{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9985775248933146,
  "eval_steps": 500,
  "global_step": 3162,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.000948316737790422,
      "grad_norm": 12.246086019020522,
      "learning_rate": 3.1545741324921134e-08,
      "loss": 0.3675,
      "step": 1
    },
    {
      "epoch": 0.001896633475580844,
      "grad_norm": 11.694026905196086,
      "learning_rate": 6.309148264984227e-08,
      "loss": 0.3513,
      "step": 2
    },
    {
      "epoch": 0.002844950213371266,
      "grad_norm": 10.688899348005608,
      "learning_rate": 9.463722397476342e-08,
      "loss": 0.3177,
      "step": 3
    },
    {
      "epoch": 0.003793266951161688,
      "grad_norm": 11.921298417211082,
      "learning_rate": 1.2618296529968454e-07,
      "loss": 0.3556,
      "step": 4
    },
    {
      "epoch": 0.00474158368895211,
      "grad_norm": 11.282902382225787,
      "learning_rate": 1.5772870662460568e-07,
      "loss": 0.3662,
      "step": 5
    },
    {
      "epoch": 0.005689900426742532,
      "grad_norm": 12.449826286939778,
      "learning_rate": 1.8927444794952683e-07,
      "loss": 0.3546,
      "step": 6
    },
    {
      "epoch": 0.006638217164532954,
      "grad_norm": 11.727097480206721,
      "learning_rate": 2.2082018927444798e-07,
      "loss": 0.3635,
      "step": 7
    },
    {
      "epoch": 0.007586533902323376,
      "grad_norm": 11.441751312661113,
      "learning_rate": 2.5236593059936907e-07,
      "loss": 0.3636,
      "step": 8
    },
    {
      "epoch": 0.008534850640113799,
      "grad_norm": 10.632930884848795,
      "learning_rate": 2.8391167192429027e-07,
      "loss": 0.2923,
      "step": 9
    },
    {
      "epoch": 0.00948316737790422,
      "grad_norm": 11.025857208188647,
      "learning_rate": 3.1545741324921137e-07,
      "loss": 0.3449,
      "step": 10
    },
    {
      "epoch": 0.010431484115694643,
      "grad_norm": 11.86359857266447,
      "learning_rate": 3.470031545741325e-07,
      "loss": 0.3354,
      "step": 11
    },
    {
      "epoch": 0.011379800853485065,
      "grad_norm": 11.01751351812872,
      "learning_rate": 3.7854889589905366e-07,
      "loss": 0.3369,
      "step": 12
    },
    {
      "epoch": 0.012328117591275486,
      "grad_norm": 9.502190495628849,
      "learning_rate": 4.100946372239748e-07,
      "loss": 0.3179,
      "step": 13
    },
    {
      "epoch": 0.013276434329065908,
      "grad_norm": 7.858408977040966,
      "learning_rate": 4.4164037854889596e-07,
      "loss": 0.2565,
      "step": 14
    },
    {
      "epoch": 0.01422475106685633,
      "grad_norm": 8.154333698814211,
      "learning_rate": 4.7318611987381705e-07,
      "loss": 0.2589,
      "step": 15
    },
    {
      "epoch": 0.015173067804646752,
      "grad_norm": 8.475444781638856,
      "learning_rate": 5.047318611987381e-07,
      "loss": 0.3001,
      "step": 16
    },
    {
      "epoch": 0.016121384542437174,
      "grad_norm": 7.13737669899092,
      "learning_rate": 5.362776025236594e-07,
      "loss": 0.2641,
      "step": 17
    },
    {
      "epoch": 0.017069701280227598,
      "grad_norm": 5.554684184392653,
      "learning_rate": 5.678233438485805e-07,
      "loss": 0.1902,
      "step": 18
    },
    {
      "epoch": 0.018018018018018018,
      "grad_norm": 4.568211300813283,
      "learning_rate": 5.993690851735017e-07,
      "loss": 0.208,
      "step": 19
    },
    {
      "epoch": 0.01896633475580844,
      "grad_norm": 4.7579569152913646,
      "learning_rate": 6.309148264984227e-07,
      "loss": 0.1994,
      "step": 20
    },
    {
      "epoch": 0.01991465149359886,
      "grad_norm": 4.7128465676673486,
      "learning_rate": 6.62460567823344e-07,
      "loss": 0.229,
      "step": 21
    },
    {
      "epoch": 0.020862968231389285,
      "grad_norm": 4.005405411985473,
      "learning_rate": 6.94006309148265e-07,
      "loss": 0.2095,
      "step": 22
    },
    {
      "epoch": 0.021811284969179705,
      "grad_norm": 4.676075338959145,
      "learning_rate": 7.255520504731863e-07,
      "loss": 0.2178,
      "step": 23
    },
    {
      "epoch": 0.02275960170697013,
      "grad_norm": 2.3652635706654435,
      "learning_rate": 7.570977917981073e-07,
      "loss": 0.1524,
      "step": 24
    },
    {
      "epoch": 0.02370791844476055,
      "grad_norm": 2.685337556789167,
      "learning_rate": 7.886435331230284e-07,
      "loss": 0.1672,
      "step": 25
    },
    {
      "epoch": 0.024656235182550973,
      "grad_norm": 2.430942973848189,
      "learning_rate": 8.201892744479496e-07,
      "loss": 0.1526,
      "step": 26
    },
    {
      "epoch": 0.025604551920341393,
      "grad_norm": 2.6467399445694286,
      "learning_rate": 8.517350157728707e-07,
      "loss": 0.1605,
      "step": 27
    },
    {
      "epoch": 0.026552868658131817,
      "grad_norm": 1.9805248826128374,
      "learning_rate": 8.832807570977919e-07,
      "loss": 0.1223,
      "step": 28
    },
    {
      "epoch": 0.027501185395922237,
      "grad_norm": 2.2695664454959785,
      "learning_rate": 9.148264984227131e-07,
      "loss": 0.1991,
      "step": 29
    },
    {
      "epoch": 0.02844950213371266,
      "grad_norm": 2.221333597086963,
      "learning_rate": 9.463722397476341e-07,
      "loss": 0.1466,
      "step": 30
    },
    {
      "epoch": 0.02939781887150308,
      "grad_norm": 2.4412316593782633,
      "learning_rate": 9.779179810725552e-07,
      "loss": 0.1757,
      "step": 31
    },
    {
      "epoch": 0.030346135609293504,
      "grad_norm": 2.3894901293863198,
      "learning_rate": 1.0094637223974763e-06,
      "loss": 0.1467,
      "step": 32
    },
    {
      "epoch": 0.031294452347083924,
      "grad_norm": 2.2254181707911003,
      "learning_rate": 1.0410094637223975e-06,
      "loss": 0.1403,
      "step": 33
    },
    {
      "epoch": 0.03224276908487435,
      "grad_norm": 2.0835670435267573,
      "learning_rate": 1.0725552050473188e-06,
      "loss": 0.1295,
      "step": 34
    },
    {
      "epoch": 0.03319108582266477,
      "grad_norm": 2.490534255553767,
      "learning_rate": 1.1041009463722398e-06,
      "loss": 0.1781,
      "step": 35
    },
    {
      "epoch": 0.034139402560455195,
      "grad_norm": 2.4852979753797526,
      "learning_rate": 1.135646687697161e-06,
      "loss": 0.1635,
      "step": 36
    },
    {
      "epoch": 0.03508771929824561,
      "grad_norm": 2.1821267836376093,
      "learning_rate": 1.1671924290220821e-06,
      "loss": 0.1536,
      "step": 37
    },
    {
      "epoch": 0.036036036036036036,
      "grad_norm": 2.0995981360286193,
      "learning_rate": 1.1987381703470034e-06,
      "loss": 0.1892,
      "step": 38
    },
    {
      "epoch": 0.03698435277382646,
      "grad_norm": 1.7671849396147046,
      "learning_rate": 1.2302839116719244e-06,
      "loss": 0.1441,
      "step": 39
    },
    {
      "epoch": 0.03793266951161688,
      "grad_norm": 1.9145592423222202,
      "learning_rate": 1.2618296529968455e-06,
      "loss": 0.1515,
      "step": 40
    },
    {
      "epoch": 0.0388809862494073,
      "grad_norm": 2.909485299628588,
      "learning_rate": 1.2933753943217667e-06,
      "loss": 0.176,
      "step": 41
    },
    {
      "epoch": 0.03982930298719772,
      "grad_norm": 1.6871888183428478,
      "learning_rate": 1.324921135646688e-06,
      "loss": 0.1215,
      "step": 42
    },
    {
      "epoch": 0.04077761972498815,
      "grad_norm": 1.5396564532901138,
      "learning_rate": 1.3564668769716088e-06,
      "loss": 0.1334,
      "step": 43
    },
    {
      "epoch": 0.04172593646277857,
      "grad_norm": 2.205033481070129,
      "learning_rate": 1.38801261829653e-06,
      "loss": 0.1397,
      "step": 44
    },
    {
      "epoch": 0.04267425320056899,
      "grad_norm": 1.8497757762613358,
      "learning_rate": 1.4195583596214513e-06,
      "loss": 0.1274,
      "step": 45
    },
    {
      "epoch": 0.04362256993835941,
      "grad_norm": 1.9376215434540043,
      "learning_rate": 1.4511041009463726e-06,
      "loss": 0.1228,
      "step": 46
    },
    {
      "epoch": 0.044570886676149835,
      "grad_norm": 1.594889970345864,
      "learning_rate": 1.4826498422712934e-06,
      "loss": 0.1137,
      "step": 47
    },
    {
      "epoch": 0.04551920341394026,
      "grad_norm": 1.7592570423176281,
      "learning_rate": 1.5141955835962146e-06,
      "loss": 0.141,
      "step": 48
    },
    {
      "epoch": 0.046467520151730675,
      "grad_norm": 1.6146283602515956,
      "learning_rate": 1.545741324921136e-06,
      "loss": 0.1428,
      "step": 49
    },
    {
      "epoch": 0.0474158368895211,
      "grad_norm": 1.503278573378982,
      "learning_rate": 1.5772870662460567e-06,
      "loss": 0.1318,
      "step": 50
    },
    {
      "epoch": 0.04836415362731152,
      "grad_norm": 1.37572777178569,
      "learning_rate": 1.608832807570978e-06,
      "loss": 0.1315,
      "step": 51
    },
    {
      "epoch": 0.049312470365101946,
      "grad_norm": 1.6002275154635794,
      "learning_rate": 1.6403785488958992e-06,
      "loss": 0.0935,
      "step": 52
    },
    {
      "epoch": 0.05026078710289237,
      "grad_norm": 1.9567696662008847,
      "learning_rate": 1.6719242902208203e-06,
      "loss": 0.1271,
      "step": 53
    },
    {
      "epoch": 0.051209103840682786,
      "grad_norm": 1.601626063178932,
      "learning_rate": 1.7034700315457413e-06,
      "loss": 0.0959,
      "step": 54
    },
    {
      "epoch": 0.05215742057847321,
      "grad_norm": 1.886431535590579,
      "learning_rate": 1.7350157728706626e-06,
      "loss": 0.1218,
      "step": 55
    },
    {
      "epoch": 0.05310573731626363,
