.0715,
      "step": 143
    },
    {
      "epoch": 0.020627417275461968,
      "grad_norm": 1.6016653776168823,
      "learning_rate": 0.00013714285714285716,
      "loss": 2.0803,
      "step": 144
    },
    {
      "epoch": 0.020770663228763787,
      "grad_norm": 1.5500822067260742,
      "learning_rate": 0.0001380952380952381,
      "loss": 2.2148,
      "step": 145
    },
    {
      "epoch": 0.020913909182065607,
      "grad_norm": 1.6780601739883423,
      "learning_rate": 0.00013904761904761905,
      "loss": 1.9264,
      "step": 146
    },
    {
      "epoch": 0.021057155135367427,
      "grad_norm": 1.4226387739181519,
      "learning_rate": 0.00014,
      "loss": 2.1784,
      "step": 147
    },
    {
      "epoch": 0.021200401088669246,
      "grad_norm": 1.3277734518051147,
      "learning_rate": 0.00014095238095238096,
      "loss": 2.1087,
      "step": 148
    },
    {
      "epoch": 0.021343647041971066,
      "grad_norm": 1.3667608499526978,
      "learning_rate": 0.00014190476190476193,
      "loss": 2.0414,
      "step": 149
    },
    {
      "epoch": 0.021486892995272882,
      "grad_norm": 1.5164493322372437,
      "learning_rate": 0.00014285714285714287,
      "loss": 2.0465,
      "step": 150
    },
    {
      "epoch": 0.0216301389485747,
      "grad_norm": 1.1949714422225952,
      "learning_rate": 0.0001438095238095238,
      "loss": 2.023,
      "step": 151
    },
    {
      "epoch": 0.02177338490187652,
      "grad_norm": 1.3489972352981567,
      "learning_rate": 0.00014476190476190475,
      "loss": 2.0367,
      "step": 152
    },
    {
      "epoch": 0.02191663085517834,
      "grad_norm": 1.3608229160308838,
      "learning_rate": 0.00014571428571428572,
      "loss": 1.9096,
      "step": 153
    },
    {
      "epoch": 0.02205987680848016,
      "grad_norm": 1.09320068359375,
      "learning_rate": 0.00014666666666666666,
      "loss": 1.9983,
      "step": 154
    },
    {
      "epoch": 0.02220312276178198,
      "grad_norm": 1.4014626741409302,
      "learning_rate": 0.00014761904761904763,
      "loss": 2.0639,
      "step": 155
    },
    {
      "epoch": 0.0223463687150838,
      "grad_norm": 1.7282015085220337,
      "learning_rate": 0.00014857142857142857,
      "loss": 1.9276,
      "step": 156
    },
    {
      "epoch": 0.02248961466838562,
      "grad_norm": 1.368768572807312,
      "learning_rate": 0.00014952380952380954,
      "loss": 1.9422,
      "step": 157
    },
    {
      "epoch": 0.02263286062168744,
      "grad_norm": 2.4440808296203613,
      "learning_rate": 0.00015047619047619048,
      "loss": 1.921,
      "step": 158
    },
    {
      "epoch": 0.022776106574989255,
      "grad_norm": 1.4210387468338013,
      "learning_rate": 0.00015142857142857143,
      "loss": 1.9634,
      "step": 159
    },
    {
      "epoch": 0.022919352528291075,
      "grad_norm": 1.8992178440093994,
      "learning_rate": 0.00015238095238095237,
      "loss": 1.9508,
      "step": 160
    },
    {
      "epoch": 0.023062598481592894,
      "grad_norm": 1.416013240814209,
      "learning_rate": 0.00015333333333333334,
      "loss": 1.9391,
      "step": 161
    },
    {
      "epoch": 0.023205844434894714,
      "grad_norm": 1.3314744234085083,
      "learning_rate": 0.0001542857142857143,
      "loss": 1.9674,
      "step": 162
    },
    {
      "epoch": 0.023349090388196533,
      "grad_norm": 1.4793882369995117,
      "learning_rate": 0.00015523809523809525,
      "loss": 2.1416,
      "step": 163
    },
    {
      "epoch": 0.023492336341498353,
      "grad_norm": 1.5465744733810425,
      "learning_rate": 0.0001561904761904762,
      "loss": 1.9835,
      "step": 164
    },
    {
      "epoch": 0.023635582294800173,
      "grad_norm": 2.0043904781341553,
      "learning_rate": 0.00015714285714285716,
      "loss": 1.9472,
      "step": 165
    },
    {
      "epoch": 0.023778828248101992,
      "grad_norm": 1.2508498430252075,
      "learning_rate": 0.0001580952380952381,
      "loss": 2.0138,
      "step": 166
    },
    {
      "epoch": 0.023922074201403812,
      "grad_norm": 1.4031628370285034,
      "learning_rate": 0.00015904761904761904,
      "loss": 2.0395,
      "step": 167
    },
    {
      "epoch": 0.024065320154705628,
      "grad_norm": 1.2188282012939453,
      "learning_rate": 0.00016,
      "loss": 1.9438,
      "step": 168
    },
    {
      "epoch": 0.024208566108007448,
      "grad_norm": 1.2908705472946167,
      "learning_rate": 0.00016095238095238096,
      "loss": 1.8868,
      "step": 169
    },
    {
      "epoch": 0.024351812061309267,
      "grad_norm": 1.6482155323028564,
      "learning_rate": 0.00016190476190476192,
      "loss": 2.0801,
      "step": 170
    },
    {
      "epoch": 0.024495058014611087,
      "grad_norm": 1.8175079822540283,
      "learning_rate": 0.00016285714285714287,
      "loss": 2.1677,
      "step": 171
    },
    {
      "epoch": 0.024638303967912906,
      "grad_norm": 1.4252662658691406,
      "learning_rate": 0.0001638095238095238,
      "loss": 1.9151,
      "step": 172
    },
    {
      "epoch": 0.024781549921214726,
      "grad_norm": 1.1491389274597168,
      "learning_rate": 0.00016476190476190475,
      "loss": 2.1041,
      "step": 173
    },
    {
      "epoch": 0.024924795874516546,
      "grad_norm": 1.263916015625,
      "learning_rate": 0.00016571428571428575,
      "loss": 1.7433,
      "step": 174
    },
    {
      "epoch": 0.025068041827818365,
      "grad_norm": 1.422499418258667,
      "learning_rate": 0.0001666666666666667,
      "loss": 2.019,
      "step": 175
    },
    {
      "epoch": 0.025211287781120185,
      "grad_norm": 1.555964708328247,
      "learning_rate": 0.00016761904761904763,
      "loss": 1.9613,
      "step": 176
    },
    {
      "epoch": 0.025354533734422,
      "grad_norm": 1.4103100299835205,
      "learning_rate": 0.00016857142857142857,
      "loss": 2.0609,
      "step": 177
    },
    {
      "epoch": 0.02549777968772382,
      "grad_norm": 1.5142751932144165,
      "learning_rate": 0.00016952380952380954,
      "loss": 2.2082,
      "step": 178
    },
    {
      "epoch": 0.02564102564102564,
      "grad_norm": 1.527642846107483,
      "learning_rate": 0.00017047619047619048,
      "loss": 1.8856,
      "step": 179
    },
    {
      "epoch": 0.02578427159432746,
      "grad_norm": 1.2264988422393799,
      "learning_rate": 0.00017142857142857143,
      "loss": 2.0158,
      "step": 180
    },
    {
      "epoch": 0.02592751754762928,
      "grad_norm": 1.1729098558425903,
      "learning_rate": 0.0001723809523809524,
      "loss": 2.1225,
      "step": 181
    },
    {
      "epoch": 0.0260707635009311,
      "grad_norm": 1.666245698928833,
      "learning_rate": 0.00017333333333333334,
      "loss": 2.1028,
      "step": 182
    },
    {
      "epoch": 0.02621400945423292,
      "grad_norm": 2.012843132019043,
      "learning_rate": 0.0001742857142857143,
      "loss": 2.1184,
      "step": 183
    },
    {
      "epoch": 0.026357255407534738,
      "grad_norm": 1.640502691268921,
      "learning_rate": 0.00017523809523809525,
      "loss": 2.0897,
      "step": 184
    },
    {
      "epoch": 0.026500501360836558,
      "grad_norm": 1.4943275451660156,
      "learning_rate": 0.0001761904761904762,
      "loss": 1.9422,
      "step": 185
    },
    {
      "epoch": 0.026643747314138374,
      "grad_norm": 1.4250091314315796,
      "learning_rate": 0.00017714285714285713,
      "loss": 1.9238,
      "step": 186
    },
    {
      "epoch": 0.026786993267440194,
      "grad_norm": 1.699051856994629,
      "learning_rate": 0.0001780952380952381,
      "loss": 2.0062,
      "step": 187
    },
    {
      "epoch": 0.026930239220742013,
      "grad_norm": 1.597662091255188,
      "learning_rate": 0.00017904761904761907,
      "loss": 2.0476,
      "step": 188
    },
    {
      "epoch": 0.027073485174043833,
      "grad_norm": 1.4776837825775146,
      "learning_rate": 0.00018,
      "loss": 2.0264,
      "step": 189
    },
    {
      "epoch": 0.027216731127345652,
      "grad_norm": 1.5483368635177612,
      "learning_rate": 0.00018095238095238095,
      "loss": 2.0707,
      "step": 190
    },
    {
      "epoch": 0.027359977080647472,
      "grad_norm": 1.2471498250961304,
      "learning_rate": 0.00018190476190476192,
      "loss": 1.9984,
      "step": 191
    },
    {
      "epoch": 0.02750322303394929,
      "grad_norm": 1.325850248336792,
      "learning_rate": 0.00018285714285714286,
      "loss": 2.011,
      "step": 192
    },
    {
      "epoch": 0.02764646898725111,
      "grad_norm": 1.6419668197631836,
      "learning_rate": 0.0001838095238095238,
      "loss": 1.8366,
      "step": 193
    },
    {
      "epoch": 0.02778971494055293,
      "grad_norm": 1.1752102375030518,
      "learning_rate": 0.00018476190476190478,
      "loss": 2.0313,
      "step": 194
    },
    {
      "epoch": 0.027932960893854747,
      "grad_norm": 1.2214974164962769,
      "learning_rate": 0.00018571428571428572,
      "loss": 2.0463,
      "step": 195
    },
    {
      "epoch": 0.028076206847156567,
      "grad_norm": 1.4102656841278076,
      "learning_rate": 0.0001866666666666667,
      "loss": 1.9033,
      "step": 196
    },
    {
      "epoch": 0.028219452800458386,
      "grad_norm": 1.3860039710998535,
      "learning_rate": 0.00018761904761904763,
      "loss": 2.1156,
      "step": 197
    },
    {
      "epoch": 0.028362698753760206,
      "grad_norm": 1.4514634609222412,
      "learning_rate": 0.00018857142857142857,
      "loss": 2.0283,
      "step": 198
    },
    {
      "epoch": 0.028505944707062025,
      "grad_norm": 1.1847095489501953,
      "learning_rate": 0.0001895238095238095,
      "loss": 2.0435,
      "step": 199
    },
    {
      "epoch": 0.028649190660363845,
      "grad_norm": 1.2426539659500122,
      "learning_rate": 0.00019047619047619048,
      "loss": 2.0134,
      "step": 200
    },
    {
      "epoch": 0.028792436613665665,
      "grad_norm": 1.1418706178665161,
      "learning_rate": 0.00019142857142857145,
      "loss": 2.0571,
      "step": 201
    },
    {
      "epoch": 0.028935682566967484,
      "grad_norm": 1.5321160554885864,
      "learning_rate": 0.0001923809523809524,
      "loss": 2.1111,
      "step": 202
    },
    {
      "epoch": 0.029078928520269304,
      "grad_norm": 1.7054376602172852,
      "learning_rate": 0.00019333333333333333,
      "loss": 1.9351,
      "step": 203
    },
    {
      "epoch": 0.02922217447357112,
      "grad_norm": 1.5790408849716187,
      "learning_rate": 0.0001942857142857143,
      "loss": 2.0556,
      "step": 204
    },
    {
      "epoch": 0.02936542042687294,
      "grad_norm": 1.5955266952514648,
      "learning_rate": 0.00019523809523809525,
      "loss": 2.1258,
      "step": 205
    },
    {
      "epoch": 0.02950866638017476,
      "grad_norm": 1.4858895540237427,
      "learning_rate": 0.0001961904761904762,
      "loss": 1.9684,
      "step": 206
    },
    {
      "epoch": 0.02965191233347658,
      "grad_norm": 1.96603524684906,
      "learning_rate": 0.00019714285714285716,
      "loss": 1.9494,
      "step": 207
    },
    {
      "epoch": 0.0297951582867784,
      "grad_norm": 1.5131514072418213,
      "learning_rate": 0.0001980952380952381,
      "loss": 2.0232,
      "step": 208
    },
    {
      "epoch": 0.029938404240080218,
      "grad_norm": 1.564658761024475,
      "learning_rate": 0.00019904761904761907,
      "loss": 1.8898,
      "step": 209
    },
    {
      "epoch": 0.030081650193382038,
      "grad_norm": 1.1527990102767944,
      "learning_rate": 0.0002,
      "loss": 1.9758,
      "step": 210
    },
    {
      "epoch": 0.030224896146683857,
      "grad_norm": 1.4897857904434204,
      "learning_rate": 0.00019999998923623733,
      "loss": 1.9385,
      "step": 211
    },
    {
      "epoch": 0.030368142099985677,
      "grad_norm": 1.3621524572372437,
      "learning_rate": 0.00019999995694495155,
      "loss": 1.9114,
      "step": 212
    },
    {
      "epoch": 0.030511388053287496,
      "grad_norm": 1.5576889514923096,
      "learning_rate": 0.00019999990312614968,
      "loss": 1.8958,
      "step": 213
    },
    {
      "epoch": 0.030654634006589312,
      "grad_norm": 1.1676753759384155,
      "learning_rate": 0.00019999982777984328,
      "loss": 1.9371,
      "step": 214
    },
    {
      "epoch": 0.030797879959891132,
      "grad_norm": 1.2504295110702515,
      "learning_rate": 0.00019999973090604857,
      "loss": 2.0819,
      "step": 215
    },
    {
      "epoch": 0.03094112591319295,
      "grad_norm": 1.1602206230163574,
      "learning_rate": 0.0001999996125047864,
      "loss": 1.9313,
      "step": 216
    },
    {
      "epoch": 0.03108437186649477,
      "grad_norm": 1.1292641162872314,
      "learning_rate": 0.00019999947257608226,
      "loss": 1.8298,
      "step": 217
    },
    {
      "epoch": 0.03122761781979659,
      "grad_norm": 1.5617598295211792,
      "learning_rate": 0.0001999993111199663,
      "loss": 1.9535,
      "step": 218
    },
    {
      "epoch": 0.03137086377309841,
      "grad_norm": 1.4218353033065796,
      "learning_rate": 0.00019999912813647321,
      "loss": 2.0776,
      "step": 219
    },
    {
      "epoch": 0.03151410972640023,
      "grad_norm": 1.4078106880187988,
      "learning_rate": 0.00019999892362564244,
      "loss": 1.9998,
      "step": 220
    },
    {
      "epoch": 0.03165735567970205,
      "grad_norm": 1.036447286605835,
      "learning_rate": 0.00019999869758751803,
      "loss": 1.9766,
      "step": 221
    },
    {
      "epoch": 0.03180060163300387,
      "grad_norm": 1.2503416538238525,
      "learning_rate": 0.00019999845002214862,
      "loss": 1.9949,
      "step": 222
    },
    {
      "epoch": 0.03194384758630569,
      "grad_norm": 1.3780275583267212,
      "learning_rate": 0.00019999818092958745,
      "loss": 1.9773,
      "step": 223
    },
    {
      "epoch": 0.03208709353960751,
      "grad_norm": 1.2220666408538818,
      "learning_rate": 0.00019999789030989256,
      "loss": 2.0193,
      "step": 224
    },
    {
      "epoch": 0.03223033949290933,
      "grad_norm": 1.5099045038223267,
      "learning_rate": 0.00019999757816312639,
      "loss": 1.9151,
      "step": 225
    },
    {
      "epoch": 0.03237358544621115,
      "grad_norm": 1.236126184463501,
      "learning_rate": 0.0001999972444893562,
      "loss": 2.0094,
      "step": 226
    },
    {
      "epoch": 0.03251683139951296,
      "grad_norm": 1.5017805099487305,
      "learning_rate": 0.00019999688928865387,
      "loss": 2.0769,
      "step": 227
    },
    {
      "epoch": 0.03266007735281478,
      "grad_norm": 1.1269341707229614,
      "learning_rate": 0.00019999651256109578,
      "loss": 2.1085,
      "step": 228
    },
    {
      "epoch": 0.0328033233061166,
      "grad_norm": 1.117932915687561,
      "learning_rate": 0.00019999611430676306,
      "loss": 2.0881,
      "step": 229
    },
    {
      "epoch": 0.03294656925941842,
      "grad_norm": 1.7601226568222046,
      "learning_rate": 0.00019999569452574148,
      "loss": 1.8408,
      "step": 230
    },
    {
      "epoch": 0.03308981521272024,
      "grad_norm": 1.252386212348938,
      "learning_rate": 0.00019999525321812134,
      "loss": 1.9917,
      "step": 231
    },
    {
      "epoch": 0.03323306116602206,
      "grad_norm": 1.3065465688705444,
      "learning_rate": 0.0001999947903839977,
      "loss": 1.8518,
      "step": 232
    },
    {
      "epoch": 0.03337630711932388,
      "grad_norm": 1.2100579738616943,
      "learning_rate": 0.00019999430602347018,
      "loss": 1.9483,
      "step": 233
    },
    {
      "epoch": 0.0335195530726257,
      "grad_norm": 1.268904447555542,
      "learning_rate": 0.00019999380013664304,
      "loss": 1.9393,
      "step": 234
    },
    {
      "epoch": 0.03366279902592752,
      "grad_norm": 1.0347895622253418,
      "learning_rate": 0.0001999932727236252,
      "loss": 1.9474,
      "step": 235
    },
    {
      "epoch": 0.03380604497922934,
      "grad_norm": 1.6240085363388062,
      "learning_rate": 0.00019999272378453019,
      "loss": 1.9719,
      "step": 236
    },
    {
      "epoch": 0.033949290932531156,
      "grad_norm": 1.3529040813446045,
      "learning_rate": 0.0001999921533194762,
      "loss": 1.8698,
      "step": 237
    },
    {
      "epoch": 0.034092536885832976,
      "grad_norm": 1.3135417699813843,
      "learning_rate": 0.00019999156132858597,
      "loss": 2.0179,
      "step": 238
    },
    {
      "epoch": 0.034235782839134796,
      "grad_norm": 1.1252350807189941,
      "learning_rate": 0.000199990947811987,
      "loss": 1.8703,
      "step": 239
    },
    {
      "epoch": 0.034379028792436615,
      "grad_norm": 1.3984723091125488,
      "learning_rate": 0.00019999031276981142,
      "loss": 2.0266,
      "step": 240
    },
    {
      "epoch": 0.034522274745738435,
      "grad_norm": 1.2189693450927734,
      "learning_rate": 0.00019998965620219585,
      "loss": 2.0681,
      "step": 241
    },
    {
      "epoch": 0.034665520699040255,
      "grad_norm": 0.9576480388641357,
      "learning_rate": 0.00019998897810928162,
      "loss": 1.8396,
      "step": 242
    },
    {
      "epoch": 0.034808766652342074,
      "grad_norm": 1.2399569749832153,
      "learning_rate": 0.0001999882784912148,
      "loss": 2.0687,
      "step": 243
    },
    {
      "epoch": 0.034952012605643894,
      "grad_norm": 1.4409414529800415,
      "learning_rate": 0.00019998755734814593,
      "loss": 1.9458,
      "step": 244
    },
    {
      "epoch": 0.035095258558945706,
      "grad_norm": 1.0613911151885986,
      "learning_rate": 0.00019998681468023027,
      "loss": 1.7906,
      "step": 245
    },
    {
      "epoch": 0.035238504512247526,
      "grad_norm": 1.1053168773651123,
      "learning_rate": 0.00019998605048762768,
      "loss": 1.9425,
      "step": 246
    },
    {
      "epoch": 0.035381750465549346,
      "grad_norm": 1.332306146621704,
      "learning_rate": 0.0001999852647705027,
      "loss": 2.1021,
      "step": 247
    },
    {
      "epoch": 0.035524996418851165,
      "grad_norm": 1.0343549251556396,
      "learning_rate": 0.0001999844575290245,
      "loss": 2.0805,
      "step": 248
    },
    {
      "epoch": 0.035668242372152985,
      "grad_norm": 1.4516009092330933,
      "learning_rate": 0.00019998362876336677,
      "loss": 2.1489,
      "step": 249
    },
    {
      "epoch": 0.035811488325454804,
      "grad_norm": 1.2583619356155396,
      "learning_rate": 0.00019998277847370802,
      "loss": 2.1472,
      "step": 250
    },
    {
      "epoch": 0.035954734278756624,
      "grad_norm": 1.2077544927597046,
      "learning_rate": 0.00019998190666023122,
      "loss": 1.8608,
      "step": 251
    },
    {
      "epoch": 0.036097980232058444,
      "grad_norm": 1.1571005582809448,
      "learning_rate": 0.00019998101332312408,
      "loss": 2.0665,
      "step": 252
    },
    {
      "epoch": 0.03624122618536026,
      "grad_norm": 1.0337742567062378,
      "learning_rate": 0.00019998009846257896,
      "loss": 1.9617,
      "step": 253
    },
    {
      "epoch": 0.03638447213866208,
      "grad_norm": 1.277536153793335,
      "learning_rate": 0.00019997916207879275,
      "loss": 2.0556,
      "step": 254
    },
    {
      "epoch": 0.0365277180919639,
      "grad_norm": 1.2914535999298096,
      "learning_rate": 0.00019997820417196703,
      "loss": 1.9707,
      "step": 255
    },
    {
      "epoch": 0.03667096404526572,
      "grad_norm": 1.0359686613082886,
      "learning_rate": 0.00019997722474230804,
      "loss": 1.8728,
      "step": 256
    },
    {
      "epoch": 0.03681420999856754,
      "grad_norm": 1.415273666381836,
      "learning_rate": 0.00019997622379002661,
      "loss": 2.0969,
      "step": 257
    },
    {
      "epoch": 0.03695745595186936,
      "grad_norm": 1.3990389108657837,
      "learning_rate": 0.00019997520131533823,
      "loss": 1.8875,
      "step": 258
    },
    {
      "epoch": 0.03710070190517118,
      "grad_norm": 1.585178017616272,
      "learning_rate": 0.00019997415731846298,
      "loss": 2.1761,
      "step": 259
    },
    {
      "epoch": 0.037243947858473,
      "grad_norm": 1.2165923118591309,
      "learning_rate": 0.0001999730917996257,
      "loss": 1.8663,
      "step": 260
    },
    {
      "epoch": 0.03738719381177482,
      "grad_norm": 1.3180073499679565,
      "learning_rate": 0.00019997200475905565,
      "loss": 2.1149,
      "step": 261
    },
    {
      "epoch": 0.03753043976507664,
      "grad_norm": 1.3815861940383911,
      "learning_rate": 0.00019997089619698693,
      "loss": 1.8873,
      "step": 262
    },
    {
      "epoch": 0.03767368571837845,
      "grad_norm": 1.2912083864212036,
      "learning_rate": 0.00019996976611365812,
      "loss": 1.9973,
      "step": 263
    },
    {
      "epoch": 0.03781693167168027,
      "grad_norm": 1.3245762586593628,
      "learning_rate": 0.00019996861450931257,
      "loss": 1.8785,
      "step": 264
    },
    {
      "epoch": 0.03796017762498209,
      "grad_norm": 1.3757787942886353,
      "learning_rate": 0.00019996744138419813,
      "loss": 1.9808,
      "step": 265
    },
    {
      "epoch": 0.03810342357828391,
      "grad_norm": 1.3698952198028564,
      "learning_rate": 0.00019996624673856737,
      "loss": 1.8661,
      "step": 266
    },
    {
      "epoch": 0.03824666953158573,
      "grad_norm": 1.2606749534606934,
      "learning_rate": 0.00019996503057267745,
      "loss": 1.9493,
      "step": 267
    },
    {
      "epoch": 0.03838991548488755,
      "grad_norm": 1.1625889539718628,
      "learning_rate": 0.00019996379288679026,
      "loss": 1.9898,
      "step": 268
    },
    {
      "epoch": 0.03853316143818937,
      "grad_norm": 1.4336519241333008,
      "learning_rate": 0.00019996253368117212,
      "loss": 1.9362,
      "step": 269
    },
    {
      "epoch": 0.03867640739149119,
      "grad_norm": 1.3553346395492554,
      "learning_rate": 0.00019996125295609421,
      "loss": 1.9553,
      "step": 270
    },
    {
      "epoch": 0.03881965334479301,
      "grad_norm": 1.1704996824264526,
      "learning_rate": 0.0001999599507118322,
      "loss": 2.0389,
      "step": 271
    },
    {
      "epoch": 0.03896289929809483,
      "grad_norm": 1.2089797258377075,
      "learning_rate": 0.00019995862694866635,
      "loss": 1.9239,
      "step": 272
    },
    {
      "epoch": 0.03910614525139665,
      "grad_norm": 1.003516435623169,
      "learning_rate": 0.00019995728166688178,
      "loss": 1.9979,
      "step": 273
    },
    {
      "epoch": 0.03924939120469847,
      "grad_norm": 1.2670575380325317,
      "learning_rate": 0.00019995591486676801,
      "loss": 1.9217,
      "step": 274
    },
    {
      "epoch": 0.03939263715800029,
      "grad_norm": 1.3995469808578491,
      "learning_rate": 0.0001999545265486193,
      "loss": 1.8742,
      "step": 275
    },
    {
      "epoch": 0.03953588311130211,
      "grad_norm": 1.4435356855392456,
      "learning_rate": 0.0001999531167127345,
      "loss": 1.9226,
      "step": 276
    },
    {
      "epoch": 0.03967912906460393,
      "grad_norm": 1.2965413331985474,
      "learning_rate": 0.00019995168535941716,
      "loss": 1.8605,
      "step": 277
    },
    {
      "epoch": 0.039822375017905746,
      "grad_norm": 1.2421996593475342,
      "learning_rate": 0.00019995023248897535,
      "loss": 1.8853,
      "step": 278
    },
    {
      "epoch": 0.039965620971207566,
      "grad_norm": 1.2318283319473267,
      "learning_rate": 0.00019994875810172186,
      "loss": 1.8703,
      "step": 279
    },
    {
      "epoch": 0.040108866924509386,
      "grad_norm": 1.2504488229751587,
      "learning_rate": 0.00019994726219797412,
      "loss": 2.0319,
      "step": 280
    },
    {
      "epoch": 0.040252112877811205,
      "grad_norm": 1.4693644046783447,
      "learning_rate": 0.00019994574477805415,
      "loss": 1.8048,
      "step": 281
    },
    {
      "epoch": 0.04039535883111302,
      "grad_norm": 1.4368122816085815,
      "learning_rate": 0.0001999442058422886,
      "loss": 1.9194,
      "step": 282
    },
    {
      "epoch": 0.04053860478441484,
      "grad_norm": 1.974845290184021,
      "learning_rate": 0.00019994264539100875,
      "loss": 2.1299,
      "step": 283
    },
    {
      "epoch": 0.04068185073771666,
      "grad_norm": 1.0318002700805664,
      "learning_rate": 0.00019994106342455053,
      "loss": 1.935,
      "step": 284
    },
    {
      "epoch": 0.04082509669101848,
      "grad_norm": 1.3428254127502441,
      "learning_rate": 0.00019993945994325454,
      "loss": 1.9015,
      "step": 285
    },
    {
      "epoch": 0.040968342644320296,
      "grad_norm": 1.4674798250198364,
      "learning_rate": 0.00019993783494746592,
      "loss": 1.8721,
      "step": 286
    },
    {
      "epoch": 0.041111588597622116,
      "grad_norm": 1.1500144004821777,
      "learning_rate": 0.00019993618843753454,
      "loss": 1.884,
      "step": 287
    },
    {
      "epoch": 0.041254834550923936,
      "grad_norm": 1.508557677268982,
      "learning_rate": 0.00019993452041381483,
      "loss": 1.8718,
      "step": 288
    },
    {
      "epoch": 0.041398080504225755,
      "grad_norm": 1.1736414432525635,
      "learning_rate": 0.00019993283087666583,
      "loss": 1.9164,
      "step": 289
    },
    {
      "epoch": 0.041541326457527575,
      "grad_norm": 1.1913667917251587,
      "learning_rate": 0.0001999311198264513,
      "loss": 1.9966,
      "step": 290
    },
    {
      "epoch": 0.041684572410829394,
      "grad_norm": 1.1760199069976807,
      "learning_rate": 0.00019992938726353963,
      "loss": 1.9862,
      "step": 291
    },
    {
      "epoch": 0.041827818364131214,
      "grad_norm": 1.116284966468811,
      "learning_rate": 0.0001999276331883037,
      "loss": 2.1363,
      "step": 292
    },
    {
      "epoch": 0.041971064317433034,
      "grad_norm": 1.3764779567718506,
      "learning_rate": 0.0001999258576011212,
      "loss": 2.0099,
      "step": 293
    },
    {
      "epoch": 0.04211431027073485,
      "grad_norm": 1.110652208328247,
      "learning_rate": 0.00019992406050237433,
      "loss": 1.9664,
      "step": 294
    },
    {
      "epoch": 0.04225755622403667,
      "grad_norm": 1.2322907447814941,
      "learning_rate": 0.00019992224189244997,
      "loss": 1.8815,
      "step": 295
    },
    {
      "epoch": 0.04240080217733849,
      "grad_norm": 1.3137946128845215,
      "learning_rate": 0.00019992040177173963,
      "loss": 1.7849,
      "step": 296
    },
    {
      "epoch": 0.04254404813064031,
      "grad_norm": 1.1724376678466797,
      "learning_rate": 0.00019991854014063943,
      "loss": 2.039,
      "step": 297
    },
    {
      "epoch": 0.04268729408394213,
      "grad_norm": 1.3877332210540771,
      "learning_rate": 0.00019991665699955013,
      "loss": 1.8806,
      "step": 298
    },
    {
      "epoch": 0.04283054003724395,
      "grad_norm": 1.1249327659606934,
      "learning_rate": 0.00019991475234887714,
      "loss": 1.9858,
      "step": 299
    },
    {
      "epoch": 0.042973785990545764,
      "grad_norm": 1.104592204093933,
      "learning_rate": 0.0001999128261890305,
      "loss": 1.8972,
      "step": 300
    },
    {
      "epoch": 0.043117031943847584,
      "grad_norm": 1.0912657976150513,
      "learning_rate": 0.00019991087852042479,
      "loss": 1.8983,
      "step": 301
    },
    {
      "epoch": 0.0432602778971494,
      "grad_norm": 1.2322807312011719,
      "learning_rate": 0.00019990890934347937,
      "loss": 1.823,
      "step": 302
    },
    {
      "epoch": 0.04340352385045122,
      "grad_norm": 1.2307558059692383,
      "learning_rate": 0.00019990691865861814,
      "loss": 1.8042,
      "step": 303
    },
    {
      "epoch": 0.04354676980375304,
      "grad_norm": 1.3651338815689087,
      "learning_rate": 0.0001999049064662696,
      "loss": 1.9489,
      "step": 304
    },
    {
      "epoch": 0.04369001575705486,
      "grad_norm": 1.2109830379486084,
      "learning_rate": 0.00019990287276686698,
      "loss": 1.9445,
      "step": 305
    },
    {
      "epoch": 0.04383326171035668,
      "grad_norm": 1.2235767841339111,
      "learning_rate": 0.00019990081756084808,
      "loss": 2.149,
      "step": 306
    },
    {
      "epoch": 0.0439765076636585,
      "grad_norm": 1.5291017293930054,
      "learning_rate": 0.00019989874084865532,
      "loss": 1.8934,
      "step": 307
    },
    {
      "epoch": 0.04411975361696032,
      "grad_norm": 1.2952134609222412,
      "learning_rate": 0.00019989664263073573,
      "loss": 1.9818,
      "step": 308
    },
    {
      "epoch": 0.04426299957026214,
      "grad_norm": 1.3951518535614014,
      "learning_rate": 0.00019989452290754107,
      "loss": 1.9474,
      "step": 309
    },
    {
      "epoch": 0.04440624552356396,
      "grad_norm": 1.39193594455719,
      "learning_rate": 0.00019989238167952765,
      "loss": 1.8892,
      "step": 310
    },
    {
      "epoch": 0.04454949147686578,
      "grad_norm": 1.589078664779663,
      "learning_rate": 0.00019989021894715637,
      "loss": 2.0554,
      "step": 311
    },
    {
      "epoch": 0.0446927374301676,
      "grad_norm": 1.2662962675094604,
      "learning_rate": 0.00019988803471089286,
      "loss": 2.0473,
      "step": 312
    },
    {
      "epoch": 0.04483598338346942,
      "grad_norm": 1.1959353685379028,
      "learning_rate": 0.00019988582897120734,
      "loss": 1.9803,
      "step": 313
    },
    {
      "epoch": 0.04497922933677124,
      "grad_norm": 1.4584497213363647,
      "learning_rate": 0.0001998836017285746,
      "loss": 1.9695,
      "step": 314
    },
    {
      "epoch": 0.04512247529007306,
      "grad_norm": 1.480107069015503,
      "learning_rate": 0.00019988135298347416,
      "loss": 1.9043,
      "step": 315
    },
    {
      "epoch": 0.04526572124337488,
      "grad_norm": 1.7253074645996094,
      "learning_rate": 0.00019987908273639011,
      "loss": 1.7338,
      "step": 316
    },
    {
      "epoch": 0.0454089671966767,
      "grad_norm": 1.3314229249954224,
      "learning_rate": 0.00019987679098781115,
      "loss": 2.0523,
      "step": 317
    },
    {
      "epoch": 0.04555221314997851,
      "grad_norm": 1.2367300987243652,
      "learning_rate": 0.00019987447773823068,
      "loss": 1.9092,
      "step": 318
    },
    {
      "epoch": 0.04569545910328033,
      "grad_norm": 1.3592756986618042,
      "learning_rate": 0.00019987214298814666,
      "loss": 1.8355,
      "step": 319
    },
    {
      "epoch": 0.04583870505658215,
      "grad_norm": 1.755040168762207,
      "learning_rate": 0.00019986978673806172,
      "loss": 1.9794,
      "step": 320
    },
    {
      "epoch": 0.04598195100988397,
      "grad_norm": 1.7361351251602173,
      "learning_rate": 0.00019986740898848306,
      "loss": 1.9147,
      "step": 321
    },
    {
      "epoch": 0.04612519696318579,
      "grad_norm": 1.2127351760864258,
      "learning_rate": 0.0001998650097399226,
      "loss": 1.9654,
      "step": 322
    },
    {
      "epoch": 0.04626844291648761,
      "grad_norm": 1.4929267168045044,
      "learning_rate": 0.0001998625889928968,
      "loss": 1.8369,
      "step": 323
    },
    {
      "epoch": 0.04641168886978943,
      "grad_norm": 1.6062040328979492,
      "learning_rate": 0.00019986014674792683,
      "loss": 2.1023,
      "step": 324
    },
    {
      "epoch": 0.04655493482309125,
      "grad_norm": 1.0715123414993286,
      "learning_rate": 0.0001998576830055384,
      "loss": 1.9211,
      "step": 325
    },
    {
      "epoch": 0.04669818077639307,
      "grad_norm": 1.1279674768447876,
      "learning_rate": 0.0001998551977662619,
      "loss": 1.7204,
      "step": 326
    },
    {
      "epoch": 0.046841426729694886,
      "grad_norm": 1.1469101905822754,
      "learning_rate": 0.00019985269103063238,
      "loss": 1.82,
      "step": 327
    },
    {
      "epoch": 0.046984672682996706,
      "grad_norm": 1.3553428649902344,
      "learning_rate": 0.00019985016279918942,
      "loss": 1.9313,
      "step": 328
    },
    {
      "epoch": 0.047127918636298526,
      "grad_norm": 1.42116117477417,
      "learning_rate": 0.00019984761307247735,
      "loss": 1.8604,
      "step": 329
    },
    {
      "epoch": 0.047271164589600345,
      "grad_norm": 1.1581860780715942,
      "learning_rate": 0.00019984504185104502,
      "loss": 1.7452,
      "step": 330
    },
    {
      "epoch": 0.047414410542902165,
      "grad_norm": 1.3027849197387695,
      "learning_rate": 0.00019984244913544597,
      "loss": 1.9309,
      "step": 331
    },
    {
      "epoch": 0.047557656496203984,
      "grad_norm": 1.2336416244506836,
      "learning_rate": 0.00019983983492623833,
      "loss": 1.9478,
      "step": 332
    },
    {
      "epoch": 0.047700902449505804,
      "grad_norm": 1.3841291666030884,
      "learning_rate": 0.00019983719922398488,
      "loss": 1.977,
      "step": 333
    },
    {
      "epoch": 0.047844148402807624,
      "grad_norm": 1.1200348138809204,
      "learning_rate": 0.00019983454202925302,
      "loss": 2.029,
      "step": 334
    },
    {
      "epoch": 0.04798739435610944,
      "grad_norm": 0.9795233011245728,
      "learning_rate": 0.0001998318633426148,
      "loss": 1.9426,
      "step": 335
    },
    {
      "epoch": 0.048130640309411256,
      "grad_norm": 1.3202542066574097,
      "learning_rate": 0.0001998291631646468,
      "loss": 1.9019,
      "step": 336
    },
    {
      "epoch": 0.048273886262713075,
      "grad_norm": 1.3691173791885376,
      "learning_rate": 0.00019982644149593043,
      "loss": 1.9393,
      "step": 337
    },
    {
      "epoch": 0.048417132216014895,
      "grad_norm": 1.2442659139633179,
      "learning_rate": 0.0001998236983370515,
      "loss": 1.8201,
      "step": 338
    },
    {
      "epoch": 0.048560378169316715,
      "grad_norm": 1.8122320175170898,
      "learning_rate": 0.00019982093368860055,
      "loss": 1.9996,
      "step": 339
    },
    {
      "epoch": 0.048703624122618534,
      "grad_norm": 1.0892632007598877,
      "learning_rate": 0.0001998181475511728,
      "loss": 1.8105,
      "step": 340
    },
    {
      "epoch": 0.048846870075920354,
      "grad_norm": 1.5804370641708374,
      "learning_rate": 0.000199815339925368,
      "loss": 2.0603,
      "step": 341
    },
    {
      "epoch": 0.048990116029222174,
      "grad_norm": 1.505051612854004,
      "learning_rate": 0.00019981251081179052,
      "loss": 1.919,
      "step": 342
    },
    {
      "epoch": 0.04913336198252399,
      "grad_norm": 1.4446977376937866,
      "learning_rate": 0.00019980966021104947,
      "loss": 1.8723,
      "step": 343
    },
    {
      "epoch": 0.04927660793582581,
      "grad_norm": 1.1602543592453003,
      "learning_rate": 0.00019980678812375847,
      "loss": 1.9026,
      "step": 344
    },
    {
      "epoch": 0.04941985388912763,
      "grad_norm": 1.362926959991455,
      "learning_rate": 0.00019980389455053583,
      "loss": 1.9745,
      "step": 345
    },
    {
      "epoch": 0.04956309984242945,
      "grad_norm": 1.3263410329818726,
      "learning_rate": 0.00019980097949200444,
      "loss": 1.7165,
      "step": 346
    },
    {
      "epoch": 0.04970634579573127,
      "grad_norm": 1.3735880851745605,
      "learning_rate": 0.00019979804294879188,
      "loss": 2.0095,
      "step": 347
    },
    {
      "epoch": 0.04984959174903309,
      "grad_norm": 1.2267096042633057,
      "learning_rate": 0.00019979508492153026,
      "loss": 1.8871,
      "step": 348
    },
    {
      "epoch": 0.04999283770233491,
      "grad_norm": 1.0966641902923584,
      "learning_rate": 0.00019979210541085644,
      "loss": 1.925,
      "step": 349
    },
    {
      "epoch": 0.05013608365563673,
      "grad_norm": 1.2373641729354858,
      "learning_rate": 0.00019978910441741175,
      "loss": 2.1047,
      "step": 350
    },
    {
      "epoch": 0.05027932960893855,
      "grad_norm": 1.0201295614242554,
      "learning_rate": 0.00019978608194184228,
      "loss": 1.9623,
      "step": 351
    },
    {
      "epoch": 0.05042257556224037,
      "grad_norm": 1.0486125946044922,
      "learning_rate": 0.00019978303798479875,
      "loss": 1.9244,
      "step": 352
    },
    {
      "epoch": 0.05056582151554219,
      "grad_norm": 1.224714994430542,
      "learning_rate": 0.00019977997254693632,
      "loss": 1.8478,
      "step": 353
    },
    {
      "epoch": 0.050709067468844,
      "grad_norm": 1.1136339902877808,
      "learning_rate": 0.00019977688562891502,
      "loss": 1.9839,
      "step": 354
    },
    {
      "epoch": 0.05085231342214582,
      "grad_norm": 1.1282410621643066,
      "learning_rate": 0.00019977377723139932,
      "loss": 2.1154,
      "step": 355
    },
    {
      "epoch": 0.05099555937544764,
      "grad_norm": 1.1078453063964844,
      "learning_rate": 0.00019977064735505836,
      "loss": 1.9597,
      "step": 356
    },
    {
      "epoch": 0.05113880532874946,
      "grad_norm": 1.3309283256530762,
      "learning_rate": 0.00019976749600056603,
      "loss": 1.851,
      "step": 357
    },
    {
      "epoch": 0.05128205128205128,
      "grad_norm": 1.3199007511138916,
      "learning_rate": 0.00019976432316860067,
      "loss": 1.9931,
      "step": 358
    },
    {
      "epoch": 0.0514252972353531,
      "grad_norm": 1.443095088005066,
      "learning_rate": 0.00019976112885984525,
      "loss": 2.0769,
      "step": 359
    },
    {
      "epoch": 0.05156854318865492,
      "grad_norm": 1.2029110193252563,
      "learning_rate": 0.00019975791307498756,
      "loss": 1.7789,
      "step": 360
    },
    {
      "epoch": 0.05171178914195674,
      "grad_norm": 1.3783962726593018,
      "learning_rate": 0.0001997546758147198,
      "loss": 2.1912,
      "step": 361
    },
    {
      "epoch": 0.05185503509525856,
      "grad_norm": 1.3991073369979858,
      "learning_rate": 0.00019975141707973882,
      "loss": 1.8083,
      "step": 362
    },
    {
      "epoch": 0.05199828104856038,
      "grad_norm": 1.2664425373077393,
      "learning_rate": 0.00019974813687074626,
      "loss": 1.9279,
      "step": 363
    },
    {
      "epoch": 0.0521415270018622,
      "grad_norm": 1.3429018259048462,
      "learning_rate": 0.0001997448351884482,
      "loss": 1.8489,
      "step": 364
    },
    {
      "epoch": 0.05228477295516402,
      "grad_norm": 1.5050594806671143,
      "learning_rate": 0.00019974151203355545,
      "loss": 2.214,
      "step": 365
    },
    {
      "epoch": 0.05242801890846584,
      "grad_norm": 1.1761552095413208,
      "learning_rate": 0.00019973816740678335,
      "loss": 1.814,
      "step": 366
    },
    {
      "epoch": 0.05257126486176766,
      "grad_norm": 1.0740997791290283,
      "learning_rate": 0.00019973480130885195,
      "loss": 1.9163,
      "step": 367
    },
    {
      "epoch": 0.052714510815069476,
      "grad_norm": 1.1985816955566406,
      "learning_rate": 0.00019973141374048588,
      "loss": 1.9949,
      "step": 368
    },
    {
      "epoch": 0.052857756768371296,
      "grad_norm": 1.2558863162994385,
      "learning_rate": 0.0001997280047024144,
      "loss": 1.7374,
      "step": 369
    },
    {
      "epoch": 0.053001002721673116,
      "grad_norm": 1.0561991930007935,
      "learning_rate": 0.00019972457419537143,
      "loss": 1.8152,
      "step": 370
    },
    {
      "epoch": 0.053144248674974935,
      "grad_norm": 1.1805089712142944,
      "learning_rate": 0.0001997211222200954,
      "loss": 1.859,
      "step": 371
    },
    {
      "epoch": 0.05328749462827675,
      "grad_norm": 1.220906138420105,
      "learning_rate": 0.0001997176487773295,
      "loss": 1.8891,
      "step": 372
    },
    {
      "epoch": 0.05343074058157857,
      "grad_norm": 1.2898125648498535,
      "learning_rate": 0.0001997141538678214,
      "loss": 2.0135,
      "step": 373
    },
    {
      "epoch": 0.05357398653488039,
      "grad_norm": 1.3417600393295288,
      "learning_rate": 0.00019971063749232352,
      "loss": 1.7994,
      "step": 374
    },
    {
      "epoch": 0.05371723248818221,
      "grad_norm": 1.2672240734100342,
      "learning_rate": 0.0001997070996515929,
      "loss": 2.0835,
      "step": 375
    },
    {
      "epoch": 0.053860478441484026,
      "grad_norm": 1.6356720924377441,
      "learning_rate": 0.00019970354034639103,
      "loss": 1.8802,
      "step": 376
    },
    {
      "epoch": 0.054003724394785846,
      "grad_norm": 1.3964208364486694,
      "learning_rate": 0.00019969995957748422,
      "loss": 1.9958,
      "step": 377
    },
    {
      "epoch": 0.054146970348087665,
      "grad_norm": 1.1695609092712402,
      "learning_rate": 0.0001996963573456433,
      "loss": 1.9214,
      "step": 378
    },
    {
      "epoch": 0.054290216301389485,
      "grad_norm": 1.5725616216659546,
      "learning_rate": 0.0001996927336516438,
      "loss": 2.0466,
      "step": 379
    },
    {
      "epoch": 0.054433462254691305,
      "grad_norm": 1.2054401636123657,
      "learning_rate": 0.00019968908849626567,
      "loss": 2.0911,
      "step": 380
    },
    {
      "epoch": 0.054576708207993124,
      "grad_norm": 1.2492305040359497,
      "learning_rate": 0.00019968542188029376,
      "loss": 1.9562,
      "step": 381
    },
    {
      "epoch": 0.054719954161294944,
      "grad_norm": 1.084065556526184,
      "learning_rate": 0.00019968173380451734,
      "loss": 1.8762,
      "step": 382
    },
    {
      "epoch": 0.054863200114596763,
      "grad_norm": 1.216033935546875,
      "learning_rate": 0.00019967802426973039,
      "loss": 1.7642,
      "step": 383
    },
    {
      "epoch": 0.05500644606789858,
      "grad_norm": 1.237306833267212,
      "learning_rate": 0.00019967429327673143,
      "loss": 2.0107,
      "step": 384
    },
    {
      "epoch": 0.0551496920212004,
      "grad_norm": 1.070798397064209,
      "learning_rate": 0.0001996705408263237,
      "loss": 1.8986,
      "step": 385
    },
    {
      "epoch": 0.05529293797450222,
      "grad_norm": 1.2178452014923096,
      "learning_rate": 0.00019966676691931498,
      "loss": 2.0638,
      "step": 386
    },
    {
      "epoch": 0.05543618392780404,
      "grad_norm": 1.8245280981063843,
      "learning_rate": 0.00019966297155651775,
      "loss": 2.0658,
      "step": 387
    },
    {
      "epoch": 0.05557942988110586,
      "grad_norm": 1.1085350513458252,
      "learning_rate": 0.00019965915473874898,
      "loss": 1.9441,
      "step": 388
    },
    {
      "epoch": 0.05572267583440768,
      "grad_norm": 1.253321886062622,
      "learning_rate": 0.0001996553164668304,
      "loss": 1.9066,
      "step": 389
    },
    {
      "epoch": 0.055865921787709494,
      "grad_norm": 1.1831797361373901,
      "learning_rate": 0.00019965145674158825,
      "loss": 1.9374,
      "step": 390
    },
    {
      "epoch": 0.05600916774101131,
      "grad_norm": 1.2637805938720703,
      "learning_rate": 0.00019964757556385348,
      "loss": 2.1155,
      "step": 391
    },
    {
      "epoch": 0.05615241369431313,
      "grad_norm": 1.1922460794448853,
      "learning_rate": 0.00019964367293446154,
      "loss": 2.0441,
      "step": 392
    },
    {
      "epoch": 0.05629565964761495,
      "grad_norm": 1.1540061235427856,
      "learning_rate": 0.00019963974885425266,
      "loss": 1.9592,
      "step": 393
    },
    {
      "epoch": 0.05643890560091677,
      "grad_norm": 1.3633723258972168,
      "learning_rate": 0.00019963580332407155,
      "loss": 1.8697,
      "step": 394
    },
    {
      "epoch": 0.05658215155421859,
      "grad_norm": 1.4259278774261475,
      "learning_rate": 0.00019963183634476756,
      "loss": 1.7931,
      "step": 395
    },
    {
      "epoch": 0.05672539750752041,
      "grad_norm": 1.4136995077133179,
      "learning_rate": 0.00019962784791719473,
      "loss": 1.9476,
      "step": 396
    },
    {
      "epoch": 0.05686864346082223,
      "grad_norm": 1.2864242792129517,
      "learning_rate": 0.00019962383804221164,
      "loss": 1.7351,
      "step": 397
    },
    {
      "epoch": 0.05701188941412405,
      "grad_norm": 1.0674023628234863,
      "learning_rate": 0.00019961980672068155,
      "loss": 1.9391,
      "step": 398
    },
    {
      "epoch": 0.05715513536742587,
      "grad_norm": 1.445884346961975,
      "learning_rate": 0.00019961575395347224,
      "loss": 1.8515,
      "step": 399
    },
    {
      "epoch": 0.05729838132072769,
      "grad_norm": 1.2526741027832031,
      "learning_rate": 0.00019961167974145625,
      "loss": 2.0766,
      "step": 400
    },
    {
      "epoch": 0.05744162727402951,
      "grad_norm": 1.2238144874572754,
      "learning_rate": 0.0001996075840855106,
      "loss": 1.785,
      "step": 401
    },
    {
      "epoch": 0.05758487322733133,
      "grad_norm": 1.074501395225525,
      "learning_rate": 0.000199603466986517,
      "loss": 1.7581,
      "step": 402
    },
    {
      "epoch": 0.05772811918063315,
      "grad_norm": 1.2030830383300781,
      "learning_rate": 0.00019959932844536177,
      "loss": 1.8801,
      "step": 403
    },
    {
      "epoch": 0.05787136513393497,
      "grad_norm": 1.5921425819396973,
      "learning_rate": 0.0001995951684629358,
      "loss": 1.7772,
      "step": 404
    },
    {
      "epoch": 0.05801461108723679,
      "grad_norm": 1.1489605903625488,
      "learning_rate": 0.0001995909870401347,
      "loss": 1.8793,
      "step": 405
    },
    {
      "epoch": 0.05815785704053861,
      "grad_norm": 1.4695888757705688,
      "learning_rate": 0.00019958678417785858,
      "loss": 1.7537,
      "step": 406
    },
    {
      "epoch": 0.05830110299384043,
      "grad_norm": 1.2490322589874268,
      "learning_rate": 0.00019958255987701217,
      "loss": 1.8742,
      "step": 407
    },
    {
      "epoch": 0.05844434894714224,
      "grad_norm": 1.002428412437439,
      "learning_rate": 0.00019957831413850493,
      "loss": 1.919,
      "step": 408
    },
    {
      "epoch": 0.05858759490044406,
      "grad_norm": 1.371732473373413,
      "learning_rate": 0.00019957404696325086,
      "loss": 1.8677,
      "step": 409
    },
    {
      "epoch": 0.05873084085374588,
      "grad_norm": 1.3012187480926514,
      "learning_rate": 0.00019956975835216852,
      "loss": 1.8269,
      "step": 410
    },
    {
      "epoch": 0.0588740868070477,
      "grad_norm": 1.0814388990402222,
      "learning_rate": 0.0001995654483061812,
      "loss": 1.994,
      "step": 411
    },
    {
      "epoch": 0.05901733276034952,
      "grad_norm": 1.2696882486343384,
      "learning_rate": 0.00019956111682621673,
      "loss": 1.9509,
      "step": 412
    },
    {
      "epoch": 0.05916057871365134,
      "grad_norm": 1.4992769956588745,
      "learning_rate": 0.00019955676391320752,
      "loss": 1.7334,
      "step": 413
    },
    {
      "epoch": 0.05930382466695316,
      "grad_norm": 1.1066880226135254,
      "learning_rate": 0.00019955238956809073,
      "loss": 2.0837,
      "step": 414
    },
    {
      "epoch": 0.05944707062025498,
      "grad_norm": 0.9836655855178833,
      "learning_rate": 0.00019954799379180797,
      "loss": 1.9481,
      "step": 415
    },
    {
      "epoch": 0.0595903165735568,
      "grad_norm": 1.1411279439926147,
      "learning_rate": 0.0001995435765853056,
      "loss": 1.9364,
      "step": 416
    },
    {
      "epoch": 0.059733562526858616,
      "grad_norm": 1.1237086057662964,
      "learning_rate": 0.00019953913794953451,
      "loss": 1.9559,
      "step": 417
    },
    {
      "epoch": 0.059876808480160436,
      "grad_norm": 1.1633782386779785,
      "learning_rate": 0.0001995346778854502,
      "loss": 1.6833,
      "step": 418
    },
    {
      "epoch": 0.060020054433462255,
      "grad_norm": 1.5859320163726807,
      "learning_rate": 0.00019953019639401288,
      "loss": 1.8446,
      "step": 419
    },
    {
      "epoch": 0.060163300386764075,
      "grad_norm": 1.499493956565857,
      "learning_rate": 0.00019952569347618727,
      "loss": 1.9842,
      "step": 420
    },
    {
      "epoch": 0.060306546340065895,
      "grad_norm": 1.256356120109558,
      "learning_rate": 0.0001995211691329427,
      "loss": 1.8953,
      "step": 421
    },
    {
      "epoch": 0.060449792293367714,
      "grad_norm": 1.6246408224105835,
      "learning_rate": 0.00019951662336525322,
      "loss": 1.9688,
      "step": 422
    },
    {
      "epoch": 0.060593038246669534,
      "grad_norm": 1.554673433303833,
      "learning_rate": 0.00019951205617409735,
      "loss": 1.9671,
      "step": 423
    },
    {
      "epoch": 0.060736284199971353,
      "grad_norm": 1.3998842239379883,
      "learning_rate": 0.00019950746756045834,
      "loss": 2.148,
      "step": 424
    },
    {
      "epoch": 0.06087953015327317,
      "grad_norm": 1.192581295967102,
      "learning_rate": 0.00019950285752532397,
      "loss": 2.0603,
      "step": 425
    },
    {
      "epoch": 0.06102277610657499,
      "grad_norm": 1.3526045083999634,
      "learning_rate": 0.00019949822606968673,
      "loss": 2.0272,
      "step": 426
    },
    {
      "epoch": 0.061166022059876805,
      "grad_norm": 1.4473577737808228,
      "learning_rate": 0.0001994935731945436,
      "loss": 1.8315,
      "step": 427
    },
    {
      "epoch": 0.061309268013178625,
      "grad_norm": 1.125598430633545,
      "learning_rate": 0.00019948889890089623,
      "loss": 2.0276,
      "step": 428
    },
    {
      "epoch": 0.061452513966480445,
      "grad_norm": 1.403790831565857,
      "learning_rate": 0.00019948420318975093,
      "loss": 2.0805,
      "step": 429
    },
    {
      "epoch": 0.061595759919782264,
      "grad_norm": 1.4935063123703003,
      "learning_rate": 0.0001994794860621185,
      "loss": 1.9328,
      "step": 430
    },
    {
      "epoch": 0.061739005873084084,
      "grad_norm": 1.2201225757598877,
      "learning_rate": 0.00019947474751901448,
      "loss": 1.8631,
      "step": 431
    },
    {
      "epoch": 0.0618822518263859,
      "grad_norm": 1.3769631385803223,
      "learning_rate": 0.0001994699875614589,
      "loss": 2.1091,
      "step": 432
    },
    {
      "epoch": 0.06202549777968772,
      "grad_norm": 1.2667357921600342,
      "learning_rate": 0.00019946520619047652,
      "loss": 1.8435,
      "step": 433
    },
    {
      "epoch": 0.06216874373298954,
      "grad_norm": 1.0790140628814697,
      "learning_rate": 0.00019946040340709665,
      "loss": 1.9343,
      "step": 434
    },
    {
      "epoch": 0.06231198968629136,
      "grad_norm": 1.2820956707000732,
      "learning_rate": 0.00019945557921235318,
      "loss": 1.9031,
      "step": 435
    },
    {
      "epoch": 0.06245523563959318,
      "grad_norm": 1.2308553457260132,
      "learning_rate": 0.00019945073360728462,
      "loss": 1.947,
      "step": 436
    },
    {
      "epoch": 0.062598481592895,
      "grad_norm": 1.1657029390335083,
      "learning_rate": 0.00019944586659293416,
      "loss": 1.8207,
      "step": 437
    },
    {
      "epoch": 0.06274172754619682,
      "grad_norm": 1.113720417022705,
      "learning_rate": 0.00019944097817034952,
      "loss": 1.9612,
      "step": 438
    },
    {
      "epoch": 0.06288497349949863,
      "grad_norm": 1.4692860841751099,
      "learning_rate": 0.00019943606834058305,
      "loss": 1.9536,
      "step": 439
    },
    {
      "epoch": 0.06302821945280046,
      "grad_norm": 1.2908706665039062,
      "learning_rate": 0.00019943113710469177,
      "loss": 1.7887,
      "step": 440
    },
    {
      "epoch": 0.06317146540610227,
      "grad_norm": 1.5681798458099365,
      "learning_rate": 0.00019942618446373717,
      "loss": 1.8364,
      "step": 441
    },
    {
      "epoch": 0.0633147113594041,
      "grad_norm": 1.4053391218185425,
      "learning_rate": 0.0001994212104187855,
      "loss": 1.998,
      "step": 442
    },
    {
      "epoch": 0.06345795731270591,
      "grad_norm": 1.3299514055252075,
      "learning_rate": 0.00019941621497090748,
      "loss": 1.7609,
      "step": 443
    },
    {
      "epoch": 0.06360120326600774,
      "grad_norm": 1.3105353116989136,
      "learning_rate": 0.0001994111981211786,
      "loss": 1.8204,
      "step": 444
    },
    {
      "epoch": 0.06374444921930955,
      "grad_norm": 1.3539772033691406,
      "learning_rate": 0.00019940615987067877,
      "loss": 2.0658,
      "step": 445
    },
    {
      "epoch": 0.06388769517261138,
      "grad_norm": 1.2422757148742676,
      "learning_rate": 0.00019940110022049264,
      "loss": 2.0037,
      "step": 446
    },
    {
      "epoch": 0.06403094112591319,
      "grad_norm": 1.1371114253997803,
      "learning_rate": 0.00019939601917170943,
      "loss": 1.8564,
      "step": 447
    },
    {
      "epoch": 0.06417418707921502,
      "grad_norm": 1.4853485822677612,
      "learning_rate": 0.00019939091672542295,
      "loss": 2.0655,
      "step": 448
    },
    {
      "epoch": 0.06431743303251683,
      "grad_norm": 1.2365187406539917,
      "learning_rate": 0.00019938579288273167,
      "loss": 1.9946,
      "step": 449
    },
    {
      "epoch": 0.06446067898581866,
      "grad_norm": 1.442976713180542,
      "learning_rate": 0.0001993806476447386,
      "loss": 1.92,
      "step": 450
    },
    {
      "epoch": 0.06460392493912047,
      "grad_norm": 1.5670967102050781,
      "learning_rate": 0.00019937548101255139,
      "loss": 1.8682,
      "step": 451
    },
    {
      "epoch": 0.0647471708924223,
      "grad_norm": 1.5802968740463257,
      "learning_rate": 0.00019937029298728226,
      "loss": 1.8258,
      "step": 452
    },
    {
      "epoch": 0.06489041684572411,
      "grad_norm": 1.2011916637420654,
      "learning_rate": 0.00019936508357004806,
      "loss": 1.7282,
      "step": 453
    },
    {
      "epoch": 0.06503366279902592,
      "grad_norm": 1.4225175380706787,
      "learning_rate": 0.0001993598527619703,
      "loss": 1.8073,
      "step": 454
    },
    {
      "epoch": 0.06517690875232775,
      "grad_norm": 1.2416908740997314,
      "learning_rate": 0.000199354600564175,
      "loss": 1.7272,
      "step": 455
    },
    {
      "epoch": 0.06532015470562956,
      "grad_norm": 1.1273763179779053,
      "learning_rate": 0.00019934932697779288,
      "loss": 1.8676,
      "step": 456
    },
    {
      "epoch": 0.06546340065893139,
      "grad_norm": 1.3882946968078613,
      "learning_rate": 0.00019934403200395914,
      "loss": 2.0666,
      "step": 457
    },
    {
      "epoch": 0.0656066466122332,
      "grad_norm": 1.3472100496292114,
      "learning_rate": 0.0001993387156438137,
      "loss": 1.8852,
      "step": 458
    },
    {
      "epoch": 0.06574989256553503,
      "grad_norm": 1.3469574451446533,
      "learning_rate": 0.000199333377898501,
      "loss": 1.6828,
      "step": 459
    },
    {
      "epoch": 0.06589313851883684,
      "grad_norm": 1.3398144245147705,
      "learning_rate": 0.00019932801876917022,
      "loss": 1.9105,
      "step": 460
    },
    {
      "epoch": 0.06603638447213867,
      "grad_norm": 1.6499736309051514,
      "learning_rate": 0.00019932263825697498,
      "loss": 1.899,
      "step": 461
    },
    {
      "epoch": 0.06617963042544048,
      "grad_norm": 1.3724581003189087,
      "learning_rate": 0.00019931723636307353,
      "loss": 1.9462,
      "step": 462
    },
    {
      "epoch": 0.0663228763787423,
      "grad_norm": 1.4621167182922363,
      "learning_rate": 0.00019931181308862885,
      "loss": 2.0525,
      "step": 463
    },
    {
      "epoch": 0.06646612233204412,
      "grad_norm": 1.922086477279663,
      "learning_rate": 0.00019930636843480838,
      "loss": 1.74,
      "step": 464
    },
    {
      "epoch": 0.06660936828534594,
      "grad_norm": 1.217824101448059,
      "learning_rate": 0.00019930090240278422,
      "loss": 1.912,
      "step": 465
    },
    {
      "epoch": 0.06675261423864776,
      "grad_norm": 1.247957706451416,
      "learning_rate": 0.0001992954149937331,
      "loss": 1.8527,
      "step": 466
    },
    {
      "epoch": 0.06689586019194958,
      "grad_norm": 1.1887818574905396,
      "learning_rate": 0.00019928990620883633,
      "loss": 1.8,
      "step": 467
    },
    {
      "epoch": 0.0670391061452514,
      "grad_norm": 1.1378886699676514,
      "learning_rate": 0.0001992843760492798,
      "loss": 1.9488,
      "step": 468
    },
    {
      "epoch": 0.06718235209855322,
      "grad_norm": 1.30185866355896,
      "learning_rate": 0.00019927882451625402,
      "loss": 1.9737,
      "step": 469
    },
    {
      "epoch": 0.06732559805185503,
      "grad_norm": 1.2080559730529785,
      "learning_rate": 0.00019927325161095408,
      "loss": 1.9244,
      "step": 470
    },
    {
      "epoch": 0.06746884400515686,
      "grad_norm": 1.2855666875839233,
      "learning_rate": 0.0001992676573345797,
      "loss": 2.0461,
      "step": 471
    },
    {
      "epoch": 0.06761208995845867,
      "grad_norm": 1.2975670099258423,
      "learning_rate": 0.00019926204168833515,
      "loss": 1.8016,
      "step": 472
    },
    {
      "epoch": 0.06775533591176049,
      "grad_norm": 1.0729694366455078,
      "learning_rate": 0.0001992564046734294,
      "loss": 1.822,
      "step": 473
    },
    {
      "epoch": 0.06789858186506231,
      "grad_norm": 1.769511342048645,
      "learning_rate": 0.000199250746291076,
      "loss": 1.9553,
      "step": 474
    },
    {
      "epoch": 0.06804182781836413,
      "grad_norm": 1.524078369140625,
      "learning_rate": 0.00019924506654249292,
      "loss": 1.7481,
      "step": 475
    },
    {
      "epoch": 0.06818507377166595,
      "grad_norm": 1.4613440036773682,
      "learning_rate": 0.00019923936542890297,
      "loss": 1.8721,
      "step": 476
    },
    {
      "epoch": 0.06832831972496776,
      "grad_norm": 1.281390905380249,
      "learning_rate": 0.00019923364295153343,
      "loss": 1.9589,
      "step": 477
    },
    {
      "epoch": 0.06847156567826959,
      "grad_norm": 1.8698230981826782,
      "learning_rate": 0.00019922789911161622,
      "loss": 1.8523,
      "step": 478
    },
    {
      "epoch": 0.0686148116315714,
      "grad_norm": 1.3571187257766724,
      "learning_rate": 0.00019922213391038786,
      "loss": 1.857,
      "step": 479
    },
    {
      "epoch": 0.06875805758487323,
      "grad_norm": 1.3558197021484375,
      "learning_rate": 0.00019921634734908943,
      "loss": 1.8722,
      "step": 480
    },
    {
      "epoch": 0.06890130353817504,
      "grad_norm": 1.2936092615127563,
      "learning_rate": 0.00019921053942896665,
      "loss": 1.9151,
      "step": 481
    },
    {
      "epoch": 0.06904454949147687,
      "grad_norm": 1.6655749082565308,
      "learning_rate": 0.0001992047101512698,
      "loss": 1.9252,
      "step": 482
    },
    {
      "epoch": 0.06918779544477868,
      "grad_norm": 1.4114404916763306,
      "learning_rate": 0.00019919885951725374,
      "loss": 1.9171,
      "step": 483
    },
    {
      "epoch": 0.06933104139808051,
      "grad_norm": 1.4054932594299316,
      "learning_rate": 0.00019919298752817807,
      "loss": 1.7685,
      "step": 484
    },
    {
      "epoch": 0.06947428735138232,
      "grad_norm": 1.5407607555389404,
      "learning_rate": 0.0001991870941853068,
      "loss": 1.9321,
      "step": 485
    },
    {
      "epoch": 0.06961753330468415,
      "grad_norm": 1.5114518404006958,
      "learning_rate": 0.00019918117948990868,
      "loss": 1.681,
      "step": 486
    },
    {
      "epoch": 0.06976077925798596,
      "grad_norm": 1.1382354497909546,
      "learning_rate": 0.00019917524344325696,
      "loss": 1.9091,
      "step": 487
    },
    {
      "epoch": 0.06990402521128779,
      "grad_norm": 1.609876036643982,
      "learning_rate": 0.00019916928604662952,
      "loss": 1.8429,
      "step": 488
    },
    {
      "epoch": 0.0700472711645896,
      "grad_norm": 1.2324614524841309,
      "learning_rate": 0.0001991633073013089,
      "loss": 1.9848,
      "step": 489
    },
    {
      "epoch": 0.07019051711789141,
      "grad_norm": 1.5840328931808472,
      "learning_rate": 0.00019915730720858208,
      "loss": 1.8343,
      "step": 490
    },
    {
      "epoch": 0.07033376307119324,
      "grad_norm": 1.394176959991455,
      "learning_rate": 0.00019915128576974077,
      "loss": 1.874,
      "step": 491
    },
    {
      "epoch": 0.07047700902449505,
      "grad_norm": 1.4741060733795166,
      "learning_rate": 0.00019914524298608128,
      "loss": 1.9982,
      "step": 492
    },
    {
      "epoch": 0.07062025497779688,
      "grad_norm": 1.872179388999939,
      "learning_rate": 0.00019913917885890442,
      "loss": 1.87,
      "step": 493
    },
    {
      "epoch": 0.07076350093109869,
      "grad_norm": 1.457639217376709,
      "learning_rate": 0.00019913309338951568,
      "loss": 1.8309,
      "step": 494
    },
    {
      "epoch": 0.07090674688440052,
      "grad_norm": 1.4248993396759033,
      "learning_rate": 0.00019912698657922507,
      "loss": 1.8919,
      "step": 495
    },
    {
      "epoch": 0.07104999283770233,
      "grad_norm": 1.6535437107086182,
      "learning_rate": 0.0001991208584293473,
      "loss": 1.8902,
      "step": 496
    },
    {
      "epoch": 0.07119323879100416,
      "grad_norm": 1.2773194313049316,
      "learning_rate": 0.00019911470894120155,
      "loss": 1.7948,
      "step": 497
    },
    {
      "epoch": 0.07133648474430597,
      "grad_norm": 1.1379121541976929,
      "learning_rate": 0.0001991085381161117,
      "loss": 1.839,
      "step": 498
    },
    {
      "epoch": 0.0714797306976078,
      "grad_norm": 1.303226351737976,
      "learning_rate": 0.0001991023459554061,
      "loss": 1.8876,
      "step": 499
    },
    {
      "epoch": 0.07162297665090961,
      "grad_norm": 1.3400683403015137,
      "learning_rate": 0.00019909613246041788,
      "loss": 1.881,
      "step": 500
    },
    {
      "epoch": 0.07176622260421144,
      "grad_norm": 1.4628103971481323,
      "learning_rate": 0.00019908989763248454,
      "loss": 1.8523,
      "step": 501
    },
    {
      "epoch": 0.07190946855751325,
      "grad_norm": 1.6308717727661133,
      "learning_rate": 0.00019908364147294835,
      "loss": 1.7968,
      "step": 502
    },
    {
      "epoch": 0.07205271451081507,
      "grad_norm": 1.3459229469299316,
      "learning_rate": 0.00019907736398315607,
      "loss": 1.7511,
      "step": 503
    },
    {
      "epoch": 0.07219596046411689,
      "grad_norm": 1.5009410381317139,
      "learning_rate": 0.00019907106516445913,
      "loss": 1.916,
      "step": 504
    },
    {
      "epoch": 0.07233920641741871,
      "grad_norm": 1.1755175590515137,
      "learning_rate": 0.00019906474501821348,
      "loss": 1.9279,
      "step": 505
    },
    {
      "epoch": 0.07248245237072053,
      "grad_norm": 1.5116721391677856,
      "learning_rate": 0.00019905840354577972,
      "loss": 1.81,
      "step": 506
    },
    {
      "epoch": 0.07262569832402235,
      "grad_norm": 1.4349769353866577,
      "learning_rate": 0.00019905204074852298,
      "loss": 1.8853,
      "step": 507
    },
    {
      "epoch": 0.07276894427732417,
      "grad_norm": 1.420134425163269,
      "learning_rate": 0.000199045656627813,
      "loss": 1.7951,
      "step": 508
    },
    {
      "epoch": 0.07291219023062598,
      "grad_norm": 1.383614182472229,
      "learning_rate": 0.00019903925118502417,
      "loss": 1.7075,
      "step": 509
    },
    {
      "epoch": 0.0730554361839278,
      "grad_norm": 1.4356929063796997,
      "learning_rate": 0.0001990328244215354,
      "loss": 1.8192,
      "step": 510
    },
    {
      "epoch": 0.07319868213722962,
      "grad_norm": 1.1821198463439941,
      "learning_rate": 0.00019902637633873021,
      "loss": 1.9616,
      "step": 511
    },
    {
      "epoch": 0.07334192809053144,
      "grad_norm": 1.1920559406280518,
      "learning_rate": 0.00019901990693799672,
      "loss": 1.767,
      "step": 512
    },
    {
      "epoch": 0.07348517404383326,
      "grad_norm": 1.2896835803985596,
      "learning_rate": 0.00019901341622072764,
      "loss": 2.155,
      "step": 513
    },
    {
      "epoch": 0.07362841999713508,
      "grad_norm": 1.6448336839675903,
      "learning_rate": 0.00019900690418832022,
      "loss": 1.6994,
      "step": 514
    },
    {
      "epoch": 0.0737716659504369,
      "grad_norm": 1.4044462442398071,
      "learning_rate": 0.00019900037084217637,
      "loss": 1.7893,
      "step": 515
    },
    {
      "epoch": 0.07391491190373872,
      "grad_norm": 1.4587130546569824,
      "learning_rate": 0.0001989938161837026,
      "loss": 1.8875,
      "step": 516
    },
    {
      "epoch": 0.07405815785704054,
      "grad_norm": 1.2490543127059937,
      "learning_rate": 0.0001989872402143099,
      "loss": 1.7479,
      "step": 517
    },
    {
      "epoch": 0.07420140381034236,
      "grad_norm": 1.501863956451416,
      "learning_rate": 0.00019898064293541392,
      "loss": 1.9654,
      "step": 518
    },
    {
      "epoch": 0.07434464976364417,
      "grad_norm": 1.215704321861267,
      "learning_rate": 0.0001989740243484349,
      "loss": 1.8862,
      "step": 519
    },
    {
      "epoch": 0.074487895716946,
      "grad_norm": 1.5387451648712158,
      "learning_rate": 0.00019896738445479768,
      "loss": 1.7493,
      "step": 520
    },
    {
      "epoch": 0.07463114167024781,
      "grad_norm": 1.77334725856781,
      "learning_rate": 0.00019896072325593166,
      "loss": 1.9767,
      "step": 521
    },
    {
      "epoch": 0.07477438762354964,
      "grad_norm": 1.5325254201889038,
      "learning_rate": 0.0001989540407532708,
      "loss": 1.7749,
      "step": 522
    },
    {
      "epoch": 0.07491763357685145,
      "grad_norm": 1.309705376625061,
      "learning_rate": 0.00019894733694825374,
      "loss": 1.8098,
      "step": 523
    },
    {
      "epoch": 0.07506087953015328,
      "grad_norm": 1.413210391998291,
      "learning_rate": 0.00019894061184232357,
      "loss": 1.6865,
      "step": 524
    },
    {
      "epoch": 0.07520412548345509,
      "grad_norm": 1.2971785068511963,
      "learning_rate": 0.00019893386543692806,
      "loss": 2.0,
      "step": 525
    },
    {
      "epoch": 0.0753473714367569,
      "grad_norm": 1.5852457284927368,
      "learning_rate": 0.00019892709773351958,
      "loss": 1.7778,
      "step": 526
    },
    {
      "epoch": 0.07549061739005873,
      "grad_norm": 1.5320847034454346,
      "learning_rate": 0.00019892030873355497,
      "loss": 1.6361,
      "step": 527
    },
    {
      "epoch": 0.07563386334336054,
      "grad_norm": 1.4010851383209229,
      "learning_rate": 0.00019891349843849585,
      "loss": 1.7991,
      "step": 528
    },
    {
      "epoch": 0.07577710929666237,
      "grad_norm": 1.3538016080856323,
      "learning_rate": 0.0001989066668498082,
      "loss": 1.8997,
      "step": 529
    },
    {
      "epoch": 0.07592035524996418,
      "grad_norm": 1.4806427955627441,
      "learning_rate": 0.00019889981396896278,
      "loss": 1.8361,
      "step": 530
    },
    {
      "epoch": 0.07606360120326601,
      "grad_norm": 1.559334635734558,
      "learning_rate": 0.00019889293979743477,
      "loss": 1.6746,
      "step": 531
    },
    {
      "epoch": 0.07620684715656782,
      "grad_norm": 1.3477094173431396,
      "learning_rate": 0.00019888604433670405,
      "loss": 1.8048,
      "step": 532
    },
    {
      "epoch": 0.07635009310986965,
      "grad_norm": 1.4291020631790161,
      "learning_rate": 0.00019887912758825504,
      "loss": 1.9826,
      "step": 533
    },
    {
      "epoch": 0.07649333906317146,
      "grad_norm": 1.3956358432769775,
      "learning_rate": 0.00019887218955357675,
      "loss": 1.8392,
      "step": 534
    },
    {
      "epoch": 0.07663658501647329,
      "grad_norm": 1.3596224784851074,
      "learning_rate": 0.00019886523023416274,
      "loss": 1.8294,
      "step": 535
    },
    {
      "epoch": 0.0767798309697751,
      "grad_norm": 1.344411849975586,
      "learning_rate": 0.00019885824963151118,
      "loss": 1.9847,
      "step": 536
    },
    {
      "epoch": 0.07692307692307693,
      "grad_norm": 1.5432337522506714,
      "learning_rate": 0.00019885124774712487,
      "loss": 1.8735,
      "step": 537
    },
    {
      "epoch": 0.07706632287637874,
      "grad_norm": 1.4612058401107788,
      "learning_rate": 0.00019884422458251109,
      "loss": 1.8228,
      "step": 538
    },
    {
      "epoch": 0.07720956882968057,
      "grad_norm": 1.412908673286438,
      "learning_rate": 0.00019883718013918178,
      "loss": 1.8775,
      "step": 539
    },
    {
      "epoch": 0.07735281478298238,
      "grad_norm": 1.836916446685791,
      "learning_rate": 0.00019883011441865342,
      "loss": 1.938,
      "step": 540
    },
    {
      "epoch": 0.0774960607362842,
      "grad_norm": 1.7175601720809937,
      "learning_rate": 0.0001988230274224471,
      "loss": 1.6055,
      "step": 541
    },
    {
      "epoch": 0.07763930668958602,
      "grad_norm": 1.8685137033462524,
      "learning_rate": 0.00019881591915208845,
      "loss": 1.7533,
      "step": 542
    },
    {
      "epoch": 0.07778255264288784,
      "grad_norm": 1.7033354043960571,
      "learning_rate": 0.00019880878960910772,
      "loss": 1.8844,
      "step": 543
    },
    {
      "epoch": 0.07792579859618966,
      "grad_norm": 1.3682303428649902,
      "learning_rate": 0.00019880163879503973,
      "loss": 1.857,
      "step": 544
    },
    {
      "epoch": 0.07806904454949147,
      "grad_norm": 1.589799404144287,
      "learning_rate": 0.00019879446671142387,
      "loss": 2.0505,
      "step": 545
    },
    {
      "epoch": 0.0782122905027933,
      "grad_norm": 1.4113374948501587,
      "learning_rate": 0.00019878727335980413,
      "loss": 1.7401,
      "step": 546
    },
    {
      "epoch": 0.07835553645609511,
      "grad_norm": 1.4761561155319214,
      "learning_rate": 0.000198780058741729,
      "loss": 1.7849,
      "step": 547
    },
    {
      "epoch": 0.07849878240939694,
      "grad_norm": 1.312427282333374,
      "learning_rate": 0.0001987728228587517,
      "loss": 1.8672,
      "step": 548
    },
    {
      "epoch": 0.07864202836269875,
      "grad_norm": 1.3250832557678223,
      "learning_rate": 0.00019876556571242986,
      "loss": 1.7832,
      "step": 549
    },
    {
      "epoch": 0.07878527431600058,
      "grad_norm": 1.4705543518066406,
      "learning_rate": 0.0001987582873043258,
      "loss": 1.7986,
      "step": 550
    },
    {
      "epoch": 0.07892852026930239,
      "grad_norm": 1.3344213962554932,
      "learning_rate": 0.00019875098763600635,
      "loss": 1.9054,
      "step": 551
    },
    {
      "epoch": 0.07907176622260421,
      "grad_norm": 1.453491449356079,
      "learning_rate": 0.00019874366670904299,
      "loss": 1.6129,
      "step": 552
    },
    {
      "epoch": 0.07921501217590603,
      "grad_norm": 1.5356755256652832,
      "learning_rate": 0.00019873632452501168,
      "loss": 1.9653,
      "step": 553
    },
    {
      "epoch": 0.07935825812920785,
      "grad_norm": 1.9556617736816406,
      "learning_rate": 0.00019872896108549308,
      "loss": 1.7465,
      "step": 554
    },
    {
      "epoch": 0.07950150408250967,
      "grad_norm": 1.4671192169189453,
      "learning_rate": 0.00019872157639207232,
      "loss": 1.7662,
      "step": 555
    },
    {
      "epoch": 0.07964475003581149,
      "grad_norm": 1.8577529191970825,
      "learning_rate": 0.00019871417044633913,
      "loss": 1.8747,
      "step": 556
    },
    {
      "epoch": 0.0797879959891133,
      "grad_norm": 1.4766255617141724,
      "learning_rate": 0.00019870674324988781,
      "loss": 1.7978,
      "step": 557
    },
    {
      "epoch": 0.07993124194241513,
      "grad_norm": 1.6983734369277954,
      "learning_rate": 0.00019869929480431734,
      "loss": 1.7239,
      "step": 558
    },
    {
      "epoch": 0.08007448789571694,
      "grad_norm": 1.6475611925125122,
      "learning_rate": 0.00019869182511123113,
      "loss": 1.8591,
      "step": 559
    },
    {
      "epoch": 0.08021773384901877,
      "grad_norm": 2.0189778804779053,
      "learning_rate": 0.00019868433417223716,
      "loss": 1.8644,
      "step": 560
    },
    {
      "epoch": 0.08036097980232058,
      "grad_norm": 1.4063400030136108,
      "learning_rate": 0.00019867682198894816,
      "loss": 1.7548,
      "step": 561
    },
    {
      "epoch": 0.08050422575562241,
      "grad_norm": 1.583289623260498,
      "learning_rate": 0.00019866928856298119,
      "loss": 1.8043,
      "step": 562
    },
    {
      "epoch": 0.08064747170892422,
      "grad_norm": 1.55256986618042,
      "learning_rate": 0.00019866173389595813,
      "loss": 1.7083,
      "step": 563
    },
    {
      "epoch": 0.08079071766222604,
      "grad_norm": 1.4576290845870972,
      "learning_rate": 0.00019865415798950525,
      "loss": 1.9382,
      "step": 564
    },
    {
      "epoch": 0.08093396361552786,
      "grad_norm": 1.7685655355453491,
      "learning_rate": 0.00019864656084525345,
      "loss": 1.8558,
      "step": 565
    },
    {
      "epoch": 0.08107720956882968,
      "grad_norm": 2.316591501235962,
      "learning_rate": 0.00019863894246483822,
      "loss": 1.6276,
      "step": 566
    },
    {
      "epoch": 0.0812204555221315,
      "grad_norm": 1.6802818775177002,
      "learning_rate": 0.00019863130284989965,
      "loss": 1.8006,
      "step": 567
    },
    {
      "epoch": 0.08136370147543331,
      "grad_norm": 1.6730725765228271,
      "learning_rate": 0.0001986236420020823,
      "loss": 1.7016,
      "step": 568
    },
    {
      "epoch": 0.08150694742873514,
      "grad_norm": 1.8522088527679443,
      "learning_rate": 0.00019861595992303537,
      "loss": 1.7074,
      "step": 569
    },
    {
      "epoch": 0.08165019338203695,
      "grad_norm": 2.0142784118652344,
      "learning_rate": 0.00019860825661441266,
      "loss": 1.7981,
      "step": 570
    },
    {
      "epoch": 0.08179343933533878,
      "grad_norm": 1.587649941444397,
      "learning_rate": 0.00019860053207787246,
      "loss": 1.6049,
      "step": 571
    },
    {
      "epoch": 0.08193668528864059,
      "grad_norm": 1.9246728420257568,
      "learning_rate": 0.0001985927863150777,
      "loss": 1.7698,
      "step": 572
    },
    {
      "epoch": 0.08207993124194242,
      "grad_norm": 1.4553924798965454,
      "learning_rate": 0.0001985850193276958,
      "loss": 1.6821,
      "step": 573
    },
    {
      "epoch": 0.08222317719524423,
      "grad_norm": 1.7816945314407349,
      "learning_rate": 0.0001985772311173989,
      "loss": 1.6999,
      "step": 574
    },
    {
      "epoch": 0.08236642314854606,
      "grad_norm": 2.0161430835723877,
      "learning_rate": 0.00019856942168586353,
      "loss": 1.721,
      "step": 575
    },
    {
      "epoch": 0.08250966910184787,
      "grad_norm": 1.947036623954773,
      "learning_rate": 0.00019856159103477086,
      "loss": 1.6449,
      "step": 576
    },
    {
      "epoch": 0.0826529150551497,
      "grad_norm": 1.5800046920776367,
      "learning_rate": 0.00019855373916580668,
      "loss": 1.8299,
      "step": 577
    },
    {
      "epoch": 0.08279616100845151,
      "grad_norm": 1.7868382930755615,
      "learning_rate": 0.00019854586608066127,
      "loss": 1.7032,
      "step": 578
    },
    {
      "epoch": 0.08293940696175334,
      "grad_norm": 1.6116896867752075,
      "learning_rate": 0.00019853797178102952,
      "loss": 1.8266,
      "step": 579
    },
    {
      "epoch": 0.08308265291505515,
      "grad_norm": 1.9898512363433838,
      "learning_rate": 0.0001985300562686109,
      "loss": 1.6916,
      "step": 580
    },
    {
      "epoch": 0.08322589886835696,
      "grad_norm": 1.758905291557312,
      "learning_rate": 0.00019852211954510943,
      "loss": 1.7571,
      "step": 581
    },
    {
      "epoch": 0.08336914482165879,
      "grad_norm": 1.9583605527877808,
      "learning_rate": 0.0001985141616122336,
      "loss": 1.7946,
      "step": 582
    },
    {
      "epoch": 0.0835123907749606,
      "grad_norm": 2.251268148422241,
      "learning_rate": 0.00019850618247169667,
      "loss": 1.7275,
      "step": 583
    },
    {
      "epoch": 0.08365563672826243,
      "grad_norm": 1.9306137561798096,
      "learning_rate": 0.0001984981821252163,
      "loss": 1.6006,
      "step": 584
    },
    {
      "epoch": 0.08379888268156424,
      "grad_norm": 2.1751723289489746,
      "learning_rate": 0.00019849016057451476,
      "loss": 1.7321,
      "step": 585
    },
    {
      "epoch": 0.08394212863486607,
      "grad_norm": 1.992838740348816,
      "learning_rate": 0.00019848211782131888,
      "loss": 1.8335,
      "step": 586
    },
    {
      "epoch": 0.08408537458816788,
      "grad_norm": 2.430884599685669,
      "learning_rate": 0.00019847405386736014,
      "loss": 1.8502,
      "step": 587
    },
    {
      "epoch": 0.0842286205414697,
      "grad_norm": 1.7405625581741333,
      "learning_rate": 0.00019846596871437441,
      "loss": 1.6762,
      "step": 588
    },
    {
      "epoch": 0.08437186649477152,
      "grad_norm": 1.7707555294036865,
      "learning_rate": 0.00019845786236410227,
      "loss": 1.7992,
      "step": 589
    },
    {
      "epoch": 0.08451511244807335,
      "grad_norm": 2.1325972080230713,
      "learning_rate": 0.00019844973481828886,
      "loss": 1.703,
      "step": 590
    },
    {
      "epoch": 0.08465835840137516,
      "grad_norm": 2.0813379287719727,
      "learning_rate": 0.00019844158607868376,
      "loss": 1.6277,
      "step": 591
    },
    {
      "epoch": 0.08480160435467698,
      "grad_norm": 1.669659972190857,
      "learning_rate": 0.00019843341614704125,
      "loss": 1.806,
      "step": 592
    },
    {
      "epoch": 0.0849448503079788,
      "grad_norm": 2.092416763305664,
      "learning_rate": 0.00019842522502512008,
      "loss": 1.8738,
      "step": 593
    },
    {
      "epoch": 0.08508809626128062,
      "grad_norm": 1.4728245735168457,
      "learning_rate": 0.0001984170127146836,
      "loss": 1.8159,
      "step": 594
    },
    {
      "epoch": 0.08523134221458244,
      "grad_norm": 2.1759033203125,
      "learning_rate": 0.00019840877921749973,
      "loss": 1.7924,
      "step": 595
    },
    {
      "epoch": 0.08537458816788426,
      "grad_norm": 1.864659070968628,
      "learning_rate": 0.00019840052453534094,
      "loss": 1.8479,
      "step": 596
    },
    {
      "epoch": 0.08551783412118608,
      "grad_norm": 2.0173592567443848,
      "learning_rate": 0.00019839224866998424,
      "loss": 1.8085,
      "step": 597
    },
    {
      "epoch": 0.0856610800744879,
      "grad_norm": 1.7638306617736816,
      "learning_rate": 0.00019838395162321125,
      "loss": 1.866,
      "step": 598
    },
    {
      "epoch": 0.08580432602778972,
      "grad_norm": 1.443919062614441,
      "learning_rate": 0.0001983756333968081,
      "loss": 1.8115,
      "step": 599
    },
    {
      "epoch": 0.08594757198109153,
      "grad_norm": 1.8691120147705078,
      "learning_rate": 0.0001983672939925655,
      "loss": 1.6545,
      "step": 600
    },
    {
      "epoch": 0.08609081793439335,
      "grad_norm": 2.423405170440674,
      "learning_rate": 0.00019835893341227872,
      "loss": 1.7101,
      "step": 601
    },
    {
      "epoch": 0.08623406388769517,
      "grad_norm": 2.1188912391662598,
      "learning_rate": 0.00019835055165774756,
      "loss": 1.5544,
      "step": 602
    },
    {
      "epoch": 0.086377309840997,
      "grad_norm": 1.7928441762924194,
      "learning_rate": 0.00019834214873077643,
      "loss": 1.804,
      "step": 603
    },
    {
      "epoch": 0.0865205557942988,
      "grad_norm": 1.8053836822509766,
      "learning_rate": 0.00019833372463317427,
      "loss": 1.7467,
      "step": 604
    },
    {
      "epoch": 0.08666380174760063,
      "grad_norm": 2.1015563011169434,
      "learning_rate": 0.0001983252793667546,
      "loss": 1.599,
      "step": 605
    },
    {
      "epoch": 0.08680704770090245,
      "grad_norm": 1.765212059020996,
      "learning_rate": 0.00019831681293333545,
      "loss": 1.7277,
      "step": 606
    },
    {
      "epoch": 0.08695029365420427,
      "grad_norm": 1.675680160522461,
      "learning_rate": 0.00019830832533473939,
      "loss": 1.6561,
      "step": 607
    },
    {
      "epoch": 0.08709353960750608,
      "grad_norm": 2.0614521503448486,
      "learning_rate": 0.0001982998165727937,
      "loss": 1.7373,
      "step": 608
    },
    {
      "epoch": 0.08723678556080791,
      "grad_norm": 1.3164680004119873,
      "learning_rate": 0.00019829128664933002,
      "loss": 1.6432,
      "step": 609
    },
    {
      "epoch": 0.08738003151410972,
      "grad_norm": 1.9503114223480225,
      "learning_rate": 0.00019828273556618466,
      "loss": 1.7732,
      "step": 610
    },
    {
      "epoch": 0.08752327746741155,
      "grad_norm": 1.5411949157714844,
      "learning_rate": 0.00019827416332519844,
      "loss": 1.8937,
      "step": 611
    },
    {
      "epoch": 0.08766652342071336,
      "grad_norm": 1.9805026054382324,
      "learning_rate": 0.00019826556992821678,
      "loss": 1.7267,
      "step": 612
    },
    {
      "epoch": 0.08780976937401519,
      "grad_norm": 2.108997106552124,
      "learning_rate": 0.0001982569553770896,
      "loss": 1.6758,
      "step": 613
    },
    {
      "epoch": 0.087953015327317,
      "grad_norm": 1.4876041412353516,
      "learning_rate": 0.0001982483196736714,
      "loss": 1.7274,
      "step": 614
    },
    {
      "epoch": 0.08809626128061883,
      "grad_norm": 1.7633486986160278,
      "learning_rate": 0.00019823966281982128,
      "loss": 1.5463,
      "step": 615
    },
    {
      "epoch": 0.08823950723392064,
      "grad_norm": 1.9821653366088867,
      "learning_rate": 0.00019823098481740276,
      "loss": 1.6797,
      "step": 616
    },
    {
      "epoch": 0.08838275318722245,
      "grad_norm": 1.4849814176559448,
      "learning_rate": 0.0001982222856682841,
      "loss": 1.7138,
      "step": 617
    },
    {
      "epoch": 0.08852599914052428,
      "grad_norm": 2.2080535888671875,
      "learning_rate": 0.0001982135653743379,
      "loss": 1.7187,
      "step": 618
    },
    {
      "epoch": 0.0886692450938261,
      "grad_norm": 1.4369357824325562,
      "learning_rate": 0.00019820482393744152,
      "loss": 1.6414,
      "step": 619
    },
    {
      "epoch": 0.08881249104712792,
      "grad_norm": 1.4783408641815186,
      "learning_rate": 0.00019819606135947676,
      "loss": 1.5684,
      "step": 620
    },
    {
      "epoch": 0.08895573700042973,
      "grad_norm": 1.8696465492248535,
      "learning_rate": 0.00019818727764232992,
      "loss": 1.6218,
      "step": 621
    },
    {
      "epoch": 0.08909898295373156,
      "grad_norm": 1.7060953378677368,
      "learning_rate": 0.00019817847278789198,
      "loss": 1.6335,
      "step": 622
    },
    {
      "epoch": 0.08924222890703337,
      "grad_norm": 1.6891498565673828,
      "learning_rate": 0.00019816964679805838,
      "loss": 1.7383,
      "step": 623
    },
    {
      "epoch": 0.0893854748603352,
      "grad_norm": 2.057689905166626,
      "learning_rate": 0.00019816079967472915,
      "loss": 1.6232,
      "step": 624
    },
    {
      "epoch": 0.08952872081363701,
      "grad_norm": 1.500284194946289,
      "learning_rate": 0.00019815193141980886,
      "loss": 1.538,
      "step": 625
    },
    {
      "epoch": 0.08967196676693884,
      "grad_norm": 1.5171517133712769,
      "learning_rate": 0.0001981430420352066,
      "loss": 1.7768,
      "step": 626
    },
    {
      "epoch": 0.08981521272024065,
      "grad_norm": 1.9033976793289185,
      "learning_rate": 0.0001981341315228361,
      "loss": 1.605,
      "step": 627
    },
    {
      "epoch": 0.08995845867354248,
      "grad_norm": 1.3244998455047607,
      "learning_rate": 0.00019812519988461548,
      "loss": 1.7871,
      "step": 628
    },
    {
      "epoch": 0.09010170462684429,
      "grad_norm": 1.9696805477142334,
      "learning_rate": 0.00019811624712246756,
      "loss": 1.7835,
      "step": 629
    },
    {
      "epoch": 0.09024495058014612,
      "grad_norm": 2.000761032104492,
      "learning_rate": 0.00019810727323831966,
      "loss": 1.5602,
      "step": 630
    },
    {
      "epoch": 0.09038819653344793,
      "grad_norm": 1.6452912092208862,
      "learning_rate": 0.00019809827823410358,
      "loss": 1.7776,
      "step": 631
    },
    {
      "epoch": 0.09053144248674976,
      "grad_norm": 2.469255208969116,
      "learning_rate": 0.00019808926211175573,
      "loss": 1.5445,
      "step": 632
    },
    {
      "epoch": 0.09067468844005157,
      "grad_norm": 1.5338579416275024,
      "learning_rate": 0.0001980802248732171,
      "loss": 1.5889,
      "step": 633
    },
    {
      "epoch": 0.0908179343933534,
      "grad_norm": 2.191380023956299,
      "learning_rate": 0.0001980711665204332,
      "loss": 1.7019,
      "step": 634
    },
    {
      "epoch": 0.09096118034665521,
      "grad_norm": 1.311061978340149,
      "learning_rate": 0.00019806208705535402,
      "loss": 1.6471,
      "step": 635
    },
    {
      "epoch": 0.09110442629995702,
      "grad_norm": 1.5730485916137695,
      "learning_rate": 0.0001980529864799341,
      "loss": 1.7731,
      "step": 636
    },
    {
      "epoch": 0.09124767225325885,
      "grad_norm": 1.9627255201339722,
      "learning_rate": 0.0001980438647961327,
      "loss": 1.7838,
      "step": 637
    },
    {
      "epoch": 0.09139091820656066,
      "grad_norm": 1.708469271659851,
      "learning_rate": 0.00019803472200591337,
      "loss": 1.7174,
      "step": 638
    },
    {
      "epoch": 0.09153416415986249,
      "grad_norm": 1.6527167558670044,
      "learning_rate": 0.00019802555811124437,
      "loss": 1.6941,
      "step": 639
    },
    {
      "epoch": 0.0916774101131643,
      "grad_norm": 1.3351771831512451,
      "learning_rate": 0.00019801637311409848,
      "loss": 1.6088,
      "step": 640
    },
    {
      "epoch": 0.09182065606646612,
      "grad_norm": 1.586374044418335,
      "learning_rate": 0.000198007167016453,
      "loss": 1.6936,
      "step": 641
    },
    {
      "epoch": 0.09196390201976794,
      "grad_norm": 1.4951837062835693,
      "learning_rate": 0.00019799793982028976,
      "loss": 1.7632,
      "step": 642
    },
    {
      "epoch": 0.09210714797306976,
      "grad_norm": 1.4759860038757324,
      "learning_rate": 0.00019798869152759513,
      "loss": 1.6381,
      "step": 643
    },
    {
      "epoch": 0.09225039392637158,
      "grad_norm": 1.6470977067947388,
      "learning_rate": 0.00019797942214036009,
      "loss": 1.4551,
      "step": 644
    },
    {
      "epoch": 0.0923936398796734,
      "grad_norm": 1.4977290630340576,
      "learning_rate": 0.00019797013166058004,
      "loss": 1.7218,
      "step": 645
    },
    {
      "epoch": 0.09253688583297522,
      "grad_norm": 2.3927955627441406,
      "learning_rate": 0.00019796082009025505,
      "loss": 1.7845,
      "step": 646
    },
    {
      "epoch": 0.09268013178627704,
      "grad_norm": 2.4020349979400635,
      "learning_rate": 0.00019795148743138966,
      "loss": 1.6147,
      "step": 647
    },
    {
      "epoch": 0.09282337773957886,
      "grad_norm": 1.9723669290542603,
      "learning_rate": 0.00019794213368599294,
      "loss": 1.7094,
      "step": 648
    },
    {
      "epoch": 0.09296662369288068,
      "grad_norm": 1.9376848936080933,
      "learning_rate": 0.0001979327588560785,
      "loss": 1.6721,
      "step": 649
    },
    {
      "epoch": 0.0931098696461825,
      "grad_norm": 2.211782693862915,
      "learning_rate": 0.00019792336294366457,
      "loss": 1.6061,
      "step": 650
    },
    {
      "epoch": 0.09325311559948432,
      "grad_norm": 1.63835608959198,
      "learning_rate": 0.00019791394595077382,
      "loss": 1.6629,
      "step": 651
    },
    {
      "epoch": 0.09339636155278613,
      "grad_norm": 2.1630845069885254,
      "learning_rate": 0.0001979045078794335,
      "loss": 1.7264,
      "step": 652
    },
    {
      "epoch": 0.09353960750608795,
      "grad_norm": 1.556217074394226,
      "learning_rate": 0.00019789504873167537,
      "loss": 1.7157,
      "step": 653
    },
    {
      "epoch": 0.09368285345938977,
      "grad_norm": 1.9371320009231567,
      "learning_rate": 0.0001978855685095358,
      "loss": 1.6862,
      "step": 654
    },
    {
      "epoch": 0.09382609941269159,
      "grad_norm": 1.8358875513076782,
      "learning_rate": 0.0001978760672150556,
      "loss": 1.6879,
      "step": 655
    },
    {
      "epoch": 0.09396934536599341,
      "grad_norm": 1.7639013528823853,
      "learning_rate": 0.0001978665448502802,
      "loss": 1.8627,
      "step": 656
    },
    {
      "epoch": 0.09411259131929522,
      "grad_norm": 2.1679720878601074,
      "learning_rate": 0.00019785700141725953,
      "loss": 1.5589,
      "step": 657
    },
    {
      "epoch": 0.09425583727259705,
      "grad_norm": 1.6932626962661743,
      "learning_rate": 0.00019784743691804804,
      "loss": 1.5579,
      "step": 658
    },
    {
      "epoch": 0.09439908322589886,
      "grad_norm": 1.5645488500595093,
      "learning_rate": 0.0001978378513547047,
      "loss": 1.5725,
      "step": 659
    },
    {
      "epoch": 0.09454232917920069,
      "grad_norm": 1.2878520488739014,
      "learning_rate": 0.00019782824472929308,
      "loss": 1.6327,
      "step": 660
    },
    {
      "epoch": 0.0946855751325025,
      "grad_norm": 1.5387991666793823,
      "learning_rate": 0.00019781861704388124,
      "loss": 1.8507,
      "step": 661
    },
    {
      "epoch": 0.09482882108580433,
      "grad_norm": 1.7131320238113403,
      "learning_rate": 0.0001978089683005418,
      "loss": 1.508,
      "step": 662
    },
    {
      "epoch": 0.09497206703910614,
      "grad_norm": 1.4265942573547363,
      "learning_rate": 0.00019779929850135189,
      "loss": 1.5197,
      "step": 663
    },
    {
      "epoch": 0.09511531299240797,
      "grad_norm": 1.8079270124435425,
      "learning_rate": 0.00019778960764839316,
      "loss": 1.6645,
      "step": 664
    },
    {
      "epoch": 0.09525855894570978,
      "grad_norm": 1.3312954902648926,
      "learning_rate": 0.00019777989574375183,
      "loss": 1.5565,
      "step": 665
    },
    {
      "epoch": 0.09540180489901161,
      "grad_norm": 1.556467056274414,
      "learning_rate": 0.0001977701627895186,
      "loss": 1.746,
      "step": 666
    },
    {
      "epoch": 0.09554505085231342,
      "grad_norm": 1.7721006870269775,
      "learning_rate": 0.00019776040878778875,
      "loss": 1.5241,
      "step": 667
    },
    {
      "epoch": 0.09568829680561525,
      "grad_norm": 1.765176773071289,
      "learning_rate": 0.00019775063374066212,
      "loss": 1.5298,
      "step": 668
    },
    {
      "epoch": 0.09583154275891706,
      "grad_norm": 1.6512705087661743,
      "learning_rate": 0.000197740837650243,
      "loss": 1.6805,
      "step": 669
    },
    {
      "epoch": 0.09597478871221889,
      "grad_norm": 1.600159764289856,
      "learning_rate": 0.00019773102051864018,
      "loss": 1.5506,
      "step": 670
    },
    {
      "epoch": 0.0961180346655207,
      "grad_norm": 1.7851074934005737,
      "learning_rate": 0.00019772118234796717,
      "loss": 1.5179,
      "step": 671
    },
    {
      "epoch": 0.09626128061882251,
      "grad_norm": 2.052377939224243,
      "learning_rate": 0.0001977113231403418,
      "loss": 1.6381,
      "step": 672
    },
    {
      "epoch": 0.09640452657212434,
      "grad_norm": 1.7640641927719116,
      "learning_rate": 0.00019770144289788655,
      "loss": 1.6755,
      "step": 673
    },
    {
      "epoch": 0.09654777252542615,
      "grad_norm": 1.3730601072311401,
      "learning_rate": 0.00019769154162272839,
      "loss": 1.4196,
      "step": 674
    },
    {
      "epoch": 0.09669101847872798,
      "grad_norm": 1.3700706958770752,
      "learning_rate": 0.0001976816193169988,
      "loss": 1.6946,
      "step": 675
    },
    {
      "epoch": 0.09683426443202979,
      "grad_norm": 1.4719998836517334,
      "learning_rate": 0.0001976716759828338,
      "loss": 1.6281,
      "step": 676
    },
    {
      "epoch": 0.09697751038533162,
      "grad_norm": 1.3043746948242188,
      "learning_rate": 0.00019766171162237397,
      "loss": 1.807,
      "step": 677
    },
    {
      "epoch": 0.09712075633863343,
      "grad_norm": 1.6509284973144531,
      "learning_rate": 0.00019765172623776437,
      "loss": 1.6569,
      "step": 678
    },
    {
      "epoch": 0.09726400229193526,
      "grad_norm": 1.9164206981658936,
      "learning_rate": 0.00019764171983115462,
      "loss": 1.6634,
      "step": 679
    },
    {
      "epoch": 0.09740724824523707,
      "grad_norm": 1.4596762657165527,
      "learning_rate": 0.00019763169240469885,
      "loss": 1.6812,
      "step": 680
    },
    {
      "epoch": 0.0975504941985389,
      "grad_norm": 1.8942707777023315,
      "learning_rate": 0.00019762164396055573,
      "loss": 1.7299,
      "step": 681
    },
    {
      "epoch": 0.09769374015184071,
      "grad_norm": 1.4766433238983154,
      "learning_rate": 0.00019761157450088842,
      "loss": 1.9527,
      "step": 682
    },
    {
      "epoch": 0.09783698610514253,
      "grad_norm": 1.445786476135254,
      "learning_rate": 0.0001976014840278646,
      "loss": 1.6917,
      "step": 683
    },
    {
      "epoch": 0.09798023205844435,
      "grad_norm": 1.5590589046478271,
      "learning_rate": 0.00019759137254365657,
      "loss": 1.5339,
      "step": 684
    },
    {
      "epoch": 0.09812347801174617,
      "grad_norm": 1.8950871229171753,
      "learning_rate": 0.000197581240050441,
      "loss": 1.4936,
      "step": 685
    },
    {
      "epoch": 0.09826672396504799,
      "grad_norm": 1.9997422695159912,
      "learning_rate": 0.00019757108655039924,
      "loss": 1.6934,
      "step": 686
    },
    {
      "epoch": 0.09840996991834981,
      "grad_norm": 1.4285129308700562,
      "learning_rate": 0.00019756091204571708,
      "loss": 1.5747,
      "step": 687
    },
    {
      "epoch": 0.09855321587165163,
      "grad_norm": 1.658769965171814,
      "learning_rate": 0.00019755071653858476,
      "loss": 1.7785,
      "step": 688
    },
    {
      "epoch": 0.09869646182495345,
      "grad_norm": 1.8999029397964478,
      "learning_rate": 0.00019754050003119723,
      "loss": 1.5716,
      "step": 689
    },
    {
      "epoch": 0.09883970777825526,
      "grad_norm": 1.6866711378097534,
      "learning_rate": 0.00019753026252575375,
      "loss": 1.8494,
      "step": 690
    },
    {
      "epoch": 0.09898295373155708,
      "grad_norm": 1.254889726638794,
      "learning_rate": 0.00019752000402445825,
      "loss": 1.7504,
      "step": 691
    },
    {
      "epoch": 0.0991261996848589,
      "grad_norm": 1.7436747550964355,
      "learning_rate": 0.00019750972452951918,
      "loss": 1.5608,
      "step": 692
    },
    {
      "epoch": 0.09926944563816072,
      "grad_norm": 1.343024492263794,
      "learning_rate": 0.00019749942404314935,
      "loss": 1.7148,
      "step": 693
    },
    {
      "epoch": 0.09941269159146254,
      "grad_norm": 1.7251167297363281,
      "learning_rate": 0.00019748910256756628,
      "loss": 1.5804,
      "step": 694
    },
    {
      "epoch": 0.09955593754476436,
      "grad_norm": 1.5857722759246826,
      "learning_rate": 0.00019747876010499192,
      "loss": 1.6637,
      "step": 695
    },
    {
      "epoch": 0.09969918349806618,
      "grad_norm": 1.2661147117614746,
      "learning_rate": 0.0001974683966576527,
      "loss": 1.6412,
      "step": 696
    },
    {
      "epoch": 0.099842429451368,
      "grad_norm": 1.3366247415542603,
      "learning_rate": 0.00019745801222777968,
      "loss": 1.6925,
      "step": 697
    },
    {
      "epoch": 0.09998567540466982,
      "grad_norm": 1.6041820049285889,
      "learning_rate": 0.00019744760681760832,
      "loss": 1.5448,
      "step": 698
    },
    {
      "epoch": 0.10012892135797163,
      "grad_norm": 1.5244828462600708,
      "learning_rate": 0.0001974371804293787,
      "loss": 1.5587,
      "step": 699
    },
    {
      "epoch": 0.10027216731127346,
      "grad_norm": 1.45596444606781,
      "learning_rate": 0.0001974267330653353,
      "loss": 1.634,
      "step": 700
    },
    {
      "epoch": 0.10041541326457527,
      "grad_norm": 1.7106397151947021,
      "learning_rate": 0.00019741626472772722,
      "loss": 1.7385,
      "step": 701
    },
    {
      "epoch": 0.1005586592178771,
      "grad_norm": 1.6318823099136353,
      "learning_rate": 0.000197405775418808,
      "loss": 1.6464,
      "step": 702
    },
    {
      "epoch": 0.10070190517117891,
      "grad_norm": 1.9525867700576782,
      "learning_rate": 0.00019739526514083578,
      "loss": 1.727,
      "step": 703
    },
    {
      "epoch": 0.10084515112448074,
      "grad_norm": 1.8090976476669312,
      "learning_rate": 0.00019738473389607314,
      "loss": 1.5947,
      "step": 704
    },
    {
      "epoch": 0.10098839707778255,
      "grad_norm": 1.7600536346435547,
      "learning_rate": 0.00019737418168678714,
      "loss": 1.5912,
      "step": 705
    },
    {
      "epoch": 0.10113164303108438,
      "grad_norm": 1.6049998998641968,
      "learning_rate": 0.0001973636085152495,
      "loss": 1.5044,
      "step": 706
    },
    {
      "epoch": 0.10127488898438619,
      "grad_norm": 1.4004878997802734,
      "learning_rate": 0.00019735301438373633,
      "loss": 1.6854,
      "step": 707
    },
    {
      "epoch": 0.101418134937688,
      "grad_norm": 1.7660269737243652,
      "learning_rate": 0.00019734239929452825,
      "loss": 1.5787,
      "step": 708
    },
    {
      "epoch": 0.10156138089098983,
      "grad_norm": 1.438454508781433,
      "learning_rate": 0.00019733176324991046,
      "loss": 1.8692,
      "step": 709
    },
    {
      "epoch": 0.10170462684429164,
      "grad_norm": 1.228301763534546,
      "learning_rate": 0.00019732110625217262,
      "loss": 1.7537,
      "step": 710
    },
    {
      "epoch": 0.10184787279759347,
      "grad_norm": 1.4191251993179321,
      "learning_rate": 0.00019731042830360896,
      "loss": 1.5946,
      "step": 711
    },
    {
      "epoch": 0.10199111875089528,
      "grad_norm": 1.5328761339187622,
      "learning_rate": 0.0001972997294065181,
      "loss": 1.6684,
      "step": 712
    },
    {
      "epoch": 0.10213436470419711,
      "grad_norm": 1.6361255645751953,
      "learning_rate": 0.0001972890095632033,
      "loss": 1.6493,
      "step": 713
    },
    {
      "epoch": 0.10227761065749892,
      "grad_norm": 1.4105583429336548,
      "learning_rate": 0.00019727826877597232,
      "loss": 1.6317,
      "step": 714
    },
    {
      "epoch": 0.10242085661080075,
      "grad_norm": 1.3408552408218384,
      "learning_rate": 0.0001972675070471373,
      "loss": 1.6468,
      "step": 715
    },
    {
      "epoch": 0.10256410256410256,
      "grad_norm": 1.877536416053772,
      "learning_rate": 0.000197256724379015,
      "loss": 1.5974,
      "step": 716
    },
    {
      "epoch": 0.10270734851740439,
      "grad_norm": 1.6134400367736816,
      "learning_rate": 0.0001972459207739267,
      "loss": 1.6142,
      "step": 717
    },
    {
      "epoch": 0.1028505944707062,
      "grad_norm": 1.2656651735305786,
      "learning_rate": 0.00019723509623419808,
      "loss": 1.7536,
      "step": 718
    },
    {
      "epoch": 0.10299384042400803,
      "grad_norm": 1.6002271175384521,
      "learning_rate": 0.00019722425076215946,
      "loss": 1.6491,
      "step": 719
    },
    {
      "epoch": 0.10313708637730984,
      "grad_norm": 1.566291093826294,
      "learning_rate": 0.00019721338436014558,
      "loss": 1.5691,
      "step": 720
    },
    {
      "epoch": 0.10328033233061167,
      "grad_norm": 1.2599036693572998,
      "learning_rate": 0.00019720249703049573,
      "loss": 1.8011,
      "step": 721
    },
    {
      "epoch": 0.10342357828391348,
      "grad_norm": 1.5381052494049072,
      "learning_rate": 0.0001971915887755536,
      "loss": 1.5195,
      "step": 722
    },
    {
      "epoch": 0.1035668242372153,
      "grad_norm": 1.434147834777832,
      "learning_rate": 0.00019718065959766756,
      "loss": 1.5305,
      "step": 723
    },
    {
      "epoch": 0.10371007019051712,
      "grad_norm": 1.3286505937576294,
      "learning_rate": 0.00019716970949919035,
      "loss": 1.6205,
      "step": 724
    },
    {
      "epoch": 0.10385331614381894,
      "grad_norm": 1.2326021194458008,
      "learning_rate": 0.00019715873848247928,
      "loss": 1.6684,
      "step": 725
    },
    {
      "epoch": 0.10399656209712076,
      "grad_norm": 1.330072283744812,
      "learning_rate": 0.0001971477465498961,
      "loss": 1.5842,
      "step": 726
    },
    {
      "epoch": 0.10413980805042257,
      "grad_norm": 1.620344877243042,
      "learning_rate": 0.00019713673370380712,
      "loss": 1.5601,
      "step": 727
    },
    {
      "epoch": 0.1042830540037244,
      "grad_norm": 1.6700549125671387,
      "learning_rate": 0.00019712569994658315,
      "loss": 1.5271,
      "step": 728
    },
    {
      "epoch": 0.10442629995702621,
      "grad_norm": 1.8878201246261597,
      "learning_rate": 0.00019711464528059946,
      "loss": 1.3555,
      "step": 729
    },
    {
      "epoch": 0.10456954591032804,
      "grad_norm": 1.4124029874801636,
      "learning_rate": 0.00019710356970823587,
      "loss": 1.6412,
      "step": 730
    },
    {
      "epoch": 0.10471279186362985,
      "grad_norm": 1.7656874656677246,
      "learning_rate": 0.00019709247323187662,
      "loss": 1.6317,
      "step": 731
    },
    {
      "epoch": 0.10485603781693167,
      "grad_norm": 1.097822904586792,
      "learning_rate": 0.0001970813558539106,
      "loss": 1.6414,
      "step": 732
    },
    {
      "epoch": 0.10499928377023349,
      "grad_norm": 1.4732197523117065,
      "learning_rate": 0.00019707021757673103,
      "loss": 1.6305,
      "step": 733
    },
    {
      "epoch": 0.10514252972353531,
      "grad_norm": 1.3448829650878906,
      "learning_rate": 0.00019705905840273572,
      "loss": 1.6061,
      "step": 734
    },
    {
      "epoch": 0.10528577567683713,
      "grad_norm": 1.3394949436187744,
      "learning_rate": 0.00019704787833432698,
      "loss": 1.6372,
      "step": 735
    },
    {
      "epoch": 0.10542902163013895,
      "grad_norm": 1.4685827493667603,
      "learning_rate": 0.00019703667737391162,
      "loss": 1.9355,
      "step": 736
    },
    {
      "epoch": 0.10557226758344077,
      "grad_norm": 1.8580946922302246,
      "learning_rate": 0.00019702545552390089,
      "loss": 1.6555,
      "step": 737
    },
    {
      "epoch": 0.10571551353674259,
      "grad_norm": 1.9116902351379395,
      "learning_rate": 0.00019701421278671058,
      "loss": 1.8081,
      "step": 738
    },
    {
      "epoch": 0.1058587594900444,
      "grad_norm": 1.2292426824569702,
      "learning_rate": 0.000197002949164761,
      "loss": 1.7138,
      "step": 739
    },
    {
      "epoch": 0.10600200544334623,
      "grad_norm": 1.3974497318267822,
      "learning_rate": 0.00019699166466047692,
      "loss": 1.5291,
      "step": 740
    },
    {
      "epoch": 0.10614525139664804,
      "grad_norm": 1.386409878730774,
      "learning_rate": 0.0001969803592762876,
      "loss": 1.652,
      "step": 741
    },
    {
      "epoch": 0.10628849734994987,
      "grad_norm": 1.6884740591049194,
      "learning_rate": 0.0001969690330146268,
      "loss": 1.7535,
      "step": 742
    },
    {
      "epoch": 0.10643174330325168,
      "grad_norm": 1.5742979049682617,
      "learning_rate": 0.00019695768587793286,
      "loss": 1.8504,
      "step": 743
    },
    {
      "epoch": 0.1065749892565535,
      "grad_norm": 1.4850658178329468,
      "learning_rate": 0.00019694631786864843,
      "loss": 1.723,
      "step": 744
    },
    {
      "epoch": 0.10671823520985532,
      "grad_norm": 1.238373875617981,
      "learning_rate": 0.00019693492898922084,
      "loss": 1.6498,
      "step": 745
    },
    {
      "epoch": 0.10686148116315713,
      "grad_norm": 2.092233180999756,
      "learning_rate": 0.00019692351924210176,
      "loss": 1.6978,
      "step": 746
    },
    {
      "epoch": 0.10700472711645896,
      "grad_norm": 2.1850333213806152,
      "learning_rate": 0.00019691208862974752,
      "loss": 1.4748,
      "step": 747
    },
    {
      "epoch": 0.10714797306976077,
      "grad_norm": 1.5780452489852905,
      "learning_rate": 0.0001969006371546188,
      "loss": 1.6114,
      "step": 748
    },
    {
      "epoch": 0.1072912190230626,
      "grad_norm": 1.6323115825653076,
      "learning_rate": 0.0001968891648191808,
      "loss": 1.5691,
      "step": 749
    },
    {
      "epoch": 0.10743446497636441,
      "grad_norm": 1.273373007774353,
      "learning_rate": 0.00019687767162590328,
      "loss": 1.6624,
      "step": 750
    },
    {
      "epoch": 0.10757771092966624,
      "grad_norm": 1.3526626825332642,
      "learning_rate": 0.00019686615757726034,
      "loss": 1.6806,
      "step": 751
    },
    {
      "epoch": 0.10772095688296805,
      "grad_norm": 1.477441668510437,
      "learning_rate": 0.0001968546226757308,
      "loss": 1.7353,
      "step": 752
    },
    {
      "epoch": 0.10786420283626988,
      "grad_norm": 1.4391818046569824,
      "learning_rate": 0.00019684306692379776,
      "loss": 1.6659,
      "step": 753
    },
    {
      "epoch": 0.10800744878957169,
      "grad_norm": 1.4352343082427979,
      "learning_rate": 0.0001968314903239489,
      "loss": 1.4964,
      "step": 754
    },
    {
      "epoch": 0.10815069474287352,
      "grad_norm": 1.4778071641921997,
      "learning_rate": 0.00019681989287867636,
      "loss": 1.6588,
      "step": 755
    },
    {
      "epoch": 0.10829394069617533,
      "grad_norm": 1.8458248376846313,
      "learning_rate": 0.00019680827459047685,
      "loss": 1.6013,
      "step": 756
    },
    {
      "epoch": 0.10843718664947716,
      "grad_norm": 1.472934365272522,
      "learning_rate": 0.00019679663546185144,
      "loss": 1.5839,
      "step": 757
    },
    {
      "epoch": 0.10858043260277897,
      "grad_norm": 1.2686278820037842,
      "learning_rate": 0.00019678497549530574,
      "loss": 1.7562,
      "step": 758
    },
    {
      "epoch": 0.1087236785560808,
      "grad_norm": 1.8563710451126099,
      "learning_rate": 0.0001967732946933499,
      "loss": 1.5606,
      "step": 759
    },
    {
      "epoch": 0.10886692450938261,
      "grad_norm": 1.3085123300552368,
      "learning_rate": 0.00019676159305849846,
      "loss": 1.6645,
      "step": 760
    },
    {
      "epoch": 0.10901017046268444,
      "grad_norm": 1.513801097869873,
      "learning_rate": 0.0001967498705932705,
      "loss": 1.6237,
      "step": 761
    },
    {
      "epoch": 0.10915341641598625,
      "grad_norm": 1.6214598417282104,
      "learning_rate": 0.0001967381273001896,
      "loss": 1.6615,
      "step": 762
    },
    {
      "epoch": 0.10929666236928806,
      "grad_norm": 1.496747374534607,
      "learning_rate": 0.00019672636318178381,
      "loss": 1.5546,
      "step": 763
    },
    {
      "epoch": 0.10943990832258989,
      "grad_norm": 1.3868905305862427,
      "learning_rate": 0.0001967145782405856,
      "loss": 1.6734,
      "step": 764
    },
    {
      "epoch": 0.1095831542758917,
      "grad_norm": 1.9769505262374878,
      "learning_rate": 0.00019670277247913205,
      "loss": 1.6059,
      "step": 765
    },
    {
      "epoch": 0.10972640022919353,
      "grad_norm": 1.4846587181091309,
      "learning_rate": 0.00019669094589996457,
      "loss": 1.6523,
      "step": 766
    },
    {
      "epoch": 0.10986964618249534,
      "grad_norm": 1.6699373722076416,
      "learning_rate": 0.00019667909850562917,
      "loss": 1.4849,
      "step": 767
    },
    {
      "epoch": 0.11001289213579717,
      "grad_norm": 1.1695612668991089,
      "learning_rate": 0.00019666723029867632,
      "loss": 1.5853,
      "step": 768
    },
    {
      "epoch": 0.11015613808909898,
      "grad_norm": 1.2621443271636963,
      "learning_rate": 0.00019665534128166092,
      "loss": 1.7378,
      "step": 769
    },
    {
      "epoch": 0.1102993840424008,
      "grad_norm": 1.462565302848816,
      "learning_rate": 0.0001966434314571424,
      "loss": 1.7238,
      "step": 770
    },
    {
      "epoch": 0.11044262999570262,
      "grad_norm": 1.6089022159576416,
      "learning_rate": 0.00019663150082768462,
      "loss": 1.6665,
      "step": 771
    },
    {
      "epoch": 0.11058587594900444,
      "grad_norm": 1.5095402002334595,
      "learning_rate": 0.000196619549395856,
      "loss": 1.5419,
      "step": 772
    },
    {
      "epoch": 0.11072912190230626,
      "grad_norm": 1.401183009147644,
      "learning_rate": 0.00019660757716422932,
      "loss": 1.4786,
      "step": 773
    },
    {
      "epoch": 0.11087236785560808,
      "grad_norm": 1.3202089071273804,
      "learning_rate": 0.00019659558413538198,
      "loss": 1.6672,
      "step": 774
    },
    {
      "epoch": 0.1110156138089099,
      "grad_norm": 1.6599305868148804,
      "learning_rate": 0.0001965835703118957,
      "loss": 1.648,
      "step": 775
    },
    {
      "epoch": 0.11115885976221172,
      "grad_norm": 1.1583521366119385,
      "learning_rate": 0.00019657153569635683,
      "loss": 1.7374,
      "step": 776
    },
    {
      "epoch": 0.11130210571551354,
      "grad_norm": 1.4404739141464233,
      "learning_rate": 0.0001965594802913561,
      "loss": 1.7739,
      "step": 777
    },
    {
      "epoch": 0.11144535166881536,
      "grad_norm": 1.8734699487686157,
      "learning_rate": 0.00019654740409948872,
      "loss": 1.5631,
      "step": 778
    },
    {
      "epoch": 0.11158859762211717,
      "grad_norm": 1.410055160522461,
      "learning_rate": 0.00019653530712335443,
      "loss": 1.5709,
      "step": 779
    },
    {
      "epoch": 0.11173184357541899,
      "grad_norm": 1.6245243549346924,
      "learning_rate": 0.00019652318936555743,
      "loss": 1.7171,
      "step": 780
    },
    {
      "epoch": 0.11187508952872081,
      "grad_norm": 1.3341032266616821,
      "learning_rate": 0.00019651105082870628,
      "loss": 1.5142,
      "step": 781
    },
    {
      "epoch": 0.11201833548202263,
      "grad_norm": 1.5533653497695923,
      "learning_rate": 0.00019649889151541417,
      "loss": 1.5558,
      "step": 782
    },
    {
      "epoch": 0.11216158143532445,
      "grad_norm": 1.6404037475585938,
      "learning_rate": 0.00019648671142829876,
      "loss": 1.6398,
      "step": 783
    },
    {
      "epoch": 0.11230482738862627,
      "grad_norm": 1.6427288055419922,
      "learning_rate": 0.000196474510569982,
      "loss": 1.618,
      "step": 784
    },
    {
      "epoch": 0.11244807334192809,
      "grad_norm": 1.4234371185302734,
      "learning_rate": 0.0001964622889430905,
      "loss": 1.3855,
      "step": 785
    },
    {
      "epoch": 0.1125913192952299,
      "grad_norm": 1.3232643604278564,
      "learning_rate": 0.00019645004655025522,
      "loss": 1.6925,
      "step": 786
    },
    {
      "epoch": 0.11273456524853173,
      "grad_norm": 1.4328253269195557,
      "learning_rate": 0.00019643778339411175,
      "loss": 1.7753,
      "step": 787
    },
    {
      "epoch": 0.11287781120183354,
      "grad_norm": 1.5762733221054077,
      "learning_rate": 0.00019642549947729992,
      "loss": 1.6083,
      "step": 788
    },
    {
      "epoch": 0.11302105715513537,
      "grad_norm": 1.3697439432144165,
      "learning_rate": 0.00019641319480246424,
      "loss": 1.5214,
      "step": 789
    },
    {
      "epoch": 0.11316430310843718,
      "grad_norm": 1.1820030212402344,
      "learning_rate": 0.0001964008693722536,
      "loss": 1.6088,
      "step": 790
    },
    {
      "epoch": 0.11330754906173901,
      "grad_norm": 1.2939660549163818,
      "learning_rate": 0.0001963885231893213,
      "loss": 1.4251,
      "step": 791
    },
    {
      "epoch": 0.11345079501504082,
      "grad_norm": 1.9614722728729248,
      "learning_rate": 0.00019637615625632524,
      "loss": 1.7844,
      "step": 792
    },
    {
      "epoch": 0.11359404096834265,
      "grad_norm": 1.4284573793411255,
      "learning_rate": 0.00019636376857592764,
      "loss": 1.8043,
      "step": 793
    },
    {
      "epoch": 0.11373728692164446,
      "grad_norm": 1.34441077709198,
      "learning_rate": 0.00019635136015079533,
      "loss": 1.7269,
      "step": 794
    },
    {
      "epoch": 0.11388053287494629,
      "grad_norm": 1.8618592023849487,
      "learning_rate": 0.00019633893098359951,
      "loss": 1.5111,
      "step": 795
    },
    {
      "epoch": 0.1140237788282481,
      "grad_norm": 1.511165738105774,
      "learning_rate": 0.00019632648107701585,
      "loss": 1.6387,
      "step": 796
    },
    {
      "epoch": 0.11416702478154993,
      "grad_norm": 1.1874998807907104,
      "learning_rate": 0.00019631401043372447,
      "loss": 1.7597,
      "step": 797
    },
    {
      "epoch": 0.11431027073485174,
      "grad_norm": 1.4529027938842773,
      "learning_rate": 0.00019630151905641012,
      "loss": 1.515,
      "step": 798
    },
    {
      "epoch": 0.11445351668815355,
      "grad_norm": 1.3633545637130737,
      "learning_rate": 0.00019628900694776177,
      "loss": 1.5391,
      "step": 799
    },
    {
      "epoch": 0.11459676264145538,
      "grad_norm": 1.3726894855499268,
      "learning_rate": 0.00019627647411047305,
      "loss": 1.5234,
      "step": 800
    },
    {
      "epoch": 0.11474000859475719,
      "grad_norm": 1.7441984415054321,
      "learning_rate": 0.0001962639205472419,
      "loss": 1.6786,
      "step": 801
    },
    {
      "epoch": 0.11488325454805902,
      "grad_norm": 1.3965880870819092,
      "learning_rate": 0.00019625134626077083,
      "loss": 1.6971,
      "step": 802
    },
    {
      "epoch": 0.11502650050136083,
      "grad_norm": 1.2910364866256714,
      "learning_rate": 0.00019623875125376674,
      "loss": 1.7357,
      "step": 803
    },
    {
      "epoch": 0.11516974645466266,
      "grad_norm": 1.306301236152649,
      "learning_rate": 0.00019622613552894107,
      "loss": 1.6423,
      "step": 804
    },
    {
      "epoch": 0.11531299240796447,
      "grad_norm": 1.4404336214065552,
      "learning_rate": 0.0001962134990890096,
      "loss": 1.4665,
      "step": 805
    },
    {
      "epoch": 0.1154562383612663,
      "grad_norm": 1.2204550504684448,
      "learning_rate": 0.00019620084193669275,
      "loss": 1.6481,
      "step": 806
    },
    {
      "epoch": 0.11559948431456811,
      "grad_norm": 1.4873749017715454,
      "learning_rate": 0.00019618816407471519,
      "loss": 1.352,
      "step": 807
    },
    {
      "epoch": 0.11574273026786994,
      "grad_norm": 1.5756500959396362,
      "learning_rate": 0.00019617546550580622,
      "loss": 1.5499,
      "step": 808
    },
    {
      "epoch": 0.11588597622117175,
      "grad_norm": 1.172492265701294,
      "learning_rate": 0.0001961627462326995,
      "loss": 1.6223,
      "step": 809
    },
    {
      "epoch": 0.11602922217447358,
      "grad_norm": 1.3477461338043213,
      "learning_rate": 0.00019615000625813314,
      "loss": 1.4655,
      "step": 810
    },
    {
      "epoch": 0.11617246812777539,
      "grad_norm": 1.4703794717788696,
      "learning_rate": 0.0001961372455848498,
      "loss": 1.6865,
      "step": 811
    },
    {
      "epoch": 0.11631571408107721,
      "grad_norm": 1.4132351875305176,
      "learning_rate": 0.0001961244642155965,
      "loss": 1.5305,
      "step": 812
    },
    {
      "epoch": 0.11645896003437903,
      "grad_norm": 1.4192314147949219,
      "learning_rate": 0.0001961116621531248,
      "loss": 1.6134,
      "step": 813
    },
    {
      "epoch": 0.11660220598768085,
      "grad_norm": 1.3303085565567017,
      "learning_rate": 0.00019609883940019057,
      "loss": 1.6348,
      "step": 814
    },
    {
      "epoch": 0.11674545194098267,
      "grad_norm": 1.469582438468933,
      "learning_rate": 0.00019608599595955436,
      "loss": 1.5101,
      "step": 815
    },
    {
      "epoch": 0.11688869789428448,
      "grad_norm": 1.5950562953948975,
      "learning_rate": 0.00019607313183398094,
      "loss": 1.6845,
      "step": 816
    },
    {
      "epoch": 0.1170319438475863,
      "grad_norm": 1.4950788021087646,
      "learning_rate": 0.0001960602470262397,
      "loss": 1.6995,
      "step": 817
    },
    {
      "epoch": 0.11717518980088812,
      "grad_norm": 1.3118116855621338,
      "learning_rate": 0.00019604734153910437,
      "loss": 1.6796,
      "step": 818
    },
    {
      "epoch": 0.11731843575418995,
      "grad_norm": 1.517374873161316,
      "learning_rate": 0.00019603441537535324,
      "loss": 1.5008,
      "step": 819
    },
    {
      "epoch": 0.11746168170749176,
      "grad_norm": 1.8473433256149292,
      "learning_rate": 0.00019602146853776894,
      "loss": 1.7425,
      "step": 820
    },
    {
      "epoch": 0.11760492766079358,
      "grad_norm": 1.3010900020599365,
      "learning_rate": 0.00019600850102913865,
      "loss": 1.6855,
      "step": 821
    },
    {
      "epoch": 0.1177481736140954,
      "grad_norm": 1.3770560026168823,
      "learning_rate": 0.00019599551285225393,
      "loss": 1.7055,
      "step": 822
    },
    {
      "epoch": 0.11789141956739722,
      "grad_norm": 1.4143677949905396,
      "learning_rate": 0.00019598250400991077,
      "loss": 1.5576,
      "step": 823
    },
    {
      "epoch": 0.11803466552069904,
      "grad_norm": 1.4342479705810547,
      "learning_rate": 0.00019596947450490975,
      "loss": 1.6647,
      "step": 824
    },
    {
      "epoch": 0.11817791147400086,
      "grad_norm": 1.639679193496704,
      "learning_rate": 0.0001959564243400557,
      "loss": 1.6066,
      "step": 825
    },
    {
      "epoch": 0.11832115742730268,
      "grad_norm": 1.459906816482544,
      "learning_rate": 0.00019594335351815807,
      "loss": 1.5834,
      "step": 826
    },
    {
      "epoch": 0.1184644033806045,
      "grad_norm": 1.5267376899719238,
      "learning_rate": 0.00019593026204203066,
      "loss": 1.555,
      "step": 827
    },
    {
      "epoch": 0.11860764933390631,
      "grad_norm": 1.8674683570861816,
      "learning_rate": 0.0001959171499144917,
      "loss": 1.5765,
      "step": 828
    },
    {
      "epoch": 0.11875089528720814,
      "grad_norm": 1.6813234090805054,
      "learning_rate": 0.00019590401713836397,
      "loss": 1.6752,
      "step": 829
    },
    {
      "epoch": 0.11889414124050995,
      "grad_norm": 1.5916492938995361,
      "learning_rate": 0.0001958908637164746,
      "loss": 1.5171,
      "step": 830
    },
    {
      "epoch": 0.11903738719381178,
      "grad_norm": 1.5751780271530151,
      "learning_rate": 0.0001958776896516552,
      "loss": 1.7203,
      "step": 831
    },
    {
      "epoch": 0.1191806331471136,
      "grad_norm": 1.3422431945800781,
      "learning_rate": 0.00019586449494674187,
      "loss": 1.5944,
      "step": 832
    },
    {
      "epoch": 0.11932387910041542,
      "grad_norm": 1.532328486442566,
      "learning_rate": 0.000195851279604575,
      "loss": 1.4551,
      "step": 833
    },
    {
      "epoch": 0.11946712505371723,
      "grad_norm": 1.6953420639038086,
      "learning_rate": 0.0001958380436279996,
      "loss": 1.5568,
      "step": 834
    },
    {
      "epoch": 0.11961037100701905,
      "grad_norm": 1.657399296760559,
      "learning_rate": 0.000195824787019865,
      "loss": 1.5253,
      "step": 835
    },
    {
      "epoch": 0.11975361696032087,
      "grad_norm": 1.3738839626312256,
      "learning_rate": 0.0001958115097830251,
      "loss": 1.4798,
      "step": 836
    },
    {
      "epoch": 0.11989686291362268,
      "grad_norm": 1.2179100513458252,
      "learning_rate": 0.0001957982119203381,
      "loss": 1.4871,
      "step": 837
    },
    {
      "epoch": 0.12004010886692451,
      "grad_norm": 1.6310373544692993,
      "learning_rate": 0.0001957848934346667,
      "loss": 1.5856,
      "step": 838
    },
    {
      "epoch": 0.12018335482022632,
      "grad_norm": 1.4978235960006714,
      "learning_rate": 0.00019577155432887804,
      "loss": 1.7682,
      "step": 839
    },
    {
      "epoch": 0.12032660077352815,
      "grad_norm": 1.4542731046676636,
      "learning_rate": 0.00019575819460584373,
      "loss": 1.4687,
      "step": 840
    },
    {
      "epoch": 0.12046984672682996,
      "grad_norm": 1.1406450271606445,
      "learning_rate": 0.00019574481426843976,
      "loss": 1.6196,
      "step": 841
    },
    {
      "epoch": 0.12061309268013179,
      "grad_norm": 1.8530430793762207,
      "learning_rate": 0.0001957314133195466,
      "loss": 1.4314,
      "step": 842
    },
    {
      "epoch": 0.1207563386334336,
      "grad_norm": 1.730963110923767,
      "learning_rate": 0.00019571799176204912,
      "loss": 1.5497,
      "step": 843
    },
    {
      "epoch": 0.12089958458673543,
      "grad_norm": 1.4124916791915894,
      "learning_rate": 0.00019570454959883668,
      "loss": 1.6568,
      "step": 844
    },
    {
      "epoch": 0.12104283054003724,
      "grad_norm": 1.5251939296722412,
      "learning_rate": 0.00019569108683280303,
      "loss": 1.608,
      "step": 845
    },
    {
      "epoch": 0.12118607649333907,
      "grad_norm": 1.287947177886963,
      "learning_rate": 0.00019567760346684638,
      "loss": 1.6271,
      "step": 846
    },
    {
      "epoch": 0.12132932244664088,
      "grad_norm": 1.4431467056274414,
      "learning_rate": 0.00019566409950386935,
      "loss": 1.273,
      "step": 847
    },
    {
      "epoch": 0.12147256839994271,
      "grad_norm": 1.4683396816253662,
      "learning_rate": 0.000195650574946779,
      "loss": 1.7495,
      "step": 848
    },
    {
      "epoch": 0.12161581435324452,
      "grad_norm": 1.6172163486480713,
      "learning_rate": 0.00019563702979848686,
      "loss": 1.6144,
      "step": 849
    },
    {
      "epoch": 0.12175906030654635,
      "grad_norm": 1.5742977857589722,
      "learning_rate": 0.00019562346406190888,
      "loss": 1.5745,
      "step": 850
    },
    {
      "epoch": 0.12190230625984816,
      "grad_norm": 1.3044545650482178,
      "learning_rate": 0.00019560987773996536,
      "loss": 1.6523,
      "step": 851
    },
    {
      "epoch": 0.12204555221314999,
      "grad_norm": 1.6706430912017822,
      "learning_rate": 0.00019559627083558115,
      "loss": 1.5829,
      "step": 852
    },
    {
      "epoch": 0.1221887981664518,
      "grad_norm": 1.5556046962738037,
      "learning_rate": 0.00019558264335168548,
      "loss": 1.5973,
      "step": 853
    },
    {
      "epoch": 0.12233204411975361,
      "grad_norm": 1.3522742986679077,
      "learning_rate": 0.000195568995291212,
      "loss": 1.5215,
      "step": 854
    },
    {
      "epoch": 0.12247529007305544,
      "grad_norm": 2.2802860736846924,
      "learning_rate": 0.00019555532665709878,
      "loss": 1.6063,
      "step": 855
    },
    {
      "epoch": 0.12261853602635725,
      "grad_norm": 1.3602712154388428,
      "learning_rate": 0.0001955416374522884,
      "loss": 1.6613,
      "step": 856
    },
    {
      "epoch": 0.12276178197965908,
      "grad_norm": 1.4493857622146606,
      "learning_rate": 0.00019552792767972771,
      "loss": 1.5722,
      "step": 857
    },
    {
      "epoch": 0.12290502793296089,
      "grad_norm": 1.3390328884124756,
      "learning_rate": 0.00019551419734236818,
      "loss": 1.593,
      "step": 858
    },
    {
      "epoch": 0.12304827388626272,
      "grad_norm": 1.2195194959640503,
      "learning_rate": 0.00019550044644316557,
      "loss": 1.598,
      "step": 859
    },
    {
      "epoch": 0.12319151983956453,
      "grad_norm": 1.470925211906433,
      "learning_rate": 0.0001954866749850801,
      "loss": 1.5284,
      "step": 860
    },
    {
      "epoch": 0.12333476579286635,
      "grad_norm": 1.6146818399429321,
      "learning_rate": 0.0001954728829710764,
      "loss": 1.4087,
      "step": 861
    },
    {
      "epoch": 0.12347801174616817,
      "grad_norm": 1.2974482774734497,
      "learning_rate": 0.00019545907040412363,
      "loss": 1.5172,
      "step": 862
    },
    {
      "epoch": 0.12362125769947,
      "grad_norm": 1.5922187566757202,
      "learning_rate": 0.00019544523728719525,
      "loss": 1.6491,
      "step": 863
    },
    {
      "epoch": 0.1237645036527718,
      "grad_norm": 1.5290395021438599,
      "learning_rate": 0.00019543138362326917,
      "loss": 1.6143,
      "step": 864
    },
    {
      "epoch": 0.12390774960607363,
      "grad_norm": 1.4779689311981201,
      "learning_rate": 0.00019541750941532774,
      "loss": 1.5974,
      "step": 865
    },
    {
      "epoch": 0.12405099555937545,
      "grad_norm": 1.5644104480743408,
      "learning_rate": 0.00019540361466635777,
      "loss": 1.5176,
      "step": 866
    },
    {
      "epoch": 0.12419424151267727,
      "grad_norm": 1.3992775678634644,
      "learning_rate": 0.00019538969937935044,
      "loss": 1.7475,
      "step": 867
    },
    {
      "epoch": 0.12433748746597909,
      "grad_norm": 1.1582415103912354,
      "learning_rate": 0.00019537576355730134,
      "loss": 1.601,
      "step": 868
    },
    {
      "epoch": 0.12448073341928091,
      "grad_norm": 1.3258936405181885,
      "learning_rate": 0.00019536180720321054,
      "loss": 1.5256,
      "step": 869
    },
    {
      "epoch": 0.12462397937258272,
      "grad_norm": 1.5257527828216553,
      "learning_rate": 0.00019534783032008248,
      "loss": 1.6944,
      "step": 870
    },
    {
      "epoch": 0.12476722532588454,
      "grad_norm": 1.726912260055542,
      "learning_rate": 0.00019533383291092606,
      "loss": 1.5629,
      "step": 871
    },
    {
      "epoch": 0.12491047127918636,
      "grad_norm": 1.3510074615478516,
      "learning_rate": 0.00019531981497875454,
      "loss": 1.5718,
      "step": 872
    },
    {
      "epoch": 0.12505371723248818,
      "grad_norm": 1.416359543800354,
      "learning_rate": 0.00019530577652658568,
      "loss": 1.3863,
      "step": 873
    },
    {
      "epoch": 0.12519696318579,
      "grad_norm": 1.2144155502319336,
      "learning_rate": 0.00019529171755744158,
      "loss": 1.5797,
      "step": 874
    },
    {
      "epoch": 0.12534020913909183,
      "grad_norm": 1.5377811193466187,
      "learning_rate": 0.00019527763807434878,
      "loss": 1.6465,
      "step": 875
    },
    {
      "epoch": 0.12548345509239364,
      "grad_norm": 1.4994487762451172,
      "learning_rate": 0.00019526353808033825,
      "loss": 1.7088,
      "step": 876
    },
    {
      "epoch": 0.12562670104569545,
      "grad_norm": 1.1898484230041504,
      "learning_rate": 0.0001952494175784454,
      "loss": 1.441,
      "step": 877
    },
    {
      "epoch": 0.12576994699899727,
      "grad_norm": 1.889299988746643,
      "learning_rate": 0.00019523527657171,
      "loss": 1.5037,
      "step": 878
    },
    {
      "epoch": 0.1259131929522991,
      "grad_norm": 1.0025755167007446,
      "learning_rate": 0.00019522111506317625,
      "loss": 1.5566,
      "step": 879
    },
    {
      "epoch": 0.12605643890560092,
      "grad_norm": 1.5120162963867188,
      "learning_rate": 0.00019520693305589282,
      "loss": 1.4998,
      "step": 880
    },
    {
      "epoch": 0.12619968485890273,
      "grad_norm": 1.5454044342041016,
      "learning_rate": 0.00019519273055291266,
      "loss": 1.5053,
      "step": 881
    },
    {
      "epoch": 0.12634293081220455,
      "grad_norm": 1.479720950126648,
      "learning_rate": 0.0001951785075572933,
      "loss": 1.3877,
      "step": 882
    },
    {
      "epoch": 0.1264861767655064,
      "grad_norm": 1.395705223083496,
      "learning_rate": 0.00019516426407209652,
      "loss": 1.5334,
      "step": 883
    },
    {
      "epoch": 0.1266294227188082,
      "grad_norm": 1.248108148574829,
      "learning_rate": 0.0001951500001003887,
      "loss": 1.6191,
      "step": 884
    },
    {
      "epoch": 0.12677266867211,
      "grad_norm": 1.399366021156311,
      "learning_rate": 0.0001951357156452404,
      "loss": 1.6154,
      "step": 885
    },
    {
      "epoch": 0.12691591462541182,
      "grad_norm": 1.459180235862732,
      "learning_rate": 0.00019512141070972678,
      "loss": 1.7004,
      "step": 886
    },
    {
      "epoch": 0.12705916057871366,
      "grad_norm": 1.526006817817688,
      "learning_rate": 0.00019510708529692735,
      "loss": 1.448,
      "step": 887
    },
    {
      "epoch": 0.12720240653201548,
      "grad_norm": 1.6083557605743408,
      "learning_rate": 0.00019509273940992596,
      "loss": 1.49,
      "step": 888
    },
    {
      "epoch": 0.1273456524853173,
      "grad_norm": 1.349533200263977,
      "learning_rate": 0.00019507837305181096,
      "loss": 1.5824,
      "step": 889
    },
    {
      "epoch": 0.1274888984386191,
      "grad_norm": 1.4738569259643555,
      "learning_rate": 0.00019506398622567509,
      "loss": 1.6849,
      "step": 890
    },
    {
      "epoch": 0.12763214439192092,
      "grad_norm": 1.4363090991973877,
      "learning_rate": 0.00019504957893461545,
      "loss": 1.4545,
      "step": 891
    },
    {
      "epoch": 0.12777539034522276,
      "grad_norm": 1.1765127182006836,
      "learning_rate": 0.00019503515118173353,
      "loss": 1.5411,
      "step": 892
    },
    {
      "epoch": 0.12791863629852457,
      "grad_norm": 1.5619832277297974,
      "learning_rate": 0.00019502070297013538,
      "loss": 1.6108,
      "step": 893
    },
    {
      "epoch": 0.12806188225182638,
      "grad_norm": 1.396733045578003,
      "learning_rate": 0.0001950062343029312,
      "loss": 1.627,
      "step": 894
    },
    {
      "epoch": 0.1282051282051282,
      "grad_norm": 1.3025591373443604,
      "learning_rate": 0.00019499174518323588,
      "loss": 1.67,
      "step": 895
    },
    {
      "epoch": 0.12834837415843003,
      "grad_norm": 1.5245089530944824,
      "learning_rate": 0.0001949772356141685,
      "loss": 1.5114,
      "step": 896
    },
    {
      "epoch": 0.12849162011173185,
      "grad_norm": 1.7307344675064087,
      "learning_rate": 0.0001949627055988526,
      "loss": 1.469,
      "step": 897
    },
    {
      "epoch": 0.12863486606503366,
      "grad_norm": 1.5148040056228638,
      "learning_rate": 0.00019494815514041613,
      "loss": 1.4896,
      "step": 898
    },
    {
      "epoch": 0.12877811201833547,
      "grad_norm": 1.3124440908432007,
      "learning_rate": 0.00019493358424199148,
      "loss": 1.6682,
      "step": 899
    },
    {
      "epoch": 0.1289213579716373,
      "grad_norm": 1.2774840593338013,
      "learning_rate": 0.00019491899290671535,
      "loss": 1.7009,
      "step": 900
    },
    {
      "epoch": 0.12906460392493913,
      "grad_norm": 1.3566539287567139,
      "learning_rate": 0.00019490438113772896,
      "loss": 1.6027,
      "step": 901
    },
    {
      "epoch": 0.12920784987824094,
      "grad_norm": 1.3069580793380737,
      "learning_rate": 0.00019488974893817784,
      "loss": 1.5501,
      "step": 902
    },
    {
      "epoch": 0.12935109583154275,
      "grad_norm": 1.5154794454574585,
      "learning_rate": 0.00019487509631121192,
      "loss": 1.7287,
      "step": 903
    },
    {
      "epoch": 0.1294943417848446,
      "grad_norm": 1.51203191280365,
      "learning_rate": 0.00019486042325998556,
      "loss": 1.4113,
      "step": 904
    },
    {
      "epoch": 0.1296375877381464,
      "grad_norm": 1.2378709316253662,
      "learning_rate": 0.0001948457297876575,
      "loss": 1.443,
      "step": 905
    },
    {
      "epoch": 0.12978083369144822,
      "grad_norm": 1.6402249336242676,
      "learning_rate": 0.00019483101589739084,
      "loss": 1.5459,
      "step": 906
    },
    {
      "epoch": 0.12992407964475003,
      "grad_norm": 1.5180134773254395,
      "learning_rate": 0.00019481628159235322,
      "loss": 1.553,
      "step": 907
    },
    {
      "epoch": 0.13006732559805184,
      "grad_norm": 1.2810626029968262,
      "learning_rate": 0.0001948015268757165,
      "loss": 1.6266,
      "step": 908
    },
    {
      "epoch": 0.13021057155135368,
      "grad_norm": 1.0726395845413208,
      "learning_rate": 0.00019478675175065702,
      "loss": 1.7729,
      "step": 909
    },
    {
      "epoch": 0.1303538175046555,
      "grad_norm": 1.4035894870758057,
      "learning_rate": 0.0001947719562203555,
      "loss": 1.5281,
      "step": 910
    },
    {
      "epoch": 0.1304970634579573,
      "grad_norm": 1.4306449890136719,
      "learning_rate": 0.00019475714028799703,
      "loss": 1.517,
      "step": 911
    },
    {
      "epoch": 0.13064030941125912,
      "grad_norm": 1.2066540718078613,
      "learning_rate": 0.0001947423039567711,
      "loss": 1.6117,
      "step": 912
    },
    {
      "epoch": 0.13078355536456096,
      "grad_norm": 1.2805730104446411,
      "learning_rate": 0.0001947274472298717,
      "loss": 1.6499,
      "step": 913
    },
    {
      "epoch": 0.13092680131786277,
      "grad_norm": 1.3742079734802246,
      "learning_rate": 0.00019471257011049702,
      "loss": 1.5715,
      "step": 914
    },
    {
      "epoch": 0.13107004727116459,
      "grad_norm": 1.5260566473007202,
      "learning_rate": 0.00019469767260184975,
      "loss": 1.5409,
      "step": 915
    },
    {
      "epoch": 0.1312132932244664,
      "grad_norm": 1.415353775024414,
      "learning_rate": 0.000194682754707137,
      "loss": 1.5905,
      "step": 916
    },
    {
      "epoch": 0.13135653917776824,
      "grad_norm": 1.3228139877319336,
      "learning_rate": 0.0001946678164295702,
      "loss": 1.4905,
      "step": 917
    },
    {
      "epoch": 0.13149978513107005,
      "grad_norm": 1.3434737920761108,
      "learning_rate": 0.0001946528577723652,
      "loss": 1.5738,
      "step": 918
    },
    {
      "epoch": 0.13164303108437186,
      "grad_norm": 1.454965591430664,
      "learning_rate": 0.00019463787873874217,
      "loss": 1.5264,
      "step": 919
    },
    {
      "epoch": 0.13178627703767368,
      "grad_norm": 1.6001193523406982,
      "learning_rate": 0.0001946228793319258,
      "loss": 1.7512,
      "step": 920
    },
    {
      "epoch": 0.13192952299097552,
      "grad_norm": 1.5035537481307983,
      "learning_rate": 0.00019460785955514504,
      "loss": 1.5072,
      "step": 921
    },
    {
      "epoch": 0.13207276894427733,
      "grad_norm": 1.1897765398025513,
      "learning_rate": 0.00019459281941163332,
      "loss": 1.498,
      "step": 922
    },
    {
      "epoch": 0.13221601489757914,
      "grad_norm": 1.2392112016677856,
      "learning_rate": 0.00019457775890462838,
      "loss": 1.3123,
      "step": 923
    },
    {
      "epoch": 0.13235926085088096,
      "grad_norm": 1.2779078483581543,
      "learning_rate": 0.00019456267803737243,
      "loss": 1.6161,
      "step": 924
    },
    {
      "epoch": 0.13250250680418277,
      "grad_norm": 1.4699326753616333,
      "learning_rate": 0.0001945475768131119,
      "loss": 1.6932,
      "step": 925
    },
    {
      "epoch": 0.1326457527574846,
      "grad_norm": 1.2128102779388428,
      "learning_rate": 0.00019453245523509777,
      "loss": 1.6362,
      "step": 926
    },
    {
      "epoch": 0.13278899871078642,
      "grad_norm": 1.7061197757720947,
      "learning_rate": 0.0001945173133065854,
      "loss": 1.57,
      "step": 927
    },
    {
      "epoch": 0.13293224466408823,
      "grad_norm": 1.182349681854248,
      "learning_rate": 0.00019450215103083437,
      "loss": 1.7028,
      "step": 928
    },
    {
      "epoch": 0.13307549061739005,
      "grad_norm": 1.5013049840927124,
      "learning_rate": 0.0001944869684111088,
      "loss": 1.4712,
      "step": 929
    },
    {
      "epoch": 0.1332187365706919,
      "grad_norm": 1.7043782472610474,
      "learning_rate": 0.00019447176545067711,
      "loss": 1.6052,
      "step": 930
    },
    {
      "epoch": 0.1333619825239937,
      "grad_norm": 1.298484206199646,
      "learning_rate": 0.00019445654215281214,
      "loss": 1.4566,
      "step": 931
    },
    {
      "epoch": 0.1335052284772955,
      "grad_norm": 1.3367338180541992,
      "learning_rate": 0.0001944412985207911,
      "loss": 1.5929,
      "step": 932
    },
    {
      "epoch": 0.13364847443059732,
      "grad_norm": 1.4428350925445557,
      "learning_rate": 0.0001944260345578955,
      "loss": 1.5432,
      "step": 933
    },
    {
      "epoch": 0.13379172038389917,
      "grad_norm": 1.440341830253601,
      "learning_rate": 0.00019441075026741138,
      "loss": 1.6273,
      "step": 934
    },
    {
      "epoch": 0.13393496633720098,
      "grad_norm": 1.4038457870483398,
      "learning_rate": 0.00019439544565262904,
      "loss": 1.6491,
      "step": 935
    },
    {
      "epoch": 0.1340782122905028,
      "grad_norm": 1.4589582681655884,
      "learning_rate": 0.00019438012071684314,
      "loss": 1.6027,
      "step": 936
    },
    {
      "epoch": 0.1342214582438046,
      "grad_norm": 1.6593842506408691,
      "learning_rate": 0.0001943647754633528,
      "loss": 1.5932,
      "step": 937
    },
    {
      "epoch": 0.13436470419710644,
      "grad_norm": 1.4873721599578857,
      "learning_rate": 0.0001943494098954615,
      "loss": 1.6837,
      "step": 938
    },
    {
      "epoch": 0.13450795015040826,
      "grad_norm": 1.2693896293640137,
      "learning_rate": 0.000194334024016477,
      "loss": 1.613,
      "step": 939
    },
    {
      "epoch": 0.13465119610371007,
      "grad_norm": 1.2949308156967163,
      "learning_rate": 0.00019431861782971156,
      "loss": 1.6487,
      "step": 940
    },
    {
      "epoch": 0.13479444205701188,
      "grad_norm": 1.2915406227111816,
      "learning_rate": 0.0001943031913384817,
      "loss": 1.6778,
      "step": 941
    },
    {
      "epoch": 0.13493768801031372,
      "grad_norm": 1.363283395767212,
      "learning_rate": 0.00019428774454610843,
      "loss": 1.6879,
      "step": 942
    },
    {
      "epoch": 0.13508093396361553,
      "grad_norm": 1.5493043661117554,
      "learning_rate": 0.000194272277455917,
      "loss": 1.5816,
      "step": 943
    },
    {
      "epoch": 0.13522417991691735,
      "grad_norm": 1.6780341863632202,
      "learning_rate": 0.0001942567900712371,
      "loss": 1.7223,
      "step": 944
    },
    {
      "epoch": 0.13536742587021916,
      "grad_norm": 1.1107063293457031,
      "learning_rate": 0.00019424128239540277,
      "loss": 1.5271,
      "step": 945
    },
    {
      "epoch": 0.13551067182352097,
      "grad_norm": 1.5544624328613281,
      "learning_rate": 0.0001942257544317525,
      "loss": 1.6354,
      "step": 946
    },
    {
      "epoch": 0.1356539177768228,
      "grad_norm": 1.163934350013733,
      "learning_rate": 0.00019421020618362898,
      "loss": 1.8291,
      "step": 947
    },
    {
      "epoch": 0.13579716373012463,
      "grad_norm": 1.355676531791687,
      "learning_rate": 0.00019419463765437943,
      "loss": 1.6046,
      "step": 948
    },
    {
      "epoch": 0.13594040968342644,
      "grad_norm": 1.3692798614501953,
      "learning_rate": 0.00019417904884735533,
      "loss": 1.551,
      "step": 949
    },
    {
      "epoch": 0.13608365563672825,
      "grad_norm": 1.3626713752746582,
      "learning_rate": 0.00019416343976591261,
      "loss": 1.4371,
      "step": 950
    },
    {
      "epoch": 0.1362269015900301,
      "grad_norm": 1.5438525676727295,
      "learning_rate": 0.0001941478104134115,
      "loss": 1.4228,
      "step": 951
    },
    {
      "epoch": 0.1363701475433319,
      "grad_norm": 1.2169398069381714,
      "learning_rate": 0.00019413216079321654,
      "loss": 1.6035,
      "step": 952
    },
    {
      "epoch": 0.13651339349663372,
      "grad_norm": 1.4688122272491455,
      "learning_rate": 0.00019411649090869684,
      "loss": 1.5359,
      "step": 953
    },
    {
      "epoch": 0.13665663944993553,
      "grad_norm": 1.2406110763549805,
      "learning_rate": 0.00019410080076322564,
      "loss": 1.5632,
      "step": 954
    },
    {
      "epoch": 0.13679988540323737,
      "grad_norm": 1.519088625907898,
      "learning_rate": 0.00019408509036018066,
      "loss": 1.4936,
      "step": 955
    },
    {
      "epoch": 0.13694313135653918,
      "grad_norm": 1.0677958726882935,
      "learning_rate": 0.00019406935970294397,
      "loss": 1.645,
      "step": 956
    },
    {
      "epoch": 0.137086377309841,
      "grad_norm": 1.0214921236038208,
      "learning_rate": 0.00019405360879490202,
      "loss": 1.5606,
      "step": 957
    },
    {
      "epoch": 0.1372296232631428,
      "grad_norm": 1.314568281173706,
      "learning_rate": 0.00019403783763944556,
      "loss": 1.6634,
      "step": 958
    },
    {
      "epoch": 0.13737286921644465,
      "grad_norm": 1.625647783279419,
      "learning_rate": 0.0001940220462399697,
      "loss": 1.6296,
      "step": 959
    },
    {
      "epoch": 0.13751611516974646,
      "grad_norm": 1.4990005493164062,
      "learning_rate": 0.000194006234599874,
      "loss": 1.6438,
      "step": 960
    },
    {
      "epoch": 0.13765936112304827,
      "grad_norm": 1.4645360708236694,
      "learning_rate": 0.00019399040272256225,
      "loss": 1.4934,
      "step": 961
    },
    {
      "epoch": 0.1378026070763501,
      "grad_norm": 1.3728829622268677,
      "learning_rate": 0.00019397455061144272,
      "loss": 1.545,
      "step": 962
    },
    {
      "epoch": 0.1379458530296519,
      "grad_norm": 1.4249588251113892,
      "learning_rate": 0.00019395867826992795,
      "loss": 1.535,
      "step": 963
    },
    {
      "epoch": 0.13808909898295374,
      "grad_norm": 1.1676795482635498,
      "learning_rate": 0.00019394278570143488,
      "loss": 1.7477,
      "step": 964
    },
    {
      "epoch": 0.13823234493625555,
      "grad_norm": 1.1489231586456299,
      "learning_rate": 0.00019392687290938475,
      "loss": 1.6569,
      "step": 965
    },
    {
      "epoch": 0.13837559088955736,
      "grad_norm": 1.5753334760665894,
      "learning_rate": 0.00019391093989720322,
      "loss": 1.4576,
      "step": 966
    },
    {
      "epoch": 0.13851883684285918,
      "grad_norm": 1.1450971364974976,
      "learning_rate": 0.00019389498666832025,
      "loss": 1.637,
      "step": 967
    },
    {
      "epoch": 0.13866208279616102,
      "grad_norm": 1.655789852142334,
      "learning_rate": 0.0001938790132261702,
      "loss": 1.6492,
      "step": 968
    },
    {
      "epoch": 0.13880532874946283,
      "grad_norm": 1.4353446960449219,
      "learning_rate": 0.00019386301957419172,
      "loss": 1.5714,
      "step": 969
    },
    {
      "epoch": 0.13894857470276464,
      "grad_norm": 1.4906525611877441,
      "learning_rate": 0.00019384700571582793,
      "loss": 1.4296,
      "step": 970
    },
    {
      "epoch": 0.13909182065606646,
      "grad_norm": 1.2638425827026367,
      "learning_rate": 0.00019383097165452613,
      "loss": 1.68,
      "step": 971
    },
    {
      "epoch": 0.1392350666093683,
      "grad_norm": 1.3439027070999146,
      "learning_rate": 0.0001938149173937381,
      "loss": 1.7358,
      "step": 972
    },
    {
      "epoch": 0.1393783125626701,
      "grad_norm": 1.5346641540527344,
      "learning_rate": 0.0001937988429369199,
      "loss": 1.6117,
      "step": 973
    },
    {
      "epoch": 0.13952155851597192,
      "grad_norm": 1.3425935506820679,
      "learning_rate": 0.000193782748287532,
      "loss": 1.7291,
      "step": 974
    },
    {
      "epoch": 0.13966480446927373,
      "grad_norm": 1.5102351903915405,
      "learning_rate": 0.00019376663344903913,
      "loss": 1.6585,
      "step": 975
    },
    {
      "epoch": 0.13980805042257557,
      "grad_norm": 1.4603413343429565,
      "learning_rate": 0.00019375049842491047,
      "loss": 1.4841,
      "step": 976
    },
    {
      "epoch": 0.1399512963758774,
      "grad_norm": 1.1883057355880737,
      "learning_rate": 0.00019373434321861942,
      "loss": 1.588,
      "step": 977
    },
    {
      "epoch": 0.1400945423291792,
      "grad_norm": 1.2276684045791626,
      "learning_rate": 0.00019371816783364388,
      "loss": 1.6725,
      "step": 978
    },
    {
      "epoch": 0.140237788282481,
      "grad_norm": 1.2830169200897217,
      "learning_rate": 0.00019370197227346596,
      "loss": 1.5397,
      "step": 979
    },
    {
      "epoch": 0.14038103423578283,
      "grad_norm": 1.1476396322250366,
      "learning_rate": 0.00019368575654157217,
      "loss": 1.5355,
      "step": 980
    },
    {
      "epoch": 0.14052428018908467,
      "grad_norm": 1.2300249338150024,
      "learning_rate": 0.00019366952064145334,
      "loss": 1.5267,
      "step": 981
    },
    {
      "epoch": 0.14066752614238648,
      "grad_norm": 1.5411263704299927,
      "learning_rate": 0.00019365326457660472,
      "loss": 1.5968,
      "step": 982
    },
    {
      "epoch": 0.1408107720956883,
      "grad_norm": 1.390707015991211,
      "learning_rate": 0.00019363698835052576,
      "loss": 1.5561,
      "step": 983
    },
    {
      "epoch": 0.1409540180489901,
      "grad_norm": 1.2429877519607544,
      "learning_rate": 0.00019362069196672037,
      "loss": 1.6717,
      "step": 984
    },
    {
      "epoch": 0.14109726400229194,
      "grad_norm": 1.4149080514907837,
      "learning_rate": 0.00019360437542869676,
      "loss": 1.5962,
      "step": 985
    },
    {
      "epoch": 0.14124050995559376,
      "grad_norm": 1.152044415473938,
      "learning_rate": 0.00019358803873996747,
      "loss": 1.5248,
      "step": 986
    },
    {
      "epoch": 0.14138375590889557,
      "grad_norm": 1.3117223978042603,
      "learning_rate": 0.00019357168190404936,
      "loss": 1.4728,
      "step": 987
    },
    {
      "epoch": 0.14152700186219738,
      "grad_norm": 1.118538737297058,
      "learning_rate": 0.0001935553049244637,
      "loss": 1.6256,
      "step": 988
    },
    {
      "epoch": 0.14167024781549922,
      "grad_norm": 1.24899160861969,
      "learning_rate": 0.00019353890780473602,
      "loss": 1.6468,
      "step": 989
    },
    {
      "epoch": 0.14181349376880104,
      "grad_norm": 1.4963958263397217,
      "learning_rate": 0.00019352249054839624,
      "loss": 1.4806,
      "step": 990
    },
    {
      "epoch": 0.14195673972210285,
      "grad_norm": 1.4605729579925537,
      "learning_rate": 0.00019350605315897852,
      "loss": 1.5374,
      "step": 991
    },
    {
      "epoch": 0.14209998567540466,
      "grad_norm": 1.2262187004089355,
      "learning_rate": 0.00019348959564002152,
      "loss": 1.4389,
      "step": 992
    },
    {
      "epoch": 0.1422432316287065,
      "grad_norm": 1.3125232458114624,
      "learning_rate": 0.00019347311799506803,
      "loss": 1.6267,
      "step": 993
    },
    {
      "epoch": 0.14238647758200831,
      "grad_norm": 1.4034234285354614,
      "learning_rate": 0.0001934566202276654,
      "loss": 1.7271,
      "step": 994
    },
    {
      "epoch": 0.14252972353531013,
      "grad_norm": 1.2048908472061157,
      "learning_rate": 0.00019344010234136508,
      "loss": 1.5188,
      "step": 995
    },
    {
      "epoch": 0.14267296948861194,
      "grad_norm": 1.2334035634994507,
      "learning_rate": 0.00019342356433972303,
      "loss": 1.5066,
      "step": 996
    },
    {
      "epoch": 0.14281621544191378,
      "grad_norm": 1.190976619720459,
      "learning_rate": 0.00019340700622629946,
      "loss": 1.7507,
      "step": 997
    },
    {
      "epoch": 0.1429594613952156,
      "grad_norm": 1.717379093170166,
      "learning_rate": 0.00019339042800465889,