      "grad_norm": 1.5354494166136305,
      "learning_rate": 1.7665615141955838e-06,
      "loss": 0.1139,
      "step": 56
    },
    {
      "epoch": 0.05405405405405406,
      "grad_norm": 2.311230053300576,
      "learning_rate": 1.7981072555205049e-06,
      "loss": 0.1426,
      "step": 57
    },
    {
      "epoch": 0.055002370791844474,
      "grad_norm": 1.6253071180005185,
      "learning_rate": 1.8296529968454261e-06,
      "loss": 0.1175,
      "step": 58
    },
    {
      "epoch": 0.0559506875296349,
      "grad_norm": 1.3821063491809322,
      "learning_rate": 1.8611987381703472e-06,
      "loss": 0.132,
      "step": 59
    },
    {
      "epoch": 0.05689900426742532,
      "grad_norm": 1.7624392868013044,
      "learning_rate": 1.8927444794952682e-06,
      "loss": 0.1221,
      "step": 60
    },
    {
      "epoch": 0.057847321005215745,
      "grad_norm": 1.3398437874784876,
      "learning_rate": 1.9242902208201892e-06,
      "loss": 0.125,
      "step": 61
    },
    {
      "epoch": 0.05879563774300616,
      "grad_norm": 1.562570182505017,
      "learning_rate": 1.9558359621451105e-06,
      "loss": 0.1413,
      "step": 62
    },
    {
      "epoch": 0.059743954480796585,
      "grad_norm": 1.6769755616188486,
      "learning_rate": 1.9873817034700317e-06,
      "loss": 0.1559,
      "step": 63
    },
    {
      "epoch": 0.06069227121858701,
      "grad_norm": 1.3917364499829268,
      "learning_rate": 2.0189274447949526e-06,
      "loss": 0.1377,
      "step": 64
    },
    {
      "epoch": 0.06164058795637743,
      "grad_norm": 1.8502674559797263,
      "learning_rate": 2.050473186119874e-06,
      "loss": 0.1487,
      "step": 65
    },
    {
      "epoch": 0.06258890469416785,
      "grad_norm": 3.158783977874437,
      "learning_rate": 2.082018927444795e-06,
      "loss": 0.119,
      "step": 66
    },
    {
      "epoch": 0.06353722143195828,
      "grad_norm": 1.811584236109641,
      "learning_rate": 2.1135646687697163e-06,
      "loss": 0.122,
      "step": 67
    },
    {
      "epoch": 0.0644855381697487,
      "grad_norm": 2.917344328319794,
      "learning_rate": 2.1451104100946376e-06,
      "loss": 0.1313,
      "step": 68
    },
    {
      "epoch": 0.06543385490753911,
      "grad_norm": 1.8029019845335916,
      "learning_rate": 2.1766561514195584e-06,
      "loss": 0.1138,
      "step": 69
    },
    {
      "epoch": 0.06638217164532954,
      "grad_norm": 1.6898543330406532,
      "learning_rate": 2.2082018927444797e-06,
      "loss": 0.1191,
      "step": 70
    },
    {
      "epoch": 0.06733048838311996,
      "grad_norm": 2.2925732127308214,
      "learning_rate": 2.239747634069401e-06,
      "loss": 0.1306,
      "step": 71
    },
    {
      "epoch": 0.06827880512091039,
      "grad_norm": 1.4433490292568716,
      "learning_rate": 2.271293375394322e-06,
      "loss": 0.1055,
      "step": 72
    },
    {
      "epoch": 0.06922712185870081,
      "grad_norm": 1.3862506183642664,
      "learning_rate": 2.302839116719243e-06,
      "loss": 0.1075,
      "step": 73
    },
    {
      "epoch": 0.07017543859649122,
      "grad_norm": 1.2816575561632197,
      "learning_rate": 2.3343848580441643e-06,
      "loss": 0.1028,
      "step": 74
    },
    {
      "epoch": 0.07112375533428165,
      "grad_norm": 1.893923472034316,
      "learning_rate": 2.3659305993690855e-06,
      "loss": 0.1011,
      "step": 75
    },
    {
      "epoch": 0.07207207207207207,
      "grad_norm": 1.6025824634915868,
      "learning_rate": 2.3974763406940068e-06,
      "loss": 0.1317,
      "step": 76
    },
    {
      "epoch": 0.07302038880986249,
      "grad_norm": 1.7176261068301808,
      "learning_rate": 2.4290220820189276e-06,
      "loss": 0.1447,
      "step": 77
    },
    {
      "epoch": 0.07396870554765292,
      "grad_norm": 2.4231160050612863,
      "learning_rate": 2.460567823343849e-06,
      "loss": 0.1384,
      "step": 78
    },
    {
      "epoch": 0.07491702228544334,
      "grad_norm": 1.2193411546548798,
      "learning_rate": 2.49211356466877e-06,
      "loss": 0.0992,
      "step": 79
    },
    {
      "epoch": 0.07586533902323377,
      "grad_norm": 1.5164983059809367,
      "learning_rate": 2.523659305993691e-06,
      "loss": 0.1001,
      "step": 80
    },
    {
      "epoch": 0.07681365576102418,
      "grad_norm": 1.6017905795769134,
      "learning_rate": 2.5552050473186126e-06,
      "loss": 0.1048,
      "step": 81
    },
    {
      "epoch": 0.0777619724988146,
      "grad_norm": 1.1836761079302904,
      "learning_rate": 2.5867507886435334e-06,
      "loss": 0.0982,
      "step": 82
    },
    {
      "epoch": 0.07871028923660503,
      "grad_norm": 3.3493572839513566,
      "learning_rate": 2.6182965299684543e-06,
      "loss": 0.1184,
      "step": 83
    },
    {
      "epoch": 0.07965860597439545,
      "grad_norm": 2.0313461386241722,
      "learning_rate": 2.649842271293376e-06,
      "loss": 0.1174,
      "step": 84
    },
    {
      "epoch": 0.08060692271218586,
      "grad_norm": 1.7152579326543271,
      "learning_rate": 2.6813880126182968e-06,
      "loss": 0.117,
      "step": 85
    },
    {
      "epoch": 0.0815552394499763,
      "grad_norm": 1.9082671591126898,
      "learning_rate": 2.7129337539432176e-06,
      "loss": 0.1538,
      "step": 86
    },
    {
      "epoch": 0.08250355618776671,
      "grad_norm": 1.1544236926306861,
      "learning_rate": 2.7444794952681393e-06,
      "loss": 0.0906,
      "step": 87
    },
    {
      "epoch": 0.08345187292555714,
      "grad_norm": 1.2516823902614436,
      "learning_rate": 2.77602523659306e-06,
      "loss": 0.1452,
      "step": 88
    },
    {
      "epoch": 0.08440018966334756,
      "grad_norm": 1.0339206815219761,
      "learning_rate": 2.807570977917981e-06,
      "loss": 0.0836,
      "step": 89
    },
    {
      "epoch": 0.08534850640113797,
      "grad_norm": 1.668394516826565,
      "learning_rate": 2.8391167192429026e-06,
      "loss": 0.1129,
      "step": 90
    },
    {
      "epoch": 0.0862968231389284,
      "grad_norm": 1.497152077632149,
      "learning_rate": 2.8706624605678234e-06,
      "loss": 0.1062,
      "step": 91
    },
    {
      "epoch": 0.08724513987671882,
      "grad_norm": 1.197731872894548,
      "learning_rate": 2.902208201892745e-06,
      "loss": 0.1102,
      "step": 92
    },
    {
      "epoch": 0.08819345661450925,
      "grad_norm": 1.4271367098608596,
      "learning_rate": 2.933753943217666e-06,
      "loss": 0.1175,
      "step": 93
    },
    {
      "epoch": 0.08914177335229967,
      "grad_norm": 1.6337936419255448,
      "learning_rate": 2.9652996845425868e-06,
      "loss": 0.1148,
      "step": 94
    },
    {
      "epoch": 0.09009009009009009,
      "grad_norm": 1.7427850789821318,
      "learning_rate": 2.9968454258675085e-06,
      "loss": 0.1246,
      "step": 95
    },
    {
      "epoch": 0.09103840682788052,
      "grad_norm": 1.2870967429199511,
      "learning_rate": 3.0283911671924293e-06,
      "loss": 0.087,
      "step": 96
    },
    {
      "epoch": 0.09198672356567093,
      "grad_norm": 1.1311991406490958,
      "learning_rate": 3.05993690851735e-06,
      "loss": 0.0996,
      "step": 97
    },
    {
      "epoch": 0.09293504030346135,
      "grad_norm": 1.3742928108454626,
      "learning_rate": 3.091482649842272e-06,
      "loss": 0.1293,
      "step": 98
    },
    {
      "epoch": 0.09388335704125178,
      "grad_norm": 1.516028333204866,
      "learning_rate": 3.1230283911671926e-06,
      "loss": 0.1078,
      "step": 99
    },
    {
      "epoch": 0.0948316737790422,
      "grad_norm": 1.1404699421620854,
      "learning_rate": 3.1545741324921135e-06,
      "loss": 0.1142,
      "step": 100
    },
    {
      "epoch": 0.09577999051683263,
      "grad_norm": 1.7924070029612504,
      "learning_rate": 3.186119873817035e-06,
      "loss": 0.1419,
      "step": 101
    },
    {
      "epoch": 0.09672830725462304,
      "grad_norm": 1.7297435544466835,
      "learning_rate": 3.217665615141956e-06,
      "loss": 0.117,
      "step": 102
    },
    {
      "epoch": 0.09767662399241346,
      "grad_norm": 1.2304625316537265,
      "learning_rate": 3.2492113564668772e-06,
      "loss": 0.0834,
      "step": 103
    },
    {
      "epoch": 0.09862494073020389,
      "grad_norm": 1.6554297434059837,
      "learning_rate": 3.2807570977917985e-06,
      "loss": 0.1251,
      "step": 104
    },
    {
      "epoch": 0.09957325746799431,
      "grad_norm": 1.9749022078409877,
      "learning_rate": 3.3123028391167193e-06,
      "loss": 0.1485,
      "step": 105
    },
    {
      "epoch": 0.10052157420578474,
      "grad_norm": 1.7816458729316766,
      "learning_rate": 3.3438485804416405e-06,
      "loss": 0.1168,
      "step": 106
    },
    {
      "epoch": 0.10146989094357516,
      "grad_norm": 1.6366795934026652,
      "learning_rate": 3.375394321766562e-06,
      "loss": 0.1221,
      "step": 107
    },
    {
      "epoch": 0.10241820768136557,
      "grad_norm": 1.0846701931516913,
      "learning_rate": 3.4069400630914826e-06,
      "loss": 0.086,
      "step": 108
    },
    {
      "epoch": 0.103366524419156,
      "grad_norm": 1.374499027535318,
      "learning_rate": 3.4384858044164043e-06,
      "loss": 0.1097,
      "step": 109
    },
    {
      "epoch": 0.10431484115694642,
      "grad_norm": 1.7733166976712489,
      "learning_rate": 3.470031545741325e-06,
      "loss": 0.1098,
      "step": 110
    },
    {
      "epoch": 0.10526315789473684,
      "grad_norm": 2.980678296553409,
      "learning_rate": 3.5015772870662464e-06,
      "loss": 0.0917,
      "step": 111
    },
    {
      "epoch": 0.10621147463252727,
      "grad_norm": 0.9904577188744437,
      "learning_rate": 3.5331230283911676e-06,
      "loss": 0.0777,
      "step": 112
    },
    {
      "epoch": 0.10715979137031768,
      "grad_norm": 1.4107631975145143,
      "learning_rate": 3.5646687697160885e-06,
      "loss": 0.0902,
      "step": 113
    },
    {
      "epoch": 0.10810810810810811,
      "grad_norm": 1.786967710835369,
      "learning_rate": 3.5962145110410097e-06,
      "loss": 0.0974,
      "step": 114
    },
    {
      "epoch": 0.10905642484589853,
      "grad_norm": 1.6278373703409408,
      "learning_rate": 3.627760252365931e-06,
      "loss": 0.1031,
      "step": 115
    },
    {
      "epoch": 0.11000474158368895,
      "grad_norm": 1.0856958566494381,
      "learning_rate": 3.6593059936908522e-06,
      "loss": 0.0872,
      "step": 116
    },
    {
      "epoch": 0.11095305832147938,
      "grad_norm": 1.0746142572780863,
      "learning_rate": 3.690851735015773e-06,
      "loss": 0.0753,
      "step": 117
    },
    {
      "epoch": 0.1119013750592698,
      "grad_norm": 1.794687772916648,
      "learning_rate": 3.7223974763406943e-06,
      "loss": 0.094,
      "step": 118
    },
    {
      "epoch": 0.11284969179706021,
      "grad_norm": 2.0574961246450543,
      "learning_rate": 3.7539432176656156e-06,
      "loss": 0.1032,
      "step": 119
    },
    {
      "epoch": 0.11379800853485064,
      "grad_norm": 1.0887603543641189,
      "learning_rate": 3.7854889589905364e-06,
      "loss": 0.0869,
      "step": 120
    },
    {
      "epoch": 0.11474632527264106,
      "grad_norm": 1.4381581196511768,
      "learning_rate": 3.817034700315458e-06,
      "loss": 0.105,
      "step": 121
    },
    {
      "epoch": 0.11569464201043149,
      "grad_norm": 2.0884869908112984,
      "learning_rate": 3.8485804416403785e-06,
      "loss": 0.1072,
      "step": 122
    },
    {
      "epoch": 0.1166429587482219,
      "grad_norm": 1.3918016525882038,
      "learning_rate": 3.8801261829653e-06,
      "loss": 0.0995,
      "step": 123
    },
    {
      "epoch": 0.11759127548601232,
      "grad_norm": 1.1199618265144746,
      "learning_rate": 3.911671924290221e-06,
      "loss": 0.0693,
      "step": 124
    },
    {
      "epoch": 0.11853959222380275,
      "grad_norm": 2.913976726787567,
      "learning_rate": 3.943217665615142e-06,
      "loss": 0.1203,
      "step": 125
    },
    {
      "epoch": 0.11948790896159317,
      "grad_norm": 1.4548880466216083,
      "learning_rate": 3.9747634069400635e-06,
      "loss": 0.0891,
      "step": 126
    },
    {
      "epoch": 0.1204362256993836,
      "grad_norm": 3.1711394720986235,
      "learning_rate": 4.006309148264985e-06,
      "loss": 0.1223,
      "step": 127
    },
    {
      "epoch": 0.12138454243717402,
      "grad_norm": 1.888765811166245,
      "learning_rate": 4.037854889589905e-06,
      "loss": 0.13,
      "step": 128
    },
    {
      "epoch": 0.12233285917496443,
      "grad_norm": 1.2398551211997078,
      "learning_rate": 4.069400630914827e-06,
      "loss": 0.1103,
      "step": 129
    },
    {
      "epoch": 0.12328117591275486,
      "grad_norm": 1.7438322556304724,
      "learning_rate": 4.100946372239748e-06,
      "loss": 0.1147,
      "step": 130
    },
    {
      "epoch": 0.12422949265054528,
      "grad_norm": 0.9363387889716617,
      "learning_rate": 4.132492113564669e-06,
      "loss": 0.0995,
      "step": 131
    },
    {
      "epoch": 0.1251778093883357,
      "grad_norm": 1.446859084810851,
      "learning_rate": 4.16403785488959e-06,
      "loss": 0.0994,
      "step": 132
    },
    {
      "epoch": 0.12612612612612611,
      "grad_norm": 1.1856203072681963,
      "learning_rate": 4.195583596214511e-06,
      "loss": 0.0927,
      "step": 133
    },
    {
      "epoch": 0.12707444286391656,
      "grad_norm": 1.103336827372462,
      "learning_rate": 4.227129337539433e-06,
      "loss": 0.0815,
      "step": 134
    },
    {
      "epoch": 0.12802275960170698,
      "grad_norm": 1.897384655096208,
      "learning_rate": 4.258675078864354e-06,
      "loss": 0.1248,
      "step": 135
    },
    {
      "epoch": 0.1289710763394974,
      "grad_norm": 1.6223901695891558,
      "learning_rate": 4.290220820189275e-06,
      "loss": 0.1456,
      "step": 136
    },
    {
      "epoch": 0.1299193930772878,
      "grad_norm": 1.93689861193564,
      "learning_rate": 4.321766561514196e-06,
      "loss": 0.1236,
      "step": 137
    },
    {
      "epoch": 0.13086770981507823,
      "grad_norm": 1.7202395942479507,
      "learning_rate": 4.353312302839117e-06,
      "loss": 0.0994,
      "step": 138
    },
    {
      "epoch": 0.13181602655286867,
      "grad_norm": 2.1336251410837717,
      "learning_rate": 4.384858044164038e-06,
      "loss": 0.0963,
      "step": 139
    },
    {
      "epoch": 0.1327643432906591,
      "grad_norm": 2.086908025505944,
      "learning_rate": 4.416403785488959e-06,
      "loss": 0.1397,
      "step": 140
    },
    {
      "epoch": 0.1337126600284495,
      "grad_norm": 1.903049841336412,
      "learning_rate": 4.447949526813881e-06,
      "loss": 0.1188,
      "step": 141
    },
    {
      "epoch": 0.13466097676623992,
      "grad_norm": 1.237639055790405,
      "learning_rate": 4.479495268138802e-06,
      "loss": 0.0864,
      "step": 142
    },
    {
      "epoch": 0.13560929350403034,
      "grad_norm": 1.533833989919448,
      "learning_rate": 4.511041009463723e-06,
      "loss": 0.1188,
      "step": 143
    },
    {
      "epoch": 0.13655761024182078,
      "grad_norm": 1.7546010414420699,
      "learning_rate": 4.542586750788644e-06,
      "loss": 0.1364,
      "step": 144
    },
    {
      "epoch": 0.1375059269796112,
      "grad_norm": 2.9799276151902645,
      "learning_rate": 4.574132492113565e-06,
      "loss": 0.1226,
      "step": 145
    },
    {
      "epoch": 0.13845424371740161,
      "grad_norm": 1.1723876001348499,
      "learning_rate": 4.605678233438486e-06,
      "loss": 0.086,
      "step": 146
    },
    {
      "epoch": 0.13940256045519203,
      "grad_norm": 2.069220754870492,
      "learning_rate": 4.637223974763407e-06,
      "loss": 0.1196,
      "step": 147
    },
    {
      "epoch": 0.14035087719298245,
      "grad_norm": 3.9795001087139124,
      "learning_rate": 4.6687697160883285e-06,
      "loss": 0.1152,
      "step": 148
    },
    {
      "epoch": 0.14129919393077287,
      "grad_norm": 1.4634422746453415,
      "learning_rate": 4.70031545741325e-06,
      "loss": 0.0916,
      "step": 149
    },
    {
      "epoch": 0.1422475106685633,
      "grad_norm": 1.3185726560010742,
      "learning_rate": 4.731861198738171e-06,
      "loss": 0.0904,
      "step": 150
    },
    {
      "epoch": 0.14319582740635373,
      "grad_norm": 1.5552910898557228,
      "learning_rate": 4.7634069400630914e-06,
      "loss": 0.0899,
      "step": 151
    },
    {
      "epoch": 0.14414414414414414,
      "grad_norm": 1.0997805514097108,
      "learning_rate": 4.7949526813880135e-06,
      "loss": 0.0795,
      "step": 152
    },
    {
      "epoch": 0.14509246088193456,
      "grad_norm": 1.7076641753438397,
      "learning_rate": 4.826498422712934e-06,
      "loss": 0.1081,
      "step": 153
    },
    {
      "epoch": 0.14604077761972498,
      "grad_norm": 1.6735518250841006,
      "learning_rate": 4.858044164037855e-06,
      "loss": 0.1068,
      "step": 154
    },
    {
      "epoch": 0.14698909435751542,
      "grad_norm": 1.2033878521779449,
      "learning_rate": 4.8895899053627764e-06,
      "loss": 0.0934,
      "step": 155
    },
    {
      "epoch": 0.14793741109530584,
      "grad_norm": 1.4908718795559122,
      "learning_rate": 4.921135646687698e-06,
      "loss": 0.1072,
      "step": 156
    },
    {
      "epoch": 0.14888572783309625,
      "grad_norm": 1.3234990953707453,
      "learning_rate": 4.952681388012618e-06,
      "loss": 0.104,
      "step": 157
    },
    {
      "epoch": 0.14983404457088667,
      "grad_norm": 1.3354249814975963,
      "learning_rate": 4.98422712933754e-06,
      "loss": 0.1189,
      "step": 158
    },
    {
      "epoch": 0.1507823613086771,
      "grad_norm": 1.224445144859879,
      "learning_rate": 5.015772870662461e-06,
      "loss": 0.1114,
      "step": 159
    },
    {
      "epoch": 0.15173067804646753,
      "grad_norm": 1.3554440133569026,
      "learning_rate": 5.047318611987382e-06,
      "loss": 0.1278,
      "step": 160
    },
    {
      "epoch": 0.15267899478425795,
      "grad_norm": 1.4393478098545054,
      "learning_rate": 5.078864353312303e-06,
      "loss": 0.1201,
      "step": 161
    },
    {
      "epoch": 0.15362731152204837,
      "grad_norm": 1.032684740456474,
      "learning_rate": 5.110410094637225e-06,
      "loss": 0.0841,
      "step": 162
    },
    {
      "epoch": 0.15457562825983878,
      "grad_norm": 1.2508286920209446,
      "learning_rate": 5.141955835962146e-06,
      "loss": 0.0863,
      "step": 163
    },
    {
      "epoch": 0.1555239449976292,
      "grad_norm": 1.899093372512286,
      "learning_rate": 5.173501577287067e-06,
      "loss": 0.1662,
      "step": 164
    },
    {
      "epoch": 0.15647226173541964,
      "grad_norm": 1.4000014551423334,
      "learning_rate": 5.205047318611987e-06,
      "loss": 0.0909,
      "step": 165
    },
    {
      "epoch": 0.15742057847321006,
      "grad_norm": 1.9418542456678585,
      "learning_rate": 5.2365930599369085e-06,
      "loss": 0.1013,
      "step": 166
    },
    {
      "epoch": 0.15836889521100048,
      "grad_norm": 1.5538903766146939,
      "learning_rate": 5.268138801261831e-06,
      "loss": 0.1177,
      "step": 167
    },
    {
      "epoch": 0.1593172119487909,