      "loss": 1.5083,
      "step": 998
    },
    {
      "epoch": 0.1431027073485174,
      "grad_norm": 1.454601526260376,
      "learning_rate": 0.00019337382967837024,
      "loss": 1.5076,
      "step": 999
    },
    {
      "epoch": 0.14324595330181922,
      "grad_norm": 1.2652497291564941,
      "learning_rate": 0.0001933572112510067,
      "loss": 1.4765,
      "step": 1000
    },
    {
      "epoch": 0.14338919925512103,
      "grad_norm": 1.0453784465789795,
      "learning_rate": 0.0001933405727261458,
      "loss": 1.6082,
      "step": 1001
    },
    {
      "epoch": 0.14353244520842287,
      "grad_norm": 1.3370368480682373,
      "learning_rate": 0.0001933239141073694,
      "loss": 1.3811,
      "step": 1002
    },
    {
      "epoch": 0.14367569116172468,
      "grad_norm": 1.414880394935608,
      "learning_rate": 0.00019330723539826375,
      "loss": 1.4487,
      "step": 1003
    },
    {
      "epoch": 0.1438189371150265,
      "grad_norm": 1.3539701700210571,
      "learning_rate": 0.00019329053660241928,
      "loss": 1.5082,
      "step": 1004
    },
    {
      "epoch": 0.1439621830683283,
      "grad_norm": 1.5669071674346924,
      "learning_rate": 0.00019327381772343087,
      "loss": 1.6623,
      "step": 1005
    },
    {
      "epoch": 0.14410542902163015,
      "grad_norm": 1.0910766124725342,
      "learning_rate": 0.00019325707876489766,
      "loss": 1.5005,
      "step": 1006
    },
    {
      "epoch": 0.14424867497493196,
      "grad_norm": 1.6274842023849487,
      "learning_rate": 0.00019324031973042317,
      "loss": 1.4766,
      "step": 1007
    },
    {
      "epoch": 0.14439192092823377,
      "grad_norm": 1.2391437292099,
      "learning_rate": 0.00019322354062361517,
      "loss": 1.5682,
      "step": 1008
    },
    {
      "epoch": 0.1445351668815356,
      "grad_norm": 1.514203667640686,
      "learning_rate": 0.0001932067414480858,
      "loss": 1.6752,
      "step": 1009
    },
    {
      "epoch": 0.14467841283483743,
      "grad_norm": 1.1302251815795898,
      "learning_rate": 0.00019318992220745149,
      "loss": 1.7205,
      "step": 1010
    },
    {
      "epoch": 0.14482165878813924,
      "grad_norm": 1.638452172279358,
      "learning_rate": 0.00019317308290533306,
      "loss": 1.5628,
      "step": 1011
    },
    {
      "epoch": 0.14496490474144105,
      "grad_norm": 1.5972185134887695,
      "learning_rate": 0.00019315622354535553,
      "loss": 1.2919,
      "step": 1012
    },
    {
      "epoch": 0.14510815069474287,
      "grad_norm": 1.8035101890563965,
      "learning_rate": 0.00019313934413114832,
      "loss": 1.4733,
      "step": 1013
    },
    {
      "epoch": 0.1452513966480447,
      "grad_norm": 1.3459478616714478,
      "learning_rate": 0.00019312244466634517,
      "loss": 1.5086,
      "step": 1014
    },
    {
      "epoch": 0.14539464260134652,
      "grad_norm": 1.4896929264068604,
      "learning_rate": 0.0001931055251545841,
      "loss": 1.7146,
      "step": 1015
    },
    {
      "epoch": 0.14553788855464833,
      "grad_norm": 1.537964105606079,
      "learning_rate": 0.00019308858559950748,
      "loss": 1.6314,
      "step": 1016
    },
    {
      "epoch": 0.14568113450795014,
      "grad_norm": 1.323756217956543,
      "learning_rate": 0.00019307162600476195,
      "loss": 1.5025,
      "step": 1017
    },
    {
      "epoch": 0.14582438046125196,
      "grad_norm": 1.3886237144470215,
      "learning_rate": 0.00019305464637399853,
      "loss": 1.4745,
      "step": 1018
    },
    {
      "epoch": 0.1459676264145538,
      "grad_norm": 1.4149054288864136,
      "learning_rate": 0.00019303764671087245,
      "loss": 1.5865,
      "step": 1019
    },
    {
      "epoch": 0.1461108723678556,
      "grad_norm": 1.5493627786636353,
      "learning_rate": 0.0001930206270190434,
      "loss": 1.5541,
      "step": 1020
    },
    {
      "epoch": 0.14625411832115742,
      "grad_norm": 1.4834001064300537,
      "learning_rate": 0.00019300358730217526,
      "loss": 1.4414,
      "step": 1021
    },
    {
      "epoch": 0.14639736427445924,
      "grad_norm": 1.583128571510315,
      "learning_rate": 0.0001929865275639362,
      "loss": 1.5788,
      "step": 1022
    },
    {
      "epoch": 0.14654061022776108,
      "grad_norm": 1.4072614908218384,
      "learning_rate": 0.00019296944780799885,
      "loss": 1.5639,
      "step": 1023
    },
    {
      "epoch": 0.1466838561810629,
      "grad_norm": 1.1666544675827026,
      "learning_rate": 0.00019295234803804004,
      "loss": 1.57,
      "step": 1024
    },
    {
      "epoch": 0.1468271021343647,
      "grad_norm": 1.5429679155349731,
      "learning_rate": 0.0001929352282577409,
      "loss": 1.551,
      "step": 1025
    },
    {
      "epoch": 0.1469703480876665,
      "grad_norm": 1.1465579271316528,
      "learning_rate": 0.0001929180884707869,
      "loss": 1.6996,
      "step": 1026
    },
    {
      "epoch": 0.14711359404096835,
      "grad_norm": 1.430687427520752,
      "learning_rate": 0.0001929009286808678,
      "loss": 1.5423,
      "step": 1027
    },
    {
      "epoch": 0.14725683999427017,
      "grad_norm": 1.2124978303909302,
      "learning_rate": 0.0001928837488916777,
      "loss": 1.4928,
      "step": 1028
    },
    {
      "epoch": 0.14740008594757198,
      "grad_norm": 1.6632754802703857,
      "learning_rate": 0.00019286654910691503,
      "loss": 1.7412,
      "step": 1029
    },
    {
      "epoch": 0.1475433319008738,
      "grad_norm": 1.1591392755508423,
      "learning_rate": 0.0001928493293302824,
      "loss": 1.609,
      "step": 1030
    },
    {
      "epoch": 0.14768657785417563,
      "grad_norm": 1.2420518398284912,
      "learning_rate": 0.0001928320895654868,
      "loss": 1.5294,
      "step": 1031
    },
    {
      "epoch": 0.14782982380747745,
      "grad_norm": 1.6039559841156006,
      "learning_rate": 0.00019281482981623957,
      "loss": 1.6582,
      "step": 1032
    },
    {
      "epoch": 0.14797306976077926,
      "grad_norm": 1.3535301685333252,
      "learning_rate": 0.0001927975500862563,
      "loss": 1.6827,
      "step": 1033
    },
    {
      "epoch": 0.14811631571408107,
      "grad_norm": 1.2901084423065186,
      "learning_rate": 0.00019278025037925689,
      "loss": 1.3468,
      "step": 1034
    },
    {
      "epoch": 0.14825956166738288,
      "grad_norm": 1.0611035823822021,
      "learning_rate": 0.00019276293069896548,
      "loss": 1.7171,
      "step": 1035
    },
    {
      "epoch": 0.14840280762068472,
      "grad_norm": 1.4382847547531128,
      "learning_rate": 0.00019274559104911067,
      "loss": 1.656,
      "step": 1036
    },
    {
      "epoch": 0.14854605357398654,
      "grad_norm": 1.5861812829971313,
      "learning_rate": 0.0001927282314334252,
      "loss": 1.6,
      "step": 1037
    },
    {
      "epoch": 0.14868929952728835,
      "grad_norm": 1.4645975828170776,
      "learning_rate": 0.00019271085185564615,
      "loss": 1.5686,
      "step": 1038
    },
    {
      "epoch": 0.14883254548059016,
      "grad_norm": 1.5574449300765991,
      "learning_rate": 0.00019269345231951493,
      "loss": 1.6133,
      "step": 1039
    },
    {
      "epoch": 0.148975791433892,
      "grad_norm": 1.3897409439086914,
      "learning_rate": 0.00019267603282877724,
      "loss": 1.4349,
      "step": 1040
    },
    {
      "epoch": 0.14911903738719381,
      "grad_norm": 1.7466847896575928,
      "learning_rate": 0.00019265859338718304,
      "loss": 1.5295,
      "step": 1041
    },
    {
      "epoch": 0.14926228334049563,
      "grad_norm": 1.3798412084579468,
      "learning_rate": 0.00019264113399848664,
      "loss": 1.543,
      "step": 1042
    },
    {
      "epoch": 0.14940552929379744,
      "grad_norm": 1.1986595392227173,
      "learning_rate": 0.0001926236546664466,
      "loss": 1.6707,
      "step": 1043
    },
    {
      "epoch": 0.14954877524709928,
      "grad_norm": 1.080202579498291,
      "learning_rate": 0.00019260615539482584,
      "loss": 1.6254,
      "step": 1044
    },
    {
      "epoch": 0.1496920212004011,
      "grad_norm": 1.2281359434127808,
      "learning_rate": 0.0001925886361873914,
      "loss": 1.6682,
      "step": 1045
    },
    {
      "epoch": 0.1498352671537029,
      "grad_norm": 1.5164756774902344,
      "learning_rate": 0.00019257109704791484,
      "loss": 1.6561,
      "step": 1046
    },
    {
      "epoch": 0.14997851310700472,
      "grad_norm": 1.232721209526062,
      "learning_rate": 0.00019255353798017184,
      "loss": 1.7319,
      "step": 1047
    },
    {
      "epoch": 0.15012175906030656,
      "grad_norm": 1.3041349649429321,
      "learning_rate": 0.00019253595898794247,
      "loss": 1.449,
      "step": 1048
    },
    {
      "epoch": 0.15026500501360837,
      "grad_norm": 1.3203157186508179,
      "learning_rate": 0.00019251836007501102,
      "loss": 1.5185,
      "step": 1049
    },
    {
      "epoch": 0.15040825096691018,
      "grad_norm": 1.2182791233062744,
      "learning_rate": 0.00019250074124516618,
      "loss": 1.6029,
      "step": 1050
    },
    {
      "epoch": 0.150551496920212,
      "grad_norm": 1.092157006263733,
      "learning_rate": 0.00019248310250220073,
      "loss": 1.4837,
      "step": 1051
    },
    {
      "epoch": 0.1506947428735138,
      "grad_norm": 1.3514152765274048,
      "learning_rate": 0.0001924654438499119,
      "loss": 1.6147,
      "step": 1052
    },
    {
      "epoch": 0.15083798882681565,
      "grad_norm": 1.3528683185577393,
      "learning_rate": 0.00019244776529210122,
      "loss": 1.3918,
      "step": 1053
    },
    {
      "epoch": 0.15098123478011746,
      "grad_norm": 1.3136584758758545,
      "learning_rate": 0.00019243006683257439,
      "loss": 1.5979,
      "step": 1054
    },
    {
      "epoch": 0.15112448073341928,
      "grad_norm": 1.66302490234375,
      "learning_rate": 0.00019241234847514144,
      "loss": 1.591,
      "step": 1055
    },
    {
      "epoch": 0.1512677266867211,
      "grad_norm": 1.2907557487487793,
      "learning_rate": 0.00019239461022361676,
      "loss": 1.5207,
      "step": 1056
    },
    {
      "epoch": 0.15141097264002293,
      "grad_norm": 1.7057480812072754,
      "learning_rate": 0.00019237685208181886,
      "loss": 1.3596,
      "step": 1057
    },
    {
      "epoch": 0.15155421859332474,
      "grad_norm": 1.4090209007263184,
      "learning_rate": 0.0001923590740535707,
      "loss": 1.8182,
      "step": 1058
    },
    {
      "epoch": 0.15169746454662655,
      "grad_norm": 1.323533058166504,
      "learning_rate": 0.00019234127614269943,
      "loss": 1.6263,
      "step": 1059
    },
    {
      "epoch": 0.15184071049992837,
      "grad_norm": 1.1065683364868164,
      "learning_rate": 0.00019232345835303648,
      "loss": 1.5436,
      "step": 1060
    },
    {
      "epoch": 0.1519839564532302,
      "grad_norm": 1.2325598001480103,
      "learning_rate": 0.0001923056206884176,
      "loss": 1.4932,
      "step": 1061
    },
    {
      "epoch": 0.15212720240653202,
      "grad_norm": 1.4053757190704346,
      "learning_rate": 0.0001922877631526828,
      "loss": 1.7089,
      "step": 1062
    },
    {
      "epoch": 0.15227044835983383,
      "grad_norm": 1.3989908695220947,
      "learning_rate": 0.0001922698857496764,
      "loss": 1.4073,
      "step": 1063
    },
    {
      "epoch": 0.15241369431313564,
      "grad_norm": 1.4045591354370117,
      "learning_rate": 0.0001922519884832469,
      "loss": 1.5376,
      "step": 1064
    },
    {
      "epoch": 0.15255694026643749,
      "grad_norm": 1.3122562170028687,
      "learning_rate": 0.00019223407135724713,
      "loss": 1.4856,
      "step": 1065
    },
    {
      "epoch": 0.1527001862197393,
      "grad_norm": 1.3495731353759766,
      "learning_rate": 0.00019221613437553426,
      "loss": 1.5055,
      "step": 1066
    },
    {
      "epoch": 0.1528434321730411,
      "grad_norm": 1.4818278551101685,
      "learning_rate": 0.00019219817754196966,
      "loss": 1.6641,
      "step": 1067
    },
    {
      "epoch": 0.15298667812634292,
      "grad_norm": 1.2352901697158813,
      "learning_rate": 0.00019218020086041898,
      "loss": 1.6724,
      "step": 1068
    },
    {
      "epoch": 0.15312992407964476,
      "grad_norm": 1.309185266494751,
      "learning_rate": 0.00019216220433475214,
      "loss": 1.5106,
      "step": 1069
    },
    {
      "epoch": 0.15327317003294658,
      "grad_norm": 1.2199681997299194,
      "learning_rate": 0.0001921441879688434,
      "loss": 1.6603,
      "step": 1070
    },
    {
      "epoch": 0.1534164159862484,
      "grad_norm": 1.2713277339935303,
      "learning_rate": 0.00019212615176657116,
      "loss": 1.4,
      "step": 1071
    },
    {
      "epoch": 0.1535596619395502,
      "grad_norm": 1.369075894355774,
      "learning_rate": 0.00019210809573181825,
      "loss": 1.5886,
      "step": 1072
    },
    {
      "epoch": 0.15370290789285201,
      "grad_norm": 1.1621137857437134,
      "learning_rate": 0.00019209001986847163,
      "loss": 1.5208,
      "step": 1073
    },
    {
      "epoch": 0.15384615384615385,
      "grad_norm": 1.0331898927688599,
      "learning_rate": 0.00019207192418042266,
      "loss": 1.4522,
      "step": 1074
    },
    {
      "epoch": 0.15398939979945567,
      "grad_norm": 1.3525487184524536,
      "learning_rate": 0.00019205380867156677,
      "loss": 1.6604,
      "step": 1075
    },
    {
      "epoch": 0.15413264575275748,
      "grad_norm": 1.1547057628631592,
      "learning_rate": 0.0001920356733458039,
      "loss": 1.8436,
      "step": 1076
    },
    {
      "epoch": 0.1542758917060593,
      "grad_norm": 1.2889951467514038,
      "learning_rate": 0.00019201751820703807,
      "loss": 1.5426,
      "step": 1077
    },
    {
      "epoch": 0.15441913765936113,
      "grad_norm": 1.2044836282730103,
      "learning_rate": 0.00019199934325917766,
      "loss": 1.5849,
      "step": 1078
    },
    {
      "epoch": 0.15456238361266295,
      "grad_norm": 1.6906284093856812,
      "learning_rate": 0.00019198114850613524,
      "loss": 1.4109,
      "step": 1079
    },
    {
      "epoch": 0.15470562956596476,
      "grad_norm": 1.5398533344268799,
      "learning_rate": 0.00019196293395182777,
      "loss": 1.5977,
      "step": 1080
    },
    {
      "epoch": 0.15484887551926657,
      "grad_norm": 1.1078940629959106,
      "learning_rate": 0.0001919446996001763,
      "loss": 1.7665,
      "step": 1081
    },
    {
      "epoch": 0.1549921214725684,
      "grad_norm": 1.2386343479156494,
      "learning_rate": 0.00019192644545510635,
      "loss": 1.7485,
      "step": 1082
    },
    {
      "epoch": 0.15513536742587022,
      "grad_norm": 1.2610774040222168,
      "learning_rate": 0.0001919081715205475,
      "loss": 1.7379,
      "step": 1083
    },
    {
      "epoch": 0.15527861337917204,
      "grad_norm": 1.390198826789856,
      "learning_rate": 0.00019188987780043365,
      "loss": 1.4863,
      "step": 1084
    },
    {
      "epoch": 0.15542185933247385,
      "grad_norm": 1.479318380355835,
      "learning_rate": 0.00019187156429870307,
      "loss": 1.5026,
      "step": 1085
    },
    {
      "epoch": 0.1555651052857757,
      "grad_norm": 1.1878783702850342,
      "learning_rate": 0.00019185323101929814,
      "loss": 1.487,
      "step": 1086
    },
    {
      "epoch": 0.1557083512390775,
      "grad_norm": 1.4197934865951538,
      "learning_rate": 0.0001918348779661656,
      "loss": 1.4418,
      "step": 1087
    },
    {
      "epoch": 0.15585159719237932,
      "grad_norm": 1.3864970207214355,
      "learning_rate": 0.00019181650514325641,
      "loss": 1.5701,
      "step": 1088
    },
    {
      "epoch": 0.15599484314568113,
      "grad_norm": 1.2309435606002808,
      "learning_rate": 0.00019179811255452575,
      "loss": 1.5973,
      "step": 1089
    },
    {
      "epoch": 0.15613808909898294,
      "grad_norm": 1.1256530284881592,
      "learning_rate": 0.00019177970020393313,
      "loss": 1.6458,
      "step": 1090
    },
    {
      "epoch": 0.15628133505228478,
      "grad_norm": 1.2956477403640747,
      "learning_rate": 0.00019176126809544223,
      "loss": 1.5702,
      "step": 1091
    },
    {
      "epoch": 0.1564245810055866,
      "grad_norm": 1.2641352415084839,
      "learning_rate": 0.00019174281623302107,
      "loss": 1.5626,
      "step": 1092
    },
    {
      "epoch": 0.1565678269588884,
      "grad_norm": 2.111788272857666,
      "learning_rate": 0.0001917243446206418,
      "loss": 1.5446,
      "step": 1093
    },
    {
      "epoch": 0.15671107291219022,
      "grad_norm": 1.5006740093231201,
      "learning_rate": 0.00019170585326228103,
      "loss": 1.4848,
      "step": 1094
    },
    {
      "epoch": 0.15685431886549206,
      "grad_norm": 1.1169055700302124,
      "learning_rate": 0.00019168734216191936,
      "loss": 1.5926,
      "step": 1095
    },
    {
      "epoch": 0.15699756481879387,
      "grad_norm": 1.3934125900268555,
      "learning_rate": 0.0001916688113235419,
      "loss": 1.467,
      "step": 1096
    },
    {
      "epoch": 0.15714081077209568,
      "grad_norm": 1.2894898653030396,
      "learning_rate": 0.00019165026075113777,
      "loss": 1.5184,
      "step": 1097
    },
    {
      "epoch": 0.1572840567253975,
      "grad_norm": 1.2575918436050415,
      "learning_rate": 0.0001916316904487005,
      "loss": 1.4199,
      "step": 1098
    },
    {
      "epoch": 0.15742730267869934,
      "grad_norm": 1.2589964866638184,
      "learning_rate": 0.00019161310042022778,
      "loss": 1.5166,
      "step": 1099
    },
    {
      "epoch": 0.15757054863200115,
      "grad_norm": 1.171149730682373,
      "learning_rate": 0.00019159449066972163,
      "loss": 1.6936,
      "step": 1100
    },
    {
      "epoch": 0.15771379458530296,
      "grad_norm": 1.1980068683624268,
      "learning_rate": 0.00019157586120118828,
      "loss": 1.6043,
      "step": 1101
    },
    {
      "epoch": 0.15785704053860478,
      "grad_norm": 1.1674002408981323,
      "learning_rate": 0.00019155721201863816,
      "loss": 1.5042,
      "step": 1102
    },
    {
      "epoch": 0.15800028649190662,
      "grad_norm": 2.0021567344665527,
      "learning_rate": 0.00019153854312608593,
      "loss": 1.5049,
      "step": 1103
    },
    {
      "epoch": 0.15814353244520843,
      "grad_norm": 1.520702600479126,
      "learning_rate": 0.00019151985452755063,
      "loss": 1.5655,
      "step": 1104
    },
    {
      "epoch": 0.15828677839851024,
      "grad_norm": 1.4473119974136353,
      "learning_rate": 0.00019150114622705543,
      "loss": 1.7629,
      "step": 1105
    },
    {
      "epoch": 0.15843002435181205,
      "grad_norm": 1.2951511144638062,
      "learning_rate": 0.0001914824182286277,
      "loss": 1.4727,
      "step": 1106
    },
    {
      "epoch": 0.15857327030511387,
      "grad_norm": 1.4768249988555908,
      "learning_rate": 0.00019146367053629918,
      "loss": 1.6218,
      "step": 1107
    },
    {
      "epoch": 0.1587165162584157,
      "grad_norm": 1.0381557941436768,
      "learning_rate": 0.0001914449031541058,
      "loss": 1.4425,
      "step": 1108
    },
    {
      "epoch": 0.15885976221171752,
      "grad_norm": 1.259022831916809,
      "learning_rate": 0.00019142611608608765,
      "loss": 1.4646,
      "step": 1109
    },
    {
      "epoch": 0.15900300816501933,
      "grad_norm": 1.2443740367889404,
      "learning_rate": 0.00019140730933628916,
      "loss": 1.5468,
      "step": 1110
    },
    {
      "epoch": 0.15914625411832115,
      "grad_norm": 1.210586428642273,
      "learning_rate": 0.00019138848290875898,
      "loss": 1.5649,
      "step": 1111
    },
    {
      "epoch": 0.15928950007162299,
      "grad_norm": 1.402271032333374,
      "learning_rate": 0.00019136963680754988,
      "loss": 1.3465,
      "step": 1112
    },
    {
      "epoch": 0.1594327460249248,
      "grad_norm": 1.1137586832046509,
      "learning_rate": 0.00019135077103671908,
      "loss": 1.5549,
      "step": 1113
    },
    {
      "epoch": 0.1595759919782266,
      "grad_norm": 1.2129237651824951,
      "learning_rate": 0.00019133188560032784,
      "loss": 1.5458,
      "step": 1114
    },
    {
      "epoch": 0.15971923793152842,
      "grad_norm": 1.4060721397399902,
      "learning_rate": 0.00019131298050244173,
      "loss": 1.5444,
      "step": 1115
    },
    {
      "epoch": 0.15986248388483026,
      "grad_norm": 1.139362096786499,
      "learning_rate": 0.00019129405574713057,
      "loss": 1.4979,
      "step": 1116
    },
    {
      "epoch": 0.16000572983813208,
      "grad_norm": 1.2220191955566406,
      "learning_rate": 0.0001912751113384684,
      "loss": 1.4723,
      "step": 1117
    },
    {
      "epoch": 0.1601489757914339,
      "grad_norm": 1.1440412998199463,
      "learning_rate": 0.00019125614728053344,
      "loss": 1.4987,
      "step": 1118
    },
    {
      "epoch": 0.1602922217447357,
      "grad_norm": 1.4038132429122925,
      "learning_rate": 0.0001912371635774082,
      "loss": 1.4667,
      "step": 1119
    },
    {
      "epoch": 0.16043546769803754,
      "grad_norm": 1.1797844171524048,
      "learning_rate": 0.00019121816023317948,
      "loss": 1.5144,
      "step": 1120
    },
    {
      "epoch": 0.16057871365133936,
      "grad_norm": 1.3288965225219727,
      "learning_rate": 0.0001911991372519381,
      "loss": 1.3749,
      "step": 1121
    },
    {
      "epoch": 0.16072195960464117,
      "grad_norm": 1.6327883005142212,
      "learning_rate": 0.0001911800946377793,
      "loss": 1.5428,
      "step": 1122
    },
    {
      "epoch": 0.16086520555794298,
      "grad_norm": 1.2628464698791504,
      "learning_rate": 0.0001911610323948025,
      "loss": 1.4392,
      "step": 1123
    },
    {
      "epoch": 0.16100845151124482,
      "grad_norm": 1.2259173393249512,
      "learning_rate": 0.0001911419505271113,
      "loss": 1.5197,
      "step": 1124
    },
    {
      "epoch": 0.16115169746454663,
      "grad_norm": 1.0414924621582031,
      "learning_rate": 0.0001911228490388136,
      "loss": 1.6483,
      "step": 1125
    },
    {
      "epoch": 0.16129494341784845,
      "grad_norm": 1.392253041267395,
      "learning_rate": 0.0001911037279340214,
      "loss": 1.5102,
      "step": 1126
    },
    {
      "epoch": 0.16143818937115026,
      "grad_norm": 0.9352260828018188,
      "learning_rate": 0.00019108458721685105,
      "loss": 1.7936,
      "step": 1127
    },
    {
      "epoch": 0.16158143532445207,
      "grad_norm": 1.005239486694336,
      "learning_rate": 0.00019106542689142306,
      "loss": 1.5246,
      "step": 1128
    },
    {
      "epoch": 0.1617246812777539,
      "grad_norm": 1.0168390274047852,
      "learning_rate": 0.0001910462469618622,
      "loss": 1.6308,
      "step": 1129
    },
    {
      "epoch": 0.16186792723105572,
      "grad_norm": 1.3340356349945068,
      "learning_rate": 0.0001910270474322974,
      "loss": 1.4908,
      "step": 1130
    },
    {
      "epoch": 0.16201117318435754,
      "grad_norm": 1.097332239151001,
      "learning_rate": 0.00019100782830686188,
      "loss": 1.503,
      "step": 1131
    },
    {
      "epoch": 0.16215441913765935,
      "grad_norm": 1.283360242843628,
      "learning_rate": 0.000190988589589693,
      "loss": 1.5463,
      "step": 1132
    },
    {
      "epoch": 0.1622976650909612,
      "grad_norm": 1.2783915996551514,
      "learning_rate": 0.00019096933128493238,
      "loss": 1.6493,
      "step": 1133
    },
    {
      "epoch": 0.162440911044263,
      "grad_norm": 1.088844656944275,
      "learning_rate": 0.0001909500533967259,
      "loss": 1.6921,
      "step": 1134
    },
    {
      "epoch": 0.16258415699756482,
      "grad_norm": 1.1864125728607178,
      "learning_rate": 0.00019093075592922358,
      "loss": 1.4782,
      "step": 1135
    },
    {
      "epoch": 0.16272740295086663,
      "grad_norm": 1.3389954566955566,
      "learning_rate": 0.0001909114388865797,
      "loss": 1.582,
      "step": 1136
    },
    {
      "epoch": 0.16287064890416847,
      "grad_norm": 1.2726613283157349,
      "learning_rate": 0.00019089210227295276,
      "loss": 1.5965,
      "step": 1137
    },
    {
      "epoch": 0.16301389485747028,
      "grad_norm": 1.3534932136535645,
      "learning_rate": 0.0001908727460925054,
      "loss": 1.5993,
      "step": 1138
    },
    {
      "epoch": 0.1631571408107721,
      "grad_norm": 1.1192597150802612,
      "learning_rate": 0.00019085337034940457,
      "loss": 1.2934,
      "step": 1139
    },
    {
      "epoch": 0.1633003867640739,
      "grad_norm": 1.1365140676498413,
      "learning_rate": 0.00019083397504782138,
      "loss": 1.6308,
      "step": 1140
    },
    {
      "epoch": 0.16344363271737575,
      "grad_norm": 1.239017367362976,
      "learning_rate": 0.00019081456019193112,
      "loss": 1.6364,
      "step": 1141
    },
    {
      "epoch": 0.16358687867067756,
      "grad_norm": 1.0785810947418213,
      "learning_rate": 0.00019079512578591337,
      "loss": 1.4922,
      "step": 1142
    },
    {
      "epoch": 0.16373012462397937,
      "grad_norm": 1.31537926197052,
      "learning_rate": 0.0001907756718339519,
      "loss": 1.5822,
      "step": 1143
    },
    {
      "epoch": 0.16387337057728119,
      "grad_norm": 1.128383994102478,
      "learning_rate": 0.0001907561983402346,
      "loss": 1.586,
      "step": 1144
    },
    {
      "epoch": 0.164016616530583,
      "grad_norm": 0.966109573841095,
      "learning_rate": 0.0001907367053089537,
      "loss": 1.6256,
      "step": 1145
    },
    {
      "epoch": 0.16415986248388484,
      "grad_norm": 1.2360272407531738,
      "learning_rate": 0.00019071719274430554,
      "loss": 1.5867,
      "step": 1146
    },
    {
      "epoch": 0.16430310843718665,
      "grad_norm": 1.1176432371139526,
      "learning_rate": 0.00019069766065049067,
      "loss": 1.5729,
      "step": 1147
    },
    {
      "epoch": 0.16444635439048846,
      "grad_norm": 1.1923054456710815,
      "learning_rate": 0.00019067810903171384,
      "loss": 1.5145,
      "step": 1148
    },
    {
      "epoch": 0.16458960034379028,
      "grad_norm": 1.384315848350525,
      "learning_rate": 0.0001906585378921841,
      "loss": 1.5497,
      "step": 1149
    },
    {
      "epoch": 0.16473284629709212,
      "grad_norm": 1.5045276880264282,
      "learning_rate": 0.0001906389472361146,
      "loss": 1.5576,
      "step": 1150
    },
    {
      "epoch": 0.16487609225039393,
      "grad_norm": 1.384010910987854,
      "learning_rate": 0.00019061933706772274,
      "loss": 1.5436,
      "step": 1151
    },
    {
      "epoch": 0.16501933820369574,
      "grad_norm": 1.8074532747268677,
      "learning_rate": 0.0001905997073912301,
      "loss": 1.4112,
      "step": 1152
    },
    {
      "epoch": 0.16516258415699756,
      "grad_norm": 1.0780630111694336,
      "learning_rate": 0.00019058005821086244,
      "loss": 1.4486,
      "step": 1153
    },
    {
      "epoch": 0.1653058301102994,
      "grad_norm": 1.2198182344436646,
      "learning_rate": 0.00019056038953084973,
      "loss": 1.6487,
      "step": 1154
    },
    {
      "epoch": 0.1654490760636012,
      "grad_norm": 1.2643805742263794,
      "learning_rate": 0.00019054070135542618,
      "loss": 1.5346,
      "step": 1155
    },
    {
      "epoch": 0.16559232201690302,
      "grad_norm": 1.2533457279205322,
      "learning_rate": 0.00019052099368883018,
      "loss": 1.5605,
      "step": 1156
    },
    {
      "epoch": 0.16573556797020483,
      "grad_norm": 1.2444400787353516,
      "learning_rate": 0.00019050126653530426,
      "loss": 1.4794,
      "step": 1157
    },
    {
      "epoch": 0.16587881392350667,
      "grad_norm": 1.3984493017196655,
      "learning_rate": 0.00019048151989909523,
      "loss": 1.5666,
      "step": 1158
    },
    {
      "epoch": 0.1660220598768085,
      "grad_norm": 1.9702389240264893,
      "learning_rate": 0.00019046175378445405,
      "loss": 1.4601,
      "step": 1159
    },
    {
      "epoch": 0.1661653058301103,
      "grad_norm": 1.6798146963119507,
      "learning_rate": 0.00019044196819563588,
      "loss": 1.6333,
      "step": 1160
    },
    {
      "epoch": 0.1663085517834121,
      "grad_norm": 1.4010471105575562,
      "learning_rate": 0.0001904221631369,
      "loss": 1.4371,
      "step": 1161
    },
    {
      "epoch": 0.16645179773671392,
      "grad_norm": 1.5270562171936035,
      "learning_rate": 0.00019040233861251002,
      "loss": 1.5629,
      "step": 1162
    },
    {
      "epoch": 0.16659504369001576,
      "grad_norm": 1.4402588605880737,
      "learning_rate": 0.00019038249462673365,
      "loss": 1.4874,
      "step": 1163
    },
    {
      "epoch": 0.16673828964331758,
      "grad_norm": 1.0173940658569336,
      "learning_rate": 0.0001903626311838428,
      "loss": 1.4896,
      "step": 1164
    },
    {
      "epoch": 0.1668815355966194,
      "grad_norm": 1.4004608392715454,
      "learning_rate": 0.00019034274828811358,
      "loss": 1.5575,
      "step": 1165
    },
    {
      "epoch": 0.1670247815499212,
      "grad_norm": 1.4390407800674438,
      "learning_rate": 0.0001903228459438263,
      "loss": 1.3593,
      "step": 1166
    },
    {
      "epoch": 0.16716802750322304,
      "grad_norm": 1.4150819778442383,
      "learning_rate": 0.00019030292415526544,
      "loss": 1.5085,
      "step": 1167
    },
    {
      "epoch": 0.16731127345652486,
      "grad_norm": 1.3124033212661743,
      "learning_rate": 0.00019028298292671965,
      "loss": 1.3581,
      "step": 1168
    },
    {
      "epoch": 0.16745451940982667,
      "grad_norm": 1.2406296730041504,
      "learning_rate": 0.0001902630222624818,
      "loss": 1.4113,
      "step": 1169
    },
    {
      "epoch": 0.16759776536312848,
      "grad_norm": 1.18865168094635,
      "learning_rate": 0.00019024304216684888,
      "loss": 1.7488,
      "step": 1170
    },
    {
      "epoch": 0.16774101131643032,
      "grad_norm": 1.2740721702575684,
      "learning_rate": 0.00019022304264412217,
      "loss": 1.514,
      "step": 1171
    },
    {
      "epoch": 0.16788425726973213,
      "grad_norm": 1.198022723197937,
      "learning_rate": 0.00019020302369860708,
      "loss": 1.5708,
      "step": 1172
    },
    {
      "epoch": 0.16802750322303395,
      "grad_norm": 1.2177618741989136,
      "learning_rate": 0.00019018298533461314,
      "loss": 1.5316,
      "step": 1173
    },
    {
      "epoch": 0.16817074917633576,
      "grad_norm": 1.2639113664627075,
      "learning_rate": 0.00019016292755645418,
      "loss": 1.6626,
      "step": 1174
    },
    {
      "epoch": 0.1683139951296376,
      "grad_norm": 1.3574819564819336,
      "learning_rate": 0.00019014285036844804,
      "loss": 1.5747,
      "step": 1175
    },
    {
      "epoch": 0.1684572410829394,
      "grad_norm": 1.2896956205368042,
      "learning_rate": 0.00019012275377491695,
      "loss": 1.5548,
      "step": 1176
    },
    {
      "epoch": 0.16860048703624123,
      "grad_norm": 1.3717323541641235,
      "learning_rate": 0.00019010263778018716,
      "loss": 1.6679,
      "step": 1177
    },
    {
      "epoch": 0.16874373298954304,
      "grad_norm": 1.316711187362671,
      "learning_rate": 0.00019008250238858914,
      "loss": 1.5343,
      "step": 1178
    },
    {
      "epoch": 0.16888697894284485,
      "grad_norm": 1.3912372589111328,
      "learning_rate": 0.00019006234760445754,
      "loss": 1.5242,
      "step": 1179
    },
    {
      "epoch": 0.1690302248961467,
      "grad_norm": 1.1723421812057495,
      "learning_rate": 0.0001900421734321312,
      "loss": 1.586,
      "step": 1180
    },
    {
      "epoch": 0.1691734708494485,
      "grad_norm": 1.4555163383483887,
      "learning_rate": 0.00019002197987595313,
      "loss": 1.6012,
      "step": 1181
    },
    {
      "epoch": 0.16931671680275032,
      "grad_norm": 1.5780225992202759,
      "learning_rate": 0.00019000176694027049,
      "loss": 1.4767,
      "step": 1182
    },
    {
      "epoch": 0.16945996275605213,
      "grad_norm": 1.1260582208633423,
      "learning_rate": 0.00018998153462943462,
      "loss": 1.502,
      "step": 1183
    },
    {
      "epoch": 0.16960320870935397,
      "grad_norm": 1.557626485824585,
      "learning_rate": 0.00018996128294780106,
      "loss": 1.5532,
      "step": 1184
    },
    {
      "epoch": 0.16974645466265578,
      "grad_norm": 1.4284131526947021,
      "learning_rate": 0.00018994101189972944,
      "loss": 1.5989,
      "step": 1185
    },
    {
      "epoch": 0.1698897006159576,
      "grad_norm": 1.2266610860824585,
      "learning_rate": 0.00018992072148958368,
      "loss": 1.6673,
      "step": 1186
    },
    {
      "epoch": 0.1700329465692594,
      "grad_norm": 1.2829517126083374,
      "learning_rate": 0.00018990041172173178,
      "loss": 1.5749,
      "step": 1187
    },
    {
      "epoch": 0.17017619252256125,
      "grad_norm": 1.4330170154571533,
      "learning_rate": 0.00018988008260054591,
      "loss": 1.6927,
      "step": 1188
    },
    {
      "epoch": 0.17031943847586306,
      "grad_norm": 1.4083924293518066,
      "learning_rate": 0.00018985973413040245,
      "loss": 1.5968,
      "step": 1189
    },
    {
      "epoch": 0.17046268442916487,
      "grad_norm": 1.2225673198699951,
      "learning_rate": 0.00018983936631568194,
      "loss": 1.7187,
      "step": 1190
    },
    {
      "epoch": 0.1706059303824667,
      "grad_norm": 1.3002638816833496,
      "learning_rate": 0.000189818979160769,
      "loss": 1.6167,
      "step": 1191
    },
    {
      "epoch": 0.17074917633576853,
      "grad_norm": 1.4452366828918457,
      "learning_rate": 0.00018979857267005255,
      "loss": 1.5847,
      "step": 1192
    },
    {
      "epoch": 0.17089242228907034,
      "grad_norm": 1.315406322479248,
      "learning_rate": 0.00018977814684792557,
      "loss": 1.5015,
      "step": 1193
    },
    {
      "epoch": 0.17103566824237215,
      "grad_norm": 1.3214411735534668,
      "learning_rate": 0.0001897577016987852,
      "loss": 1.5417,
      "step": 1194
    },
    {
      "epoch": 0.17117891419567396,
      "grad_norm": 1.129775881767273,
      "learning_rate": 0.00018973723722703285,
      "loss": 1.3749,
      "step": 1195
    },
    {
      "epoch": 0.1713221601489758,
      "grad_norm": 1.1681206226348877,
      "learning_rate": 0.00018971675343707397,
      "loss": 1.5046,
      "step": 1196
    },
    {
      "epoch": 0.17146540610227762,
      "grad_norm": 1.353795051574707,
      "learning_rate": 0.00018969625033331822,
      "loss": 1.6288,
      "step": 1197
    },
    {
      "epoch": 0.17160865205557943,
      "grad_norm": 1.6154727935791016,
      "learning_rate": 0.0001896757279201794,
      "loss": 1.6408,
      "step": 1198
    },
    {
      "epoch": 0.17175189800888124,
      "grad_norm": 1.6788369417190552,
      "learning_rate": 0.00018965518620207549,
      "loss": 1.5689,
      "step": 1199
    },
    {
      "epoch": 0.17189514396218306,
      "grad_norm": 1.5394922494888306,
      "learning_rate": 0.00018963462518342862,
      "loss": 1.4842,
      "step": 1200
    },
    {
      "epoch": 0.1720383899154849,
      "grad_norm": 1.4983444213867188,
      "learning_rate": 0.00018961404486866508,
      "loss": 1.4745,
      "step": 1201
    },
    {
      "epoch": 0.1721816358687867,
      "grad_norm": 1.1654797792434692,
      "learning_rate": 0.00018959344526221525,
      "loss": 1.4511,
      "step": 1202
    },
    {
      "epoch": 0.17232488182208852,
      "grad_norm": 1.3940114974975586,
      "learning_rate": 0.00018957282636851376,
      "loss": 1.5774,
      "step": 1203
    },
    {
      "epoch": 0.17246812777539033,
      "grad_norm": 1.5903339385986328,
      "learning_rate": 0.00018955218819199937,
      "loss": 1.636,
      "step": 1204
    },
    {
      "epoch": 0.17261137372869217,
      "grad_norm": 1.554872989654541,
      "learning_rate": 0.00018953153073711487,
      "loss": 1.5397,
      "step": 1205
    },
    {
      "epoch": 0.172754619681994,
      "grad_norm": 1.257312536239624,
      "learning_rate": 0.0001895108540083074,
      "loss": 1.6073,
      "step": 1206
    },
    {
      "epoch": 0.1728978656352958,
      "grad_norm": 1.4855549335479736,
      "learning_rate": 0.0001894901580100281,
      "loss": 1.5972,
      "step": 1207
    },
    {
      "epoch": 0.1730411115885976,
      "grad_norm": 1.3067160844802856,
      "learning_rate": 0.00018946944274673234,
      "loss": 1.5547,
      "step": 1208
    },
    {
      "epoch": 0.17318435754189945,
      "grad_norm": 1.513991117477417,
      "learning_rate": 0.00018944870822287956,
      "loss": 1.5859,
      "step": 1209
    },
    {
      "epoch": 0.17332760349520127,
      "grad_norm": 1.20363187789917,
      "learning_rate": 0.00018942795444293342,
      "loss": 1.5133,
      "step": 1210
    },
    {
      "epoch": 0.17347084944850308,
      "grad_norm": 1.4597328901290894,
      "learning_rate": 0.00018940718141136168,
      "loss": 1.5437,
      "step": 1211
    },
    {
      "epoch": 0.1736140954018049,
      "grad_norm": 1.5445020198822021,
      "learning_rate": 0.0001893863891326363,
      "loss": 1.4522,
      "step": 1212
    },
    {
      "epoch": 0.17375734135510673,
      "grad_norm": 1.2678565979003906,
      "learning_rate": 0.00018936557761123327,
      "loss": 1.5093,
      "step": 1213
    },
    {
      "epoch": 0.17390058730840854,
      "grad_norm": 1.2605350017547607,
      "learning_rate": 0.00018934474685163285,
      "loss": 1.64,
      "step": 1214
    },
    {
      "epoch": 0.17404383326171036,
      "grad_norm": 1.2464580535888672,
      "learning_rate": 0.00018932389685831936,
      "loss": 1.5262,
      "step": 1215
    },
    {
      "epoch": 0.17418707921501217,
      "grad_norm": 1.2617474794387817,
      "learning_rate": 0.00018930302763578132,
      "loss": 1.6288,
      "step": 1216
    },
    {
      "epoch": 0.17433032516831398,
      "grad_norm": 1.3272151947021484,
      "learning_rate": 0.0001892821391885113,
      "loss": 1.5683,
      "step": 1217
    },
    {
      "epoch": 0.17447357112161582,
      "grad_norm": 1.3451011180877686,
      "learning_rate": 0.00018926123152100615,
      "loss": 1.4513,
      "step": 1218
    },
    {
      "epoch": 0.17461681707491764,
      "grad_norm": 1.1489102840423584,
      "learning_rate": 0.0001892403046377667,
      "loss": 1.6464,
      "step": 1219
    },
    {
      "epoch": 0.17476006302821945,
      "grad_norm": 1.304760217666626,
      "learning_rate": 0.00018921935854329802,
      "loss": 1.3769,
      "step": 1220
    },
    {
      "epoch": 0.17490330898152126,
      "grad_norm": 1.2934755086898804,
      "learning_rate": 0.00018919839324210927,
      "loss": 1.5776,
      "step": 1221
    },
    {
      "epoch": 0.1750465549348231,
      "grad_norm": 1.4807761907577515,
      "learning_rate": 0.00018917740873871378,
      "loss": 1.5134,
      "step": 1222
    },
    {
      "epoch": 0.1751898008881249,
      "grad_norm": 1.3128358125686646,
      "learning_rate": 0.000189156405037629,
      "loss": 1.4788,
      "step": 1223
    },
    {
      "epoch": 0.17533304684142673,
      "grad_norm": 1.6213737726211548,
      "learning_rate": 0.0001891353821433765,
      "loss": 1.5816,
      "step": 1224
    },
    {
      "epoch": 0.17547629279472854,
      "grad_norm": 1.4727811813354492,
      "learning_rate": 0.00018911434006048196,
      "loss": 1.4297,
      "step": 1225
    },
    {
      "epoch": 0.17561953874803038,
      "grad_norm": 1.4761892557144165,
      "learning_rate": 0.00018909327879347524,
      "loss": 1.4269,
      "step": 1226
    },
    {
      "epoch": 0.1757627847013322,
      "grad_norm": 1.0209108591079712,
      "learning_rate": 0.00018907219834689033,
      "loss": 1.5705,
      "step": 1227
    },
    {
      "epoch": 0.175906030654634,
      "grad_norm": 1.575480341911316,
      "learning_rate": 0.00018905109872526532,
      "loss": 1.5731,
      "step": 1228
    },
    {
      "epoch": 0.17604927660793582,
      "grad_norm": 1.4144480228424072,
      "learning_rate": 0.0001890299799331424,
      "loss": 1.4263,
      "step": 1229
    },
    {
      "epoch": 0.17619252256123766,
      "grad_norm": 1.124971866607666,
      "learning_rate": 0.00018900884197506796,
      "loss": 1.5052,
      "step": 1230
    },
    {
      "epoch": 0.17633576851453947,
      "grad_norm": 1.4364194869995117,
      "learning_rate": 0.00018898768485559248,
      "loss": 1.5004,
      "step": 1231
    },
    {
      "epoch": 0.17647901446784128,
      "grad_norm": 1.3133175373077393,
      "learning_rate": 0.00018896650857927054,
      "loss": 1.5857,
      "step": 1232
    },
    {
      "epoch": 0.1766222604211431,
      "grad_norm": 1.299665093421936,
      "learning_rate": 0.00018894531315066088,
      "loss": 1.5527,
      "step": 1233
    },
    {
      "epoch": 0.1767655063744449,
      "grad_norm": 1.4516721963882446,
      "learning_rate": 0.00018892409857432636,
      "loss": 1.272,
      "step": 1234
    },
    {
      "epoch": 0.17690875232774675,
      "grad_norm": 1.3323822021484375,
      "learning_rate": 0.00018890286485483395,
      "loss": 1.5578,
      "step": 1235
    },
    {
      "epoch": 0.17705199828104856,
      "grad_norm": 1.1122660636901855,
      "learning_rate": 0.00018888161199675474,
      "loss": 1.6342,
      "step": 1236
    },
    {
      "epoch": 0.17719524423435037,
      "grad_norm": 1.2867708206176758,
      "learning_rate": 0.00018886034000466391,
      "loss": 1.456,
      "step": 1237
    },
    {
      "epoch": 0.1773384901876522,
      "grad_norm": 1.4764132499694824,
      "learning_rate": 0.0001888390488831409,
      "loss": 1.3443,
      "step": 1238
    },
    {
      "epoch": 0.17748173614095403,
      "grad_norm": 1.1385782957077026,
      "learning_rate": 0.00018881773863676905,
      "loss": 1.7634,
      "step": 1239
    },
    {
      "epoch": 0.17762498209425584,
      "grad_norm": 1.081588625907898,
      "learning_rate": 0.00018879640927013598,
      "loss": 1.474,
      "step": 1240
    },
    {
      "epoch": 0.17776822804755765,
      "grad_norm": 1.0898150205612183,
      "learning_rate": 0.00018877506078783337,
      "loss": 1.5884,
      "step": 1241
    },
    {
      "epoch": 0.17791147400085947,
      "grad_norm": 1.514620065689087,
      "learning_rate": 0.000188753693194457,
      "loss": 1.388,
      "step": 1242
    },
    {
      "epoch": 0.1780547199541613,
      "grad_norm": 1.4251086711883545,
      "learning_rate": 0.0001887323064946068,
      "loss": 1.6073,
      "step": 1243
    },
    {
      "epoch": 0.17819796590746312,
      "grad_norm": 1.2940031290054321,
      "learning_rate": 0.00018871090069288678,
      "loss": 1.7119,
      "step": 1244
    },
    {
      "epoch": 0.17834121186076493,
      "grad_norm": 1.2243354320526123,
      "learning_rate": 0.0001886894757939051,
      "loss": 1.4685,
      "step": 1245
    },
    {
      "epoch": 0.17848445781406674,
      "grad_norm": 1.044276237487793,
      "learning_rate": 0.00018866803180227402,
      "loss": 1.6356,
      "step": 1246
    },
    {
      "epoch": 0.17862770376736858,
      "grad_norm": 1.1098270416259766,
      "learning_rate": 0.00018864656872260985,
      "loss": 1.6102,
      "step": 1247
    },
    {
      "epoch": 0.1787709497206704,
      "grad_norm": 1.2511557340621948,
      "learning_rate": 0.00018862508655953316,
      "loss": 1.4131,
      "step": 1248
    },
    {
      "epoch": 0.1789141956739722,
      "grad_norm": 1.3002887964248657,
      "learning_rate": 0.0001886035853176684,
      "loss": 1.8038,
      "step": 1249
    },
    {
      "epoch": 0.17905744162727402,
      "grad_norm": 1.2917287349700928,
      "learning_rate": 0.00018858206500164436,
      "loss": 1.4838,
      "step": 1250
    },
    {
      "epoch": 0.17920068758057586,
      "grad_norm": 1.290832757949829,
      "learning_rate": 0.00018856052561609378,
      "loss": 1.6101,
      "step": 1251
    },
    {
      "epoch": 0.17934393353387768,
      "grad_norm": 0.9336103200912476,
      "learning_rate": 0.00018853896716565358,
      "loss": 1.6037,
      "step": 1252
    },
    {
      "epoch": 0.1794871794871795,
      "grad_norm": 1.3125022649765015,
      "learning_rate": 0.00018851738965496476,
      "loss": 1.4673,
      "step": 1253
    },
    {
      "epoch": 0.1796304254404813,
      "grad_norm": 1.1871716976165771,
      "learning_rate": 0.00018849579308867238,
      "loss": 1.5099,
      "step": 1254
    },
    {
      "epoch": 0.1797736713937831,
      "grad_norm": 1.5702705383300781,
      "learning_rate": 0.00018847417747142568,
      "loss": 1.5679,
      "step": 1255
    },
    {
      "epoch": 0.17991691734708495,
      "grad_norm": 1.1900712251663208,
      "learning_rate": 0.00018845254280787797,
      "loss": 1.4986,
      "step": 1256
    },
    {
      "epoch": 0.18006016330038677,
      "grad_norm": 1.5937440395355225,
      "learning_rate": 0.00018843088910268664,
      "loss": 1.6138,
      "step": 1257
    },
    {
      "epoch": 0.18020340925368858,
      "grad_norm": 1.2704687118530273,
      "learning_rate": 0.00018840921636051325,
      "loss": 1.5548,
      "step": 1258
    },
    {
      "epoch": 0.1803466552069904,
      "grad_norm": 1.2647334337234497,
      "learning_rate": 0.00018838752458602334,
      "loss": 1.521,
      "step": 1259
    },
    {
      "epoch": 0.18048990116029223,
      "grad_norm": 1.4259322881698608,
      "learning_rate": 0.00018836581378388665,
      "loss": 1.5202,
      "step": 1260
    },
    {
      "epoch": 0.18063314711359404,
      "grad_norm": 1.2759722471237183,
      "learning_rate": 0.00018834408395877693,
      "loss": 1.5498,
      "step": 1261
    },
    {
      "epoch": 0.18077639306689586,
      "grad_norm": 1.0806058645248413,
      "learning_rate": 0.00018832233511537216,
      "loss": 1.4809,
      "step": 1262
    },
    {
      "epoch": 0.18091963902019767,
      "grad_norm": 1.004581093788147,
      "learning_rate": 0.00018830056725835424,
      "loss": 1.6035,
      "step": 1263
    },
    {
      "epoch": 0.1810628849734995,
      "grad_norm": 1.1678098440170288,
      "learning_rate": 0.00018827878039240933,
      "loss": 1.4315,
      "step": 1264
    },
    {
      "epoch": 0.18120613092680132,
      "grad_norm": 0.9221726655960083,
      "learning_rate": 0.00018825697452222754,
      "loss": 1.5908,
      "step": 1265
    },
    {
      "epoch": 0.18134937688010314,
      "grad_norm": 1.002099871635437,
      "learning_rate": 0.00018823514965250317,
      "loss": 1.5367,
      "step": 1266
    },
    {
      "epoch": 0.18149262283340495,
      "grad_norm": 1.2065305709838867,
      "learning_rate": 0.00018821330578793453,
      "loss": 1.3529,
      "step": 1267
    },
    {
      "epoch": 0.1816358687867068,
      "grad_norm": 1.3609874248504639,
      "learning_rate": 0.0001881914429332241,
      "loss": 1.5409,
      "step": 1268
    },
    {
      "epoch": 0.1817791147400086,
      "grad_norm": 1.1752578020095825,
      "learning_rate": 0.00018816956109307843,
      "loss": 1.6717,
      "step": 1269
    },
    {
      "epoch": 0.18192236069331041,
      "grad_norm": 1.355809211730957,
      "learning_rate": 0.0001881476602722081,
      "loss": 1.4575,
      "step": 1270
    },
    {
      "epoch": 0.18206560664661223,
      "grad_norm": 1.412541151046753,
      "learning_rate": 0.00018812574047532782,
      "loss": 1.4629,
      "step": 1271
    },
    {
      "epoch": 0.18220885259991404,
      "grad_norm": 1.0576441287994385,
      "learning_rate": 0.00018810380170715643,
      "loss": 1.4376,
      "step": 1272
    },
    {
      "epoch": 0.18235209855321588,
      "grad_norm": 1.1732678413391113,
      "learning_rate": 0.00018808184397241674,
      "loss": 1.5155,
      "step": 1273
    },
    {
      "epoch": 0.1824953445065177,
      "grad_norm": 1.0407720804214478,
      "learning_rate": 0.0001880598672758357,
      "loss": 1.6492,
      "step": 1274
    },
    {
      "epoch": 0.1826385904598195,
      "grad_norm": 0.9912101626396179,
      "learning_rate": 0.00018803787162214442,
      "loss": 1.3754,
      "step": 1275
    },
    {
      "epoch": 0.18278183641312132,
      "grad_norm": 1.3536498546600342,
      "learning_rate": 0.00018801585701607793,
      "loss": 1.535,
      "step": 1276
    },
    {
      "epoch": 0.18292508236642316,
      "grad_norm": 1.4388937950134277,
      "learning_rate": 0.00018799382346237553,
      "loss": 1.6789,
      "step": 1277
    },
    {
      "epoch": 0.18306832831972497,
      "grad_norm": 1.400286078453064,
      "learning_rate": 0.0001879717709657804,
      "loss": 1.6491,
      "step": 1278
    },
    {
      "epoch": 0.18321157427302678,
      "grad_norm": 1.3085813522338867,
      "learning_rate": 0.00018794969953104,
      "loss": 1.4905,
      "step": 1279
    },
    {
      "epoch": 0.1833548202263286,
      "grad_norm": 0.994949996471405,
      "learning_rate": 0.00018792760916290567,
      "loss": 1.4742,
      "step": 1280
    },
    {
      "epoch": 0.18349806617963044,
      "grad_norm": 1.4386508464813232,
      "learning_rate": 0.00018790549986613298,
      "loss": 1.6211,
      "step": 1281
    },
    {
      "epoch": 0.18364131213293225,
      "grad_norm": 1.4195054769515991,
      "learning_rate": 0.0001878833716454815,
      "loss": 1.5751,
      "step": 1282
    },
    {
      "epoch": 0.18378455808623406,
      "grad_norm": 1.2426488399505615,
      "learning_rate": 0.00018786122450571485,
      "loss": 1.5271,
      "step": 1283
    },
    {
      "epoch": 0.18392780403953587,
      "grad_norm": 1.2151144742965698,
      "learning_rate": 0.0001878390584516008,
      "loss": 1.5028,
      "step": 1284
    },
    {
      "epoch": 0.18407104999283772,
      "grad_norm": 1.4324997663497925,
      "learning_rate": 0.00018781687348791114,
      "loss": 1.5807,
      "step": 1285
    },
    {
      "epoch": 0.18421429594613953,
      "grad_norm": 1.0086147785186768,
      "learning_rate": 0.00018779466961942176,
      "loss": 1.4979,
      "step": 1286
    },
    {
      "epoch": 0.18435754189944134,
      "grad_norm": 1.231493592262268,
      "learning_rate": 0.00018777244685091259,
      "loss": 1.3738,
      "step": 1287
    },
    {
      "epoch": 0.18450078785274315,
      "grad_norm": 1.3142328262329102,
      "learning_rate": 0.00018775020518716761,
      "loss": 1.3915,
      "step": 1288
    },
    {
      "epoch": 0.18464403380604497,
      "grad_norm": 1.110546588897705,
      "learning_rate": 0.00018772794463297498,
      "loss": 1.4936,
      "step": 1289
    },
    {
      "epoch": 0.1847872797593468,
      "grad_norm": 1.082787036895752,
      "learning_rate": 0.00018770566519312677,
      "loss": 1.5962,
      "step": 1290
    },
    {
      "epoch": 0.18493052571264862,
      "grad_norm": 1.1838253736495972,
      "learning_rate": 0.00018768336687241926,
      "loss": 1.4919,
      "step": 1291
    },
    {
      "epoch": 0.18507377166595043,
      "grad_norm": 1.272300362586975,
      "learning_rate": 0.00018766104967565266,
      "loss": 1.5732,
      "step": 1292
    },
    {
      "epoch": 0.18521701761925224,
      "grad_norm": 1.4428644180297852,
      "learning_rate": 0.00018763871360763136,
      "loss": 1.5202,
      "step": 1293
    },
    {
      "epoch": 0.18536026357255408,
      "grad_norm": 1.3102538585662842,
      "learning_rate": 0.00018761635867316372,
      "loss": 1.5307,
      "step": 1294
    },
    {
      "epoch": 0.1855035095258559,
      "grad_norm": 1.2022842168807983,
      "learning_rate": 0.0001875939848770622,
      "loss": 1.5179,
      "step": 1295
    },
    {
      "epoch": 0.1856467554791577,
      "grad_norm": 1.2640762329101562,
      "learning_rate": 0.0001875715922241434,
      "loss": 1.581,
      "step": 1296
    },
    {
      "epoch": 0.18579000143245952,
      "grad_norm": 1.1838408708572388,
      "learning_rate": 0.00018754918071922782,
      "loss": 1.4242,
      "step": 1297
    },
    {
      "epoch": 0.18593324738576136,
      "grad_norm": 1.2696324586868286,
      "learning_rate": 0.00018752675036714015,
      "loss": 1.5717,
      "step": 1298
    },
    {
      "epoch": 0.18607649333906318,
      "grad_norm": 1.3920341730117798,
      "learning_rate": 0.00018750430117270913,
      "loss": 1.4246,
      "step": 1299
    },
    {
      "epoch": 0.186219739292365,
      "grad_norm": 1.1627073287963867,
      "learning_rate": 0.00018748183314076737,
      "loss": 1.5879,
      "step": 1300
    },
    {
      "epoch": 0.1863629852456668,
      "grad_norm": 1.2461504936218262,
      "learning_rate": 0.00018745934627615186,
      "loss": 1.6001,
      "step": 1301
    },
    {
      "epoch": 0.18650623119896864,
      "grad_norm": 1.5189625024795532,
      "learning_rate": 0.0001874368405837033,
      "loss": 1.458,
      "step": 1302
    },
    {
      "epoch": 0.18664947715227045,
      "grad_norm": 1.3910939693450928,
      "learning_rate": 0.00018741431606826672,
      "loss": 1.437,
      "step": 1303
    },
    {
      "epoch": 0.18679272310557227,
      "grad_norm": 1.3026515245437622,
      "learning_rate": 0.0001873917727346911,
      "loss": 1.6072,
      "step": 1304
    },
    {
      "epoch": 0.18693596905887408,
      "grad_norm": 1.2518068552017212,
      "learning_rate": 0.00018736921058782935,
      "loss": 1.4938,
      "step": 1305
    },
    {
      "epoch": 0.1870792150121759,
      "grad_norm": 1.391098976135254,
      "learning_rate": 0.00018734662963253867,
      "loss": 1.4608,
      "step": 1306
    },
    {
      "epoch": 0.18722246096547773,
      "grad_norm": 1.2206428050994873,
      "learning_rate": 0.0001873240298736801,
      "loss": 1.3664,
      "step": 1307
    },
    {
      "epoch": 0.18736570691877955,
      "grad_norm": 1.2746695280075073,
      "learning_rate": 0.00018730141131611882,
      "loss": 1.5006,
      "step": 1308
    },
    {
      "epoch": 0.18750895287208136,
      "grad_norm": 1.448061227798462,
      "learning_rate": 0.00018727877396472408,
      "loss": 1.4169,
      "step": 1309
    },
    {
      "epoch": 0.18765219882538317,
      "grad_norm": 1.2230345010757446,
      "learning_rate": 0.00018725611782436911,
      "loss": 1.4712,
      "step": 1310
    },
    {
      "epoch": 0.187795444778685,
      "grad_norm": 1.2678667306900024,
      "learning_rate": 0.00018723344289993122,
      "loss": 1.3659,
      "step": 1311
    },
    {
      "epoch": 0.18793869073198682,
      "grad_norm": 1.203021764755249,
      "learning_rate": 0.00018721074919629177,
      "loss": 1.5056,
      "step": 1312
    },
    {
      "epoch": 0.18808193668528864,
      "grad_norm": 1.3539254665374756,
      "learning_rate": 0.00018718803671833616,
      "loss": 1.4424,
      "step": 1313
    },
    {
      "epoch": 0.18822518263859045,
      "grad_norm": 1.0582293272018433,
      "learning_rate": 0.0001871653054709538,
      "loss": 1.5905,
      "step": 1314
    },
    {
      "epoch": 0.1883684285918923,
      "grad_norm": 1.0168813467025757,
      "learning_rate": 0.0001871425554590382,
      "loss": 1.6622,
      "step": 1315
    },
    {
      "epoch": 0.1885116745451941,
      "grad_norm": 1.4426645040512085,
      "learning_rate": 0.00018711978668748685,
      "loss": 1.4436,
      "step": 1316
    },
    {
      "epoch": 0.18865492049849591,
      "grad_norm": 1.4375327825546265,
      "learning_rate": 0.00018709699916120127,
      "loss": 1.4814,
      "step": 1317
    },
    {
      "epoch": 0.18879816645179773,
      "grad_norm": 1.2645779848098755,
      "learning_rate": 0.00018707419288508713,
      "loss": 1.5021,
      "step": 1318
    },
    {
      "epoch": 0.18894141240509957,
      "grad_norm": 1.1306257247924805,
      "learning_rate": 0.00018705136786405398,
      "loss": 1.4533,
      "step": 1319
    },
    {
      "epoch": 0.18908465835840138,
      "grad_norm": 1.6387449502944946,
      "learning_rate": 0.00018702852410301554,
      "loss": 1.3,
      "step": 1320
    },
    {
      "epoch": 0.1892279043117032,
      "grad_norm": 1.2490342855453491,
      "learning_rate": 0.00018700566160688946,
      "loss": 1.6539,
      "step": 1321
    },
    {
      "epoch": 0.189371150265005,
      "grad_norm": 1.1145542860031128,
      "learning_rate": 0.00018698278038059752,
      "loss": 1.6368,
      "step": 1322
    },
    {
      "epoch": 0.18951439621830685,
      "grad_norm": 1.474271297454834,
      "learning_rate": 0.00018695988042906542,
      "loss": 1.3596,
      "step": 1323
    },
    {
      "epoch": 0.18965764217160866,
      "grad_norm": 1.216958999633789,
      "learning_rate": 0.00018693696175722303,
      "loss": 1.4823,
      "step": 1324
    },
    {
      "epoch": 0.18980088812491047,
      "grad_norm": 1.1959834098815918,
      "learning_rate": 0.00018691402437000408,
      "loss": 1.5421,
      "step": 1325
    },
    {
      "epoch": 0.18994413407821228,
      "grad_norm": 1.1801759004592896,
      "learning_rate": 0.0001868910682723465,
      "loss": 1.4484,
      "step": 1326
    },
    {
      "epoch": 0.1900873800315141,
      "grad_norm": 1.584790825843811,
      "learning_rate": 0.00018686809346919213,
      "loss": 1.5239,
      "step": 1327
    },
    {
      "epoch": 0.19023062598481594,
      "grad_norm": 1.0420902967453003,
      "learning_rate": 0.0001868450999654869,
      "loss": 1.5239,
      "step": 1328
    },
    {
      "epoch": 0.19037387193811775,
      "grad_norm": 1.1460515260696411,
      "learning_rate": 0.00018682208776618072,
      "loss": 1.6759,
      "step": 1329
    },
    {
      "epoch": 0.19051711789141956,
      "grad_norm": 1.3263332843780518,
      "learning_rate": 0.00018679905687622758,
      "loss": 1.4208,
      "step": 1330
    },
    {
      "epoch": 0.19066036384472138,
      "grad_norm": 1.160867691040039,
      "learning_rate": 0.0001867760073005854,
      "loss": 1.4879,
      "step": 1331
    },
    {
      "epoch": 0.19080360979802322,
      "grad_norm": 1.2872892618179321,
      "learning_rate": 0.00018675293904421624,
      "loss": 1.6284,
      "step": 1332
    },
    {
      "epoch": 0.19094685575132503,
      "grad_norm": 1.4801182746887207,
      "learning_rate": 0.0001867298521120861,
      "loss": 1.3618,
      "step": 1333
    },
    {
      "epoch": 0.19109010170462684,
      "grad_norm": 1.6616297960281372,
      "learning_rate": 0.00018670674650916506,
      "loss": 1.5601,
      "step": 1334
    },
    {
      "epoch": 0.19123334765792865,
      "grad_norm": 1.1809343099594116,
      "learning_rate": 0.0001866836222404271,
      "loss": 1.6917,
      "step": 1335
    },
    {
      "epoch": 0.1913765936112305,
      "grad_norm": 1.2287925481796265,
      "learning_rate": 0.0001866604793108504,
      "loss": 1.6374,
      "step": 1336
    },
    {
      "epoch": 0.1915198395645323,
      "grad_norm": 1.015394687652588,
      "learning_rate": 0.00018663731772541702,
      "loss": 1.4899,
      "step": 1337
    },
    {
      "epoch": 0.19166308551783412,
      "grad_norm": 1.2321321964263916,
      "learning_rate": 0.00018661413748911304,
      "loss": 1.5549,
      "step": 1338
    },
    {
      "epoch": 0.19180633147113593,
      "grad_norm": 1.0912636518478394,
      "learning_rate": 0.00018659093860692866,
      "loss": 1.301,
      "step": 1339
    },
    {
      "epoch": 0.19194957742443777,
      "grad_norm": 1.2184662818908691,
      "learning_rate": 0.000186567721083858,
      "loss": 1.4529,
      "step": 1340
    },
    {
      "epoch": 0.19209282337773959,
      "grad_norm": 1.1306071281433105,
      "learning_rate": 0.00018654448492489917,
      "loss": 1.5536,
      "step": 1341
    },
    {
      "epoch": 0.1922360693310414,
      "grad_norm": 1.3233520984649658,
      "learning_rate": 0.0001865212301350544,
      "loss": 1.5416,
      "step": 1342
    },
    {
      "epoch": 0.1923793152843432,
      "grad_norm": 1.313374638557434,
      "learning_rate": 0.00018649795671932986,
      "loss": 1.4261,
      "step": 1343
    },
    {
      "epoch": 0.19252256123764502,
      "grad_norm": 1.1718517541885376,
      "learning_rate": 0.0001864746646827357,
      "loss": 1.5037,
      "step": 1344
    },
    {
      "epoch": 0.19266580719094686,
      "grad_norm": 1.3711566925048828,
      "learning_rate": 0.00018645135403028617,
      "loss": 1.4132,
      "step": 1345
    },
    {
      "epoch": 0.19280905314424868,
      "grad_norm": 1.4926767349243164,
      "learning_rate": 0.00018642802476699944,
      "loss": 1.5733,
      "step": 1346
    },
    {
      "epoch": 0.1929522990975505,
      "grad_norm": 1.3704196214675903,
      "learning_rate": 0.00018640467689789775,
      "loss": 1.403,
      "step": 1347
    },
    {
      "epoch": 0.1930955450508523,
      "grad_norm": 1.1321619749069214,
      "learning_rate": 0.00018638131042800733,
      "loss": 1.7169,
      "step": 1348
    },
    {
      "epoch": 0.19323879100415414,
      "grad_norm": 1.237526297569275,
      "learning_rate": 0.00018635792536235836,
      "loss": 1.5314,
      "step": 1349
    },
    {
      "epoch": 0.19338203695745596,
      "grad_norm": 1.1408690214157104,
      "learning_rate": 0.00018633452170598508,
      "loss": 1.6169,
      "step": 1350
    },
    {
      "epoch": 0.19352528291075777,
      "grad_norm": 1.1132980585098267,
      "learning_rate": 0.00018631109946392574,
      "loss": 1.3108,
      "step": 1351
    },
    {
      "epoch": 0.19366852886405958,
      "grad_norm": 1.0329493284225464,
      "learning_rate": 0.00018628765864122255,
      "loss": 1.384,
      "step": 1352
    },
    {
      "epoch": 0.19381177481736142,
      "grad_norm": 1.2056758403778076,
      "learning_rate": 0.00018626419924292173,
      "loss": 1.4361,
      "step": 1353
    },
    {
      "epoch": 0.19395502077066323,
      "grad_norm": 1.3188925981521606,
      "learning_rate": 0.00018624072127407351,
      "loss": 1.5968,
      "step": 1354
    },
    {
      "epoch": 0.19409826672396505,
      "grad_norm": 1.3645230531692505,
      "learning_rate": 0.00018621722473973216,
      "loss": 1.5848,
      "step": 1355
    },
    {
      "epoch": 0.19424151267726686,
      "grad_norm": 1.1375141143798828,
      "learning_rate": 0.00018619370964495586,
      "loss": 1.5364,
      "step": 1356
    },
    {
      "epoch": 0.1943847586305687,
      "grad_norm": 1.3814671039581299,
      "learning_rate": 0.00018617017599480682,
      "loss": 1.4141,
      "step": 1357
    },
    {
      "epoch": 0.1945280045838705,
      "grad_norm": 1.353074550628662,
      "learning_rate": 0.00018614662379435129,
      "loss": 1.4439,
      "step": 1358
    },
    {
      "epoch": 0.19467125053717232,
      "grad_norm": 1.1491092443466187,
      "learning_rate": 0.0001861230530486594,
      "loss": 1.6599,
      "step": 1359
    },
    {
      "epoch": 0.19481449649047414,
      "grad_norm": 1.1675071716308594,
      "learning_rate": 0.00018609946376280548,
      "loss": 1.4965,
      "step": 1360
    },
    {
      "epoch": 0.19495774244377595,
      "grad_norm": 1.0877861976623535,
      "learning_rate": 0.0001860758559418676,
      "loss": 1.4455,
      "step": 1361
    },
    {
      "epoch": 0.1951009883970778,
      "grad_norm": 1.1705178022384644,
      "learning_rate": 0.000186052229590928,
      "loss": 1.4463,
      "step": 1362
    },
    {
      "epoch": 0.1952442343503796,
      "grad_norm": 1.2562415599822998,
      "learning_rate": 0.00018602858471507283,
      "loss": 1.5457,
      "step": 1363
    },
    {
      "epoch": 0.19538748030368142,
      "grad_norm": 1.2922163009643555,
      "learning_rate": 0.00018600492131939225,
      "loss": 1.4898,
      "step": 1364
    },
    {
      "epoch": 0.19553072625698323,
      "grad_norm": 1.102820873260498,
      "learning_rate": 0.00018598123940898037,
      "loss": 1.4937,
      "step": 1365
    },
    {
      "epoch": 0.19567397221028507,
      "grad_norm": 0.9924041628837585,
      "learning_rate": 0.00018595753898893537,
      "loss": 1.6517,
      "step": 1366
    },
    {
      "epoch": 0.19581721816358688,
      "grad_norm": 1.1563290357589722,
      "learning_rate": 0.00018593382006435935,
      "loss": 1.4809,
      "step": 1367
    },
    {
      "epoch": 0.1959604641168887,
      "grad_norm": 1.358456015586853,
      "learning_rate": 0.0001859100826403584,
      "loss": 1.3629,
      "step": 1368
    },
    {
      "epoch": 0.1961037100701905,
      "grad_norm": 1.2798407077789307,
      "learning_rate": 0.00018588632672204264,
      "loss": 1.2177,
      "step": 1369
    },
    {
      "epoch": 0.19624695602349235,
      "grad_norm": 2.063000440597534,
      "learning_rate": 0.00018586255231452605,
      "loss": 1.5109,
      "step": 1370
    },
    {
      "epoch": 0.19639020197679416,
      "grad_norm": 1.2859746217727661,
      "learning_rate": 0.0001858387594229267,
      "loss": 1.3825,
      "step": 1371
    },
    {
      "epoch": 0.19653344793009597,
      "grad_norm": 1.2065383195877075,
      "learning_rate": 0.00018581494805236667,
      "loss": 1.6827,
      "step": 1372
    },
    {
      "epoch": 0.19667669388339779,
      "grad_norm": 1.323138952255249,
      "learning_rate": 0.00018579111820797185,
      "loss": 1.3593,
      "step": 1373
    },
    {
      "epoch": 0.19681993983669963,
      "grad_norm": 1.2470442056655884,
      "learning_rate": 0.00018576726989487233,
      "loss": 1.4548,
      "step": 1374
    },
    {
      "epoch": 0.19696318579000144,
      "grad_norm": 1.1567609310150146,
      "learning_rate": 0.00018574340311820203,
      "loss": 1.5156,
      "step": 1375
    },
    {
      "epoch": 0.19710643174330325,
      "grad_norm": 1.2888273000717163,
      "learning_rate": 0.00018571951788309883,
      "loss": 1.5113,
      "step": 1376
    },
    {
      "epoch": 0.19724967769660506,
      "grad_norm": 1.0796053409576416,
      "learning_rate": 0.00018569561419470466,
      "loss": 1.4605,
      "step": 1377
    },
    {
      "epoch": 0.1973929236499069,
      "grad_norm": 0.9260571002960205,
      "learning_rate": 0.00018567169205816538,
      "loss": 1.7155,
      "step": 1378
    },
    {
      "epoch": 0.19753616960320872,
      "grad_norm": 1.3426995277404785,
      "learning_rate": 0.00018564775147863086,
      "loss": 1.435,
      "step": 1379
    },
    {
      "epoch": 0.19767941555651053,
      "grad_norm": 1.362938642501831,
      "learning_rate": 0.0001856237924612549,
      "loss": 1.3506,
      "step": 1380
    },
    {
      "epoch": 0.19782266150981234,
      "grad_norm": 1.527593731880188,
      "learning_rate": 0.00018559981501119525,
      "loss": 1.4524,
      "step": 1381
    },
    {
      "epoch": 0.19796590746311415,
      "grad_norm": 1.4399865865707397,
      "learning_rate": 0.00018557581913361372,
      "loss": 1.7006,
      "step": 1382
    },
    {
      "epoch": 0.198109153416416,
      "grad_norm": 1.4785088300704956,
      "learning_rate": 0.000185551804833676,
      "loss": 1.4817,
      "step": 1383
    },
    {
      "epoch": 0.1982523993697178,
      "grad_norm": 1.270953893661499,
      "learning_rate": 0.00018552777211655182,
      "loss": 1.4826,
      "step": 1384
    },
    {
      "epoch": 0.19839564532301962,
      "grad_norm": 1.1148852109909058,
      "learning_rate": 0.00018550372098741474,
      "loss": 1.6778,
      "step": 1385
    },
    {
      "epoch": 0.19853889127632143,
      "grad_norm": 1.291720986366272,
      "learning_rate": 0.00018547965145144244,
      "loss": 1.5262,
      "step": 1386
    },
    {
      "epoch": 0.19868213722962327,
      "grad_norm": 0.9664726257324219,
      "learning_rate": 0.00018545556351381643,
      "loss": 1.5423,
      "step": 1387
    },
    {
      "epoch": 0.1988253831829251,
      "grad_norm": 1.152539610862732,
      "learning_rate": 0.00018543145717972234,
      "loss": 1.5401,
      "step": 1388
    },
    {
      "epoch": 0.1989686291362269,
      "grad_norm": 1.0582084655761719,
      "learning_rate": 0.00018540733245434962,
      "loss": 1.2952,
      "step": 1389
    },
    {
      "epoch": 0.1991118750895287,
      "grad_norm": 1.1961158514022827,
      "learning_rate": 0.00018538318934289172,
      "loss": 1.4996,
      "step": 1390
    },
    {
      "epoch": 0.19925512104283055,
      "grad_norm": 1.396264910697937,
      "learning_rate": 0.00018535902785054605,
      "loss": 1.5477,
      "step": 1391
    },
    {
      "epoch": 0.19939836699613236,
      "grad_norm": 1.3177614212036133,
      "learning_rate": 0.00018533484798251398,
      "loss": 1.6327,
      "step": 1392
    },
    {
      "epoch": 0.19954161294943418,
      "grad_norm": 1.250672698020935,
      "learning_rate": 0.00018531064974400087,
      "loss": 1.4355,
      "step": 1393
    },
    {
      "epoch": 0.199684858902736,
      "grad_norm": 1.361416220664978,
      "learning_rate": 0.000185286433140216,
      "loss": 1.4801,
      "step": 1394
    },
    {
      "epoch": 0.19982810485603783,
      "grad_norm": 1.3723920583724976,
      "learning_rate": 0.00018526219817637256,
      "loss": 1.5312,
      "step": 1395
    },
    {
      "epoch": 0.19997135080933964,
      "grad_norm": 1.6589053869247437,
      "learning_rate": 0.00018523794485768774,
      "loss": 1.5444,
      "step": 1396
    },
    {
      "epoch": 0.20011459676264146,
      "grad_norm": 1.036561369895935,
      "learning_rate": 0.00018521367318938275,
      "loss": 1.6317,
      "step": 1397
    },
    {
      "epoch": 0.20025784271594327,
      "grad_norm": 1.1682296991348267,
      "learning_rate": 0.00018518938317668262,
      "loss": 1.6605,
      "step": 1398
    },
    {
      "epoch": 0.20040108866924508,
      "grad_norm": 1.135780930519104,
      "learning_rate": 0.00018516507482481637,
      "loss": 1.558,
      "step": 1399
    },
    {
      "epoch": 0.20054433462254692,
      "grad_norm": 1.2439090013504028,
      "learning_rate": 0.00018514074813901705,
      "loss": 1.6033,
      "step": 1400
    },
    {
      "epoch": 0.20068758057584873,
      "grad_norm": 1.1552304029464722,
      "learning_rate": 0.00018511640312452156,
      "loss": 1.5208,
      "step": 1401
    },
    {
      "epoch": 0.20083082652915055,
      "grad_norm": 1.5979442596435547,
      "learning_rate": 0.0001850920397865708,
      "loss": 1.3715,
      "step": 1402
    },
    {
      "epoch": 0.20097407248245236,
      "grad_norm": 1.4474765062332153,
      "learning_rate": 0.00018506765813040954,
      "loss": 1.4695,
      "step": 1403
    },
    {
      "epoch": 0.2011173184357542,
      "grad_norm": 1.1645184755325317,
      "learning_rate": 0.00018504325816128662,
      "loss": 1.3906,
      "step": 1404
    },
    {
      "epoch": 0.201260564389056,
      "grad_norm": 1.2833558320999146,
      "learning_rate": 0.00018501883988445466,
      "loss": 1.5119,
      "step": 1405
    },
    {
      "epoch": 0.20140381034235783,
      "grad_norm": 1.0259755849838257,
      "learning_rate": 0.00018499440330517039,
      "loss": 1.5412,
      "step": 1406
    },
    {
      "epoch": 0.20154705629565964,
      "grad_norm": 1.1516698598861694,
      "learning_rate": 0.00018496994842869438,
      "loss": 1.4942,
      "step": 1407
    },
    {
      "epoch": 0.20169030224896148,
      "grad_norm": 1.1080873012542725,
      "learning_rate": 0.00018494547526029114,
      "loss": 1.5325,
      "step": 1408
    },
    {
      "epoch": 0.2018335482022633,
      "grad_norm": 1.0680632591247559,
      "learning_rate": 0.00018492098380522916,
      "loss": 1.6113,
      "step": 1409
    },
    {
      "epoch": 0.2019767941555651,
      "grad_norm": 1.056185245513916,
      "learning_rate": 0.0001848964740687808,
      "loss": 1.585,
      "step": 1410
    },
    {
      "epoch": 0.20212004010886692,
      "grad_norm": 1.4105743169784546,
      "learning_rate": 0.00018487194605622248,
      "loss": 1.5146,
      "step": 1411
    },
    {
      "epoch": 0.20226328606216876,
      "grad_norm": 1.1395865678787231,
      "learning_rate": 0.00018484739977283444,
      "loss": 1.6744,
      "step": 1412
    },
    {
      "epoch": 0.20240653201547057,
      "grad_norm": 0.964874804019928,
      "learning_rate": 0.00018482283522390085,
      "loss": 1.6188,
      "step": 1413
    },
    {
      "epoch": 0.20254977796877238,
      "grad_norm": 1.5245773792266846,
      "learning_rate": 0.00018479825241470986,
      "loss": 1.5057,
      "step": 1414
    },
    {
      "epoch": 0.2026930239220742,
      "grad_norm": 1.4016399383544922,
      "learning_rate": 0.00018477365135055357,
      "loss": 1.469,
      "step": 1415
    },
    {
      "epoch": 0.202836269875376,
      "grad_norm": 1.0630128383636475,
      "learning_rate": 0.00018474903203672796,
      "loss": 1.554,
      "step": 1416
    },
    {
      "epoch": 0.20297951582867785,
      "grad_norm": 1.1299080848693848,
      "learning_rate": 0.00018472439447853297,
      "loss": 1.6916,
      "step": 1417
    },
    {
      "epoch": 0.20312276178197966,
      "grad_norm": 1.0836881399154663,
      "learning_rate": 0.00018469973868127246,
      "loss": 1.5345,
      "step": 1418
    },
    {
      "epoch": 0.20326600773528147,
      "grad_norm": 1.3375639915466309,
      "learning_rate": 0.0001846750646502542,
      "loss": 1.4818,
      "step": 1419
    },
    {
      "epoch": 0.20340925368858329,
      "grad_norm": 1.122757911682129,
      "learning_rate": 0.0001846503723907899,
      "loss": 1.4005,
      "step": 1420
    },
    {
      "epoch": 0.20355249964188513,
      "grad_norm": 1.1756877899169922,
      "learning_rate": 0.00018462566190819522,
      "loss": 1.703,
      "step": 1421
    },
    {
      "epoch": 0.20369574559518694,
      "grad_norm": 1.2671613693237305,
      "learning_rate": 0.00018460093320778968,
      "loss": 1.665,
      "step": 1422
    },
    {
      "epoch": 0.20383899154848875,
      "grad_norm": 1.4893039464950562,
      "learning_rate": 0.00018457618629489673,
      "loss": 1.5189,
      "step": 1423
    },
    {
      "epoch": 0.20398223750179056,
      "grad_norm": 1.146140217781067,
      "learning_rate": 0.00018455142117484386,
      "loss": 1.5304,
      "step": 1424
    },
    {
      "epoch": 0.2041254834550924,
      "grad_norm": 1.17046058177948,
      "learning_rate": 0.0001845266378529623,
      "loss": 1.7892,
      "step": 1425
    },
    {
      "epoch": 0.20426872940839422,
      "grad_norm": 1.1615841388702393,
      "learning_rate": 0.00018450183633458733,
      "loss": 1.5839,
      "step": 1426
    },
    {
      "epoch": 0.20441197536169603,
      "grad_norm": 1.7441776990890503,
      "learning_rate": 0.0001844770166250581,
      "loss": 1.8265,
      "step": 1427
    },
    {
      "epoch": 0.20455522131499784,
      "grad_norm": 1.0648633241653442,
      "learning_rate": 0.00018445217872971767,
      "loss": 1.6114,
      "step": 1428
    },
    {
      "epoch": 0.20469846726829968,
      "grad_norm": 1.2427387237548828,
      "learning_rate": 0.000184427322653913,
      "loss": 1.5721,
      "step": 1429
    },
    {
      "epoch": 0.2048417132216015,
      "grad_norm": 1.0513112545013428,
      "learning_rate": 0.00018440244840299506,
      "loss": 1.5165,
      "step": 1430
    },
    {
      "epoch": 0.2049849591749033,
      "grad_norm": 1.2392395734786987,
      "learning_rate": 0.00018437755598231856,
      "loss": 1.4319,
      "step": 1431
    },
    {
      "epoch": 0.20512820512820512,
      "grad_norm": 1.514569640159607,
      "learning_rate": 0.00018435264539724234,
      "loss": 1.4045,
      "step": 1432
    },
    {
      "epoch": 0.20527145108150693,
      "grad_norm": 1.050420880317688,
      "learning_rate": 0.00018432771665312893,
      "loss": 1.519,
      "step": 1433
    },
    {
      "epoch": 0.20541469703480877,
      "grad_norm": 1.2610220909118652,
      "learning_rate": 0.0001843027697553449,
      "loss": 1.4593,
      "step": 1434
    },
    {
      "epoch": 0.2055579429881106,
      "grad_norm": 1.1642498970031738,
      "learning_rate": 0.00018427780470926073,
      "loss": 1.5324,
      "step": 1435
    },
    {
      "epoch": 0.2057011889414124,
      "grad_norm": 1.2829734086990356,
      "learning_rate": 0.00018425282152025076,
      "loss": 1.5211,
      "step": 1436
    },
    {
      "epoch": 0.2058444348947142,
      "grad_norm": 1.4340356588363647,
      "learning_rate": 0.00018422782019369323,
      "loss": 1.3409,
      "step": 1437
    },
    {
      "epoch": 0.20598768084801605,
      "grad_norm": 1.5705552101135254,
      "learning_rate": 0.0001842028007349704,
      "loss": 1.4458,
      "step": 1438
    },
    {
      "epoch": 0.20613092680131787,
      "grad_norm": 0.8794950842857361,
      "learning_rate": 0.0001841777631494682,
      "loss": 1.7866,
      "step": 1439
    },
    {
      "epoch": 0.20627417275461968,
      "grad_norm": 1.0647470951080322,
      "learning_rate": 0.00018415270744257667,
      "loss": 1.4891,
      "step": 1440
    },
    {
      "epoch": 0.2064174187079215,
      "grad_norm": 1.3066332340240479,
      "learning_rate": 0.0001841276336196897,
      "loss": 1.5876,
      "step": 1441
    },
    {
      "epoch": 0.20656066466122333,
      "grad_norm": 1.3406274318695068,
      "learning_rate": 0.00018410254168620504,
      "loss": 1.4268,
      "step": 1442
    },
    {
      "epoch": 0.20670391061452514,
      "grad_norm": 1.0838818550109863,
      "learning_rate": 0.00018407743164752438,
      "loss": 1.4043,
      "step": 1443
    },
    {
      "epoch": 0.20684715656782696,
      "grad_norm": 1.149341344833374,
      "learning_rate": 0.0001840523035090533,
      "loss": 1.446,
      "step": 1444
    },
    {
      "epoch": 0.20699040252112877,
      "grad_norm": 1.3102495670318604,
      "learning_rate": 0.0001840271572762012,
      "loss": 1.6143,
      "step": 1445
    },
    {
      "epoch": 0.2071336484744306,
      "grad_norm": 1.3083823919296265,
      "learning_rate": 0.00018400199295438152,
      "loss": 1.3857,
      "step": 1446
    },
    {
      "epoch": 0.20727689442773242,
      "grad_norm": 1.2183609008789062,
      "learning_rate": 0.00018397681054901146,
      "loss": 1.3833,
      "step": 1447
    },
    {
      "epoch": 0.20742014038103423,
      "grad_norm": 1.3878487348556519,
      "learning_rate": 0.0001839516100655122,
      "loss": 1.5222,
      "step": 1448
    },
    {
      "epoch": 0.20756338633433605,
      "grad_norm": 1.1470175981521606,
      "learning_rate": 0.0001839263915093088,
      "loss": 1.5015,
      "step": 1449
    },
    {
      "epoch": 0.2077066322876379,
      "grad_norm": 1.291468620300293,
      "learning_rate": 0.00018390115488583014,
      "loss": 1.5208,
      "step": 1450
    },
    {
      "epoch": 0.2078498782409397,
      "grad_norm": 1.122896671295166,
      "learning_rate": 0.00018387590020050904,
      "loss": 1.4962,
      "step": 1451
    },
    {
      "epoch": 0.2079931241942415,
      "grad_norm": 1.886612057685852,
      "learning_rate": 0.00018385062745878225,
      "loss": 1.4617,
      "step": 1452
    },
    {
      "epoch": 0.20813637014754333,
      "grad_norm": 1.3602330684661865,
      "learning_rate": 0.0001838253366660904,
      "loss": 1.5006,
      "step": 1453
    },
    {
      "epoch": 0.20827961610084514,
      "grad_norm": 1.2793840169906616,
      "learning_rate": 0.00018380002782787783,
      "loss": 1.604,
      "step": 1454
    },
    {
      "epoch": 0.20842286205414698,
      "grad_norm": 1.1887341737747192,
      "learning_rate": 0.00018377470094959307,
      "loss": 1.5572,
      "step": 1455
    },
    {
      "epoch": 0.2085661080074488,
      "grad_norm": 1.2990611791610718,
      "learning_rate": 0.00018374935603668826,
      "loss": 1.6974,
      "step": 1456
    },
    {
      "epoch": 0.2087093539607506,
      "grad_norm": 1.1259386539459229,
      "learning_rate": 0.00018372399309461962,
      "loss": 1.645,
      "step": 1457
    },
    {
      "epoch": 0.20885259991405242,
      "grad_norm": 1.1804717779159546,
      "learning_rate": 0.00018369861212884706,
      "loss": 1.6709,
      "step": 1458
    },
    {
      "epoch": 0.20899584586735426,
      "grad_norm": 1.2612565755844116,
      "learning_rate": 0.00018367321314483452,
      "loss": 1.5517,
      "step": 1459
    },
    {
      "epoch": 0.20913909182065607,
      "grad_norm": 1.1413666009902954,
      "learning_rate": 0.0001836477961480498,
      "loss": 1.5906,
      "step": 1460
    },
    {
      "epoch": 0.20928233777395788,
      "grad_norm": 1.257929801940918,
      "learning_rate": 0.00018362236114396457,
      "loss": 1.7267,
      "step": 1461
    },
    {
      "epoch": 0.2094255837272597,
      "grad_norm": 1.2549214363098145,
      "learning_rate": 0.00018359690813805427,
      "loss": 1.4221,
      "step": 1462
    },
    {
      "epoch": 0.20956882968056154,
      "grad_norm": 1.2725675106048584,
      "learning_rate": 0.00018357143713579837,
      "loss": 1.5109,
      "step": 1463
    },
    {
      "epoch": 0.20971207563386335,
      "grad_norm": 1.3137381076812744,
      "learning_rate": 0.0001835459481426801,
      "loss": 1.3757,
      "step": 1464
    },
    {
      "epoch": 0.20985532158716516,
      "grad_norm": 1.3137515783309937,
      "learning_rate": 0.00018352044116418668,
      "loss": 1.5882,
      "step": 1465
    },
    {
      "epoch": 0.20999856754046697,
      "grad_norm": 1.1340367794036865,
      "learning_rate": 0.00018349491620580906,
      "loss": 1.5905,
      "step": 1466
    },
    {
      "epoch": 0.21014181349376881,
      "grad_norm": 1.4691225290298462,
      "learning_rate": 0.00018346937327304216,
      "loss": 1.3293,
      "step": 1467
    },
    {
      "epoch": 0.21028505944707063,
      "grad_norm": 1.1611536741256714,
      "learning_rate": 0.00018344381237138472,
      "loss": 1.4509,
      "step": 1468
    },
    {
      "epoch": 0.21042830540037244,
      "grad_norm": 1.252271056175232,
      "learning_rate": 0.00018341823350633942,
      "loss": 1.5841,
      "step": 1469
    },
    {
      "epoch": 0.21057155135367425,
      "grad_norm": 1.3695937395095825,
      "learning_rate": 0.00018339263668341275,
      "loss": 1.6684,
      "step": 1470
    },
    {
      "epoch": 0.21071479730697606,
      "grad_norm": 1.146183729171753,
      "learning_rate": 0.00018336702190811498,
      "loss": 1.6243,
      "step": 1471
    },
    {
      "epoch": 0.2108580432602779,
      "grad_norm": 1.1380293369293213,
      "learning_rate": 0.00018334138918596046,
      "loss": 1.5668,
      "step": 1472
    },
    {
      "epoch": 0.21100128921357972,
      "grad_norm": 1.1479898691177368,
      "learning_rate": 0.00018331573852246722,
      "loss": 1.4759,
      "step": 1473
    },
    {
      "epoch": 0.21114453516688153,
      "grad_norm": 0.9555863738059998,
      "learning_rate": 0.00018329006992315723,
      "loss": 1.4898,
      "step": 1474
    },
    {
      "epoch": 0.21128778112018334,
      "grad_norm": 0.9860863089561462,
      "learning_rate": 0.00018326438339355628,
      "loss": 1.3669,
      "step": 1475
    },
    {
      "epoch": 0.21143102707348518,
      "grad_norm": 1.2669332027435303,
      "learning_rate": 0.00018323867893919405,
      "loss": 1.3633,
      "step": 1476
    },
    {
      "epoch": 0.211574273026787,
      "grad_norm": 1.15382719039917,
      "learning_rate": 0.00018321295656560408,
      "loss": 1.5659,
      "step": 1477
    },
    {
      "epoch": 0.2117175189800888,
      "grad_norm": 1.1834615468978882,
      "learning_rate": 0.00018318721627832377,
      "loss": 1.4881,
      "step": 1478
    },
    {
      "epoch": 0.21186076493339062,
      "grad_norm": 1.2377066612243652,
      "learning_rate": 0.0001831614580828944,
      "loss": 1.5968,
      "step": 1479
    },
    {
      "epoch": 0.21200401088669246,
      "grad_norm": 1.5492805242538452,
      "learning_rate": 0.000183135681984861,
      "loss": 1.3493,
      "step": 1480
    },
    {
      "epoch": 0.21214725683999427,
      "grad_norm": 1.1334848403930664,
      "learning_rate": 0.00018310988798977255,
      "loss": 1.7096,
      "step": 1481
    },
    {
      "epoch": 0.2122905027932961,
      "grad_norm": 1.097018837928772,
      "learning_rate": 0.00018308407610318183,
      "loss": 1.3083,
      "step": 1482
    },
    {
      "epoch": 0.2124337487465979,
      "grad_norm": 1.1167300939559937,
      "learning_rate": 0.00018305824633064557,
      "loss": 1.4817,
      "step": 1483
    },
    {
      "epoch": 0.21257699469989974,
      "grad_norm": 1.1967469453811646,
      "learning_rate": 0.00018303239867772426,
      "loss": 1.4046,
      "step": 1484
    },
    {
      "epoch": 0.21272024065320155,
      "grad_norm": 1.3095366954803467,
      "learning_rate": 0.00018300653314998224,
      "loss": 1.5943,
      "step": 1485
    },
    {
      "epoch": 0.21286348660650337,
      "grad_norm": 1.5506595373153687,
      "learning_rate": 0.00018298064975298773,
      "loss": 1.4374,
      "step": 1486
    },
    {
      "epoch": 0.21300673255980518,
      "grad_norm": 1.0912269353866577,
      "learning_rate": 0.0001829547484923128,
      "loss": 1.6117,
      "step": 1487
    },
    {
      "epoch": 0.213149978513107,
      "grad_norm": 1.237838625907898,
      "learning_rate": 0.00018292882937353326,
      "loss": 1.4815,
      "step": 1488
    },
    {
      "epoch": 0.21329322446640883,
      "grad_norm": 1.545166254043579,
      "learning_rate": 0.000182902892402229,
      "loss": 1.4443,
      "step": 1489
    },
    {
      "epoch": 0.21343647041971064,
      "grad_norm": 1.1019178628921509,
      "learning_rate": 0.0001828769375839835,
      "loss": 1.3695,
      "step": 1490
    },
    {
      "epoch": 0.21357971637301246,
      "grad_norm": 1.0788493156433105,
      "learning_rate": 0.00018285096492438424,
      "loss": 1.6622,
      "step": 1491
    },
    {
      "epoch": 0.21372296232631427,
      "grad_norm": 1.4582874774932861,
      "learning_rate": 0.00018282497442902244,
      "loss": 1.4786,
      "step": 1492
    },
    {
      "epoch": 0.2138662082796161,
      "grad_norm": 1.1142699718475342,
      "learning_rate": 0.0001827989661034933,
      "loss": 1.4218,
      "step": 1493
    },
    {
      "epoch": 0.21400945423291792,
      "grad_norm": 1.1761949062347412,
      "learning_rate": 0.00018277293995339565,
      "loss": 1.6064,
      "step": 1494
    },
    {
      "epoch": 0.21415270018621974,
      "grad_norm": 1.1660184860229492,
      "learning_rate": 0.00018274689598433237,
      "loss": 1.5647,
      "step": 1495
    },
    {
      "epoch": 0.21429594613952155,
      "grad_norm": 1.3660680055618286,
      "learning_rate": 0.0001827208342019101,
      "loss": 1.5302,
      "step": 1496
    },
    {
      "epoch": 0.2144391920928234,
      "grad_norm": 0.960205078125,
      "learning_rate": 0.00018269475461173918,
      "loss": 1.5379,
      "step": 1497
    },
    {
      "epoch": 0.2145824380461252,
      "grad_norm": 1.1545640230178833,
      "learning_rate": 0.000182668657219434,
      "loss": 1.61,
      "step": 1498
    },
    {
      "epoch": 0.21472568399942701,
      "grad_norm": 1.2488517761230469,
      "learning_rate": 0.00018264254203061264,
      "loss": 1.3874,
      "step": 1499
    },
    {
      "epoch": 0.21486892995272883,
      "grad_norm": 1.0373151302337646,
      "learning_rate": 0.00018261640905089708,
      "loss": 1.6404,
      "step": 1500
    },
    {
      "epoch": 0.21501217590603067,
      "grad_norm": 1.083553671836853,
      "learning_rate": 0.00018259025828591308,
      "loss": 1.4594,
      "step": 1501
    },
    {
      "epoch": 0.21515542185933248,
      "grad_norm": 1.507617473602295,
      "learning_rate": 0.00018256408974129027,
      "loss": 1.5502,
      "step": 1502
    },
    {
      "epoch": 0.2152986678126343,
      "grad_norm": 1.1398086547851562,
      "learning_rate": 0.00018253790342266207,
      "loss": 1.6201,
      "step": 1503
    },
    {
      "epoch": 0.2154419137659361,
      "grad_norm": 1.2936040163040161,
      "learning_rate": 0.00018251169933566577,
      "loss": 1.5975,
      "step": 1504
    },
    {
      "epoch": 0.21558515971923792,
      "grad_norm": 1.0466448068618774,
      "learning_rate": 0.00018248547748594244,
      "loss": 1.4843,
      "step": 1505
    },
    {
      "epoch": 0.21572840567253976,
      "grad_norm": 1.195206642150879,
      "learning_rate": 0.00018245923787913704,
      "loss": 1.1496,
      "step": 1506
    },
    {
      "epoch": 0.21587165162584157,
      "grad_norm": 1.3630177974700928,
      "learning_rate": 0.00018243298052089823,
      "loss": 1.4804,
      "step": 1507
    },
    {
      "epoch": 0.21601489757914338,
      "grad_norm": 1.1443976163864136,
      "learning_rate": 0.00018240670541687864,
      "loss": 1.6369,
      "step": 1508
    },
    {
      "epoch": 0.2161581435324452,
      "grad_norm": 1.5311540365219116,
      "learning_rate": 0.00018238041257273463,
      "loss": 1.4739,
      "step": 1509
    },
    {
      "epoch": 0.21630138948574704,
      "grad_norm": 1.1093631982803345,
      "learning_rate": 0.00018235410199412636,
      "loss": 1.4604,
      "step": 1510
    },
    {
      "epoch": 0.21644463543904885,
      "grad_norm": 1.2489898204803467,
      "learning_rate": 0.0001823277736867179,
      "loss": 1.3782,
      "step": 1511
    },
    {
      "epoch": 0.21658788139235066,
      "grad_norm": 1.1612844467163086,
      "learning_rate": 0.00018230142765617705,
      "loss": 1.7413,
      "step": 1512
    },
    {
      "epoch": 0.21673112734565247,
      "grad_norm": 1.0746690034866333,
      "learning_rate": 0.0001822750639081755,
      "loss": 1.569,
      "step": 1513
    },
    {
      "epoch": 0.21687437329895431,
      "grad_norm": 1.423208475112915,
      "learning_rate": 0.0001822486824483886,
      "loss": 1.6716,
      "step": 1514
    },
    {
      "epoch": 0.21701761925225613,
      "grad_norm": 1.1169368028640747,
      "learning_rate": 0.0001822222832824958,
      "loss": 1.3822,
      "step": 1515
    },
    {
      "epoch": 0.21716086520555794,
      "grad_norm": 1.347452163696289,
      "learning_rate": 0.00018219586641618005,
      "loss": 1.4939,
      "step": 1516
    },
    {
      "epoch": 0.21730411115885975,
      "grad_norm": 1.5060954093933105,
      "learning_rate": 0.0001821694318551283,
      "loss": 1.5256,
      "step": 1517
    },
    {
      "epoch": 0.2174473571121616,
      "grad_norm": 1.2865513563156128,
      "learning_rate": 0.00018214297960503125,
      "loss": 1.561,
      "step": 1518
    },
    {
      "epoch": 0.2175906030654634,
      "grad_norm": 1.2649914026260376,
      "learning_rate": 0.00018211650967158344,
      "loss": 1.4917,
      "step": 1519
    },
    {
      "epoch": 0.21773384901876522,
      "grad_norm": 1.3209419250488281,
      "learning_rate": 0.00018209002206048315,
      "loss": 1.6617,
      "step": 1520
    },
    {
      "epoch": 0.21787709497206703,
      "grad_norm": 1.1756374835968018,
      "learning_rate": 0.0001820635167774325,
      "loss": 1.3885,
      "step": 1521
    },
    {
      "epoch": 0.21802034092536887,
      "grad_norm": 1.0940485000610352,
      "learning_rate": 0.00018203699382813746,
      "loss": 1.6826,
      "step": 1522
    },
    {
      "epoch": 0.21816358687867068,
      "grad_norm": 1.1051404476165771,
      "learning_rate": 0.00018201045321830775,
      "loss": 1.3603,
      "step": 1523
    },
    {
      "epoch": 0.2183068328319725,
      "grad_norm": 1.2450709342956543,
      "learning_rate": 0.0001819838949536569,
      "loss": 1.628,
      "step": 1524
    },
    {
      "epoch": 0.2184500787852743,
      "grad_norm": 1.1604392528533936,
      "learning_rate": 0.00018195731903990225,
      "loss": 1.4947,
      "step": 1525
    },
    {
      "epoch": 0.21859332473857612,
      "grad_norm": 1.123118281364441,
      "learning_rate": 0.00018193072548276494,
      "loss": 1.5528,
      "step": 1526
    },
    {
      "epoch": 0.21873657069187796,
      "grad_norm": 1.1758928298950195,
      "learning_rate": 0.00018190411428796991,
      "loss": 1.4828,
      "step": 1527
    },
    {
      "epoch": 0.21887981664517978,
      "grad_norm": 1.14836585521698,
      "learning_rate": 0.0001818774854612459,
      "loss": 1.6918,
      "step": 1528
    },
    {
      "epoch": 0.2190230625984816,
      "grad_norm": 1.1046864986419678,
      "learning_rate": 0.00018185083900832544,
      "loss": 1.5385,
      "step": 1529
    },
    {
      "epoch": 0.2191663085517834,
      "grad_norm": 1.1961406469345093,
      "learning_rate": 0.00018182417493494478,
      "loss": 1.5787,
      "step": 1530
    },
    {
      "epoch": 0.21930955450508524,
      "grad_norm": 1.1146355867385864,
      "learning_rate": 0.00018179749324684412,
      "loss": 1.6995,
      "step": 1531
    },
    {
      "epoch": 0.21945280045838705,
      "grad_norm": 1.060309886932373,
      "learning_rate": 0.00018177079394976736,
      "loss": 1.4269,
      "step": 1532
    },
    {
      "epoch": 0.21959604641168887,
      "grad_norm": 1.0514110326766968,
      "learning_rate": 0.00018174407704946212,
      "loss": 1.5451,
      "step": 1533
    },
    {
      "epoch": 0.21973929236499068,
      "grad_norm": 1.3559415340423584,
      "learning_rate": 0.00018171734255167997,
      "loss": 1.3571,
      "step": 1534
    },
    {
      "epoch": 0.21988253831829252,
      "grad_norm": 1.224799394607544,
      "learning_rate": 0.00018169059046217617,
      "loss": 1.5162,
      "step": 1535
    },
    {
      "epoch": 0.22002578427159433,
      "grad_norm": 1.1984279155731201,
      "learning_rate": 0.00018166382078670977,
      "loss": 1.2535,
      "step": 1536
    },
    {
      "epoch": 0.22016903022489615,
      "grad_norm": 1.5035969018936157,
      "learning_rate": 0.0001816370335310436,
      "loss": 1.4727,
      "step": 1537
    },
    {
      "epoch": 0.22031227617819796,
      "grad_norm": 1.2182730436325073,
      "learning_rate": 0.00018161022870094432,
      "loss": 1.626,
      "step": 1538
    },
    {
      "epoch": 0.2204555221314998,
      "grad_norm": 1.2001169919967651,
      "learning_rate": 0.00018158340630218235,
      "loss": 1.6117,
      "step": 1539
    },
    {
      "epoch": 0.2205987680848016,
      "grad_norm": 1.2035704851150513,
      "learning_rate": 0.0001815565663405319,
      "loss": 1.4128,
      "step": 1540
    },
    {
      "epoch": 0.22074201403810342,
      "grad_norm": 1.2661808729171753,
      "learning_rate": 0.0001815297088217709,
      "loss": 1.5654,
      "step": 1541
    },
    {
      "epoch": 0.22088525999140524,
      "grad_norm": 1.1710599660873413,
      "learning_rate": 0.00018150283375168114,
      "loss": 1.462,
      "step": 1542
    },
    {
      "epoch": 0.22102850594470705,
      "grad_norm": 1.1481469869613647,
      "learning_rate": 0.00018147594113604817,
      "loss": 1.3322,
      "step": 1543
    },
    {
      "epoch": 0.2211717518980089,
      "grad_norm": 1.0221550464630127,
      "learning_rate": 0.00018144903098066126,
      "loss": 1.5989,
      "step": 1544
    },
    {
      "epoch": 0.2213149978513107,
      "grad_norm": 1.0604640245437622,
      "learning_rate": 0.00018142210329131358,
      "loss": 1.3891,
      "step": 1545
    },
    {
      "epoch": 0.22145824380461251,
      "grad_norm": 1.3637282848358154,
      "learning_rate": 0.0001813951580738019,
      "loss": 1.4772,
      "step": 1546
    },
    {
      "epoch": 0.22160148975791433,
      "grad_norm": 1.3984220027923584,
      "learning_rate": 0.00018136819533392693,
      "loss": 1.3779,
      "step": 1547
    },
    {
      "epoch": 0.22174473571121617,
      "grad_norm": 1.1531269550323486,
      "learning_rate": 0.00018134121507749304,
      "loss": 1.4907,
      "step": 1548
    },
    {
      "epoch": 0.22188798166451798,
      "grad_norm": 1.262723684310913,
      "learning_rate": 0.0001813142173103084,
      "loss": 1.5794,
      "step": 1549
    },
    {
      "epoch": 0.2220312276178198,
      "grad_norm": 1.372413992881775,
      "learning_rate": 0.00018128720203818504,
      "loss": 1.591,
      "step": 1550
    },
    {
      "epoch": 0.2221744735711216,
      "grad_norm": 1.2786822319030762,
      "learning_rate": 0.0001812601692669386,
      "loss": 1.6051,
      "step": 1551
    },
    {
      "epoch": 0.22231771952442345,
      "grad_norm": 1.2073897123336792,
      "learning_rate": 0.0001812331190023886,
      "loss": 1.6959,
      "step": 1552
    },
    {
      "epoch": 0.22246096547772526,
      "grad_norm": 1.0315661430358887,
      "learning_rate": 0.00018120605125035829,
      "loss": 1.55,
      "step": 1553
    },
    {
      "epoch": 0.22260421143102707,
      "grad_norm": 1.2762595415115356,
      "learning_rate": 0.00018117896601667469,
      "loss": 1.5479,
      "step": 1554
    },
    {
      "epoch": 0.22274745738432888,
      "grad_norm": 1.0317778587341309,
      "learning_rate": 0.00018115186330716854,
      "loss": 1.5641,
      "step": 1555
    },
    {
      "epoch": 0.22289070333763072,
      "grad_norm": 1.1282223463058472,
      "learning_rate": 0.00018112474312767445,
      "loss": 1.4049,
      "step": 1556
    },
    {
      "epoch": 0.22303394929093254,
      "grad_norm": 1.262637972831726,
      "learning_rate": 0.00018109760548403066,
      "loss": 1.6847,
      "step": 1557
    },
    {
      "epoch": 0.22317719524423435,
      "grad_norm": 1.115439534187317,
      "learning_rate": 0.00018107045038207931,
      "loss": 1.5362,
      "step": 1558
    },
    {
      "epoch": 0.22332044119753616,
      "grad_norm": 1.4673593044281006,
      "learning_rate": 0.00018104327782766615,
      "loss": 1.3601,
      "step": 1559
    },
    {
      "epoch": 0.22346368715083798,
      "grad_norm": 1.3072248697280884,
      "learning_rate": 0.00018101608782664078,
      "loss": 1.5443,
      "step": 1560
    },
    {
      "epoch": 0.22360693310413982,
      "grad_norm": 1.3873815536499023,
      "learning_rate": 0.00018098888038485652,
      "loss": 1.3851,
      "step": 1561
    },
    {
      "epoch": 0.22375017905744163,
      "grad_norm": 1.455453872680664,
      "learning_rate": 0.0001809616555081705,
      "loss": 1.555,
      "step": 1562
    },
    {
      "epoch": 0.22389342501074344,
      "grad_norm": 1.2670179605484009,
      "learning_rate": 0.00018093441320244353,
      "loss": 1.4691,
      "step": 1563
    },
    {
      "epoch": 0.22403667096404525,
      "grad_norm": 1.1840388774871826,
      "learning_rate": 0.00018090715347354023,
      "loss": 1.4754,
      "step": 1564
    },
    {
      "epoch": 0.2241799169173471,
      "grad_norm": 1.1661396026611328,
      "learning_rate": 0.0001808798763273289,
      "loss": 1.4071,
      "step": 1565
    },
    {
      "epoch": 0.2243231628706489,
      "grad_norm": 1.1518869400024414,
      "learning_rate": 0.0001808525817696817,
      "loss": 1.6451,
      "step": 1566
    },
    {
      "epoch": 0.22446640882395072,
      "grad_norm": 1.1735764741897583,
      "learning_rate": 0.0001808252698064744,
      "loss": 1.6268,
      "step": 1567
    },
    {
      "epoch": 0.22460965477725253,
      "grad_norm": 1.5874416828155518,
      "learning_rate": 0.00018079794044358668,
      "loss": 1.4834,
      "step": 1568
    },
    {
      "epoch": 0.22475290073055437,
      "grad_norm": 1.0743379592895508,
      "learning_rate": 0.00018077059368690174,
      "loss": 1.6421,
      "step": 1569
    },
    {
      "epoch": 0.22489614668385619,
      "grad_norm": 1.1248879432678223,
      "learning_rate": 0.00018074322954230677,
      "loss": 1.4636,
      "step": 1570
    },
    {
      "epoch": 0.225039392637158,
      "grad_norm": 1.206067442893982,
      "learning_rate": 0.0001807158480156926,
      "loss": 1.5657,
      "step": 1571
    },
    {
      "epoch": 0.2251826385904598,
      "grad_norm": 1.5736825466156006,
      "learning_rate": 0.00018068844911295372,
      "loss": 1.5803,
      "step": 1572
    },
    {
      "epoch": 0.22532588454376165,
      "grad_norm": 1.0210436582565308,
      "learning_rate": 0.00018066103283998852,
      "loss": 1.4687,
      "step": 1573
    },
    {
      "epoch": 0.22546913049706346,
      "grad_norm": 1.370393991470337,
      "learning_rate": 0.00018063359920269896,
      "loss": 1.4709,
      "step": 1574
    },
    {
      "epoch": 0.22561237645036528,
      "grad_norm": 1.1733421087265015,
      "learning_rate": 0.00018060614820699085,
      "loss": 1.5174,
      "step": 1575
    },
    {
      "epoch": 0.2257556224036671,
      "grad_norm": 1.3706817626953125,
      "learning_rate": 0.00018057867985877377,
      "loss": 1.5157,
      "step": 1576
    },
    {
      "epoch": 0.22589886835696893,
      "grad_norm": 1.1381696462631226,
      "learning_rate": 0.0001805511941639609,
      "loss": 1.6675,
      "step": 1577
    },
    {
      "epoch": 0.22604211431027074,
      "grad_norm": 1.1196262836456299,
      "learning_rate": 0.0001805236911284693,
      "loss": 1.4569,
      "step": 1578
    },
    {
      "epoch": 0.22618536026357255,
      "grad_norm": 1.0801070928573608,
      "learning_rate": 0.00018049617075821962,
      "loss": 1.3081,
      "step": 1579
    },
    {
      "epoch": 0.22632860621687437,
      "grad_norm": 1.0941091775894165,
      "learning_rate": 0.00018046863305913632,
      "loss": 1.4623,
      "step": 1580
    },
    {
      "epoch": 0.22647185217017618,
      "grad_norm": 1.41851806640625,
      "learning_rate": 0.00018044107803714764,
      "loss": 1.6527,
      "step": 1581
    },
    {
      "epoch": 0.22661509812347802,
      "grad_norm": 1.2155038118362427,
      "learning_rate": 0.00018041350569818546,
      "loss": 1.5218,
      "step": 1582
    },
    {
      "epoch": 0.22675834407677983,
      "grad_norm": 1.4116519689559937,
      "learning_rate": 0.00018038591604818544,
      "loss": 1.3852,
      "step": 1583
    },
    {
      "epoch": 0.22690159003008165,
      "grad_norm": 1.3979288339614868,
      "learning_rate": 0.00018035830909308694,
      "loss": 1.5874,
      "step": 1584
    },
    {
      "epoch": 0.22704483598338346,
      "grad_norm": 1.2356114387512207,
      "learning_rate": 0.00018033068483883305,
      "loss": 1.5841,
      "step": 1585
    },
    {
      "epoch": 0.2271880819366853,
      "grad_norm": 1.0980898141860962,
      "learning_rate": 0.00018030304329137055,
      "loss": 1.4187,
      "step": 1586
    },
    {
      "epoch": 0.2273313278899871,
      "grad_norm": 1.2686175107955933,
      "learning_rate": 0.00018027538445665006,
      "loss": 1.3722,
      "step": 1587
    },
    {
      "epoch": 0.22747457384328892,
      "grad_norm": 1.0811446905136108,
      "learning_rate": 0.00018024770834062582,
      "loss": 1.5286,
      "step": 1588
    },
    {
      "epoch": 0.22761781979659074,
      "grad_norm": 1.1239668130874634,
      "learning_rate": 0.00018022001494925576,
      "loss": 1.5191,
      "step": 1589
    },
    {
      "epoch": 0.22776106574989258,
      "grad_norm": 1.0656206607818604,
      "learning_rate": 0.00018019230428850165,
      "loss": 1.7238,
      "step": 1590
    },
    {
      "epoch": 0.2279043117031944,
      "grad_norm": 1.3077001571655273,
      "learning_rate": 0.00018016457636432884,
      "loss": 1.5561,
      "step": 1591
    },
    {
      "epoch": 0.2280475576564962,
      "grad_norm": 1.1455917358398438,
      "learning_rate": 0.00018013683118270652,
      "loss": 1.5712,
      "step": 1592
    },
    {
      "epoch": 0.22819080360979802,
      "grad_norm": 1.208787441253662,
      "learning_rate": 0.00018010906874960754,
      "loss": 1.4216,
      "step": 1593
    },
    {
      "epoch": 0.22833404956309986,
      "grad_norm": 1.0071102380752563,
      "learning_rate": 0.00018008128907100844,
      "loss": 1.5824,
      "step": 1594
    },
    {
      "epoch": 0.22847729551640167,
      "grad_norm": 1.1912972927093506,
      "learning_rate": 0.0001800534921528895,
      "loss": 1.6535,
      "step": 1595
    },
    {
      "epoch": 0.22862054146970348,
      "grad_norm": 1.4216176271438599,
      "learning_rate": 0.00018002567800123474,
      "loss": 1.4237,
      "step": 1596
    },
    {
      "epoch": 0.2287637874230053,
      "grad_norm": 1.387319803237915,
      "learning_rate": 0.00017999784662203178,
      "loss": 1.2825,
      "step": 1597
    },
    {
      "epoch": 0.2289070333763071,
      "grad_norm": 1.398067831993103,
      "learning_rate": 0.00017996999802127214,
      "loss": 1.5442,
      "step": 1598
    },
    {
      "epoch": 0.22905027932960895,
      "grad_norm": 1.1122944355010986,
      "learning_rate": 0.00017994213220495084,
      "loss": 1.6582,
      "step": 1599
    },
    {
      "epoch": 0.22919352528291076,
      "grad_norm": 1.4073266983032227,
      "learning_rate": 0.00017991424917906677,
      "loss": 1.3174,
      "step": 1600
    },
    {
      "epoch": 0.22933677123621257,
      "grad_norm": 1.1288211345672607,
      "learning_rate": 0.0001798863489496224,
      "loss": 1.5664,
      "step": 1601
    },
    {
      "epoch": 0.22948001718951438,
      "grad_norm": 1.2065292596817017,
      "learning_rate": 0.00017985843152262397,
      "loss": 1.538,
      "step": 1602
    },
    {
      "epoch": 0.22962326314281623,
      "grad_norm": 1.2226641178131104,
      "learning_rate": 0.00017983049690408146,
      "loss": 1.5167,
      "step": 1603
    },
    {
      "epoch": 0.22976650909611804,
      "grad_norm": 1.327951431274414,
      "learning_rate": 0.00017980254510000844,
      "loss": 1.5346,
      "step": 1604
    },
    {
      "epoch": 0.22990975504941985,
      "grad_norm": 1.0875424146652222,
      "learning_rate": 0.00017977457611642226,
      "loss": 1.5092,
      "step": 1605
    },
    {
      "epoch": 0.23005300100272166,
      "grad_norm": 1.277587652206421,
      "learning_rate": 0.00017974658995934396,
      "loss": 1.4052,
      "step": 1606
    },
    {
      "epoch": 0.2301962469560235,
      "grad_norm": 1.008635401725769,
      "learning_rate": 0.00017971858663479826,
      "loss": 1.5983,
      "step": 1607
    },
    {
      "epoch": 0.23033949290932532,
      "grad_norm": 1.410136342048645,
      "learning_rate": 0.0001796905661488136,
      "loss": 1.4776,
      "step": 1608
    },
    {
      "epoch": 0.23048273886262713,
      "grad_norm": 1.2685760259628296,
      "learning_rate": 0.0001796625285074221,
      "loss": 1.3976,
      "step": 1609
    },
    {
      "epoch": 0.23062598481592894,
      "grad_norm": 1.0443456172943115,
      "learning_rate": 0.0001796344737166595,
      "loss": 1.6023,
      "step": 1610
    },
    {
      "epoch": 0.23076923076923078,
      "grad_norm": 1.1500216722488403,
      "learning_rate": 0.0001796064017825654,
      "loss": 1.5464,
      "step": 1611
    },
    {
      "epoch": 0.2309124767225326,
      "grad_norm": 1.2528773546218872,
      "learning_rate": 0.00017957831271118294,
      "loss": 1.6987,
      "step": 1612
    },
    {
      "epoch": 0.2310557226758344,
      "grad_norm": 1.1374362707138062,
      "learning_rate": 0.000179550206508559,
      "loss": 1.607,
      "step": 1613
    },
    {
      "epoch": 0.23119896862913622,
      "grad_norm": 1.2306231260299683,
      "learning_rate": 0.00017952208318074416,
      "loss": 1.4653,
      "step": 1614
    },
    {
      "epoch": 0.23134221458243803,
      "grad_norm": 1.2008354663848877,
      "learning_rate": 0.0001794939427337927,
      "loss": 1.501,
      "step": 1615
    },
    {
      "epoch": 0.23148546053573987,
      "grad_norm": 1.3857522010803223,
      "learning_rate": 0.0001794657851737625,
      "loss": 1.3772,
      "step": 1616
    },
    {
      "epoch": 0.23162870648904169,
      "grad_norm": 1.1792274713516235,
      "learning_rate": 0.00017943761050671526,
      "loss": 1.4868,
      "step": 1617
    },
    {
      "epoch": 0.2317719524423435,
      "grad_norm": 1.3106101751327515,
      "learning_rate": 0.00017940941873871626,
      "loss": 1.495,
      "step": 1618
    },
    {
      "epoch": 0.2319151983956453,
      "grad_norm": 1.1484472751617432,
      "learning_rate": 0.00017938120987583445,
      "loss": 1.6104,
      "step": 1619
    },
    {
      "epoch": 0.23205844434894715,
      "grad_norm": 1.16348397731781,
      "learning_rate": 0.00017935298392414257,
      "loss": 1.5255,
      "step": 1620
    },
    {
      "epoch": 0.23220169030224896,
      "grad_norm": 1.138157606124878,
      "learning_rate": 0.00017932474088971692,
      "loss": 1.5399,
      "step": 1621
    },
    {
      "epoch": 0.23234493625555078,
      "grad_norm": 1.0980589389801025,
      "learning_rate": 0.0001792964807786375,
      "loss": 1.5398,
      "step": 1622
    },
    {
      "epoch": 0.2324881822088526,
      "grad_norm": 1.0768204927444458,
      "learning_rate": 0.0001792682035969881,
      "loss": 1.5361,
      "step": 1623
    },
    {
      "epoch": 0.23263142816215443,
      "grad_norm": 0.9410218000411987,
      "learning_rate": 0.00017923990935085602,
      "loss": 1.5161,
      "step": 1624
    },
    {
      "epoch": 0.23277467411545624,
      "grad_norm": 1.1880786418914795,
      "learning_rate": 0.00017921159804633238,
      "loss": 1.315,
      "step": 1625
    },
    {
      "epoch": 0.23291792006875806,
      "grad_norm": 1.1521860361099243,
      "learning_rate": 0.00017918326968951182,
      "loss": 1.5468,
      "step": 1626
    },
    {
      "epoch": 0.23306116602205987,
      "grad_norm": 1.2211812734603882,
      "learning_rate": 0.00017915492428649277,
      "loss": 1.7284,
      "step": 1627
    },
    {
      "epoch": 0.2332044119753617,
      "grad_norm": 1.0255587100982666,
      "learning_rate": 0.00017912656184337734,
      "loss": 1.7276,
      "step": 1628
    },
    {
      "epoch": 0.23334765792866352,
      "grad_norm": 1.038635492324829,
      "learning_rate": 0.00017909818236627124,
      "loss": 1.5494,
      "step": 1629
    },
    {
      "epoch": 0.23349090388196533,
      "grad_norm": 1.132211685180664,
      "learning_rate": 0.0001790697858612838,
      "loss": 1.411,
      "step": 1630
    },
    {
      "epoch": 0.23363414983526715,
      "grad_norm": 1.2159178256988525,
      "learning_rate": 0.00017904137233452815,
      "loss": 1.46,
      "step": 1631
    },
    {
      "epoch": 0.23377739578856896,
      "grad_norm": 1.3370416164398193,
      "learning_rate": 0.000179012941792121,
      "loss": 1.4356,
      "step": 1632
    },
    {
      "epoch": 0.2339206417418708,
      "grad_norm": 1.1122897863388062,
      "learning_rate": 0.00017898449424018278,
      "loss": 1.4177,
      "step": 1633
    },
    {
      "epoch": 0.2340638876951726,
      "grad_norm": 1.1317368745803833,
      "learning_rate": 0.0001789560296848375,
      "loss": 1.4307,
      "step": 1634
    },
    {
      "epoch": 0.23420713364847442,
      "grad_norm": 1.1846390962600708,
      "learning_rate": 0.00017892754813221288,
      "loss": 1.389,
      "step": 1635
    },
    {
      "epoch": 0.23435037960177624,
      "grad_norm": 1.259944200515747,
      "learning_rate": 0.0001788990495884403,
      "loss": 1.496,
      "step": 1636
    },
    {
      "epoch": 0.23449362555507808,
      "grad_norm": 1.4734100103378296,
      "learning_rate": 0.00017887053405965482,
      "loss": 1.4312,
      "step": 1637
    },
    {
      "epoch": 0.2346368715083799,
      "grad_norm": 1.0750943422317505,
      "learning_rate": 0.00017884200155199507,
      "loss": 1.3276,
      "step": 1638
    },
    {
      "epoch": 0.2347801174616817,
      "grad_norm": 1.1855219602584839,
      "learning_rate": 0.00017881345207160343,
      "loss": 1.6613,
      "step": 1639
    },
    {
      "epoch": 0.23492336341498352,
      "grad_norm": 0.9636190533638,
      "learning_rate": 0.00017878488562462588,
      "loss": 1.5859,
      "step": 1640
    },
    {
      "epoch": 0.23506660936828536,
      "grad_norm": 1.403967022895813,
      "learning_rate": 0.00017875630221721208,
      "loss": 1.3911,
      "step": 1641
    },
    {
      "epoch": 0.23520985532158717,
      "grad_norm": 0.9366437792778015,
      "learning_rate": 0.00017872770185551535,
      "loss": 1.5734,
      "step": 1642
    },
    {
      "epoch": 0.23535310127488898,
      "grad_norm": 1.1883364915847778,
      "learning_rate": 0.00017869908454569257,
      "loss": 1.3841,
      "step": 1643
    },
    {
      "epoch": 0.2354963472281908,
      "grad_norm": 1.170893907546997,
      "learning_rate": 0.00017867045029390445,
      "loss": 1.5987,
      "step": 1644
    },
    {
      "epoch": 0.23563959318149263,
      "grad_norm": 1.1183254718780518,
      "learning_rate": 0.00017864179910631514,
      "loss": 1.545,
      "step": 1645
    },
    {
      "epoch": 0.23578283913479445,
      "grad_norm": 1.2897900342941284,
      "learning_rate": 0.00017861313098909256,
      "loss": 1.3466,
      "step": 1646
    },
    {
      "epoch": 0.23592608508809626,
      "grad_norm": 1.139197826385498,
      "learning_rate": 0.00017858444594840827,
      "loss": 1.5891,
      "step": 1647
    },
    {
      "epoch": 0.23606933104139807,
      "grad_norm": 1.2418596744537354,
      "learning_rate": 0.00017855574399043743,
      "loss": 1.5439,
      "step": 1648
    },
    {
      "epoch": 0.2362125769946999,
      "grad_norm": 1.104661226272583,
      "learning_rate": 0.00017852702512135884,
      "loss": 1.5013,
      "step": 1649
    },
    {
      "epoch": 0.23635582294800173,
      "grad_norm": 1.2875605821609497,
      "learning_rate": 0.00017849828934735497,
      "loss": 1.5453,
      "step": 1650
    },
    {
      "epoch": 0.23649906890130354,
      "grad_norm": 1.6004868745803833,
      "learning_rate": 0.00017846953667461198,
      "loss": 1.3004,
      "step": 1651
    },
    {
      "epoch": 0.23664231485460535,
      "grad_norm": 1.215222954750061,
      "learning_rate": 0.00017844076710931954,
      "loss": 1.2455,
      "step": 1652
    },
    {
      "epoch": 0.23678556080790716,
      "grad_norm": 1.2152013778686523,
      "learning_rate": 0.00017841198065767107,
      "loss": 1.4268,
      "step": 1653
    },
    {
      "epoch": 0.236928806761209,
      "grad_norm": 1.139962077140808,
      "learning_rate": 0.00017838317732586352,
      "loss": 1.5386,
      "step": 1654
    },
    {
      "epoch": 0.23707205271451082,
      "grad_norm": 0.9816656708717346,
      "learning_rate": 0.0001783543571200976,
      "loss": 1.4273,
      "step": 1655
    },
    {
      "epoch": 0.23721529866781263,
      "grad_norm": 1.1998710632324219,
      "learning_rate": 0.00017832552004657756,
      "loss": 1.45,
      "step": 1656
    },
    {
      "epoch": 0.23735854462111444,
      "grad_norm": 1.3308093547821045,
      "learning_rate": 0.0001782966661115113,
      "loss": 1.5399,
      "step": 1657
    },
    {
      "epoch": 0.23750179057441628,
      "grad_norm": 1.0545663833618164,
      "learning_rate": 0.00017826779532111037,
      "loss": 1.6803,
      "step": 1658
    },
    {
      "epoch": 0.2376450365277181,
      "grad_norm": 1.4090311527252197,
      "learning_rate": 0.00017823890768158996,
      "loss": 1.3388,
      "step": 1659
    },
    {
      "epoch": 0.2377882824810199,
      "grad_norm": 1.0986031293869019,
      "learning_rate": 0.0001782100031991688,
      "loss": 1.5565,
      "step": 1660
    },
    {
      "epoch": 0.23793152843432172,
      "grad_norm": 1.23817777633667,
      "learning_rate": 0.00017818108188006937,
      "loss": 1.3763,
      "step": 1661
    },
    {
      "epoch": 0.23807477438762356,
      "grad_norm": 1.0550546646118164,
      "learning_rate": 0.0001781521437305177,
      "loss": 1.5913,
      "step": 1662
    },
    {
      "epoch": 0.23821802034092537,
      "grad_norm": 1.0696789026260376,
      "learning_rate": 0.00017812318875674342,
      "loss": 1.4448,
      "step": 1663
    },
    {
      "epoch": 0.2383612662942272,
      "grad_norm": 0.9307172894477844,
      "learning_rate": 0.00017809421696497987,
      "loss": 1.5858,
      "step": 1664
    },
    {
      "epoch": 0.238504512247529,
      "grad_norm": 1.2478545904159546,
      "learning_rate": 0.00017806522836146395,
      "loss": 1.4587,
      "step": 1665
    },
    {
      "epoch": 0.23864775820083084,
      "grad_norm": 1.224277377128601,
      "learning_rate": 0.00017803622295243615,
      "loss": 1.3342,
      "step": 1666
    },
    {
      "epoch": 0.23879100415413265,
      "grad_norm": 1.021478533744812,
      "learning_rate": 0.00017800720074414068,
      "loss": 1.6131,
      "step": 1667
    },
    {
      "epoch": 0.23893425010743446,
      "grad_norm": 1.0162409543991089,
      "learning_rate": 0.00017797816174282524,
      "loss": 1.6935,
      "step": 1668
    },
    {
      "epoch": 0.23907749606073628,
      "grad_norm": 1.1760752201080322,
      "learning_rate": 0.00017794910595474126,
      "loss": 1.3912,
      "step": 1669
    },
    {
      "epoch": 0.2392207420140381,
      "grad_norm": 1.0430537462234497,
      "learning_rate": 0.00017792003338614368,
      "loss": 1.3529,
      "step": 1670
    },
    {
      "epoch": 0.23936398796733993,
      "grad_norm": 1.2220804691314697,
      "learning_rate": 0.00017789094404329115,
      "loss": 1.5991,
      "step": 1671
    },
    {
      "epoch": 0.23950723392064174,
      "grad_norm": 1.2312926054000854,
      "learning_rate": 0.0001778618379324459,
      "loss": 1.4976,
      "step": 1672
    },
    {
      "epoch": 0.23965047987394356,
      "grad_norm": 1.2138832807540894,
      "learning_rate": 0.00017783271505987367,
      "loss": 1.5458,
      "step": 1673
    },
    {
      "epoch": 0.23979372582724537,
      "grad_norm": 1.0128521919250488,
      "learning_rate": 0.00017780357543184397,
      "loss": 1.5235,
      "step": 1674
    },
    {
      "epoch": 0.2399369717805472,
      "grad_norm": 1.0863696336746216,
      "learning_rate": 0.00017777441905462982,
      "loss": 1.646,
      "step": 1675
    },
    {
      "epoch": 0.24008021773384902,
      "grad_norm": 1.2562332153320312,
      "learning_rate": 0.00017774524593450785,
      "loss": 1.5221,
      "step": 1676
    },
    {
      "epoch": 0.24022346368715083,
      "grad_norm": 1.2026699781417847,
      "learning_rate": 0.00017771605607775834,
      "loss": 1.4357,
      "step": 1677
    },
    {
      "epoch": 0.24036670964045265,
      "grad_norm": 1.2922749519348145,
      "learning_rate": 0.00017768684949066515,
      "loss": 1.556,
      "step": 1678
    },
    {
      "epoch": 0.2405099555937545,
      "grad_norm": 1.2074483633041382,
      "learning_rate": 0.0001776576261795157,
      "loss": 1.5278,
      "step": 1679
    },
    {
      "epoch": 0.2406532015470563,
      "grad_norm": 1.2819236516952515,
      "learning_rate": 0.00017762838615060104,
      "loss": 1.5061,
      "step": 1680
    },
    {
      "epoch": 0.2407964475003581,
      "grad_norm": 1.1819390058517456,
      "learning_rate": 0.00017759912941021585,
      "loss": 1.3712,
      "step": 1681
    },
    {
      "epoch": 0.24093969345365993,
      "grad_norm": 1.1596397161483765,
      "learning_rate": 0.0001775698559646584,
      "loss": 1.5251,
      "step": 1682
    },
    {
      "epoch": 0.24108293940696177,
      "grad_norm": 1.276223063468933,
      "learning_rate": 0.00017754056582023052,
      "loss": 1.5252,
      "step": 1683
    },
    {
      "epoch": 0.24122618536026358,
      "grad_norm": 1.2092808485031128,
      "learning_rate": 0.00017751125898323765,
      "loss": 1.4462,
      "step": 1684
    },
    {
      "epoch": 0.2413694313135654,
      "grad_norm": 1.1401129961013794,
      "learning_rate": 0.00017748193545998883,
      "loss": 1.517,
      "step": 1685
    },
    {
      "epoch": 0.2415126772668672,
      "grad_norm": 1.019573450088501,
      "learning_rate": 0.00017745259525679666,
      "loss": 1.6774,
      "step": 1686
    },
    {
      "epoch": 0.24165592322016902,
      "grad_norm": 1.0991650819778442,
      "learning_rate": 0.0001774232383799774,
      "loss": 1.5439,
      "step": 1687
    },
    {
      "epoch": 0.24179916917347086,
      "grad_norm": 1.1929291486740112,
      "learning_rate": 0.00017739386483585083,
      "loss": 1.5512,
      "step": 1688
    },
    {
      "epoch": 0.24194241512677267,
      "grad_norm": 1.2396111488342285,
      "learning_rate": 0.00017736447463074037,
      "loss": 1.4041,
      "step": 1689
    },
    {
      "epoch": 0.24208566108007448,
      "grad_norm": 1.1199657917022705,
      "learning_rate": 0.000177335067770973,
      "loss": 1.5615,
      "step": 1690
    },
    {
      "epoch": 0.2422289070333763,
      "grad_norm": 1.2150777578353882,
      "learning_rate": 0.0001773056442628793,
      "loss": 1.5394,
      "step": 1691
    },
    {
      "epoch": 0.24237215298667814,
      "grad_norm": 1.1789854764938354,
      "learning_rate": 0.00017727620411279337,
      "loss": 1.5883,
      "step": 1692
    },
    {
      "epoch": 0.24251539893997995,
      "grad_norm": 1.1000840663909912,
      "learning_rate": 0.00017724674732705301,
      "loss": 1.5895,
      "step": 1693
    },
    {
      "epoch": 0.24265864489328176,
      "grad_norm": 1.2763997316360474,
      "learning_rate": 0.0001772172739119995,
      "loss": 1.6236,
      "step": 1694
    },
    {
      "epoch": 0.24280189084658357,
      "grad_norm": 1.0187008380889893,
      "learning_rate": 0.00017718778387397775,
      "loss": 1.486,
      "step": 1695
    },
    {
      "epoch": 0.24294513679988541,
      "grad_norm": 1.150473713874817,
      "learning_rate": 0.00017715827721933623,
      "loss": 1.4881,
      "step": 1696
    },
    {
      "epoch": 0.24308838275318723,
      "grad_norm": 1.050354242324829,
      "learning_rate": 0.00017712875395442702,
      "loss": 1.6785,
      "step": 1697
    },
    {
      "epoch": 0.24323162870648904,
      "grad_norm": 1.1723244190216064,
      "learning_rate": 0.0001770992140856057,
      "loss": 1.3832,
      "step": 1698
    },
    {
      "epoch": 0.24337487465979085,
      "grad_norm": 1.0287948846817017,
      "learning_rate": 0.0001770696576192315,
      "loss": 1.4747,
      "step": 1699
    },
    {
      "epoch": 0.2435181206130927,
      "grad_norm": 1.2437360286712646,
      "learning_rate": 0.00017704008456166718,
      "loss": 1.5576,
      "step": 1700
    },
    {
      "epoch": 0.2436613665663945,
      "grad_norm": 1.2614781856536865,
      "learning_rate": 0.00017701049491927916,
      "loss": 1.5181,
      "step": 1701
    },
    {
      "epoch": 0.24380461251969632,
      "grad_norm": 1.356305718421936,
      "learning_rate": 0.00017698088869843722,
      "loss": 1.2966,
      "step": 1702
    },
    {
      "epoch": 0.24394785847299813,
      "grad_norm": 1.1504061222076416,
      "learning_rate": 0.00017695126590551495,
      "loss": 1.5534,
      "step": 1703
    },
    {
      "epoch": 0.24409110442629997,
      "grad_norm": 1.174885630607605,
      "learning_rate": 0.00017692162654688943,
      "loss": 1.5194,
      "step": 1704
    },
    {
      "epoch": 0.24423435037960178,
      "grad_norm": 1.1576462984085083,
      "learning_rate": 0.00017689197062894116,
      "loss": 1.5961,
      "step": 1705
    },
    {
      "epoch": 0.2443775963329036,
      "grad_norm": 1.4788256883621216,
      "learning_rate": 0.00017686229815805441,
      "loss": 1.4422,
      "step": 1706
    },
    {
      "epoch": 0.2445208422862054,
      "grad_norm": 1.0189123153686523,
      "learning_rate": 0.00017683260914061694,
      "loss": 1.5468,
      "step": 1707
    },
    {
      "epoch": 0.24466408823950722,
      "grad_norm": 1.221017837524414,
      "learning_rate": 0.00017680290358302,
      "loss": 1.5931,
      "step": 1708
    },
    {
      "epoch": 0.24480733419280906,
      "grad_norm": 1.057818055152893,
      "learning_rate": 0.00017677318149165853,
      "loss": 1.5742,
      "step": 1709
    },
    {
      "epoch": 0.24495058014611087,
      "grad_norm": 1.3094794750213623,
      "learning_rate": 0.00017674344287293089,
      "loss": 1.6952,
      "step": 1710
    },
    {
      "epoch": 0.2450938260994127,
      "grad_norm": 1.2983174324035645,
      "learning_rate": 0.00017671368773323912,
      "loss": 1.5533,
      "step": 1711
    },
    {
      "epoch": 0.2452370720527145,
      "grad_norm": 1.2301416397094727,
      "learning_rate": 0.00017668391607898878,
      "loss": 1.6913,
      "step": 1712
    },
    {
      "epoch": 0.24538031800601634,
      "grad_norm": 1.3136086463928223,
      "learning_rate": 0.00017665412791658888,
      "loss": 1.4652,
      "step": 1713
    },
    {
      "epoch": 0.24552356395931815,
      "grad_norm": 1.0584815740585327,
      "learning_rate": 0.00017662432325245218,
      "loss": 1.5523,
      "step": 1714
    },
    {
      "epoch": 0.24566680991261997,
      "grad_norm": 1.1548343896865845,
      "learning_rate": 0.00017659450209299484,
      "loss": 1.4768,
      "step": 1715
    },
    {
      "epoch": 0.24581005586592178,
      "grad_norm": 1.2268909215927124,
      "learning_rate": 0.00017656466444463658,
      "loss": 1.4709,
      "step": 1716
    },
    {
      "epoch": 0.24595330181922362,
      "grad_norm": 1.257816195487976,
      "learning_rate": 0.0001765348103138008,
      "loss": 1.4373,
      "step": 1717
    },
    {
      "epoch": 0.24609654777252543,
      "grad_norm": 1.1751478910446167,
      "learning_rate": 0.0001765049397069143,
      "loss": 1.4415,
      "step": 1718
    },
    {
      "epoch": 0.24623979372582724,
      "grad_norm": 1.1696817874908447,
      "learning_rate": 0.00017647505263040743,
      "loss": 1.5036,
      "step": 1719
    },
    {
      "epoch": 0.24638303967912906,
      "grad_norm": 1.1908512115478516,
      "learning_rate": 0.00017644514909071423,
      "loss": 1.5009,
      "step": 1720
    },
    {
      "epoch": 0.2465262856324309,
      "grad_norm": 1.2070001363754272,
      "learning_rate": 0.00017641522909427214,
      "loss": 1.4688,
      "step": 1721
    },
    {
      "epoch": 0.2466695315857327,
      "grad_norm": 0.9402499198913574,
      "learning_rate": 0.00017638529264752225,
      "loss": 1.5808,
      "step": 1722
    },
    {
      "epoch": 0.24681277753903452,
      "grad_norm": 1.0196526050567627,
      "learning_rate": 0.00017635533975690905,
      "loss": 1.5524,
      "step": 1723
    },
    {
      "epoch": 0.24695602349233634,
      "grad_norm": 1.006087064743042,
      "learning_rate": 0.0001763253704288807,
      "loss": 1.62,
      "step": 1724
    },
    {
      "epoch": 0.24709926944563815,
      "grad_norm": 1.0902975797653198,
      "learning_rate": 0.00017629538466988888,
      "loss": 1.6567,
      "step": 1725
    },
    {
      "epoch": 0.24724251539894,
      "grad_norm": 1.2466589212417603,
      "learning_rate": 0.00017626538248638872,
      "loss": 1.5318,
      "step": 1726
    },
    {
      "epoch": 0.2473857613522418,
      "grad_norm": 1.2524625062942505,
      "learning_rate": 0.00017623536388483905,
      "loss": 1.6266,
      "step": 1727
    },
    {
      "epoch": 0.2475290073055436,
      "grad_norm": 1.467842936515808,
      "learning_rate": 0.000176205328871702,
      "loss": 1.5037,
      "step": 1728
    },
    {
      "epoch": 0.24767225325884543,
      "grad_norm": 0.9722418189048767,
      "learning_rate": 0.00017617527745344348,
      "loss": 1.6168,
      "step": 1729
    },
    {
      "epoch": 0.24781549921214727,
      "grad_norm": 1.1312811374664307,
      "learning_rate": 0.00017614520963653277,
      "loss": 1.7052,
      "step": 1730
    },
    {
      "epoch": 0.24795874516544908,
      "grad_norm": 1.1263035535812378,
      "learning_rate": 0.0001761151254274427,
      "loss": 1.5386,
      "step": 1731
    },
    {
      "epoch": 0.2481019911187509,
      "grad_norm": 1.2759530544281006,
      "learning_rate": 0.0001760850248326497,
      "loss": 1.4436,
      "step": 1732
    },
    {
      "epoch": 0.2482452370720527,
      "grad_norm": 1.198973298072815,
      "learning_rate": 0.00017605490785863368,
      "loss": 1.6428,
      "step": 1733
    },
    {
      "epoch": 0.24838848302535455,
      "grad_norm": 1.0669065713882446,
      "learning_rate": 0.00017602477451187802,
      "loss": 1.4647,
      "step": 1734
    },
    {
      "epoch": 0.24853172897865636,
      "grad_norm": 1.157007098197937,
      "learning_rate": 0.00017599462479886974,
      "loss": 1.4387,
      "step": 1735
    },
    {
      "epoch": 0.24867497493195817,
      "grad_norm": 1.0468538999557495,
      "learning_rate": 0.00017596445872609934,
      "loss": 1.4636,
      "step": 1736
    },
    {
      "epoch": 0.24881822088525998,
      "grad_norm": 0.9225525259971619,
      "learning_rate": 0.0001759342763000608,
      "loss": 1.6646,
      "step": 1737
    },
    {
      "epoch": 0.24896146683856182,
      "grad_norm": 1.3613708019256592,
      "learning_rate": 0.00017590407752725166,
      "loss": 1.5077,
      "step": 1738
    },
    {
      "epoch": 0.24910471279186364,
      "grad_norm": 1.1746326684951782,
      "learning_rate": 0.00017587386241417293,
      "loss": 1.4209,
      "step": 1739
    },
    {
      "epoch": 0.24924795874516545,
      "grad_norm": 1.1406930685043335,
      "learning_rate": 0.00017584363096732926,
      "loss": 1.4518,
      "step": 1740
    },
    {
      "epoch": 0.24939120469846726,
      "grad_norm": 1.1395173072814941,
      "learning_rate": 0.00017581338319322862,
      "loss": 1.7543,
      "step": 1741
    },
    {
      "epoch": 0.24953445065176907,
      "grad_norm": 0.9967325925827026,
      "learning_rate": 0.0001757831190983827,
      "loss": 1.5416,
      "step": 1742
    },
    {
      "epoch": 0.24967769660507091,
      "grad_norm": 1.0967161655426025,
      "learning_rate": 0.00017575283868930658,
      "loss": 1.5241,
      "step": 1743
    },
    {
      "epoch": 0.24982094255837273,
      "grad_norm": 1.072513461112976,
      "learning_rate": 0.0001757225419725189,
      "loss": 1.4348,
      "step": 1744
    },
    {
      "epoch": 0.24996418851167454,
      "grad_norm": 1.2924058437347412,
      "learning_rate": 0.00017569222895454174,
      "loss": 1.5236,
      "step": 1745
    },
    {
      "epoch": 0.25010743446497635,
      "grad_norm": 1.1272467374801636,
      "learning_rate": 0.0001756618996419008,
      "loss": 1.5072,
      "step": 1746
    },
    {
      "epoch": 0.2502506804182782,
      "grad_norm": 1.1392260789871216,
      "learning_rate": 0.00017563155404112522,
      "loss": 1.5721,
      "step": 1747
    },
    {
      "epoch": 0.25039392637158,
      "grad_norm": 1.1709660291671753,
      "learning_rate": 0.00017560119215874763,
      "loss": 1.4299,
      "step": 1748
    },
    {
      "epoch": 0.2505371723248818,
      "grad_norm": 1.3311039209365845,
      "learning_rate": 0.0001755708140013042,
      "loss": 1.5245,
      "step": 1749
    },
    {
      "epoch": 0.25068041827818366,
      "grad_norm": 1.2343685626983643,
      "learning_rate": 0.0001755404195753346,
      "loss": 1.6023,
      "step": 1750
    },
    {
      "epoch": 0.25082366423148544,
      "grad_norm": 1.0501316785812378,
      "learning_rate": 0.000175510008887382,
      "loss": 1.558,
      "step": 1751
    },
    {
      "epoch": 0.2509669101847873,
      "grad_norm": 0.9963275194168091,
      "learning_rate": 0.00017547958194399308,
      "loss": 1.5513,
      "step": 1752
    },
    {
      "epoch": 0.2511101561380891,
      "grad_norm": 1.0529346466064453,
      "learning_rate": 0.000175449138751718,
      "loss": 1.7666,
      "step": 1753
    },
    {
      "epoch": 0.2512534020913909,
      "grad_norm": 1.128468632698059,
      "learning_rate": 0.00017541867931711042,
      "loss": 1.4772,
      "step": 1754
    },
    {
      "epoch": 0.25139664804469275,
      "grad_norm": 1.3378150463104248,
      "learning_rate": 0.0001753882036467275,
      "loss": 1.4729,
      "step": 1755
    },
    {
      "epoch": 0.25153989399799453,
      "grad_norm": 0.9767134189605713,
      "learning_rate": 0.00017535771174712988,
      "loss": 1.4067,
      "step": 1756
    },
    {
      "epoch": 0.2516831399512964,
      "grad_norm": 1.0687772035598755,
      "learning_rate": 0.0001753272036248818,
      "loss": 1.5425,
      "step": 1757
    },
    {
      "epoch": 0.2518263859045982,
      "grad_norm": 1.1526464223861694,
      "learning_rate": 0.00017529667928655078,
      "loss": 1.513,
      "step": 1758
    },
    {
      "epoch": 0.2519696318579,
      "grad_norm": 1.2871372699737549,
      "learning_rate": 0.00017526613873870806,
      "loss": 1.5584,
      "step": 1759
    },
    {
      "epoch": 0.25211287781120184,
      "grad_norm": 1.1934125423431396,
      "learning_rate": 0.00017523558198792816,
      "loss": 1.4383,
      "step": 1760
    },
    {
      "epoch": 0.2522561237645036,
      "grad_norm": 1.0536959171295166,
      "learning_rate": 0.0001752050090407893,
      "loss": 1.565,
      "step": 1761
    },
    {
      "epoch": 0.25239936971780547,
      "grad_norm": 1.4676578044891357,
      "learning_rate": 0.000175174419903873,
      "loss": 1.4668,
      "step": 1762
    },
    {
      "epoch": 0.2525426156711073,
      "grad_norm": 1.408983588218689,
      "learning_rate": 0.00017514381458376437,
      "loss": 1.4313,
      "step": 1763
    },
    {
      "epoch": 0.2526858616244091,
      "grad_norm": 1.0105700492858887,
      "learning_rate": 0.00017511319308705198,
      "loss": 1.5167,
      "step": 1764
    },
    {
      "epoch": 0.25282910757771093,
      "grad_norm": 1.017472505569458,
      "learning_rate": 0.0001750825554203279,
      "loss": 1.5071,
      "step": 1765
    },
    {
      "epoch": 0.2529723535310128,
      "grad_norm": 1.105825662612915,
      "learning_rate": 0.00017505190159018764,
      "loss": 1.4043,
      "step": 1766
    },
    {
      "epoch": 0.25311559948431456,
      "grad_norm": 1.534285306930542,
      "learning_rate": 0.0001750212316032302,
      "loss": 1.4089,
      "step": 1767
    },
    {
      "epoch": 0.2532588454376164,
      "grad_norm": 1.4834009408950806,
      "learning_rate": 0.00017499054546605812,
      "loss": 1.4654,
      "step": 1768
    },
    {
      "epoch": 0.2534020913909182,
      "grad_norm": 0.9281497597694397,
      "learning_rate": 0.0001749598431852773,
      "loss": 1.5428,
      "step": 1769
    },
    {
      "epoch": 0.25354533734422,
      "grad_norm": 1.2559762001037598,
      "learning_rate": 0.00017492912476749722,
      "loss": 1.5603,
      "step": 1770
    },
    {
      "epoch": 0.25368858329752186,
      "grad_norm": 0.982282817363739,
      "learning_rate": 0.00017489839021933077,
      "loss": 1.4609,
      "step": 1771
    },
    {
      "epoch": 0.25383182925082365,
      "grad_norm": 1.3256710767745972,
      "learning_rate": 0.00017486763954739436,
      "loss": 1.4468,
      "step": 1772
    },
    {
      "epoch": 0.2539750752041255,
      "grad_norm": 0.9639436602592468,
      "learning_rate": 0.00017483687275830783,
      "loss": 1.352,
      "step": 1773
    },
    {
      "epoch": 0.25411832115742733,
      "grad_norm": 1.022750973701477,
      "learning_rate": 0.00017480608985869452,
      "loss": 1.5195,
      "step": 1774
    },
    {
      "epoch": 0.2542615671107291,
      "grad_norm": 1.0688238143920898,
      "learning_rate": 0.00017477529085518128,
      "loss": 1.4605,
      "step": 1775
    },
    {
      "epoch": 0.25440481306403095,
      "grad_norm": 1.2416025400161743,
      "learning_rate": 0.00017474447575439826,
      "loss": 1.4796,
      "step": 1776
    },
    {
      "epoch": 0.25454805901733274,
      "grad_norm": 1.5684871673583984,
      "learning_rate": 0.00017471364456297925,
      "loss": 1.5657,
      "step": 1777
    },
    {
      "epoch": 0.2546913049706346,
      "grad_norm": 1.3178629875183105,
      "learning_rate": 0.00017468279728756147,
      "loss": 1.3458,
      "step": 1778
    },
    {
      "epoch": 0.2548345509239364,
      "grad_norm": 0.9080681204795837,
      "learning_rate": 0.00017465193393478555,
      "loss": 1.5403,
      "step": 1779
    },
    {
      "epoch": 0.2549777968772382,
      "grad_norm": 1.1326262950897217,
      "learning_rate": 0.0001746210545112956,
      "loss": 1.4565,
      "step": 1780
    },
    {
      "epoch": 0.25512104283054005,
      "grad_norm": 1.1706405878067017,
      "learning_rate": 0.00017459015902373916,
      "loss": 1.3311,
      "step": 1781
    },
    {
      "epoch": 0.25526428878384183,
      "grad_norm": 1.2679508924484253,
      "learning_rate": 0.00017455924747876734,
      "loss": 1.5045,
      "step": 1782
    },
    {
      "epoch": 0.25540753473714367,
      "grad_norm": 1.0095635652542114,
      "learning_rate": 0.00017452831988303458,
      "loss": 1.5044,
      "step": 1783
    },
    {
      "epoch": 0.2555507806904455,
      "grad_norm": 1.1225614547729492,
      "learning_rate": 0.0001744973762431988,
      "loss": 1.5232,
      "step": 1784
    },
    {
      "epoch": 0.2556940266437473,
      "grad_norm": 1.2272361516952515,
      "learning_rate": 0.0001744664165659215,
      "loss": 1.3776,
      "step": 1785
    },
    {
      "epoch": 0.25583727259704914,
      "grad_norm": 1.1880406141281128,
      "learning_rate": 0.00017443544085786746,
      "loss": 1.5442,
      "step": 1786
    },
    {
      "epoch": 0.255980518550351,
      "grad_norm": 1.1158430576324463,
      "learning_rate": 0.000174404449125705,
      "loss": 1.6801,
      "step": 1787
    },
    {
      "epoch": 0.25612376450365276,
      "grad_norm": 1.192739725112915,
      "learning_rate": 0.00017437344137610582,
      "loss": 1.4992,
      "step": 1788
    },
    {
      "epoch": 0.2562670104569546,
      "grad_norm": 1.2527509927749634,
      "learning_rate": 0.0001743424176157452,
      "loss": 1.4861,
      "step": 1789
    },
    {
      "epoch": 0.2564102564102564,
      "grad_norm": 1.2842694520950317,
      "learning_rate": 0.00017431137785130178,
      "loss": 1.5096,
      "step": 1790
    },
    {
      "epoch": 0.25655350236355823,
      "grad_norm": 1.2836097478866577,
      "learning_rate": 0.00017428032208945758,
      "loss": 1.569,
      "step": 1791
    },
    {
      "epoch": 0.25669674831686007,
      "grad_norm": 1.1481493711471558,
      "learning_rate": 0.00017424925033689826,
      "loss": 1.5641,
      "step": 1792
    },
    {
      "epoch": 0.25683999427016185,
      "grad_norm": 1.244872808456421,
      "learning_rate": 0.00017421816260031265,
      "loss": 1.5275,
      "step": 1793
    },
    {
      "epoch": 0.2569832402234637,
      "grad_norm": 1.20480215549469,
      "learning_rate": 0.00017418705888639328,
      "loss": 1.3315,
      "step": 1794
    },
    {
      "epoch": 0.25712648617676553,
      "grad_norm": 1.0637485980987549,
      "learning_rate": 0.00017415593920183596,
      "loss": 1.4912,
      "step": 1795
    },
    {
      "epoch": 0.2572697321300673,
      "grad_norm": 1.3856927156448364,
      "learning_rate": 0.00017412480355334005,
      "loss": 1.5383,
      "step": 1796
    },
    {
      "epoch": 0.25741297808336916,
      "grad_norm": 1.0708727836608887,
      "learning_rate": 0.00017409365194760818,
      "loss": 1.5053,
      "step": 1797
    },
    {
      "epoch": 0.25755622403667094,
      "grad_norm": 1.5909911394119263,
      "learning_rate": 0.00017406248439134662,
      "loss": 1.4566,
      "step": 1798
    },
    {
      "epoch": 0.2576994699899728,
      "grad_norm": 1.2129840850830078,
      "learning_rate": 0.0001740313008912649,
      "loss": 1.5322,
      "step": 1799
    },
    {
      "epoch": 0.2578427159432746,
      "grad_norm": 1.1843641996383667,
      "learning_rate": 0.0001740001014540761,
      "loss": 1.4607,
      "step": 1800
    },
    {
      "epoch": 0.2579859618965764,
      "grad_norm": 1.1128901243209839,
      "learning_rate": 0.0001739688860864967,
      "loss": 1.3882,
      "step": 1801
    },
    {
      "epoch": 0.25812920784987825,
      "grad_norm": 1.0887901782989502,
      "learning_rate": 0.00017393765479524652,
      "loss": 1.5362,
      "step": 1802
    },
    {
      "epoch": 0.25827245380318004,
      "grad_norm": 1.221447229385376,
      "learning_rate": 0.000173906407587049,
      "loss": 1.7137,
      "step": 1803
    },
    {
      "epoch": 0.2584156997564819,
      "grad_norm": 1.2307121753692627,
      "learning_rate": 0.00017387514446863078,
      "loss": 1.3769,
      "step": 1804
    },
    {
      "epoch": 0.2585589457097837,
      "grad_norm": 1.1925400495529175,
      "learning_rate": 0.0001738438654467221,
      "loss": 1.7529,
      "step": 1805
    },
    {
      "epoch": 0.2587021916630855,
      "grad_norm": 1.0972455739974976,
      "learning_rate": 0.00017381257052805652,
      "loss": 1.6218,
      "step": 1806
    },
    {
      "epoch": 0.25884543761638734,
      "grad_norm": 0.8442767262458801,
      "learning_rate": 0.0001737812597193711,
      "loss": 1.5325,
      "step": 1807
    },
    {
      "epoch": 0.2589886835696892,
      "grad_norm": 1.13351571559906,
      "learning_rate": 0.0001737499330274063,
      "loss": 1.3995,
      "step": 1808
    },
    {
      "epoch": 0.25913192952299097,
      "grad_norm": 1.4445209503173828,
      "learning_rate": 0.0001737185904589059,
      "loss": 1.5527,
      "step": 1809
    },
    {
      "epoch": 0.2592751754762928,
      "grad_norm": 1.145247459411621,
      "learning_rate": 0.00017368723202061724,
      "loss": 1.4185,
      "step": 1810
    },
    {
      "epoch": 0.2594184214295946,
      "grad_norm": 1.223759412765503,
      "learning_rate": 0.000173655857719291,
      "loss": 1.5832,
      "step": 1811
    },
    {
      "epoch": 0.25956166738289643,
      "grad_norm": 1.2860922813415527,
      "learning_rate": 0.00017362446756168128,
      "loss": 1.4601,
      "step": 1812
    },
    {
      "epoch": 0.2597049133361983,
      "grad_norm": 1.2196576595306396,
      "learning_rate": 0.00017359306155454563,
      "loss": 1.5545,
      "step": 1813
    },
    {
      "epoch": 0.25984815928950006,
      "grad_norm": 1.0808385610580444,
      "learning_rate": 0.00017356163970464497,
      "loss": 1.4089,
      "step": 1814
    },
    {
      "epoch": 0.2599914052428019,
      "grad_norm": 1.0863580703735352,
      "learning_rate": 0.00017353020201874367,
      "loss": 1.5375,
      "step": 1815
    },
    {
      "epoch": 0.2601346511961037,
      "grad_norm": 1.0849562883377075,
      "learning_rate": 0.00017349874850360942,
      "loss": 1.7483,
      "step": 1816
    },
    {
      "epoch": 0.2602778971494055,
      "grad_norm": 0.9827321767807007,
      "learning_rate": 0.00017346727916601345,
      "loss": 1.6488,
      "step": 1817
    },
    {
      "epoch": 0.26042114310270736,
      "grad_norm": 1.4656785726547241,
      "learning_rate": 0.00017343579401273034,
      "loss": 1.4981,
      "step": 1818
    },
    {
      "epoch": 0.26056438905600915,
      "grad_norm": 1.0635572671890259,
      "learning_rate": 0.00017340429305053801,
      "loss": 1.6685,
      "step": 1819
    },
    {
      "epoch": 0.260707635009311,
      "grad_norm": 1.2255606651306152,
      "learning_rate": 0.00017337277628621785,
      "loss": 1.5059,
      "step": 1820
    },
    {
      "epoch": 0.26085088096261283,
      "grad_norm": 1.016801357269287,
      "learning_rate": 0.00017334124372655465,
      "loss": 1.5405,
      "step": 1821
    },
    {
      "epoch": 0.2609941269159146,
      "grad_norm": 1.137406826019287,
      "learning_rate": 0.00017330969537833658,
      "loss": 1.5363,
      "step": 1822
    },
    {
      "epoch": 0.26113737286921646,
      "grad_norm": 1.0736136436462402,
      "learning_rate": 0.00017327813124835525,
      "loss": 1.3897,
      "step": 1823
    },
    {
      "epoch": 0.26128061882251824,
      "grad_norm": 1.13479745388031,
      "learning_rate": 0.0001732465513434056,
      "loss": 1.5586,
      "step": 1824
    },
    {
      "epoch": 0.2614238647758201,
      "grad_norm": 1.194003701210022,
      "learning_rate": 0.00017321495567028606,
      "loss": 1.3579,
      "step": 1825
    },
    {
      "epoch": 0.2615671107291219,
      "grad_norm": 1.1439099311828613,
      "learning_rate": 0.0001731833442357983,
      "loss": 1.4746,
      "step": 1826
    },
    {
      "epoch": 0.2617103566824237,
      "grad_norm": 1.3568291664123535,
      "learning_rate": 0.00017315171704674755,
      "loss": 1.3914,
      "step": 1827
    },
    {
      "epoch": 0.26185360263572555,
      "grad_norm": 1.0243639945983887,
      "learning_rate": 0.00017312007410994235,
      "loss": 1.699,
      "step": 1828
    },
    {
      "epoch": 0.2619968485890274,
      "grad_norm": 1.2641977071762085,
      "learning_rate": 0.0001730884154321946,
      "loss": 1.5485,
      "step": 1829
    },
    {
      "epoch": 0.26214009454232917,
      "grad_norm": 1.1948013305664062,
      "learning_rate": 0.0001730567410203197,
      "loss": 1.4031,
      "step": 1830
    },
    {
      "epoch": 0.262283340495631,
      "grad_norm": 1.1585381031036377,
      "learning_rate": 0.00017302505088113634,
      "loss": 1.5109,
      "step": 1831
    },
    {
      "epoch": 0.2624265864489328,
      "grad_norm": 1.291182279586792,
      "learning_rate": 0.0001729933450214666,
      "loss": 1.5202,
      "step": 1832
    },
    {
      "epoch": 0.26256983240223464,
      "grad_norm": 1.3056670427322388,
      "learning_rate": 0.00017296162344813598,
      "loss": 1.6215,
      "step": 1833
    },
    {
      "epoch": 0.2627130783555365,
      "grad_norm": 1.1388869285583496,
      "learning_rate": 0.00017292988616797336,
      "loss": 1.5668,
      "step": 1834
    },
    {
      "epoch": 0.26285632430883826,
      "grad_norm": 0.9789540767669678,
      "learning_rate": 0.00017289813318781098,
      "loss": 1.5762,
      "step": 1835
    },
    {
      "epoch": 0.2629995702621401,
      "grad_norm": 1.2062697410583496,
      "learning_rate": 0.00017286636451448447,
      "loss": 1.3662,
      "step": 1836
    },
    {
      "epoch": 0.2631428162154419,
      "grad_norm": 1.0115976333618164,
      "learning_rate": 0.00017283458015483283,
      "loss": 1.7368,
      "step": 1837
    },
    {
      "epoch": 0.26328606216874373,
      "grad_norm": 1.151848316192627,
      "learning_rate": 0.00017280278011569847,
      "loss": 1.5255,
      "step": 1838
    },
    {
      "epoch": 0.26342930812204557,
      "grad_norm": 1.246985673904419,
      "learning_rate": 0.00017277096440392717,
      "loss": 1.377,
      "step": 1839
    },
    {
      "epoch": 0.26357255407534735,
      "grad_norm": 1.1437475681304932,
      "learning_rate": 0.00017273913302636798,
      "loss": 1.5936,
      "step": 1840
    },
    {
      "epoch": 0.2637158000286492,
      "grad_norm": 1.1832574605941772,
      "learning_rate": 0.00017270728598987352,
      "loss": 1.5206,
      "step": 1841
    },
    {
      "epoch": 0.26385904598195103,
      "grad_norm": 1.4420394897460938,
      "learning_rate": 0.00017267542330129957,
      "loss": 1.4458,
      "step": 1842
    },
    {
      "epoch": 0.2640022919352528,
      "grad_norm": 1.001261830329895,
      "learning_rate": 0.00017264354496750544,
      "loss": 1.4021,
      "step": 1843
    },
    {
      "epoch": 0.26414553788855466,
      "grad_norm": 1.3838437795639038,
      "learning_rate": 0.0001726116509953537,
      "loss": 1.6433,
      "step": 1844
    },
    {
      "epoch": 0.26428878384185645,
      "grad_norm": 1.4447978734970093,
      "learning_rate": 0.0001725797413917104,
      "loss": 1.2929,
      "step": 1845
    },
    {
      "epoch": 0.2644320297951583,
      "grad_norm": 1.1281362771987915,
      "learning_rate": 0.00017254781616344485,
      "loss": 1.4892,
      "step": 1846
    },
    {
      "epoch": 0.2645752757484601,
      "grad_norm": 1.043124794960022,
      "learning_rate": 0.00017251587531742971,
      "loss": 1.6243,
      "step": 1847
    },
    {
      "epoch": 0.2647185217017619,
      "grad_norm": 1.2842302322387695,
      "learning_rate": 0.00017248391886054114,
      "loss": 1.5771,
      "step": 1848
    },
    {
      "epoch": 0.26486176765506375,
      "grad_norm": 1.451529622077942,
      "learning_rate": 0.00017245194679965854,
      "loss": 1.4743,
      "step": 1849
    },