      "grad_norm": 1.3035129364423688,
      "learning_rate": 5.299684542586752e-06,
      "loss": 0.0961,
      "step": 168
    },
    {
      "epoch": 0.1602655286865813,
      "grad_norm": 1.273421849890499,
      "learning_rate": 5.331230283911672e-06,
      "loss": 0.1252,
      "step": 169
    },
    {
      "epoch": 0.16121384542437173,
      "grad_norm": 1.123016604976548,
      "learning_rate": 5.3627760252365935e-06,
      "loss": 0.0999,
      "step": 170
    },
    {
      "epoch": 0.16216216216216217,
      "grad_norm": 1.2409364166994,
      "learning_rate": 5.394321766561515e-06,
      "loss": 0.1095,
      "step": 171
    },
    {
      "epoch": 0.1631104788999526,
      "grad_norm": 1.1082140455460585,
      "learning_rate": 5.425867507886435e-06,
      "loss": 0.0736,
      "step": 172
    },
    {
      "epoch": 0.164058795637743,
      "grad_norm": 1.2872459579560394,
      "learning_rate": 5.457413249211357e-06,
      "loss": 0.0928,
      "step": 173
    },
    {
      "epoch": 0.16500711237553342,
      "grad_norm": 1.3830237110418746,
      "learning_rate": 5.4889589905362786e-06,
      "loss": 0.0973,
      "step": 174
    },
    {
      "epoch": 0.16595542911332384,
      "grad_norm": 1.2546887092347754,
      "learning_rate": 5.520504731861199e-06,
      "loss": 0.0832,
      "step": 175
    },
    {
      "epoch": 0.16690374585111428,
      "grad_norm": 1.1708284069676944,
      "learning_rate": 5.55205047318612e-06,
      "loss": 0.1075,
      "step": 176
    },
    {
      "epoch": 0.1678520625889047,
      "grad_norm": 1.101853335061695,
      "learning_rate": 5.5835962145110415e-06,
      "loss": 0.0897,
      "step": 177
    },
    {
      "epoch": 0.16880037932669512,
      "grad_norm": 1.015907357215909,
      "learning_rate": 5.615141955835962e-06,
      "loss": 0.0819,
      "step": 178
    },
    {
      "epoch": 0.16974869606448553,
      "grad_norm": 1.8752154604515816,
      "learning_rate": 5.646687697160884e-06,
      "loss": 0.1021,
      "step": 179
    },
    {
      "epoch": 0.17069701280227595,
      "grad_norm": 1.6971011710183759,
      "learning_rate": 5.678233438485805e-06,
      "loss": 0.0996,
      "step": 180
    },
    {
      "epoch": 0.1716453295400664,
      "grad_norm": 1.2212507178791898,
      "learning_rate": 5.709779179810726e-06,
      "loss": 0.1079,
      "step": 181
    },
    {
      "epoch": 0.1725936462778568,
      "grad_norm": 1.7343284525300247,
      "learning_rate": 5.741324921135647e-06,
      "loss": 0.1292,
      "step": 182
    },
    {
      "epoch": 0.17354196301564723,
      "grad_norm": 1.4376592014404461,
      "learning_rate": 5.772870662460568e-06,
      "loss": 0.1312,
      "step": 183
    },
    {
      "epoch": 0.17449027975343764,
      "grad_norm": 1.2528619821880524,
      "learning_rate": 5.80441640378549e-06,
      "loss": 0.0762,
      "step": 184
    },
    {
      "epoch": 0.17543859649122806,
      "grad_norm": 1.9247297159171304,
      "learning_rate": 5.835962145110411e-06,
      "loss": 0.1403,
      "step": 185
    },
    {
      "epoch": 0.1763869132290185,
      "grad_norm": 1.5028101353474104,
      "learning_rate": 5.867507886435332e-06,
      "loss": 0.1147,
      "step": 186
    },
    {
      "epoch": 0.17733522996680892,
      "grad_norm": 2.4179600186213714,
      "learning_rate": 5.899053627760253e-06,
      "loss": 0.0913,
      "step": 187
    },
    {
      "epoch": 0.17828354670459934,
      "grad_norm": 1.518835105924909,
      "learning_rate": 5.9305993690851736e-06,
      "loss": 0.0918,
      "step": 188
    },
    {
      "epoch": 0.17923186344238975,
      "grad_norm": 1.6543687104918372,
      "learning_rate": 5.962145110410095e-06,
      "loss": 0.122,
      "step": 189
    },
    {
      "epoch": 0.18018018018018017,
      "grad_norm": 1.4531807393638785,
      "learning_rate": 5.993690851735017e-06,
      "loss": 0.1228,
      "step": 190
    },
    {
      "epoch": 0.1811284969179706,
      "grad_norm": 1.4665808153812976,
      "learning_rate": 6.025236593059937e-06,
      "loss": 0.1014,
      "step": 191
    },
    {
      "epoch": 0.18207681365576103,
      "grad_norm": 1.2889682170490027,
      "learning_rate": 6.056782334384859e-06,
      "loss": 0.1055,
      "step": 192
    },
    {
      "epoch": 0.18302513039355145,
      "grad_norm": 1.3310497561635966,
      "learning_rate": 6.08832807570978e-06,
      "loss": 0.119,
      "step": 193
    },
    {
      "epoch": 0.18397344713134187,
      "grad_norm": 1.3246051325093873,
      "learning_rate": 6.1198738170347e-06,
      "loss": 0.1288,
      "step": 194
    },
    {
      "epoch": 0.18492176386913228,
      "grad_norm": 1.1979924093987135,
      "learning_rate": 6.1514195583596215e-06,
      "loss": 0.0877,
      "step": 195
    },
    {
      "epoch": 0.1858700806069227,
      "grad_norm": 1.1280419900810446,
      "learning_rate": 6.182965299684544e-06,
      "loss": 0.1085,
      "step": 196
    },
    {
      "epoch": 0.18681839734471314,
      "grad_norm": 1.3307017446168579,
      "learning_rate": 6.214511041009465e-06,
      "loss": 0.0853,
      "step": 197
    },
    {
      "epoch": 0.18776671408250356,
      "grad_norm": 1.1814823672365349,
      "learning_rate": 6.246056782334385e-06,
      "loss": 0.1066,
      "step": 198
    },
    {
      "epoch": 0.18871503082029398,
      "grad_norm": 0.7829348670836794,
      "learning_rate": 6.2776025236593065e-06,
      "loss": 0.0662,
      "step": 199
    },
    {
      "epoch": 0.1896633475580844,
      "grad_norm": 1.2435224715978643,
      "learning_rate": 6.309148264984227e-06,
      "loss": 0.088,
      "step": 200
    },
    {
      "epoch": 0.1906116642958748,
      "grad_norm": 1.0014149948809556,
      "learning_rate": 6.340694006309149e-06,
      "loss": 0.0975,
      "step": 201
    },
    {
      "epoch": 0.19155998103366526,
      "grad_norm": 0.9250673471848995,
      "learning_rate": 6.37223974763407e-06,
      "loss": 0.0877,
      "step": 202
    },
    {
      "epoch": 0.19250829777145567,
      "grad_norm": 1.056412139362465,
      "learning_rate": 6.4037854889589915e-06,
      "loss": 0.0763,
      "step": 203
    },
    {
      "epoch": 0.1934566145092461,
      "grad_norm": 0.9891782097788515,
      "learning_rate": 6.435331230283912e-06,
      "loss": 0.0834,
      "step": 204
    },
    {
      "epoch": 0.1944049312470365,
      "grad_norm": 1.0792725374885792,
      "learning_rate": 6.466876971608833e-06,
      "loss": 0.0885,
      "step": 205
    },
    {
      "epoch": 0.19535324798482692,
      "grad_norm": 1.2366811021393578,
      "learning_rate": 6.4984227129337544e-06,
      "loss": 0.0954,
      "step": 206
    },
    {
      "epoch": 0.19630156472261737,
      "grad_norm": 1.024115365006771,
      "learning_rate": 6.529968454258676e-06,
      "loss": 0.1215,
      "step": 207
    },
    {
      "epoch": 0.19724988146040778,
      "grad_norm": 1.2203185957532192,
      "learning_rate": 6.561514195583597e-06,
      "loss": 0.1202,
      "step": 208
    },
    {
      "epoch": 0.1981981981981982,
      "grad_norm": 0.9501403270885721,
      "learning_rate": 6.593059936908518e-06,
      "loss": 0.0715,
      "step": 209
    },
    {
      "epoch": 0.19914651493598862,
      "grad_norm": 1.5511308370546482,
      "learning_rate": 6.624605678233439e-06,
      "loss": 0.1089,
      "step": 210
    },
    {
      "epoch": 0.20009483167377903,
      "grad_norm": 0.9433860573102355,
      "learning_rate": 6.65615141955836e-06,
      "loss": 0.0648,
      "step": 211
    },
    {
      "epoch": 0.20104314841156948,
      "grad_norm": 1.0981902231687461,
      "learning_rate": 6.687697160883281e-06,
      "loss": 0.0663,
      "step": 212
    },
    {
      "epoch": 0.2019914651493599,
      "grad_norm": 1.064443363672458,
      "learning_rate": 6.719242902208203e-06,
      "loss": 0.077,
      "step": 213
    },
    {
      "epoch": 0.2029397818871503,
      "grad_norm": 1.3753290546304533,
      "learning_rate": 6.750788643533124e-06,
      "loss": 0.1093,
      "step": 214
    },
    {
      "epoch": 0.20388809862494073,
      "grad_norm": 1.2200081175269764,
      "learning_rate": 6.782334384858045e-06,
      "loss": 0.1094,
      "step": 215
    },
    {
      "epoch": 0.20483641536273114,
      "grad_norm": 0.9141258918864384,
      "learning_rate": 6.813880126182965e-06,
      "loss": 0.0911,
      "step": 216
    },
    {
      "epoch": 0.20578473210052156,
      "grad_norm": 2.528170753397052,
      "learning_rate": 6.8454258675078865e-06,
      "loss": 0.1079,
      "step": 217
    },
    {
      "epoch": 0.206733048838312,
      "grad_norm": 1.4430688823297448,
      "learning_rate": 6.876971608832809e-06,
      "loss": 0.1053,
      "step": 218
    },
    {
      "epoch": 0.20768136557610242,
      "grad_norm": 1.0186932336289805,
      "learning_rate": 6.90851735015773e-06,
      "loss": 0.0861,
      "step": 219
    },
    {
      "epoch": 0.20862968231389284,
      "grad_norm": 1.1420742589304766,
      "learning_rate": 6.94006309148265e-06,
      "loss": 0.094,
      "step": 220
    },
    {
      "epoch": 0.20957799905168326,
      "grad_norm": 1.2741420533987797,
      "learning_rate": 6.9716088328075715e-06,
      "loss": 0.0951,
      "step": 221
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 0.9075216722351295,
      "learning_rate": 7.003154574132493e-06,
      "loss": 0.0866,
      "step": 222
    },
    {
      "epoch": 0.21147463252726412,
      "grad_norm": 1.1980754719122302,
      "learning_rate": 7.034700315457413e-06,
      "loss": 0.0914,
      "step": 223
    },
    {
      "epoch": 0.21242294926505453,
      "grad_norm": 1.1939921471415105,
      "learning_rate": 7.066246056782335e-06,
      "loss": 0.1047,
      "step": 224
    },
    {
      "epoch": 0.21337126600284495,
      "grad_norm": 0.8519438677271276,
      "learning_rate": 7.0977917981072565e-06,
      "loss": 0.0941,
      "step": 225
    },
    {
      "epoch": 0.21431958274063537,
      "grad_norm": 0.789532854502906,
      "learning_rate": 7.129337539432177e-06,
      "loss": 0.0819,
      "step": 226
    },
    {
      "epoch": 0.21526789947842578,
      "grad_norm": 1.2111156014392817,
      "learning_rate": 7.160883280757098e-06,
      "loss": 0.1027,
      "step": 227
    },
    {
      "epoch": 0.21621621621621623,
      "grad_norm": 1.0588737043402552,
      "learning_rate": 7.1924290220820195e-06,
      "loss": 0.0952,
      "step": 228
    },
    {
      "epoch": 0.21716453295400664,
      "grad_norm": 0.933483217055125,
      "learning_rate": 7.22397476340694e-06,
      "loss": 0.0763,
      "step": 229
    },
    {
      "epoch": 0.21811284969179706,
      "grad_norm": 1.049586247769339,
      "learning_rate": 7.255520504731862e-06,
      "loss": 0.0789,
      "step": 230
    },
    {
      "epoch": 0.21906116642958748,
      "grad_norm": 1.1220808424289264,
      "learning_rate": 7.287066246056783e-06,
      "loss": 0.074,
      "step": 231
    },
    {
      "epoch": 0.2200094831673779,
      "grad_norm": 1.254391611101815,
      "learning_rate": 7.3186119873817045e-06,
      "loss": 0.093,
      "step": 232
    },
    {
      "epoch": 0.22095779990516834,
      "grad_norm": 1.274839766592392,
      "learning_rate": 7.350157728706625e-06,
      "loss": 0.0938,
      "step": 233
    },
    {
      "epoch": 0.22190611664295876,
      "grad_norm": 1.2629251738997191,
      "learning_rate": 7.381703470031546e-06,
      "loss": 0.1129,
      "step": 234
    },
    {
      "epoch": 0.22285443338074917,
      "grad_norm": 1.3595829605121952,
      "learning_rate": 7.413249211356468e-06,
      "loss": 0.1062,
      "step": 235
    },
    {
      "epoch": 0.2238027501185396,
      "grad_norm": 1.353026352957774,
      "learning_rate": 7.444794952681389e-06,
      "loss": 0.117,
      "step": 236
    },
    {
      "epoch": 0.22475106685633,
      "grad_norm": 1.3472351125895725,
      "learning_rate": 7.47634069400631e-06,
      "loss": 0.0827,
      "step": 237
    },
    {
      "epoch": 0.22569938359412042,
      "grad_norm": 0.9510770172761661,
      "learning_rate": 7.507886435331231e-06,
      "loss": 0.0759,
      "step": 238
    },
    {
      "epoch": 0.22664770033191087,
      "grad_norm": 1.2025915899822757,
      "learning_rate": 7.5394321766561515e-06,
      "loss": 0.0807,
      "step": 239
    },
    {
      "epoch": 0.22759601706970128,
      "grad_norm": 1.1640028047547857,
      "learning_rate": 7.570977917981073e-06,
      "loss": 0.0709,
      "step": 240
    },
    {
      "epoch": 0.2285443338074917,
      "grad_norm": 1.5223127858935517,
      "learning_rate": 7.602523659305995e-06,
      "loss": 0.1018,
      "step": 241
    },
    {
      "epoch": 0.22949265054528212,
      "grad_norm": 1.8495916864800697,
      "learning_rate": 7.634069400630916e-06,
      "loss": 0.0968,
      "step": 242
    },
    {
      "epoch": 0.23044096728307253,
      "grad_norm": 1.8476848640745251,
      "learning_rate": 7.665615141955837e-06,
      "loss": 0.086,
      "step": 243
    },
    {
      "epoch": 0.23138928402086298,
      "grad_norm": 1.4644626825262619,
      "learning_rate": 7.697160883280757e-06,
      "loss": 0.0974,
      "step": 244
    },
    {
      "epoch": 0.2323376007586534,
      "grad_norm": 1.8857810882326624,
      "learning_rate": 7.728706624605679e-06,
      "loss": 0.1036,
      "step": 245
    },
    {
      "epoch": 0.2332859174964438,
      "grad_norm": 1.7638762752182895,
      "learning_rate": 7.7602523659306e-06,
      "loss": 0.1097,
      "step": 246
    },
    {
      "epoch": 0.23423423423423423,
      "grad_norm": 1.2348758426158113,
      "learning_rate": 7.791798107255522e-06,
      "loss": 0.0866,
      "step": 247
    },
    {
      "epoch": 0.23518255097202465,
      "grad_norm": 1.1223471436540764,
      "learning_rate": 7.823343848580442e-06,
      "loss": 0.0564,
      "step": 248
    },
    {
      "epoch": 0.2361308677098151,
      "grad_norm": 0.8821001750676984,
      "learning_rate": 7.854889589905364e-06,
      "loss": 0.0696,
      "step": 249
    },
    {
      "epoch": 0.2370791844476055,
      "grad_norm": 0.9899264223411232,
      "learning_rate": 7.886435331230284e-06,
      "loss": 0.0702,
      "step": 250
    },
    {
      "epoch": 0.23802750118539592,
      "grad_norm": 0.9289219027994224,
      "learning_rate": 7.917981072555205e-06,
      "loss": 0.0843,
      "step": 251
    },
    {
      "epoch": 0.23897581792318634,
      "grad_norm": 1.0579670590751298,
      "learning_rate": 7.949526813880127e-06,
      "loss": 0.0921,
      "step": 252
    },
    {
      "epoch": 0.23992413466097676,
      "grad_norm": 1.4593486745973783,
      "learning_rate": 7.981072555205049e-06,
      "loss": 0.1229,
      "step": 253
    },
    {
      "epoch": 0.2408724513987672,
      "grad_norm": 0.9496576247693762,
      "learning_rate": 8.01261829652997e-06,
      "loss": 0.0861,
      "step": 254
    },
    {
      "epoch": 0.24182076813655762,
      "grad_norm": 1.1030565317688061,
      "learning_rate": 8.04416403785489e-06,
      "loss": 0.0893,
      "step": 255
    },
    {
      "epoch": 0.24276908487434803,
      "grad_norm": 0.9907604990146169,
      "learning_rate": 8.07570977917981e-06,
      "loss": 0.0928,
      "step": 256
    },
    {
      "epoch": 0.24371740161213845,
      "grad_norm": 0.9460810229319789,
      "learning_rate": 8.107255520504732e-06,
      "loss": 0.0974,
      "step": 257
    },
    {
      "epoch": 0.24466571834992887,
      "grad_norm": 0.8329291976282354,
      "learning_rate": 8.138801261829655e-06,
      "loss": 0.077,
      "step": 258
    },
    {
      "epoch": 0.24561403508771928,
      "grad_norm": 0.8587085474520708,
      "learning_rate": 8.170347003154575e-06,
      "loss": 0.0837,
      "step": 259
    },
    {
      "epoch": 0.24656235182550973,
      "grad_norm": 0.9113223159844124,
      "learning_rate": 8.201892744479495e-06,
      "loss": 0.088,
      "step": 260
    },
    {
      "epoch": 0.24751066856330015,
      "grad_norm": 0.8328940868524983,
      "learning_rate": 8.233438485804417e-06,
      "loss": 0.091,
      "step": 261
    },
    {
      "epoch": 0.24845898530109056,
      "grad_norm": 1.4264090310082065,
      "learning_rate": 8.264984227129338e-06,
      "loss": 0.1354,
      "step": 262
    },
    {
      "epoch": 0.24940730203888098,
      "grad_norm": 1.0550225951223755,
      "learning_rate": 8.296529968454258e-06,
      "loss": 0.0972,
      "step": 263
    },
    {
      "epoch": 0.2503556187766714,
      "grad_norm": 1.053508559451355,
      "learning_rate": 8.32807570977918e-06,
      "loss": 0.1035,
      "step": 264
    },
    {
      "epoch": 0.25130393551446184,
      "grad_norm": 1.4971087544821369,
      "learning_rate": 8.359621451104102e-06,
      "loss": 0.1001,
      "step": 265
    },
    {
      "epoch": 0.25225225225225223,
      "grad_norm": 1.075521297085326,
      "learning_rate": 8.391167192429023e-06,
      "loss": 0.0923,
      "step": 266
    },
    {
      "epoch": 0.2532005689900427,
      "grad_norm": 1.6910075728505873,
      "learning_rate": 8.422712933753943e-06,
      "loss": 0.1212,
      "step": 267
    },
    {
      "epoch": 0.2541488857278331,
      "grad_norm": 1.5073460991202734,
      "learning_rate": 8.454258675078865e-06,
      "loss": 0.087,
      "step": 268
    },
    {
      "epoch": 0.2550972024656235,
      "grad_norm": 1.0201575671512444,
      "learning_rate": 8.485804416403787e-06,
      "loss": 0.0871,
      "step": 269
    },
    {
      "epoch": 0.25604551920341395,
      "grad_norm": 1.1193230353064818,
      "learning_rate": 8.517350157728708e-06,
      "loss": 0.1031,
      "step": 270
    },
    {
      "epoch": 0.25699383594120434,
      "grad_norm": 1.3593779355277376,
      "learning_rate": 8.548895899053628e-06,
      "loss": 0.0861,
      "step": 271
    },
    {
      "epoch": 0.2579421526789948,
      "grad_norm": 1.5824627519870196,
      "learning_rate": 8.58044164037855e-06,
      "loss": 0.0998,
      "step": 272
    },
    {
      "epoch": 0.25889046941678523,
      "grad_norm": 2.316620691088296,
      "learning_rate": 8.61198738170347e-06,
      "loss": 0.1237,
      "step": 273
    },
    {
      "epoch": 0.2598387861545756,
      "grad_norm": 1.3708391836342668,
      "learning_rate": 8.643533123028391e-06,
      "loss": 0.0806,
      "step": 274
    },
    {
      "epoch": 0.26078710289236606,
      "grad_norm": 1.259879695037933,
      "learning_rate": 8.675078864353313e-06,
      "loss": 0.088,
      "step": 275
    },
    {
      "epoch": 0.26173541963015645,
      "grad_norm": 1.236718933875791,
      "learning_rate": 8.706624605678234e-06,
      "loss": 0.0842,
      "step": 276
    },
    {
      "epoch": 0.2626837363679469,
      "grad_norm": 1.438488419989871,
      "learning_rate": 8.738170347003156e-06,
      "loss": 0.0955,
      "step": 277
    },
    {
      "epoch": 0.26363205310573734,
      "grad_norm": 0.9563516338397714,
      "learning_rate": 8.769716088328076e-06,
      "loss": 0.0761,
      "step": 278
    },
    {
      "epoch": 0.26458036984352773,
      "grad_norm": 1.2728124128011007,
      "learning_rate": 8.801261829652997e-06,
      "loss": 0.0805,
      "step": 279
    },
    {
      "epoch": 0.2655286865813182,
      "grad_norm": 1.2205595373118223,