    {
      "epoch": 0.26500501360836554,
      "grad_norm": 1.3679934740066528,
      "learning_rate": 0.00017241995914166468,
      "loss": 1.4831,
      "step": 1850
    },
    {
      "epoch": 0.2651482595616674,
      "grad_norm": 0.9345340132713318,
      "learning_rate": 0.00017238795589344575,
      "loss": 1.7174,
      "step": 1851
    },
    {
      "epoch": 0.2652915055149692,
      "grad_norm": 1.5027233362197876,
      "learning_rate": 0.00017235593706189123,
      "loss": 1.3468,
      "step": 1852
    },
    {
      "epoch": 0.265434751468271,
      "grad_norm": 0.987837553024292,
      "learning_rate": 0.000172323902653894,
      "loss": 1.6088,
      "step": 1853
    },
    {
      "epoch": 0.26557799742157284,
      "grad_norm": 1.037922978401184,
      "learning_rate": 0.00017229185267635027,
      "loss": 1.5212,
      "step": 1854
    },
    {
      "epoch": 0.2657212433748747,
      "grad_norm": 1.0385154485702515,
      "learning_rate": 0.00017225978713615958,
      "loss": 1.5651,
      "step": 1855
    },
    {
      "epoch": 0.26586448932817647,
      "grad_norm": 1.1097460985183716,
      "learning_rate": 0.0001722277060402249,
      "loss": 1.6475,
      "step": 1856
    },
    {
      "epoch": 0.2660077352814783,
      "grad_norm": 1.4183669090270996,
      "learning_rate": 0.00017219560939545246,
      "loss": 1.4513,
      "step": 1857
    },
    {
      "epoch": 0.2661509812347801,
      "grad_norm": 1.0849918127059937,
      "learning_rate": 0.0001721634972087519,
      "loss": 1.5405,
      "step": 1858
    },
    {
      "epoch": 0.26629422718808193,
      "grad_norm": 0.98914635181427,
      "learning_rate": 0.0001721313694870361,
      "loss": 1.556,
      "step": 1859
    },
    {
      "epoch": 0.2664374731413838,
      "grad_norm": 1.02264404296875,
      "learning_rate": 0.00017209922623722147,
      "loss": 1.4175,
      "step": 1860
    },
    {
      "epoch": 0.26658071909468556,
      "grad_norm": 1.0332555770874023,
      "learning_rate": 0.0001720670674662276,
      "loss": 1.5259,
      "step": 1861
    },
    {
      "epoch": 0.2667239650479874,
      "grad_norm": 1.1522632837295532,
      "learning_rate": 0.00017203489318097753,
      "loss": 1.54,
      "step": 1862
    },
    {
      "epoch": 0.26686721100128924,
      "grad_norm": 1.1492314338684082,
      "learning_rate": 0.00017200270338839748,
      "loss": 1.5314,
      "step": 1863
    },
    {
      "epoch": 0.267010456954591,
      "grad_norm": 1.0583293437957764,
      "learning_rate": 0.00017197049809541717,
      "loss": 1.4963,
      "step": 1864
    },
    {
      "epoch": 0.26715370290789286,
      "grad_norm": 1.292770266532898,
      "learning_rate": 0.00017193827730896968,
      "loss": 1.567,
      "step": 1865
    },
    {
      "epoch": 0.26729694886119465,
      "grad_norm": 1.2695982456207275,
      "learning_rate": 0.00017190604103599127,
      "loss": 1.4201,
      "step": 1866
    },
    {
      "epoch": 0.2674401948144965,
      "grad_norm": 1.3297650814056396,
      "learning_rate": 0.0001718737892834216,
      "loss": 1.4772,
      "step": 1867
    },
    {
      "epoch": 0.26758344076779833,
      "grad_norm": 1.0892633199691772,
      "learning_rate": 0.00017184152205820368,
      "loss": 1.5657,
      "step": 1868
    },
    {
      "epoch": 0.2677266867211001,
      "grad_norm": 1.2263191938400269,
      "learning_rate": 0.00017180923936728387,
      "loss": 1.4843,
      "step": 1869
    },
    {
      "epoch": 0.26786993267440196,
      "grad_norm": 1.264400839805603,
      "learning_rate": 0.00017177694121761188,
      "loss": 1.5874,
      "step": 1870
    },
    {
      "epoch": 0.26801317862770374,
      "grad_norm": 1.2720884084701538,
      "learning_rate": 0.00017174462761614058,
      "loss": 1.4386,
      "step": 1871
    },
    {
      "epoch": 0.2681564245810056,
      "grad_norm": 1.2213882207870483,
      "learning_rate": 0.00017171229856982639,
      "loss": 1.3395,
      "step": 1872
    },
    {
      "epoch": 0.2682996705343074,
      "grad_norm": 1.2335898876190186,
      "learning_rate": 0.0001716799540856289,
      "loss": 1.2524,
      "step": 1873
    },
    {
      "epoch": 0.2684429164876092,
      "grad_norm": 1.3253637552261353,
      "learning_rate": 0.00017164759417051114,
      "loss": 1.4424,
      "step": 1874
    },
    {
      "epoch": 0.26858616244091105,
      "grad_norm": 1.155497431755066,
      "learning_rate": 0.00017161521883143934,
      "loss": 1.5293,
      "step": 1875
    },
    {
      "epoch": 0.2687294083942129,
      "grad_norm": 1.0171724557876587,
      "learning_rate": 0.00017158282807538312,
      "loss": 1.5982,
      "step": 1876
    },
    {
      "epoch": 0.2688726543475147,
      "grad_norm": 1.1201980113983154,
      "learning_rate": 0.00017155042190931542,
      "loss": 1.4978,
      "step": 1877
    },
    {
      "epoch": 0.2690159003008165,
      "grad_norm": 1.346705436706543,
      "learning_rate": 0.00017151800034021248,
      "loss": 1.6491,
      "step": 1878
    },
    {
      "epoch": 0.2691591462541183,
      "grad_norm": 1.1991596221923828,
      "learning_rate": 0.00017148556337505386,
      "loss": 1.4552,
      "step": 1879
    },
    {
      "epoch": 0.26930239220742014,
      "grad_norm": 1.1959863901138306,
      "learning_rate": 0.00017145311102082243,
      "loss": 1.4665,
      "step": 1880
    },
    {
      "epoch": 0.269445638160722,
      "grad_norm": 1.2051140069961548,
      "learning_rate": 0.00017142064328450442,
      "loss": 1.564,
      "step": 1881
    },
    {
      "epoch": 0.26958888411402376,
      "grad_norm": 1.2838658094406128,
      "learning_rate": 0.00017138816017308925,
      "loss": 1.392,
      "step": 1882
    },
    {
      "epoch": 0.2697321300673256,
      "grad_norm": 1.3614393472671509,
      "learning_rate": 0.00017135566169356983,
      "loss": 1.4089,
      "step": 1883
    },
    {
      "epoch": 0.26987537602062744,
      "grad_norm": 1.3757820129394531,
      "learning_rate": 0.00017132314785294218,
      "loss": 1.3859,
      "step": 1884
    },
    {
      "epoch": 0.27001862197392923,
      "grad_norm": 1.2666202783584595,
      "learning_rate": 0.00017129061865820582,
      "loss": 1.6483,
      "step": 1885
    },
    {
      "epoch": 0.27016186792723107,
      "grad_norm": 1.3503568172454834,
      "learning_rate": 0.0001712580741163634,
      "loss": 1.4915,
      "step": 1886
    },
    {
      "epoch": 0.27030511388053285,
      "grad_norm": 1.2183852195739746,
      "learning_rate": 0.00017122551423442097,
      "loss": 1.5049,
      "step": 1887
    },
    {
      "epoch": 0.2704483598338347,
      "grad_norm": 0.9913017749786377,
      "learning_rate": 0.0001711929390193879,
      "loss": 1.2689,
      "step": 1888
    },
    {
      "epoch": 0.27059160578713654,
      "grad_norm": 1.223230004310608,
      "learning_rate": 0.0001711603484782768,
      "loss": 1.3459,
      "step": 1889
    },
    {
      "epoch": 0.2707348517404383,
      "grad_norm": 1.1682639122009277,
      "learning_rate": 0.00017112774261810365,
      "loss": 1.4451,
      "step": 1890
    },
    {
      "epoch": 0.27087809769374016,
      "grad_norm": 1.2320588827133179,
      "learning_rate": 0.00017109512144588762,
      "loss": 1.471,
      "step": 1891
    },
    {
      "epoch": 0.27102134364704195,
      "grad_norm": 1.2000607252120972,
      "learning_rate": 0.0001710624849686513,
      "loss": 1.4219,
      "step": 1892
    },
    {
      "epoch": 0.2711645896003438,
      "grad_norm": 1.3722758293151855,
      "learning_rate": 0.00017102983319342046,
      "loss": 1.5727,
      "step": 1893
    },
    {
      "epoch": 0.2713078355536456,
      "grad_norm": 1.1525709629058838,
      "learning_rate": 0.00017099716612722427,
      "loss": 1.6695,
      "step": 1894
    },
    {
      "epoch": 0.2714510815069474,
      "grad_norm": 1.111646056175232,
      "learning_rate": 0.00017096448377709514,
      "loss": 1.5913,
      "step": 1895
    },
    {
      "epoch": 0.27159432746024925,
      "grad_norm": 1.3374725580215454,
      "learning_rate": 0.00017093178615006872,
      "loss": 1.4504,
      "step": 1896
    },
    {
      "epoch": 0.2717375734135511,
      "grad_norm": 1.209067702293396,
      "learning_rate": 0.00017089907325318403,
      "loss": 1.4779,
      "step": 1897
    },
    {
      "epoch": 0.2718808193668529,
      "grad_norm": 1.0878421068191528,
      "learning_rate": 0.00017086634509348337,
      "loss": 1.3987,
      "step": 1898
    },
    {
      "epoch": 0.2720240653201547,
      "grad_norm": 1.3177785873413086,
      "learning_rate": 0.00017083360167801225,
      "loss": 1.6204,
      "step": 1899
    },
    {
      "epoch": 0.2721673112734565,
      "grad_norm": 1.2348480224609375,
      "learning_rate": 0.00017080084301381956,
      "loss": 1.4488,
      "step": 1900
    },
    {
      "epoch": 0.27231055722675834,
      "grad_norm": 1.1994290351867676,
      "learning_rate": 0.0001707680691079574,
      "loss": 1.4117,
      "step": 1901
    },
    {
      "epoch": 0.2724538031800602,
      "grad_norm": 1.2006165981292725,
      "learning_rate": 0.00017073527996748123,
      "loss": 1.442,
      "step": 1902
    },
    {
      "epoch": 0.27259704913336197,
      "grad_norm": 1.205885887145996,
      "learning_rate": 0.0001707024755994497,
      "loss": 1.5712,
      "step": 1903
    },
    {
      "epoch": 0.2727402950866638,
      "grad_norm": 1.3983968496322632,
      "learning_rate": 0.00017066965601092478,
      "loss": 1.4837,
      "step": 1904
    },
    {
      "epoch": 0.2728835410399656,
      "grad_norm": 1.0385433435440063,
      "learning_rate": 0.00017063682120897175,
      "loss": 1.6284,
      "step": 1905
    },
    {
      "epoch": 0.27302678699326743,
      "grad_norm": 1.2428748607635498,
      "learning_rate": 0.00017060397120065909,
      "loss": 1.5576,
      "step": 1906
    },
    {
      "epoch": 0.2731700329465693,
      "grad_norm": 1.2136071920394897,
      "learning_rate": 0.0001705711059930586,
      "loss": 1.495,
      "step": 1907
    },
    {
      "epoch": 0.27331327889987106,
      "grad_norm": 1.1357100009918213,
      "learning_rate": 0.00017053822559324536,
      "loss": 1.5327,
      "step": 1908
    },
    {
      "epoch": 0.2734565248531729,
      "grad_norm": 1.517285943031311,
      "learning_rate": 0.0001705053300082977,
      "loss": 1.4914,
      "step": 1909
    },
    {
      "epoch": 0.27359977080647474,
      "grad_norm": 1.0948734283447266,
      "learning_rate": 0.00017047241924529721,
      "loss": 1.559,
      "step": 1910
    },
    {
      "epoch": 0.2737430167597765,
      "grad_norm": 1.1471765041351318,
      "learning_rate": 0.00017043949331132878,
      "loss": 1.4555,
      "step": 1911
    },
    {
      "epoch": 0.27388626271307837,
      "grad_norm": 1.1361476182937622,
      "learning_rate": 0.00017040655221348057,
      "loss": 1.5873,
      "step": 1912
    },
    {
      "epoch": 0.27402950866638015,
      "grad_norm": 1.1316759586334229,
      "learning_rate": 0.00017037359595884394,
      "loss": 1.4994,
      "step": 1913
    },
    {
      "epoch": 0.274172754619682,
      "grad_norm": 1.2050585746765137,
      "learning_rate": 0.0001703406245545136,
      "loss": 1.6988,
      "step": 1914
    },
    {
      "epoch": 0.27431600057298383,
      "grad_norm": 1.2752808332443237,
      "learning_rate": 0.00017030763800758743,
      "loss": 1.4632,
      "step": 1915
    },
    {
      "epoch": 0.2744592465262856,
      "grad_norm": 1.11107337474823,
      "learning_rate": 0.00017027463632516665,
      "loss": 1.5772,
      "step": 1916
    },
    {
      "epoch": 0.27460249247958746,
      "grad_norm": 1.1216106414794922,
      "learning_rate": 0.00017024161951435568,
      "loss": 1.2876,
      "step": 1917
    },
    {
      "epoch": 0.2747457384328893,
      "grad_norm": 1.3822041749954224,
      "learning_rate": 0.00017020858758226229,
      "loss": 1.5395,
      "step": 1918
    },
    {
      "epoch": 0.2748889843861911,
      "grad_norm": 0.9759470224380493,
      "learning_rate": 0.0001701755405359973,
      "loss": 1.5713,
      "step": 1919
    },
    {
      "epoch": 0.2750322303394929,
      "grad_norm": 1.0533623695373535,
      "learning_rate": 0.00017014247838267508,
      "loss": 1.6032,
      "step": 1920
    },
    {
      "epoch": 0.2751754762927947,
      "grad_norm": 1.5107101202011108,
      "learning_rate": 0.000170109401129413,
      "loss": 1.4686,
      "step": 1921
    },
    {
      "epoch": 0.27531872224609655,
      "grad_norm": 1.0671602487564087,
      "learning_rate": 0.0001700763087833318,
      "loss": 1.5392,
      "step": 1922
    },
    {
      "epoch": 0.2754619681993984,
      "grad_norm": 1.1020092964172363,
      "learning_rate": 0.00017004320135155542,
      "loss": 1.6509,
      "step": 1923
    },
    {
      "epoch": 0.2756052141527002,
      "grad_norm": 1.03842294216156,
      "learning_rate": 0.0001700100788412111,
      "loss": 1.5069,
      "step": 1924
    },
    {
      "epoch": 0.275748460106002,
      "grad_norm": 1.1630704402923584,
      "learning_rate": 0.0001699769412594293,
      "loss": 1.3128,
      "step": 1925
    },
    {
      "epoch": 0.2758917060593038,
      "grad_norm": 1.168697476387024,
      "learning_rate": 0.0001699437886133437,
      "loss": 1.3766,
      "step": 1926
    },
    {
      "epoch": 0.27603495201260564,
      "grad_norm": 1.1691465377807617,
      "learning_rate": 0.00016991062091009122,
      "loss": 1.4515,
      "step": 1927
    },
    {
      "epoch": 0.2761781979659075,
      "grad_norm": 1.1161352396011353,
      "learning_rate": 0.0001698774381568121,
      "loss": 1.4423,
      "step": 1928
    },
    {
      "epoch": 0.27632144391920926,
      "grad_norm": 1.354994535446167,
      "learning_rate": 0.00016984424036064975,
      "loss": 1.5517,
      "step": 1929
    },
    {
      "epoch": 0.2764646898725111,
      "grad_norm": 1.1025282144546509,
      "learning_rate": 0.0001698110275287508,
      "loss": 1.6438,
      "step": 1930
    },
    {
      "epoch": 0.27660793582581295,
      "grad_norm": 1.0699341297149658,
      "learning_rate": 0.00016977779966826522,
      "loss": 1.3712,
      "step": 1931
    },
    {
      "epoch": 0.27675118177911473,
      "grad_norm": 1.0224467515945435,
      "learning_rate": 0.00016974455678634608,
      "loss": 1.5837,
      "step": 1932
    },
    {
      "epoch": 0.27689442773241657,
      "grad_norm": 1.265955924987793,
      "learning_rate": 0.0001697112988901498,
      "loss": 1.5958,
      "step": 1933
    },
    {
      "epoch": 0.27703767368571836,
      "grad_norm": 0.9183480143547058,
      "learning_rate": 0.00016967802598683593,
      "loss": 1.4942,
      "step": 1934
    },
    {
      "epoch": 0.2771809196390202,
      "grad_norm": 1.1272201538085938,
      "learning_rate": 0.00016964473808356735,
      "loss": 1.6687,
      "step": 1935
    },
    {
      "epoch": 0.27732416559232204,
      "grad_norm": 0.9888427257537842,
      "learning_rate": 0.00016961143518751004,
      "loss": 1.6476,
      "step": 1936
    },
    {
      "epoch": 0.2774674115456238,
      "grad_norm": 1.028112530708313,
      "learning_rate": 0.0001695781173058334,
      "loss": 1.4354,
      "step": 1937
    },
    {
      "epoch": 0.27761065749892566,
      "grad_norm": 1.298352837562561,
      "learning_rate": 0.0001695447844457099,
      "loss": 1.4704,
      "step": 1938
    },
    {
      "epoch": 0.2777539034522275,
      "grad_norm": 1.4349368810653687,
      "learning_rate": 0.00016951143661431524,
      "loss": 1.3916,
      "step": 1939
    },
    {
      "epoch": 0.2778971494055293,
      "grad_norm": 1.2475242614746094,
      "learning_rate": 0.00016947807381882844,
      "loss": 1.4985,
      "step": 1940
    },
    {
      "epoch": 0.2780403953588311,
      "grad_norm": 1.2409940958023071,
      "learning_rate": 0.00016944469606643167,
      "loss": 1.5647,
      "step": 1941
    },
    {
      "epoch": 0.2781836413121329,
      "grad_norm": 1.1128641366958618,
      "learning_rate": 0.00016941130336431032,
      "loss": 1.5453,
      "step": 1942
    },
    {
      "epoch": 0.27832688726543475,
      "grad_norm": 1.2181445360183716,
      "learning_rate": 0.00016937789571965303,
      "loss": 1.4346,
      "step": 1943
    },
    {
      "epoch": 0.2784701332187366,
      "grad_norm": 1.1452646255493164,
      "learning_rate": 0.0001693444731396516,
      "loss": 1.5195,
      "step": 1944
    },
    {
      "epoch": 0.2786133791720384,
      "grad_norm": 1.1039472818374634,
      "learning_rate": 0.00016931103563150112,
      "loss": 1.5302,
      "step": 1945
    },
    {
      "epoch": 0.2787566251253402,
      "grad_norm": 1.3316044807434082,
      "learning_rate": 0.00016927758320239988,
      "loss": 1.2911,
      "step": 1946
    },
    {
      "epoch": 0.278899871078642,
      "grad_norm": 1.1934200525283813,
      "learning_rate": 0.00016924411585954928,
      "loss": 1.4202,
      "step": 1947
    },
    {
      "epoch": 0.27904311703194384,
      "grad_norm": 1.0135799646377563,
      "learning_rate": 0.00016921063361015413,
      "loss": 1.5612,
      "step": 1948
    },
    {
      "epoch": 0.2791863629852457,
      "grad_norm": 0.9171785116195679,
      "learning_rate": 0.00016917713646142222,
      "loss": 1.5431,
      "step": 1949
    },
    {
      "epoch": 0.27932960893854747,
      "grad_norm": 1.1183679103851318,
      "learning_rate": 0.0001691436244205647,
      "loss": 1.5461,
      "step": 1950
    },
    {
      "epoch": 0.2794728548918493,
      "grad_norm": 1.3267743587493896,
      "learning_rate": 0.00016911009749479586,
      "loss": 1.5698,
      "step": 1951
    },
    {
      "epoch": 0.27961610084515115,
      "grad_norm": 1.3369734287261963,
      "learning_rate": 0.00016907655569133327,
      "loss": 1.437,
      "step": 1952
    },
    {
      "epoch": 0.27975934679845293,
      "grad_norm": 1.0967153310775757,
      "learning_rate": 0.0001690429990173976,
      "loss": 1.587,
      "step": 1953
    },
    {
      "epoch": 0.2799025927517548,
      "grad_norm": 1.1468194723129272,
      "learning_rate": 0.0001690094274802128,
      "loss": 1.5188,
      "step": 1954
    },
    {
      "epoch": 0.28004583870505656,
      "grad_norm": 1.0509079694747925,
      "learning_rate": 0.00016897584108700598,
      "loss": 1.6605,
      "step": 1955
    },
    {
      "epoch": 0.2801890846583584,
      "grad_norm": 1.168953776359558,
      "learning_rate": 0.00016894223984500746,
      "loss": 1.5138,
      "step": 1956
    },
    {
      "epoch": 0.28033233061166024,
      "grad_norm": 0.9107540249824524,
      "learning_rate": 0.00016890862376145075,
      "loss": 1.5842,
      "step": 1957
    },
    {
      "epoch": 0.280475576564962,
      "grad_norm": 0.8969243168830872,
      "learning_rate": 0.00016887499284357258,
      "loss": 1.5129,
      "step": 1958
    },
    {
      "epoch": 0.28061882251826387,
      "grad_norm": 1.2624499797821045,
      "learning_rate": 0.00016884134709861285,
      "loss": 1.4811,
      "step": 1959
    },
    {
      "epoch": 0.28076206847156565,
      "grad_norm": 1.0120142698287964,
      "learning_rate": 0.00016880768653381462,
      "loss": 1.4556,
      "step": 1960
    },
    {
      "epoch": 0.2809053144248675,
      "grad_norm": 1.2255308628082275,
      "learning_rate": 0.0001687740111564242,
      "loss": 1.4382,
      "step": 1961
    },
    {
      "epoch": 0.28104856037816933,
      "grad_norm": 1.2514064311981201,
      "learning_rate": 0.00016874032097369113,
      "loss": 1.4465,
      "step": 1962
    },
    {
      "epoch": 0.2811918063314711,
      "grad_norm": 1.1582660675048828,
      "learning_rate": 0.00016870661599286798,
      "loss": 1.3946,
      "step": 1963
    },
    {
      "epoch": 0.28133505228477296,
      "grad_norm": 1.0066133737564087,
      "learning_rate": 0.0001686728962212106,
      "loss": 1.5363,
      "step": 1964
    },
    {
      "epoch": 0.2814782982380748,
      "grad_norm": 1.346342921257019,
      "learning_rate": 0.0001686391616659781,
      "loss": 1.5577,
      "step": 1965
    },
    {
      "epoch": 0.2816215441913766,
      "grad_norm": 0.9517983198165894,
      "learning_rate": 0.00016860541233443263,
      "loss": 1.4439,
      "step": 1966
    },
    {
      "epoch": 0.2817647901446784,
      "grad_norm": 1.2255555391311646,
      "learning_rate": 0.0001685716482338396,
      "loss": 1.5787,
      "step": 1967
    },
    {
      "epoch": 0.2819080360979802,
      "grad_norm": 1.2086429595947266,
      "learning_rate": 0.0001685378693714676,
      "loss": 1.6298,
      "step": 1968
    },
    {
      "epoch": 0.28205128205128205,
      "grad_norm": 0.9574936628341675,
      "learning_rate": 0.00016850407575458838,
      "loss": 1.5551,
      "step": 1969
    },
    {
      "epoch": 0.2821945280045839,
      "grad_norm": 1.1391602754592896,
      "learning_rate": 0.00016847026739047683,
      "loss": 1.4595,
      "step": 1970
    },
    {
      "epoch": 0.2823377739578857,
      "grad_norm": 1.208785891532898,
      "learning_rate": 0.00016843644428641113,
      "loss": 1.4375,
      "step": 1971
    },
    {
      "epoch": 0.2824810199111875,
      "grad_norm": 1.231567621231079,
      "learning_rate": 0.00016840260644967247,
      "loss": 1.542,
      "step": 1972
    },
    {
      "epoch": 0.28262426586448935,
      "grad_norm": 0.9703326225280762,
      "learning_rate": 0.00016836875388754537,
      "loss": 1.4323,
      "step": 1973
    },
    {
      "epoch": 0.28276751181779114,
      "grad_norm": 1.1931341886520386,
      "learning_rate": 0.00016833488660731742,
      "loss": 1.4877,
      "step": 1974
    },
    {
      "epoch": 0.282910757771093,
      "grad_norm": 1.4200971126556396,
      "learning_rate": 0.0001683010046162794,
      "loss": 1.5347,
      "step": 1975
    },
    {
      "epoch": 0.28305400372439476,
      "grad_norm": 1.2097086906433105,
      "learning_rate": 0.0001682671079217253,
      "loss": 1.3702,
      "step": 1976
    },
    {
      "epoch": 0.2831972496776966,
      "grad_norm": 1.409982442855835,
      "learning_rate": 0.00016823319653095218,
      "loss": 1.4491,
      "step": 1977
    },
    {
      "epoch": 0.28334049563099845,
      "grad_norm": 1.1274410486221313,
      "learning_rate": 0.00016819927045126035,
      "loss": 1.4181,
      "step": 1978
    },
    {
      "epoch": 0.28348374158430023,
      "grad_norm": 1.1649516820907593,
      "learning_rate": 0.00016816532968995328,
      "loss": 1.4905,
      "step": 1979
    },
    {
      "epoch": 0.28362698753760207,
      "grad_norm": 1.1671640872955322,
      "learning_rate": 0.00016813137425433758,
      "loss": 1.471,
      "step": 1980
    },
    {
      "epoch": 0.28377023349090386,
      "grad_norm": 1.5519614219665527,
      "learning_rate": 0.00016809740415172297,
      "loss": 1.5533,
      "step": 1981
    },
    {
      "epoch": 0.2839134794442057,
      "grad_norm": 1.2579014301300049,
      "learning_rate": 0.0001680634193894224,
      "loss": 1.3766,
      "step": 1982
    },
    {
      "epoch": 0.28405672539750754,
      "grad_norm": 1.148878812789917,
      "learning_rate": 0.00016802941997475196,
      "loss": 1.3156,
      "step": 1983
    },
    {
      "epoch": 0.2841999713508093,
      "grad_norm": 1.2862945795059204,
      "learning_rate": 0.00016799540591503083,
      "loss": 1.6057,
      "step": 1984
    },
    {
      "epoch": 0.28434321730411116,
      "grad_norm": 1.0098503828048706,
      "learning_rate": 0.00016796137721758142,
      "loss": 1.5406,
      "step": 1985
    },
    {
      "epoch": 0.284486463257413,
      "grad_norm": 1.2319185733795166,
      "learning_rate": 0.00016792733388972932,
      "loss": 1.5553,
      "step": 1986
    },
    {
      "epoch": 0.2846297092107148,
      "grad_norm": 1.1229959726333618,
      "learning_rate": 0.00016789327593880318,
      "loss": 1.511,
      "step": 1987
    },
    {
      "epoch": 0.28477295516401663,
      "grad_norm": 1.0639524459838867,
      "learning_rate": 0.0001678592033721348,
      "loss": 1.4794,
      "step": 1988
    },
    {
      "epoch": 0.2849162011173184,
      "grad_norm": 1.2658641338348389,
      "learning_rate": 0.0001678251161970592,
      "loss": 1.6231,
      "step": 1989
    },
    {
      "epoch": 0.28505944707062025,
      "grad_norm": 1.185575246810913,
      "learning_rate": 0.00016779101442091447,
      "loss": 1.5016,
      "step": 1990
    },
    {
      "epoch": 0.2852026930239221,
      "grad_norm": 1.1228439807891846,
      "learning_rate": 0.00016775689805104192,
      "loss": 1.5584,
      "step": 1991
    },
    {
      "epoch": 0.2853459389772239,
      "grad_norm": 1.2919049263000488,
      "learning_rate": 0.00016772276709478597,
      "loss": 1.3909,
      "step": 1992
    },
    {
      "epoch": 0.2854891849305257,
      "grad_norm": 0.9523124694824219,
      "learning_rate": 0.00016768862155949413,
      "loss": 1.3926,
      "step": 1993
    },
    {
      "epoch": 0.28563243088382756,
      "grad_norm": 1.056960940361023,
      "learning_rate": 0.00016765446145251706,
      "loss": 1.6344,
      "step": 1994
    },
    {
      "epoch": 0.28577567683712934,
      "grad_norm": 1.2078131437301636,
      "learning_rate": 0.00016762028678120867,
      "loss": 1.6735,
      "step": 1995
    },
    {
      "epoch": 0.2859189227904312,
      "grad_norm": 1.0921590328216553,
      "learning_rate": 0.00016758609755292584,
      "loss": 1.729,
      "step": 1996
    },
    {
      "epoch": 0.28606216874373297,
      "grad_norm": 1.2056080102920532,
      "learning_rate": 0.00016755189377502876,
      "loss": 1.393,
      "step": 1997
    },
    {
      "epoch": 0.2862054146970348,
      "grad_norm": 1.1325101852416992,
      "learning_rate": 0.00016751767545488056,
      "loss": 1.4909,
      "step": 1998
    },
    {
      "epoch": 0.28634866065033665,
      "grad_norm": 1.0890710353851318,
      "learning_rate": 0.00016748344259984762,
      "loss": 1.5392,
      "step": 1999
    },
    {
      "epoch": 0.28649190660363844,
      "grad_norm": 1.3951345682144165,
      "learning_rate": 0.00016744919521729948,
      "loss": 1.4531,
      "step": 2000
    },
    {
      "epoch": 0.2866351525569403,
      "grad_norm": 1.0557584762573242,
      "learning_rate": 0.0001674149333146087,
      "loss": 1.4296,
      "step": 2001
    },
    {
      "epoch": 0.28677839851024206,
      "grad_norm": 1.0310229063034058,
      "learning_rate": 0.00016738065689915106,
      "loss": 1.3732,
      "step": 2002
    },
    {
      "epoch": 0.2869216444635439,
      "grad_norm": 1.0658046007156372,
      "learning_rate": 0.0001673463659783054,
      "loss": 1.3561,
      "step": 2003
    },
    {
      "epoch": 0.28706489041684574,
      "grad_norm": 1.177260160446167,
      "learning_rate": 0.00016731206055945366,
      "loss": 1.3102,
      "step": 2004
    },
    {
      "epoch": 0.2872081363701475,
      "grad_norm": 0.9633142948150635,
      "learning_rate": 0.00016727774064998106,
      "loss": 1.5386,
      "step": 2005
    },
    {
      "epoch": 0.28735138232344937,
      "grad_norm": 1.1098424196243286,
      "learning_rate": 0.00016724340625727573,
      "loss": 1.4217,
      "step": 2006
    },
    {
      "epoch": 0.2874946282767512,
      "grad_norm": 0.9937300086021423,
      "learning_rate": 0.00016720905738872905,
      "loss": 1.3578,
      "step": 2007
    },
    {
      "epoch": 0.287637874230053,
      "grad_norm": 1.161625862121582,
      "learning_rate": 0.00016717469405173549,
      "loss": 1.3959,
      "step": 2008
    },
    {
      "epoch": 0.28778112018335483,
      "grad_norm": 0.9400433897972107,
      "learning_rate": 0.00016714031625369264,
      "loss": 1.6437,
      "step": 2009
    },
    {
      "epoch": 0.2879243661366566,
      "grad_norm": 1.4957982301712036,
      "learning_rate": 0.0001671059240020011,
      "loss": 1.3158,
      "step": 2010
    },
    {
      "epoch": 0.28806761208995846,
      "grad_norm": 1.1123487949371338,
      "learning_rate": 0.00016707151730406482,
      "loss": 1.717,
      "step": 2011
    },
    {
      "epoch": 0.2882108580432603,
      "grad_norm": 1.150346040725708,
      "learning_rate": 0.00016703709616729055,
      "loss": 1.4951,
      "step": 2012
    },
    {
      "epoch": 0.2883541039965621,
      "grad_norm": 1.0847374200820923,
      "learning_rate": 0.00016700266059908842,
      "loss": 1.4901,
      "step": 2013
    },
    {
      "epoch": 0.2884973499498639,
      "grad_norm": 0.90064936876297,
      "learning_rate": 0.00016696821060687155,
      "loss": 1.5383,
      "step": 2014
    },
    {
      "epoch": 0.2886405959031657,
      "grad_norm": 1.0576918125152588,
      "learning_rate": 0.00016693374619805606,
      "loss": 1.4958,
      "step": 2015
    },
    {
      "epoch": 0.28878384185646755,
      "grad_norm": 1.3058196306228638,
      "learning_rate": 0.00016689926738006143,
      "loss": 1.5537,
      "step": 2016
    },
    {
      "epoch": 0.2889270878097694,
      "grad_norm": 1.2564175128936768,
      "learning_rate": 0.00016686477416031004,
      "loss": 1.5773,
      "step": 2017
    },
    {
      "epoch": 0.2890703337630712,
      "grad_norm": 1.1657989025115967,
      "learning_rate": 0.0001668302665462274,
      "loss": 1.5286,
      "step": 2018
    },
    {
      "epoch": 0.289213579716373,
      "grad_norm": 1.1100525856018066,
      "learning_rate": 0.00016679574454524213,
      "loss": 1.5379,
      "step": 2019
    },
    {
      "epoch": 0.28935682566967486,
      "grad_norm": 1.007617473602295,
      "learning_rate": 0.00016676120816478605,
      "loss": 1.5477,
      "step": 2020
    },
    {
      "epoch": 0.28950007162297664,
      "grad_norm": 0.9686825275421143,
      "learning_rate": 0.00016672665741229392,
      "loss": 1.6948,
      "step": 2021
    },
    {
      "epoch": 0.2896433175762785,
      "grad_norm": 1.1076608896255493,
      "learning_rate": 0.00016669209229520367,
      "loss": 1.4572,
      "step": 2022
    },
    {
      "epoch": 0.28978656352958027,
      "grad_norm": 0.9844187498092651,
      "learning_rate": 0.00016665751282095634,
      "loss": 1.3964,
      "step": 2023
    },
    {
      "epoch": 0.2899298094828821,
      "grad_norm": 1.141664981842041,
      "learning_rate": 0.000166622918996996,
      "loss": 1.5967,
      "step": 2024
    },
    {
      "epoch": 0.29007305543618395,
      "grad_norm": 1.1901179552078247,
      "learning_rate": 0.00016658831083076985,
      "loss": 1.5506,
      "step": 2025
    },
    {
      "epoch": 0.29021630138948573,
      "grad_norm": 0.9641210436820984,
      "learning_rate": 0.0001665536883297282,
      "loss": 1.5155,
      "step": 2026
    },
    {
      "epoch": 0.29035954734278757,
      "grad_norm": 1.2455899715423584,
      "learning_rate": 0.0001665190515013244,
      "loss": 1.3626,
      "step": 2027
    },
    {
      "epoch": 0.2905027932960894,
      "grad_norm": 1.0462678670883179,
      "learning_rate": 0.0001664844003530149,
      "loss": 1.55,
      "step": 2028
    },
    {
      "epoch": 0.2906460392493912,
      "grad_norm": 1.0634565353393555,
      "learning_rate": 0.00016644973489225926,
      "loss": 1.5275,
      "step": 2029
    },
    {
      "epoch": 0.29078928520269304,
      "grad_norm": 0.9982190728187561,
      "learning_rate": 0.00016641505512652005,
      "loss": 1.5624,
      "step": 2030
    },
    {
      "epoch": 0.2909325311559948,
      "grad_norm": 1.0724668502807617,
      "learning_rate": 0.00016638036106326296,
      "loss": 1.2859,
      "step": 2031
    },
    {
      "epoch": 0.29107577710929666,
      "grad_norm": 1.1475642919540405,
      "learning_rate": 0.00016634565270995684,
      "loss": 1.3716,
      "step": 2032
    },
    {
      "epoch": 0.2912190230625985,
      "grad_norm": 1.275540828704834,
      "learning_rate": 0.00016631093007407347,
      "loss": 1.4838,
      "step": 2033
    },
    {
      "epoch": 0.2913622690159003,
      "grad_norm": 1.113063931465149,
      "learning_rate": 0.0001662761931630878,
      "loss": 1.463,
      "step": 2034
    },
    {
      "epoch": 0.29150551496920213,
      "grad_norm": 0.8122738599777222,
      "learning_rate": 0.00016624144198447782,
      "loss": 1.5566,
      "step": 2035
    },
    {
      "epoch": 0.2916487609225039,
      "grad_norm": 1.1239324808120728,
      "learning_rate": 0.0001662066765457246,
      "loss": 1.3179,
      "step": 2036
    },
    {
      "epoch": 0.29179200687580575,
      "grad_norm": 1.2529375553131104,
      "learning_rate": 0.00016617189685431228,
      "loss": 1.5301,
      "step": 2037
    },
    {
      "epoch": 0.2919352528291076,
      "grad_norm": 1.1471854448318481,
      "learning_rate": 0.00016613710291772812,
      "loss": 1.4626,
      "step": 2038
    },
    {
      "epoch": 0.2920784987824094,
      "grad_norm": 1.1331422328948975,
      "learning_rate": 0.00016610229474346228,
      "loss": 1.4181,
      "step": 2039
    },
    {
      "epoch": 0.2922217447357112,
      "grad_norm": 1.0694282054901123,
      "learning_rate": 0.00016606747233900815,
      "loss": 1.5129,
      "step": 2040
    },
    {
      "epoch": 0.29236499068901306,
      "grad_norm": 1.1972836256027222,
      "learning_rate": 0.00016603263571186218,
      "loss": 1.5453,
      "step": 2041
    },
    {
      "epoch": 0.29250823664231484,
      "grad_norm": 1.229253888130188,
      "learning_rate": 0.0001659977848695238,
      "loss": 1.4401,
      "step": 2042
    },
    {
      "epoch": 0.2926514825956167,
      "grad_norm": 1.1006466150283813,
      "learning_rate": 0.0001659629198194955,
      "loss": 1.4402,
      "step": 2043
    },
    {
      "epoch": 0.29279472854891847,
      "grad_norm": 1.2078962326049805,
      "learning_rate": 0.00016592804056928285,
      "loss": 1.1869,
      "step": 2044
    },
    {
      "epoch": 0.2929379745022203,
      "grad_norm": 1.0909185409545898,
      "learning_rate": 0.0001658931471263946,
      "loss": 1.5252,
      "step": 2045
    },
    {
      "epoch": 0.29308122045552215,
      "grad_norm": 1.2717727422714233,
      "learning_rate": 0.0001658582394983424,
      "loss": 1.3947,
      "step": 2046
    },
    {
      "epoch": 0.29322446640882394,
      "grad_norm": 1.0644125938415527,
      "learning_rate": 0.0001658233176926409,
      "loss": 1.518,
      "step": 2047
    },
    {
      "epoch": 0.2933677123621258,
      "grad_norm": 1.2992795705795288,
      "learning_rate": 0.000165788381716808,
      "loss": 1.5907,
      "step": 2048
    },
    {
      "epoch": 0.29351095831542756,
      "grad_norm": 1.2054922580718994,
      "learning_rate": 0.00016575343157836455,
      "loss": 1.2582,
      "step": 2049
    },
    {
      "epoch": 0.2936542042687294,
      "grad_norm": 1.2525360584259033,
      "learning_rate": 0.00016571846728483442,
      "loss": 1.4705,
      "step": 2050
    },
    {
      "epoch": 0.29379745022203124,
      "grad_norm": 1.3702844381332397,
      "learning_rate": 0.00016568348884374453,
      "loss": 1.3646,
      "step": 2051
    },
    {
      "epoch": 0.293940696175333,
      "grad_norm": 1.1646143198013306,
      "learning_rate": 0.00016564849626262492,
      "loss": 1.5212,
      "step": 2052
    },
    {
      "epoch": 0.29408394212863487,
      "grad_norm": 1.1041207313537598,
      "learning_rate": 0.00016561348954900865,
      "loss": 1.5302,
      "step": 2053
    },
    {
      "epoch": 0.2942271880819367,
      "grad_norm": 1.0390639305114746,
      "learning_rate": 0.00016557846871043173,
      "loss": 1.3909,
      "step": 2054
    },
    {
      "epoch": 0.2943704340352385,
      "grad_norm": 1.0313633680343628,
      "learning_rate": 0.00016554343375443331,
      "loss": 1.6806,
      "step": 2055
    },
    {
      "epoch": 0.29451367998854033,
      "grad_norm": 1.3112484216690063,
      "learning_rate": 0.00016550838468855553,
      "loss": 1.578,
      "step": 2056
    },
    {
      "epoch": 0.2946569259418421,
      "grad_norm": 1.255183458328247,
      "learning_rate": 0.00016547332152034365,
      "loss": 1.5373,
      "step": 2057
    },
    {
      "epoch": 0.29480017189514396,
      "grad_norm": 1.277254343032837,
      "learning_rate": 0.00016543824425734583,
      "loss": 1.5696,
      "step": 2058
    },
    {
      "epoch": 0.2949434178484458,
      "grad_norm": 1.0825320482254028,
      "learning_rate": 0.00016540315290711338,
      "loss": 1.344,
      "step": 2059
    },
    {
      "epoch": 0.2950866638017476,
      "grad_norm": 1.1773914098739624,
      "learning_rate": 0.0001653680474772006,
      "loss": 1.5834,
      "step": 2060
    },
    {
      "epoch": 0.2952299097550494,
      "grad_norm": 1.3338125944137573,
      "learning_rate": 0.0001653329279751648,
      "loss": 1.3147,
      "step": 2061
    },
    {
      "epoch": 0.29537315570835126,
      "grad_norm": 1.0381402969360352,
      "learning_rate": 0.00016529779440856634,
      "loss": 1.5587,
      "step": 2062
    },
    {
      "epoch": 0.29551640166165305,
      "grad_norm": 1.1725177764892578,
      "learning_rate": 0.0001652626467849686,
      "loss": 1.5401,
      "step": 2063
    },
    {
      "epoch": 0.2956596476149549,
      "grad_norm": 1.2530937194824219,
      "learning_rate": 0.00016522748511193803,
      "loss": 1.4864,
      "step": 2064
    },
    {
      "epoch": 0.2958028935682567,
      "grad_norm": 1.280437707901001,
      "learning_rate": 0.00016519230939704402,
      "loss": 1.5546,
      "step": 2065
    },
    {
      "epoch": 0.2959461395215585,
      "grad_norm": 1.2502906322479248,
      "learning_rate": 0.00016515711964785906,
      "loss": 1.5615,
      "step": 2066
    },
    {
      "epoch": 0.29608938547486036,
      "grad_norm": 1.2065811157226562,
      "learning_rate": 0.00016512191587195866,
      "loss": 1.5018,
      "step": 2067
    },
    {
      "epoch": 0.29623263142816214,
      "grad_norm": 1.1206344366073608,
      "learning_rate": 0.00016508669807692127,
      "loss": 1.4378,
      "step": 2068
    },
    {
      "epoch": 0.296375877381464,
      "grad_norm": 1.1130290031433105,
      "learning_rate": 0.0001650514662703284,
      "loss": 1.5022,
      "step": 2069
    },
    {
      "epoch": 0.29651912333476577,
      "grad_norm": 1.1376516819000244,
      "learning_rate": 0.00016501622045976463,
      "loss": 1.5937,
      "step": 2070
    },
    {
      "epoch": 0.2966623692880676,
      "grad_norm": 0.9944444894790649,
      "learning_rate": 0.00016498096065281753,
      "loss": 1.5888,
      "step": 2071
    },
    {
      "epoch": 0.29680561524136945,
      "grad_norm": 1.234916090965271,
      "learning_rate": 0.00016494568685707762,
      "loss": 1.3764,
      "step": 2072
    },
    {
      "epoch": 0.29694886119467123,
      "grad_norm": 1.0958075523376465,
      "learning_rate": 0.00016491039908013847,
      "loss": 1.4478,
      "step": 2073
    },
    {
      "epoch": 0.2970921071479731,
      "grad_norm": 1.1120153665542603,
      "learning_rate": 0.00016487509732959668,
      "loss": 1.5635,
      "step": 2074
    },
    {
      "epoch": 0.2972353531012749,
      "grad_norm": 1.3901543617248535,
      "learning_rate": 0.00016483978161305188,
      "loss": 1.4523,
      "step": 2075
    },
    {
      "epoch": 0.2973785990545767,
      "grad_norm": 1.255515694618225,
      "learning_rate": 0.0001648044519381066,
      "loss": 1.5609,
      "step": 2076
    },
    {
      "epoch": 0.29752184500787854,
      "grad_norm": 1.0801960229873657,
      "learning_rate": 0.00016476910831236648,
      "loss": 1.4023,
      "step": 2077
    },
    {
      "epoch": 0.2976650909611803,
      "grad_norm": 1.359724998474121,
      "learning_rate": 0.0001647337507434401,
      "loss": 1.6331,
      "step": 2078
    },
    {
      "epoch": 0.29780833691448216,
      "grad_norm": 0.9753069877624512,
      "learning_rate": 0.00016469837923893915,
      "loss": 1.5339,
      "step": 2079
    },
    {
      "epoch": 0.297951582867784,
      "grad_norm": 1.0298526287078857,
      "learning_rate": 0.00016466299380647818,
      "loss": 1.463,
      "step": 2080
    },
    {
      "epoch": 0.2980948288210858,
      "grad_norm": 0.9451895356178284,
      "learning_rate": 0.00016462759445367477,
      "loss": 1.3969,
      "step": 2081
    },
    {
      "epoch": 0.29823807477438763,
      "grad_norm": 1.15664803981781,
      "learning_rate": 0.00016459218118814958,
      "loss": 1.4679,
      "step": 2082
    },
    {
      "epoch": 0.29838132072768947,
      "grad_norm": 1.0966644287109375,
      "learning_rate": 0.00016455675401752618,
      "loss": 1.2798,
      "step": 2083
    },
    {
      "epoch": 0.29852456668099125,
      "grad_norm": 1.2720539569854736,
      "learning_rate": 0.00016452131294943117,
      "loss": 1.4518,
      "step": 2084
    },
    {
      "epoch": 0.2986678126342931,
      "grad_norm": 1.0670486688613892,
      "learning_rate": 0.00016448585799149417,
      "loss": 1.5698,
      "step": 2085
    },
    {
      "epoch": 0.2988110585875949,
      "grad_norm": 1.2468509674072266,
      "learning_rate": 0.00016445038915134766,
      "loss": 1.5231,
      "step": 2086
    },
    {
      "epoch": 0.2989543045408967,
      "grad_norm": 1.2199418544769287,
      "learning_rate": 0.00016441490643662732,
      "loss": 1.4772,
      "step": 2087
    },
    {
      "epoch": 0.29909755049419856,
      "grad_norm": 1.2481852769851685,
      "learning_rate": 0.0001643794098549716,
      "loss": 1.5118,
      "step": 2088
    },
    {
      "epoch": 0.29924079644750035,
      "grad_norm": 1.1355080604553223,
      "learning_rate": 0.0001643438994140221,
      "loss": 1.6856,
      "step": 2089
    },
    {
      "epoch": 0.2993840424008022,
      "grad_norm": 0.9006658792495728,
      "learning_rate": 0.00016430837512142334,
      "loss": 1.6043,
      "step": 2090
    },
    {
      "epoch": 0.29952728835410397,
      "grad_norm": 1.1728410720825195,
      "learning_rate": 0.00016427283698482281,
      "loss": 1.4023,
      "step": 2091
    },
    {
      "epoch": 0.2996705343074058,
      "grad_norm": 1.1827518939971924,
      "learning_rate": 0.00016423728501187098,
      "loss": 1.6054,
      "step": 2092
    },
    {
      "epoch": 0.29981378026070765,
      "grad_norm": 0.9066771268844604,
      "learning_rate": 0.0001642017192102213,
      "loss": 1.4302,
      "step": 2093
    },
    {
      "epoch": 0.29995702621400944,
      "grad_norm": 0.9178932309150696,
      "learning_rate": 0.0001641661395875302,
      "loss": 1.4626,
      "step": 2094
    },
    {
      "epoch": 0.3001002721673113,
      "grad_norm": 1.1124746799468994,
      "learning_rate": 0.0001641305461514571,
      "loss": 1.5385,
      "step": 2095
    },
    {
      "epoch": 0.3002435181206131,
      "grad_norm": 1.167476773262024,
      "learning_rate": 0.00016409493890966442,
      "loss": 1.4419,
      "step": 2096
    },
    {
      "epoch": 0.3003867640739149,
      "grad_norm": 1.2050307989120483,
      "learning_rate": 0.00016405931786981755,
      "loss": 1.5183,
      "step": 2097
    },
    {
      "epoch": 0.30053001002721674,
      "grad_norm": 1.2343586683273315,
      "learning_rate": 0.00016402368303958468,
      "loss": 1.2939,
      "step": 2098
    },
    {
      "epoch": 0.30067325598051853,
      "grad_norm": 0.9498979449272156,
      "learning_rate": 0.00016398803442663724,
      "loss": 1.3712,
      "step": 2099
    },
    {
      "epoch": 0.30081650193382037,
      "grad_norm": 1.2889301776885986,
      "learning_rate": 0.00016395237203864943,
      "loss": 1.2372,
      "step": 2100
    },
    {
      "epoch": 0.3009597478871222,
      "grad_norm": 1.2563834190368652,
      "learning_rate": 0.0001639166958832985,
      "loss": 1.349,
      "step": 2101
    },
    {
      "epoch": 0.301102993840424,
      "grad_norm": 1.070231318473816,
      "learning_rate": 0.00016388100596826465,
      "loss": 1.3202,
      "step": 2102
    },
    {
      "epoch": 0.30124623979372583,
      "grad_norm": 0.9682635068893433,
      "learning_rate": 0.000163845302301231,
      "loss": 1.4629,
      "step": 2103
    },
    {
      "epoch": 0.3013894857470276,
      "grad_norm": 1.239965796470642,
      "learning_rate": 0.00016380958488988368,
      "loss": 1.4767,
      "step": 2104
    },
    {
      "epoch": 0.30153273170032946,
      "grad_norm": 1.1767518520355225,
      "learning_rate": 0.00016377385374191183,
      "loss": 1.3701,
      "step": 2105
    },
    {
      "epoch": 0.3016759776536313,
      "grad_norm": 0.9566437602043152,
      "learning_rate": 0.0001637381088650074,
      "loss": 1.5249,
      "step": 2106
    },
    {
      "epoch": 0.3018192236069331,
      "grad_norm": 1.1168581247329712,
      "learning_rate": 0.0001637023502668654,
      "loss": 1.5557,
      "step": 2107
    },
    {
      "epoch": 0.3019624695602349,
      "grad_norm": 1.2364850044250488,
      "learning_rate": 0.00016366657795518377,
      "loss": 1.248,
      "step": 2108
    },
    {
      "epoch": 0.30210571551353677,
      "grad_norm": 1.1146354675292969,
      "learning_rate": 0.00016363079193766345,
      "loss": 1.5762,
      "step": 2109
    },
    {
      "epoch": 0.30224896146683855,
      "grad_norm": 1.2413594722747803,
      "learning_rate": 0.00016359499222200818,
      "loss": 1.5448,
      "step": 2110
    },
    {
      "epoch": 0.3023922074201404,
      "grad_norm": 1.1661057472229004,
      "learning_rate": 0.00016355917881592485,
      "loss": 1.6454,
      "step": 2111
    },
    {
      "epoch": 0.3025354533734422,
      "grad_norm": 1.074952483177185,
      "learning_rate": 0.00016352335172712317,
      "loss": 1.6959,
      "step": 2112
    },
    {
      "epoch": 0.302678699326744,
      "grad_norm": 1.5178353786468506,
      "learning_rate": 0.0001634875109633158,
      "loss": 1.5195,
      "step": 2113
    },
    {
      "epoch": 0.30282194528004586,
      "grad_norm": 1.0164861679077148,
      "learning_rate": 0.00016345165653221838,
      "loss": 1.5207,
      "step": 2114
    },
    {
      "epoch": 0.30296519123334764,
      "grad_norm": 1.103379249572754,
      "learning_rate": 0.00016341578844154955,
      "loss": 1.5232,
      "step": 2115
    },
    {
      "epoch": 0.3031084371866495,
      "grad_norm": 1.0848551988601685,
      "learning_rate": 0.00016337990669903073,
      "loss": 1.5539,
      "step": 2116
    },
    {
      "epoch": 0.3032516831399513,
      "grad_norm": 1.0429604053497314,
      "learning_rate": 0.00016334401131238644,
      "loss": 1.6297,
      "step": 2117
    },
    {
      "epoch": 0.3033949290932531,
      "grad_norm": 1.0036472082138062,
      "learning_rate": 0.000163308102289344,
      "loss": 1.3732,
      "step": 2118
    },
    {
      "epoch": 0.30353817504655495,
      "grad_norm": 1.1866599321365356,
      "learning_rate": 0.00016327217963763374,
      "loss": 1.3574,
      "step": 2119
    },
    {
      "epoch": 0.30368142099985673,
      "grad_norm": 1.2575559616088867,
      "learning_rate": 0.000163236243364989,
      "loss": 1.6162,
      "step": 2120
    },
    {
      "epoch": 0.3038246669531586,
      "grad_norm": 0.9871320724487305,
      "learning_rate": 0.0001632002934791459,
      "loss": 1.5079,
      "step": 2121
    },
    {
      "epoch": 0.3039679129064604,
      "grad_norm": 1.0725313425064087,
      "learning_rate": 0.0001631643299878436,
      "loss": 1.5054,
      "step": 2122
    },
    {
      "epoch": 0.3041111588597622,
      "grad_norm": 1.0979975461959839,
      "learning_rate": 0.00016312835289882408,
      "loss": 1.6496,
      "step": 2123
    },
    {
      "epoch": 0.30425440481306404,
      "grad_norm": 1.101214051246643,
      "learning_rate": 0.0001630923622198324,
      "loss": 1.3688,
      "step": 2124
    },
    {
      "epoch": 0.3043976507663658,
      "grad_norm": 1.305351972579956,
      "learning_rate": 0.00016305635795861643,
      "loss": 1.3986,
      "step": 2125
    },
    {
      "epoch": 0.30454089671966766,
      "grad_norm": 1.2938451766967773,
      "learning_rate": 0.00016302034012292697,
      "loss": 1.3696,
      "step": 2126
    },
    {
      "epoch": 0.3046841426729695,
      "grad_norm": 1.298505425453186,
      "learning_rate": 0.0001629843087205178,
      "loss": 1.5394,
      "step": 2127
    },
    {
      "epoch": 0.3048273886262713,
      "grad_norm": 1.2057178020477295,
      "learning_rate": 0.0001629482637591456,
      "loss": 1.4988,
      "step": 2128
    },
    {
      "epoch": 0.30497063457957313,
      "grad_norm": 1.249611496925354,
      "learning_rate": 0.00016291220524656993,
      "loss": 1.5572,
      "step": 2129
    },
    {
      "epoch": 0.30511388053287497,
      "grad_norm": 1.1235121488571167,
      "learning_rate": 0.00016287613319055332,
      "loss": 1.5192,
      "step": 2130
    },
    {
      "epoch": 0.30525712648617676,
      "grad_norm": 1.4710801839828491,
      "learning_rate": 0.00016284004759886114,
      "loss": 1.4539,
      "step": 2131
    },
    {
      "epoch": 0.3054003724394786,
      "grad_norm": 0.978493332862854,
      "learning_rate": 0.00016280394847926178,
      "loss": 1.4619,
      "step": 2132
    },
    {
      "epoch": 0.3055436183927804,
      "grad_norm": 1.0383172035217285,
      "learning_rate": 0.0001627678358395265,
      "loss": 1.4563,
      "step": 2133
    },
    {
      "epoch": 0.3056868643460822,
      "grad_norm": 1.1714521646499634,
      "learning_rate": 0.0001627317096874294,
      "loss": 1.3108,
      "step": 2134
    },
    {
      "epoch": 0.30583011029938406,
      "grad_norm": 1.1852748394012451,
      "learning_rate": 0.00016269557003074757,
      "loss": 1.6014,
      "step": 2135
    },
    {
      "epoch": 0.30597335625268585,
      "grad_norm": 1.0720202922821045,
      "learning_rate": 0.00016265941687726099,
      "loss": 1.4473,
      "step": 2136
    },
    {
      "epoch": 0.3061166022059877,
      "grad_norm": 1.0016392469406128,
      "learning_rate": 0.00016262325023475252,
      "loss": 1.3899,
      "step": 2137
    },
    {
      "epoch": 0.3062598481592895,
      "grad_norm": 1.2735028266906738,
      "learning_rate": 0.000162587070111008,
      "loss": 1.5014,
      "step": 2138
    },
    {
      "epoch": 0.3064030941125913,
      "grad_norm": 1.3058689832687378,
      "learning_rate": 0.00016255087651381603,
      "loss": 1.3919,
      "step": 2139
    },
    {
      "epoch": 0.30654634006589315,
      "grad_norm": 1.1695564985275269,
      "learning_rate": 0.00016251466945096826,
      "loss": 1.3818,
      "step": 2140
    },
    {
      "epoch": 0.30668958601919494,
      "grad_norm": 1.0100617408752441,
      "learning_rate": 0.00016247844893025918,
      "loss": 1.4555,
      "step": 2141
    },
    {
      "epoch": 0.3068328319724968,
      "grad_norm": 1.0054876804351807,
      "learning_rate": 0.0001624422149594861,
      "loss": 1.4446,
      "step": 2142
    },
    {
      "epoch": 0.3069760779257986,
      "grad_norm": 1.1957088708877563,
      "learning_rate": 0.00016240596754644936,
      "loss": 1.3559,
      "step": 2143
    },
    {
      "epoch": 0.3071193238791004,
      "grad_norm": 1.2647818326950073,
      "learning_rate": 0.00016236970669895214,
      "loss": 1.6286,
      "step": 2144
    },
    {
      "epoch": 0.30726256983240224,
      "grad_norm": 1.10520339012146,
      "learning_rate": 0.00016233343242480047,
      "loss": 1.5808,
      "step": 2145
    },
    {
      "epoch": 0.30740581578570403,
      "grad_norm": 1.2879663705825806,
      "learning_rate": 0.00016229714473180328,
      "loss": 1.5655,
      "step": 2146
    },
    {
      "epoch": 0.30754906173900587,
      "grad_norm": 1.2024028301239014,
      "learning_rate": 0.00016226084362777247,
      "loss": 1.4058,
      "step": 2147
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": 1.2346138954162598,
      "learning_rate": 0.00016222452912052272,
      "loss": 1.3247,
      "step": 2148
    },
    {
      "epoch": 0.3078355536456095,
      "grad_norm": 1.369972825050354,
      "learning_rate": 0.0001621882012178717,
      "loss": 1.4775,
      "step": 2149
    },
    {
      "epoch": 0.30797879959891133,
      "grad_norm": 1.1432427167892456,
      "learning_rate": 0.0001621518599276399,
      "loss": 1.4419,
      "step": 2150
    },
    {
      "epoch": 0.3081220455522132,
      "grad_norm": 1.015098214149475,
      "learning_rate": 0.00016211550525765063,
      "loss": 1.5148,
      "step": 2151
    },
    {
      "epoch": 0.30826529150551496,
      "grad_norm": 1.3381190299987793,
      "learning_rate": 0.0001620791372157302,
      "loss": 1.4458,
      "step": 2152
    },
    {
      "epoch": 0.3084085374588168,
      "grad_norm": 1.2559499740600586,
      "learning_rate": 0.00016204275580970773,
      "loss": 1.3879,
      "step": 2153
    },
    {
      "epoch": 0.3085517834121186,
      "grad_norm": 1.3334771394729614,
      "learning_rate": 0.00016200636104741532,
      "loss": 1.499,
      "step": 2154
    },
    {
      "epoch": 0.3086950293654204,
      "grad_norm": 1.1180161237716675,
      "learning_rate": 0.00016196995293668774,
      "loss": 1.6303,
      "step": 2155
    },
    {
      "epoch": 0.30883827531872227,
      "grad_norm": 1.1852434873580933,
      "learning_rate": 0.0001619335314853628,
      "loss": 1.4955,
      "step": 2156
    },
    {
      "epoch": 0.30898152127202405,
      "grad_norm": 1.0549120903015137,
      "learning_rate": 0.00016189709670128122,
      "loss": 1.4196,
      "step": 2157
    },
    {
      "epoch": 0.3091247672253259,
      "grad_norm": 1.0799815654754639,
      "learning_rate": 0.00016186064859228638,
      "loss": 1.4356,
      "step": 2158
    },
    {
      "epoch": 0.3092680131786277,
      "grad_norm": 0.9937320947647095,
      "learning_rate": 0.0001618241871662247,
      "loss": 1.4904,
      "step": 2159
    },
    {
      "epoch": 0.3094112591319295,
      "grad_norm": 1.2357194423675537,
      "learning_rate": 0.00016178771243094547,
      "loss": 1.4504,
      "step": 2160
    },
    {
      "epoch": 0.30955450508523136,
      "grad_norm": 1.3001954555511475,
      "learning_rate": 0.00016175122439430077,
      "loss": 1.5392,
      "step": 2161
    },
    {
      "epoch": 0.30969775103853314,
      "grad_norm": 1.202223300933838,
      "learning_rate": 0.00016171472306414554,
      "loss": 1.4394,
      "step": 2162
    },
    {
      "epoch": 0.309840996991835,
      "grad_norm": 1.0657094717025757,
      "learning_rate": 0.00016167820844833764,
      "loss": 1.4745,
      "step": 2163
    },
    {
      "epoch": 0.3099842429451368,
      "grad_norm": 0.9404382109642029,
      "learning_rate": 0.00016164168055473775,
      "loss": 1.5796,
      "step": 2164
    },
    {
      "epoch": 0.3101274888984386,
      "grad_norm": 1.0849108695983887,
      "learning_rate": 0.00016160513939120943,
      "loss": 1.6945,
      "step": 2165
    },
    {
      "epoch": 0.31027073485174045,
      "grad_norm": 1.1392961740493774,
      "learning_rate": 0.0001615685849656191,
      "loss": 1.5728,
      "step": 2166
    },
    {
      "epoch": 0.31041398080504223,
      "grad_norm": 0.9921726584434509,
      "learning_rate": 0.00016153201728583602,
      "loss": 1.6066,
      "step": 2167
    },
    {
      "epoch": 0.3105572267583441,
      "grad_norm": 1.2363765239715576,
      "learning_rate": 0.0001614954363597323,
      "loss": 1.5385,
      "step": 2168
    },
    {
      "epoch": 0.3107004727116459,
      "grad_norm": 1.2025961875915527,
      "learning_rate": 0.0001614588421951829,
      "loss": 1.5741,
      "step": 2169
    },
    {
      "epoch": 0.3108437186649477,
      "grad_norm": 1.1037356853485107,
      "learning_rate": 0.00016142223480006563,
      "loss": 1.4146,
      "step": 2170
    },
    {
      "epoch": 0.31098696461824954,
      "grad_norm": 1.018072247505188,
      "learning_rate": 0.0001613856141822612,
      "loss": 1.6295,
      "step": 2171
    },
    {
      "epoch": 0.3111302105715514,
      "grad_norm": 1.3242359161376953,
      "learning_rate": 0.0001613489803496531,
      "loss": 1.6588,
      "step": 2172
    },
    {
      "epoch": 0.31127345652485316,
      "grad_norm": 1.4029531478881836,
      "learning_rate": 0.00016131233331012762,
      "loss": 1.5318,
      "step": 2173
    },
    {
      "epoch": 0.311416702478155,
      "grad_norm": 1.3314331769943237,
      "learning_rate": 0.0001612756730715741,
      "loss": 1.437,
      "step": 2174
    },
    {
      "epoch": 0.3115599484314568,
      "grad_norm": 1.1653331518173218,
      "learning_rate": 0.00016123899964188448,
      "loss": 1.4383,
      "step": 2175
    },
    {
      "epoch": 0.31170319438475863,
      "grad_norm": 1.15831458568573,
      "learning_rate": 0.00016120231302895366,
      "loss": 1.4641,
      "step": 2176
    },
    {
      "epoch": 0.31184644033806047,
      "grad_norm": 1.431643009185791,
      "learning_rate": 0.00016116561324067935,
      "loss": 1.3607,
      "step": 2177
    },
    {
      "epoch": 0.31198968629136226,
      "grad_norm": 1.15006422996521,
      "learning_rate": 0.00016112890028496216,
      "loss": 1.5628,
      "step": 2178
    },
    {
      "epoch": 0.3121329322446641,
      "grad_norm": 1.111311435699463,
      "learning_rate": 0.0001610921741697054,
      "loss": 1.5959,
      "step": 2179
    },
    {
      "epoch": 0.3122761781979659,
      "grad_norm": 1.1853933334350586,
      "learning_rate": 0.00016105543490281535,
      "loss": 1.6071,
      "step": 2180
    },
    {
      "epoch": 0.3124194241512677,
      "grad_norm": 1.5622564554214478,
      "learning_rate": 0.0001610186824922011,
      "loss": 1.6407,
      "step": 2181
    },
    {
      "epoch": 0.31256267010456956,
      "grad_norm": 1.170989990234375,
      "learning_rate": 0.00016098191694577442,
      "loss": 1.7186,
      "step": 2182
    },
    {
      "epoch": 0.31270591605787135,
      "grad_norm": 1.3031423091888428,
      "learning_rate": 0.00016094513827145016,
      "loss": 1.4696,
      "step": 2183
    },
    {
      "epoch": 0.3128491620111732,
      "grad_norm": 1.1214048862457275,
      "learning_rate": 0.00016090834647714575,
      "loss": 1.5491,
      "step": 2184
    },
    {
      "epoch": 0.31299240796447503,
      "grad_norm": 1.1624176502227783,
      "learning_rate": 0.00016087154157078156,
      "loss": 1.3835,
      "step": 2185
    },
    {
      "epoch": 0.3131356539177768,
      "grad_norm": 1.067884087562561,
      "learning_rate": 0.00016083472356028086,
      "loss": 1.7291,
      "step": 2186
    },
    {
      "epoch": 0.31327889987107865,
      "grad_norm": 1.4405581951141357,
      "learning_rate": 0.00016079789245356958,
      "loss": 1.3773,
      "step": 2187
    },
    {
      "epoch": 0.31342214582438044,
      "grad_norm": 1.1195565462112427,
      "learning_rate": 0.00016076104825857657,
      "loss": 1.5359,
      "step": 2188
    },
    {
      "epoch": 0.3135653917776823,
      "grad_norm": 1.4966022968292236,
      "learning_rate": 0.00016072419098323346,
      "loss": 1.456,
      "step": 2189
    },
    {
      "epoch": 0.3137086377309841,
      "grad_norm": 1.0616106986999512,
      "learning_rate": 0.00016068732063547473,
      "loss": 1.4872,
      "step": 2190
    },
    {
      "epoch": 0.3138518836842859,
      "grad_norm": 1.1995162963867188,
      "learning_rate": 0.00016065043722323768,
      "loss": 1.5527,
      "step": 2191
    },
    {
      "epoch": 0.31399512963758774,
      "grad_norm": 1.1095688343048096,
      "learning_rate": 0.0001606135407544623,
      "loss": 1.4138,
      "step": 2192
    },
    {
      "epoch": 0.3141383755908896,
      "grad_norm": 1.0261632204055786,
      "learning_rate": 0.0001605766312370916,
      "loss": 1.4086,
      "step": 2193
    },
    {
      "epoch": 0.31428162154419137,
      "grad_norm": 1.0876903533935547,
      "learning_rate": 0.0001605397086790712,
      "loss": 1.5084,
      "step": 2194
    },
    {
      "epoch": 0.3144248674974932,
      "grad_norm": 1.123974323272705,
      "learning_rate": 0.00016050277308834966,
      "loss": 1.492,
      "step": 2195
    },
    {
      "epoch": 0.314568113450795,
      "grad_norm": 1.466103434562683,
      "learning_rate": 0.00016046582447287828,
      "loss": 1.3834,
      "step": 2196
    },
    {
      "epoch": 0.31471135940409684,
      "grad_norm": 1.1065490245819092,
      "learning_rate": 0.0001604288628406112,
      "loss": 1.3981,
      "step": 2197
    },
    {
      "epoch": 0.3148546053573987,
      "grad_norm": 1.269467830657959,
      "learning_rate": 0.0001603918881995053,
      "loss": 1.5069,
      "step": 2198
    },
    {
      "epoch": 0.31499785131070046,
      "grad_norm": 1.3922014236450195,
      "learning_rate": 0.00016035490055752035,
      "loss": 1.3686,
      "step": 2199
    },
    {
      "epoch": 0.3151410972640023,
      "grad_norm": 1.3569414615631104,
      "learning_rate": 0.0001603178999226189,
      "loss": 1.3837,
      "step": 2200
    },
    {
      "epoch": 0.3152843432173041,
      "grad_norm": 1.118282437324524,
      "learning_rate": 0.0001602808863027662,
      "loss": 1.4184,
      "step": 2201
    },
    {
      "epoch": 0.3154275891706059,
      "grad_norm": 1.2146110534667969,
      "learning_rate": 0.0001602438597059304,
      "loss": 1.5382,
      "step": 2202
    },
    {
      "epoch": 0.31557083512390777,
      "grad_norm": 1.2688679695129395,
      "learning_rate": 0.00016020682014008242,
      "loss": 1.5125,
      "step": 2203
    },
    {
      "epoch": 0.31571408107720955,
      "grad_norm": 1.3139230012893677,
      "learning_rate": 0.00016016976761319595,
      "loss": 1.3806,
      "step": 2204
    },
    {
      "epoch": 0.3158573270305114,
      "grad_norm": 1.0200927257537842,
      "learning_rate": 0.0001601327021332475,
      "loss": 1.2959,
      "step": 2205
    },
    {
      "epoch": 0.31600057298381323,
      "grad_norm": 1.351724624633789,
      "learning_rate": 0.0001600956237082163,
      "loss": 1.3509,
      "step": 2206
    },
    {
      "epoch": 0.316143818937115,
      "grad_norm": 1.0325864553451538,
      "learning_rate": 0.00016005853234608446,
      "loss": 1.5371,
      "step": 2207
    },
    {
      "epoch": 0.31628706489041686,
      "grad_norm": 1.2085545063018799,
      "learning_rate": 0.00016002142805483685,
      "loss": 1.4455,
      "step": 2208
    },
    {
      "epoch": 0.31643031084371864,
      "grad_norm": 1.042733907699585,
      "learning_rate": 0.00015998431084246107,
      "loss": 1.4025,
      "step": 2209
    },
    {
      "epoch": 0.3165735567970205,
      "grad_norm": 1.0463552474975586,
      "learning_rate": 0.0001599471807169475,
      "loss": 1.5703,
      "step": 2210
    },
    {
      "epoch": 0.3167168027503223,
      "grad_norm": 1.1998884677886963,
      "learning_rate": 0.00015991003768628942,
      "loss": 1.4712,
      "step": 2211
    },
    {
      "epoch": 0.3168600487036241,
      "grad_norm": 0.9740208983421326,
      "learning_rate": 0.00015987288175848277,
      "loss": 1.5639,
      "step": 2212
    },
    {
      "epoch": 0.31700329465692595,
      "grad_norm": 1.4041082859039307,
      "learning_rate": 0.0001598357129415263,
      "loss": 1.474,
      "step": 2213
    },
    {
      "epoch": 0.31714654061022773,
      "grad_norm": 1.0597376823425293,
      "learning_rate": 0.00015979853124342154,
      "loss": 1.5728,
      "step": 2214
    },
    {
      "epoch": 0.3172897865635296,
      "grad_norm": 1.0904829502105713,
      "learning_rate": 0.00015976133667217277,
      "loss": 1.3362,
      "step": 2215
    },
    {
      "epoch": 0.3174330325168314,
      "grad_norm": 1.0378105640411377,
      "learning_rate": 0.00015972412923578712,
      "loss": 1.5321,
      "step": 2216
    },
    {
      "epoch": 0.3175762784701332,
      "grad_norm": 0.957038938999176,
      "learning_rate": 0.00015968690894227437,
      "loss": 1.3742,
      "step": 2217
    },
    {
      "epoch": 0.31771952442343504,
      "grad_norm": 1.2085511684417725,
      "learning_rate": 0.0001596496757996471,
      "loss": 1.4769,
      "step": 2218
    },
    {
      "epoch": 0.3178627703767369,
      "grad_norm": 1.1358758211135864,
      "learning_rate": 0.00015961242981592076,
      "loss": 1.3116,
      "step": 2219
    },
    {
      "epoch": 0.31800601633003867,
      "grad_norm": 1.0222986936569214,
      "learning_rate": 0.00015957517099911349,
      "loss": 1.6448,
      "step": 2220
    },
    {
      "epoch": 0.3181492622833405,
      "grad_norm": 1.0717604160308838,
      "learning_rate": 0.00015953789935724613,
      "loss": 1.7298,
      "step": 2221
    },
    {
      "epoch": 0.3182925082366423,
      "grad_norm": 1.1606909036636353,
      "learning_rate": 0.0001595006148983424,
      "loss": 1.3108,
      "step": 2222
    },
    {
      "epoch": 0.31843575418994413,
      "grad_norm": 1.3621573448181152,
      "learning_rate": 0.00015946331763042867,
      "loss": 1.3623,
      "step": 2223
    },
    {
      "epoch": 0.31857900014324597,
      "grad_norm": 1.155504584312439,
      "learning_rate": 0.00015942600756153414,
      "loss": 1.356,
      "step": 2224
    },
    {
      "epoch": 0.31872224609654776,
      "grad_norm": 1.0299174785614014,
      "learning_rate": 0.00015938868469969077,
      "loss": 1.4444,
      "step": 2225
    },
    {
      "epoch": 0.3188654920498496,
      "grad_norm": 1.0487732887268066,
      "learning_rate": 0.00015935134905293322,
      "loss": 1.6577,
      "step": 2226
    },
    {
      "epoch": 0.31900873800315144,
      "grad_norm": 1.1563888788223267,
      "learning_rate": 0.0001593140006292989,
      "loss": 1.3916,
      "step": 2227
    },
    {
      "epoch": 0.3191519839564532,
      "grad_norm": 1.2340785264968872,
      "learning_rate": 0.00015927663943682808,
      "loss": 1.3571,
      "step": 2228
    },
    {
      "epoch": 0.31929522990975506,
      "grad_norm": 1.4335360527038574,
      "learning_rate": 0.00015923926548356364,
      "loss": 1.4666,
      "step": 2229
    },
    {
      "epoch": 0.31943847586305685,
      "grad_norm": 0.9405779242515564,
      "learning_rate": 0.00015920187877755128,
      "loss": 1.4326,
      "step": 2230
    },
    {
      "epoch": 0.3195817218163587,
      "grad_norm": 1.0857157707214355,
      "learning_rate": 0.00015916447932683947,
      "loss": 1.4451,
      "step": 2231
    },
    {
      "epoch": 0.31972496776966053,
      "grad_norm": 1.1476625204086304,
      "learning_rate": 0.00015912706713947932,
      "loss": 1.5027,
      "step": 2232
    },
    {
      "epoch": 0.3198682137229623,
      "grad_norm": 1.0132766962051392,
      "learning_rate": 0.0001590896422235248,
      "loss": 1.5499,
      "step": 2233
    },
    {
      "epoch": 0.32001145967626415,
      "grad_norm": 1.142949104309082,
      "learning_rate": 0.00015905220458703253,
      "loss": 1.3891,
      "step": 2234
    },
    {
      "epoch": 0.32015470562956594,
      "grad_norm": 0.990533173084259,
      "learning_rate": 0.00015901475423806195,
      "loss": 1.4042,
      "step": 2235
    },
    {
      "epoch": 0.3202979515828678,
      "grad_norm": 1.1556705236434937,
      "learning_rate": 0.00015897729118467517,
      "loss": 1.4879,
      "step": 2236
    },
    {
      "epoch": 0.3204411975361696,
      "grad_norm": 1.2752677202224731,
      "learning_rate": 0.00015893981543493705,
      "loss": 1.4728,
      "step": 2237
    },
    {
      "epoch": 0.3205844434894714,
      "grad_norm": 1.0377402305603027,
      "learning_rate": 0.0001589023269969152,
      "loss": 1.4546,
      "step": 2238
    },
    {
      "epoch": 0.32072768944277324,
      "grad_norm": 0.9539663791656494,
      "learning_rate": 0.00015886482587867997,
      "loss": 1.5698,
      "step": 2239
    },
    {
      "epoch": 0.3208709353960751,
      "grad_norm": 0.9791962504386902,
      "learning_rate": 0.0001588273120883044,
      "loss": 1.4965,
      "step": 2240
    },
    {
      "epoch": 0.32101418134937687,
      "grad_norm": 1.2567335367202759,
      "learning_rate": 0.00015878978563386425,
      "loss": 1.6168,
      "step": 2241
    },
    {
      "epoch": 0.3211574273026787,
      "grad_norm": 0.9634516835212708,
      "learning_rate": 0.00015875224652343815,
      "loss": 1.4496,
      "step": 2242
    },
    {
      "epoch": 0.3213006732559805,
      "grad_norm": 1.3572815656661987,
      "learning_rate": 0.00015871469476510722,
      "loss": 1.3976,
      "step": 2243
    },
    {
      "epoch": 0.32144391920928234,
      "grad_norm": 1.1213479042053223,
      "learning_rate": 0.00015867713036695546,
      "loss": 1.4437,
      "step": 2244
    },
    {
      "epoch": 0.3215871651625842,
      "grad_norm": 1.2937026023864746,
      "learning_rate": 0.00015863955333706957,
      "loss": 1.416,
      "step": 2245
    },
    {
      "epoch": 0.32173041111588596,
      "grad_norm": 1.1191105842590332,
      "learning_rate": 0.00015860196368353897,
      "loss": 1.4718,
      "step": 2246
    },
    {
      "epoch": 0.3218736570691878,
      "grad_norm": 0.9822176694869995,
      "learning_rate": 0.00015856436141445577,
      "loss": 1.5605,
      "step": 2247
    },
    {
      "epoch": 0.32201690302248964,
      "grad_norm": 0.9939042329788208,
      "learning_rate": 0.00015852674653791477,
      "loss": 1.3622,
      "step": 2248
    },
    {
      "epoch": 0.3221601489757914,
      "grad_norm": 1.1110790967941284,
      "learning_rate": 0.00015848911906201355,
      "loss": 1.5047,
      "step": 2249
    },
    {
      "epoch": 0.32230339492909327,
      "grad_norm": 1.171623945236206,
      "learning_rate": 0.0001584514789948524,
      "loss": 1.571,
      "step": 2250
    },
    {
      "epoch": 0.32244664088239505,
      "grad_norm": 1.0576767921447754,
      "learning_rate": 0.00015841382634453427,
      "loss": 1.4366,
      "step": 2251
    },
    {
      "epoch": 0.3225898868356969,
      "grad_norm": 1.228026032447815,
      "learning_rate": 0.00015837616111916483,
      "loss": 1.3506,
      "step": 2252
    },
    {
      "epoch": 0.32273313278899873,
      "grad_norm": 1.191338300704956,
      "learning_rate": 0.0001583384833268525,
      "loss": 1.6866,
      "step": 2253
    },
    {
      "epoch": 0.3228763787423005,
      "grad_norm": 1.400315284729004,
      "learning_rate": 0.00015830079297570837,
      "loss": 1.4451,
      "step": 2254
    },
    {
      "epoch": 0.32301962469560236,
      "grad_norm": 1.071686029434204,
      "learning_rate": 0.0001582630900738462,
      "loss": 1.3625,
      "step": 2255
    },
    {
      "epoch": 0.32316287064890414,
      "grad_norm": 1.098414421081543,
      "learning_rate": 0.00015822537462938254,
      "loss": 1.4156,
      "step": 2256
    },
    {
      "epoch": 0.323306116602206,
      "grad_norm": 1.3927415609359741,
      "learning_rate": 0.0001581876466504366,
      "loss": 1.4602,
      "step": 2257
    },
    {
      "epoch": 0.3234493625555078,
      "grad_norm": 1.1547175645828247,
      "learning_rate": 0.00015814990614513024,
      "loss": 1.3088,
      "step": 2258
    },
    {
      "epoch": 0.3235926085088096,
      "grad_norm": 1.1632882356643677,
      "learning_rate": 0.00015811215312158807,
      "loss": 1.3343,
      "step": 2259
    },
    {
      "epoch": 0.32373585446211145,
      "grad_norm": 1.2297313213348389,
      "learning_rate": 0.00015807438758793735,
      "loss": 1.5297,
      "step": 2260
    },
    {
      "epoch": 0.3238791004154133,
      "grad_norm": 1.1143958568572998,
      "learning_rate": 0.00015803660955230817,
      "loss": 1.4774,
      "step": 2261
    },
    {
      "epoch": 0.3240223463687151,
      "grad_norm": 1.2539821863174438,
      "learning_rate": 0.0001579988190228331,
      "loss": 1.4504,
      "step": 2262
    },
    {
      "epoch": 0.3241655923220169,
      "grad_norm": 1.1935629844665527,
      "learning_rate": 0.00015796101600764755,
      "loss": 1.5425,
      "step": 2263
    },
    {
      "epoch": 0.3243088382753187,
      "grad_norm": 1.5752179622650146,
      "learning_rate": 0.00015792320051488955,
      "loss": 1.4837,
      "step": 2264
    },
    {
      "epoch": 0.32445208422862054,
      "grad_norm": 1.1541529893875122,
      "learning_rate": 0.0001578853725526999,
      "loss": 1.6657,
      "step": 2265
    },
    {
      "epoch": 0.3245953301819224,
      "grad_norm": 1.1816414594650269,
      "learning_rate": 0.00015784753212922192,
      "loss": 1.4886,
      "step": 2266
    },
    {
      "epoch": 0.32473857613522417,
      "grad_norm": 1.0843379497528076,
      "learning_rate": 0.0001578096792526018,
      "loss": 1.5176,
      "step": 2267
    },
    {
      "epoch": 0.324881822088526,
      "grad_norm": 1.0409349203109741,
      "learning_rate": 0.00015777181393098833,
      "loss": 1.4746,
      "step": 2268
    },
    {
      "epoch": 0.3250250680418278,
      "grad_norm": 1.1670563220977783,
      "learning_rate": 0.0001577339361725329,
      "loss": 1.4648,
      "step": 2269
    },
    {
      "epoch": 0.32516831399512963,
      "grad_norm": 1.6981006860733032,
      "learning_rate": 0.0001576960459853898,
      "loss": 1.3805,
      "step": 2270
    },
    {
      "epoch": 0.3253115599484315,
      "grad_norm": 1.0562732219696045,
      "learning_rate": 0.00015765814337771565,
      "loss": 1.4242,
      "step": 2271
    },
    {
      "epoch": 0.32545480590173326,
      "grad_norm": 1.0475010871887207,
      "learning_rate": 0.00015762022835767012,
      "loss": 1.6098,
      "step": 2272
    },
    {
      "epoch": 0.3255980518550351,
      "grad_norm": 1.4024945497512817,
      "learning_rate": 0.00015758230093341523,
      "loss": 1.6323,
      "step": 2273
    },
    {
      "epoch": 0.32574129780833694,
      "grad_norm": 0.9667457342147827,
      "learning_rate": 0.00015754436111311594,
      "loss": 1.4453,
      "step": 2274
    },
    {
      "epoch": 0.3258845437616387,
      "grad_norm": 1.225041389465332,
      "learning_rate": 0.0001575064089049397,
      "loss": 1.2455,
      "step": 2275
    },
    {
      "epoch": 0.32602778971494056,
      "grad_norm": 1.1015914678573608,
      "learning_rate": 0.0001574684443170567,
      "loss": 1.5497,
      "step": 2276
    },
    {
      "epoch": 0.32617103566824235,
      "grad_norm": 1.1255090236663818,
      "learning_rate": 0.00015743046735763975,
      "loss": 1.4143,
      "step": 2277
    },
    {
      "epoch": 0.3263142816215442,
      "grad_norm": 1.0734444856643677,
      "learning_rate": 0.00015739247803486434,
      "loss": 1.269,
      "step": 2278
    },
    {
      "epoch": 0.32645752757484603,
      "grad_norm": 1.330520749092102,
      "learning_rate": 0.00015735447635690868,
      "loss": 1.3547,
      "step": 2279
    },
    {
      "epoch": 0.3266007735281478,
      "grad_norm": 1.29810631275177,
      "learning_rate": 0.00015731646233195357,
      "loss": 1.5087,
      "step": 2280
    },
    {
      "epoch": 0.32674401948144965,
      "grad_norm": 1.3337150812149048,
      "learning_rate": 0.00015727843596818243,
      "loss": 1.4006,
      "step": 2281
    },
    {
      "epoch": 0.3268872654347515,
      "grad_norm": 0.9875546097755432,
      "learning_rate": 0.00015724039727378148,
      "loss": 1.6777,
      "step": 2282
    },
    {
      "epoch": 0.3270305113880533,
      "grad_norm": 1.3372234106063843,
      "learning_rate": 0.0001572023462569395,
      "loss": 1.554,
      "step": 2283
    },
    {
      "epoch": 0.3271737573413551,
      "grad_norm": 1.1482656002044678,
      "learning_rate": 0.00015716428292584787,
      "loss": 1.5611,
      "step": 2284
    },
    {
      "epoch": 0.3273170032946569,
      "grad_norm": 1.0315511226654053,
      "learning_rate": 0.00015712620728870074,
      "loss": 1.5942,
      "step": 2285
    },
    {
      "epoch": 0.32746024924795875,
      "grad_norm": 1.0177346467971802,
      "learning_rate": 0.00015708811935369484,
      "loss": 1.3909,
      "step": 2286
    },
    {
      "epoch": 0.3276034952012606,
      "grad_norm": 1.2984817028045654,
      "learning_rate": 0.00015705001912902957,
      "loss": 1.2198,
      "step": 2287
    },
    {
      "epoch": 0.32774674115456237,
      "grad_norm": 1.277712106704712,
      "learning_rate": 0.00015701190662290693,
      "loss": 1.4536,
      "step": 2288
    },
    {
      "epoch": 0.3278899871078642,
      "grad_norm": 1.0963908433914185,
      "learning_rate": 0.0001569737818435316,
      "loss": 1.5342,
      "step": 2289
    },
    {
      "epoch": 0.328033233061166,
      "grad_norm": 1.3517429828643799,
      "learning_rate": 0.00015693564479911097,
      "loss": 1.4373,
      "step": 2290
    },
    {
      "epoch": 0.32817647901446784,
      "grad_norm": 1.0697838068008423,
      "learning_rate": 0.0001568974954978549,
      "loss": 1.4788,
      "step": 2291
    },
    {
      "epoch": 0.3283197249677697,
      "grad_norm": 1.1097118854522705,
      "learning_rate": 0.00015685933394797607,
      "loss": 1.5807,
      "step": 2292
    },
    {
      "epoch": 0.32846297092107146,
      "grad_norm": 1.1345523595809937,
      "learning_rate": 0.00015682116015768965,
      "loss": 1.3191,
      "step": 2293
    },
    {
      "epoch": 0.3286062168743733,
      "grad_norm": 1.0295121669769287,
      "learning_rate": 0.00015678297413521363,
      "loss": 1.6011,
      "step": 2294
    },
    {
      "epoch": 0.32874946282767514,
      "grad_norm": 1.204581379890442,
      "learning_rate": 0.00015674477588876838,
      "loss": 1.4564,
      "step": 2295
    },
    {
      "epoch": 0.32889270878097693,
      "grad_norm": 1.2863401174545288,
      "learning_rate": 0.00015670656542657714,
      "loss": 1.606,
      "step": 2296
    },
    {
      "epoch": 0.32903595473427877,
      "grad_norm": 0.9845897555351257,
      "learning_rate": 0.0001566683427568656,
      "loss": 1.6463,
      "step": 2297
    },
    {
      "epoch": 0.32917920068758055,
      "grad_norm": 1.2133171558380127,
      "learning_rate": 0.00015663010788786221,
      "loss": 1.453,
      "step": 2298
    },
    {
      "epoch": 0.3293224466408824,
      "grad_norm": 1.4354416131973267,
      "learning_rate": 0.00015659186082779797,
      "loss": 1.5182,
      "step": 2299
    },
    {
      "epoch": 0.32946569259418423,
      "grad_norm": 1.2099313735961914,
      "learning_rate": 0.00015655360158490651,
      "loss": 1.4823,
      "step": 2300
    },
    {
      "epoch": 0.329608938547486,
      "grad_norm": 1.1526235342025757,
      "learning_rate": 0.00015651533016742414,
      "loss": 1.7675,
      "step": 2301
    },
    {
      "epoch": 0.32975218450078786,
      "grad_norm": 1.070486307144165,
      "learning_rate": 0.00015647704658358966,
      "loss": 1.4568,
      "step": 2302
    },
    {
      "epoch": 0.32989543045408964,
      "grad_norm": 0.9689673781394958,
      "learning_rate": 0.00015643875084164467,
      "loss": 1.5509,
      "step": 2303
    },
    {
      "epoch": 0.3300386764073915,
      "grad_norm": 1.198616862297058,
      "learning_rate": 0.00015640044294983329,
      "loss": 1.7433,
      "step": 2304
    },
    {
      "epoch": 0.3301819223606933,
      "grad_norm": 1.1746795177459717,
      "learning_rate": 0.00015636212291640224,
      "loss": 1.4049,
      "step": 2305
    },
    {
      "epoch": 0.3303251683139951,
      "grad_norm": 1.2097768783569336,
      "learning_rate": 0.00015632379074960082,
      "loss": 1.6876,
      "step": 2306
    },
    {
      "epoch": 0.33046841426729695,
      "grad_norm": 1.1446720361709595,
      "learning_rate": 0.0001562854464576811,
      "loss": 1.5306,
      "step": 2307
    },
    {
      "epoch": 0.3306116602205988,
      "grad_norm": 1.4527058601379395,
      "learning_rate": 0.00015624709004889758,
      "loss": 1.7195,
      "step": 2308
    },
    {
      "epoch": 0.3307549061739006,
      "grad_norm": 1.2266982793807983,
      "learning_rate": 0.0001562087215315075,
      "loss": 1.5028,
      "step": 2309
    },
    {
      "epoch": 0.3308981521272024,
      "grad_norm": 0.8389855027198792,
      "learning_rate": 0.0001561703409137706,
      "loss": 1.4199,
      "step": 2310
    },
    {
      "epoch": 0.3310413980805042,
      "grad_norm": 1.0936599969863892,
      "learning_rate": 0.0001561319482039493,
      "loss": 1.3433,
      "step": 2311
    },
    {
      "epoch": 0.33118464403380604,
      "grad_norm": 1.170125126838684,
      "learning_rate": 0.0001560935434103086,
      "loss": 1.3587,
      "step": 2312
    },
    {
      "epoch": 0.3313278899871079,
      "grad_norm": 1.2092872858047485,
      "learning_rate": 0.00015605512654111615,
      "loss": 1.427,
      "step": 2313
    },
    {
      "epoch": 0.33147113594040967,
      "grad_norm": 0.9599800109863281,
      "learning_rate": 0.00015601669760464203,
      "loss": 1.4219,
      "step": 2314
    },
    {
      "epoch": 0.3316143818937115,
      "grad_norm": 1.182982325553894,
      "learning_rate": 0.00015597825660915916,
      "loss": 1.4012,
      "step": 2315
    },
    {
      "epoch": 0.33175762784701335,
      "grad_norm": 1.0515201091766357,
      "learning_rate": 0.0001559398035629429,
      "loss": 1.388,
      "step": 2316
    },
    {
      "epoch": 0.33190087380031513,
      "grad_norm": 1.3263829946517944,
      "learning_rate": 0.00015590133847427116,
      "loss": 1.7417,
      "step": 2317
    },
    {
      "epoch": 0.332044119753617,
      "grad_norm": 0.9802058935165405,
      "learning_rate": 0.00015586286135142467,
      "loss": 1.4557,
      "step": 2318
    },
    {
      "epoch": 0.33218736570691876,
      "grad_norm": 0.9632196426391602,
      "learning_rate": 0.00015582437220268647,
      "loss": 1.6805,
      "step": 2319
    },
    {
      "epoch": 0.3323306116602206,
      "grad_norm": 0.9994145035743713,
      "learning_rate": 0.00015578587103634242,
      "loss": 1.4888,
      "step": 2320
    },
    {
      "epoch": 0.33247385761352244,
      "grad_norm": 1.0613521337509155,
      "learning_rate": 0.00015574735786068078,
      "loss": 1.6683,
      "step": 2321
    },
    {
      "epoch": 0.3326171035668242,
      "grad_norm": 1.0585310459136963,
      "learning_rate": 0.00015570883268399257,
      "loss": 1.4463,
      "step": 2322
    },
    {
      "epoch": 0.33276034952012606,
      "grad_norm": 1.0785530805587769,
      "learning_rate": 0.0001556702955145712,
      "loss": 1.4671,
      "step": 2323
    },
    {
      "epoch": 0.33290359547342785,
      "grad_norm": 1.5648711919784546,
      "learning_rate": 0.00015563174636071286,
      "loss": 1.4255,
      "step": 2324
    },
    {
      "epoch": 0.3330468414267297,
      "grad_norm": 1.15556800365448,
      "learning_rate": 0.00015559318523071622,
      "loss": 1.3972,
      "step": 2325
    },
    {
      "epoch": 0.33319008738003153,
      "grad_norm": 1.247313380241394,
      "learning_rate": 0.0001555546121328825,
      "loss": 1.3692,
      "step": 2326
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 1.0710923671722412,
      "learning_rate": 0.00015551602707551557,
      "loss": 1.5111,
      "step": 2327
    },
    {
      "epoch": 0.33347657928663516,
      "grad_norm": 1.0387035608291626,
      "learning_rate": 0.00015547743006692177,
      "loss": 1.4257,
      "step": 2328
    },
    {
      "epoch": 0.333619825239937,
      "grad_norm": 1.3260657787322998,
      "learning_rate": 0.00015543882111541016,
      "loss": 1.4778,
      "step": 2329
    },
    {
      "epoch": 0.3337630711932388,
      "grad_norm": 1.0877724885940552,
      "learning_rate": 0.0001554002002292923,
      "loss": 1.4053,
      "step": 2330
    },
    {
      "epoch": 0.3339063171465406,
      "grad_norm": 1.1008830070495605,
      "learning_rate": 0.00015536156741688222,
      "loss": 1.628,
      "step": 2331
    },
    {
      "epoch": 0.3340495630998424,
      "grad_norm": 1.6206791400909424,
      "learning_rate": 0.00015532292268649668,
      "loss": 1.3678,
      "step": 2332
    },
    {
      "epoch": 0.33419280905314425,
      "grad_norm": 1.35749089717865,
      "learning_rate": 0.0001552842660464549,
      "loss": 1.5505,
      "step": 2333
    },
    {
      "epoch": 0.3343360550064461,
      "grad_norm": 0.9546893835067749,
      "learning_rate": 0.00015524559750507873,
      "loss": 1.4559,
      "step": 2334
    },
    {
      "epoch": 0.33447930095974787,
      "grad_norm": 1.160882592201233,
      "learning_rate": 0.00015520691707069254,
      "loss": 1.2959,
      "step": 2335
    },
    {
      "epoch": 0.3346225469130497,
      "grad_norm": 1.0720781087875366,
      "learning_rate": 0.00015516822475162325,
      "loss": 1.5045,
      "step": 2336
    },
    {
      "epoch": 0.33476579286635155,
      "grad_norm": 1.3178445100784302,
      "learning_rate": 0.0001551295205562004,
      "loss": 1.4813,
      "step": 2337
    },
    {
      "epoch": 0.33490903881965334,
      "grad_norm": 1.1616284847259521,
      "learning_rate": 0.000155090804492756,
      "loss": 1.5558,
      "step": 2338
    },
    {
      "epoch": 0.3350522847729552,
      "grad_norm": 1.1865198612213135,
      "learning_rate": 0.00015505207656962467,
      "loss": 1.536,
      "step": 2339
    },
    {
      "epoch": 0.33519553072625696,
      "grad_norm": 1.0836961269378662,
      "learning_rate": 0.00015501333679514357,
      "loss": 1.3854,
      "step": 2340
    },
    {
      "epoch": 0.3353387766795588,
      "grad_norm": 1.137036681175232,
      "learning_rate": 0.00015497458517765245,
      "loss": 1.381,
      "step": 2341
    },
    {
      "epoch": 0.33548202263286064,
      "grad_norm": 1.2110360860824585,
      "learning_rate": 0.00015493582172549354,
      "loss": 1.3714,
      "step": 2342
    },
    {
      "epoch": 0.33562526858616243,
      "grad_norm": 1.1092302799224854,
      "learning_rate": 0.00015489704644701166,
      "loss": 1.4093,
      "step": 2343
    },
    {
      "epoch": 0.33576851453946427,
      "grad_norm": 1.2615810632705688,
      "learning_rate": 0.00015485825935055418,
      "loss": 1.4011,
      "step": 2344
    },
    {
      "epoch": 0.33591176049276605,
      "grad_norm": 1.0309126377105713,
      "learning_rate": 0.00015481946044447099,
      "loss": 1.4938,
      "step": 2345
    },
    {
      "epoch": 0.3360550064460679,
      "grad_norm": 1.0209304094314575,
      "learning_rate": 0.0001547806497371145,
      "loss": 1.398,
      "step": 2346
    },
    {
      "epoch": 0.33619825239936973,
      "grad_norm": 1.1450188159942627,
      "learning_rate": 0.0001547418272368398,
      "loss": 1.4812,
      "step": 2347
    },
    {
      "epoch": 0.3363414983526715,
      "grad_norm": 1.2547285556793213,
      "learning_rate": 0.00015470299295200434,
      "loss": 1.491,
      "step": 2348
    },
    {
      "epoch": 0.33648474430597336,
      "grad_norm": 1.1752949953079224,
      "learning_rate": 0.00015466414689096816,
      "loss": 1.4352,
      "step": 2349
    },
    {
      "epoch": 0.3366279902592752,
      "grad_norm": 1.191088318824768,
      "learning_rate": 0.00015462528906209387,
      "loss": 1.4359,
      "step": 2350
    },
    {
      "epoch": 0.336771236212577,
      "grad_norm": 1.062980055809021,
      "learning_rate": 0.0001545864194737466,
      "loss": 1.5149,
      "step": 2351
    },
    {
      "epoch": 0.3369144821658788,
      "grad_norm": 1.3870859146118164,
      "learning_rate": 0.00015454753813429407,
      "loss": 1.2917,
      "step": 2352
    },
    {
      "epoch": 0.3370577281191806,
      "grad_norm": 1.1073658466339111,
      "learning_rate": 0.0001545086450521064,
      "loss": 1.5295,
      "step": 2353
    },
    {
      "epoch": 0.33720097407248245,
      "grad_norm": 1.0015462636947632,
      "learning_rate": 0.00015446974023555632,
      "loss": 1.3628,
      "step": 2354
    },
    {
      "epoch": 0.3373442200257843,
      "grad_norm": 1.0080305337905884,
      "learning_rate": 0.00015443082369301912,
      "loss": 1.3981,
      "step": 2355
    },
    {
      "epoch": 0.3374874659790861,
      "grad_norm": 1.0880720615386963,
      "learning_rate": 0.00015439189543287247,
      "loss": 1.4368,
      "step": 2356
    },
    {
      "epoch": 0.3376307119323879,
      "grad_norm": 1.1552070379257202,
      "learning_rate": 0.00015435295546349678,
      "loss": 1.3068,
      "step": 2357
    },
    {
      "epoch": 0.3377739578856897,
      "grad_norm": 1.0665514469146729,
      "learning_rate": 0.0001543140037932748,
      "loss": 1.5995,
      "step": 2358
    },
    {
      "epoch": 0.33791720383899154,
      "grad_norm": 1.5568562746047974,
      "learning_rate": 0.0001542750404305918,
      "loss": 1.4746,
      "step": 2359
    },
    {
      "epoch": 0.3380604497922934,
      "grad_norm": 0.9816345572471619,
      "learning_rate": 0.00015423606538383577,
      "loss": 1.521,
      "step": 2360
    },
    {
      "epoch": 0.33820369574559517,
      "grad_norm": 0.9525036215782166,
      "learning_rate": 0.00015419707866139696,
      "loss": 1.4592,
      "step": 2361
    },
    {
      "epoch": 0.338346941698897,
      "grad_norm": 1.0205943584442139,
      "learning_rate": 0.00015415808027166832,
      "loss": 1.3816,
      "step": 2362
    },
    {
      "epoch": 0.33849018765219885,
      "grad_norm": 1.1603480577468872,
      "learning_rate": 0.00015411907022304516,
      "loss": 1.3569,
      "step": 2363
    },
    {
      "epoch": 0.33863343360550063,
      "grad_norm": 1.1628680229187012,
      "learning_rate": 0.00015408004852392543,
      "loss": 1.4764,
      "step": 2364
    },
    {
      "epoch": 0.3387766795588025,
      "grad_norm": 1.1466549634933472,
      "learning_rate": 0.00015404101518270955,
      "loss": 1.5823,
      "step": 2365
    },
    {
      "epoch": 0.33891992551210426,
      "grad_norm": 1.175576090812683,
      "learning_rate": 0.00015400197020780038,
      "loss": 1.4831,
      "step": 2366
    },
    {
      "epoch": 0.3390631714654061,
      "grad_norm": 1.0311682224273682,
      "learning_rate": 0.0001539629136076034,
      "loss": 1.4604,
      "step": 2367
    },
    {
      "epoch": 0.33920641741870794,
      "grad_norm": 1.004539966583252,
      "learning_rate": 0.00015392384539052642,
      "loss": 1.4496,
      "step": 2368
    },
    {
      "epoch": 0.3393496633720097,
      "grad_norm": 1.1400903463363647,
      "learning_rate": 0.00015388476556498003,
      "loss": 1.2634,
      "step": 2369
    },
    {
      "epoch": 0.33949290932531156,
      "grad_norm": 1.298755407333374,
      "learning_rate": 0.000153845674139377,
      "loss": 1.3708,
      "step": 2370
    },
    {
      "epoch": 0.3396361552786134,
      "grad_norm": 1.0015780925750732,
      "learning_rate": 0.0001538065711221328,
      "loss": 1.6126,
      "step": 2371
    },
    {
      "epoch": 0.3397794012319152,
      "grad_norm": 0.9902368783950806,
      "learning_rate": 0.00015376745652166535,
      "loss": 1.55,
      "step": 2372
    },
    {
      "epoch": 0.33992264718521703,
      "grad_norm": 1.1694645881652832,
      "learning_rate": 0.0001537283303463951,
      "loss": 1.3771,
      "step": 2373
    },
    {
      "epoch": 0.3400658931385188,
      "grad_norm": 1.2839577198028564,
      "learning_rate": 0.00015368919260474483,
      "loss": 1.4093,
      "step": 2374
    },
    {
      "epoch": 0.34020913909182066,
      "grad_norm": 1.049832820892334,
      "learning_rate": 0.00015365004330514,
      "loss": 1.5767,
      "step": 2375
    },
    {
      "epoch": 0.3403523850451225,
      "grad_norm": 1.3304349184036255,
      "learning_rate": 0.0001536108824560085,
      "loss": 1.5181,
      "step": 2376
    },
    {
      "epoch": 0.3404956309984243,
      "grad_norm": 1.0783777236938477,
      "learning_rate": 0.00015357171006578067,
      "loss": 1.4845,
      "step": 2377
    },
    {
      "epoch": 0.3406388769517261,
      "grad_norm": 0.9464433789253235,
      "learning_rate": 0.00015353252614288935,
      "loss": 1.5138,
      "step": 2378
    },
    {
      "epoch": 0.3407821229050279,
      "grad_norm": 0.9589561820030212,
      "learning_rate": 0.0001534933306957699,
      "loss": 1.5157,
      "step": 2379
    },
    {
      "epoch": 0.34092536885832975,
      "grad_norm": 0.9670271277427673,
      "learning_rate": 0.00015345412373286008,
      "loss": 1.5581,
      "step": 2380
    },
    {
      "epoch": 0.3410686148116316,
      "grad_norm": 1.1025387048721313,
      "learning_rate": 0.00015341490526260017,
      "loss": 1.512,
      "step": 2381
    },
    {
      "epoch": 0.3412118607649334,
      "grad_norm": 1.3430148363113403,
      "learning_rate": 0.00015337567529343302,
      "loss": 1.4264,
      "step": 2382
    },
    {
      "epoch": 0.3413551067182352,
      "grad_norm": 1.1018345355987549,
      "learning_rate": 0.00015333643383380383,
      "loss": 1.5427,
      "step": 2383
    },
    {
      "epoch": 0.34149835267153705,
      "grad_norm": 1.0232715606689453,
      "learning_rate": 0.0001532971808921603,
      "loss": 1.4403,
      "step": 2384
    },
    {
      "epoch": 0.34164159862483884,
      "grad_norm": 1.0198861360549927,
      "learning_rate": 0.00015325791647695258,
      "loss": 1.5244,
      "step": 2385
    },
    {
      "epoch": 0.3417848445781407,
      "grad_norm": 1.1815303564071655,
      "learning_rate": 0.00015321864059663343,
      "loss": 1.5148,
      "step": 2386
    },
    {
      "epoch": 0.34192809053144246,
      "grad_norm": 1.2060497999191284,
      "learning_rate": 0.00015317935325965789,
      "loss": 1.6608,
      "step": 2387
    },
    {
      "epoch": 0.3420713364847443,
      "grad_norm": 1.074978232383728,
      "learning_rate": 0.00015314005447448356,
      "loss": 1.4645,
      "step": 2388
    },
    {
      "epoch": 0.34221458243804614,
      "grad_norm": 1.297765851020813,
      "learning_rate": 0.00015310074424957054,
      "loss": 1.5866,
      "step": 2389
    },
    {
      "epoch": 0.34235782839134793,
      "grad_norm": 1.212799310684204,
      "learning_rate": 0.00015306142259338129,
      "loss": 1.5388,
      "step": 2390
    },
    {
      "epoch": 0.34250107434464977,
      "grad_norm": 1.194262981414795,
      "learning_rate": 0.00015302208951438085,
      "loss": 1.3649,
      "step": 2391
    },
    {
      "epoch": 0.3426443202979516,
      "grad_norm": 1.2334723472595215,
      "learning_rate": 0.00015298274502103664,
      "loss": 1.3383,
      "step": 2392
    },
    {
      "epoch": 0.3427875662512534,
      "grad_norm": 1.1157070398330688,
      "learning_rate": 0.0001529433891218185,
      "loss": 1.4272,
      "step": 2393
    },
    {
      "epoch": 0.34293081220455524,
      "grad_norm": 1.2604163885116577,
      "learning_rate": 0.0001529040218251989,
      "loss": 1.4221,
      "step": 2394
    },
    {
      "epoch": 0.343074058157857,
      "grad_norm": 1.0845160484313965,
      "learning_rate": 0.00015286464313965254,
      "loss": 1.4989,
      "step": 2395
    },
    {
      "epoch": 0.34321730411115886,
      "grad_norm": 1.1471657752990723,
      "learning_rate": 0.0001528252530736567,
      "loss": 1.5288,
      "step": 2396
    },
    {
      "epoch": 0.3433605500644607,
      "grad_norm": 1.1633987426757812,
      "learning_rate": 0.00015278585163569107,
      "loss": 1.3403,
      "step": 2397
    },
    {
      "epoch": 0.3435037960177625,
      "grad_norm": 0.9312422275543213,
      "learning_rate": 0.0001527464388342379,
      "loss": 1.497,
      "step": 2398
    },
    {
      "epoch": 0.3436470419710643,
      "grad_norm": 1.208841323852539,
      "learning_rate": 0.00015270701467778167,
      "loss": 1.6432,
      "step": 2399
    },
    {
      "epoch": 0.3437902879243661,
      "grad_norm": 1.2214096784591675,
      "learning_rate": 0.00015266757917480948,
      "loss": 1.7182,
      "step": 2400
    },
    {
      "epoch": 0.34393353387766795,
      "grad_norm": 1.165998935699463,
      "learning_rate": 0.00015262813233381082,
      "loss": 1.5253,
      "step": 2401
    },
    {
      "epoch": 0.3440767798309698,
      "grad_norm": 1.301129937171936,
      "learning_rate": 0.00015258867416327758,
      "loss": 1.6513,
      "step": 2402
    },
    {
      "epoch": 0.3442200257842716,
      "grad_norm": 1.3936134576797485,
      "learning_rate": 0.0001525492046717042,
      "loss": 1.342,
      "step": 2403
    },
    {
      "epoch": 0.3443632717375734,
      "grad_norm": 1.006881594657898,
      "learning_rate": 0.00015250972386758745,
      "loss": 1.3893,
      "step": 2404
    },
    {
      "epoch": 0.34450651769087526,
      "grad_norm": 1.1349140405654907,
      "learning_rate": 0.00015247023175942657,
      "loss": 1.5365,
      "step": 2405
    },
    {
      "epoch": 0.34464976364417704,
      "grad_norm": 1.3732683658599854,
      "learning_rate": 0.00015243072835572318,
      "loss": 1.5158,
      "step": 2406
    },
    {
      "epoch": 0.3447930095974789,
      "grad_norm": 1.371100664138794,
      "learning_rate": 0.00015239121366498147,
      "loss": 1.4272,
      "step": 2407
    },
    {
      "epoch": 0.34493625555078067,
      "grad_norm": 1.141140341758728,
      "learning_rate": 0.00015235168769570792,
      "loss": 1.4489,
      "step": 2408
    },
    {
      "epoch": 0.3450795015040825,
      "grad_norm": 1.2566807270050049,
      "learning_rate": 0.0001523121504564115,
      "loss": 1.3978,
      "step": 2409
    },
    {
      "epoch": 0.34522274745738435,
      "grad_norm": 1.0878827571868896,
      "learning_rate": 0.00015227260195560367,
      "loss": 1.4564,
      "step": 2410
    },
    {
      "epoch": 0.34536599341068613,
      "grad_norm": 1.0803250074386597,
      "learning_rate": 0.00015223304220179812,
      "loss": 1.5442,
      "step": 2411
    },
    {
      "epoch": 0.345509239363988,
      "grad_norm": 1.2518631219863892,
      "learning_rate": 0.00015219347120351123,
      "loss": 1.6102,
      "step": 2412
    },
    {
      "epoch": 0.34565248531728976,
      "grad_norm": 1.0459473133087158,
      "learning_rate": 0.00015215388896926152,
      "loss": 1.4808,
      "step": 2413
    },
    {
      "epoch": 0.3457957312705916,
      "grad_norm": 1.0294448137283325,
      "learning_rate": 0.00015211429550757012,
      "loss": 1.5121,
      "step": 2414
    },
    {
      "epoch": 0.34593897722389344,
      "grad_norm": 1.2770122289657593,
      "learning_rate": 0.00015207469082696053,
      "loss": 1.4725,
      "step": 2415
    },
    {
      "epoch": 0.3460822231771952,
      "grad_norm": 0.9999268651008606,
      "learning_rate": 0.00015203507493595866,
      "loss": 1.4866,
      "step": 2416
    },
    {
      "epoch": 0.34622546913049707,
      "grad_norm": 0.9365148544311523,
      "learning_rate": 0.00015199544784309286,
      "loss": 1.5976,
      "step": 2417
    },
    {
      "epoch": 0.3463687150837989,
      "grad_norm": 1.1144593954086304,
      "learning_rate": 0.00015195580955689378,
      "loss": 1.269,
      "step": 2418
    },
    {
      "epoch": 0.3465119610371007,
      "grad_norm": 1.1058090925216675,
      "learning_rate": 0.00015191616008589461,
      "loss": 1.4277,
      "step": 2419
    },
    {
      "epoch": 0.34665520699040253,
      "grad_norm": 0.8692079186439514,
      "learning_rate": 0.00015187649943863095,
      "loss": 1.6169,
      "step": 2420
    },
    {
      "epoch": 0.3467984529437043,
      "grad_norm": 1.1412423849105835,
      "learning_rate": 0.00015183682762364066,
      "loss": 1.477,
      "step": 2421
    },
    {
      "epoch": 0.34694169889700616,
      "grad_norm": 0.9700475931167603,
      "learning_rate": 0.0001517971446494641,
      "loss": 1.5336,
      "step": 2422
    },
    {
      "epoch": 0.347084944850308,
      "grad_norm": 1.2372982501983643,
      "learning_rate": 0.00015175745052464415,
      "loss": 1.4172,
      "step": 2423
    },
    {
      "epoch": 0.3472281908036098,
      "grad_norm": 1.3604100942611694,
      "learning_rate": 0.00015171774525772592,
      "loss": 1.4789,
      "step": 2424
    },
    {
      "epoch": 0.3473714367569116,
      "grad_norm": 1.130306601524353,
      "learning_rate": 0.00015167802885725687,
      "loss": 1.5878,
      "step": 2425
    },
    {
      "epoch": 0.34751468271021346,
      "grad_norm": 1.3036110401153564,
      "learning_rate": 0.0001516383013317871,
      "loss": 1.4351,
      "step": 2426
    },
    {
      "epoch": 0.34765792866351525,
      "grad_norm": 1.4934039115905762,
      "learning_rate": 0.00015159856268986887,
      "loss": 1.4764,
      "step": 2427
    },
    {
      "epoch": 0.3478011746168171,
      "grad_norm": 1.0250859260559082,
      "learning_rate": 0.00015155881294005697,
      "loss": 1.5109,
      "step": 2428
    },
    {
      "epoch": 0.3479444205701189,
      "grad_norm": 1.089347243309021,
      "learning_rate": 0.00015151905209090854,
      "loss": 1.3593,
      "step": 2429
    },
    {
      "epoch": 0.3480876665234207,
      "grad_norm": 1.2811094522476196,
      "learning_rate": 0.0001514792801509831,
      "loss": 1.4851,
      "step": 2430
    },
    {
      "epoch": 0.34823091247672255,
      "grad_norm": 1.446021318435669,
      "learning_rate": 0.00015143949712884252,
      "loss": 1.4453,
      "step": 2431
    },
    {
      "epoch": 0.34837415843002434,
      "grad_norm": 1.3093568086624146,
      "learning_rate": 0.00015139970303305119,
      "loss": 1.429,
      "step": 2432
    },
    {
      "epoch": 0.3485174043833262,
      "grad_norm": 1.2506701946258545,
      "learning_rate": 0.00015135989787217567,
      "loss": 1.2644,
      "step": 2433
    },
    {
      "epoch": 0.34866065033662796,
      "grad_norm": 1.0441100597381592,
      "learning_rate": 0.00015132008165478516,
      "loss": 1.3778,
      "step": 2434
    },
    {
      "epoch": 0.3488038962899298,
      "grad_norm": 0.9274334907531738,
      "learning_rate": 0.00015128025438945102,
      "loss": 1.5475,
      "step": 2435
    },
    {
      "epoch": 0.34894714224323164,
      "grad_norm": 1.1874804496765137,
      "learning_rate": 0.0001512404160847471,
      "loss": 1.5333,
      "step": 2436
    },
    {
      "epoch": 0.34909038819653343,
      "grad_norm": 1.101256251335144,
      "learning_rate": 0.0001512005667492496,
      "loss": 1.4267,
      "step": 2437
    },
    {
      "epoch": 0.34923363414983527,
      "grad_norm": 0.9447908997535706,
      "learning_rate": 0.0001511607063915371,
      "loss": 1.4612,
      "step": 2438
    },
    {
      "epoch": 0.3493768801031371,
      "grad_norm": 0.9819499254226685,
      "learning_rate": 0.00015112083502019056,
      "loss": 1.5346,
      "step": 2439
    },
    {
      "epoch": 0.3495201260564389,
      "grad_norm": 1.075096607208252,
      "learning_rate": 0.00015108095264379325,
      "loss": 1.6348,
      "step": 2440
    },
    {
      "epoch": 0.34966337200974074,
      "grad_norm": 1.1388366222381592,
      "learning_rate": 0.00015104105927093092,
      "loss": 1.2484,
      "step": 2441
    },
    {
      "epoch": 0.3498066179630425,
      "grad_norm": 1.578209400177002,
      "learning_rate": 0.0001510011549101916,
      "loss": 1.508,
      "step": 2442
    },
    {
      "epoch": 0.34994986391634436,
      "grad_norm": 1.0529425144195557,
      "learning_rate": 0.00015096123957016565,
      "loss": 1.5795,
      "step": 2443
    },
    {
      "epoch": 0.3500931098696462,
      "grad_norm": 1.1954271793365479,
      "learning_rate": 0.000150921313259446,
      "loss": 1.4587,
      "step": 2444
    },
    {
      "epoch": 0.350236355822948,
      "grad_norm": 1.1856555938720703,
      "learning_rate": 0.0001508813759866277,
      "loss": 1.4059,
      "step": 2445
    },
    {
      "epoch": 0.3503796017762498,
      "grad_norm": 1.0036766529083252,
      "learning_rate": 0.00015084142776030824,
      "loss": 1.4509,
      "step": 2446
    },
    {
      "epoch": 0.35052284772955167,
      "grad_norm": 0.9608394503593445,
      "learning_rate": 0.0001508014685890875,
      "loss": 1.4979,
      "step": 2447
    },
    {
      "epoch": 0.35066609368285345,
      "grad_norm": 1.2596564292907715,
      "learning_rate": 0.00015076149848156775,
      "loss": 1.6497,
      "step": 2448
    },
    {
      "epoch": 0.3508093396361553,
      "grad_norm": 0.9816955924034119,
      "learning_rate": 0.00015072151744635352,
      "loss": 1.3047,
      "step": 2449
    },
    {
      "epoch": 0.3509525855894571,
      "grad_norm": 1.047676682472229,
      "learning_rate": 0.00015068152549205173,
      "loss": 1.4998,
      "step": 2450
    },
    {
      "epoch": 0.3510958315427589,
      "grad_norm": 1.022405743598938,
      "learning_rate": 0.0001506415226272717,
      "loss": 1.4455,
      "step": 2451
    },
    {
      "epoch": 0.35123907749606076,
      "grad_norm": 1.0790386199951172,
      "learning_rate": 0.000150601508860625,
      "loss": 1.4453,
      "step": 2452
    },
    {
      "epoch": 0.35138232344936254,
      "grad_norm": 0.9402562379837036,
      "learning_rate": 0.00015056148420072564,
      "loss": 1.3687,
      "step": 2453
    },
    {
      "epoch": 0.3515255694026644,
      "grad_norm": 1.2030194997787476,
      "learning_rate": 0.00015052144865618995,
      "loss": 1.3809,
      "step": 2454
    },
    {
      "epoch": 0.35166881535596617,
      "grad_norm": 1.1288782358169556,
      "learning_rate": 0.0001504814022356366,
      "loss": 1.588,
      "step": 2455
    },
    {
      "epoch": 0.351812061309268,
      "grad_norm": 1.2761846780776978,
      "learning_rate": 0.0001504413449476865,
      "loss": 1.6743,
      "step": 2456
    },
    {
      "epoch": 0.35195530726256985,
      "grad_norm": 0.9686523079872131,
      "learning_rate": 0.00015040127680096313,
      "loss": 1.4878,
      "step": 2457
    },
    {
      "epoch": 0.35209855321587163,
      "grad_norm": 1.1332261562347412,
      "learning_rate": 0.00015036119780409207,
      "loss": 1.4049,
      "step": 2458
    },
    {
      "epoch": 0.3522417991691735,
      "grad_norm": 1.0721514225006104,
      "learning_rate": 0.00015032110796570137,
      "loss": 1.3662,
      "step": 2459
    },
    {
      "epoch": 0.3523850451224753,
      "grad_norm": 1.0312167406082153,
      "learning_rate": 0.00015028100729442138,
      "loss": 1.216,
      "step": 2460
    },
    {
      "epoch": 0.3525282910757771,
      "grad_norm": 1.0709669589996338,
      "learning_rate": 0.00015024089579888478,
      "loss": 1.3847,
      "step": 2461
    },
    {
      "epoch": 0.35267153702907894,
      "grad_norm": 1.0403696298599243,
      "learning_rate": 0.0001502007734877266,
      "loss": 1.5603,
      "step": 2462
    },
    {
      "epoch": 0.3528147829823807,
      "grad_norm": 1.136157512664795,
      "learning_rate": 0.00015016064036958414,
      "loss": 1.2572,
      "step": 2463
    },
    {
      "epoch": 0.35295802893568257,
      "grad_norm": 1.1979122161865234,
      "learning_rate": 0.00015012049645309712,
      "loss": 1.4414,
      "step": 2464
    },
    {
      "epoch": 0.3531012748889844,
      "grad_norm": 1.1076048612594604,
      "learning_rate": 0.0001500803417469075,
      "loss": 1.3954,
      "step": 2465
    },
    {
      "epoch": 0.3532445208422862,
      "grad_norm": 1.1115305423736572,
      "learning_rate": 0.00015004017625965958,
      "loss": 1.5678,
      "step": 2466
    },
    {
      "epoch": 0.35338776679558803,
      "grad_norm": 1.1326771974563599,
      "learning_rate": 0.00015000000000000001,
      "loss": 1.448,
      "step": 2467
    },
    {
      "epoch": 0.3535310127488898,
      "grad_norm": 1.0988236665725708,
      "learning_rate": 0.00014995981297657776,
      "loss": 1.5052,
      "step": 2468
    },
    {
      "epoch": 0.35367425870219166,
      "grad_norm": 1.1608054637908936,
      "learning_rate": 0.00014991961519804408,
      "loss": 1.4011,
      "step": 2469
    },
    {
      "epoch": 0.3538175046554935,
      "grad_norm": 1.0278706550598145,
      "learning_rate": 0.00014987940667305258,
      "loss": 1.6579,
      "step": 2470
    },
    {
      "epoch": 0.3539607506087953,
      "grad_norm": 0.9136257767677307,
      "learning_rate": 0.00014983918741025916,
      "loss": 1.2434,
      "step": 2471
    },
    {
      "epoch": 0.3541039965620971,
      "grad_norm": 1.0890923738479614,
      "learning_rate": 0.00014979895741832198,
      "loss": 1.3587,
      "step": 2472
    },
    {
      "epoch": 0.35424724251539896,
      "grad_norm": 1.4404780864715576,
      "learning_rate": 0.00014975871670590163,
      "loss": 1.6476,
      "step": 2473
    },
    {
      "epoch": 0.35439048846870075,
      "grad_norm": 1.3689135313034058,
      "learning_rate": 0.0001497184652816609,
      "loss": 1.5361,
      "step": 2474
    },
    {
      "epoch": 0.3545337344220026,
      "grad_norm": 1.3381685018539429,
      "learning_rate": 0.0001496782031542649,
      "loss": 1.4194,
      "step": 2475
    },
    {
      "epoch": 0.3546769803753044,
      "grad_norm": 1.2028056383132935,
      "learning_rate": 0.0001496379303323812,
      "loss": 1.6739,
      "step": 2476
    },
    {
      "epoch": 0.3548202263286062,
      "grad_norm": 1.3352667093276978,
      "learning_rate": 0.00014959764682467933,
      "loss": 1.4492,
      "step": 2477
    },
    {
      "epoch": 0.35496347228190805,
      "grad_norm": 1.2482444047927856,
      "learning_rate": 0.00014955735263983154,
      "loss": 1.569,
      "step": 2478
    },
    {
      "epoch": 0.35510671823520984,
      "grad_norm": 1.0980826616287231,
      "learning_rate": 0.00014951704778651202,
      "loss": 1.506,
      "step": 2479
    },
    {
      "epoch": 0.3552499641885117,
      "grad_norm": 1.3518425226211548,
      "learning_rate": 0.00014947673227339755,
      "loss": 1.453,
      "step": 2480
    },
    {
      "epoch": 0.3553932101418135,
      "grad_norm": 1.0751659870147705,
      "learning_rate": 0.00014943640610916688,
      "loss": 1.468,
      "step": 2481
    },
    {
      "epoch": 0.3555364560951153,
      "grad_norm": 1.1745363473892212,
      "learning_rate": 0.00014939606930250142,
      "loss": 1.3486,
      "step": 2482
    },
    {
      "epoch": 0.35567970204841715,
      "grad_norm": 0.9499922394752502,
      "learning_rate": 0.00014935572186208456,
      "loss": 1.3732,
      "step": 2483
    },
    {
      "epoch": 0.35582294800171893,
      "grad_norm": 1.098778486251831,
      "learning_rate": 0.00014931536379660213,
      "loss": 1.611,
      "step": 2484
    },
    {
      "epoch": 0.35596619395502077,
      "grad_norm": 1.059660792350769,
      "learning_rate": 0.00014927499511474228,
      "loss": 1.4004,
      "step": 2485
    },
    {
      "epoch": 0.3561094399083226,
      "grad_norm": 1.108597993850708,
      "learning_rate": 0.00014923461582519532,
      "loss": 1.4137,
      "step": 2486
    },
    {
      "epoch": 0.3562526858616244,
      "grad_norm": 1.218572974205017,
      "learning_rate": 0.00014919422593665397,
      "loss": 1.3978,
      "step": 2487
    },
    {
      "epoch": 0.35639593181492624,
      "grad_norm": 1.0528897047042847,
      "learning_rate": 0.00014915382545781315,
      "loss": 1.59,
      "step": 2488
    },
    {
      "epoch": 0.356539177768228,
      "grad_norm": 1.4149610996246338,
      "learning_rate": 0.00014911341439737002,
      "loss": 1.3505,
      "step": 2489
    },
    {
      "epoch": 0.35668242372152986,
      "grad_norm": 1.1221262216567993,
      "learning_rate": 0.00014907299276402418,
      "loss": 1.5635,
      "step": 2490
    },
    {
      "epoch": 0.3568256696748317,
      "grad_norm": 1.222782850265503,
      "learning_rate": 0.00014903256056647736,
      "loss": 1.4488,
      "step": 2491
    },
    {
      "epoch": 0.3569689156281335,
      "grad_norm": 1.1195868253707886,
      "learning_rate": 0.0001489921178134336,
      "loss": 1.5141,
      "step": 2492
    },
    {
      "epoch": 0.35711216158143533,
      "grad_norm": 1.2428107261657715,
      "learning_rate": 0.0001489516645135993,
      "loss": 1.4349,
      "step": 2493
    },
    {
      "epoch": 0.35725540753473717,
      "grad_norm": 1.0416040420532227,
      "learning_rate": 0.00014891120067568294,
      "loss": 1.4898,
      "step": 2494
    },
    {
      "epoch": 0.35739865348803895,
      "grad_norm": 1.1553033590316772,
      "learning_rate": 0.00014887072630839546,
      "loss": 1.457,
      "step": 2495
    },
    {
      "epoch": 0.3575418994413408,
      "grad_norm": 0.9414494037628174,
      "learning_rate": 0.00014883024142044995,
      "loss": 1.5649,
      "step": 2496
    },
    {
      "epoch": 0.3576851453946426,
      "grad_norm": 1.128084659576416,
      "learning_rate": 0.00014878974602056181,
      "loss": 1.3616,
      "step": 2497
    },
    {
      "epoch": 0.3578283913479444,
      "grad_norm": 1.351125717163086,
      "learning_rate": 0.00014874924011744876,
      "loss": 1.3059,
      "step": 2498
    },
    {
      "epoch": 0.35797163730124626,
      "grad_norm": 1.060027837753296,
      "learning_rate": 0.00014870872371983062,
      "loss": 1.4246,
      "step": 2499
    },
    {
      "epoch": 0.35811488325454804,
      "grad_norm": 1.2175381183624268,
      "learning_rate": 0.00014866819683642966,
      "loss": 1.391,
      "step": 2500
    },
    {
      "epoch": 0.3582581292078499,
      "grad_norm": 1.1252983808517456,
      "learning_rate": 0.00014862765947597025,
      "loss": 1.3124,
      "step": 2501
    },
    {
      "epoch": 0.3584013751611517,
      "grad_norm": 1.0511078834533691,
      "learning_rate": 0.00014858711164717912,
      "loss": 1.5625,
      "step": 2502
    },
    {
      "epoch": 0.3585446211144535,
      "grad_norm": 1.038171648979187,
      "learning_rate": 0.00014854655335878517,
      "loss": 1.5304,
      "step": 2503
    },
    {
      "epoch": 0.35868786706775535,
      "grad_norm": 1.052503228187561,
      "learning_rate": 0.00014850598461951963,
      "loss": 1.3118,
      "step": 2504
    },
    {
      "epoch": 0.35883111302105714,
      "grad_norm": 1.0193835496902466,
      "learning_rate": 0.00014846540543811596,
      "loss": 1.4586,
      "step": 2505
    },
    {
      "epoch": 0.358974358974359,
      "grad_norm": 1.1015459299087524,
      "learning_rate": 0.0001484248158233098,
      "loss": 1.3647,
      "step": 2506
    },
    {
      "epoch": 0.3591176049276608,
      "grad_norm": 1.1612577438354492,
      "learning_rate": 0.00014838421578383914,
      "loss": 1.4625,
      "step": 2507
    },
    {
      "epoch": 0.3592608508809626,
      "grad_norm": 1.0630440711975098,
      "learning_rate": 0.00014834360532844413,
      "loss": 1.3518,
      "step": 2508
    },
    {
      "epoch": 0.35940409683426444,
      "grad_norm": 0.9513444304466248,
      "learning_rate": 0.0001483029844658672,
      "loss": 1.4603,
      "step": 2509
    },
    {
      "epoch": 0.3595473427875662,
      "grad_norm": 1.243257761001587,
      "learning_rate": 0.00014826235320485306,
      "loss": 1.4823,
      "step": 2510
    },
    {
      "epoch": 0.35969058874086807,
      "grad_norm": 0.9204735159873962,
      "learning_rate": 0.00014822171155414856,
      "loss": 1.585,
      "step": 2511
    },
    {
      "epoch": 0.3598338346941699,
      "grad_norm": 1.1022237539291382,
      "learning_rate": 0.00014818105952250292,
      "loss": 1.3876,
      "step": 2512
    },
    {
      "epoch": 0.3599770806474717,
      "grad_norm": 1.1986637115478516,
      "learning_rate": 0.0001481403971186674,
      "loss": 1.3128,
      "step": 2513
    },
    {
      "epoch": 0.36012032660077353,
      "grad_norm": 1.3053021430969238,
      "learning_rate": 0.00014809972435139568,
      "loss": 1.4194,
      "step": 2514
    },
    {
      "epoch": 0.3602635725540754,
      "grad_norm": 1.248697280883789,
      "learning_rate": 0.0001480590412294436,
      "loss": 1.2864,
      "step": 2515
    },
    {
      "epoch": 0.36040681850737716,
      "grad_norm": 0.9447510242462158,
      "learning_rate": 0.00014801834776156925,
      "loss": 1.6017,
      "step": 2516
    },
    {
      "epoch": 0.360550064460679,
      "grad_norm": 1.0043679475784302,
      "learning_rate": 0.00014797764395653283,
      "loss": 1.4786,
      "step": 2517
    },
    {
      "epoch": 0.3606933104139808,
      "grad_norm": 1.1880528926849365,
      "learning_rate": 0.000147936929823097,
      "loss": 1.3449,
      "step": 2518
    },
    {
      "epoch": 0.3608365563672826,
      "grad_norm": 1.0238244533538818,
      "learning_rate": 0.00014789620537002639,
      "loss": 1.4868,
      "step": 2519
    },
    {
      "epoch": 0.36097980232058446,
      "grad_norm": 1.4210186004638672,
      "learning_rate": 0.000147855470606088,
      "loss": 1.5587,
      "step": 2520
    },
    {
      "epoch": 0.36112304827388625,
      "grad_norm": 1.2720956802368164,
      "learning_rate": 0.00014781472554005107,
      "loss": 1.4834,
      "step": 2521
    },
    {
      "epoch": 0.3612662942271881,
      "grad_norm": 1.132568597793579,
      "learning_rate": 0.0001477739701806869,
      "loss": 1.4971,
      "step": 2522
    },
    {
      "epoch": 0.3614095401804899,
      "grad_norm": 0.9899293780326843,
      "learning_rate": 0.00014773320453676924,
      "loss": 1.3419,
      "step": 2523
    },
    {
      "epoch": 0.3615527861337917,
      "grad_norm": 1.2437946796417236,
      "learning_rate": 0.00014769242861707382,
      "loss": 1.4447,
      "step": 2524
    },
    {
      "epoch": 0.36169603208709356,
      "grad_norm": 1.1225162744522095,
      "learning_rate": 0.00014765164243037875,
      "loss": 1.4871,
      "step": 2525
    },
    {
      "epoch": 0.36183927804039534,
      "grad_norm": 1.014938473701477,
      "learning_rate": 0.0001476108459854642,
      "loss": 1.4887,
      "step": 2526
    },
    {
      "epoch": 0.3619825239936972,
      "grad_norm": 1.2254246473312378,
      "learning_rate": 0.00014757003929111276,
      "loss": 1.4382,
      "step": 2527
    },
    {
      "epoch": 0.362125769946999,
      "grad_norm": 1.2509281635284424,
      "learning_rate": 0.000147529222356109,
      "loss": 1.4276,
      "step": 2528
    },
    {
      "epoch": 0.3622690159003008,
      "grad_norm": 1.1786195039749146,
      "learning_rate": 0.00014748839518923985,
      "loss": 1.4581,
      "step": 2529
    },
    {
      "epoch": 0.36241226185360265,
      "grad_norm": 1.2477809190750122,
      "learning_rate": 0.00014744755779929437,
      "loss": 1.5629,
      "step": 2530
    },
    {
      "epoch": 0.36255550780690443,
      "grad_norm": 1.0423568487167358,
      "learning_rate": 0.00014740671019506383,
      "loss": 1.4088,
      "step": 2531
    },
    {
      "epoch": 0.36269875376020627,
      "grad_norm": 0.9987547397613525,
      "learning_rate": 0.00014736585238534172,
      "loss": 1.5414,
      "step": 2532
    },
    {
      "epoch": 0.3628419997135081,
      "grad_norm": 0.8615167737007141,
      "learning_rate": 0.00014732498437892373,
      "loss": 1.5009,
      "step": 2533
    },
    {
      "epoch": 0.3629852456668099,
      "grad_norm": 1.2286875247955322,
      "learning_rate": 0.0001472841061846077,
      "loss": 1.4962,
      "step": 2534
    },
    {
      "epoch": 0.36312849162011174,
      "grad_norm": 0.9477217197418213,
      "learning_rate": 0.0001472432178111937,
      "loss": 1.4642,
      "step": 2535
    },
    {
      "epoch": 0.3632717375734136,
      "grad_norm": 1.3960607051849365,
      "learning_rate": 0.000147202319267484,
      "loss": 1.5182,
      "step": 2536
    },
    {
      "epoch": 0.36341498352671536,
      "grad_norm": 1.061038851737976,
      "learning_rate": 0.00014716141056228305,
      "loss": 1.495,
      "step": 2537
    },
    {
      "epoch": 0.3635582294800172,
      "grad_norm": 1.4418344497680664,
      "learning_rate": 0.00014712049170439748,
      "loss": 1.2843,
      "step": 2538
    },
    {
      "epoch": 0.363701475433319,
      "grad_norm": 1.023033857345581,
      "learning_rate": 0.00014707956270263605,
      "loss": 1.4014,
      "step": 2539
    },
    {
      "epoch": 0.36384472138662083,
      "grad_norm": 1.067755937576294,
      "learning_rate": 0.00014703862356580985,
      "loss": 1.4649,
      "step": 2540
    },
    {
      "epoch": 0.36398796733992267,
      "grad_norm": 1.038650631904602,
      "learning_rate": 0.000146997674302732,
      "loss": 1.4716,
      "step": 2541
    },
    {
      "epoch": 0.36413121329322445,
      "grad_norm": 1.2082433700561523,
      "learning_rate": 0.00014695671492221792,
      "loss": 1.4394,
      "step": 2542
    },
    {
      "epoch": 0.3642744592465263,
      "grad_norm": 0.9982503652572632,
      "learning_rate": 0.0001469157454330851,
      "loss": 1.5247,
      "step": 2543
    },
    {
      "epoch": 0.3644177051998281,
      "grad_norm": 0.9793524146080017,
      "learning_rate": 0.00014687476584415325,
      "loss": 1.3516,
      "step": 2544
    },
    {
      "epoch": 0.3645609511531299,
      "grad_norm": 1.0603388547897339,
      "learning_rate": 0.00014683377616424428,
      "loss": 1.6374,
      "step": 2545
    },
    {
      "epoch": 0.36470419710643176,
      "grad_norm": 0.996548056602478,
      "learning_rate": 0.0001467927764021823,
      "loss": 1.3796,
      "step": 2546
    },
    {
      "epoch": 0.36484744305973354,
      "grad_norm": 1.0186673402786255,
      "learning_rate": 0.00014675176656679345,
      "loss": 1.3512,
      "step": 2547
    },
    {
      "epoch": 0.3649906890130354,
      "grad_norm": 1.0116450786590576,
      "learning_rate": 0.0001467107466669062,
      "loss": 1.4934,
      "step": 2548
    },
    {
      "epoch": 0.3651339349663372,
      "grad_norm": 1.010191559791565,
      "learning_rate": 0.00014666971671135112,
      "loss": 1.3588,
      "step": 2549
    },
    {
      "epoch": 0.365277180919639,
      "grad_norm": 1.325577974319458,
      "learning_rate": 0.00014662867670896094,
      "loss": 1.3208,
      "step": 2550
    },
    {
      "epoch": 0.36542042687294085,
      "grad_norm": 1.0123873949050903,
      "learning_rate": 0.00014658762666857052,
      "loss": 1.5638,
      "step": 2551
    },
    {
      "epoch": 0.36556367282624264,
      "grad_norm": 1.0866594314575195,
      "learning_rate": 0.00014654656659901695,
      "loss": 1.4287,
      "step": 2552
    },
    {
      "epoch": 0.3657069187795445,
      "grad_norm": 1.2401455640792847,
      "learning_rate": 0.00014650549650913945,
      "loss": 1.4929,
      "step": 2553
    },
    {
      "epoch": 0.3658501647328463,
      "grad_norm": 1.2523385286331177,
      "learning_rate": 0.00014646441640777936,
      "loss": 1.3921,
      "step": 2554
    },
    {
      "epoch": 0.3659934106861481,
      "grad_norm": 1.0730689764022827,
      "learning_rate": 0.0001464233263037803,
      "loss": 1.4909,
      "step": 2555
    },
    {
      "epoch": 0.36613665663944994,
      "grad_norm": 1.4879099130630493,
      "learning_rate": 0.00014638222620598777,
      "loss": 1.4058,
      "step": 2556
    },
    {
      "epoch": 0.3662799025927517,
      "grad_norm": 1.0155384540557861,
      "learning_rate": 0.00014634111612324982,
      "loss": 1.4914,
      "step": 2557
    },
    {
      "epoch": 0.36642314854605357,
      "grad_norm": 1.0073775053024292,
      "learning_rate": 0.0001462999960644163,
      "loss": 1.5356,
      "step": 2558
    },
    {
      "epoch": 0.3665663944993554,
      "grad_norm": 0.9533057808876038,
      "learning_rate": 0.00014625886603833937,
      "loss": 1.4409,
      "step": 2559
    },
    {
      "epoch": 0.3667096404526572,
      "grad_norm": 1.0768482685089111,
      "learning_rate": 0.0001462177260538733,
      "loss": 1.7473,
      "step": 2560
    },
    {
      "epoch": 0.36685288640595903,
      "grad_norm": 1.1366655826568604,
      "learning_rate": 0.00014617657611987455,
      "loss": 1.3921,
      "step": 2561
    },
    {
      "epoch": 0.3669961323592609,
      "grad_norm": 1.1692472696304321,
      "learning_rate": 0.00014613541624520165,
      "loss": 1.4196,
      "step": 2562
    },
    {
      "epoch": 0.36713937831256266,
      "grad_norm": 0.9287104606628418,
      "learning_rate": 0.0001460942464387153,
      "loss": 1.4302,
      "step": 2563
    },
    {
      "epoch": 0.3672826242658645,
      "grad_norm": 1.336983561515808,
      "learning_rate": 0.0001460530667092783,
      "loss": 1.3601,
      "step": 2564
    },
    {
      "epoch": 0.3674258702191663,
      "grad_norm": 1.1765611171722412,
      "learning_rate": 0.00014601187706575572,
      "loss": 1.4402,
      "step": 2565
    },
    {
      "epoch": 0.3675691161724681,
      "grad_norm": 1.1721441745758057,
      "learning_rate": 0.00014597067751701465,
      "loss": 1.5211,
      "step": 2566
    },
    {
      "epoch": 0.36771236212576996,
      "grad_norm": 1.1252217292785645,
      "learning_rate": 0.00014592946807192426,
      "loss": 1.548,
      "step": 2567
    },
    {
      "epoch": 0.36785560807907175,
      "grad_norm": 1.0455622673034668,
      "learning_rate": 0.000145888248739356,
      "loss": 1.4642,
      "step": 2568
    },
    {
      "epoch": 0.3679988540323736,
      "grad_norm": 0.9194879531860352,
      "learning_rate": 0.00014584701952818333,
      "loss": 1.4646,
      "step": 2569
    },
    {
      "epoch": 0.36814209998567543,
      "grad_norm": 1.1840286254882812,
      "learning_rate": 0.00014580578044728188,
      "loss": 1.431,
      "step": 2570
    },
    {
      "epoch": 0.3682853459389772,
      "grad_norm": 1.011466383934021,
      "learning_rate": 0.00014576453150552947,
      "loss": 1.5775,
      "step": 2571
    },
    {
      "epoch": 0.36842859189227906,
      "grad_norm": 1.0589476823806763,
      "learning_rate": 0.00014572327271180586,
      "loss": 1.2801,
      "step": 2572
    },
    {
      "epoch": 0.36857183784558084,
      "grad_norm": 1.0883053541183472,
      "learning_rate": 0.00014568200407499314,
      "loss": 1.4236,
      "step": 2573
    },
    {
      "epoch": 0.3687150837988827,
      "grad_norm": 1.0136991739273071,
      "learning_rate": 0.00014564072560397542,
      "loss": 1.4541,
      "step": 2574
    },
    {
      "epoch": 0.3688583297521845,
      "grad_norm": 1.1944260597229004,
      "learning_rate": 0.00014559943730763889,
      "loss": 1.5655,
      "step": 2575
    },
    {
      "epoch": 0.3690015757054863,
      "grad_norm": 1.1560734510421753,
      "learning_rate": 0.0001455581391948719,
      "loss": 1.425,
      "step": 2576
    },
    {
      "epoch": 0.36914482165878815,
      "grad_norm": 1.2052932977676392,
      "learning_rate": 0.00014551683127456494,
      "loss": 1.474,
      "step": 2577
    },
    {
      "epoch": 0.36928806761208993,
      "grad_norm": 0.9979791641235352,
      "learning_rate": 0.0001454755135556106,
      "loss": 1.3691,
      "step": 2578
    },
    {
      "epoch": 0.3694313135653918,
      "grad_norm": 1.409983515739441,
      "learning_rate": 0.0001454341860469035,
      "loss": 1.3881,
      "step": 2579
    },
    {
      "epoch": 0.3695745595186936,
      "grad_norm": 1.2113723754882812,
      "learning_rate": 0.0001453928487573405,
      "loss": 1.4127,
      "step": 2580
    },
    {
      "epoch": 0.3697178054719954,
      "grad_norm": 0.9448793530464172,
      "learning_rate": 0.0001453515016958204,
      "loss": 1.3904,
      "step": 2581
    },
    {
      "epoch": 0.36986105142529724,
      "grad_norm": 1.1924959421157837,
      "learning_rate": 0.00014531014487124432,
      "loss": 1.3369,
      "step": 2582
    },
    {
      "epoch": 0.3700042973785991,
      "grad_norm": 1.3233561515808105,
      "learning_rate": 0.00014526877829251528,
      "loss": 1.4517,
      "step": 2583
    },
    {
      "epoch": 0.37014754333190086,
      "grad_norm": 1.2325234413146973,
      "learning_rate": 0.00014522740196853853,
      "loss": 1.4014,
      "step": 2584
    },
    {
      "epoch": 0.3702907892852027,
      "grad_norm": 1.2031364440917969,
      "learning_rate": 0.0001451860159082213,
      "loss": 1.4983,
      "step": 2585
    },
    {
      "epoch": 0.3704340352385045,
      "grad_norm": 1.0210764408111572,
      "learning_rate": 0.00014514462012047306,
      "loss": 1.4157,
      "step": 2586
    },
    {
      "epoch": 0.37057728119180633,
      "grad_norm": 0.9976498484611511,
      "learning_rate": 0.00014510321461420523,
      "loss": 1.5828,
      "step": 2587
    },
    {
      "epoch": 0.37072052714510817,
      "grad_norm": 1.2574971914291382,
      "learning_rate": 0.00014506179939833142,
      "loss": 1.521,
      "step": 2588
    },
    {
      "epoch": 0.37086377309840995,
      "grad_norm": 0.9519497156143188,
      "learning_rate": 0.00014502037448176734,
      "loss": 1.6336,
      "step": 2589
    },
    {
      "epoch": 0.3710070190517118,
      "grad_norm": 0.994592547416687,
      "learning_rate": 0.0001449789398734307,
      "loss": 1.2408,
      "step": 2590
    },
    {
      "epoch": 0.37115026500501364,
      "grad_norm": 1.2184882164001465,
      "learning_rate": 0.00014493749558224138,
      "loss": 1.4688,
      "step": 2591
    },
    {
      "epoch": 0.3712935109583154,
      "grad_norm": 1.0672271251678467,
      "learning_rate": 0.00014489604161712128,
      "loss": 1.4611,
      "step": 2592
    },
    {
      "epoch": 0.37143675691161726,
      "grad_norm": 1.1550872325897217,
      "learning_rate": 0.0001448545779869944,
      "loss": 1.3873,
      "step": 2593
    },
    {
      "epoch": 0.37158000286491905,
      "grad_norm": 1.1462234258651733,
      "learning_rate": 0.00014481310470078687,
      "loss": 1.5359,
      "step": 2594
    },
    {
      "epoch": 0.3717232488182209,
      "grad_norm": 1.1339164972305298,
      "learning_rate": 0.00014477162176742688,
      "loss": 1.4471,
      "step": 2595
    },
    {
      "epoch": 0.3718664947715227,
      "grad_norm": 1.2721680402755737,
      "learning_rate": 0.00014473012919584462,
      "loss": 1.4699,
      "step": 2596
    },
    {
      "epoch": 0.3720097407248245,
      "grad_norm": 1.2865521907806396,
      "learning_rate": 0.00014468862699497243,
      "loss": 1.7196,
      "step": 2597
    },
    {
      "epoch": 0.37215298667812635,
      "grad_norm": 0.9609204530715942,
      "learning_rate": 0.00014464711517374475,
      "loss": 1.4763,
      "step": 2598
    },
    {
      "epoch": 0.37229623263142814,
      "grad_norm": 1.0643188953399658,
      "learning_rate": 0.000144605593741098,
      "loss": 1.4065,
      "step": 2599
    },
    {
      "epoch": 0.37243947858473,
      "grad_norm": 1.081160545349121,
      "learning_rate": 0.00014456406270597073,
      "loss": 1.4513,
      "step": 2600
    },
    {
      "epoch": 0.3725827245380318,
      "grad_norm": 1.2300240993499756,
      "learning_rate": 0.00014452252207730354,
      "loss": 1.5522,
      "step": 2601
    },
    {
      "epoch": 0.3727259704913336,
      "grad_norm": 1.0324400663375854,
      "learning_rate": 0.00014448097186403914,
      "loss": 1.3137,
      "step": 2602
    },
    {
      "epoch": 0.37286921644463544,
      "grad_norm": 1.4336881637573242,
      "learning_rate": 0.0001444394120751222,
      "loss": 1.4175,
      "step": 2603
    },
    {
      "epoch": 0.3730124623979373,
      "grad_norm": 1.417158603668213,
      "learning_rate": 0.0001443978427194996,
      "loss": 1.4492,
      "step": 2604
    },
    {
      "epoch": 0.37315570835123907,
      "grad_norm": 1.1301422119140625,
      "learning_rate": 0.0001443562638061201,
      "loss": 1.3633,
      "step": 2605
    },
    {
      "epoch": 0.3732989543045409,
      "grad_norm": 1.3131352663040161,
      "learning_rate": 0.00014431467534393463,
      "loss": 1.315,
      "step": 2606
    },
    {
      "epoch": 0.3734422002578427,
      "grad_norm": 1.1095761060714722,
      "learning_rate": 0.0001442730773418962,
      "loss": 1.3941,
      "step": 2607
    },
    {
      "epoch": 0.37358544621114453,
      "grad_norm": 0.9951792359352112,
      "learning_rate": 0.0001442314698089598,
      "loss": 1.5122,
      "step": 2608
    },
    {
      "epoch": 0.3737286921644464,
      "grad_norm": 1.4373977184295654,
      "learning_rate": 0.00014418985275408254,
      "loss": 1.3482,
      "step": 2609
    },
    {
      "epoch": 0.37387193811774816,
      "grad_norm": 1.4116655588150024,
      "learning_rate": 0.00014414822618622345,
      "loss": 1.2703,
      "step": 2610
    },
    {
      "epoch": 0.37401518407105,
      "grad_norm": 1.2531754970550537,
      "learning_rate": 0.00014410659011434383,
      "loss": 1.5549,
      "step": 2611
    },
    {
      "epoch": 0.3741584300243518,
      "grad_norm": 1.1813033819198608,
      "learning_rate": 0.00014406494454740677,
      "loss": 1.444,
      "step": 2612
    },
    {
      "epoch": 0.3743016759776536,
      "grad_norm": 1.1966707706451416,
      "learning_rate": 0.0001440232894943776,
      "loss": 1.4333,
      "step": 2613
    },
    {
      "epoch": 0.37444492193095547,
      "grad_norm": 1.2637412548065186,
      "learning_rate": 0.00014398162496422363,
      "loss": 1.3979,
      "step": 2614
    },
    {
      "epoch": 0.37458816788425725,
      "grad_norm": 1.1271389722824097,
      "learning_rate": 0.00014393995096591416,
      "loss": 1.3936,
      "step": 2615
    },
    {
      "epoch": 0.3747314138375591,
      "grad_norm": 1.082987666130066,
      "learning_rate": 0.0001438982675084206,
      "loss": 1.5254,
      "step": 2616
    },
    {
      "epoch": 0.37487465979086093,
      "grad_norm": 1.1699860095977783,
      "learning_rate": 0.00014385657460071639,
      "loss": 1.3008,
      "step": 2617
    },
    {
      "epoch": 0.3750179057441627,
      "grad_norm": 1.2353825569152832,
      "learning_rate": 0.0001438148722517769,
      "loss": 1.6122,
      "step": 2618
    },
    {
      "epoch": 0.37516115169746456,
      "grad_norm": 1.0600746870040894,
      "learning_rate": 0.0001437731604705797,
      "loss": 1.438,
      "step": 2619
    },
    {
      "epoch": 0.37530439765076634,
      "grad_norm": 1.2206144332885742,
      "learning_rate": 0.00014373143926610425,
      "loss": 1.2056,
      "step": 2620
    },
    {
      "epoch": 0.3754476436040682,
      "grad_norm": 1.0833669900894165,
      "learning_rate": 0.0001436897086473321,
      "loss": 1.3367,
      "step": 2621
    },
    {
      "epoch": 0.37559088955737,
      "grad_norm": 1.298516869544983,
      "learning_rate": 0.00014364796862324685,
      "loss": 1.5758,
      "step": 2622
    },
    {
      "epoch": 0.3757341355106718,
      "grad_norm": 1.0810481309890747,
      "learning_rate": 0.00014360621920283406,
      "loss": 1.4324,
      "step": 2623
    },
    {
      "epoch": 0.37587738146397365,
      "grad_norm": 1.0148545503616333,
      "learning_rate": 0.00014356446039508138,
      "loss": 1.5731,
      "step": 2624
    },
    {
      "epoch": 0.3760206274172755,
      "grad_norm": 1.1309711933135986,
      "learning_rate": 0.00014352269220897844,
      "loss": 1.3841,
      "step": 2625
    },
    {
      "epoch": 0.3761638733705773,
      "grad_norm": 1.2754181623458862,
      "learning_rate": 0.00014348091465351683,
      "loss": 1.4201,
      "step": 2626
    },
    {
      "epoch": 0.3763071193238791,
      "grad_norm": 1.0657199621200562,
      "learning_rate": 0.00014343912773769036,
      "loss": 1.477,
      "step": 2627
    },
    {
      "epoch": 0.3764503652771809,
      "grad_norm": 1.155584454536438,
      "learning_rate": 0.00014339733147049458,
      "loss": 1.559,
      "step": 2628
    },
    {
      "epoch": 0.37659361123048274,
      "grad_norm": 0.9936962723731995,
      "learning_rate": 0.0001433555258609273,
      "loss": 1.5949,
      "step": 2629
    },
    {
      "epoch": 0.3767368571837846,
      "grad_norm": 1.0808379650115967,
      "learning_rate": 0.0001433137109179881,
      "loss": 1.643,
      "step": 2630
    },
    {
      "epoch": 0.37688010313708636,
      "grad_norm": 1.1006921529769897,
      "learning_rate": 0.00014327188665067887,
      "loss": 1.4214,
      "step": 2631
    },
    {
      "epoch": 0.3770233490903882,
      "grad_norm": 1.2167447805404663,
      "learning_rate": 0.00014323005306800322,
      "loss": 1.5325,
      "step": 2632
    },
    {
      "epoch": 0.37716659504369,
      "grad_norm": 1.020889163017273,
      "learning_rate": 0.00014318821017896693,
      "loss": 1.4068,
      "step": 2633
    },
    {
      "epoch": 0.37730984099699183,
      "grad_norm": 1.093817949295044,
      "learning_rate": 0.00014314635799257775,
      "loss": 1.5973,
      "step": 2634
    },
    {
      "epoch": 0.37745308695029367,
      "grad_norm": 1.0956398248672485,
      "learning_rate": 0.00014310449651784536,
      "loss": 1.4462,
      "step": 2635
    },
    {
      "epoch": 0.37759633290359546,
      "grad_norm": 0.9416504502296448,
      "learning_rate": 0.00014306262576378157,
      "loss": 1.5593,
      "step": 2636
    },
    {
      "epoch": 0.3777395788568973,
      "grad_norm": 1.121124267578125,
      "learning_rate": 0.00014302074573940008,
      "loss": 1.3173,
      "step": 2637
    },
    {
      "epoch": 0.37788282481019914,
      "grad_norm": 1.0044233798980713,
      "learning_rate": 0.00014297885645371663,
      "loss": 1.6411,
      "step": 2638
    },
    {
      "epoch": 0.3780260707635009,
      "grad_norm": 0.9457429051399231,
      "learning_rate": 0.00014293695791574895,
      "loss": 1.5112,
      "step": 2639
    },
    {
      "epoch": 0.37816931671680276,
      "grad_norm": 1.0443371534347534,
      "learning_rate": 0.00014289505013451677,
      "loss": 1.4621,
      "step": 2640
    },
    {
      "epoch": 0.37831256267010455,
      "grad_norm": 1.0153977870941162,
      "learning_rate": 0.00014285313311904177,
      "loss": 1.472,
      "step": 2641
    },
    {
      "epoch": 0.3784558086234064,
      "grad_norm": 1.3298951387405396,
      "learning_rate": 0.00014281120687834764,
      "loss": 1.445,
      "step": 2642
    },
    {
      "epoch": 0.3785990545767082,
      "grad_norm": 1.1219143867492676,
      "learning_rate": 0.0001427692714214601,
      "loss": 1.3953,
      "step": 2643
    },
    {
      "epoch": 0.37874230053001,
      "grad_norm": 1.1074706315994263,
      "learning_rate": 0.0001427273267574068,
      "loss": 1.4122,
      "step": 2644
    },
    {
      "epoch": 0.37888554648331185,
      "grad_norm": 1.0301778316497803,
      "learning_rate": 0.0001426853728952174,
      "loss": 1.3201,
      "step": 2645
    },
    {
      "epoch": 0.3790287924366137,
      "grad_norm": 1.1880346536636353,
      "learning_rate": 0.0001426434098439235,
      "loss": 1.4723,
      "step": 2646
    },
    {
      "epoch": 0.3791720383899155,
      "grad_norm": 1.14169180393219,
      "learning_rate": 0.0001426014376125587,
      "loss": 1.4408,
      "step": 2647
    },
    {
      "epoch": 0.3793152843432173,
      "grad_norm": 1.1174323558807373,
      "learning_rate": 0.00014255945621015863,
      "loss": 1.5359,
      "step": 2648
    },
    {
      "epoch": 0.3794585302965191,
      "grad_norm": 1.0801656246185303,
      "learning_rate": 0.00014251746564576082,
      "loss": 1.4907,
      "step": 2649
    },
    {
      "epoch": 0.37960177624982094,
      "grad_norm": 0.8966488242149353,
      "learning_rate": 0.0001424754659284048,
      "loss": 1.3018,
      "step": 2650
    },
    {
      "epoch": 0.3797450222031228,
      "grad_norm": 1.1092829704284668,
      "learning_rate": 0.00014243345706713205,
      "loss": 1.3202,
      "step": 2651
    },
    {
      "epoch": 0.37988826815642457,
      "grad_norm": 1.1917495727539062,
      "learning_rate": 0.0001423914390709861,
      "loss": 1.4054,
      "step": 2652
    },
    {
      "epoch": 0.3800315141097264,
      "grad_norm": 0.9716973304748535,
      "learning_rate": 0.0001423494119490123,
      "loss": 1.5545,
      "step": 2653
    },
    {
      "epoch": 0.3801747600630282,
      "grad_norm": 0.965319037437439,
      "learning_rate": 0.00014230737571025812,
      "loss": 1.4721,
      "step": 2654
    },
    {
      "epoch": 0.38031800601633003,
      "grad_norm": 1.1608227491378784,
      "learning_rate": 0.00014226533036377286,
      "loss": 1.3839,
      "step": 2655
    },
    {
      "epoch": 0.3804612519696319,
      "grad_norm": 1.4573674201965332,
      "learning_rate": 0.00014222327591860792,
      "loss": 1.4683,
      "step": 2656
    },
    {
      "epoch": 0.38060449792293366,
      "grad_norm": 1.3096299171447754,
      "learning_rate": 0.00014218121238381652,
      "loss": 1.3719,
      "step": 2657
    },
    {
      "epoch": 0.3807477438762355,
      "grad_norm": 1.1627764701843262,
      "learning_rate": 0.0001421391397684539,
      "loss": 1.4346,
      "step": 2658
    },
    {
      "epoch": 0.38089098982953734,
      "grad_norm": 1.2307974100112915,
      "learning_rate": 0.00014209705808157733,
      "loss": 1.5122,
      "step": 2659
    },
    {
      "epoch": 0.3810342357828391,
      "grad_norm": 1.1243153810501099,
      "learning_rate": 0.00014205496733224582,
      "loss": 1.5507,
      "step": 2660
    },
    {
      "epoch": 0.38117748173614097,
      "grad_norm": 1.220015048980713,
      "learning_rate": 0.00014201286752952056,
      "loss": 1.5266,
      "step": 2661
    },
    {
      "epoch": 0.38132072768944275,
      "grad_norm": 1.0911564826965332,
      "learning_rate": 0.00014197075868246461,
      "loss": 1.4817,
      "step": 2662
    },
    {
      "epoch": 0.3814639736427446,
      "grad_norm": 1.0004096031188965,
      "learning_rate": 0.0001419286408001429,
      "loss": 1.5482,
      "step": 2663
    },
    {
      "epoch": 0.38160721959604643,
      "grad_norm": 1.1626720428466797,
      "learning_rate": 0.00014188651389162243,
      "loss": 1.5116,
      "step": 2664
    },
    {
      "epoch": 0.3817504655493482,
      "grad_norm": 1.1454691886901855,
      "learning_rate": 0.00014184437796597202,
      "loss": 1.4704,
      "step": 2665
    },
    {
      "epoch": 0.38189371150265006,
      "grad_norm": 1.0907492637634277,
      "learning_rate": 0.00014180223303226255,
      "loss": 1.5029,
      "step": 2666
    },
    {
      "epoch": 0.38203695745595184,
      "grad_norm": 1.0236293077468872,
      "learning_rate": 0.0001417600790995667,
      "loss": 1.5527,
      "step": 2667
    },
    {
      "epoch": 0.3821802034092537,
      "grad_norm": 1.1283305883407593,
      "learning_rate": 0.00014171791617695927,
      "loss": 1.3616,
      "step": 2668
    },
    {
      "epoch": 0.3823234493625555,
      "grad_norm": 1.5132068395614624,
      "learning_rate": 0.00014167574427351683,
      "loss": 1.4162,
      "step": 2669
    },
    {
      "epoch": 0.3824666953158573,
      "grad_norm": 1.174524188041687,
      "learning_rate": 0.00014163356339831797,
      "loss": 1.4974,
      "step": 2670
    },
    {
      "epoch": 0.38260994126915915,
      "grad_norm": 1.099651575088501,
      "learning_rate": 0.00014159137356044318,
      "loss": 1.4267,
      "step": 2671
    },
    {
      "epoch": 0.382753187222461,
      "grad_norm": 0.9730185866355896,
      "learning_rate": 0.00014154917476897486,
      "loss": 1.4964,
      "step": 2672
    },
    {
      "epoch": 0.3828964331757628,
      "grad_norm": 1.252204418182373,
      "learning_rate": 0.00014150696703299744,
      "loss": 1.502,
      "step": 2673
    },
    {
      "epoch": 0.3830396791290646,
      "grad_norm": 1.2112748622894287,
      "learning_rate": 0.00014146475036159713,
      "loss": 1.5899,
      "step": 2674
    },
    {
      "epoch": 0.3831829250823664,
      "grad_norm": 0.9682261347770691,
      "learning_rate": 0.00014142252476386218,
      "loss": 1.5922,
      "step": 2675
    },
    {
      "epoch": 0.38332617103566824,
      "grad_norm": 1.1810462474822998,
      "learning_rate": 0.00014138029024888263,
      "loss": 1.5129,
      "step": 2676
    },
    {
      "epoch": 0.3834694169889701,
      "grad_norm": 1.1216154098510742,
      "learning_rate": 0.00014133804682575068,
      "loss": 1.3655,
      "step": 2677
    },
    {
      "epoch": 0.38361266294227186,
      "grad_norm": 1.2673734426498413,
      "learning_rate": 0.00014129579450356016,
      "loss": 1.3562,
      "step": 2678
    },
    {
      "epoch": 0.3837559088955737,
      "grad_norm": 1.0625911951065063,
      "learning_rate": 0.00014125353329140703,
      "loss": 1.5199,
      "step": 2679
    },
    {
      "epoch": 0.38389915484887555,
      "grad_norm": 1.0846940279006958,
      "learning_rate": 0.000141211263198389,
      "loss": 1.5895,
      "step": 2680
    },
    {
      "epoch": 0.38404240080217733,
      "grad_norm": 1.1811267137527466,
      "learning_rate": 0.00014116898423360586,
      "loss": 1.4873,
      "step": 2681
    },
    {
      "epoch": 0.38418564675547917,
      "grad_norm": 1.2285857200622559,
      "learning_rate": 0.00014112669640615918,
      "loss": 1.4566,
      "step": 2682
    },
    {
      "epoch": 0.38432889270878096,
      "grad_norm": 1.1813266277313232,
      "learning_rate": 0.00014108439972515248,
      "loss": 1.4262,
      "step": 2683
    },
    {
      "epoch": 0.3844721386620828,
      "grad_norm": 1.395022988319397,
      "learning_rate": 0.00014104209419969123,
      "loss": 1.4795,
      "step": 2684
    },
    {
      "epoch": 0.38461538461538464,
      "grad_norm": 1.0140002965927124,
      "learning_rate": 0.00014099977983888267,
      "loss": 1.4915,
      "step": 2685
    },
    {
      "epoch": 0.3847586305686864,
      "grad_norm": 1.1237748861312866,
      "learning_rate": 0.00014095745665183618,
      "loss": 1.4766,
      "step": 2686
    },
    {
      "epoch": 0.38490187652198826,
      "grad_norm": 0.9617596864700317,
      "learning_rate": 0.00014091512464766277,
      "loss": 1.3649,
      "step": 2687
    },
    {
      "epoch": 0.38504512247529005,
      "grad_norm": 0.9155746698379517,
      "learning_rate": 0.00014087278383547553,
      "loss": 1.4637,
      "step": 2688
    },
    {
      "epoch": 0.3851883684285919,
      "grad_norm": 1.0307685136795044,
      "learning_rate": 0.00014083043422438935,
      "loss": 1.4593,
      "step": 2689
    },
    {
      "epoch": 0.38533161438189373,
      "grad_norm": 1.1862280368804932,
      "learning_rate": 0.00014078807582352108,
      "loss": 1.4602,
      "step": 2690
    },
    {
      "epoch": 0.3854748603351955,
      "grad_norm": 1.167630910873413,
      "learning_rate": 0.00014074570864198947,
      "loss": 1.3992,
      "step": 2691
    },
    {
      "epoch": 0.38561810628849735,
      "grad_norm": 0.993311882019043,
      "learning_rate": 0.00014070333268891504,
      "loss": 1.5854,
      "step": 2692
    },
    {
      "epoch": 0.3857613522417992,
      "grad_norm": 1.0830954313278198,
      "learning_rate": 0.00014066094797342036,
      "loss": 1.5101,
      "step": 2693
    },
    {
      "epoch": 0.385904598195101,
      "grad_norm": 1.4246561527252197,
      "learning_rate": 0.00014061855450462978,
      "loss": 1.424,
      "step": 2694
    },
    {
      "epoch": 0.3860478441484028,
      "grad_norm": 0.9677397608757019,
      "learning_rate": 0.0001405761522916696,
      "loss": 1.5294,
      "step": 2695
    },
    {
      "epoch": 0.3861910901017046,
      "grad_norm": 1.0421785116195679,
      "learning_rate": 0.00014053374134366788,
      "loss": 1.4412,
      "step": 2696
    },
    {
      "epoch": 0.38633433605500644,
      "grad_norm": 1.1148985624313354,
      "learning_rate": 0.0001404913216697547,
      "loss": 1.539,
      "step": 2697
    },
    {
      "epoch": 0.3864775820083083,
      "grad_norm": 1.039480209350586,
      "learning_rate": 0.00014044889327906202,
      "loss": 1.5344,
      "step": 2698
    },
    {
      "epoch": 0.38662082796161007,
      "grad_norm": 1.0137301683425903,
      "learning_rate": 0.00014040645618072355,
      "loss": 1.5247,
      "step": 2699
    },
    {
      "epoch": 0.3867640739149119,
      "grad_norm": 1.2161946296691895,
      "learning_rate": 0.00014036401038387497,
      "loss": 1.3838,
      "step": 2700
    },
    {
      "epoch": 0.38690731986821375,
      "grad_norm": 0.9870474338531494,
      "learning_rate": 0.00014032155589765378,
      "loss": 1.5133,
      "step": 2701
    },
    {
      "epoch": 0.38705056582151554,
      "grad_norm": 0.9808529615402222,
      "learning_rate": 0.00014027909273119944,