      "learning_rate": 8.832807570977919e-06,
      "loss": 0.0879,
      "step": 280
    },
    {
      "epoch": 0.26647700331910856,
      "grad_norm": 0.959493141925286,
      "learning_rate": 8.86435331230284e-06,
      "loss": 0.0728,
      "step": 281
    },
    {
      "epoch": 0.267425320056899,
      "grad_norm": 1.4340945839201555,
      "learning_rate": 8.895899053627761e-06,
      "loss": 0.0897,
      "step": 282
    },
    {
      "epoch": 0.26837363679468945,
      "grad_norm": 1.0061297486879381,
      "learning_rate": 8.927444794952682e-06,
      "loss": 0.0857,
      "step": 283
    },
    {
      "epoch": 0.26932195353247984,
      "grad_norm": 1.5459293734675696,
      "learning_rate": 8.958990536277604e-06,
      "loss": 0.1029,
      "step": 284
    },
    {
      "epoch": 0.2702702702702703,
      "grad_norm": 1.3222303946698841,
      "learning_rate": 8.990536277602524e-06,
      "loss": 0.084,
      "step": 285
    },
    {
      "epoch": 0.2712185870080607,
      "grad_norm": 1.185863549947665,
      "learning_rate": 9.022082018927446e-06,
      "loss": 0.1311,
      "step": 286
    },
    {
      "epoch": 0.2721669037458511,
      "grad_norm": 0.8959238307125761,
      "learning_rate": 9.053627760252367e-06,
      "loss": 0.067,
      "step": 287
    },
    {
      "epoch": 0.27311522048364156,
      "grad_norm": 1.369443136318961,
      "learning_rate": 9.085173501577289e-06,
      "loss": 0.1093,
      "step": 288
    },
    {
      "epoch": 0.27406353722143195,
      "grad_norm": 1.1052390238476015,
      "learning_rate": 9.116719242902209e-06,
      "loss": 0.103,
      "step": 289
    },
    {
      "epoch": 0.2750118539592224,
      "grad_norm": 1.325059650748033,
      "learning_rate": 9.14826498422713e-06,
      "loss": 0.1111,
      "step": 290
    },
    {
      "epoch": 0.2759601706970128,
      "grad_norm": 1.3248936963910136,
      "learning_rate": 9.17981072555205e-06,
      "loss": 0.0933,
      "step": 291
    },
    {
      "epoch": 0.27690848743480323,
      "grad_norm": 1.127118183479871,
      "learning_rate": 9.211356466876972e-06,
      "loss": 0.0891,
      "step": 292
    },
    {
      "epoch": 0.2778568041725936,
      "grad_norm": 1.3108916887707827,
      "learning_rate": 9.242902208201894e-06,
      "loss": 0.0939,
      "step": 293
    },
    {
      "epoch": 0.27880512091038406,
      "grad_norm": 1.0013886049046197,
      "learning_rate": 9.274447949526815e-06,
      "loss": 0.0692,
      "step": 294
    },
    {
      "epoch": 0.2797534376481745,
      "grad_norm": 1.1156101698361054,
      "learning_rate": 9.305993690851735e-06,
      "loss": 0.0868,
      "step": 295
    },
    {
      "epoch": 0.2807017543859649,
      "grad_norm": 1.2522202479933553,
      "learning_rate": 9.337539432176657e-06,
      "loss": 0.0914,
      "step": 296
    },
    {
      "epoch": 0.28165007112375534,
      "grad_norm": 1.3755827124206237,
      "learning_rate": 9.369085173501577e-06,
      "loss": 0.0936,
      "step": 297
    },
    {
      "epoch": 0.28259838786154573,
      "grad_norm": 1.4694162511089293,
      "learning_rate": 9.4006309148265e-06,
      "loss": 0.1071,
      "step": 298
    },
    {
      "epoch": 0.2835467045993362,
      "grad_norm": 1.255879045911956,
      "learning_rate": 9.43217665615142e-06,
      "loss": 0.0815,
      "step": 299
    },
    {
      "epoch": 0.2844950213371266,
      "grad_norm": 1.560204819302283,
      "learning_rate": 9.463722397476342e-06,
      "loss": 0.1234,
      "step": 300
    },
    {
      "epoch": 0.285443338074917,
      "grad_norm": 1.0121817898281276,
      "learning_rate": 9.495268138801262e-06,
      "loss": 0.0595,
      "step": 301
    },
    {
      "epoch": 0.28639165481270745,
      "grad_norm": 1.0711466156341418,
      "learning_rate": 9.526813880126183e-06,
      "loss": 0.0641,
      "step": 302
    },
    {
      "epoch": 0.28733997155049784,
      "grad_norm": 1.1496695710149105,
      "learning_rate": 9.558359621451105e-06,
      "loss": 0.0761,
      "step": 303
    },
    {
      "epoch": 0.2882882882882883,
      "grad_norm": 1.2059272704315518,
      "learning_rate": 9.589905362776027e-06,
      "loss": 0.0756,
      "step": 304
    },
    {
      "epoch": 0.28923660502607873,
      "grad_norm": 1.0424292745296735,
      "learning_rate": 9.621451104100947e-06,
      "loss": 0.0855,
      "step": 305
    },
    {
      "epoch": 0.2901849217638691,
      "grad_norm": 1.1497786768197902,
      "learning_rate": 9.652996845425868e-06,
      "loss": 0.071,
      "step": 306
    },
    {
      "epoch": 0.29113323850165956,
      "grad_norm": 1.3472444992692172,
      "learning_rate": 9.68454258675079e-06,
      "loss": 0.0934,
      "step": 307
    },
    {
      "epoch": 0.29208155523944995,
      "grad_norm": 1.3345310370843513,
      "learning_rate": 9.71608832807571e-06,
      "loss": 0.0998,
      "step": 308
    },
    {
      "epoch": 0.2930298719772404,
      "grad_norm": 1.01109508034154,
      "learning_rate": 9.747634069400632e-06,
      "loss": 0.0762,
      "step": 309
    },
    {
      "epoch": 0.29397818871503084,
      "grad_norm": 0.9249973635125475,
      "learning_rate": 9.779179810725553e-06,
      "loss": 0.074,
      "step": 310
    },
    {
      "epoch": 0.29492650545282123,
      "grad_norm": 0.804446344253587,
      "learning_rate": 9.810725552050473e-06,
      "loss": 0.0517,
      "step": 311
    },
    {
      "epoch": 0.2958748221906117,
      "grad_norm": 0.965596925556689,
      "learning_rate": 9.842271293375395e-06,
      "loss": 0.098,
      "step": 312
    },
    {
      "epoch": 0.29682313892840206,
      "grad_norm": 2.012807451707843,
      "learning_rate": 9.873817034700316e-06,
      "loss": 0.1038,
      "step": 313
    },
    {
      "epoch": 0.2977714556661925,
      "grad_norm": 1.2864066063043205,
      "learning_rate": 9.905362776025236e-06,
      "loss": 0.1102,
      "step": 314
    },
    {
      "epoch": 0.29871977240398295,
      "grad_norm": 0.8775284858258785,
      "learning_rate": 9.936908517350158e-06,
      "loss": 0.0913,
      "step": 315
    },
    {
      "epoch": 0.29966808914177334,
      "grad_norm": 0.9395466275555749,
      "learning_rate": 9.96845425867508e-06,
      "loss": 0.1156,
      "step": 316
    },
    {
      "epoch": 0.3006164058795638,
      "grad_norm": 1.031977177693936,
      "learning_rate": 1e-05,
      "loss": 0.0772,
      "step": 317
    },
    {
      "epoch": 0.3015647226173542,
      "grad_norm": 0.906696222035988,
      "learning_rate": 9.999996951577431e-06,
      "loss": 0.0745,
      "step": 318
    },
    {
      "epoch": 0.3025130393551446,
      "grad_norm": 1.6486632782552955,
      "learning_rate": 9.999987806313436e-06,
      "loss": 0.1295,
      "step": 319
    },
    {
      "epoch": 0.30346135609293506,
      "grad_norm": 1.0682004904191784,
      "learning_rate": 9.999972564219169e-06,
      "loss": 0.089,
      "step": 320
    },
    {
      "epoch": 0.30440967283072545,
      "grad_norm": 1.0160084965418597,
      "learning_rate": 9.999951225313217e-06,
      "loss": 0.0795,
      "step": 321
    },
    {
      "epoch": 0.3053579895685159,
      "grad_norm": 1.1229797355618714,
      "learning_rate": 9.999923789621598e-06,
      "loss": 0.0924,
      "step": 322
    },
    {
      "epoch": 0.3063063063063063,
      "grad_norm": 0.9925832526069106,
      "learning_rate": 9.999890257177766e-06,
      "loss": 0.0803,
      "step": 323
    },
    {
      "epoch": 0.30725462304409673,
      "grad_norm": 1.1785860516178814,
      "learning_rate": 9.999850628022611e-06,
      "loss": 0.0797,
      "step": 324
    },
    {
      "epoch": 0.3082029397818872,
      "grad_norm": 1.1520304204509717,
      "learning_rate": 9.999804902204455e-06,
      "loss": 0.0775,
      "step": 325
    },
    {
      "epoch": 0.30915125651967756,
      "grad_norm": 1.0880132191910508,
      "learning_rate": 9.999753079779054e-06,
      "loss": 0.0906,
      "step": 326
    },
    {
      "epoch": 0.310099573257468,
      "grad_norm": 1.5767657455822397,
      "learning_rate": 9.999695160809598e-06,
      "loss": 0.0956,
      "step": 327
    },
    {
      "epoch": 0.3110478899952584,
      "grad_norm": 0.7125012678361342,
      "learning_rate": 9.999631145366713e-06,
      "loss": 0.0661,
      "step": 328
    },
    {
      "epoch": 0.31199620673304884,
      "grad_norm": 1.088584252037159,
      "learning_rate": 9.999561033528457e-06,
      "loss": 0.1149,
      "step": 329
    },
    {
      "epoch": 0.3129445234708393,
      "grad_norm": 0.8523222222870042,
      "learning_rate": 9.999484825380323e-06,
      "loss": 0.0913,
      "step": 330
    },
    {
      "epoch": 0.3138928402086297,
      "grad_norm": 1.0164571883774136,
      "learning_rate": 9.999402521015236e-06,
      "loss": 0.0878,
      "step": 331
    },
    {
      "epoch": 0.3148411569464201,
      "grad_norm": 0.7164573705993513,
      "learning_rate": 9.999314120533557e-06,
      "loss": 0.0866,
      "step": 332
    },
    {
      "epoch": 0.3157894736842105,
      "grad_norm": 0.7954216406429697,
      "learning_rate": 9.999219624043075e-06,
      "loss": 0.0702,
      "step": 333
    },
    {
      "epoch": 0.31673779042200095,
      "grad_norm": 0.7996263107367133,
      "learning_rate": 9.99911903165902e-06,
      "loss": 0.0758,
      "step": 334
    },
    {
      "epoch": 0.3176861071597914,
      "grad_norm": 1.101451187378474,
      "learning_rate": 9.999012343504049e-06,
      "loss": 0.0957,
      "step": 335
    },
    {
      "epoch": 0.3186344238975818,
      "grad_norm": 0.7265535166036453,
      "learning_rate": 9.998899559708254e-06,
      "loss": 0.0743,