      "loss": 1.3012,
      "step": 2702
    },
    {
      "epoch": 0.3871938117748174,
      "grad_norm": 0.986569881439209,
      "learning_rate": 0.00014023662089365319,
      "loss": 1.5178,
      "step": 2703
    },
    {
      "epoch": 0.38733705772811916,
      "grad_norm": 1.529737114906311,
      "learning_rate": 0.00014019414039415817,
      "loss": 1.4995,
      "step": 2704
    },
    {
      "epoch": 0.387480303681421,
      "grad_norm": 1.100267767906189,
      "learning_rate": 0.00014015165124185933,
      "loss": 1.3576,
      "step": 2705
    },
    {
      "epoch": 0.38762354963472284,
      "grad_norm": 1.2582354545593262,
      "learning_rate": 0.00014010915344590363,
      "loss": 1.4032,
      "step": 2706
    },
    {
      "epoch": 0.3877667955880246,
      "grad_norm": 1.3152642250061035,
      "learning_rate": 0.0001400666470154397,
      "loss": 1.5016,
      "step": 2707
    },
    {
      "epoch": 0.38791004154132647,
      "grad_norm": 1.2205884456634521,
      "learning_rate": 0.00014002413195961819,
      "loss": 1.3578,
      "step": 2708
    },
    {
      "epoch": 0.38805328749462825,
      "grad_norm": 1.118962287902832,
      "learning_rate": 0.0001399816082875915,
      "loss": 1.4669,
      "step": 2709
    },
    {
      "epoch": 0.3881965334479301,
      "grad_norm": 1.0975672006607056,
      "learning_rate": 0.0001399390760085139,
      "loss": 1.5507,
      "step": 2710
    },
    {
      "epoch": 0.38833977940123193,
      "grad_norm": 0.8327113389968872,
      "learning_rate": 0.00013989653513154165,
      "loss": 1.5713,
      "step": 2711
    },
    {
      "epoch": 0.3884830253545337,
      "grad_norm": 1.1631824970245361,
      "learning_rate": 0.00013985398566583262,
      "loss": 1.4363,
      "step": 2712
    },
    {
      "epoch": 0.38862627130783556,
      "grad_norm": 1.2302311658859253,
      "learning_rate": 0.00013981142762054674,
      "loss": 1.3535,
      "step": 2713
    },
    {
      "epoch": 0.3887695172611374,
      "grad_norm": 1.2719974517822266,
      "learning_rate": 0.00013976886100484562,
      "loss": 1.4649,
      "step": 2714
    },
    {
      "epoch": 0.3889127632144392,
      "grad_norm": 1.0551207065582275,
      "learning_rate": 0.00013972628582789294,
      "loss": 1.4807,
      "step": 2715
    },
    {
      "epoch": 0.389056009167741,
      "grad_norm": 1.2198597192764282,
      "learning_rate": 0.00013968370209885392,
      "loss": 1.4442,
      "step": 2716
    },
    {
      "epoch": 0.3891992551210428,
      "grad_norm": 1.2216465473175049,
      "learning_rate": 0.0001396411098268959,
      "loss": 1.4289,
      "step": 2717
    },
    {
      "epoch": 0.38934250107434465,
      "grad_norm": 1.29093599319458,
      "learning_rate": 0.00013959850902118786,
      "loss": 1.2032,
      "step": 2718
    },
    {
      "epoch": 0.3894857470276465,
      "grad_norm": 1.2781002521514893,
      "learning_rate": 0.00013955589969090075,
      "loss": 1.4226,
      "step": 2719
    },
    {
      "epoch": 0.3896289929809483,
      "grad_norm": 1.0744190216064453,
      "learning_rate": 0.00013951328184520732,
      "loss": 1.544,
      "step": 2720
    },
    {
      "epoch": 0.3897722389342501,
      "grad_norm": 1.1971217393875122,
      "learning_rate": 0.0001394706554932821,
      "loss": 1.4484,
      "step": 2721
    },
    {
      "epoch": 0.3899154848875519,
      "grad_norm": 1.0613747835159302,
      "learning_rate": 0.00013942802064430146,
      "loss": 1.5457,
      "step": 2722
    },
    {
      "epoch": 0.39005873084085374,
      "grad_norm": 1.0913233757019043,
      "learning_rate": 0.0001393853773074437,
      "loss": 1.3368,
      "step": 2723
    },
    {
      "epoch": 0.3902019767941556,
      "grad_norm": 1.1087960004806519,
      "learning_rate": 0.00013934272549188888,
      "loss": 1.3487,
      "step": 2724
    },
    {
      "epoch": 0.39034522274745737,
      "grad_norm": 1.001875638961792,
      "learning_rate": 0.0001393000652068188,
      "loss": 1.5003,
      "step": 2725
    },
    {
      "epoch": 0.3904884687007592,
      "grad_norm": 1.1012027263641357,
      "learning_rate": 0.0001392573964614172,
      "loss": 1.459,
      "step": 2726
    },
    {
      "epoch": 0.39063171465406105,
      "grad_norm": 1.2598809003829956,
      "learning_rate": 0.00013921471926486961,
      "loss": 1.4978,
      "step": 2727
    },
    {
      "epoch": 0.39077496060736283,
      "grad_norm": 1.0599154233932495,
      "learning_rate": 0.0001391720336263634,
      "loss": 1.5258,
      "step": 2728
    },
    {
      "epoch": 0.39091820656066467,
      "grad_norm": 1.2188338041305542,
      "learning_rate": 0.0001391293395550877,
      "loss": 1.3219,
      "step": 2729
    },
    {
      "epoch": 0.39106145251396646,
      "grad_norm": 1.2200278043746948,
      "learning_rate": 0.00013908663706023347,
      "loss": 1.4668,
      "step": 2730
    },
    {
      "epoch": 0.3912046984672683,
      "grad_norm": 0.9860569834709167,
      "learning_rate": 0.00013904392615099356,
      "loss": 1.4106,
      "step": 2731
    },
    {
      "epoch": 0.39134794442057014,
      "grad_norm": 1.3161883354187012,
      "learning_rate": 0.00013900120683656253,
      "loss": 1.3509,
      "step": 2732
    },
    {
      "epoch": 0.3914911903738719,
      "grad_norm": 1.2047829627990723,
      "learning_rate": 0.00013895847912613678,
      "loss": 1.3227,
      "step": 2733
    },
    {
      "epoch": 0.39163443632717376,
      "grad_norm": 1.234537959098816,
      "learning_rate": 0.00013891574302891458,
      "loss": 1.497,
      "step": 2734
    },
    {
      "epoch": 0.3917776822804756,
      "grad_norm": 1.2812901735305786,
      "learning_rate": 0.00013887299855409586,
      "loss": 1.4334,
      "step": 2735
    },
    {
      "epoch": 0.3919209282337774,
      "grad_norm": 1.0933613777160645,
      "learning_rate": 0.00013883024571088257,
      "loss": 1.3088,
      "step": 2736
    },
    {
      "epoch": 0.39206417418707923,
      "grad_norm": 1.1715500354766846,
      "learning_rate": 0.00013878748450847826,
      "loss": 1.4636,
      "step": 2737
    },
    {
      "epoch": 0.392207420140381,
      "grad_norm": 1.0626330375671387,
      "learning_rate": 0.00013874471495608836,
      "loss": 1.458,
      "step": 2738
    },
    {
      "epoch": 0.39235066609368285,
      "grad_norm": 1.1760331392288208,
      "learning_rate": 0.00013870193706292012,
      "loss": 1.6347,
      "step": 2739
    },
    {
      "epoch": 0.3924939120469847,
      "grad_norm": 1.1869091987609863,
      "learning_rate": 0.00013865915083818256,
      "loss": 1.5342,
      "step": 2740
    },
    {
      "epoch": 0.3926371580002865,
      "grad_norm": 1.1295634508132935,
      "learning_rate": 0.0001386163562910865,
      "loss": 1.563,
      "step": 2741
    },
    {
      "epoch": 0.3927804039535883,
      "grad_norm": 1.0127500295639038,
      "learning_rate": 0.00013857355343084452,
      "loss": 1.55,
      "step": 2742
    },
    {
      "epoch": 0.3929236499068901,
      "grad_norm": 1.2653014659881592,
      "learning_rate": 0.00013853074226667102,
      "loss": 1.3498,
      "step": 2743
    },
    {
      "epoch": 0.39306689586019194,
      "grad_norm": 1.2269583940505981,
      "learning_rate": 0.00013848792280778222,
      "loss": 1.5557,
      "step": 2744
    },
    {
      "epoch": 0.3932101418134938,
      "grad_norm": 1.3298572301864624,
      "learning_rate": 0.0001384450950633961,
      "loss": 1.527,
      "step": 2745
    },
    {
      "epoch": 0.39335338776679557,
      "grad_norm": 1.0951743125915527,
      "learning_rate": 0.00013840225904273234,
      "loss": 1.5201,
      "step": 2746
    },
    {
      "epoch": 0.3934966337200974,
      "grad_norm": 1.1128827333450317,
      "learning_rate": 0.00013835941475501251,
      "loss": 1.381,
      "step": 2747
    },
    {
      "epoch": 0.39363987967339925,
      "grad_norm": 1.1959412097930908,
      "learning_rate": 0.00013831656220945993,
      "loss": 1.3947,
      "step": 2748
    },
    {
      "epoch": 0.39378312562670104,
      "grad_norm": 1.0881004333496094,
      "learning_rate": 0.0001382737014152997,
      "loss": 1.5294,
      "step": 2749
    },
    {
      "epoch": 0.3939263715800029,
      "grad_norm": 1.0680370330810547,
      "learning_rate": 0.00013823083238175872,
      "loss": 1.3959,
      "step": 2750
    },
    {
      "epoch": 0.39406961753330466,
      "grad_norm": 1.1935373544692993,
      "learning_rate": 0.00013818795511806554,
      "loss": 1.5015,
      "step": 2751
    },
    {
      "epoch": 0.3942128634866065,
      "grad_norm": 1.1426711082458496,
      "learning_rate": 0.00013814506963345067,
      "loss": 1.3803,
      "step": 2752
    },
    {
      "epoch": 0.39435610943990834,
      "grad_norm": 1.140364646911621,
      "learning_rate": 0.00013810217593714623,
      "loss": 1.6693,
      "step": 2753
    },
    {
      "epoch": 0.3944993553932101,
      "grad_norm": 0.9717327952384949,
      "learning_rate": 0.00013805927403838622,
      "loss": 1.4261,
      "step": 2754
    },
    {
      "epoch": 0.39464260134651197,
      "grad_norm": 1.185996174812317,
      "learning_rate": 0.00013801636394640627,
      "loss": 1.3365,
      "step": 2755
    },
    {
      "epoch": 0.3947858472998138,
      "grad_norm": 1.1653282642364502,
      "learning_rate": 0.00013797344567044396,
      "loss": 1.4204,
      "step": 2756
    },
    {
      "epoch": 0.3949290932531156,
      "grad_norm": 1.1007578372955322,
      "learning_rate": 0.00013793051921973852,
      "loss": 1.5159,
      "step": 2757
    },
    {
      "epoch": 0.39507233920641743,
      "grad_norm": 1.0245144367218018,
      "learning_rate": 0.00013788758460353087,
      "loss": 1.5361,
      "step": 2758
    },
    {
      "epoch": 0.3952155851597192,
      "grad_norm": 1.1577740907669067,
      "learning_rate": 0.00013784464183106389,
      "loss": 1.4572,
      "step": 2759
    },
    {
      "epoch": 0.39535883111302106,
      "grad_norm": 0.9496899247169495,
      "learning_rate": 0.00013780169091158197,
      "loss": 1.3499,
      "step": 2760
    },
    {
      "epoch": 0.3955020770663229,
      "grad_norm": 1.0334111452102661,
      "learning_rate": 0.0001377587318543315,
      "loss": 1.3178,
      "step": 2761
    },
    {
      "epoch": 0.3956453230196247,
      "grad_norm": 1.073441982269287,
      "learning_rate": 0.0001377157646685604,
      "loss": 1.5489,
      "step": 2762
    },
    {
      "epoch": 0.3957885689729265,
      "grad_norm": 1.1558798551559448,
      "learning_rate": 0.00013767278936351854,
      "loss": 1.5191,
      "step": 2763
    },
    {
      "epoch": 0.3959318149262283,
      "grad_norm": 1.1839429140090942,
      "learning_rate": 0.0001376298059484573,
      "loss": 1.4471,
      "step": 2764
    },
    {
      "epoch": 0.39607506087953015,
      "grad_norm": 0.9303610324859619,
      "learning_rate": 0.00013758681443263012,
      "loss": 1.52,
      "step": 2765
    },
    {
      "epoch": 0.396218306832832,
      "grad_norm": 1.0678328275680542,
      "learning_rate": 0.00013754381482529188,
      "loss": 1.5048,
      "step": 2766
    },
    {
      "epoch": 0.3963615527861338,
      "grad_norm": 1.188411831855774,
      "learning_rate": 0.0001375008071356994,
      "loss": 1.6107,
      "step": 2767
    },
    {
      "epoch": 0.3965047987394356,
      "grad_norm": 0.9718554019927979,
      "learning_rate": 0.0001374577913731111,
      "loss": 1.4725,
      "step": 2768
    },
    {
      "epoch": 0.39664804469273746,
      "grad_norm": 0.9430140852928162,
      "learning_rate": 0.0001374147675467873,
      "loss": 1.347,
      "step": 2769
    },
    {
      "epoch": 0.39679129064603924,
      "grad_norm": 1.3940454721450806,
      "learning_rate": 0.00013737173566598991,
      "loss": 1.4218,
      "step": 2770
    },
    {
      "epoch": 0.3969345365993411,
      "grad_norm": 1.2685039043426514,
      "learning_rate": 0.00013732869573998262,
      "loss": 1.4834,
      "step": 2771
    },
    {
      "epoch": 0.39707778255264287,
      "grad_norm": 0.9639720320701599,
      "learning_rate": 0.00013728564777803088,
      "loss": 1.4831,
      "step": 2772
    },
    {
      "epoch": 0.3972210285059447,
      "grad_norm": 1.4262007474899292,
      "learning_rate": 0.00013724259178940184,
      "loss": 1.6025,
      "step": 2773
    },
    {
      "epoch": 0.39736427445924655,
      "grad_norm": 1.1385854482650757,
      "learning_rate": 0.00013719952778336442,
      "loss": 1.4198,
      "step": 2774
    },
    {
      "epoch": 0.39750752041254833,
      "grad_norm": 1.6640490293502808,
      "learning_rate": 0.0001371564557691892,
      "loss": 1.4862,
      "step": 2775
    },
    {
      "epoch": 0.3976507663658502,
      "grad_norm": 1.1516785621643066,
      "learning_rate": 0.00013711337575614857,
      "loss": 1.4584,
      "step": 2776
    },
    {
      "epoch": 0.39779401231915196,
      "grad_norm": 1.0535268783569336,
      "learning_rate": 0.0001370702877535165,
      "loss": 1.3723,
      "step": 2777
    },
    {
      "epoch": 0.3979372582724538,
      "grad_norm": 1.0283282995224,
      "learning_rate": 0.00013702719177056884,
      "loss": 1.7369,
      "step": 2778
    },
    {
      "epoch": 0.39808050422575564,
      "grad_norm": 1.1601848602294922,
      "learning_rate": 0.0001369840878165831,
      "loss": 1.4379,
      "step": 2779
    },
    {
      "epoch": 0.3982237501790574,
      "grad_norm": 1.1825915575027466,
      "learning_rate": 0.00013694097590083844,
      "loss": 1.4749,
      "step": 2780
    },
    {
      "epoch": 0.39836699613235926,
      "grad_norm": 1.1129486560821533,
      "learning_rate": 0.00013689785603261583,
      "loss": 1.5742,
      "step": 2781
    },
    {
      "epoch": 0.3985102420856611,
      "grad_norm": 1.1634058952331543,
      "learning_rate": 0.00013685472822119786,
      "loss": 1.4949,
      "step": 2782
    },
    {
      "epoch": 0.3986534880389629,
      "grad_norm": 1.1009522676467896,
      "learning_rate": 0.00013681159247586896,
      "loss": 1.5692,
      "step": 2783
    },
    {
      "epoch": 0.39879673399226473,
      "grad_norm": 1.083465337753296,
      "learning_rate": 0.00013676844880591512,
      "loss": 1.6274,
      "step": 2784
    },
    {
      "epoch": 0.3989399799455665,
      "grad_norm": 0.9692406058311462,
      "learning_rate": 0.00013672529722062415,
      "loss": 1.568,
      "step": 2785
    },
    {
      "epoch": 0.39908322589886835,
      "grad_norm": 1.1225932836532593,
      "learning_rate": 0.0001366821377292855,
      "loss": 1.5064,
      "step": 2786
    },
    {
      "epoch": 0.3992264718521702,
      "grad_norm": 1.116595983505249,
      "learning_rate": 0.0001366389703411903,
      "loss": 1.472,
      "step": 2787
    },
    {
      "epoch": 0.399369717805472,
      "grad_norm": 1.1441022157669067,
      "learning_rate": 0.0001365957950656315,
      "loss": 1.6116,
      "step": 2788
    },
    {
      "epoch": 0.3995129637587738,
      "grad_norm": 1.054233193397522,
      "learning_rate": 0.0001365526119119036,
      "loss": 1.5448,
      "step": 2789
    },
    {
      "epoch": 0.39965620971207566,
      "grad_norm": 0.8752365708351135,
      "learning_rate": 0.00013650942088930295,
      "loss": 1.4324,
      "step": 2790
    },
    {
      "epoch": 0.39979945566537745,
      "grad_norm": 1.4697915315628052,
      "learning_rate": 0.00013646622200712738,
      "loss": 1.3729,
      "step": 2791
    },
    {
      "epoch": 0.3999427016186793,
      "grad_norm": 1.1016391515731812,
      "learning_rate": 0.00013642301527467664,
      "loss": 1.2313,
      "step": 2792
    },
    {
      "epoch": 0.40008594757198107,
      "grad_norm": 1.0731985569000244,
      "learning_rate": 0.00013637980070125205,
      "loss": 1.5159,
      "step": 2793
    },
    {
      "epoch": 0.4002291935252829,
      "grad_norm": 1.120406150817871,
      "learning_rate": 0.0001363365782961566,
      "loss": 1.3374,
      "step": 2794
    },
    {
      "epoch": 0.40037243947858475,
      "grad_norm": 1.0571527481079102,
      "learning_rate": 0.00013629334806869507,
      "loss": 1.5812,
      "step": 2795
    },
    {
      "epoch": 0.40051568543188654,
      "grad_norm": 1.041649580001831,
      "learning_rate": 0.0001362501100281738,
      "loss": 1.3901,
      "step": 2796
    },
    {
      "epoch": 0.4006589313851884,
      "grad_norm": 1.134366750717163,
      "learning_rate": 0.0001362068641839009,
      "loss": 1.6005,
      "step": 2797
    },
    {
      "epoch": 0.40080217733849016,
      "grad_norm": 1.573654055595398,
      "learning_rate": 0.0001361636105451861,
      "loss": 1.3569,
      "step": 2798
    },
    {
      "epoch": 0.400945423291792,
      "grad_norm": 1.1163673400878906,
      "learning_rate": 0.0001361203491213409,
      "loss": 1.5145,
      "step": 2799
    },
    {
      "epoch": 0.40108866924509384,
      "grad_norm": 1.1613695621490479,
      "learning_rate": 0.00013607707992167834,
      "loss": 1.3559,
      "step": 2800
    },
    {
      "epoch": 0.40123191519839563,
      "grad_norm": 1.2973524332046509,
      "learning_rate": 0.00013603380295551328,
      "loss": 1.4852,
      "step": 2801
    },
    {
      "epoch": 0.40137516115169747,
      "grad_norm": 1.0951015949249268,
      "learning_rate": 0.0001359905182321621,
      "loss": 1.2779,
      "step": 2802
    },
    {
      "epoch": 0.4015184071049993,
      "grad_norm": 1.144309163093567,
      "learning_rate": 0.00013594722576094296,
      "loss": 1.4245,
      "step": 2803
    },
    {
      "epoch": 0.4016616530583011,
      "grad_norm": 1.2852388620376587,
      "learning_rate": 0.00013590392555117573,
      "loss": 1.5809,
      "step": 2804
    },
    {
      "epoch": 0.40180489901160293,
      "grad_norm": 1.2533011436462402,
      "learning_rate": 0.00013586061761218176,
      "loss": 1.4776,
      "step": 2805
    },
    {
      "epoch": 0.4019481449649047,
      "grad_norm": 1.0780951976776123,
      "learning_rate": 0.00013581730195328425,
      "loss": 1.5449,
      "step": 2806
    },
    {
      "epoch": 0.40209139091820656,
      "grad_norm": 1.12636137008667,
      "learning_rate": 0.00013577397858380798,
      "loss": 1.6181,
      "step": 2807
    },
    {
      "epoch": 0.4022346368715084,
      "grad_norm": 1.159757375717163,
      "learning_rate": 0.0001357306475130794,
      "loss": 1.4516,
      "step": 2808
    },
    {
      "epoch": 0.4023778828248102,
      "grad_norm": 1.234001636505127,
      "learning_rate": 0.00013568730875042654,
      "loss": 1.4359,
      "step": 2809
    },
    {
      "epoch": 0.402521128778112,
      "grad_norm": 1.2021113634109497,
      "learning_rate": 0.0001356439623051793,
      "loss": 1.5114,
      "step": 2810
    },
    {
      "epoch": 0.4026643747314138,
      "grad_norm": 1.043145775794983,
      "learning_rate": 0.000135600608186669,
      "loss": 1.5188,
      "step": 2811
    },
    {
      "epoch": 0.40280762068471565,
      "grad_norm": 1.0737450122833252,
      "learning_rate": 0.00013555724640422874,
      "loss": 1.5434,
      "step": 2812
    },
    {
      "epoch": 0.4029508666380175,
      "grad_norm": 1.1912723779678345,
      "learning_rate": 0.00013551387696719326,
      "loss": 1.5644,
      "step": 2813
    },
    {
      "epoch": 0.4030941125913193,
      "grad_norm": 0.937789261341095,
      "learning_rate": 0.00013547049988489883,
      "loss": 1.7395,
      "step": 2814
    },
    {
      "epoch": 0.4032373585446211,
      "grad_norm": 1.2506660223007202,
      "learning_rate": 0.0001354271151666836,
      "loss": 1.5189,
      "step": 2815
    },
    {
      "epoch": 0.40338060449792296,
      "grad_norm": 1.3174413442611694,
      "learning_rate": 0.00013538372282188713,
      "loss": 1.5846,
      "step": 2816
    },
    {
      "epoch": 0.40352385045122474,
      "grad_norm": 1.0043715238571167,
      "learning_rate": 0.0001353403228598508,
      "loss": 1.3538,
      "step": 2817
    },
    {
      "epoch": 0.4036670964045266,
      "grad_norm": 1.2512023448944092,
      "learning_rate": 0.00013529691528991743,
      "loss": 1.5064,
      "step": 2818
    },
    {
      "epoch": 0.40381034235782837,
      "grad_norm": 1.122894048690796,
      "learning_rate": 0.00013525350012143168,
      "loss": 1.4153,
      "step": 2819
    },
    {
      "epoch": 0.4039535883111302,
      "grad_norm": 1.2257254123687744,
      "learning_rate": 0.00013521007736373974,
      "loss": 1.479,
      "step": 2820
    },
    {
      "epoch": 0.40409683426443205,
      "grad_norm": 1.211525559425354,
      "learning_rate": 0.00013516664702618948,
      "loss": 1.4493,
      "step": 2821
    },
    {
      "epoch": 0.40424008021773383,
      "grad_norm": 0.9623228907585144,
      "learning_rate": 0.0001351232091181303,
      "loss": 1.4962,
      "step": 2822
    },
    {
      "epoch": 0.4043833261710357,
      "grad_norm": 1.1287436485290527,
      "learning_rate": 0.0001350797636489134,
      "loss": 1.4036,
      "step": 2823
    },
    {
      "epoch": 0.4045265721243375,
      "grad_norm": 1.3107287883758545,
      "learning_rate": 0.0001350363106278915,
      "loss": 1.5864,
      "step": 2824
    },
    {
      "epoch": 0.4046698180776393,
      "grad_norm": 0.955753743648529,
      "learning_rate": 0.00013499285006441888,
      "loss": 1.3797,
      "step": 2825
    },
    {
      "epoch": 0.40481306403094114,
      "grad_norm": 0.9739630818367004,
      "learning_rate": 0.00013494938196785162,
      "loss": 1.4027,
      "step": 2826
    },
    {
      "epoch": 0.4049563099842429,
      "grad_norm": 1.1018555164337158,
      "learning_rate": 0.00013490590634754728,
      "loss": 1.4327,
      "step": 2827
    },
    {
      "epoch": 0.40509955593754476,
      "grad_norm": 1.1858118772506714,
      "learning_rate": 0.00013486242321286506,
      "loss": 1.473,
      "step": 2828
    },
    {
      "epoch": 0.4052428018908466,
      "grad_norm": 1.1405820846557617,
      "learning_rate": 0.0001348189325731659,
      "loss": 1.3526,
      "step": 2829
    },
    {
      "epoch": 0.4053860478441484,
      "grad_norm": 0.9196799397468567,
      "learning_rate": 0.0001347754344378121,
      "loss": 1.5678,
      "step": 2830
    },
    {
      "epoch": 0.40552929379745023,
      "grad_norm": 1.2381303310394287,
      "learning_rate": 0.0001347319288161679,
      "loss": 1.5034,
      "step": 2831
    },
    {
      "epoch": 0.405672539750752,
      "grad_norm": 1.251454472541809,
      "learning_rate": 0.00013468841571759888,
      "loss": 1.2357,
      "step": 2832
    },
    {
      "epoch": 0.40581578570405386,
      "grad_norm": 1.2669380903244019,
      "learning_rate": 0.00013464489515147238,
      "loss": 1.3635,
      "step": 2833
    },
    {
      "epoch": 0.4059590316573557,
      "grad_norm": 1.0958219766616821,
      "learning_rate": 0.00013460136712715724,
      "loss": 1.6209,
      "step": 2834
    },
    {
      "epoch": 0.4061022776106575,
      "grad_norm": 1.2327121496200562,
      "learning_rate": 0.00013455783165402404,
      "loss": 1.4954,
      "step": 2835
    },
    {
      "epoch": 0.4062455235639593,
      "grad_norm": 1.2271267175674438,
      "learning_rate": 0.00013451428874144484,
      "loss": 1.37,
      "step": 2836
    },
    {
      "epoch": 0.40638876951726116,
      "grad_norm": 1.0246933698654175,
      "learning_rate": 0.0001344707383987934,
      "loss": 1.4175,
      "step": 2837
    },
    {
      "epoch": 0.40653201547056295,
      "grad_norm": 1.249424934387207,
      "learning_rate": 0.000134427180635445,
      "loss": 1.338,
      "step": 2838
    },
    {
      "epoch": 0.4066752614238648,
      "grad_norm": 1.0454884767532349,
      "learning_rate": 0.0001343836154607765,
      "loss": 1.4687,
      "step": 2839
    },
    {
      "epoch": 0.40681850737716657,
      "grad_norm": 1.1392124891281128,
      "learning_rate": 0.0001343400428841665,
      "loss": 1.6099,
      "step": 2840
    },
    {
      "epoch": 0.4069617533304684,
      "grad_norm": 1.1528297662734985,
      "learning_rate": 0.00013429646291499502,
      "loss": 1.4681,
      "step": 2841
    },
    {
      "epoch": 0.40710499928377025,
      "grad_norm": 1.300648808479309,
      "learning_rate": 0.00013425287556264383,
      "loss": 1.41,
      "step": 2842
    },
    {
      "epoch": 0.40724824523707204,
      "grad_norm": 1.127364993095398,
      "learning_rate": 0.00013420928083649608,
      "loss": 1.5362,
      "step": 2843
    },
    {
      "epoch": 0.4073914911903739,
      "grad_norm": 1.1325756311416626,
      "learning_rate": 0.0001341656787459368,
      "loss": 1.5431,
      "step": 2844
    },
    {
      "epoch": 0.4075347371436757,
      "grad_norm": 0.8674351572990417,
      "learning_rate": 0.0001341220693003523,
      "loss": 1.424,
      "step": 2845
    },
    {
      "epoch": 0.4076779830969775,
      "grad_norm": 1.3473536968231201,
      "learning_rate": 0.00013407845250913066,
      "loss": 1.3906,
      "step": 2846
    },
    {
      "epoch": 0.40782122905027934,
      "grad_norm": 1.1625980138778687,
      "learning_rate": 0.00013403482838166155,
      "loss": 1.5181,
      "step": 2847
    },
    {
      "epoch": 0.40796447500358113,
      "grad_norm": 1.3195536136627197,
      "learning_rate": 0.00013399119692733612,
      "loss": 1.5212,
      "step": 2848
    },
    {
      "epoch": 0.40810772095688297,
      "grad_norm": 0.977924644947052,
      "learning_rate": 0.00013394755815554713,
      "loss": 1.5501,
      "step": 2849
    },
    {
      "epoch": 0.4082509669101848,
      "grad_norm": 1.2794781923294067,
      "learning_rate": 0.0001339039120756889,
      "loss": 1.4029,
      "step": 2850
    },
    {
      "epoch": 0.4083942128634866,
      "grad_norm": 1.2609822750091553,
      "learning_rate": 0.00013386025869715744,
      "loss": 1.6376,
      "step": 2851
    },
    {
      "epoch": 0.40853745881678843,
      "grad_norm": 1.1048805713653564,
      "learning_rate": 0.00013381659802935017,
      "loss": 1.3858,
      "step": 2852
    },
    {
      "epoch": 0.4086807047700902,
      "grad_norm": 1.3309123516082764,
      "learning_rate": 0.0001337729300816662,
      "loss": 1.4825,
      "step": 2853
    },
    {
      "epoch": 0.40882395072339206,
      "grad_norm": 1.2099355459213257,
      "learning_rate": 0.0001337292548635061,
      "loss": 1.2859,
      "step": 2854
    },
    {
      "epoch": 0.4089671966766939,
      "grad_norm": 1.0338795185089111,
      "learning_rate": 0.00013368557238427212,
      "loss": 1.4519,
      "step": 2855
    },
    {
      "epoch": 0.4091104426299957,
      "grad_norm": 1.0662775039672852,
      "learning_rate": 0.000133641882653368,
      "loss": 1.5718,
      "step": 2856
    },
    {
      "epoch": 0.4092536885832975,
      "grad_norm": 1.3427984714508057,
      "learning_rate": 0.00013359818568019904,
      "loss": 1.4128,
      "step": 2857
    },
    {
      "epoch": 0.40939693453659937,
      "grad_norm": 1.0407174825668335,
      "learning_rate": 0.00013355448147417214,
      "loss": 1.4963,
      "step": 2858
    },
    {
      "epoch": 0.40954018048990115,
      "grad_norm": 1.2363427877426147,
      "learning_rate": 0.0001335107700446957,
      "loss": 1.5084,
      "step": 2859
    },
    {
      "epoch": 0.409683426443203,
      "grad_norm": 0.9856411218643188,
      "learning_rate": 0.00013346705140117977,
      "loss": 1.5685,
      "step": 2860
    },
    {
      "epoch": 0.4098266723965048,
      "grad_norm": 1.1373803615570068,
      "learning_rate": 0.0001334233255530358,
      "loss": 1.5162,
      "step": 2861
    },
    {
      "epoch": 0.4099699183498066,
      "grad_norm": 1.0047681331634521,
      "learning_rate": 0.00013337959250967697,
      "loss": 1.3596,
      "step": 2862
    },
    {
      "epoch": 0.41011316430310846,
      "grad_norm": 1.3044847249984741,
      "learning_rate": 0.0001333358522805179,
      "loss": 1.4707,
      "step": 2863
    },
    {
      "epoch": 0.41025641025641024,
      "grad_norm": 1.3690197467803955,
      "learning_rate": 0.00013329210487497475,
      "loss": 1.4703,
      "step": 2864
    },
    {
      "epoch": 0.4103996562097121,
      "grad_norm": 0.9814923405647278,
      "learning_rate": 0.0001332483503024653,
      "loss": 1.5002,
      "step": 2865
    },
    {
      "epoch": 0.41054290216301387,
      "grad_norm": 1.3000876903533936,
      "learning_rate": 0.00013320458857240877,
      "loss": 1.542,
      "step": 2866
    },
    {
      "epoch": 0.4106861481163157,
      "grad_norm": 1.0864509344100952,
      "learning_rate": 0.00013316081969422602,
      "loss": 1.383,
      "step": 2867
    },
    {
      "epoch": 0.41082939406961755,
      "grad_norm": 0.9772033095359802,
      "learning_rate": 0.00013311704367733937,
      "loss": 1.51,
      "step": 2868
    },
    {
      "epoch": 0.41097264002291933,
      "grad_norm": 1.0879658460617065,
      "learning_rate": 0.00013307326053117277,
      "loss": 1.3992,
      "step": 2869
    },
    {
      "epoch": 0.4111158859762212,
      "grad_norm": 1.232767105102539,
      "learning_rate": 0.00013302947026515158,
      "loss": 1.6992,
      "step": 2870
    },
    {
      "epoch": 0.411259131929523,
      "grad_norm": 0.9638648629188538,
      "learning_rate": 0.0001329856728887028,
      "loss": 1.5831,
      "step": 2871
    },
    {
      "epoch": 0.4114023778828248,
      "grad_norm": 1.0368996858596802,
      "learning_rate": 0.00013294186841125493,
      "loss": 1.559,
      "step": 2872
    },
    {
      "epoch": 0.41154562383612664,
      "grad_norm": 1.0689702033996582,
      "learning_rate": 0.00013289805684223798,
      "loss": 1.494,
      "step": 2873
    },
    {
      "epoch": 0.4116888697894284,
      "grad_norm": 0.9811191558837891,
      "learning_rate": 0.0001328542381910835,
      "loss": 1.4314,
      "step": 2874
    },
    {
      "epoch": 0.41183211574273026,
      "grad_norm": 1.2971858978271484,
      "learning_rate": 0.00013281041246722452,
      "loss": 1.3218,
      "step": 2875
    },
    {
      "epoch": 0.4119753616960321,
      "grad_norm": 1.0183264017105103,
      "learning_rate": 0.00013276657968009568,
      "loss": 1.4108,
      "step": 2876
    },
    {
      "epoch": 0.4121186076493339,
      "grad_norm": 1.0744152069091797,
      "learning_rate": 0.0001327227398391331,
      "loss": 1.5798,
      "step": 2877
    },
    {
      "epoch": 0.41226185360263573,
      "grad_norm": 1.0109795331954956,
      "learning_rate": 0.00013267889295377438,
      "loss": 1.3757,
      "step": 2878
    },
    {
      "epoch": 0.41240509955593757,
      "grad_norm": 1.0059654712677002,
      "learning_rate": 0.00013263503903345868,
      "loss": 1.4304,
      "step": 2879
    },
    {
      "epoch": 0.41254834550923936,
      "grad_norm": 0.9491704106330872,
      "learning_rate": 0.0001325911780876267,
      "loss": 1.3791,
      "step": 2880
    },
    {
      "epoch": 0.4126915914625412,
      "grad_norm": 1.0815435647964478,
      "learning_rate": 0.00013254731012572052,
      "loss": 1.4149,
      "step": 2881
    },
    {
      "epoch": 0.412834837415843,
      "grad_norm": 0.9652405381202698,
      "learning_rate": 0.00013250343515718393,
      "loss": 1.3955,
      "step": 2882
    },
    {
      "epoch": 0.4129780833691448,
      "grad_norm": 1.0151206254959106,
      "learning_rate": 0.0001324595531914621,
      "loss": 1.3202,
      "step": 2883
    },
    {
      "epoch": 0.41312132932244666,
      "grad_norm": 1.0841914415359497,
      "learning_rate": 0.0001324156642380017,
      "loss": 1.3594,
      "step": 2884
    },
    {
      "epoch": 0.41326457527574845,
      "grad_norm": 1.0119885206222534,
      "learning_rate": 0.00013237176830625095,
      "loss": 1.4671,
      "step": 2885
    },
    {
      "epoch": 0.4134078212290503,
      "grad_norm": 1.1591053009033203,
      "learning_rate": 0.00013232786540565955,
      "loss": 1.4805,
      "step": 2886
    },
    {
      "epoch": 0.4135510671823521,
      "grad_norm": 1.3846744298934937,
      "learning_rate": 0.00013228395554567874,
      "loss": 1.3099,
      "step": 2887
    },
    {
      "epoch": 0.4136943131356539,
      "grad_norm": 1.0866940021514893,
      "learning_rate": 0.00013224003873576123,
      "loss": 1.2355,
      "step": 2888
    },
    {
      "epoch": 0.41383755908895575,
      "grad_norm": 1.1971701383590698,
      "learning_rate": 0.00013219611498536114,
      "loss": 1.4966,
      "step": 2889
    },
    {
      "epoch": 0.41398080504225754,
      "grad_norm": 0.9875962734222412,
      "learning_rate": 0.00013215218430393425,
      "loss": 1.4,
      "step": 2890
    },
    {
      "epoch": 0.4141240509955594,
      "grad_norm": 1.0728237628936768,
      "learning_rate": 0.0001321082467009377,
      "loss": 1.4172,
      "step": 2891
    },
    {
      "epoch": 0.4142672969488612,
      "grad_norm": 1.2615853548049927,
      "learning_rate": 0.00013206430218583025,
      "loss": 1.3558,
      "step": 2892
    },
    {
      "epoch": 0.414410542902163,
      "grad_norm": 1.1126292943954468,
      "learning_rate": 0.00013202035076807195,
      "loss": 1.5324,
      "step": 2893
    },
    {
      "epoch": 0.41455378885546484,
      "grad_norm": 1.0649443864822388,
      "learning_rate": 0.00013197639245712454,
      "loss": 1.4024,
      "step": 2894
    },
    {
      "epoch": 0.41469703480876663,
      "grad_norm": 1.183631420135498,
      "learning_rate": 0.0001319324272624511,
      "loss": 1.4497,
      "step": 2895
    },
    {
      "epoch": 0.41484028076206847,
      "grad_norm": 1.069616436958313,
      "learning_rate": 0.00013188845519351632,
      "loss": 1.3838,
      "step": 2896
    },
    {
      "epoch": 0.4149835267153703,
      "grad_norm": 1.0213828086853027,
      "learning_rate": 0.0001318444762597862,
      "loss": 1.2517,
      "step": 2897
    },
    {
      "epoch": 0.4151267726686721,
      "grad_norm": 1.1460169553756714,
      "learning_rate": 0.0001318004904707284,
      "loss": 1.3841,
      "step": 2898
    },
    {
      "epoch": 0.41527001862197394,
      "grad_norm": 1.0720977783203125,
      "learning_rate": 0.00013175649783581195,
      "loss": 1.7008,
      "step": 2899
    },
    {
      "epoch": 0.4154132645752758,
      "grad_norm": 1.2382010221481323,
      "learning_rate": 0.00013171249836450735,
      "loss": 1.6062,
      "step": 2900
    },
    {
      "epoch": 0.41555651052857756,
      "grad_norm": 1.060104489326477,
      "learning_rate": 0.00013166849206628663,
      "loss": 1.586,
      "step": 2901
    },
    {
      "epoch": 0.4156997564818794,
      "grad_norm": 1.1118627786636353,
      "learning_rate": 0.00013162447895062322,
      "loss": 1.5309,
      "step": 2902
    },
    {
      "epoch": 0.4158430024351812,
      "grad_norm": 0.8764147758483887,
      "learning_rate": 0.0001315804590269921,
      "loss": 1.5577,
      "step": 2903
    },
    {
      "epoch": 0.415986248388483,
      "grad_norm": 1.3989410400390625,
      "learning_rate": 0.00013153643230486963,
      "loss": 1.6303,
      "step": 2904
    },
    {
      "epoch": 0.41612949434178487,
      "grad_norm": 1.2069836854934692,
      "learning_rate": 0.0001314923987937337,
      "loss": 1.7139,
      "step": 2905
    },
    {
      "epoch": 0.41627274029508665,
      "grad_norm": 1.0896302461624146,
      "learning_rate": 0.0001314483585030636,
      "loss": 1.6546,
      "step": 2906
    },
    {
      "epoch": 0.4164159862483885,
      "grad_norm": 1.156543254852295,
      "learning_rate": 0.00013140431144234018,
      "loss": 1.4581,
      "step": 2907
    },
    {
      "epoch": 0.4165592322016903,
      "grad_norm": 1.2816578149795532,
      "learning_rate": 0.00013136025762104563,
      "loss": 1.5644,
      "step": 2908
    },
    {
      "epoch": 0.4167024781549921,
      "grad_norm": 0.8713088035583496,
      "learning_rate": 0.00013131619704866363,
      "loss": 1.5475,
      "step": 2909
    },
    {
      "epoch": 0.41684572410829396,
      "grad_norm": 0.9993542432785034,
      "learning_rate": 0.0001312721297346794,
      "loss": 1.4377,
      "step": 2910
    },
    {
      "epoch": 0.41698897006159574,
      "grad_norm": 1.2636206150054932,
      "learning_rate": 0.00013122805568857948,
      "loss": 1.4936,
      "step": 2911
    },
    {
      "epoch": 0.4171322160148976,
      "grad_norm": 1.3151670694351196,
      "learning_rate": 0.00013118397491985198,
      "loss": 1.4931,
      "step": 2912
    },
    {
      "epoch": 0.4172754619681994,
      "grad_norm": 1.1698790788650513,
      "learning_rate": 0.00013113988743798628,
      "loss": 1.5435,
      "step": 2913
    },
    {
      "epoch": 0.4174187079215012,
      "grad_norm": 0.9804201722145081,
      "learning_rate": 0.0001310957932524735,
      "loss": 1.4183,
      "step": 2914
    },
    {
      "epoch": 0.41756195387480305,
      "grad_norm": 1.2931900024414062,
      "learning_rate": 0.00013105169237280587,
      "loss": 1.5388,
      "step": 2915
    },
    {
      "epoch": 0.41770519982810483,
      "grad_norm": 1.1844921112060547,
      "learning_rate": 0.00013100758480847732,
      "loss": 1.4067,
      "step": 2916
    },
    {
      "epoch": 0.4178484457814067,
      "grad_norm": 1.134605884552002,
      "learning_rate": 0.00013096347056898308,
      "loss": 1.5314,
      "step": 2917
    },
    {
      "epoch": 0.4179916917347085,
      "grad_norm": 0.9807714223861694,
      "learning_rate": 0.00013091934966381983,
      "loss": 1.3936,
      "step": 2918
    },
    {
      "epoch": 0.4181349376880103,
      "grad_norm": 1.1086297035217285,
      "learning_rate": 0.00013087522210248576,
      "loss": 1.4978,
      "step": 2919
    },
    {
      "epoch": 0.41827818364131214,
      "grad_norm": 1.0083820819854736,
      "learning_rate": 0.00013083108789448038,
      "loss": 1.4995,
      "step": 2920
    },
    {
      "epoch": 0.4184214295946139,
      "grad_norm": 1.4726800918579102,
      "learning_rate": 0.00013078694704930473,
      "loss": 1.6217,
      "step": 2921
    },
    {
      "epoch": 0.41856467554791577,
      "grad_norm": 1.0786010026931763,
      "learning_rate": 0.00013074279957646126,
      "loss": 1.4239,
      "step": 2922
    },
    {
      "epoch": 0.4187079215012176,
      "grad_norm": 1.0577421188354492,
      "learning_rate": 0.0001306986454854538,
      "loss": 1.3987,
      "step": 2923
    },
    {
      "epoch": 0.4188511674545194,
      "grad_norm": 1.080257534980774,
      "learning_rate": 0.00013065448478578764,
      "loss": 1.5135,
      "step": 2924
    },
    {
      "epoch": 0.41899441340782123,
      "grad_norm": 1.2067092657089233,
      "learning_rate": 0.0001306103174869695,
      "loss": 1.5282,
      "step": 2925
    },
    {
      "epoch": 0.41913765936112307,
      "grad_norm": 1.0792328119277954,
      "learning_rate": 0.0001305661435985075,
      "loss": 1.3832,
      "step": 2926
    },
    {
      "epoch": 0.41928090531442486,
      "grad_norm": 0.9527145624160767,
      "learning_rate": 0.00013052196312991114,
      "loss": 1.3426,
      "step": 2927
    },
    {
      "epoch": 0.4194241512677267,
      "grad_norm": 0.9496454000473022,
      "learning_rate": 0.00013047777609069146,
      "loss": 1.4468,
      "step": 2928
    },
    {
      "epoch": 0.4195673972210285,
      "grad_norm": 1.0863112211227417,
      "learning_rate": 0.00013043358249036077,
      "loss": 1.4123,
      "step": 2929
    },
    {
      "epoch": 0.4197106431743303,
      "grad_norm": 1.052915096282959,
      "learning_rate": 0.00013038938233843287,
      "loss": 1.5495,
      "step": 2930
    },
    {
      "epoch": 0.41985388912763216,
      "grad_norm": 1.0582530498504639,
      "learning_rate": 0.000130345175644423,
      "loss": 1.4478,
      "step": 2931
    },
    {
      "epoch": 0.41999713508093395,
      "grad_norm": 1.008997917175293,
      "learning_rate": 0.00013030096241784772,
      "loss": 1.4705,
      "step": 2932
    },
    {
      "epoch": 0.4201403810342358,
      "grad_norm": 1.0599136352539062,
      "learning_rate": 0.0001302567426682251,
      "loss": 1.4005,
      "step": 2933
    },
    {
      "epoch": 0.42028362698753763,
      "grad_norm": 1.0790364742279053,
      "learning_rate": 0.00013021251640507448,
      "loss": 1.3247,
      "step": 2934
    },
    {
      "epoch": 0.4204268729408394,
      "grad_norm": 1.065268874168396,
      "learning_rate": 0.00013016828363791673,
      "loss": 1.5966,
      "step": 2935
    },
    {
      "epoch": 0.42057011889414125,
      "grad_norm": 1.0589488744735718,
      "learning_rate": 0.00013012404437627407,
      "loss": 1.5706,
      "step": 2936
    },
    {
      "epoch": 0.42071336484744304,
      "grad_norm": 1.2438745498657227,
      "learning_rate": 0.00013007979862967014,
      "loss": 1.4395,
      "step": 2937
    },
    {
      "epoch": 0.4208566108007449,
      "grad_norm": 1.17854905128479,
      "learning_rate": 0.00013003554640762986,
      "loss": 1.5951,
      "step": 2938
    },
    {
      "epoch": 0.4209998567540467,
      "grad_norm": 0.9444679617881775,
      "learning_rate": 0.00012999128771967977,
      "loss": 1.5462,
      "step": 2939
    },
    {
      "epoch": 0.4211431027073485,
      "grad_norm": 1.0070197582244873,
      "learning_rate": 0.00012994702257534756,
      "loss": 1.6002,
      "step": 2940
    },
    {
      "epoch": 0.42128634866065034,
      "grad_norm": 1.1389254331588745,
      "learning_rate": 0.00012990275098416246,
      "loss": 1.5906,
      "step": 2941
    },
    {
      "epoch": 0.42142959461395213,
      "grad_norm": 1.1540002822875977,
      "learning_rate": 0.00012985847295565508,
      "loss": 1.4798,
      "step": 2942
    },
    {
      "epoch": 0.42157284056725397,
      "grad_norm": 1.148228406906128,
      "learning_rate": 0.00012981418849935733,
      "loss": 1.5471,
      "step": 2943
    },
    {
      "epoch": 0.4217160865205558,
      "grad_norm": 1.1624430418014526,
      "learning_rate": 0.0001297698976248026,
      "loss": 1.2949,
      "step": 2944
    },
    {
      "epoch": 0.4218593324738576,
      "grad_norm": 1.074743628501892,
      "learning_rate": 0.00012972560034152562,
      "loss": 1.4902,
      "step": 2945
    },
    {
      "epoch": 0.42200257842715944,
      "grad_norm": 1.004320502281189,
      "learning_rate": 0.00012968129665906242,
      "loss": 1.4159,
      "step": 2946
    },
    {
      "epoch": 0.4221458243804613,
      "grad_norm": 1.2791447639465332,
      "learning_rate": 0.0001296369865869506,
      "loss": 1.4524,
      "step": 2947
    },
    {
      "epoch": 0.42228907033376306,
      "grad_norm": 1.2297550439834595,
      "learning_rate": 0.00012959267013472892,
      "loss": 1.4057,
      "step": 2948
    },
    {
      "epoch": 0.4224323162870649,
      "grad_norm": 1.2139707803726196,
      "learning_rate": 0.00012954834731193775,
      "loss": 1.4629,
      "step": 2949
    },
    {
      "epoch": 0.4225755622403667,
      "grad_norm": 0.8826777935028076,
      "learning_rate": 0.00012950401812811854,
      "loss": 1.3843,
      "step": 2950
    },
    {
      "epoch": 0.4227188081936685,
      "grad_norm": 1.01175856590271,
      "learning_rate": 0.00012945968259281437,
      "loss": 1.3502,
      "step": 2951
    },
    {
      "epoch": 0.42286205414697037,
      "grad_norm": 1.467247724533081,
      "learning_rate": 0.00012941534071556952,
      "loss": 1.3557,
      "step": 2952
    },
    {
      "epoch": 0.42300530010027215,
      "grad_norm": 1.1977813243865967,
      "learning_rate": 0.00012937099250592975,
      "loss": 1.4897,
      "step": 2953
    },
    {
      "epoch": 0.423148546053574,
      "grad_norm": 1.0589356422424316,
      "learning_rate": 0.00012932663797344214,
      "loss": 1.5279,
      "step": 2954
    },
    {
      "epoch": 0.42329179200687583,
      "grad_norm": 1.0807514190673828,
      "learning_rate": 0.00012928227712765504,
      "loss": 1.4172,
      "step": 2955
    },
    {
      "epoch": 0.4234350379601776,
      "grad_norm": 1.0285794734954834,
      "learning_rate": 0.00012923790997811834,
      "loss": 1.5824,
      "step": 2956
    },
    {
      "epoch": 0.42357828391347946,
      "grad_norm": 1.0368008613586426,
      "learning_rate": 0.00012919353653438313,
      "loss": 1.7461,
      "step": 2957
    },
    {
      "epoch": 0.42372152986678124,
      "grad_norm": 1.0917025804519653,
      "learning_rate": 0.00012914915680600195,
      "loss": 1.3869,
      "step": 2958
    },
    {
      "epoch": 0.4238647758200831,
      "grad_norm": 1.2777318954467773,
      "learning_rate": 0.00012910477080252858,
      "loss": 1.5453,
      "step": 2959
    },
    {
      "epoch": 0.4240080217733849,
      "grad_norm": 1.106281042098999,
      "learning_rate": 0.00012906037853351835,
      "loss": 1.286,
      "step": 2960
    },
    {
      "epoch": 0.4241512677266867,
      "grad_norm": 1.4309871196746826,
      "learning_rate": 0.00012901598000852774,
      "loss": 1.4638,
      "step": 2961
    },
    {
      "epoch": 0.42429451367998855,
      "grad_norm": 0.9520977735519409,
      "learning_rate": 0.0001289715752371147,
      "loss": 1.6378,
      "step": 2962
    },
    {
      "epoch": 0.42443775963329033,
      "grad_norm": 1.0166503190994263,
      "learning_rate": 0.00012892716422883838,
      "loss": 1.6441,
      "step": 2963
    },
    {
      "epoch": 0.4245810055865922,
      "grad_norm": 0.9065435528755188,
      "learning_rate": 0.0001288827469932595,
      "loss": 1.5039,
      "step": 2964
    },
    {
      "epoch": 0.424724251539894,
      "grad_norm": 0.9011521339416504,
      "learning_rate": 0.00012883832353993986,
      "loss": 1.4887,
      "step": 2965
    },
    {
      "epoch": 0.4248674974931958,
      "grad_norm": 1.3229097127914429,
      "learning_rate": 0.00012879389387844285,
      "loss": 1.4829,
      "step": 2966
    },
    {
      "epoch": 0.42501074344649764,
      "grad_norm": 1.1755478382110596,
      "learning_rate": 0.00012874945801833306,
      "loss": 1.2967,
      "step": 2967
    },
    {
      "epoch": 0.4251539893997995,
      "grad_norm": 1.1517176628112793,
      "learning_rate": 0.00012870501596917632,
      "loss": 1.6385,
      "step": 2968
    },
    {
      "epoch": 0.42529723535310127,
      "grad_norm": 1.0937844514846802,
      "learning_rate": 0.00012866056774054002,
      "loss": 1.5154,
      "step": 2969
    },
    {
      "epoch": 0.4254404813064031,
      "grad_norm": 1.2419525384902954,
      "learning_rate": 0.0001286161133419927,
      "loss": 1.5851,
      "step": 2970
    },
    {
      "epoch": 0.4255837272597049,
      "grad_norm": 1.0323566198349,
      "learning_rate": 0.00012857165278310436,
      "loss": 1.672,
      "step": 2971
    },
    {
      "epoch": 0.42572697321300673,
      "grad_norm": 1.0173742771148682,
      "learning_rate": 0.00012852718607344617,
      "loss": 1.4588,
      "step": 2972
    },
    {
      "epoch": 0.4258702191663086,
      "grad_norm": 1.046380639076233,
      "learning_rate": 0.0001284827132225908,
      "loss": 1.3932,
      "step": 2973
    },
    {
      "epoch": 0.42601346511961036,
      "grad_norm": 0.9603387713432312,
      "learning_rate": 0.00012843823424011207,
      "loss": 1.4069,
      "step": 2974
    },
    {
      "epoch": 0.4261567110729122,
      "grad_norm": 1.0051465034484863,
      "learning_rate": 0.00012839374913558527,
      "loss": 1.5149,
      "step": 2975
    },
    {
      "epoch": 0.426299957026214,
      "grad_norm": 1.2073255777359009,
      "learning_rate": 0.0001283492579185869,
      "loss": 1.4158,
      "step": 2976
    },
    {
      "epoch": 0.4264432029795158,
      "grad_norm": 0.9852157831192017,
      "learning_rate": 0.00012830476059869482,
      "loss": 1.4799,
      "step": 2977
    },
    {
      "epoch": 0.42658644893281766,
      "grad_norm": 1.0413848161697388,
      "learning_rate": 0.0001282602571854883,
      "loss": 1.343,
      "step": 2978
    },
    {
      "epoch": 0.42672969488611945,
      "grad_norm": 1.1822084188461304,
      "learning_rate": 0.00012821574768854765,
      "loss": 1.5502,
      "step": 2979
    },
    {
      "epoch": 0.4268729408394213,
      "grad_norm": 0.9565073251724243,
      "learning_rate": 0.0001281712321174548,
      "loss": 1.5571,
      "step": 2980
    },
    {
      "epoch": 0.42701618679272313,
      "grad_norm": 1.3902720212936401,
      "learning_rate": 0.00012812671048179276,
      "loss": 1.4454,
      "step": 2981
    },
    {
      "epoch": 0.4271594327460249,
      "grad_norm": 1.109741449356079,
      "learning_rate": 0.00012808218279114603,
      "loss": 1.5342,
      "step": 2982
    },
    {
      "epoch": 0.42730267869932675,
      "grad_norm": 1.0324418544769287,
      "learning_rate": 0.00012803764905510025,
      "loss": 1.4393,
      "step": 2983
    },
    {
      "epoch": 0.42744592465262854,
      "grad_norm": 1.1344586610794067,
      "learning_rate": 0.00012799310928324244,
      "loss": 1.4596,
      "step": 2984
    },
    {
      "epoch": 0.4275891706059304,
      "grad_norm": 1.2849620580673218,
      "learning_rate": 0.00012794856348516095,
      "loss": 1.376,
      "step": 2985
    },
    {
      "epoch": 0.4277324165592322,
      "grad_norm": 0.877849280834198,
      "learning_rate": 0.00012790401167044535,
      "loss": 1.3614,
      "step": 2986
    },
    {
      "epoch": 0.427875662512534,
      "grad_norm": 1.3338195085525513,
      "learning_rate": 0.00012785945384868658,
      "loss": 1.312,
      "step": 2987
    },
    {
      "epoch": 0.42801890846583585,
      "grad_norm": 0.9819057583808899,
      "learning_rate": 0.00012781489002947676,
      "loss": 1.2258,
      "step": 2988
    },
    {
      "epoch": 0.4281621544191377,
      "grad_norm": 1.21397864818573,
      "learning_rate": 0.00012777032022240947,
      "loss": 1.4147,
      "step": 2989
    },
    {
      "epoch": 0.42830540037243947,
      "grad_norm": 1.0348050594329834,
      "learning_rate": 0.00012772574443707942,
      "loss": 1.43,
      "step": 2990
    },
    {
      "epoch": 0.4284486463257413,
      "grad_norm": 1.079624891281128,
      "learning_rate": 0.0001276811626830827,
      "loss": 1.4671,
      "step": 2991
    },
    {
      "epoch": 0.4285918922790431,
      "grad_norm": 1.2637993097305298,
      "learning_rate": 0.00012763657497001668,
      "loss": 1.3709,
      "step": 2992
    },
    {
      "epoch": 0.42873513823234494,
      "grad_norm": 1.3295892477035522,
      "learning_rate": 0.00012759198130747994,
      "loss": 1.2281,
      "step": 2993
    },
    {
      "epoch": 0.4288783841856468,
      "grad_norm": 1.0518338680267334,
      "learning_rate": 0.00012754738170507243,
      "loss": 1.4146,
      "step": 2994
    },
    {
      "epoch": 0.42902163013894856,
      "grad_norm": 1.0216201543807983,
      "learning_rate": 0.00012750277617239533,
      "loss": 1.4083,
      "step": 2995
    },
    {
      "epoch": 0.4291648760922504,
      "grad_norm": 1.0603731870651245,
      "learning_rate": 0.00012745816471905114,
      "loss": 1.5712,
      "step": 2996
    },
    {
      "epoch": 0.4293081220455522,
      "grad_norm": 1.243438482284546,
      "learning_rate": 0.0001274135473546435,
      "loss": 1.5306,
      "step": 2997
    },
    {
      "epoch": 0.42945136799885403,
      "grad_norm": 1.1117500066757202,
      "learning_rate": 0.00012736892408877758,
      "loss": 1.4488,
      "step": 2998
    },
    {
      "epoch": 0.42959461395215587,
      "grad_norm": 1.16628897190094,
      "learning_rate": 0.0001273242949310595,
      "loss": 1.4502,
      "step": 2999
    },
    {
      "epoch": 0.42973785990545765,
      "grad_norm": 1.5213935375213623,
      "learning_rate": 0.00012727965989109692,
      "loss": 1.311,
      "step": 3000
    },
    {
      "epoch": 0.4298811058587595,
      "grad_norm": 1.4517474174499512,
      "learning_rate": 0.00012723501897849866,
      "loss": 1.5248,
      "step": 3001
    },
    {
      "epoch": 0.43002435181206133,
      "grad_norm": 1.2523372173309326,
      "learning_rate": 0.00012719037220287476,
      "loss": 1.225,
      "step": 3002
    },
    {
      "epoch": 0.4301675977653631,
      "grad_norm": 1.0150803327560425,
      "learning_rate": 0.00012714571957383658,
      "loss": 1.5225,
      "step": 3003
    },
    {
      "epoch": 0.43031084371866496,
      "grad_norm": 1.1522403955459595,
      "learning_rate": 0.0001271010611009967,
      "loss": 1.509,
      "step": 3004
    },
    {
      "epoch": 0.43045408967196674,
      "grad_norm": 1.2456793785095215,
      "learning_rate": 0.00012705639679396906,
      "loss": 1.5249,
      "step": 3005
    },
    {
      "epoch": 0.4305973356252686,
      "grad_norm": 1.0238828659057617,
      "learning_rate": 0.00012701172666236869,
      "loss": 1.5292,
      "step": 3006
    },
    {
      "epoch": 0.4307405815785704,
      "grad_norm": 1.2697727680206299,
      "learning_rate": 0.00012696705071581205,
      "loss": 1.45,
      "step": 3007
    },
    {
      "epoch": 0.4308838275318722,
      "grad_norm": 1.0821571350097656,
      "learning_rate": 0.0001269223689639167,
      "loss": 1.4024,
      "step": 3008
    },
    {
      "epoch": 0.43102707348517405,
      "grad_norm": 1.0689291954040527,
      "learning_rate": 0.00012687768141630152,
      "loss": 1.4816,
      "step": 3009
    },
    {
      "epoch": 0.43117031943847584,
      "grad_norm": 1.1147712469100952,
      "learning_rate": 0.00012683298808258665,
      "loss": 1.5372,
      "step": 3010
    },
    {
      "epoch": 0.4313135653917777,
      "grad_norm": 1.0224536657333374,
      "learning_rate": 0.00012678828897239348,
      "loss": 1.5718,
      "step": 3011
    },
    {
      "epoch": 0.4314568113450795,
      "grad_norm": 1.2083054780960083,
      "learning_rate": 0.00012674358409534462,
      "loss": 1.3513,
      "step": 3012
    },
    {
      "epoch": 0.4316000572983813,
      "grad_norm": 1.0714867115020752,
      "learning_rate": 0.00012669887346106385,
      "loss": 1.5939,
      "step": 3013
    },
    {
      "epoch": 0.43174330325168314,
      "grad_norm": 1.3239539861679077,
      "learning_rate": 0.00012665415707917638,
      "loss": 1.4316,
      "step": 3014
    },
    {
      "epoch": 0.431886549204985,
      "grad_norm": 1.2914295196533203,
      "learning_rate": 0.00012660943495930845,
      "loss": 1.3322,
      "step": 3015
    },
    {
      "epoch": 0.43202979515828677,
      "grad_norm": 1.2645832300186157,
      "learning_rate": 0.00012656470711108764,
      "loss": 1.5821,
      "step": 3016
    },
    {
      "epoch": 0.4321730411115886,
      "grad_norm": 1.0351508855819702,
      "learning_rate": 0.00012651997354414278,
      "loss": 1.4938,
      "step": 3017
    },
    {
      "epoch": 0.4323162870648904,
      "grad_norm": 1.1938503980636597,
      "learning_rate": 0.00012647523426810386,
      "loss": 1.3562,
      "step": 3018
    },
    {
      "epoch": 0.43245953301819223,
      "grad_norm": 0.9267191290855408,
      "learning_rate": 0.00012643048929260222,
      "loss": 1.5197,
      "step": 3019
    },
    {
      "epoch": 0.4326027789714941,
      "grad_norm": 0.9674130082130432,
      "learning_rate": 0.00012638573862727023,
      "loss": 1.3878,
      "step": 3020
    },
    {
      "epoch": 0.43274602492479586,
      "grad_norm": 1.1477378606796265,
      "learning_rate": 0.0001263409822817417,
      "loss": 1.4516,
      "step": 3021
    },
    {
      "epoch": 0.4328892708780977,
      "grad_norm": 1.2554242610931396,
      "learning_rate": 0.00012629622026565147,
      "loss": 1.3433,
      "step": 3022
    },
    {
      "epoch": 0.43303251683139954,
      "grad_norm": 1.2385050058364868,
      "learning_rate": 0.0001262514525886358,
      "loss": 1.5042,
      "step": 3023
    },
    {
      "epoch": 0.4331757627847013,
      "grad_norm": 1.1519674062728882,
      "learning_rate": 0.00012620667926033197,
      "loss": 1.3635,
      "step": 3024
    },
    {
      "epoch": 0.43331900873800316,
      "grad_norm": 1.0280259847640991,
      "learning_rate": 0.00012616190029037864,
      "loss": 1.5123,
      "step": 3025
    },
    {
      "epoch": 0.43346225469130495,
      "grad_norm": 1.2098954916000366,
      "learning_rate": 0.00012611711568841558,
      "loss": 1.5378,
      "step": 3026
    },
    {
      "epoch": 0.4336055006446068,
      "grad_norm": 1.0682852268218994,
      "learning_rate": 0.0001260723254640838,
      "loss": 1.368,
      "step": 3027
    },
    {
      "epoch": 0.43374874659790863,
      "grad_norm": 1.0201326608657837,
      "learning_rate": 0.00012602752962702555,
      "loss": 1.2167,
      "step": 3028
    },
    {
      "epoch": 0.4338919925512104,
      "grad_norm": 0.9865608215332031,
      "learning_rate": 0.00012598272818688423,
      "loss": 1.724,
      "step": 3029
    },
    {
      "epoch": 0.43403523850451226,
      "grad_norm": 1.0396380424499512,
      "learning_rate": 0.00012593792115330455,
      "loss": 1.3534,
      "step": 3030
    },
    {
      "epoch": 0.43417848445781404,
      "grad_norm": 1.1980247497558594,
      "learning_rate": 0.00012589310853593224,
      "loss": 1.4163,
      "step": 3031
    },
    {
      "epoch": 0.4343217304111159,
      "grad_norm": 1.1176769733428955,
      "learning_rate": 0.00012584829034441446,
      "loss": 1.4575,
      "step": 3032
    },
    {
      "epoch": 0.4344649763644177,
      "grad_norm": 1.1439476013183594,
      "learning_rate": 0.00012580346658839944,
      "loss": 1.368,
      "step": 3033
    },
    {
      "epoch": 0.4346082223177195,
      "grad_norm": 1.2689814567565918,
      "learning_rate": 0.00012575863727753655,
      "loss": 1.2865,
      "step": 3034
    },
    {
      "epoch": 0.43475146827102135,
      "grad_norm": 1.260332465171814,
      "learning_rate": 0.0001257138024214765,
      "loss": 1.3918,
      "step": 3035
    },
    {
      "epoch": 0.4348947142243232,
      "grad_norm": 1.0823941230773926,
      "learning_rate": 0.0001256689620298711,
      "loss": 1.5517,
      "step": 3036
    },
    {
      "epoch": 0.43503796017762497,
      "grad_norm": 1.1936004161834717,
      "learning_rate": 0.00012562411611237342,
      "loss": 1.5122,
      "step": 3037
    },
    {
      "epoch": 0.4351812061309268,
      "grad_norm": 0.9758128523826599,
      "learning_rate": 0.00012557926467863758,
      "loss": 1.4986,
      "step": 3038
    },
    {
      "epoch": 0.4353244520842286,
      "grad_norm": 1.4188203811645508,
      "learning_rate": 0.0001255344077383191,
      "loss": 1.471,
      "step": 3039
    },
    {
      "epoch": 0.43546769803753044,
      "grad_norm": 1.0242336988449097,
      "learning_rate": 0.00012548954530107448,
      "loss": 1.4611,
      "step": 3040
    },
    {
      "epoch": 0.4356109439908323,
      "grad_norm": 1.3733683824539185,
      "learning_rate": 0.00012544467737656153,
      "loss": 1.5593,
      "step": 3041
    },
    {
      "epoch": 0.43575418994413406,
      "grad_norm": 1.0285357236862183,
      "learning_rate": 0.0001253998039744392,
      "loss": 1.3345,
      "step": 3042
    },
    {
      "epoch": 0.4358974358974359,
      "grad_norm": 0.9556021690368652,
      "learning_rate": 0.00012535492510436762,
      "loss": 1.4751,
      "step": 3043
    },
    {
      "epoch": 0.43604068185073774,
      "grad_norm": 1.094390630722046,
      "learning_rate": 0.0001253100407760081,
      "loss": 1.4566,
      "step": 3044
    },
    {
      "epoch": 0.43618392780403953,
      "grad_norm": 1.3010364770889282,
      "learning_rate": 0.00012526515099902313,
      "loss": 1.4804,
      "step": 3045
    },
    {
      "epoch": 0.43632717375734137,
      "grad_norm": 1.2758674621582031,
      "learning_rate": 0.00012522025578307638,
      "loss": 1.2753,
      "step": 3046
    },
    {
      "epoch": 0.43647041971064315,
      "grad_norm": 0.996068000793457,
      "learning_rate": 0.00012517535513783263,
      "loss": 1.5083,
      "step": 3047
    },
    {
      "epoch": 0.436613665663945,
      "grad_norm": 1.0624992847442627,
      "learning_rate": 0.00012513044907295795,
      "loss": 1.43,
      "step": 3048
    },
    {
      "epoch": 0.43675691161724683,
      "grad_norm": 1.090915322303772,
      "learning_rate": 0.00012508553759811946,
      "loss": 1.3936,
      "step": 3049
    },
    {
      "epoch": 0.4369001575705486,
      "grad_norm": 1.141618013381958,
      "learning_rate": 0.00012504062072298544,
      "loss": 1.4458,
      "step": 3050
    },
    {
      "epoch": 0.43704340352385046,
      "grad_norm": 1.0165143013000488,
      "learning_rate": 0.0001249956984572255,
      "loss": 1.3128,
      "step": 3051
    },
    {
      "epoch": 0.43718664947715224,
      "grad_norm": 0.9935486912727356,
      "learning_rate": 0.0001249507708105102,
      "loss": 1.4738,
      "step": 3052
    },
    {
      "epoch": 0.4373298954304541,
      "grad_norm": 1.0991129875183105,
      "learning_rate": 0.00012490583779251142,
      "loss": 1.547,
      "step": 3053
    },
    {
      "epoch": 0.4374731413837559,
      "grad_norm": 0.9572281241416931,
      "learning_rate": 0.00012486089941290206,
      "loss": 1.4446,
      "step": 3054
    },
    {
      "epoch": 0.4376163873370577,
      "grad_norm": 1.2859573364257812,
      "learning_rate": 0.00012481595568135628,
      "loss": 1.3339,
      "step": 3055
    },
    {
      "epoch": 0.43775963329035955,
      "grad_norm": 1.1983587741851807,
      "learning_rate": 0.00012477100660754933,
      "loss": 1.3494,
      "step": 3056
    },
    {
      "epoch": 0.4379028792436614,
      "grad_norm": 1.1122018098831177,
      "learning_rate": 0.00012472605220115765,
      "loss": 1.5065,
      "step": 3057
    },
    {
      "epoch": 0.4380461251969632,
      "grad_norm": 1.138338327407837,
      "learning_rate": 0.0001246810924718588,
      "loss": 1.4528,
      "step": 3058
    },
    {
      "epoch": 0.438189371150265,
      "grad_norm": 1.239978313446045,
      "learning_rate": 0.00012463612742933148,
      "loss": 1.4125,
      "step": 3059
    },
    {
      "epoch": 0.4383326171035668,
      "grad_norm": 1.0537867546081543,
      "learning_rate": 0.0001245911570832556,
      "loss": 1.2958,
      "step": 3060
    },
    {
      "epoch": 0.43847586305686864,
      "grad_norm": 0.9920015335083008,
      "learning_rate": 0.00012454618144331213,
      "loss": 1.3586,
      "step": 3061
    },
    {
      "epoch": 0.4386191090101705,
      "grad_norm": 1.1013517379760742,
      "learning_rate": 0.00012450120051918324,
      "loss": 1.468,
      "step": 3062
    },
    {
      "epoch": 0.43876235496347227,
      "grad_norm": 0.9714090824127197,
      "learning_rate": 0.00012445621432055214,
      "loss": 1.4832,
      "step": 3063
    },
    {
      "epoch": 0.4389056009167741,
      "grad_norm": 0.9250453114509583,
      "learning_rate": 0.00012441122285710335,
      "loss": 1.5898,
      "step": 3064
    },
    {
      "epoch": 0.4390488468700759,
      "grad_norm": 1.21511709690094,
      "learning_rate": 0.00012436622613852234,
      "loss": 1.391,
      "step": 3065
    },
    {
      "epoch": 0.43919209282337773,
      "grad_norm": 1.0050164461135864,
      "learning_rate": 0.0001243212241744958,
      "loss": 1.4469,
      "step": 3066
    },
    {
      "epoch": 0.4393353387766796,
      "grad_norm": 1.2763065099716187,
      "learning_rate": 0.00012427621697471157,
      "loss": 1.3759,
      "step": 3067
    },
    {
      "epoch": 0.43947858472998136,
      "grad_norm": 1.1281070709228516,
      "learning_rate": 0.00012423120454885857,
      "loss": 1.4813,
      "step": 3068
    },
    {
      "epoch": 0.4396218306832832,
      "grad_norm": 1.0350903272628784,
      "learning_rate": 0.00012418618690662685,
      "loss": 1.3968,
      "step": 3069
    },
    {
      "epoch": 0.43976507663658504,
      "grad_norm": 1.120147466659546,
      "learning_rate": 0.00012414116405770758,
      "loss": 1.3834,
      "step": 3070
    },
    {
      "epoch": 0.4399083225898868,
      "grad_norm": 0.9993228912353516,
      "learning_rate": 0.00012409613601179316,
      "loss": 1.5349,
      "step": 3071
    },
    {
      "epoch": 0.44005156854318866,
      "grad_norm": 1.1676287651062012,
      "learning_rate": 0.00012405110277857685,
      "loss": 1.4776,
      "step": 3072
    },
    {
      "epoch": 0.44019481449649045,
      "grad_norm": 1.1972886323928833,
      "learning_rate": 0.00012400606436775336,
      "loss": 1.4927,
      "step": 3073
    },
    {
      "epoch": 0.4403380604497923,
      "grad_norm": 1.2024924755096436,
      "learning_rate": 0.00012396102078901823,
      "loss": 1.5038,
      "step": 3074
    },
    {
      "epoch": 0.44048130640309413,
      "grad_norm": 0.9534226655960083,
      "learning_rate": 0.00012391597205206828,
      "loss": 1.2454,
      "step": 3075
    },
    {
      "epoch": 0.4406245523563959,
      "grad_norm": 1.0901786088943481,
      "learning_rate": 0.00012387091816660136,
      "loss": 1.4556,
      "step": 3076
    },
    {
      "epoch": 0.44076779830969776,
      "grad_norm": 1.1119468212127686,
      "learning_rate": 0.0001238258591423165,
      "loss": 1.4172,
      "step": 3077
    },
    {
      "epoch": 0.4409110442629996,
      "grad_norm": 1.366089105606079,
      "learning_rate": 0.00012378079498891377,
      "loss": 1.4831,
      "step": 3078
    },
    {
      "epoch": 0.4410542902163014,
      "grad_norm": 1.2530357837677002,
      "learning_rate": 0.00012373572571609432,
      "loss": 1.2852,
      "step": 3079
    },
    {
      "epoch": 0.4411975361696032,
      "grad_norm": 1.0945103168487549,
      "learning_rate": 0.00012369065133356052,
      "loss": 1.4229,
      "step": 3080
    },
    {
      "epoch": 0.441340782122905,
      "grad_norm": 1.0497305393218994,
      "learning_rate": 0.00012364557185101576,
      "loss": 1.3797,
      "step": 3081
    },
    {
      "epoch": 0.44148402807620685,
      "grad_norm": 1.1022846698760986,
      "learning_rate": 0.00012360048727816448,
      "loss": 1.4378,
      "step": 3082
    },
    {
      "epoch": 0.4416272740295087,
      "grad_norm": 1.064237356185913,
      "learning_rate": 0.00012355539762471234,
      "loss": 1.5383,
      "step": 3083
    },
    {
      "epoch": 0.4417705199828105,
      "grad_norm": 1.0266364812850952,
      "learning_rate": 0.000123510302900366,
      "loss": 1.3943,
      "step": 3084
    },
    {
      "epoch": 0.4419137659361123,
      "grad_norm": 1.1215288639068604,
      "learning_rate": 0.00012346520311483318,
      "loss": 1.5223,
      "step": 3085
    },
    {
      "epoch": 0.4420570118894141,
      "grad_norm": 1.3592530488967896,
      "learning_rate": 0.00012342009827782284,
      "loss": 1.2945,
      "step": 3086
    },
    {
      "epoch": 0.44220025784271594,
      "grad_norm": 1.1827378273010254,
      "learning_rate": 0.00012337498839904492,
      "loss": 1.379,
      "step": 3087
    },
    {
      "epoch": 0.4423435037960178,
      "grad_norm": 1.3468207120895386,
      "learning_rate": 0.0001233298734882104,
      "loss": 1.4093,
      "step": 3088
    },
    {
      "epoch": 0.44248674974931956,
      "grad_norm": 1.1650348901748657,
      "learning_rate": 0.00012328475355503145,
      "loss": 1.3834,
      "step": 3089
    },
    {
      "epoch": 0.4426299957026214,
      "grad_norm": 1.31522798538208,
      "learning_rate": 0.0001232396286092213,
      "loss": 1.4305,
      "step": 3090
    },
    {
      "epoch": 0.44277324165592324,
      "grad_norm": 1.1427093744277954,
      "learning_rate": 0.00012319449866049416,
      "loss": 1.4521,
      "step": 3091
    },
    {
      "epoch": 0.44291648760922503,
      "grad_norm": 1.3706148862838745,
      "learning_rate": 0.00012314936371856543,
      "loss": 1.5094,
      "step": 3092
    },
    {
      "epoch": 0.44305973356252687,
      "grad_norm": 1.0026769638061523,
      "learning_rate": 0.00012310422379315162,
      "loss": 1.5859,
      "step": 3093
    },
    {
      "epoch": 0.44320297951582865,
      "grad_norm": 1.142472267150879,
      "learning_rate": 0.0001230590788939701,
      "loss": 1.5192,
      "step": 3094
    },
    {
      "epoch": 0.4433462254691305,
      "grad_norm": 1.11885666847229,
      "learning_rate": 0.00012301392903073954,
      "loss": 1.4231,
      "step": 3095
    },
    {
      "epoch": 0.44348947142243234,
      "grad_norm": 1.2014662027359009,
      "learning_rate": 0.0001229687742131796,
      "loss": 1.5498,
      "step": 3096
    },
    {
      "epoch": 0.4436327173757341,
      "grad_norm": 1.4500850439071655,
      "learning_rate": 0.0001229236144510109,
      "loss": 1.4122,
      "step": 3097
    },
    {
      "epoch": 0.44377596332903596,
      "grad_norm": 1.1103951930999756,
      "learning_rate": 0.0001228784497539553,
      "loss": 1.4943,
      "step": 3098
    },
    {
      "epoch": 0.4439192092823378,
      "grad_norm": 1.1672239303588867,
      "learning_rate": 0.00012283328013173563,
      "loss": 1.4736,
      "step": 3099
    },
    {
      "epoch": 0.4440624552356396,
      "grad_norm": 1.0671937465667725,
      "learning_rate": 0.00012278810559407578,
      "loss": 1.4096,
      "step": 3100
    },
    {
      "epoch": 0.4442057011889414,
      "grad_norm": 1.1332948207855225,
      "learning_rate": 0.00012274292615070068,
      "loss": 1.4526,
      "step": 3101
    },
    {
      "epoch": 0.4443489471422432,
      "grad_norm": 1.0210916996002197,
      "learning_rate": 0.00012269774181133643,
      "loss": 1.6282,
      "step": 3102
    },
    {
      "epoch": 0.44449219309554505,
      "grad_norm": 1.0761029720306396,
      "learning_rate": 0.00012265255258571005,
      "loss": 1.299,
      "step": 3103
    },
    {
      "epoch": 0.4446354390488469,
      "grad_norm": 0.9963146448135376,
      "learning_rate": 0.00012260735848354962,
      "loss": 1.3939,
      "step": 3104
    },
    {
      "epoch": 0.4447786850021487,
      "grad_norm": 1.0436007976531982,
      "learning_rate": 0.00012256215951458437,
      "loss": 1.562,
      "step": 3105
    },
    {
      "epoch": 0.4449219309554505,
      "grad_norm": 1.1525242328643799,
      "learning_rate": 0.00012251695568854453,
      "loss": 1.347,
      "step": 3106
    },
    {
      "epoch": 0.4450651769087523,
      "grad_norm": 1.0592950582504272,
      "learning_rate": 0.00012247174701516134,
      "loss": 1.3837,
      "step": 3107
    },
    {
      "epoch": 0.44520842286205414,
      "grad_norm": 1.1663986444473267,
      "learning_rate": 0.00012242653350416708,
      "loss": 1.4266,
      "step": 3108
    },
    {
      "epoch": 0.445351668815356,
      "grad_norm": 1.423520565032959,
      "learning_rate": 0.00012238131516529514,
      "loss": 1.448,
      "step": 3109
    },
    {
      "epoch": 0.44549491476865777,
      "grad_norm": 1.1414059400558472,
      "learning_rate": 0.00012233609200827986,
      "loss": 1.339,
      "step": 3110
    },
    {
      "epoch": 0.4456381607219596,
      "grad_norm": 1.3034290075302124,
      "learning_rate": 0.00012229086404285674,
      "loss": 1.403,
      "step": 3111
    },
    {
      "epoch": 0.44578140667526145,
      "grad_norm": 1.1641031503677368,
      "learning_rate": 0.00012224563127876222,
      "loss": 1.5103,
      "step": 3112
    },
    {
      "epoch": 0.44592465262856323,
      "grad_norm": 1.1962319612503052,
      "learning_rate": 0.00012220039372573373,
      "loss": 1.2811,
      "step": 3113
    },
    {
      "epoch": 0.4460678985818651,
      "grad_norm": 1.1172598600387573,
      "learning_rate": 0.0001221551513935099,
      "loss": 1.4634,
      "step": 3114
    },
    {
      "epoch": 0.44621114453516686,
      "grad_norm": 1.0949052572250366,
      "learning_rate": 0.0001221099042918302,
      "loss": 1.4911,
      "step": 3115
    },
    {
      "epoch": 0.4463543904884687,
      "grad_norm": 1.384107232093811,
      "learning_rate": 0.00012206465243043525,
      "loss": 1.4994,
      "step": 3116
    },
    {
      "epoch": 0.44649763644177054,
      "grad_norm": 1.1620814800262451,
      "learning_rate": 0.00012201939581906662,
      "loss": 1.4314,
      "step": 3117
    },
    {
      "epoch": 0.4466408823950723,
      "grad_norm": 1.1559195518493652,
      "learning_rate": 0.00012197413446746702,
      "loss": 1.4812,
      "step": 3118
    },
    {
      "epoch": 0.44678412834837417,
      "grad_norm": 1.0727044343948364,
      "learning_rate": 0.00012192886838538,
      "loss": 1.4508,
      "step": 3119
    },
    {
      "epoch": 0.44692737430167595,
      "grad_norm": 1.341267704963684,
      "learning_rate": 0.00012188359758255028,
      "loss": 1.5079,
      "step": 3120
    },
    {
      "epoch": 0.4470706202549778,
      "grad_norm": 1.0778058767318726,
      "learning_rate": 0.00012183832206872355,
      "loss": 1.4923,
      "step": 3121
    },
    {
      "epoch": 0.44721386620827963,
      "grad_norm": 1.0122088193893433,
      "learning_rate": 0.00012179304185364646,
      "loss": 1.5925,
      "step": 3122
    },
    {
      "epoch": 0.4473571121615814,
      "grad_norm": 1.0031112432479858,
      "learning_rate": 0.00012174775694706679,
      "loss": 1.516,
      "step": 3123
    },
    {
      "epoch": 0.44750035811488326,
      "grad_norm": 1.1362916231155396,
      "learning_rate": 0.00012170246735873321,
      "loss": 1.4884,
      "step": 3124
    },
    {
      "epoch": 0.4476436040681851,
      "grad_norm": 1.112246036529541,
      "learning_rate": 0.00012165717309839548,
      "loss": 1.4312,
      "step": 3125
    },
    {
      "epoch": 0.4477868500214869,
      "grad_norm": 1.0987281799316406,
      "learning_rate": 0.00012161187417580427,
      "loss": 1.5845,
      "step": 3126
    },
    {
      "epoch": 0.4479300959747887,
      "grad_norm": 1.0815194845199585,
      "learning_rate": 0.0001215665706007114,
      "loss": 1.4914,
      "step": 3127
    },
    {
      "epoch": 0.4480733419280905,
      "grad_norm": 1.2938653230667114,
      "learning_rate": 0.00012152126238286953,
      "loss": 1.5168,
      "step": 3128
    },
    {
      "epoch": 0.44821658788139235,
      "grad_norm": 1.1902167797088623,
      "learning_rate": 0.00012147594953203247,
      "loss": 1.5087,
      "step": 3129
    },
    {
      "epoch": 0.4483598338346942,
      "grad_norm": 1.146147608757019,
      "learning_rate": 0.0001214306320579549,
      "loss": 1.448,
      "step": 3130
    },
    {
      "epoch": 0.448503079787996,
      "grad_norm": 1.1689090728759766,
      "learning_rate": 0.00012138530997039259,
      "loss": 1.6199,
      "step": 3131
    },
    {
      "epoch": 0.4486463257412978,
      "grad_norm": 1.1992183923721313,
      "learning_rate": 0.00012133998327910225,
      "loss": 1.4455,
      "step": 3132
    },
    {
      "epoch": 0.44878957169459965,
      "grad_norm": 1.2008085250854492,
      "learning_rate": 0.00012129465199384157,
      "loss": 1.5091,
      "step": 3133
    },
    {
      "epoch": 0.44893281764790144,
      "grad_norm": 1.1654683351516724,
      "learning_rate": 0.00012124931612436932,
      "loss": 1.494,
      "step": 3134
    },
    {
      "epoch": 0.4490760636012033,
      "grad_norm": 1.1424264907836914,
      "learning_rate": 0.0001212039756804451,
      "loss": 1.4086,
      "step": 3135
    },
    {
      "epoch": 0.44921930955450506,
      "grad_norm": 1.245821475982666,
      "learning_rate": 0.00012115863067182967,
      "loss": 1.4057,
      "step": 3136
    },
    {
      "epoch": 0.4493625555078069,
      "grad_norm": 1.4618412256240845,
      "learning_rate": 0.00012111328110828464,
      "loss": 1.3526,
      "step": 3137
    },
    {
      "epoch": 0.44950580146110874,
      "grad_norm": 1.0155736207962036,
      "learning_rate": 0.00012106792699957263,
      "loss": 1.6852,
      "step": 3138
    },
    {
      "epoch": 0.44964904741441053,
      "grad_norm": 1.0592164993286133,
      "learning_rate": 0.00012102256835545734,
      "loss": 1.3375,
      "step": 3139
    },
    {
      "epoch": 0.44979229336771237,
      "grad_norm": 1.0476815700531006,
      "learning_rate": 0.00012097720518570326,
      "loss": 1.441,
      "step": 3140
    },
    {
      "epoch": 0.44993553932101416,
      "grad_norm": 1.0410012006759644,
      "learning_rate": 0.00012093183750007606,
      "loss": 1.4251,
      "step": 3141
    },
    {
      "epoch": 0.450078785274316,
      "grad_norm": 1.078812599182129,
      "learning_rate": 0.00012088646530834218,
      "loss": 1.5098,
      "step": 3142
    },
    {
      "epoch": 0.45022203122761784,
      "grad_norm": 1.173561930656433,
      "learning_rate": 0.0001208410886202692,
      "loss": 1.4976,
      "step": 3143
    },
    {
      "epoch": 0.4503652771809196,
      "grad_norm": 0.9869365692138672,
      "learning_rate": 0.00012079570744562558,
      "loss": 1.3745,
      "step": 3144
    },
    {
      "epoch": 0.45050852313422146,
      "grad_norm": 1.0315371751785278,
      "learning_rate": 0.00012075032179418076,
      "loss": 1.4201,
      "step": 3145
    },
    {
      "epoch": 0.4506517690875233,
      "grad_norm": 1.137511968612671,
      "learning_rate": 0.00012070493167570516,
      "loss": 1.6021,
      "step": 3146
    },
    {
      "epoch": 0.4507950150408251,
      "grad_norm": 1.209009051322937,
      "learning_rate": 0.00012065953709997009,
      "loss": 1.5155,
      "step": 3147
    },
    {
      "epoch": 0.4509382609941269,
      "grad_norm": 1.0661952495574951,
      "learning_rate": 0.00012061413807674797,
      "loss": 1.3594,
      "step": 3148
    },
    {
      "epoch": 0.4510815069474287,
      "grad_norm": 0.9781462550163269,
      "learning_rate": 0.00012056873461581204,
      "loss": 1.4059,
      "step": 3149
    },
    {
      "epoch": 0.45122475290073055,
      "grad_norm": 1.1417714357376099,
      "learning_rate": 0.00012052332672693656,
      "loss": 1.3588,
      "step": 3150
    },
    {
      "epoch": 0.4513679988540324,
      "grad_norm": 1.4227643013000488,
      "learning_rate": 0.00012047791441989665,
      "loss": 1.6042,
      "step": 3151
    },
    {
      "epoch": 0.4515112448073342,
      "grad_norm": 1.178970456123352,
      "learning_rate": 0.00012043249770446856,
      "loss": 1.3268,
      "step": 3152
    },
    {
      "epoch": 0.451654490760636,
      "grad_norm": 1.204473614692688,
      "learning_rate": 0.00012038707659042934,
      "loss": 1.5764,
      "step": 3153
    },
    {
      "epoch": 0.45179773671393786,
      "grad_norm": 0.9864242076873779,
      "learning_rate": 0.00012034165108755702,
      "loss": 1.5221,
      "step": 3154
    },
    {
      "epoch": 0.45194098266723964,
      "grad_norm": 1.1069221496582031,
      "learning_rate": 0.0001202962212056306,
      "loss": 1.544,
      "step": 3155
    },
    {
      "epoch": 0.4520842286205415,
      "grad_norm": 0.9879958629608154,
      "learning_rate": 0.00012025078695442999,
      "loss": 1.4061,
      "step": 3156
    },
    {
      "epoch": 0.45222747457384327,
      "grad_norm": 1.103913426399231,
      "learning_rate": 0.00012020534834373612,
      "loss": 1.6199,
      "step": 3157
    },
    {
      "epoch": 0.4523707205271451,
      "grad_norm": 1.313738226890564,
      "learning_rate": 0.00012015990538333073,
      "loss": 1.6016,
      "step": 3158
    },
    {
      "epoch": 0.45251396648044695,
      "grad_norm": 1.0389853715896606,
      "learning_rate": 0.00012011445808299659,
      "loss": 1.5377,
      "step": 3159
    },
    {
      "epoch": 0.45265721243374873,
      "grad_norm": 1.0367650985717773,
      "learning_rate": 0.0001200690064525174,
      "loss": 1.3871,
      "step": 3160
    },
    {
      "epoch": 0.4528004583870506,
      "grad_norm": 1.2739322185516357,
      "learning_rate": 0.00012002355050167776,
      "loss": 1.2641,
      "step": 3161
    },
    {
      "epoch": 0.45294370434035236,
      "grad_norm": 1.089424967765808,
      "learning_rate": 0.00011997809024026316,
      "loss": 1.4099,
      "step": 3162
    },
    {
      "epoch": 0.4530869502936542,
      "grad_norm": 1.3017481565475464,
      "learning_rate": 0.00011993262567806012,
      "loss": 1.4638,
      "step": 3163
    },
    {
      "epoch": 0.45323019624695604,
      "grad_norm": 1.4463725090026855,
      "learning_rate": 0.00011988715682485602,
      "loss": 1.4422,
      "step": 3164
    },
    {
      "epoch": 0.4533734422002578,
      "grad_norm": 0.990341067314148,
      "learning_rate": 0.00011984168369043922,
      "loss": 1.5195,
      "step": 3165
    },
    {
      "epoch": 0.45351668815355967,
      "grad_norm": 1.113857388496399,
      "learning_rate": 0.00011979620628459893,
      "loss": 1.5846,
      "step": 3166
    },
    {
      "epoch": 0.4536599341068615,
      "grad_norm": 1.1528223752975464,
      "learning_rate": 0.00011975072461712527,
      "loss": 1.5592,
      "step": 3167
    },
    {
      "epoch": 0.4538031800601633,
      "grad_norm": 1.2071025371551514,
      "learning_rate": 0.00011970523869780938,
      "loss": 1.278,
      "step": 3168
    },
    {
      "epoch": 0.45394642601346513,
      "grad_norm": 1.0715287923812866,
      "learning_rate": 0.00011965974853644321,
      "loss": 1.3596,
      "step": 3169
    },
    {
      "epoch": 0.4540896719667669,
      "grad_norm": 1.0038518905639648,
      "learning_rate": 0.0001196142541428197,
      "loss": 1.5137,
      "step": 3170
    },
    {
      "epoch": 0.45423291792006876,
      "grad_norm": 0.8300107717514038,
      "learning_rate": 0.00011956875552673268,
      "loss": 1.4952,
      "step": 3171
    },
    {
      "epoch": 0.4543761638733706,
      "grad_norm": 0.9304805994033813,
      "learning_rate": 0.00011952325269797677,
      "loss": 1.361,
      "step": 3172
    },
    {
      "epoch": 0.4545194098266724,
      "grad_norm": 0.8712587952613831,
      "learning_rate": 0.00011947774566634775,
      "loss": 1.5969,
      "step": 3173
    },
    {
      "epoch": 0.4546626557799742,
      "grad_norm": 1.1067285537719727,
      "learning_rate": 0.00011943223444164205,
      "loss": 1.5286,
      "step": 3174
    },
    {
      "epoch": 0.454805901733276,
      "grad_norm": 1.157167911529541,
      "learning_rate": 0.00011938671903365717,
      "loss": 1.3667,
      "step": 3175
    },
    {
      "epoch": 0.45494914768657785,
      "grad_norm": 0.8720133304595947,
      "learning_rate": 0.0001193411994521914,
      "loss": 1.4309,
      "step": 3176
    },
    {
      "epoch": 0.4550923936398797,
      "grad_norm": 1.030298113822937,
      "learning_rate": 0.00011929567570704403,
      "loss": 1.3805,
      "step": 3177
    },
    {
      "epoch": 0.4552356395931815,
      "grad_norm": 1.0273535251617432,
      "learning_rate": 0.00011925014780801516,
      "loss": 1.3773,
      "step": 3178
    },
    {
      "epoch": 0.4553788855464833,
      "grad_norm": 0.9274500012397766,
      "learning_rate": 0.00011920461576490584,
      "loss": 1.4026,
      "step": 3179
    },
    {
      "epoch": 0.45552213149978515,
      "grad_norm": 1.0567808151245117,
      "learning_rate": 0.000119159079587518,
      "loss": 1.5002,
      "step": 3180
    },
    {
      "epoch": 0.45566537745308694,
      "grad_norm": 1.0315468311309814,
      "learning_rate": 0.0001191135392856544,
      "loss": 1.5916,
      "step": 3181
    },
    {
      "epoch": 0.4558086234063888,
      "grad_norm": 1.203462839126587,
      "learning_rate": 0.00011906799486911884,
      "loss": 1.5059,
      "step": 3182
    },
    {
      "epoch": 0.45595186935969056,
      "grad_norm": 1.271903157234192,
      "learning_rate": 0.0001190224463477158,
      "loss": 1.389,
      "step": 3183
    },
    {
      "epoch": 0.4560951153129924,
      "grad_norm": 1.1927798986434937,
      "learning_rate": 0.00011897689373125081,
      "loss": 1.3348,
      "step": 3184
    },
    {
      "epoch": 0.45623836126629425,
      "grad_norm": 1.182523250579834,
      "learning_rate": 0.00011893133702953018,
      "loss": 1.4619,
      "step": 3185
    },
    {
      "epoch": 0.45638160721959603,
      "grad_norm": 1.153469204902649,
      "learning_rate": 0.00011888577625236122,
      "loss": 1.3313,
      "step": 3186
    },
    {
      "epoch": 0.45652485317289787,
      "grad_norm": 1.1265850067138672,
      "learning_rate": 0.00011884021140955193,
      "loss": 1.4595,
      "step": 3187
    },
    {
      "epoch": 0.4566680991261997,
      "grad_norm": 0.9332097172737122,
      "learning_rate": 0.00011879464251091135,
      "loss": 1.495,
      "step": 3188
    },
    {
      "epoch": 0.4568113450795015,
      "grad_norm": 1.0890226364135742,
      "learning_rate": 0.00011874906956624934,
      "loss": 1.6299,
      "step": 3189
    },
    {
      "epoch": 0.45695459103280334,
      "grad_norm": 1.1782417297363281,
      "learning_rate": 0.00011870349258537663,
      "loss": 1.424,
      "step": 3190
    },
    {
      "epoch": 0.4570978369861051,
      "grad_norm": 1.16672682762146,
      "learning_rate": 0.00011865791157810482,
      "loss": 1.4099,
      "step": 3191
    },
    {
      "epoch": 0.45724108293940696,
      "grad_norm": 1.0685315132141113,
      "learning_rate": 0.00011861232655424633,
      "loss": 1.4614,
      "step": 3192
    },
    {
      "epoch": 0.4573843288927088,
      "grad_norm": 1.0267964601516724,
      "learning_rate": 0.00011856673752361453,
      "loss": 1.5197,
      "step": 3193
    },
    {
      "epoch": 0.4575275748460106,
      "grad_norm": 0.9944993853569031,
      "learning_rate": 0.00011852114449602358,
      "loss": 1.4405,
      "step": 3194
    },
    {
      "epoch": 0.45767082079931243,
      "grad_norm": 1.0819752216339111,
      "learning_rate": 0.0001184755474812886,
      "loss": 1.1958,
      "step": 3195
    },
    {
      "epoch": 0.4578140667526142,
      "grad_norm": 1.08761465549469,
      "learning_rate": 0.00011842994648922536,
      "loss": 1.5107,
      "step": 3196
    },
    {
      "epoch": 0.45795731270591605,
      "grad_norm": 1.0154162645339966,
      "learning_rate": 0.00011838434152965079,
      "loss": 1.4732,
      "step": 3197
    },
    {
      "epoch": 0.4581005586592179,
      "grad_norm": 1.2753942012786865,
      "learning_rate": 0.0001183387326123824,
      "loss": 1.6183,
      "step": 3198
    },
    {
      "epoch": 0.4582438046125197,
      "grad_norm": 0.9973735809326172,
      "learning_rate": 0.00011829311974723867,
      "loss": 1.3681,
      "step": 3199
    },
    {
      "epoch": 0.4583870505658215,
      "grad_norm": 1.1050772666931152,
      "learning_rate": 0.00011824750294403899,
      "loss": 1.5178,
      "step": 3200
    },
    {
      "epoch": 0.45853029651912336,
      "grad_norm": 1.323988676071167,
      "learning_rate": 0.00011820188221260341,
      "loss": 1.3708,
      "step": 3201
    },
    {
      "epoch": 0.45867354247242514,
      "grad_norm": 1.262052059173584,
      "learning_rate": 0.0001181562575627531,
      "loss": 1.1791,
      "step": 3202
    },
    {
      "epoch": 0.458816788425727,
      "grad_norm": 1.0263917446136475,
      "learning_rate": 0.00011811062900430978,
      "loss": 1.388,
      "step": 3203
    },
    {
      "epoch": 0.45896003437902877,
      "grad_norm": 1.0436040163040161,
      "learning_rate": 0.00011806499654709621,
      "loss": 1.5122,
      "step": 3204
    },
    {
      "epoch": 0.4591032803323306,
      "grad_norm": 0.9282182455062866,
      "learning_rate": 0.00011801936020093594,
      "loss": 1.379,
      "step": 3205
    },
    {
      "epoch": 0.45924652628563245,
      "grad_norm": 1.119687557220459,
      "learning_rate": 0.00011797371997565332,
      "loss": 1.5841,
      "step": 3206
    },
    {
      "epoch": 0.45938977223893424,
      "grad_norm": 1.2043997049331665,
      "learning_rate": 0.00011792807588107357,
      "loss": 1.2598,
      "step": 3207
    },
    {
      "epoch": 0.4595330181922361,
      "grad_norm": 1.0688637495040894,
      "learning_rate": 0.00011788242792702275,
      "loss": 1.4565,
      "step": 3208
    },
    {
      "epoch": 0.4596762641455379,
      "grad_norm": 0.9249189496040344,
      "learning_rate": 0.00011783677612332769,
      "loss": 1.4079,
      "step": 3209
    },
    {
      "epoch": 0.4598195100988397,
      "grad_norm": 1.0595513582229614,
      "learning_rate": 0.00011779112047981613,
      "loss": 1.4279,
      "step": 3210
    },
    {
      "epoch": 0.45996275605214154,
      "grad_norm": 0.9271141290664673,
      "learning_rate": 0.00011774546100631662,
      "loss": 1.4195,
      "step": 3211
    },
    {
      "epoch": 0.4601060020054433,
      "grad_norm": 1.0756725072860718,
      "learning_rate": 0.00011769979771265846,
      "loss": 1.4055,
      "step": 3212
    },
    {
      "epoch": 0.46024924795874517,
      "grad_norm": 1.086134433746338,
      "learning_rate": 0.00011765413060867185,
      "loss": 1.4966,
      "step": 3213
    },
    {
      "epoch": 0.460392493912047,
      "grad_norm": 1.2458773851394653,
      "learning_rate": 0.00011760845970418782,
      "loss": 1.3551,
      "step": 3214
    },
    {
      "epoch": 0.4605357398653488,
      "grad_norm": 1.055637001991272,
      "learning_rate": 0.00011756278500903812,
      "loss": 1.4505,
      "step": 3215
    },
    {
      "epoch": 0.46067898581865063,
      "grad_norm": 1.1264960765838623,
      "learning_rate": 0.00011751710653305546,
      "loss": 1.5455,
      "step": 3216
    },
    {
      "epoch": 0.4608222317719524,
      "grad_norm": 0.9137976765632629,
      "learning_rate": 0.00011747142428607318,
      "loss": 1.5716,
      "step": 3217
    },
    {
      "epoch": 0.46096547772525426,
      "grad_norm": 1.4584661722183228,
      "learning_rate": 0.00011742573827792567,
      "loss": 1.438,
      "step": 3218
    },
    {
      "epoch": 0.4611087236785561,
      "grad_norm": 1.1215171813964844,
      "learning_rate": 0.0001173800485184479,
      "loss": 1.2736,
      "step": 3219
    },
    {
      "epoch": 0.4612519696318579,
      "grad_norm": 1.2485909461975098,
      "learning_rate": 0.00011733435501747578,
      "loss": 1.4094,
      "step": 3220
    },
    {
      "epoch": 0.4613952155851597,
      "grad_norm": 1.0997424125671387,
      "learning_rate": 0.00011728865778484597,
      "loss": 1.3956,
      "step": 3221
    },
    {
      "epoch": 0.46153846153846156,
      "grad_norm": 0.9940543174743652,
      "learning_rate": 0.00011724295683039599,
      "loss": 1.5838,
      "step": 3222
    },
    {
      "epoch": 0.46168170749176335,
      "grad_norm": 1.0352367162704468,
      "learning_rate": 0.00011719725216396409,
      "loss": 1.4961,
      "step": 3223
    },
    {
      "epoch": 0.4618249534450652,
      "grad_norm": 1.2422566413879395,
      "learning_rate": 0.00011715154379538935,
      "loss": 1.3742,
      "step": 3224
    },
    {
      "epoch": 0.461968199398367,
      "grad_norm": 1.1647753715515137,
      "learning_rate": 0.0001171058317345117,
      "loss": 1.408,
      "step": 3225
    },
    {
      "epoch": 0.4621114453516688,
      "grad_norm": 1.2552604675292969,
      "learning_rate": 0.00011706011599117173,
      "loss": 1.6675,
      "step": 3226
    },
    {
      "epoch": 0.46225469130497066,
      "grad_norm": 1.030846118927002,
      "learning_rate": 0.000117014396575211,
      "loss": 1.3646,
      "step": 3227
    },
    {
      "epoch": 0.46239793725827244,
      "grad_norm": 1.085591435432434,
      "learning_rate": 0.00011696867349647171,
      "loss": 1.4564,
      "step": 3228
    },
    {
      "epoch": 0.4625411832115743,
      "grad_norm": 1.2605788707733154,
      "learning_rate": 0.00011692294676479696,
      "loss": 1.474,
      "step": 3229
    },
    {
      "epoch": 0.46268442916487607,
      "grad_norm": 0.988599956035614,
      "learning_rate": 0.00011687721639003051,
      "loss": 1.3967,
      "step": 3230
    },
    {
      "epoch": 0.4628276751181779,
      "grad_norm": 1.2275782823562622,
      "learning_rate": 0.00011683148238201704,
      "loss": 1.361,
      "step": 3231
    },
    {
      "epoch": 0.46297092107147975,
      "grad_norm": 1.096309781074524,
      "learning_rate": 0.00011678574475060191,
      "loss": 1.2918,
      "step": 3232
    },
    {
      "epoch": 0.46311416702478153,
      "grad_norm": 1.1264857053756714,
      "learning_rate": 0.00011674000350563133,
      "loss": 1.2893,
      "step": 3233
    },
    {
      "epoch": 0.46325741297808337,
      "grad_norm": 1.0949419736862183,
      "learning_rate": 0.00011669425865695223,
      "loss": 1.4873,
      "step": 3234
    },
    {
      "epoch": 0.4634006589313852,
      "grad_norm": 1.0669584274291992,
      "learning_rate": 0.00011664851021441237,
      "loss": 1.4774,
      "step": 3235
    },
    {
      "epoch": 0.463543904884687,
      "grad_norm": 1.2258151769638062,
      "learning_rate": 0.00011660275818786027,
      "loss": 1.443,
      "step": 3236
    },
    {
      "epoch": 0.46368715083798884,
      "grad_norm": 1.3936161994934082,
      "learning_rate": 0.00011655700258714517,
      "loss": 1.4439,
      "step": 3237
    },
    {
      "epoch": 0.4638303967912906,
      "grad_norm": 1.1493662595748901,
      "learning_rate": 0.00011651124342211712,
      "loss": 1.4592,
      "step": 3238
    },
    {
      "epoch": 0.46397364274459246,
      "grad_norm": 1.0491929054260254,
      "learning_rate": 0.00011646548070262695,
      "loss": 1.4272,
      "step": 3239
    },
    {
      "epoch": 0.4641168886978943,
      "grad_norm": 1.2662327289581299,
      "learning_rate": 0.00011641971443852627,
      "loss": 1.4197,
      "step": 3240
    },
    {
      "epoch": 0.4642601346511961,
      "grad_norm": 1.0402677059173584,
      "learning_rate": 0.00011637394463966737,
      "loss": 1.3199,
      "step": 3241
    },
    {
      "epoch": 0.46440338060449793,
      "grad_norm": 1.0888824462890625,
      "learning_rate": 0.00011632817131590339,
      "loss": 1.5224,
      "step": 3242
    },
    {
      "epoch": 0.46454662655779977,
      "grad_norm": 0.9686537384986877,
      "learning_rate": 0.0001162823944770882,
      "loss": 1.3672,
      "step": 3243
    },
    {
      "epoch": 0.46468987251110155,
      "grad_norm": 1.2487537860870361,
      "learning_rate": 0.00011623661413307639,
      "loss": 1.3844,
      "step": 3244
    },
    {
      "epoch": 0.4648331184644034,
      "grad_norm": 1.1117708683013916,
      "learning_rate": 0.00011619083029372338,
      "loss": 1.2275,
      "step": 3245
    },
    {
      "epoch": 0.4649763644177052,
      "grad_norm": 1.2130790948867798,
      "learning_rate": 0.0001161450429688852,
      "loss": 1.5362,
      "step": 3246
    },
    {
      "epoch": 0.465119610371007,
      "grad_norm": 1.1776996850967407,
      "learning_rate": 0.00011609925216841886,
      "loss": 1.503,
      "step": 3247
    },
    {
      "epoch": 0.46526285632430886,
      "grad_norm": 1.0639768838882446,
      "learning_rate": 0.00011605345790218189,
      "loss": 1.2413,
      "step": 3248
    },
    {
      "epoch": 0.46540610227761064,
      "grad_norm": 1.097795009613037,
      "learning_rate": 0.0001160076601800327,
      "loss": 1.6393,
      "step": 3249
    },
    {
      "epoch": 0.4655493482309125,
      "grad_norm": 0.9664180874824524,
      "learning_rate": 0.00011596185901183043,
      "loss": 1.5207,
      "step": 3250
    },
    {
      "epoch": 0.46569259418421427,
      "grad_norm": 1.0399538278579712,
      "learning_rate": 0.00011591605440743488,
      "loss": 1.6099,
      "step": 3251
    },
    {
      "epoch": 0.4658358401375161,
      "grad_norm": 1.1419068574905396,
      "learning_rate": 0.00011587024637670669,
      "loss": 1.4907,
      "step": 3252
    },
    {
      "epoch": 0.46597908609081795,
      "grad_norm": 0.92630535364151,
      "learning_rate": 0.00011582443492950716,
      "loss": 1.4683,
      "step": 3253
    },
    {
      "epoch": 0.46612233204411974,
      "grad_norm": 1.26869535446167,
      "learning_rate": 0.00011577862007569842,
      "loss": 1.5627,
      "step": 3254
    },
    {
      "epoch": 0.4662655779974216,
      "grad_norm": 1.1297804117202759,
      "learning_rate": 0.00011573280182514321,
      "loss": 1.5184,
      "step": 3255
    },
    {
      "epoch": 0.4664088239507234,
      "grad_norm": 1.5263715982437134,
      "learning_rate": 0.00011568698018770512,
      "loss": 1.4713,
      "step": 3256
    },
    {
      "epoch": 0.4665520699040252,
      "grad_norm": 1.0389893054962158,
      "learning_rate": 0.00011564115517324836,
      "loss": 1.5073,
      "step": 3257
    },
    {
      "epoch": 0.46669531585732704,
      "grad_norm": 1.1668983697891235,
      "learning_rate": 0.00011559532679163796,
      "loss": 1.3207,
      "step": 3258
    },
    {
      "epoch": 0.4668385618106288,
      "grad_norm": 1.0060150623321533,
      "learning_rate": 0.00011554949505273962,
      "loss": 1.2718,
      "step": 3259
    },
    {
      "epoch": 0.46698180776393067,
      "grad_norm": 1.1631263494491577,
      "learning_rate": 0.00011550365996641979,
      "loss": 1.3085,
      "step": 3260
    },
    {
      "epoch": 0.4671250537172325,
      "grad_norm": 1.244558334350586,
      "learning_rate": 0.00011545782154254565,
      "loss": 1.5046,
      "step": 3261
    },
    {
      "epoch": 0.4672682996705343,
      "grad_norm": 1.0249354839324951,
      "learning_rate": 0.00011541197979098501,
      "loss": 1.4909,
      "step": 3262
    },
    {
      "epoch": 0.46741154562383613,
      "grad_norm": 1.0131266117095947,
      "learning_rate": 0.00011536613472160653,
      "loss": 1.2331,
      "step": 3263
    },
    {
      "epoch": 0.4675547915771379,
      "grad_norm": 1.1467456817626953,
      "learning_rate": 0.00011532028634427949,
      "loss": 1.4885,
      "step": 3264
    },
    {
      "epoch": 0.46769803753043976,
      "grad_norm": 0.9395096302032471,
      "learning_rate": 0.00011527443466887393,
      "loss": 1.1577,
      "step": 3265
    },
    {
      "epoch": 0.4678412834837416,
      "grad_norm": 1.2805176973342896,
      "learning_rate": 0.00011522857970526058,
      "loss": 1.4491,
      "step": 3266
    },
    {
      "epoch": 0.4679845294370434,
      "grad_norm": 1.0709673166275024,
      "learning_rate": 0.00011518272146331082,
      "loss": 1.3827,
      "step": 3267
    },
    {
      "epoch": 0.4681277753903452,
      "grad_norm": 1.2241272926330566,
      "learning_rate": 0.00011513685995289689,
      "loss": 1.5112,
      "step": 3268
    },
    {
      "epoch": 0.46827102134364706,
      "grad_norm": 1.0638458728790283,
      "learning_rate": 0.00011509099518389156,
      "loss": 1.4051,
      "step": 3269
    },
    {
      "epoch": 0.46841426729694885,
      "grad_norm": 1.0161278247833252,
      "learning_rate": 0.00011504512716616846,
      "loss": 1.5547,
      "step": 3270
    },
    {
      "epoch": 0.4685575132502507,
      "grad_norm": 0.9485410451889038,
      "learning_rate": 0.00011499925590960172,
      "loss": 1.4866,
      "step": 3271
    },
    {
      "epoch": 0.4687007592035525,
      "grad_norm": 1.0508158206939697,
      "learning_rate": 0.00011495338142406643,
      "loss": 1.3563,
      "step": 3272
    },
    {
      "epoch": 0.4688440051568543,
      "grad_norm": 1.0126726627349854,
      "learning_rate": 0.00011490750371943813,
      "loss": 1.2948,
      "step": 3273
    },
    {
      "epoch": 0.46898725111015616,
      "grad_norm": 1.0145466327667236,
      "learning_rate": 0.00011486162280559316,
      "loss": 1.5088,
      "step": 3274
    },
    {
      "epoch": 0.46913049706345794,
      "grad_norm": 1.1184942722320557,
      "learning_rate": 0.0001148157386924086,
      "loss": 1.2719,
      "step": 3275
    },
    {
      "epoch": 0.4692737430167598,
      "grad_norm": 1.19002366065979,
      "learning_rate": 0.00011476985138976209,
      "loss": 1.4071,
      "step": 3276
    },
    {
      "epoch": 0.4694169889700616,
      "grad_norm": 1.0520473718643188,
      "learning_rate": 0.0001147239609075321,
      "loss": 1.354,
      "step": 3277
    },
    {
      "epoch": 0.4695602349233634,
      "grad_norm": 1.0657501220703125,
      "learning_rate": 0.00011467806725559769,
      "loss": 1.3836,
      "step": 3278
    },
    {
      "epoch": 0.46970348087666525,
      "grad_norm": 1.15292227268219,
      "learning_rate": 0.00011463217044383865,
      "loss": 1.4323,
      "step": 3279
    },
    {
      "epoch": 0.46984672682996703,
      "grad_norm": 1.1347779035568237,
      "learning_rate": 0.00011458627048213535,
      "loss": 1.3662,
      "step": 3280
    },
    {
      "epoch": 0.4699899727832689,
      "grad_norm": 1.0751999616622925,
      "learning_rate": 0.00011454036738036899,
      "loss": 1.338,
      "step": 3281
    },
    {
      "epoch": 0.4701332187365707,
      "grad_norm": 1.1605064868927002,
      "learning_rate": 0.00011449446114842137,
      "loss": 1.3779,
      "step": 3282
    },
    {
      "epoch": 0.4702764646898725,
      "grad_norm": 0.9580484628677368,
      "learning_rate": 0.00011444855179617493,
      "loss": 1.6082,
      "step": 3283
    },
    {
      "epoch": 0.47041971064317434,
      "grad_norm": 1.0737966299057007,
      "learning_rate": 0.00011440263933351283,
      "loss": 1.4823,
      "step": 3284
    },
    {
      "epoch": 0.4705629565964761,
      "grad_norm": 1.0176937580108643,
      "learning_rate": 0.00011435672377031889,
      "loss": 1.4786,
      "step": 3285
    },
    {
      "epoch": 0.47070620254977796,
      "grad_norm": 1.1679856777191162,
      "learning_rate": 0.00011431080511647763,
      "loss": 1.2433,
      "step": 3286
    },
    {
      "epoch": 0.4708494485030798,
      "grad_norm": 1.0988603830337524,
      "learning_rate": 0.00011426488338187414,
      "loss": 1.4521,
      "step": 3287
    },
    {
      "epoch": 0.4709926944563816,
      "grad_norm": 1.0639088153839111,
      "learning_rate": 0.00011421895857639424,
      "loss": 1.4075,
      "step": 3288
    },
    {
      "epoch": 0.47113594040968343,
      "grad_norm": 0.9400182962417603,
      "learning_rate": 0.00011417303070992445,
      "loss": 1.4765,
      "step": 3289
    },
    {
      "epoch": 0.47127918636298527,
      "grad_norm": 1.0353741645812988,
      "learning_rate": 0.00011412709979235187,
      "loss": 1.4377,
      "step": 3290
    },
    {
      "epoch": 0.47142243231628705,
      "grad_norm": 1.0382957458496094,
      "learning_rate": 0.0001140811658335643,
      "loss": 1.4198,
      "step": 3291
    },
    {
      "epoch": 0.4715656782695889,
      "grad_norm": 1.119065761566162,
      "learning_rate": 0.00011403522884345017,
      "loss": 1.5738,
      "step": 3292
    },
    {
      "epoch": 0.4717089242228907,
      "grad_norm": 0.8763836026191711,
      "learning_rate": 0.00011398928883189859,
      "loss": 1.469,
      "step": 3293
    },
    {
      "epoch": 0.4718521701761925,
      "grad_norm": 0.9945670962333679,
      "learning_rate": 0.00011394334580879931,
      "loss": 1.4831,
      "step": 3294
    },
    {
      "epoch": 0.47199541612949436,
      "grad_norm": 1.1996240615844727,
      "learning_rate": 0.00011389739978404273,
      "loss": 1.499,
      "step": 3295
    },
    {
      "epoch": 0.47213866208279615,
      "grad_norm": 1.0718663930892944,
      "learning_rate": 0.00011385145076751986,
      "loss": 1.3571,
      "step": 3296
    },
    {
      "epoch": 0.472281908036098,
      "grad_norm": 1.35163152217865,
      "learning_rate": 0.00011380549876912244,
      "loss": 1.5648,
      "step": 3297
    },
    {
      "epoch": 0.4724251539893998,
      "grad_norm": 1.1435168981552124,
      "learning_rate": 0.00011375954379874274,
      "loss": 1.369,
      "step": 3298
    },
    {
      "epoch": 0.4725683999427016,
      "grad_norm": 0.876053512096405,
      "learning_rate": 0.00011371358586627376,
      "loss": 1.3048,
      "step": 3299
    },
    {
      "epoch": 0.47271164589600345,
      "grad_norm": 0.9831293821334839,
      "learning_rate": 0.00011366762498160914,
      "loss": 1.3607,
      "step": 3300
    },
    {
      "epoch": 0.47285489184930524,
      "grad_norm": 1.059891939163208,
      "learning_rate": 0.00011362166115464304,
      "loss": 1.5376,
      "step": 3301
    },
    {
      "epoch": 0.4729981378026071,
      "grad_norm": 1.1816636323928833,
      "learning_rate": 0.00011357569439527038,
      "loss": 1.4659,
      "step": 3302
    },
    {
      "epoch": 0.4731413837559089,
      "grad_norm": 1.1875196695327759,
      "learning_rate": 0.00011352972471338668,
      "loss": 1.2874,
      "step": 3303
    },
    {
      "epoch": 0.4732846297092107,
      "grad_norm": 1.2097115516662598,
      "learning_rate": 0.00011348375211888807,
      "loss": 1.4972,
      "step": 3304
    },
    {
      "epoch": 0.47342787566251254,
      "grad_norm": 1.065178632736206,
      "learning_rate": 0.00011343777662167126,
      "loss": 1.3953,
      "step": 3305
    },
    {
      "epoch": 0.4735711216158143,
      "grad_norm": 1.0165363550186157,
      "learning_rate": 0.0001133917982316337,
      "loss": 1.6532,
      "step": 3306
    },
    {
      "epoch": 0.47371436756911617,
      "grad_norm": 1.3845362663269043,
      "learning_rate": 0.00011334581695867339,
      "loss": 1.3454,
      "step": 3307
    },
    {
      "epoch": 0.473857613522418,
      "grad_norm": 1.085850715637207,
      "learning_rate": 0.00011329983281268892,
      "loss": 1.5056,
      "step": 3308
    },
    {
      "epoch": 0.4740008594757198,
      "grad_norm": 0.8846843242645264,
      "learning_rate": 0.00011325384580357957,
      "loss": 1.519,
      "step": 3309
    },
    {
      "epoch": 0.47414410542902163,
      "grad_norm": 1.1567100286483765,
      "learning_rate": 0.00011320785594124518,
      "loss": 1.6023,
      "step": 3310
    },
    {
      "epoch": 0.4742873513823235,
      "grad_norm": 1.1920970678329468,
      "learning_rate": 0.00011316186323558631,
      "loss": 1.3781,
      "step": 3311
    },
    {
      "epoch": 0.47443059733562526,
      "grad_norm": 1.0693937540054321,
      "learning_rate": 0.00011311586769650395,
      "loss": 1.6848,
      "step": 3312
    },
    {
      "epoch": 0.4745738432889271,
      "grad_norm": 1.0067671537399292,
      "learning_rate": 0.00011306986933389984,
      "loss": 1.4515,
      "step": 3313
    },
    {
      "epoch": 0.4747170892422289,
      "grad_norm": 1.185122013092041,
      "learning_rate": 0.00011302386815767629,
      "loss": 1.6257,
      "step": 3314
    },
    {
      "epoch": 0.4748603351955307,
      "grad_norm": 1.0342673063278198,
      "learning_rate": 0.00011297786417773626,
      "loss": 1.3211,
      "step": 3315
    },
    {
      "epoch": 0.47500358114883257,
      "grad_norm": 1.136813759803772,
      "learning_rate": 0.00011293185740398317,
      "loss": 1.419,
      "step": 3316
    },
    {
      "epoch": 0.47514682710213435,
      "grad_norm": 0.9932136535644531,
      "learning_rate": 0.00011288584784632124,
      "loss": 1.4535,
      "step": 3317
    },
    {
      "epoch": 0.4752900730554362,
      "grad_norm": 1.0629267692565918,
      "learning_rate": 0.00011283983551465511,
      "loss": 1.4108,
      "step": 3318
    },
    {
      "epoch": 0.475433319008738,
      "grad_norm": 1.08879554271698,
      "learning_rate": 0.00011279382041889013,
      "loss": 1.4213,
      "step": 3319
    },
    {
      "epoch": 0.4755765649620398,
      "grad_norm": 1.1810017824172974,
      "learning_rate": 0.00011274780256893225,
      "loss": 1.4481,
      "step": 3320
    },
    {
      "epoch": 0.47571981091534166,
      "grad_norm": 1.1317284107208252,
      "learning_rate": 0.00011270178197468789,
      "loss": 1.3384,
      "step": 3321
    },
    {
      "epoch": 0.47586305686864344,
      "grad_norm": 1.1286649703979492,
      "learning_rate": 0.00011265575864606421,
      "loss": 1.3778,
      "step": 3322
    },
    {
      "epoch": 0.4760063028219453,
      "grad_norm": 1.1449493169784546,
      "learning_rate": 0.00011260973259296888,
      "loss": 1.2873,
      "step": 3323
    },
    {
      "epoch": 0.4761495487752471,
      "grad_norm": 1.3122366666793823,
      "learning_rate": 0.00011256370382531017,
      "loss": 1.6179,
      "step": 3324
    },
    {
      "epoch": 0.4762927947285489,
      "grad_norm": 0.9566621780395508,
      "learning_rate": 0.00011251767235299688,
      "loss": 1.4554,
      "step": 3325
    },
    {
      "epoch": 0.47643604068185075,
      "grad_norm": 0.8679423928260803,
      "learning_rate": 0.00011247163818593856,
      "loss": 1.3446,
      "step": 3326
    },
    {
      "epoch": 0.47657928663515253,
      "grad_norm": 1.2780513763427734,
      "learning_rate": 0.00011242560133404513,
      "loss": 1.5271,
      "step": 3327
    },
    {
      "epoch": 0.4767225325884544,
      "grad_norm": 1.3613779544830322,
      "learning_rate": 0.00011237956180722722,
      "loss": 1.4222,
      "step": 3328
    },
    {
      "epoch": 0.4768657785417562,
      "grad_norm": 0.9809669852256775,
      "learning_rate": 0.00011233351961539605,
      "loss": 1.4102,
      "step": 3329
    },
    {
      "epoch": 0.477009024495058,
      "grad_norm": 1.221941351890564,
      "learning_rate": 0.00011228747476846322,
      "loss": 1.5474,
      "step": 3330
    },
    {
      "epoch": 0.47715227044835984,
      "grad_norm": 1.1364434957504272,
      "learning_rate": 0.00011224142727634122,
      "loss": 1.5003,
      "step": 3331
    },
    {
      "epoch": 0.4772955164016617,
      "grad_norm": 1.062312126159668,
      "learning_rate": 0.00011219537714894282,
      "loss": 1.4746,
      "step": 3332
    },
    {
      "epoch": 0.47743876235496346,
      "grad_norm": 0.9702563881874084,
      "learning_rate": 0.00011214932439618151,
      "loss": 1.6234,
      "step": 3333
    },
    {
      "epoch": 0.4775820083082653,
      "grad_norm": 1.1632903814315796,
      "learning_rate": 0.00011210326902797131,
      "loss": 1.4115,
      "step": 3334
    },
    {
      "epoch": 0.4777252542615671,
      "grad_norm": 1.0407432317733765,
      "learning_rate": 0.00011205721105422679,
      "loss": 1.3137,
      "step": 3335
    },
    {
      "epoch": 0.47786850021486893,
      "grad_norm": 0.9921265244483948,
      "learning_rate": 0.00011201115048486313,
      "loss": 1.3385,
      "step": 3336
    },
    {
      "epoch": 0.47801174616817077,
      "grad_norm": 1.224592924118042,
      "learning_rate": 0.00011196508732979599,
      "loss": 1.443,
      "step": 3337
    },
    {
      "epoch": 0.47815499212147256,
      "grad_norm": 0.97806316614151,
      "learning_rate": 0.00011191902159894161,
      "loss": 1.505,
      "step": 3338
    },
    {
      "epoch": 0.4782982380747744,
      "grad_norm": 1.1465346813201904,
      "learning_rate": 0.00011187295330221686,
      "loss": 1.5355,
      "step": 3339
    },
    {
      "epoch": 0.4784414840280762,
      "grad_norm": 1.0608819723129272,
      "learning_rate": 0.00011182688244953907,
      "loss": 1.5581,
      "step": 3340
    },
    {
      "epoch": 0.478584729981378,
      "grad_norm": 1.2137082815170288,
      "learning_rate": 0.00011178080905082615,
      "loss": 1.4183,
      "step": 3341
    },
    {
      "epoch": 0.47872797593467986,
      "grad_norm": 0.9721546769142151,
      "learning_rate": 0.00011173473311599656,
      "loss": 1.3798,
      "step": 3342
    },
    {
      "epoch": 0.47887122188798165,
      "grad_norm": 1.270617961883545,
      "learning_rate": 0.00011168865465496932,
      "loss": 1.4943,
      "step": 3343
    },
    {
      "epoch": 0.4790144678412835,
      "grad_norm": 1.1930001974105835,
      "learning_rate": 0.000111642573677664,
      "loss": 1.3323,
      "step": 3344
    },
    {
      "epoch": 0.4791577137945853,
      "grad_norm": 1.035975456237793,
      "learning_rate": 0.00011159649019400069,
      "loss": 1.4723,
      "step": 3345
    },
    {
      "epoch": 0.4793009597478871,
      "grad_norm": 1.2159650325775146,
      "learning_rate": 0.00011155040421389996,
      "loss": 1.3959,
      "step": 3346
    },
    {
      "epoch": 0.47944420570118895,
      "grad_norm": 1.190435767173767,
      "learning_rate": 0.00011150431574728308,
      "loss": 1.4425,
      "step": 3347
    },
    {
      "epoch": 0.47958745165449074,
      "grad_norm": 1.0722824335098267,
      "learning_rate": 0.00011145822480407168,
      "loss": 1.3259,
      "step": 3348
    },
    {
      "epoch": 0.4797306976077926,
      "grad_norm": 1.1439940929412842,
      "learning_rate": 0.00011141213139418805,
      "loss": 1.6281,
      "step": 3349
    },
    {
      "epoch": 0.4798739435610944,
      "grad_norm": 1.2694685459136963,
      "learning_rate": 0.00011136603552755489,
      "loss": 1.4223,
      "step": 3350
    },
    {
      "epoch": 0.4800171895143962,
      "grad_norm": 1.099033236503601,
      "learning_rate": 0.00011131993721409559,
      "loss": 1.5037,
      "step": 3351
    },
    {
      "epoch": 0.48016043546769804,
      "grad_norm": 0.9075151681900024,
      "learning_rate": 0.00011127383646373393,
      "loss": 1.3515,
      "step": 3352
    },
    {
      "epoch": 0.4803036814209999,
      "grad_norm": 0.9200207591056824,
      "learning_rate": 0.00011122773328639424,
      "loss": 1.3497,
      "step": 3353
    },
    {
      "epoch": 0.48044692737430167,
      "grad_norm": 1.4254951477050781,
      "learning_rate": 0.00011118162769200146,
      "loss": 1.4599,
      "step": 3354
    },
    {
      "epoch": 0.4805901733276035,
      "grad_norm": 1.005355954170227,
      "learning_rate": 0.00011113551969048089,
      "loss": 1.3563,
      "step": 3355
    },
    {
      "epoch": 0.4807334192809053,
      "grad_norm": 1.6896955966949463,
      "learning_rate": 0.00011108940929175853,
      "loss": 1.52,
      "step": 3356
    },
    {
      "epoch": 0.48087666523420713,
      "grad_norm": 0.9772291779518127,
      "learning_rate": 0.00011104329650576073,
      "loss": 1.4418,
      "step": 3357
    },
    {
      "epoch": 0.481019911187509,
      "grad_norm": 1.6666598320007324,
      "learning_rate": 0.00011099718134241451,
      "loss": 1.3506,
      "step": 3358
    },
    {
      "epoch": 0.48116315714081076,
      "grad_norm": 0.9738074541091919,
      "learning_rate": 0.00011095106381164727,
      "loss": 1.4856,
      "step": 3359
    },
    {
      "epoch": 0.4813064030941126,
      "grad_norm": 1.0006673336029053,
      "learning_rate": 0.00011090494392338697,
      "loss": 1.539,
      "step": 3360
    },
    {
      "epoch": 0.4814496490474144,
      "grad_norm": 1.3746137619018555,
      "learning_rate": 0.00011085882168756212,
      "loss": 1.4118,
      "step": 3361
    },
    {
      "epoch": 0.4815928950007162,
      "grad_norm": 1.0909067392349243,
      "learning_rate": 0.00011081269711410167,
      "loss": 1.4434,
      "step": 3362
    },
    {
      "epoch": 0.48173614095401807,
      "grad_norm": 1.0215246677398682,
      "learning_rate": 0.0001107665702129351,
      "loss": 1.3228,
      "step": 3363
    },
    {
      "epoch": 0.48187938690731985,
      "grad_norm": 1.223340392112732,
      "learning_rate": 0.00011072044099399242,
      "loss": 1.5585,
      "step": 3364
    },
    {
      "epoch": 0.4820226328606217,
      "grad_norm": 1.0300712585449219,
      "learning_rate": 0.00011067430946720408,
      "loss": 1.4543,
      "step": 3365
    },
    {
      "epoch": 0.48216587881392353,
      "grad_norm": 1.1743842363357544,
      "learning_rate": 0.00011062817564250103,
      "loss": 1.3898,
      "step": 3366
    },
    {
      "epoch": 0.4823091247672253,
      "grad_norm": 1.2403898239135742,
      "learning_rate": 0.00011058203952981476,
      "loss": 1.5558,
      "step": 3367
    },
    {
      "epoch": 0.48245237072052716,
      "grad_norm": 1.311026930809021,
      "learning_rate": 0.00011053590113907728,
      "loss": 1.4761,
      "step": 3368
    },
    {
      "epoch": 0.48259561667382894,
      "grad_norm": 1.1651062965393066,
      "learning_rate": 0.000110489760480221,
      "loss": 1.5753,
      "step": 3369
    },
    {
      "epoch": 0.4827388626271308,
      "grad_norm": 1.0940343141555786,
      "learning_rate": 0.00011044361756317887,
      "loss": 1.4922,
      "step": 3370
    },
    {
      "epoch": 0.4828821085804326,
      "grad_norm": 1.2779818773269653,
      "learning_rate": 0.0001103974723978843,
      "loss": 1.3541,
      "step": 3371
    },
    {
      "epoch": 0.4830253545337344,
      "grad_norm": 1.5476937294006348,
      "learning_rate": 0.00011035132499427123,
      "loss": 1.4233,
      "step": 3372
    },
    {
      "epoch": 0.48316860048703625,
      "grad_norm": 0.9741626381874084,
      "learning_rate": 0.00011030517536227405,
      "loss": 1.2097,
      "step": 3373
    },
    {
      "epoch": 0.48331184644033803,
      "grad_norm": 1.1045200824737549,
      "learning_rate": 0.00011025902351182765,
      "loss": 1.3024,
      "step": 3374
    },
    {
      "epoch": 0.4834550923936399,
      "grad_norm": 1.168770670890808,
      "learning_rate": 0.00011021286945286731,
      "loss": 1.5221,
      "step": 3375
    },
    {
      "epoch": 0.4835983383469417,
      "grad_norm": 1.171493649482727,
      "learning_rate": 0.00011016671319532894,
      "loss": 1.6083,
      "step": 3376
    },
    {
      "epoch": 0.4837415843002435,
      "grad_norm": 1.0298054218292236,
      "learning_rate": 0.0001101205547491488,
      "loss": 1.428,
      "step": 3377
    },
    {
      "epoch": 0.48388483025354534,
      "grad_norm": 1.0101838111877441,
      "learning_rate": 0.00011007439412426365,
      "loss": 1.4277,
      "step": 3378
    },
    {
      "epoch": 0.4840280762068472,
      "grad_norm": 1.11522376537323,
      "learning_rate": 0.00011002823133061079,
      "loss": 1.4073,
      "step": 3379
    },
    {
      "epoch": 0.48417132216014896,
      "grad_norm": 1.1002283096313477,
      "learning_rate": 0.00010998206637812783,
      "loss": 1.4094,
      "step": 3380
    },
    {
      "epoch": 0.4843145681134508,
      "grad_norm": 1.1121855974197388,
      "learning_rate": 0.00010993589927675305,
      "loss": 1.424,
      "step": 3381
    },
    {
      "epoch": 0.4844578140667526,
      "grad_norm": 1.1498883962631226,
      "learning_rate": 0.00010988973003642499,
      "loss": 1.4892,
      "step": 3382
    },
    {
      "epoch": 0.48460106002005443,
      "grad_norm": 1.2971811294555664,
      "learning_rate": 0.00010984355866708282,
      "loss": 1.3987,
      "step": 3383
    },
    {
      "epoch": 0.48474430597335627,
      "grad_norm": 0.994145393371582,
      "learning_rate": 0.000109797385178666,
      "loss": 1.5989,
      "step": 3384
    },
    {
      "epoch": 0.48488755192665806,
      "grad_norm": 1.0696685314178467,
      "learning_rate": 0.00010975120958111467,
      "loss": 1.5071,
      "step": 3385
    },
    {
      "epoch": 0.4850307978799599,
      "grad_norm": 0.9798629879951477,
      "learning_rate": 0.00010970503188436918,
      "loss": 1.7237,
      "step": 3386
    },
    {
      "epoch": 0.48517404383326174,
      "grad_norm": 1.067047357559204,
      "learning_rate": 0.0001096588520983705,
      "loss": 1.5126,
      "step": 3387
    },
    {
      "epoch": 0.4853172897865635,
      "grad_norm": 1.094775915145874,
      "learning_rate": 0.00010961267023305996,
      "loss": 1.3751,
      "step": 3388
    },
    {
      "epoch": 0.48546053573986536,
      "grad_norm": 0.995444118976593,
      "learning_rate": 0.00010956648629837943,
      "loss": 1.3402,
      "step": 3389
    },
    {
      "epoch": 0.48560378169316715,
      "grad_norm": 1.1569347381591797,
      "learning_rate": 0.00010952030030427114,
      "loss": 1.4272,
      "step": 3390
    },
    {
      "epoch": 0.485747027646469,
      "grad_norm": 1.3626402616500854,
      "learning_rate": 0.00010947411226067777,
      "loss": 1.46,
      "step": 3391
    },
    {
      "epoch": 0.48589027359977083,
      "grad_norm": 1.355678677558899,
      "learning_rate": 0.00010942792217754245,
      "loss": 1.3429,
      "step": 3392
    },
    {
      "epoch": 0.4860335195530726,
      "grad_norm": 1.210789680480957,
      "learning_rate": 0.00010938173006480881,
      "loss": 1.6074,
      "step": 3393
    },
    {
      "epoch": 0.48617676550637445,
      "grad_norm": 0.9643422365188599,
      "learning_rate": 0.00010933553593242085,
      "loss": 1.4335,
      "step": 3394
    },
    {
      "epoch": 0.48632001145967624,
      "grad_norm": 1.046739935874939,
      "learning_rate": 0.00010928933979032305,
      "loss": 1.4613,
      "step": 3395
    },
    {
      "epoch": 0.4864632574129781,
      "grad_norm": 1.1703240871429443,
      "learning_rate": 0.00010924314164846021,
      "loss": 1.4556,
      "step": 3396
    },
    {
      "epoch": 0.4866065033662799,
      "grad_norm": 0.9878906011581421,
      "learning_rate": 0.00010919694151677778,
      "loss": 1.4051,
      "step": 3397
    },
    {
      "epoch": 0.4867497493195817,
      "grad_norm": 1.3072961568832397,
      "learning_rate": 0.00010915073940522136,
      "loss": 1.2739,
      "step": 3398
    },
    {
      "epoch": 0.48689299527288354,
      "grad_norm": 1.1261928081512451,
      "learning_rate": 0.00010910453532373726,
      "loss": 1.4672,
      "step": 3399
    },
    {
      "epoch": 0.4870362412261854,
      "grad_norm": 1.211167812347412,
      "learning_rate": 0.00010905832928227193,
      "loss": 1.4634,
      "step": 3400
    },
    {
      "epoch": 0.48717948717948717,
      "grad_norm": 1.0091819763183594,
      "learning_rate": 0.00010901212129077252,
      "loss": 1.51,
      "step": 3401
    },
    {
      "epoch": 0.487322733132789,
      "grad_norm": 1.2708327770233154,
      "learning_rate": 0.00010896591135918638,
      "loss": 1.4039,
      "step": 3402
    },
    {
      "epoch": 0.4874659790860908,
      "grad_norm": 1.1145243644714355,
      "learning_rate": 0.00010891969949746141,
      "loss": 1.3797,
      "step": 3403
    },
    {
      "epoch": 0.48760922503939264,
      "grad_norm": 1.0332392454147339,
      "learning_rate": 0.0001088734857155459,
      "loss": 1.4836,
      "step": 3404
    },
    {
      "epoch": 0.4877524709926945,
      "grad_norm": 1.1336630582809448,
      "learning_rate": 0.00010882727002338842,
      "loss": 1.5205,
      "step": 3405
    },
    {
      "epoch": 0.48789571694599626,
      "grad_norm": 1.1838899850845337,
      "learning_rate": 0.00010878105243093821,
      "loss": 1.4271,
      "step": 3406
    },
    {
      "epoch": 0.4880389628992981,
      "grad_norm": 1.197563648223877,
      "learning_rate": 0.00010873483294814471,
      "loss": 1.4667,
      "step": 3407
    },
    {
      "epoch": 0.48818220885259994,
      "grad_norm": 1.0738539695739746,
      "learning_rate": 0.00010868861158495782,
      "loss": 1.3432,
      "step": 3408
    },
    {
      "epoch": 0.4883254548059017,
      "grad_norm": 1.0344204902648926,
      "learning_rate": 0.00010864238835132783,
      "loss": 1.4408,
      "step": 3409
    },
    {
      "epoch": 0.48846870075920357,
      "grad_norm": 1.133379340171814,
      "learning_rate": 0.00010859616325720554,
      "loss": 1.4679,
      "step": 3410
    },
    {
      "epoch": 0.48861194671250535,
      "grad_norm": 1.1490164995193481,
      "learning_rate": 0.000108549936312542,
      "loss": 1.4928,
      "step": 3411
    },
    {
      "epoch": 0.4887551926658072,
      "grad_norm": 1.095466136932373,
      "learning_rate": 0.00010850370752728874,
      "loss": 1.2896,
      "step": 3412
    },
    {
      "epoch": 0.48889843861910903,
      "grad_norm": 1.038574457168579,
      "learning_rate": 0.0001084574769113977,
      "loss": 1.4865,
      "step": 3413
    },
    {
      "epoch": 0.4890416845724108,
      "grad_norm": 1.2063612937927246,
      "learning_rate": 0.00010841124447482115,
      "loss": 1.439,
      "step": 3414
    },
    {
      "epoch": 0.48918493052571266,
      "grad_norm": 1.1815158128738403,
      "learning_rate": 0.00010836501022751184,
      "loss": 1.2932,
      "step": 3415
    },
    {
      "epoch": 0.48932817647901444,
      "grad_norm": 1.062836766242981,
      "learning_rate": 0.00010831877417942283,
      "loss": 1.4812,
      "step": 3416
    },
    {
      "epoch": 0.4894714224323163,
      "grad_norm": 1.1808329820632935,
      "learning_rate": 0.00010827253634050758,
      "loss": 1.2662,
      "step": 3417
    },
    {
      "epoch": 0.4896146683856181,
      "grad_norm": 1.1382912397384644,
      "learning_rate": 0.00010822629672071995,
      "loss": 1.5181,
      "step": 3418
    },
    {
      "epoch": 0.4897579143389199,
      "grad_norm": 1.048482894897461,
      "learning_rate": 0.00010818005533001425,
      "loss": 1.5647,
      "step": 3419
    },
    {
      "epoch": 0.48990116029222175,
      "grad_norm": 1.0185450315475464,
      "learning_rate": 0.00010813381217834503,
      "loss": 1.3724,
      "step": 3420
    },
    {
      "epoch": 0.4900444062455236,
      "grad_norm": 1.147403359413147,
      "learning_rate": 0.00010808756727566736,
      "loss": 1.413,
      "step": 3421
    },
    {
      "epoch": 0.4901876521988254,
      "grad_norm": 1.2361677885055542,
      "learning_rate": 0.00010804132063193655,
      "loss": 1.4378,
      "step": 3422
    },
    {
      "epoch": 0.4903308981521272,
      "grad_norm": 1.0963706970214844,
      "learning_rate": 0.00010799507225710843,
      "loss": 1.468,
      "step": 3423
    },
    {
      "epoch": 0.490474144105429,
      "grad_norm": 1.054884433746338,
      "learning_rate": 0.00010794882216113911,
      "loss": 1.3274,
      "step": 3424
    },
    {
      "epoch": 0.49061739005873084,
      "grad_norm": 1.1649534702301025,
      "learning_rate": 0.00010790257035398503,
      "loss": 1.5202,
      "step": 3425
    },
    {
      "epoch": 0.4907606360120327,
      "grad_norm": 1.1683424711227417,
      "learning_rate": 0.00010785631684560316,
      "loss": 1.4954,
      "step": 3426
    },
    {
      "epoch": 0.49090388196533447,
      "grad_norm": 0.9950129389762878,
      "learning_rate": 0.00010781006164595067,
      "loss": 1.3923,
      "step": 3427
    },
    {
      "epoch": 0.4910471279186363,
      "grad_norm": 1.3926845788955688,
      "learning_rate": 0.00010776380476498518,
      "loss": 1.3753,
      "step": 3428
    },
    {
      "epoch": 0.4911903738719381,
      "grad_norm": 1.040183186531067,
      "learning_rate": 0.00010771754621266466,
      "loss": 1.4767,
      "step": 3429
    },
    {
      "epoch": 0.49133361982523993,
      "grad_norm": 1.1992692947387695,
      "learning_rate": 0.00010767128599894738,
      "loss": 1.4342,
      "step": 3430
    },
    {
      "epoch": 0.49147686577854177,
      "grad_norm": 1.1832858324050903,
      "learning_rate": 0.00010762502413379209,
      "loss": 1.4657,
      "step": 3431
    },
    {
      "epoch": 0.49162011173184356,
      "grad_norm": 1.2628129720687866,
      "learning_rate": 0.0001075787606271578,
      "loss": 1.4361,
      "step": 3432
    },
    {
      "epoch": 0.4917633576851454,
      "grad_norm": 1.2780241966247559,
      "learning_rate": 0.0001075324954890039,
      "loss": 1.5173,
      "step": 3433
    },
    {
      "epoch": 0.49190660363844724,
      "grad_norm": 1.3416266441345215,
      "learning_rate": 0.00010748622872929009,
      "loss": 1.4028,
      "step": 3434
    },
    {
      "epoch": 0.492049849591749,
      "grad_norm": 1.0095843076705933,
      "learning_rate": 0.0001074399603579765,
      "loss": 1.5194,
      "step": 3435
    },
    {
      "epoch": 0.49219309554505086,
      "grad_norm": 1.0416206121444702,
      "learning_rate": 0.00010739369038502356,
      "loss": 1.5138,
      "step": 3436
    },
    {
      "epoch": 0.49233634149835265,
      "grad_norm": 1.1671234369277954,
      "learning_rate": 0.00010734741882039204,
      "loss": 1.4027,
      "step": 3437
    },
    {
      "epoch": 0.4924795874516545,
      "grad_norm": 1.050166368484497,
      "learning_rate": 0.00010730114567404305,
      "loss": 1.5624,
      "step": 3438
    },
    {
      "epoch": 0.49262283340495633,
      "grad_norm": 1.0378574132919312,
      "learning_rate": 0.00010725487095593811,
      "loss": 1.5012,
      "step": 3439
    },
    {
      "epoch": 0.4927660793582581,
      "grad_norm": 0.9915941953659058,
      "learning_rate": 0.00010720859467603898,
      "loss": 1.4393,
      "step": 3440
    },
    {
      "epoch": 0.49290932531155995,
      "grad_norm": 1.4908393621444702,
      "learning_rate": 0.00010716231684430779,
      "loss": 1.4421,
      "step": 3441
    },
    {
      "epoch": 0.4930525712648618,
      "grad_norm": 1.087257742881775,
      "learning_rate": 0.00010711603747070702,
      "loss": 1.1705,
      "step": 3442
    },
    {
      "epoch": 0.4931958172181636,
      "grad_norm": 1.2150148153305054,
      "learning_rate": 0.00010706975656519946,
      "loss": 1.3682,
      "step": 3443
    },
    {
      "epoch": 0.4933390631714654,
      "grad_norm": 1.1109732389450073,
      "learning_rate": 0.00010702347413774832,
      "loss": 1.4389,
      "step": 3444
    },
    {
      "epoch": 0.4934823091247672,
      "grad_norm": 0.9174085259437561,
      "learning_rate": 0.00010697719019831695,
      "loss": 1.5606,
      "step": 3445
    },
    {
      "epoch": 0.49362555507806904,
      "grad_norm": 1.0272855758666992,
      "learning_rate": 0.0001069309047568692,
      "loss": 1.3576,
      "step": 3446
    },
    {
      "epoch": 0.4937688010313709,
      "grad_norm": 0.8774657845497131,
      "learning_rate": 0.00010688461782336915,
      "loss": 1.6035,
      "step": 3447
    },
    {
      "epoch": 0.49391204698467267,
      "grad_norm": 1.0023385286331177,
      "learning_rate": 0.00010683832940778127,
      "loss": 1.2459,
      "step": 3448
    },
    {
      "epoch": 0.4940552929379745,
      "grad_norm": 0.9568058252334595,
      "learning_rate": 0.00010679203952007031,
      "loss": 1.2008,
      "step": 3449
    },
    {
      "epoch": 0.4941985388912763,
      "grad_norm": 1.1014618873596191,
      "learning_rate": 0.00010674574817020128,
      "loss": 1.5294,
      "step": 3450
    },
    {
      "epoch": 0.49434178484457814,
      "grad_norm": 1.0991486310958862,
      "learning_rate": 0.00010669945536813963,
      "loss": 1.307,
      "step": 3451
    },
    {
      "epoch": 0.49448503079788,
      "grad_norm": 1.1589651107788086,
      "learning_rate": 0.00010665316112385102,
      "loss": 1.4186,
      "step": 3452
    },
    {
      "epoch": 0.49462827675118176,
      "grad_norm": 1.1061111688613892,
      "learning_rate": 0.00010660686544730145,
      "loss": 1.3606,
      "step": 3453
    },
    {
      "epoch": 0.4947715227044836,
      "grad_norm": 1.2987325191497803,
      "learning_rate": 0.00010656056834845727,
      "loss": 1.5352,
      "step": 3454
    },
    {
      "epoch": 0.49491476865778544,
      "grad_norm": 1.111465334892273,
      "learning_rate": 0.00010651426983728503,
      "loss": 1.2665,
      "step": 3455
    },
    {
      "epoch": 0.4950580146110872,
      "grad_norm": 1.312867283821106,
      "learning_rate": 0.00010646796992375172,
      "loss": 1.4199,
      "step": 3456
    },
    {
      "epoch": 0.49520126056438907,
      "grad_norm": 1.2852320671081543,
      "learning_rate": 0.00010642166861782455,
      "loss": 1.4076,
      "step": 3457
    },
    {
      "epoch": 0.49534450651769085,
      "grad_norm": 1.0688996315002441,
      "learning_rate": 0.00010637536592947103,
      "loss": 1.5563,
      "step": 3458
    },
    {
      "epoch": 0.4954877524709927,
      "grad_norm": 1.1064202785491943,
      "learning_rate": 0.00010632906186865899,
      "loss": 1.4824,
      "step": 3459
    },
    {
      "epoch": 0.49563099842429453,
      "grad_norm": 1.1301168203353882,
      "learning_rate": 0.00010628275644535657,
      "loss": 1.3545,
      "step": 3460
    },
    {
      "epoch": 0.4957742443775963,
      "grad_norm": 1.0639129877090454,
      "learning_rate": 0.00010623644966953212,
      "loss": 1.4714,
      "step": 3461
    },
    {
      "epoch": 0.49591749033089816,
      "grad_norm": 0.9906801581382751,
      "learning_rate": 0.00010619014155115441,
      "loss": 1.4358,
      "step": 3462
    },
    {
      "epoch": 0.4960607362842,
      "grad_norm": 1.2321364879608154,
      "learning_rate": 0.00010614383210019241,
      "loss": 1.3419,
      "step": 3463
    },
    {
      "epoch": 0.4962039822375018,
      "grad_norm": 1.1029038429260254,
      "learning_rate": 0.00010609752132661539,
      "loss": 1.3464,
      "step": 3464
    },
    {
      "epoch": 0.4963472281908036,
      "grad_norm": 1.0389331579208374,
      "learning_rate": 0.00010605120924039293,
      "loss": 1.4859,
      "step": 3465
    },
    {
      "epoch": 0.4964904741441054,
      "grad_norm": 1.1372735500335693,
      "learning_rate": 0.00010600489585149484,
      "loss": 1.3899,
      "step": 3466
    },
    {
      "epoch": 0.49663372009740725,
      "grad_norm": 1.1853569746017456,
      "learning_rate": 0.00010595858116989128,
      "loss": 1.3445,
      "step": 3467
    },
    {
      "epoch": 0.4967769660507091,
      "grad_norm": 1.0772411823272705,
      "learning_rate": 0.00010591226520555264,
      "loss": 1.2438,
      "step": 3468
    },
    {
      "epoch": 0.4969202120040109,
      "grad_norm": 1.1271501779556274,
      "learning_rate": 0.00010586594796844965,
      "loss": 1.427,
      "step": 3469
    },
    {
      "epoch": 0.4970634579573127,
      "grad_norm": 1.1908570528030396,
      "learning_rate": 0.00010581962946855317,
      "loss": 1.6222,
      "step": 3470
    },
    {
      "epoch": 0.4972067039106145,
      "grad_norm": 1.278064489364624,
      "learning_rate": 0.0001057733097158345,
      "loss": 1.6879,
      "step": 3471
    },
    {
      "epoch": 0.49734994986391634,
      "grad_norm": 1.2117416858673096,
      "learning_rate": 0.0001057269887202651,
      "loss": 1.5391,
      "step": 3472
    },
    {
      "epoch": 0.4974931958172182,
      "grad_norm": 1.0972899198532104,
      "learning_rate": 0.00010568066649181676,
      "loss": 1.3929,
      "step": 3473
    },
    {
      "epoch": 0.49763644177051997,
      "grad_norm": 1.1661982536315918,
      "learning_rate": 0.00010563434304046151,
      "loss": 1.335,
      "step": 3474
    },
    {
      "epoch": 0.4977796877238218,
      "grad_norm": 1.2792906761169434,
      "learning_rate": 0.0001055880183761716,
      "loss": 1.4514,
      "step": 3475
    },
    {
      "epoch": 0.49792293367712365,
      "grad_norm": 1.221048355102539,
      "learning_rate": 0.00010554169250891967,
      "loss": 1.2078,
      "step": 3476
    },
    {
      "epoch": 0.49806617963042543,
      "grad_norm": 1.480916142463684,
      "learning_rate": 0.00010549536544867845,
      "loss": 1.387,
      "step": 3477
    },
    {
      "epoch": 0.4982094255837273,
      "grad_norm": 1.1148792505264282,
      "learning_rate": 0.00010544903720542105,
      "loss": 1.3677,
      "step": 3478
    },
    {
      "epoch": 0.49835267153702906,
      "grad_norm": 0.9390967488288879,
      "learning_rate": 0.00010540270778912073,
      "loss": 1.386,
      "step": 3479
    },
    {
      "epoch": 0.4984959174903309,
      "grad_norm": 1.0526762008666992,
      "learning_rate": 0.00010535637720975117,
      "loss": 1.4109,
      "step": 3480
    },
    {
      "epoch": 0.49863916344363274,
      "grad_norm": 1.1913859844207764,
      "learning_rate": 0.00010531004547728613,
      "loss": 1.4696,
      "step": 3481
    },
    {
      "epoch": 0.4987824093969345,
      "grad_norm": 0.969770073890686,
      "learning_rate": 0.00010526371260169972,
      "loss": 1.5182,
      "step": 3482
    },
    {
      "epoch": 0.49892565535023636,
      "grad_norm": 1.2126184701919556,
      "learning_rate": 0.00010521737859296623,
      "loss": 1.4371,
      "step": 3483
    },
    {
      "epoch": 0.49906890130353815,
      "grad_norm": 1.1347695589065552,
      "learning_rate": 0.00010517104346106022,
      "loss": 1.3374,
      "step": 3484
    },
    {
      "epoch": 0.49921214725684,
      "grad_norm": 1.031693458557129,
      "learning_rate": 0.00010512470721595655,
      "loss": 1.1568,
      "step": 3485
    },
    {
      "epoch": 0.49935539321014183,
      "grad_norm": 0.972343921661377,
      "learning_rate": 0.00010507836986763022,
      "loss": 1.5731,
      "step": 3486
    },
    {
      "epoch": 0.4994986391634436,
      "grad_norm": 1.110434651374817,
      "learning_rate": 0.0001050320314260565,
      "loss": 1.3145,
      "step": 3487
    },
    {
      "epoch": 0.49964188511674545,
      "grad_norm": 1.256438136100769,
      "learning_rate": 0.00010498569190121097,
      "loss": 1.4113,
      "step": 3488
    },
    {
      "epoch": 0.4997851310700473,
      "grad_norm": 1.2458997964859009,
      "learning_rate": 0.00010493935130306934,
      "loss": 1.2825,
      "step": 3489
    },
    {
      "epoch": 0.4999283770233491,
      "grad_norm": 1.125307321548462,
      "learning_rate": 0.00010489300964160762,
      "loss": 1.3764,
      "step": 3490
    },
    {
      "epoch": 0.5000716229766509,
      "grad_norm": 1.0271503925323486,
      "learning_rate": 0.00010484666692680201,
      "loss": 1.3052,
      "step": 3491
    },
    {
      "epoch": 0.5002148689299527,
      "grad_norm": 1.138046383857727,
      "learning_rate": 0.00010480032316862891,
      "loss": 1.3657,
      "step": 3492
    },
    {
      "epoch": 0.5003581148832545,
      "grad_norm": 1.1509279012680054,
      "learning_rate": 0.00010475397837706504,
      "loss": 1.3596,
      "step": 3493
    },
    {
      "epoch": 0.5005013608365564,
      "grad_norm": 0.9271268844604492,
      "learning_rate": 0.00010470763256208729,
      "loss": 1.4179,
      "step": 3494
    },
    {
      "epoch": 0.5006446067898582,
      "grad_norm": 0.9962475299835205,
      "learning_rate": 0.00010466128573367273,
      "loss": 1.5723,
      "step": 3495
    },
    {
      "epoch": 0.50078785274316,
      "grad_norm": 1.202447533607483,
      "learning_rate": 0.00010461493790179868,
      "loss": 1.4853,
      "step": 3496
    },
    {
      "epoch": 0.5009310986964618,
      "grad_norm": 0.9652479887008667,
      "learning_rate": 0.00010456858907644271,
      "loss": 1.3837,
      "step": 3497
    },
    {
      "epoch": 0.5010743446497636,
      "grad_norm": 0.9198633432388306,
      "learning_rate": 0.00010452223926758258,
      "loss": 1.5054,
      "step": 3498
    },
    {
      "epoch": 0.5012175906030655,
      "grad_norm": 1.013107180595398,
      "learning_rate": 0.00010447588848519625,
      "loss": 1.5808,
      "step": 3499
    },
    {
      "epoch": 0.5013608365563673,
      "grad_norm": 1.0811415910720825,
      "learning_rate": 0.00010442953673926185,
      "loss": 1.4595,
      "step": 3500
    },
    {
      "epoch": 0.501504082509669,
      "grad_norm": 1.044018030166626,
      "learning_rate": 0.00010438318403975786,
      "loss": 1.653,
      "step": 3501
    },
    {
      "epoch": 0.5016473284629709,
      "grad_norm": 1.00082266330719,
      "learning_rate": 0.00010433683039666278,
      "loss": 1.311,
      "step": 3502
    },
    {
      "epoch": 0.5017905744162727,
      "grad_norm": 1.2591431140899658,
      "learning_rate": 0.00010429047581995546,
      "loss": 1.4552,
      "step": 3503
    },
    {
      "epoch": 0.5019338203695746,
      "grad_norm": 1.1139363050460815,
      "learning_rate": 0.00010424412031961484,
      "loss": 1.4517,
      "step": 3504
    },
    {
      "epoch": 0.5020770663228764,
      "grad_norm": 1.0364805459976196,
      "learning_rate": 0.00010419776390562015,
      "loss": 1.4931,
      "step": 3505
    },
    {
      "epoch": 0.5022203122761782,
      "grad_norm": 1.1609759330749512,
      "learning_rate": 0.00010415140658795077,
      "loss": 1.3019,
      "step": 3506
    },
    {
      "epoch": 0.50236355822948,
      "grad_norm": 1.2251722812652588,
      "learning_rate": 0.00010410504837658627,
      "loss": 1.435,
      "step": 3507
    },
    {
      "epoch": 0.5025068041827818,
      "grad_norm": 1.0920590162277222,
      "learning_rate": 0.00010405868928150648,
      "loss": 1.3149,
      "step": 3508
    },
    {
      "epoch": 0.5026500501360837,
      "grad_norm": 1.1456358432769775,
      "learning_rate": 0.00010401232931269127,
      "loss": 1.5039,
      "step": 3509
    },
    {
      "epoch": 0.5027932960893855,
      "grad_norm": 0.9957374334335327,
      "learning_rate": 0.0001039659684801209,
      "loss": 1.4093,
      "step": 3510
    },
    {
      "epoch": 0.5029365420426873,
      "grad_norm": 1.0901576280593872,
      "learning_rate": 0.00010391960679377563,
      "loss": 1.2113,
      "step": 3511
    },
    {
      "epoch": 0.5030797879959891,
      "grad_norm": 1.248495101928711,
      "learning_rate": 0.00010387324426363605,
      "loss": 1.4261,
      "step": 3512
    },
    {
      "epoch": 0.5032230339492909,
      "grad_norm": 1.0495734214782715,
      "learning_rate": 0.00010382688089968275,
      "loss": 1.4713,
      "step": 3513
    },
    {
      "epoch": 0.5033662799025928,
      "grad_norm": 1.0612696409225464,
      "learning_rate": 0.00010378051671189677,
      "loss": 1.3575,
      "step": 3514
    },
    {
      "epoch": 0.5035095258558946,
      "grad_norm": 1.1795153617858887,
      "learning_rate": 0.00010373415171025904,
      "loss": 1.5452,
      "step": 3515
    },
    {
      "epoch": 0.5036527718091964,
      "grad_norm": 0.9429616928100586,
      "learning_rate": 0.00010368778590475088,
      "loss": 1.6492,
      "step": 3516
    },
    {
      "epoch": 0.5037960177624982,
      "grad_norm": 1.1151853799819946,
      "learning_rate": 0.00010364141930535367,
      "loss": 1.3705,
      "step": 3517
    },
    {
      "epoch": 0.5039392637158,
      "grad_norm": 1.1622681617736816,
      "learning_rate": 0.00010359505192204899,
      "loss": 1.3159,
      "step": 3518
    },
    {
      "epoch": 0.5040825096691018,
      "grad_norm": 1.12083101272583,
      "learning_rate": 0.00010354868376481862,
      "loss": 1.426,
      "step": 3519
    },
    {
      "epoch": 0.5042257556224037,
      "grad_norm": 1.304024338722229,
      "learning_rate": 0.00010350231484364443,
      "loss": 1.4219,
      "step": 3520
    },
    {
      "epoch": 0.5043690015757055,
      "grad_norm": 1.0422699451446533,
      "learning_rate": 0.00010345594516850851,
      "loss": 1.4692,
      "step": 3521
    },
    {
      "epoch": 0.5045122475290073,
      "grad_norm": 0.9824684858322144,
      "learning_rate": 0.00010340957474939312,
      "loss": 1.4901,
      "step": 3522
    },
    {
      "epoch": 0.5046554934823091,
      "grad_norm": 1.0426971912384033,
      "learning_rate": 0.00010336320359628067,
      "loss": 1.1901,
      "step": 3523
    },
    {
      "epoch": 0.5047987394356109,
      "grad_norm": 1.111275315284729,
      "learning_rate": 0.00010331683171915374,
      "loss": 1.4754,
      "step": 3524
    },
    {
      "epoch": 0.5049419853889128,
      "grad_norm": 1.085099458694458,
      "learning_rate": 0.00010327045912799496,
      "loss": 1.24,
      "step": 3525
    },
    {
      "epoch": 0.5050852313422146,
      "grad_norm": 1.129228115081787,
      "learning_rate": 0.00010322408583278732,
      "loss": 1.3735,
      "step": 3526
    },
    {
      "epoch": 0.5052284772955165,
      "grad_norm": 1.2121059894561768,
      "learning_rate": 0.00010317771184351375,
      "loss": 1.4364,
      "step": 3527
    },
    {
      "epoch": 0.5053717232488182,
      "grad_norm": 1.0406516790390015,
      "learning_rate": 0.00010313133717015749,
      "loss": 1.4341,
      "step": 3528
    },
    {
      "epoch": 0.50551496920212,
      "grad_norm": 1.0965200662612915,
      "learning_rate": 0.00010308496182270176,
      "loss": 1.4157,
      "step": 3529
    },
    {
      "epoch": 0.5056582151554219,
      "grad_norm": 1.0915249586105347,
      "learning_rate": 0.00010303858581113015,
      "loss": 1.3564,
      "step": 3530
    },
    {
      "epoch": 0.5058014611087237,
      "grad_norm": 1.229333519935608,
      "learning_rate": 0.00010299220914542618,
      "loss": 1.2827,
      "step": 3531
    },
    {
      "epoch": 0.5059447070620255,
      "grad_norm": 1.2443373203277588,
      "learning_rate": 0.00010294583183557362,
      "loss": 1.3208,
      "step": 3532
    },
    {
      "epoch": 0.5060879530153273,
      "grad_norm": 1.2086431980133057,
      "learning_rate": 0.00010289945389155643,
      "loss": 1.4478,
      "step": 3533
    },
    {
      "epoch": 0.5062311989686291,
      "grad_norm": 1.1349196434020996,
      "learning_rate": 0.00010285307532335846,
      "loss": 1.4209,
      "step": 3534
    },
    {
      "epoch": 0.506374444921931,
      "grad_norm": 1.244154930114746,
      "learning_rate": 0.00010280669614096405,
      "loss": 1.6094,
      "step": 3535
    },
    {
      "epoch": 0.5065176908752328,
      "grad_norm": 1.1485425233840942,
      "learning_rate": 0.00010276031635435741,
      "loss": 1.371,
      "step": 3536
    },
    {
      "epoch": 0.5066609368285346,
      "grad_norm": 1.0762343406677246,
      "learning_rate": 0.00010271393597352297,
      "loss": 1.5967,
      "step": 3537
    },
    {
      "epoch": 0.5068041827818364,
      "grad_norm": 1.2191520929336548,
      "learning_rate": 0.00010266755500844523,
      "loss": 1.4247,
      "step": 3538
    },
    {
      "epoch": 0.5069474287351382,
      "grad_norm": 1.1406378746032715,
      "learning_rate": 0.00010262117346910896,
      "loss": 1.43,
      "step": 3539
    },
    {
      "epoch": 0.50709067468844,
      "grad_norm": 1.1244142055511475,
      "learning_rate": 0.00010257479136549889,
      "loss": 1.3693,
      "step": 3540
    },
    {
      "epoch": 0.5072339206417419,
      "grad_norm": 1.1609869003295898,
      "learning_rate": 0.00010252840870759993,
      "loss": 1.4445,
      "step": 3541
    },
    {
      "epoch": 0.5073771665950437,
      "grad_norm": 1.0453346967697144,
      "learning_rate": 0.00010248202550539716,
      "loss": 1.5308,
      "step": 3542
    },
    {
      "epoch": 0.5075204125483455,
      "grad_norm": 1.098109245300293,
      "learning_rate": 0.0001024356417688757,
      "loss": 1.3971,
      "step": 3543
    },
    {
      "epoch": 0.5076636585016473,
      "grad_norm": 0.9849418997764587,
      "learning_rate": 0.00010238925750802089,
      "loss": 1.347,
      "step": 3544
    },
    {
      "epoch": 0.5078069044549491,
      "grad_norm": 1.385166049003601,
      "learning_rate": 0.00010234287273281802,
      "loss": 1.3742,
      "step": 3545
    },
    {
      "epoch": 0.507950150408251,
      "grad_norm": 1.1497560739517212,
      "learning_rate": 0.00010229648745325265,
      "loss": 1.285,
      "step": 3546
    },
    {
      "epoch": 0.5080933963615528,
      "grad_norm": 1.158395767211914,
      "learning_rate": 0.00010225010167931035,
      "loss": 1.5339,
      "step": 3547
    },
    {
      "epoch": 0.5082366423148547,
      "grad_norm": 0.9182134866714478,
      "learning_rate": 0.00010220371542097682,
      "loss": 1.3372,
      "step": 3548
    },
    {
      "epoch": 0.5083798882681564,
      "grad_norm": 1.1867213249206543,
      "learning_rate": 0.00010215732868823795,
      "loss": 1.444,
      "step": 3549
    },
    {
      "epoch": 0.5085231342214582,
      "grad_norm": 1.001237154006958,
      "learning_rate": 0.00010211094149107954,
      "loss": 1.4755,
      "step": 3550
    },
    {
      "epoch": 0.5086663801747601,
      "grad_norm": 1.096407413482666,
      "learning_rate": 0.0001020645538394877,
      "loss": 1.4568,
      "step": 3551
    },
    {
      "epoch": 0.5088096261280619,
      "grad_norm": 0.9367827773094177,
      "learning_rate": 0.00010201816574344849,
      "loss": 1.4318,
      "step": 3552
    },
    {
      "epoch": 0.5089528720813637,
      "grad_norm": 1.0399775505065918,
      "learning_rate": 0.00010197177721294818,
      "loss": 1.4125,
      "step": 3553
    },
    {
      "epoch": 0.5090961180346655,
      "grad_norm": 1.0926892757415771,
      "learning_rate": 0.00010192538825797296,
      "loss": 1.3252,
      "step": 3554
    },
    {