      "step": 336
    },
    {
      "epoch": 0.31958274063537223,
      "grad_norm": 1.272801256055057,
      "learning_rate": 9.998780680409161e-06,
      "loss": 0.0952,
      "step": 337
    },
    {
      "epoch": 0.3205310573731626,
      "grad_norm": 0.8770881337944402,
      "learning_rate": 9.99865570575173e-06,
      "loss": 0.066,
      "step": 338
    },
    {
      "epoch": 0.32147937411095306,
      "grad_norm": 1.0607119132841634,
      "learning_rate": 9.998524635888347e-06,
      "loss": 0.0913,
      "step": 339
    },
    {
      "epoch": 0.32242769084874345,
      "grad_norm": 0.9189346974278031,
      "learning_rate": 9.998387470978837e-06,
      "loss": 0.0881,
      "step": 340
    },
    {
      "epoch": 0.3233760075865339,
      "grad_norm": 0.7272168469454553,
      "learning_rate": 9.998244211190454e-06,
      "loss": 0.0713,
      "step": 341
    },
    {
      "epoch": 0.32432432432432434,
      "grad_norm": 0.9819255696828616,
      "learning_rate": 9.998094856697885e-06,
      "loss": 0.0834,
      "step": 342
    },
    {
      "epoch": 0.32527264106211473,
      "grad_norm": 0.6857773270509248,
      "learning_rate": 9.997939407683249e-06,
      "loss": 0.0524,
      "step": 343
    },
    {
      "epoch": 0.3262209577999052,
      "grad_norm": 1.0324591704355464,
      "learning_rate": 9.99777786433609e-06,
      "loss": 0.1108,
      "step": 344
    },
    {
      "epoch": 0.32716927453769556,
      "grad_norm": 1.1264206703681527,
      "learning_rate": 9.997610226853399e-06,
      "loss": 0.0987,
      "step": 345
    },
    {
      "epoch": 0.328117591275486,
      "grad_norm": 0.95789066514891,
      "learning_rate": 9.997436495439581e-06,
      "loss": 0.093,
      "step": 346
    },
    {
      "epoch": 0.32906590801327645,
      "grad_norm": 1.0448222803112024,
      "learning_rate": 9.997256670306478e-06,
      "loss": 0.0983,
      "step": 347
    },
    {
      "epoch": 0.33001422475106684,
      "grad_norm": 0.7737283316563024,
      "learning_rate": 9.997070751673367e-06,
      "loss": 0.0706,
      "step": 348
    },
    {
      "epoch": 0.3309625414888573,
      "grad_norm": 0.9596984880180834,
      "learning_rate": 9.99687873976695e-06,
      "loss": 0.0991,
      "step": 349
    },
    {
      "epoch": 0.3319108582266477,
      "grad_norm": 0.8411109119380658,
      "learning_rate": 9.99668063482136e-06,
      "loss": 0.0678,
      "step": 350
    },
    {
      "epoch": 0.3328591749644381,
      "grad_norm": 1.136491883808786,
      "learning_rate": 9.996476437078162e-06,
      "loss": 0.0986,
      "step": 351
    },
    {
      "epoch": 0.33380749170222856,
      "grad_norm": 3.03438587624818,
      "learning_rate": 9.996266146786344e-06,
      "loss": 0.0969,
      "step": 352
    },
    {
      "epoch": 0.33475580844001895,
      "grad_norm": 1.2333568047254937,
      "learning_rate": 9.996049764202332e-06,
      "loss": 0.0832,
      "step": 353
    },
    {
      "epoch": 0.3357041251778094,
      "grad_norm": 1.1301139087376384,
      "learning_rate": 9.995827289589974e-06,
      "loss": 0.0994,
      "step": 354
    },
    {
      "epoch": 0.3366524419155998,
      "grad_norm": 1.0303329732235522,
      "learning_rate": 9.995598723220548e-06,
      "loss": 0.0757,
      "step": 355
    },
    {
      "epoch": 0.33760075865339023,
      "grad_norm": 1.0605991674508604,
      "learning_rate": 9.995364065372762e-06,
      "loss": 0.0815,
      "step": 356
    },
    {
      "epoch": 0.3385490753911807,
      "grad_norm": 0.7941030771981634,
      "learning_rate": 9.995123316332752e-06,
      "loss": 0.0747,
      "step": 357
    },
    {
      "epoch": 0.33949739212897106,
      "grad_norm": 1.2313896272302265,
      "learning_rate": 9.994876476394075e-06,
      "loss": 0.0769,
      "step": 358
    },
    {
      "epoch": 0.3404457088667615,
      "grad_norm": 1.1944743493159886,
      "learning_rate": 9.994623545857727e-06,
      "loss": 0.0979,
      "step": 359
    },
    {
      "epoch": 0.3413940256045519,
      "grad_norm": 0.8285281294809631,
      "learning_rate": 9.994364525032116e-06,
      "loss": 0.0793,
      "step": 360
    },
    {
      "epoch": 0.34234234234234234,
      "grad_norm": 1.4761389910370195,
      "learning_rate": 9.994099414233091e-06,
      "loss": 0.0913,
      "step": 361
    },
    {
      "epoch": 0.3432906590801328,
      "grad_norm": 1.5408966458771916,
      "learning_rate": 9.993828213783915e-06,
      "loss": 0.0973,
      "step": 362
    },
    {
      "epoch": 0.3442389758179232,
      "grad_norm": 1.4559933930399096,
      "learning_rate": 9.993550924015283e-06,
      "loss": 0.0999,
      "step": 363
    },
    {
      "epoch": 0.3451872925557136,
      "grad_norm": 0.8454336561992738,
      "learning_rate": 9.993267545265314e-06,
      "loss": 0.0655,
      "step": 364
    },
    {
      "epoch": 0.346135609293504,
      "grad_norm": 0.796992439441769,
      "learning_rate": 9.992978077879552e-06,
      "loss": 0.0696,
      "step": 365
    },
    {
      "epoch": 0.34708392603129445,
      "grad_norm": 1.0553149426590827,
      "learning_rate": 9.992682522210963e-06,
      "loss": 0.0787,
      "step": 366
    },
    {
      "epoch": 0.3480322427690849,
      "grad_norm": 1.4860431297237584,
      "learning_rate": 9.992380878619939e-06,
      "loss": 0.106,
      "step": 367
    },
    {
      "epoch": 0.3489805595068753,
      "grad_norm": 1.3032907057151817,
      "learning_rate": 9.992073147474292e-06,
      "loss": 0.1021,
      "step": 368
    },
    {
      "epoch": 0.34992887624466573,
      "grad_norm": 1.0894704335759804,
      "learning_rate": 9.991759329149266e-06,
      "loss": 0.0905,
      "step": 369
    },
    {
      "epoch": 0.3508771929824561,
      "grad_norm": 1.1130576081628205,
      "learning_rate": 9.991439424027518e-06,
      "loss": 0.0846,
      "step": 370
    },
    {
      "epoch": 0.35182550972024657,
      "grad_norm": 0.9253664091514998,
      "learning_rate": 9.991113432499128e-06,
      "loss": 0.0882,
      "step": 371
    },
    {
      "epoch": 0.352773826458037,
      "grad_norm": 0.841899923853967,
      "learning_rate": 9.990781354961605e-06,
      "loss": 0.0806,
      "step": 372
    },
    {
      "epoch": 0.3537221431958274,
      "grad_norm": 0.9407729946270026,
      "learning_rate": 9.99044319181987e-06,
      "loss": 0.0939,
      "step": 373
    },
    {
      "epoch": 0.35467045993361784,
      "grad_norm": 0.9090058769044609,
      "learning_rate": 9.99009894348627e-06,
      "loss": 0.0891,
      "step": 374
    },
    {
      "epoch": 0.35561877667140823,
      "grad_norm": 0.6294083333837054,
      "learning_rate": 9.989748610380571e-06,
      "loss": 0.0706,
      "step": 375
    },
    {
      "epoch": 0.3565670934091987,
      "grad_norm": 0.9163781177038506,
      "learning_rate": 9.98939219292996e-06,
      "loss": 0.0697,
      "step": 376
    },
    {
      "epoch": 0.3575154101469891,
      "grad_norm": 1.1693511630739546,
      "learning_rate": 9.989029691569037e-06,
      "loss": 0.1056,
      "step": 377
    },
    {
      "epoch": 0.3584637268847795,
      "grad_norm": 1.0414233510818562,
      "learning_rate": 9.988661106739827e-06,
      "loss": 0.0988,
      "step": 378
    },
    {
      "epoch": 0.35941204362256995,
      "grad_norm": 1.2822153621266594,
      "learning_rate": 9.988286438891774e-06,
      "loss": 0.1189,
      "step": 379
    },
    {
      "epoch": 0.36036036036036034,
      "grad_norm": 0.63669429794073,
      "learning_rate": 9.987905688481732e-06,
      "loss": 0.0828,
      "step": 380
    },
    {
      "epoch": 0.3613086770981508,
      "grad_norm": 0.826754093590745,
      "learning_rate": 9.98751885597398e-06,
      "loss": 0.0848,
      "step": 381
    },
    {
      "epoch": 0.3622569938359412,
      "grad_norm": 0.8825949393702691,
      "learning_rate": 9.987125941840205e-06,
      "loss": 0.092,
      "step": 382
    },
    {
      "epoch": 0.3632053105737316,
      "grad_norm": 0.6103241173744877,
      "learning_rate": 9.986726946559517e-06,
      "loss": 0.08,
      "step": 383
    },
    {
      "epoch": 0.36415362731152207,
      "grad_norm": 0.7105367439957658,
      "learning_rate": 9.986321870618441e-06,
      "loss": 0.0685,
      "step": 384
    },
    {
      "epoch": 0.36510194404931245,
      "grad_norm": 1.802287343988455,
      "learning_rate": 9.985910714510908e-06,
      "loss": 0.0818,
      "step": 385
    },
    {
      "epoch": 0.3660502607871029,
      "grad_norm": 0.7732813708584271,
      "learning_rate": 9.985493478738275e-06,
      "loss": 0.07,
      "step": 386
    },
    {
      "epoch": 0.3669985775248933,
      "grad_norm": 0.8451643375246307,
      "learning_rate": 9.985070163809306e-06,
      "loss": 0.0744,
      "step": 387
    },
    {
      "epoch": 0.36794689426268373,
      "grad_norm": 1.126067442650852,
      "learning_rate": 9.984640770240173e-06,
      "loss": 0.1101,
      "step": 388
    },
    {
      "epoch": 0.3688952110004742,
      "grad_norm": 0.6652401258855057,
      "learning_rate": 9.984205298554467e-06,
      "loss": 0.0663,
      "step": 389
    },
    {
      "epoch": 0.36984352773826457,
      "grad_norm": 1.0802552975196003,
      "learning_rate": 9.983763749283193e-06,
      "loss": 0.0975,
      "step": 390
    },
    {
      "epoch": 0.370791844476055,
      "grad_norm": 0.7496808510910429,
      "learning_rate": 9.983316122964757e-06,
      "loss": 0.0701,
      "step": 391
    },
    {
      "epoch": 0.3717401612138454,
      "grad_norm": 0.6248602765248035,
      "learning_rate": 9.982862420144986e-06,
      "loss": 0.0643,
      "step": 392
    },
    {