      "epoch": 0.5092393639879673,
      "grad_norm": 1.0777487754821777,
      "learning_rate": 0.00010187899888850933,
      "loss": 1.4171,
      "step": 3555
    },
    {
      "epoch": 0.5093826099412692,
      "grad_norm": 1.1289820671081543,
      "learning_rate": 0.00010183260911454373,
      "loss": 1.3153,
      "step": 3556
    },
    {
      "epoch": 0.509525855894571,
      "grad_norm": 1.0153565406799316,
      "learning_rate": 0.00010178621894606275,
      "loss": 1.5408,
      "step": 3557
    },
    {
      "epoch": 0.5096691018478728,
      "grad_norm": 1.171958565711975,
      "learning_rate": 0.00010173982839305304,
      "loss": 1.3925,
      "step": 3558
    },
    {
      "epoch": 0.5098123478011746,
      "grad_norm": 1.178712248802185,
      "learning_rate": 0.0001016934374655013,
      "loss": 1.4596,
      "step": 3559
    },
    {
      "epoch": 0.5099555937544764,
      "grad_norm": 0.9625037312507629,
      "learning_rate": 0.00010164704617339442,
      "loss": 1.5827,
      "step": 3560
    },
    {
      "epoch": 0.5100988397077783,
      "grad_norm": 1.018351435661316,
      "learning_rate": 0.00010160065452671923,
      "loss": 1.4315,
      "step": 3561
    },
    {
      "epoch": 0.5102420856610801,
      "grad_norm": 1.138958215713501,
      "learning_rate": 0.00010155426253546274,
      "loss": 1.4114,
      "step": 3562
    },
    {
      "epoch": 0.5103853316143819,
      "grad_norm": 1.1471924781799316,
      "learning_rate": 0.00010150787020961197,
      "loss": 1.4659,
      "step": 3563
    },
    {
      "epoch": 0.5105285775676837,
      "grad_norm": 1.1197329759597778,
      "learning_rate": 0.00010146147755915407,
      "loss": 1.5327,
      "step": 3564
    },
    {
      "epoch": 0.5106718235209855,
      "grad_norm": 1.0745993852615356,
      "learning_rate": 0.00010141508459407623,
      "loss": 1.5338,
      "step": 3565
    },
    {
      "epoch": 0.5108150694742873,
      "grad_norm": 1.2184962034225464,
      "learning_rate": 0.00010136869132436568,
      "loss": 1.2889,
      "step": 3566
    },
    {
      "epoch": 0.5109583154275892,
      "grad_norm": 0.8949021100997925,
      "learning_rate": 0.00010132229776000974,
      "loss": 1.428,
      "step": 3567
    },
    {
      "epoch": 0.511101561380891,
      "grad_norm": 1.0946015119552612,
      "learning_rate": 0.00010127590391099584,
      "loss": 1.4546,
      "step": 3568
    },
    {
      "epoch": 0.5112448073341929,
      "grad_norm": 0.9700352549552917,
      "learning_rate": 0.00010122950978731141,
      "loss": 1.4379,
      "step": 3569
    },
    {
      "epoch": 0.5113880532874946,
      "grad_norm": 1.1097909212112427,
      "learning_rate": 0.00010118311539894394,
      "loss": 1.5445,
      "step": 3570
    },
    {
      "epoch": 0.5115312992407964,
      "grad_norm": 1.1244127750396729,
      "learning_rate": 0.00010113672075588099,
      "loss": 1.4041,
      "step": 3571
    },
    {
      "epoch": 0.5116745451940983,
      "grad_norm": 1.1424570083618164,
      "learning_rate": 0.0001010903258681102,
      "loss": 1.3175,
      "step": 3572
    },
    {
      "epoch": 0.5118177911474001,
      "grad_norm": 1.2052518129348755,
      "learning_rate": 0.00010104393074561924,
      "loss": 1.4077,
      "step": 3573
    },
    {
      "epoch": 0.511961037100702,
      "grad_norm": 1.0825068950653076,
      "learning_rate": 0.0001009975353983958,
      "loss": 1.4789,
      "step": 3574
    },
    {
      "epoch": 0.5121042830540037,
      "grad_norm": 0.9709814190864563,
      "learning_rate": 0.0001009511398364277,
      "loss": 1.3619,
      "step": 3575
    },
    {
      "epoch": 0.5122475290073055,
      "grad_norm": 1.0578655004501343,
      "learning_rate": 0.00010090474406970271,
      "loss": 1.3785,
      "step": 3576
    },
    {
      "epoch": 0.5123907749606074,
      "grad_norm": 0.9182901382446289,
      "learning_rate": 0.00010085834810820871,
      "loss": 1.5544,
      "step": 3577
    },
    {
      "epoch": 0.5125340209139092,
      "grad_norm": 1.026823878288269,
      "learning_rate": 0.00010081195196193362,
      "loss": 1.6281,
      "step": 3578
    },
    {
      "epoch": 0.512677266867211,
      "grad_norm": 1.0636945962905884,
      "learning_rate": 0.00010076555564086534,
      "loss": 1.6485,
      "step": 3579
    },
    {
      "epoch": 0.5128205128205128,
      "grad_norm": 0.8647247552871704,
      "learning_rate": 0.0001007191591549919,
      "loss": 1.4047,
      "step": 3580
    },
    {
      "epoch": 0.5129637587738146,
      "grad_norm": 1.25603449344635,
      "learning_rate": 0.0001006727625143013,
      "loss": 1.3918,
      "step": 3581
    },
    {
      "epoch": 0.5131070047271165,
      "grad_norm": 1.4147133827209473,
      "learning_rate": 0.00010062636572878155,
      "loss": 1.4178,
      "step": 3582
    },
    {
      "epoch": 0.5132502506804183,
      "grad_norm": 0.952659010887146,
      "learning_rate": 0.00010057996880842078,
      "loss": 1.4683,
      "step": 3583
    },
    {
      "epoch": 0.5133934966337201,
      "grad_norm": 1.19082772731781,
      "learning_rate": 0.00010053357176320703,
      "loss": 1.34,
      "step": 3584
    },
    {
      "epoch": 0.5135367425870219,
      "grad_norm": 1.41098153591156,
      "learning_rate": 0.00010048717460312855,
      "loss": 1.3464,
      "step": 3585
    },
    {
      "epoch": 0.5136799885403237,
      "grad_norm": 0.9611150622367859,
      "learning_rate": 0.00010044077733817341,
      "loss": 1.5962,
      "step": 3586
    },
    {
      "epoch": 0.5138232344936255,
      "grad_norm": 1.1712177991867065,
      "learning_rate": 0.00010039437997832984,
      "loss": 1.3965,
      "step": 3587
    },
    {
      "epoch": 0.5139664804469274,
      "grad_norm": 1.274287223815918,
      "learning_rate": 0.00010034798253358595,
      "loss": 1.4191,
      "step": 3588
    },
    {
      "epoch": 0.5141097264002292,
      "grad_norm": 1.1152019500732422,
      "learning_rate": 0.00010030158501393008,
      "loss": 1.5161,
      "step": 3589
    },
    {
      "epoch": 0.5142529723535311,
      "grad_norm": 1.252150535583496,
      "learning_rate": 0.00010025518742935041,
      "loss": 1.2252,
      "step": 3590
    },
    {
      "epoch": 0.5143962183068328,
      "grad_norm": 0.9764754176139832,
      "learning_rate": 0.00010020878978983522,
      "loss": 1.5607,
      "step": 3591
    },
    {
      "epoch": 0.5145394642601346,
      "grad_norm": 1.1876507997512817,
      "learning_rate": 0.00010016239210537273,
      "loss": 1.5723,
      "step": 3592
    },
    {
      "epoch": 0.5146827102134365,
      "grad_norm": 1.0151089429855347,
      "learning_rate": 0.00010011599438595123,
      "loss": 1.5332,
      "step": 3593
    },
    {
      "epoch": 0.5148259561667383,
      "grad_norm": 0.9354745149612427,
      "learning_rate": 0.00010006959664155904,
      "loss": 1.4294,
      "step": 3594
    },
    {
      "epoch": 0.5149692021200402,
      "grad_norm": 0.9792630672454834,
      "learning_rate": 0.0001000231988821844,
      "loss": 1.617,
      "step": 3595
    },
    {
      "epoch": 0.5151124480733419,
      "grad_norm": 1.1594178676605225,
      "learning_rate": 9.997680111781562e-05,
      "loss": 1.5172,
      "step": 3596
    },
    {
      "epoch": 0.5152556940266437,
      "grad_norm": 1.0626468658447266,
      "learning_rate": 9.9930403358441e-05,
      "loss": 1.3427,
      "step": 3597
    },
    {
      "epoch": 0.5153989399799456,
      "grad_norm": 1.099827527999878,
      "learning_rate": 9.98840056140488e-05,
      "loss": 1.2924,
      "step": 3598
    },
    {
      "epoch": 0.5155421859332474,
      "grad_norm": 1.1610121726989746,
      "learning_rate": 9.983760789462728e-05,
      "loss": 1.4958,
      "step": 3599
    },
    {
      "epoch": 0.5156854318865493,
      "grad_norm": 0.9326909184455872,
      "learning_rate": 9.979121021016482e-05,
      "loss": 1.2922,
      "step": 3600
    },
    {
      "epoch": 0.515828677839851,
      "grad_norm": 1.0037531852722168,
      "learning_rate": 9.97448125706496e-05,
      "loss": 1.5875,
      "step": 3601
    },
    {
      "epoch": 0.5159719237931528,
      "grad_norm": 1.0887730121612549,
      "learning_rate": 9.969841498606993e-05,
      "loss": 1.5401,
      "step": 3602
    },
    {
      "epoch": 0.5161151697464547,
      "grad_norm": 0.9652317762374878,
      "learning_rate": 9.965201746641407e-05,
      "loss": 1.386,
      "step": 3603
    },
    {
      "epoch": 0.5162584156997565,
      "grad_norm": 1.1576441526412964,
      "learning_rate": 9.96056200216702e-05,
      "loss": 1.4251,
      "step": 3604
    },
    {
      "epoch": 0.5164016616530583,
      "grad_norm": 0.9793096780776978,
      "learning_rate": 9.955922266182664e-05,
      "loss": 1.473,
      "step": 3605
    },
    {
      "epoch": 0.5165449076063601,
      "grad_norm": 1.2000478506088257,
      "learning_rate": 9.951282539687146e-05,
      "loss": 1.2743,
      "step": 3606
    },
    {
      "epoch": 0.5166881535596619,
      "grad_norm": 1.299761176109314,
      "learning_rate": 9.946642823679295e-05,
      "loss": 1.3419,
      "step": 3607
    },
    {
      "epoch": 0.5168313995129638,
      "grad_norm": 1.00190269947052,
      "learning_rate": 9.942003119157926e-05,
      "loss": 1.3603,
      "step": 3608
    },
    {
      "epoch": 0.5169746454662656,
      "grad_norm": 1.1703273057937622,
      "learning_rate": 9.937363427121847e-05,
      "loss": 1.5272,
      "step": 3609
    },
    {
      "epoch": 0.5171178914195674,
      "grad_norm": 0.9897456169128418,
      "learning_rate": 9.932723748569876e-05,
      "loss": 1.3854,
      "step": 3610
    },
    {
      "epoch": 0.5172611373728692,
      "grad_norm": 1.143493890762329,
      "learning_rate": 9.928084084500812e-05,
      "loss": 1.4402,
      "step": 3611
    },
    {
      "epoch": 0.517404383326171,
      "grad_norm": 0.9592913389205933,
      "learning_rate": 9.923444435913466e-05,
      "loss": 1.2905,
      "step": 3612
    },
    {
      "epoch": 0.5175476292794728,
      "grad_norm": 1.0832074880599976,
      "learning_rate": 9.918804803806642e-05,
      "loss": 1.3416,
      "step": 3613
    },
    {
      "epoch": 0.5176908752327747,
      "grad_norm": 1.1010894775390625,
      "learning_rate": 9.914165189179131e-05,
      "loss": 1.2985,
      "step": 3614
    },
    {
      "epoch": 0.5178341211860765,
      "grad_norm": 1.1934311389923096,
      "learning_rate": 9.90952559302973e-05,
      "loss": 1.6325,
      "step": 3615
    },
    {
      "epoch": 0.5179773671393784,
      "grad_norm": 1.0829458236694336,
      "learning_rate": 9.904886016357233e-05,
      "loss": 1.5473,
      "step": 3616
    },
    {
      "epoch": 0.5181206130926801,
      "grad_norm": 1.0482755899429321,
      "learning_rate": 9.900246460160422e-05,
      "loss": 1.5872,
      "step": 3617
    },
    {
      "epoch": 0.5182638590459819,
      "grad_norm": 1.1596132516860962,
      "learning_rate": 9.89560692543808e-05,
      "loss": 1.4246,
      "step": 3618
    },
    {
      "epoch": 0.5184071049992838,
      "grad_norm": 1.3710565567016602,
      "learning_rate": 9.890967413188983e-05,
      "loss": 1.4766,
      "step": 3619
    },
    {
      "epoch": 0.5185503509525856,
      "grad_norm": 0.9798039793968201,
      "learning_rate": 9.886327924411902e-05,
      "loss": 1.4038,
      "step": 3620
    },
    {
      "epoch": 0.5186935969058875,
      "grad_norm": 1.2228951454162598,
      "learning_rate": 9.88168846010561e-05,
      "loss": 1.6637,
      "step": 3621
    },
    {
      "epoch": 0.5188368428591892,
      "grad_norm": 1.1269609928131104,
      "learning_rate": 9.87704902126886e-05,
      "loss": 1.3064,
      "step": 3622
    },
    {
      "epoch": 0.518980088812491,
      "grad_norm": 1.1092020273208618,
      "learning_rate": 9.872409608900416e-05,
      "loss": 1.3595,
      "step": 3623
    },
    {
      "epoch": 0.5191233347657929,
      "grad_norm": 1.135745644569397,
      "learning_rate": 9.867770223999028e-05,
      "loss": 1.2588,
      "step": 3624
    },
    {
      "epoch": 0.5192665807190947,
      "grad_norm": 0.9903700947761536,
      "learning_rate": 9.863130867563435e-05,
      "loss": 1.5104,
      "step": 3625
    },
    {
      "epoch": 0.5194098266723965,
      "grad_norm": 1.039293885231018,
      "learning_rate": 9.858491540592382e-05,
      "loss": 1.5577,
      "step": 3626
    },
    {
      "epoch": 0.5195530726256983,
      "grad_norm": 0.9378083944320679,
      "learning_rate": 9.853852244084594e-05,
      "loss": 1.4388,
      "step": 3627
    },
    {
      "epoch": 0.5196963185790001,
      "grad_norm": 1.2949731349945068,
      "learning_rate": 9.849212979038804e-05,
      "loss": 1.4602,
      "step": 3628
    },
    {
      "epoch": 0.519839564532302,
      "grad_norm": 1.0818771123886108,
      "learning_rate": 9.84457374645373e-05,
      "loss": 1.3401,
      "step": 3629
    },
    {
      "epoch": 0.5199828104856038,
      "grad_norm": 1.213658094406128,
      "learning_rate": 9.83993454732808e-05,
      "loss": 1.4178,
      "step": 3630
    },
    {
      "epoch": 0.5201260564389056,
      "grad_norm": 0.924899697303772,
      "learning_rate": 9.835295382660559e-05,
      "loss": 1.397,
      "step": 3631
    },
    {
      "epoch": 0.5202693023922074,
      "grad_norm": 1.1707476377487183,
      "learning_rate": 9.830656253449872e-05,
      "loss": 1.4536,
      "step": 3632
    },
    {
      "epoch": 0.5204125483455092,
      "grad_norm": 1.1964712142944336,
      "learning_rate": 9.826017160694697e-05,
      "loss": 1.3359,
      "step": 3633
    },
    {
      "epoch": 0.520555794298811,
      "grad_norm": 1.0140048265457153,
      "learning_rate": 9.821378105393727e-05,
      "loss": 1.4623,
      "step": 3634
    },
    {
      "epoch": 0.5206990402521129,
      "grad_norm": 0.9919286966323853,
      "learning_rate": 9.816739088545628e-05,
      "loss": 1.4354,
      "step": 3635
    },
    {
      "epoch": 0.5208422862054147,
      "grad_norm": 1.0083461999893188,
      "learning_rate": 9.812100111149068e-05,
      "loss": 1.5466,
      "step": 3636
    },
    {
      "epoch": 0.5209855321587166,
      "grad_norm": 1.0954113006591797,
      "learning_rate": 9.807461174202707e-05,
      "loss": 1.3662,
      "step": 3637
    },
    {
      "epoch": 0.5211287781120183,
      "grad_norm": 1.0055007934570312,
      "learning_rate": 9.802822278705186e-05,
      "loss": 1.3512,
      "step": 3638
    },
    {
      "epoch": 0.5212720240653201,
      "grad_norm": 1.1127763986587524,
      "learning_rate": 9.798183425655156e-05,
      "loss": 1.6607,
      "step": 3639
    },
    {
      "epoch": 0.521415270018622,
      "grad_norm": 1.178781509399414,
      "learning_rate": 9.793544616051232e-05,
      "loss": 1.4596,
      "step": 3640
    },
    {
      "epoch": 0.5215585159719238,
      "grad_norm": 0.979458212852478,
      "learning_rate": 9.788905850892047e-05,
      "loss": 1.5598,
      "step": 3641
    },
    {
      "epoch": 0.5217017619252257,
      "grad_norm": 1.0697314739227295,
      "learning_rate": 9.78426713117621e-05,
      "loss": 1.3801,
      "step": 3642
    },
    {
      "epoch": 0.5218450078785274,
      "grad_norm": 1.142638921737671,
      "learning_rate": 9.779628457902319e-05,
      "loss": 1.3785,
      "step": 3643
    },
    {
      "epoch": 0.5219882538318292,
      "grad_norm": 1.3159234523773193,
      "learning_rate": 9.774989832068966e-05,
      "loss": 1.4439,
      "step": 3644
    },
    {
      "epoch": 0.5221314997851311,
      "grad_norm": 1.4077343940734863,
      "learning_rate": 9.770351254674738e-05,
      "loss": 1.4663,
      "step": 3645
    },
    {
      "epoch": 0.5222747457384329,
      "grad_norm": 1.0821431875228882,
      "learning_rate": 9.765712726718199e-05,
      "loss": 1.3828,
      "step": 3646
    },
    {
      "epoch": 0.5224179916917348,
      "grad_norm": 0.982258141040802,
      "learning_rate": 9.761074249197915e-05,
      "loss": 1.4064,
      "step": 3647
    },
    {
      "epoch": 0.5225612376450365,
      "grad_norm": 1.0557477474212646,
      "learning_rate": 9.75643582311243e-05,
      "loss": 1.4764,
      "step": 3648
    },
    {
      "epoch": 0.5227044835983383,
      "grad_norm": 1.007712721824646,
      "learning_rate": 9.751797449460285e-05,
      "loss": 1.5898,
      "step": 3649
    },
    {
      "epoch": 0.5228477295516402,
      "grad_norm": 0.9973738789558411,
      "learning_rate": 9.74715912924001e-05,
      "loss": 1.366,
      "step": 3650
    },
    {
      "epoch": 0.522990975504942,
      "grad_norm": 1.09053635597229,
      "learning_rate": 9.742520863450115e-05,
      "loss": 1.3529,
      "step": 3651
    },
    {
      "epoch": 0.5231342214582438,
      "grad_norm": 1.0612603425979614,
      "learning_rate": 9.737882653089107e-05,
      "loss": 1.3934,
      "step": 3652
    },
    {
      "epoch": 0.5232774674115456,
      "grad_norm": 1.039867877960205,
      "learning_rate": 9.733244499155479e-05,
      "loss": 1.5296,
      "step": 3653
    },
    {
      "epoch": 0.5234207133648474,
      "grad_norm": 1.1256277561187744,
      "learning_rate": 9.728606402647705e-05,
      "loss": 1.485,
      "step": 3654
    },
    {
      "epoch": 0.5235639593181493,
      "grad_norm": 1.45452880859375,
      "learning_rate": 9.723968364564264e-05,
      "loss": 1.5192,
      "step": 3655
    },
    {
      "epoch": 0.5237072052714511,
      "grad_norm": 1.5885354280471802,
      "learning_rate": 9.719330385903596e-05,
      "loss": 1.5644,
      "step": 3656
    },
    {
      "epoch": 0.5238504512247529,
      "grad_norm": 1.1291733980178833,
      "learning_rate": 9.714692467664152e-05,
      "loss": 1.3879,
      "step": 3657
    },
    {
      "epoch": 0.5239936971780548,
      "grad_norm": 1.141465663909912,
      "learning_rate": 9.710054610844364e-05,
      "loss": 1.6012,
      "step": 3658
    },
    {
      "epoch": 0.5241369431313565,
      "grad_norm": 1.2138019800186157,
      "learning_rate": 9.705416816442639e-05,
      "loss": 1.4978,
      "step": 3659
    },
    {
      "epoch": 0.5242801890846583,
      "grad_norm": 0.9463986158370972,
      "learning_rate": 9.700779085457386e-05,
      "loss": 1.4724,
      "step": 3660
    },
    {
      "epoch": 0.5244234350379602,
      "grad_norm": 1.2441587448120117,
      "learning_rate": 9.696141418886987e-05,
      "loss": 1.3979,
      "step": 3661
    },
    {
      "epoch": 0.524566680991262,
      "grad_norm": 1.083832859992981,
      "learning_rate": 9.691503817729824e-05,
      "loss": 1.5718,
      "step": 3662
    },
    {
      "epoch": 0.5247099269445639,
      "grad_norm": 1.232864499092102,
      "learning_rate": 9.686866282984256e-05,
      "loss": 1.2951,
      "step": 3663
    },
    {
      "epoch": 0.5248531728978656,
      "grad_norm": 0.9618181586265564,
      "learning_rate": 9.682228815648628e-05,
      "loss": 1.3801,
      "step": 3664
    },
    {
      "epoch": 0.5249964188511674,
      "grad_norm": 1.0909799337387085,
      "learning_rate": 9.67759141672127e-05,
      "loss": 1.3373,
      "step": 3665
    },
    {
      "epoch": 0.5251396648044693,
      "grad_norm": 0.9654986262321472,
      "learning_rate": 9.672954087200506e-05,
      "loss": 1.3922,
      "step": 3666
    },
    {
      "epoch": 0.5252829107577711,
      "grad_norm": 0.9920533895492554,
      "learning_rate": 9.668316828084628e-05,
      "loss": 1.389,
      "step": 3667
    },
    {
      "epoch": 0.525426156711073,
      "grad_norm": 1.0261787176132202,
      "learning_rate": 9.663679640371934e-05,
      "loss": 1.3685,
      "step": 3668
    },
    {
      "epoch": 0.5255694026643747,
      "grad_norm": 0.9672529101371765,
      "learning_rate": 9.65904252506069e-05,
      "loss": 1.5902,
      "step": 3669
    },
    {
      "epoch": 0.5257126486176765,
      "grad_norm": 1.0289409160614014,
      "learning_rate": 9.654405483149151e-05,
      "loss": 1.6726,
      "step": 3670
    },
    {
      "epoch": 0.5258558945709784,
      "grad_norm": 0.9697561860084534,
      "learning_rate": 9.649768515635562e-05,
      "loss": 1.4761,
      "step": 3671
    },
    {
      "epoch": 0.5259991405242802,
      "grad_norm": 1.0699905157089233,
      "learning_rate": 9.64513162351814e-05,
      "loss": 1.3822,
      "step": 3672
    },
    {
      "epoch": 0.526142386477582,
      "grad_norm": 1.0676175355911255,
      "learning_rate": 9.6404948077951e-05,
      "loss": 1.6505,
      "step": 3673
    },
    {
      "epoch": 0.5262856324308838,
      "grad_norm": 1.2039189338684082,
      "learning_rate": 9.635858069464635e-05,
      "loss": 1.5004,
      "step": 3674
    },
    {
      "epoch": 0.5264288783841856,
      "grad_norm": 1.1918784379959106,
      "learning_rate": 9.631221409524913e-05,
      "loss": 1.3343,
      "step": 3675
    },
    {
      "epoch": 0.5265721243374875,
      "grad_norm": 0.9595719575881958,
      "learning_rate": 9.6265848289741e-05,
      "loss": 1.4458,
      "step": 3676
    },
    {
      "epoch": 0.5267153702907893,
      "grad_norm": 1.1812224388122559,
      "learning_rate": 9.621948328810326e-05,
      "loss": 1.4513,
      "step": 3677
    },
    {
      "epoch": 0.5268586162440911,
      "grad_norm": 1.1946989297866821,
      "learning_rate": 9.617311910031724e-05,
      "loss": 1.4736,
      "step": 3678
    },
    {
      "epoch": 0.527001862197393,
      "grad_norm": 1.0645179748535156,
      "learning_rate": 9.6126755736364e-05,
      "loss": 1.4159,
      "step": 3679
    },
    {
      "epoch": 0.5271451081506947,
      "grad_norm": 1.4286433458328247,
      "learning_rate": 9.608039320622439e-05,
      "loss": 1.5471,
      "step": 3680
    },
    {
      "epoch": 0.5272883541039965,
      "grad_norm": 1.2019009590148926,
      "learning_rate": 9.60340315198791e-05,
      "loss": 1.5456,
      "step": 3681
    },
    {
      "epoch": 0.5274316000572984,
      "grad_norm": 1.0661293268203735,
      "learning_rate": 9.598767068730874e-05,
      "loss": 1.4934,
      "step": 3682
    },
    {
      "epoch": 0.5275748460106002,
      "grad_norm": 1.045753836631775,
      "learning_rate": 9.594131071849353e-05,
      "loss": 1.5277,
      "step": 3683
    },
    {
      "epoch": 0.5277180919639021,
      "grad_norm": 0.8283714652061462,
      "learning_rate": 9.589495162341374e-05,
      "loss": 1.4843,
      "step": 3684
    },
    {
      "epoch": 0.5278613379172038,
      "grad_norm": 1.2093822956085205,
      "learning_rate": 9.584859341204924e-05,
      "loss": 1.3494,
      "step": 3685
    },
    {
      "epoch": 0.5280045838705056,
      "grad_norm": 1.0161635875701904,
      "learning_rate": 9.580223609437986e-05,
      "loss": 1.4181,
      "step": 3686
    },
    {
      "epoch": 0.5281478298238075,
      "grad_norm": 1.0097510814666748,
      "learning_rate": 9.57558796803852e-05,
      "loss": 1.3536,
      "step": 3687
    },
    {
      "epoch": 0.5282910757771093,
      "grad_norm": 0.9038384556770325,
      "learning_rate": 9.570952418004455e-05,
      "loss": 1.492,
      "step": 3688
    },
    {
      "epoch": 0.5284343217304112,
      "grad_norm": 0.9352546334266663,
      "learning_rate": 9.566316960333726e-05,
      "loss": 1.4586,
      "step": 3689
    },
    {
      "epoch": 0.5285775676837129,
      "grad_norm": 1.070955514907837,
      "learning_rate": 9.561681596024215e-05,
      "loss": 1.5079,
      "step": 3690
    },
    {
      "epoch": 0.5287208136370147,
      "grad_norm": 1.0143582820892334,
      "learning_rate": 9.557046326073814e-05,
      "loss": 1.4311,
      "step": 3691
    },
    {
      "epoch": 0.5288640595903166,
      "grad_norm": 1.1470767259597778,
      "learning_rate": 9.552411151480378e-05,
      "loss": 1.5105,
      "step": 3692
    },
    {
      "epoch": 0.5290073055436184,
      "grad_norm": 1.3744410276412964,
      "learning_rate": 9.547776073241744e-05,
      "loss": 1.5286,
      "step": 3693
    },
    {
      "epoch": 0.5291505514969203,
      "grad_norm": 1.1075444221496582,
      "learning_rate": 9.543141092355727e-05,
      "loss": 1.2798,
      "step": 3694
    },
    {
      "epoch": 0.529293797450222,
      "grad_norm": 0.998053252696991,
      "learning_rate": 9.538506209820133e-05,
      "loss": 1.3604,
      "step": 3695
    },
    {
      "epoch": 0.5294370434035238,
      "grad_norm": 0.9933651685714722,
      "learning_rate": 9.53387142663273e-05,
      "loss": 1.4591,
      "step": 3696
    },
    {
      "epoch": 0.5295802893568257,
      "grad_norm": 1.2047055959701538,
      "learning_rate": 9.529236743791276e-05,
      "loss": 1.4494,
      "step": 3697
    },
    {
      "epoch": 0.5297235353101275,
      "grad_norm": 1.3577064275741577,
      "learning_rate": 9.524602162293497e-05,
      "loss": 1.6059,
      "step": 3698
    },
    {
      "epoch": 0.5298667812634293,
      "grad_norm": 0.9883847832679749,
      "learning_rate": 9.519967683137108e-05,
      "loss": 1.4963,
      "step": 3699
    },
    {
      "epoch": 0.5300100272167311,
      "grad_norm": 1.151240587234497,
      "learning_rate": 9.515333307319805e-05,
      "loss": 1.4894,
      "step": 3700
    },
    {
      "epoch": 0.5301532731700329,
      "grad_norm": 1.2949846982955933,
      "learning_rate": 9.51069903583924e-05,
      "loss": 1.3232,
      "step": 3701
    },
    {
      "epoch": 0.5302965191233348,
      "grad_norm": 1.2442485094070435,
      "learning_rate": 9.506064869693066e-05,
      "loss": 1.4509,
      "step": 3702
    },
    {
      "epoch": 0.5304397650766366,
      "grad_norm": 1.0899699926376343,
      "learning_rate": 9.501430809878906e-05,
      "loss": 1.394,
      "step": 3703
    },
    {
      "epoch": 0.5305830110299384,
      "grad_norm": 1.0936716794967651,
      "learning_rate": 9.496796857394352e-05,
      "loss": 1.5165,
      "step": 3704
    },
    {
      "epoch": 0.5307262569832403,
      "grad_norm": 0.9987739324569702,
      "learning_rate": 9.492163013236983e-05,
      "loss": 1.4658,
      "step": 3705
    },
    {
      "epoch": 0.530869502936542,
      "grad_norm": 1.4368233680725098,
      "learning_rate": 9.487529278404347e-05,
      "loss": 1.3469,
      "step": 3706
    },
    {
      "epoch": 0.5310127488898438,
      "grad_norm": 1.1306756734848022,
      "learning_rate": 9.482895653893978e-05,
      "loss": 1.3995,
      "step": 3707
    },
    {
      "epoch": 0.5311559948431457,
      "grad_norm": 0.9543665647506714,
      "learning_rate": 9.478262140703381e-05,
      "loss": 1.544,
      "step": 3708
    },
    {
      "epoch": 0.5312992407964475,
      "grad_norm": 1.261277198791504,
      "learning_rate": 9.473628739830032e-05,
      "loss": 1.2997,
      "step": 3709
    },
    {
      "epoch": 0.5314424867497494,
      "grad_norm": 1.018303394317627,
      "learning_rate": 9.468995452271387e-05,
      "loss": 1.4486,
      "step": 3710
    },
    {
      "epoch": 0.5315857327030511,
      "grad_norm": 1.0480808019638062,
      "learning_rate": 9.464362279024884e-05,
      "loss": 1.5603,
      "step": 3711
    },
    {
      "epoch": 0.5317289786563529,
      "grad_norm": 1.0268921852111816,
      "learning_rate": 9.459729221087926e-05,
      "loss": 1.504,
      "step": 3712
    },
    {
      "epoch": 0.5318722246096548,
      "grad_norm": 1.113730788230896,
      "learning_rate": 9.4550962794579e-05,
      "loss": 1.4887,
      "step": 3713
    },
    {
      "epoch": 0.5320154705629566,
      "grad_norm": 1.1313713788986206,
      "learning_rate": 9.450463455132158e-05,
      "loss": 1.3447,
      "step": 3714
    },
    {
      "epoch": 0.5321587165162585,
      "grad_norm": 0.9995348453521729,
      "learning_rate": 9.445830749108034e-05,
      "loss": 1.3711,
      "step": 3715
    },
    {
      "epoch": 0.5323019624695602,
      "grad_norm": 1.106647253036499,
      "learning_rate": 9.44119816238284e-05,
      "loss": 1.2895,
      "step": 3716
    },
    {
      "epoch": 0.532445208422862,
      "grad_norm": 1.117048978805542,
      "learning_rate": 9.43656569595385e-05,
      "loss": 1.147,
      "step": 3717
    },
    {
      "epoch": 0.5325884543761639,
      "grad_norm": 1.1832622289657593,
      "learning_rate": 9.431933350818326e-05,
      "loss": 1.5899,
      "step": 3718
    },
    {
      "epoch": 0.5327317003294657,
      "grad_norm": 0.983793318271637,
      "learning_rate": 9.427301127973491e-05,
      "loss": 1.5206,
      "step": 3719
    },
    {
      "epoch": 0.5328749462827675,
      "grad_norm": 1.0098230838775635,
      "learning_rate": 9.422669028416552e-05,
      "loss": 1.3397,
      "step": 3720
    },
    {
      "epoch": 0.5330181922360693,
      "grad_norm": 1.0970121622085571,
      "learning_rate": 9.418037053144687e-05,
      "loss": 1.4693,
      "step": 3721
    },
    {
      "epoch": 0.5331614381893711,
      "grad_norm": 1.0659592151641846,
      "learning_rate": 9.413405203155038e-05,
      "loss": 1.4068,
      "step": 3722
    },
    {
      "epoch": 0.533304684142673,
      "grad_norm": 1.0793672800064087,
      "learning_rate": 9.408773479444736e-05,
      "loss": 1.4173,
      "step": 3723
    },
    {
      "epoch": 0.5334479300959748,
      "grad_norm": 1.0774383544921875,
      "learning_rate": 9.404141883010874e-05,
      "loss": 1.3618,
      "step": 3724
    },
    {
      "epoch": 0.5335911760492766,
      "grad_norm": 1.050866961479187,
      "learning_rate": 9.399510414850518e-05,
      "loss": 1.3432,
      "step": 3725
    },
    {
      "epoch": 0.5337344220025785,
      "grad_norm": 1.3098955154418945,
      "learning_rate": 9.394879075960712e-05,
      "loss": 1.4692,
      "step": 3726
    },
    {
      "epoch": 0.5338776679558802,
      "grad_norm": 1.521201491355896,
      "learning_rate": 9.390247867338464e-05,
      "loss": 1.3366,
      "step": 3727
    },
    {
      "epoch": 0.534020913909182,
      "grad_norm": 1.0459558963775635,
      "learning_rate": 9.385616789980759e-05,
      "loss": 1.3019,
      "step": 3728
    },
    {
      "epoch": 0.5341641598624839,
      "grad_norm": 1.0965683460235596,
      "learning_rate": 9.380985844884561e-05,
      "loss": 1.2818,
      "step": 3729
    },
    {
      "epoch": 0.5343074058157857,
      "grad_norm": 0.9563542604446411,
      "learning_rate": 9.376355033046789e-05,
      "loss": 1.516,
      "step": 3730
    },
    {
      "epoch": 0.5344506517690876,
      "grad_norm": 1.326822280883789,
      "learning_rate": 9.371724355464346e-05,
      "loss": 1.5117,
      "step": 3731
    },
    {
      "epoch": 0.5345938977223893,
      "grad_norm": 0.9649562239646912,
      "learning_rate": 9.367093813134103e-05,
      "loss": 1.5129,
      "step": 3732
    },
    {
      "epoch": 0.5347371436756911,
      "grad_norm": 0.9666440486907959,
      "learning_rate": 9.362463407052898e-05,
      "loss": 1.5485,
      "step": 3733
    },
    {
      "epoch": 0.534880389628993,
      "grad_norm": 1.1279494762420654,
      "learning_rate": 9.357833138217549e-05,
      "loss": 1.4396,
      "step": 3734
    },
    {
      "epoch": 0.5350236355822948,
      "grad_norm": 1.0130126476287842,
      "learning_rate": 9.35320300762483e-05,
      "loss": 1.3434,
      "step": 3735
    },
    {
      "epoch": 0.5351668815355967,
      "grad_norm": 1.3631035089492798,
      "learning_rate": 9.348573016271498e-05,
      "loss": 1.6268,
      "step": 3736
    },
    {
      "epoch": 0.5353101274888984,
      "grad_norm": 1.3924418687820435,
      "learning_rate": 9.343943165154278e-05,
      "loss": 1.6327,
      "step": 3737
    },
    {
      "epoch": 0.5354533734422002,
      "grad_norm": 1.0600417852401733,
      "learning_rate": 9.339313455269856e-05,
      "loss": 1.4602,
      "step": 3738
    },
    {
      "epoch": 0.5355966193955021,
      "grad_norm": 0.9704450368881226,
      "learning_rate": 9.334683887614902e-05,
      "loss": 1.4972,
      "step": 3739
    },
    {
      "epoch": 0.5357398653488039,
      "grad_norm": 1.1102216243743896,
      "learning_rate": 9.330054463186038e-05,
      "loss": 1.4775,
      "step": 3740
    },
    {
      "epoch": 0.5358831113021058,
      "grad_norm": 1.1245002746582031,
      "learning_rate": 9.325425182979873e-05,
      "loss": 1.4571,
      "step": 3741
    },
    {
      "epoch": 0.5360263572554075,
      "grad_norm": 1.1927329301834106,
      "learning_rate": 9.320796047992973e-05,
      "loss": 1.2818,
      "step": 3742
    },
    {
      "epoch": 0.5361696032087093,
      "grad_norm": 0.9984963536262512,
      "learning_rate": 9.316167059221874e-05,
      "loss": 1.5256,
      "step": 3743
    },
    {
      "epoch": 0.5363128491620112,
      "grad_norm": 1.1356059312820435,
      "learning_rate": 9.311538217663084e-05,
      "loss": 1.4571,
      "step": 3744
    },
    {
      "epoch": 0.536456095115313,
      "grad_norm": 1.1551594734191895,
      "learning_rate": 9.306909524313083e-05,
      "loss": 1.4098,
      "step": 3745
    },
    {
      "epoch": 0.5365993410686148,
      "grad_norm": 1.1143728494644165,
      "learning_rate": 9.302280980168307e-05,
      "loss": 1.3824,
      "step": 3746
    },
    {
      "epoch": 0.5367425870219167,
      "grad_norm": 0.9751464128494263,
      "learning_rate": 9.297652586225175e-05,
      "loss": 1.505,
      "step": 3747
    },
    {
      "epoch": 0.5368858329752184,
      "grad_norm": 1.0940778255462646,
      "learning_rate": 9.293024343480055e-05,
      "loss": 1.3833,
      "step": 3748
    },
    {
      "epoch": 0.5370290789285203,
      "grad_norm": 1.0687774419784546,
      "learning_rate": 9.288396252929299e-05,
      "loss": 1.3458,
      "step": 3749
    },
    {
      "epoch": 0.5371723248818221,
      "grad_norm": 1.0949299335479736,
      "learning_rate": 9.283768315569226e-05,
      "loss": 1.5347,
      "step": 3750
    },
    {
      "epoch": 0.5373155708351239,
      "grad_norm": 1.0210727453231812,
      "learning_rate": 9.279140532396104e-05,
      "loss": 1.4049,
      "step": 3751
    },
    {
      "epoch": 0.5374588167884258,
      "grad_norm": 1.2851144075393677,
      "learning_rate": 9.27451290440619e-05,
      "loss": 1.4596,
      "step": 3752
    },
    {
      "epoch": 0.5376020627417275,
      "grad_norm": 1.1687676906585693,
      "learning_rate": 9.269885432595696e-05,
      "loss": 1.6354,
      "step": 3753
    },
    {
      "epoch": 0.5377453086950293,
      "grad_norm": 1.020384669303894,
      "learning_rate": 9.265258117960797e-05,
      "loss": 1.5364,
      "step": 3754
    },
    {
      "epoch": 0.5378885546483312,
      "grad_norm": 1.1265329122543335,
      "learning_rate": 9.260630961497648e-05,
      "loss": 1.384,
      "step": 3755
    },
    {
      "epoch": 0.538031800601633,
      "grad_norm": 0.8979662656784058,
      "learning_rate": 9.256003964202352e-05,
      "loss": 1.5755,
      "step": 3756
    },
    {
      "epoch": 0.5381750465549349,
      "grad_norm": 1.2778605222702026,
      "learning_rate": 9.251377127070992e-05,
      "loss": 1.3791,
      "step": 3757
    },
    {
      "epoch": 0.5383182925082366,
      "grad_norm": 0.974638819694519,
      "learning_rate": 9.246750451099615e-05,
      "loss": 1.3681,
      "step": 3758
    },
    {
      "epoch": 0.5384615384615384,
      "grad_norm": 1.0232633352279663,
      "learning_rate": 9.242123937284223e-05,
      "loss": 1.5524,
      "step": 3759
    },
    {
      "epoch": 0.5386047844148403,
      "grad_norm": 1.1448662281036377,
      "learning_rate": 9.23749758662079e-05,
      "loss": 1.402,
      "step": 3760
    },
    {
      "epoch": 0.5387480303681421,
      "grad_norm": 1.3974542617797852,
      "learning_rate": 9.232871400105265e-05,
      "loss": 1.3393,
      "step": 3761
    },
    {
      "epoch": 0.538891276321444,
      "grad_norm": 1.007043480873108,
      "learning_rate": 9.228245378733537e-05,
      "loss": 1.2798,
      "step": 3762
    },
    {
      "epoch": 0.5390345222747457,
      "grad_norm": 1.0703458786010742,
      "learning_rate": 9.223619523501484e-05,
      "loss": 1.4011,
      "step": 3763
    },
    {
      "epoch": 0.5391777682280475,
      "grad_norm": 1.2995879650115967,
      "learning_rate": 9.218993835404935e-05,
      "loss": 1.4441,
      "step": 3764
    },
    {
      "epoch": 0.5393210141813494,
      "grad_norm": 1.1272830963134766,
      "learning_rate": 9.214368315439685e-05,
      "loss": 1.4058,
      "step": 3765
    },
    {
      "epoch": 0.5394642601346512,
      "grad_norm": 1.0471073389053345,
      "learning_rate": 9.209742964601498e-05,
      "loss": 1.4818,
      "step": 3766
    },
    {
      "epoch": 0.539607506087953,
      "grad_norm": 1.1284985542297363,
      "learning_rate": 9.205117783886091e-05,
      "loss": 1.4006,
      "step": 3767
    },
    {
      "epoch": 0.5397507520412549,
      "grad_norm": 1.1525839567184448,
      "learning_rate": 9.20049277428916e-05,
      "loss": 1.4398,
      "step": 3768
    },
    {
      "epoch": 0.5398939979945566,
      "grad_norm": 0.9529622197151184,
      "learning_rate": 9.195867936806346e-05,
      "loss": 1.3445,
      "step": 3769
    },
    {
      "epoch": 0.5400372439478585,
      "grad_norm": 1.156707525253296,
      "learning_rate": 9.191243272433268e-05,
      "loss": 1.3769,
      "step": 3770
    },
    {
      "epoch": 0.5401804899011603,
      "grad_norm": 1.3044835329055786,
      "learning_rate": 9.1866187821655e-05,
      "loss": 1.5264,
      "step": 3771
    },
    {
      "epoch": 0.5403237358544621,
      "grad_norm": 1.090537190437317,
      "learning_rate": 9.181994466998577e-05,
      "loss": 1.2196,
      "step": 3772
    },
    {
      "epoch": 0.540466981807764,
      "grad_norm": 1.139548659324646,
      "learning_rate": 9.177370327928004e-05,
      "loss": 1.4651,
      "step": 3773
    },
    {
      "epoch": 0.5406102277610657,
      "grad_norm": 1.1102821826934814,
      "learning_rate": 9.172746365949246e-05,
      "loss": 1.3472,
      "step": 3774
    },
    {
      "epoch": 0.5407534737143675,
      "grad_norm": 1.2802883386611938,
      "learning_rate": 9.16812258205772e-05,
      "loss": 1.4484,
      "step": 3775
    },
    {
      "epoch": 0.5408967196676694,
      "grad_norm": 0.9637173414230347,
      "learning_rate": 9.16349897724882e-05,
      "loss": 1.2357,
      "step": 3776
    },
    {
      "epoch": 0.5410399656209712,
      "grad_norm": 0.9517381191253662,
      "learning_rate": 9.158875552517887e-05,
      "loss": 1.3801,
      "step": 3777
    },
    {
      "epoch": 0.5411832115742731,
      "grad_norm": 1.0572885274887085,
      "learning_rate": 9.154252308860231e-05,
      "loss": 1.2741,
      "step": 3778
    },
    {
      "epoch": 0.5413264575275748,
      "grad_norm": 1.0451692342758179,
      "learning_rate": 9.149629247271128e-05,
      "loss": 1.5178,
      "step": 3779
    },
    {
      "epoch": 0.5414697034808766,
      "grad_norm": 1.0329433679580688,
      "learning_rate": 9.145006368745803e-05,
      "loss": 1.2116,
      "step": 3780
    },
    {
      "epoch": 0.5416129494341785,
      "grad_norm": 1.1981528997421265,
      "learning_rate": 9.14038367427945e-05,
      "loss": 1.4681,
      "step": 3781
    },
    {
      "epoch": 0.5417561953874803,
      "grad_norm": 1.1192377805709839,
      "learning_rate": 9.135761164867221e-05,
      "loss": 1.5535,
      "step": 3782
    },
    {
      "epoch": 0.5418994413407822,
      "grad_norm": 1.038725733757019,
      "learning_rate": 9.131138841504222e-05,
      "loss": 1.3324,
      "step": 3783
    },
    {
      "epoch": 0.5420426872940839,
      "grad_norm": 1.2320663928985596,
      "learning_rate": 9.126516705185535e-05,
      "loss": 1.4358,
      "step": 3784
    },
    {
      "epoch": 0.5421859332473857,
      "grad_norm": 1.1724793910980225,
      "learning_rate": 9.121894756906181e-05,
      "loss": 1.5002,
      "step": 3785
    },
    {
      "epoch": 0.5423291792006876,
      "grad_norm": 0.9963299036026001,
      "learning_rate": 9.117272997661157e-05,
      "loss": 1.4899,
      "step": 3786
    },
    {
      "epoch": 0.5424724251539894,
      "grad_norm": 1.2009650468826294,
      "learning_rate": 9.112651428445416e-05,
      "loss": 1.5703,
      "step": 3787
    },
    {
      "epoch": 0.5426156711072913,
      "grad_norm": 1.2211109399795532,
      "learning_rate": 9.108030050253861e-05,
      "loss": 1.4439,
      "step": 3788
    },
    {
      "epoch": 0.5427589170605931,
      "grad_norm": 1.1849770545959473,
      "learning_rate": 9.103408864081365e-05,
      "loss": 1.3528,
      "step": 3789
    },
    {
      "epoch": 0.5429021630138948,
      "grad_norm": 1.1851774454116821,
      "learning_rate": 9.098787870922751e-05,
      "loss": 1.5033,
      "step": 3790
    },
    {
      "epoch": 0.5430454089671967,
      "grad_norm": 1.2212743759155273,
      "learning_rate": 9.094167071772808e-05,
      "loss": 1.4093,
      "step": 3791
    },
    {
      "epoch": 0.5431886549204985,
      "grad_norm": 1.0389831066131592,
      "learning_rate": 9.08954646762628e-05,
      "loss": 1.4082,
      "step": 3792
    },
    {
      "epoch": 0.5433319008738003,
      "grad_norm": 0.9905836582183838,
      "learning_rate": 9.084926059477865e-05,
      "loss": 1.3448,
      "step": 3793
    },
    {
      "epoch": 0.5434751468271022,
      "grad_norm": 1.2270208597183228,
      "learning_rate": 9.080305848322223e-05,
      "loss": 1.4081,
      "step": 3794
    },
    {
      "epoch": 0.5436183927804039,
      "grad_norm": 0.9713574051856995,
      "learning_rate": 9.075685835153981e-05,
      "loss": 1.4037,
      "step": 3795
    },
    {
      "epoch": 0.5437616387337058,
      "grad_norm": 0.9763878583908081,
      "learning_rate": 9.071066020967698e-05,
      "loss": 1.4406,
      "step": 3796
    },
    {
      "epoch": 0.5439048846870076,
      "grad_norm": 1.0408380031585693,
      "learning_rate": 9.066446406757916e-05,
      "loss": 1.503,
      "step": 3797
    },
    {
      "epoch": 0.5440481306403094,
      "grad_norm": 1.399655818939209,
      "learning_rate": 9.06182699351912e-05,
      "loss": 1.2813,
      "step": 3798
    },
    {
      "epoch": 0.5441913765936113,
      "grad_norm": 1.4461750984191895,
      "learning_rate": 9.057207782245757e-05,
      "loss": 1.4684,
      "step": 3799
    },
    {
      "epoch": 0.544334622546913,
      "grad_norm": 1.0886634588241577,
      "learning_rate": 9.052588773932228e-05,
      "loss": 1.2758,
      "step": 3800
    },
    {
      "epoch": 0.5444778685002148,
      "grad_norm": 1.2171721458435059,
      "learning_rate": 9.047969969572889e-05,
      "loss": 1.5761,
      "step": 3801
    },
    {
      "epoch": 0.5446211144535167,
      "grad_norm": 1.094599723815918,
      "learning_rate": 9.043351370162058e-05,
      "loss": 1.3756,
      "step": 3802
    },
    {
      "epoch": 0.5447643604068185,
      "grad_norm": 0.988581120967865,
      "learning_rate": 9.038732976694005e-05,
      "loss": 1.3801,
      "step": 3803
    },
    {
      "epoch": 0.5449076063601204,
      "grad_norm": 1.0670268535614014,
      "learning_rate": 9.034114790162952e-05,
      "loss": 1.5492,
      "step": 3804
    },
    {
      "epoch": 0.5450508523134221,
      "grad_norm": 1.0462517738342285,
      "learning_rate": 9.029496811563086e-05,
      "loss": 1.4568,
      "step": 3805
    },
    {
      "epoch": 0.5451940982667239,
      "grad_norm": 1.123539924621582,
      "learning_rate": 9.024879041888535e-05,
      "loss": 1.6635,
      "step": 3806
    },
    {
      "epoch": 0.5453373442200258,
      "grad_norm": 0.9439468383789062,
      "learning_rate": 9.020261482133398e-05,
      "loss": 1.3692,
      "step": 3807
    },
    {
      "epoch": 0.5454805901733276,
      "grad_norm": 1.149488925933838,
      "learning_rate": 9.015644133291723e-05,
      "loss": 1.316,
      "step": 3808
    },
    {
      "epoch": 0.5456238361266295,
      "grad_norm": 1.1205295324325562,
      "learning_rate": 9.011026996357503e-05,
      "loss": 1.5297,
      "step": 3809
    },
    {
      "epoch": 0.5457670820799312,
      "grad_norm": 1.0543869733810425,
      "learning_rate": 9.006410072324696e-05,
      "loss": 1.4257,
      "step": 3810
    },
    {
      "epoch": 0.545910328033233,
      "grad_norm": 1.0220791101455688,
      "learning_rate": 9.00179336218722e-05,
      "loss": 1.5224,
      "step": 3811
    },
    {
      "epoch": 0.5460535739865349,
      "grad_norm": 1.0441293716430664,
      "learning_rate": 8.997176866938923e-05,
      "loss": 1.6382,
      "step": 3812
    },
    {
      "epoch": 0.5461968199398367,
      "grad_norm": 1.1052736043930054,
      "learning_rate": 8.992560587573636e-05,
      "loss": 1.4608,
      "step": 3813
    },
    {
      "epoch": 0.5463400658931385,
      "grad_norm": 1.0017259120941162,
      "learning_rate": 8.987944525085123e-05,
      "loss": 1.5593,
      "step": 3814
    },
    {
      "epoch": 0.5464833118464404,
      "grad_norm": 1.206753134727478,
      "learning_rate": 8.983328680467108e-05,
      "loss": 1.4171,
      "step": 3815
    },
    {
      "epoch": 0.5466265577997421,
      "grad_norm": 1.1038222312927246,
      "learning_rate": 8.978713054713271e-05,
      "loss": 1.5701,
      "step": 3816
    },
    {
      "epoch": 0.546769803753044,
      "grad_norm": 1.0201430320739746,
      "learning_rate": 8.974097648817238e-05,
      "loss": 1.449,
      "step": 3817
    },
    {
      "epoch": 0.5469130497063458,
      "grad_norm": 0.9676105976104736,
      "learning_rate": 8.9694824637726e-05,
      "loss": 1.4471,
      "step": 3818
    },
    {
      "epoch": 0.5470562956596476,
      "grad_norm": 1.0683642625808716,
      "learning_rate": 8.964867500572878e-05,
      "loss": 1.4505,
      "step": 3819
    },
    {
      "epoch": 0.5471995416129495,
      "grad_norm": 1.0126845836639404,
      "learning_rate": 8.960252760211571e-05,
      "loss": 1.2989,
      "step": 3820
    },
    {
      "epoch": 0.5473427875662512,
      "grad_norm": 1.1245943307876587,
      "learning_rate": 8.955638243682118e-05,
      "loss": 1.2839,
      "step": 3821
    },
    {
      "epoch": 0.547486033519553,
      "grad_norm": 1.096720814704895,
      "learning_rate": 8.951023951977904e-05,
      "loss": 1.4595,
      "step": 3822
    },
    {
      "epoch": 0.5476292794728549,
      "grad_norm": 1.269492268562317,
      "learning_rate": 8.946409886092273e-05,
      "loss": 1.2394,
      "step": 3823
    },
    {
      "epoch": 0.5477725254261567,
      "grad_norm": 1.0266296863555908,
      "learning_rate": 8.941796047018525e-05,
      "loss": 1.5995,
      "step": 3824
    },
    {
      "epoch": 0.5479157713794586,
      "grad_norm": 1.0450046062469482,
      "learning_rate": 8.9371824357499e-05,
      "loss": 1.3744,
      "step": 3825
    },
    {
      "epoch": 0.5480590173327603,
      "grad_norm": 1.100357174873352,
      "learning_rate": 8.932569053279599e-05,
      "loss": 1.2157,
      "step": 3826
    },
    {
      "epoch": 0.5482022632860621,
      "grad_norm": 1.2342329025268555,
      "learning_rate": 8.927955900600762e-05,
      "loss": 1.4972,
      "step": 3827
    },
    {
      "epoch": 0.548345509239364,
      "grad_norm": 1.069729208946228,
      "learning_rate": 8.92334297870649e-05,
      "loss": 1.6715,
      "step": 3828
    },
    {
      "epoch": 0.5484887551926658,
      "grad_norm": 1.1644538640975952,
      "learning_rate": 8.918730288589835e-05,
      "loss": 1.3111,
      "step": 3829
    },
    {
      "epoch": 0.5486320011459677,
      "grad_norm": 0.9721007943153381,
      "learning_rate": 8.914117831243789e-05,
      "loss": 1.381,
      "step": 3830
    },
    {
      "epoch": 0.5487752470992694,
      "grad_norm": 1.3485933542251587,
      "learning_rate": 8.909505607661304e-05,
      "loss": 1.3944,
      "step": 3831
    },
    {
      "epoch": 0.5489184930525712,
      "grad_norm": 0.9442260265350342,
      "learning_rate": 8.904893618835277e-05,
      "loss": 1.5614,
      "step": 3832
    },
    {
      "epoch": 0.5490617390058731,
      "grad_norm": 0.9943949580192566,
      "learning_rate": 8.900281865758551e-05,
      "loss": 1.4188,
      "step": 3833
    },
    {
      "epoch": 0.5492049849591749,
      "grad_norm": 1.1620928049087524,
      "learning_rate": 8.89567034942393e-05,
      "loss": 1.4703,
      "step": 3834
    },
    {
      "epoch": 0.5493482309124768,
      "grad_norm": 1.2041715383529663,
      "learning_rate": 8.89105907082415e-05,
      "loss": 1.3214,
      "step": 3835
    },
    {
      "epoch": 0.5494914768657786,
      "grad_norm": 1.1206190586090088,
      "learning_rate": 8.886448030951912e-05,
      "loss": 1.4794,
      "step": 3836
    },
    {
      "epoch": 0.5496347228190803,
      "grad_norm": 1.107960820198059,
      "learning_rate": 8.881837230799859e-05,
      "loss": 1.4781,
      "step": 3837
    },
    {
      "epoch": 0.5497779687723822,
      "grad_norm": 1.121025800704956,
      "learning_rate": 8.877226671360577e-05,
      "loss": 1.5051,
      "step": 3838
    },
    {
      "epoch": 0.549921214725684,
      "grad_norm": 1.0006877183914185,
      "learning_rate": 8.872616353626608e-05,
      "loss": 1.4129,
      "step": 3839
    },
    {
      "epoch": 0.5500644606789858,
      "grad_norm": 1.0987566709518433,
      "learning_rate": 8.868006278590442e-05,
      "loss": 1.428,
      "step": 3840
    },
    {
      "epoch": 0.5502077066322877,
      "grad_norm": 1.187184453010559,
      "learning_rate": 8.863396447244511e-05,
      "loss": 1.4837,
      "step": 3841
    },
    {
      "epoch": 0.5503509525855894,
      "grad_norm": 0.9451871514320374,
      "learning_rate": 8.8587868605812e-05,
      "loss": 1.4214,
      "step": 3842
    },
    {
      "epoch": 0.5504941985388913,
      "grad_norm": 1.081092119216919,
      "learning_rate": 8.854177519592834e-05,
      "loss": 1.5579,
      "step": 3843
    },
    {
      "epoch": 0.5506374444921931,
      "grad_norm": 1.0740270614624023,
      "learning_rate": 8.849568425271693e-05,
      "loss": 1.4188,
      "step": 3844
    },
    {
      "epoch": 0.5507806904454949,
      "grad_norm": 1.0851736068725586,
      "learning_rate": 8.844959578610005e-05,
      "loss": 1.3686,
      "step": 3845
    },
    {
      "epoch": 0.5509239363987968,
      "grad_norm": 1.0271912813186646,
      "learning_rate": 8.840350980599934e-05,
      "loss": 1.4093,
      "step": 3846
    },
    {
      "epoch": 0.5510671823520985,
      "grad_norm": 1.0875070095062256,
      "learning_rate": 8.835742632233603e-05,
      "loss": 1.6076,
      "step": 3847
    },
    {
      "epoch": 0.5512104283054003,
      "grad_norm": 0.9868462085723877,
      "learning_rate": 8.831134534503069e-05,
      "loss": 1.4214,
      "step": 3848
    },
    {
      "epoch": 0.5513536742587022,
      "grad_norm": 1.0932023525238037,
      "learning_rate": 8.826526688400346e-05,
      "loss": 1.516,
      "step": 3849
    },
    {
      "epoch": 0.551496920212004,
      "grad_norm": 1.0740574598312378,
      "learning_rate": 8.82191909491739e-05,
      "loss": 1.4841,
      "step": 3850
    },
    {
      "epoch": 0.5516401661653059,
      "grad_norm": 1.0741970539093018,
      "learning_rate": 8.817311755046095e-05,
      "loss": 1.3124,
      "step": 3851
    },
    {
      "epoch": 0.5517834121186076,
      "grad_norm": 1.0681806802749634,
      "learning_rate": 8.812704669778315e-05,
      "loss": 1.3405,
      "step": 3852
    },
    {
      "epoch": 0.5519266580719094,
      "grad_norm": 1.030573844909668,
      "learning_rate": 8.808097840105841e-05,
      "loss": 1.4681,
      "step": 3853
    },
    {
      "epoch": 0.5520699040252113,
      "grad_norm": 1.0412677526474,
      "learning_rate": 8.803491267020404e-05,
      "loss": 1.4991,
      "step": 3854
    },
    {
      "epoch": 0.5522131499785131,
      "grad_norm": 1.028397798538208,
      "learning_rate": 8.79888495151369e-05,
      "loss": 1.3751,
      "step": 3855
    },
    {
      "epoch": 0.552356395931815,
      "grad_norm": 1.0425646305084229,
      "learning_rate": 8.794278894577322e-05,
      "loss": 1.4308,
      "step": 3856
    },
    {
      "epoch": 0.5524996418851168,
      "grad_norm": 1.0469813346862793,
      "learning_rate": 8.789673097202869e-05,
      "loss": 1.4158,
      "step": 3857
    },
    {
      "epoch": 0.5526428878384185,
      "grad_norm": 1.0224820375442505,
      "learning_rate": 8.785067560381852e-05,
      "loss": 1.4871,
      "step": 3858
    },
    {
      "epoch": 0.5527861337917204,
      "grad_norm": 1.2438688278198242,
      "learning_rate": 8.78046228510572e-05,
      "loss": 1.5247,
      "step": 3859
    },
    {
      "epoch": 0.5529293797450222,
      "grad_norm": 1.1570391654968262,
      "learning_rate": 8.77585727236588e-05,
      "loss": 1.4033,
      "step": 3860
    },
    {
      "epoch": 0.553072625698324,
      "grad_norm": 1.131225347518921,
      "learning_rate": 8.771252523153679e-05,
      "loss": 1.4037,
      "step": 3861
    },
    {
      "epoch": 0.5532158716516259,
      "grad_norm": 1.3698474168777466,
      "learning_rate": 8.766648038460399e-05,
      "loss": 1.529,
      "step": 3862
    },
    {
      "epoch": 0.5533591176049276,
      "grad_norm": 1.0556474924087524,
      "learning_rate": 8.762043819277279e-05,
      "loss": 1.5553,
      "step": 3863
    },
    {
      "epoch": 0.5535023635582295,
      "grad_norm": 1.063498854637146,
      "learning_rate": 8.757439866595489e-05,
      "loss": 1.4272,
      "step": 3864
    },
    {
      "epoch": 0.5536456095115313,
      "grad_norm": 1.147944450378418,
      "learning_rate": 8.752836181406147e-05,
      "loss": 1.4297,
      "step": 3865
    },
    {
      "epoch": 0.5537888554648331,
      "grad_norm": 1.038952350616455,
      "learning_rate": 8.748232764700314e-05,
      "loss": 1.4522,
      "step": 3866
    },
    {
      "epoch": 0.553932101418135,
      "grad_norm": 1.064469814300537,
      "learning_rate": 8.743629617468986e-05,
      "loss": 1.4506,
      "step": 3867
    },
    {
      "epoch": 0.5540753473714367,
      "grad_norm": 1.461004376411438,
      "learning_rate": 8.739026740703119e-05,
      "loss": 1.3769,
      "step": 3868
    },
    {
      "epoch": 0.5542185933247386,
      "grad_norm": 1.429337739944458,
      "learning_rate": 8.734424135393582e-05,
      "loss": 1.4826,
      "step": 3869
    },
    {
      "epoch": 0.5543618392780404,
      "grad_norm": 1.0875413417816162,
      "learning_rate": 8.729821802531212e-05,
      "loss": 1.4437,
      "step": 3870
    },
    {
      "epoch": 0.5545050852313422,
      "grad_norm": 1.3868958950042725,
      "learning_rate": 8.72521974310678e-05,
      "loss": 1.4189,
      "step": 3871
    },
    {
      "epoch": 0.5546483311846441,
      "grad_norm": 1.039495587348938,
      "learning_rate": 8.720617958110989e-05,
      "loss": 1.3494,
      "step": 3872
    },
    {
      "epoch": 0.5547915771379458,
      "grad_norm": 1.0088516473770142,
      "learning_rate": 8.71601644853449e-05,
      "loss": 1.3424,
      "step": 3873
    },
    {
      "epoch": 0.5549348230912476,
      "grad_norm": 1.179149866104126,
      "learning_rate": 8.71141521536788e-05,
      "loss": 1.4375,
      "step": 3874
    },
    {
      "epoch": 0.5550780690445495,
      "grad_norm": 0.9701326489448547,
      "learning_rate": 8.706814259601684e-05,
      "loss": 1.3753,
      "step": 3875
    },
    {
      "epoch": 0.5552213149978513,
      "grad_norm": 1.1500099897384644,
      "learning_rate": 8.702213582226379e-05,
      "loss": 1.5272,
      "step": 3876
    },
    {
      "epoch": 0.5553645609511532,
      "grad_norm": 0.8950748443603516,
      "learning_rate": 8.697613184232372e-05,
      "loss": 1.4617,
      "step": 3877
    },
    {
      "epoch": 0.555507806904455,
      "grad_norm": 1.3198305368423462,
      "learning_rate": 8.693013066610016e-05,
      "loss": 1.4492,
      "step": 3878
    },
    {
      "epoch": 0.5556510528577567,
      "grad_norm": 0.9568539261817932,
      "learning_rate": 8.68841323034961e-05,
      "loss": 1.398,
      "step": 3879
    },
    {
      "epoch": 0.5557942988110586,
      "grad_norm": 1.0182719230651855,
      "learning_rate": 8.68381367644137e-05,
      "loss": 1.3819,
      "step": 3880
    },
    {
      "epoch": 0.5559375447643604,
      "grad_norm": 1.1024657487869263,
      "learning_rate": 8.67921440587548e-05,
      "loss": 1.4614,
      "step": 3881
    },
    {
      "epoch": 0.5560807907176623,
      "grad_norm": 1.0097256898880005,
      "learning_rate": 8.674615419642044e-05,
      "loss": 1.5333,
      "step": 3882
    },
    {
      "epoch": 0.5562240366709641,
      "grad_norm": 1.2199742794036865,
      "learning_rate": 8.670016718731111e-05,
      "loss": 1.3474,
      "step": 3883
    },
    {
      "epoch": 0.5563672826242658,
      "grad_norm": 1.0519652366638184,
      "learning_rate": 8.665418304132666e-05,
      "loss": 1.5538,
      "step": 3884
    },
    {
      "epoch": 0.5565105285775677,
      "grad_norm": 1.143084168434143,
      "learning_rate": 8.66082017683663e-05,
      "loss": 1.2621,
      "step": 3885
    },
    {
      "epoch": 0.5566537745308695,
      "grad_norm": 1.058337688446045,
      "learning_rate": 8.656222337832875e-05,
      "loss": 1.4939,
      "step": 3886
    },
    {
      "epoch": 0.5567970204841713,
      "grad_norm": 1.0118240118026733,
      "learning_rate": 8.651624788111197e-05,
      "loss": 1.394,
      "step": 3887
    },
    {
      "epoch": 0.5569402664374732,
      "grad_norm": 1.1734325885772705,
      "learning_rate": 8.647027528661333e-05,
      "loss": 1.4291,
      "step": 3888
    },
    {
      "epoch": 0.5570835123907749,
      "grad_norm": 1.0719181299209595,
      "learning_rate": 8.64243056047296e-05,
      "loss": 1.4236,
      "step": 3889
    },
    {
      "epoch": 0.5572267583440768,
      "grad_norm": 1.1030477285385132,
      "learning_rate": 8.6378338845357e-05,
      "loss": 1.4701,
      "step": 3890
    },
    {
      "epoch": 0.5573700042973786,
      "grad_norm": 1.020727515220642,
      "learning_rate": 8.633237501839089e-05,
      "loss": 1.418,
      "step": 3891
    },
    {
      "epoch": 0.5575132502506804,
      "grad_norm": 0.9887786507606506,
      "learning_rate": 8.628641413372625e-05,
      "loss": 1.4229,
      "step": 3892
    },
    {
      "epoch": 0.5576564962039823,
      "grad_norm": 1.0122346878051758,
      "learning_rate": 8.624045620125727e-05,
      "loss": 1.463,
      "step": 3893
    },
    {
      "epoch": 0.557799742157284,
      "grad_norm": 1.035226821899414,
      "learning_rate": 8.619450123087757e-05,
      "loss": 1.458,
      "step": 3894
    },
    {
      "epoch": 0.5579429881105858,
      "grad_norm": 1.1406611204147339,
      "learning_rate": 8.614854923248016e-05,
      "loss": 1.3109,
      "step": 3895
    },
    {
      "epoch": 0.5580862340638877,
      "grad_norm": 1.2442231178283691,
      "learning_rate": 8.610260021595728e-05,
      "loss": 1.5573,
      "step": 3896
    },
    {
      "epoch": 0.5582294800171895,
      "grad_norm": 1.1228947639465332,
      "learning_rate": 8.605665419120071e-05,
      "loss": 1.623,
      "step": 3897
    },
    {
      "epoch": 0.5583727259704914,
      "grad_norm": 0.9782788753509521,
      "learning_rate": 8.601071116810143e-05,
      "loss": 1.2369,
      "step": 3898
    },
    {
      "epoch": 0.5585159719237931,
      "grad_norm": 1.1443662643432617,
      "learning_rate": 8.596477115654985e-05,
      "loss": 1.4317,
      "step": 3899
    },
    {
      "epoch": 0.5586592178770949,
      "grad_norm": 1.0493360757827759,
      "learning_rate": 8.591883416643574e-05,
      "loss": 1.4359,
      "step": 3900
    },
    {
      "epoch": 0.5588024638303968,
      "grad_norm": 0.9904102087020874,
      "learning_rate": 8.587290020764814e-05,
      "loss": 1.467,
      "step": 3901
    },
    {
      "epoch": 0.5589457097836986,
      "grad_norm": 1.177286148071289,
      "learning_rate": 8.582696929007556e-05,
      "loss": 1.6094,
      "step": 3902
    },
    {
      "epoch": 0.5590889557370005,
      "grad_norm": 1.3895848989486694,
      "learning_rate": 8.578104142360578e-05,
      "loss": 1.2424,
      "step": 3903
    },
    {
      "epoch": 0.5592322016903023,
      "grad_norm": 1.1811342239379883,
      "learning_rate": 8.573511661812589e-05,
      "loss": 1.4958,
      "step": 3904
    },
    {
      "epoch": 0.559375447643604,
      "grad_norm": 1.180285930633545,
      "learning_rate": 8.568919488352242e-05,
      "loss": 1.3743,
      "step": 3905
    },
    {
      "epoch": 0.5595186935969059,
      "grad_norm": 1.1766390800476074,
      "learning_rate": 8.564327622968113e-05,
      "loss": 1.5588,
      "step": 3906
    },
    {
      "epoch": 0.5596619395502077,
      "grad_norm": 1.0139480829238892,
      "learning_rate": 8.559736066648717e-05,
      "loss": 1.4818,
      "step": 3907
    },
    {
      "epoch": 0.5598051855035096,
      "grad_norm": 1.019096851348877,
      "learning_rate": 8.55514482038251e-05,
      "loss": 1.4169,
      "step": 3908
    },
    {
      "epoch": 0.5599484314568114,
      "grad_norm": 1.0620943307876587,
      "learning_rate": 8.550553885157866e-05,
      "loss": 1.3312,
      "step": 3909
    },
    {
      "epoch": 0.5600916774101131,
      "grad_norm": 1.053501009941101,
      "learning_rate": 8.545963261963102e-05,
      "loss": 1.5649,
      "step": 3910
    },
    {
      "epoch": 0.560234923363415,
      "grad_norm": 0.9656388759613037,
      "learning_rate": 8.541372951786469e-05,
      "loss": 1.4338,
      "step": 3911
    },
    {
      "epoch": 0.5603781693167168,
      "grad_norm": 0.9396020770072937,
      "learning_rate": 8.536782955616138e-05,
      "loss": 1.3519,
      "step": 3912
    },
    {
      "epoch": 0.5605214152700186,
      "grad_norm": 0.9551534056663513,
      "learning_rate": 8.532193274440235e-05,
      "loss": 1.52,
      "step": 3913
    },
    {
      "epoch": 0.5606646612233205,
      "grad_norm": 1.2763575315475464,
      "learning_rate": 8.527603909246791e-05,
      "loss": 1.2783,
      "step": 3914
    },
    {
      "epoch": 0.5608079071766222,
      "grad_norm": 1.0461019277572632,
      "learning_rate": 8.523014861023791e-05,
      "loss": 1.3851,
      "step": 3915
    },
    {
      "epoch": 0.560951153129924,
      "grad_norm": 1.0487090349197388,
      "learning_rate": 8.518426130759145e-05,
      "loss": 1.3792,
      "step": 3916
    },
    {
      "epoch": 0.5610943990832259,
      "grad_norm": 1.0350587368011475,
      "learning_rate": 8.513837719440688e-05,
      "loss": 1.3428,
      "step": 3917
    },
    {
      "epoch": 0.5612376450365277,
      "grad_norm": 0.9404356479644775,
      "learning_rate": 8.509249628056192e-05,
      "loss": 1.3942,
      "step": 3918
    },
    {
      "epoch": 0.5613808909898296,
      "grad_norm": 1.0931662321090698,
      "learning_rate": 8.504661857593358e-05,
      "loss": 1.4072,
      "step": 3919
    },
    {
      "epoch": 0.5615241369431313,
      "grad_norm": 1.213301181793213,
      "learning_rate": 8.500074409039827e-05,
      "loss": 1.4583,
      "step": 3920
    },
    {
      "epoch": 0.5616673828964331,
      "grad_norm": 0.9574490785598755,
      "learning_rate": 8.495487283383158e-05,
      "loss": 1.4549,
      "step": 3921
    },
    {
      "epoch": 0.561810628849735,
      "grad_norm": 1.125637412071228,
      "learning_rate": 8.490900481610845e-05,
      "loss": 1.3733,
      "step": 3922
    },
    {
      "epoch": 0.5619538748030368,
      "grad_norm": 1.0171054601669312,
      "learning_rate": 8.486314004710311e-05,
      "loss": 1.3631,
      "step": 3923
    },
    {
      "epoch": 0.5620971207563387,
      "grad_norm": 1.153102159500122,
      "learning_rate": 8.48172785366892e-05,
      "loss": 1.4226,
      "step": 3924
    },
    {
      "epoch": 0.5622403667096405,
      "grad_norm": 1.125910758972168,
      "learning_rate": 8.477142029473945e-05,
      "loss": 1.3307,
      "step": 3925
    },
    {
      "epoch": 0.5623836126629422,
      "grad_norm": 1.0122894048690796,
      "learning_rate": 8.47255653311261e-05,
      "loss": 1.5344,
      "step": 3926
    },
    {
      "epoch": 0.5625268586162441,
      "grad_norm": 1.0655895471572876,
      "learning_rate": 8.467971365572053e-05,
      "loss": 1.4975,
      "step": 3927
    },
    {
      "epoch": 0.5626701045695459,
      "grad_norm": 1.19464111328125,
      "learning_rate": 8.463386527839349e-05,
      "loss": 1.3688,
      "step": 3928
    },
    {
      "epoch": 0.5628133505228478,
      "grad_norm": 1.0846012830734253,
      "learning_rate": 8.458802020901503e-05,
      "loss": 1.34,
      "step": 3929
    },
    {
      "epoch": 0.5629565964761496,
      "grad_norm": 1.254239797592163,
      "learning_rate": 8.454217845745439e-05,
      "loss": 1.5473,
      "step": 3930
    },
    {
      "epoch": 0.5630998424294513,
      "grad_norm": 1.2421461343765259,
      "learning_rate": 8.449634003358022e-05,
      "loss": 1.3548,
      "step": 3931
    },
    {
      "epoch": 0.5632430883827532,
      "grad_norm": 1.0915660858154297,
      "learning_rate": 8.44505049472604e-05,
      "loss": 1.4373,
      "step": 3932
    },
    {
      "epoch": 0.563386334336055,
      "grad_norm": 1.2855892181396484,
      "learning_rate": 8.440467320836208e-05,
      "loss": 1.4674,
      "step": 3933
    },
    {
      "epoch": 0.5635295802893568,
      "grad_norm": 0.9798267483711243,
      "learning_rate": 8.435884482675168e-05,
      "loss": 1.382,
      "step": 3934
    },
    {
      "epoch": 0.5636728262426587,
      "grad_norm": 1.1581510305404663,
      "learning_rate": 8.431301981229492e-05,
      "loss": 1.3357,
      "step": 3935
    },
    {
      "epoch": 0.5638160721959604,
      "grad_norm": 1.156049132347107,
      "learning_rate": 8.42671981748568e-05,
      "loss": 1.3556,
      "step": 3936
    },
    {
      "epoch": 0.5639593181492623,
      "grad_norm": 0.9260801076889038,
      "learning_rate": 8.422137992430162e-05,
      "loss": 1.3248,
      "step": 3937
    },
    {
      "epoch": 0.5641025641025641,
      "grad_norm": 1.0331379175186157,
      "learning_rate": 8.417556507049285e-05,
      "loss": 1.3735,
      "step": 3938
    },
    {
      "epoch": 0.5642458100558659,
      "grad_norm": 1.1090047359466553,
      "learning_rate": 8.412975362329332e-05,
      "loss": 1.3409,
      "step": 3939
    },
    {
      "epoch": 0.5643890560091678,
      "grad_norm": 1.098628282546997,
      "learning_rate": 8.408394559256516e-05,
      "loss": 1.3075,
      "step": 3940
    },
    {
      "epoch": 0.5645323019624695,
      "grad_norm": 1.0496476888656616,
      "learning_rate": 8.40381409881696e-05,
      "loss": 1.4826,
      "step": 3941
    },
    {
      "epoch": 0.5646755479157713,
      "grad_norm": 0.9968166947364807,
      "learning_rate": 8.399233981996731e-05,
      "loss": 1.4571,
      "step": 3942
    },
    {
      "epoch": 0.5648187938690732,
      "grad_norm": 1.21078622341156,
      "learning_rate": 8.394654209781813e-05,
      "loss": 1.4344,
      "step": 3943
    },
    {
      "epoch": 0.564962039822375,
      "grad_norm": 1.0950098037719727,
      "learning_rate": 8.390074783158116e-05,
      "loss": 1.4021,
      "step": 3944
    },
    {
      "epoch": 0.5651052857756769,
      "grad_norm": 0.9033725261688232,
      "learning_rate": 8.385495703111483e-05,
      "loss": 1.268,
      "step": 3945
    },
    {
      "epoch": 0.5652485317289787,
      "grad_norm": 1.0935604572296143,
      "learning_rate": 8.380916970627666e-05,
      "loss": 1.4423,
      "step": 3946
    },
    {
      "epoch": 0.5653917776822804,
      "grad_norm": 0.9992495775222778,
      "learning_rate": 8.376338586692366e-05,
      "loss": 1.3864,
      "step": 3947
    },
    {
      "epoch": 0.5655350236355823,
      "grad_norm": 1.1999882459640503,
      "learning_rate": 8.371760552291183e-05,
      "loss": 1.4,
      "step": 3948
    },
    {
      "epoch": 0.5656782695888841,
      "grad_norm": 1.177969217300415,
      "learning_rate": 8.367182868409662e-05,
      "loss": 1.3745,
      "step": 3949
    },
    {
      "epoch": 0.565821515542186,
      "grad_norm": 1.396816611289978,
      "learning_rate": 8.362605536033265e-05,
      "loss": 1.4306,
      "step": 3950
    },
    {
      "epoch": 0.5659647614954878,
      "grad_norm": 1.2220689058303833,
      "learning_rate": 8.358028556147377e-05,
      "loss": 1.2499,
      "step": 3951
    },
    {
      "epoch": 0.5661080074487895,
      "grad_norm": 1.132494330406189,
      "learning_rate": 8.353451929737304e-05,
      "loss": 1.4422,
      "step": 3952
    },
    {
      "epoch": 0.5662512534020914,
      "grad_norm": 1.2047371864318848,
      "learning_rate": 8.348875657788291e-05,
      "loss": 1.3615,
      "step": 3953
    },
    {
      "epoch": 0.5663944993553932,
      "grad_norm": 1.0638824701309204,
      "learning_rate": 8.344299741285486e-05,
      "loss": 1.3023,
      "step": 3954
    },
    {
      "epoch": 0.566537745308695,
      "grad_norm": 1.0057721138000488,
      "learning_rate": 8.339724181213977e-05,
      "loss": 1.3085,
      "step": 3955
    },
    {
      "epoch": 0.5666809912619969,
      "grad_norm": 1.2296563386917114,
      "learning_rate": 8.335148978558764e-05,
      "loss": 1.4909,
      "step": 3956
    },
    {
      "epoch": 0.5668242372152986,
      "grad_norm": 1.0066629648208618,
      "learning_rate": 8.330574134304776e-05,
      "loss": 1.3211,
      "step": 3957
    },
    {
      "epoch": 0.5669674831686005,
      "grad_norm": 1.301001787185669,
      "learning_rate": 8.32599964943687e-05,
      "loss": 1.2835,
      "step": 3958
    },
    {
      "epoch": 0.5671107291219023,
      "grad_norm": 1.3211230039596558,
      "learning_rate": 8.32142552493981e-05,
      "loss": 1.4301,
      "step": 3959
    },
    {
      "epoch": 0.5672539750752041,
      "grad_norm": 1.1973659992218018,
      "learning_rate": 8.316851761798298e-05,
      "loss": 1.4096,
      "step": 3960
    },
    {
      "epoch": 0.567397221028506,
      "grad_norm": 1.138844609260559,
      "learning_rate": 8.312278360996952e-05,
      "loss": 1.3388,
      "step": 3961
    },
    {
      "epoch": 0.5675404669818077,
      "grad_norm": 1.3326082229614258,
      "learning_rate": 8.307705323520305e-05,
      "loss": 1.4922,
      "step": 3962
    },
    {
      "epoch": 0.5676837129351096,
      "grad_norm": 1.052769422531128,
      "learning_rate": 8.303132650352831e-05,
      "loss": 1.2773,
      "step": 3963
    },
    {
      "epoch": 0.5678269588884114,
      "grad_norm": 1.1518619060516357,
      "learning_rate": 8.298560342478901e-05,
      "loss": 1.5905,
      "step": 3964
    },
    {
      "epoch": 0.5679702048417132,
      "grad_norm": 1.367877721786499,
      "learning_rate": 8.293988400882826e-05,
      "loss": 1.4601,
      "step": 3965
    },
    {
      "epoch": 0.5681134507950151,
      "grad_norm": 1.0225945711135864,
      "learning_rate": 8.289416826548834e-05,
      "loss": 1.487,
      "step": 3966
    },
    {
      "epoch": 0.5682566967483169,
      "grad_norm": 1.0794260501861572,
      "learning_rate": 8.284845620461067e-05,
      "loss": 1.6319,
      "step": 3967
    },
    {
      "epoch": 0.5683999427016186,
      "grad_norm": 1.0988457202911377,
      "learning_rate": 8.28027478360359e-05,
      "loss": 1.3488,
      "step": 3968
    },
    {
      "epoch": 0.5685431886549205,
      "grad_norm": 1.0937910079956055,
      "learning_rate": 8.275704316960403e-05,
      "loss": 1.1671,
      "step": 3969
    },
    {
      "epoch": 0.5686864346082223,
      "grad_norm": 1.015166997909546,
      "learning_rate": 8.271134221515403e-05,
      "loss": 1.3646,
      "step": 3970
    },
    {
      "epoch": 0.5688296805615242,
      "grad_norm": 1.0831773281097412,
      "learning_rate": 8.266564498252425e-05,
      "loss": 1.6984,
      "step": 3971
    },
    {
      "epoch": 0.568972926514826,
      "grad_norm": 1.0808922052383423,
      "learning_rate": 8.261995148155213e-05,
      "loss": 1.5173,
      "step": 3972
    },
    {
      "epoch": 0.5691161724681277,
      "grad_norm": 1.053452730178833,
      "learning_rate": 8.257426172207434e-05,
      "loss": 1.3327,
      "step": 3973
    },
    {
      "epoch": 0.5692594184214296,
      "grad_norm": 1.1618472337722778,
      "learning_rate": 8.252857571392683e-05,
      "loss": 1.5115,
      "step": 3974
    },
    {
      "epoch": 0.5694026643747314,
      "grad_norm": 1.021773099899292,
      "learning_rate": 8.248289346694457e-05,
      "loss": 1.5194,
      "step": 3975
    },
    {
      "epoch": 0.5695459103280333,
      "grad_norm": 1.0760945081710815,
      "learning_rate": 8.24372149909619e-05,
      "loss": 1.4145,
      "step": 3976
    },
    {
      "epoch": 0.5696891562813351,
      "grad_norm": 1.0001275539398193,
      "learning_rate": 8.239154029581222e-05,
      "loss": 1.383,
      "step": 3977
    },
    {
      "epoch": 0.5698324022346368,
      "grad_norm": 1.0532678365707397,
      "learning_rate": 8.234586939132817e-05,
      "loss": 1.3821,
      "step": 3978
    },
    {
      "epoch": 0.5699756481879387,
      "grad_norm": 0.9672591090202332,
      "learning_rate": 8.230020228734159e-05,
      "loss": 1.5348,
      "step": 3979
    },
    {
      "epoch": 0.5701188941412405,
      "grad_norm": 1.1478209495544434,
      "learning_rate": 8.22545389936834e-05,
      "loss": 1.4502,
      "step": 3980
    },
    {
      "epoch": 0.5702621400945423,
      "grad_norm": 1.231196641921997,
      "learning_rate": 8.220887952018387e-05,
      "loss": 1.421,
      "step": 3981
    },
    {
      "epoch": 0.5704053860478442,
      "grad_norm": 1.0397428274154663,
      "learning_rate": 8.216322387667232e-05,
      "loss": 1.3013,
      "step": 3982
    },
    {
      "epoch": 0.5705486320011459,
      "grad_norm": 1.0203744173049927,
      "learning_rate": 8.211757207297727e-05,
      "loss": 1.6783,
      "step": 3983
    },
    {
      "epoch": 0.5706918779544478,
      "grad_norm": 1.0634477138519287,
      "learning_rate": 8.207192411892646e-05,
      "loss": 1.3888,
      "step": 3984
    },
    {
      "epoch": 0.5708351239077496,
      "grad_norm": 0.9715363383293152,
      "learning_rate": 8.202628002434672e-05,
      "loss": 1.5194,
      "step": 3985
    },
    {
      "epoch": 0.5709783698610514,
      "grad_norm": 1.0375038385391235,
      "learning_rate": 8.198063979906407e-05,
      "loss": 1.4894,
      "step": 3986
    },
    {
      "epoch": 0.5711216158143533,
      "grad_norm": 1.3240010738372803,
      "learning_rate": 8.19350034529038e-05,
      "loss": 1.3202,
      "step": 3987
    },
    {
      "epoch": 0.5712648617676551,
      "grad_norm": 1.0191293954849243,
      "learning_rate": 8.188937099569024e-05,
      "loss": 1.354,
      "step": 3988
    },
    {
      "epoch": 0.5714081077209568,
      "grad_norm": 1.102787733078003,
      "learning_rate": 8.184374243724693e-05,
      "loss": 1.3427,
      "step": 3989
    },
    {
      "epoch": 0.5715513536742587,
      "grad_norm": 1.0335856676101685,
      "learning_rate": 8.17981177873966e-05,
      "loss": 1.4533,
      "step": 3990
    },
    {
      "epoch": 0.5716945996275605,
      "grad_norm": 1.0352792739868164,
      "learning_rate": 8.175249705596105e-05,
      "loss": 1.3224,
      "step": 3991
    },
    {
      "epoch": 0.5718378455808624,
      "grad_norm": 1.1521683931350708,
      "learning_rate": 8.170688025276134e-05,
      "loss": 1.3636,
      "step": 3992
    },
    {
      "epoch": 0.5719810915341642,
      "grad_norm": 1.0134963989257812,
      "learning_rate": 8.166126738761764e-05,
      "loss": 1.4362,
      "step": 3993
    },
    {
      "epoch": 0.5721243374874659,
      "grad_norm": 1.052530288696289,
      "learning_rate": 8.161565847034925e-05,
      "loss": 1.3769,
      "step": 3994
    },
    {
      "epoch": 0.5722675834407678,
      "grad_norm": 1.054436206817627,
      "learning_rate": 8.157005351077465e-05,
      "loss": 1.3641,
      "step": 3995
    },
    {
      "epoch": 0.5724108293940696,
      "grad_norm": 0.9436133503913879,
      "learning_rate": 8.152445251871144e-05,
      "loss": 1.4507,
      "step": 3996
    },
    {
      "epoch": 0.5725540753473715,
      "grad_norm": 1.3076878786087036,
      "learning_rate": 8.147885550397645e-05,
      "loss": 1.4484,
      "step": 3997
    },
    {
      "epoch": 0.5726973213006733,
      "grad_norm": 1.1398749351501465,
      "learning_rate": 8.143326247638549e-05,
      "loss": 1.4209,
      "step": 3998
    },
    {
      "epoch": 0.572840567253975,
      "grad_norm": 1.0094306468963623,
      "learning_rate": 8.138767344575368e-05,
      "loss": 1.5894,
      "step": 3999
    },
    {
      "epoch": 0.5729838132072769,
      "grad_norm": 1.1019415855407715,
      "learning_rate": 8.134208842189522e-05,
      "loss": 1.4945,
      "step": 4000
    },
    {
      "epoch": 0.5731270591605787,
      "grad_norm": 1.3345929384231567,
      "learning_rate": 8.12965074146234e-05,
      "loss": 1.4221,
      "step": 4001
    },
    {
      "epoch": 0.5732703051138806,
      "grad_norm": 1.0852625370025635,
      "learning_rate": 8.125093043375064e-05,
      "loss": 1.3058,
      "step": 4002
    },
    {
      "epoch": 0.5734135510671824,
      "grad_norm": 1.083067774772644,
      "learning_rate": 8.120535748908866e-05,
      "loss": 1.504,
      "step": 4003
    },
    {
      "epoch": 0.5735567970204841,
      "grad_norm": 1.0089505910873413,
      "learning_rate": 8.11597885904481e-05,
      "loss": 1.3618,
      "step": 4004
    },
    {
      "epoch": 0.573700042973786,
      "grad_norm": 1.242389440536499,
      "learning_rate": 8.111422374763883e-05,
      "loss": 1.5587,
      "step": 4005
    },
    {
      "epoch": 0.5738432889270878,
      "grad_norm": 1.1502662897109985,
      "learning_rate": 8.106866297046983e-05,
      "loss": 1.4469,
      "step": 4006
    },
    {
      "epoch": 0.5739865348803896,
      "grad_norm": 1.0065104961395264,
      "learning_rate": 8.10231062687492e-05,
      "loss": 1.4672,
      "step": 4007
    },
    {
      "epoch": 0.5741297808336915,
      "grad_norm": 1.0659387111663818,
      "learning_rate": 8.097755365228425e-05,
      "loss": 1.4156,
      "step": 4008
    },
    {
      "epoch": 0.5742730267869932,
      "grad_norm": 1.1952720880508423,
      "learning_rate": 8.093200513088118e-05,
      "loss": 1.492,
      "step": 4009
    },
    {
      "epoch": 0.574416272740295,
      "grad_norm": 1.0429340600967407,
      "learning_rate": 8.088646071434559e-05,
      "loss": 1.4798,
      "step": 4010
    },
    {
      "epoch": 0.5745595186935969,
      "grad_norm": 1.0296752452850342,
      "learning_rate": 8.084092041248203e-05,
      "loss": 1.2927,
      "step": 4011
    },
    {
      "epoch": 0.5747027646468987,
      "grad_norm": 0.9452112913131714,
      "learning_rate": 8.079538423509417e-05,
      "loss": 1.7685,
      "step": 4012
    },
    {
      "epoch": 0.5748460106002006,
      "grad_norm": 1.1228902339935303,
      "learning_rate": 8.074985219198488e-05,
      "loss": 1.4869,
      "step": 4013
    },
    {
      "epoch": 0.5749892565535024,
      "grad_norm": 1.2515939474105835,
      "learning_rate": 8.070432429295599e-05,
      "loss": 1.3411,
      "step": 4014
    },
    {
      "epoch": 0.5751325025068041,
      "grad_norm": 1.3844817876815796,
      "learning_rate": 8.065880054780861e-05,
      "loss": 1.4273,
      "step": 4015
    },
    {
      "epoch": 0.575275748460106,
      "grad_norm": 1.2724149227142334,
      "learning_rate": 8.061328096634288e-05,
      "loss": 1.4378,
      "step": 4016
    },
    {
      "epoch": 0.5754189944134078,
      "grad_norm": 1.1060429811477661,
      "learning_rate": 8.056776555835798e-05,
      "loss": 1.3972,
      "step": 4017
    },
    {
      "epoch": 0.5755622403667097,
      "grad_norm": 1.1961658000946045,
      "learning_rate": 8.052225433365226e-05,
      "loss": 1.43,
      "step": 4018
    },
    {
      "epoch": 0.5757054863200115,
      "grad_norm": 1.1288669109344482,
      "learning_rate": 8.047674730202325e-05,
      "loss": 1.514,
      "step": 4019
    },
    {
      "epoch": 0.5758487322733132,
      "grad_norm": 1.0612808465957642,
      "learning_rate": 8.043124447326736e-05,
      "loss": 1.4434,
      "step": 4020
    },
    {
      "epoch": 0.5759919782266151,
      "grad_norm": 1.1025935411453247,
      "learning_rate": 8.038574585718032e-05,
      "loss": 1.278,
      "step": 4021
    },
    {
      "epoch": 0.5761352241799169,
      "grad_norm": 1.148199439048767,
      "learning_rate": 8.03402514635568e-05,
      "loss": 1.5157,
      "step": 4022
    },
    {
      "epoch": 0.5762784701332188,
      "grad_norm": 1.1931822299957275,
      "learning_rate": 8.029476130219064e-05,
      "loss": 1.33,
      "step": 4023
    },
    {
      "epoch": 0.5764217160865206,
      "grad_norm": 1.1266332864761353,
      "learning_rate": 8.024927538287476e-05,
      "loss": 1.3397,
      "step": 4024
    },
    {
      "epoch": 0.5765649620398223,
      "grad_norm": 1.1183571815490723,
      "learning_rate": 8.02037937154011e-05,
      "loss": 1.4459,
      "step": 4025
    },
    {
      "epoch": 0.5767082079931242,
      "grad_norm": 1.1886775493621826,
      "learning_rate": 8.015831630956079e-05,
      "loss": 1.5075,
      "step": 4026
    },
    {
      "epoch": 0.576851453946426,
      "grad_norm": 1.073609709739685,
      "learning_rate": 8.011284317514398e-05,
      "loss": 1.3052,
      "step": 4027
    },
    {
      "epoch": 0.5769946998997278,
      "grad_norm": 1.1282099485397339,
      "learning_rate": 8.006737432193989e-05,
      "loss": 1.4253,
      "step": 4028
    },
    {
      "epoch": 0.5771379458530297,
      "grad_norm": 1.2322652339935303,
      "learning_rate": 8.002190975973689e-05,
      "loss": 1.3272,
      "step": 4029
    },
    {
      "epoch": 0.5772811918063314,
      "grad_norm": 1.1895850896835327,
      "learning_rate": 7.997644949832228e-05,
      "loss": 1.4017,
      "step": 4030
    },
    {
      "epoch": 0.5774244377596333,
      "grad_norm": 1.0355268716812134,
      "learning_rate": 7.993099354748261e-05,
      "loss": 1.382,
      "step": 4031
    },
    {
      "epoch": 0.5775676837129351,
      "grad_norm": 1.2945213317871094,
      "learning_rate": 7.988554191700343e-05,
      "loss": 1.5764,
      "step": 4032
    },
    {
      "epoch": 0.5777109296662369,
      "grad_norm": 0.952515184879303,
      "learning_rate": 7.98400946166693e-05,
      "loss": 1.4734,
      "step": 4033
    },
    {
      "epoch": 0.5778541756195388,
      "grad_norm": 1.1861485242843628,
      "learning_rate": 7.979465165626392e-05,
      "loss": 1.3564,
      "step": 4034
    },
    {
      "epoch": 0.5779974215728406,
      "grad_norm": 0.9773125648498535,
      "learning_rate": 7.974921304557002e-05,
      "loss": 1.3533,
      "step": 4035
    },
    {
      "epoch": 0.5781406675261423,
      "grad_norm": 1.1982392072677612,
      "learning_rate": 7.970377879436941e-05,
      "loss": 1.3583,
      "step": 4036
    },
    {
      "epoch": 0.5782839134794442,
      "grad_norm": 1.1087453365325928,
      "learning_rate": 7.965834891244301e-05,
      "loss": 1.4546,
      "step": 4037
    },
    {
      "epoch": 0.578427159432746,
      "grad_norm": 1.1653337478637695,
      "learning_rate": 7.961292340957069e-05,
      "loss": 1.3689,
      "step": 4038
    },
    {
      "epoch": 0.5785704053860479,
      "grad_norm": 0.925621509552002,
      "learning_rate": 7.956750229553145e-05,
      "loss": 1.5919,
      "step": 4039
    },
    {
      "epoch": 0.5787136513393497,
      "grad_norm": 1.099250316619873,
      "learning_rate": 7.952208558010336e-05,
      "loss": 1.5213,
      "step": 4040
    },
    {
      "epoch": 0.5788568972926514,
      "grad_norm": 0.9960086941719055,
      "learning_rate": 7.947667327306348e-05,
      "loss": 1.4788,
      "step": 4041
    },
    {
      "epoch": 0.5790001432459533,
      "grad_norm": 0.9202391505241394,
      "learning_rate": 7.943126538418802e-05,
      "loss": 1.3521,
      "step": 4042
    },
    {
      "epoch": 0.5791433891992551,
      "grad_norm": 1.127137541770935,
      "learning_rate": 7.938586192325205e-05,
      "loss": 1.5062,
      "step": 4043
    },
    {
      "epoch": 0.579286635152557,
      "grad_norm": 1.044081211090088,
      "learning_rate": 7.934046290002991e-05,
      "loss": 1.4451,
      "step": 4044
    },
    {
      "epoch": 0.5794298811058588,
      "grad_norm": 1.0595406293869019,
      "learning_rate": 7.929506832429489e-05,
      "loss": 1.5446,
      "step": 4045
    },
    {
      "epoch": 0.5795731270591605,
      "grad_norm": 1.1960043907165527,
      "learning_rate": 7.924967820581928e-05,
      "loss": 1.4994,
      "step": 4046
    },
    {
      "epoch": 0.5797163730124624,
      "grad_norm": 0.968834638595581,
      "learning_rate": 7.920429255437447e-05,
      "loss": 1.4074,
      "step": 4047
    },
    {
      "epoch": 0.5798596189657642,
      "grad_norm": 1.2260991334915161,
      "learning_rate": 7.915891137973082e-05,
      "loss": 1.3711,
      "step": 4048
    },
    {
      "epoch": 0.580002864919066,
      "grad_norm": 1.111136794090271,
      "learning_rate": 7.911353469165782e-05,
      "loss": 1.4278,
      "step": 4049
    },
    {
      "epoch": 0.5801461108723679,
      "grad_norm": 0.9621102213859558,
      "learning_rate": 7.9068162499924e-05,
      "loss": 1.5178,
      "step": 4050
    },
    {
      "epoch": 0.5802893568256696,
      "grad_norm": 1.1072837114334106,
      "learning_rate": 7.902279481429675e-05,
      "loss": 1.3059,
      "step": 4051
    },
    {
      "epoch": 0.5804326027789715,
      "grad_norm": 1.0482300519943237,
      "learning_rate": 7.897743164454267e-05,
      "loss": 1.435,
      "step": 4052
    },
    {
      "epoch": 0.5805758487322733,
      "grad_norm": 1.0429731607437134,
      "learning_rate": 7.89320730004274e-05,
      "loss": 1.428,
      "step": 4053
    },
    {
      "epoch": 0.5807190946855751,
      "grad_norm": 1.0610629320144653,
      "learning_rate": 7.888671889171538e-05,
      "loss": 1.6398,
      "step": 4054
    },
    {
      "epoch": 0.580862340638877,
      "grad_norm": 1.0603666305541992,
      "learning_rate": 7.884136932817037e-05,
      "loss": 1.5053,
      "step": 4055
    },
    {
      "epoch": 0.5810055865921788,
      "grad_norm": 1.2090955972671509,
      "learning_rate": 7.879602431955492e-05,
      "loss": 1.3053,
      "step": 4056
    },
    {
      "epoch": 0.5811488325454806,
      "grad_norm": 1.138738989830017,
      "learning_rate": 7.875068387563069e-05,
      "loss": 1.2628,
      "step": 4057
    },
    {
      "epoch": 0.5812920784987824,
      "grad_norm": 1.163888692855835,
      "learning_rate": 7.870534800615845e-05,
      "loss": 1.2563,
      "step": 4058
    },
    {
      "epoch": 0.5814353244520842,
      "grad_norm": 1.1937397718429565,
      "learning_rate": 7.866001672089776e-05,
      "loss": 1.5258,
      "step": 4059
    },
    {
      "epoch": 0.5815785704053861,
      "grad_norm": 1.0095622539520264,
      "learning_rate": 7.861469002960742e-05,
      "loss": 1.3591,
      "step": 4060
    },
    {
      "epoch": 0.5817218163586879,
      "grad_norm": 1.1307746171951294,
      "learning_rate": 7.856936794204513e-05,
      "loss": 1.4926,
      "step": 4061
    },
    {
      "epoch": 0.5818650623119896,
      "grad_norm": 1.050833821296692,
      "learning_rate": 7.852405046796756e-05,
      "loss": 1.51,
      "step": 4062
    },
    {
      "epoch": 0.5820083082652915,
      "grad_norm": 1.0171937942504883,
      "learning_rate": 7.84787376171305e-05,
      "loss": 1.3694,
      "step": 4063
    },
    {
      "epoch": 0.5821515542185933,
      "grad_norm": 1.167988896369934,
      "learning_rate": 7.843342939928864e-05,
      "loss": 1.4884,
      "step": 4064
    },
    {
      "epoch": 0.5822948001718952,
      "grad_norm": 1.1452760696411133,
      "learning_rate": 7.838812582419574e-05,
      "loss": 1.4827,
      "step": 4065
    },
    {
      "epoch": 0.582438046125197,
      "grad_norm": 1.161697268486023,
      "learning_rate": 7.834282690160457e-05,
      "loss": 1.3815,
      "step": 4066
    },
    {
      "epoch": 0.5825812920784987,
      "grad_norm": 0.9736148118972778,
      "learning_rate": 7.829753264126681e-05,
      "loss": 1.3525,
      "step": 4067
    },
    {
      "epoch": 0.5827245380318006,
      "grad_norm": 1.1203678846359253,
      "learning_rate": 7.82522430529332e-05,
      "loss": 1.4147,
      "step": 4068
    },
    {
      "epoch": 0.5828677839851024,
      "grad_norm": 1.2531877756118774,
      "learning_rate": 7.820695814635356e-05,
      "loss": 1.4253,
      "step": 4069
    },
    {
      "epoch": 0.5830110299384043,
      "grad_norm": 1.0416511297225952,
      "learning_rate": 7.816167793127646e-05,
      "loss": 1.4336,
      "step": 4070
    },
    {
      "epoch": 0.5831542758917061,
      "grad_norm": 0.962336003780365,
      "learning_rate": 7.811640241744975e-05,
      "loss": 1.2869,
      "step": 4071
    },
    {
      "epoch": 0.5832975218450078,
      "grad_norm": 1.1722517013549805,
      "learning_rate": 7.807113161462003e-05,
      "loss": 1.3873,
      "step": 4072
    },
    {
      "epoch": 0.5834407677983097,
      "grad_norm": 0.9998098611831665,
      "learning_rate": 7.802586553253301e-05,
      "loss": 1.4721,
      "step": 4073
    },
    {
      "epoch": 0.5835840137516115,
      "grad_norm": 0.999030351638794,
      "learning_rate": 7.79806041809334e-05,
      "loss": 1.4494,
      "step": 4074
    },
    {
      "epoch": 0.5837272597049133,
      "grad_norm": 1.364902138710022,
      "learning_rate": 7.793534756956477e-05,
      "loss": 1.4415,
      "step": 4075
    },
    {
      "epoch": 0.5838705056582152,
      "grad_norm": 1.0513964891433716,
      "learning_rate": 7.789009570816985e-05,
      "loss": 1.5833,
      "step": 4076
    },
    {
      "epoch": 0.584013751611517,
      "grad_norm": 1.21407949924469,
      "learning_rate": 7.784484860649013e-05,
      "loss": 1.4014,
      "step": 4077
    },
    {
      "epoch": 0.5841569975648188,
      "grad_norm": 1.1745800971984863,
      "learning_rate": 7.779960627426627e-05,
      "loss": 1.5007,
      "step": 4078
    },
    {
      "epoch": 0.5843002435181206,
      "grad_norm": 1.0718715190887451,
      "learning_rate": 7.775436872123783e-05,
      "loss": 1.5661,
      "step": 4079
    },
    {
      "epoch": 0.5844434894714224,
      "grad_norm": 0.9887749552726746,
      "learning_rate": 7.770913595714327e-05,
      "loss": 1.3546,
      "step": 4080
    },
    {
      "epoch": 0.5845867354247243,
      "grad_norm": 1.1965467929840088,
      "learning_rate": 7.766390799172012e-05,
      "loss": 1.3469,
      "step": 4081
    },
    {
      "epoch": 0.5847299813780261,
      "grad_norm": 1.1297179460525513,
      "learning_rate": 7.76186848347049e-05,
      "loss": 1.2296,
      "step": 4082
    },
    {
      "epoch": 0.5848732273313278,
      "grad_norm": 1.093031406402588,
      "learning_rate": 7.757346649583294e-05,
      "loss": 1.5112,
      "step": 4083
    },
    {
      "epoch": 0.5850164732846297,
      "grad_norm": 1.2159183025360107,
      "learning_rate": 7.75282529848387e-05,
      "loss": 1.324,
      "step": 4084
    },
    {
      "epoch": 0.5851597192379315,
      "grad_norm": 0.9650058746337891,
      "learning_rate": 7.74830443114555e-05,
      "loss": 1.5063,
      "step": 4085
    },
    {
      "epoch": 0.5853029651912334,
      "grad_norm": 1.1315046548843384,
      "learning_rate": 7.743784048541561e-05,
      "loss": 1.2774,
      "step": 4086
    },
    {
      "epoch": 0.5854462111445352,
      "grad_norm": 1.0977623462677002,
      "learning_rate": 7.739264151645039e-05,
      "loss": 1.4178,
      "step": 4087
    },
    {
      "epoch": 0.5855894570978369,
      "grad_norm": 1.045953392982483,
      "learning_rate": 7.734744741428997e-05,
      "loss": 1.4494,
      "step": 4088
    },
    {
      "epoch": 0.5857327030511388,
      "grad_norm": 1.086333990097046,
      "learning_rate": 7.730225818866358e-05,
      "loss": 1.435,
      "step": 4089
    },
    {
      "epoch": 0.5858759490044406,
      "grad_norm": 1.2137300968170166,
      "learning_rate": 7.725707384929932e-05,
      "loss": 1.4743,
      "step": 4090
    },
    {
      "epoch": 0.5860191949577425,
      "grad_norm": 0.9116840362548828,
      "learning_rate": 7.721189440592423e-05,
      "loss": 1.4257,
      "step": 4091
    },
    {
      "epoch": 0.5861624409110443,
      "grad_norm": 0.9345767498016357,
      "learning_rate": 7.716671986826443e-05,
      "loss": 1.3673,
      "step": 4092
    },
    {
      "epoch": 0.586305686864346,
      "grad_norm": 1.1537500619888306,
      "learning_rate": 7.712155024604473e-05,
      "loss": 1.4915,
      "step": 4093
    },
    {
      "epoch": 0.5864489328176479,
      "grad_norm": 0.9780703783035278,
      "learning_rate": 7.707638554898913e-05,
      "loss": 1.4099,
      "step": 4094
    },
    {
      "epoch": 0.5865921787709497,
      "grad_norm": 1.0122650861740112,
      "learning_rate": 7.703122578682046e-05,
      "loss": 1.4124,
      "step": 4095
    },
    {
      "epoch": 0.5867354247242516,
      "grad_norm": 1.1379693746566772,
      "learning_rate": 7.698607096926048e-05,
      "loss": 1.4942,
      "step": 4096
    },
    {
      "epoch": 0.5868786706775534,
      "grad_norm": 0.9861220121383667,
      "learning_rate": 7.69409211060299e-05,
      "loss": 1.442,
      "step": 4097
    },
    {
      "epoch": 0.5870219166308551,
      "grad_norm": 1.0483198165893555,
      "learning_rate": 7.689577620684842e-05,
      "loss": 1.5484,
      "step": 4098
    },
    {
      "epoch": 0.587165162584157,
      "grad_norm": 0.9657735824584961,
      "learning_rate": 7.685063628143455e-05,
      "loss": 1.4297,
      "step": 4099
    },
    {
      "epoch": 0.5873084085374588,
      "grad_norm": 1.0831466913223267,
      "learning_rate": 7.680550133950586e-05,
      "loss": 1.4629,
      "step": 4100
    },
    {
      "epoch": 0.5874516544907606,
      "grad_norm": 0.9783119559288025,
      "learning_rate": 7.676037139077874e-05,
      "loss": 1.4518,
      "step": 4101
    },
    {
      "epoch": 0.5875949004440625,
      "grad_norm": 0.9771605134010315,
      "learning_rate": 7.671524644496853e-05,
      "loss": 1.298,
      "step": 4102
    },
    {
      "epoch": 0.5877381463973643,
      "grad_norm": 1.1136901378631592,
      "learning_rate": 7.667012651178963e-05,
      "loss": 1.4304,
      "step": 4103
    },
    {
      "epoch": 0.587881392350666,
      "grad_norm": 1.159201741218567,
      "learning_rate": 7.662501160095509e-05,
      "loss": 1.5095,
      "step": 4104
    },
    {
      "epoch": 0.5880246383039679,
      "grad_norm": 0.9954947233200073,
      "learning_rate": 7.657990172217718e-05,
      "loss": 1.4314,
      "step": 4105
    },
    {
      "epoch": 0.5881678842572697,
      "grad_norm": 1.3003029823303223,
      "learning_rate": 7.653479688516683e-05,
      "loss": 1.3413,
      "step": 4106
    },
    {
      "epoch": 0.5883111302105716,
      "grad_norm": 1.0706342458724976,
      "learning_rate": 7.648969709963405e-05,
      "loss": 1.4444,
      "step": 4107
    },
    {
      "epoch": 0.5884543761638734,
      "grad_norm": 1.2283451557159424,
      "learning_rate": 7.644460237528771e-05,
      "loss": 1.379,
      "step": 4108
    },
    {
      "epoch": 0.5885976221171751,
      "grad_norm": 1.0265097618103027,
      "learning_rate": 7.639951272183551e-05,
      "loss": 1.4598,
      "step": 4109
    },
    {
      "epoch": 0.588740868070477,
      "grad_norm": 1.102243185043335,
      "learning_rate": 7.635442814898426e-05,
      "loss": 1.4984,
      "step": 4110
    },
    {
      "epoch": 0.5888841140237788,
      "grad_norm": 1.187455177307129,
      "learning_rate": 7.630934866643949e-05,
      "loss": 1.3111,
      "step": 4111
    },
    {
      "epoch": 0.5890273599770807,
      "grad_norm": 1.2127901315689087,
      "learning_rate": 7.626427428390567e-05,
      "loss": 1.3828,
      "step": 4112
    },
    {
      "epoch": 0.5891706059303825,
      "grad_norm": 1.2042181491851807,
      "learning_rate": 7.621920501108627e-05,
      "loss": 1.4411,
      "step": 4113
    },
    {
      "epoch": 0.5893138518836842,
      "grad_norm": 1.066982388496399,
      "learning_rate": 7.617414085768351e-05,
      "loss": 1.3189,
      "step": 4114
    },
    {
      "epoch": 0.5894570978369861,
      "grad_norm": 1.051342487335205,
      "learning_rate": 7.612908183339862e-05,
      "loss": 1.3499,
      "step": 4115
    },
    {
      "epoch": 0.5896003437902879,
      "grad_norm": 1.2656009197235107,
      "learning_rate": 7.608402794793174e-05,
      "loss": 1.4651,
      "step": 4116
    },
    {
      "epoch": 0.5897435897435898,
      "grad_norm": 0.9503814578056335,
      "learning_rate": 7.603897921098177e-05,
      "loss": 1.5977,
      "step": 4117
    },
    {
      "epoch": 0.5898868356968916,
      "grad_norm": 1.4044922590255737,
      "learning_rate": 7.599393563224666e-05,
      "loss": 1.5585,
      "step": 4118
    },
    {
      "epoch": 0.5900300816501933,
      "grad_norm": 0.955354630947113,
      "learning_rate": 7.594889722142316e-05,
      "loss": 1.4689,
      "step": 4119
    },
    {
      "epoch": 0.5901733276034952,
      "grad_norm": 1.0804277658462524,
      "learning_rate": 7.590386398820687e-05,
      "loss": 1.398,
      "step": 4120
    },
    {
      "epoch": 0.590316573556797,
      "grad_norm": 1.2909722328186035,
      "learning_rate": 7.585883594229243e-05,
      "loss": 1.4093,
      "step": 4121
    },
    {
      "epoch": 0.5904598195100988,
      "grad_norm": 1.1141750812530518,
      "learning_rate": 7.581381309337318e-05,
      "loss": 1.5269,
      "step": 4122
    },
    {
      "epoch": 0.5906030654634007,
      "grad_norm": 0.9842514395713806,
      "learning_rate": 7.576879545114145e-05,
      "loss": 1.4431,
      "step": 4123
    },
    {
      "epoch": 0.5907463114167025,
      "grad_norm": 1.1852399110794067,
      "learning_rate": 7.572378302528847e-05,
      "loss": 1.512,
      "step": 4124
    },
    {
      "epoch": 0.5908895573700043,
      "grad_norm": 1.1678013801574707,
      "learning_rate": 7.56787758255042e-05,
      "loss": 1.3506,
      "step": 4125
    },
    {
      "epoch": 0.5910328033233061,
      "grad_norm": 0.9724790453910828,
      "learning_rate": 7.56337738614777e-05,
      "loss": 1.4405,
      "step": 4126
    },
    {
      "epoch": 0.5911760492766079,
      "grad_norm": 1.127126693725586,
      "learning_rate": 7.558877714289667e-05,
      "loss": 1.3432,
      "step": 4127
    },
    {
      "epoch": 0.5913192952299098,
      "grad_norm": 0.8662317991256714,
      "learning_rate": 7.554378567944786e-05,
      "loss": 1.3631,
      "step": 4128
    },
    {
      "epoch": 0.5914625411832116,
      "grad_norm": 1.2070472240447998,
      "learning_rate": 7.549879948081681e-05,
      "loss": 1.6516,
      "step": 4129
    },
    {
      "epoch": 0.5916057871365134,
      "grad_norm": 1.0595580339431763,
      "learning_rate": 7.54538185566879e-05,
      "loss": 1.469,
      "step": 4130
    },
    {
      "epoch": 0.5917490330898152,
      "grad_norm": 0.9995733499526978,
      "learning_rate": 7.54088429167444e-05,
      "loss": 1.4044,
      "step": 4131
    },
    {
      "epoch": 0.591892279043117,
      "grad_norm": 1.281429648399353,
      "learning_rate": 7.536387257066854e-05,
      "loss": 1.2951,
      "step": 4132
    },
    {
      "epoch": 0.5920355249964189,
      "grad_norm": 1.0896403789520264,
      "learning_rate": 7.531890752814123e-05,
      "loss": 1.4317,
      "step": 4133
    },
    {
      "epoch": 0.5921787709497207,
      "grad_norm": 1.200994849205017,
      "learning_rate": 7.52739477988424e-05,
      "loss": 1.5888,
      "step": 4134
    },
    {
      "epoch": 0.5923220169030224,
      "grad_norm": 1.0673589706420898,
      "learning_rate": 7.52289933924507e-05,
      "loss": 1.4086,
      "step": 4135
    },
    {
      "epoch": 0.5924652628563243,
      "grad_norm": 1.1783102750778198,
      "learning_rate": 7.518404431864373e-05,
      "loss": 1.297,
      "step": 4136
    },
    {
      "epoch": 0.5926085088096261,
      "grad_norm": 0.9957106709480286,
      "learning_rate": 7.513910058709798e-05,
      "loss": 1.5496,
      "step": 4137
    },
    {
      "epoch": 0.592751754762928,
      "grad_norm": 1.2256033420562744,
      "learning_rate": 7.50941622074886e-05,
      "loss": 1.3981,
      "step": 4138
    },
    {
      "epoch": 0.5928950007162298,
      "grad_norm": 0.9200305342674255,
      "learning_rate": 7.504922918948978e-05,
      "loss": 1.4156,
      "step": 4139
    },
    {
      "epoch": 0.5930382466695315,
      "grad_norm": 1.1790897846221924,
      "learning_rate": 7.500430154277452e-05,
      "loss": 1.4273,
      "step": 4140
    },
    {
      "epoch": 0.5931814926228334,
      "grad_norm": 1.0630582571029663,
      "learning_rate": 7.495937927701455e-05,
      "loss": 1.3231,
      "step": 4141
    },
    {
      "epoch": 0.5933247385761352,
      "grad_norm": 1.1541399955749512,
      "learning_rate": 7.49144624018806e-05,
      "loss": 1.3764,
      "step": 4142
    },
    {
      "epoch": 0.593467984529437,
      "grad_norm": 1.188480257987976,
      "learning_rate": 7.486955092704207e-05,
      "loss": 1.495,
      "step": 4143
    },
    {
      "epoch": 0.5936112304827389,
      "grad_norm": 1.1780165433883667,
      "learning_rate": 7.482464486216737e-05,
      "loss": 1.262,
      "step": 4144
    },
    {
      "epoch": 0.5937544764360407,
      "grad_norm": 1.0949242115020752,
      "learning_rate": 7.477974421692367e-05,
      "loss": 1.3558,
      "step": 4145
    },
    {
      "epoch": 0.5938977223893425,
      "grad_norm": 1.138108730316162,
      "learning_rate": 7.47348490009769e-05,
      "loss": 1.3837,
      "step": 4146
    },
    {
      "epoch": 0.5940409683426443,
      "grad_norm": 0.995722770690918,
      "learning_rate": 7.468995922399189e-05,
      "loss": 1.5003,
      "step": 4147
    },
    {
      "epoch": 0.5941842142959461,
      "grad_norm": 1.2390016317367554,
      "learning_rate": 7.464507489563242e-05,
      "loss": 1.5789,
      "step": 4148
    },
    {
      "epoch": 0.594327460249248,
      "grad_norm": 1.035523772239685,
      "learning_rate": 7.460019602556081e-05,
      "loss": 1.5099,
      "step": 4149
    },
    {
      "epoch": 0.5944707062025498,
      "grad_norm": 1.2457205057144165,
      "learning_rate": 7.45553226234385e-05,
      "loss": 1.3649,
      "step": 4150
    },
    {
      "epoch": 0.5946139521558516,
      "grad_norm": 1.1202598810195923,
      "learning_rate": 7.451045469892554e-05,
      "loss": 1.262,
      "step": 4151
    },
    {
      "epoch": 0.5947571981091534,
      "grad_norm": 1.224773645401001,
      "learning_rate": 7.44655922616809e-05,
      "loss": 1.3971,
      "step": 4152
    },
    {
      "epoch": 0.5949004440624552,
      "grad_norm": 1.2140729427337646,
      "learning_rate": 7.442073532136244e-05,
      "loss": 1.5063,
      "step": 4153
    },
    {
      "epoch": 0.5950436900157571,
      "grad_norm": 0.8836366534233093,
      "learning_rate": 7.43758838876266e-05,
      "loss": 1.4853,
      "step": 4154
    },
    {
      "epoch": 0.5951869359690589,
      "grad_norm": 1.0752472877502441,
      "learning_rate": 7.433103797012892e-05,
      "loss": 1.456,
      "step": 4155
    },
    {
      "epoch": 0.5953301819223606,
      "grad_norm": 1.1957483291625977,
      "learning_rate": 7.428619757852352e-05,
      "loss": 1.5008,
      "step": 4156
    },
    {
      "epoch": 0.5954734278756625,
      "grad_norm": 1.1404597759246826,
      "learning_rate": 7.424136272246347e-05,
      "loss": 1.4522,
      "step": 4157
    },
    {
      "epoch": 0.5956166738289643,
      "grad_norm": 1.217352271080017,
      "learning_rate": 7.419653341160062e-05,
      "loss": 1.4309,
      "step": 4158
    },
    {
      "epoch": 0.5957599197822662,
      "grad_norm": 1.3097559213638306,
      "learning_rate": 7.415170965558553e-05,
      "loss": 1.4148,
      "step": 4159
    },
    {
      "epoch": 0.595903165735568,
      "grad_norm": 1.1112788915634155,
      "learning_rate": 7.410689146406775e-05,
      "loss": 1.3385,
      "step": 4160
    },
    {
      "epoch": 0.5960464116888697,
      "grad_norm": 1.4310517311096191,
      "learning_rate": 7.40620788466955e-05,
      "loss": 1.3791,
      "step": 4161
    },
    {
      "epoch": 0.5961896576421716,
      "grad_norm": 1.0071412324905396,
      "learning_rate": 7.401727181311578e-05,
      "loss": 1.4749,
      "step": 4162
    },
    {
      "epoch": 0.5963329035954734,
      "grad_norm": 1.010916829109192,
      "learning_rate": 7.397247037297449e-05,
      "loss": 1.386,
      "step": 4163
    },
    {
      "epoch": 0.5964761495487753,
      "grad_norm": 1.0271905660629272,
      "learning_rate": 7.392767453591623e-05,
      "loss": 1.3405,
      "step": 4164
    },
    {
      "epoch": 0.5966193955020771,
      "grad_norm": 1.1605325937271118,
      "learning_rate": 7.388288431158443e-05,
      "loss": 1.43,
      "step": 4165
    },
    {
      "epoch": 0.5967626414553789,
      "grad_norm": 1.067371129989624,
      "learning_rate": 7.383809970962137e-05,
      "loss": 1.422,
      "step": 4166
    },
    {
      "epoch": 0.5969058874086807,
      "grad_norm": 1.1182775497436523,
      "learning_rate": 7.379332073966804e-05,
      "loss": 1.1895,
      "step": 4167
    },
    {
      "epoch": 0.5970491333619825,
      "grad_norm": 1.0228168964385986,
      "learning_rate": 7.374854741136422e-05,
      "loss": 1.5937,
      "step": 4168
    },
    {
      "epoch": 0.5971923793152843,
      "grad_norm": 1.2450546026229858,
      "learning_rate": 7.370377973434855e-05,
      "loss": 1.3985,
      "step": 4169
    },
    {
      "epoch": 0.5973356252685862,
      "grad_norm": 1.225685477256775,
      "learning_rate": 7.365901771825833e-05,
      "loss": 1.5442,
      "step": 4170
    },
    {
      "epoch": 0.597478871221888,
      "grad_norm": 1.0614421367645264,
      "learning_rate": 7.361426137272982e-05,
      "loss": 1.4296,
      "step": 4171
    },
    {
      "epoch": 0.5976221171751898,
      "grad_norm": 1.2749584913253784,
      "learning_rate": 7.356951070739781e-05,
      "loss": 1.2497,
      "step": 4172
    },
    {
      "epoch": 0.5977653631284916,
      "grad_norm": 1.076539158821106,
      "learning_rate": 7.352476573189614e-05,
      "loss": 1.5376,
      "step": 4173
    },
    {
      "epoch": 0.5979086090817934,
      "grad_norm": 1.1491841077804565,
      "learning_rate": 7.348002645585725e-05,
      "loss": 1.3928,
      "step": 4174
    },
    {
      "epoch": 0.5980518550350953,
      "grad_norm": 1.2570098638534546,
      "learning_rate": 7.343529288891239e-05,
      "loss": 1.2062,
      "step": 4175
    },
    {
      "epoch": 0.5981951009883971,
      "grad_norm": 1.2708749771118164,
      "learning_rate": 7.33905650406916e-05,
      "loss": 1.3741,
      "step": 4176
    },
    {
      "epoch": 0.5983383469416989,
      "grad_norm": 1.116259217262268,
      "learning_rate": 7.334584292082365e-05,
      "loss": 1.4933,
      "step": 4177
    },
    {
      "epoch": 0.5984815928950007,
      "grad_norm": 1.0042681694030762,
      "learning_rate": 7.330112653893614e-05,
      "loss": 1.5124,
      "step": 4178
    },
    {
      "epoch": 0.5986248388483025,
      "grad_norm": 1.110458493232727,
      "learning_rate": 7.325641590465542e-05,
      "loss": 1.4164,
      "step": 4179
    },
    {
      "epoch": 0.5987680848016044,
      "grad_norm": 1.1291543245315552,
      "learning_rate": 7.321171102760653e-05,
      "loss": 1.4849,
      "step": 4180
    },
    {
      "epoch": 0.5989113307549062,
      "grad_norm": 1.1526830196380615,
      "learning_rate": 7.316701191741333e-05,
      "loss": 1.3997,
      "step": 4181
    },
    {
      "epoch": 0.5990545767082079,
      "grad_norm": 1.0835102796554565,
      "learning_rate": 7.31223185836985e-05,
      "loss": 1.5266,
      "step": 4182
    },
    {
      "epoch": 0.5991978226615098,
      "grad_norm": 1.0526680946350098,
      "learning_rate": 7.307763103608332e-05,
      "loss": 1.5701,
      "step": 4183
    },
    {
      "epoch": 0.5993410686148116,
      "grad_norm": 1.0574432611465454,
      "learning_rate": 7.303294928418799e-05,
      "loss": 1.3117,
      "step": 4184
    },
    {
      "epoch": 0.5994843145681135,
      "grad_norm": 1.101569652557373,
      "learning_rate": 7.298827333763132e-05,
      "loss": 1.5238,
      "step": 4185
    },
    {
      "epoch": 0.5996275605214153,
      "grad_norm": 1.1604955196380615,
      "learning_rate": 7.294360320603095e-05,
      "loss": 1.6361,
      "step": 4186
    },
    {
      "epoch": 0.5997708064747171,
      "grad_norm": 1.101251244544983,
      "learning_rate": 7.289893889900332e-05,
      "loss": 1.2875,
      "step": 4187
    },
    {
      "epoch": 0.5999140524280189,
      "grad_norm": 1.1338562965393066,
      "learning_rate": 7.285428042616344e-05,
      "loss": 1.433,
      "step": 4188
    },
    {
      "epoch": 0.6000572983813207,
      "grad_norm": 1.2323518991470337,
      "learning_rate": 7.280962779712525e-05,
      "loss": 1.4939,
      "step": 4189
    },
    {
      "epoch": 0.6002005443346226,
      "grad_norm": 1.0256216526031494,
      "learning_rate": 7.276498102150138e-05,
      "loss": 1.3425,
      "step": 4190
    },
    {
      "epoch": 0.6003437902879244,
      "grad_norm": 0.9391832947731018,
      "learning_rate": 7.272034010890309e-05,
      "loss": 1.555,
      "step": 4191
    },
    {
      "epoch": 0.6004870362412262,
      "grad_norm": 1.264958143234253,
      "learning_rate": 7.267570506894052e-05,
      "loss": 1.3776,
      "step": 4192
    },
    {
      "epoch": 0.600630282194528,
      "grad_norm": 0.943480908870697,
      "learning_rate": 7.263107591122246e-05,
      "loss": 1.4483,
      "step": 4193
    },
    {
      "epoch": 0.6007735281478298,
      "grad_norm": 1.3064343929290771,
      "learning_rate": 7.258645264535649e-05,
      "loss": 1.5057,
      "step": 4194
    },
    {
      "epoch": 0.6009167741011316,
      "grad_norm": 1.0172686576843262,
      "learning_rate": 7.254183528094891e-05,
      "loss": 1.3136,
      "step": 4195
    },
    {
      "epoch": 0.6010600200544335,
      "grad_norm": 1.185977816581726,
      "learning_rate": 7.249722382760469e-05,
      "loss": 1.395,
      "step": 4196
    },
    {
      "epoch": 0.6012032660077353,
      "grad_norm": 0.9948939085006714,
      "learning_rate": 7.245261829492755e-05,
      "loss": 1.404,
      "step": 4197
    },
    {
      "epoch": 0.6013465119610371,
      "grad_norm": 1.046208381652832,
      "learning_rate": 7.24080186925201e-05,
      "loss": 1.5871,
      "step": 4198
    },
    {
      "epoch": 0.6014897579143389,
      "grad_norm": 1.0895016193389893,
      "learning_rate": 7.236342502998335e-05,
      "loss": 1.5129,
      "step": 4199
    },
    {
      "epoch": 0.6016330038676407,
      "grad_norm": 1.0407259464263916,
      "learning_rate": 7.231883731691732e-05,
      "loss": 1.7093,
      "step": 4200
    },
    {
      "epoch": 0.6017762498209426,
      "grad_norm": 1.0808143615722656,
      "learning_rate": 7.22742555629206e-05,
      "loss": 1.3784,
      "step": 4201
    },
    {
      "epoch": 0.6019194957742444,
      "grad_norm": 1.1595443487167358,
      "learning_rate": 7.222967977759056e-05,
      "loss": 1.4016,
      "step": 4202
    },
    {
      "epoch": 0.6020627417275461,
      "grad_norm": 0.9473376274108887,
      "learning_rate": 7.218510997052326e-05,
      "loss": 1.3154,
      "step": 4203
    },
    {
      "epoch": 0.602205987680848,
      "grad_norm": 1.0887020826339722,
      "learning_rate": 7.214054615131345e-05,
      "loss": 1.2286,
      "step": 4204
    },
    {
      "epoch": 0.6023492336341498,
      "grad_norm": 1.2252492904663086,
      "learning_rate": 7.209598832955469e-05,
      "loss": 1.4846,
      "step": 4205
    },
    {
      "epoch": 0.6024924795874517,
      "grad_norm": 0.9583574533462524,
      "learning_rate": 7.205143651483906e-05,
      "loss": 1.406,
      "step": 4206
    },
    {
      "epoch": 0.6026357255407535,
      "grad_norm": 1.072060227394104,
      "learning_rate": 7.200689071675755e-05,
      "loss": 1.4619,
      "step": 4207
    },
    {
      "epoch": 0.6027789714940552,
      "grad_norm": 1.3092440366744995,
      "learning_rate": 7.196235094489978e-05,
      "loss": 1.3726,
      "step": 4208
    },
    {
      "epoch": 0.6029222174473571,
      "grad_norm": 1.1472665071487427,
      "learning_rate": 7.1917817208854e-05,
      "loss": 1.4685,
      "step": 4209
    },
    {
      "epoch": 0.6030654634006589,
      "grad_norm": 0.9766077995300293,
      "learning_rate": 7.187328951820723e-05,
      "loss": 1.4257,
      "step": 4210
    },
    {
      "epoch": 0.6032087093539608,
      "grad_norm": 1.1146196126937866,
      "learning_rate": 7.182876788254525e-05,
      "loss": 1.5676,
      "step": 4211
    },
    {
      "epoch": 0.6033519553072626,
      "grad_norm": 0.9668264985084534,
      "learning_rate": 7.178425231145236e-05,
      "loss": 1.2464,
      "step": 4212
    },
    {
      "epoch": 0.6034952012605644,
      "grad_norm": 1.0392547845840454,
      "learning_rate": 7.173974281451176e-05,
      "loss": 1.6053,
      "step": 4213
    },
    {
      "epoch": 0.6036384472138662,
      "grad_norm": 1.0427402257919312,
      "learning_rate": 7.169523940130519e-05,
      "loss": 1.3612,
      "step": 4214
    },
    {
      "epoch": 0.603781693167168,
      "grad_norm": 0.9183353781700134,
      "learning_rate": 7.16507420814131e-05,
      "loss": 1.6304,
      "step": 4215
    },
    {
      "epoch": 0.6039249391204699,
      "grad_norm": 1.1151803731918335,
      "learning_rate": 7.160625086441476e-05,
      "loss": 1.5209,
      "step": 4216
    },
    {
      "epoch": 0.6040681850737717,
      "grad_norm": 1.2650302648544312,
      "learning_rate": 7.156176575988794e-05,
      "loss": 1.2602,
      "step": 4217
    },
    {
      "epoch": 0.6042114310270735,
      "grad_norm": 1.0805094242095947,
      "learning_rate": 7.151728677740923e-05,
      "loss": 1.2032,
      "step": 4218
    },
    {
      "epoch": 0.6043546769803753,
      "grad_norm": 0.9683591723442078,
      "learning_rate": 7.147281392655385e-05,
      "loss": 1.5932,
      "step": 4219
    },
    {
      "epoch": 0.6044979229336771,
      "grad_norm": 1.0664361715316772,
      "learning_rate": 7.142834721689565e-05,
      "loss": 1.534,
      "step": 4220
    },
    {
      "epoch": 0.6046411688869789,
      "grad_norm": 1.0504882335662842,
      "learning_rate": 7.138388665800733e-05,
      "loss": 1.3112,
      "step": 4221
    },
    {
      "epoch": 0.6047844148402808,
      "grad_norm": 1.1913233995437622,
      "learning_rate": 7.133943225946e-05,
      "loss": 1.2581,
      "step": 4222
    },
    {
      "epoch": 0.6049276607935826,
      "grad_norm": 1.0433645248413086,
      "learning_rate": 7.129498403082369e-05,
      "loss": 1.4469,
      "step": 4223
    },
    {
      "epoch": 0.6050709067468844,
      "grad_norm": 1.1232019662857056,
      "learning_rate": 7.125054198166701e-05,
      "loss": 1.4002,
      "step": 4224
    },
    {
      "epoch": 0.6052141527001862,
      "grad_norm": 1.151464581489563,
      "learning_rate": 7.120610612155716e-05,
      "loss": 1.5652,
      "step": 4225
    },
    {
      "epoch": 0.605357398653488,
      "grad_norm": 0.8970821499824524,
      "learning_rate": 7.11616764600601e-05,
      "loss": 1.7904,
      "step": 4226
    },
    {
      "epoch": 0.6055006446067899,
      "grad_norm": 1.2024282217025757,
      "learning_rate": 7.111725300674052e-05,
      "loss": 1.4129,
      "step": 4227
    },
    {
      "epoch": 0.6056438905600917,
      "grad_norm": 1.0418825149536133,
      "learning_rate": 7.107283577116161e-05,
      "loss": 1.3158,
      "step": 4228
    },
    {
      "epoch": 0.6057871365133934,
      "grad_norm": 1.041850209236145,
      "learning_rate": 7.102842476288534e-05,
      "loss": 1.4392,
      "step": 4229
    },
    {
      "epoch": 0.6059303824666953,
      "grad_norm": 1.0541305541992188,
      "learning_rate": 7.098401999147226e-05,
      "loss": 1.3479,
      "step": 4230
    },
    {
      "epoch": 0.6060736284199971,
      "grad_norm": 0.9854674339294434,
      "learning_rate": 7.093962146648164e-05,
      "loss": 1.5298,
      "step": 4231
    },
    {
      "epoch": 0.606216874373299,
      "grad_norm": 1.0490758419036865,
      "learning_rate": 7.089522919747142e-05,
      "loss": 1.5571,
      "step": 4232
    },
    {
      "epoch": 0.6063601203266008,
      "grad_norm": 1.1308553218841553,
      "learning_rate": 7.085084319399808e-05,
      "loss": 1.4666,
      "step": 4233
    },
    {
      "epoch": 0.6065033662799026,
      "grad_norm": 1.142994999885559,
      "learning_rate": 7.08064634656169e-05,
      "loss": 1.308,
      "step": 4234
    },
    {
      "epoch": 0.6066466122332044,
      "grad_norm": 1.0846341848373413,
      "learning_rate": 7.076209002188168e-05,
      "loss": 1.3641,
      "step": 4235
    },
    {
      "epoch": 0.6067898581865062,
      "grad_norm": 1.8414368629455566,
      "learning_rate": 7.071772287234497e-05,
      "loss": 1.5357,
      "step": 4236
    },
    {
      "epoch": 0.606933104139808,
      "grad_norm": 1.0243433713912964,
      "learning_rate": 7.067336202655792e-05,
      "loss": 1.5019,
      "step": 4237
    },
    {
      "epoch": 0.6070763500931099,
      "grad_norm": 1.1277682781219482,
      "learning_rate": 7.062900749407026e-05,
      "loss": 1.2696,
      "step": 4238
    },
    {
      "epoch": 0.6072195960464117,
      "grad_norm": 1.0590330362319946,
      "learning_rate": 7.058465928443048e-05,
      "loss": 1.5483,
      "step": 4239
    },
    {
      "epoch": 0.6073628419997135,
      "grad_norm": 1.019850730895996,
      "learning_rate": 7.054031740718567e-05,
      "loss": 1.4716,
      "step": 4240
    },
    {
      "epoch": 0.6075060879530153,
      "grad_norm": 0.9222798943519592,
      "learning_rate": 7.049598187188148e-05,
      "loss": 1.4367,
      "step": 4241
    },
    {
      "epoch": 0.6076493339063171,
      "grad_norm": 0.9350789189338684,
      "learning_rate": 7.045165268806231e-05,
      "loss": 1.39,
      "step": 4242
    },
    {
      "epoch": 0.607792579859619,
      "grad_norm": 1.0299463272094727,
      "learning_rate": 7.040732986527108e-05,
      "loss": 1.4142,
      "step": 4243
    },
    {
      "epoch": 0.6079358258129208,
      "grad_norm": 1.1178940534591675,
      "learning_rate": 7.03630134130494e-05,
      "loss": 1.3787,
      "step": 4244
    },
    {
      "epoch": 0.6080790717662226,
      "grad_norm": 0.9920247197151184,
      "learning_rate": 7.03187033409376e-05,
      "loss": 1.3926,
      "step": 4245
    },
    {
      "epoch": 0.6082223177195244,
      "grad_norm": 1.0793055295944214,
      "learning_rate": 7.027439965847442e-05,
      "loss": 1.4906,
      "step": 4246
    },
    {
      "epoch": 0.6083655636728262,
      "grad_norm": 1.327938437461853,
      "learning_rate": 7.023010237519739e-05,
      "loss": 1.3738,
      "step": 4247
    },
    {
      "epoch": 0.6085088096261281,
      "grad_norm": 0.9713819622993469,
      "learning_rate": 7.018581150064269e-05,
      "loss": 1.5806,
      "step": 4248
    },
    {
      "epoch": 0.6086520555794299,
      "grad_norm": 1.1489806175231934,
      "learning_rate": 7.014152704434494e-05,
      "loss": 1.3484,
      "step": 4249
    },
    {
      "epoch": 0.6087953015327316,
      "grad_norm": 0.9832737445831299,
      "learning_rate": 7.009724901583755e-05,
      "loss": 1.3335,
      "step": 4250
    },
    {
      "epoch": 0.6089385474860335,
      "grad_norm": 1.181135892868042,
      "learning_rate": 7.005297742465247e-05,
      "loss": 1.3087,
      "step": 4251
    },
    {
      "epoch": 0.6090817934393353,
      "grad_norm": 1.2180101871490479,
      "learning_rate": 7.000871228032027e-05,
      "loss": 1.3228,
      "step": 4252
    },
    {
      "epoch": 0.6092250393926372,
      "grad_norm": 0.8919790387153625,
      "learning_rate": 6.996445359237016e-05,
      "loss": 1.5048,
      "step": 4253
    },
    {
      "epoch": 0.609368285345939,
      "grad_norm": 0.9915915131568909,
      "learning_rate": 6.992020137032988e-05,
      "loss": 1.5087,
      "step": 4254
    },
    {
      "epoch": 0.6095115312992408,
      "grad_norm": 1.1417628526687622,
      "learning_rate": 6.987595562372596e-05,
      "loss": 1.35,
      "step": 4255
    },
    {
      "epoch": 0.6096547772525426,
      "grad_norm": 1.0542162656784058,
      "learning_rate": 6.983171636208328e-05,
      "loss": 1.4884,
      "step": 4256
    },
    {
      "epoch": 0.6097980232058444,
      "grad_norm": 1.1912367343902588,
      "learning_rate": 6.978748359492553e-05,
      "loss": 1.5181,
      "step": 4257
    },
    {
      "epoch": 0.6099412691591463,
      "grad_norm": 1.161149024963379,
      "learning_rate": 6.974325733177495e-05,
      "loss": 1.4194,
      "step": 4258
    },
    {
      "epoch": 0.6100845151124481,
      "grad_norm": 1.0689343214035034,
      "learning_rate": 6.96990375821523e-05,
      "loss": 1.3595,
      "step": 4259
    },
    {
      "epoch": 0.6102277610657499,
      "grad_norm": 1.454522967338562,
      "learning_rate": 6.9654824355577e-05,
      "loss": 1.2879,
      "step": 4260
    },
    {
      "epoch": 0.6103710070190517,
      "grad_norm": 1.1314560174942017,
      "learning_rate": 6.961061766156715e-05,
      "loss": 1.5118,
      "step": 4261
    },
    {
      "epoch": 0.6105142529723535,
      "grad_norm": 0.9639393091201782,
      "learning_rate": 6.956641750963927e-05,
      "loss": 1.4141,
      "step": 4262
    },
    {
      "epoch": 0.6106574989256554,
      "grad_norm": 1.0022913217544556,
      "learning_rate": 6.952222390930858e-05,
      "loss": 1.431,
      "step": 4263
    },
    {
      "epoch": 0.6108007448789572,
      "grad_norm": 0.959291934967041,
      "learning_rate": 6.947803687008888e-05,
      "loss": 1.5013,
      "step": 4264
    },
    {
      "epoch": 0.610943990832259,
      "grad_norm": 1.1029284000396729,
      "learning_rate": 6.943385640149251e-05,
      "loss": 1.4265,
      "step": 4265
    },
    {
      "epoch": 0.6110872367855608,
      "grad_norm": 1.052175521850586,
      "learning_rate": 6.938968251303053e-05,
      "loss": 1.4451,
      "step": 4266
    },
    {
      "epoch": 0.6112304827388626,
      "grad_norm": 1.0712398290634155,
      "learning_rate": 6.934551521421235e-05,
      "loss": 1.3255,
      "step": 4267
    },
    {
      "epoch": 0.6113737286921644,
      "grad_norm": 0.9968187212944031,
      "learning_rate": 6.93013545145462e-05,
      "loss": 1.4126,
      "step": 4268
    },
    {
      "epoch": 0.6115169746454663,
      "grad_norm": 1.038417935371399,
      "learning_rate": 6.925720042353876e-05,
      "loss": 1.6231,
      "step": 4269
    },
    {
      "epoch": 0.6116602205987681,
      "grad_norm": 1.0528451204299927,
      "learning_rate": 6.921305295069528e-05,
      "loss": 1.4078,
      "step": 4270
    },
    {
      "epoch": 0.6118034665520699,
      "grad_norm": 1.3477317094802856,
      "learning_rate": 6.916891210551965e-05,
      "loss": 1.192,
      "step": 4271
    },
    {
      "epoch": 0.6119467125053717,
      "grad_norm": 1.0575848817825317,
      "learning_rate": 6.912477789751426e-05,
      "loss": 1.2557,
      "step": 4272
    },
    {
      "epoch": 0.6120899584586735,
      "grad_norm": 1.0288550853729248,
      "learning_rate": 6.908065033618018e-05,
      "loss": 1.54,
      "step": 4273
    },
    {
      "epoch": 0.6122332044119754,
      "grad_norm": 1.1299831867218018,
      "learning_rate": 6.903652943101697e-05,
      "loss": 1.3998,
      "step": 4274
    },
    {
      "epoch": 0.6123764503652772,
      "grad_norm": 0.9560255408287048,
      "learning_rate": 6.89924151915227e-05,
      "loss": 1.408,
      "step": 4275
    },
    {
      "epoch": 0.612519696318579,
      "grad_norm": 0.9701712131500244,
      "learning_rate": 6.894830762719411e-05,
      "loss": 1.4396,
      "step": 4276
    },
    {
      "epoch": 0.6126629422718808,
      "grad_norm": 1.0037755966186523,
      "learning_rate": 6.890420674752653e-05,
      "loss": 1.4684,
      "step": 4277
    },
    {
      "epoch": 0.6128061882251826,
      "grad_norm": 1.1921309232711792,
      "learning_rate": 6.886011256201371e-05,
      "loss": 1.4515,
      "step": 4278
    },
    {
      "epoch": 0.6129494341784845,
      "grad_norm": 1.0522853136062622,
      "learning_rate": 6.881602508014808e-05,
      "loss": 1.492,
      "step": 4279
    },
    {
      "epoch": 0.6130926801317863,
      "grad_norm": 1.0163192749023438,
      "learning_rate": 6.877194431142055e-05,
      "loss": 1.5184,
      "step": 4280
    },
    {
      "epoch": 0.6132359260850881,
      "grad_norm": 0.9095003604888916,
      "learning_rate": 6.872787026532062e-05,
      "loss": 1.447,
      "step": 4281
    },
    {
      "epoch": 0.6133791720383899,
      "grad_norm": 1.3466076850891113,
      "learning_rate": 6.868380295133641e-05,
      "loss": 1.2762,
      "step": 4282
    },
    {
      "epoch": 0.6135224179916917,
      "grad_norm": 1.1882424354553223,
      "learning_rate": 6.86397423789544e-05,
      "loss": 1.4759,
      "step": 4283
    },
    {
      "epoch": 0.6136656639449936,
      "grad_norm": 1.255839228630066,
      "learning_rate": 6.859568855765985e-05,
      "loss": 1.4593,
      "step": 4284
    },
    {
      "epoch": 0.6138089098982954,
      "grad_norm": 1.0026546716690063,
      "learning_rate": 6.855164149693641e-05,
      "loss": 1.3282,
      "step": 4285
    },
    {
      "epoch": 0.6139521558515972,
      "grad_norm": 1.0240823030471802,
      "learning_rate": 6.850760120626633e-05,
      "loss": 1.3893,
      "step": 4286
    },
    {
      "epoch": 0.614095401804899,
      "grad_norm": 1.1057519912719727,
      "learning_rate": 6.84635676951304e-05,
      "loss": 1.4381,
      "step": 4287
    },
    {
      "epoch": 0.6142386477582008,
      "grad_norm": 0.9514822363853455,
      "learning_rate": 6.841954097300791e-05,
      "loss": 1.4008,
      "step": 4288
    },
    {
      "epoch": 0.6143818937115026,
      "grad_norm": 1.039219617843628,
      "learning_rate": 6.837552104937679e-05,
      "loss": 1.4591,
      "step": 4289
    },
    {
      "epoch": 0.6145251396648045,
      "grad_norm": 1.2077304124832153,
      "learning_rate": 6.833150793371341e-05,
      "loss": 1.2696,
      "step": 4290
    },
    {
      "epoch": 0.6146683856181063,
      "grad_norm": 0.9501237869262695,
      "learning_rate": 6.828750163549267e-05,
      "loss": 1.4338,
      "step": 4291
    },
    {
      "epoch": 0.6148116315714081,
      "grad_norm": 0.8435130715370178,
      "learning_rate": 6.824350216418808e-05,
      "loss": 1.3838,
      "step": 4292
    },
    {
      "epoch": 0.6149548775247099,
      "grad_norm": 0.9703522324562073,
      "learning_rate": 6.819950952927161e-05,
      "loss": 1.2938,
      "step": 4293
    },
    {
      "epoch": 0.6150981234780117,
      "grad_norm": 1.0569028854370117,
      "learning_rate": 6.815552374021378e-05,
      "loss": 1.4936,
      "step": 4294
    },
    {
      "epoch": 0.6152413694313136,
      "grad_norm": 0.9793410897254944,
      "learning_rate": 6.811154480648371e-05,
      "loss": 1.5457,
      "step": 4295
    },
    {
      "epoch": 0.6153846153846154,
      "grad_norm": 1.0446767807006836,
      "learning_rate": 6.80675727375489e-05,
      "loss": 1.5509,
      "step": 4296
    },
    {
      "epoch": 0.6155278613379173,
      "grad_norm": 1.0542500019073486,
      "learning_rate": 6.802360754287547e-05,
      "loss": 1.3835,
      "step": 4297
    },
    {
      "epoch": 0.615671107291219,
      "grad_norm": 1.2045929431915283,
      "learning_rate": 6.797964923192807e-05,
      "loss": 1.2305,
      "step": 4298
    },
    {
      "epoch": 0.6158143532445208,
      "grad_norm": 1.0417587757110596,
      "learning_rate": 6.793569781416978e-05,
      "loss": 1.4487,
      "step": 4299
    },
    {
      "epoch": 0.6159575991978227,
      "grad_norm": 1.017083764076233,
      "learning_rate": 6.789175329906232e-05,
      "loss": 1.5643,
      "step": 4300
    },
    {
      "epoch": 0.6161008451511245,
      "grad_norm": 1.3205528259277344,
      "learning_rate": 6.784781569606576e-05,
      "loss": 1.3747,
      "step": 4301
    },
    {
      "epoch": 0.6162440911044264,
      "grad_norm": 1.064139723777771,
      "learning_rate": 6.780388501463887e-05,
      "loss": 1.3241,
      "step": 4302
    },
    {
      "epoch": 0.6163873370577281,
      "grad_norm": 1.2266569137573242,
      "learning_rate": 6.775996126423882e-05,
      "loss": 1.3339,
      "step": 4303
    },
    {
      "epoch": 0.6165305830110299,
      "grad_norm": 1.1888982057571411,
      "learning_rate": 6.771604445432127e-05,
      "loss": 1.4214,
      "step": 4304
    },
    {
      "epoch": 0.6166738289643318,
      "grad_norm": 1.0135010480880737,
      "learning_rate": 6.767213459434047e-05,
      "loss": 1.43,
      "step": 4305
    },
    {
      "epoch": 0.6168170749176336,
      "grad_norm": 1.1999993324279785,
      "learning_rate": 6.762823169374906e-05,
      "loss": 1.3565,
      "step": 4306
    },
    {
      "epoch": 0.6169603208709354,
      "grad_norm": 1.078255295753479,
      "learning_rate": 6.758433576199832e-05,
      "loss": 1.5479,
      "step": 4307
    },
    {
      "epoch": 0.6171035668242372,
      "grad_norm": 1.3791784048080444,
      "learning_rate": 6.754044680853794e-05,
      "loss": 1.4519,
      "step": 4308
    },
    {
      "epoch": 0.617246812777539,
      "grad_norm": 0.9597680568695068,
      "learning_rate": 6.749656484281608e-05,
      "loss": 1.3832,
      "step": 4309
    },
    {
      "epoch": 0.6173900587308409,
      "grad_norm": 1.0660167932510376,
      "learning_rate": 6.745268987427946e-05,
      "loss": 1.4179,
      "step": 4310
    },
    {
      "epoch": 0.6175333046841427,
      "grad_norm": 0.9361695051193237,
      "learning_rate": 6.740882191237334e-05,
      "loss": 1.5645,
      "step": 4311
    },
    {
      "epoch": 0.6176765506374445,
      "grad_norm": 1.1283692121505737,
      "learning_rate": 6.736496096654134e-05,
      "loss": 1.2424,
      "step": 4312
    },
    {
      "epoch": 0.6178197965907463,
      "grad_norm": 1.091731309890747,
      "learning_rate": 6.732110704622564e-05,
      "loss": 1.4349,
      "step": 4313
    },
    {
      "epoch": 0.6179630425440481,
      "grad_norm": 0.9673663973808289,
      "learning_rate": 6.727726016086693e-05,
      "loss": 1.4077,
      "step": 4314
    },
    {
      "epoch": 0.6181062884973499,
      "grad_norm": 0.9439156651496887,
      "learning_rate": 6.723342031990431e-05,
      "loss": 1.4306,
      "step": 4315
    },
    {
      "epoch": 0.6182495344506518,
      "grad_norm": 0.9377850294113159,
      "learning_rate": 6.71895875327755e-05,
      "loss": 1.2805,
      "step": 4316
    },
    {
      "epoch": 0.6183927804039536,
      "grad_norm": 1.2535158395767212,
      "learning_rate": 6.714576180891654e-05,
      "loss": 1.4505,
      "step": 4317
    },
    {
      "epoch": 0.6185360263572554,
      "grad_norm": 0.9291293025016785,
      "learning_rate": 6.710194315776203e-05,
      "loss": 1.5022,
      "step": 4318
    },
    {
      "epoch": 0.6186792723105572,
      "grad_norm": 1.030538558959961,
      "learning_rate": 6.705813158874509e-05,
      "loss": 1.4603,
      "step": 4319
    },
    {
      "epoch": 0.618822518263859,
      "grad_norm": 1.0712991952896118,
      "learning_rate": 6.70143271112972e-05,
      "loss": 1.2475,
      "step": 4320
    },
    {
      "epoch": 0.6189657642171609,
      "grad_norm": 1.2680820226669312,
      "learning_rate": 6.697052973484845e-05,
      "loss": 1.2832,
      "step": 4321
    },
    {
      "epoch": 0.6191090101704627,
      "grad_norm": 1.0325183868408203,
      "learning_rate": 6.692673946882727e-05,
      "loss": 1.4727,
      "step": 4322
    },
    {
      "epoch": 0.6192522561237646,
      "grad_norm": 1.106132984161377,
      "learning_rate": 6.688295632266064e-05,
      "loss": 1.4177,
      "step": 4323
    },
    {
      "epoch": 0.6193955020770663,
      "grad_norm": 1.0620390176773071,
      "learning_rate": 6.683918030577402e-05,
      "loss": 1.4906,
      "step": 4324
    },
    {
      "epoch": 0.6195387480303681,
      "grad_norm": 1.1507896184921265,
      "learning_rate": 6.679541142759126e-05,
      "loss": 1.4685,
      "step": 4325
    },
    {
      "epoch": 0.61968199398367,
      "grad_norm": 1.0380597114562988,
      "learning_rate": 6.675164969753472e-05,
      "loss": 1.3283,
      "step": 4326
    },
    {
      "epoch": 0.6198252399369718,
      "grad_norm": 1.0493804216384888,
      "learning_rate": 6.670789512502527e-05,
      "loss": 1.4178,
      "step": 4327
    },
    {
      "epoch": 0.6199684858902736,
      "grad_norm": 1.0655943155288696,
      "learning_rate": 6.666414771948211e-05,
      "loss": 1.355,
      "step": 4328
    },
    {
      "epoch": 0.6201117318435754,
      "grad_norm": 0.961478054523468,
      "learning_rate": 6.662040749032303e-05,
      "loss": 1.5201,
      "step": 4329
    },
    {
      "epoch": 0.6202549777968772,
      "grad_norm": 1.170172095298767,
      "learning_rate": 6.65766744469642e-05,
      "loss": 1.5383,
      "step": 4330
    },
    {
      "epoch": 0.6203982237501791,
      "grad_norm": 1.3746247291564941,
      "learning_rate": 6.653294859882027e-05,
      "loss": 1.1033,
      "step": 4331
    },
    {
      "epoch": 0.6205414697034809,
      "grad_norm": 1.3051588535308838,
      "learning_rate": 6.648922995530433e-05,
      "loss": 1.4448,
      "step": 4332
    },
    {
      "epoch": 0.6206847156567827,
      "grad_norm": 1.041752815246582,
      "learning_rate": 6.644551852582787e-05,
      "loss": 1.5704,
      "step": 4333
    },
    {
      "epoch": 0.6208279616100845,
      "grad_norm": 0.9749700427055359,
      "learning_rate": 6.6401814319801e-05,
      "loss": 1.0998,
      "step": 4334
    },
    {
      "epoch": 0.6209712075633863,
      "grad_norm": 1.117542028427124,
      "learning_rate": 6.635811734663202e-05,
      "loss": 1.4643,
      "step": 4335
    },
    {
      "epoch": 0.6211144535166881,
      "grad_norm": 1.2276312112808228,
      "learning_rate": 6.631442761572788e-05,
      "loss": 1.3601,
      "step": 4336
    },
    {
      "epoch": 0.62125769946999,
      "grad_norm": 1.165616512298584,
      "learning_rate": 6.627074513649392e-05,
      "loss": 1.3909,
      "step": 4337
    },
    {
      "epoch": 0.6214009454232918,
      "grad_norm": 1.1910256147384644,
      "learning_rate": 6.622706991833383e-05,
      "loss": 1.3367,
      "step": 4338
    },
    {
      "epoch": 0.6215441913765936,
      "grad_norm": 1.0784599781036377,
      "learning_rate": 6.618340197064983e-05,
      "loss": 1.3365,
      "step": 4339
    },
    {
      "epoch": 0.6216874373298954,
      "grad_norm": 1.2258021831512451,
      "learning_rate": 6.613974130284258e-05,
      "loss": 1.4701,
      "step": 4340
    },
    {
      "epoch": 0.6218306832831972,
      "grad_norm": 0.9904068112373352,
      "learning_rate": 6.60960879243111e-05,
      "loss": 1.5921,
      "step": 4341
    },
    {
      "epoch": 0.6219739292364991,
      "grad_norm": 1.0260205268859863,
      "learning_rate": 6.605244184445292e-05,
      "loss": 1.2602,
      "step": 4342
    },
    {
      "epoch": 0.6221171751898009,
      "grad_norm": 1.2487592697143555,
      "learning_rate": 6.600880307266393e-05,
      "loss": 1.4236,
      "step": 4343
    },
    {
      "epoch": 0.6222604211431028,
      "grad_norm": 0.9697585701942444,
      "learning_rate": 6.596517161833845e-05,
      "loss": 1.3647,
      "step": 4344
    },
    {
      "epoch": 0.6224036670964045,
      "grad_norm": 1.0318689346313477,
      "learning_rate": 6.592154749086934e-05,
      "loss": 1.4249,
      "step": 4345
    },
    {
      "epoch": 0.6225469130497063,
      "grad_norm": 1.2537623643875122,
      "learning_rate": 6.587793069964771e-05,
      "loss": 1.4691,
      "step": 4346
    },
    {
      "epoch": 0.6226901590030082,
      "grad_norm": 0.9639636874198914,
      "learning_rate": 6.583432125406323e-05,
      "loss": 1.2842,
      "step": 4347
    },
    {
      "epoch": 0.62283340495631,
      "grad_norm": 1.045020580291748,
      "learning_rate": 6.579071916350393e-05,
      "loss": 1.6159,
      "step": 4348
    },
    {
      "epoch": 0.6229766509096119,
      "grad_norm": 0.9704081416130066,
      "learning_rate": 6.57471244373562e-05,
      "loss": 1.3843,
      "step": 4349
    },
    {
      "epoch": 0.6231198968629136,
      "grad_norm": 0.9892087578773499,
      "learning_rate": 6.5703537085005e-05,
      "loss": 1.458,
      "step": 4350
    },
    {
      "epoch": 0.6232631428162154,
      "grad_norm": 1.1427958011627197,
      "learning_rate": 6.565995711583353e-05,
      "loss": 1.4811,
      "step": 4351
    },
    {
      "epoch": 0.6234063887695173,
      "grad_norm": 1.0081120729446411,
      "learning_rate": 6.561638453922349e-05,
      "loss": 1.4028,
      "step": 4352
    },
    {
      "epoch": 0.6235496347228191,
      "grad_norm": 0.9824202060699463,
      "learning_rate": 6.557281936455506e-05,
      "loss": 1.3992,
      "step": 4353
    },
    {
      "epoch": 0.6236928806761209,
      "grad_norm": 1.043860912322998,
      "learning_rate": 6.552926160120663e-05,
      "loss": 1.3606,
      "step": 4354
    },
    {
      "epoch": 0.6238361266294227,
      "grad_norm": 0.9067694544792175,
      "learning_rate": 6.548571125855519e-05,
      "loss": 1.5557,
      "step": 4355
    },
    {
      "epoch": 0.6239793725827245,
      "grad_norm": 1.0535426139831543,
      "learning_rate": 6.544216834597597e-05,
      "loss": 1.3799,
      "step": 4356
    },
    {
      "epoch": 0.6241226185360264,
      "grad_norm": 1.2056607007980347,
      "learning_rate": 6.539863287284275e-05,
      "loss": 1.4768,
      "step": 4357
    },
    {
      "epoch": 0.6242658644893282,
      "grad_norm": 0.9185512661933899,
      "learning_rate": 6.535510484852767e-05,
      "loss": 1.5403,
      "step": 4358
    },
    {
      "epoch": 0.62440911044263,
      "grad_norm": 0.9265509247779846,
      "learning_rate": 6.531158428240113e-05,
      "loss": 1.6739,
      "step": 4359
    },
    {
      "epoch": 0.6245523563959318,
      "grad_norm": 1.0269006490707397,
      "learning_rate": 6.52680711838321e-05,
      "loss": 1.3949,
      "step": 4360
    },
    {
      "epoch": 0.6246956023492336,
      "grad_norm": 1.1948872804641724,
      "learning_rate": 6.522456556218791e-05,
      "loss": 1.3824,
      "step": 4361
    },
    {
      "epoch": 0.6248388483025354,
      "grad_norm": 1.0690702199935913,
      "learning_rate": 6.518106742683415e-05,
      "loss": 1.4393,
      "step": 4362
    },
    {
      "epoch": 0.6249820942558373,
      "grad_norm": 1.0668222904205322,
      "learning_rate": 6.513757678713495e-05,
      "loss": 1.518,
      "step": 4363
    },
    {
      "epoch": 0.6251253402091391,
      "grad_norm": 1.1756402254104614,
      "learning_rate": 6.509409365245276e-05,
      "loss": 1.33,
      "step": 4364
    },
    {
      "epoch": 0.625268586162441,
      "grad_norm": 1.1705524921417236,
      "learning_rate": 6.50506180321484e-05,
      "loss": 1.2453,
      "step": 4365
    },
    {
      "epoch": 0.6254118321157427,
      "grad_norm": 1.2229130268096924,
      "learning_rate": 6.500714993558115e-05,
      "loss": 1.397,
      "step": 4366
    },
    {
      "epoch": 0.6255550780690445,
      "grad_norm": 0.9605799317359924,
      "learning_rate": 6.496368937210853e-05,
      "loss": 1.406,
      "step": 4367
    },
    {
      "epoch": 0.6256983240223464,
      "grad_norm": 0.9673969149589539,
      "learning_rate": 6.49202363510866e-05,
      "loss": 1.4262,
      "step": 4368
    },
    {
      "epoch": 0.6258415699756482,
      "grad_norm": 1.1713780164718628,
      "learning_rate": 6.487679088186973e-05,
      "loss": 1.3311,
      "step": 4369
    },
    {
      "epoch": 0.6259848159289501,
      "grad_norm": 1.0765293836593628,
      "learning_rate": 6.483335297381057e-05,
      "loss": 1.4639,
      "step": 4370
    },
    {
      "epoch": 0.6261280618822518,
      "grad_norm": 0.9536416530609131,
      "learning_rate": 6.478992263626031e-05,
      "loss": 1.4857,
      "step": 4371
    },
    {
      "epoch": 0.6262713078355536,
      "grad_norm": 1.0394986867904663,
      "learning_rate": 6.474649987856834e-05,
      "loss": 1.3282,
      "step": 4372
    },
    {
      "epoch": 0.6264145537888555,
      "grad_norm": 0.9428933262825012,
      "learning_rate": 6.47030847100826e-05,
      "loss": 1.5142,
      "step": 4373
    },
    {
      "epoch": 0.6265577997421573,
      "grad_norm": 1.1511598825454712,
      "learning_rate": 6.465967714014927e-05,
      "loss": 1.341,
      "step": 4374
    },
    {
      "epoch": 0.6267010456954591,
      "grad_norm": 0.9480801224708557,
      "learning_rate": 6.461627717811288e-05,
      "loss": 1.33,
      "step": 4375
    },
    {
      "epoch": 0.6268442916487609,
      "grad_norm": 1.1816121339797974,
      "learning_rate": 6.457288483331639e-05,
      "loss": 1.3985,
      "step": 4376
    },
    {
      "epoch": 0.6269875376020627,
      "grad_norm": 1.0404813289642334,
      "learning_rate": 6.452950011510118e-05,
      "loss": 1.5845,
      "step": 4377
    },
    {
      "epoch": 0.6271307835553646,
      "grad_norm": 0.9674251079559326,
      "learning_rate": 6.448612303280677e-05,
      "loss": 1.4529,
      "step": 4378
    },
    {
      "epoch": 0.6272740295086664,
      "grad_norm": 1.1875535249710083,
      "learning_rate": 6.444275359577128e-05,
      "loss": 1.4005,
      "step": 4379
    },
    {
      "epoch": 0.6274172754619682,
      "grad_norm": 0.9839823842048645,
      "learning_rate": 6.439939181333101e-05,
      "loss": 1.4954,
      "step": 4380
    },
    {
      "epoch": 0.62756052141527,
      "grad_norm": 1.040177822113037,
      "learning_rate": 6.435603769482071e-05,
      "loss": 1.0878,
      "step": 4381
    },
    {
      "epoch": 0.6277037673685718,
      "grad_norm": 0.8938812017440796,
      "learning_rate": 6.431269124957347e-05,
      "loss": 1.6431,
      "step": 4382
    },
    {
      "epoch": 0.6278470133218736,
      "grad_norm": 1.0711067914962769,
      "learning_rate": 6.426935248692064e-05,
      "loss": 1.2503,
      "step": 4383
    },
    {
      "epoch": 0.6279902592751755,
      "grad_norm": 1.0661498308181763,
      "learning_rate": 6.422602141619207e-05,
      "loss": 1.4776,
      "step": 4384
    },
    {
      "epoch": 0.6281335052284773,
      "grad_norm": 1.0704522132873535,
      "learning_rate": 6.418269804671576e-05,
      "loss": 1.5105,
      "step": 4385
    },
    {
      "epoch": 0.6282767511817792,
      "grad_norm": 1.1132880449295044,
      "learning_rate": 6.413938238781824e-05,
      "loss": 1.3982,
      "step": 4386
    },
    {
      "epoch": 0.6284199971350809,
      "grad_norm": 1.1062909364700317,
      "learning_rate": 6.409607444882431e-05,
      "loss": 1.4186,
      "step": 4387
    },
    {
      "epoch": 0.6285632430883827,
      "grad_norm": 1.0716989040374756,
      "learning_rate": 6.405277423905705e-05,
      "loss": 1.3269,
      "step": 4388
    },
    {
      "epoch": 0.6287064890416846,
      "grad_norm": 1.0510997772216797,
      "learning_rate": 6.40094817678379e-05,
      "loss": 1.5088,
      "step": 4389
    },
    {
      "epoch": 0.6288497349949864,
      "grad_norm": 1.1650594472885132,
      "learning_rate": 6.396619704448677e-05,
      "loss": 1.4777,
      "step": 4390
    },
    {
      "epoch": 0.6289929809482883,
      "grad_norm": 1.041411280632019,
      "learning_rate": 6.392292007832168e-05,
      "loss": 1.4776,
      "step": 4391
    },
    {
      "epoch": 0.62913622690159,
      "grad_norm": 0.9809937477111816,
      "learning_rate": 6.387965087865915e-05,
      "loss": 1.4002,
      "step": 4392
    },
    {
      "epoch": 0.6292794728548918,
      "grad_norm": 1.183027982711792,
      "learning_rate": 6.383638945481391e-05,
      "loss": 1.4181,
      "step": 4393
    },
    {
      "epoch": 0.6294227188081937,
      "grad_norm": 1.1031701564788818,
      "learning_rate": 6.379313581609912e-05,
      "loss": 1.5543,
      "step": 4394
    },
    {
      "epoch": 0.6295659647614955,
      "grad_norm": 1.1924076080322266,
      "learning_rate": 6.374988997182623e-05,
      "loss": 1.4058,
      "step": 4395
    },
    {
      "epoch": 0.6297092107147974,
      "grad_norm": 1.3014189004898071,
      "learning_rate": 6.370665193130495e-05,
      "loss": 1.4321,
      "step": 4396
    },
    {
      "epoch": 0.6298524566680991,
      "grad_norm": 1.1979700326919556,
      "learning_rate": 6.36634217038434e-05,
      "loss": 1.4608,
      "step": 4397
    },
    {
      "epoch": 0.6299957026214009,
      "grad_norm": 1.238476276397705,
      "learning_rate": 6.362019929874799e-05,
      "loss": 1.2799,
      "step": 4398
    },
    {
      "epoch": 0.6301389485747028,
      "grad_norm": 1.080533504486084,
      "learning_rate": 6.357698472532338e-05,
      "loss": 1.3547,
      "step": 4399
    },
    {
      "epoch": 0.6302821945280046,
      "grad_norm": 0.9394568800926208,
      "learning_rate": 6.353377799287266e-05,
      "loss": 1.5745,
      "step": 4400
    },
    {
      "epoch": 0.6304254404813064,
      "grad_norm": 0.9222342371940613,
      "learning_rate": 6.349057911069709e-05,
      "loss": 1.4971,
      "step": 4401
    },
    {
      "epoch": 0.6305686864346082,
      "grad_norm": 1.2611083984375,
      "learning_rate": 6.344738808809639e-05,
      "loss": 1.5,
      "step": 4402
    },
    {
      "epoch": 0.63071193238791,
      "grad_norm": 1.0464383363723755,
      "learning_rate": 6.340420493436851e-05,
      "loss": 1.1798,
      "step": 4403
    },
    {
      "epoch": 0.6308551783412119,
      "grad_norm": 1.0896241664886475,
      "learning_rate": 6.33610296588097e-05,
      "loss": 1.3448,
      "step": 4404
    },
    {
      "epoch": 0.6309984242945137,
      "grad_norm": 1.097690463066101,
      "learning_rate": 6.33178622707145e-05,
      "loss": 1.3587,
      "step": 4405
    },
    {
      "epoch": 0.6311416702478155,
      "grad_norm": 1.1287853717803955,
      "learning_rate": 6.327470277937586e-05,
      "loss": 1.4861,
      "step": 4406
    },
    {
      "epoch": 0.6312849162011173,
      "grad_norm": 1.175378680229187,
      "learning_rate": 6.323155119408489e-05,
      "loss": 1.568,
      "step": 4407
    },
    {
      "epoch": 0.6314281621544191,
      "grad_norm": 0.937500536441803,
      "learning_rate": 6.318840752413106e-05,
      "loss": 1.3875,
      "step": 4408
    },
    {
      "epoch": 0.6315714081077209,
      "grad_norm": 1.0182346105575562,
      "learning_rate": 6.314527177880215e-05,
      "loss": 1.3817,
      "step": 4409
    },
    {
      "epoch": 0.6317146540610228,
      "grad_norm": 1.0185761451721191,
      "learning_rate": 6.310214396738419e-05,
      "loss": 1.5255,
      "step": 4410
    },
    {
      "epoch": 0.6318579000143246,
      "grad_norm": 1.3752186298370361,
      "learning_rate": 6.30590240991616e-05,
      "loss": 1.5994,
      "step": 4411
    },
    {
      "epoch": 0.6320011459676265,
      "grad_norm": 1.1595021486282349,
      "learning_rate": 6.301591218341693e-05,
      "loss": 1.4939,
      "step": 4412
    },
    {
      "epoch": 0.6321443919209282,
      "grad_norm": 1.0067163705825806,
      "learning_rate": 6.297280822943118e-05,
      "loss": 1.3161,
      "step": 4413
    },
    {
      "epoch": 0.63228763787423,
      "grad_norm": 1.0694667100906372,
      "learning_rate": 6.292971224648352e-05,
      "loss": 1.5245,
      "step": 4414
    },
    {
      "epoch": 0.6324308838275319,
      "grad_norm": 1.0529897212982178,
      "learning_rate": 6.288662424385148e-05,
      "loss": 1.4631,
      "step": 4415
    },
    {
      "epoch": 0.6325741297808337,
      "grad_norm": 1.196317434310913,
      "learning_rate": 6.284354423081083e-05,
      "loss": 1.4685,
      "step": 4416
    },
    {
      "epoch": 0.6327173757341356,
      "grad_norm": 1.1030393838882446,
      "learning_rate": 6.280047221663558e-05,
      "loss": 1.3836,
      "step": 4417
    },
    {
      "epoch": 0.6328606216874373,
      "grad_norm": 1.1806074380874634,
      "learning_rate": 6.275740821059817e-05,
      "loss": 1.3812,
      "step": 4418
    },
    {
      "epoch": 0.6330038676407391,
      "grad_norm": 1.0087110996246338,
      "learning_rate": 6.271435222196916e-05,
      "loss": 1.4381,
      "step": 4419
    },
    {
      "epoch": 0.633147113594041,
      "grad_norm": 1.0950722694396973,
      "learning_rate": 6.267130426001742e-05,
      "loss": 1.4961,
      "step": 4420
    },
    {
      "epoch": 0.6332903595473428,
      "grad_norm": 1.0456870794296265,
      "learning_rate": 6.262826433401015e-05,
      "loss": 1.467,
      "step": 4421
    },
    {
      "epoch": 0.6334336055006446,
      "grad_norm": 1.065233588218689,
      "learning_rate": 6.258523245321274e-05,
      "loss": 1.3606,
      "step": 4422
    },
    {
      "epoch": 0.6335768514539464,
      "grad_norm": 0.9999430775642395,
      "learning_rate": 6.254220862688889e-05,
      "loss": 1.4896,
      "step": 4423
    },
    {
      "epoch": 0.6337200974072482,
      "grad_norm": 0.8572548627853394,
      "learning_rate": 6.249919286430063e-05,
      "loss": 1.3753,
      "step": 4424
    },
    {
      "epoch": 0.6338633433605501,
      "grad_norm": 1.059081792831421,
      "learning_rate": 6.245618517470813e-05,
      "loss": 1.3788,
      "step": 4425
    },
    {
      "epoch": 0.6340065893138519,
      "grad_norm": 1.0721956491470337,
      "learning_rate": 6.24131855673699e-05,
      "loss": 1.24,
      "step": 4426
    },
    {
      "epoch": 0.6341498352671537,
      "grad_norm": 1.187609314918518,
      "learning_rate": 6.23701940515427e-05,
      "loss": 1.3354,
      "step": 4427
    },
    {
      "epoch": 0.6342930812204555,
      "grad_norm": 1.0395785570144653,
      "learning_rate": 6.232721063648148e-05,
      "loss": 1.4325,
      "step": 4428
    },
    {
      "epoch": 0.6344363271737573,
      "grad_norm": 1.2745407819747925,
      "learning_rate": 6.228423533143963e-05,
      "loss": 1.4736,
      "step": 4429
    },
    {
      "epoch": 0.6345795731270591,
      "grad_norm": 1.2750049829483032,
      "learning_rate": 6.224126814566853e-05,
      "loss": 1.5035,
      "step": 4430
    },
    {
      "epoch": 0.634722819080361,
      "grad_norm": 1.1220617294311523,
      "learning_rate": 6.219830908841802e-05,
      "loss": 1.4499,
      "step": 4431
    },
    {
      "epoch": 0.6348660650336628,
      "grad_norm": 1.1072580814361572,
      "learning_rate": 6.215535816893615e-05,
      "loss": 1.3874,
      "step": 4432
    },
    {
      "epoch": 0.6350093109869647,
      "grad_norm": 0.966542661190033,
      "learning_rate": 6.211241539646913e-05,
      "loss": 1.386,
      "step": 4433
    },
    {
      "epoch": 0.6351525569402664,
      "grad_norm": 1.1650657653808594,
      "learning_rate": 6.206948078026154e-05,
      "loss": 1.4699,
      "step": 4434
    },
    {
      "epoch": 0.6352958028935682,
      "grad_norm": 0.9655526280403137,
      "learning_rate": 6.202655432955604e-05,
      "loss": 1.4294,
      "step": 4435
    },
    {
      "epoch": 0.6354390488468701,
      "grad_norm": 1.0863144397735596,
      "learning_rate": 6.198363605359373e-05,
      "loss": 1.3428,
      "step": 4436
    },
    {
      "epoch": 0.6355822948001719,
      "grad_norm": 1.1797558069229126,
      "learning_rate": 6.194072596161383e-05,
      "loss": 1.3891,
      "step": 4437
    },
    {
      "epoch": 0.6357255407534738,
      "grad_norm": 0.9600929617881775,
      "learning_rate": 6.18978240628538e-05,
      "loss": 1.3947,
      "step": 4438
    },
    {
      "epoch": 0.6358687867067755,
      "grad_norm": 1.1644914150238037,
      "learning_rate": 6.185493036654934e-05,
      "loss": 1.3252,
      "step": 4439
    },
    {
      "epoch": 0.6360120326600773,
      "grad_norm": 1.5045429468154907,
      "learning_rate": 6.181204488193446e-05,
      "loss": 1.5303,
      "step": 4440
    },
    {
      "epoch": 0.6361552786133792,
      "grad_norm": 0.9693203568458557,
      "learning_rate": 6.176916761824129e-05,
      "loss": 1.464,
      "step": 4441
    },
    {
      "epoch": 0.636298524566681,
      "grad_norm": 1.0141738653182983,
      "learning_rate": 6.172629858470031e-05,
      "loss": 1.5092,
      "step": 4442
    },
    {
      "epoch": 0.6364417705199829,
      "grad_norm": 1.377707600593567,
      "learning_rate": 6.168343779054009e-05,
      "loss": 1.4183,
      "step": 4443
    },
    {
      "epoch": 0.6365850164732846,
      "grad_norm": 1.1417499780654907,
      "learning_rate": 6.16405852449875e-05,
      "loss": 1.3681,
      "step": 4444
    },
    {
      "epoch": 0.6367282624265864,
      "grad_norm": 1.6364623308181763,
      "learning_rate": 6.159774095726771e-05,
      "loss": 1.4026,
      "step": 4445
    },
    {
      "epoch": 0.6368715083798883,
      "grad_norm": 1.557728886604309,
      "learning_rate": 6.155490493660394e-05,
      "loss": 1.4005,
      "step": 4446
    },
    {
      "epoch": 0.6370147543331901,
      "grad_norm": 1.1836497783660889,
      "learning_rate": 6.151207719221778e-05,
      "loss": 1.4201,
      "step": 4447
    },
    {
      "epoch": 0.6371580002864919,
      "grad_norm": 1.101131796836853,
      "learning_rate": 6.146925773332899e-05,
      "loss": 1.4411,
      "step": 4448
    },
    {
      "epoch": 0.6373012462397937,
      "grad_norm": 1.0415374040603638,
      "learning_rate": 6.14264465691555e-05,
      "loss": 1.5427,
      "step": 4449
    },
    {
      "epoch": 0.6374444921930955,
      "grad_norm": 1.0730514526367188,
      "learning_rate": 6.138364370891354e-05,
      "loss": 1.5127,
      "step": 4450
    },
    {
      "epoch": 0.6375877381463974,
      "grad_norm": 1.064699411392212,
      "learning_rate": 6.134084916181746e-05,
      "loss": 1.4927,
      "step": 4451
    },
    {
      "epoch": 0.6377309840996992,
      "grad_norm": 1.001696228981018,
      "learning_rate": 6.129806293707989e-05,
      "loss": 1.3577,
      "step": 4452
    },
    {
      "epoch": 0.637874230053001,
      "grad_norm": 0.982567548751831,
      "learning_rate": 6.125528504391167e-05,
      "loss": 1.4494,
      "step": 4453
    },
    {
      "epoch": 0.6380174760063029,
      "grad_norm": 1.213464379310608,
      "learning_rate": 6.121251549152178e-05,
      "loss": 1.3442,
      "step": 4454
    },
    {
      "epoch": 0.6381607219596046,
      "grad_norm": 1.1963921785354614,
      "learning_rate": 6.116975428911744e-05,
      "loss": 1.4634,
      "step": 4455
    },
    {
      "epoch": 0.6383039679129064,
      "grad_norm": 0.9238240122795105,
      "learning_rate": 6.112700144590416e-05,
      "loss": 1.5199,
      "step": 4456
    },
    {
      "epoch": 0.6384472138662083,
      "grad_norm": 1.1160914897918701,
      "learning_rate": 6.108425697108546e-05,
      "loss": 1.5136,
      "step": 4457
    },
    {
      "epoch": 0.6385904598195101,
      "grad_norm": 1.106880784034729,
      "learning_rate": 6.104152087386325e-05,
      "loss": 1.4349,
      "step": 4458
    },
    {
      "epoch": 0.638733705772812,
      "grad_norm": 1.2798396348953247,
      "learning_rate": 6.0998793163437505e-05,
      "loss": 1.4765,
      "step": 4459
    },
    {
      "epoch": 0.6388769517261137,
      "grad_norm": 0.9928169250488281,
      "learning_rate": 6.0956073849006456e-05,
      "loss": 1.4168,
      "step": 4460
    },
    {
      "epoch": 0.6390201976794155,
      "grad_norm": 1.0597409009933472,
      "learning_rate": 6.091336293976655e-05,
      "loss": 1.3773,
      "step": 4461
    },
    {
      "epoch": 0.6391634436327174,
      "grad_norm": 1.1796460151672363,
      "learning_rate": 6.087066044491232e-05,
      "loss": 1.3276,
      "step": 4462
    },
    {
      "epoch": 0.6393066895860192,
      "grad_norm": 1.371848225593567,
      "learning_rate": 6.0827966373636656e-05,
      "loss": 1.3714,
      "step": 4463
    },
    {
      "epoch": 0.6394499355393211,
      "grad_norm": 1.1110001802444458,
      "learning_rate": 6.078528073513041e-05,
      "loss": 1.5113,
      "step": 4464
    },
    {
      "epoch": 0.6395931814926228,
      "grad_norm": 1.0719579458236694,
      "learning_rate": 6.0742603538582835e-05,
      "loss": 1.4711,
      "step": 4465
    },
    {
      "epoch": 0.6397364274459246,
      "grad_norm": 1.136971354484558,
      "learning_rate": 6.069993479318126e-05,
      "loss": 1.3748,
      "step": 4466
    },
    {
      "epoch": 0.6398796733992265,
      "grad_norm": 1.0547744035720825,
      "learning_rate": 6.065727450811115e-05,
      "loss": 1.4879,
      "step": 4467
    },
    {
      "epoch": 0.6400229193525283,
      "grad_norm": 1.0067002773284912,
      "learning_rate": 6.061462269255629e-05,
      "loss": 1.4133,
      "step": 4468
    },
    {
      "epoch": 0.6401661653058301,
      "grad_norm": 1.071082353591919,
      "learning_rate": 6.057197935569854e-05,
      "loss": 1.5086,
      "step": 4469
    },
    {
      "epoch": 0.6403094112591319,
      "grad_norm": 1.0285546779632568,
      "learning_rate": 6.0529344506717935e-05,
      "loss": 1.4429,
      "step": 4470
    },
    {
      "epoch": 0.6404526572124337,
      "grad_norm": 1.017897129058838,
      "learning_rate": 6.0486718154792724e-05,
      "loss": 1.3756,
      "step": 4471
    },
    {
      "epoch": 0.6405959031657356,
      "grad_norm": 1.3890084028244019,
      "learning_rate": 6.044410030909926e-05,
      "loss": 1.4051,
      "step": 4472
    },
    {
      "epoch": 0.6407391491190374,
      "grad_norm": 0.9783182740211487,
      "learning_rate": 6.040149097881214e-05,
      "loss": 1.4816,
      "step": 4473
    },
    {
      "epoch": 0.6408823950723392,
      "grad_norm": 1.0839159488677979,
      "learning_rate": 6.035889017310414e-05,
      "loss": 1.5641,
      "step": 4474
    },
    {
      "epoch": 0.6410256410256411,
      "grad_norm": 1.076854944229126,
      "learning_rate": 6.0316297901146103e-05,
      "loss": 1.513,
      "step": 4475
    },
    {
      "epoch": 0.6411688869789428,
      "grad_norm": 1.2682785987854004,
      "learning_rate": 6.02737141721071e-05,
      "loss": 1.5,
      "step": 4476
    },
    {
      "epoch": 0.6413121329322446,
      "grad_norm": 1.1027729511260986,
      "learning_rate": 6.023113899515438e-05,
      "loss": 1.5581,
      "step": 4477
    },
    {
      "epoch": 0.6414553788855465,
      "grad_norm": 1.2295446395874023,
      "learning_rate": 6.0188572379453276e-05,
      "loss": 1.3184,
      "step": 4478
    },
    {
      "epoch": 0.6415986248388483,
      "grad_norm": 0.8571249842643738,
      "learning_rate": 6.014601433416741e-05,
      "loss": 1.4203,
      "step": 4479
    },
    {
      "epoch": 0.6417418707921502,
      "grad_norm": 1.0159034729003906,
      "learning_rate": 6.010346486845837e-05,
      "loss": 1.3116,
      "step": 4480
    },
    {
      "epoch": 0.6418851167454519,
      "grad_norm": 1.0799195766448975,
      "learning_rate": 6.0060923991486084e-05,
      "loss": 1.2659,
      "step": 4481
    },
    {
      "epoch": 0.6420283626987537,
      "grad_norm": 1.0707416534423828,
      "learning_rate": 6.001839171240853e-05,
      "loss": 1.3262,
      "step": 4482
    },
    {
      "epoch": 0.6421716086520556,
      "grad_norm": 1.0587345361709595,
      "learning_rate": 5.9975868040381844e-05,
      "loss": 1.4757,
      "step": 4483
    },
    {
      "epoch": 0.6423148546053574,
      "grad_norm": 1.024596095085144,
      "learning_rate": 5.9933352984560334e-05,
      "loss": 1.51,
      "step": 4484
    },
    {
      "epoch": 0.6424581005586593,
      "grad_norm": 0.9928714632987976,
      "learning_rate": 5.98908465540964e-05,
      "loss": 1.4354,
      "step": 4485
    },
    {
      "epoch": 0.642601346511961,
      "grad_norm": 1.0642262697219849,
      "learning_rate": 5.9848348758140674e-05,
      "loss": 1.4156,
      "step": 4486
    },
    {
      "epoch": 0.6427445924652628,
      "grad_norm": 0.9713102579116821,
      "learning_rate": 5.980585960584187e-05,
      "loss": 1.5137,
      "step": 4487
    },
    {
      "epoch": 0.6428878384185647,
      "grad_norm": 1.1548058986663818,
      "learning_rate": 5.976337910634684e-05,
      "loss": 1.4686,
      "step": 4488
    },
    {
      "epoch": 0.6430310843718665,
      "grad_norm": 1.0420300960540771,
      "learning_rate": 5.972090726880055e-05,
      "loss": 1.3066,
      "step": 4489
    },
    {
      "epoch": 0.6431743303251684,
      "grad_norm": 1.2319544553756714,
      "learning_rate": 5.967844410234624e-05,
      "loss": 1.3301,
      "step": 4490
    },
    {
      "epoch": 0.6433175762784701,
      "grad_norm": 1.076306939125061,
      "learning_rate": 5.963598961612506e-05,
      "loss": 1.5036,
      "step": 4491
    },
    {
      "epoch": 0.6434608222317719,
      "grad_norm": 1.0033247470855713,
      "learning_rate": 5.9593543819276486e-05,
      "loss": 1.3613,
      "step": 4492
    },
    {
      "epoch": 0.6436040681850738,
      "grad_norm": 0.965067982673645,
      "learning_rate": 5.955110672093801e-05,
      "loss": 1.399,
      "step": 4493
    },
    {
      "epoch": 0.6437473141383756,
      "grad_norm": 0.9814190864562988,
      "learning_rate": 5.950867833024529e-05,
      "loss": 1.464,
      "step": 4494
    },
    {
      "epoch": 0.6438905600916774,
      "grad_norm": 0.9549692273139954,
      "learning_rate": 5.946625865633216e-05,
      "loss": 1.4978,
      "step": 4495
    },
    {
      "epoch": 0.6440338060449793,
      "grad_norm": 1.1571122407913208,
      "learning_rate": 5.942384770833045e-05,
      "loss": 1.6527,
      "step": 4496
    },
    {
      "epoch": 0.644177051998281,
      "grad_norm": 1.2206703424453735,
      "learning_rate": 5.938144549537023e-05,
      "loss": 1.383,
      "step": 4497
    },
    {
      "epoch": 0.6443202979515829,
      "grad_norm": 1.016944169998169,
      "learning_rate": 5.9339052026579654e-05,
      "loss": 1.5799,
      "step": 4498
    },
    {
      "epoch": 0.6444635439048847,
      "grad_norm": 1.0843195915222168,
      "learning_rate": 5.929666731108497e-05,
      "loss": 1.3369,
      "step": 4499
    },
    {
      "epoch": 0.6446067898581865,
      "grad_norm": 0.9656053781509399,
      "learning_rate": 5.9254291358010586e-05,
      "loss": 1.5175,
      "step": 4500
    },
    {
      "epoch": 0.6447500358114884,
      "grad_norm": 1.0099884271621704,
      "learning_rate": 5.9211924176478915e-05,
      "loss": 1.4064,
      "step": 4501
    },
    {
      "epoch": 0.6448932817647901,
      "grad_norm": 0.9601247906684875,
      "learning_rate": 5.9169565775610656e-05,
      "loss": 1.3782,
      "step": 4502
    },
    {
      "epoch": 0.645036527718092,
      "grad_norm": 0.9413478970527649,
      "learning_rate": 5.9127216164524504e-05,
      "loss": 1.5851,
      "step": 4503
    },
    {
      "epoch": 0.6451797736713938,
      "grad_norm": 0.9864361882209778,
      "learning_rate": 5.908487535233725e-05,
      "loss": 1.3832,
      "step": 4504
    },
    {
      "epoch": 0.6453230196246956,
      "grad_norm": 1.1628612279891968,
      "learning_rate": 5.904254334816381e-05,
      "loss": 1.3238,
      "step": 4505
    },
    {
      "epoch": 0.6454662655779975,
      "grad_norm": 1.2943830490112305,
      "learning_rate": 5.900022016111733e-05,
      "loss": 1.5005,
      "step": 4506
    },
    {
      "epoch": 0.6456095115312992,
      "grad_norm": 1.0280368328094482,
      "learning_rate": 5.895790580030879e-05,
      "loss": 1.4286,
      "step": 4507
    },
    {
      "epoch": 0.645752757484601,
      "grad_norm": 1.1297389268875122,
      "learning_rate": 5.891560027484753e-05,
      "loss": 1.3872,
      "step": 4508
    },
    {
      "epoch": 0.6458960034379029,
      "grad_norm": 1.0245985984802246,
      "learning_rate": 5.8873303593840846e-05,
      "loss": 1.5299,
      "step": 4509
    },
    {
      "epoch": 0.6460392493912047,
      "grad_norm": 1.1804975271224976,
      "learning_rate": 5.883101576639415e-05,
      "loss": 1.3902,
      "step": 4510
    },
    {
      "epoch": 0.6461824953445066,
      "grad_norm": 0.9928570985794067,
      "learning_rate": 5.878873680161101e-05,
      "loss": 1.4612,
      "step": 4511
    },
    {
      "epoch": 0.6463257412978083,
      "grad_norm": 1.0377628803253174,
      "learning_rate": 5.8746466708592986e-05,
      "loss": 1.4976,
      "step": 4512
    },
    {
      "epoch": 0.6464689872511101,
      "grad_norm": 1.104115605354309,
      "learning_rate": 5.870420549643987e-05,
      "loss": 1.4478,
      "step": 4513
    },
    {
      "epoch": 0.646612233204412,
      "grad_norm": 1.1387014389038086,
      "learning_rate": 5.866195317424934e-05,
      "loss": 1.4754,
      "step": 4514
    },
    {
      "epoch": 0.6467554791577138,
      "grad_norm": 0.9884274005889893,
      "learning_rate": 5.8619709751117344e-05,
      "loss": 1.4509,
      "step": 4515
    },
    {
      "epoch": 0.6468987251110156,
      "grad_norm": 1.0251950025558472,
      "learning_rate": 5.8577475236137855e-05,
      "loss": 1.3572,
      "step": 4516
    },
    {
      "epoch": 0.6470419710643174,
      "grad_norm": 1.1747344732284546,
      "learning_rate": 5.853524963840289e-05,
      "loss": 1.4893,
      "step": 4517
    },
    {
      "epoch": 0.6471852170176192,
      "grad_norm": 1.1077133417129517,
      "learning_rate": 5.849303296700257e-05,
      "loss": 1.4538,
      "step": 4518
    },
    {
      "epoch": 0.6473284629709211,
      "grad_norm": 1.3253967761993408,
      "learning_rate": 5.845082523102514e-05,
      "loss": 1.3669,
      "step": 4519
    },
    {
      "epoch": 0.6474717089242229,
      "grad_norm": 0.9842604994773865,
      "learning_rate": 5.8408626439556845e-05,
      "loss": 1.4237,
      "step": 4520
    },
    {
      "epoch": 0.6476149548775247,
      "grad_norm": 1.1810357570648193,
      "learning_rate": 5.8366436601682084e-05,
      "loss": 1.3325,
      "step": 4521
    },
    {
      "epoch": 0.6477582008308266,
      "grad_norm": 1.1851726770401,
      "learning_rate": 5.832425572648317e-05,
      "loss": 1.3467,
      "step": 4522
    },
    {
      "epoch": 0.6479014467841283,
      "grad_norm": 1.0546855926513672,
      "learning_rate": 5.828208382304072e-05,
      "loss": 1.3371,
      "step": 4523
    },
    {
      "epoch": 0.6480446927374302,
      "grad_norm": 0.9989160895347595,
      "learning_rate": 5.823992090043333e-05,
      "loss": 1.3331,
      "step": 4524
    },
    {
      "epoch": 0.648187938690732,
      "grad_norm": 1.0630459785461426,
      "learning_rate": 5.819776696773751e-05,
      "loss": 1.4482,
      "step": 4525
    },
    {
      "epoch": 0.6483311846440338,
      "grad_norm": 1.0202382802963257,
      "learning_rate": 5.815562203402798e-05,
      "loss": 1.6199,
      "step": 4526
    },
    {
      "epoch": 0.6484744305973357,
      "grad_norm": 1.162812352180481,
      "learning_rate": 5.8113486108377615e-05,
      "loss": 1.2535,
      "step": 4527
    },
    {
      "epoch": 0.6486176765506374,
      "grad_norm": 1.006460428237915,
      "learning_rate": 5.8071359199857114e-05,
      "loss": 1.4937,
      "step": 4528
    },
    {
      "epoch": 0.6487609225039392,
      "grad_norm": 1.1499429941177368,
      "learning_rate": 5.802924131753542e-05,
      "loss": 1.6045,
      "step": 4529
    },
    {
      "epoch": 0.6489041684572411,
      "grad_norm": 1.06486177444458,
      "learning_rate": 5.798713247047944e-05,
      "loss": 1.4043,
      "step": 4530
    },
    {
      "epoch": 0.6490474144105429,
      "grad_norm": 1.292765736579895,
      "learning_rate": 5.79450326677542e-05,
      "loss": 1.5262,
      "step": 4531
    },
    {
      "epoch": 0.6491906603638448,
      "grad_norm": 1.0108109712600708,
      "learning_rate": 5.790294191842276e-05,
      "loss": 1.3823,
      "step": 4532
    },
    {
      "epoch": 0.6493339063171465,
      "grad_norm": 0.9800224304199219,
      "learning_rate": 5.786086023154609e-05,
      "loss": 1.4338,
      "step": 4533
    },
    {
      "epoch": 0.6494771522704483,
      "grad_norm": 0.8256798386573792,
      "learning_rate": 5.781878761618349e-05,
      "loss": 1.2739,
      "step": 4534
    },
    {
      "epoch": 0.6496203982237502,
      "grad_norm": 1.148642897605896,
      "learning_rate": 5.777672408139212e-05,
      "loss": 1.4234,
      "step": 4535
    },
    {
      "epoch": 0.649763644177052,
      "grad_norm": 1.1265605688095093,
      "learning_rate": 5.773466963622716e-05,
      "loss": 1.4703,
      "step": 4536
    },
    {
      "epoch": 0.6499068901303539,
      "grad_norm": 1.0741684436798096,
      "learning_rate": 5.7692624289741914e-05,
      "loss": 1.4628,
      "step": 4537
    },
    {
      "epoch": 0.6500501360836556,
      "grad_norm": 1.138808250427246,
      "learning_rate": 5.765058805098773e-05,
      "loss": 1.3964,
      "step": 4538
    },
    {
      "epoch": 0.6501933820369574,
      "grad_norm": 1.1394202709197998,
      "learning_rate": 5.7608560929013946e-05,
      "loss": 1.2188,
      "step": 4539
    },
    {
      "epoch": 0.6503366279902593,
      "grad_norm": 0.9526465535163879,
      "learning_rate": 5.756654293286796e-05,
      "loss": 1.4596,
      "step": 4540
    },
    {
      "epoch": 0.6504798739435611,
      "grad_norm": 1.0236263275146484,
      "learning_rate": 5.752453407159522e-05,
      "loss": 1.5331,
      "step": 4541
    },
    {
      "epoch": 0.650623119896863,
      "grad_norm": 1.0928754806518555,
      "learning_rate": 5.7482534354239225e-05,
      "loss": 1.4247,
      "step": 4542
    },
    {
      "epoch": 0.6507663658501648,
      "grad_norm": 1.0806907415390015,
      "learning_rate": 5.74405437898414e-05,
      "loss": 1.4387,
      "step": 4543
    },
    {
      "epoch": 0.6509096118034665,
      "grad_norm": 1.1278156042099,
      "learning_rate": 5.739856238744129e-05,
      "loss": 1.3394,
      "step": 4544
    },
    {
      "epoch": 0.6510528577567684,
      "grad_norm": 1.2492507696151733,
      "learning_rate": 5.735659015607655e-05,
      "loss": 1.3531,
      "step": 4545
    },
    {
      "epoch": 0.6511961037100702,
      "grad_norm": 1.0405765771865845,
      "learning_rate": 5.731462710478264e-05,
      "loss": 1.4454,
      "step": 4546
    },
    {
      "epoch": 0.651339349663372,
      "grad_norm": 1.0472185611724854,
      "learning_rate": 5.7272673242593174e-05,
      "loss": 1.4223,
      "step": 4547
    },
    {
      "epoch": 0.6514825956166739,
      "grad_norm": 1.1658077239990234,
      "learning_rate": 5.723072857853992e-05,
      "loss": 1.4263,
      "step": 4548
    },
    {
      "epoch": 0.6516258415699756,
      "grad_norm": 1.0355273485183716,
      "learning_rate": 5.7188793121652374e-05,
      "loss": 1.3195,
      "step": 4549
    },
    {
      "epoch": 0.6517690875232774,
      "grad_norm": 1.2060174942016602,
      "learning_rate": 5.714686688095825e-05,
      "loss": 1.3627,
      "step": 4550
    },
    {
      "epoch": 0.6519123334765793,
      "grad_norm": 1.2119323015213013,
      "learning_rate": 5.7104949865483246e-05,
      "loss": 1.3354,
      "step": 4551
    },
    {
      "epoch": 0.6520555794298811,
      "grad_norm": 0.9646314978599548,
      "learning_rate": 5.706304208425105e-05,
      "loss": 1.3961,
      "step": 4552
    },
    {
      "epoch": 0.652198825383183,
      "grad_norm": 1.051529049873352,
      "learning_rate": 5.702114354628341e-05,
      "loss": 1.4272,
      "step": 4553
    },
    {
      "epoch": 0.6523420713364847,
      "grad_norm": 0.9479196667671204,
      "learning_rate": 5.697925426059991e-05,
      "loss": 1.2748,
      "step": 4554
    },
    {
      "epoch": 0.6524853172897865,
      "grad_norm": 1.058199405670166,
      "learning_rate": 5.6937374236218424e-05,
      "loss": 1.3929,
      "step": 4555
    },
    {
      "epoch": 0.6526285632430884,
      "grad_norm": 1.1558740139007568,
      "learning_rate": 5.6895503482154666e-05,
      "loss": 1.2779,
      "step": 4556
    },
    {
      "epoch": 0.6527718091963902,
      "grad_norm": 1.0622003078460693,
      "learning_rate": 5.6853642007422294e-05,
      "loss": 1.3791,
      "step": 4557
    },
    {
      "epoch": 0.6529150551496921,
      "grad_norm": 1.126219391822815,
      "learning_rate": 5.681178982103309e-05,
      "loss": 1.4501,
      "step": 4558
    },
    {
      "epoch": 0.6530583011029938,
      "grad_norm": 1.1512702703475952,
      "learning_rate": 5.6769946931996795e-05,
      "loss": 1.3336,
      "step": 4559
    },
    {
      "epoch": 0.6532015470562956,
      "grad_norm": 1.2613657712936401,
      "learning_rate": 5.672811334932116e-05,
      "loss": 1.2297,
      "step": 4560
    },
    {
      "epoch": 0.6533447930095975,
      "grad_norm": 1.188358187675476,
      "learning_rate": 5.668628908201189e-05,
      "loss": 1.4524,
      "step": 4561
    },
    {
      "epoch": 0.6534880389628993,
      "grad_norm": 0.9419224858283997,
      "learning_rate": 5.6644474139072746e-05,
      "loss": 1.3315,
      "step": 4562
    },
    {
      "epoch": 0.6536312849162011,
      "grad_norm": 1.0754724740982056,
      "learning_rate": 5.660266852950547e-05,
      "loss": 1.4844,
      "step": 4563
    },
    {
      "epoch": 0.653774530869503,
      "grad_norm": 1.0817204713821411,
      "learning_rate": 5.6560872262309704e-05,
      "loss": 1.4058,
      "step": 4564
    },
    {
      "epoch": 0.6539177768228047,
      "grad_norm": 1.0191084146499634,
      "learning_rate": 5.651908534648315e-05,
      "loss": 1.523,
      "step": 4565
    },
    {
      "epoch": 0.6540610227761066,
      "grad_norm": 0.9848802089691162,
      "learning_rate": 5.647730779102161e-05,
      "loss": 1.2908,
      "step": 4566
    },
    {
      "epoch": 0.6542042687294084,
      "grad_norm": 1.0660760402679443,
      "learning_rate": 5.6435539604918654e-05,
      "loss": 1.4286,
      "step": 4567
    },
    {
      "epoch": 0.6543475146827102,
      "grad_norm": 1.0634881258010864,
      "learning_rate": 5.639378079716595e-05,
      "loss": 1.3289,
      "step": 4568
    },
    {
      "epoch": 0.6544907606360121,
      "grad_norm": 1.004914402961731,
      "learning_rate": 5.635203137675318e-05,
      "loss": 1.4854,
      "step": 4569
    },
    {
      "epoch": 0.6546340065893138,
      "grad_norm": 0.9726465344429016,
      "learning_rate": 5.631029135266791e-05,
      "loss": 1.4301,
      "step": 4570
    },
    {
      "epoch": 0.6547772525426157,
      "grad_norm": 1.1482840776443481,
      "learning_rate": 5.6268560733895816e-05,
      "loss": 1.3698,
      "step": 4571
    },
    {
      "epoch": 0.6549204984959175,
      "grad_norm": 1.2183526754379272,
      "learning_rate": 5.6226839529420314e-05,
      "loss": 1.4579,
      "step": 4572
    },
    {
      "epoch": 0.6550637444492193,
      "grad_norm": 1.2847543954849243,
      "learning_rate": 5.618512774822311e-05,
      "loss": 1.3118,
      "step": 4573
    },
    {
      "epoch": 0.6552069904025212,
      "grad_norm": 1.0733232498168945,
      "learning_rate": 5.6143425399283664e-05,
      "loss": 1.3983,
      "step": 4574
    },
    {
      "epoch": 0.6553502363558229,
      "grad_norm": 1.1844935417175293,
      "learning_rate": 5.610173249157942e-05,
      "loss": 1.435,
      "step": 4575
    },
    {
      "epoch": 0.6554934823091247,
      "grad_norm": 1.1102830171585083,
      "learning_rate": 5.6060049034085815e-05,
      "loss": 1.4206,
      "step": 4576
    },
    {
      "epoch": 0.6556367282624266,
      "grad_norm": 1.1408616304397583,
      "learning_rate": 5.6018375035776406e-05,
      "loss": 1.422,
      "step": 4577
    },
    {
      "epoch": 0.6557799742157284,
      "grad_norm": 1.0308042764663696,
      "learning_rate": 5.597671050562241e-05,
      "loss": 1.5397,
      "step": 4578
    },
    {
      "epoch": 0.6559232201690303,
      "grad_norm": 1.3452590703964233,
      "learning_rate": 5.5935055452593254e-05,
      "loss": 1.2149,
      "step": 4579
    },
    {
      "epoch": 0.656066466122332,
      "grad_norm": 1.1442689895629883,
      "learning_rate": 5.5893409885656214e-05,
      "loss": 1.4176,
      "step": 4580
    },
    {
      "epoch": 0.6562097120756338,
      "grad_norm": 1.1220859289169312,
      "learning_rate": 5.5851773813776556e-05,
      "loss": 1.5515,
      "step": 4581
    },
    {
      "epoch": 0.6563529580289357,
      "grad_norm": 1.0263195037841797,
      "learning_rate": 5.5810147245917535e-05,
      "loss": 1.4533,
      "step": 4582
    },
    {
      "epoch": 0.6564962039822375,
      "grad_norm": 1.2668074369430542,
      "learning_rate": 5.5768530191040206e-05,
      "loss": 1.3813,
      "step": 4583
    },
    {
      "epoch": 0.6566394499355394,
      "grad_norm": 1.0619176626205444,
      "learning_rate": 5.57269226581038e-05,
      "loss": 1.3198,
      "step": 4584
    },
    {
      "epoch": 0.6567826958888412,
      "grad_norm": 1.3197219371795654,
      "learning_rate": 5.56853246560654e-05,
      "loss": 1.4631,
      "step": 4585
    },
    {
      "epoch": 0.6569259418421429,
      "grad_norm": 1.0135531425476074,
      "learning_rate": 5.564373619387995e-05,
      "loss": 1.4171,
      "step": 4586
    },
    {
      "epoch": 0.6570691877954448,
      "grad_norm": 1.0145251750946045,
      "learning_rate": 5.5602157280500446e-05,
      "loss": 1.3297,
      "step": 4587
    },
    {
      "epoch": 0.6572124337487466,
      "grad_norm": 0.9064403772354126,
      "learning_rate": 5.556058792487779e-05,
      "loss": 1.4661,
      "step": 4588
    },
    {
      "epoch": 0.6573556797020484,
      "grad_norm": 1.383042812347412,
      "learning_rate": 5.551902813596087e-05,
      "loss": 1.268,
      "step": 4589
    },
    {
      "epoch": 0.6574989256553503,
      "grad_norm": 0.9954293370246887,
      "learning_rate": 5.5477477922696465e-05,
      "loss": 1.4083,
      "step": 4590
    },
    {
      "epoch": 0.657642171608652,
      "grad_norm": 1.0450252294540405,
      "learning_rate": 5.543593729402927e-05,
      "loss": 1.4777,
      "step": 4591
    },
    {
      "epoch": 0.6577854175619539,
      "grad_norm": 0.8920850157737732,
      "learning_rate": 5.5394406258902054e-05,
      "loss": 1.4769,
      "step": 4592
    },
    {
      "epoch": 0.6579286635152557,
      "grad_norm": 1.0448344945907593,
      "learning_rate": 5.5352884826255317e-05,
      "loss": 1.423,
      "step": 4593
    },
    {
      "epoch": 0.6580719094685575,
      "grad_norm": 1.1138440370559692,
      "learning_rate": 5.531137300502758e-05,
      "loss": 1.5368,
      "step": 4594
    },
    {
      "epoch": 0.6582151554218594,
      "grad_norm": 1.153704047203064,
      "learning_rate": 5.526987080415543e-05,
      "loss": 1.4418,
      "step": 4595
    },
    {
      "epoch": 0.6583584013751611,
      "grad_norm": 1.075050711631775,
      "learning_rate": 5.522837823257317e-05,
      "loss": 1.5503,
      "step": 4596
    },
    {
      "epoch": 0.658501647328463,
      "grad_norm": 1.0047483444213867,
      "learning_rate": 5.518689529921315e-05,
      "loss": 1.5202,
      "step": 4597
    },
    {
      "epoch": 0.6586448932817648,
      "grad_norm": 1.3084261417388916,
      "learning_rate": 5.514542201300563e-05,
      "loss": 1.4243,
      "step": 4598
    },
    {
      "epoch": 0.6587881392350666,
      "grad_norm": 1.054765224456787,
      "learning_rate": 5.5103958382878765e-05,
      "loss": 1.4511,
      "step": 4599
    },
    {
      "epoch": 0.6589313851883685,
      "grad_norm": 1.2251580953598022,
      "learning_rate": 5.5062504417758684e-05,
      "loss": 1.5065,
      "step": 4600
    },
    {
      "epoch": 0.6590746311416702,
      "grad_norm": 1.1300359964370728,
      "learning_rate": 5.502106012656931e-05,
      "loss": 1.3949,
      "step": 4601
    },
    {
      "epoch": 0.659217877094972,
      "grad_norm": 1.0543614625930786,
      "learning_rate": 5.497962551823266e-05,
      "loss": 1.5467,
      "step": 4602
    },
    {
      "epoch": 0.6593611230482739,
      "grad_norm": 1.142508625984192,
      "learning_rate": 5.493820060166861e-05,
      "loss": 1.3815,
      "step": 4603
    },
    {
      "epoch": 0.6595043690015757,
      "grad_norm": 1.1749300956726074,
      "learning_rate": 5.4896785385794815e-05,
      "loss": 1.2552,
      "step": 4604
    },
    {
      "epoch": 0.6596476149548776,
      "grad_norm": 1.2161883115768433,
      "learning_rate": 5.485537987952696e-05,
      "loss": 1.5366,
      "step": 4605
    },
    {
      "epoch": 0.6597908609081793,
      "grad_norm": 1.0623385906219482,
      "learning_rate": 5.4813984091778734e-05,
      "loss": 1.3996,
      "step": 4606
    },
    {
      "epoch": 0.6599341068614811,
      "grad_norm": 1.2170624732971191,
      "learning_rate": 5.4772598031461507e-05,
      "loss": 1.3891,
      "step": 4607
    },
    {
      "epoch": 0.660077352814783,
      "grad_norm": 0.9561026692390442,
      "learning_rate": 5.473122170748472e-05,
      "loss": 1.4362,
      "step": 4608
    },
    {
      "epoch": 0.6602205987680848,
      "grad_norm": 1.2127102613449097,
      "learning_rate": 5.4689855128755686e-05,
      "loss": 1.4595,
      "step": 4609
    },
    {
      "epoch": 0.6603638447213867,
      "grad_norm": 1.1160438060760498,
      "learning_rate": 5.4648498304179585e-05,
      "loss": 1.2667,
      "step": 4610
    },
    {
      "epoch": 0.6605070906746885,
      "grad_norm": 0.9959889054298401,
      "learning_rate": 5.4607151242659524e-05,
      "loss": 1.354,
      "step": 4611
    },
    {
      "epoch": 0.6606503366279902,
      "grad_norm": 1.1645419597625732,
      "learning_rate": 5.45658139530965e-05,
      "loss": 1.5773,
      "step": 4612
    },
    {
      "epoch": 0.6607935825812921,
      "grad_norm": 1.195703148841858,
      "learning_rate": 5.452448644438946e-05,
      "loss": 1.5745,
      "step": 4613
    },
    {
      "epoch": 0.6609368285345939,
      "grad_norm": 1.0137521028518677,
      "learning_rate": 5.4483168725435086e-05,
      "loss": 1.3337,
      "step": 4614
    },
    {
      "epoch": 0.6610800744878957,
      "grad_norm": 1.0424811840057373,
      "learning_rate": 5.444186080512809e-05,
      "loss": 1.3422,
      "step": 4615
    },
    {
      "epoch": 0.6612233204411976,
      "grad_norm": 0.9815017580986023,
      "learning_rate": 5.4400562692361145e-05,
      "loss": 1.3115,
      "step": 4616
    },
    {
      "epoch": 0.6613665663944993,
      "grad_norm": 1.0448380708694458,
      "learning_rate": 5.435927439602462e-05,
      "loss": 1.5276,
      "step": 4617
    },
    {
      "epoch": 0.6615098123478012,
      "grad_norm": 1.0571798086166382,
      "learning_rate": 5.431799592500686e-05,
      "loss": 1.3613,
      "step": 4618
    },
    {
      "epoch": 0.661653058301103,
      "grad_norm": 1.1727370023727417,
      "learning_rate": 5.427672728819414e-05,
      "loss": 1.4281,
      "step": 4619
    },
    {
      "epoch": 0.6617963042544048,
      "grad_norm": 1.043731451034546,
      "learning_rate": 5.423546849447055e-05,
      "loss": 1.4378,
      "step": 4620
    },
    {
      "epoch": 0.6619395502077067,
      "grad_norm": 0.9515613913536072,
      "learning_rate": 5.419421955271815e-05,
      "loss": 1.4298,
      "step": 4621
    },
    {
      "epoch": 0.6620827961610084,
      "grad_norm": 1.2299749851226807,
      "learning_rate": 5.415298047181667e-05,
      "loss": 1.6592,
      "step": 4622
    },
    {
      "epoch": 0.6622260421143102,
      "grad_norm": 1.2164242267608643,
      "learning_rate": 5.4111751260644e-05,
      "loss": 1.4575,
      "step": 4623
    },
    {
      "epoch": 0.6623692880676121,
      "grad_norm": 0.9539395570755005,
      "learning_rate": 5.407053192807576e-05,
      "loss": 1.258,
      "step": 4624
    },
    {
      "epoch": 0.6625125340209139,
      "grad_norm": 1.069482445716858,
      "learning_rate": 5.40293224829854e-05,
      "loss": 1.3364,
      "step": 4625
    },
    {
      "epoch": 0.6626557799742158,
      "grad_norm": 1.0834589004516602,
      "learning_rate": 5.398812293424426e-05,
      "loss": 1.3881,
      "step": 4626
    },
    {
      "epoch": 0.6627990259275175,
      "grad_norm": 0.9015188217163086,
      "learning_rate": 5.394693329072171e-05,
      "loss": 1.3456,
      "step": 4627
    },
    {
      "epoch": 0.6629422718808193,
      "grad_norm": 1.1651769876480103,
      "learning_rate": 5.390575356128474e-05,
      "loss": 1.422,
      "step": 4628
    },
    {
      "epoch": 0.6630855178341212,
      "grad_norm": 1.197929859161377,
      "learning_rate": 5.386458375479839e-05,
      "loss": 1.3686,
      "step": 4629
    },
    {
      "epoch": 0.663228763787423,
      "grad_norm": 1.0539158582687378,
      "learning_rate": 5.382342388012547e-05,
      "loss": 1.6378,
      "step": 4630
    },
    {
      "epoch": 0.6633720097407249,
      "grad_norm": 0.9805400371551514,
      "learning_rate": 5.37822739461267e-05,
      "loss": 1.3745,
      "step": 4631
    },
    {
      "epoch": 0.6635152556940267,
      "grad_norm": 1.1135749816894531,
      "learning_rate": 5.3741133961660686e-05,
      "loss": 1.3122,
      "step": 4632
    },
    {
      "epoch": 0.6636585016473284,
      "grad_norm": 1.2728681564331055,
      "learning_rate": 5.370000393558371e-05,
      "loss": 1.6586,
      "step": 4633
    },
    {
      "epoch": 0.6638017476006303,
      "grad_norm": 1.1571128368377686,
      "learning_rate": 5.365888387675018e-05,
      "loss": 1.3125,
      "step": 4634
    },
    {
      "epoch": 0.6639449935539321,
      "grad_norm": 1.2323644161224365,
      "learning_rate": 5.361777379401223e-05,
      "loss": 1.4344,
      "step": 4635
    },
    {
      "epoch": 0.664088239507234,
      "grad_norm": 0.979188859462738,
      "learning_rate": 5.357667369621977e-05,
      "loss": 1.482,
      "step": 4636
    },
    {
      "epoch": 0.6642314854605358,
      "grad_norm": 0.9748269319534302,
      "learning_rate": 5.353558359222065e-05,
      "loss": 1.415,
      "step": 4637
    },
    {
      "epoch": 0.6643747314138375,
      "grad_norm": 1.4010051488876343,
      "learning_rate": 5.349450349086057e-05,
      "loss": 1.4324,
      "step": 4638
    },
    {
      "epoch": 0.6645179773671394,
      "grad_norm": 1.0288825035095215,
      "learning_rate": 5.3453433400983055e-05,
      "loss": 1.409,
      "step": 4639
    },
    {
      "epoch": 0.6646612233204412,
      "grad_norm": 0.9494627714157104,
      "learning_rate": 5.3412373331429474e-05,
      "loss": 1.4856,
      "step": 4640
    },
    {
      "epoch": 0.664804469273743,
      "grad_norm": 1.1552248001098633,
      "learning_rate": 5.337132329103907e-05,
      "loss": 1.4849,
      "step": 4641
    },
    {
      "epoch": 0.6649477152270449,
      "grad_norm": 1.1238632202148438,
      "learning_rate": 5.3330283288648906e-05,
      "loss": 1.5129,
      "step": 4642
    },
    {
      "epoch": 0.6650909611803466,
      "grad_norm": 1.0481737852096558,
      "learning_rate": 5.3289253333093826e-05,
      "loss": 1.5223,
      "step": 4643
    },
    {
      "epoch": 0.6652342071336484,
      "grad_norm": 1.0552860498428345,
      "learning_rate": 5.324823343320654e-05,
      "loss": 1.4586,
      "step": 4644
    },
    {
      "epoch": 0.6653774530869503,
      "grad_norm": 1.1617873907089233,
      "learning_rate": 5.320722359781776e-05,
      "loss": 1.3824,
      "step": 4645
    },
    {
      "epoch": 0.6655206990402521,
      "grad_norm": 0.9579392075538635,
      "learning_rate": 5.3166223835755736e-05,
      "loss": 1.5059,
      "step": 4646
    },
    {
      "epoch": 0.665663944993554,
      "grad_norm": 1.015425682067871,
      "learning_rate": 5.312523415584678e-05,
      "loss": 1.4629,
      "step": 4647
    },
    {
      "epoch": 0.6658071909468557,
      "grad_norm": 1.1427415609359741,
      "learning_rate": 5.308425456691495e-05,
      "loss": 1.3648,
      "step": 4648
    },
    {
      "epoch": 0.6659504369001575,
      "grad_norm": 0.9164599776268005,
      "learning_rate": 5.3043285077782114e-05,
      "loss": 1.3849,
      "step": 4649
    },
    {
      "epoch": 0.6660936828534594,
      "grad_norm": 1.0506738424301147,
      "learning_rate": 5.300232569726804e-05,
      "loss": 1.4166,
      "step": 4650
    },
    {
      "epoch": 0.6662369288067612,
      "grad_norm": 1.1794301271438599,
      "learning_rate": 5.2961376434190144e-05,
      "loss": 1.2502,
      "step": 4651
    },
    {
      "epoch": 0.6663801747600631,
      "grad_norm": 0.9880931377410889,
      "learning_rate": 5.292043729736394e-05,
      "loss": 1.4938,
      "step": 4652
    },
    {
      "epoch": 0.6665234207133649,
      "grad_norm": 1.02407705783844,
      "learning_rate": 5.2879508295602575e-05,
      "loss": 1.4923,
      "step": 4653
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 1.1631250381469727,
      "learning_rate": 5.283858943771698e-05,
      "loss": 1.4391,
      "step": 4654
    },
    {
      "epoch": 0.6668099126199685,
      "grad_norm": 1.2295377254486084,
      "learning_rate": 5.2797680732515986e-05,
      "loss": 1.5335,
      "step": 4655
    },
    {
      "epoch": 0.6669531585732703,
      "grad_norm": 0.9972642660140991,
      "learning_rate": 5.275678218880632e-05,
      "loss": 1.4697,
      "step": 4656
    },
    {
      "epoch": 0.6670964045265722,
      "grad_norm": 1.0502122640609741,
      "learning_rate": 5.2715893815392325e-05,
      "loss": 1.3583,
      "step": 4657
    },
    {
      "epoch": 0.667239650479874,
      "grad_norm": 1.0837233066558838,
      "learning_rate": 5.26750156210763e-05,
      "loss": 1.4316,
      "step": 4658
    },
    {
      "epoch": 0.6673828964331757,
      "grad_norm": 1.1484932899475098,
      "learning_rate": 5.2634147614658294e-05,
      "loss": 1.4295,
      "step": 4659
    },
    {
      "epoch": 0.6675261423864776,
      "grad_norm": 1.1339203119277954,
      "learning_rate": 5.259328980493618e-05,
      "loss": 1.189,
      "step": 4660
    },
    {
      "epoch": 0.6676693883397794,
      "grad_norm": 0.9505112767219543,
      "learning_rate": 5.255244220070567e-05,
      "loss": 1.4944,
      "step": 4661
    },
    {
      "epoch": 0.6678126342930812,
      "grad_norm": 1.0537834167480469,
      "learning_rate": 5.251160481076016e-05,
      "loss": 1.2769,
      "step": 4662
    },
    {
      "epoch": 0.6679558802463831,
      "grad_norm": 1.139582633972168,
      "learning_rate": 5.247077764389099e-05,
      "loss": 1.4404,
      "step": 4663
    },
    {
      "epoch": 0.6680991261996848,
      "grad_norm": 1.0137072801589966,
      "learning_rate": 5.242996070888728e-05,
      "loss": 1.3511,
      "step": 4664
    },
    {
      "epoch": 0.6682423721529867,
      "grad_norm": 1.0986104011535645,
      "learning_rate": 5.2389154014535814e-05,
      "loss": 1.3368,
      "step": 4665
    },
    {
      "epoch": 0.6683856181062885,
      "grad_norm": 1.0190446376800537,
      "learning_rate": 5.23483575696213e-05,
      "loss": 1.4032,
      "step": 4666
    },
    {
      "epoch": 0.6685288640595903,
      "grad_norm": 1.0660560131072998,
      "learning_rate": 5.2307571382926215e-05,
      "loss": 1.3215,
      "step": 4667
    },
    {
      "epoch": 0.6686721100128922,
      "grad_norm": 1.0610193014144897,
      "learning_rate": 5.226679546323079e-05,
      "loss": 1.3175,
      "step": 4668
    },
    {
      "epoch": 0.6688153559661939,
      "grad_norm": 1.042181372642517,
      "learning_rate": 5.222602981931309e-05,
      "loss": 1.3867,
      "step": 4669
    },
    {
      "epoch": 0.6689586019194957,
      "grad_norm": 0.9854345917701721,
      "learning_rate": 5.2185274459948965e-05,
      "loss": 1.4576,
      "step": 4670
    },
    {
      "epoch": 0.6691018478727976,
      "grad_norm": 0.9518062472343445,
      "learning_rate": 5.2144529393912036e-05,
      "loss": 1.4949,
      "step": 4671
    },
    {
      "epoch": 0.6692450938260994,
      "grad_norm": 0.9983185529708862,
      "learning_rate": 5.210379462997364e-05,
      "loss": 1.6027,
      "step": 4672
    },
    {
      "epoch": 0.6693883397794013,
      "grad_norm": 1.495338797569275,
      "learning_rate": 5.206307017690302e-05,
      "loss": 1.5165,
      "step": 4673
    },
    {
      "epoch": 0.6695315857327031,
      "grad_norm": 0.9914500713348389,
      "learning_rate": 5.202235604346719e-05,
      "loss": 1.3573,
      "step": 4674
    },
    {
      "epoch": 0.6696748316860048,
      "grad_norm": 1.0878039598464966,
      "learning_rate": 5.19816522384308e-05,
      "loss": 1.4245,
      "step": 4675
    },
    {
      "epoch": 0.6698180776393067,
      "grad_norm": 1.0673552751541138,
      "learning_rate": 5.1940958770556383e-05,
      "loss": 1.3157,
      "step": 4676
    },
    {
      "epoch": 0.6699613235926085,
      "grad_norm": 1.0680954456329346,
      "learning_rate": 5.190027564860433e-05,
      "loss": 1.4977,
      "step": 4677
    },
    {
      "epoch": 0.6701045695459104,
      "grad_norm": 0.9051030874252319,
      "learning_rate": 5.185960288133261e-05,
      "loss": 1.355,
      "step": 4678
    },
    {
      "epoch": 0.6702478154992122,
      "grad_norm": 1.0885728597640991,
      "learning_rate": 5.181894047749711e-05,
      "loss": 1.5478,
      "step": 4679
    },
    {
      "epoch": 0.6703910614525139,
      "grad_norm": 1.2315906286239624,
      "learning_rate": 5.177828844585142e-05,
      "loss": 1.2399,
      "step": 4680
    },
    {
      "epoch": 0.6705343074058158,
      "grad_norm": 1.2804405689239502,
      "learning_rate": 5.173764679514693e-05,
      "loss": 1.2929,
      "step": 4681
    },
    {
      "epoch": 0.6706775533591176,
      "grad_norm": 1.2457267045974731,
      "learning_rate": 5.169701553413282e-05,
      "loss": 1.4978,
      "step": 4682
    },
    {
      "epoch": 0.6708207993124194,
      "grad_norm": 1.1037180423736572,
      "learning_rate": 5.1656394671555875e-05,
      "loss": 1.3802,
      "step": 4683
    },
    {
      "epoch": 0.6709640452657213,
      "grad_norm": 1.25758695602417,
      "learning_rate": 5.161578421616087e-05,
      "loss": 1.5493,
      "step": 4684
    },
    {
      "epoch": 0.671107291219023,
      "grad_norm": 0.997965931892395,
      "learning_rate": 5.157518417669023e-05,
      "loss": 1.1578,
      "step": 4685
    },
    {
      "epoch": 0.6712505371723249,
      "grad_norm": 1.0033360719680786,
      "learning_rate": 5.1534594561884086e-05,
      "loss": 1.3032,
      "step": 4686
    },
    {
      "epoch": 0.6713937831256267,
      "grad_norm": 1.1388763189315796,
      "learning_rate": 5.1494015380480396e-05,
      "loss": 1.1793,
      "step": 4687
    },
    {
      "epoch": 0.6715370290789285,
      "grad_norm": 1.1157748699188232,
      "learning_rate": 5.1453446641214854e-05,
      "loss": 1.2793,
      "step": 4688
    },
    {
      "epoch": 0.6716802750322304,
      "grad_norm": 1.1203961372375488,
      "learning_rate": 5.141288835282092e-05,
      "loss": 1.4593,
      "step": 4689
    },
    {
      "epoch": 0.6718235209855321,
      "grad_norm": 1.197005271911621,
      "learning_rate": 5.137234052402977e-05,
      "loss": 1.4407,
      "step": 4690
    },
    {
      "epoch": 0.671966766938834,
      "grad_norm": 1.1796092987060547,
      "learning_rate": 5.133180316357036e-05,
      "loss": 1.308,
      "step": 4691
    },
    {
      "epoch": 0.6721100128921358,
      "grad_norm": 1.0350306034088135,
      "learning_rate": 5.129127628016941e-05,
      "loss": 1.2719,
      "step": 4692
    },
    {
      "epoch": 0.6722532588454376,
      "grad_norm": 1.1337361335754395,
      "learning_rate": 5.1250759882551284e-05,
      "loss": 1.3773,
      "step": 4693
    },
    {
      "epoch": 0.6723965047987395,
      "grad_norm": 1.2108922004699707,
      "learning_rate": 5.1210253979438174e-05,
      "loss": 1.427,
      "step": 4694
    },
    {
      "epoch": 0.6725397507520413,
      "grad_norm": 1.15889310836792,
      "learning_rate": 5.1169758579550084e-05,
      "loss": 1.2235,
      "step": 4695
    },
    {
      "epoch": 0.672682996705343,
      "grad_norm": 0.9420252442359924,
      "learning_rate": 5.112927369160458e-05,
      "loss": 1.5244,
      "step": 4696
    },
    {
      "epoch": 0.6728262426586449,
      "grad_norm": 0.9551871418952942,
      "learning_rate": 5.108879932431709e-05,
      "loss": 1.5211,
      "step": 4697
    },
    {
      "epoch": 0.6729694886119467,
      "grad_norm": 1.2036726474761963,
      "learning_rate": 5.104833548640074e-05,
      "loss": 1.2473,
      "step": 4698
    },
    {
      "epoch": 0.6731127345652486,
      "grad_norm": 1.0814273357391357,
      "learning_rate": 5.10078821865664e-05,
      "loss": 1.2721,
      "step": 4699
    },
    {
      "epoch": 0.6732559805185504,
      "grad_norm": 0.9679111838340759,
      "learning_rate": 5.096743943352269e-05,
      "loss": 1.4724,
      "step": 4700
    },
    {
      "epoch": 0.6733992264718521,
      "grad_norm": 1.0734469890594482,
      "learning_rate": 5.092700723597583e-05,
      "loss": 1.4583,
      "step": 4701
    },
    {
      "epoch": 0.673542472425154,
      "grad_norm": 1.2774831056594849,
      "learning_rate": 5.088658560262998e-05,
      "loss": 1.4771,
      "step": 4702
    },
    {
      "epoch": 0.6736857183784558,
      "grad_norm": 1.0236616134643555,
      "learning_rate": 5.084617454218692e-05,
      "loss": 1.3778,
      "step": 4703
    },
    {
      "epoch": 0.6738289643317577,
      "grad_norm": 1.3890806436538696,
      "learning_rate": 5.080577406334607e-05,
      "loss": 1.424,
      "step": 4704
    },
    {
      "epoch": 0.6739722102850595,
      "grad_norm": 1.048463225364685,
      "learning_rate": 5.076538417480465e-05,
      "loss": 1.6065,
      "step": 4705
    },
    {
      "epoch": 0.6741154562383612,
      "grad_norm": 1.2354425191879272,
      "learning_rate": 5.072500488525775e-05,
      "loss": 1.3801,
      "step": 4706
    },
    {
      "epoch": 0.6742587021916631,
      "grad_norm": 1.0977635383605957,
      "learning_rate": 5.068463620339787e-05,
      "loss": 1.3023,
      "step": 4707
    },
    {
      "epoch": 0.6744019481449649,
      "grad_norm": 0.9004481434822083,
      "learning_rate": 5.064427813791547e-05,
      "loss": 1.4373,
      "step": 4708
    },
    {
      "epoch": 0.6745451940982667,
      "grad_norm": 1.1374307870864868,
      "learning_rate": 5.0603930697498605e-05,
      "loss": 1.6981,
      "step": 4709
    },
    {
      "epoch": 0.6746884400515686,
      "grad_norm": 1.0072087049484253,
      "learning_rate": 5.0563593890833115e-05,
      "loss": 1.3042,
      "step": 4710
    },
    {
      "epoch": 0.6748316860048703,
      "grad_norm": 1.370752215385437,
      "learning_rate": 5.052326772660252e-05,
      "loss": 1.4036,
      "step": 4711
    },
    {
      "epoch": 0.6749749319581722,
      "grad_norm": 1.3025410175323486,
      "learning_rate": 5.048295221348795e-05,
      "loss": 1.4296,
      "step": 4712
    },
    {
      "epoch": 0.675118177911474,
      "grad_norm": 1.000956416130066,
      "learning_rate": 5.044264736016846e-05,
      "loss": 1.4186,
      "step": 4713
    },
    {
      "epoch": 0.6752614238647758,
      "grad_norm": 1.4790866374969482,
      "learning_rate": 5.040235317532067e-05,
      "loss": 1.4009,
      "step": 4714
    },
    {
      "epoch": 0.6754046698180777,
      "grad_norm": 1.0980147123336792,
      "learning_rate": 5.036206966761887e-05,
      "loss": 1.3314,
      "step": 4715
    },
    {
      "epoch": 0.6755479157713794,
      "grad_norm": 1.0527515411376953,
      "learning_rate": 5.03217968457351e-05,
      "loss": 1.3224,
      "step": 4716
    },
    {
      "epoch": 0.6756911617246812,
      "grad_norm": 1.2290985584259033,
      "learning_rate": 5.028153471833914e-05,
      "loss": 1.3974,
      "step": 4717
    },
    {
      "epoch": 0.6758344076779831,
      "grad_norm": 1.0816236734390259,
      "learning_rate": 5.024128329409841e-05,
      "loss": 1.4561,
      "step": 4718
    },
    {
      "epoch": 0.6759776536312849,
      "grad_norm": 1.0314141511917114,
      "learning_rate": 5.0201042581678034e-05,
      "loss": 1.5268,
      "step": 4719
    },
    {
      "epoch": 0.6761208995845868,
      "grad_norm": 1.0151269435882568,
      "learning_rate": 5.016081258974088e-05,
      "loss": 1.2441,
      "step": 4720
    },
    {
      "epoch": 0.6762641455378886,
      "grad_norm": 1.0514795780181885,
      "learning_rate": 5.012059332694747e-05,
      "loss": 1.3433,
      "step": 4721
    },
    {
      "epoch": 0.6764073914911903,
      "grad_norm": 1.162483811378479,
      "learning_rate": 5.008038480195596e-05,
      "loss": 1.5176,
      "step": 4722
    },
    {
      "epoch": 0.6765506374444922,
      "grad_norm": 0.9874649047851562,
      "learning_rate": 5.0040187023422237e-05,
      "loss": 1.4608,
      "step": 4723
    },
    {
      "epoch": 0.676693883397794,
      "grad_norm": 1.122450828552246,
      "learning_rate": 5.000000000000002e-05,
      "loss": 1.2824,
      "step": 4724
    },
    {
      "epoch": 0.6768371293510959,
      "grad_norm": 1.218967080116272,
      "learning_rate": 4.995982374034046e-05,
      "loss": 1.4422,
      "step": 4725
    },
    {
      "epoch": 0.6769803753043977,
      "grad_norm": 1.1774418354034424,
      "learning_rate": 4.991965825309254e-05,
      "loss": 1.4671,
      "step": 4726
    },
    {
      "epoch": 0.6771236212576994,
      "grad_norm": 1.1773892641067505,
      "learning_rate": 4.9879503546902906e-05,
      "loss": 1.4382,
      "step": 4727
    },
    {
      "epoch": 0.6772668672110013,
      "grad_norm": 1.0624511241912842,
      "learning_rate": 4.983935963041586e-05,
      "loss": 1.398,
      "step": 4728
    },
    {
      "epoch": 0.6774101131643031,
      "grad_norm": 1.0673298835754395,
      "learning_rate": 4.979922651227346e-05,
      "loss": 1.2745,
      "step": 4729
    },
    {
      "epoch": 0.677553359117605,
      "grad_norm": 1.0545941591262817,
      "learning_rate": 4.9759104201115214e-05,
      "loss": 1.5376,
      "step": 4730
    },
    {
      "epoch": 0.6776966050709068,
      "grad_norm": 1.152147889137268,
      "learning_rate": 4.9718992705578615e-05,
      "loss": 1.401,
      "step": 4731
    },
    {
      "epoch": 0.6778398510242085,
      "grad_norm": 1.0504961013793945,
      "learning_rate": 4.967889203429866e-05,
      "loss": 1.2607,
      "step": 4732
    },
    {
      "epoch": 0.6779830969775104,
      "grad_norm": 1.0604902505874634,
      "learning_rate": 4.9638802195907976e-05,
      "loss": 1.3426,
      "step": 4733
    },
    {
      "epoch": 0.6781263429308122,
      "grad_norm": 0.9292310476303101,
      "learning_rate": 4.9598723199036865e-05,
      "loss": 1.3828,
      "step": 4734
    },
    {
      "epoch": 0.678269588884114,
      "grad_norm": 1.0685092210769653,
      "learning_rate": 4.955865505231351e-05,
      "loss": 1.397,
      "step": 4735
    },
    {
      "epoch": 0.6784128348374159,
      "grad_norm": 1.057086706161499,
      "learning_rate": 4.951859776436345e-05,
      "loss": 1.5747,
      "step": 4736
    },
    {
      "epoch": 0.6785560807907176,
      "grad_norm": 1.0746482610702515,
      "learning_rate": 4.947855134381007e-05,
      "loss": 1.4448,
      "step": 4737
    },
    {
      "epoch": 0.6786993267440194,
      "grad_norm": 0.9867801666259766,
      "learning_rate": 4.943851579927436e-05,
      "loss": 1.4783,
      "step": 4738
    },
    {
      "epoch": 0.6788425726973213,
      "grad_norm": 1.0670701265335083,
      "learning_rate": 4.9398491139375016e-05,
      "loss": 1.3113,
      "step": 4739
    },
    {
      "epoch": 0.6789858186506231,
      "grad_norm": 1.1904017925262451,
      "learning_rate": 4.9358477372728326e-05,
      "loss": 1.51,
      "step": 4740
    },
    {
      "epoch": 0.679129064603925,
      "grad_norm": 1.051641821861267,
      "learning_rate": 4.931847450794828e-05,
      "loss": 1.4968,
      "step": 4741
    },
    {
      "epoch": 0.6792723105572268,
      "grad_norm": 1.1551966667175293,
      "learning_rate": 4.9278482553646535e-05,
      "loss": 1.3866,
      "step": 4742
    },
    {
      "epoch": 0.6794155565105285,
      "grad_norm": 1.1434199810028076,
      "learning_rate": 4.923850151843229e-05,
      "loss": 1.5198,
      "step": 4743
    },
    {
      "epoch": 0.6795588024638304,
      "grad_norm": 0.9995795488357544,
      "learning_rate": 4.9198531410912486e-05,
      "loss": 1.4389,
      "step": 4744
    },
    {
      "epoch": 0.6797020484171322,
      "grad_norm": 1.159527063369751,
      "learning_rate": 4.91585722396918e-05,
      "loss": 1.6115,
      "step": 4745
    },
    {
      "epoch": 0.6798452943704341,
      "grad_norm": 0.994971513748169,
      "learning_rate": 4.911862401337234e-05,
      "loss": 1.3416,
      "step": 4746
    },
    {
      "epoch": 0.6799885403237359,
      "grad_norm": 1.267449975013733,
      "learning_rate": 4.907868674055401e-05,
      "loss": 1.5255,
      "step": 4747
    },
    {
      "epoch": 0.6801317862770376,
      "grad_norm": 1.2041118144989014,
      "learning_rate": 4.903876042983433e-05,
      "loss": 1.3445,
      "step": 4748
    },
    {
      "epoch": 0.6802750322303395,
      "grad_norm": 1.215191125869751,
      "learning_rate": 4.899884508980843e-05,
      "loss": 1.1965,
      "step": 4749
    },
    {
      "epoch": 0.6804182781836413,
      "grad_norm": 1.2949573993682861,
      "learning_rate": 4.8958940729069134e-05,
      "loss": 1.3193,
      "step": 4750
    },
    {
      "epoch": 0.6805615241369432,
      "grad_norm": 1.178986668586731,
      "learning_rate": 4.891904735620675e-05,
      "loss": 1.501,
      "step": 4751
    },
    {
      "epoch": 0.680704770090245,
      "grad_norm": 1.1801514625549316,
      "learning_rate": 4.8879164979809454e-05,
      "loss": 1.4221,
      "step": 4752
    },
    {
      "epoch": 0.6808480160435467,
      "grad_norm": 1.0119770765304565,
      "learning_rate": 4.883929360846293e-05,
      "loss": 1.4934,
      "step": 4753
    },
    {
      "epoch": 0.6809912619968486,
      "grad_norm": 1.076130986213684,
      "learning_rate": 4.8799433250750425e-05,
      "loss": 1.5046,
      "step": 4754
    },
    {
      "epoch": 0.6811345079501504,
      "grad_norm": 0.8642748594284058,
      "learning_rate": 4.875958391525289e-05,
      "loss": 1.5181,
      "step": 4755
    },
    {
      "epoch": 0.6812777539034522,
      "grad_norm": 1.0306544303894043,
      "learning_rate": 4.871974561054901e-05,
      "loss": 1.3963,
      "step": 4756
    },
    {
      "epoch": 0.6814209998567541,
      "grad_norm": 1.233858585357666,
      "learning_rate": 4.867991834521487e-05,
      "loss": 1.4668,
      "step": 4757
    },
    {
      "epoch": 0.6815642458100558,
      "grad_norm": 1.132620930671692,
      "learning_rate": 4.864010212782434e-05,
      "loss": 1.4424,
      "step": 4758
    },
    {
      "epoch": 0.6817074917633577,
      "grad_norm": 1.2647408246994019,
      "learning_rate": 4.860029696694886e-05,
      "loss": 1.352,
      "step": 4759
    },
    {
      "epoch": 0.6818507377166595,
      "grad_norm": 1.1134819984436035,
      "learning_rate": 4.8560502871157485e-05,
      "loss": 1.5445,
      "step": 4760
    },
    {
      "epoch": 0.6819939836699613,
      "grad_norm": 0.9433553218841553,
      "learning_rate": 4.852071984901696e-05,
      "loss": 1.3118,
      "step": 4761
    },
    {
      "epoch": 0.6821372296232632,
      "grad_norm": 1.1843516826629639,
      "learning_rate": 4.8480947909091454e-05,
      "loss": 1.2485,
      "step": 4762
    },
    {
      "epoch": 0.682280475576565,
      "grad_norm": 1.1037076711654663,
      "learning_rate": 4.8441187059943015e-05,
      "loss": 1.4771,
      "step": 4763
    },
    {
      "epoch": 0.6824237215298667,
      "grad_norm": 1.0096940994262695,
      "learning_rate": 4.840143731013115e-05,
      "loss": 1.5292,
      "step": 4764
    },
    {
      "epoch": 0.6825669674831686,
      "grad_norm": 1.1632788181304932,
      "learning_rate": 4.836169866821293e-05,
      "loss": 1.4896,
      "step": 4765
    },
    {
      "epoch": 0.6827102134364704,
      "grad_norm": 0.9995408654212952,
      "learning_rate": 4.832197114274314e-05,
      "loss": 1.3944,
      "step": 4766
    },
    {
      "epoch": 0.6828534593897723,
      "grad_norm": 0.9000145196914673,
      "learning_rate": 4.8282254742274126e-05,
      "loss": 1.4949,
      "step": 4767
    },
    {
      "epoch": 0.6829967053430741,
      "grad_norm": 1.1202291250228882,
      "learning_rate": 4.8242549475355844e-05,
      "loss": 1.3469,
      "step": 4768
    },
    {
      "epoch": 0.6831399512963758,
      "grad_norm": 1.1277642250061035,
      "learning_rate": 4.820285535053588e-05,
      "loss": 1.5005,
      "step": 4769
    },
    {
      "epoch": 0.6832831972496777,
      "grad_norm": 1.1232801675796509,
      "learning_rate": 4.816317237635937e-05,
      "loss": 1.3975,
      "step": 4770
    },
    {
      "epoch": 0.6834264432029795,
      "grad_norm": 1.0864691734313965,
      "learning_rate": 4.8123500561369115e-05,
      "loss": 1.3348,
      "step": 4771
    },
    {
      "epoch": 0.6835696891562814,
      "grad_norm": 1.14359712600708,
      "learning_rate": 4.808383991410542e-05,
      "loss": 1.3851,
      "step": 4772
    },
    {
      "epoch": 0.6837129351095832,
      "grad_norm": 1.1836166381835938,
      "learning_rate": 4.804419044310622e-05,
      "loss": 1.4476,
      "step": 4773
    },
    {
      "epoch": 0.6838561810628849,
      "grad_norm": 1.291280746459961,
      "learning_rate": 4.800455215690719e-05,
      "loss": 1.3281,
      "step": 4774
    },
    {
      "epoch": 0.6839994270161868,
      "grad_norm": 1.0990039110183716,
      "learning_rate": 4.7964925064041356e-05,
      "loss": 1.4437,
      "step": 4775
    },
    {
      "epoch": 0.6841426729694886,
      "grad_norm": 0.929581344127655,
      "learning_rate": 4.792530917303948e-05,
      "loss": 1.4478,
      "step": 4776
    },
    {
      "epoch": 0.6842859189227904,
      "grad_norm": 1.1568092107772827,
      "learning_rate": 4.7885704492429894e-05,
      "loss": 1.3816,
      "step": 4777
    },
    {
      "epoch": 0.6844291648760923,
      "grad_norm": 1.091101884841919,
      "learning_rate": 4.7846111030738506e-05,
      "loss": 1.4976,
      "step": 4778
    },
    {
      "epoch": 0.684572410829394,
      "grad_norm": 0.9966301321983337,
      "learning_rate": 4.780652879648884e-05,
      "loss": 1.5102,
      "step": 4779
    },
    {
      "epoch": 0.6847156567826959,
      "grad_norm": 1.1916595697402954,
      "learning_rate": 4.776695779820186e-05,
      "loss": 1.4293,
      "step": 4780
    },
    {
      "epoch": 0.6848589027359977,
      "grad_norm": 1.0427385568618774,
      "learning_rate": 4.772739804439633e-05,
      "loss": 1.4752,
      "step": 4781
    },
    {
      "epoch": 0.6850021486892995,
      "grad_norm": 1.01448655128479,
      "learning_rate": 4.7687849543588505e-05,
      "loss": 1.4508,
      "step": 4782
    },
    {
      "epoch": 0.6851453946426014,
      "grad_norm": 1.192375659942627,
      "learning_rate": 4.764831230429211e-05,
      "loss": 1.3124,
      "step": 4783
    },
    {
      "epoch": 0.6852886405959032,
      "grad_norm": 0.9594294428825378,
      "learning_rate": 4.760878633501853e-05,
      "loss": 1.3555,
      "step": 4784
    },
    {
      "epoch": 0.685431886549205,
      "grad_norm": 1.0502291917800903,
      "learning_rate": 4.756927164427685e-05,
      "loss": 1.6175,
      "step": 4785
    },
    {
      "epoch": 0.6855751325025068,
      "grad_norm": 1.211198329925537,
      "learning_rate": 4.752976824057348e-05,
      "loss": 1.476,
      "step": 4786
    },
    {
      "epoch": 0.6857183784558086,
      "grad_norm": 1.0386450290679932,
      "learning_rate": 4.7490276132412584e-05,
      "loss": 1.4373,
      "step": 4787
    },
    {
      "epoch": 0.6858616244091105,
      "grad_norm": 1.0045961141586304,
      "learning_rate": 4.745079532829582e-05,
      "loss": 1.2074,
      "step": 4788
    },
    {
      "epoch": 0.6860048703624123,
      "grad_norm": 1.094271183013916,
      "learning_rate": 4.741132583672242e-05,
      "loss": 1.5173,
      "step": 4789
    },
    {
      "epoch": 0.686148116315714,
      "grad_norm": 1.0525481700897217,
      "learning_rate": 4.737186766618924e-05,
      "loss": 1.0464,
      "step": 4790
    },
    {
      "epoch": 0.6862913622690159,
      "grad_norm": 1.023247480392456,
      "learning_rate": 4.733242082519054e-05,
      "loss": 1.3911,
      "step": 4791
    },
    {
      "epoch": 0.6864346082223177,
      "grad_norm": 1.1460680961608887,
      "learning_rate": 4.7292985322218355e-05,
      "loss": 1.3252,
      "step": 4792
    },
    {
      "epoch": 0.6865778541756196,
      "grad_norm": 1.239241361618042,
      "learning_rate": 4.7253561165762163e-05,
      "loss": 1.3775,
      "step": 4793
    },
    {
      "epoch": 0.6867211001289214,
      "grad_norm": 1.1763759851455688,
      "learning_rate": 4.721414836430894e-05,
      "loss": 1.4646,
      "step": 4794
    },
    {
      "epoch": 0.6868643460822231,
      "grad_norm": 0.8961011171340942,
      "learning_rate": 4.717474692634334e-05,
      "loss": 1.4793,
      "step": 4795
    },
    {
      "epoch": 0.687007592035525,
      "grad_norm": 1.0425609350204468,
      "learning_rate": 4.713535686034749e-05,
      "loss": 1.4133,
      "step": 4796
    },
    {
      "epoch": 0.6871508379888268,
      "grad_norm": 1.2007516622543335,
      "learning_rate": 4.709597817480112e-05,
      "loss": 1.547,
      "step": 4797
    },
    {
      "epoch": 0.6872940839421287,
      "grad_norm": 1.124093770980835,
      "learning_rate": 4.7056610878181486e-05,
      "loss": 1.3413,
      "step": 4798
    },
    {
      "epoch": 0.6874373298954305,
      "grad_norm": 1.3353739976882935,
      "learning_rate": 4.7017254978963366e-05,
      "loss": 1.452,
      "step": 4799
    },
    {
      "epoch": 0.6875805758487322,
      "grad_norm": 1.1405316591262817,
      "learning_rate": 4.697791048561918e-05,
      "loss": 1.2804,
      "step": 4800
    },
    {
      "epoch": 0.6877238218020341,
      "grad_norm": 1.0872316360473633,
      "learning_rate": 4.693857740661869e-05,
      "loss": 1.4985,
      "step": 4801
    },
    {
      "epoch": 0.6878670677553359,
      "grad_norm": 1.1340214014053345,
      "learning_rate": 4.689925575042946e-05,
      "loss": 1.1228,
      "step": 4802
    },
    {
      "epoch": 0.6880103137086377,
      "grad_norm": 1.212959885597229,
      "learning_rate": 4.685994552551647e-05,
      "loss": 1.4991,
      "step": 4803
    },
    {
      "epoch": 0.6881535596619396,
      "grad_norm": 1.1516879796981812,
      "learning_rate": 4.682064674034216e-05,
      "loss": 1.4218,
      "step": 4804
    },
    {
      "epoch": 0.6882968056152414,
      "grad_norm": 1.0645337104797363,
      "learning_rate": 4.6781359403366576e-05,
      "loss": 1.4217,
      "step": 4805
    },
    {
      "epoch": 0.6884400515685432,
      "grad_norm": 1.0520826578140259,
      "learning_rate": 4.6742083523047434e-05,
      "loss": 1.3315,
      "step": 4806
    },
    {
      "epoch": 0.688583297521845,
      "grad_norm": 1.155551552772522,
      "learning_rate": 4.670281910783974e-05,
      "loss": 1.5684,
      "step": 4807
    },
    {
      "epoch": 0.6887265434751468,
      "grad_norm": 1.1492711305618286,
      "learning_rate": 4.666356616619619e-05,
      "loss": 1.3786,
      "step": 4808
    },
    {
      "epoch": 0.6888697894284487,
      "grad_norm": 0.9987225532531738,
      "learning_rate": 4.662432470656698e-05,
      "loss": 1.4129,
      "step": 4809
    },
    {
      "epoch": 0.6890130353817505,
      "grad_norm": 1.320759892463684,
      "learning_rate": 4.658509473739981e-05,
      "loss": 1.3739,
      "step": 4810
    },
    {
      "epoch": 0.6891562813350522,
      "grad_norm": 0.9801371693611145,
      "learning_rate": 4.6545876267139974e-05,
      "loss": 1.3402,
      "step": 4811
    },
    {
      "epoch": 0.6892995272883541,
      "grad_norm": 0.953399121761322,
      "learning_rate": 4.650666930423012e-05,
      "loss": 1.51,
      "step": 4812
    },
    {
      "epoch": 0.6894427732416559,
      "grad_norm": 0.9664946794509888,
      "learning_rate": 4.646747385711064e-05,
      "loss": 1.2917,
      "step": 4813
    },
    {
      "epoch": 0.6895860191949578,
      "grad_norm": 1.1409845352172852,
      "learning_rate": 4.642828993421936e-05,
      "loss": 1.3537,
      "step": 4814
    },
    {
      "epoch": 0.6897292651482596,
      "grad_norm": 1.049520492553711,
      "learning_rate": 4.638911754399152e-05,
      "loss": 1.5072,
      "step": 4815
    },
    {
      "epoch": 0.6898725111015613,
      "grad_norm": 1.217037320137024,
      "learning_rate": 4.634995669486001e-05,
      "loss": 1.3279,
      "step": 4816
    },
    {
      "epoch": 0.6900157570548632,
      "grad_norm": 1.029850959777832,
      "learning_rate": 4.631080739525518e-05,
      "loss": 1.1781,
      "step": 4817
    },
    {
      "epoch": 0.690159003008165,
      "grad_norm": 0.9795482158660889,
      "learning_rate": 4.6271669653604945e-05,
      "loss": 1.3603,
      "step": 4818
    },
    {
      "epoch": 0.6903022489614669,
      "grad_norm": 1.1098555326461792,
      "learning_rate": 4.623254347833464e-05,
      "loss": 1.3102,
      "step": 4819
    },
    {
      "epoch": 0.6904454949147687,
      "grad_norm": 1.0522459745407104,
      "learning_rate": 4.61934288778672e-05,
      "loss": 1.3217,
      "step": 4820
    },
    {
      "epoch": 0.6905887408680704,
      "grad_norm": 0.9890515208244324,
      "learning_rate": 4.6154325860623037e-05,
      "loss": 1.3785,
      "step": 4821
    },
    {
      "epoch": 0.6907319868213723,
      "grad_norm": 0.9646671414375305,
      "learning_rate": 4.6115234435020016e-05,
      "loss": 1.4027,
      "step": 4822
    },
    {
      "epoch": 0.6908752327746741,
      "grad_norm": 1.178161382675171,
      "learning_rate": 4.607615460947354e-05,
      "loss": 1.4829,
      "step": 4823
    },
    {
      "epoch": 0.691018478727976,
      "grad_norm": 1.2263121604919434,
      "learning_rate": 4.6037086392396654e-05,
      "loss": 1.4178,
      "step": 4824
    },
    {
      "epoch": 0.6911617246812778,
      "grad_norm": 1.1153571605682373,
      "learning_rate": 4.599802979219965e-05,
      "loss": 1.4907,
      "step": 4825
    },
    {
      "epoch": 0.6913049706345795,
      "grad_norm": 0.9988410472869873,
      "learning_rate": 4.5958984817290484e-05,
      "loss": 1.4451,
      "step": 4826
    },
    {
      "epoch": 0.6914482165878814,
      "grad_norm": 0.966171383857727,
      "learning_rate": 4.5919951476074583e-05,
      "loss": 1.4059,
      "step": 4827
    },
    {
      "epoch": 0.6915914625411832,
      "grad_norm": 1.328002691268921,
      "learning_rate": 4.588092977695485e-05,
      "loss": 1.3912,
      "step": 4828
    },
    {
      "epoch": 0.691734708494485,
      "grad_norm": 1.2720357179641724,
      "learning_rate": 4.584191972833175e-05,
      "loss": 1.4637,
      "step": 4829
    },
    {
      "epoch": 0.6918779544477869,
      "grad_norm": 1.2456566095352173,
      "learning_rate": 4.580292133860303e-05,
      "loss": 1.5103,
      "step": 4830
    },
    {
      "epoch": 0.6920212004010887,
      "grad_norm": 0.9940177798271179,
      "learning_rate": 4.576393461616424e-05,
      "loss": 1.4939,
      "step": 4831
    },
    {
      "epoch": 0.6921644463543905,
      "grad_norm": 1.3750369548797607,
      "learning_rate": 4.572495956940821e-05,
      "loss": 1.5017,
      "step": 4832
    },
    {
      "epoch": 0.6923076923076923,
      "grad_norm": 1.0775599479675293,
      "learning_rate": 4.5685996206725253e-05,
      "loss": 1.3319,
      "step": 4833
    },
    {
      "epoch": 0.6924509382609941,
      "grad_norm": 1.0888468027114868,
      "learning_rate": 4.564704453650323e-05,
      "loss": 1.5101,
      "step": 4834
    },
    {
      "epoch": 0.692594184214296,
      "grad_norm": 1.3396648168563843,
      "learning_rate": 4.560810456712754e-05,
      "loss": 1.5322,
      "step": 4835
    },
    {
      "epoch": 0.6927374301675978,
      "grad_norm": 1.0036617517471313,
      "learning_rate": 4.556917630698092e-05,
      "loss": 1.5268,
      "step": 4836
    },
    {
      "epoch": 0.6928806761208995,
      "grad_norm": 1.1821216344833374,
      "learning_rate": 4.553025976444369e-05,
      "loss": 1.189,
      "step": 4837
    },
    {
      "epoch": 0.6930239220742014,
      "grad_norm": 1.213313102722168,
      "learning_rate": 4.5491354947893616e-05,
      "loss": 1.4063,
      "step": 4838
    },
    {
      "epoch": 0.6931671680275032,
      "grad_norm": 1.1589694023132324,
      "learning_rate": 4.545246186570594e-05,
      "loss": 1.3108,
      "step": 4839
    },
    {
      "epoch": 0.6933104139808051,
      "grad_norm": 0.9477056264877319,
      "learning_rate": 4.541358052625342e-05,
      "loss": 1.6081,
      "step": 4840
    },
    {
      "epoch": 0.6934536599341069,
      "grad_norm": 0.9669190049171448,
      "learning_rate": 4.537471093790614e-05,
      "loss": 1.426,
      "step": 4841
    },
    {
      "epoch": 0.6935969058874086,
      "grad_norm": 0.9902364611625671,
      "learning_rate": 4.5335853109031855e-05,
      "loss": 1.4248,
      "step": 4842
    },
    {
      "epoch": 0.6937401518407105,
      "grad_norm": 1.1749876737594604,
      "learning_rate": 4.529700704799571e-05,
      "loss": 1.5017,
      "step": 4843
    },
    {
      "epoch": 0.6938833977940123,
      "grad_norm": 1.015008568763733,
      "learning_rate": 4.525817276316022e-05,
      "loss": 1.4043,
      "step": 4844
    },
    {
      "epoch": 0.6940266437473142,
      "grad_norm": 1.004389762878418,
      "learning_rate": 4.5219350262885484e-05,
      "loss": 1.5983,
      "step": 4845
    },
    {
      "epoch": 0.694169889700616,
      "grad_norm": 1.1781011819839478,
      "learning_rate": 4.518053955552903e-05,
      "loss": 1.5675,
      "step": 4846
    },
    {
      "epoch": 0.6943131356539177,
      "grad_norm": 1.3226794004440308,
      "learning_rate": 4.514174064944584e-05,
      "loss": 1.3031,
      "step": 4847
    },
    {
      "epoch": 0.6944563816072196,
      "grad_norm": 0.9350221753120422,
      "learning_rate": 4.510295355298835e-05,
      "loss": 1.2796,
      "step": 4848
    },
    {
      "epoch": 0.6945996275605214,
      "grad_norm": 1.1182013750076294,
      "learning_rate": 4.506417827450647e-05,
      "loss": 1.3785,
      "step": 4849
    },
    {
      "epoch": 0.6947428735138232,
      "grad_norm": 1.0845601558685303,
      "learning_rate": 4.50254148223476e-05,
      "loss": 1.1681,
      "step": 4850
    },
    {
      "epoch": 0.6948861194671251,
      "grad_norm": 0.9920936822891235,
      "learning_rate": 4.498666320485647e-05,
      "loss": 1.3027,
      "step": 4851
    },
    {
      "epoch": 0.6950293654204269,
      "grad_norm": 0.974132776260376,
      "learning_rate": 4.494792343037535e-05,
      "loss": 1.5396,
      "step": 4852
    },
    {
      "epoch": 0.6951726113737287,
      "grad_norm": 0.9596220850944519,
      "learning_rate": 4.490919550724405e-05,
      "loss": 1.4654,
      "step": 4853
    },
    {
      "epoch": 0.6953158573270305,
      "grad_norm": 1.3003628253936768,
      "learning_rate": 4.487047944379964e-05,
      "loss": 1.2778,
      "step": 4854
    },
    {
      "epoch": 0.6954591032803323,
      "grad_norm": 1.043830394744873,
      "learning_rate": 4.4831775248376764e-05,
      "loss": 1.4356,
      "step": 4855
    },
    {
      "epoch": 0.6956023492336342,
      "grad_norm": 1.0722198486328125,
      "learning_rate": 4.4793082929307474e-05,
      "loss": 1.4904,
      "step": 4856
    },
    {
      "epoch": 0.695745595186936,
      "grad_norm": 1.1482727527618408,
      "learning_rate": 4.4754402494921275e-05,
      "loss": 1.3148,
      "step": 4857
    },
    {
      "epoch": 0.6958888411402377,
      "grad_norm": 1.0990558862686157,
      "learning_rate": 4.471573395354514e-05,
      "loss": 1.3117,
      "step": 4858
    },
    {
      "epoch": 0.6960320870935396,
      "grad_norm": 1.0018548965454102,
      "learning_rate": 4.467707731350332e-05,
      "loss": 1.4614,
      "step": 4859
    },
    {
      "epoch": 0.6961753330468414,
      "grad_norm": 1.168513298034668,
      "learning_rate": 4.4638432583117786e-05,
      "loss": 1.5093,
      "step": 4860
    },
    {
      "epoch": 0.6963185790001433,
      "grad_norm": 1.1008967161178589,
      "learning_rate": 4.4599799770707754e-05,
      "loss": 1.4191,
      "step": 4861
    },
    {
      "epoch": 0.6964618249534451,
      "grad_norm": 1.1140096187591553,
      "learning_rate": 4.4561178884589815e-05,
      "loss": 1.3664,
      "step": 4862
    },
    {
      "epoch": 0.6966050709067468,
      "grad_norm": 1.2130491733551025,
      "learning_rate": 4.452256993307821e-05,
      "loss": 1.4676,
      "step": 4863
    },
    {
      "epoch": 0.6967483168600487,
      "grad_norm": 0.9866144061088562,
      "learning_rate": 4.4483972924484474e-05,
      "loss": 1.5902,
      "step": 4864
    },
    {
      "epoch": 0.6968915628133505,
      "grad_norm": 1.0489447116851807,
      "learning_rate": 4.4445387867117526e-05,
      "loss": 1.4313,
      "step": 4865
    },
    {
      "epoch": 0.6970348087666524,
      "grad_norm": 0.9920719861984253,
      "learning_rate": 4.44068147692838e-05,
      "loss": 1.4739,
      "step": 4866
    },
    {
      "epoch": 0.6971780547199542,
      "grad_norm": 0.8992570638656616,
      "learning_rate": 4.436825363928714e-05,
      "loss": 1.3678,
      "step": 4867
    },
    {
      "epoch": 0.6973213006732559,
      "grad_norm": 1.0031867027282715,
      "learning_rate": 4.43297044854288e-05,
      "loss": 1.4566,
      "step": 4868
    },
    {
      "epoch": 0.6974645466265578,
      "grad_norm": 0.9616906642913818,
      "learning_rate": 4.4291167316007464e-05,
      "loss": 1.4792,
      "step": 4869
    },
    {
      "epoch": 0.6976077925798596,
      "grad_norm": 1.168142557144165,
      "learning_rate": 4.4252642139319226e-05,
      "loss": 1.2891,
      "step": 4870
    },
    {
      "epoch": 0.6977510385331614,
      "grad_norm": 1.1153199672698975,
      "learning_rate": 4.421412896365763e-05,
      "loss": 1.217,
      "step": 4871
    },
    {
      "epoch": 0.6978942844864633,
      "grad_norm": 1.0973092317581177,
      "learning_rate": 4.417562779731355e-05,
      "loss": 1.6411,
      "step": 4872
    },
    {
      "epoch": 0.6980375304397651,
      "grad_norm": 1.0448617935180664,
      "learning_rate": 4.413713864857533e-05,
      "loss": 1.3711,
      "step": 4873
    },
    {
      "epoch": 0.6981807763930669,
      "grad_norm": 1.1163080930709839,
      "learning_rate": 4.409866152572883e-05,
      "loss": 1.64,
      "step": 4874
    },
    {
      "epoch": 0.6983240223463687,
      "grad_norm": 1.247266411781311,
      "learning_rate": 4.406019643705715e-05,
      "loss": 1.4342,
      "step": 4875
    },
    {
      "epoch": 0.6984672682996705,
      "grad_norm": 1.0371639728546143,
      "learning_rate": 4.4021743390840866e-05,
      "loss": 1.5044,
      "step": 4876
    },
    {
      "epoch": 0.6986105142529724,
      "grad_norm": 1.0786230564117432,
      "learning_rate": 4.398330239535797e-05,
      "loss": 1.5325,
      "step": 4877
    },
    {
      "epoch": 0.6987537602062742,
      "grad_norm": 1.1284019947052002,
      "learning_rate": 4.3944873458883887e-05,
      "loss": 1.5008,
      "step": 4878
    },
    {
      "epoch": 0.698897006159576,
      "grad_norm": 1.1389825344085693,
      "learning_rate": 4.390645658969143e-05,
      "loss": 1.3201,
      "step": 4879
    },
    {
      "epoch": 0.6990402521128778,
      "grad_norm": 1.1710530519485474,
      "learning_rate": 4.3868051796050705e-05,
      "loss": 1.2706,
      "step": 4880
    },
    {
      "epoch": 0.6991834980661796,
      "grad_norm": 1.2138415575027466,
      "learning_rate": 4.3829659086229403e-05,
      "loss": 1.4588,
      "step": 4881
    },
    {
      "epoch": 0.6993267440194815,
      "grad_norm": 1.1283677816390991,
      "learning_rate": 4.379127846849255e-05,
      "loss": 1.5379,
      "step": 4882
    },
    {
      "epoch": 0.6994699899727833,
      "grad_norm": 1.1636452674865723,
      "learning_rate": 4.375290995110244e-05,
      "loss": 1.286,
      "step": 4883
    },
    {
      "epoch": 0.699613235926085,
      "grad_norm": 1.17005455493927,
      "learning_rate": 4.371455354231889e-05,
      "loss": 1.3387,
      "step": 4884
    },
    {
      "epoch": 0.6997564818793869,
      "grad_norm": 1.1313883066177368,
      "learning_rate": 4.3676209250399194e-05,
      "loss": 1.3364,
      "step": 4885
    },
    {
      "epoch": 0.6998997278326887,
      "grad_norm": 1.0579769611358643,
      "learning_rate": 4.36378770835978e-05,
      "loss": 1.5664,
      "step": 4886
    },
    {
      "epoch": 0.7000429737859906,
      "grad_norm": 1.3555585145950317,
      "learning_rate": 4.359955705016673e-05,
      "loss": 1.5081,
      "step": 4887
    },
    {
      "epoch": 0.7001862197392924,
      "grad_norm": 1.1886733770370483,
      "learning_rate": 4.356124915835533e-05,
      "loss": 1.4471,
      "step": 4888
    },
    {
      "epoch": 0.7003294656925941,
      "grad_norm": 1.4983137845993042,
      "learning_rate": 4.352295341641035e-05,
      "loss": 1.2348,
      "step": 4889
    },
    {
      "epoch": 0.700472711645896,
      "grad_norm": 1.0202289819717407,
      "learning_rate": 4.348466983257594e-05,
      "loss": 1.4595,
      "step": 4890
    },
    {
      "epoch": 0.7006159575991978,
      "grad_norm": 1.1768124103546143,
      "learning_rate": 4.34463984150935e-05,
      "loss": 1.3208,
      "step": 4891
    },
    {
      "epoch": 0.7007592035524997,
      "grad_norm": 1.109227180480957,
      "learning_rate": 4.3408139172202044e-05,
      "loss": 1.4597,
      "step": 4892
    },
    {
      "epoch": 0.7009024495058015,
      "grad_norm": 1.1736289262771606,
      "learning_rate": 4.336989211213782e-05,
      "loss": 1.434,
      "step": 4893
    },
    {
      "epoch": 0.7010456954591033,
      "grad_norm": 0.9314563274383545,
      "learning_rate": 4.333165724313442e-05,
      "loss": 1.3474,
      "step": 4894
    },
    {
      "epoch": 0.7011889414124051,
      "grad_norm": 1.1512372493743896,
      "learning_rate": 4.329343457342289e-05,
      "loss": 1.4563,
      "step": 4895
    },
    {
      "epoch": 0.7013321873657069,
      "grad_norm": 1.0794873237609863,
      "learning_rate": 4.325522411123162e-05,
      "loss": 1.3951,
      "step": 4896
    },
    {
      "epoch": 0.7014754333190087,
      "grad_norm": 1.1541352272033691,
      "learning_rate": 4.321702586478639e-05,
      "loss": 1.2084,
      "step": 4897
    },
    {
      "epoch": 0.7016186792723106,
      "grad_norm": 1.1582107543945312,
      "learning_rate": 4.3178839842310326e-05,
      "loss": 1.3965,
      "step": 4898
    },
    {
      "epoch": 0.7017619252256124,
      "grad_norm": 1.1217702627182007,
      "learning_rate": 4.3140666052023936e-05,
      "loss": 1.353,
      "step": 4899
    },
    {
      "epoch": 0.7019051711789142,
      "grad_norm": 1.0962918996810913,
      "learning_rate": 4.3102504502145147e-05,
      "loss": 1.3657,
      "step": 4900
    },
    {
      "epoch": 0.702048417132216,
      "grad_norm": 0.8652566075325012,
      "learning_rate": 4.3064355200889094e-05,
      "loss": 1.2515,
      "step": 4901
    },
    {
      "epoch": 0.7021916630855178,
      "grad_norm": 0.9534581303596497,
      "learning_rate": 4.302621815646839e-05,
      "loss": 1.4601,
      "step": 4902
    },
    {
      "epoch": 0.7023349090388197,
      "grad_norm": 1.2013819217681885,
      "learning_rate": 4.298809337709312e-05,
      "loss": 1.521,
      "step": 4903
    },
    {
      "epoch": 0.7024781549921215,
      "grad_norm": 0.9915494918823242,
      "learning_rate": 4.294998087097048e-05,
      "loss": 1.4599,
      "step": 4904
    },
    {
      "epoch": 0.7026214009454232,
      "grad_norm": 0.9526614546775818,
      "learning_rate": 4.2911880646305184e-05,
      "loss": 1.1174,
      "step": 4905
    },
    {
      "epoch": 0.7027646468987251,
      "grad_norm": 1.0566946268081665,
      "learning_rate": 4.2873792711299266e-05,
      "loss": 1.3874,
      "step": 4906
    },
    {
      "epoch": 0.7029078928520269,
      "grad_norm": 1.1422756910324097,
      "learning_rate": 4.283571707415214e-05,
      "loss": 1.3858,
      "step": 4907
    },
    {
      "epoch": 0.7030511388053288,
      "grad_norm": 1.0093618631362915,
      "learning_rate": 4.279765374306055e-05,
      "loss": 1.612,
      "step": 4908
    },
    {
      "epoch": 0.7031943847586306,
      "grad_norm": 1.385175108909607,
      "learning_rate": 4.275960272621852e-05,
      "loss": 1.4079,
      "step": 4909
    },
    {
      "epoch": 0.7033376307119323,
      "grad_norm": 0.9692065715789795,
      "learning_rate": 4.272156403181756e-05,
      "loss": 1.3884,
      "step": 4910
    },
    {
      "epoch": 0.7034808766652342,
      "grad_norm": 0.905538022518158,
      "learning_rate": 4.268353766804649e-05,
      "loss": 1.3402,
      "step": 4911
    },
    {
      "epoch": 0.703624122618536,
      "grad_norm": 1.0803182125091553,
      "learning_rate": 4.2645523643091354e-05,
      "loss": 1.4981,
      "step": 4912
    },
    {
      "epoch": 0.7037673685718379,
      "grad_norm": 1.11542809009552,
      "learning_rate": 4.2607521965135645e-05,
      "loss": 1.2637,
      "step": 4913
    },
    {
      "epoch": 0.7039106145251397,
      "grad_norm": 1.3580089807510376,
      "learning_rate": 4.256953264236029e-05,
      "loss": 1.4892,
      "step": 4914
    },
    {
      "epoch": 0.7040538604784414,
      "grad_norm": 1.0030322074890137,
      "learning_rate": 4.253155568294333e-05,
      "loss": 1.5783,
      "step": 4915
    },
    {
      "epoch": 0.7041971064317433,
      "grad_norm": 1.003808617591858,
      "learning_rate": 4.24935910950603e-05,
      "loss": 1.3485,
      "step": 4916
    },
    {
      "epoch": 0.7043403523850451,
      "grad_norm": 1.302235722541809,
      "learning_rate": 4.245563888688405e-05,
      "loss": 1.4738,
      "step": 4917
    },
    {
      "epoch": 0.704483598338347,
      "grad_norm": 1.2146676778793335,
      "learning_rate": 4.241769906658476e-05,
      "loss": 1.4636,
      "step": 4918
    },
    {
      "epoch": 0.7046268442916488,
      "grad_norm": 1.1777825355529785,
      "learning_rate": 4.2379771642329944e-05,
      "loss": 1.5306,
      "step": 4919
    },
    {
      "epoch": 0.7047700902449506,
      "grad_norm": 1.0241835117340088,
      "learning_rate": 4.234185662228435e-05,
      "loss": 1.4563,
      "step": 4920
    },
    {
      "epoch": 0.7049133361982524,
      "grad_norm": 1.1451879739761353,
      "learning_rate": 4.2303954014610235e-05,
      "loss": 1.35,
      "step": 4921
    },
    {
      "epoch": 0.7050565821515542,
      "grad_norm": 1.0630794763565063,
      "learning_rate": 4.226606382746711e-05,
      "loss": 1.4199,
      "step": 4922
    },
    {
      "epoch": 0.705199828104856,
      "grad_norm": 1.169994831085205,
      "learning_rate": 4.2228186069011696e-05,
      "loss": 1.4908,
      "step": 4923
    },
    {
      "epoch": 0.7053430740581579,
      "grad_norm": 1.0153543949127197,
      "learning_rate": 4.2190320747398206e-05,
      "loss": 1.5536,
      "step": 4924
    },
    {
      "epoch": 0.7054863200114597,
      "grad_norm": 1.1942942142486572,
      "learning_rate": 4.215246787077809e-05,
      "loss": 1.2849,
      "step": 4925
    },
    {
      "epoch": 0.7056295659647615,
      "grad_norm": 1.0324727296829224,
      "learning_rate": 4.2114627447300134e-05,
      "loss": 1.5281,
      "step": 4926
    },
    {
      "epoch": 0.7057728119180633,
      "grad_norm": 0.956970751285553,
      "learning_rate": 4.2076799485110454e-05,
      "loss": 1.4555,
      "step": 4927
    },
    {
      "epoch": 0.7059160578713651,
      "grad_norm": 1.093381404876709,
      "learning_rate": 4.203898399235246e-05,
      "loss": 1.4237,
      "step": 4928
    },
    {
      "epoch": 0.706059303824667,
      "grad_norm": 0.9523422718048096,
      "learning_rate": 4.200118097716694e-05,
      "loss": 1.4837,
      "step": 4929
    },
    {
      "epoch": 0.7062025497779688,
      "grad_norm": 1.0700111389160156,
      "learning_rate": 4.196339044769184e-05,
      "loss": 1.525,
      "step": 4930
    },
    {
      "epoch": 0.7063457957312705,
      "grad_norm": 1.2418214082717896,
      "learning_rate": 4.192561241206262e-05,
      "loss": 1.4357,
      "step": 4931
    },
    {
      "epoch": 0.7064890416845724,
      "grad_norm": 1.148093581199646,
      "learning_rate": 4.1887846878411975e-05,
      "loss": 1.5205,
      "step": 4932
    },
    {
      "epoch": 0.7066322876378742,
      "grad_norm": 1.325565218925476,
      "learning_rate": 4.18500938548698e-05,
      "loss": 1.5901,
      "step": 4933
    },
    {
      "epoch": 0.7067755335911761,
      "grad_norm": 1.026126503944397,
      "learning_rate": 4.181235334956341e-05,
      "loss": 1.4371,
      "step": 4934
    },
    {
      "epoch": 0.7069187795444779,
      "grad_norm": 0.9592928290367126,
      "learning_rate": 4.1774625370617484e-05,
      "loss": 1.7092,
      "step": 4935
    },
    {
      "epoch": 0.7070620254977796,
      "grad_norm": 1.3618437051773071,
      "learning_rate": 4.173690992615382e-05,
      "loss": 1.261,
      "step": 4936
    },
    {
      "epoch": 0.7072052714510815,
      "grad_norm": 1.0414167642593384,
      "learning_rate": 4.1699207024291666e-05,
      "loss": 1.5409,
      "step": 4937
    },
    {
      "epoch": 0.7073485174043833,
      "grad_norm": 1.0105513334274292,
      "learning_rate": 4.1661516673147517e-05,
      "loss": 1.4434,
      "step": 4938
    },
    {
      "epoch": 0.7074917633576852,
      "grad_norm": 1.0188097953796387,
      "learning_rate": 4.162383888083519e-05,
      "loss": 1.4705,
      "step": 4939
    },
    {
      "epoch": 0.707635009310987,
      "grad_norm": 1.1254373788833618,
      "learning_rate": 4.1586173655465785e-05,
      "loss": 1.5872,
      "step": 4940
    },
    {
      "epoch": 0.7077782552642888,
      "grad_norm": 1.2008291482925415,
      "learning_rate": 4.154852100514761e-05,
      "loss": 1.6866,
      "step": 4941
    },
    {
      "epoch": 0.7079215012175906,
      "grad_norm": 1.146675944328308,
      "learning_rate": 4.151088093798644e-05,
      "loss": 1.3855,
      "step": 4942
    },
    {
      "epoch": 0.7080647471708924,
      "grad_norm": 1.0632609128952026,
      "learning_rate": 4.147325346208527e-05,
      "loss": 1.3466,
      "step": 4943
    },
    {
      "epoch": 0.7082079931241942,
      "grad_norm": 0.9820842742919922,
      "learning_rate": 4.143563858554428e-05,
      "loss": 1.4555,
      "step": 4944
    },
    {
      "epoch": 0.7083512390774961,
      "grad_norm": 1.0430525541305542,
      "learning_rate": 4.1398036316461054e-05,
      "loss": 1.4242,
      "step": 4945
    },
    {
      "epoch": 0.7084944850307979,
      "grad_norm": 1.210944414138794,
      "learning_rate": 4.136044666293044e-05,
      "loss": 1.2304,
      "step": 4946
    },
    {
      "epoch": 0.7086377309840997,
      "grad_norm": 0.9258332252502441,
      "learning_rate": 4.132286963304456e-05,
      "loss": 1.3622,
      "step": 4947
    },
    {
      "epoch": 0.7087809769374015,
      "grad_norm": 1.2679623365402222,
      "learning_rate": 4.1285305234892803e-05,
      "loss": 1.5023,
      "step": 4948
    },
    {
      "epoch": 0.7089242228907033,
      "grad_norm": 1.226791262626648,
      "learning_rate": 4.124775347656187e-05,
      "loss": 1.2274,
      "step": 4949
    },
    {
      "epoch": 0.7090674688440052,
      "grad_norm": 1.104860544204712,
      "learning_rate": 4.121021436613576e-05,
      "loss": 1.3914,
      "step": 4950
    },
    {
      "epoch": 0.709210714797307,
      "grad_norm": 1.2253968715667725,
      "learning_rate": 4.117268791169564e-05,
      "loss": 1.2296,
      "step": 4951
    },
    {
      "epoch": 0.7093539607506087,
      "grad_norm": 1.125827670097351,
      "learning_rate": 4.113517412132003e-05,
      "loss": 1.2573,
      "step": 4952
    },
    {
      "epoch": 0.7094972067039106,
      "grad_norm": 0.9368603825569153,
      "learning_rate": 4.109767300308481e-05,
      "loss": 1.5181,
      "step": 4953
    },
    {
      "epoch": 0.7096404526572124,
      "grad_norm": 1.0678833723068237,
      "learning_rate": 4.1060184565062963e-05,
      "loss": 1.4978,
      "step": 4954
    },
    {
      "epoch": 0.7097836986105143,
      "grad_norm": 0.97263503074646,
      "learning_rate": 4.102270881532485e-05,
      "loss": 1.4448,
      "step": 4955
    },
    {
      "epoch": 0.7099269445638161,
      "grad_norm": 1.110864281654358,
      "learning_rate": 4.0985245761938064e-05,
      "loss": 1.4616,
      "step": 4956
    },
    {
      "epoch": 0.7100701905171178,
      "grad_norm": 1.060823678970337,
      "learning_rate": 4.094779541296747e-05,
      "loss": 1.4392,
      "step": 4957
    },
    {
      "epoch": 0.7102134364704197,
      "grad_norm": 1.0522089004516602,
      "learning_rate": 4.091035777647525e-05,
      "loss": 1.5224,
      "step": 4958
    },
    {
      "epoch": 0.7103566824237215,
      "grad_norm": 1.0164978504180908,
      "learning_rate": 4.087293286052069e-05,
      "loss": 1.4234,
      "step": 4959
    },
    {
      "epoch": 0.7104999283770234,
      "grad_norm": 1.5885467529296875,
      "learning_rate": 4.083552067316054e-05,
      "loss": 1.4112,
      "step": 4960
    },
    {
      "epoch": 0.7106431743303252,
      "grad_norm": 1.0973089933395386,
      "learning_rate": 4.079812122244874e-05,
      "loss": 1.2094,
      "step": 4961
    },
    {
      "epoch": 0.710786420283627,
      "grad_norm": 1.2338813543319702,
      "learning_rate": 4.076073451643639e-05,
      "loss": 1.2646,
      "step": 4962
    },
    {
      "epoch": 0.7109296662369288,
      "grad_norm": 1.263325572013855,
      "learning_rate": 4.0723360563171906e-05,
      "loss": 1.5651,
      "step": 4963
    },
    {
      "epoch": 0.7110729121902306,
      "grad_norm": 1.002611756324768,
      "learning_rate": 4.068599937070111e-05,
      "loss": 1.2438,
      "step": 4964
    },
    {
      "epoch": 0.7112161581435325,
      "grad_norm": 1.083384394645691,
      "learning_rate": 4.064865094706681e-05,
      "loss": 1.3277,
      "step": 4965
    },
    {
      "epoch": 0.7113594040968343,
      "grad_norm": 1.0699213743209839,
      "learning_rate": 4.0611315300309246e-05,
      "loss": 1.5035,
      "step": 4966
    },
    {
      "epoch": 0.7115026500501361,
      "grad_norm": 0.9197946786880493,
      "learning_rate": 4.0573992438465866e-05,
      "loss": 1.5064,
      "step": 4967
    },
    {
      "epoch": 0.7116458960034379,
      "grad_norm": 1.0107725858688354,
      "learning_rate": 4.053668236957134e-05,
      "loss": 1.2366,
      "step": 4968
    },
    {
      "epoch": 0.7117891419567397,
      "grad_norm": 1.0047035217285156,
      "learning_rate": 4.049938510165765e-05,
      "loss": 1.3793,
      "step": 4969
    },
    {
      "epoch": 0.7119323879100415,
      "grad_norm": 1.089130163192749,
      "learning_rate": 4.0462100642753856e-05,
      "loss": 1.3201,
      "step": 4970
    },
    {
      "epoch": 0.7120756338633434,
      "grad_norm": 1.0256634950637817,
      "learning_rate": 4.04248290008865e-05,
      "loss": 1.4263,
      "step": 4971
    },
    {
      "epoch": 0.7122188798166452,
      "grad_norm": 0.9777927994728088,
      "learning_rate": 4.038757018407924e-05,
      "loss": 1.3401,
      "step": 4972
    },
    {
      "epoch": 0.712362125769947,
      "grad_norm": 1.1109718084335327,
      "learning_rate": 4.035032420035291e-05,
      "loss": 1.2865,
      "step": 4973
    },
    {
      "epoch": 0.7125053717232488,
      "grad_norm": 1.0624184608459473,
      "learning_rate": 4.031309105772567e-05,
      "loss": 1.45,
      "step": 4974
    },
    {
      "epoch": 0.7126486176765506,
      "grad_norm": 1.1249079704284668,
      "learning_rate": 4.027587076421291e-05,
      "loss": 1.5996,
      "step": 4975
    },
    {
      "epoch": 0.7127918636298525,
      "grad_norm": 1.3260663747787476,
      "learning_rate": 4.023866332782723e-05,
      "loss": 1.3964,
      "step": 4976
    },
    {
      "epoch": 0.7129351095831543,
      "grad_norm": 1.0054802894592285,
      "learning_rate": 4.0201468756578465e-05,
      "loss": 1.435,
      "step": 4977
    },
    {
      "epoch": 0.713078355536456,
      "grad_norm": 1.0342075824737549,
      "learning_rate": 4.01642870584737e-05,
      "loss": 1.433,
      "step": 4978
    },
    {
      "epoch": 0.7132216014897579,
      "grad_norm": 1.0962247848510742,
      "learning_rate": 4.012711824151727e-05,
      "loss": 1.4263,
      "step": 4979
    },
    {
      "epoch": 0.7133648474430597,
      "grad_norm": 1.3745440244674683,
      "learning_rate": 4.008996231371062e-05,
      "loss": 1.4615,
      "step": 4980
    },
    {
      "epoch": 0.7135080933963616,
      "grad_norm": 1.053774118423462,
      "learning_rate": 4.005281928305249e-05,
      "loss": 1.5309,
      "step": 4981
    },
    {
      "epoch": 0.7136513393496634,
      "grad_norm": 0.9896112084388733,
      "learning_rate": 4.0015689157538994e-05,
      "loss": 1.6359,
      "step": 4982
    },
    {
      "epoch": 0.7137945853029652,
      "grad_norm": 0.9917699098587036,
      "learning_rate": 3.997857194516319e-05,
      "loss": 1.4501,
      "step": 4983
    },
    {
      "epoch": 0.713937831256267,
      "grad_norm": 1.0307033061981201,
      "learning_rate": 3.994146765391557e-05,
      "loss": 1.4745,
      "step": 4984
    },
    {
      "epoch": 0.7140810772095688,
      "grad_norm": 1.1613492965698242,
      "learning_rate": 3.990437629178372e-05,
      "loss": 1.3274,
      "step": 4985
    },
    {
      "epoch": 0.7142243231628707,
      "grad_norm": 1.281091332435608,
      "learning_rate": 3.986729786675253e-05,
      "loss": 1.5138,
      "step": 4986
    },
    {
      "epoch": 0.7143675691161725,
      "grad_norm": 1.1112534999847412,
      "learning_rate": 3.983023238680406e-05,
      "loss": 1.3925,
      "step": 4987
    },
    {
      "epoch": 0.7145108150694743,
      "grad_norm": 0.9237897396087646,
      "learning_rate": 3.9793179859917585e-05,
      "loss": 1.2174,
      "step": 4988
    },
    {
      "epoch": 0.7146540610227761,
      "grad_norm": 1.222542405128479,
      "learning_rate": 3.97561402940696e-05,
      "loss": 1.3699,
      "step": 4989
    },
    {
      "epoch": 0.7147973069760779,
      "grad_norm": 0.8941904902458191,
      "learning_rate": 3.9719113697233835e-05,
      "loss": 1.4045,
      "step": 4990
    },
    {
      "epoch": 0.7149405529293797,
      "grad_norm": 0.9991548657417297,
      "learning_rate": 3.96821000773811e-05,
      "loss": 1.4922,
      "step": 4991
    },
    {
      "epoch": 0.7150837988826816,
      "grad_norm": 1.3762160539627075,
      "learning_rate": 3.964509944247962e-05,
      "loss": 1.5094,
      "step": 4992
    },
    {
      "epoch": 0.7152270448359834,
      "grad_norm": 1.1509665250778198,
      "learning_rate": 3.9608111800494715e-05,
      "loss": 1.5775,
      "step": 4993
    },
    {
      "epoch": 0.7153702907892852,
      "grad_norm": 1.111878514289856,
      "learning_rate": 3.957113715938884e-05,
      "loss": 1.5462,
      "step": 4994
    },
    {
      "epoch": 0.715513536742587,
      "grad_norm": 1.0730425119400024,
      "learning_rate": 3.953417552712174e-05,
      "loss": 1.4371,
      "step": 4995
    },
    {
      "epoch": 0.7156567826958888,
      "grad_norm": 1.0492804050445557,
      "learning_rate": 3.949722691165036e-05,
      "loss": 1.3903,
      "step": 4996
    },
    {
      "epoch": 0.7158000286491907,
      "grad_norm": 1.172573208808899,
      "learning_rate": 3.9460291320928825e-05,
      "loss": 1.4616,
      "step": 4997
    },
    {
      "epoch": 0.7159432746024925,
      "grad_norm": 1.247603178024292,
      "learning_rate": 3.942336876290843e-05,
      "loss": 1.4334,
      "step": 4998
    },
    {
      "epoch": 0.7160865205557942,
      "grad_norm": 1.048155426979065,
      "learning_rate": 3.93864592455377e-05,
      "loss": 1.5198,
      "step": 4999
    },
    {
      "epoch": 0.7162297665090961,
      "grad_norm": 0.9921081066131592,
      "learning_rate": 3.9349562776762385e-05,
      "loss": 1.3085,
      "step": 5000
    },
    {
      "epoch": 0.7163730124623979,
      "grad_norm": 1.171191930770874,
      "learning_rate": 3.9312679364525294e-05,
      "loss": 1.2653,
      "step": 5001
    },
    {
      "epoch": 0.7165162584156998,
      "grad_norm": 1.1787899732589722,
      "learning_rate": 3.927580901676653e-05,
      "loss": 1.4823,
      "step": 5002
    },
    {
      "epoch": 0.7166595043690016,
      "grad_norm": 1.0939325094223022,
      "learning_rate": 3.923895174142347e-05,
      "loss": 1.3794,
      "step": 5003
    },
    {
      "epoch": 0.7168027503223035,
      "grad_norm": 0.9014527797698975,
      "learning_rate": 3.9202107546430456e-05,
      "loss": 1.3886,
      "step": 5004
    },
    {
      "epoch": 0.7169459962756052,
      "grad_norm": 0.928516149520874,
      "learning_rate": 3.916527643971917e-05,
      "loss": 1.4484,
      "step": 5005
    },
    {
      "epoch": 0.717089242228907,
      "grad_norm": 1.04092538356781,
      "learning_rate": 3.912845842921844e-05,
      "loss": 1.3436,
      "step": 5006
    },
    {
      "epoch": 0.7172324881822089,
      "grad_norm": 1.0407545566558838,
      "learning_rate": 3.909165352285429e-05,
      "loss": 1.3823,
      "step": 5007
    },
    {
      "epoch": 0.7173757341355107,
      "grad_norm": 0.9373136758804321,
      "learning_rate": 3.905486172854991e-05,
      "loss": 1.3684,
      "step": 5008
    },
    {
      "epoch": 0.7175189800888125,
      "grad_norm": 0.9039185047149658,
      "learning_rate": 3.9018083054225564e-05,
      "loss": 1.4995,
      "step": 5009
    },
    {
      "epoch": 0.7176622260421143,
      "grad_norm": 1.133274793624878,
      "learning_rate": 3.898131750779892e-05,
      "loss": 1.2873,
      "step": 5010
    },
    {
      "epoch": 0.7178054719954161,
      "grad_norm": 0.9841527938842773,
      "learning_rate": 3.894456509718466e-05,
      "loss": 1.3205,
      "step": 5011
    },
    {
      "epoch": 0.717948717948718,
      "grad_norm": 0.927627682685852,
      "learning_rate": 3.890782583029462e-05,
      "loss": 1.3358,
      "step": 5012
    },
    {
      "epoch": 0.7180919639020198,
      "grad_norm": 1.21440851688385,
      "learning_rate": 3.887109971503785e-05,
      "loss": 1.5089,
      "step": 5013
    },
    {
      "epoch": 0.7182352098553216,
      "grad_norm": 0.9294370412826538,
      "learning_rate": 3.883438675932067e-05,
      "loss": 1.3146,
      "step": 5014
    },
    {
      "epoch": 0.7183784558086234,
      "grad_norm": 0.9559040069580078,
      "learning_rate": 3.8797686971046375e-05,
      "loss": 1.548,
      "step": 5015
    },
    {
      "epoch": 0.7185217017619252,
      "grad_norm": 0.9770917892456055,
      "learning_rate": 3.876100035811555e-05,
      "loss": 1.3953,
      "step": 5016
    },
    {
      "epoch": 0.718664947715227,
      "grad_norm": 1.0446397066116333,
      "learning_rate": 3.872432692842591e-05,
      "loss": 1.6165,
      "step": 5017
    },
    {
      "epoch": 0.7188081936685289,
      "grad_norm": 1.059687852859497,
      "learning_rate": 3.868766668987236e-05,
      "loss": 1.4572,
      "step": 5018
    },
    {
      "epoch": 0.7189514396218307,
      "grad_norm": 1.1223117113113403,
      "learning_rate": 3.8651019650346955e-05,
      "loss": 1.4853,
      "step": 5019
    },
    {
      "epoch": 0.7190946855751325,
      "grad_norm": 1.1878575086593628,
      "learning_rate": 3.8614385817738794e-05,
      "loss": 1.485,
      "step": 5020
    },
    {
      "epoch": 0.7192379315284343,
      "grad_norm": 1.1482911109924316,
      "learning_rate": 3.8577765199934354e-05,
      "loss": 1.4172,
      "step": 5021
    },
    {
      "epoch": 0.7193811774817361,
      "grad_norm": 1.0261939764022827,
      "learning_rate": 3.854115780481713e-05,
      "loss": 1.515,
      "step": 5022
    },
    {
      "epoch": 0.719524423435038,
      "grad_norm": 1.0533255338668823,
      "learning_rate": 3.8504563640267735e-05,
      "loss": 1.521,
      "step": 5023
    },
    {
      "epoch": 0.7196676693883398,
      "grad_norm": 1.0538556575775146,
      "learning_rate": 3.8467982714163994e-05,
      "loss": 1.561,
      "step": 5024
    },
    {
      "epoch": 0.7198109153416415,
      "grad_norm": 1.0932462215423584,
      "learning_rate": 3.843141503438091e-05,
      "loss": 1.3337,
      "step": 5025
    },
    {
      "epoch": 0.7199541612949434,
      "grad_norm": 0.9990554451942444,
      "learning_rate": 3.839486060879057e-05,
      "loss": 1.2824,
      "step": 5026
    },
    {
      "epoch": 0.7200974072482452,
      "grad_norm": 1.1891356706619263,
      "learning_rate": 3.8358319445262256e-05,
      "loss": 1.2647,
      "step": 5027
    },
    {
      "epoch": 0.7202406532015471,
      "grad_norm": 1.3223845958709717,
      "learning_rate": 3.832179155166238e-05,
      "loss": 1.3523,
      "step": 5028
    },
    {
      "epoch": 0.7203838991548489,
      "grad_norm": 1.1276562213897705,
      "learning_rate": 3.828527693585451e-05,
      "loss": 1.3522,
      "step": 5029
    },
    {
      "epoch": 0.7205271451081507,
      "grad_norm": 1.06482994556427,
      "learning_rate": 3.8248775605699285e-05,
      "loss": 1.3855,
      "step": 5030
    },
    {
      "epoch": 0.7206703910614525,
      "grad_norm": 1.0633624792099,
      "learning_rate": 3.821228756905452e-05,
      "loss": 1.3052,
      "step": 5031
    },
    {
      "epoch": 0.7208136370147543,
      "grad_norm": 1.0099753141403198,
      "learning_rate": 3.817581283377532e-05,
      "loss": 1.3413,
      "step": 5032
    },
    {
      "epoch": 0.7209568829680562,
      "grad_norm": 1.0035628080368042,
      "learning_rate": 3.813935140771365e-05,
      "loss": 1.1599,
      "step": 5033
    },
    {
      "epoch": 0.721100128921358,
      "grad_norm": 1.297598123550415,
      "learning_rate": 3.810290329871882e-05,
      "loss": 1.2384,
      "step": 5034
    },
    {
      "epoch": 0.7212433748746598,
      "grad_norm": 1.1277319192886353,
      "learning_rate": 3.8066468514637186e-05,
      "loss": 1.3573,
      "step": 5035
    },
    {
      "epoch": 0.7213866208279616,
      "grad_norm": 1.2562358379364014,
      "learning_rate": 3.803004706331228e-05,
      "loss": 1.544,
      "step": 5036
    },
    {
      "epoch": 0.7215298667812634,
      "grad_norm": 1.0702641010284424,
      "learning_rate": 3.7993638952584744e-05,
      "loss": 1.3725,
      "step": 5037
    },
    {
      "epoch": 0.7216731127345652,
      "grad_norm": 1.4319472312927246,
      "learning_rate": 3.7957244190292264e-05,
      "loss": 1.4949,
      "step": 5038
    },
    {
      "epoch": 0.7218163586878671,
      "grad_norm": 1.0116281509399414,
      "learning_rate": 3.792086278426982e-05,
      "loss": 1.3818,
      "step": 5039
    },
    {
      "epoch": 0.7219596046411689,
      "grad_norm": 1.0370063781738281,
      "learning_rate": 3.788449474234943e-05,
      "loss": 1.4451,
      "step": 5040
    },
    {
      "epoch": 0.7221028505944707,
      "grad_norm": 1.1651767492294312,
      "learning_rate": 3.784814007236016e-05,
      "loss": 1.4135,
      "step": 5041
    },
    {
      "epoch": 0.7222460965477725,
      "grad_norm": 1.0801609754562378,
      "learning_rate": 3.781179878212829e-05,
      "loss": 1.3637,
      "step": 5042
    },
    {
      "epoch": 0.7223893425010743,
      "grad_norm": 1.114863634109497,
      "learning_rate": 3.777547087947729e-05,
      "loss": 1.3898,
      "step": 5043
    },
    {
      "epoch": 0.7225325884543762,
      "grad_norm": 1.1830052137374878,
      "learning_rate": 3.773915637222756e-05,
      "loss": 1.3423,
      "step": 5044
    },
    {
      "epoch": 0.722675834407678,
      "grad_norm": 1.0711772441864014,
      "learning_rate": 3.770285526819674e-05,
      "loss": 1.3016,
      "step": 5045
    },
    {
      "epoch": 0.7228190803609797,
      "grad_norm": 1.1650713682174683,
      "learning_rate": 3.766656757519956e-05,
      "loss": 1.3565,
      "step": 5046
    },
    {
      "epoch": 0.7229623263142816,
      "grad_norm": 0.9908804297447205,
      "learning_rate": 3.763029330104788e-05,
      "loss": 1.3839,
      "step": 5047
    },
    {
      "epoch": 0.7231055722675834,
      "grad_norm": 1.3818883895874023,
      "learning_rate": 3.759403245355068e-05,
      "loss": 1.2427,
      "step": 5048
    },
    {
      "epoch": 0.7232488182208853,
      "grad_norm": 0.8899428248405457,
      "learning_rate": 3.7557785040513905e-05,
      "loss": 1.2043,
      "step": 5049
    },
    {
      "epoch": 0.7233920641741871,
      "grad_norm": 1.0241647958755493,
      "learning_rate": 3.752155106974085e-05,
      "loss": 1.4297,
      "step": 5050
    },
    {
      "epoch": 0.723535310127489,
      "grad_norm": 1.2215874195098877,
      "learning_rate": 3.7485330549031775e-05,
      "loss": 1.3857,
      "step": 5051
    },
    {
      "epoch": 0.7236785560807907,
      "grad_norm": 1.0548607110977173,
      "learning_rate": 3.744912348618399e-05,
      "loss": 1.4524,
      "step": 5052
    },
    {
      "epoch": 0.7238218020340925,
      "grad_norm": 1.242562174797058,
      "learning_rate": 3.741292988899204e-05,
      "loss": 1.5793,
      "step": 5053
    },
    {
      "epoch": 0.7239650479873944,
      "grad_norm": 0.9969272613525391,
      "learning_rate": 3.737674976524749e-05,
      "loss": 1.4581,
      "step": 5054
    },
    {
      "epoch": 0.7241082939406962,
      "grad_norm": 0.9887746572494507,
      "learning_rate": 3.734058312273904e-05,
      "loss": 1.4328,
      "step": 5055
    },
    {
      "epoch": 0.724251539893998,
      "grad_norm": 1.2633026838302612,
      "learning_rate": 3.730442996925245e-05,
      "loss": 1.5669,
      "step": 5056
    },
    {
      "epoch": 0.7243947858472998,
      "grad_norm": 1.0923923254013062,
      "learning_rate": 3.726829031257062e-05,
      "loss": 1.315,
      "step": 5057
    },
    {
      "epoch": 0.7245380318006016,
      "grad_norm": 1.240658164024353,
      "learning_rate": 3.723216416047355e-05,
      "loss": 1.352,
      "step": 5058
    },
    {
      "epoch": 0.7246812777539035,
      "grad_norm": 0.872370183467865,
      "learning_rate": 3.7196051520738204e-05,
      "loss": 1.5188,
      "step": 5059
    },
    {
      "epoch": 0.7248245237072053,
      "grad_norm": 1.0220781564712524,
      "learning_rate": 3.7159952401138844e-05,
      "loss": 1.3101,
      "step": 5060
    },
    {
      "epoch": 0.7249677696605071,
      "grad_norm": 1.0093969106674194,
      "learning_rate": 3.712386680944672e-05,
      "loss": 1.4861,
      "step": 5061
    },
    {
      "epoch": 0.7251110156138089,
      "grad_norm": 1.1113320589065552,
      "learning_rate": 3.708779475343009e-05,
      "loss": 1.4861,
      "step": 5062
    },
    {
      "epoch": 0.7252542615671107,
      "grad_norm": 1.0081902742385864,
      "learning_rate": 3.705173624085438e-05,
      "loss": 1.3892,
      "step": 5063
    },
    {
      "epoch": 0.7253975075204125,
      "grad_norm": 1.1471179723739624,
      "learning_rate": 3.70156912794822e-05,
      "loss": 1.1921,
      "step": 5064
    },
    {
      "epoch": 0.7255407534737144,
      "grad_norm": 1.2230408191680908,
      "learning_rate": 3.6979659877073034e-05,
      "loss": 1.2967,
      "step": 5065
    },
    {
      "epoch": 0.7256839994270162,
      "grad_norm": 1.2109131813049316,
      "learning_rate": 3.694364204138359e-05,
      "loss": 1.4671,
      "step": 5066
    },
    {
      "epoch": 0.725827245380318,
      "grad_norm": 1.0959734916687012,
      "learning_rate": 3.6907637780167604e-05,
      "loss": 1.3604,
      "step": 5067
    },
    {
      "epoch": 0.7259704913336198,
      "grad_norm": 1.0213323831558228,
      "learning_rate": 3.687164710117592e-05,
      "loss": 1.3356,
      "step": 5068
    },
    {
      "epoch": 0.7261137372869216,
      "grad_norm": 1.0476126670837402,
      "learning_rate": 3.6835670012156456e-05,
      "loss": 1.4403,
      "step": 5069
    },
    {
      "epoch": 0.7262569832402235,
      "grad_norm": 1.0093845129013062,
      "learning_rate": 3.6799706520854094e-05,
      "loss": 1.5003,
      "step": 5070
    },
    {
      "epoch": 0.7264002291935253,
      "grad_norm": 1.210913062095642,
      "learning_rate": 3.6763756635010993e-05,
      "loss": 1.5885,
      "step": 5071
    },
    {
      "epoch": 0.7265434751468272,
      "grad_norm": 0.9551475644111633,
      "learning_rate": 3.672782036236627e-05,
      "loss": 1.4741,
      "step": 5072
    },
    {
      "epoch": 0.7266867211001289,
      "grad_norm": 1.0842825174331665,
      "learning_rate": 3.669189771065604e-05,
      "loss": 1.4197,
      "step": 5073
    },
    {
      "epoch": 0.7268299670534307,
      "grad_norm": 1.1217559576034546,
      "learning_rate": 3.6655988687613605e-05,
      "loss": 1.4862,
      "step": 5074
    },
    {
      "epoch": 0.7269732130067326,
      "grad_norm": 1.084631323814392,
      "learning_rate": 3.6620093300969284e-05,
      "loss": 1.3352,
      "step": 5075
    },
    {
      "epoch": 0.7271164589600344,
      "grad_norm": 1.1429111957550049,
      "learning_rate": 3.658421155845047e-05,
      "loss": 1.5375,
      "step": 5076
    },
    {
      "epoch": 0.7272597049133362,
      "grad_norm": 1.1932175159454346,
      "learning_rate": 3.654834346778161e-05,
      "loss": 1.3544,
      "step": 5077
    },
    {
      "epoch": 0.727402950866638,
      "grad_norm": 1.2409740686416626,
      "learning_rate": 3.651248903668421e-05,
      "loss": 1.2834,
      "step": 5078
    },
    {
      "epoch": 0.7275461968199398,
      "grad_norm": 1.1142911911010742,
      "learning_rate": 3.647664827287688e-05,
      "loss": 1.2448,
      "step": 5079
    },
    {
      "epoch": 0.7276894427732417,
      "grad_norm": 1.0831637382507324,
      "learning_rate": 3.644082118407519e-05,
      "loss": 1.5494,
      "step": 5080
    },
    {
      "epoch": 0.7278326887265435,
      "grad_norm": 1.108092188835144,
      "learning_rate": 3.640500777799182e-05,
      "loss": 1.309,
      "step": 5081
    },
    {
      "epoch": 0.7279759346798453,
      "grad_norm": 1.0170727968215942,
      "learning_rate": 3.636920806233661e-05,
      "loss": 1.4854,
      "step": 5082
    },
    {
      "epoch": 0.7281191806331471,
      "grad_norm": 1.1278061866760254,
      "learning_rate": 3.6333422044816256e-05,
      "loss": 1.3182,
      "step": 5083
    },
    {
      "epoch": 0.7282624265864489,
      "grad_norm": 1.0599029064178467,
      "learning_rate": 3.629764973313463e-05,
      "loss": 1.4213,
      "step": 5084
    },
    {
      "epoch": 0.7284056725397507,
      "grad_norm": 1.0957692861557007,
      "learning_rate": 3.626189113499262e-05,
      "loss": 1.4757,
      "step": 5085
    },
    {
      "epoch": 0.7285489184930526,
      "grad_norm": 1.0854219198226929,
      "learning_rate": 3.622614625808819e-05,
      "loss": 1.4341,
      "step": 5086
    },
    {
      "epoch": 0.7286921644463544,
      "grad_norm": 1.1914632320404053,
      "learning_rate": 3.6190415110116336e-05,
      "loss": 1.207,
      "step": 5087
    },
    {
      "epoch": 0.7288354103996562,
      "grad_norm": 0.8960670828819275,
      "learning_rate": 3.6154697698769005e-05,
      "loss": 1.5671,
      "step": 5088
    },
    {
      "epoch": 0.728978656352958,
      "grad_norm": 1.4607712030410767,
      "learning_rate": 3.6118994031735365e-05,
      "loss": 1.3311,
      "step": 5089
    },
    {
      "epoch": 0.7291219023062598,
      "grad_norm": 1.1879698038101196,
      "learning_rate": 3.6083304116701535e-05,
      "loss": 1.4848,
      "step": 5090
    },
    {
      "epoch": 0.7292651482595617,
      "grad_norm": 1.1448615789413452,
      "learning_rate": 3.604762796135059e-05,
      "loss": 1.3068,
      "step": 5091
    },
    {
      "epoch": 0.7294083942128635,
      "grad_norm": 1.0488063097000122,
      "learning_rate": 3.601196557336275e-05,
      "loss": 1.2553,
      "step": 5092
    },
    {
      "epoch": 0.7295516401661654,
      "grad_norm": 1.0632097721099854,
      "learning_rate": 3.5976316960415334e-05,
      "loss": 1.3257,
      "step": 5093
    },
    {
      "epoch": 0.7296948861194671,
      "grad_norm": 1.0820939540863037,
      "learning_rate": 3.594068213018249e-05,
      "loss": 1.4148,
      "step": 5094
    },
    {
      "epoch": 0.7298381320727689,
      "grad_norm": 1.1446694135665894,
      "learning_rate": 3.590506109033558e-05,
      "loss": 1.4865,
      "step": 5095
    },
    {
      "epoch": 0.7299813780260708,
      "grad_norm": 1.1402205228805542,
      "learning_rate": 3.58694538485429e-05,
      "loss": 1.4474,
      "step": 5096
    },
    {
      "epoch": 0.7301246239793726,
      "grad_norm": 1.3481671810150146,
      "learning_rate": 3.583386041246982e-05,
      "loss": 1.6087,
      "step": 5097
    },
    {
      "epoch": 0.7302678699326745,
      "grad_norm": 1.242810606956482,
      "learning_rate": 3.579828078977877e-05,
      "loss": 1.2383,
      "step": 5098
    },
    {
      "epoch": 0.7304111158859762,
      "grad_norm": 1.1956006288528442,
      "learning_rate": 3.576271498812904e-05,
      "loss": 1.4584,
      "step": 5099
    },
    {
      "epoch": 0.730554361839278,
      "grad_norm": 1.2181980609893799,
      "learning_rate": 3.572716301517719e-05,
      "loss": 1.2328,
      "step": 5100
    },
    {
      "epoch": 0.7306976077925799,
      "grad_norm": 1.0998504161834717,
      "learning_rate": 3.5691624878576666e-05,
      "loss": 1.5332,
      "step": 5101
    },
    {
      "epoch": 0.7308408537458817,
      "grad_norm": 1.0666316747665405,
      "learning_rate": 3.565610058597789e-05,
      "loss": 1.4041,
      "step": 5102
    },
    {
      "epoch": 0.7309840996991835,
      "grad_norm": 1.1092159748077393,
      "learning_rate": 3.56205901450284e-05,
      "loss": 1.3183,
      "step": 5103
    },
    {
      "epoch": 0.7311273456524853,
      "grad_norm": 0.8322303295135498,
      "learning_rate": 3.5585093563372706e-05,
      "loss": 1.4843,
      "step": 5104
    },
    {
      "epoch": 0.7312705916057871,
      "grad_norm": 1.0714200735092163,
      "learning_rate": 3.5549610848652335e-05,
      "loss": 1.4965,
      "step": 5105
    },
    {
      "epoch": 0.731413837559089,
      "grad_norm": 1.228163480758667,
      "learning_rate": 3.551414200850586e-05,
      "loss": 1.5391,
      "step": 5106
    },
    {
      "epoch": 0.7315570835123908,
      "grad_norm": 1.0059279203414917,
      "learning_rate": 3.547868705056882e-05,
      "loss": 1.3441,
      "step": 5107
    },
    {
      "epoch": 0.7317003294656926,
      "grad_norm": 1.0050758123397827,
      "learning_rate": 3.544324598247386e-05,
      "loss": 1.2361,
      "step": 5108
    },
    {
      "epoch": 0.7318435754189944,
      "grad_norm": 1.0628576278686523,
      "learning_rate": 3.540781881185046e-05,
      "loss": 1.5581,
      "step": 5109
    },
    {
      "epoch": 0.7319868213722962,
      "grad_norm": 0.9868745803833008,
      "learning_rate": 3.537240554632523e-05,
      "loss": 1.253,
      "step": 5110
    },
    {
      "epoch": 0.732130067325598,
      "grad_norm": 1.0466634035110474,
      "learning_rate": 3.533700619352187e-05,
      "loss": 1.2713,
      "step": 5111
    },
    {
      "epoch": 0.7322733132788999,
      "grad_norm": 1.196458339691162,
      "learning_rate": 3.530162076106088e-05,
      "loss": 1.2757,
      "step": 5112
    },
    {
      "epoch": 0.7324165592322017,
      "grad_norm": 1.0627546310424805,
      "learning_rate": 3.52662492565599e-05,
      "loss": 1.2646,
      "step": 5113
    },
    {
      "epoch": 0.7325598051855035,
      "grad_norm": 1.017972469329834,
      "learning_rate": 3.523089168763355e-05,
      "loss": 1.3142,
      "step": 5114
    },
    {
      "epoch": 0.7327030511388053,
      "grad_norm": 0.9428439736366272,
      "learning_rate": 3.519554806189343e-05,
      "loss": 1.4131,
      "step": 5115
    },
    {
      "epoch": 0.7328462970921071,
      "grad_norm": 1.3508042097091675,
      "learning_rate": 3.516021838694815e-05,
      "loss": 1.378,
      "step": 5116
    },
    {
      "epoch": 0.732989543045409,
      "grad_norm": 1.1308386325836182,
      "learning_rate": 3.512490267040333e-05,
      "loss": 1.2781,
      "step": 5117
    },
    {
      "epoch": 0.7331327889987108,
      "grad_norm": 1.1428152322769165,
      "learning_rate": 3.5089600919861535e-05,
      "loss": 1.1993,
      "step": 5118
    },
    {
      "epoch": 0.7332760349520127,
      "grad_norm": 1.0723371505737305,
      "learning_rate": 3.505431314292243e-05,
      "loss": 1.4097,
      "step": 5119
    },
    {
      "epoch": 0.7334192809053144,
      "grad_norm": 1.0461530685424805,
      "learning_rate": 3.501903934718247e-05,
      "loss": 1.4394,
      "step": 5120
    },
    {
      "epoch": 0.7335625268586162,
      "grad_norm": 1.0470033884048462,
      "learning_rate": 3.498377954023534e-05,
      "loss": 1.5315,
      "step": 5121
    },
    {
      "epoch": 0.7337057728119181,
      "grad_norm": 1.0908406972885132,
      "learning_rate": 3.4948533729671616e-05,
      "loss": 1.2682,
      "step": 5122
    },
    {
      "epoch": 0.7338490187652199,
      "grad_norm": 1.1356652975082397,
      "learning_rate": 3.4913301923078765e-05,
      "loss": 1.3389,
      "step": 5123
    },
    {
      "epoch": 0.7339922647185217,
      "grad_norm": 0.9959984421730042,
      "learning_rate": 3.4878084128041366e-05,
      "loss": 1.4553,
      "step": 5124
    },
    {
      "epoch": 0.7341355106718235,
      "grad_norm": 1.1513054370880127,
      "learning_rate": 3.484288035214095e-05,
      "loss": 1.3737,
      "step": 5125
    },
    {
      "epoch": 0.7342787566251253,
      "grad_norm": 1.2114496231079102,
      "learning_rate": 3.480769060295599e-05,
      "loss": 1.4182,
      "step": 5126
    },
    {
      "epoch": 0.7344220025784272,
      "grad_norm": 1.0501726865768433,
      "learning_rate": 3.477251488806199e-05,
      "loss": 1.3188,
      "step": 5127
    },
    {
      "epoch": 0.734565248531729,
      "grad_norm": 1.175553560256958,
      "learning_rate": 3.47373532150314e-05,
      "loss": 1.4866,
      "step": 5128
    },
    {
      "epoch": 0.7347084944850308,
      "grad_norm": 1.1490381956100464,
      "learning_rate": 3.47022055914337e-05,
      "loss": 1.4673,
      "step": 5129
    },
    {
      "epoch": 0.7348517404383326,
      "grad_norm": 0.9892501831054688,
      "learning_rate": 3.466707202483523e-05,
      "loss": 1.462,
      "step": 5130
    },
    {
      "epoch": 0.7349949863916344,
      "grad_norm": 1.251485824584961,
      "learning_rate": 3.463195252279939e-05,
      "loss": 1.4227,
      "step": 5131
    },
    {
      "epoch": 0.7351382323449362,
      "grad_norm": 1.1222338676452637,
      "learning_rate": 3.459684709288663e-05,
      "loss": 1.4306,
      "step": 5132
    },
    {
      "epoch": 0.7352814782982381,
      "grad_norm": 1.1060682535171509,
      "learning_rate": 3.456175574265418e-05,
      "loss": 1.4484,
      "step": 5133
    },
    {
      "epoch": 0.7354247242515399,
      "grad_norm": 1.0340923070907593,
      "learning_rate": 3.452667847965636e-05,
      "loss": 1.4221,
      "step": 5134
    },
    {
      "epoch": 0.7355679702048417,
      "grad_norm": 1.0253523588180542,
      "learning_rate": 3.449161531144447e-05,
      "loss": 1.3562,
      "step": 5135
    },
    {
      "epoch": 0.7357112161581435,
      "grad_norm": 1.041393518447876,
      "learning_rate": 3.4456566245566715e-05,
      "loss": 1.4077,
      "step": 5136
    },
    {
      "epoch": 0.7358544621114453,
      "grad_norm": 0.9283429980278015,
      "learning_rate": 3.4421531289568324e-05,
      "loss": 1.4198,
      "step": 5137
    },
    {
      "epoch": 0.7359977080647472,
      "grad_norm": 1.090798258781433,
      "learning_rate": 3.438651045099137e-05,
      "loss": 1.4804,
      "step": 5138
    },
    {
      "epoch": 0.736140954018049,
      "grad_norm": 1.0441683530807495,
      "learning_rate": 3.4351503737375065e-05,
      "loss": 1.5021,
      "step": 5139
    },
    {
      "epoch": 0.7362841999713509,
      "grad_norm": 1.1803879737854004,
      "learning_rate": 3.4316511156255494e-05,
      "loss": 1.187,
      "step": 5140
    },
    {
      "epoch": 0.7364274459246526,
      "grad_norm": 1.0683974027633667,
      "learning_rate": 3.428153271516562e-05,
      "loss": 1.2241,
      "step": 5141
    },
    {
      "epoch": 0.7365706918779544,
      "grad_norm": 1.0174381732940674,
      "learning_rate": 3.424656842163545e-05,
      "loss": 1.2744,
      "step": 5142
    },
    {
      "epoch": 0.7367139378312563,
      "grad_norm": 0.9443196654319763,
      "learning_rate": 3.4211618283192014e-05,
      "loss": 1.3846,
      "step": 5143
    },
    {
      "epoch": 0.7368571837845581,
      "grad_norm": 0.9626579284667969,
      "learning_rate": 3.417668230735912e-05,
      "loss": 1.343,
      "step": 5144
    },
    {
      "epoch": 0.73700042973786,
      "grad_norm": 1.0951868295669556,
      "learning_rate": 3.414176050165765e-05,
      "loss": 1.4742,
      "step": 5145
    },
    {
      "epoch": 0.7371436756911617,
      "grad_norm": 1.2620041370391846,
      "learning_rate": 3.41068528736054e-05,
      "loss": 1.5067,
      "step": 5146
    },
    {
      "epoch": 0.7372869216444635,
      "grad_norm": 1.128720760345459,
      "learning_rate": 3.4071959430717124e-05,
      "loss": 1.3316,
      "step": 5147
    },
    {
      "epoch": 0.7374301675977654,
      "grad_norm": 1.0559282302856445,
      "learning_rate": 3.403708018050456e-05,
      "loss": 1.127,
      "step": 5148
    },
    {
      "epoch": 0.7375734135510672,
      "grad_norm": 1.3836275339126587,
      "learning_rate": 3.4002215130476236e-05,
      "loss": 1.2041,
      "step": 5149
    },
    {
      "epoch": 0.737716659504369,
      "grad_norm": 1.0276132822036743,
      "learning_rate": 3.3967364288137824e-05,
      "loss": 1.2263,
      "step": 5150
    },
    {
      "epoch": 0.7378599054576708,
      "grad_norm": 1.2269060611724854,
      "learning_rate": 3.393252766099187e-05,
      "loss": 1.4776,
      "step": 5151
    },
    {
      "epoch": 0.7380031514109726,
      "grad_norm": 1.0339391231536865,
      "learning_rate": 3.389770525653777e-05,
      "loss": 1.3279,
      "step": 5152
    },
    {
      "epoch": 0.7381463973642745,
      "grad_norm": 1.130679965019226,
      "learning_rate": 3.386289708227194e-05,
      "loss": 1.6257,
      "step": 5153
    },
    {
      "epoch": 0.7382896433175763,
      "grad_norm": 1.0850118398666382,
      "learning_rate": 3.3828103145687726e-05,
      "loss": 1.286,
      "step": 5154
    },
    {
      "epoch": 0.7384328892708781,
      "grad_norm": 1.1260206699371338,
      "learning_rate": 3.379332345427541e-05,
      "loss": 1.3295,
      "step": 5155
    },
    {
      "epoch": 0.7385761352241799,
      "grad_norm": 1.1195250749588013,
      "learning_rate": 3.375855801552219e-05,
      "loss": 1.4368,
      "step": 5156
    },
    {
      "epoch": 0.7387193811774817,
      "grad_norm": 0.9693967700004578,
      "learning_rate": 3.372380683691221e-05,
      "loss": 1.2948,
      "step": 5157
    },
    {
      "epoch": 0.7388626271307835,
      "grad_norm": 0.9901983737945557,
      "learning_rate": 3.368906992592656e-05,
      "loss": 1.4092,
      "step": 5158
    },
    {
      "epoch": 0.7390058730840854,
      "grad_norm": 1.116674542427063,
      "learning_rate": 3.36543472900432e-05,
      "loss": 1.3243,
      "step": 5159
    },
    {
      "epoch": 0.7391491190373872,
      "grad_norm": 0.9755721688270569,
      "learning_rate": 3.361963893673703e-05,
      "loss": 1.4994,
      "step": 5160
    },
    {
      "epoch": 0.7392923649906891,
      "grad_norm": 1.034799575805664,
      "learning_rate": 3.358494487348e-05,
      "loss": 1.311,
      "step": 5161
    },
    {
      "epoch": 0.7394356109439908,
      "grad_norm": 0.9473602175712585,
      "learning_rate": 3.355026510774079e-05,
      "loss": 1.4483,
      "step": 5162
    },
    {
      "epoch": 0.7395788568972926,
      "grad_norm": 1.32988440990448,
      "learning_rate": 3.3515599646985116e-05,
      "loss": 1.3638,
      "step": 5163
    },
    {
      "epoch": 0.7397221028505945,
      "grad_norm": 0.9939714670181274,
      "learning_rate": 3.3480948498675616e-05,
      "loss": 1.2631,
      "step": 5164
    },
    {
      "epoch": 0.7398653488038963,
      "grad_norm": 1.0577690601348877,
      "learning_rate": 3.344631167027181e-05,
      "loss": 1.3235,
      "step": 5165
    },
    {
      "epoch": 0.7400085947571982,
      "grad_norm": 1.0976585149765015,
      "learning_rate": 3.341168916923019e-05,
      "loss": 1.4051,
      "step": 5166
    },
    {
      "epoch": 0.7401518407104999,
      "grad_norm": 1.1382545232772827,
      "learning_rate": 3.3377081003004017e-05,
      "loss": 1.468,
      "step": 5167
    },
    {
      "epoch": 0.7402950866638017,
      "grad_norm": 1.0567196607589722,
      "learning_rate": 3.334248717904368e-05,
      "loss": 1.3457,
      "step": 5168
    },
    {
      "epoch": 0.7404383326171036,
      "grad_norm": 1.3578484058380127,
      "learning_rate": 3.330790770479636e-05,
      "loss": 1.3705,
      "step": 5169
    },
    {
      "epoch": 0.7405815785704054,
      "grad_norm": 1.237394094467163,
      "learning_rate": 3.327334258770611e-05,
      "loss": 1.4007,
      "step": 5170
    },
    {
      "epoch": 0.7407248245237072,
      "grad_norm": 1.1391876935958862,
      "learning_rate": 3.3238791835213944e-05,
      "loss": 1.4434,
      "step": 5171
    },
    {
      "epoch": 0.740868070477009,
      "grad_norm": 0.958648145198822,
      "learning_rate": 3.320425545475787e-05,
      "loss": 1.6004,
      "step": 5172
    },
    {
      "epoch": 0.7410113164303108,
      "grad_norm": 1.034090518951416,
      "learning_rate": 3.316973345377263e-05,
      "loss": 1.4909,
      "step": 5173
    },
    {
      "epoch": 0.7411545623836127,
      "grad_norm": 0.9962390065193176,
      "learning_rate": 3.313522583969e-05,
      "loss": 1.501,
      "step": 5174
    },
    {
      "epoch": 0.7412978083369145,
      "grad_norm": 1.2294760942459106,
      "learning_rate": 3.3100732619938576e-05,
      "loss": 1.3623,
      "step": 5175
    },
    {
      "epoch": 0.7414410542902163,
      "grad_norm": 1.0550763607025146,
      "learning_rate": 3.306625380194394e-05,
      "loss": 1.3483,
      "step": 5176
    },
    {
      "epoch": 0.7415843002435181,
      "grad_norm": 0.9557855129241943,
      "learning_rate": 3.303178939312849e-05,
      "loss": 1.6285,
      "step": 5177
    },
    {
      "epoch": 0.7417275461968199,
      "grad_norm": 1.0615184307098389,
      "learning_rate": 3.299733940091159e-05,
      "loss": 1.2086,
      "step": 5178
    },
    {
      "epoch": 0.7418707921501217,
      "grad_norm": 1.1278741359710693,
      "learning_rate": 3.296290383270948e-05,
      "loss": 1.4017,
      "step": 5179
    },
    {
      "epoch": 0.7420140381034236,
      "grad_norm": 1.0902199745178223,
      "learning_rate": 3.292848269593524e-05,
      "loss": 1.287,
      "step": 5180
    },
    {
      "epoch": 0.7421572840567254,
      "grad_norm": 0.8778855800628662,
      "learning_rate": 3.2894075997998876e-05,
      "loss": 1.4942,
      "step": 5181
    },
    {
      "epoch": 0.7423005300100273,
      "grad_norm": 1.0814718008041382,
      "learning_rate": 3.28596837463074e-05,
      "loss": 1.5752,
      "step": 5182
    },
    {
      "epoch": 0.742443775963329,
      "grad_norm": 1.0368082523345947,
      "learning_rate": 3.282530594826452e-05,
      "loss": 1.4614,
      "step": 5183
    },
    {
      "epoch": 0.7425870219166308,
      "grad_norm": 1.0122549533843994,
      "learning_rate": 3.2790942611270955e-05,
      "loss": 1.4966,
      "step": 5184
    },
    {
      "epoch": 0.7427302678699327,
      "grad_norm": 1.2061089277267456,
      "learning_rate": 3.2756593742724274e-05,
      "loss": 1.4116,
      "step": 5185
    },
    {
      "epoch": 0.7428735138232345,
      "grad_norm": 1.0547289848327637,
      "learning_rate": 3.272225935001895e-05,
      "loss": 1.4862,
      "step": 5186
    },
    {
      "epoch": 0.7430167597765364,
      "grad_norm": 1.1018575429916382,
      "learning_rate": 3.268793944054636e-05,
      "loss": 1.4485,
      "step": 5187
    },
    {
      "epoch": 0.7431600057298381,
      "grad_norm": 1.1554844379425049,
      "learning_rate": 3.265363402169461e-05,
      "loss": 1.3146,
      "step": 5188
    },
    {
      "epoch": 0.7433032516831399,
      "grad_norm": 1.0814515352249146,
      "learning_rate": 3.261934310084894e-05,
      "loss": 1.4899,
      "step": 5189
    },
    {
      "epoch": 0.7434464976364418,
      "grad_norm": 0.9772927761077881,
      "learning_rate": 3.258506668539132e-05,
      "loss": 1.4447,
      "step": 5190
    },
    {
      "epoch": 0.7435897435897436,
      "grad_norm": 0.9453019499778748,
      "learning_rate": 3.255080478270054e-05,
      "loss": 1.3508,
      "step": 5191
    },
    {
      "epoch": 0.7437329895430455,
      "grad_norm": 1.1947277784347534,
      "learning_rate": 3.251655740015236e-05,
      "loss": 1.7345,
      "step": 5192
    },
    {
      "epoch": 0.7438762354963472,
      "grad_norm": 1.1774908304214478,
      "learning_rate": 3.248232454511947e-05,
      "loss": 1.4239,
      "step": 5193
    },
    {
      "epoch": 0.744019481449649,
      "grad_norm": 1.0628496408462524,
      "learning_rate": 3.2448106224971275e-05,
      "loss": 1.4381,
      "step": 5194
    },
    {
      "epoch": 0.7441627274029509,
      "grad_norm": 1.1268256902694702,
      "learning_rate": 3.2413902447074164e-05,
      "loss": 1.4024,
      "step": 5195
    },
    {
      "epoch": 0.7443059733562527,
      "grad_norm": 0.9839563369750977,
      "learning_rate": 3.2379713218791355e-05,
      "loss": 1.2616,
      "step": 5196
    },
    {
      "epoch": 0.7444492193095545,
      "grad_norm": 1.020828127861023,
      "learning_rate": 3.2345538547482945e-05,
      "loss": 1.2967,
      "step": 5197
    },
    {
      "epoch": 0.7445924652628563,
      "grad_norm": 1.2020199298858643,
      "learning_rate": 3.231137844050593e-05,
      "loss": 1.4465,
      "step": 5198
    },
    {
      "epoch": 0.7447357112161581,
      "grad_norm": 1.198256254196167,
      "learning_rate": 3.227723290521405e-05,
      "loss": 1.5287,
      "step": 5199
    },
    {
      "epoch": 0.74487895716946,
      "grad_norm": 1.0634926557540894,
      "learning_rate": 3.224310194895807e-05,
      "loss": 1.3536,
      "step": 5200
    },
    {
      "epoch": 0.7450222031227618,
      "grad_norm": 1.2458806037902832,
      "learning_rate": 3.220898557908555e-05,
      "loss": 1.4251,
      "step": 5201
    },
    {
      "epoch": 0.7451654490760636,
      "grad_norm": 1.0965760946273804,
      "learning_rate": 3.217488380294083e-05,
      "loss": 1.4223,
      "step": 5202
    },
    {
      "epoch": 0.7453086950293655,
      "grad_norm": 1.2408809661865234,
      "learning_rate": 3.214079662786523e-05,
      "loss": 1.4341,
      "step": 5203
    },
    {
      "epoch": 0.7454519409826672,
      "grad_norm": 0.9288257956504822,
      "learning_rate": 3.210672406119686e-05,
      "loss": 1.4318,
      "step": 5204
    },
    {
      "epoch": 0.745595186935969,
      "grad_norm": 1.2427265644073486,
      "learning_rate": 3.207266611027069e-05,
      "loss": 1.3204,
      "step": 5205
    },
    {
      "epoch": 0.7457384328892709,
      "grad_norm": 1.1435372829437256,
      "learning_rate": 3.203862278241857e-05,
      "loss": 1.329,
      "step": 5206
    },
    {
      "epoch": 0.7458816788425727,
      "grad_norm": 0.9676604866981506,
      "learning_rate": 3.200459408496919e-05,
      "loss": 1.4805,
      "step": 5207
    },
    {
      "epoch": 0.7460249247958746,
      "grad_norm": 1.1434203386306763,
      "learning_rate": 3.197058002524811e-05,
      "loss": 1.2603,
      "step": 5208
    },
    {
      "epoch": 0.7461681707491763,
      "grad_norm": 1.0523406267166138,
      "learning_rate": 3.1936580610577636e-05,
      "loss": 1.4589,
      "step": 5209
    },
    {
      "epoch": 0.7463114167024781,
      "grad_norm": 1.3663837909698486,
      "learning_rate": 3.190259584827704e-05,
      "loss": 1.359,
      "step": 5210
    },
    {
      "epoch": 0.74645466265578,
      "grad_norm": 1.2006990909576416,
      "learning_rate": 3.186862574566245e-05,
      "loss": 1.4907,
      "step": 5211
    },
    {
      "epoch": 0.7465979086090818,
      "grad_norm": 1.1332813501358032,
      "learning_rate": 3.1834670310046734e-05,
      "loss": 1.4201,
      "step": 5212
    },
    {
      "epoch": 0.7467411545623837,
      "grad_norm": 1.0703095197677612,
      "learning_rate": 3.180072954873966e-05,
      "loss": 1.3224,
      "step": 5213
    },
    {
      "epoch": 0.7468844005156854,
      "grad_norm": 1.0745210647583008,
      "learning_rate": 3.1766803469047846e-05,
      "loss": 1.3616,
      "step": 5214
    },
    {
      "epoch": 0.7470276464689872,
      "grad_norm": 1.0645780563354492,
      "learning_rate": 3.1732892078274735e-05,
      "loss": 1.4676,
      "step": 5215
    },
    {
      "epoch": 0.7471708924222891,
      "grad_norm": 1.211129903793335,
      "learning_rate": 3.1698995383720645e-05,
      "loss": 1.4181,
      "step": 5216
    },
    {
      "epoch": 0.7473141383755909,
      "grad_norm": 1.1194862127304077,
      "learning_rate": 3.166511339268259e-05,
      "loss": 1.4322,
      "step": 5217
    },
    {
      "epoch": 0.7474573843288927,
      "grad_norm": 1.154075026512146,
      "learning_rate": 3.163124611245464e-05,
      "loss": 1.4809,
      "step": 5218
    },
    {
      "epoch": 0.7476006302821945,
      "grad_norm": 1.2102794647216797,
      "learning_rate": 3.1597393550327556e-05,
      "loss": 1.308,
      "step": 5219
    },
    {
      "epoch": 0.7477438762354963,
      "grad_norm": 1.0487525463104248,
      "learning_rate": 3.1563555713588924e-05,
      "loss": 1.4788,
      "step": 5220
    },
    {
      "epoch": 0.7478871221887982,
      "grad_norm": 1.1613438129425049,
      "learning_rate": 3.152973260952315e-05,
      "loss": 1.5997,
      "step": 5221
    },
    {
      "epoch": 0.7480303681421,
      "grad_norm": 1.1017931699752808,
      "learning_rate": 3.149592424541166e-05,
      "loss": 1.4539,
      "step": 5222
    },
    {
      "epoch": 0.7481736140954018,
      "grad_norm": 1.0918571949005127,
      "learning_rate": 3.146213062853243e-05,
      "loss": 1.4629,
      "step": 5223
    },
    {
      "epoch": 0.7483168600487036,
      "grad_norm": 1.0165950059890747,
      "learning_rate": 3.1428351766160415e-05,
      "loss": 1.3913,
      "step": 5224
    },
    {
      "epoch": 0.7484601060020054,
      "grad_norm": 0.9308567643165588,
      "learning_rate": 3.139458766556739e-05,
      "loss": 1.4879,
      "step": 5225
    },
    {
      "epoch": 0.7486033519553073,
      "grad_norm": 1.1123707294464111,
      "learning_rate": 3.136083833402192e-05,
      "loss": 1.3793,
      "step": 5226
    },
    {
      "epoch": 0.7487465979086091,
      "grad_norm": 1.0654362440109253,
      "learning_rate": 3.132710377878942e-05,
      "loss": 1.5298,
      "step": 5227
    },
    {
      "epoch": 0.7488898438619109,
      "grad_norm": 0.9506818056106567,
      "learning_rate": 3.1293384007132035e-05,
      "loss": 1.6356,
      "step": 5228
    },
    {
      "epoch": 0.7490330898152128,
      "grad_norm": 1.0375168323516846,
      "learning_rate": 3.1259679026308875e-05,
      "loss": 1.3733,
      "step": 5229
    },
    {
      "epoch": 0.7491763357685145,
      "grad_norm": 1.2448631525039673,
      "learning_rate": 3.12259888435758e-05,
      "loss": 1.3093,
      "step": 5230
    },
    {
      "epoch": 0.7493195817218163,
      "grad_norm": 1.1992628574371338,
      "learning_rate": 3.11923134661854e-05,
      "loss": 1.3638,
      "step": 5231
    },
    {
      "epoch": 0.7494628276751182,
      "grad_norm": 1.0951753854751587,
      "learning_rate": 3.1158652901387186e-05,
      "loss": 1.3924,
      "step": 5232
    },
    {
      "epoch": 0.74960607362842,
      "grad_norm": 0.9657092094421387,
      "learning_rate": 3.112500715642743e-05,
      "loss": 1.4188,
      "step": 5233
    },
    {
      "epoch": 0.7497493195817219,
      "grad_norm": 1.0975433588027954,
      "learning_rate": 3.1091376238549265e-05,
      "loss": 1.4857,
      "step": 5234
    },
    {
      "epoch": 0.7498925655350236,
      "grad_norm": 1.3587557077407837,
      "learning_rate": 3.105776015499255e-05,
      "loss": 1.2851,
      "step": 5235
    },
    {
      "epoch": 0.7500358114883254,
      "grad_norm": 1.4095728397369385,
      "learning_rate": 3.102415891299403e-05,
      "loss": 1.2907,
      "step": 5236
    },
    {
      "epoch": 0.7501790574416273,
      "grad_norm": 1.101123571395874,
      "learning_rate": 3.099057251978724e-05,
      "loss": 1.5271,
      "step": 5237
    },
    {
      "epoch": 0.7503223033949291,
      "grad_norm": 1.1078753471374512,
      "learning_rate": 3.0957000982602436e-05,
      "loss": 1.2573,
      "step": 5238
    },
    {
      "epoch": 0.750465549348231,
      "grad_norm": 1.1320340633392334,
      "learning_rate": 3.092344430866674e-05,
      "loss": 1.4062,
      "step": 5239
    },
    {
      "epoch": 0.7506087953015327,
      "grad_norm": 1.1358118057250977,
      "learning_rate": 3.088990250520417e-05,
      "loss": 1.3566,
      "step": 5240
    },
    {
      "epoch": 0.7507520412548345,
      "grad_norm": 0.9678964614868164,
      "learning_rate": 3.085637557943535e-05,
      "loss": 1.4855,
      "step": 5241
    },
    {
      "epoch": 0.7508952872081364,
      "grad_norm": 0.896219789981842,
      "learning_rate": 3.082286353857782e-05,
      "loss": 1.3572,
      "step": 5242
    },
    {
      "epoch": 0.7510385331614382,
      "grad_norm": 1.0818207263946533,
      "learning_rate": 3.0789366389845905e-05,
      "loss": 1.4082,
      "step": 5243
    },
    {
      "epoch": 0.75118177911474,
      "grad_norm": 1.2900702953338623,
      "learning_rate": 3.0755884140450705e-05,
      "loss": 1.3919,
      "step": 5244
    },
    {
      "epoch": 0.7513250250680418,
      "grad_norm": 0.9471254944801331,
      "learning_rate": 3.072241679760014e-05,
      "loss": 1.4226,
      "step": 5245
    },
    {
      "epoch": 0.7514682710213436,
      "grad_norm": 1.1165205240249634,
      "learning_rate": 3.068896436849888e-05,
      "loss": 1.3225,
      "step": 5246
    },
    {
      "epoch": 0.7516115169746455,
      "grad_norm": 1.321968674659729,
      "learning_rate": 3.0655526860348396e-05,
      "loss": 1.3786,
      "step": 5247
    },
    {
      "epoch": 0.7517547629279473,
      "grad_norm": 1.009322166442871,
      "learning_rate": 3.062210428034701e-05,
      "loss": 1.3402,
      "step": 5248
    },
    {
      "epoch": 0.7518980088812491,
      "grad_norm": 1.2682723999023438,
      "learning_rate": 3.058869663568967e-05,
      "loss": 1.393,
      "step": 5249
    },
    {
      "epoch": 0.752041254834551,
      "grad_norm": 1.3009908199310303,
      "learning_rate": 3.055530393356831e-05,
      "loss": 1.4349,
      "step": 5250
    },
    {
      "epoch": 0.7521845007878527,
      "grad_norm": 1.046252727508545,
      "learning_rate": 3.0521926181171566e-05,
      "loss": 1.3119,
      "step": 5251
    },
    {
      "epoch": 0.7523277467411545,
      "grad_norm": 1.127808928489685,
      "learning_rate": 3.0488563385684764e-05,
      "loss": 1.215,
      "step": 5252
    },
    {
      "epoch": 0.7524709926944564,
      "grad_norm": 1.0358679294586182,
      "learning_rate": 3.0455215554290128e-05,
      "loss": 1.2998,
      "step": 5253
    },
    {
      "epoch": 0.7526142386477582,
      "grad_norm": 1.1149992942810059,
      "learning_rate": 3.0421882694166602e-05,
      "loss": 1.5254,
      "step": 5254
    },
    {
      "epoch": 0.7527574846010601,
      "grad_norm": 1.065881609916687,
      "learning_rate": 3.038856481248996e-05,
      "loss": 1.3185,
      "step": 5255
    },
    {
      "epoch": 0.7529007305543618,
      "grad_norm": 0.9617979526519775,
      "learning_rate": 3.0355261916432688e-05,
      "loss": 1.3173,
      "step": 5256
    },
    {
      "epoch": 0.7530439765076636,
      "grad_norm": 1.0790705680847168,
      "learning_rate": 3.032197401316409e-05,
      "loss": 1.3111,
      "step": 5257
    },
    {
      "epoch": 0.7531872224609655,
      "grad_norm": 1.1235885620117188,
      "learning_rate": 3.028870110985025e-05,
      "loss": 1.395,
      "step": 5258
    },
    {
      "epoch": 0.7533304684142673,
      "grad_norm": 1.0750095844268799,
      "learning_rate": 3.0255443213653943e-05,
      "loss": 1.2471,
      "step": 5259
    },
    {
      "epoch": 0.7534737143675692,
      "grad_norm": 1.0138497352600098,
      "learning_rate": 3.0222200331734772e-05,
      "loss": 1.3654,
      "step": 5260
    },
    {
      "epoch": 0.7536169603208709,
      "grad_norm": 1.2020118236541748,
      "learning_rate": 3.0188972471249198e-05,
      "loss": 1.494,
      "step": 5261
    },
    {
      "epoch": 0.7537602062741727,
      "grad_norm": 1.0309174060821533,
      "learning_rate": 3.015575963935027e-05,
      "loss": 1.2139,
      "step": 5262
    },
    {
      "epoch": 0.7539034522274746,
      "grad_norm": 0.9640597105026245,
      "learning_rate": 3.0122561843187914e-05,
      "loss": 1.4568,
      "step": 5263
    },
    {
      "epoch": 0.7540466981807764,
      "grad_norm": 1.1459474563598633,
      "learning_rate": 3.0089379089908786e-05,
      "loss": 1.4002,
      "step": 5264
    },
    {
      "epoch": 0.7541899441340782,
      "grad_norm": 1.067918062210083,
      "learning_rate": 3.005621138665633e-05,
      "loss": 1.5095,
      "step": 5265
    },
    {
      "epoch": 0.75433319008738,
      "grad_norm": 0.9664956331253052,
      "learning_rate": 3.0023058740570754e-05,
      "loss": 1.4564,
      "step": 5266
    },
    {
      "epoch": 0.7544764360406818,
      "grad_norm": 1.1434638500213623,
      "learning_rate": 2.9989921158788902e-05,
      "loss": 1.4323,
      "step": 5267
    },
    {
      "epoch": 0.7546196819939837,
      "grad_norm": 1.0754722356796265,
      "learning_rate": 2.9956798648444584e-05,
      "loss": 1.307,
      "step": 5268
    },
    {
      "epoch": 0.7547629279472855,
      "grad_norm": 1.0213178396224976,
      "learning_rate": 2.9923691216668238e-05,
      "loss": 1.5007,
      "step": 5269
    },
    {
      "epoch": 0.7549061739005873,
      "grad_norm": 1.0263917446136475,
      "learning_rate": 2.989059887058703e-05,
      "loss": 1.4063,
      "step": 5270
    },
    {
      "epoch": 0.7550494198538892,
      "grad_norm": 1.573898434638977,
      "learning_rate": 2.9857521617324914e-05,
      "loss": 1.3927,
      "step": 5271
    },
    {
      "epoch": 0.7551926658071909,
      "grad_norm": 1.0526155233383179,
      "learning_rate": 2.9824459464002697e-05,
      "loss": 1.2306,
      "step": 5272
    },
    {
      "epoch": 0.7553359117604928,
      "grad_norm": 0.9442198276519775,
      "learning_rate": 2.979141241773775e-05,
      "loss": 1.4535,
      "step": 5273
    },
    {
      "epoch": 0.7554791577137946,
      "grad_norm": 1.1698716878890991,
      "learning_rate": 2.9758380485644323e-05,
      "loss": 1.3061,
      "step": 5274
    },
    {
      "epoch": 0.7556224036670964,
      "grad_norm": 1.0386513471603394,
      "learning_rate": 2.9725363674833362e-05,
      "loss": 1.3918,
      "step": 5275
    },
    {
      "epoch": 0.7557656496203983,
      "grad_norm": 0.9259045124053955,
      "learning_rate": 2.9692361992412577e-05,
      "loss": 1.3563,
      "step": 5276
    },
    {
      "epoch": 0.7559088955737,
      "grad_norm": 0.883860170841217,
      "learning_rate": 2.965937544548645e-05,
      "loss": 1.5507,
      "step": 5277
    },
    {
      "epoch": 0.7560521415270018,
      "grad_norm": 1.2496848106384277,
      "learning_rate": 2.9626404041156053e-05,
      "loss": 1.4985,
      "step": 5278
    },
    {
      "epoch": 0.7561953874803037,
      "grad_norm": 1.0576503276824951,
      "learning_rate": 2.9593447786519425e-05,
      "loss": 1.4358,
      "step": 5279
    },
    {
      "epoch": 0.7563386334336055,
      "grad_norm": 1.0769339799880981,
      "learning_rate": 2.956050668867123e-05,
      "loss": 1.47,
      "step": 5280
    },
    {
      "epoch": 0.7564818793869074,
      "grad_norm": 1.1733505725860596,
      "learning_rate": 2.952758075470281e-05,
      "loss": 1.4854,
      "step": 5281
    },
    {
      "epoch": 0.7566251253402091,
      "grad_norm": 1.1533260345458984,
      "learning_rate": 2.949466999170233e-05,
      "loss": 1.4455,
      "step": 5282
    },
    {
      "epoch": 0.7567683712935109,
      "grad_norm": 1.0212055444717407,
      "learning_rate": 2.946177440675466e-05,
      "loss": 1.4097,
      "step": 5283
    },
    {
      "epoch": 0.7569116172468128,
      "grad_norm": 1.1643275022506714,
      "learning_rate": 2.942889400694141e-05,
      "loss": 1.3051,
      "step": 5284
    },
    {
      "epoch": 0.7570548632001146,
      "grad_norm": 1.2512731552124023,
      "learning_rate": 2.9396028799340924e-05,
      "loss": 1.4109,
      "step": 5285
    },
    {
      "epoch": 0.7571981091534165,
      "grad_norm": 1.0877684354782104,
      "learning_rate": 2.9363178791028257e-05,
      "loss": 1.3648,
      "step": 5286
    },
    {
      "epoch": 0.7573413551067182,
      "grad_norm": 1.0833747386932373,
      "learning_rate": 2.9330343989075236e-05,
      "loss": 1.5236,
      "step": 5287
    },
    {
      "epoch": 0.75748460106002,
      "grad_norm": 1.006443738937378,
      "learning_rate": 2.9297524400550325e-05,
      "loss": 1.3639,
      "step": 5288
    },
    {
      "epoch": 0.7576278470133219,
      "grad_norm": 1.2001993656158447,
      "learning_rate": 2.9264720032518756e-05,
      "loss": 1.34,
      "step": 5289
    },
    {
      "epoch": 0.7577710929666237,
      "grad_norm": 1.043632984161377,
      "learning_rate": 2.9231930892042614e-05,
      "loss": 1.3043,
      "step": 5290
    },
    {
      "epoch": 0.7579143389199255,
      "grad_norm": 1.0485841035842896,
      "learning_rate": 2.9199156986180463e-05,
      "loss": 1.371,
      "step": 5291
    },
    {
      "epoch": 0.7580575848732274,
      "grad_norm": 1.188919186592102,
      "learning_rate": 2.9166398321987774e-05,
      "loss": 1.4428,
      "step": 5292
    },
    {
      "epoch": 0.7582008308265291,
      "grad_norm": 0.9221444725990295,
      "learning_rate": 2.9133654906516672e-05,
      "loss": 1.5253,
      "step": 5293
    },
    {
      "epoch": 0.758344076779831,
      "grad_norm": 0.9905220866203308,
      "learning_rate": 2.9100926746815992e-05,
      "loss": 1.5086,
      "step": 5294
    },
    {
      "epoch": 0.7584873227331328,
      "grad_norm": 0.9922628998756409,
      "learning_rate": 2.9068213849931338e-05,
      "loss": 1.5098,
      "step": 5295
    },
    {
      "epoch": 0.7586305686864346,
      "grad_norm": 1.106137752532959,
      "learning_rate": 2.903551622290489e-05,
      "loss": 1.5416,
      "step": 5296
    },
    {
      "epoch": 0.7587738146397365,
      "grad_norm": 1.348939299583435,
      "learning_rate": 2.9002833872775735e-05,
      "loss": 1.2668,
      "step": 5297
    },
    {
      "epoch": 0.7589170605930382,
      "grad_norm": 1.1245379447937012,
      "learning_rate": 2.8970166806579568e-05,
      "loss": 1.1444,
      "step": 5298
    },
    {
      "epoch": 0.75906030654634,
      "grad_norm": 1.4945236444473267,
      "learning_rate": 2.893751503134874e-05,
      "loss": 1.39,
      "step": 5299
    },
    {
      "epoch": 0.7592035524996419,
      "grad_norm": 1.1722297668457031,
      "learning_rate": 2.8904878554112367e-05,
      "loss": 1.2366,
      "step": 5300
    },
    {
      "epoch": 0.7593467984529437,
      "grad_norm": 1.0624130964279175,
      "learning_rate": 2.8872257381896385e-05,
      "loss": 1.4595,
      "step": 5301
    },
    {
      "epoch": 0.7594900444062456,
      "grad_norm": 1.006986379623413,
      "learning_rate": 2.883965152172321e-05,
      "loss": 1.3608,
      "step": 5302
    },
    {
      "epoch": 0.7596332903595473,
      "grad_norm": 1.014585018157959,
      "learning_rate": 2.880706098061211e-05,
      "loss": 1.4577,
      "step": 5303
    },
    {
      "epoch": 0.7597765363128491,
      "grad_norm": 0.8617925047874451,
      "learning_rate": 2.877448576557904e-05,
      "loss": 1.3414,
      "step": 5304
    },
    {
      "epoch": 0.759919782266151,
      "grad_norm": 1.217434048652649,
      "learning_rate": 2.874192588363662e-05,
      "loss": 1.264,
      "step": 5305
    },
    {
      "epoch": 0.7600630282194528,
      "grad_norm": 0.9415032267570496,
      "learning_rate": 2.8709381341794204e-05,
      "loss": 1.4105,
      "step": 5306
    },
    {
      "epoch": 0.7602062741727547,
      "grad_norm": 1.0786097049713135,
      "learning_rate": 2.867685214705781e-05,
      "loss": 1.4373,
      "step": 5307
    },
    {
      "epoch": 0.7603495201260564,
      "grad_norm": 1.1815820932388306,
      "learning_rate": 2.8644338306430208e-05,
      "loss": 1.3294,
      "step": 5308
    },
    {
      "epoch": 0.7604927660793582,
      "grad_norm": 1.1950478553771973,
      "learning_rate": 2.8611839826910757e-05,
      "loss": 1.3058,
      "step": 5309
    },
    {
      "epoch": 0.7606360120326601,
      "grad_norm": 1.2802622318267822,
      "learning_rate": 2.8579356715495577e-05,
      "loss": 1.4427,
      "step": 5310
    },
    {
      "epoch": 0.7607792579859619,
      "grad_norm": 1.1187304258346558,
      "learning_rate": 2.8546888979177578e-05,
      "loss": 1.5523,
      "step": 5311
    },
    {
      "epoch": 0.7609225039392638,
      "grad_norm": 1.20724356174469,
      "learning_rate": 2.851443662494615e-05,
      "loss": 1.3855,
      "step": 5312
    },
    {
      "epoch": 0.7610657498925655,
      "grad_norm": 1.4193682670593262,
      "learning_rate": 2.848199965978753e-05,
      "loss": 1.4285,
      "step": 5313
    },
    {
      "epoch": 0.7612089958458673,
      "grad_norm": 1.416662335395813,
      "learning_rate": 2.8449578090684593e-05,
      "loss": 1.4097,
      "step": 5314
    },
    {
      "epoch": 0.7613522417991692,
      "grad_norm": 1.062666654586792,
      "learning_rate": 2.841717192461688e-05,
      "loss": 1.3727,
      "step": 5315
    },
    {
      "epoch": 0.761495487752471,
      "grad_norm": 0.9586269855499268,
      "learning_rate": 2.8384781168560693e-05,
      "loss": 1.4306,
      "step": 5316
    },
    {
      "epoch": 0.7616387337057728,
      "grad_norm": 1.1697758436203003,
      "learning_rate": 2.835240582948886e-05,
      "loss": 1.357,
      "step": 5317
    },
    {
      "epoch": 0.7617819796590747,
      "grad_norm": 1.032918095588684,
      "learning_rate": 2.8320045914371074e-05,
      "loss": 1.5393,
      "step": 5318
    },
    {
      "epoch": 0.7619252256123764,
      "grad_norm": 1.2699965238571167,
      "learning_rate": 2.828770143017363e-05,
      "loss": 1.1875,
      "step": 5319
    },
    {
      "epoch": 0.7620684715656783,
      "grad_norm": 1.2660740613937378,
      "learning_rate": 2.8255372383859435e-05,
      "loss": 1.4097,
      "step": 5320
    },
    {
      "epoch": 0.7622117175189801,
      "grad_norm": 1.082439661026001,
      "learning_rate": 2.8223058782388134e-05,
      "loss": 1.5422,
      "step": 5321
    },
    {
      "epoch": 0.7623549634722819,
      "grad_norm": 1.01420259475708,
      "learning_rate": 2.8190760632716127e-05,
      "loss": 1.4008,
      "step": 5322
    },
    {
      "epoch": 0.7624982094255838,
      "grad_norm": 1.111799955368042,
      "learning_rate": 2.8158477941796336e-05,
      "loss": 1.337,
      "step": 5323
    },
    {
      "epoch": 0.7626414553788855,
      "grad_norm": 1.2761080265045166,
      "learning_rate": 2.8126210716578427e-05,
      "loss": 1.2787,
      "step": 5324
    },
    {
      "epoch": 0.7627847013321873,
      "grad_norm": 1.1700340509414673,
      "learning_rate": 2.809395896400876e-05,
      "loss": 1.341,
      "step": 5325
    },
    {
      "epoch": 0.7629279472854892,
      "grad_norm": 1.0529197454452515,
      "learning_rate": 2.8061722691030335e-05,
      "loss": 1.2499,
      "step": 5326
    },
    {
      "epoch": 0.763071193238791,
      "grad_norm": 1.132962703704834,
      "learning_rate": 2.8029501904582835e-05,
      "loss": 1.4047,
      "step": 5327
    },
    {
      "epoch": 0.7632144391920929,
      "grad_norm": 1.0735987424850464,
      "learning_rate": 2.799729661160253e-05,
      "loss": 1.4283,
      "step": 5328
    },
    {
      "epoch": 0.7633576851453946,
      "grad_norm": 1.1575231552124023,
      "learning_rate": 2.7965106819022504e-05,
      "loss": 1.5334,
      "step": 5329
    },
    {
      "epoch": 0.7635009310986964,
      "grad_norm": 1.0861072540283203,
      "learning_rate": 2.7932932533772417e-05,
      "loss": 1.5401,
      "step": 5330
    },
    {
      "epoch": 0.7636441770519983,
      "grad_norm": 1.032377004623413,
      "learning_rate": 2.790077376277854e-05,
      "loss": 1.3929,
      "step": 5331
    },
    {
      "epoch": 0.7637874230053001,
      "grad_norm": 1.1348316669464111,
      "learning_rate": 2.786863051296391e-05,
      "loss": 1.4727,
      "step": 5332
    },
    {
      "epoch": 0.763930668958602,
      "grad_norm": 1.0201689004898071,
      "learning_rate": 2.7836502791248142e-05,
      "loss": 1.5093,
      "step": 5333
    },
    {
      "epoch": 0.7640739149119037,
      "grad_norm": 1.0382293462753296,
      "learning_rate": 2.7804390604547557e-05,
      "loss": 1.4964,
      "step": 5334
    },
    {
      "epoch": 0.7642171608652055,
      "grad_norm": 1.093854546546936,
      "learning_rate": 2.777229395977511e-05,
      "loss": 1.4272,
      "step": 5335
    },
    {
      "epoch": 0.7643604068185074,
      "grad_norm": 1.0585594177246094,
      "learning_rate": 2.774021286384042e-05,
      "loss": 1.4228,
      "step": 5336
    },
    {
      "epoch": 0.7645036527718092,
      "grad_norm": 1.0755082368850708,
      "learning_rate": 2.770814732364978e-05,
      "loss": 1.3671,
      "step": 5337
    },
    {
      "epoch": 0.764646898725111,
      "grad_norm": 1.0224575996398926,
      "learning_rate": 2.7676097346106034e-05,
      "loss": 1.3841,
      "step": 5338
    },
    {
      "epoch": 0.7647901446784129,
      "grad_norm": 1.1585297584533691,
      "learning_rate": 2.7644062938108774e-05,
      "loss": 1.2674,
      "step": 5339
    },
    {
      "epoch": 0.7649333906317146,
      "grad_norm": 0.9601103067398071,
      "learning_rate": 2.761204410655428e-05,
      "loss": 1.5596,
      "step": 5340
    },
    {
      "epoch": 0.7650766365850165,
      "grad_norm": 1.2528530359268188,
      "learning_rate": 2.7580040858335345e-05,
      "loss": 1.3788,
      "step": 5341
    },
    {
      "epoch": 0.7652198825383183,
      "grad_norm": 1.1307536363601685,
      "learning_rate": 2.7548053200341496e-05,
      "loss": 1.3003,
      "step": 5342
    },
    {
      "epoch": 0.7653631284916201,
      "grad_norm": 1.1641361713409424,
      "learning_rate": 2.7516081139458883e-05,
      "loss": 1.5079,
      "step": 5343
    },
    {
      "epoch": 0.765506374444922,
      "grad_norm": 1.0204488039016724,
      "learning_rate": 2.7484124682570302e-05,
      "loss": 1.4381,
      "step": 5344
    },
    {
      "epoch": 0.7656496203982237,
      "grad_norm": 1.0582654476165771,
      "learning_rate": 2.7452183836555212e-05,
      "loss": 1.3036,
      "step": 5345
    },
    {
      "epoch": 0.7657928663515255,
      "grad_norm": 1.0235873460769653,
      "learning_rate": 2.7420258608289607e-05,
      "loss": 1.3728,
      "step": 5346
    },
    {
      "epoch": 0.7659361123048274,
      "grad_norm": 1.0706382989883423,
      "learning_rate": 2.7388349004646285e-05,
      "loss": 1.3289,
      "step": 5347
    },
    {
      "epoch": 0.7660793582581292,
      "grad_norm": 1.0654786825180054,
      "learning_rate": 2.7356455032494598e-05,
      "loss": 1.373,
      "step": 5348
    },
    {
      "epoch": 0.7662226042114311,
      "grad_norm": 1.123969316482544,
      "learning_rate": 2.7324576698700453e-05,
      "loss": 1.4721,
      "step": 5349
    },
    {
      "epoch": 0.7663658501647328,
      "grad_norm": 1.0574196577072144,
      "learning_rate": 2.7292714010126484e-05,
      "loss": 1.4121,
      "step": 5350
    },
    {
      "epoch": 0.7665090961180346,
      "grad_norm": 1.080439805984497,
      "learning_rate": 2.7260866973632025e-05,
      "loss": 1.2691,
      "step": 5351
    },
    {
      "epoch": 0.7666523420713365,
      "grad_norm": 1.0477899312973022,
      "learning_rate": 2.722903559607286e-05,
      "loss": 1.4546,
      "step": 5352
    },
    {
      "epoch": 0.7667955880246383,
      "grad_norm": 1.2967371940612793,
      "learning_rate": 2.719721988430153e-05,
      "loss": 1.2958,
      "step": 5353
    },
    {
      "epoch": 0.7669388339779402,
      "grad_norm": 1.31163489818573,
      "learning_rate": 2.7165419845167172e-05,
      "loss": 1.5292,
      "step": 5354
    },
    {
      "epoch": 0.7670820799312419,
      "grad_norm": 0.9819673299789429,
      "learning_rate": 2.713363548551554e-05,
      "loss": 1.4179,
      "step": 5355
    },
    {
      "epoch": 0.7672253258845437,
      "grad_norm": 1.3790018558502197,
      "learning_rate": 2.7101866812189057e-05,
      "loss": 1.248,
      "step": 5356
    },
    {
      "epoch": 0.7673685718378456,
      "grad_norm": 1.0061616897583008,
      "learning_rate": 2.7070113832026643e-05,
      "loss": 1.4502,
      "step": 5357
    },
    {
      "epoch": 0.7675118177911474,
      "grad_norm": 1.0803977251052856,
      "learning_rate": 2.7038376551864008e-05,
      "loss": 1.498,
      "step": 5358
    },
    {
      "epoch": 0.7676550637444493,
      "grad_norm": 1.0983235836029053,
      "learning_rate": 2.7006654978533417e-05,
      "loss": 1.4305,
      "step": 5359
    },
    {
      "epoch": 0.7677983096977511,
      "grad_norm": 1.0291062593460083,
      "learning_rate": 2.697494911886368e-05,
      "loss": 1.3164,
      "step": 5360
    },
    {
      "epoch": 0.7679415556510528,
      "grad_norm": 1.154427170753479,
      "learning_rate": 2.6943258979680308e-05,
      "loss": 1.4109,
      "step": 5361
    },
    {
      "epoch": 0.7680848016043547,
      "grad_norm": 1.14201021194458,
      "learning_rate": 2.6911584567805393e-05,
      "loss": 1.5377,
      "step": 5362
    },
    {
      "epoch": 0.7682280475576565,
      "grad_norm": 1.0191935300827026,
      "learning_rate": 2.687992589005768e-05,
      "loss": 1.3696,
      "step": 5363
    },
    {
      "epoch": 0.7683712935109583,
      "grad_norm": 1.2176258563995361,
      "learning_rate": 2.6848282953252467e-05,
      "loss": 1.5087,
      "step": 5364
    },
    {
      "epoch": 0.7685145394642602,
      "grad_norm": 1.2263424396514893,
      "learning_rate": 2.6816655764201714e-05,
      "loss": 1.3312,
      "step": 5365
    },
    {
      "epoch": 0.7686577854175619,
      "grad_norm": 0.982971727848053,
      "learning_rate": 2.6785044329714004e-05,
      "loss": 1.4073,
      "step": 5366
    },
    {
      "epoch": 0.7688010313708638,
      "grad_norm": 1.1179509162902832,
      "learning_rate": 2.6753448656594393e-05,
      "loss": 1.2986,
      "step": 5367
    },
    {
      "epoch": 0.7689442773241656,
      "grad_norm": 1.0811750888824463,
      "learning_rate": 2.672186875164475e-05,
      "loss": 1.3693,
      "step": 5368
    },
    {
      "epoch": 0.7690875232774674,
      "grad_norm": 0.9156041145324707,
      "learning_rate": 2.6690304621663442e-05,
      "loss": 1.6285,
      "step": 5369
    },
    {
      "epoch": 0.7692307692307693,
      "grad_norm": 1.072241187095642,
      "learning_rate": 2.6658756273445386e-05,
      "loss": 1.4609,
      "step": 5370
    },
    {
      "epoch": 0.769374015184071,
      "grad_norm": 1.5767362117767334,
      "learning_rate": 2.6627223713782157e-05,
      "loss": 1.3948,
      "step": 5371
    },
    {
      "epoch": 0.7695172611373728,
      "grad_norm": 1.1711581945419312,
      "learning_rate": 2.659570694946203e-05,
      "loss": 1.4621,
      "step": 5372
    },
    {
      "epoch": 0.7696605070906747,
      "grad_norm": 1.2863517999649048,
      "learning_rate": 2.6564205987269696e-05,
      "loss": 1.3688,
      "step": 5373
    },
    {
      "epoch": 0.7698037530439765,
      "grad_norm": 1.351944923400879,
      "learning_rate": 2.653272083398656e-05,
      "loss": 1.3103,
      "step": 5374
    },
    {
      "epoch": 0.7699469989972784,
      "grad_norm": 1.1083176136016846,
      "learning_rate": 2.650125149639059e-05,
      "loss": 1.1606,
      "step": 5375
    },
    {
      "epoch": 0.7700902449505801,
      "grad_norm": 1.153162956237793,
      "learning_rate": 2.646979798125636e-05,
      "loss": 1.5447,
      "step": 5376
    },
    {
      "epoch": 0.7702334909038819,
      "grad_norm": 1.1354740858078003,
      "learning_rate": 2.643836029535507e-05,
      "loss": 1.5613,
      "step": 5377
    },
    {
      "epoch": 0.7703767368571838,
      "grad_norm": 1.1255152225494385,
      "learning_rate": 2.6406938445454376e-05,
      "loss": 1.3382,
      "step": 5378
    },
    {
      "epoch": 0.7705199828104856,
      "grad_norm": 1.068695306777954,
      "learning_rate": 2.6375532438318716e-05,
      "loss": 1.4676,
      "step": 5379
    },
    {
      "epoch": 0.7706632287637875,
      "grad_norm": 0.9634365439414978,
      "learning_rate": 2.634414228070904e-05,
      "loss": 1.4151,
      "step": 5380
    },
    {
      "epoch": 0.7708064747170893,
      "grad_norm": 1.0725656747817993,
      "learning_rate": 2.631276797938279e-05,
      "loss": 1.4625,
      "step": 5381
    },
    {
      "epoch": 0.770949720670391,
      "grad_norm": 1.060855507850647,
      "learning_rate": 2.6281409541094127e-05,
      "loss": 1.3603,
      "step": 5382
    },
    {
      "epoch": 0.7710929666236929,
      "grad_norm": 1.0964360237121582,
      "learning_rate": 2.6250066972593735e-05,
      "loss": 1.3861,
      "step": 5383
    },
    {
      "epoch": 0.7712362125769947,
      "grad_norm": 1.2195426225662231,
      "learning_rate": 2.6218740280628896e-05,
      "loss": 1.4804,
      "step": 5384
    },
    {
      "epoch": 0.7713794585302965,
      "grad_norm": 0.9624045491218567,
      "learning_rate": 2.618742947194347e-05,
      "loss": 1.4278,
      "step": 5385
    },
    {
      "epoch": 0.7715227044835984,
      "grad_norm": 1.1158347129821777,
      "learning_rate": 2.615613455327791e-05,
      "loss": 1.6594,
      "step": 5386
    },
    {
      "epoch": 0.7716659504369001,
      "grad_norm": 1.26571524143219,
      "learning_rate": 2.612485553136925e-05,
      "loss": 1.3648,
      "step": 5387
    },
    {
      "epoch": 0.771809196390202,
      "grad_norm": 1.1605147123336792,
      "learning_rate": 2.609359241295104e-05,
      "loss": 1.4667,
      "step": 5388
    },
    {
      "epoch": 0.7719524423435038,
      "grad_norm": 1.053276538848877,
      "learning_rate": 2.6062345204753457e-05,
      "loss": 1.2711,
      "step": 5389
    },
    {
      "epoch": 0.7720956882968056,
      "grad_norm": 0.9543305039405823,
      "learning_rate": 2.6031113913503337e-05,
      "loss": 1.514,
      "step": 5390
    },
    {
      "epoch": 0.7722389342501075,
      "grad_norm": 1.0924980640411377,
      "learning_rate": 2.5999898545923908e-05,
      "loss": 1.4086,
      "step": 5391
    },
    {
      "epoch": 0.7723821802034092,
      "grad_norm": 1.0135517120361328,
      "learning_rate": 2.596869910873512e-05,
      "loss": 1.4942,
      "step": 5392
    },
    {
      "epoch": 0.772525426156711,
      "grad_norm": 0.9889259338378906,
      "learning_rate": 2.5937515608653408e-05,
      "loss": 1.3339,
      "step": 5393
    },
    {
      "epoch": 0.7726686721100129,
      "grad_norm": 1.1338982582092285,
      "learning_rate": 2.5906348052391828e-05,
      "loss": 1.5144,
      "step": 5394
    },
    {
      "epoch": 0.7728119180633147,
      "grad_norm": 0.9133114814758301,
      "learning_rate": 2.587519644666001e-05,
      "loss": 1.3932,
      "step": 5395
    },
    {
      "epoch": 0.7729551640166166,
      "grad_norm": 1.1685774326324463,
      "learning_rate": 2.5844060798164038e-05,
      "loss": 1.3839,
      "step": 5396
    },
    {
      "epoch": 0.7730984099699183,
      "grad_norm": 1.1761701107025146,
      "learning_rate": 2.5812941113606726e-05,
      "loss": 1.388,
      "step": 5397
    },
    {
      "epoch": 0.7732416559232201,
      "grad_norm": 1.1376409530639648,
      "learning_rate": 2.578183739968738e-05,
      "loss": 1.3842,
      "step": 5398
    },
    {
      "epoch": 0.773384901876522,
      "grad_norm": 1.0157896280288696,
      "learning_rate": 2.575074966310179e-05,
      "loss": 1.3521,
      "step": 5399
    },
    {
      "epoch": 0.7735281478298238,
      "grad_norm": 1.0421208143234253,
      "learning_rate": 2.5719677910542394e-05,
      "loss": 1.4851,
      "step": 5400
    },
    {
      "epoch": 0.7736713937831257,
      "grad_norm": 1.0282460451126099,
      "learning_rate": 2.568862214869825e-05,
      "loss": 1.212,
      "step": 5401
    },
    {
      "epoch": 0.7738146397364275,
      "grad_norm": 1.0647400617599487,
      "learning_rate": 2.5657582384254796e-05,
      "loss": 1.4748,
      "step": 5402
    },
    {
      "epoch": 0.7739578856897292,
      "grad_norm": 1.051937222480774,
      "learning_rate": 2.562655862389418e-05,
      "loss": 1.4836,
      "step": 5403
    },
    {
      "epoch": 0.7741011316430311,
      "grad_norm": 1.0603488683700562,
      "learning_rate": 2.5595550874295027e-05,
      "loss": 1.6119,
      "step": 5404
    },
    {
      "epoch": 0.7742443775963329,
      "grad_norm": 1.0364676713943481,
      "learning_rate": 2.556455914213255e-05,
      "loss": 1.3752,
      "step": 5405
    },
    {
      "epoch": 0.7743876235496348,
      "grad_norm": 0.9507237672805786,
      "learning_rate": 2.5533583434078523e-05,
      "loss": 1.0841,
      "step": 5406
    },
    {
      "epoch": 0.7745308695029366,
      "grad_norm": 1.0436803102493286,
      "learning_rate": 2.550262375680117e-05,
      "loss": 1.1878,
      "step": 5407
    },
    {
      "epoch": 0.7746741154562383,
      "grad_norm": 1.3421522378921509,
      "learning_rate": 2.5471680116965425e-05,
      "loss": 1.3296,
      "step": 5408
    },
    {
      "epoch": 0.7748173614095402,
      "grad_norm": 0.9931652545928955,
      "learning_rate": 2.5440752521232692e-05,
      "loss": 1.3149,
      "step": 5409
    },
    {
      "epoch": 0.774960607362842,
      "grad_norm": 1.006361722946167,
      "learning_rate": 2.5409840976260855e-05,
      "loss": 1.4635,
      "step": 5410
    },
    {
      "epoch": 0.7751038533161438,
      "grad_norm": 1.1905099153518677,
      "learning_rate": 2.5378945488704443e-05,
      "loss": 1.5726,
      "step": 5411
    },
    {
      "epoch": 0.7752470992694457,
      "grad_norm": 1.0121513605117798,
      "learning_rate": 2.5348066065214482e-05,
      "loss": 1.4586,
      "step": 5412
    },
    {
      "epoch": 0.7753903452227474,
      "grad_norm": 1.0191130638122559,
      "learning_rate": 2.5317202712438535e-05,
      "loss": 1.255,
      "step": 5413
    },
    {
      "epoch": 0.7755335911760493,
      "grad_norm": 1.1988415718078613,
      "learning_rate": 2.5286355437020746e-05,
      "loss": 1.3644,
      "step": 5414
    },
    {
      "epoch": 0.7756768371293511,
      "grad_norm": 1.2554223537445068,
      "learning_rate": 2.5255524245601748e-05,
      "loss": 1.4097,
      "step": 5415
    },
    {
      "epoch": 0.7758200830826529,
      "grad_norm": 0.9802923798561096,
      "learning_rate": 2.5224709144818782e-05,
      "loss": 1.4447,
      "step": 5416
    },
    {
      "epoch": 0.7759633290359548,
      "grad_norm": 1.3789926767349243,
      "learning_rate": 2.51939101413055e-05,
      "loss": 1.5304,
      "step": 5417
    },
    {
      "epoch": 0.7761065749892565,
      "grad_norm": 1.0385291576385498,
      "learning_rate": 2.5163127241692165e-05,
      "loss": 1.4261,
      "step": 5418
    },
    {
      "epoch": 0.7762498209425583,
      "grad_norm": 0.8672406077384949,
      "learning_rate": 2.5132360452605673e-05,
      "loss": 1.3072,
      "step": 5419
    },
    {
      "epoch": 0.7763930668958602,
      "grad_norm": 1.1061148643493652,
      "learning_rate": 2.5101609780669266e-05,
      "loss": 1.3123,
      "step": 5420
    },
    {
      "epoch": 0.776536312849162,
      "grad_norm": 1.088796854019165,
      "learning_rate": 2.507087523250282e-05,
      "loss": 1.4215,
      "step": 5421
    },
    {
      "epoch": 0.7766795588024639,
      "grad_norm": 1.0772571563720703,
      "learning_rate": 2.5040156814722727e-05,
      "loss": 1.4445,
      "step": 5422
    },
    {
      "epoch": 0.7768228047557656,
      "grad_norm": 1.0032541751861572,
      "learning_rate": 2.50094545339419e-05,
      "loss": 1.4058,
      "step": 5423
    },
    {
      "epoch": 0.7769660507090674,
      "grad_norm": 1.3033168315887451,
      "learning_rate": 2.4978768396769824e-05,
      "loss": 1.3154,
      "step": 5424
    },
    {
      "epoch": 0.7771092966623693,
      "grad_norm": 1.084582805633545,
      "learning_rate": 2.494809840981236e-05,
      "loss": 1.4181,
      "step": 5425
    },
    {
      "epoch": 0.7772525426156711,
      "grad_norm": 1.025909423828125,
      "learning_rate": 2.491744457967209e-05,
      "loss": 1.4027,
      "step": 5426
    },
    {
      "epoch": 0.777395788568973,
      "grad_norm": 1.0471577644348145,
      "learning_rate": 2.4886806912948035e-05,
      "loss": 1.4609,
      "step": 5427
    },
    {
      "epoch": 0.7775390345222748,
      "grad_norm": 1.0506747961044312,
      "learning_rate": 2.4856185416235656e-05,
      "loss": 1.4481,
      "step": 5428
    },
    {
      "epoch": 0.7776822804755765,
      "grad_norm": 1.1074923276901245,
      "learning_rate": 2.4825580096126998e-05,
      "loss": 1.4442,
      "step": 5429
    },
    {
      "epoch": 0.7778255264288784,
      "grad_norm": 1.0668188333511353,
      "learning_rate": 2.479499095921074e-05,
      "loss": 1.4821,
      "step": 5430
    },
    {
      "epoch": 0.7779687723821802,
      "grad_norm": 1.0391645431518555,
      "learning_rate": 2.4764418012071855e-05,
      "loss": 1.406,
      "step": 5431
    },
    {
      "epoch": 0.778112018335482,
      "grad_norm": 1.0142494440078735,
      "learning_rate": 2.473386126129198e-05,
      "loss": 1.37,
      "step": 5432
    },
    {
      "epoch": 0.7782552642887839,
      "grad_norm": 0.8885267972946167,
      "learning_rate": 2.470332071344923e-05,
      "loss": 1.3655,
      "step": 5433
    },
    {
      "epoch": 0.7783985102420856,
      "grad_norm": 1.2007206678390503,
      "learning_rate": 2.4672796375118225e-05,
      "loss": 1.4813,
      "step": 5434
    },
    {
      "epoch": 0.7785417561953875,
      "grad_norm": 1.0071890354156494,
      "learning_rate": 2.4642288252870106e-05,
      "loss": 1.3395,
      "step": 5435
    },
    {
      "epoch": 0.7786850021486893,
      "grad_norm": 1.0513888597488403,
      "learning_rate": 2.461179635327251e-05,
      "loss": 1.3781,
      "step": 5436
    },
    {
      "epoch": 0.7788282481019911,
      "grad_norm": 1.2440940141677856,
      "learning_rate": 2.458132068288962e-05,
      "loss": 1.3451,
      "step": 5437
    },
    {
      "epoch": 0.778971494055293,
      "grad_norm": 1.2658860683441162,
      "learning_rate": 2.4550861248282032e-05,
      "loss": 1.2586,
      "step": 5438
    },
    {
      "epoch": 0.7791147400085947,
      "grad_norm": 1.0538208484649658,
      "learning_rate": 2.4520418056006912e-05,
      "loss": 1.4797,
      "step": 5439
    },
    {
      "epoch": 0.7792579859618965,
      "grad_norm": 1.1901886463165283,
      "learning_rate": 2.4489991112618017e-05,
      "loss": 1.4614,
      "step": 5440
    },
    {
      "epoch": 0.7794012319151984,
      "grad_norm": 0.9623347520828247,
      "learning_rate": 2.4459580424665417e-05,
      "loss": 1.3402,
      "step": 5441
    },
    {
      "epoch": 0.7795444778685002,
      "grad_norm": 1.0573067665100098,
      "learning_rate": 2.4429185998695825e-05,
      "loss": 1.4963,
      "step": 5442
    },
    {
      "epoch": 0.7796877238218021,
      "grad_norm": 1.0454970598220825,
      "learning_rate": 2.43988078412524e-05,
      "loss": 1.2832,
      "step": 5443
    },
    {
      "epoch": 0.7798309697751038,
      "grad_norm": 0.9679948687553406,
      "learning_rate": 2.4368445958874807e-05,
      "loss": 1.4636,
      "step": 5444
    },
    {
      "epoch": 0.7799742157284056,
      "grad_norm": 1.2437002658843994,
      "learning_rate": 2.4338100358099235e-05,
      "loss": 1.5036,
      "step": 5445
    },
    {
      "epoch": 0.7801174616817075,
      "grad_norm": 1.1896706819534302,
      "learning_rate": 2.430777104545826e-05,
      "loss": 1.346,
      "step": 5446
    },
    {
      "epoch": 0.7802607076350093,
      "grad_norm": 1.0673450231552124,
      "learning_rate": 2.4277458027481104e-05,
      "loss": 1.4702,
      "step": 5447
    },
    {
      "epoch": 0.7804039535883112,
      "grad_norm": 1.200490117073059,
      "learning_rate": 2.4247161310693434e-05,
      "loss": 1.327,
      "step": 5448
    },
    {
      "epoch": 0.780547199541613,
      "grad_norm": 1.1671743392944336,
      "learning_rate": 2.4216880901617313e-05,
      "loss": 1.4504,
      "step": 5449
    },
    {
      "epoch": 0.7806904454949147,
      "grad_norm": 1.1693484783172607,
      "learning_rate": 2.4186616806771357e-05,
      "loss": 1.3291,
      "step": 5450
    },
    {
      "epoch": 0.7808336914482166,
      "grad_norm": 1.1167196035385132,
      "learning_rate": 2.415636903267078e-05,
      "loss": 1.4,
      "step": 5451
    },
    {
      "epoch": 0.7809769374015184,
      "grad_norm": 1.4478158950805664,
      "learning_rate": 2.412613758582707e-05,
      "loss": 1.5248,
      "step": 5452
    },
    {
      "epoch": 0.7811201833548203,
      "grad_norm": 0.9325673580169678,
      "learning_rate": 2.4095922472748367e-05,
      "loss": 1.2945,
      "step": 5453
    },
    {
      "epoch": 0.7812634293081221,
      "grad_norm": 0.9552785158157349,
      "learning_rate": 2.4065723699939203e-05,
      "loss": 1.6263,
      "step": 5454
    },
    {
      "epoch": 0.7814066752614238,
      "grad_norm": 0.9998942017555237,
      "learning_rate": 2.4035541273900663e-05,
      "loss": 1.5621,
      "step": 5455
    },
    {
      "epoch": 0.7815499212147257,
      "grad_norm": 1.0076566934585571,
      "learning_rate": 2.4005375201130274e-05,
      "loss": 1.3258,
      "step": 5456
    },
    {
      "epoch": 0.7816931671680275,
      "grad_norm": 1.1170464754104614,
      "learning_rate": 2.3975225488121976e-05,
      "loss": 1.4238,
      "step": 5457
    },
    {
      "epoch": 0.7818364131213293,
      "grad_norm": 1.0396312475204468,
      "learning_rate": 2.3945092141366343e-05,
      "loss": 1.3998,
      "step": 5458
    },
    {
      "epoch": 0.7819796590746312,
      "grad_norm": 1.184221863746643,
      "learning_rate": 2.3914975167350328e-05,
      "loss": 1.4981,
      "step": 5459
    },
    {
      "epoch": 0.7821229050279329,
      "grad_norm": 1.058962106704712,
      "learning_rate": 2.3884874572557316e-05,
      "loss": 1.4708,
      "step": 5460
    },
    {
      "epoch": 0.7822661509812348,
      "grad_norm": 1.1053345203399658,
      "learning_rate": 2.3854790363467262e-05,
      "loss": 1.3753,
      "step": 5461
    },
    {
      "epoch": 0.7824093969345366,
      "grad_norm": 1.124083161354065,
      "learning_rate": 2.3824722546556533e-05,
      "loss": 1.1979,
      "step": 5462
    },
    {
      "epoch": 0.7825526428878384,
      "grad_norm": 1.1387380361557007,
      "learning_rate": 2.3794671128297995e-05,
      "loss": 1.1768,
      "step": 5463
    },
    {
      "epoch": 0.7826958888411403,
      "grad_norm": 1.0596282482147217,
      "learning_rate": 2.3764636115160978e-05,
      "loss": 1.3623,
      "step": 5464
    },
    {
      "epoch": 0.782839134794442,
      "grad_norm": 1.2180850505828857,
      "learning_rate": 2.3734617513611266e-05,
      "loss": 1.2866,
      "step": 5465
    },
    {
      "epoch": 0.7829823807477438,
      "grad_norm": 1.1552622318267822,
      "learning_rate": 2.3704615330111156e-05,
      "loss": 1.3754,
      "step": 5466
    },
    {
      "epoch": 0.7831256267010457,
      "grad_norm": 1.1800016164779663,
      "learning_rate": 2.3674629571119332e-05,
      "loss": 1.4913,
      "step": 5467
    },
    {
      "epoch": 0.7832688726543475,
      "grad_norm": 1.2159316539764404,
      "learning_rate": 2.3644660243090966e-05,
      "loss": 1.3708,
      "step": 5468
    },
    {
      "epoch": 0.7834121186076494,
      "grad_norm": 1.1430907249450684,
      "learning_rate": 2.3614707352477804e-05,
      "loss": 1.2621,
      "step": 5469
    },
    {
      "epoch": 0.7835553645609512,
      "grad_norm": 0.9268341064453125,
      "learning_rate": 2.358477090572787e-05,
      "loss": 1.3888,
      "step": 5470
    },
    {
      "epoch": 0.7836986105142529,
      "grad_norm": 1.1056623458862305,
      "learning_rate": 2.3554850909285786e-05,
      "loss": 1.3127,
      "step": 5471
    },
    {
      "epoch": 0.7838418564675548,
      "grad_norm": 1.0710093975067139,
      "learning_rate": 2.3524947369592575e-05,
      "loss": 1.1727,
      "step": 5472
    },
    {
      "epoch": 0.7839851024208566,
      "grad_norm": 1.2467036247253418,
      "learning_rate": 2.3495060293085735e-05,
      "loss": 1.4337,
      "step": 5473
    },
    {
      "epoch": 0.7841283483741585,
      "grad_norm": 1.0214260816574097,
      "learning_rate": 2.346518968619924e-05,
      "loss": 1.299,
      "step": 5474
    },
    {
      "epoch": 0.7842715943274603,
      "grad_norm": 1.049080729484558,
      "learning_rate": 2.34353355553634e-05,
      "loss": 1.3697,
      "step": 5475
    },
    {
      "epoch": 0.784414840280762,
      "grad_norm": 1.2326804399490356,
      "learning_rate": 2.340549790700517e-05,
      "loss": 1.5754,
      "step": 5476
    },
    {
      "epoch": 0.7845580862340639,
      "grad_norm": 1.212562084197998,
      "learning_rate": 2.337567674754785e-05,
      "loss": 1.2214,
      "step": 5477
    },
    {
      "epoch": 0.7847013321873657,
      "grad_norm": 1.1864029169082642,
      "learning_rate": 2.3345872083411135e-05,
      "loss": 1.444,
      "step": 5478
    },
    {
      "epoch": 0.7848445781406675,
      "grad_norm": 1.181270956993103,
      "learning_rate": 2.3316083921011235e-05,
      "loss": 1.3331,
      "step": 5479
    },
    {
      "epoch": 0.7849878240939694,
      "grad_norm": 1.059954047203064,
      "learning_rate": 2.328631226676089e-05,
      "loss": 1.2838,
      "step": 5480
    },
    {
      "epoch": 0.7851310700472711,
      "grad_norm": 1.1604628562927246,
      "learning_rate": 2.3256557127069124e-05,
      "loss": 1.3718,
      "step": 5481
    },
    {
      "epoch": 0.785274316000573,
      "grad_norm": 1.1982344388961792,
      "learning_rate": 2.3226818508341496e-05,
      "loss": 1.5401,
      "step": 5482
    },
    {
      "epoch": 0.7854175619538748,
      "grad_norm": 1.0517563819885254,
      "learning_rate": 2.3197096416980013e-05,
      "loss": 1.4108,
      "step": 5483
    },
    {
      "epoch": 0.7855608079071766,
      "grad_norm": 0.9944028854370117,
      "learning_rate": 2.3167390859383088e-05,
      "loss": 1.2913,
      "step": 5484
    },
    {
      "epoch": 0.7857040538604785,
      "grad_norm": 1.021971344947815,
      "learning_rate": 2.3137701841945627e-05,
      "loss": 1.383,
      "step": 5485
    },
    {
      "epoch": 0.7858472998137802,
      "grad_norm": 1.0083472728729248,
      "learning_rate": 2.3108029371058848e-05,
      "loss": 1.3285,
      "step": 5486
    },
    {
      "epoch": 0.785990545767082,
      "grad_norm": 1.1019871234893799,
      "learning_rate": 2.30783734531106e-05,
      "loss": 1.4371,
      "step": 5487
    },
    {
      "epoch": 0.7861337917203839,
      "grad_norm": 1.0592209100723267,
      "learning_rate": 2.304873409448506e-05,
      "loss": 1.2965,
      "step": 5488
    },
    {
      "epoch": 0.7862770376736857,
      "grad_norm": 1.2254985570907593,
      "learning_rate": 2.3019111301562802e-05,
      "loss": 1.3391,
      "step": 5489
    },
    {
      "epoch": 0.7864202836269876,
      "grad_norm": 0.9758579730987549,
      "learning_rate": 2.2989505080720886e-05,
      "loss": 1.2837,
      "step": 5490
    },
    {
      "epoch": 0.7865635295802894,
      "grad_norm": 0.9957001805305481,
      "learning_rate": 2.2959915438332823e-05,
      "loss": 1.2922,
      "step": 5491
    },
    {
      "epoch": 0.7867067755335911,
      "grad_norm": 1.3815044164657593,
      "learning_rate": 2.293034238076851e-05,
      "loss": 1.2947,
      "step": 5492
    },
    {
      "epoch": 0.786850021486893,
      "grad_norm": 1.1347591876983643,
      "learning_rate": 2.2900785914394308e-05,
      "loss": 1.4194,
      "step": 5493
    },
    {
      "epoch": 0.7869932674401948,
      "grad_norm": 1.013049840927124,
      "learning_rate": 2.2871246045572993e-05,
      "loss": 1.4999,
      "step": 5494
    },
    {
      "epoch": 0.7871365133934967,
      "grad_norm": 1.357895851135254,
      "learning_rate": 2.2841722780663788e-05,
      "loss": 1.4436,
      "step": 5495
    },
    {
      "epoch": 0.7872797593467985,
      "grad_norm": 1.189643383026123,
      "learning_rate": 2.2812216126022245e-05,
      "loss": 1.4037,
      "step": 5496
    },
    {
      "epoch": 0.7874230053001002,
      "grad_norm": 0.9385799765586853,
      "learning_rate": 2.2782726088000495e-05,
      "loss": 1.4793,
      "step": 5497
    },
    {
      "epoch": 0.7875662512534021,
      "grad_norm": 0.9452044367790222,
      "learning_rate": 2.2753252672947022e-05,
      "loss": 1.3812,
      "step": 5498
    },
    {
      "epoch": 0.7877094972067039,
      "grad_norm": 1.0751274824142456,
      "learning_rate": 2.2723795887206657e-05,
      "loss": 1.5765,
      "step": 5499
    },
    {
      "epoch": 0.7878527431600058,
      "grad_norm": 1.1663727760314941,
      "learning_rate": 2.269435573712071e-05,
      "loss": 1.4953,
      "step": 5500
    },
    {
      "epoch": 0.7879959891133076,
      "grad_norm": 0.9630807638168335,
      "learning_rate": 2.2664932229027024e-05,
      "loss": 1.414,
      "step": 5501
    },
    {
      "epoch": 0.7881392350666093,
      "grad_norm": 1.1125819683074951,
      "learning_rate": 2.2635525369259648e-05,
      "loss": 1.4275,
      "step": 5502
    },
    {
      "epoch": 0.7882824810199112,
      "grad_norm": 1.153269648551941,
      "learning_rate": 2.260613516414919e-05,
      "loss": 1.4855,
      "step": 5503
    },
    {
      "epoch": 0.788425726973213,
      "grad_norm": 1.0586724281311035,
      "learning_rate": 2.2576761620022626e-05,
      "loss": 1.3624,
      "step": 5504
    },
    {
      "epoch": 0.7885689729265148,
      "grad_norm": 1.0902937650680542,
      "learning_rate": 2.254740474320336e-05,
      "loss": 1.3233,
      "step": 5505
    },
    {
      "epoch": 0.7887122188798167,
      "grad_norm": 1.1037815809249878,
      "learning_rate": 2.251806454001123e-05,
      "loss": 1.2333,
      "step": 5506
    },
    {
      "epoch": 0.7888554648331184,
      "grad_norm": 1.0841825008392334,
      "learning_rate": 2.248874101676236e-05,
      "loss": 1.5312,
      "step": 5507
    },
    {
      "epoch": 0.7889987107864203,
      "grad_norm": 1.0819047689437866,
      "learning_rate": 2.245943417976948e-05,
      "loss": 1.3083,
      "step": 5508
    },
    {
      "epoch": 0.7891419567397221,
      "grad_norm": 1.2178468704223633,
      "learning_rate": 2.2430144035341617e-05,
      "loss": 1.3101,
      "step": 5509
    },
    {
      "epoch": 0.7892852026930239,
      "grad_norm": 1.0202821493148804,
      "learning_rate": 2.240087058978415e-05,
      "loss": 1.3679,
      "step": 5510
    },
    {
      "epoch": 0.7894284486463258,
      "grad_norm": 1.1028978824615479,
      "learning_rate": 2.2371613849398975e-05,
      "loss": 1.4634,
      "step": 5511
    },
    {
      "epoch": 0.7895716945996276,
      "grad_norm": 1.057798147201538,
      "learning_rate": 2.234237382048433e-05,
      "loss": 1.3211,
      "step": 5512
    },
    {
      "epoch": 0.7897149405529293,
      "grad_norm": 0.9596744775772095,
      "learning_rate": 2.231315050933488e-05,
      "loss": 1.3493,
      "step": 5513
    },
    {
      "epoch": 0.7898581865062312,
      "grad_norm": 1.0085740089416504,
      "learning_rate": 2.228394392224167e-05,
      "loss": 1.4692,
      "step": 5514
    },
    {
      "epoch": 0.790001432459533,
      "grad_norm": 1.0538007020950317,
      "learning_rate": 2.2254754065492157e-05,
      "loss": 1.3279,
      "step": 5515
    },
    {
      "epoch": 0.7901446784128349,
      "grad_norm": 0.9678391814231873,
      "learning_rate": 2.2225580945370228e-05,
      "loss": 1.442,
      "step": 5516
    },
    {
      "epoch": 0.7902879243661367,
      "grad_norm": 1.027978777885437,
      "learning_rate": 2.2196424568156073e-05,
      "loss": 1.4822,
      "step": 5517
    },
    {
      "epoch": 0.7904311703194384,
      "grad_norm": 1.1491197347640991,
      "learning_rate": 2.2167284940126344e-05,
      "loss": 1.3251,
      "step": 5518
    },
    {
      "epoch": 0.7905744162727403,
      "grad_norm": 1.1693757772445679,
      "learning_rate": 2.2138162067554158e-05,
      "loss": 1.5452,
      "step": 5519
    },
    {
      "epoch": 0.7907176622260421,
      "grad_norm": 0.9138875603675842,
      "learning_rate": 2.210905595670887e-05,
      "loss": 1.3953,
      "step": 5520
    },
    {
      "epoch": 0.790860908179344,
      "grad_norm": 1.3093968629837036,
      "learning_rate": 2.207996661385634e-05,
      "loss": 1.3597,
      "step": 5521
    },
    {
      "epoch": 0.7910041541326458,
      "grad_norm": 0.940271258354187,
      "learning_rate": 2.2050894045258762e-05,
      "loss": 1.3858,
      "step": 5522
    },
    {
      "epoch": 0.7911474000859475,
      "grad_norm": 1.0557079315185547,
      "learning_rate": 2.2021838257174765e-05,
      "loss": 1.3408,
      "step": 5523
    },
    {
      "epoch": 0.7912906460392494,
      "grad_norm": 1.1002577543258667,
      "learning_rate": 2.1992799255859364e-05,
      "loss": 1.4487,
      "step": 5524
    },
    {
      "epoch": 0.7914338919925512,
      "grad_norm": 1.179927945137024,
      "learning_rate": 2.196377704756385e-05,
      "loss": 1.4006,
      "step": 5525
    },
    {
      "epoch": 0.791577137945853,
      "grad_norm": 0.9497984051704407,
      "learning_rate": 2.1934771638536054e-05,
      "loss": 1.3803,
      "step": 5526
    },
    {
      "epoch": 0.7917203838991549,
      "grad_norm": 1.2308974266052246,
      "learning_rate": 2.1905783035020157e-05,
      "loss": 1.3779,
      "step": 5527
    },
    {
      "epoch": 0.7918636298524566,
      "grad_norm": 1.0323413610458374,
      "learning_rate": 2.18768112432566e-05,
      "loss": 1.5662,
      "step": 5528
    },
    {
      "epoch": 0.7920068758057585,
      "grad_norm": 0.9781385064125061,
      "learning_rate": 2.1847856269482303e-05,
      "loss": 1.4688,
      "step": 5529
    },
    {
      "epoch": 0.7921501217590603,
      "grad_norm": 1.124023199081421,
      "learning_rate": 2.1818918119930644e-05,
      "loss": 1.3386,
      "step": 5530
    },
    {
      "epoch": 0.7922933677123621,
      "grad_norm": 1.148544192314148,
      "learning_rate": 2.1789996800831215e-05,
      "loss": 1.3778,
      "step": 5531
    },
    {
      "epoch": 0.792436613665664,
      "grad_norm": 1.3279540538787842,
      "learning_rate": 2.1761092318410072e-05,
      "loss": 1.3828,
      "step": 5532
    },
    {
      "epoch": 0.7925798596189657,
      "grad_norm": 0.8857930898666382,
      "learning_rate": 2.1732204678889632e-05,
      "loss": 1.3991,
      "step": 5533
    },
    {
      "epoch": 0.7927231055722676,
      "grad_norm": 0.9784302115440369,
      "learning_rate": 2.1703333888488708e-05,
      "loss": 1.3432,
      "step": 5534
    },
    {
      "epoch": 0.7928663515255694,
      "grad_norm": 1.1189216375350952,
      "learning_rate": 2.1674479953422477e-05,
      "loss": 1.3938,
      "step": 5535
    },
    {
      "epoch": 0.7930095974788712,
      "grad_norm": 1.2653486728668213,
      "learning_rate": 2.1645642879902406e-05,
      "loss": 1.2727,
      "step": 5536
    },
    {
      "epoch": 0.7931528434321731,
      "grad_norm": 1.0303517580032349,
      "learning_rate": 2.1616822674136473e-05,
      "loss": 1.3938,
      "step": 5537
    },
    {
      "epoch": 0.7932960893854749,
      "grad_norm": 0.9577755331993103,
      "learning_rate": 2.1588019342328968e-05,
      "loss": 1.3339,
      "step": 5538
    },
    {
      "epoch": 0.7934393353387766,
      "grad_norm": 1.4209814071655273,
      "learning_rate": 2.155923289068048e-05,
      "loss": 1.4727,
      "step": 5539
    },
    {
      "epoch": 0.7935825812920785,
      "grad_norm": 1.0015904903411865,
      "learning_rate": 2.153046332538804e-05,
      "loss": 1.5235,
      "step": 5540
    },
    {
      "epoch": 0.7937258272453803,
      "grad_norm": 0.9848486185073853,
      "learning_rate": 2.1501710652645034e-05,
      "loss": 1.3215,
      "step": 5541
    },
    {
      "epoch": 0.7938690731986822,
      "grad_norm": 1.1095030307769775,
      "learning_rate": 2.1472974878641183e-05,
      "loss": 1.0968,
      "step": 5542
    },
    {
      "epoch": 0.794012319151984,
      "grad_norm": 1.1565525531768799,
      "learning_rate": 2.1444256009562602e-05,
      "loss": 1.4391,
      "step": 5543
    },
    {
      "epoch": 0.7941555651052857,
      "grad_norm": 1.272929072380066,
      "learning_rate": 2.1415554051591746e-05,
      "loss": 1.5163,
      "step": 5544
    },
    {
      "epoch": 0.7942988110585876,
      "grad_norm": 1.0581684112548828,
      "learning_rate": 2.1386869010907472e-05,
      "loss": 1.2667,
      "step": 5545
    },
    {
      "epoch": 0.7944420570118894,
      "grad_norm": 1.1608738899230957,
      "learning_rate": 2.1358200893684898e-05,
      "loss": 1.4381,
      "step": 5546
    },
    {
      "epoch": 0.7945853029651913,
      "grad_norm": 1.0682607889175415,
      "learning_rate": 2.1329549706095562e-05,
      "loss": 1.3557,
      "step": 5547
    },
    {
      "epoch": 0.7947285489184931,
      "grad_norm": 0.9385896325111389,
      "learning_rate": 2.1300915454307435e-05,
      "loss": 1.4743,
      "step": 5548
    },
    {
      "epoch": 0.7948717948717948,
      "grad_norm": 1.1274384260177612,
      "learning_rate": 2.1272298144484682e-05,
      "loss": 1.3438,
      "step": 5549
    },
    {
      "epoch": 0.7950150408250967,
      "grad_norm": 1.1000089645385742,
      "learning_rate": 2.124369778278794e-05,
      "loss": 1.5051,
      "step": 5550
    },
    {
      "epoch": 0.7951582867783985,
      "grad_norm": 1.0289571285247803,
      "learning_rate": 2.1215114375374144e-05,
      "loss": 1.5719,
      "step": 5551
    },
    {
      "epoch": 0.7953015327317003,
      "grad_norm": 1.1477240324020386,
      "learning_rate": 2.1186547928396594e-05,
      "loss": 1.383,
      "step": 5552
    },
    {
      "epoch": 0.7954447786850022,
      "grad_norm": 1.076686978340149,
      "learning_rate": 2.115799844800498e-05,
      "loss": 1.292,
      "step": 5553
    },
    {
      "epoch": 0.7955880246383039,
      "grad_norm": 1.0538661479949951,
      "learning_rate": 2.1129465940345206e-05,
      "loss": 1.3934,
      "step": 5554
    },
    {
      "epoch": 0.7957312705916058,
      "grad_norm": 1.0516835451126099,
      "learning_rate": 2.1100950411559706e-05,
      "loss": 1.4139,
      "step": 5555
    },
    {
      "epoch": 0.7958745165449076,
      "grad_norm": 1.0002833604812622,
      "learning_rate": 2.1072451867787146e-05,
      "loss": 1.3807,
      "step": 5556
    },
    {
      "epoch": 0.7960177624982094,
      "grad_norm": 1.0306679010391235,
      "learning_rate": 2.104397031516253e-05,
      "loss": 1.3901,
      "step": 5557
    },
    {
      "epoch": 0.7961610084515113,
      "grad_norm": 1.416664719581604,
      "learning_rate": 2.1015505759817223e-05,
      "loss": 1.3207,
      "step": 5558
    },
    {
      "epoch": 0.7963042544048131,
      "grad_norm": 1.305562138557434,
      "learning_rate": 2.098705820787901e-05,
      "loss": 1.4521,
      "step": 5559
    },
    {
      "epoch": 0.7964475003581148,
      "grad_norm": 1.0682446956634521,
      "learning_rate": 2.0958627665471865e-05,
      "loss": 1.3705,
      "step": 5560
    },
    {
      "epoch": 0.7965907463114167,
      "grad_norm": 1.4303133487701416,
      "learning_rate": 2.093021413871622e-05,
      "loss": 1.5328,
      "step": 5561
    },
    {
      "epoch": 0.7967339922647185,
      "grad_norm": 1.2248668670654297,
      "learning_rate": 2.0901817633728804e-05,
      "loss": 1.5614,
      "step": 5562
    },
    {
      "epoch": 0.7968772382180204,
      "grad_norm": 1.2573518753051758,
      "learning_rate": 2.087343815662267e-05,
      "loss": 1.2541,
      "step": 5563
    },
    {
      "epoch": 0.7970204841713222,
      "grad_norm": 1.0623184442520142,
      "learning_rate": 2.0845075713507222e-05,
      "loss": 1.4891,
      "step": 5564
    },
    {
      "epoch": 0.7971637301246239,
      "grad_norm": 1.1186460256576538,
      "learning_rate": 2.0816730310488186e-05,
      "loss": 1.2532,
      "step": 5565
    },
    {
      "epoch": 0.7973069760779258,
      "grad_norm": 1.0472114086151123,
      "learning_rate": 2.0788401953667668e-05,
      "loss": 1.477,
      "step": 5566
    },
    {
      "epoch": 0.7974502220312276,
      "grad_norm": 1.25272798538208,
      "learning_rate": 2.0760090649144005e-05,
      "loss": 1.1717,
      "step": 5567
    },
    {
      "epoch": 0.7975934679845295,
      "grad_norm": 1.0973775386810303,
      "learning_rate": 2.0731796403011906e-05,
      "loss": 1.5596,
      "step": 5568
    },
    {
      "epoch": 0.7977367139378313,
      "grad_norm": 1.1017022132873535,
      "learning_rate": 2.070351922136251e-05,
      "loss": 1.4256,
      "step": 5569
    },
    {
      "epoch": 0.797879959891133,
      "grad_norm": 1.1429880857467651,
      "learning_rate": 2.0675259110283117e-05,
      "loss": 1.5247,
      "step": 5570
    },
    {
      "epoch": 0.7980232058444349,
      "grad_norm": 1.049401879310608,
      "learning_rate": 2.0647016075857463e-05,
      "loss": 1.5903,
      "step": 5571
    },
    {
      "epoch": 0.7981664517977367,
      "grad_norm": 0.9920111894607544,
      "learning_rate": 2.0618790124165556e-05,
      "loss": 1.4359,
      "step": 5572
    },
    {
      "epoch": 0.7983096977510385,
      "grad_norm": 0.9339282512664795,
      "learning_rate": 2.059058126128376e-05,
      "loss": 1.378,
      "step": 5573
    },
    {
      "epoch": 0.7984529437043404,
      "grad_norm": 0.9228371977806091,
      "learning_rate": 2.0562389493284763e-05,
      "loss": 1.4882,
      "step": 5574
    },
    {
      "epoch": 0.7985961896576421,
      "grad_norm": 1.1618367433547974,
      "learning_rate": 2.0534214826237484e-05,
      "loss": 1.3113,
      "step": 5575
    },
    {
      "epoch": 0.798739435610944,
      "grad_norm": 1.3761992454528809,
      "learning_rate": 2.0506057266207313e-05,
      "loss": 1.4591,
      "step": 5576
    },
    {
      "epoch": 0.7988826815642458,
      "grad_norm": 0.9985799789428711,
      "learning_rate": 2.047791681925586e-05,
      "loss": 1.297,
      "step": 5577
    },
    {
      "epoch": 0.7990259275175476,
      "grad_norm": 1.0619736909866333,
      "learning_rate": 2.0449793491441028e-05,
      "loss": 1.2337,
      "step": 5578
    },
    {
      "epoch": 0.7991691734708495,
      "grad_norm": 1.008886456489563,
      "learning_rate": 2.0421687288817058e-05,
      "loss": 1.3267,
      "step": 5579
    },
    {
      "epoch": 0.7993124194241513,
      "grad_norm": 1.0852863788604736,
      "learning_rate": 2.0393598217434616e-05,
      "loss": 1.4054,
      "step": 5580
    },
    {
      "epoch": 0.799455665377453,
      "grad_norm": 1.1519148349761963,
      "learning_rate": 2.0365526283340508e-05,
      "loss": 1.418,
      "step": 5581
    },
    {
      "epoch": 0.7995989113307549,
      "grad_norm": 1.1324946880340576,
      "learning_rate": 2.033747149257793e-05,
      "loss": 1.3423,
      "step": 5582
    },
    {
      "epoch": 0.7997421572840567,
      "grad_norm": 1.0217424631118774,
      "learning_rate": 2.030943385118641e-05,
      "loss": 1.5086,
      "step": 5583
    },
    {
      "epoch": 0.7998854032373586,
      "grad_norm": 1.222145676612854,
      "learning_rate": 2.028141336520174e-05,
      "loss": 1.3491,
      "step": 5584
    },
    {
      "epoch": 0.8000286491906604,
      "grad_norm": 0.9856448173522949,
      "learning_rate": 2.0253410040656073e-05,
      "loss": 1.3835,
      "step": 5585
    },
    {
      "epoch": 0.8001718951439621,
      "grad_norm": 1.0147873163223267,
      "learning_rate": 2.0225423883577754e-05,
      "loss": 1.3378,
      "step": 5586
    },
    {
      "epoch": 0.800315141097264,
      "grad_norm": 1.2144800424575806,
      "learning_rate": 2.0197454899991573e-05,
      "loss": 1.3332,
      "step": 5587
    },
    {
      "epoch": 0.8004583870505658,
      "grad_norm": 1.1955982446670532,
      "learning_rate": 2.0169503095918586e-05,
      "loss": 1.301,
      "step": 5588
    },
    {
      "epoch": 0.8006016330038677,
      "grad_norm": 1.2471030950546265,
      "learning_rate": 2.014156847737605e-05,
      "loss": 1.1921,
      "step": 5589
    },
    {
      "epoch": 0.8007448789571695,
      "grad_norm": 0.966519832611084,
      "learning_rate": 2.0113651050377623e-05,
      "loss": 1.4768,
      "step": 5590
    },
    {
      "epoch": 0.8008881249104712,
      "grad_norm": 1.0456572771072388,
      "learning_rate": 2.0085750820933257e-05,
      "loss": 1.4942,
      "step": 5591
    },
    {
      "epoch": 0.8010313708637731,
      "grad_norm": 1.0134669542312622,
      "learning_rate": 2.005786779504917e-05,
      "loss": 1.3711,
      "step": 5592
    },
    {
      "epoch": 0.8011746168170749,
      "grad_norm": 1.0319499969482422,
      "learning_rate": 2.0030001978727874e-05,
      "loss": 1.4634,
      "step": 5593
    },
    {
      "epoch": 0.8013178627703768,
      "grad_norm": 1.1356571912765503,
      "learning_rate": 2.0002153377968213e-05,
      "loss": 1.377,
      "step": 5594
    },
    {
      "epoch": 0.8014611087236786,
      "grad_norm": 1.1101000308990479,
      "learning_rate": 1.997432199876531e-05,
      "loss": 1.3712,
      "step": 5595
    },
    {
      "epoch": 0.8016043546769803,
      "grad_norm": 1.2884794473648071,
      "learning_rate": 1.994650784711053e-05,
      "loss": 1.5032,
      "step": 5596
    },
    {
      "epoch": 0.8017476006302822,
      "grad_norm": 1.0528610944747925,
      "learning_rate": 1.9918710928991567e-05,
      "loss": 1.2989,
      "step": 5597
    },
    {
      "epoch": 0.801890846583584,
      "grad_norm": 1.105637788772583,
      "learning_rate": 1.9890931250392498e-05,
      "loss": 1.3532,
      "step": 5598
    },
    {
      "epoch": 0.8020340925368858,
      "grad_norm": 0.9321838021278381,
      "learning_rate": 1.9863168817293497e-05,
      "loss": 1.4691,
      "step": 5599
    },
    {
      "epoch": 0.8021773384901877,
      "grad_norm": 1.2732150554656982,
      "learning_rate": 1.983542363567118e-05,
      "loss": 1.314,
      "step": 5600
    },
    {
      "epoch": 0.8023205844434895,
      "grad_norm": 1.095652461051941,
      "learning_rate": 1.9807695711498385e-05,
      "loss": 1.5321,
      "step": 5601
    },
    {
      "epoch": 0.8024638303967913,
      "grad_norm": 1.01155424118042,
      "learning_rate": 1.9779985050744256e-05,
      "loss": 1.4224,
      "step": 5602
    },
    {
      "epoch": 0.8026070763500931,
      "grad_norm": 1.1057287454605103,
      "learning_rate": 1.9752291659374234e-05,
      "loss": 1.5719,
      "step": 5603
    },
    {
      "epoch": 0.8027503223033949,
      "grad_norm": 1.1701321601867676,
      "learning_rate": 1.9724615543349943e-05,
      "loss": 1.3276,
      "step": 5604
    },
    {
      "epoch": 0.8028935682566968,
      "grad_norm": 1.1319361925125122,
      "learning_rate": 1.9696956708629445e-05,
      "loss": 1.4887,
      "step": 5605
    },
    {
      "epoch": 0.8030368142099986,
      "grad_norm": 1.2079811096191406,
      "learning_rate": 1.9669315161167e-05,
      "loss": 1.2583,
      "step": 5606
    },
    {
      "epoch": 0.8031800601633003,
      "grad_norm": 1.1055455207824707,
      "learning_rate": 1.964169090691309e-05,
      "loss": 1.4886,
      "step": 5607
    },
    {
      "epoch": 0.8033233061166022,
      "grad_norm": 1.126715064048767,
      "learning_rate": 1.9614083951814554e-05,
      "loss": 1.2918,
      "step": 5608
    },
    {
      "epoch": 0.803466552069904,
      "grad_norm": 1.1366931200027466,
      "learning_rate": 1.958649430181455e-05,
      "loss": 1.4162,
      "step": 5609
    },
    {
      "epoch": 0.8036097980232059,
      "grad_norm": 1.0904452800750732,
      "learning_rate": 1.955892196285237e-05,
      "loss": 1.4557,
      "step": 5610
    },
    {
      "epoch": 0.8037530439765077,
      "grad_norm": 1.1547094583511353,
      "learning_rate": 1.9531366940863694e-05,
      "loss": 1.4017,
      "step": 5611
    },
    {
      "epoch": 0.8038962899298094,
      "grad_norm": 1.2169826030731201,
      "learning_rate": 1.9503829241780412e-05,
      "loss": 1.3618,
      "step": 5612
    },
    {
      "epoch": 0.8040395358831113,
      "grad_norm": 1.0908526182174683,
      "learning_rate": 1.9476308871530723e-05,
      "loss": 1.3414,
      "step": 5613
    },
    {
      "epoch": 0.8041827818364131,
      "grad_norm": 0.9590395092964172,
      "learning_rate": 1.944880583603912e-05,
      "loss": 1.6114,
      "step": 5614
    },
    {
      "epoch": 0.804326027789715,
      "grad_norm": 1.3139491081237793,
      "learning_rate": 1.9421320141226228e-05,
      "loss": 1.32,
      "step": 5615
    },
    {
      "epoch": 0.8044692737430168,
      "grad_norm": 1.2910867929458618,
      "learning_rate": 1.939385179300912e-05,
      "loss": 1.5992,
      "step": 5616
    },
    {
      "epoch": 0.8046125196963185,
      "grad_norm": 1.0733730792999268,
      "learning_rate": 1.9366400797301066e-05,
      "loss": 1.3882,
      "step": 5617
    },
    {
      "epoch": 0.8047557656496204,
      "grad_norm": 1.1911749839782715,
      "learning_rate": 1.9338967160011512e-05,
      "loss": 1.3702,
      "step": 5618
    },
    {
      "epoch": 0.8048990116029222,
      "grad_norm": 1.1243938207626343,
      "learning_rate": 1.9311550887046282e-05,
      "loss": 1.3426,
      "step": 5619
    },
    {
      "epoch": 0.805042257556224,
      "grad_norm": 1.0868338346481323,
      "learning_rate": 1.928415198430742e-05,
      "loss": 1.4727,
      "step": 5620
    },
    {
      "epoch": 0.8051855035095259,
      "grad_norm": 0.9629149436950684,
      "learning_rate": 1.925677045769322e-05,
      "loss": 1.574,
      "step": 5621
    },
    {
      "epoch": 0.8053287494628276,
      "grad_norm": 0.9636642336845398,
      "learning_rate": 1.9229406313098264e-05,
      "loss": 1.4913,
      "step": 5622
    },
    {
      "epoch": 0.8054719954161295,
      "grad_norm": 1.1252906322479248,
      "learning_rate": 1.9202059556413366e-05,
      "loss": 1.4437,
      "step": 5623
    },
    {
      "epoch": 0.8056152413694313,
      "grad_norm": 1.0094927549362183,
      "learning_rate": 1.9174730193525626e-05,
      "loss": 1.3751,
      "step": 5624
    },
    {
      "epoch": 0.8057584873227331,
      "grad_norm": 1.037348985671997,
      "learning_rate": 1.9147418230318316e-05,
      "loss": 1.3615,
      "step": 5625
    },
    {
      "epoch": 0.805901733276035,
      "grad_norm": 1.2424753904342651,
      "learning_rate": 1.9120123672671086e-05,
      "loss": 1.4923,
      "step": 5626
    },
    {
      "epoch": 0.8060449792293368,
      "grad_norm": 1.0879966020584106,
      "learning_rate": 1.9092846526459797e-05,
      "loss": 1.4476,
      "step": 5627
    },
    {
      "epoch": 0.8061882251826386,
      "grad_norm": 0.9539452195167542,
      "learning_rate": 1.906558679755649e-05,
      "loss": 1.4783,
      "step": 5628
    },
    {
      "epoch": 0.8063314711359404,
      "grad_norm": 1.1715091466903687,
      "learning_rate": 1.9038344491829495e-05,
      "loss": 1.5123,
      "step": 5629
    },
    {
      "epoch": 0.8064747170892422,
      "grad_norm": 0.9144424796104431,
      "learning_rate": 1.9011119615143492e-05,
      "loss": 1.597,
      "step": 5630
    },
    {
      "epoch": 0.8066179630425441,
      "grad_norm": 1.0063729286193848,
      "learning_rate": 1.898391217335924e-05,
      "loss": 1.5628,
      "step": 5631
    },
    {
      "epoch": 0.8067612089958459,
      "grad_norm": 1.1106654405593872,
      "learning_rate": 1.8956722172333875e-05,
      "loss": 1.3737,
      "step": 5632
    },
    {
      "epoch": 0.8069044549491476,
      "grad_norm": 1.0304570198059082,
      "learning_rate": 1.8929549617920716e-05,
      "loss": 1.4638,
      "step": 5633
    },
    {
      "epoch": 0.8070477009024495,
      "grad_norm": 1.1444005966186523,
      "learning_rate": 1.8902394515969335e-05,
      "loss": 1.3828,
      "step": 5634
    },
    {
      "epoch": 0.8071909468557513,
      "grad_norm": 1.0587037801742554,
      "learning_rate": 1.8875256872325587e-05,
      "loss": 1.3434,
      "step": 5635
    },
    {
      "epoch": 0.8073341928090532,
      "grad_norm": 1.166538119316101,
      "learning_rate": 1.884813669283145e-05,
      "loss": 1.4208,
      "step": 5636
    },
    {
      "epoch": 0.807477438762355,
      "grad_norm": 0.9851470589637756,
      "learning_rate": 1.882103398332533e-05,
      "loss": 1.4393,
      "step": 5637
    },
    {
      "epoch": 0.8076206847156567,
      "grad_norm": 1.0796467065811157,
      "learning_rate": 1.8793948749641744e-05,
      "loss": 1.5337,
      "step": 5638
    },
    {
      "epoch": 0.8077639306689586,
      "grad_norm": 1.5454179048538208,
      "learning_rate": 1.8766880997611424e-05,
      "loss": 1.3759,
      "step": 5639
    },
    {
      "epoch": 0.8079071766222604,
      "grad_norm": 1.0702461004257202,
      "learning_rate": 1.8739830733061413e-05,
      "loss": 1.4806,
      "step": 5640
    },
    {
      "epoch": 0.8080504225755623,
      "grad_norm": 1.007646918296814,
      "learning_rate": 1.8712797961814975e-05,
      "loss": 1.3136,
      "step": 5641
    },
    {
      "epoch": 0.8081936685288641,
      "grad_norm": 1.0816047191619873,
      "learning_rate": 1.8685782689691587e-05,
      "loss": 1.4315,
      "step": 5642
    },
    {
      "epoch": 0.8083369144821658,
      "grad_norm": 0.9248777031898499,
      "learning_rate": 1.865878492250698e-05,
      "loss": 1.4139,
      "step": 5643
    },
    {
      "epoch": 0.8084801604354677,
      "grad_norm": 1.1594047546386719,
      "learning_rate": 1.8631804666073094e-05,
      "loss": 1.3085,
      "step": 5644
    },
    {
      "epoch": 0.8086234063887695,
      "grad_norm": 1.2599555253982544,
      "learning_rate": 1.8604841926198135e-05,
      "loss": 1.2833,
      "step": 5645
    },
    {
      "epoch": 0.8087666523420713,
      "grad_norm": 1.0698325634002686,
      "learning_rate": 1.857789670868647e-05,
      "loss": 1.5376,
      "step": 5646
    },
    {
      "epoch": 0.8089098982953732,
      "grad_norm": 1.0841995477676392,
      "learning_rate": 1.8550969019338725e-05,
      "loss": 1.3287,
      "step": 5647
    },
    {
      "epoch": 0.809053144248675,
      "grad_norm": 1.1984270811080933,
      "learning_rate": 1.8524058863951854e-05,
      "loss": 1.2035,
      "step": 5648
    },
    {
      "epoch": 0.8091963902019768,
      "grad_norm": 1.0743377208709717,
      "learning_rate": 1.8497166248318876e-05,
      "loss": 1.4306,
      "step": 5649
    },
    {
      "epoch": 0.8093396361552786,
      "grad_norm": 1.3287091255187988,
      "learning_rate": 1.8470291178229116e-05,
      "loss": 1.4562,
      "step": 5650
    },
    {
      "epoch": 0.8094828821085804,
      "grad_norm": 1.0762503147125244,
      "learning_rate": 1.8443433659468123e-05,
      "loss": 1.4059,
      "step": 5651
    },
    {
      "epoch": 0.8096261280618823,
      "grad_norm": 1.1955372095108032,
      "learning_rate": 1.841659369781764e-05,
      "loss": 1.2773,
      "step": 5652
    },
    {
      "epoch": 0.8097693740151841,
      "grad_norm": 1.196042537689209,
      "learning_rate": 1.838977129905569e-05,
      "loss": 1.3151,
      "step": 5653
    },
    {
      "epoch": 0.8099126199684858,
      "grad_norm": 1.1636439561843872,
      "learning_rate": 1.83629664689564e-05,
      "loss": 1.3429,
      "step": 5654
    },
    {
      "epoch": 0.8100558659217877,
      "grad_norm": 1.153066873550415,
      "learning_rate": 1.833617921329024e-05,
      "loss": 1.4177,
      "step": 5655
    },
    {
      "epoch": 0.8101991118750895,
      "grad_norm": 1.0415980815887451,
      "learning_rate": 1.830940953782385e-05,
      "loss": 1.3,
      "step": 5656
    },
    {
      "epoch": 0.8103423578283914,
      "grad_norm": 1.0865952968597412,
      "learning_rate": 1.828265744832004e-05,
      "loss": 1.4011,
      "step": 5657
    },
    {
      "epoch": 0.8104856037816932,
      "grad_norm": 1.185973048210144,
      "learning_rate": 1.8255922950537872e-05,
      "loss": 1.4024,
      "step": 5658
    },
    {
      "epoch": 0.8106288497349949,
      "grad_norm": 1.2533318996429443,
      "learning_rate": 1.8229206050232684e-05,
      "loss": 1.2632,
      "step": 5659
    },
    {
      "epoch": 0.8107720956882968,
      "grad_norm": 1.1986994743347168,
      "learning_rate": 1.820250675315589e-05,
      "loss": 1.3096,
      "step": 5660
    },
    {
      "epoch": 0.8109153416415986,
      "grad_norm": 1.0425302982330322,
      "learning_rate": 1.817582506505523e-05,
      "loss": 1.2316,
      "step": 5661
    },
    {
      "epoch": 0.8110585875949005,
      "grad_norm": 1.0855088233947754,
      "learning_rate": 1.8149160991674597e-05,
      "loss": 1.4734,
      "step": 5662
    },
    {
      "epoch": 0.8112018335482023,
      "grad_norm": 1.2144453525543213,
      "learning_rate": 1.81225145387541e-05,
      "loss": 1.471,
      "step": 5663
    },
    {
      "epoch": 0.811345079501504,
      "grad_norm": 1.061293363571167,
      "learning_rate": 1.80958857120301e-05,
      "loss": 1.2886,
      "step": 5664
    },
    {
      "epoch": 0.8114883254548059,
      "grad_norm": 0.9511222243309021,
      "learning_rate": 1.8069274517235047e-05,
      "loss": 1.4152,
      "step": 5665
    },
    {
      "epoch": 0.8116315714081077,
      "grad_norm": 1.3765077590942383,
      "learning_rate": 1.8042680960097735e-05,
      "loss": 1.4206,
      "step": 5666
    },
    {
      "epoch": 0.8117748173614096,
      "grad_norm": 1.3805012702941895,
      "learning_rate": 1.8016105046343123e-05,
      "loss": 1.4367,
      "step": 5667
    },
    {
      "epoch": 0.8119180633147114,
      "grad_norm": 0.9719917178153992,
      "learning_rate": 1.798954678169228e-05,
      "loss": 1.223,
      "step": 5668
    },
    {
      "epoch": 0.8120613092680132,
      "grad_norm": 1.0138293504714966,
      "learning_rate": 1.7963006171862562e-05,
      "loss": 1.3879,
      "step": 5669
    },
    {
      "epoch": 0.812204555221315,
      "grad_norm": 1.1095738410949707,
      "learning_rate": 1.7936483222567523e-05,
      "loss": 1.4462,
      "step": 5670
    },
    {
      "epoch": 0.8123478011746168,
      "grad_norm": 0.9974596500396729,
      "learning_rate": 1.7909977939516887e-05,
      "loss": 1.3172,
      "step": 5671
    },
    {
      "epoch": 0.8124910471279186,
      "grad_norm": 1.0114816427230835,
      "learning_rate": 1.7883490328416587e-05,
      "loss": 1.368,
      "step": 5672
    },
    {
      "epoch": 0.8126342930812205,
      "grad_norm": 1.3137753009796143,
      "learning_rate": 1.785702039496875e-05,
      "loss": 1.4381,
      "step": 5673
    },
    {
      "epoch": 0.8127775390345223,
      "grad_norm": 1.0969570875167847,
      "learning_rate": 1.783056814487172e-05,
      "loss": 1.6914,
      "step": 5674
    },
    {
      "epoch": 0.812920784987824,
      "grad_norm": 1.1589391231536865,
      "learning_rate": 1.780413358381997e-05,
      "loss": 1.2729,
      "step": 5675
    },
    {
      "epoch": 0.8130640309411259,
      "grad_norm": 0.9956794381141663,
      "learning_rate": 1.7777716717504213e-05,
      "loss": 1.3966,
      "step": 5676
    },
    {
      "epoch": 0.8132072768944277,
      "grad_norm": 1.1816606521606445,
      "learning_rate": 1.775131755161139e-05,
      "loss": 1.2947,
      "step": 5677
    },
    {
      "epoch": 0.8133505228477296,
      "grad_norm": 0.9762355089187622,
      "learning_rate": 1.772493609182455e-05,
      "loss": 1.5738,
      "step": 5678
    },
    {
      "epoch": 0.8134937688010314,
      "grad_norm": 0.9318240284919739,
      "learning_rate": 1.7698572343822973e-05,
      "loss": 1.5495,
      "step": 5679
    },
    {
      "epoch": 0.8136370147543331,
      "grad_norm": 1.272712230682373,
      "learning_rate": 1.7672226313282126e-05,
      "loss": 1.3791,
      "step": 5680
    },
    {
      "epoch": 0.813780260707635,
      "grad_norm": 0.9423567652702332,
      "learning_rate": 1.7645898005873663e-05,
      "loss": 1.389,
      "step": 5681
    },
    {
      "epoch": 0.8139235066609368,
      "grad_norm": 1.0018702745437622,
      "learning_rate": 1.7619587427265405e-05,
      "loss": 1.327,
      "step": 5682
    },
    {
      "epoch": 0.8140667526142387,
      "grad_norm": 0.9877719283103943,
      "learning_rate": 1.7593294583121377e-05,
      "loss": 1.4499,
      "step": 5683
    },
    {
      "epoch": 0.8142099985675405,
      "grad_norm": 0.9803469181060791,
      "learning_rate": 1.7567019479101776e-05,
      "loss": 1.4247,
      "step": 5684
    },
    {
      "epoch": 0.8143532445208422,
      "grad_norm": 1.0692263841629028,
      "learning_rate": 1.754076212086301e-05,
      "loss": 1.2784,
      "step": 5685
    },
    {
      "epoch": 0.8144964904741441,
      "grad_norm": 1.0264326333999634,
      "learning_rate": 1.7514522514057553e-05,
      "loss": 1.3225,
      "step": 5686
    },
    {
      "epoch": 0.8146397364274459,
      "grad_norm": 1.051205039024353,
      "learning_rate": 1.7488300664334236e-05,
      "loss": 1.4064,
      "step": 5687
    },
    {
      "epoch": 0.8147829823807478,
      "grad_norm": 0.9736294746398926,
      "learning_rate": 1.746209657733795e-05,
      "loss": 1.2956,
      "step": 5688
    },
    {
      "epoch": 0.8149262283340496,
      "grad_norm": 1.0348933935165405,
      "learning_rate": 1.7435910258709752e-05,
      "loss": 1.3572,
      "step": 5689
    },
    {
      "epoch": 0.8150694742873514,
      "grad_norm": 0.9968773722648621,
      "learning_rate": 1.7409741714086948e-05,
      "loss": 1.4103,
      "step": 5690
    },
    {
      "epoch": 0.8152127202406532,
      "grad_norm": 1.0300939083099365,
      "learning_rate": 1.7383590949102945e-05,
      "loss": 1.3176,
      "step": 5691
    },
    {
      "epoch": 0.815355966193955,
      "grad_norm": 1.0646156072616577,
      "learning_rate": 1.7357457969387368e-05,
      "loss": 1.2789,
      "step": 5692
    },
    {
      "epoch": 0.8154992121472568,
      "grad_norm": 1.1840293407440186,
      "learning_rate": 1.7331342780566017e-05,
      "loss": 1.2092,
      "step": 5693
    },
    {
      "epoch": 0.8156424581005587,
      "grad_norm": 0.9098217487335205,
      "learning_rate": 1.7305245388260826e-05,
      "loss": 1.4562,
      "step": 5694
    },
    {
      "epoch": 0.8157857040538605,
      "grad_norm": 1.110976219177246,
      "learning_rate": 1.7279165798089957e-05,
      "loss": 1.5646,
      "step": 5695
    },
    {
      "epoch": 0.8159289500071623,
      "grad_norm": 1.040792465209961,
      "learning_rate": 1.7253104015667643e-05,
      "loss": 1.5351,
      "step": 5696
    },
    {
      "epoch": 0.8160721959604641,
      "grad_norm": 1.0433226823806763,
      "learning_rate": 1.7227060046604336e-05,
      "loss": 1.2707,
      "step": 5697
    },
    {
      "epoch": 0.8162154419137659,
      "grad_norm": 0.9411223530769348,
      "learning_rate": 1.7201033896506746e-05,
      "loss": 1.3303,
      "step": 5698
    },
    {
      "epoch": 0.8163586878670678,
      "grad_norm": 1.2658718824386597,
      "learning_rate": 1.7175025570977577e-05,
      "loss": 1.417,
      "step": 5699
    },
    {
      "epoch": 0.8165019338203696,
      "grad_norm": 1.0115282535552979,
      "learning_rate": 1.7149035075615794e-05,
      "loss": 1.4125,
      "step": 5700
    },
    {
      "epoch": 0.8166451797736713,
      "grad_norm": 1.124245524406433,
      "learning_rate": 1.7123062416016524e-05,
      "loss": 1.3277,
      "step": 5701
    },
    {
      "epoch": 0.8167884257269732,
      "grad_norm": 0.9520460367202759,
      "learning_rate": 1.7097107597771024e-05,
      "loss": 1.3215,
      "step": 5702
    },
    {
      "epoch": 0.816931671680275,
      "grad_norm": 1.044511079788208,
      "learning_rate": 1.707117062646676e-05,
      "loss": 1.291,
      "step": 5703
    },
    {
      "epoch": 0.8170749176335769,
      "grad_norm": 1.072210431098938,
      "learning_rate": 1.7045251507687232e-05,
      "loss": 1.2702,
      "step": 5704
    },
    {
      "epoch": 0.8172181635868787,
      "grad_norm": 1.1078007221221924,
      "learning_rate": 1.7019350247012278e-05,
      "loss": 1.2887,
      "step": 5705
    },
    {
      "epoch": 0.8173614095401804,
      "grad_norm": 1.2150322198867798,
      "learning_rate": 1.699346685001778e-05,
      "loss": 1.4202,
      "step": 5706
    },
    {
      "epoch": 0.8175046554934823,
      "grad_norm": 1.187727451324463,
      "learning_rate": 1.696760132227576e-05,
      "loss": 1.4804,
      "step": 5707
    },
    {
      "epoch": 0.8176479014467841,
      "grad_norm": 1.188535451889038,
      "learning_rate": 1.694175366935442e-05,
      "loss": 1.2612,
      "step": 5708
    },
    {
      "epoch": 0.817791147400086,
      "grad_norm": 1.278638243675232,
      "learning_rate": 1.6915923896818188e-05,
      "loss": 1.5195,
      "step": 5709
    },
    {
      "epoch": 0.8179343933533878,
      "grad_norm": 1.0303146839141846,
      "learning_rate": 1.6890112010227498e-05,
      "loss": 1.4401,
      "step": 5710
    },
    {
      "epoch": 0.8180776393066896,
      "grad_norm": 1.033528447151184,
      "learning_rate": 1.6864318015139047e-05,
      "loss": 1.2995,
      "step": 5711
    },
    {
      "epoch": 0.8182208852599914,
      "grad_norm": 1.2521709203720093,
      "learning_rate": 1.6838541917105632e-05,
      "loss": 1.3439,
      "step": 5712
    },
    {
      "epoch": 0.8183641312132932,
      "grad_norm": 1.2076901197433472,
      "learning_rate": 1.6812783721676228e-05,
      "loss": 1.3262,
      "step": 5713
    },
    {
      "epoch": 0.818507377166595,
      "grad_norm": 1.058674693107605,
      "learning_rate": 1.6787043434395942e-05,
      "loss": 1.4615,
      "step": 5714
    },
    {
      "epoch": 0.8186506231198969,
      "grad_norm": 1.0597326755523682,
      "learning_rate": 1.6761321060805957e-05,
      "loss": 1.3264,
      "step": 5715
    },
    {
      "epoch": 0.8187938690731987,
      "grad_norm": 1.0054484605789185,
      "learning_rate": 1.6735616606443728e-05,
      "loss": 1.1676,
      "step": 5716
    },
    {
      "epoch": 0.8189371150265005,
      "grad_norm": 1.0756042003631592,
      "learning_rate": 1.6709930076842805e-05,
      "loss": 1.4556,
      "step": 5717
    },
    {
      "epoch": 0.8190803609798023,
      "grad_norm": 1.020453929901123,
      "learning_rate": 1.6684261477532793e-05,
      "loss": 1.3462,
      "step": 5718
    },
    {
      "epoch": 0.8192236069331041,
      "grad_norm": 1.1664289236068726,
      "learning_rate": 1.6658610814039544e-05,
      "loss": 1.4291,
      "step": 5719
    },
    {
      "epoch": 0.819366852886406,
      "grad_norm": 1.3196290731430054,
      "learning_rate": 1.663297809188501e-05,
      "loss": 1.5667,
      "step": 5720
    },
    {
      "epoch": 0.8195100988397078,
      "grad_norm": 1.2542836666107178,
      "learning_rate": 1.6607363316587277e-05,
      "loss": 1.3736,
      "step": 5721
    },
    {
      "epoch": 0.8196533447930096,
      "grad_norm": 1.0937902927398682,
      "learning_rate": 1.6581766493660578e-05,
      "loss": 1.2699,
      "step": 5722
    },
    {
      "epoch": 0.8197965907463114,
      "grad_norm": 0.9938157200813293,
      "learning_rate": 1.6556187628615273e-05,
      "loss": 1.3481,
      "step": 5723
    },
    {
      "epoch": 0.8199398366996132,
      "grad_norm": 1.0735880136489868,
      "learning_rate": 1.6530626726957877e-05,
      "loss": 1.471,
      "step": 5724
    },
    {
      "epoch": 0.8200830826529151,
      "grad_norm": 1.0578898191452026,
      "learning_rate": 1.650508379419098e-05,
      "loss": 1.5773,
      "step": 5725
    },
    {
      "epoch": 0.8202263286062169,
      "grad_norm": 1.1096580028533936,
      "learning_rate": 1.6479558835813334e-05,
      "loss": 1.4956,
      "step": 5726
    },
    {
      "epoch": 0.8203695745595186,
      "grad_norm": 1.0737659931182861,
      "learning_rate": 1.6454051857319906e-05,
      "loss": 1.58,
      "step": 5727
    },
    {
      "epoch": 0.8205128205128205,
      "grad_norm": 1.0251860618591309,
      "learning_rate": 1.6428562864201658e-05,
      "loss": 1.4514,
      "step": 5728
    },
    {
      "epoch": 0.8206560664661223,
      "grad_norm": 0.9603286981582642,
      "learning_rate": 1.6403091861945753e-05,
      "loss": 1.4944,
      "step": 5729
    },
    {
      "epoch": 0.8207993124194242,
      "grad_norm": 1.2171103954315186,
      "learning_rate": 1.637763885603546e-05,
      "loss": 1.4356,
      "step": 5730
    },
    {
      "epoch": 0.820942558372726,
      "grad_norm": 0.9659716486930847,
      "learning_rate": 1.6352203851950198e-05,
      "loss": 1.4968,
      "step": 5731
    },
    {
      "epoch": 0.8210858043260277,
      "grad_norm": 1.1402076482772827,
      "learning_rate": 1.63267868551655e-05,
      "loss": 1.4284,
      "step": 5732
    },
    {
      "epoch": 0.8212290502793296,
      "grad_norm": 1.033719539642334,
      "learning_rate": 1.6301387871152963e-05,
      "loss": 1.2219,
      "step": 5733
    },
    {
      "epoch": 0.8213722962326314,
      "grad_norm": 1.0863749980926514,
      "learning_rate": 1.6276006905380413e-05,
      "loss": 1.5239,
      "step": 5734
    },
    {
      "epoch": 0.8215155421859333,
      "grad_norm": 0.996165931224823,
      "learning_rate": 1.625064396331176e-05,
      "loss": 1.3887,
      "step": 5735
    },
    {
      "epoch": 0.8216587881392351,
      "grad_norm": 1.191558599472046,
      "learning_rate": 1.622529905040696e-05,
      "loss": 1.5328,
      "step": 5736
    },
    {
      "epoch": 0.8218020340925369,
      "grad_norm": 1.1870588064193726,
      "learning_rate": 1.6199972172122147e-05,
      "loss": 1.4848,
      "step": 5737
    },
    {
      "epoch": 0.8219452800458387,
      "grad_norm": 0.956283450126648,
      "learning_rate": 1.6174663333909646e-05,
      "loss": 1.4075,
      "step": 5738
    },
    {
      "epoch": 0.8220885259991405,
      "grad_norm": 1.1312839984893799,
      "learning_rate": 1.6149372541217755e-05,
      "loss": 1.458,
      "step": 5739
    },
    {
      "epoch": 0.8222317719524423,
      "grad_norm": 1.3449602127075195,
      "learning_rate": 1.6124099799490968e-05,
      "loss": 1.4481,
      "step": 5740
    },
    {
      "epoch": 0.8223750179057442,
      "grad_norm": 1.3037852048873901,
      "learning_rate": 1.6098845114169893e-05,
      "loss": 1.4553,
      "step": 5741
    },
    {
      "epoch": 0.822518263859046,
      "grad_norm": 1.0383838415145874,
      "learning_rate": 1.6073608490691228e-05,
      "loss": 1.538,
      "step": 5742
    },
    {
      "epoch": 0.8226615098123478,
      "grad_norm": 1.051731824874878,
      "learning_rate": 1.604838993448783e-05,
      "loss": 1.328,
      "step": 5743
    },
    {
      "epoch": 0.8228047557656496,
      "grad_norm": 1.1407042741775513,
      "learning_rate": 1.602318945098855e-05,
      "loss": 1.285,
      "step": 5744
    },
    {
      "epoch": 0.8229480017189514,
      "grad_norm": 1.0576179027557373,
      "learning_rate": 1.5998007045618502e-05,
      "loss": 1.2916,
      "step": 5745
    },
    {
      "epoch": 0.8230912476722533,
      "grad_norm": 1.2697982788085938,
      "learning_rate": 1.597284272379883e-05,
      "loss": 1.3054,
      "step": 5746
    },
    {
      "epoch": 0.8232344936255551,
      "grad_norm": 1.0967398881912231,
      "learning_rate": 1.594769649094675e-05,
      "loss": 1.3967,
      "step": 5747
    },
    {
      "epoch": 0.8233777395788568,
      "grad_norm": 1.1864540576934814,
      "learning_rate": 1.5922568352475642e-05,
      "loss": 1.5568,
      "step": 5748
    },
    {
      "epoch": 0.8235209855321587,
      "grad_norm": 1.9493398666381836,
      "learning_rate": 1.5897458313794966e-05,
      "loss": 1.3124,
      "step": 5749
    },
    {
      "epoch": 0.8236642314854605,
      "grad_norm": 1.1356884241104126,
      "learning_rate": 1.587236638031031e-05,
      "loss": 1.3017,
      "step": 5750
    },
    {
      "epoch": 0.8238074774387624,
      "grad_norm": 1.2949925661087036,
      "learning_rate": 1.5847292557423344e-05,
      "loss": 1.289,
      "step": 5751
    },
    {
      "epoch": 0.8239507233920642,
      "grad_norm": 0.91130131483078,
      "learning_rate": 1.5822236850531824e-05,
      "loss": 1.2107,
      "step": 5752
    },
    {
      "epoch": 0.8240939693453659,
      "grad_norm": 0.8751494288444519,
      "learning_rate": 1.579719926502966e-05,
      "loss": 1.4364,
      "step": 5753
    },
    {
      "epoch": 0.8242372152986678,
      "grad_norm": 1.0112683773040771,
      "learning_rate": 1.5772179806306743e-05,
      "loss": 1.333,
      "step": 5754
    },
    {
      "epoch": 0.8243804612519696,
      "grad_norm": 1.006009817123413,
      "learning_rate": 1.5747178479749236e-05,
      "loss": 1.2797,
      "step": 5755
    },
    {
      "epoch": 0.8245237072052715,
      "grad_norm": 0.9960899949073792,
      "learning_rate": 1.5722195290739285e-05,
      "loss": 1.3951,
      "step": 5756
    },
    {
      "epoch": 0.8246669531585733,
      "grad_norm": 1.0580023527145386,
      "learning_rate": 1.5697230244655114e-05,
      "loss": 1.3871,
      "step": 5757
    },
    {
      "epoch": 0.8248101991118751,
      "grad_norm": 1.1296244859695435,
      "learning_rate": 1.5672283346871074e-05,
      "loss": 1.406,
      "step": 5758
    },
    {
      "epoch": 0.8249534450651769,
      "grad_norm": 1.1733819246292114,
      "learning_rate": 1.564735460275769e-05,
      "loss": 1.4868,
      "step": 5759
    },
    {
      "epoch": 0.8250966910184787,
      "grad_norm": 1.0858979225158691,
      "learning_rate": 1.562244401768144e-05,
      "loss": 1.2394,
      "step": 5760
    },
    {
      "epoch": 0.8252399369717806,
      "grad_norm": 1.012631893157959,
      "learning_rate": 1.5597551597004966e-05,
      "loss": 1.3404,
      "step": 5761
    },
    {
      "epoch": 0.8253831829250824,
      "grad_norm": 1.171512484550476,
      "learning_rate": 1.5572677346087004e-05,
      "loss": 1.2798,
      "step": 5762
    },
    {
      "epoch": 0.8255264288783842,
      "grad_norm": 1.2161369323730469,
      "learning_rate": 1.5547821270282346e-05,
      "loss": 1.5199,
      "step": 5763
    },
    {
      "epoch": 0.825669674831686,
      "grad_norm": 1.1851780414581299,
      "learning_rate": 1.5522983374941937e-05,
      "loss": 1.2706,
      "step": 5764
    },
    {
      "epoch": 0.8258129207849878,
      "grad_norm": 1.0883551836013794,
      "learning_rate": 1.549816366541268e-05,
      "loss": 1.1912,
      "step": 5765
    },
    {
      "epoch": 0.8259561667382896,
      "grad_norm": 1.0480178594589233,
      "learning_rate": 1.5473362147037706e-05,
      "loss": 1.4259,
      "step": 5766
    },
    {
      "epoch": 0.8260994126915915,
      "grad_norm": 1.109197735786438,
      "learning_rate": 1.544857882515617e-05,
      "loss": 1.3174,
      "step": 5767
    },
    {
      "epoch": 0.8262426586448933,
      "grad_norm": 1.0364460945129395,
      "learning_rate": 1.5423813705103275e-05,
      "loss": 1.4802,
      "step": 5768
    },
    {
      "epoch": 0.826385904598195,
      "grad_norm": 1.0377068519592285,
      "learning_rate": 1.539906679221035e-05,
      "loss": 1.5347,
      "step": 5769
    },
    {
      "epoch": 0.8265291505514969,
      "grad_norm": 1.0012344121932983,
      "learning_rate": 1.537433809180481e-05,
      "loss": 1.4465,
      "step": 5770
    },
    {
      "epoch": 0.8266723965047987,
      "grad_norm": 1.3207192420959473,
      "learning_rate": 1.5349627609210104e-05,
      "loss": 1.3257,
      "step": 5771
    },
    {
      "epoch": 0.8268156424581006,
      "grad_norm": 1.2061342000961304,
      "learning_rate": 1.5324935349745805e-05,
      "loss": 1.4627,
      "step": 5772
    },
    {
      "epoch": 0.8269588884114024,
      "grad_norm": 1.111749291419983,
      "learning_rate": 1.5300261318727537e-05,
      "loss": 1.3454,
      "step": 5773
    },
    {
      "epoch": 0.8271021343647041,
      "grad_norm": 0.9952139258384705,
      "learning_rate": 1.5275605521467052e-05,
      "loss": 1.3543,
      "step": 5774
    },
    {
      "epoch": 0.827245380318006,
      "grad_norm": 0.9418818950653076,
      "learning_rate": 1.5250967963272056e-05,
      "loss": 1.4071,
      "step": 5775
    },
    {
      "epoch": 0.8273886262713078,
      "grad_norm": 1.0336658954620361,
      "learning_rate": 1.5226348649446432e-05,
      "loss": 1.5182,
      "step": 5776
    },
    {
      "epoch": 0.8275318722246097,
      "grad_norm": 1.0942680835723877,
      "learning_rate": 1.520174758529016e-05,
      "loss": 1.531,
      "step": 5777
    },
    {
      "epoch": 0.8276751181779115,
      "grad_norm": 1.027611494064331,
      "learning_rate": 1.5177164776099184e-05,
      "loss": 1.4072,
      "step": 5778
    },
    {
      "epoch": 0.8278183641312133,
      "grad_norm": 1.1368470191955566,
      "learning_rate": 1.5152600227165591e-05,
      "loss": 1.3798,
      "step": 5779
    },
    {
      "epoch": 0.8279616100845151,
      "grad_norm": 1.0480561256408691,
      "learning_rate": 1.5128053943777532e-05,
      "loss": 1.4848,
      "step": 5780
    },
    {
      "epoch": 0.8281048560378169,
      "grad_norm": 0.920052170753479,
      "learning_rate": 1.5103525931219186e-05,
      "loss": 1.3972,
      "step": 5781
    },
    {
      "epoch": 0.8282481019911188,
      "grad_norm": 1.0311025381088257,
      "learning_rate": 1.5079016194770889e-05,
      "loss": 1.3737,
      "step": 5782
    },
    {
      "epoch": 0.8283913479444206,
      "grad_norm": 1.2496471405029297,
      "learning_rate": 1.5054524739708876e-05,
      "loss": 1.4451,
      "step": 5783
    },
    {
      "epoch": 0.8285345938977224,
      "grad_norm": 1.0876752138137817,
      "learning_rate": 1.5030051571305637e-05,
      "loss": 1.4355,
      "step": 5784
    },
    {
      "epoch": 0.8286778398510242,
      "grad_norm": 1.1764819622039795,
      "learning_rate": 1.5005596694829637e-05,
      "loss": 1.6363,
      "step": 5785
    },
    {
      "epoch": 0.828821085804326,
      "grad_norm": 1.175207495689392,
      "learning_rate": 1.4981160115545367e-05,
      "loss": 1.1861,
      "step": 5786
    },
    {
      "epoch": 0.8289643317576278,
      "grad_norm": 1.0424830913543701,
      "learning_rate": 1.4956741838713406e-05,
      "loss": 1.4001,
      "step": 5787
    },
    {
      "epoch": 0.8291075777109297,
      "grad_norm": 1.267825961112976,
      "learning_rate": 1.4932341869590483e-05,
      "loss": 1.3516,
      "step": 5788
    },
    {
      "epoch": 0.8292508236642315,
      "grad_norm": 0.9715421199798584,
      "learning_rate": 1.4907960213429239e-05,
      "loss": 1.7048,
      "step": 5789
    },
    {
      "epoch": 0.8293940696175333,
      "grad_norm": 1.1823830604553223,
      "learning_rate": 1.4883596875478457e-05,
      "loss": 1.4406,
      "step": 5790
    },
    {
      "epoch": 0.8295373155708351,
      "grad_norm": 1.1496598720550537,
      "learning_rate": 1.485925186098296e-05,
      "loss": 1.2746,
      "step": 5791
    },
    {
      "epoch": 0.8296805615241369,
      "grad_norm": 0.9640195369720459,
      "learning_rate": 1.4834925175183635e-05,
      "loss": 1.3106,
      "step": 5792
    },
    {
      "epoch": 0.8298238074774388,
      "grad_norm": 0.9685528874397278,
      "learning_rate": 1.4810616823317425e-05,
      "loss": 1.4178,
      "step": 5793
    },
    {
      "epoch": 0.8299670534307406,
      "grad_norm": 1.0400193929672241,
      "learning_rate": 1.4786326810617268e-05,
      "loss": 1.2435,
      "step": 5794
    },
    {
      "epoch": 0.8301102993840423,
      "grad_norm": 1.1492332220077515,
      "learning_rate": 1.476205514231226e-05,
      "loss": 1.462,
      "step": 5795
    },
    {
      "epoch": 0.8302535453373442,
      "grad_norm": 1.2634183168411255,
      "learning_rate": 1.4737801823627485e-05,
      "loss": 1.3799,
      "step": 5796
    },
    {
      "epoch": 0.830396791290646,
      "grad_norm": 1.0642378330230713,
      "learning_rate": 1.4713566859784045e-05,
      "loss": 1.5513,
      "step": 5797
    },
    {
      "epoch": 0.8305400372439479,
      "grad_norm": 1.0573594570159912,
      "learning_rate": 1.4689350255999146e-05,
      "loss": 1.3058,
      "step": 5798
    },
    {
      "epoch": 0.8306832831972497,
      "grad_norm": 1.164003610610962,
      "learning_rate": 1.4665152017486028e-05,
      "loss": 1.3271,
      "step": 5799
    },
    {
      "epoch": 0.8308265291505516,
      "grad_norm": 1.1846450567245483,
      "learning_rate": 1.4640972149453969e-05,
      "loss": 1.3016,
      "step": 5800
    },
    {
      "epoch": 0.8309697751038533,
      "grad_norm": 1.4839425086975098,
      "learning_rate": 1.4616810657108304e-05,
      "loss": 1.272,
      "step": 5801
    },
    {
      "epoch": 0.8311130210571551,
      "grad_norm": 1.2670466899871826,
      "learning_rate": 1.4592667545650396e-05,
      "loss": 1.2919,
      "step": 5802
    },
    {
      "epoch": 0.831256267010457,
      "grad_norm": 1.0132611989974976,
      "learning_rate": 1.4568542820277686e-05,
      "loss": 1.4328,
      "step": 5803
    },
    {
      "epoch": 0.8313995129637588,
      "grad_norm": 1.1387439966201782,
      "learning_rate": 1.4544436486183577e-05,
      "loss": 1.4944,
      "step": 5804
    },
    {
      "epoch": 0.8315427589170606,
      "grad_norm": 1.0245195627212524,
      "learning_rate": 1.4520348548557583e-05,
      "loss": 1.3372,
      "step": 5805
    },
    {
      "epoch": 0.8316860048703624,
      "grad_norm": 1.5224705934524536,
      "learning_rate": 1.449627901258529e-05,
      "loss": 1.4102,
      "step": 5806
    },
    {
      "epoch": 0.8318292508236642,
      "grad_norm": 1.0465983152389526,
      "learning_rate": 1.4472227883448219e-05,
      "loss": 1.6267,
      "step": 5807
    },
    {
      "epoch": 0.831972496776966,
      "grad_norm": 1.0657180547714233,
      "learning_rate": 1.4448195166324008e-05,
      "loss": 1.4945,
      "step": 5808
    },
    {
      "epoch": 0.8321157427302679,
      "grad_norm": 0.9429823756217957,
      "learning_rate": 1.4424180866386283e-05,
      "loss": 1.524,
      "step": 5809
    },
    {
      "epoch": 0.8322589886835697,
      "grad_norm": 1.0257388353347778,
      "learning_rate": 1.4400184988804754e-05,
      "loss": 1.4285,
      "step": 5810
    },
    {
      "epoch": 0.8324022346368715,
      "grad_norm": 1.042864203453064,
      "learning_rate": 1.4376207538745134e-05,
      "loss": 1.4724,
      "step": 5811
    },
    {
      "epoch": 0.8325454805901733,
      "grad_norm": 1.0017151832580566,
      "learning_rate": 1.4352248521369161e-05,
      "loss": 1.2386,
      "step": 5812
    },
    {
      "epoch": 0.8326887265434751,
      "grad_norm": 1.0717041492462158,
      "learning_rate": 1.432830794183464e-05,
      "loss": 1.3436,
      "step": 5813
    },
    {
      "epoch": 0.832831972496777,
      "grad_norm": 1.0257350206375122,
      "learning_rate": 1.4304385805295384e-05,
      "loss": 1.4052,
      "step": 5814
    },
    {
      "epoch": 0.8329752184500788,
      "grad_norm": 1.087152361869812,
      "learning_rate": 1.4280482116901195e-05,
      "loss": 1.6241,
      "step": 5815
    },
    {
      "epoch": 0.8331184644033806,
      "grad_norm": 1.0746204853057861,
      "learning_rate": 1.425659688179799e-05,
      "loss": 1.3253,
      "step": 5816
    },
    {
      "epoch": 0.8332617103566824,
      "grad_norm": 1.1218425035476685,
      "learning_rate": 1.4232730105127689e-05,
      "loss": 1.4443,
      "step": 5817
    },
    {
      "epoch": 0.8334049563099842,
      "grad_norm": 1.0890130996704102,
      "learning_rate": 1.420888179202815e-05,
      "loss": 1.309,
      "step": 5818
    },
    {
      "epoch": 0.8335482022632861,
      "grad_norm": 1.0417159795761108,
      "learning_rate": 1.4185051947633377e-05,
      "loss": 1.4817,
      "step": 5819
    },
    {
      "epoch": 0.8336914482165879,
      "grad_norm": 1.1337549686431885,
      "learning_rate": 1.416124057707331e-05,
      "loss": 1.3495,
      "step": 5820
    },
    {
      "epoch": 0.8338346941698896,
      "grad_norm": 1.3896255493164062,
      "learning_rate": 1.413744768547398e-05,
      "loss": 1.3021,
      "step": 5821
    },
    {
      "epoch": 0.8339779401231915,
      "grad_norm": 1.1626311540603638,
      "learning_rate": 1.4113673277957395e-05,
      "loss": 1.4148,
      "step": 5822
    },
    {
      "epoch": 0.8341211860764933,
      "grad_norm": 1.144176959991455,
      "learning_rate": 1.408991735964159e-05,
      "loss": 1.4197,
      "step": 5823
    },
    {
      "epoch": 0.8342644320297952,
      "grad_norm": 1.0574923753738403,
      "learning_rate": 1.4066179935640666e-05,
      "loss": 1.3387,
      "step": 5824
    },
    {
      "epoch": 0.834407677983097,
      "grad_norm": 1.0092427730560303,
      "learning_rate": 1.4042461011064634e-05,
      "loss": 1.3207,
      "step": 5825
    },
    {
      "epoch": 0.8345509239363988,
      "grad_norm": 1.3478448390960693,
      "learning_rate": 1.401876059101962e-05,
      "loss": 1.4324,
      "step": 5826
    },
    {
      "epoch": 0.8346941698897006,
      "grad_norm": 1.2492777109146118,
      "learning_rate": 1.3995078680607776e-05,
      "loss": 1.331,
      "step": 5827
    },
    {
      "epoch": 0.8348374158430024,
      "grad_norm": 0.9944934844970703,
      "learning_rate": 1.3971415284927192e-05,
      "loss": 1.4097,
      "step": 5828
    },
    {
      "epoch": 0.8349806617963043,
      "grad_norm": 1.1373724937438965,
      "learning_rate": 1.3947770409072014e-05,
      "loss": 1.352,
      "step": 5829
    },
    {
      "epoch": 0.8351239077496061,
      "grad_norm": 1.0326415300369263,
      "learning_rate": 1.3924144058132405e-05,
      "loss": 1.5714,
      "step": 5830
    },
    {
      "epoch": 0.8352671537029079,
      "grad_norm": 1.0562578439712524,
      "learning_rate": 1.3900536237194528e-05,
      "loss": 1.444,
      "step": 5831
    },
    {
      "epoch": 0.8354103996562097,
      "grad_norm": 0.9469882249832153,
      "learning_rate": 1.3876946951340598e-05,
      "loss": 1.5159,
      "step": 5832
    },
    {
      "epoch": 0.8355536456095115,
      "grad_norm": 1.2857955694198608,
      "learning_rate": 1.3853376205648727e-05,
      "loss": 1.2951,
      "step": 5833
    },
    {
      "epoch": 0.8356968915628133,
      "grad_norm": 1.136573314666748,
      "learning_rate": 1.3829824005193181e-05,
      "loss": 1.4218,
      "step": 5834
    },
    {
      "epoch": 0.8358401375161152,
      "grad_norm": 1.0780973434448242,
      "learning_rate": 1.3806290355044171e-05,
      "loss": 1.3424,
      "step": 5835
    },
    {
      "epoch": 0.835983383469417,
      "grad_norm": 1.01009202003479,
      "learning_rate": 1.3782775260267856e-05,
      "loss": 1.4121,
      "step": 5836
    },
    {
      "epoch": 0.8361266294227188,
      "grad_norm": 0.9678208827972412,
      "learning_rate": 1.3759278725926472e-05,
      "loss": 1.4011,
      "step": 5837
    },
    {
      "epoch": 0.8362698753760206,
      "grad_norm": 1.0487544536590576,
      "learning_rate": 1.3735800757078287e-05,
      "loss": 1.3903,
      "step": 5838
    },
    {
      "epoch": 0.8364131213293224,
      "grad_norm": 1.0286619663238525,
      "learning_rate": 1.3712341358777469e-05,
      "loss": 1.3095,
      "step": 5839
    },
    {
      "epoch": 0.8365563672826243,
      "grad_norm": 0.9621546864509583,
      "learning_rate": 1.3688900536074278e-05,
      "loss": 1.6125,
      "step": 5840
    },
    {
      "epoch": 0.8366996132359261,
      "grad_norm": 1.0732231140136719,
      "learning_rate": 1.3665478294014922e-05,
      "loss": 1.3224,
      "step": 5841
    },
    {
      "epoch": 0.8368428591892279,
      "grad_norm": 1.0345354080200195,
      "learning_rate": 1.3642074637641644e-05,
      "loss": 1.3836,
      "step": 5842
    },
    {
      "epoch": 0.8369861051425297,
      "grad_norm": 1.2423455715179443,
      "learning_rate": 1.3618689571992705e-05,
      "loss": 1.4345,
      "step": 5843
    },
    {
      "epoch": 0.8371293510958315,
      "grad_norm": 0.9972796440124512,
      "learning_rate": 1.359532310210223e-05,
      "loss": 1.4318,
      "step": 5844
    },
    {
      "epoch": 0.8372725970491334,
      "grad_norm": 1.119672417640686,
      "learning_rate": 1.3571975233000544e-05,
      "loss": 1.3708,
      "step": 5845
    },
    {
      "epoch": 0.8374158430024352,
      "grad_norm": 1.4175831079483032,
      "learning_rate": 1.3548645969713848e-05,
      "loss": 1.3858,
      "step": 5846
    },
    {
      "epoch": 0.837559088955737,
      "grad_norm": 1.1079649925231934,
      "learning_rate": 1.352533531726432e-05,
      "loss": 1.3008,
      "step": 5847
    },
    {
      "epoch": 0.8377023349090388,
      "grad_norm": 1.0787467956542969,
      "learning_rate": 1.3502043280670174e-05,
      "loss": 1.2124,
      "step": 5848
    },
    {
      "epoch": 0.8378455808623406,
      "grad_norm": 1.1681524515151978,
      "learning_rate": 1.3478769864945618e-05,
      "loss": 1.3824,
      "step": 5849
    },
    {
      "epoch": 0.8379888268156425,
      "grad_norm": 1.1513557434082031,
      "learning_rate": 1.3455515075100844e-05,
      "loss": 1.4161,
      "step": 5850
    },
    {
      "epoch": 0.8381320727689443,
      "grad_norm": 0.9557899832725525,
      "learning_rate": 1.3432278916142027e-05,
      "loss": 1.307,
      "step": 5851
    },
    {
      "epoch": 0.8382753187222461,
      "grad_norm": 1.0192763805389404,
      "learning_rate": 1.3409061393071343e-05,
      "loss": 1.3341,
      "step": 5852
    },
    {
      "epoch": 0.8384185646755479,
      "grad_norm": 1.173835277557373,
      "learning_rate": 1.3385862510886971e-05,
      "loss": 1.4787,
      "step": 5853
    },
    {
      "epoch": 0.8385618106288497,
      "grad_norm": 1.1230459213256836,
      "learning_rate": 1.3362682274583017e-05,
      "loss": 1.2567,
      "step": 5854
    },
    {
      "epoch": 0.8387050565821516,
      "grad_norm": 1.2405790090560913,
      "learning_rate": 1.3339520689149599e-05,
      "loss": 1.3182,
      "step": 5855
    },
    {
      "epoch": 0.8388483025354534,
      "grad_norm": 1.0970443487167358,
      "learning_rate": 1.3316377759572906e-05,
      "loss": 1.4368,
      "step": 5856
    },
    {
      "epoch": 0.8389915484887552,
      "grad_norm": 1.1253119707107544,
      "learning_rate": 1.3293253490834978e-05,
      "loss": 1.4241,
      "step": 5857
    },
    {
      "epoch": 0.839134794442057,
      "grad_norm": 1.1183799505233765,
      "learning_rate": 1.3270147887913909e-05,
      "loss": 1.4552,
      "step": 5858
    },
    {
      "epoch": 0.8392780403953588,
      "grad_norm": 1.2072148323059082,
      "learning_rate": 1.3247060955783775e-05,
      "loss": 1.3884,
      "step": 5859
    },
    {
      "epoch": 0.8394212863486606,
      "grad_norm": 1.349250316619873,
      "learning_rate": 1.322399269941461e-05,
      "loss": 1.4938,
      "step": 5860
    },
    {
      "epoch": 0.8395645323019625,
      "grad_norm": 1.1933029890060425,
      "learning_rate": 1.320094312377247e-05,
      "loss": 1.3254,
      "step": 5861
    },
    {
      "epoch": 0.8397077782552643,
      "grad_norm": 1.1124521493911743,
      "learning_rate": 1.3177912233819289e-05,
      "loss": 1.3491,
      "step": 5862
    },
    {
      "epoch": 0.839851024208566,
      "grad_norm": 1.0690639019012451,
      "learning_rate": 1.3154900034513106e-05,
      "loss": 1.5199,
      "step": 5863
    },
    {
      "epoch": 0.8399942701618679,
      "grad_norm": 0.9684329628944397,
      "learning_rate": 1.3131906530807891e-05,
      "loss": 1.5549,
      "step": 5864
    },
    {
      "epoch": 0.8401375161151697,
      "grad_norm": 1.0842161178588867,
      "learning_rate": 1.3108931727653529e-05,
      "loss": 1.2659,
      "step": 5865
    },
    {
      "epoch": 0.8402807620684716,
      "grad_norm": 1.289391279220581,
      "learning_rate": 1.3085975629995916e-05,
      "loss": 1.2047,
      "step": 5866
    },
    {
      "epoch": 0.8404240080217734,
      "grad_norm": 0.9037125706672668,
      "learning_rate": 1.3063038242777014e-05,
      "loss": 1.4522,
      "step": 5867
    },
    {
      "epoch": 0.8405672539750753,
      "grad_norm": 1.1377625465393066,
      "learning_rate": 1.3040119570934595e-05,
      "loss": 1.199,
      "step": 5868
    },
    {
      "epoch": 0.840710499928377,
      "grad_norm": 1.0364843606948853,
      "learning_rate": 1.3017219619402509e-05,
      "loss": 1.481,
      "step": 5869
    },
    {
      "epoch": 0.8408537458816788,
      "grad_norm": 1.1615277528762817,
      "learning_rate": 1.299433839311055e-05,
      "loss": 1.4197,
      "step": 5870
    },
    {
      "epoch": 0.8409969918349807,
      "grad_norm": 0.972763180732727,
      "learning_rate": 1.2971475896984475e-05,
      "loss": 1.4199,
      "step": 5871
    },
    {
      "epoch": 0.8411402377882825,
      "grad_norm": 0.9330835342407227,
      "learning_rate": 1.2948632135946048e-05,
      "loss": 1.4107,
      "step": 5872
    },
    {
      "epoch": 0.8412834837415843,
      "grad_norm": 0.9538688659667969,
      "learning_rate": 1.292580711491289e-05,
      "loss": 1.4094,
      "step": 5873
    },
    {
      "epoch": 0.8414267296948861,
      "grad_norm": 1.1188175678253174,
      "learning_rate": 1.2903000838798752e-05,
      "loss": 1.4263,
      "step": 5874
    },
    {
      "epoch": 0.8415699756481879,
      "grad_norm": 0.95613032579422,
      "learning_rate": 1.288021331251319e-05,
      "loss": 1.3848,
      "step": 5875
    },
    {
      "epoch": 0.8417132216014898,
      "grad_norm": 1.1459360122680664,
      "learning_rate": 1.2857444540961805e-05,
      "loss": 1.2825,
      "step": 5876
    },
    {
      "epoch": 0.8418564675547916,
      "grad_norm": 0.9626212120056152,
      "learning_rate": 1.2834694529046209e-05,
      "loss": 1.2882,
      "step": 5877
    },
    {
      "epoch": 0.8419997135080934,
      "grad_norm": 1.1761460304260254,
      "learning_rate": 1.2811963281663853e-05,
      "loss": 1.3349,
      "step": 5878
    },
    {
      "epoch": 0.8421429594613952,
      "grad_norm": 1.0111474990844727,
      "learning_rate": 1.278925080370823e-05,
      "loss": 1.5591,
      "step": 5879
    },
    {
      "epoch": 0.842286205414697,
      "grad_norm": 1.220505952835083,
      "learning_rate": 1.2766557100068787e-05,
      "loss": 1.2215,
      "step": 5880
    },
    {
      "epoch": 0.8424294513679988,
      "grad_norm": 1.0335100889205933,
      "learning_rate": 1.2743882175630905e-05,
      "loss": 1.5031,
      "step": 5881
    },
    {
      "epoch": 0.8425726973213007,
      "grad_norm": 1.2098928689956665,
      "learning_rate": 1.2721226035275957e-05,
      "loss": 1.3362,
      "step": 5882
    },
    {
      "epoch": 0.8427159432746025,
      "grad_norm": 1.0982825756072998,
      "learning_rate": 1.2698588683881186e-05,
      "loss": 1.3165,
      "step": 5883
    },
    {
      "epoch": 0.8428591892279043,
      "grad_norm": 1.0473443269729614,
      "learning_rate": 1.2675970126319914e-05,
      "loss": 1.4683,
      "step": 5884
    },
    {
      "epoch": 0.8430024351812061,
      "grad_norm": 1.0436902046203613,
      "learning_rate": 1.2653370367461359e-05,
      "loss": 1.4847,
      "step": 5885
    },
    {
      "epoch": 0.8431456811345079,
      "grad_norm": 1.2056255340576172,
      "learning_rate": 1.2630789412170663e-05,
      "loss": 1.266,
      "step": 5886
    },
    {
      "epoch": 0.8432889270878098,
      "grad_norm": 1.0287574529647827,
      "learning_rate": 1.2608227265308925e-05,
      "loss": 1.1773,
      "step": 5887
    },
    {
      "epoch": 0.8434321730411116,
      "grad_norm": 1.2423624992370605,
      "learning_rate": 1.2585683931733283e-05,
      "loss": 1.2254,
      "step": 5888
    },
    {
      "epoch": 0.8435754189944135,
      "grad_norm": 1.3301151990890503,
      "learning_rate": 1.256315941629671e-05,
      "loss": 1.4898,
      "step": 5889
    },
    {
      "epoch": 0.8437186649477152,
      "grad_norm": 1.149516224861145,
      "learning_rate": 1.2540653723848183e-05,
      "loss": 1.3118,
      "step": 5890
    },
    {
      "epoch": 0.843861910901017,
      "grad_norm": 1.135151982307434,
      "learning_rate": 1.2518166859232627e-05,
      "loss": 1.5089,
      "step": 5891
    },
    {
      "epoch": 0.8440051568543189,
      "grad_norm": 1.168349027633667,
      "learning_rate": 1.2495698827290902e-05,
      "loss": 1.2862,
      "step": 5892
    },
    {
      "epoch": 0.8441484028076207,
      "grad_norm": 1.0435901880264282,
      "learning_rate": 1.2473249632859862e-05,
      "loss": 1.3335,
      "step": 5893
    },
    {
      "epoch": 0.8442916487609226,
      "grad_norm": 1.2024805545806885,
      "learning_rate": 1.2450819280772164e-05,
      "loss": 1.5125,
      "step": 5894
    },
    {
      "epoch": 0.8444348947142243,
      "grad_norm": 1.0005171298980713,
      "learning_rate": 1.2428407775856598e-05,
      "loss": 1.4868,
      "step": 5895
    },
    {
      "epoch": 0.8445781406675261,
      "grad_norm": 0.9306319355964661,
      "learning_rate": 1.2406015122937809e-05,
      "loss": 1.4259,
      "step": 5896
    },
    {
      "epoch": 0.844721386620828,
      "grad_norm": 1.1300095319747925,
      "learning_rate": 1.2383641326836314e-05,
      "loss": 1.3294,
      "step": 5897
    },
    {
      "epoch": 0.8448646325741298,
      "grad_norm": 1.4026776552200317,
      "learning_rate": 1.2361286392368676e-05,
      "loss": 1.3565,
      "step": 5898
    },
    {
      "epoch": 0.8450078785274316,
      "grad_norm": 1.4191770553588867,
      "learning_rate": 1.2338950324347354e-05,
      "loss": 1.2097,
      "step": 5899
    },
    {
      "epoch": 0.8451511244807334,
      "grad_norm": 0.9725842475891113,
      "learning_rate": 1.2316633127580757e-05,
      "loss": 1.3567,
      "step": 5900
    },
    {
      "epoch": 0.8452943704340352,
      "grad_norm": 0.9735100865364075,
      "learning_rate": 1.2294334806873231e-05,
      "loss": 1.223,
      "step": 5901
    },
    {
      "epoch": 0.845437616387337,
      "grad_norm": 0.9836328029632568,
      "learning_rate": 1.2272055367025027e-05,
      "loss": 1.4428,
      "step": 5902
    },
    {
      "epoch": 0.8455808623406389,
      "grad_norm": 1.0228328704833984,
      "learning_rate": 1.2249794812832404e-05,
      "loss": 1.5348,
      "step": 5903
    },
    {
      "epoch": 0.8457241082939407,
      "grad_norm": 1.0916717052459717,
      "learning_rate": 1.2227553149087446e-05,
      "loss": 1.4367,
      "step": 5904
    },
    {
      "epoch": 0.8458673542472425,
      "grad_norm": 1.0190634727478027,
      "learning_rate": 1.2205330380578251e-05,
      "loss": 1.6492,
      "step": 5905
    },
    {
      "epoch": 0.8460106002005443,
      "grad_norm": 1.0167508125305176,
      "learning_rate": 1.2183126512088882e-05,
      "loss": 1.1656,
      "step": 5906
    },
    {
      "epoch": 0.8461538461538461,
      "grad_norm": 1.166505217552185,
      "learning_rate": 1.216094154839923e-05,
      "loss": 1.4668,
      "step": 5907
    },
    {
      "epoch": 0.846297092107148,
      "grad_norm": 1.1457016468048096,
      "learning_rate": 1.2138775494285182e-05,
      "loss": 1.432,
      "step": 5908
    },
    {
      "epoch": 0.8464403380604498,
      "grad_norm": 1.2262722253799438,
      "learning_rate": 1.2116628354518544e-05,
      "loss": 1.3973,
      "step": 5909
    },
    {
      "epoch": 0.8465835840137517,
      "grad_norm": 0.9604052901268005,
      "learning_rate": 1.2094500133867038e-05,
      "loss": 1.3984,
      "step": 5910
    },
    {
      "epoch": 0.8467268299670534,
      "grad_norm": 1.1697322130203247,
      "learning_rate": 1.207239083709435e-05,
      "loss": 1.1937,
      "step": 5911
    },
    {
      "epoch": 0.8468700759203552,
      "grad_norm": 0.9779345393180847,
      "learning_rate": 1.2050300468960007e-05,
      "loss": 1.3652,
      "step": 5912
    },
    {
      "epoch": 0.8470133218736571,
      "grad_norm": 1.23443603515625,
      "learning_rate": 1.2028229034219584e-05,
      "loss": 1.5366,
      "step": 5913
    },
    {
      "epoch": 0.8471565678269589,
      "grad_norm": 1.264336347579956,
      "learning_rate": 1.2006176537624491e-05,
      "loss": 1.3569,
      "step": 5914
    },
    {
      "epoch": 0.8472998137802608,
      "grad_norm": 1.0483895540237427,
      "learning_rate": 1.1984142983922075e-05,
      "loss": 1.316,
      "step": 5915
    },
    {
      "epoch": 0.8474430597335625,
      "grad_norm": 1.1203054189682007,
      "learning_rate": 1.1962128377855597e-05,
      "loss": 1.3518,
      "step": 5916
    },
    {
      "epoch": 0.8475863056868643,
      "grad_norm": 1.14857017993927,
      "learning_rate": 1.194013272416431e-05,
      "loss": 1.5212,
      "step": 5917
    },
    {
      "epoch": 0.8477295516401662,
      "grad_norm": 1.077927827835083,
      "learning_rate": 1.1918156027583294e-05,
      "loss": 1.3337,
      "step": 5918
    },
    {
      "epoch": 0.847872797593468,
      "grad_norm": 0.9853607416152954,
      "learning_rate": 1.1896198292843597e-05,
      "loss": 1.3714,
      "step": 5919
    },
    {
      "epoch": 0.8480160435467698,
      "grad_norm": 1.0781772136688232,
      "learning_rate": 1.187425952467217e-05,
      "loss": 1.5929,
      "step": 5920
    },
    {
      "epoch": 0.8481592895000716,
      "grad_norm": 1.2823963165283203,
      "learning_rate": 1.1852339727791906e-05,
      "loss": 1.52,
      "step": 5921
    },
    {
      "epoch": 0.8483025354533734,
      "grad_norm": 1.1340436935424805,
      "learning_rate": 1.1830438906921592e-05,
      "loss": 1.5589,
      "step": 5922
    },
    {
      "epoch": 0.8484457814066753,
      "grad_norm": 1.2404865026474,
      "learning_rate": 1.1808557066775883e-05,
      "loss": 1.3667,
      "step": 5923
    },
    {
      "epoch": 0.8485890273599771,
      "grad_norm": 0.9128482341766357,
      "learning_rate": 1.1786694212065463e-05,
      "loss": 1.2467,
      "step": 5924
    },
    {
      "epoch": 0.8487322733132789,
      "grad_norm": 1.0264840126037598,
      "learning_rate": 1.1764850347496859e-05,
      "loss": 1.3486,
      "step": 5925
    },
    {
      "epoch": 0.8488755192665807,
      "grad_norm": 1.2550785541534424,
      "learning_rate": 1.1743025477772474e-05,
      "loss": 1.4195,
      "step": 5926
    },
    {
      "epoch": 0.8490187652198825,
      "grad_norm": 1.2878550291061401,
      "learning_rate": 1.1721219607590683e-05,
      "loss": 1.3839,
      "step": 5927
    },
    {
      "epoch": 0.8491620111731844,
      "grad_norm": 1.253336787223816,
      "learning_rate": 1.1699432741645754e-05,
      "loss": 1.3767,
      "step": 5928
    },
    {
      "epoch": 0.8493052571264862,
      "grad_norm": 0.9495038986206055,
      "learning_rate": 1.1677664884627847e-05,
      "loss": 1.3349,
      "step": 5929
    },
    {
      "epoch": 0.849448503079788,
      "grad_norm": 1.0434354543685913,
      "learning_rate": 1.1655916041223059e-05,
      "loss": 1.3824,
      "step": 5930
    },
    {
      "epoch": 0.8495917490330898,
      "grad_norm": 1.0651259422302246,
      "learning_rate": 1.1634186216113363e-05,
      "loss": 1.1771,
      "step": 5931
    },
    {
      "epoch": 0.8497349949863916,
      "grad_norm": 1.1817960739135742,
      "learning_rate": 1.1612475413976686e-05,
      "loss": 1.1969,
      "step": 5932
    },
    {
      "epoch": 0.8498782409396934,
      "grad_norm": 1.1750916242599487,
      "learning_rate": 1.159078363948678e-05,
      "loss": 1.4054,
      "step": 5933
    },
    {
      "epoch": 0.8500214868929953,
      "grad_norm": 1.1518166065216064,
      "learning_rate": 1.1569110897313351e-05,
      "loss": 1.5648,
      "step": 5934
    },
    {
      "epoch": 0.8501647328462971,
      "grad_norm": 1.1537971496582031,
      "learning_rate": 1.1547457192122058e-05,
      "loss": 1.4468,
      "step": 5935
    },
    {
      "epoch": 0.850307978799599,
      "grad_norm": 0.9919856786727905,
      "learning_rate": 1.1525822528574349e-05,
      "loss": 1.2949,
      "step": 5936
    },
    {
      "epoch": 0.8504512247529007,
      "grad_norm": 1.0180913209915161,
      "learning_rate": 1.1504206911327653e-05,
      "loss": 1.4565,
      "step": 5937
    },
    {
      "epoch": 0.8505944707062025,
      "grad_norm": 0.9992722272872925,
      "learning_rate": 1.1482610345035283e-05,
      "loss": 1.5096,
      "step": 5938
    },
    {
      "epoch": 0.8507377166595044,
      "grad_norm": 1.1832038164138794,
      "learning_rate": 1.146103283434643e-05,
      "loss": 1.3783,
      "step": 5939
    },
    {
      "epoch": 0.8508809626128062,
      "grad_norm": 1.0289227962493896,
      "learning_rate": 1.1439474383906223e-05,
      "loss": 1.5543,
      "step": 5940
    },
    {
      "epoch": 0.851024208566108,
      "grad_norm": 1.0689349174499512,
      "learning_rate": 1.141793499835564e-05,
      "loss": 1.3577,
      "step": 5941
    },
    {
      "epoch": 0.8511674545194098,
      "grad_norm": 1.068604826927185,
      "learning_rate": 1.1396414682331592e-05,
      "loss": 1.5291,
      "step": 5942
    },
    {
      "epoch": 0.8513107004727116,
      "grad_norm": 1.0408991575241089,
      "learning_rate": 1.1374913440466872e-05,
      "loss": 1.2585,
      "step": 5943
    },
    {
      "epoch": 0.8514539464260135,
      "grad_norm": 1.0943585634231567,
      "learning_rate": 1.1353431277390126e-05,
      "loss": 1.3339,
      "step": 5944
    },
    {
      "epoch": 0.8515971923793153,
      "grad_norm": 1.0280113220214844,
      "learning_rate": 1.1331968197725984e-05,
      "loss": 1.4563,
      "step": 5945
    },
    {
      "epoch": 0.8517404383326171,
      "grad_norm": 1.4389575719833374,
      "learning_rate": 1.1310524206094919e-05,
      "loss": 1.5263,
      "step": 5946
    },
    {
      "epoch": 0.8518836842859189,
      "grad_norm": 1.0397403240203857,
      "learning_rate": 1.128909930711325e-05,
      "loss": 1.4598,
      "step": 5947
    },
    {
      "epoch": 0.8520269302392207,
      "grad_norm": 0.9412740468978882,
      "learning_rate": 1.126769350539324e-05,
      "loss": 1.7607,
      "step": 5948
    },
    {
      "epoch": 0.8521701761925226,
      "grad_norm": 0.9969056844711304,
      "learning_rate": 1.1246306805543038e-05,
      "loss": 1.3967,
      "step": 5949
    },
    {
      "epoch": 0.8523134221458244,
      "grad_norm": 1.126189947128296,
      "learning_rate": 1.1224939212166662e-05,
      "loss": 1.4216,
      "step": 5950
    },
    {
      "epoch": 0.8524566680991262,
      "grad_norm": 1.1168287992477417,
      "learning_rate": 1.1203590729864033e-05,
      "loss": 1.4634,
      "step": 5951
    },
    {
      "epoch": 0.852599914052428,
      "grad_norm": 1.0573464632034302,
      "learning_rate": 1.1182261363230962e-05,
      "loss": 1.4072,
      "step": 5952
    },
    {
      "epoch": 0.8527431600057298,
      "grad_norm": 1.0440900325775146,
      "learning_rate": 1.116095111685913e-05,
      "loss": 1.2846,
      "step": 5953
    },
    {
      "epoch": 0.8528864059590316,
      "grad_norm": 1.2952592372894287,
      "learning_rate": 1.1139659995336082e-05,
      "loss": 1.3101,
      "step": 5954
    },
    {
      "epoch": 0.8530296519123335,
      "grad_norm": 1.1916544437408447,
      "learning_rate": 1.1118388003245262e-05,
      "loss": 1.2321,
      "step": 5955
    },
    {
      "epoch": 0.8531728978656353,
      "grad_norm": 1.1573394536972046,
      "learning_rate": 1.1097135145166072e-05,
      "loss": 1.3454,
      "step": 5956
    },
    {
      "epoch": 0.8533161438189372,
      "grad_norm": 1.0574326515197754,
      "learning_rate": 1.1075901425673651e-05,
      "loss": 1.5053,
      "step": 5957
    },
    {
      "epoch": 0.8534593897722389,
      "grad_norm": 1.1780495643615723,
      "learning_rate": 1.1054686849339124e-05,
      "loss": 1.4924,
      "step": 5958
    },
    {
      "epoch": 0.8536026357255407,
      "grad_norm": 1.141882061958313,
      "learning_rate": 1.103349142072947e-05,
      "loss": 1.501,
      "step": 5959
    },
    {
      "epoch": 0.8537458816788426,
      "grad_norm": 0.8827534317970276,
      "learning_rate": 1.1012315144407538e-05,
      "loss": 1.4229,
      "step": 5960
    },
    {
      "epoch": 0.8538891276321444,
      "grad_norm": 1.001202940940857,
      "learning_rate": 1.0991158024932068e-05,
      "loss": 1.201,
      "step": 5961
    },
    {
      "epoch": 0.8540323735854463,
      "grad_norm": 1.0827302932739258,
      "learning_rate": 1.0970020066857601e-05,
      "loss": 1.4203,
      "step": 5962
    },
    {
      "epoch": 0.854175619538748,
      "grad_norm": 1.2331846952438354,
      "learning_rate": 1.0948901274734691e-05,
      "loss": 1.3337,
      "step": 5963
    },
    {
      "epoch": 0.8543188654920498,
      "grad_norm": 1.075698971748352,
      "learning_rate": 1.0927801653109682e-05,
      "loss": 1.3747,
      "step": 5964
    },
    {
      "epoch": 0.8544621114453517,
      "grad_norm": 1.0469216108322144,
      "learning_rate": 1.0906721206524762e-05,
      "loss": 1.1162,
      "step": 5965
    },
    {
      "epoch": 0.8546053573986535,
      "grad_norm": 1.0299696922302246,
      "learning_rate": 1.0885659939518033e-05,
      "loss": 1.4588,
      "step": 5966
    },
    {
      "epoch": 0.8547486033519553,
      "grad_norm": 0.8674678206443787,
      "learning_rate": 1.0864617856623516e-05,
      "loss": 1.4715,
      "step": 5967
    },
    {
      "epoch": 0.8548918493052571,
      "grad_norm": 0.99725741147995,
      "learning_rate": 1.0843594962371007e-05,
      "loss": 1.5728,
      "step": 5968
    },
    {
      "epoch": 0.8550350952585589,
      "grad_norm": 1.1514919996261597,
      "learning_rate": 1.0822591261286207e-05,
      "loss": 1.3922,
      "step": 5969
    },
    {
      "epoch": 0.8551783412118608,
      "grad_norm": 1.228461503982544,
      "learning_rate": 1.0801606757890725e-05,
      "loss": 1.5458,
      "step": 5970
    },
    {
      "epoch": 0.8553215871651626,
      "grad_norm": 1.1286852359771729,
      "learning_rate": 1.078064145670199e-05,
      "loss": 1.4287,
      "step": 5971
    },
    {
      "epoch": 0.8554648331184644,
      "grad_norm": 1.0499244928359985,
      "learning_rate": 1.0759695362233324e-05,
      "loss": 1.1998,
      "step": 5972
    },
    {
      "epoch": 0.8556080790717662,
      "grad_norm": 1.0609266757965088,
      "learning_rate": 1.0738768478993854e-05,
      "loss": 1.3644,
      "step": 5973
    },
    {
      "epoch": 0.855751325025068,
      "grad_norm": 1.0579861402511597,
      "learning_rate": 1.0717860811488689e-05,
      "loss": 1.4787,
      "step": 5974
    },
    {
      "epoch": 0.8558945709783699,
      "grad_norm": 1.088641881942749,
      "learning_rate": 1.0696972364218705e-05,
      "loss": 1.4341,
      "step": 5975
    },
    {
      "epoch": 0.8560378169316717,
      "grad_norm": 1.151374101638794,
      "learning_rate": 1.0676103141680649e-05,
      "loss": 1.4171,
      "step": 5976
    },
    {
      "epoch": 0.8561810628849735,
      "grad_norm": 1.2851057052612305,
      "learning_rate": 1.0655253148367173e-05,
      "loss": 1.4876,
      "step": 5977
    },
    {
      "epoch": 0.8563243088382754,
      "grad_norm": 1.0351163148880005,
      "learning_rate": 1.0634422388766752e-05,
      "loss": 1.3716,
      "step": 5978
    },
    {
      "epoch": 0.8564675547915771,
      "grad_norm": 1.0411376953125,
      "learning_rate": 1.0613610867363722e-05,
      "loss": 1.3855,
      "step": 5979
    },
    {
      "epoch": 0.8566108007448789,
      "grad_norm": 0.9885153770446777,
      "learning_rate": 1.059281858863832e-05,
      "loss": 1.1327,
      "step": 5980
    },
    {
      "epoch": 0.8567540466981808,
      "grad_norm": 1.0979875326156616,
      "learning_rate": 1.0572045557066579e-05,
      "loss": 1.315,
      "step": 5981
    },
    {
      "epoch": 0.8568972926514826,
      "grad_norm": 1.2584151029586792,
      "learning_rate": 1.0551291777120464e-05,
      "loss": 1.3016,
      "step": 5982
    },
    {
      "epoch": 0.8570405386047845,
      "grad_norm": 1.2990772724151611,
      "learning_rate": 1.053055725326768e-05,
      "loss": 1.315,
      "step": 5983
    },
    {
      "epoch": 0.8571837845580862,
      "grad_norm": 0.9196381568908691,
      "learning_rate": 1.0509841989971891e-05,
      "loss": 1.5631,
      "step": 5984
    },
    {
      "epoch": 0.857327030511388,
      "grad_norm": 1.2640011310577393,
      "learning_rate": 1.0489145991692618e-05,
      "loss": 1.496,
      "step": 5985
    },
    {
      "epoch": 0.8574702764646899,
      "grad_norm": 1.1008282899856567,
      "learning_rate": 1.0468469262885139e-05,
      "loss": 1.3345,
      "step": 5986
    },
    {
      "epoch": 0.8576135224179917,
      "grad_norm": 0.9982852339744568,
      "learning_rate": 1.0447811808000673e-05,
      "loss": 1.5105,
      "step": 5987
    },
    {
      "epoch": 0.8577567683712936,
      "grad_norm": 0.9490407705307007,
      "learning_rate": 1.0427173631486254e-05,
      "loss": 1.218,
      "step": 5988
    },
    {
      "epoch": 0.8579000143245953,
      "grad_norm": 1.120341181755066,
      "learning_rate": 1.0406554737784757e-05,
      "loss": 1.5099,
      "step": 5989
    },
    {
      "epoch": 0.8580432602778971,
      "grad_norm": 0.9381440281867981,
      "learning_rate": 1.0385955131334958e-05,
      "loss": 1.4984,
      "step": 5990
    },
    {
      "epoch": 0.858186506231199,
      "grad_norm": 0.8875547647476196,
      "learning_rate": 1.0365374816571382e-05,
      "loss": 1.3073,
      "step": 5991
    },
    {
      "epoch": 0.8583297521845008,
      "grad_norm": 1.1631871461868286,
      "learning_rate": 1.0344813797924501e-05,
      "loss": 1.3985,
      "step": 5992
    },
    {
      "epoch": 0.8584729981378026,
      "grad_norm": 1.0492743253707886,
      "learning_rate": 1.0324272079820618e-05,
      "loss": 1.173,
      "step": 5993
    },
    {
      "epoch": 0.8586162440911044,
      "grad_norm": 1.0785988569259644,
      "learning_rate": 1.0303749666681806e-05,
      "loss": 1.3367,
      "step": 5994
    },
    {
      "epoch": 0.8587594900444062,
      "grad_norm": 1.0852069854736328,
      "learning_rate": 1.0283246562926041e-05,
      "loss": 1.2261,
      "step": 5995
    },
    {
      "epoch": 0.8589027359977081,
      "grad_norm": 1.0551471710205078,
      "learning_rate": 1.0262762772967172e-05,
      "loss": 1.4964,
      "step": 5996
    },
    {
      "epoch": 0.8590459819510099,
      "grad_norm": 1.0236765146255493,
      "learning_rate": 1.0242298301214814e-05,
      "loss": 1.342,
      "step": 5997
    },
    {
      "epoch": 0.8591892279043117,
      "grad_norm": 1.1315295696258545,
      "learning_rate": 1.0221853152074468e-05,
      "loss": 1.468,
      "step": 5998
    },
    {
      "epoch": 0.8593324738576136,
      "grad_norm": 1.445736050605774,
      "learning_rate": 1.0201427329947477e-05,
      "loss": 1.2237,
      "step": 5999
    },
    {
      "epoch": 0.8594757198109153,
      "grad_norm": 1.1562144756317139,
      "learning_rate": 1.0181020839231014e-05,
      "loss": 1.2791,
      "step": 6000
    },
    {
      "epoch": 0.8596189657642171,
      "grad_norm": 1.061842441558838,
      "learning_rate": 1.0160633684318088e-05,
      "loss": 1.3866,
      "step": 6001
    },
    {
      "epoch": 0.859762211717519,
      "grad_norm": 1.0045994520187378,
      "learning_rate": 1.0140265869597554e-05,
      "loss": 1.3542,
      "step": 6002
    },
    {
      "epoch": 0.8599054576708208,
      "grad_norm": 1.0613834857940674,
      "learning_rate": 1.0119917399454105e-05,
      "loss": 1.5329,
      "step": 6003
    },
    {
      "epoch": 0.8600487036241227,
      "grad_norm": 1.2301291227340698,
      "learning_rate": 1.0099588278268246e-05,
      "loss": 1.4949,
      "step": 6004
    },
    {
      "epoch": 0.8601919495774244,
      "grad_norm": 0.9867377281188965,
      "learning_rate": 1.0079278510416313e-05,
      "loss": 1.3574,
      "step": 6005
    },
    {
      "epoch": 0.8603351955307262,
      "grad_norm": 1.4309240579605103,
      "learning_rate": 1.0058988100270561e-05,
      "loss": 1.3548,
      "step": 6006
    },
    {
      "epoch": 0.8604784414840281,
      "grad_norm": 1.0366567373275757,
      "learning_rate": 1.0038717052198965e-05,
      "loss": 1.3271,
      "step": 6007
    },
    {
      "epoch": 0.8606216874373299,
      "grad_norm": 1.2306365966796875,
      "learning_rate": 1.0018465370565388e-05,
      "loss": 1.3101,
      "step": 6008
    },
    {
      "epoch": 0.8607649333906318,
      "grad_norm": 1.1581447124481201,
      "learning_rate": 9.998233059729523e-06,
      "loss": 1.3947,
      "step": 6009
    },
    {
      "epoch": 0.8609081793439335,
      "grad_norm": 1.0660598278045654,
      "learning_rate": 9.978020124046882e-06,
      "loss": 1.3289,
      "step": 6010
    },
    {
      "epoch": 0.8610514252972353,
      "grad_norm": 1.0561665296554565,
      "learning_rate": 9.957826567868822e-06,
      "loss": 1.6065,
      "step": 6011
    },
    {
      "epoch": 0.8611946712505372,
      "grad_norm": 1.0173041820526123,
      "learning_rate": 9.937652395542474e-06,
      "loss": 1.4442,
      "step": 6012
    },
    {
      "epoch": 0.861337917203839,
      "grad_norm": 1.059556245803833,
      "learning_rate": 9.917497611410875e-06,
      "loss": 1.4359,
      "step": 6013
    },
    {
      "epoch": 0.8614811631571409,
      "grad_norm": 1.2258607149124146,
      "learning_rate": 9.897362219812879e-06,
      "loss": 1.3255,
      "step": 6014
    },
    {
      "epoch": 0.8616244091104426,
      "grad_norm": 0.9293045997619629,
      "learning_rate": 9.877246225083082e-06,
      "loss": 1.4518,
      "step": 6015
    },
    {
      "epoch": 0.8617676550637444,
      "grad_norm": 1.060230016708374,
      "learning_rate": 9.857149631551955e-06,
      "loss": 1.517,
      "step": 6016
    },
    {
      "epoch": 0.8619109010170463,
      "grad_norm": 1.045097827911377,
      "learning_rate": 9.837072443545858e-06,
      "loss": 1.3968,
      "step": 6017
    },
    {
      "epoch": 0.8620541469703481,
      "grad_norm": 0.9710952639579773,
      "learning_rate": 9.81701466538687e-06,
      "loss": 1.3088,
      "step": 6018
    },
    {
      "epoch": 0.8621973929236499,
      "grad_norm": 0.9889569878578186,
      "learning_rate": 9.796976301392934e-06,
      "loss": 1.6315,
      "step": 6019
    },
    {
      "epoch": 0.8623406388769517,
      "grad_norm": 1.4043844938278198,
      "learning_rate": 9.776957355877825e-06,
      "loss": 1.4982,
      "step": 6020
    },
    {
      "epoch": 0.8624838848302535,
      "grad_norm": 1.0389546155929565,
      "learning_rate": 9.756957833151115e-06,
      "loss": 1.4787,
      "step": 6021
    },
    {
      "epoch": 0.8626271307835554,
      "grad_norm": 1.059665322303772,
      "learning_rate": 9.736977737518249e-06,
      "loss": 1.2921,
      "step": 6022
    },
    {
      "epoch": 0.8627703767368572,
      "grad_norm": 1.1262787580490112,
      "learning_rate": 9.717017073280366e-06,
      "loss": 1.4134,
      "step": 6023
    },
    {
      "epoch": 0.862913622690159,
      "grad_norm": 1.1173547506332397,
      "learning_rate": 9.697075844734571e-06,
      "loss": 1.2979,
      "step": 6024
    },
    {
      "epoch": 0.8630568686434609,
      "grad_norm": 1.2087897062301636,
      "learning_rate": 9.67715405617371e-06,
      "loss": 1.2733,
      "step": 6025
    },
    {
      "epoch": 0.8632001145967626,
      "grad_norm": 1.0024877786636353,
      "learning_rate": 9.657251711886427e-06,
      "loss": 1.3509,
      "step": 6026
    },
    {
      "epoch": 0.8633433605500644,
      "grad_norm": 0.9519560933113098,
      "learning_rate": 9.63736881615721e-06,
      "loss": 1.151,
      "step": 6027
    },
    {
      "epoch": 0.8634866065033663,
      "grad_norm": 1.2101024389266968,
      "learning_rate": 9.617505373266355e-06,
      "loss": 1.317,
      "step": 6028
    },
    {
      "epoch": 0.8636298524566681,
      "grad_norm": 1.1807539463043213,
      "learning_rate": 9.597661387489987e-06,
      "loss": 1.5106,
      "step": 6029
    },
    {
      "epoch": 0.86377309840997,
      "grad_norm": 1.0933598279953003,
      "learning_rate": 9.577836863099999e-06,
      "loss": 1.5389,
      "step": 6030
    },
    {
      "epoch": 0.8639163443632717,
      "grad_norm": 1.077688217163086,
      "learning_rate": 9.558031804364142e-06,
      "loss": 1.5173,
      "step": 6031
    },
    {
      "epoch": 0.8640595903165735,
      "grad_norm": 1.1793551445007324,
      "learning_rate": 9.538246215545953e-06,
      "loss": 1.4578,
      "step": 6032
    },
    {
      "epoch": 0.8642028362698754,
      "grad_norm": 1.1994574069976807,
      "learning_rate": 9.518480100904769e-06,
      "loss": 1.3562,
      "step": 6033
    },
    {
      "epoch": 0.8643460822231772,
      "grad_norm": 1.1399985551834106,
      "learning_rate": 9.498733464695719e-06,
      "loss": 1.3796,
      "step": 6034
    },
    {
      "epoch": 0.864489328176479,
      "grad_norm": 1.0083770751953125,
      "learning_rate": 9.479006311169835e-06,
      "loss": 1.2143,
      "step": 6035
    },
    {
      "epoch": 0.8646325741297808,
      "grad_norm": 1.0196020603179932,
      "learning_rate": 9.459298644573833e-06,
      "loss": 1.4158,
      "step": 6036
    },
    {
      "epoch": 0.8647758200830826,
      "grad_norm": 0.9705145359039307,
      "learning_rate": 9.439610469150295e-06,
      "loss": 1.3836,
      "step": 6037
    },
    {
      "epoch": 0.8649190660363845,
      "grad_norm": 0.9643406867980957,
      "learning_rate": 9.419941789137598e-06,
      "loss": 1.4293,
      "step": 6038
    },
    {
      "epoch": 0.8650623119896863,
      "grad_norm": 1.2338895797729492,
      "learning_rate": 9.400292608769933e-06,
      "loss": 1.3262,
      "step": 6039
    },
    {
      "epoch": 0.8652055579429881,
      "grad_norm": 1.0757371187210083,
      "learning_rate": 9.380662932277285e-06,
      "loss": 1.4438,
      "step": 6040
    },
    {
      "epoch": 0.8653488038962899,
      "grad_norm": 1.120244026184082,
      "learning_rate": 9.361052763885392e-06,
      "loss": 1.2985,
      "step": 6041
    },
    {
      "epoch": 0.8654920498495917,
      "grad_norm": 1.031241536140442,
      "learning_rate": 9.341462107815891e-06,
      "loss": 1.4219,
      "step": 6042
    },
    {
      "epoch": 0.8656352958028936,
      "grad_norm": 1.0601811408996582,
      "learning_rate": 9.321890968286173e-06,
      "loss": 1.2912,
      "step": 6043
    },
    {
      "epoch": 0.8657785417561954,
      "grad_norm": 1.145644187927246,
      "learning_rate": 9.302339349509371e-06,
      "loss": 1.3257,
      "step": 6044
    },
    {
      "epoch": 0.8659217877094972,
      "grad_norm": 1.062959909439087,
      "learning_rate": 9.28280725569447e-06,
      "loss": 1.4619,
      "step": 6045
    },
    {
      "epoch": 0.8660650336627991,
      "grad_norm": 1.0325846672058105,
      "learning_rate": 9.263294691046297e-06,
      "loss": 1.3487,
      "step": 6046
    },
    {
      "epoch": 0.8662082796161008,
      "grad_norm": 1.0434093475341797,
      "learning_rate": 9.243801659765383e-06,
      "loss": 1.4288,
      "step": 6047
    },
    {
      "epoch": 0.8663515255694026,
      "grad_norm": 0.9925659894943237,
      "learning_rate": 9.224328166048101e-06,
      "loss": 1.4375,
      "step": 6048
    },
    {
      "epoch": 0.8664947715227045,
      "grad_norm": 1.2259210348129272,
      "learning_rate": 9.204874214086611e-06,
      "loss": 1.4381,
      "step": 6049
    },
    {
      "epoch": 0.8666380174760063,
      "grad_norm": 0.8642308712005615,
      "learning_rate": 9.185439808068885e-06,
      "loss": 1.3681,
      "step": 6050
    },
    {
      "epoch": 0.8667812634293082,
      "grad_norm": 0.990006148815155,
      "learning_rate": 9.166024952178665e-06,
      "loss": 1.3661,
      "step": 6051
    },
    {
      "epoch": 0.8669245093826099,
      "grad_norm": 1.1940925121307373,
      "learning_rate": 9.14662965059544e-06,
      "loss": 1.5103,
      "step": 6052
    },
    {
      "epoch": 0.8670677553359117,
      "grad_norm": 1.0651370286941528,
      "learning_rate": 9.127253907494604e-06,
      "loss": 1.3593,
      "step": 6053
    },
    {
      "epoch": 0.8672110012892136,
      "grad_norm": 1.1549527645111084,
      "learning_rate": 9.107897727047276e-06,
      "loss": 1.2538,
      "step": 6054
    },
    {
      "epoch": 0.8673542472425154,
      "grad_norm": 1.012184739112854,
      "learning_rate": 9.088561113420302e-06,
      "loss": 1.4339,
      "step": 6055
    },
    {
      "epoch": 0.8674974931958173,
      "grad_norm": 1.136501669883728,
      "learning_rate": 9.069244070776428e-06,
      "loss": 1.49,
      "step": 6056
    },
    {
      "epoch": 0.867640739149119,
      "grad_norm": 1.0919582843780518,
      "learning_rate": 9.049946603274106e-06,
      "loss": 1.4035,
      "step": 6057
    },
    {
      "epoch": 0.8677839851024208,
      "grad_norm": 1.2973735332489014,
      "learning_rate": 9.030668715067625e-06,
      "loss": 1.3846,
      "step": 6058
    },
    {
      "epoch": 0.8679272310557227,
      "grad_norm": 0.9568747282028198,
      "learning_rate": 9.011410410307019e-06,
      "loss": 1.4862,
      "step": 6059
    },
    {
      "epoch": 0.8680704770090245,
      "grad_norm": 1.1177923679351807,
      "learning_rate": 8.99217169313814e-06,
      "loss": 1.4924,
      "step": 6060
    },
    {
      "epoch": 0.8682137229623264,
      "grad_norm": 1.0631240606307983,
      "learning_rate": 8.972952567702619e-06,
      "loss": 1.5309,
      "step": 6061
    },
    {
      "epoch": 0.8683569689156281,
      "grad_norm": 0.8184772729873657,
      "learning_rate": 8.953753038137825e-06,
      "loss": 1.4991,
      "step": 6062
    },
    {
      "epoch": 0.8685002148689299,
      "grad_norm": 0.9700676202774048,
      "learning_rate": 8.934573108576937e-06,
      "loss": 1.4232,
      "step": 6063
    },
    {
      "epoch": 0.8686434608222318,
      "grad_norm": 1.1821479797363281,
      "learning_rate": 8.915412783148969e-06,
      "loss": 1.4882,
      "step": 6064
    },
    {
      "epoch": 0.8687867067755336,
      "grad_norm": 1.0869390964508057,
      "learning_rate": 8.896272065978628e-06,
      "loss": 1.3837,
      "step": 6065
    },
    {
      "epoch": 0.8689299527288354,
      "grad_norm": 1.1160736083984375,
      "learning_rate": 8.87715096118642e-06,
      "loss": 1.4212,
      "step": 6066
    },
    {
      "epoch": 0.8690731986821373,
      "grad_norm": 1.1133400201797485,
      "learning_rate": 8.858049472888707e-06,
      "loss": 1.3743,
      "step": 6067
    },
    {
      "epoch": 0.869216444635439,
      "grad_norm": 1.0873199701309204,
      "learning_rate": 8.838967605197512e-06,
      "loss": 1.3837,
      "step": 6068
    },
    {
      "epoch": 0.8693596905887409,
      "grad_norm": 1.088273525238037,
      "learning_rate": 8.819905362220704e-06,
      "loss": 1.4743,
      "step": 6069
    },
    {
      "epoch": 0.8695029365420427,
      "grad_norm": 1.0866031646728516,
      "learning_rate": 8.800862748061911e-06,
      "loss": 1.4741,
      "step": 6070
    },
    {
      "epoch": 0.8696461824953445,
      "grad_norm": 1.2942227125167847,
      "learning_rate": 8.781839766820543e-06,
      "loss": 1.4369,
      "step": 6071
    },
    {
      "epoch": 0.8697894284486464,
      "grad_norm": 1.0024195909500122,
      "learning_rate": 8.762836422591802e-06,
      "loss": 1.3551,
      "step": 6072
    },
    {
      "epoch": 0.8699326744019481,
      "grad_norm": 1.078922152519226,
      "learning_rate": 8.743852719466561e-06,
      "loss": 1.2113,
      "step": 6073
    },
    {
      "epoch": 0.8700759203552499,
      "grad_norm": 1.08243727684021,
      "learning_rate": 8.724888661531616e-06,
      "loss": 1.3558,
      "step": 6074
    },
    {
      "epoch": 0.8702191663085518,
      "grad_norm": 1.016839861869812,
      "learning_rate": 8.705944252869446e-06,
      "loss": 1.4482,
      "step": 6075
    },
    {
      "epoch": 0.8703624122618536,
      "grad_norm": 1.0688939094543457,
      "learning_rate": 8.68701949755829e-06,
      "loss": 1.3105,
      "step": 6076
    },
    {
      "epoch": 0.8705056582151555,
      "grad_norm": 1.0628862380981445,
      "learning_rate": 8.668114399672179e-06,
      "loss": 1.3064,
      "step": 6077
    },
    {
      "epoch": 0.8706489041684572,
      "grad_norm": 1.0293161869049072,
      "learning_rate": 8.649228963280931e-06,
      "loss": 1.3684,
      "step": 6078
    },
    {
      "epoch": 0.870792150121759,
      "grad_norm": 1.1159167289733887,
      "learning_rate": 8.630363192450109e-06,
      "loss": 1.305,
      "step": 6079
    },
    {
      "epoch": 0.8709353960750609,
      "grad_norm": 1.0850244760513306,
      "learning_rate": 8.611517091241051e-06,
      "loss": 1.5661,
      "step": 6080
    },
    {
      "epoch": 0.8710786420283627,
      "grad_norm": 1.0197510719299316,
      "learning_rate": 8.592690663710834e-06,
      "loss": 1.3131,
      "step": 6081
    },
    {
      "epoch": 0.8712218879816646,
      "grad_norm": 1.0595096349716187,
      "learning_rate": 8.573883913912361e-06,
      "loss": 1.2878,
      "step": 6082
    },
    {
      "epoch": 0.8713651339349663,
      "grad_norm": 1.305173635482788,
      "learning_rate": 8.555096845894228e-06,
      "loss": 1.4644,
      "step": 6083
    },
    {
      "epoch": 0.8715083798882681,
      "grad_norm": 1.0326265096664429,
      "learning_rate": 8.53632946370081e-06,
      "loss": 1.4134,
      "step": 6084
    },
    {
      "epoch": 0.87165162584157,
      "grad_norm": 1.087653398513794,
      "learning_rate": 8.517581771372318e-06,
      "loss": 1.1652,
      "step": 6085
    },
    {
      "epoch": 0.8717948717948718,
      "grad_norm": 0.9684380888938904,
      "learning_rate": 8.498853772944614e-06,
      "loss": 1.571,
      "step": 6086
    },
    {
      "epoch": 0.8719381177481736,
      "grad_norm": 1.153364658355713,
      "learning_rate": 8.480145472449386e-06,
      "loss": 1.2698,
      "step": 6087
    },
    {
      "epoch": 0.8720813637014755,
      "grad_norm": 1.1119121313095093,
      "learning_rate": 8.461456873914087e-06,
      "loss": 1.5464,
      "step": 6088
    },
    {
      "epoch": 0.8722246096547772,
      "grad_norm": 1.0873825550079346,
      "learning_rate": 8.442787981361877e-06,
      "loss": 1.3236,
      "step": 6089
    },
    {
      "epoch": 0.8723678556080791,
      "grad_norm": 1.0571669340133667,
      "learning_rate": 8.424138798811753e-06,
      "loss": 1.344,
      "step": 6090
    },
    {
      "epoch": 0.8725111015613809,
      "grad_norm": 1.2416207790374756,
      "learning_rate": 8.40550933027836e-06,
      "loss": 1.1794,
      "step": 6091
    },
    {
      "epoch": 0.8726543475146827,
      "grad_norm": 1.1220674514770508,
      "learning_rate": 8.386899579772223e-06,
      "loss": 1.3089,
      "step": 6092
    },
    {
      "epoch": 0.8727975934679846,
      "grad_norm": 1.4793025255203247,
      "learning_rate": 8.368309551299536e-06,
      "loss": 1.3771,
      "step": 6093
    },
    {
      "epoch": 0.8729408394212863,
      "grad_norm": 1.256934404373169,
      "learning_rate": 8.349739248862265e-06,
      "loss": 1.3173,
      "step": 6094
    },
    {
      "epoch": 0.8730840853745881,
      "grad_norm": 1.1983052492141724,
      "learning_rate": 8.331188676458113e-06,
      "loss": 1.4946,
      "step": 6095
    },
    {
      "epoch": 0.87322733132789,
      "grad_norm": 1.003615379333496,
      "learning_rate": 8.312657838080629e-06,
      "loss": 1.4727,
      "step": 6096
    },
    {
      "epoch": 0.8733705772811918,
      "grad_norm": 0.9718660116195679,
      "learning_rate": 8.294146737718988e-06,
      "loss": 1.3032,
      "step": 6097
    },
    {
      "epoch": 0.8735138232344937,
      "grad_norm": 1.0460987091064453,
      "learning_rate": 8.275655379358183e-06,
      "loss": 1.4446,
      "step": 6098
    },
    {
      "epoch": 0.8736570691877954,
      "grad_norm": 1.1340272426605225,
      "learning_rate": 8.257183766978948e-06,
      "loss": 1.5376,
      "step": 6099
    },
    {
      "epoch": 0.8738003151410972,
      "grad_norm": 1.082486867904663,
      "learning_rate": 8.238731904557773e-06,
      "loss": 1.2932,
      "step": 6100
    },
    {
      "epoch": 0.8739435610943991,
      "grad_norm": 1.0665994882583618,
      "learning_rate": 8.220299796066899e-06,
      "loss": 1.2606,
      "step": 6101
    },
    {
      "epoch": 0.8740868070477009,
      "grad_norm": 1.072098970413208,
      "learning_rate": 8.201887445474244e-06,
      "loss": 1.2838,
      "step": 6102
    },
    {
      "epoch": 0.8742300530010028,
      "grad_norm": 1.0588206052780151,
      "learning_rate": 8.18349485674359e-06,
      "loss": 1.4828,
      "step": 6103
    },
    {
      "epoch": 0.8743732989543045,
      "grad_norm": 1.1506743431091309,
      "learning_rate": 8.165122033834393e-06,
      "loss": 1.4904,
      "step": 6104
    },
    {
      "epoch": 0.8745165449076063,
      "grad_norm": 1.2753790616989136,
      "learning_rate": 8.146768980701858e-06,
      "loss": 1.304,
      "step": 6105
    },
    {
      "epoch": 0.8746597908609082,
      "grad_norm": 1.1703521013259888,
      "learning_rate": 8.128435701296943e-06,
      "loss": 1.3782,
      "step": 6106
    },
    {
      "epoch": 0.87480303681421,
      "grad_norm": 1.530335783958435,
      "learning_rate": 8.110122199566362e-06,
      "loss": 1.2855,
      "step": 6107
    },
    {
      "epoch": 0.8749462827675119,
      "grad_norm": 1.3372070789337158,
      "learning_rate": 8.091828479452535e-06,
      "loss": 1.4336,
      "step": 6108
    },
    {
      "epoch": 0.8750895287208137,
      "grad_norm": 1.0343393087387085,
      "learning_rate": 8.073554544893668e-06,
      "loss": 1.3849,
      "step": 6109
    },
    {
      "epoch": 0.8752327746741154,
      "grad_norm": 1.2149019241333008,
      "learning_rate": 8.055300399823684e-06,
      "loss": 1.342,
      "step": 6110
    },
    {
      "epoch": 0.8753760206274173,
      "grad_norm": 1.1708725690841675,
      "learning_rate": 8.03706604817226e-06,
      "loss": 1.2752,
      "step": 6111
    },
    {
      "epoch": 0.8755192665807191,
      "grad_norm": 0.9658020734786987,
      "learning_rate": 8.018851493864777e-06,
      "loss": 1.6042,
      "step": 6112
    },
    {
      "epoch": 0.8756625125340209,
      "grad_norm": 1.0494414567947388,
      "learning_rate": 8.000656740822365e-06,
      "loss": 1.2579,
      "step": 6113
    },
    {
      "epoch": 0.8758057584873228,
      "grad_norm": 1.2507928609848022,
      "learning_rate": 7.98248179296196e-06,
      "loss": 1.3525,
      "step": 6114
    },
    {
      "epoch": 0.8759490044406245,
      "grad_norm": 1.2697725296020508,
      "learning_rate": 7.964326654196119e-06,
      "loss": 1.358,
      "step": 6115
    },
    {
      "epoch": 0.8760922503939264,
      "grad_norm": 1.1196260452270508,
      "learning_rate": 7.946191328433228e-06,
      "loss": 1.374,
      "step": 6116
    },
    {
      "epoch": 0.8762354963472282,
      "grad_norm": 1.3061590194702148,
      "learning_rate": 7.928075819577375e-06,
      "loss": 1.3393,
      "step": 6117
    },
    {
      "epoch": 0.87637874230053,
      "grad_norm": 1.0822381973266602,
      "learning_rate": 7.909980131528361e-06,
      "loss": 1.4585,
      "step": 6118
    },
    {
      "epoch": 0.8765219882538319,
      "grad_norm": 1.1793339252471924,
      "learning_rate": 7.891904268181772e-06,
      "loss": 1.2471,
      "step": 6119
    },
    {
      "epoch": 0.8766652342071336,
      "grad_norm": 1.130366325378418,
      "learning_rate": 7.873848233428826e-06,
      "loss": 1.3817,
      "step": 6120
    },
    {
      "epoch": 0.8768084801604354,
      "grad_norm": 0.9909486174583435,
      "learning_rate": 7.855812031156618e-06,
      "loss": 1.4232,
      "step": 6121
    },
    {
      "epoch": 0.8769517261137373,
      "grad_norm": 1.0860902070999146,
      "learning_rate": 7.837795665247882e-06,
      "loss": 1.4136,
      "step": 6122
    },
    {
      "epoch": 0.8770949720670391,
      "grad_norm": 1.145939588546753,
      "learning_rate": 7.819799139581051e-06,
      "loss": 1.5638,
      "step": 6123
    },
    {
      "epoch": 0.877238218020341,
      "grad_norm": 1.0899732112884521,
      "learning_rate": 7.80182245803035e-06,
      "loss": 1.2457,
      "step": 6124
    },
    {
      "epoch": 0.8773814639736427,
      "grad_norm": 1.195052981376648,
      "learning_rate": 7.783865624465758e-06,
      "loss": 1.4705,
      "step": 6125
    },
    {
      "epoch": 0.8775247099269445,
      "grad_norm": 1.0247740745544434,
      "learning_rate": 7.765928642752884e-06,
      "loss": 1.5351,
      "step": 6126
    },
    {
      "epoch": 0.8776679558802464,
      "grad_norm": 0.9951050281524658,
      "learning_rate": 7.74801151675314e-06,
      "loss": 1.3996,
      "step": 6127
    },
    {
      "epoch": 0.8778112018335482,
      "grad_norm": 0.9022167921066284,
      "learning_rate": 7.730114250323627e-06,
      "loss": 1.4954,
      "step": 6128
    },
    {
      "epoch": 0.8779544477868501,
      "grad_norm": 0.9744040369987488,
      "learning_rate": 7.712236847317188e-06,
      "loss": 1.4274,
      "step": 6129
    },
    {
      "epoch": 0.8780976937401518,
      "grad_norm": 1.1184483766555786,
      "learning_rate": 7.6943793115824e-06,
      "loss": 1.3383,
      "step": 6130
    },
    {
      "epoch": 0.8782409396934536,
      "grad_norm": 1.1000092029571533,
      "learning_rate": 7.67654164696352e-06,
      "loss": 1.5549,
      "step": 6131
    },
    {
      "epoch": 0.8783841856467555,
      "grad_norm": 1.0265151262283325,
      "learning_rate": 7.658723857300599e-06,
      "loss": 1.4362,
      "step": 6132
    },
    {
      "epoch": 0.8785274316000573,
      "grad_norm": 1.0515469312667847,
      "learning_rate": 7.640925946429323e-06,
      "loss": 1.49,
      "step": 6133
    },
    {
      "epoch": 0.8786706775533591,
      "grad_norm": 1.2573654651641846,
      "learning_rate": 7.623147918181139e-06,
      "loss": 1.2438,
      "step": 6134
    },
    {
      "epoch": 0.878813923506661,
      "grad_norm": 1.1230283975601196,
      "learning_rate": 7.605389776383276e-06,
      "loss": 1.5643,
      "step": 6135
    },
    {
      "epoch": 0.8789571694599627,
      "grad_norm": 1.0806825160980225,
      "learning_rate": 7.587651524858564e-06,
      "loss": 1.4948,
      "step": 6136
    },
    {
      "epoch": 0.8791004154132646,
      "grad_norm": 1.056382417678833,
      "learning_rate": 7.569933167425625e-06,
      "loss": 1.4369,
      "step": 6137
    },
    {
      "epoch": 0.8792436613665664,
      "grad_norm": 1.0525827407836914,
      "learning_rate": 7.552234707898787e-06,
      "loss": 1.2996,
      "step": 6138
    },
    {
      "epoch": 0.8793869073198682,
      "grad_norm": 1.069077968597412,
      "learning_rate": 7.53455615008809e-06,
      "loss": 1.4151,
      "step": 6139
    },
    {
      "epoch": 0.8795301532731701,
      "grad_norm": 1.147133469581604,
      "learning_rate": 7.516897497799302e-06,
      "loss": 1.3841,
      "step": 6140
    },
    {
      "epoch": 0.8796733992264718,
      "grad_norm": 0.9963047504425049,
      "learning_rate": 7.4992587548338485e-06,
      "loss": 1.3876,
      "step": 6141
    },
    {
      "epoch": 0.8798166451797736,
      "grad_norm": 1.0968685150146484,
      "learning_rate": 7.48163992498897e-06,
      "loss": 1.4206,
      "step": 6142
    },
    {
      "epoch": 0.8799598911330755,
      "grad_norm": 1.0552432537078857,
      "learning_rate": 7.464041012057554e-06,
      "loss": 1.3697,
      "step": 6143
    },
    {
      "epoch": 0.8801031370863773,
      "grad_norm": 1.0118249654769897,
      "learning_rate": 7.446462019828182e-06,
      "loss": 1.2755,
      "step": 6144
    },
    {
      "epoch": 0.8802463830396792,
      "grad_norm": 1.1240556240081787,
      "learning_rate": 7.428902952085171e-06,
      "loss": 1.2693,
      "step": 6145
    },
    {
      "epoch": 0.8803896289929809,
      "grad_norm": 1.1441346406936646,
      "learning_rate": 7.411363812608618e-06,
      "loss": 1.3218,
      "step": 6146
    },
    {
      "epoch": 0.8805328749462827,
      "grad_norm": 1.3975311517715454,
      "learning_rate": 7.393844605174205e-06,
      "loss": 1.2018,
      "step": 6147
    },
    {
      "epoch": 0.8806761208995846,
      "grad_norm": 0.9469305276870728,
      "learning_rate": 7.376345333553403e-06,
      "loss": 1.365,
      "step": 6148
    },
    {
      "epoch": 0.8808193668528864,
      "grad_norm": 1.0788648128509521,
      "learning_rate": 7.358866001513364e-06,
      "loss": 1.3918,
      "step": 6149
    },
    {
      "epoch": 0.8809626128061883,
      "grad_norm": 1.1089503765106201,
      "learning_rate": 7.3414066128169705e-06,
      "loss": 1.4112,
      "step": 6150
    },
    {
      "epoch": 0.88110585875949,
      "grad_norm": 1.0184285640716553,
      "learning_rate": 7.323967171222801e-06,
      "loss": 1.3326,
      "step": 6151
    },
    {
      "epoch": 0.8812491047127918,
      "grad_norm": 1.087239146232605,
      "learning_rate": 7.306547680485088e-06,
      "loss": 1.4886,
      "step": 6152
    },
    {
      "epoch": 0.8813923506660937,
      "grad_norm": 1.04493248462677,
      "learning_rate": 7.289148144353875e-06,
      "loss": 1.3661,
      "step": 6153
    },
    {
      "epoch": 0.8815355966193955,
      "grad_norm": 0.9697790741920471,
      "learning_rate": 7.271768566574843e-06,
      "loss": 1.316,
      "step": 6154
    },
    {
      "epoch": 0.8816788425726974,
      "grad_norm": 1.4127305746078491,
      "learning_rate": 7.2544089508893575e-06,
      "loss": 1.3447,
      "step": 6155
    },
    {
      "epoch": 0.8818220885259992,
      "grad_norm": 1.0601903200149536,
      "learning_rate": 7.237069301034516e-06,
      "loss": 1.4913,
      "step": 6156
    },
    {
      "epoch": 0.8819653344793009,
      "grad_norm": 1.2036266326904297,
      "learning_rate": 7.219749620743144e-06,
      "loss": 1.2353,
      "step": 6157
    },
    {
      "epoch": 0.8821085804326028,
      "grad_norm": 1.3706302642822266,
      "learning_rate": 7.202449913743714e-06,
      "loss": 1.4462,
      "step": 6158
    },
    {
      "epoch": 0.8822518263859046,
      "grad_norm": 1.033921241760254,
      "learning_rate": 7.185170183760437e-06,
      "loss": 1.3452,
      "step": 6159
    },
    {
      "epoch": 0.8823950723392064,
      "grad_norm": 0.9142951965332031,
      "learning_rate": 7.167910434513214e-06,
      "loss": 1.3113,
      "step": 6160
    },
    {
      "epoch": 0.8825383182925083,
      "grad_norm": 1.0199739933013916,
      "learning_rate": 7.15067066971764e-06,
      "loss": 1.5048,
      "step": 6161
    },
    {
      "epoch": 0.88268156424581,
      "grad_norm": 1.2085869312286377,
      "learning_rate": 7.133450893085003e-06,
      "loss": 1.476,
      "step": 6162
    },
    {
      "epoch": 0.8828248101991119,
      "grad_norm": 1.1762008666992188,
      "learning_rate": 7.116251108322281e-06,
      "loss": 1.434,
      "step": 6163
    },
    {
      "epoch": 0.8829680561524137,
      "grad_norm": 1.0077625513076782,
      "learning_rate": 7.099071319132211e-06,
      "loss": 1.4332,
      "step": 6164
    },
    {
      "epoch": 0.8831113021057155,
      "grad_norm": 1.0314888954162598,
      "learning_rate": 7.081911529213126e-06,
      "loss": 1.2855,
      "step": 6165
    },
    {
      "epoch": 0.8832545480590174,
      "grad_norm": 1.039973497390747,
      "learning_rate": 7.064771742259124e-06,
      "loss": 1.5249,
      "step": 6166
    },
    {
      "epoch": 0.8833977940123191,
      "grad_norm": 1.0815176963806152,
      "learning_rate": 7.047651961959978e-06,
      "loss": 1.4006,
      "step": 6167
    },
    {
      "epoch": 0.883541039965621,
      "grad_norm": 1.2100239992141724,
      "learning_rate": 7.03055219200115e-06,
      "loss": 1.3623,
      "step": 6168
    },
    {
      "epoch": 0.8836842859189228,
      "grad_norm": 1.2314960956573486,
      "learning_rate": 7.013472436063817e-06,
      "loss": 1.271,
      "step": 6169
    },
    {
      "epoch": 0.8838275318722246,
      "grad_norm": 1.0538991689682007,
      "learning_rate": 6.996412697824772e-06,
      "loss": 1.3367,
      "step": 6170
    },
    {
      "epoch": 0.8839707778255265,
      "grad_norm": 1.0745586156845093,
      "learning_rate": 6.979372980956611e-06,
      "loss": 1.3359,
      "step": 6171
    },
    {
      "epoch": 0.8841140237788282,
      "grad_norm": 1.1423468589782715,
      "learning_rate": 6.962353289127555e-06,
      "loss": 1.2749,
      "step": 6172
    },
    {
      "epoch": 0.88425726973213,
      "grad_norm": 1.3436617851257324,
      "learning_rate": 6.945353626001494e-06,
      "loss": 1.3652,
      "step": 6173
    },
    {
      "epoch": 0.8844005156854319,
      "grad_norm": 1.0537688732147217,
      "learning_rate": 6.928373995238047e-06,
      "loss": 1.4774,
      "step": 6174
    },
    {
      "epoch": 0.8845437616387337,
      "grad_norm": 0.9605186581611633,
      "learning_rate": 6.911414400492544e-06,
      "loss": 1.4939,
      "step": 6175
    },
    {
      "epoch": 0.8846870075920356,
      "grad_norm": 1.0892926454544067,
      "learning_rate": 6.89447484541591e-06,
      "loss": 1.5917,
      "step": 6176
    },
    {
      "epoch": 0.8848302535453374,
      "grad_norm": 1.176377534866333,
      "learning_rate": 6.87755533365485e-06,
      "loss": 1.5501,
      "step": 6177
    },
    {
      "epoch": 0.8849734994986391,
      "grad_norm": 1.0803096294403076,
      "learning_rate": 6.860655868851696e-06,
      "loss": 1.4181,
      "step": 6178
    },
    {
      "epoch": 0.885116745451941,
      "grad_norm": 1.1753509044647217,
      "learning_rate": 6.84377645464449e-06,
      "loss": 1.333,
      "step": 6179
    },
    {
      "epoch": 0.8852599914052428,
      "grad_norm": 1.0639952421188354,
      "learning_rate": 6.826917094666973e-06,
      "loss": 1.3638,
      "step": 6180
    },
    {
      "epoch": 0.8854032373585446,
      "grad_norm": 1.0984638929367065,
      "learning_rate": 6.810077792548508e-06,
      "loss": 1.3114,
      "step": 6181
    },
    {
      "epoch": 0.8855464833118465,
      "grad_norm": 1.081886649131775,
      "learning_rate": 6.793258551914206e-06,
      "loss": 1.4676,
      "step": 6182
    },
    {
      "epoch": 0.8856897292651482,
      "grad_norm": 1.2206023931503296,
      "learning_rate": 6.776459376384847e-06,
      "loss": 1.3366,
      "step": 6183
    },
    {
      "epoch": 0.8858329752184501,
      "grad_norm": 1.1014976501464844,
      "learning_rate": 6.759680269576852e-06,
      "loss": 1.5732,
      "step": 6184
    },
    {
      "epoch": 0.8859762211717519,
      "grad_norm": 1.1496285200119019,
      "learning_rate": 6.74292123510234e-06,
      "loss": 1.3276,
      "step": 6185
    },
    {
      "epoch": 0.8861194671250537,
      "grad_norm": 1.2254159450531006,
      "learning_rate": 6.726182276569148e-06,
      "loss": 1.386,
      "step": 6186
    },
    {
      "epoch": 0.8862627130783556,
      "grad_norm": 1.1976909637451172,
      "learning_rate": 6.709463397580728e-06,
      "loss": 1.3907,
      "step": 6187
    },
    {
      "epoch": 0.8864059590316573,
      "grad_norm": 1.1922142505645752,
      "learning_rate": 6.692764601736268e-06,
      "loss": 1.3916,
      "step": 6188
    },
    {
      "epoch": 0.8865492049849591,
      "grad_norm": 0.9329336285591125,
      "learning_rate": 6.67608589263059e-06,
      "loss": 1.4402,
      "step": 6189
    },
    {
      "epoch": 0.886692450938261,
      "grad_norm": 1.071075677871704,
      "learning_rate": 6.659427273854224e-06,
      "loss": 1.4704,
      "step": 6190
    },
    {
      "epoch": 0.8868356968915628,
      "grad_norm": 0.9965466260910034,
      "learning_rate": 6.642788748993323e-06,
      "loss": 1.3147,
      "step": 6191
    },
    {
      "epoch": 0.8869789428448647,
      "grad_norm": 0.9360291957855225,
      "learning_rate": 6.626170321629776e-06,
      "loss": 1.5308,
      "step": 6192
    },
    {
      "epoch": 0.8871221887981664,
      "grad_norm": 1.0450764894485474,
      "learning_rate": 6.609571995341135e-06,
      "loss": 1.3125,
      "step": 6193
    },
    {
      "epoch": 0.8872654347514682,
      "grad_norm": 1.097119688987732,
      "learning_rate": 6.592993773700573e-06,
      "loss": 1.4344,
      "step": 6194
    },
    {
      "epoch": 0.8874086807047701,
      "grad_norm": 1.11098051071167,
      "learning_rate": 6.576435660276969e-06,
      "loss": 1.3367,
      "step": 6195
    },
    {
      "epoch": 0.8875519266580719,
      "grad_norm": 1.2049585580825806,
      "learning_rate": 6.559897658634928e-06,
      "loss": 1.3774,
      "step": 6196
    },
    {
      "epoch": 0.8876951726113738,
      "grad_norm": 0.9086595177650452,
      "learning_rate": 6.5433797723346235e-06,
      "loss": 1.469,
      "step": 6197
    },
    {
      "epoch": 0.8878384185646756,
      "grad_norm": 1.185227632522583,
      "learning_rate": 6.526882004931967e-06,
      "loss": 1.3401,
      "step": 6198
    },
    {
      "epoch": 0.8879816645179773,
      "grad_norm": 1.064042329788208,
      "learning_rate": 6.510404359978506e-06,
      "loss": 1.3726,
      "step": 6199
    },
    {
      "epoch": 0.8881249104712792,
      "grad_norm": 1.0553909540176392,
      "learning_rate": 6.493946841021481e-06,
      "loss": 1.4104,
      "step": 6200
    },
    {
      "epoch": 0.888268156424581,
      "grad_norm": 0.935221254825592,
      "learning_rate": 6.477509451603791e-06,
      "loss": 1.4899,
      "step": 6201
    },
    {
      "epoch": 0.8884114023778829,
      "grad_norm": 0.9904655814170837,
      "learning_rate": 6.461092195263973e-06,
      "loss": 1.5958,
      "step": 6202
    },
    {
      "epoch": 0.8885546483311847,
      "grad_norm": 1.0689826011657715,
      "learning_rate": 6.444695075536289e-06,
      "loss": 1.3991,
      "step": 6203
    },
    {
      "epoch": 0.8886978942844864,
      "grad_norm": 1.0686777830123901,
      "learning_rate": 6.428318095950647e-06,
      "loss": 1.4976,
      "step": 6204
    },
    {
      "epoch": 0.8888411402377883,
      "grad_norm": 0.9992170929908752,
      "learning_rate": 6.411961260032551e-06,
      "loss": 1.4536,
      "step": 6205
    },
    {
      "epoch": 0.8889843861910901,
      "grad_norm": 0.9994930028915405,
      "learning_rate": 6.39562457130326e-06,
      "loss": 1.4394,
      "step": 6206
    },
    {
      "epoch": 0.8891276321443919,
      "grad_norm": 1.087852120399475,
      "learning_rate": 6.379308033279641e-06,
      "loss": 1.4414,
      "step": 6207
    },
    {
      "epoch": 0.8892708780976938,
      "grad_norm": 1.007928729057312,
      "learning_rate": 6.363011649474249e-06,
      "loss": 1.426,
      "step": 6208
    },
    {
      "epoch": 0.8894141240509955,
      "grad_norm": 1.0900826454162598,
      "learning_rate": 6.3467354233953e-06,
      "loss": 1.2698,
      "step": 6209
    },
    {
      "epoch": 0.8895573700042974,
      "grad_norm": 1.2544763088226318,
      "learning_rate": 6.3304793585466595e-06,
      "loss": 1.3357,
      "step": 6210
    },
    {
      "epoch": 0.8897006159575992,
      "grad_norm": 1.0173091888427734,
      "learning_rate": 6.3142434584278596e-06,
      "loss": 1.3894,
      "step": 6211
    },
    {
      "epoch": 0.889843861910901,
      "grad_norm": 1.051266074180603,
      "learning_rate": 6.2980277265340726e-06,
      "loss": 1.3084,
      "step": 6212
    },
    {
      "epoch": 0.8899871078642029,
      "grad_norm": 1.135063886642456,
      "learning_rate": 6.281832166356127e-06,
      "loss": 1.4347,
      "step": 6213
    },
    {
      "epoch": 0.8901303538175046,
      "grad_norm": 1.0632984638214111,
      "learning_rate": 6.265656781380591e-06,
      "loss": 1.3738,
      "step": 6214
    },
    {
      "epoch": 0.8902735997708064,
      "grad_norm": 0.927520751953125,
      "learning_rate": 6.249501575089567e-06,
      "loss": 1.3379,
      "step": 6215
    },
    {
      "epoch": 0.8904168457241083,
      "grad_norm": 1.1920615434646606,
      "learning_rate": 6.233366550960884e-06,
      "loss": 1.2079,
      "step": 6216
    },
    {
      "epoch": 0.8905600916774101,
      "grad_norm": 1.1692345142364502,
      "learning_rate": 6.2172517124680305e-06,
      "loss": 1.4754,
      "step": 6217
    },
    {
      "epoch": 0.890703337630712,
      "grad_norm": 1.0433944463729858,
      "learning_rate": 6.20115706308011e-06,
      "loss": 1.352,
      "step": 6218
    },
    {
      "epoch": 0.8908465835840138,
      "grad_norm": 0.9226456880569458,
      "learning_rate": 6.185082606261927e-06,
      "loss": 1.4443,
      "step": 6219
    },
    {
      "epoch": 0.8909898295373155,
      "grad_norm": 1.0460840463638306,
      "learning_rate": 6.169028345473881e-06,
      "loss": 1.446,
      "step": 6220
    },
    {
      "epoch": 0.8911330754906174,
      "grad_norm": 1.1735236644744873,
      "learning_rate": 6.152994284172076e-06,
      "loss": 1.2339,
      "step": 6221
    },
    {
      "epoch": 0.8912763214439192,
      "grad_norm": 1.2027605772018433,
      "learning_rate": 6.136980425808281e-06,
      "loss": 1.3602,
      "step": 6222
    },
    {
      "epoch": 0.8914195673972211,
      "grad_norm": 1.135356068611145,
      "learning_rate": 6.1209867738298286e-06,
      "loss": 1.3209,
      "step": 6223
    },
    {
      "epoch": 0.8915628133505229,
      "grad_norm": 1.140559434890747,
      "learning_rate": 6.105013331679754e-06,
      "loss": 1.4739,
      "step": 6224
    },
    {
      "epoch": 0.8917060593038246,
      "grad_norm": 1.170394778251648,
      "learning_rate": 6.089060102796807e-06,
      "loss": 1.3255,
      "step": 6225
    },
    {
      "epoch": 0.8918493052571265,
      "grad_norm": 1.0642247200012207,
      "learning_rate": 6.073127090615271e-06,
      "loss": 1.3135,
      "step": 6226
    },
    {
      "epoch": 0.8919925512104283,
      "grad_norm": 1.0005143880844116,
      "learning_rate": 6.057214298565151e-06,
      "loss": 1.4081,
      "step": 6227
    },
    {
      "epoch": 0.8921357971637301,
      "grad_norm": 1.0217949151992798,
      "learning_rate": 6.041321730072058e-06,
      "loss": 1.3107,
      "step": 6228
    },
    {
      "epoch": 0.892279043117032,
      "grad_norm": 1.41926908493042,
      "learning_rate": 6.025449388557281e-06,
      "loss": 1.3773,
      "step": 6229
    },
    {
      "epoch": 0.8924222890703337,
      "grad_norm": 1.2625526189804077,
      "learning_rate": 6.009597277437762e-06,
      "loss": 1.4789,
      "step": 6230
    },
    {
      "epoch": 0.8925655350236356,
      "grad_norm": 1.2700263261795044,
      "learning_rate": 5.9937654001260164e-06,
      "loss": 1.2987,
      "step": 6231
    },
    {
      "epoch": 0.8927087809769374,
      "grad_norm": 1.1416374444961548,
      "learning_rate": 5.977953760030297e-06,
      "loss": 1.2169,
      "step": 6232
    },
    {
      "epoch": 0.8928520269302392,
      "grad_norm": 1.0866599082946777,
      "learning_rate": 5.962162360554469e-06,
      "loss": 1.2683,
      "step": 6233
    },
    {
      "epoch": 0.8929952728835411,
      "grad_norm": 1.0845009088516235,
      "learning_rate": 5.9463912050979895e-06,
      "loss": 1.503,
      "step": 6234
    },
    {
      "epoch": 0.8931385188368428,
      "grad_norm": 1.125756025314331,
      "learning_rate": 5.930640297056022e-06,
      "loss": 1.4683,
      "step": 6235
    },
    {
      "epoch": 0.8932817647901447,
      "grad_norm": 1.233464241027832,
      "learning_rate": 5.914909639819344e-06,
      "loss": 1.3926,
      "step": 6236
    },
    {
      "epoch": 0.8934250107434465,
      "grad_norm": 1.1848875284194946,
      "learning_rate": 5.899199236774377e-06,
      "loss": 1.4587,
      "step": 6237
    },
    {
      "epoch": 0.8935682566967483,
      "grad_norm": 1.0152925252914429,
      "learning_rate": 5.883509091303174e-06,
      "loss": 1.4759,
      "step": 6238
    },
    {
      "epoch": 0.8937115026500502,
      "grad_norm": 0.9698759317398071,
      "learning_rate": 5.867839206783454e-06,
      "loss": 1.4933,
      "step": 6239
    },
    {
      "epoch": 0.8938547486033519,
      "grad_norm": 1.1864376068115234,
      "learning_rate": 5.852189586588552e-06,
      "loss": 1.2575,
      "step": 6240
    },
    {
      "epoch": 0.8939979945566537,
      "grad_norm": 1.143431544303894,
      "learning_rate": 5.836560234087418e-06,
      "loss": 1.441,
      "step": 6241
    },
    {
      "epoch": 0.8941412405099556,
      "grad_norm": 1.1181886196136475,
      "learning_rate": 5.820951152644671e-06,
      "loss": 1.2594,
      "step": 6242
    },
    {
      "epoch": 0.8942844864632574,
      "grad_norm": 1.0237746238708496,
      "learning_rate": 5.805362345620602e-06,
      "loss": 1.2846,
      "step": 6243
    },
    {
      "epoch": 0.8944277324165593,
      "grad_norm": 1.157672643661499,
      "learning_rate": 5.7897938163710365e-06,
      "loss": 1.2023,
      "step": 6244
    },
    {
      "epoch": 0.8945709783698611,
      "grad_norm": 0.9851744771003723,
      "learning_rate": 5.7742455682475384e-06,
      "loss": 1.1994,
      "step": 6245
    },
    {
      "epoch": 0.8947142243231628,
      "grad_norm": 1.0079066753387451,
      "learning_rate": 5.758717604597242e-06,
      "loss": 1.5645,
      "step": 6246
    },
    {
      "epoch": 0.8948574702764647,
      "grad_norm": 1.1544268131256104,
      "learning_rate": 5.743209928762927e-06,
      "loss": 1.3554,
      "step": 6247
    },
    {
      "epoch": 0.8950007162297665,
      "grad_norm": 1.0780153274536133,
      "learning_rate": 5.727722544083036e-06,
      "loss": 1.3706,
      "step": 6248
    },
    {
      "epoch": 0.8951439621830684,
      "grad_norm": 1.1990412473678589,
      "learning_rate": 5.71225545389158e-06,
      "loss": 1.2863,
      "step": 6249
    },
    {
      "epoch": 0.8952872081363702,
      "grad_norm": 1.0910295248031616,
      "learning_rate": 5.696808661518283e-06,
      "loss": 1.4708,
      "step": 6250
    },
    {
      "epoch": 0.8954304540896719,
      "grad_norm": 0.9389721155166626,
      "learning_rate": 5.681382170288452e-06,
      "loss": 1.3056,
      "step": 6251
    },
    {
      "epoch": 0.8955737000429738,
      "grad_norm": 1.16049063205719,
      "learning_rate": 5.665975983522997e-06,
      "loss": 1.4097,
      "step": 6252
    },
    {
      "epoch": 0.8957169459962756,
      "grad_norm": 1.180406928062439,
      "learning_rate": 5.6505901045385e-06,
      "loss": 1.4388,
      "step": 6253
    },
    {
      "epoch": 0.8958601919495774,
      "grad_norm": 1.3661245107650757,
      "learning_rate": 5.635224536647188e-06,
      "loss": 1.5029,
      "step": 6254
    },
    {
      "epoch": 0.8960034379028793,
      "grad_norm": 1.1520068645477295,
      "learning_rate": 5.619879283156859e-06,
      "loss": 1.2789,
      "step": 6255
    },
    {
      "epoch": 0.896146683856181,
      "grad_norm": 1.054107904434204,
      "learning_rate": 5.604554347370983e-06,
      "loss": 1.3756,
      "step": 6256
    },
    {
      "epoch": 0.8962899298094829,
      "grad_norm": 1.1637687683105469,
      "learning_rate": 5.5892497325886195e-06,
      "loss": 1.4209,
      "step": 6257
    },
    {
      "epoch": 0.8964331757627847,
      "grad_norm": 1.2004553079605103,
      "learning_rate": 5.573965442104489e-06,
      "loss": 1.4547,
      "step": 6258
    },
    {
      "epoch": 0.8965764217160865,
      "grad_norm": 1.2279695272445679,
      "learning_rate": 5.558701479208916e-06,
      "loss": 1.3911,
      "step": 6259
    },
    {
      "epoch": 0.8967196676693884,
      "grad_norm": 1.06075918674469,
      "learning_rate": 5.5434578471878605e-06,
      "loss": 1.4286,
      "step": 6260
    },
    {
      "epoch": 0.8968629136226901,
      "grad_norm": 1.3761261701583862,
      "learning_rate": 5.528234549322908e-06,
      "loss": 1.4348,
      "step": 6261
    },
    {
      "epoch": 0.897006159575992,
      "grad_norm": 1.0343631505966187,
      "learning_rate": 5.513031588891226e-06,
      "loss": 1.2258,
      "step": 6262
    },
    {
      "epoch": 0.8971494055292938,
      "grad_norm": 1.0171232223510742,
      "learning_rate": 5.497848969165631e-06,
      "loss": 1.5321,
      "step": 6263
    },
    {
      "epoch": 0.8972926514825956,
      "grad_norm": 1.214233636856079,
      "learning_rate": 5.482686693414629e-06,
      "loss": 1.266,
      "step": 6264
    },
    {
      "epoch": 0.8974358974358975,
      "grad_norm": 1.2685524225234985,
      "learning_rate": 5.467544764902221e-06,
      "loss": 1.3513,
      "step": 6265
    },
    {
      "epoch": 0.8975791433891993,
      "grad_norm": 1.0766239166259766,
      "learning_rate": 5.452423186888111e-06,
      "loss": 1.3664,
      "step": 6266
    },
    {
      "epoch": 0.897722389342501,
      "grad_norm": 1.140155553817749,
      "learning_rate": 5.437321962627606e-06,
      "loss": 1.2976,
      "step": 6267
    },
    {
      "epoch": 0.8978656352958029,
      "grad_norm": 1.0092066526412964,
      "learning_rate": 5.422241095371605e-06,
      "loss": 1.2941,
      "step": 6268
    },
    {
      "epoch": 0.8980088812491047,
      "grad_norm": 1.015629768371582,
      "learning_rate": 5.407180588366689e-06,
      "loss": 1.4672,
      "step": 6269
    },
    {
      "epoch": 0.8981521272024066,
      "grad_norm": 1.1754333972930908,
      "learning_rate": 5.392140444854954e-06,
      "loss": 1.4973,
      "step": 6270
    },
    {
      "epoch": 0.8982953731557084,
      "grad_norm": 1.0000135898590088,
      "learning_rate": 5.377120668074209e-06,
      "loss": 1.3792,
      "step": 6271
    },
    {
      "epoch": 0.8984386191090101,
      "grad_norm": 1.2560063600540161,
      "learning_rate": 5.362121261257847e-06,
      "loss": 1.5642,
      "step": 6272
    },
    {
      "epoch": 0.898581865062312,
      "grad_norm": 1.1916520595550537,
      "learning_rate": 5.3471422276348385e-06,
      "loss": 1.4755,
      "step": 6273
    },
    {
      "epoch": 0.8987251110156138,
      "grad_norm": 1.1774929761886597,
      "learning_rate": 5.332183570429794e-06,
      "loss": 1.3062,
      "step": 6274
    },
    {
      "epoch": 0.8988683569689156,
      "grad_norm": 1.1466996669769287,
      "learning_rate": 5.317245292862994e-06,
      "loss": 1.5093,
      "step": 6275
    },
    {
      "epoch": 0.8990116029222175,
      "grad_norm": 1.0106556415557861,
      "learning_rate": 5.302327398150242e-06,
      "loss": 1.4178,
      "step": 6276
    },
    {
      "epoch": 0.8991548488755192,
      "grad_norm": 1.1131858825683594,
      "learning_rate": 5.287429889502993e-06,
      "loss": 1.5591,
      "step": 6277
    },
    {
      "epoch": 0.8992980948288211,
      "grad_norm": 1.1433442831039429,
      "learning_rate": 5.272552770128314e-06,
      "loss": 1.3623,
      "step": 6278
    },
    {
      "epoch": 0.8994413407821229,
      "grad_norm": 1.1273397207260132,
      "learning_rate": 5.257696043228888e-06,
      "loss": 1.5303,
      "step": 6279
    },
    {
      "epoch": 0.8995845867354247,
      "grad_norm": 1.0660606622695923,
      "learning_rate": 5.242859712003001e-06,
      "loss": 1.333,
      "step": 6280
    },
    {
      "epoch": 0.8997278326887266,
      "grad_norm": 1.046756386756897,
      "learning_rate": 5.22804377964452e-06,
      "loss": 1.3777,
      "step": 6281
    },
    {
      "epoch": 0.8998710786420283,
      "grad_norm": 1.1891934871673584,
      "learning_rate": 5.2132482493429835e-06,
      "loss": 1.5024,
      "step": 6282
    },
    {
      "epoch": 0.9000143245953302,
      "grad_norm": 1.0168426036834717,
      "learning_rate": 5.198473124283509e-06,
      "loss": 1.5215,
      "step": 6283
    },
    {
      "epoch": 0.900157570548632,
      "grad_norm": 1.2852948904037476,
      "learning_rate": 5.183718407646787e-06,
      "loss": 1.3111,
      "step": 6284
    },
    {
      "epoch": 0.9003008165019338,
      "grad_norm": 1.2262461185455322,
      "learning_rate": 5.168984102609142e-06,
      "loss": 1.3744,
      "step": 6285
    },
    {
      "epoch": 0.9004440624552357,
      "grad_norm": 0.970107913017273,
      "learning_rate": 5.1542702123425264e-06,
      "loss": 1.5221,
      "step": 6286
    },
    {
      "epoch": 0.9005873084085375,
      "grad_norm": 0.9678454995155334,
      "learning_rate": 5.139576740014462e-06,
      "loss": 1.4439,
      "step": 6287
    },
    {
      "epoch": 0.9007305543618392,
      "grad_norm": 1.179451823234558,
      "learning_rate": 5.124903688788096e-06,
      "loss": 1.4788,
      "step": 6288
    },
    {
      "epoch": 0.9008738003151411,
      "grad_norm": 0.9689302444458008,
      "learning_rate": 5.1102510618221686e-06,
      "loss": 1.4789,
      "step": 6289
    },
    {
      "epoch": 0.9010170462684429,
      "grad_norm": 1.1082656383514404,
      "learning_rate": 5.0956188622710455e-06,
      "loss": 1.2672,
      "step": 6290
    },
    {
      "epoch": 0.9011602922217448,
      "grad_norm": 0.9578465819358826,
      "learning_rate": 5.081007093284651e-06,
      "loss": 1.3596,
      "step": 6291
    },
    {
      "epoch": 0.9013035381750466,
      "grad_norm": 0.9995458722114563,
      "learning_rate": 5.066415758008536e-06,
      "loss": 1.4438,
      "step": 6292
    },
    {
      "epoch": 0.9014467841283483,
      "grad_norm": 1.2014294862747192,
      "learning_rate": 5.051844859583888e-06,
      "loss": 1.3551,
      "step": 6293
    },
    {
      "epoch": 0.9015900300816502,
      "grad_norm": 1.06302011013031,
      "learning_rate": 5.03729440114743e-06,
      "loss": 1.217,
      "step": 6294
    },
    {
      "epoch": 0.901733276034952,
      "grad_norm": 1.2783023118972778,
      "learning_rate": 5.022764385831524e-06,
      "loss": 1.4079,
      "step": 6295
    },
    {
      "epoch": 0.9018765219882539,
      "grad_norm": 1.0324749946594238,
      "learning_rate": 5.008254816764124e-06,
      "loss": 1.5073,
      "step": 6296
    },
    {
      "epoch": 0.9020197679415557,
      "grad_norm": 1.0042519569396973,
      "learning_rate": 4.993765697068787e-06,
      "loss": 1.5058,
      "step": 6297
    },
    {
      "epoch": 0.9021630138948574,
      "grad_norm": 1.0395405292510986,
      "learning_rate": 4.979297029864672e-06,
      "loss": 1.4014,
      "step": 6298
    },
    {
      "epoch": 0.9023062598481593,
      "grad_norm": 0.962011456489563,
      "learning_rate": 4.964848818266466e-06,
      "loss": 1.3273,
      "step": 6299
    },
    {
      "epoch": 0.9024495058014611,
      "grad_norm": 1.288873314857483,
      "learning_rate": 4.950421065384581e-06,
      "loss": 1.2289,
      "step": 6300
    },
    {
      "epoch": 0.902592751754763,
      "grad_norm": 1.252167820930481,
      "learning_rate": 4.936013774324944e-06,
      "loss": 1.4425,
      "step": 6301
    },
    {
      "epoch": 0.9027359977080648,
      "grad_norm": 1.2269313335418701,
      "learning_rate": 4.9216269481890505e-06,
      "loss": 1.4974,
      "step": 6302
    },
    {
      "epoch": 0.9028792436613665,
      "grad_norm": 1.2080711126327515,
      "learning_rate": 4.907260590074048e-06,
      "loss": 1.3452,
      "step": 6303
    },
    {
      "epoch": 0.9030224896146684,
      "grad_norm": 1.1367989778518677,
      "learning_rate": 4.892914703072671e-06,
      "loss": 1.3002,
      "step": 6304
    },
    {
      "epoch": 0.9031657355679702,
      "grad_norm": 1.2795456647872925,
      "learning_rate": 4.8785892902732275e-06,
      "loss": 1.2931,
      "step": 6305
    },
    {
      "epoch": 0.903308981521272,
      "grad_norm": 1.1423330307006836,
      "learning_rate": 4.864284354759607e-06,
      "loss": 1.352,
      "step": 6306
    },
    {
      "epoch": 0.9034522274745739,
      "grad_norm": 1.1512235403060913,
      "learning_rate": 4.849999899611324e-06,
      "loss": 1.244,
      "step": 6307
    },
    {
      "epoch": 0.9035954734278757,
      "grad_norm": 1.193757176399231,
      "learning_rate": 4.835735927903473e-06,
      "loss": 1.2428,
      "step": 6308
    },
    {
      "epoch": 0.9037387193811774,
      "grad_norm": 1.1099021434783936,
      "learning_rate": 4.821492442706732e-06,
      "loss": 1.4622,
      "step": 6309
    },
    {
      "epoch": 0.9038819653344793,
      "grad_norm": 0.9713549613952637,
      "learning_rate": 4.807269447087348e-06,
      "loss": 1.3404,
      "step": 6310
    },
    {
      "epoch": 0.9040252112877811,
      "grad_norm": 1.155503273010254,
      "learning_rate": 4.793066944107205e-06,
      "loss": 1.4044,
      "step": 6311
    },
    {
      "epoch": 0.904168457241083,
      "grad_norm": 1.0077558755874634,
      "learning_rate": 4.778884936823758e-06,
      "loss": 1.5516,
      "step": 6312
    },
    {
      "epoch": 0.9043117031943848,
      "grad_norm": 1.0336101055145264,
      "learning_rate": 4.764723428290019e-06,
      "loss": 1.3172,
      "step": 6313
    },
    {
      "epoch": 0.9044549491476865,
      "grad_norm": 1.1511133909225464,
      "learning_rate": 4.750582421554605e-06,
      "loss": 1.4852,
      "step": 6314
    },
    {
      "epoch": 0.9045981951009884,
      "grad_norm": 1.1120917797088623,
      "learning_rate": 4.7364619196617495e-06,
      "loss": 1.3315,
      "step": 6315
    },
    {
      "epoch": 0.9047414410542902,
      "grad_norm": 1.1380597352981567,
      "learning_rate": 4.722361925651231e-06,
      "loss": 1.4127,
      "step": 6316
    },
    {
      "epoch": 0.9048846870075921,
      "grad_norm": 0.8841583728790283,
      "learning_rate": 4.708282442558443e-06,
      "loss": 1.3319,
      "step": 6317
    },
    {
      "epoch": 0.9050279329608939,
      "grad_norm": 1.1729316711425781,
      "learning_rate": 4.694223473414328e-06,
      "loss": 1.2857,
      "step": 6318
    },
    {
      "epoch": 0.9051711789141956,
      "grad_norm": 0.9037086963653564,
      "learning_rate": 4.6801850212454755e-06,
      "loss": 1.3834,
      "step": 6319
    },
    {
      "epoch": 0.9053144248674975,
      "grad_norm": 1.21295166015625,
      "learning_rate": 4.6661670890739475e-06,
      "loss": 1.4177,
      "step": 6320
    },
    {
      "epoch": 0.9054576708207993,
      "grad_norm": 1.466363549232483,
      "learning_rate": 4.652169679917517e-06,
      "loss": 1.4996,
      "step": 6321
    },
    {
      "epoch": 0.9056009167741012,
      "grad_norm": 1.1179686784744263,
      "learning_rate": 4.638192796789487e-06,
      "loss": 1.594,
      "step": 6322
    },
    {
      "epoch": 0.905744162727403,
      "grad_norm": 1.0876028537750244,
      "learning_rate": 4.6242364426986815e-06,
      "loss": 1.4474,
      "step": 6323
    },
    {
      "epoch": 0.9058874086807047,
      "grad_norm": 1.1088327169418335,
      "learning_rate": 4.610300620649577e-06,
      "loss": 1.1431,
      "step": 6324
    },
    {
      "epoch": 0.9060306546340066,
      "grad_norm": 1.0077004432678223,
      "learning_rate": 4.59638533364225e-06,
      "loss": 1.4191,
      "step": 6325
    },
    {
      "epoch": 0.9061739005873084,
      "grad_norm": 1.439351201057434,
      "learning_rate": 4.582490584672272e-06,
      "loss": 1.3493,
      "step": 6326
    },
    {
      "epoch": 0.9063171465406102,
      "grad_norm": 1.2281850576400757,
      "learning_rate": 4.5686163767308606e-06,
      "loss": 1.4246,
      "step": 6327
    },
    {
      "epoch": 0.9064603924939121,
      "grad_norm": 1.1451550722122192,
      "learning_rate": 4.554762712804772e-06,
      "loss": 1.3007,
      "step": 6328
    },
    {
      "epoch": 0.9066036384472138,
      "grad_norm": 1.092812180519104,
      "learning_rate": 4.540929595876376e-06,
      "loss": 1.4011,
      "step": 6329
    },
    {
      "epoch": 0.9067468844005157,
      "grad_norm": 0.9452556371688843,
      "learning_rate": 4.527117028923599e-06,
      "loss": 1.4344,
      "step": 6330
    },
    {
      "epoch": 0.9068901303538175,
      "grad_norm": 0.9995648264884949,
      "learning_rate": 4.513325014919923e-06,
      "loss": 1.5025,
      "step": 6331
    },
    {
      "epoch": 0.9070333763071193,
      "grad_norm": 1.0484440326690674,
      "learning_rate": 4.499553556834446e-06,
      "loss": 1.3382,
      "step": 6332
    },
    {
      "epoch": 0.9071766222604212,
      "grad_norm": 1.012657642364502,
      "learning_rate": 4.4858026576318435e-06,
      "loss": 1.3602,
      "step": 6333
    },
    {
      "epoch": 0.907319868213723,
      "grad_norm": 1.147506594657898,
      "learning_rate": 4.472072320272292e-06,
      "loss": 1.4667,
      "step": 6334
    },
    {
      "epoch": 0.9074631141670247,
      "grad_norm": 1.0223970413208008,
      "learning_rate": 4.4583625477116256e-06,
      "loss": 1.4221,
      "step": 6335
    },
    {
      "epoch": 0.9076063601203266,
      "grad_norm": 1.0855114459991455,
      "learning_rate": 4.444673342901218e-06,
      "loss": 1.2877,
      "step": 6336
    },
    {
      "epoch": 0.9077496060736284,
      "grad_norm": 1.2820749282836914,
      "learning_rate": 4.431004708788011e-06,
      "loss": 1.2337,
      "step": 6337
    },
    {
      "epoch": 0.9078928520269303,
      "grad_norm": 1.1012824773788452,
      "learning_rate": 4.41735664831453e-06,
      "loss": 1.3096,
      "step": 6338
    },
    {
      "epoch": 0.9080360979802321,
      "grad_norm": 1.0830519199371338,
      "learning_rate": 4.403729164418857e-06,
      "loss": 1.4016,
      "step": 6339
    },
    {
      "epoch": 0.9081793439335338,
      "grad_norm": 1.1054964065551758,
      "learning_rate": 4.390122260034657e-06,
      "loss": 1.427,
      "step": 6340
    },
    {
      "epoch": 0.9083225898868357,
      "grad_norm": 1.2707315683364868,
      "learning_rate": 4.376535938091153e-06,
      "loss": 1.1683,
      "step": 6341
    },
    {
      "epoch": 0.9084658358401375,
      "grad_norm": 1.4266351461410522,
      "learning_rate": 4.362970201513139e-06,
      "loss": 1.3057,
      "step": 6342
    },
    {
      "epoch": 0.9086090817934394,
      "grad_norm": 1.1522914171218872,
      "learning_rate": 4.3494250532210124e-06,
      "loss": 1.6687,
      "step": 6343
    },
    {
      "epoch": 0.9087523277467412,
      "grad_norm": 1.1509004831314087,
      "learning_rate": 4.335900496130674e-06,
      "loss": 1.4543,
      "step": 6344
    },
    {
      "epoch": 0.9088955737000429,
      "grad_norm": 1.037503719329834,
      "learning_rate": 4.3223965331536386e-06,
      "loss": 1.3607,
      "step": 6345
    },
    {
      "epoch": 0.9090388196533448,
      "grad_norm": 1.185473084449768,
      "learning_rate": 4.308913167196982e-06,
      "loss": 1.2868,
      "step": 6346
    },
    {
      "epoch": 0.9091820656066466,
      "grad_norm": 1.037713646888733,
      "learning_rate": 4.2954504011633255e-06,
      "loss": 1.4476,
      "step": 6347
    },
    {
      "epoch": 0.9093253115599484,
      "grad_norm": 1.029164433479309,
      "learning_rate": 4.282008237950896e-06,
      "loss": 1.3455,
      "step": 6348
    },
    {
      "epoch": 0.9094685575132503,
      "grad_norm": 1.2151432037353516,
      "learning_rate": 4.2685866804534236e-06,
      "loss": 1.315,
      "step": 6349
    },
    {
      "epoch": 0.909611803466552,
      "grad_norm": 0.9240227341651917,
      "learning_rate": 4.255185731560252e-06,
      "loss": 1.4948,
      "step": 6350
    },
    {
      "epoch": 0.9097550494198539,
      "grad_norm": 1.0023682117462158,
      "learning_rate": 4.241805394156295e-06,
      "loss": 1.5267,
      "step": 6351
    },
    {
      "epoch": 0.9098982953731557,
      "grad_norm": 1.3643170595169067,
      "learning_rate": 4.228445671121972e-06,
      "loss": 1.5083,
      "step": 6352
    },
    {
      "epoch": 0.9100415413264575,
      "grad_norm": 0.9547325372695923,
      "learning_rate": 4.215106565333316e-06,
      "loss": 1.4505,
      "step": 6353
    },
    {
      "epoch": 0.9101847872797594,
      "grad_norm": 0.9599502086639404,
      "learning_rate": 4.201788079661928e-06,
      "loss": 1.4625,
      "step": 6354
    },
    {
      "epoch": 0.9103280332330612,
      "grad_norm": 1.4839260578155518,
      "learning_rate": 4.188490216974916e-06,
      "loss": 1.4722,
      "step": 6355
    },
    {
      "epoch": 0.910471279186363,
      "grad_norm": 0.9959442019462585,
      "learning_rate": 4.17521298013499e-06,
      "loss": 1.4661,
      "step": 6356
    },
    {
      "epoch": 0.9106145251396648,
      "grad_norm": 1.2105557918548584,
      "learning_rate": 4.161956372000419e-06,
      "loss": 1.4673,
      "step": 6357
    },
    {
      "epoch": 0.9107577710929666,
      "grad_norm": 1.1694109439849854,
      "learning_rate": 4.14872039542501e-06,
      "loss": 1.3651,
      "step": 6358
    },
    {
      "epoch": 0.9109010170462685,
      "grad_norm": 0.937365710735321,
      "learning_rate": 4.135505053258171e-06,
      "loss": 1.4675,
      "step": 6359
    },
    {
      "epoch": 0.9110442629995703,
      "grad_norm": 0.9310875535011292,
      "learning_rate": 4.122310348344782e-06,
      "loss": 1.3986,
      "step": 6360
    },
    {
      "epoch": 0.911187508952872,
      "grad_norm": 0.9632576107978821,
      "learning_rate": 4.109136283525394e-06,
      "loss": 1.5452,
      "step": 6361
    },
    {
      "epoch": 0.9113307549061739,
      "grad_norm": 1.23027765750885,
      "learning_rate": 4.0959828616360385e-06,
      "loss": 1.1532,
      "step": 6362
    },
    {
      "epoch": 0.9114740008594757,
      "grad_norm": 1.1082545518875122,
      "learning_rate": 4.082850085508305e-06,
      "loss": 1.326,
      "step": 6363
    },
    {
      "epoch": 0.9116172468127776,
      "grad_norm": 1.0489147901535034,
      "learning_rate": 4.069737957969366e-06,
      "loss": 1.3453,
      "step": 6364
    },
    {
      "epoch": 0.9117604927660794,
      "grad_norm": 1.1038568019866943,
      "learning_rate": 4.056646481841952e-06,
      "loss": 1.3133,
      "step": 6365
    },
    {
      "epoch": 0.9119037387193811,
      "grad_norm": 0.9686883091926575,
      "learning_rate": 4.04357565994431e-06,
      "loss": 1.4335,
      "step": 6366
    },
    {
      "epoch": 0.912046984672683,
      "grad_norm": 1.1946451663970947,
      "learning_rate": 4.030525495090276e-06,
      "loss": 1.3979,
      "step": 6367
    },
    {
      "epoch": 0.9121902306259848,
      "grad_norm": 1.1127737760543823,
      "learning_rate": 4.017495990089227e-06,
      "loss": 1.472,
      "step": 6368
    },
    {
      "epoch": 0.9123334765792867,
      "grad_norm": 0.9373816847801208,
      "learning_rate": 4.004487147746105e-06,
      "loss": 1.3799,
      "step": 6369
    },
    {
      "epoch": 0.9124767225325885,
      "grad_norm": 1.2301857471466064,
      "learning_rate": 3.991498970861373e-06,
      "loss": 1.3138,
      "step": 6370
    },
    {
      "epoch": 0.9126199684858902,
      "grad_norm": 1.12795889377594,
      "learning_rate": 3.9785314622310495e-06,
      "loss": 1.3839,
      "step": 6371
    },
    {
      "epoch": 0.9127632144391921,
      "grad_norm": 1.0495474338531494,
      "learning_rate": 3.965584624646768e-06,
      "loss": 1.2327,
      "step": 6372
    },
    {
      "epoch": 0.9129064603924939,
      "grad_norm": 1.1497222185134888,
      "learning_rate": 3.9526584608956196e-06,
      "loss": 1.449,
      "step": 6373
    },
    {
      "epoch": 0.9130497063457957,
      "grad_norm": 1.0091516971588135,
      "learning_rate": 3.939752973760313e-06,
      "loss": 1.3958,
      "step": 6374
    },
    {
      "epoch": 0.9131929522990976,
      "grad_norm": 1.1240473985671997,
      "learning_rate": 3.9268681660190595e-06,
      "loss": 1.3291,
      "step": 6375
    },
    {
      "epoch": 0.9133361982523994,
      "grad_norm": 1.1444019079208374,
      "learning_rate": 3.9140040404456515e-06,
      "loss": 1.3045,
      "step": 6376
    },
    {
      "epoch": 0.9134794442057012,
      "grad_norm": 1.0356318950653076,
      "learning_rate": 3.901160599809428e-06,
      "loss": 1.3282,
      "step": 6377
    },
    {
      "epoch": 0.913622690159003,
      "grad_norm": 1.0371900796890259,
      "learning_rate": 3.888337846875223e-06,
      "loss": 1.4139,
      "step": 6378
    },
    {
      "epoch": 0.9137659361123048,
      "grad_norm": 1.1123536825180054,
      "learning_rate": 3.875535784403495e-06,
      "loss": 1.2246,
      "step": 6379
    },
    {
      "epoch": 0.9139091820656067,
      "grad_norm": 1.2111045122146606,
      "learning_rate": 3.862754415150216e-06,
      "loss": 1.3707,
      "step": 6380
    },
    {
      "epoch": 0.9140524280189085,
      "grad_norm": 1.1193556785583496,
      "learning_rate": 3.849993741866864e-06,
      "loss": 1.4247,
      "step": 6381
    },
    {
      "epoch": 0.9141956739722102,
      "grad_norm": 1.1268606185913086,
      "learning_rate": 3.837253767300519e-06,
      "loss": 1.3626,
      "step": 6382
    },
    {
      "epoch": 0.9143389199255121,
      "grad_norm": 1.5920072793960571,
      "learning_rate": 3.824534494193799e-06,
      "loss": 1.4573,
      "step": 6383
    },
    {
      "epoch": 0.9144821658788139,
      "grad_norm": 1.0998446941375732,
      "learning_rate": 3.8118359252848236e-06,
      "loss": 1.249,
      "step": 6384
    },
    {
      "epoch": 0.9146254118321158,
      "grad_norm": 1.1941128969192505,
      "learning_rate": 3.799158063307273e-06,
      "loss": 1.3874,
      "step": 6385
    },
    {
      "epoch": 0.9147686577854176,
      "grad_norm": 0.946759045124054,
      "learning_rate": 3.786500910990398e-06,
      "loss": 1.4146,
      "step": 6386
    },
    {
      "epoch": 0.9149119037387193,
      "grad_norm": 1.0596836805343628,
      "learning_rate": 3.773864471058963e-06,
      "loss": 1.1985,
      "step": 6387
    },
    {
      "epoch": 0.9150551496920212,
      "grad_norm": 1.305836796760559,
      "learning_rate": 3.76124874623327e-06,
      "loss": 1.3179,
      "step": 6388
    },
    {
      "epoch": 0.915198395645323,
      "grad_norm": 1.0566449165344238,
      "learning_rate": 3.748653739229191e-06,
      "loss": 1.336,
      "step": 6389
    },
    {
      "epoch": 0.9153416415986249,
      "grad_norm": 1.0194088220596313,
      "learning_rate": 3.7360794527581234e-06,
      "loss": 1.4013,
      "step": 6390
    },
    {
      "epoch": 0.9154848875519267,
      "grad_norm": 0.9489338994026184,
      "learning_rate": 3.7235258895269686e-06,
      "loss": 1.1851,
      "step": 6391
    },
    {
      "epoch": 0.9156281335052284,
      "grad_norm": 1.189700722694397,
      "learning_rate": 3.7109930522382097e-06,
      "loss": 1.4238,
      "step": 6392
    },
    {
      "epoch": 0.9157713794585303,
      "grad_norm": 1.3931171894073486,
      "learning_rate": 3.698480943589888e-06,
      "loss": 1.4314,
      "step": 6393
    },
    {
      "epoch": 0.9159146254118321,
      "grad_norm": 1.0614889860153198,
      "learning_rate": 3.685989566275516e-06,
      "loss": 1.4658,
      "step": 6394
    },
    {
      "epoch": 0.916057871365134,
      "grad_norm": 0.9896694421768188,
      "learning_rate": 3.673518922984187e-06,
      "loss": 1.3688,
      "step": 6395
    },
    {
      "epoch": 0.9162011173184358,
      "grad_norm": 1.1887366771697998,
      "learning_rate": 3.66106901640052e-06,
      "loss": 1.3087,
      "step": 6396
    },
    {
      "epoch": 0.9163443632717376,
      "grad_norm": 1.233389973640442,
      "learning_rate": 3.6486398492046827e-06,
      "loss": 1.3968,
      "step": 6397
    },
    {
      "epoch": 0.9164876092250394,
      "grad_norm": 1.0355703830718994,
      "learning_rate": 3.636231424072367e-06,
      "loss": 1.3336,
      "step": 6398
    },
    {
      "epoch": 0.9166308551783412,
      "grad_norm": 1.1142059564590454,
      "learning_rate": 3.623843743674771e-06,
      "loss": 1.4192,
      "step": 6399
    },
    {
      "epoch": 0.916774101131643,
      "grad_norm": 1.3490045070648193,
      "learning_rate": 3.611476810678693e-06,
      "loss": 1.4034,
      "step": 6400
    },
    {
      "epoch": 0.9169173470849449,
      "grad_norm": 0.9730175733566284,
      "learning_rate": 3.5991306277464167e-06,
      "loss": 1.4531,
      "step": 6401
    },
    {
      "epoch": 0.9170605930382467,
      "grad_norm": 1.1233919858932495,
      "learning_rate": 3.5868051975357587e-06,
      "loss": 1.3041,
      "step": 6402
    },
    {
      "epoch": 0.9172038389915484,
      "grad_norm": 1.0261316299438477,
      "learning_rate": 3.5745005227000637e-06,
      "loss": 1.4572,
      "step": 6403
    },
    {
      "epoch": 0.9173470849448503,
      "grad_norm": 1.193352460861206,
      "learning_rate": 3.56221660588828e-06,
      "loss": 1.3143,
      "step": 6404
    },
    {
      "epoch": 0.9174903308981521,
      "grad_norm": 1.0419745445251465,
      "learning_rate": 3.5499534497447807e-06,
      "loss": 1.2528,
      "step": 6405
    },
    {
      "epoch": 0.917633576851454,
      "grad_norm": 1.0759903192520142,
      "learning_rate": 3.5377110569095316e-06,
      "loss": 1.3717,
      "step": 6406
    },
    {
      "epoch": 0.9177768228047558,
      "grad_norm": 1.2016997337341309,
      "learning_rate": 3.5254894300180252e-06,
      "loss": 1.4165,
      "step": 6407
    },
    {
      "epoch": 0.9179200687580575,
      "grad_norm": 1.1434471607208252,
      "learning_rate": 3.5132885717012675e-06,
      "loss": 1.2155,
      "step": 6408
    },
    {
      "epoch": 0.9180633147113594,
      "grad_norm": 1.4297231435775757,
      "learning_rate": 3.5011084845858246e-06,
      "loss": 1.2848,
      "step": 6409
    },
    {
      "epoch": 0.9182065606646612,
      "grad_norm": 1.0412758588790894,
      "learning_rate": 3.4889491712937205e-06,
      "loss": 1.319,
      "step": 6410
    },
    {
      "epoch": 0.9183498066179631,
      "grad_norm": 0.9930747151374817,
      "learning_rate": 3.4768106344425956e-06,
      "loss": 1.4622,
      "step": 6411
    },
    {
      "epoch": 0.9184930525712649,
      "grad_norm": 0.957931637763977,
      "learning_rate": 3.46469287664557e-06,
      "loss": 1.3093,
      "step": 6412
    },
    {
      "epoch": 0.9186362985245666,
      "grad_norm": 1.179731845855713,
      "learning_rate": 3.4525959005112794e-06,
      "loss": 1.4868,
      "step": 6413
    },
    {
      "epoch": 0.9187795444778685,
      "grad_norm": 0.889260470867157,
      "learning_rate": 3.440519708643919e-06,
      "loss": 1.3579,
      "step": 6414
    },
    {
      "epoch": 0.9189227904311703,
      "grad_norm": 1.1458030939102173,
      "learning_rate": 3.428464303643186e-06,
      "loss": 1.3403,
      "step": 6415
    },
    {
      "epoch": 0.9190660363844722,
      "grad_norm": 1.189430594444275,
      "learning_rate": 3.4164296881043055e-06,
      "loss": 1.5304,
      "step": 6416
    },
    {
      "epoch": 0.919209282337774,
      "grad_norm": 1.0229490995407104,
      "learning_rate": 3.4044158646180604e-06,
      "loss": 1.3845,
      "step": 6417
    },
    {
      "epoch": 0.9193525282910758,
      "grad_norm": 1.1848504543304443,
      "learning_rate": 3.3924228357706922e-06,
      "loss": 1.4941,
      "step": 6418
    },
    {
      "epoch": 0.9194957742443776,
      "grad_norm": 1.21671724319458,
      "learning_rate": 3.3804506041440363e-06,
      "loss": 1.5864,
      "step": 6419
    },
    {
      "epoch": 0.9196390201976794,
      "grad_norm": 1.372091293334961,
      "learning_rate": 3.368499172315398e-06,
      "loss": 1.36,
      "step": 6420
    },
    {
      "epoch": 0.9197822661509812,
      "grad_norm": 1.164242148399353,
      "learning_rate": 3.356568542857619e-06,
      "loss": 1.2125,
      "step": 6421
    },
    {
      "epoch": 0.9199255121042831,
      "grad_norm": 1.1633225679397583,
      "learning_rate": 3.3446587183390888e-06,
      "loss": 1.4637,
      "step": 6422
    },
    {
      "epoch": 0.9200687580575849,
      "grad_norm": 0.9969565868377686,
      "learning_rate": 3.3327697013236904e-06,
      "loss": 1.4894,
      "step": 6423
    },
    {
      "epoch": 0.9202120040108867,
      "grad_norm": 1.3349730968475342,
      "learning_rate": 3.3209014943708204e-06,
      "loss": 1.2023,
      "step": 6424
    },
    {
      "epoch": 0.9203552499641885,
      "grad_norm": 1.0232676267623901,
      "learning_rate": 3.3090541000354358e-06,
      "loss": 1.4571,
      "step": 6425
    },
    {
      "epoch": 0.9204984959174903,
      "grad_norm": 1.0784478187561035,
      "learning_rate": 3.2972275208679625e-06,
      "loss": 1.4151,
      "step": 6426
    },
    {
      "epoch": 0.9206417418707922,
      "grad_norm": 0.9066776633262634,
      "learning_rate": 3.2854217594143975e-06,
      "loss": 1.3459,
      "step": 6427
    },
    {
      "epoch": 0.920784987824094,
      "grad_norm": 1.1936663389205933,
      "learning_rate": 3.273636818216197e-06,
      "loss": 1.468,
      "step": 6428
    },
    {
      "epoch": 0.9209282337773957,
      "grad_norm": 1.1588611602783203,
      "learning_rate": 3.2618726998103867e-06,
      "loss": 1.6047,
      "step": 6429
    },
    {
      "epoch": 0.9210714797306976,
      "grad_norm": 1.0405627489089966,
      "learning_rate": 3.2501294067295073e-06,
      "loss": 1.6571,
      "step": 6430
    },
    {
      "epoch": 0.9212147256839994,
      "grad_norm": 1.163070797920227,
      "learning_rate": 3.2384069415015594e-06,
      "loss": 1.2721,
      "step": 6431
    },
    {
      "epoch": 0.9213579716373013,
      "grad_norm": 1.0822196006774902,
      "learning_rate": 3.226705306650113e-06,
      "loss": 1.3982,
      "step": 6432
    },
    {
      "epoch": 0.9215012175906031,
      "grad_norm": 1.1785391569137573,
      "learning_rate": 3.215024504694264e-06,
      "loss": 1.4575,
      "step": 6433
    },
    {
      "epoch": 0.9216444635439048,
      "grad_norm": 1.0128576755523682,
      "learning_rate": 3.203364538148579e-06,
      "loss": 1.3587,
      "step": 6434
    },
    {
      "epoch": 0.9217877094972067,
      "grad_norm": 0.9469732642173767,
      "learning_rate": 3.1917254095231606e-06,
      "loss": 1.3952,
      "step": 6435
    },
    {
      "epoch": 0.9219309554505085,
      "grad_norm": 1.11313796043396,
      "learning_rate": 3.1801071213236277e-06,
      "loss": 1.4033,
      "step": 6436
    },
    {
      "epoch": 0.9220742014038104,
      "grad_norm": 0.9706379771232605,
      "learning_rate": 3.1685096760511123e-06,
      "loss": 1.5194,
      "step": 6437
    },
    {
      "epoch": 0.9222174473571122,
      "grad_norm": 1.0394903421401978,
      "learning_rate": 3.156933076202262e-06,
      "loss": 1.2867,
      "step": 6438
    },
    {
      "epoch": 0.9223606933104139,
      "grad_norm": 1.1205493211746216,
      "learning_rate": 3.145377324269205e-06,
      "loss": 1.5495,
      "step": 6439
    },
    {
      "epoch": 0.9225039392637158,
      "grad_norm": 1.1452409029006958,
      "learning_rate": 3.1338424227396524e-06,
      "loss": 1.4917,
      "step": 6440
    },
    {
      "epoch": 0.9226471852170176,
      "grad_norm": 0.8819735646247864,
      "learning_rate": 3.122328374096761e-06,
      "loss": 1.3784,
      "step": 6441
    },
    {
      "epoch": 0.9227904311703194,
      "grad_norm": 1.0010957717895508,
      "learning_rate": 3.110835180819216e-06,
      "loss": 1.3729,
      "step": 6442
    },
    {
      "epoch": 0.9229336771236213,
      "grad_norm": 1.091059684753418,
      "learning_rate": 3.099362845381215e-06,
      "loss": 1.3659,
      "step": 6443
    },
    {
      "epoch": 0.9230769230769231,
      "grad_norm": 1.1242009401321411,
      "learning_rate": 3.0879113702524832e-06,
      "loss": 1.3277,
      "step": 6444
    },
    {
      "epoch": 0.9232201690302249,
      "grad_norm": 1.8432196378707886,
      "learning_rate": 3.0764807578982256e-06,
      "loss": 1.207,
      "step": 6445
    },
    {
      "epoch": 0.9233634149835267,
      "grad_norm": 1.1757209300994873,
      "learning_rate": 3.0650710107791748e-06,
      "loss": 1.4039,
      "step": 6446
    },
    {
      "epoch": 0.9235066609368285,
      "grad_norm": 0.9085772037506104,
      "learning_rate": 3.053682131351576e-06,
      "loss": 1.4077,
      "step": 6447
    },
    {
      "epoch": 0.9236499068901304,
      "grad_norm": 1.0273581743240356,
      "learning_rate": 3.042314122067169e-06,
      "loss": 1.3291,
      "step": 6448
    },
    {
      "epoch": 0.9237931528434322,
      "grad_norm": 1.4282138347625732,
      "learning_rate": 3.0309669853731848e-06,
      "loss": 1.363,
      "step": 6449
    },
    {
      "epoch": 0.923936398796734,
      "grad_norm": 1.030496597290039,
      "learning_rate": 3.0196407237124024e-06,
      "loss": 1.358,
      "step": 6450
    },
    {
      "epoch": 0.9240796447500358,
      "grad_norm": 1.0387153625488281,
      "learning_rate": 3.008335339523105e-06,
      "loss": 1.41,
      "step": 6451
    },
    {
      "epoch": 0.9242228907033376,
      "grad_norm": 1.1078928709030151,
      "learning_rate": 2.9970508352390125e-06,
      "loss": 1.3474,
      "step": 6452
    },
    {
      "epoch": 0.9243661366566395,
      "grad_norm": 1.038006067276001,
      "learning_rate": 2.985787213289415e-06,
      "loss": 1.3773,
      "step": 6453
    },
    {
      "epoch": 0.9245093826099413,
      "grad_norm": 0.962173581123352,
      "learning_rate": 2.9745444760991283e-06,
      "loss": 1.3373,
      "step": 6454
    },
    {
      "epoch": 0.924652628563243,
      "grad_norm": 0.9636121988296509,
      "learning_rate": 2.963322626088405e-06,
      "loss": 1.3639,
      "step": 6455
    },
    {
      "epoch": 0.9247958745165449,
      "grad_norm": 1.2030967473983765,
      "learning_rate": 2.9521216656730242e-06,
      "loss": 1.4641,
      "step": 6456
    },
    {
      "epoch": 0.9249391204698467,
      "grad_norm": 1.2235599756240845,
      "learning_rate": 2.9409415972642905e-06,
      "loss": 1.4006,
      "step": 6457
    },
    {
      "epoch": 0.9250823664231486,
      "grad_norm": 1.1225051879882812,
      "learning_rate": 2.9297824232689895e-06,
      "loss": 1.4989,
      "step": 6458
    },
    {
      "epoch": 0.9252256123764504,
      "grad_norm": 1.066063404083252,
      "learning_rate": 2.9186441460894333e-06,
      "loss": 1.3654,
      "step": 6459
    },
    {
      "epoch": 0.9253688583297521,
      "grad_norm": 1.2072542905807495,
      "learning_rate": 2.9075267681233697e-06,
      "loss": 1.2892,
      "step": 6460
    },
    {
      "epoch": 0.925512104283054,
      "grad_norm": 1.2591460943222046,
      "learning_rate": 2.896430291764152e-06,
      "loss": 1.3125,
      "step": 6461
    },
    {
      "epoch": 0.9256553502363558,
      "grad_norm": 1.2391941547393799,
      "learning_rate": 2.885354719400557e-06,
      "loss": 1.3301,
      "step": 6462
    },
    {
      "epoch": 0.9257985961896577,
      "grad_norm": 1.0523512363433838,
      "learning_rate": 2.8743000534168675e-06,
      "loss": 1.5114,
      "step": 6463
    },
    {
      "epoch": 0.9259418421429595,
      "grad_norm": 0.9606161713600159,
      "learning_rate": 2.863266296192879e-06,
      "loss": 1.5302,
      "step": 6464
    },
    {
      "epoch": 0.9260850880962613,
      "grad_norm": 0.963270366191864,
      "learning_rate": 2.8522534501039035e-06,
      "loss": 1.3593,
      "step": 6465
    },
    {
      "epoch": 0.9262283340495631,
      "grad_norm": 1.2613978385925293,
      "learning_rate": 2.8412615175207324e-06,
      "loss": 1.4532,
      "step": 6466
    },
    {
      "epoch": 0.9263715800028649,
      "grad_norm": 0.9834369421005249,
      "learning_rate": 2.8302905008096403e-06,
      "loss": 1.3304,
      "step": 6467
    },
    {
      "epoch": 0.9265148259561667,
      "grad_norm": 1.0687365531921387,
      "learning_rate": 2.8193404023324376e-06,
      "loss": 1.4737,
      "step": 6468
    },
    {
      "epoch": 0.9266580719094686,
      "grad_norm": 1.2179547548294067,
      "learning_rate": 2.8084112244464056e-06,
      "loss": 1.425,
      "step": 6469
    },
    {
      "epoch": 0.9268013178627704,
      "grad_norm": 1.090914011001587,
      "learning_rate": 2.7975029695043064e-06,
      "loss": 1.3174,
      "step": 6470
    },
    {
      "epoch": 0.9269445638160722,
      "grad_norm": 1.0008254051208496,
      "learning_rate": 2.7866156398544176e-06,
      "loss": 1.3894,
      "step": 6471
    },
    {
      "epoch": 0.927087809769374,
      "grad_norm": 1.2196687459945679,
      "learning_rate": 2.7757492378405414e-06,
      "loss": 1.303,
      "step": 6472
    },
    {
      "epoch": 0.9272310557226758,
      "grad_norm": 1.0364189147949219,
      "learning_rate": 2.7649037658019183e-06,
      "loss": 1.3859,
      "step": 6473
    },
    {
      "epoch": 0.9273743016759777,
      "grad_norm": 1.0551917552947998,
      "learning_rate": 2.754079226073325e-06,
      "loss": 1.5722,
      "step": 6474
    },
    {
      "epoch": 0.9275175476292795,
      "grad_norm": 0.9522950053215027,
      "learning_rate": 2.7432756209850084e-06,
      "loss": 1.4138,
      "step": 6475
    },
    {
      "epoch": 0.9276607935825812,
      "grad_norm": 1.098632574081421,
      "learning_rate": 2.7324929528627195e-06,
      "loss": 1.4097,
      "step": 6476
    },
    {
      "epoch": 0.9278040395358831,
      "grad_norm": 0.9926488995552063,
      "learning_rate": 2.7217312240277127e-06,
      "loss": 1.2624,
      "step": 6477
    },
    {
      "epoch": 0.9279472854891849,
      "grad_norm": 1.0938059091567993,
      "learning_rate": 2.71099043679669e-06,
      "loss": 1.3838,
      "step": 6478
    },
    {
      "epoch": 0.9280905314424868,
      "grad_norm": 1.2125778198242188,
      "learning_rate": 2.7002705934819018e-06,
      "loss": 1.2383,
      "step": 6479
    },
    {
      "epoch": 0.9282337773957886,
      "grad_norm": 1.1910125017166138,
      "learning_rate": 2.68957169639108e-06,
      "loss": 1.3472,
      "step": 6480
    },
    {
      "epoch": 0.9283770233490903,
      "grad_norm": 1.1332073211669922,
      "learning_rate": 2.6788937478273934e-06,
      "loss": 1.4529,
      "step": 6481
    },
    {
      "epoch": 0.9285202693023922,
      "grad_norm": 1.0849745273590088,
      "learning_rate": 2.6682367500895587e-06,
      "loss": 1.3518,
      "step": 6482
    },
    {
      "epoch": 0.928663515255694,
      "grad_norm": 1.0638624429702759,
      "learning_rate": 2.6576007054717746e-06,
      "loss": 1.4478,
      "step": 6483
    },
    {
      "epoch": 0.9288067612089959,
      "grad_norm": 1.1152218580245972,
      "learning_rate": 2.6469856162637086e-06,
      "loss": 1.5815,
      "step": 6484
    },
    {
      "epoch": 0.9289500071622977,
      "grad_norm": 1.0590484142303467,
      "learning_rate": 2.6363914847505113e-06,
      "loss": 1.3776,
      "step": 6485
    },
    {
      "epoch": 0.9290932531155995,
      "grad_norm": 1.0281152725219727,
      "learning_rate": 2.625818313212869e-06,
      "loss": 1.4886,
      "step": 6486
    },
    {
      "epoch": 0.9292364990689013,
      "grad_norm": 1.170256495475769,
      "learning_rate": 2.6152661039268943e-06,
      "loss": 1.4909,
      "step": 6487
    },
    {
      "epoch": 0.9293797450222031,
      "grad_norm": 1.1871517896652222,
      "learning_rate": 2.6047348591642483e-06,
      "loss": 1.381,
      "step": 6488
    },
    {
      "epoch": 0.929522990975505,
      "grad_norm": 1.199926733970642,
      "learning_rate": 2.5942245811920065e-06,
      "loss": 1.5402,
      "step": 6489
    },
    {
      "epoch": 0.9296662369288068,
      "grad_norm": 1.009153127670288,
      "learning_rate": 2.5837352722728026e-06,
      "loss": 1.5732,
      "step": 6490
    },
    {
      "epoch": 0.9298094828821086,
      "grad_norm": 0.829253077507019,
      "learning_rate": 2.57326693466472e-06,
      "loss": 1.4724,
      "step": 6491
    },
    {
      "epoch": 0.9299527288354104,
      "grad_norm": 1.084157109260559,
      "learning_rate": 2.562819570621322e-06,
      "loss": 1.4095,
      "step": 6492
    },
    {
      "epoch": 0.9300959747887122,
      "grad_norm": 1.1585246324539185,
      "learning_rate": 2.552393182391677e-06,
      "loss": 1.3877,
      "step": 6493
    },
    {
      "epoch": 0.930239220742014,
      "grad_norm": 0.9959848523139954,
      "learning_rate": 2.5419877722203333e-06,
      "loss": 1.3872,
      "step": 6494
    },
    {
      "epoch": 0.9303824666953159,
      "grad_norm": 1.3616734743118286,
      "learning_rate": 2.5316033423472997e-06,
      "loss": 1.3783,
      "step": 6495
    },
    {
      "epoch": 0.9305257126486177,
      "grad_norm": 1.2964977025985718,
      "learning_rate": 2.5212398950081096e-06,
      "loss": 1.3801,
      "step": 6496
    },
    {
      "epoch": 0.9306689586019194,
      "grad_norm": 1.019456386566162,
      "learning_rate": 2.510897432433734e-06,
      "loss": 1.3676,
      "step": 6497
    },
    {
      "epoch": 0.9308122045552213,
      "grad_norm": 0.9017200469970703,
      "learning_rate": 2.5005759568506704e-06,
      "loss": 1.3527,
      "step": 6498
    },
    {
      "epoch": 0.9309554505085231,
      "grad_norm": 0.9133973717689514,
      "learning_rate": 2.490275470480863e-06,
      "loss": 1.4446,
      "step": 6499
    },
    {
      "epoch": 0.931098696461825,
      "grad_norm": 0.961898922920227,
      "learning_rate": 2.479995975541749e-06,
      "loss": 1.4438,
      "step": 6500
    },
    {
      "epoch": 0.9312419424151268,
      "grad_norm": 1.2371913194656372,
      "learning_rate": 2.4697374742462698e-06,
      "loss": 1.2757,
      "step": 6501
    },
    {
      "epoch": 0.9313851883684285,
      "grad_norm": 1.0603811740875244,
      "learning_rate": 2.4594999688028032e-06,
      "loss": 1.3934,
      "step": 6502
    },
    {
      "epoch": 0.9315284343217304,
      "grad_norm": 0.9528247714042664,
      "learning_rate": 2.4492834614152414e-06,
      "loss": 1.2006,
      "step": 6503
    },
    {
      "epoch": 0.9316716802750322,
      "grad_norm": 1.3816990852355957,
      "learning_rate": 2.439087954282948e-06,
      "loss": 1.4039,
      "step": 6504
    },
    {
      "epoch": 0.9318149262283341,
      "grad_norm": 1.1050955057144165,
      "learning_rate": 2.428913449600756e-06,
      "loss": 1.1844,
      "step": 6505
    },
    {
      "epoch": 0.9319581721816359,
      "grad_norm": 1.1582280397415161,
      "learning_rate": 2.418759949558991e-06,
      "loss": 1.4193,
      "step": 6506
    },
    {
      "epoch": 0.9321014181349377,
      "grad_norm": 1.1322112083435059,
      "learning_rate": 2.4086274563434488e-06,
      "loss": 1.4639,
      "step": 6507
    },
    {
      "epoch": 0.9322446640882395,
      "grad_norm": 1.1956231594085693,
      "learning_rate": 2.3985159721353956e-06,
      "loss": 1.3419,
      "step": 6508
    },
    {
      "epoch": 0.9323879100415413,
      "grad_norm": 0.9906167387962341,
      "learning_rate": 2.388425499111613e-06,
      "loss": 1.5061,
      "step": 6509
    },
    {
      "epoch": 0.9325311559948432,
      "grad_norm": 1.0036982297897339,
      "learning_rate": 2.378356039444285e-06,
      "loss": 1.3708,
      "step": 6510
    },
    {
      "epoch": 0.932674401948145,
      "grad_norm": 0.9920417666435242,
      "learning_rate": 2.3683075953011558e-06,
      "loss": 1.6787,
      "step": 6511
    },
    {
      "epoch": 0.9328176479014468,
      "grad_norm": 1.249273657798767,
      "learning_rate": 2.3582801688453948e-06,
      "loss": 1.4878,
      "step": 6512
    },
    {
      "epoch": 0.9329608938547486,
      "grad_norm": 0.9275979399681091,
      "learning_rate": 2.348273762235642e-06,
      "loss": 1.3736,
      "step": 6513
    },
    {
      "epoch": 0.9331041398080504,
      "grad_norm": 1.1471542119979858,
      "learning_rate": 2.3382883776260524e-06,
      "loss": 1.4489,
      "step": 6514
    },
    {
      "epoch": 0.9332473857613522,
      "grad_norm": 1.1553690433502197,
      "learning_rate": 2.3283240171662167e-06,
      "loss": 1.3614,
      "step": 6515
    },
    {
      "epoch": 0.9333906317146541,
      "grad_norm": 1.105751633644104,
      "learning_rate": 2.318380683001231e-06,
      "loss": 1.3318,
      "step": 6516
    },
    {
      "epoch": 0.9335338776679559,
      "grad_norm": 1.143595814704895,
      "learning_rate": 2.3084583772716275e-06,
      "loss": 1.5478,
      "step": 6517
    },
    {
      "epoch": 0.9336771236212577,
      "grad_norm": 1.1402645111083984,
      "learning_rate": 2.2985571021134524e-06,
      "loss": 1.4338,
      "step": 6518
    },
    {
      "epoch": 0.9338203695745595,
      "grad_norm": 1.2224313020706177,
      "learning_rate": 2.288676859658212e-06,
      "loss": 1.3265,
      "step": 6519
    },
    {
      "epoch": 0.9339636155278613,
      "grad_norm": 1.1123749017715454,
      "learning_rate": 2.27881765203285e-06,
      "loss": 1.4388,
      "step": 6520
    },
    {
      "epoch": 0.9341068614811632,
      "grad_norm": 1.0451598167419434,
      "learning_rate": 2.268979481359812e-06,
      "loss": 1.4712,
      "step": 6521
    },
    {
      "epoch": 0.934250107434465,
      "grad_norm": 1.1520843505859375,
      "learning_rate": 2.259162349757038e-06,
      "loss": 1.3638,
      "step": 6522
    },
    {
      "epoch": 0.9343933533877667,
      "grad_norm": 0.9486491680145264,
      "learning_rate": 2.249366259337893e-06,
      "loss": 1.437,
      "step": 6523
    },
    {
      "epoch": 0.9345365993410686,
      "grad_norm": 1.0022501945495605,
      "learning_rate": 2.239591212211245e-06,
      "loss": 1.3158,
      "step": 6524
    },
    {
      "epoch": 0.9346798452943704,
      "grad_norm": 1.1213300228118896,
      "learning_rate": 2.2298372104814115e-06,
      "loss": 1.294,
      "step": 6525
    },
    {
      "epoch": 0.9348230912476723,
      "grad_norm": 1.2453380823135376,
      "learning_rate": 2.22010425624819e-06,
      "loss": 1.4299,
      "step": 6526
    },
    {
      "epoch": 0.9349663372009741,
      "grad_norm": 0.9873899221420288,
      "learning_rate": 2.2103923516068605e-06,
      "loss": 1.3196,
      "step": 6527
    },
    {
      "epoch": 0.9351095831542758,
      "grad_norm": 0.9710806012153625,
      "learning_rate": 2.2007014986481167e-06,
      "loss": 1.3067,
      "step": 6528
    },
    {
      "epoch": 0.9352528291075777,
      "grad_norm": 1.000098466873169,
      "learning_rate": 2.1910316994581904e-06,
      "loss": 1.4584,
      "step": 6529
    },
    {
      "epoch": 0.9353960750608795,
      "grad_norm": 1.0472301244735718,
      "learning_rate": 2.1813829561187604e-06,
      "loss": 1.2174,
      "step": 6530
    },
    {
      "epoch": 0.9355393210141814,
      "grad_norm": 1.1677483320236206,
      "learning_rate": 2.1717552707069323e-06,
      "loss": 1.2664,
      "step": 6531
    },
    {
      "epoch": 0.9356825669674832,
      "grad_norm": 0.9195520877838135,
      "learning_rate": 2.1621486452953144e-06,
      "loss": 1.5163,
      "step": 6532
    },
    {
      "epoch": 0.935825812920785,
      "grad_norm": 1.0236012935638428,
      "learning_rate": 2.152563081951997e-06,
      "loss": 1.2329,
      "step": 6533
    },
    {
      "epoch": 0.9359690588740868,
      "grad_norm": 1.543801188468933,
      "learning_rate": 2.1429985827404853e-06,
      "loss": 1.295,
      "step": 6534
    },
    {
      "epoch": 0.9361123048273886,
      "grad_norm": 1.1117914915084839,
      "learning_rate": 2.1334551497197987e-06,
      "loss": 1.4779,
      "step": 6535
    },
    {
      "epoch": 0.9362555507806904,
      "grad_norm": 1.0583205223083496,
      "learning_rate": 2.1239327849444045e-06,
      "loss": 1.2041,
      "step": 6536
    },
    {
      "epoch": 0.9363987967339923,
      "grad_norm": 1.1168370246887207,
      "learning_rate": 2.1144314904642195e-06,
      "loss": 1.3311,
      "step": 6537
    },
    {
      "epoch": 0.9365420426872941,
      "grad_norm": 1.0833189487457275,
      "learning_rate": 2.104951268324651e-06,
      "loss": 1.4011,
      "step": 6538
    },
    {
      "epoch": 0.9366852886405959,
      "grad_norm": 1.1070337295532227,
      "learning_rate": 2.095492120566522e-06,
      "loss": 1.2256,
      "step": 6539
    },
    {
      "epoch": 0.9368285345938977,
      "grad_norm": 1.0879265069961548,
      "learning_rate": 2.0860540492262047e-06,
      "loss": 1.2924,
      "step": 6540
    },
    {
      "epoch": 0.9369717805471995,
      "grad_norm": 1.113938570022583,
      "learning_rate": 2.0766370563354508e-06,
      "loss": 1.4909,
      "step": 6541
    },
    {
      "epoch": 0.9371150265005014,
      "grad_norm": 0.9503293037414551,
      "learning_rate": 2.0672411439215165e-06,
      "loss": 1.3318,
      "step": 6542
    },
    {
      "epoch": 0.9372582724538032,
      "grad_norm": 1.1616195440292358,
      "learning_rate": 2.0578663140070954e-06,
      "loss": 1.3118,
      "step": 6543
    },
    {
      "epoch": 0.937401518407105,
      "grad_norm": 1.1152852773666382,
      "learning_rate": 2.048512568610361e-06,
      "loss": 1.2262,
      "step": 6544
    },
    {
      "epoch": 0.9375447643604068,
      "grad_norm": 1.106892704963684,
      "learning_rate": 2.0391799097449593e-06,
      "loss": 1.1471,
      "step": 6545
    },
    {
      "epoch": 0.9376880103137086,
      "grad_norm": 1.2196919918060303,
      "learning_rate": 2.02986833941996e-06,
      "loss": 1.4286,
      "step": 6546
    },
    {
      "epoch": 0.9378312562670105,
      "grad_norm": 1.0149511098861694,
      "learning_rate": 2.020577859639927e-06,
      "loss": 1.2843,
      "step": 6547
    },
    {
      "epoch": 0.9379745022203123,
      "grad_norm": 1.0943989753723145,
      "learning_rate": 2.0113084724048715e-06,
      "loss": 1.2628,
      "step": 6548
    },
    {
      "epoch": 0.938117748173614,
      "grad_norm": 0.9725356101989746,
      "learning_rate": 2.0020601797102523e-06,
      "loss": 1.4806,
      "step": 6549
    },
    {
      "epoch": 0.9382609941269159,
      "grad_norm": 1.04365074634552,
      "learning_rate": 1.9928329835469996e-06,
      "loss": 1.4893,
      "step": 6550
    },
    {
      "epoch": 0.9384042400802177,
      "grad_norm": 0.9421520233154297,
      "learning_rate": 1.983626885901513e-06,
      "loss": 1.4679,
      "step": 6551
    },
    {
      "epoch": 0.9385474860335196,
      "grad_norm": 1.0554935932159424,
      "learning_rate": 1.974441888755629e-06,
      "loss": 1.3595,
      "step": 6552
    },
    {
      "epoch": 0.9386907319868214,
      "grad_norm": 1.0590242147445679,
      "learning_rate": 1.965277994086645e-06,
      "loss": 1.3376,
      "step": 6553
    },
    {
      "epoch": 0.9388339779401232,
      "grad_norm": 0.9728385210037231,
      "learning_rate": 1.9561352038673263e-06,
      "loss": 1.3972,
      "step": 6554
    },
    {
      "epoch": 0.938977223893425,
      "grad_norm": 1.1325448751449585,
      "learning_rate": 1.947013520065899e-06,
      "loss": 1.4875,
      "step": 6555
    },
    {
      "epoch": 0.9391204698467268,
      "grad_norm": 1.1805099248886108,
      "learning_rate": 1.9379129446460253e-06,
      "loss": 1.2584,
      "step": 6556
    },
    {
      "epoch": 0.9392637158000287,
      "grad_norm": 1.02956223487854,
      "learning_rate": 1.9288334795668163e-06,
      "loss": 1.4759,
      "step": 6557
    },
    {
      "epoch": 0.9394069617533305,
      "grad_norm": 1.0579255819320679,
      "learning_rate": 1.9197751267828855e-06,
      "loss": 1.4481,
      "step": 6558
    },
    {
      "epoch": 0.9395502077066323,
      "grad_norm": 1.195554494857788,
      "learning_rate": 1.910737888244274e-06,
      "loss": 1.448,
      "step": 6559
    },
    {
      "epoch": 0.9396934536599341,
      "grad_norm": 1.2533451318740845,
      "learning_rate": 1.901721765896447e-06,
      "loss": 1.3288,
      "step": 6560
    },
    {
      "epoch": 0.9398366996132359,
      "grad_norm": 0.9402534365653992,
      "learning_rate": 1.8927267616803634e-06,
      "loss": 1.3684,
      "step": 6561
    },
    {
      "epoch": 0.9399799455665377,
      "grad_norm": 1.0570292472839355,
      "learning_rate": 1.883752877532452e-06,
      "loss": 1.444,
      "step": 6562
    },
    {
      "epoch": 0.9401231915198396,
      "grad_norm": 1.218531847000122,
      "learning_rate": 1.874800115384523e-06,
      "loss": 1.2587,
      "step": 6563
    },
    {
      "epoch": 0.9402664374731414,
      "grad_norm": 0.9667032957077026,
      "learning_rate": 1.8658684771639234e-06,
      "loss": 1.5308,
      "step": 6564
    },
    {
      "epoch": 0.9404096834264432,
      "grad_norm": 1.2847278118133545,
      "learning_rate": 1.8569579647933933e-06,
      "loss": 1.5103,
      "step": 6565
    },
    {
      "epoch": 0.940552929379745,
      "grad_norm": 1.2722854614257812,
      "learning_rate": 1.8480685801911424e-06,
      "loss": 1.4827,
      "step": 6566
    },
    {
      "epoch": 0.9406961753330468,
      "grad_norm": 1.0711419582366943,
      "learning_rate": 1.8392003252708622e-06,
      "loss": 1.3001,
      "step": 6567
    },
    {
      "epoch": 0.9408394212863487,
      "grad_norm": 1.0763605833053589,
      "learning_rate": 1.8303532019416258e-06,
      "loss": 1.3421,
      "step": 6568
    },
    {
      "epoch": 0.9409826672396505,
      "grad_norm": 0.9694857001304626,
      "learning_rate": 1.8215272121080317e-06,
      "loss": 1.35,
      "step": 6569
    },
    {
      "epoch": 0.9411259131929522,
      "grad_norm": 1.085087776184082,
      "learning_rate": 1.8127223576701046e-06,
      "loss": 1.28,
      "step": 6570
    },
    {
      "epoch": 0.9412691591462541,
      "grad_norm": 1.1587072610855103,
      "learning_rate": 1.8039386405232728e-06,
      "loss": 1.2014,
      "step": 6571
    },
    {
      "epoch": 0.9414124050995559,
      "grad_norm": 1.0818008184432983,
      "learning_rate": 1.79517606255849e-06,
      "loss": 1.3786,
      "step": 6572
    },
    {
      "epoch": 0.9415556510528578,
      "grad_norm": 0.9864383935928345,
      "learning_rate": 1.7864346256621033e-06,
      "loss": 1.4213,
      "step": 6573
    },
    {
      "epoch": 0.9416988970061596,
      "grad_norm": 1.424342155456543,
      "learning_rate": 1.7777143317159406e-06,
      "loss": 1.3157,
      "step": 6574
    },
    {
      "epoch": 0.9418421429594614,
      "grad_norm": 1.0142978429794312,
      "learning_rate": 1.7690151825972446e-06,
      "loss": 1.3288,
      "step": 6575
    },
    {
      "epoch": 0.9419853889127632,
      "grad_norm": 1.090136170387268,
      "learning_rate": 1.7603371801787505e-06,
      "loss": 1.3036,
      "step": 6576
    },
    {
      "epoch": 0.942128634866065,
      "grad_norm": 1.0465911626815796,
      "learning_rate": 1.7516803263286086e-06,
      "loss": 1.4323,
      "step": 6577
    },
    {
      "epoch": 0.9422718808193669,
      "grad_norm": 0.8737765550613403,
      "learning_rate": 1.743044622910417e-06,
      "loss": 1.4544,
      "step": 6578
    },
    {
      "epoch": 0.9424151267726687,
      "grad_norm": 1.2587758302688599,
      "learning_rate": 1.734430071783244e-06,
      "loss": 1.4904,
      "step": 6579
    },
    {
      "epoch": 0.9425583727259705,
      "grad_norm": 1.3229172229766846,
      "learning_rate": 1.7258366748015842e-06,
      "loss": 1.3992,
      "step": 6580
    },
    {
      "epoch": 0.9427016186792723,
      "grad_norm": 1.1931192874908447,
      "learning_rate": 1.7172644338153686e-06,
      "loss": 1.4733,
      "step": 6581
    },
    {
      "epoch": 0.9428448646325741,
      "grad_norm": 1.0861363410949707,
      "learning_rate": 1.7087133506699992e-06,
      "loss": 1.329,
      "step": 6582
    },
    {
      "epoch": 0.942988110585876,
      "grad_norm": 1.0513147115707397,
      "learning_rate": 1.7001834272063255e-06,
      "loss": 1.4861,
      "step": 6583
    },
    {
      "epoch": 0.9431313565391778,
      "grad_norm": 1.1442103385925293,
      "learning_rate": 1.6916746652606119e-06,
      "loss": 1.5508,
      "step": 6584
    },
    {
      "epoch": 0.9432746024924796,
      "grad_norm": 1.0573018789291382,
      "learning_rate": 1.6831870666645822e-06,
      "loss": 1.2921,
      "step": 6585
    },
    {
      "epoch": 0.9434178484457814,
      "grad_norm": 1.0331250429153442,
      "learning_rate": 1.6747206332454191e-06,
      "loss": 1.3437,
      "step": 6586
    },
    {
      "epoch": 0.9435610943990832,
      "grad_norm": 1.190805196762085,
      "learning_rate": 1.6662753668257314e-06,
      "loss": 1.2865,
      "step": 6587
    },
    {
      "epoch": 0.943704340352385,
      "grad_norm": 1.1967812776565552,
      "learning_rate": 1.657851269223587e-06,
      "loss": 1.2588,
      "step": 6588
    },
    {
      "epoch": 0.9438475863056869,
      "grad_norm": 1.2354650497436523,
      "learning_rate": 1.6494483422524466e-06,
      "loss": 1.2766,
      "step": 6589
    },
    {
      "epoch": 0.9439908322589887,
      "grad_norm": 0.9557923674583435,
      "learning_rate": 1.641066587721296e-06,
      "loss": 1.193,
      "step": 6590
    },
    {
      "epoch": 0.9441340782122905,
      "grad_norm": 1.3335047960281372,
      "learning_rate": 1.6327060074345147e-06,
      "loss": 1.4057,
      "step": 6591
    },
    {
      "epoch": 0.9442773241655923,
      "grad_norm": 1.118815541267395,
      "learning_rate": 1.624366603191907e-06,
      "loss": 1.1424,
      "step": 6592
    },
    {
      "epoch": 0.9444205701188941,
      "grad_norm": 0.9733049273490906,
      "learning_rate": 1.616048376788748e-06,
      "loss": 1.402,
      "step": 6593
    },
    {
      "epoch": 0.944563816072196,
      "grad_norm": 1.3634076118469238,
      "learning_rate": 1.6077513300157499e-06,
      "loss": 1.3282,
      "step": 6594
    },
    {
      "epoch": 0.9447070620254978,
      "grad_norm": 1.0222326517105103,
      "learning_rate": 1.5994754646590615e-06,
      "loss": 1.2558,
      "step": 6595
    },
    {
      "epoch": 0.9448503079787997,
      "grad_norm": 0.9995129704475403,
      "learning_rate": 1.591220782500269e-06,
      "loss": 1.4475,
      "step": 6596
    },
    {
      "epoch": 0.9449935539321014,
      "grad_norm": 1.0908689498901367,
      "learning_rate": 1.5829872853163952e-06,
      "loss": 1.3801,
      "step": 6597
    },
    {
      "epoch": 0.9451367998854032,
      "grad_norm": 1.081181287765503,
      "learning_rate": 1.5747749748799334e-06,
      "loss": 1.2232,
      "step": 6598
    },
    {
      "epoch": 0.9452800458387051,
      "grad_norm": 0.9398199319839478,
      "learning_rate": 1.5665838529587695e-06,
      "loss": 1.3441,
      "step": 6599
    },
    {
      "epoch": 0.9454232917920069,
      "grad_norm": 1.1726338863372803,
      "learning_rate": 1.5584139213162374e-06,
      "loss": 1.29,
      "step": 6600
    },
    {
      "epoch": 0.9455665377453087,
      "grad_norm": 0.967519223690033,
      "learning_rate": 1.5502651817111524e-06,
      "loss": 1.5401,
      "step": 6601
    },
    {
      "epoch": 0.9457097836986105,
      "grad_norm": 1.0717920064926147,
      "learning_rate": 1.5421376358977224e-06,
      "loss": 1.4192,
      "step": 6602
    },
    {
      "epoch": 0.9458530296519123,
      "grad_norm": 1.906010389328003,
      "learning_rate": 1.5340312856255922e-06,
      "loss": 1.2618,
      "step": 6603
    },
    {
      "epoch": 0.9459962756052142,
      "grad_norm": 1.0776267051696777,
      "learning_rate": 1.5259461326398772e-06,
      "loss": 1.5682,
      "step": 6604
    },
    {
      "epoch": 0.946139521558516,
      "grad_norm": 1.0738922357559204,
      "learning_rate": 1.517882178681107e-06,
      "loss": 1.3782,
      "step": 6605
    },
    {
      "epoch": 0.9462827675118178,
      "grad_norm": 0.9170629978179932,
      "learning_rate": 1.50983942548526e-06,
      "loss": 1.3167,
      "step": 6606
    },
    {
      "epoch": 0.9464260134651196,
      "grad_norm": 1.1050288677215576,
      "learning_rate": 1.5018178747837174e-06,
      "loss": 1.2376,
      "step": 6607
    },
    {
      "epoch": 0.9465692594184214,
      "grad_norm": 1.2266547679901123,
      "learning_rate": 1.4938175283033319e-06,
      "loss": 1.3362,
      "step": 6608
    },
    {
      "epoch": 0.9467125053717232,
      "grad_norm": 1.1739236116409302,
      "learning_rate": 1.4858383877664029e-06,
      "loss": 1.367,
      "step": 6609
    },
    {
      "epoch": 0.9468557513250251,
      "grad_norm": 0.9569011330604553,
      "learning_rate": 1.477880454890601e-06,
      "loss": 1.3771,
      "step": 6610
    },
    {
      "epoch": 0.9469989972783269,
      "grad_norm": 1.0979208946228027,
      "learning_rate": 1.4699437313891007e-06,
      "loss": 1.439,
      "step": 6611
    },
    {
      "epoch": 0.9471422432316287,
      "grad_norm": 1.0229387283325195,
      "learning_rate": 1.4620282189704793e-06,
      "loss": 1.5583,
      "step": 6612
    },
    {
      "epoch": 0.9472854891849305,
      "grad_norm": 0.9753866195678711,
      "learning_rate": 1.4541339193387404e-06,
      "loss": 1.2007,
      "step": 6613
    },
    {
      "epoch": 0.9474287351382323,
      "grad_norm": 1.0789514780044556,
      "learning_rate": 1.446260834193336e-06,
      "loss": 1.5225,
      "step": 6614
    },
    {
      "epoch": 0.9475719810915342,
      "grad_norm": 1.0734541416168213,
      "learning_rate": 1.4384089652291543e-06,
      "loss": 1.4904,
      "step": 6615
    },
    {
      "epoch": 0.947715227044836,
      "grad_norm": 1.1250686645507812,
      "learning_rate": 1.4305783141364992e-06,
      "loss": 1.4637,
      "step": 6616
    },
    {
      "epoch": 0.9478584729981379,
      "grad_norm": 1.107622504234314,
      "learning_rate": 1.422768882601122e-06,
      "loss": 1.4989,
      "step": 6617
    },
    {
      "epoch": 0.9480017189514396,
      "grad_norm": 0.9898762702941895,
      "learning_rate": 1.4149806723041892e-06,
      "loss": 1.3872,
      "step": 6618
    },
    {
      "epoch": 0.9481449649047414,
      "grad_norm": 1.1569151878356934,
      "learning_rate": 1.407213684922315e-06,
      "loss": 1.3477,
      "step": 6619
    },
    {
      "epoch": 0.9482882108580433,
      "grad_norm": 1.061189889907837,
      "learning_rate": 1.3994679221275509e-06,
      "loss": 1.3899,
      "step": 6620
    },
    {
      "epoch": 0.9484314568113451,
      "grad_norm": 1.0550694465637207,
      "learning_rate": 1.391743385587363e-06,
      "loss": 1.4333,
      "step": 6621
    },
    {
      "epoch": 0.948574702764647,
      "grad_norm": 1.4299442768096924,
      "learning_rate": 1.3840400769646322e-06,
      "loss": 1.3545,
      "step": 6622
    },
    {
      "epoch": 0.9487179487179487,
      "grad_norm": 1.113505482673645,
      "learning_rate": 1.376357997917721e-06,
      "loss": 1.2025,
      "step": 6623
    },
    {
      "epoch": 0.9488611946712505,
      "grad_norm": 0.9943010210990906,
      "learning_rate": 1.3686971501003621e-06,
      "loss": 1.4224,
      "step": 6624
    },
    {
      "epoch": 0.9490044406245524,
      "grad_norm": 0.9991193413734436,
      "learning_rate": 1.3610575351617693e-06,
      "loss": 1.2803,
      "step": 6625
    },
    {
      "epoch": 0.9491476865778542,
      "grad_norm": 1.141183853149414,
      "learning_rate": 1.3534391547465608e-06,
      "loss": 1.299,
      "step": 6626
    },
    {
      "epoch": 0.949290932531156,
      "grad_norm": 1.2125797271728516,
      "learning_rate": 1.34584201049478e-06,
      "loss": 1.4634,
      "step": 6627
    },
    {
      "epoch": 0.9494341784844578,
      "grad_norm": 1.142867088317871,
      "learning_rate": 1.3382661040418964e-06,
      "loss": 1.4659,
      "step": 6628
    },
    {
      "epoch": 0.9495774244377596,
      "grad_norm": 1.1054949760437012,
      "learning_rate": 1.3307114370188057e-06,
      "loss": 1.3877,
      "step": 6629
    },
    {
      "epoch": 0.9497206703910615,
      "grad_norm": 1.033025860786438,
      "learning_rate": 1.3231780110518844e-06,
      "loss": 1.5872,
      "step": 6630
    },
    {
      "epoch": 0.9498639163443633,
      "grad_norm": 0.9341566562652588,
      "learning_rate": 1.3156658277628463e-06,
      "loss": 1.3833,
      "step": 6631
    },
    {
      "epoch": 0.9500071622976651,
      "grad_norm": 1.021216630935669,
      "learning_rate": 1.3081748887689094e-06,
      "loss": 1.337,
      "step": 6632
    },
    {
      "epoch": 0.9501504082509669,
      "grad_norm": 0.95775306224823,
      "learning_rate": 1.3007051956826611e-06,
      "loss": 1.3069,
      "step": 6633
    },
    {
      "epoch": 0.9502936542042687,
      "grad_norm": 0.963055431842804,
      "learning_rate": 1.2932567501121707e-06,
      "loss": 1.4067,
      "step": 6634
    },
    {
      "epoch": 0.9504369001575705,
      "grad_norm": 1.080165147781372,
      "learning_rate": 1.2858295536608778e-06,
      "loss": 1.3757,
      "step": 6635
    },
    {
      "epoch": 0.9505801461108724,
      "grad_norm": 0.9608297944068909,
      "learning_rate": 1.2784236079276924e-06,
      "loss": 1.5086,
      "step": 6636
    },
    {
      "epoch": 0.9507233920641742,
      "grad_norm": 1.15897798538208,
      "learning_rate": 1.271038914506928e-06,
      "loss": 1.3479,
      "step": 6637
    },
    {
      "epoch": 0.950866638017476,
      "grad_norm": 1.2203125953674316,
      "learning_rate": 1.263675474988324e-06,
      "loss": 1.4616,
      "step": 6638
    },
    {
      "epoch": 0.9510098839707778,
      "grad_norm": 1.0418373346328735,
      "learning_rate": 1.2563332909570346e-06,
      "loss": 1.5018,
      "step": 6639
    },
    {
      "epoch": 0.9511531299240796,
      "grad_norm": 1.0435278415679932,
      "learning_rate": 1.2490123639936625e-06,
      "loss": 1.3865,
      "step": 6640
    },
    {
      "epoch": 0.9512963758773815,
      "grad_norm": 1.0926727056503296,
      "learning_rate": 1.2417126956742241e-06,
      "loss": 1.391,
      "step": 6641
    },
    {
      "epoch": 0.9514396218306833,
      "grad_norm": 0.9789139032363892,
      "learning_rate": 1.234434287570163e-06,
      "loss": 1.4996,
      "step": 6642
    },
    {
      "epoch": 0.9515828677839852,
      "grad_norm": 1.1839205026626587,
      "learning_rate": 1.2271771412483146e-06,
      "loss": 1.4096,
      "step": 6643
    },
    {
      "epoch": 0.9517261137372869,
      "grad_norm": 1.034575343132019,
      "learning_rate": 1.2199412582709956e-06,
      "loss": 1.3697,
      "step": 6644
    },
    {
      "epoch": 0.9518693596905887,
      "grad_norm": 0.9773863554000854,
      "learning_rate": 1.2127266401958826e-06,
      "loss": 1.4784,
      "step": 6645
    },
    {
      "epoch": 0.9520126056438906,
      "grad_norm": 1.2207194566726685,
      "learning_rate": 1.2055332885761327e-06,
      "loss": 1.4042,
      "step": 6646
    },
    {
      "epoch": 0.9521558515971924,
      "grad_norm": 1.030057430267334,
      "learning_rate": 1.1983612049602744e-06,
      "loss": 1.4045,
      "step": 6647
    },
    {
      "epoch": 0.9522990975504942,
      "grad_norm": 1.1055970191955566,
      "learning_rate": 1.1912103908922945e-06,
      "loss": 1.4057,
      "step": 6648
    },
    {
      "epoch": 0.952442343503796,
      "grad_norm": 1.0934616327285767,
      "learning_rate": 1.1840808479115727e-06,
      "loss": 1.4209,
      "step": 6649
    },
    {
      "epoch": 0.9525855894570978,
      "grad_norm": 1.080164909362793,
      "learning_rate": 1.176972577552915e-06,
      "loss": 1.4804,
      "step": 6650
    },
    {
      "epoch": 0.9527288354103997,
      "grad_norm": 0.9830626845359802,
      "learning_rate": 1.169885581346597e-06,
      "loss": 1.4227,
      "step": 6651
    },
    {
      "epoch": 0.9528720813637015,
      "grad_norm": 1.0269112586975098,
      "learning_rate": 1.1628198608182429e-06,
      "loss": 1.4547,
      "step": 6652
    },
    {
      "epoch": 0.9530153273170033,
      "grad_norm": 1.2169865369796753,
      "learning_rate": 1.1557754174889247e-06,
      "loss": 1.2574,
      "step": 6653
    },
    {
      "epoch": 0.9531585732703051,
      "grad_norm": 1.085243582725525,
      "learning_rate": 1.1487522528751404e-06,
      "loss": 1.4427,
      "step": 6654
    },
    {
      "epoch": 0.9533018192236069,
      "grad_norm": 1.0422412157058716,
      "learning_rate": 1.1417503684888142e-06,
      "loss": 1.3612,
      "step": 6655
    },
    {
      "epoch": 0.9534450651769087,
      "grad_norm": 0.9516494870185852,
      "learning_rate": 1.1347697658372847e-06,
      "loss": 1.3363,
      "step": 6656
    },
    {
      "epoch": 0.9535883111302106,
      "grad_norm": 0.9706204533576965,
      "learning_rate": 1.127810446423261e-06,
      "loss": 1.3554,
      "step": 6657
    },
    {
      "epoch": 0.9537315570835124,
      "grad_norm": 1.0176167488098145,
      "learning_rate": 1.1208724117449554e-06,
      "loss": 1.5602,
      "step": 6658
    },
    {
      "epoch": 0.9538748030368142,
      "grad_norm": 1.0628103017807007,
      "learning_rate": 1.1139556632959515e-06,
      "loss": 1.3962,
      "step": 6659
    },
    {
      "epoch": 0.954018048990116,
      "grad_norm": 0.9588499665260315,
      "learning_rate": 1.1070602025652355e-06,
      "loss": 1.4275,
      "step": 6660
    },
    {
      "epoch": 0.9541612949434178,
      "grad_norm": 1.1299690008163452,
      "learning_rate": 1.1001860310372314e-06,
      "loss": 1.3483,
      "step": 6661
    },
    {
      "epoch": 0.9543045408967197,
      "grad_norm": 0.9739903211593628,
      "learning_rate": 1.0933331501917998e-06,
      "loss": 1.5693,
      "step": 6662
    },
    {
      "epoch": 0.9544477868500215,
      "grad_norm": 0.9503809809684753,
      "learning_rate": 1.0865015615041607e-06,
      "loss": 1.4149,
      "step": 6663
    },
    {
      "epoch": 0.9545910328033234,
      "grad_norm": 1.2064324617385864,
      "learning_rate": 1.0796912664450265e-06,
      "loss": 1.4276,
      "step": 6664
    },
    {
      "epoch": 0.9547342787566251,
      "grad_norm": 0.9394369721412659,
      "learning_rate": 1.0729022664804467e-06,
      "loss": 1.5726,
      "step": 6665
    },
    {
      "epoch": 0.9548775247099269,
      "grad_norm": 1.1692692041397095,
      "learning_rate": 1.0661345630719522e-06,
      "loss": 1.3959,
      "step": 6666
    },
    {
      "epoch": 0.9550207706632288,
      "grad_norm": 0.9987855553627014,
      "learning_rate": 1.0593881576764664e-06,
      "loss": 1.3795,
      "step": 6667
    },
    {
      "epoch": 0.9551640166165306,
      "grad_norm": 1.0725371837615967,
      "learning_rate": 1.0526630517462833e-06,
      "loss": 1.55,
      "step": 6668
    },
    {
      "epoch": 0.9553072625698324,
      "grad_norm": 1.0539506673812866,
      "learning_rate": 1.045959246729189e-06,
      "loss": 1.315,
      "step": 6669
    },
    {
      "epoch": 0.9554505085231342,
      "grad_norm": 1.0855098962783813,
      "learning_rate": 1.0392767440683516e-06,
      "loss": 1.4658,
      "step": 6670
    },
    {
      "epoch": 0.955593754476436,
      "grad_norm": 1.0522102117538452,
      "learning_rate": 1.03261554520232e-06,
      "loss": 1.3525,
      "step": 6671
    },
    {
      "epoch": 0.9557370004297379,
      "grad_norm": 1.0024594068527222,
      "learning_rate": 1.0259756515651032e-06,
      "loss": 1.5479,
      "step": 6672
    },
    {
      "epoch": 0.9558802463830397,
      "grad_norm": 1.199232816696167,
      "learning_rate": 1.0193570645860905e-06,
      "loss": 1.2984,
      "step": 6673
    },
    {
      "epoch": 0.9560234923363415,
      "grad_norm": 0.950915515422821,
      "learning_rate": 1.0127597856901205e-06,
      "loss": 1.361,
      "step": 6674
    },
    {
      "epoch": 0.9561667382896433,
      "grad_norm": 1.133409023284912,
      "learning_rate": 1.0061838162974235e-06,
      "loss": 1.3439,
      "step": 6675
    },
    {
      "epoch": 0.9563099842429451,
      "grad_norm": 1.2114832401275635,
      "learning_rate": 9.996291578236228e-07,
      "loss": 1.2736,
      "step": 6676
    },
    {
      "epoch": 0.956453230196247,
      "grad_norm": 1.8214119672775269,
      "learning_rate": 9.930958116797895e-07,
      "loss": 1.5906,
      "step": 6677
    },
    {
      "epoch": 0.9565964761495488,
      "grad_norm": 1.2845548391342163,
      "learning_rate": 9.865837792723875e-07,
      "loss": 1.3198,
      "step": 6678
    },
    {
      "epoch": 0.9567397221028506,
      "grad_norm": 1.0146381855010986,
      "learning_rate": 9.800930620032843e-07,
      "loss": 1.5795,
      "step": 6679
    },
    {
      "epoch": 0.9568829680561524,
      "grad_norm": 1.0424240827560425,
      "learning_rate": 9.736236612697957e-07,
      "loss": 1.3348,
      "step": 6680
    },
    {
      "epoch": 0.9570262140094542,
      "grad_norm": 1.1131378412246704,
      "learning_rate": 9.671755784646075e-07,
      "loss": 1.2732,
      "step": 6681
    },
    {
      "epoch": 0.957169459962756,
      "grad_norm": 0.9339991211891174,
      "learning_rate": 9.607488149758425e-07,
      "loss": 1.2275,
      "step": 6682
    },
    {
      "epoch": 0.9573127059160579,
      "grad_norm": 1.1991435289382935,
      "learning_rate": 9.543433721870055e-07,
      "loss": 1.2605,
      "step": 6683
    },
    {
      "epoch": 0.9574559518693597,
      "grad_norm": 0.9194568395614624,
      "learning_rate": 9.479592514770486e-07,
      "loss": 1.3574,
      "step": 6684
    },
    {
      "epoch": 0.9575991978226616,
      "grad_norm": 1.0245776176452637,
      "learning_rate": 9.415964542203059e-07,
      "loss": 1.3816,
      "step": 6685
    },
    {
      "epoch": 0.9577424437759633,
      "grad_norm": 0.9724869728088379,
      "learning_rate": 9.352549817865263e-07,
      "loss": 1.5024,
      "step": 6686
    },
    {
      "epoch": 0.9578856897292651,
      "grad_norm": 1.0989818572998047,
      "learning_rate": 9.289348355408734e-07,
      "loss": 1.3091,
      "step": 6687
    },
    {
      "epoch": 0.958028935682567,
      "grad_norm": 1.2178518772125244,
      "learning_rate": 9.226360168439363e-07,
      "loss": 1.4937,
      "step": 6688
    },
    {
      "epoch": 0.9581721816358688,
      "grad_norm": 0.9448843002319336,
      "learning_rate": 9.16358527051675e-07,
      "loss": 1.5251,
      "step": 6689
    },
    {
      "epoch": 0.9583154275891707,
      "grad_norm": 1.1422120332717896,
      "learning_rate": 9.101023675154751e-07,
      "loss": 1.382,
      "step": 6690
    },
    {
      "epoch": 0.9584586735424724,
      "grad_norm": 1.2850912809371948,
      "learning_rate": 9.038675395821594e-07,
      "loss": 1.3588,
      "step": 6691
    },
    {
      "epoch": 0.9586019194957742,
      "grad_norm": 0.9747809767723083,
      "learning_rate": 8.976540445938986e-07,
      "loss": 1.3269,
      "step": 6692
    },
    {
      "epoch": 0.9587451654490761,
      "grad_norm": 0.9804068803787231,
      "learning_rate": 8.914618838883226e-07,
      "loss": 1.2533,
      "step": 6693
    },
    {
      "epoch": 0.9588884114023779,
      "grad_norm": 1.092737078666687,
      "learning_rate": 8.852910587984542e-07,
      "loss": 1.7149,
      "step": 6694
    },
    {
      "epoch": 0.9590316573556797,
      "grad_norm": 0.9056750535964966,
      "learning_rate": 8.791415706526973e-07,
      "loss": 1.4214,
      "step": 6695
    },
    {
      "epoch": 0.9591749033089815,
      "grad_norm": 1.11558198928833,
      "learning_rate": 8.730134207749152e-07,
      "loss": 1.4036,
      "step": 6696
    },
    {
      "epoch": 0.9593181492622833,
      "grad_norm": 1.0779403448104858,
      "learning_rate": 8.669066104843304e-07,
      "loss": 1.2249,
      "step": 6697
    },
    {
      "epoch": 0.9594613952155852,
      "grad_norm": 1.0764225721359253,
      "learning_rate": 8.608211410955802e-07,
      "loss": 1.3402,
      "step": 6698
    },
    {
      "epoch": 0.959604641168887,
      "grad_norm": 1.0803821086883545,
      "learning_rate": 8.547570139187388e-07,
      "loss": 1.3921,
      "step": 6699
    },
    {
      "epoch": 0.9597478871221888,
      "grad_norm": 1.080828070640564,
      "learning_rate": 8.487142302592288e-07,
      "loss": 1.4322,
      "step": 6700
    },
    {
      "epoch": 0.9598911330754906,
      "grad_norm": 1.2060621976852417,
      "learning_rate": 8.426927914179428e-07,
      "loss": 1.5657,
      "step": 6701
    },
    {
      "epoch": 0.9600343790287924,
      "grad_norm": 1.0757522583007812,
      "learning_rate": 8.366926986911328e-07,
      "loss": 1.596,
      "step": 6702
    },
    {
      "epoch": 0.9601776249820942,
      "grad_norm": 1.2298121452331543,
      "learning_rate": 8.307139533704766e-07,
      "loss": 1.3426,
      "step": 6703
    },
    {
      "epoch": 0.9603208709353961,
      "grad_norm": 1.142180323600769,
      "learning_rate": 8.247565567430559e-07,
      "loss": 1.3139,
      "step": 6704
    },
    {
      "epoch": 0.9604641168886979,
      "grad_norm": 1.0267170667648315,
      "learning_rate": 8.188205100913337e-07,
      "loss": 1.3529,
      "step": 6705
    },
    {
      "epoch": 0.9606073628419998,
      "grad_norm": 1.1094212532043457,
      "learning_rate": 8.12905814693199e-07,
      "loss": 1.51,
      "step": 6706
    },
    {
      "epoch": 0.9607506087953015,
      "grad_norm": 1.110493540763855,
      "learning_rate": 8.07012471821944e-07,
      "loss": 1.4262,
      "step": 6707
    },
    {
      "epoch": 0.9608938547486033,
      "grad_norm": 1.1183924674987793,
      "learning_rate": 8.011404827462654e-07,
      "loss": 1.3114,
      "step": 6708
    },
    {
      "epoch": 0.9610371007019052,
      "grad_norm": 0.925645112991333,
      "learning_rate": 7.952898487302518e-07,
      "loss": 1.43,
      "step": 6709
    },
    {
      "epoch": 0.961180346655207,
      "grad_norm": 1.1463572978973389,
      "learning_rate": 7.894605710333847e-07,
      "loss": 1.4017,
      "step": 6710
    },
    {
      "epoch": 0.9613235926085089,
      "grad_norm": 1.06252920627594,
      "learning_rate": 7.836526509105824e-07,
      "loss": 1.4132,
      "step": 6711
    },
    {
      "epoch": 0.9614668385618106,
      "grad_norm": 0.9852263331413269,
      "learning_rate": 7.778660896121448e-07,
      "loss": 1.6288,
      "step": 6712
    },
    {
      "epoch": 0.9616100845151124,
      "grad_norm": 1.1114845275878906,
      "learning_rate": 7.721008883837755e-07,
      "loss": 1.3084,
      "step": 6713
    },
    {
      "epoch": 0.9617533304684143,
      "grad_norm": 1.4312448501586914,
      "learning_rate": 7.663570484665705e-07,
      "loss": 1.387,
      "step": 6714
    },
    {
      "epoch": 0.9618965764217161,
      "grad_norm": 1.0424330234527588,
      "learning_rate": 7.606345710970409e-07,
      "loss": 1.5029,
      "step": 6715
    },
    {
      "epoch": 0.962039822375018,
      "grad_norm": 1.2997428178787231,
      "learning_rate": 7.549334575070899e-07,
      "loss": 1.3587,
      "step": 6716
    },
    {
      "epoch": 0.9621830683283197,
      "grad_norm": 0.85928875207901,
      "learning_rate": 7.492537089240471e-07,
      "loss": 1.4165,
      "step": 6717
    },
    {
      "epoch": 0.9623263142816215,
      "grad_norm": 1.2010200023651123,
      "learning_rate": 7.435953265705897e-07,
      "loss": 1.5883,
      "step": 6718
    },
    {
      "epoch": 0.9624695602349234,
      "grad_norm": 1.2231497764587402,
      "learning_rate": 7.379583116648436e-07,
      "loss": 1.2676,
      "step": 6719
    },
    {
      "epoch": 0.9626128061882252,
      "grad_norm": 1.2439926862716675,
      "learning_rate": 7.32342665420338e-07,
      "loss": 1.4369,
      "step": 6720
    },
    {
      "epoch": 0.962756052141527,
      "grad_norm": 1.0502089262008667,
      "learning_rate": 7.267483890459503e-07,
      "loss": 1.2661,
      "step": 6721
    },
    {
      "epoch": 0.9628992980948288,
      "grad_norm": 1.2538540363311768,
      "learning_rate": 7.21175483745995e-07,
      "loss": 1.2551,
      "step": 6722
    },
    {
      "epoch": 0.9630425440481306,
      "grad_norm": 1.2815312147140503,
      "learning_rate": 7.156239507202011e-07,
      "loss": 1.3625,
      "step": 6723
    },
    {
      "epoch": 0.9631857900014325,
      "grad_norm": 0.9382988214492798,
      "learning_rate": 7.100937911636574e-07,
      "loss": 1.2515,
      "step": 6724
    },
    {
      "epoch": 0.9633290359547343,
      "grad_norm": 1.0944647789001465,
      "learning_rate": 7.04585006266878e-07,
      "loss": 1.529,
      "step": 6725
    },
    {
      "epoch": 0.9634722819080361,
      "grad_norm": 1.2149970531463623,
      "learning_rate": 6.990975972157699e-07,
      "loss": 1.31,
      "step": 6726
    },
    {
      "epoch": 0.963615527861338,
      "grad_norm": 0.974987268447876,
      "learning_rate": 6.936315651916325e-07,
      "loss": 1.4224,
      "step": 6727
    },
    {
      "epoch": 0.9637587738146397,
      "grad_norm": 1.0514013767242432,
      "learning_rate": 6.881869113711692e-07,
      "loss": 1.3358,
      "step": 6728
    },
    {
      "epoch": 0.9639020197679415,
      "grad_norm": 0.8235097527503967,
      "learning_rate": 6.827636369264645e-07,
      "loss": 1.6115,
      "step": 6729
    },
    {
      "epoch": 0.9640452657212434,
      "grad_norm": 1.0267082452774048,
      "learning_rate": 6.773617430250512e-07,
      "loss": 1.5101,
      "step": 6730
    },
    {
      "epoch": 0.9641885116745452,
      "grad_norm": 1.089717984199524,
      "learning_rate": 6.71981230829788e-07,
      "loss": 1.5461,
      "step": 6731
    },
    {
      "epoch": 0.9643317576278471,
      "grad_norm": 1.0367276668548584,
      "learning_rate": 6.666221014989815e-07,
      "loss": 1.5668,
      "step": 6732
    },
    {
      "epoch": 0.9644750035811488,
      "grad_norm": 1.1212315559387207,
      "learning_rate": 6.612843561863092e-07,
      "loss": 1.2067,
      "step": 6733
    },
    {
      "epoch": 0.9646182495344506,
      "grad_norm": 1.1231820583343506,
      "learning_rate": 6.559679960408738e-07,
      "loss": 1.4505,
      "step": 6734
    },
    {
      "epoch": 0.9647614954877525,
      "grad_norm": 1.0434908866882324,
      "learning_rate": 6.506730222071488e-07,
      "loss": 1.4701,
      "step": 6735
    },
    {
      "epoch": 0.9649047414410543,
      "grad_norm": 0.9652475714683533,
      "learning_rate": 6.453994358249893e-07,
      "loss": 1.4284,
      "step": 6736
    },
    {
      "epoch": 0.9650479873943562,
      "grad_norm": 0.9612234830856323,
      "learning_rate": 6.401472380297091e-07,
      "loss": 1.4568,
      "step": 6737
    },
    {
      "epoch": 0.9651912333476579,
      "grad_norm": 0.8843510150909424,
      "learning_rate": 6.349164299519483e-07,
      "loss": 1.4741,
      "step": 6738
    },
    {
      "epoch": 0.9653344793009597,
      "grad_norm": 1.0622419118881226,
      "learning_rate": 6.297070127177617e-07,
      "loss": 1.2744,
      "step": 6739
    },
    {
      "epoch": 0.9654777252542616,
      "grad_norm": 0.9046955108642578,
      "learning_rate": 6.245189874486301e-07,
      "loss": 1.3447,
      "step": 6740
    },
    {
      "epoch": 0.9656209712075634,
      "grad_norm": 1.0797584056854248,
      "learning_rate": 6.193523552614044e-07,
      "loss": 1.3195,
      "step": 6741
    },
    {
      "epoch": 0.9657642171608652,
      "grad_norm": 1.1649394035339355,
      "learning_rate": 6.142071172683284e-07,
      "loss": 1.3349,
      "step": 6742
    },
    {
      "epoch": 0.965907463114167,
      "grad_norm": 0.915144145488739,
      "learning_rate": 6.09083274577038e-07,
      "loss": 1.1392,
      "step": 6743
    },
    {
      "epoch": 0.9660507090674688,
      "grad_norm": 1.137065052986145,
      "learning_rate": 6.039808282905735e-07,
      "loss": 1.3218,
      "step": 6744
    },
    {
      "epoch": 0.9661939550207707,
      "grad_norm": 1.0155001878738403,
      "learning_rate": 5.98899779507367e-07,
      "loss": 1.6545,
      "step": 6745
    },
    {
      "epoch": 0.9663372009740725,
      "grad_norm": 1.5141016244888306,
      "learning_rate": 5.938401293212547e-07,
      "loss": 1.2473,
      "step": 6746
    },
    {
      "epoch": 0.9664804469273743,
      "grad_norm": 1.0211079120635986,
      "learning_rate": 5.888018788214322e-07,
      "loss": 1.4887,
      "step": 6747
    },
    {
      "epoch": 0.9666236928806761,
      "grad_norm": 0.946195125579834,
      "learning_rate": 5.837850290925206e-07,
      "loss": 1.3365,
      "step": 6748
    },
    {
      "epoch": 0.9667669388339779,
      "grad_norm": 1.2444173097610474,
      "learning_rate": 5.787895812145227e-07,
      "loss": 1.4039,
      "step": 6749
    },
    {
      "epoch": 0.9669101847872797,
      "grad_norm": 1.1345558166503906,
      "learning_rate": 5.738155362628339e-07,
      "loss": 1.5199,
      "step": 6750
    },
    {
      "epoch": 0.9670534307405816,
      "grad_norm": 1.2315428256988525,
      "learning_rate": 5.68862895308242e-07,
      "loss": 1.2671,
      "step": 6751
    },
    {
      "epoch": 0.9671966766938834,
      "grad_norm": 1.031372308731079,
      "learning_rate": 5.639316594169386e-07,
      "loss": 1.6054,
      "step": 6752
    },
    {
      "epoch": 0.9673399226471853,
      "grad_norm": 0.985232412815094,
      "learning_rate": 5.590218296504857e-07,
      "loss": 1.5288,
      "step": 6753
    },
    {
      "epoch": 0.967483168600487,
      "grad_norm": 1.05845308303833,
      "learning_rate": 5.541334070658488e-07,
      "loss": 1.5907,
      "step": 6754
    },
    {
      "epoch": 0.9676264145537888,
      "grad_norm": 1.0177547931671143,
      "learning_rate": 5.492663927153863e-07,
      "loss": 1.2761,
      "step": 6755
    },
    {
      "epoch": 0.9677696605070907,
      "grad_norm": 1.0305039882659912,
      "learning_rate": 5.444207876468488e-07,
      "loss": 1.4256,
      "step": 6756
    },
    {
      "epoch": 0.9679129064603925,
      "grad_norm": 1.1531683206558228,
      "learning_rate": 5.395965929033686e-07,
      "loss": 1.4734,
      "step": 6757
    },
    {
      "epoch": 0.9680561524136944,
      "grad_norm": 1.1651809215545654,
      "learning_rate": 5.347938095234705e-07,
      "loss": 1.4751,
      "step": 6758
    },
    {
      "epoch": 0.9681993983669961,
      "grad_norm": 1.2625993490219116,
      "learning_rate": 5.300124385410943e-07,
      "loss": 1.3226,
      "step": 6759
    },
    {
      "epoch": 0.9683426443202979,
      "grad_norm": 1.1251366138458252,
      "learning_rate": 5.252524809855386e-07,
      "loss": 1.423,
      "step": 6760
    },
    {
      "epoch": 0.9684858902735998,
      "grad_norm": 1.1552636623382568,
      "learning_rate": 5.20513937881506e-07,
      "loss": 1.4261,
      "step": 6761
    },
    {
      "epoch": 0.9686291362269016,
      "grad_norm": 0.8800671696662903,
      "learning_rate": 5.157968102490918e-07,
      "loss": 1.365,
      "step": 6762
    },
    {
      "epoch": 0.9687723821802035,
      "grad_norm": 0.9991808533668518,
      "learning_rate": 5.111010991037613e-07,
      "loss": 1.4096,
      "step": 6763
    },
    {
      "epoch": 0.9689156281335052,
      "grad_norm": 0.8865735530853271,
      "learning_rate": 5.064268054564059e-07,
      "loss": 1.4666,
      "step": 6764
    },
    {
      "epoch": 0.969058874086807,
      "grad_norm": 0.9733909964561462,
      "learning_rate": 5.017739303132763e-07,
      "loss": 1.3083,
      "step": 6765
    },
    {
      "epoch": 0.9692021200401089,
      "grad_norm": 1.055842638015747,
      "learning_rate": 4.971424746760156e-07,
      "loss": 1.3502,
      "step": 6766
    },
    {
      "epoch": 0.9693453659934107,
      "grad_norm": 1.0827281475067139,
      "learning_rate": 4.925324395416709e-07,
      "loss": 1.6086,
      "step": 6767
    },
    {
      "epoch": 0.9694886119467125,
      "grad_norm": 1.003227949142456,
      "learning_rate": 4.879438259026592e-07,
      "loss": 1.3716,
      "step": 6768
    },
    {
      "epoch": 0.9696318579000143,
      "grad_norm": 1.0578702688217163,
      "learning_rate": 4.833766347468016e-07,
      "loss": 1.4366,
      "step": 6769
    },
    {
      "epoch": 0.9697751038533161,
      "grad_norm": 0.9403130412101746,
      "learning_rate": 4.788308670573005e-07,
      "loss": 1.256,
      "step": 6770
    },
    {
      "epoch": 0.969918349806618,
      "grad_norm": 1.1767295598983765,
      "learning_rate": 4.743065238127509e-07,
      "loss": 1.3124,
      "step": 6771
    },
    {
      "epoch": 0.9700615957599198,
      "grad_norm": 1.0601083040237427,
      "learning_rate": 4.698036059871291e-07,
      "loss": 1.5397,
      "step": 6772
    },
    {
      "epoch": 0.9702048417132216,
      "grad_norm": 0.9455981254577637,
      "learning_rate": 4.6532211454979324e-07,
      "loss": 1.4451,
      "step": 6773
    },
    {
      "epoch": 0.9703480876665235,
      "grad_norm": 0.8330476880073547,
      "learning_rate": 4.608620504655048e-07,
      "loss": 1.4231,
      "step": 6774
    },
    {
      "epoch": 0.9704913336198252,
      "grad_norm": 1.0068690776824951,
      "learning_rate": 4.5642341469441794e-07,
      "loss": 1.3338,
      "step": 6775
    },
    {
      "epoch": 0.970634579573127,
      "grad_norm": 0.9365617632865906,
      "learning_rate": 4.520062081920351e-07,
      "loss": 1.4194,
      "step": 6776
    },
    {
      "epoch": 0.9707778255264289,
      "grad_norm": 0.9804052710533142,
      "learning_rate": 4.4761043190929553e-07,
      "loss": 1.4141,
      "step": 6777
    },
    {
      "epoch": 0.9709210714797307,
      "grad_norm": 1.0885045528411865,
      "learning_rate": 4.4323608679248676e-07,
      "loss": 1.4991,
      "step": 6778
    },
    {
      "epoch": 0.9710643174330326,
      "grad_norm": 1.026187777519226,
      "learning_rate": 4.388831737832999e-07,
      "loss": 1.4373,
      "step": 6779
    },
    {
      "epoch": 0.9712075633863343,
      "grad_norm": 1.1540892124176025,
      "learning_rate": 4.345516938188188e-07,
      "loss": 1.2668,
      "step": 6780
    },
    {
      "epoch": 0.9713508093396361,
      "grad_norm": 1.0256913900375366,
      "learning_rate": 4.3024164783148636e-07,
      "loss": 1.4697,
      "step": 6781
    },
    {
      "epoch": 0.971494055292938,
      "grad_norm": 1.1743240356445312,
      "learning_rate": 4.2595303674916044e-07,
      "loss": 1.451,
      "step": 6782
    },
    {
      "epoch": 0.9716373012462398,
      "grad_norm": 1.140163779258728,
      "learning_rate": 4.216858614950692e-07,
      "loss": 1.5215,
      "step": 6783
    },
    {
      "epoch": 0.9717805471995417,
      "grad_norm": 0.940036416053772,
      "learning_rate": 4.174401229878333e-07,
      "loss": 1.4223,
      "step": 6784
    },
    {
      "epoch": 0.9719237931528434,
      "grad_norm": 0.931448221206665,
      "learning_rate": 4.1321582214145506e-07,
      "loss": 1.5661,
      "step": 6785
    },
    {
      "epoch": 0.9720670391061452,
      "grad_norm": 1.0401630401611328,
      "learning_rate": 4.090129598653181e-07,
      "loss": 1.3189,
      "step": 6786
    },
    {
      "epoch": 0.9722102850594471,
      "grad_norm": 1.2921048402786255,
      "learning_rate": 4.048315370641986e-07,
      "loss": 1.4964,
      "step": 6787
    },
    {
      "epoch": 0.9723535310127489,
      "grad_norm": 1.021125078201294,
      "learning_rate": 4.006715546382434e-07,
      "loss": 1.432,
      "step": 6788
    },
    {
      "epoch": 0.9724967769660507,
      "grad_norm": 1.0780339241027832,
      "learning_rate": 3.9653301348301364e-07,
      "loss": 1.3225,
      "step": 6789
    },
    {
      "epoch": 0.9726400229193525,
      "grad_norm": 1.312149167060852,
      "learning_rate": 3.92415914489408e-07,
      "loss": 1.4005,
      "step": 6790
    },
    {
      "epoch": 0.9727832688726543,
      "grad_norm": 1.1159489154815674,
      "learning_rate": 3.8832025854376176e-07,
      "loss": 1.3355,
      "step": 6791
    },
    {
      "epoch": 0.9729265148259562,
      "grad_norm": 0.9776840209960938,
      "learning_rate": 3.842460465277586e-07,
      "loss": 1.4045,
      "step": 6792
    },
    {
      "epoch": 0.973069760779258,
      "grad_norm": 1.1597959995269775,
      "learning_rate": 3.801932793184748e-07,
      "loss": 1.2606,
      "step": 6793
    },
    {
      "epoch": 0.9732130067325598,
      "grad_norm": 1.2329657077789307,
      "learning_rate": 3.7616195778836793e-07,
      "loss": 1.2383,
      "step": 6794
    },
    {
      "epoch": 0.9733562526858617,
      "grad_norm": 1.1108812093734741,
      "learning_rate": 3.721520828052771e-07,
      "loss": 1.4166,
      "step": 6795
    },
    {
      "epoch": 0.9734994986391634,
      "grad_norm": 0.9824594259262085,
      "learning_rate": 3.681636552324452e-07,
      "loss": 1.3938,
      "step": 6796
    },
    {
      "epoch": 0.9736427445924652,
      "grad_norm": 0.9686800241470337,
      "learning_rate": 3.6419667592847427e-07,
      "loss": 1.3513,
      "step": 6797
    },
    {
      "epoch": 0.9737859905457671,
      "grad_norm": 1.0292178392410278,
      "learning_rate": 3.6025114574734785e-07,
      "loss": 1.2447,
      "step": 6798
    },
    {
      "epoch": 0.9739292364990689,
      "grad_norm": 0.9551668167114258,
      "learning_rate": 3.563270655384532e-07,
      "loss": 1.3317,
      "step": 6799
    },
    {
      "epoch": 0.9740724824523708,
      "grad_norm": 0.9626041054725647,
      "learning_rate": 3.5242443614654784e-07,
      "loss": 1.5247,
      "step": 6800
    },
    {
      "epoch": 0.9742157284056725,
      "grad_norm": 1.173126220703125,
      "learning_rate": 3.4854325841175985e-07,
      "loss": 1.414,
      "step": 6801
    },
    {
      "epoch": 0.9743589743589743,
      "grad_norm": 1.0089622735977173,
      "learning_rate": 3.446835331696208e-07,
      "loss": 1.3249,
      "step": 6802
    },
    {
      "epoch": 0.9745022203122762,
      "grad_norm": 1.021348476409912,
      "learning_rate": 3.4084526125103267e-07,
      "loss": 1.5224,
      "step": 6803
    },
    {
      "epoch": 0.974645466265578,
      "grad_norm": 0.8531768918037415,
      "learning_rate": 3.370284434822679e-07,
      "loss": 1.4358,
      "step": 6804
    },
    {
      "epoch": 0.9747887122188799,
      "grad_norm": 1.0392380952835083,
      "learning_rate": 3.332330806850137e-07,
      "loss": 1.5288,
      "step": 6805
    },
    {
      "epoch": 0.9749319581721816,
      "grad_norm": 1.064247965812683,
      "learning_rate": 3.294591736763164e-07,
      "loss": 1.3342,
      "step": 6806
    },
    {
      "epoch": 0.9750752041254834,
      "grad_norm": 1.1191128492355347,
      "learning_rate": 3.2570672326858175e-07,
      "loss": 1.4196,
      "step": 6807
    },
    {
      "epoch": 0.9752184500787853,
      "grad_norm": 1.041569471359253,
      "learning_rate": 3.219757302696302e-07,
      "loss": 1.4812,
      "step": 6808
    },
    {
      "epoch": 0.9753616960320871,
      "grad_norm": 1.1266881227493286,
      "learning_rate": 3.182661954826638e-07,
      "loss": 1.2498,
      "step": 6809
    },
    {
      "epoch": 0.975504941985389,
      "grad_norm": 1.055417776107788,
      "learning_rate": 3.1457811970624364e-07,
      "loss": 1.3558,
      "step": 6810
    },
    {
      "epoch": 0.9756481879386907,
      "grad_norm": 1.0740935802459717,
      "learning_rate": 3.1091150373433465e-07,
      "loss": 1.2596,
      "step": 6811
    },
    {
      "epoch": 0.9757914338919925,
      "grad_norm": 1.1849511861801147,
      "learning_rate": 3.072663483562388e-07,
      "loss": 1.2935,
      "step": 6812
    },
    {
      "epoch": 0.9759346798452944,
      "grad_norm": 1.1551735401153564,
      "learning_rate": 3.0364265435669503e-07,
      "loss": 1.3887,
      "step": 6813
    },
    {
      "epoch": 0.9760779257985962,
      "grad_norm": 1.2335840463638306,
      "learning_rate": 3.0004042251579047e-07,
      "loss": 1.4372,
      "step": 6814
    },
    {
      "epoch": 0.976221171751898,
      "grad_norm": 1.142035961151123,
      "learning_rate": 2.9645965360898255e-07,
      "loss": 1.3081,
      "step": 6815
    },
    {
      "epoch": 0.9763644177051999,
      "grad_norm": 0.942271888256073,
      "learning_rate": 2.9290034840713245e-07,
      "loss": 1.4419,
      "step": 6816
    },
    {
      "epoch": 0.9765076636585016,
      "grad_norm": 0.9963545203208923,
      "learning_rate": 2.8936250767647167e-07,
      "loss": 1.4326,
      "step": 6817
    },
    {
      "epoch": 0.9766509096118035,
      "grad_norm": 1.1144119501113892,
      "learning_rate": 2.8584613217861324e-07,
      "loss": 1.3214,
      "step": 6818
    },
    {
      "epoch": 0.9767941555651053,
      "grad_norm": 0.9936862587928772,
      "learning_rate": 2.8235122267052936e-07,
      "loss": 1.3324,
      "step": 6819
    },
    {
      "epoch": 0.9769374015184071,
      "grad_norm": 1.0080571174621582,
      "learning_rate": 2.7887777990460716e-07,
      "loss": 1.343,
      "step": 6820
    },
    {
      "epoch": 0.977080647471709,
      "grad_norm": 1.0317273139953613,
      "learning_rate": 2.754258046285818e-07,
      "loss": 1.3758,
      "step": 6821
    },
    {
      "epoch": 0.9772238934250107,
      "grad_norm": 1.3107060194015503,
      "learning_rate": 2.719952975855811e-07,
      "loss": 1.3296,
      "step": 6822
    },
    {
      "epoch": 0.9773671393783125,
      "grad_norm": 1.0547453165054321,
      "learning_rate": 2.685862595141142e-07,
      "loss": 1.4308,
      "step": 6823
    },
    {
      "epoch": 0.9775103853316144,
      "grad_norm": 1.2521612644195557,
      "learning_rate": 2.6519869114804975e-07,
      "loss": 1.4205,
      "step": 6824
    },
    {
      "epoch": 0.9776536312849162,
      "grad_norm": 1.0133212804794312,
      "learning_rate": 2.618325932166488e-07,
      "loss": 1.3525,
      "step": 6825
    },
    {
      "epoch": 0.9777968772382181,
      "grad_norm": 1.0904204845428467,
      "learning_rate": 2.58487966444565e-07,
      "loss": 1.4972,
      "step": 6826
    },
    {
      "epoch": 0.9779401231915198,
      "grad_norm": 1.1512699127197266,
      "learning_rate": 2.5516481155180014e-07,
      "loss": 1.3731,
      "step": 6827
    },
    {
      "epoch": 0.9780833691448216,
      "grad_norm": 1.0409411191940308,
      "learning_rate": 2.518631292537488e-07,
      "loss": 1.4045,
      "step": 6828
    },
    {
      "epoch": 0.9782266150981235,
      "grad_norm": 1.3136940002441406,
      "learning_rate": 2.485829202611756e-07,
      "loss": 1.3718,
      "step": 6829
    },
    {
      "epoch": 0.9783698610514253,
      "grad_norm": 1.012670874595642,
      "learning_rate": 2.453241852802379e-07,
      "loss": 1.4019,
      "step": 6830
    },
    {
      "epoch": 0.9785131070047272,
      "grad_norm": 1.157504916191101,
      "learning_rate": 2.4208692501246354e-07,
      "loss": 1.2291,
      "step": 6831
    },
    {
      "epoch": 0.9786563529580289,
      "grad_norm": 0.9912055730819702,
      "learning_rate": 2.3887114015475056e-07,
      "loss": 1.3806,
      "step": 6832
    },
    {
      "epoch": 0.9787995989113307,
      "grad_norm": 1.1572833061218262,
      "learning_rate": 2.3567683139936735e-07,
      "loss": 1.6286,
      "step": 6833
    },
    {
      "epoch": 0.9789428448646326,
      "grad_norm": 1.0619378089904785,
      "learning_rate": 2.3250399943398614e-07,
      "loss": 1.3827,
      "step": 6834
    },
    {
      "epoch": 0.9790860908179344,
      "grad_norm": 1.1415191888809204,
      "learning_rate": 2.2935264494162724e-07,
      "loss": 1.4628,
      "step": 6835
    },
    {
      "epoch": 0.9792293367712362,
      "grad_norm": 1.050657868385315,
      "learning_rate": 2.2622276860070346e-07,
      "loss": 1.4915,
      "step": 6836
    },
    {
      "epoch": 0.979372582724538,
      "grad_norm": 1.2451061010360718,
      "learning_rate": 2.231143710849981e-07,
      "loss": 1.4248,
      "step": 6837
    },
    {
      "epoch": 0.9795158286778398,
      "grad_norm": 0.9803173542022705,
      "learning_rate": 2.20027453063687e-07,
      "loss": 1.4441,
      "step": 6838
    },
    {
      "epoch": 0.9796590746311417,
      "grad_norm": 1.0542032718658447,
      "learning_rate": 2.1696201520128302e-07,
      "loss": 1.3416,
      "step": 6839
    },
    {
      "epoch": 0.9798023205844435,
      "grad_norm": 1.1661409139633179,
      "learning_rate": 2.1391805815771382e-07,
      "loss": 1.2723,
      "step": 6840
    },
    {
      "epoch": 0.9799455665377453,
      "grad_norm": 1.1447373628616333,
      "learning_rate": 2.1089558258826637e-07,
      "loss": 1.3319,
      "step": 6841
    },
    {
      "epoch": 0.9800888124910472,
      "grad_norm": 0.9953629374504089,
      "learning_rate": 2.0789458914359793e-07,
      "loss": 1.3527,
      "step": 6842
    },
    {
      "epoch": 0.9802320584443489,
      "grad_norm": 1.0353646278381348,
      "learning_rate": 2.0491507846975843e-07,
      "loss": 1.4133,
      "step": 6843
    },
    {
      "epoch": 0.9803753043976507,
      "grad_norm": 1.1344181299209595,
      "learning_rate": 2.019570512081459e-07,
      "loss": 1.5246,
      "step": 6844
    },
    {
      "epoch": 0.9805185503509526,
      "grad_norm": 1.0790821313858032,
      "learning_rate": 1.9902050799557315e-07,
      "loss": 1.3916,
      "step": 6845
    },
    {
      "epoch": 0.9806617963042544,
      "grad_norm": 0.9657145738601685,
      "learning_rate": 1.9610544946420117e-07,
      "loss": 1.3994,
      "step": 6846
    },
    {
      "epoch": 0.9808050422575563,
      "grad_norm": 1.096034049987793,
      "learning_rate": 1.9321187624155024e-07,
      "loss": 1.4486,
      "step": 6847
    },
    {
      "epoch": 0.980948288210858,
      "grad_norm": 1.2756056785583496,
      "learning_rate": 1.9033978895054429e-07,
      "loss": 1.3731,
      "step": 6848
    },
    {
      "epoch": 0.9810915341641598,
      "grad_norm": 1.0752605199813843,
      "learning_rate": 1.8748918820948868e-07,
      "loss": 1.3842,
      "step": 6849
    },
    {
      "epoch": 0.9812347801174617,
      "grad_norm": 1.198684811592102,
      "learning_rate": 1.8466007463202596e-07,
      "loss": 1.4298,
      "step": 6850
    },
    {
      "epoch": 0.9813780260707635,
      "grad_norm": 1.0399978160858154,
      "learning_rate": 1.8185244882721332e-07,
      "loss": 1.5014,
      "step": 6851
    },
    {
      "epoch": 0.9815212720240654,
      "grad_norm": 1.0486814975738525,
      "learning_rate": 1.7906631139944508e-07,
      "loss": 1.2973,
      "step": 6852
    },
    {
      "epoch": 0.9816645179773671,
      "grad_norm": 1.1483771800994873,
      "learning_rate": 1.7630166294850813e-07,
      "loss": 1.3234,
      "step": 6853
    },
    {
      "epoch": 0.9818077639306689,
      "grad_norm": 1.4123873710632324,
      "learning_rate": 1.7355850406958196e-07,
      "loss": 1.3593,
      "step": 6854
    },
    {
      "epoch": 0.9819510098839708,
      "grad_norm": 1.0696710348129272,
      "learning_rate": 1.7083683535318306e-07,
      "loss": 1.3982,
      "step": 6855
    },
    {
      "epoch": 0.9820942558372726,
      "grad_norm": 1.224077582359314,
      "learning_rate": 1.6813665738523166e-07,
      "loss": 1.5042,
      "step": 6856
    },
    {
      "epoch": 0.9822375017905745,
      "grad_norm": 1.0116801261901855,
      "learning_rate": 1.654579707469961e-07,
      "loss": 1.297,
      "step": 6857
    },
    {
      "epoch": 0.9823807477438762,
      "grad_norm": 1.322712779045105,
      "learning_rate": 1.6280077601513734e-07,
      "loss": 1.2341,
      "step": 6858
    },
    {
      "epoch": 0.982523993697178,
      "grad_norm": 1.1745147705078125,
      "learning_rate": 1.6016507376169777e-07,
      "loss": 1.2589,
      "step": 6859
    },
    {
      "epoch": 0.9826672396504799,
      "grad_norm": 1.1023924350738525,
      "learning_rate": 1.5755086455404577e-07,
      "loss": 1.3073,
      "step": 6860
    },
    {
      "epoch": 0.9828104856037817,
      "grad_norm": 0.9869914650917053,
      "learning_rate": 1.5495814895498673e-07,
      "loss": 1.3961,
      "step": 6861
    },
    {
      "epoch": 0.9829537315570835,
      "grad_norm": 1.1217172145843506,
      "learning_rate": 1.5238692752266303e-07,
      "loss": 1.4043,
      "step": 6862
    },
    {
      "epoch": 0.9830969775103854,
      "grad_norm": 0.9376956820487976,
      "learning_rate": 1.498372008105764e-07,
      "loss": 1.3973,
      "step": 6863
    },
    {
      "epoch": 0.9832402234636871,
      "grad_norm": 1.1462827920913696,
      "learning_rate": 1.4730896936764327e-07,
      "loss": 1.2721,
      "step": 6864
    },
    {
      "epoch": 0.983383469416989,
      "grad_norm": 1.0549758672714233,
      "learning_rate": 1.448022337381061e-07,
      "loss": 1.4016,
      "step": 6865
    },
    {
      "epoch": 0.9835267153702908,
      "grad_norm": 1.310039758682251,
      "learning_rate": 1.4231699446162205e-07,
      "loss": 1.3705,
      "step": 6866
    },
    {
      "epoch": 0.9836699613235926,
      "grad_norm": 1.0545952320098877,
      "learning_rate": 1.3985325207319655e-07,
      "loss": 1.5126,
      "step": 6867
    },
    {
      "epoch": 0.9838132072768945,
      "grad_norm": 0.9783623218536377,
      "learning_rate": 1.3741100710321643e-07,
      "loss": 1.2753,
      "step": 6868
    },
    {
      "epoch": 0.9839564532301962,
      "grad_norm": 0.9266145825386047,
      "learning_rate": 1.3499026007741665e-07,
      "loss": 1.3952,
      "step": 6869
    },
    {
      "epoch": 0.984099699183498,
      "grad_norm": 0.94971764087677,
      "learning_rate": 1.3259101151694708e-07,
      "loss": 1.3221,
      "step": 6870
    },
    {
      "epoch": 0.9842429451367999,
      "grad_norm": 1.1135625839233398,
      "learning_rate": 1.3021326193830564e-07,
      "loss": 1.3664,
      "step": 6871
    },
    {
      "epoch": 0.9843861910901017,
      "grad_norm": 1.0000101327896118,
      "learning_rate": 1.2785701185333844e-07,
      "loss": 1.4034,
      "step": 6872
    },
    {
      "epoch": 0.9845294370434036,
      "grad_norm": 0.9760690927505493,
      "learning_rate": 1.2552226176931746e-07,
      "loss": 1.2539,
      "step": 6873
    },
    {
      "epoch": 0.9846726829967053,
      "grad_norm": 0.8988303542137146,
      "learning_rate": 1.2320901218884072e-07,
      "loss": 1.3908,
      "step": 6874
    },
    {
      "epoch": 0.9848159289500071,
      "grad_norm": 1.4073565006256104,
      "learning_rate": 1.2091726360989874e-07,
      "loss": 1.2897,
      "step": 6875
    },
    {
      "epoch": 0.984959174903309,
      "grad_norm": 1.0398823022842407,
      "learning_rate": 1.1864701652584132e-07,
      "loss": 1.657,
      "step": 6876
    },
    {
      "epoch": 0.9851024208566108,
      "grad_norm": 1.1171183586120605,
      "learning_rate": 1.1639827142539972e-07,
      "loss": 1.38,
      "step": 6877
    },
    {
      "epoch": 0.9852456668099127,
      "grad_norm": 1.0871721506118774,
      "learning_rate": 1.1417102879268671e-07,
      "loss": 1.4262,
      "step": 6878
    },
    {
      "epoch": 0.9853889127632144,
      "grad_norm": 1.1369484663009644,
      "learning_rate": 1.1196528910715209e-07,
      "loss": 1.3316,
      "step": 6879
    },
    {
      "epoch": 0.9855321587165162,
      "grad_norm": 0.960265576839447,
      "learning_rate": 1.0978105284363826e-07,
      "loss": 1.4356,
      "step": 6880
    },
    {
      "epoch": 0.9856754046698181,
      "grad_norm": 1.1848841905593872,
      "learning_rate": 1.0761832047238019e-07,
      "loss": 1.5113,
      "step": 6881
    },
    {
      "epoch": 0.9858186506231199,
      "grad_norm": 1.0930269956588745,
      "learning_rate": 1.0547709245893877e-07,
      "loss": 1.2834,
      "step": 6882
    },
    {
      "epoch": 0.9859618965764217,
      "grad_norm": 1.0428828001022339,
      "learning_rate": 1.0335736926426754e-07,
      "loss": 1.39,
      "step": 6883
    },
    {
      "epoch": 0.9861051425297236,
      "grad_norm": 1.0184683799743652,
      "learning_rate": 1.0125915134470143e-07,
      "loss": 1.4095,
      "step": 6884
    },
    {
      "epoch": 0.9862483884830253,
      "grad_norm": 0.9889693260192871,
      "learning_rate": 9.918243915193471e-08,
      "loss": 1.265,
      "step": 6885
    },
    {
      "epoch": 0.9863916344363272,
      "grad_norm": 1.1571595668792725,
      "learning_rate": 9.712723313302085e-08,
      "loss": 1.2631,
      "step": 6886
    },
    {
      "epoch": 0.986534880389629,
      "grad_norm": 1.2174410820007324,
      "learning_rate": 9.509353373040596e-08,
      "loss": 1.3352,
      "step": 6887
    },
    {
      "epoch": 0.9866781263429308,
      "grad_norm": 1.156304955482483,
      "learning_rate": 9.308134138188429e-08,
      "loss": 1.585,
      "step": 6888
    },
    {
      "epoch": 0.9868213722962327,
      "grad_norm": 1.062891960144043,
      "learning_rate": 9.109065652064263e-08,
      "loss": 1.4425,
      "step": 6889
    },
    {
      "epoch": 0.9869646182495344,
      "grad_norm": 1.059097409248352,
      "learning_rate": 8.9121479575216e-08,
      "loss": 1.4094,
      "step": 6890
    },
    {
      "epoch": 0.9871078642028362,
      "grad_norm": 0.9478225708007812,
      "learning_rate": 8.717381096953192e-08,
      "loss": 1.5065,
      "step": 6891
    },
    {
      "epoch": 0.9872511101561381,
      "grad_norm": 1.0617536306381226,
      "learning_rate": 8.524765112286614e-08,
      "loss": 1.3645,
      "step": 6892
    },
    {
      "epoch": 0.9873943561094399,
      "grad_norm": 1.1142922639846802,
      "learning_rate": 8.334300044987587e-08,
      "loss": 1.4638,
      "step": 6893
    },
    {
      "epoch": 0.9875376020627418,
      "grad_norm": 1.1490753889083862,
      "learning_rate": 8.145985936057754e-08,
      "loss": 1.344,
      "step": 6894
    },
    {
      "epoch": 0.9876808480160435,
      "grad_norm": 1.0488166809082031,
      "learning_rate": 7.959822826038022e-08,
      "loss": 1.4072,
      "step": 6895
    },
    {
      "epoch": 0.9878240939693453,
      "grad_norm": 1.1462736129760742,
      "learning_rate": 7.775810755003e-08,
      "loss": 1.4117,
      "step": 6896
    },
    {
      "epoch": 0.9879673399226472,
      "grad_norm": 1.0138330459594727,
      "learning_rate": 7.593949762567664e-08,
      "loss": 1.4115,
      "step": 6897
    },
    {
      "epoch": 0.988110585875949,
      "grad_norm": 1.1845062971115112,
      "learning_rate": 7.414239887880702e-08,
      "loss": 1.2574,
      "step": 6898
    },
    {
      "epoch": 0.9882538318292509,
      "grad_norm": 0.9185782670974731,
      "learning_rate": 7.236681169628945e-08,
      "loss": 1.4322,
      "step": 6899
    },
    {
      "epoch": 0.9883970777825526,
      "grad_norm": 1.1274644136428833,
      "learning_rate": 7.06127364603848e-08,
      "loss": 1.2809,
      "step": 6900
    },
    {
      "epoch": 0.9885403237358544,
      "grad_norm": 0.9798005819320679,
      "learning_rate": 6.888017354869103e-08,
      "loss": 1.3656,
      "step": 6901
    },
    {
      "epoch": 0.9886835696891563,
      "grad_norm": 0.8769481778144836,
      "learning_rate": 6.716912333417646e-08,
      "loss": 1.5398,
      "step": 6902
    },
    {
      "epoch": 0.9888268156424581,
      "grad_norm": 0.9875479340553284,
      "learning_rate": 6.5479586185202e-08,
      "loss": 1.323,
      "step": 6903
    },
    {
      "epoch": 0.98897006159576,
      "grad_norm": 1.095491886138916,
      "learning_rate": 6.38115624654656e-08,
      "loss": 1.3066,
      "step": 6904
    },
    {
      "epoch": 0.9891133075490618,
      "grad_norm": 1.2760183811187744,
      "learning_rate": 6.216505253408e-08,
      "loss": 1.4963,
      "step": 6905
    },
    {
      "epoch": 0.9892565535023635,
      "grad_norm": 1.1250486373901367,
      "learning_rate": 6.05400567454728e-08,
      "loss": 1.4826,
      "step": 6906
    },
    {
      "epoch": 0.9893997994556654,
      "grad_norm": 1.08491051197052,
      "learning_rate": 5.893657544947528e-08,
      "loss": 1.1679,
      "step": 6907
    },
    {
      "epoch": 0.9895430454089672,
      "grad_norm": 1.0493440628051758,
      "learning_rate": 5.7354608991266876e-08,
      "loss": 1.3012,
      "step": 6908
    },
    {
      "epoch": 0.989686291362269,
      "grad_norm": 1.173628330230713,
      "learning_rate": 5.5794157711430705e-08,
      "loss": 1.3258,
      "step": 6909
    },
    {
      "epoch": 0.9898295373155709,
      "grad_norm": 0.9683168530464172,
      "learning_rate": 5.4255221945864744e-08,
      "loss": 1.475,
      "step": 6910
    },
    {
      "epoch": 0.9899727832688726,
      "grad_norm": 0.9912375211715698,
      "learning_rate": 5.273780202588174e-08,
      "loss": 1.4734,
      "step": 6911
    },
    {
      "epoch": 0.9901160292221745,
      "grad_norm": 1.07449471950531,
      "learning_rate": 5.124189827813153e-08,
      "loss": 1.3338,
      "step": 6912
    },
    {
      "epoch": 0.9902592751754763,
      "grad_norm": 1.2173659801483154,
      "learning_rate": 4.9767511024656486e-08,
      "loss": 1.3669,
      "step": 6913
    },
    {
      "epoch": 0.9904025211287781,
      "grad_norm": 1.1356706619262695,
      "learning_rate": 4.8314640582858284e-08,
      "loss": 1.3088,
      "step": 6914
    },
    {
      "epoch": 0.99054576708208,
      "grad_norm": 0.9836452603340149,
      "learning_rate": 4.6883287265497844e-08,
      "loss": 1.4552,
      "step": 6915
    },
    {
      "epoch": 0.9906890130353817,
      "grad_norm": 1.0289698839187622,
      "learning_rate": 4.5473451380706463e-08,
      "loss": 1.3499,
      "step": 6916
    },
    {
      "epoch": 0.9908322589886835,
      "grad_norm": 1.0422101020812988,
      "learning_rate": 4.408513323198582e-08,
      "loss": 1.3797,
      "step": 6917
    },
    {
      "epoch": 0.9909755049419854,
      "grad_norm": 1.0672037601470947,
      "learning_rate": 4.271833311821905e-08,
      "loss": 1.4964,
      "step": 6918
    },
    {
      "epoch": 0.9911187508952872,
      "grad_norm": 1.1386151313781738,
      "learning_rate": 4.137305133362634e-08,
      "loss": 1.3855,
      "step": 6919
    },
    {
      "epoch": 0.9912619968485891,
      "grad_norm": 1.0069860219955444,
      "learning_rate": 4.0049288167842705e-08,
      "loss": 1.45,
      "step": 6920
    },
    {
      "epoch": 0.9914052428018908,
      "grad_norm": 1.1581194400787354,
      "learning_rate": 3.8747043905806856e-08,
      "loss": 1.3311,
      "step": 6921
    },
    {
      "epoch": 0.9915484887551926,
      "grad_norm": 1.049757480621338,
      "learning_rate": 3.746631882787233e-08,
      "loss": 1.2218,
      "step": 6922
    },
    {
      "epoch": 0.9916917347084945,
      "grad_norm": 1.0640041828155518,
      "learning_rate": 3.6207113209763e-08,
      "loss": 1.4512,
      "step": 6923
    },
    {
      "epoch": 0.9918349806617963,
      "grad_norm": 1.1585804224014282,
      "learning_rate": 3.496942732253983e-08,
      "loss": 1.3157,
      "step": 6924
    },
    {
      "epoch": 0.9919782266150982,
      "grad_norm": 1.1739224195480347,
      "learning_rate": 3.375326143264523e-08,
      "loss": 1.2783,
      "step": 6925
    },
    {
      "epoch": 0.9921214725684,
      "grad_norm": 1.1030296087265015,
      "learning_rate": 3.2558615801892014e-08,
      "loss": 1.3843,
      "step": 6926
    },
    {
      "epoch": 0.9922647185217017,
      "grad_norm": 1.0498274564743042,
      "learning_rate": 3.138549068745222e-08,
      "loss": 1.4051,
      "step": 6927
    },
    {
      "epoch": 0.9924079644750036,
      "grad_norm": 1.021190881729126,
      "learning_rate": 3.0233886341890504e-08,
      "loss": 1.4729,
      "step": 6928
    },
    {
      "epoch": 0.9925512104283054,
      "grad_norm": 1.3140523433685303,
      "learning_rate": 2.9103803013097453e-08,
      "loss": 1.1676,
      "step": 6929
    },
    {
      "epoch": 0.9926944563816072,
      "grad_norm": 1.0635639429092407,
      "learning_rate": 2.799524094436734e-08,
      "loss": 1.3982,
      "step": 6930
    },
    {
      "epoch": 0.9928377023349091,
      "grad_norm": 1.047019600868225,
      "learning_rate": 2.6908200374331503e-08,
      "loss": 1.3862,
      "step": 6931
    },
    {
      "epoch": 0.9929809482882108,
      "grad_norm": 1.0426539182662964,
      "learning_rate": 2.584268153701386e-08,
      "loss": 1.3052,
      "step": 6932
    },
    {
      "epoch": 0.9931241942415127,
      "grad_norm": 1.2257951498031616,
      "learning_rate": 2.4798684661786476e-08,
      "loss": 1.5279,
      "step": 6933
    },
    {
      "epoch": 0.9932674401948145,
      "grad_norm": 1.1203614473342896,
      "learning_rate": 2.377620997340291e-08,
      "loss": 1.3,
      "step": 6934
    },
    {
      "epoch": 0.9934106861481163,
      "grad_norm": 1.0221182107925415,
      "learning_rate": 2.2775257691975972e-08,
      "loss": 1.3475,
      "step": 6935
    },
    {
      "epoch": 0.9935539321014182,
      "grad_norm": 1.2252392768859863,
      "learning_rate": 2.179582803297775e-08,
      "loss": 1.5167,
      "step": 6936
    },
    {
      "epoch": 0.9936971780547199,
      "grad_norm": 1.054181456565857,
      "learning_rate": 2.0837921207272902e-08,
      "loss": 1.2551,
      "step": 6937
    },
    {
      "epoch": 0.9938404240080217,
      "grad_norm": 0.995672881603241,
      "learning_rate": 1.990153742105205e-08,
      "loss": 1.3657,
      "step": 6938
    },
    {
      "epoch": 0.9939836699613236,
      "grad_norm": 1.1049494743347168,
      "learning_rate": 1.8986676875909494e-08,
      "loss": 1.5315,
      "step": 6939
    },
    {
      "epoch": 0.9941269159146254,
      "grad_norm": 1.1547455787658691,
      "learning_rate": 1.8093339768798788e-08,
      "loss": 1.5175,
      "step": 6940
    },
    {
      "epoch": 0.9942701618679273,
      "grad_norm": 1.09352707862854,
      "learning_rate": 1.722152629201057e-08,
      "loss": 1.3678,
      "step": 6941
    },
    {
      "epoch": 0.994413407821229,
      "grad_norm": 1.1882737874984741,
      "learning_rate": 1.637123663323914e-08,
      "loss": 1.3243,
      "step": 6942
    },
    {
      "epoch": 0.9945566537745308,
      "grad_norm": 1.1435630321502686,
      "learning_rate": 1.554247097553807e-08,
      "loss": 1.469,
      "step": 6943
    },
    {
      "epoch": 0.9946998997278327,
      "grad_norm": 1.0475714206695557,
      "learning_rate": 1.47352294973091e-08,
      "loss": 1.4755,
      "step": 6944
    },
    {
      "epoch": 0.9948431456811345,
      "grad_norm": 1.0620249509811401,
      "learning_rate": 1.3949512372335438e-08,
      "loss": 1.6501,
      "step": 6945
    },
    {
      "epoch": 0.9949863916344364,
      "grad_norm": 1.0384702682495117,
      "learning_rate": 1.3185319769759564e-08,
      "loss": 1.334,
      "step": 6946
    },
    {
      "epoch": 0.9951296375877381,
      "grad_norm": 1.0790804624557495,
      "learning_rate": 1.2442651854094322e-08,
      "loss": 1.3465,
      "step": 6947
    },
    {
      "epoch": 0.9952728835410399,
      "grad_norm": 1.114069938659668,
      "learning_rate": 1.1721508785211832e-08,
      "loss": 1.4863,
      "step": 6948
    },
    {
      "epoch": 0.9954161294943418,
      "grad_norm": 1.0187630653381348,
      "learning_rate": 1.1021890718376781e-08,
      "loss": 1.3024,
      "step": 6949
    },
    {
      "epoch": 0.9955593754476436,
      "grad_norm": 1.1869088411331177,
      "learning_rate": 1.034379780416872e-08,
      "loss": 1.4357,
      "step": 6950
    },
    {
      "epoch": 0.9957026214009455,
      "grad_norm": 0.9803943634033203,
      "learning_rate": 9.687230188593077e-09,
      "loss": 1.3271,
      "step": 6951
    },
    {
      "epoch": 0.9958458673542473,
      "grad_norm": 1.0960941314697266,
      "learning_rate": 9.052188012981244e-09,
      "loss": 1.3742,
      "step": 6952
    },
    {
      "epoch": 0.995989113307549,
      "grad_norm": 1.0931395292282104,
      "learning_rate": 8.438671414034982e-09,
      "loss": 1.4412,
      "step": 6953
    },
    {
      "epoch": 0.9961323592608509,
      "grad_norm": 1.0480124950408936,
      "learning_rate": 7.846680523837524e-09,
      "loss": 1.219,
      "step": 6954
    },
    {
      "epoch": 0.9962756052141527,
      "grad_norm": 1.047921061515808,
      "learning_rate": 7.276215469831371e-09,
      "loss": 1.39,
      "step": 6955
    },
    {
      "epoch": 0.9964188511674545,
      "grad_norm": 1.0247312784194946,
      "learning_rate": 6.727276374818292e-09,
      "loss": 1.4031,
      "step": 6956
    },
    {
      "epoch": 0.9965620971207564,
      "grad_norm": 1.0511873960494995,
      "learning_rate": 6.1998633569704256e-09,
      "loss": 1.3113,
      "step": 6957
    },
    {
      "epoch": 0.9967053430740581,
      "grad_norm": 1.1090106964111328,
      "learning_rate": 5.693976529841383e-09,
      "loss": 1.3447,
      "step": 6958
    },
    {
      "epoch": 0.99684858902736,
      "grad_norm": 1.1817384958267212,
      "learning_rate": 5.209616002310736e-09,
      "loss": 1.3977,
      "step": 6959
    },
    {
      "epoch": 0.9969918349806618,
      "grad_norm": 1.1059447526931763,
      "learning_rate": 4.746781878672835e-09,
      "loss": 1.4947,
      "step": 6960
    },
    {
      "epoch": 0.9971350809339636,
      "grad_norm": 1.1532925367355347,
      "learning_rate": 4.305474258547993e-09,
      "loss": 1.5286,
      "step": 6961
    },
    {
      "epoch": 0.9972783268872655,
      "grad_norm": 1.181623935699463,
      "learning_rate": 3.885693236949095e-09,
      "loss": 1.3611,
      "step": 6962
    },
    {
      "epoch": 0.9974215728405672,
      "grad_norm": 1.2167772054672241,
      "learning_rate": 3.487438904237195e-09,
      "loss": 1.5892,
      "step": 6963
    },
    {
      "epoch": 0.997564818793869,
      "grad_norm": 1.1761642694473267,
      "learning_rate": 3.1107113461437133e-09,
      "loss": 1.421,
      "step": 6964
    },
    {
      "epoch": 0.9977080647471709,
      "grad_norm": 1.1366292238235474,
      "learning_rate": 2.755510643792647e-09,
      "loss": 1.4673,
      "step": 6965
    },
    {
      "epoch": 0.9978513107004727,
      "grad_norm": 1.089664340019226,
      "learning_rate": 2.4218368736228512e-09,
      "loss": 1.3498,
      "step": 6966
    },
    {
      "epoch": 0.9979945566537746,
      "grad_norm": 0.9447208046913147,
      "learning_rate": 2.109690107465756e-09,
      "loss": 1.3743,
      "step": 6967
    },
    {
      "epoch": 0.9981378026070763,
      "grad_norm": 1.1080571413040161,
      "learning_rate": 1.819070412545365e-09,
      "loss": 1.419,
      "step": 6968
    },
    {
      "epoch": 0.9982810485603781,
      "grad_norm": 1.1412129402160645,
      "learning_rate": 1.5499778514005415e-09,
      "loss": 1.4905,
      "step": 6969
    },
    {
      "epoch": 0.99842429451368,
      "grad_norm": 1.0053709745407104,
      "learning_rate": 1.3024124819738247e-09,
      "loss": 1.3341,
      "step": 6970
    },
    {
      "epoch": 0.9985675404669818,
      "grad_norm": 1.057105541229248,
      "learning_rate": 1.0763743575448182e-09,
      "loss": 1.3919,
      "step": 6971
    },
    {
      "epoch": 0.9987107864202837,
      "grad_norm": 1.2198081016540527,
      "learning_rate": 8.718635267856989e-10,
      "loss": 1.2662,
      "step": 6972
    },
    {
      "epoch": 0.9988540323735855,
      "grad_norm": 0.923511803150177,
      "learning_rate": 6.888800337279122e-10,
      "loss": 1.5828,
      "step": 6973
    },
    {
      "epoch": 0.9989972783268872,
      "grad_norm": 1.0067452192306519,
      "learning_rate": 5.274239177510687e-10,
      "loss": 1.4352,
      "step": 6974
    },
    {
      "epoch": 0.9991405242801891,
      "grad_norm": 1.035966396331787,
      "learning_rate": 3.874952136162513e-10,
      "loss": 1.3927,
      "step": 6975
    },
    {
      "epoch": 0.9992837702334909,
      "grad_norm": 1.0147660970687866,
      "learning_rate": 2.690939514438107e-10,
      "loss": 1.3475,
      "step": 6976
    },
    {
      "epoch": 0.9994270161867927,
      "grad_norm": 1.023337483406067,
      "learning_rate": 1.7222015673556968e-10,
      "loss": 1.4416,
      "step": 6977
    },
    {
      "epoch": 0.9995702621400946,
      "grad_norm": 1.0672696828842163,
      "learning_rate": 9.687385033041451e-11,
      "loss": 1.459,
      "step": 6978
    },
    {
      "epoch": 0.9997135080933963,
      "grad_norm": 1.103571891784668,
      "learning_rate": 4.305504844870356e-11,
      "loss": 1.2565,
      "step": 6979
    },
    {
      "epoch": 0.9998567540466982,
      "grad_norm": 1.0865168571472168,
      "learning_rate": 1.0763762692267421e-11,
      "loss": 1.4758,
      "step": 6980
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2903202772140503,
      "learning_rate": 0.0,
      "loss": 1.432,
      "step": 6981
    },
    {
      "epoch": 1.0,
      "step": 6981,
      "total_flos": 2.4986378761037414e+17,
      "train_loss": 1.5009833613741177,
      "train_runtime": 12243.7828,
      "train_samples_per_second": 18.245,
      "train_steps_per_second": 0.57
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 6981,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.4986378761037414e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}