      "epoch": 0.37268847795163584,
      "grad_norm": 1.7058022738803864,
      "learning_rate": 9.982402641377105e-06,
      "loss": 0.0936,
      "step": 393
    },
    {
      "epoch": 0.3736367946894263,
      "grad_norm": 1.205579756742393,
      "learning_rate": 9.98193678722176e-06,
      "loss": 0.0811,
      "step": 394
    },
    {
      "epoch": 0.3745851114272167,
      "grad_norm": 0.8021701752607538,
      "learning_rate": 9.981464858246993e-06,
      "loss": 0.0719,
      "step": 395
    },
    {
      "epoch": 0.3755334281650071,
      "grad_norm": 0.9210208736552777,
      "learning_rate": 9.980986855028267e-06,
      "loss": 0.0589,
      "step": 396
    },
    {
      "epoch": 0.3764817449027975,
      "grad_norm": 1.0458476195224804,
      "learning_rate": 9.980502778148438e-06,
      "loss": 0.0696,
      "step": 397
    },
    {
      "epoch": 0.37743006164058795,
      "grad_norm": 1.5095103680379303,
      "learning_rate": 9.980012628197778e-06,
      "loss": 0.0909,
      "step": 398
    },
    {
      "epoch": 0.3783783783783784,
      "grad_norm": 0.9521689001456719,
      "learning_rate": 9.979516405773956e-06,
      "loss": 0.0844,
      "step": 399
    },
    {
      "epoch": 0.3793266951161688,
      "grad_norm": 0.9909335290642662,
      "learning_rate": 9.979014111482057e-06,
      "loss": 0.079,
      "step": 400
    },
    {
      "epoch": 0.38027501185395923,
      "grad_norm": 1.300023515267878,
      "learning_rate": 9.978505745934559e-06,
      "loss": 0.1087,
      "step": 401
    },
    {
      "epoch": 0.3812233285917496,
      "grad_norm": 0.8905160216053487,
      "learning_rate": 9.977991309751347e-06,
      "loss": 0.0654,
      "step": 402
    },
    {
      "epoch": 0.38217164532954007,
      "grad_norm": 0.7908744916198801,
      "learning_rate": 9.97747080355971e-06,
      "loss": 0.0697,
      "step": 403
    },
    {
      "epoch": 0.3831199620673305,
      "grad_norm": 1.0819522254088034,
      "learning_rate": 9.976944227994337e-06,
      "loss": 0.0729,
      "step": 404
    },
    {
      "epoch": 0.3840682788051209,
      "grad_norm": 0.9319836261266163,
      "learning_rate": 9.976411583697316e-06,
      "loss": 0.077,
      "step": 405
    },
    {
      "epoch": 0.38501659554291134,
      "grad_norm": 0.7209233770781128,
      "learning_rate": 9.97587287131814e-06,
      "loss": 0.0708,
      "step": 406
    },
    {
      "epoch": 0.38596491228070173,
      "grad_norm": 0.8430932582390814,
      "learning_rate": 9.975328091513696e-06,
      "loss": 0.07,
      "step": 407
    },
    {
      "epoch": 0.3869132290184922,
      "grad_norm": 0.7932090811238357,
      "learning_rate": 9.974777244948271e-06,
      "loss": 0.0648,
      "step": 408
    },
    {
      "epoch": 0.3878615457562826,
      "grad_norm": 0.9213278429313838,
      "learning_rate": 9.974220332293554e-06,
      "loss": 0.0737,
      "step": 409
    },
    {
      "epoch": 0.388809862494073,
      "grad_norm": 0.4369389269684112,
      "learning_rate": 9.973657354228623e-06,
      "loss": 0.0509,
      "step": 410
    },
    {
      "epoch": 0.38975817923186346,
      "grad_norm": 0.7988805293653696,
      "learning_rate": 9.973088311439957e-06,
      "loss": 0.0684,
      "step": 411
    },
    {
      "epoch": 0.39070649596965384,
      "grad_norm": 0.9648310793568026,
      "learning_rate": 9.97251320462143e-06,
      "loss": 0.0849,
      "step": 412
    },
    {
      "epoch": 0.3916548127074443,
      "grad_norm": 0.7585613690692753,
      "learning_rate": 9.97193203447431e-06,
      "loss": 0.077,
      "step": 413
    },
    {
      "epoch": 0.39260312944523473,
      "grad_norm": 0.9380377046145346,
      "learning_rate": 9.971344801707256e-06,
      "loss": 0.0771,
      "step": 414
    },
    {
      "epoch": 0.3935514461830251,
      "grad_norm": 0.9822247506181627,
      "learning_rate": 9.970751507036323e-06,
      "loss": 0.1123,
      "step": 415
    },
    {
      "epoch": 0.39449976292081557,
      "grad_norm": 0.7156423865364446,
      "learning_rate": 9.970152151184956e-06,
      "loss": 0.0801,
      "step": 416
    },
    {
      "epoch": 0.39544807965860596,
      "grad_norm": 1.05912629502688,
      "learning_rate": 9.96954673488399e-06,
      "loss": 0.0804,
      "step": 417
    },
    {
      "epoch": 0.3963963963963964,
      "grad_norm": 1.1230479850270394,
      "learning_rate": 9.968935258871652e-06,
      "loss": 0.0799,
      "step": 418
    },
    {
      "epoch": 0.39734471313418684,
      "grad_norm": 1.0054642393242061,
      "learning_rate": 9.968317723893556e-06,
      "loss": 0.082,
      "step": 419
    },
    {
      "epoch": 0.39829302987197723,
      "grad_norm": 1.227859524837509,
      "learning_rate": 9.967694130702706e-06,
      "loss": 0.1069,
      "step": 420
    },
    {
      "epoch": 0.3992413466097677,
      "grad_norm": 1.2136272659300074,
      "learning_rate": 9.96706448005949e-06,
      "loss": 0.1112,
      "step": 421
    },
    {
      "epoch": 0.40018966334755807,
      "grad_norm": 0.9692912194018656,
      "learning_rate": 9.96642877273169e-06,
      "loss": 0.0837,
      "step": 422
    },
    {
      "epoch": 0.4011379800853485,
      "grad_norm": 0.7181203670103851,
      "learning_rate": 9.965787009494458e-06,
      "loss": 0.0648,
      "step": 423
    },
    {
      "epoch": 0.40208629682313896,
      "grad_norm": 0.9389223502528147,
      "learning_rate": 9.96513919113035e-06,
      "loss": 0.0846,
      "step": 424
    },
    {
      "epoch": 0.40303461356092934,
      "grad_norm": 0.6566856036851983,
      "learning_rate": 9.964485318429292e-06,
      "loss": 0.0776,
      "step": 425
    },
    {
      "epoch": 0.4039829302987198,
      "grad_norm": 1.0028156563396406,
      "learning_rate": 9.963825392188595e-06,
      "loss": 0.0719,
      "step": 426
    },
    {
      "epoch": 0.4049312470365102,
      "grad_norm": 0.9682157984093804,
      "learning_rate": 9.963159413212952e-06,
      "loss": 0.1058,
      "step": 427
    },
    {
      "epoch": 0.4058795637743006,
      "grad_norm": 1.1561667939356075,
      "learning_rate": 9.96248738231444e-06,
      "loss": 0.0982,
      "step": 428
    },
    {
      "epoch": 0.406827880512091,
      "grad_norm": 0.7960344078481167,
      "learning_rate": 9.961809300312512e-06,
      "loss": 0.0643,
      "step": 429
    },
    {
      "epoch": 0.40777619724988146,
      "grad_norm": 0.914323773268032,
      "learning_rate": 9.961125168034e-06,
      "loss": 0.0835,
      "step": 430
    },
    {
      "epoch": 0.4087245139876719,
      "grad_norm": 0.7441869330920762,
      "learning_rate": 9.960434986313113e-06,
      "loss": 0.0559,
      "step": 431
    },
    {
      "epoch": 0.4096728307254623,
      "grad_norm": 2.4732017252552367,
      "learning_rate": 9.959738755991437e-06,
      "loss": 0.1445,
      "step": 432
    },
    {
      "epoch": 0.41062114746325273,
      "grad_norm": 0.8533585342555405,
      "learning_rate": 9.959036477917935e-06,
      "loss": 0.0575,
      "step": 433
    },
    {
      "epoch": 0.4115694642010431,
      "grad_norm": 0.8190438451317316,
      "learning_rate": 9.95832815294894e-06,
      "loss": 0.0794,
      "step": 434
    },
    {
      "epoch": 0.41251778093883357,
      "grad_norm": 1.0046620676404385,
      "learning_rate": 9.957613781948164e-06,
      "loss": 0.0686,
      "step": 435
    },
    {
      "epoch": 0.413466097676624,
      "grad_norm": 0.9887051267008984,
      "learning_rate": 9.956893365786691e-06,
      "loss": 0.0618,
      "step": 436
    },
    {
      "epoch": 0.4144144144144144,
      "grad_norm": 0.6105909207601089,
      "learning_rate": 9.95616690534297e-06,
      "loss": 0.0572,
      "step": 437
    },
    {
      "epoch": 0.41536273115220484,
      "grad_norm": 1.5234824479103468,
      "learning_rate": 9.955434401502825e-06,
      "loss": 0.0994,
      "step": 438
    },
    {
      "epoch": 0.41631104788999523,
      "grad_norm": 1.1295839815001452,
      "learning_rate": 9.954695855159454e-06,
      "loss": 0.073,
      "step": 439
    },
    {
      "epoch": 0.4172593646277857,
      "grad_norm": 0.6583329952843571,
      "learning_rate": 9.95395126721341e-06,
      "loss": 0.0699,
      "step": 440
    },
    {
      "epoch": 0.4182076813655761,
      "grad_norm": 0.955937586299997,
      "learning_rate": 9.953200638572625e-06,
      "loss": 0.0815,
      "step": 441
    },
    {
      "epoch": 0.4191559981033665,
      "grad_norm": 1.5323108400108396,
      "learning_rate": 9.95244397015239e-06,
      "loss": 0.0732,
      "step": 442
    },
    {
      "epoch": 0.42010431484115696,
      "grad_norm": 1.677920724371183,
      "learning_rate": 9.951681262875365e-06,
      "loss": 0.0944,
      "step": 443
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 0.8926328574943209,
      "learning_rate": 9.95091251767157e-06,
      "loss": 0.0731,
      "step": 444
    },
    {
      "epoch": 0.4220009483167378,
      "grad_norm": 1.2692898943255297,
      "learning_rate": 9.950137735478389e-06,
      "loss": 0.1029,
      "step": 445
    },
    {
      "epoch": 0.42294926505452823,
      "grad_norm": 0.7345506207483801,
      "learning_rate": 9.949356917240569e-06,
      "loss": 0.0748,
      "step": 446
    },
    {
      "epoch": 0.4238975817923186,
      "grad_norm": 1.2435473519034808,
      "learning_rate": 9.948570063910216e-06,
      "loss": 0.1009,
      "step": 447
    },
    {
      "epoch": 0.42484589853010907,
      "grad_norm": 0.7650866909769807,
      "learning_rate": 9.947777176446792e-06,
      "loss": 0.0746,
      "step": 448
    },
    {
      "epoch": 0.42579421526789946,
      "grad_norm": 1.3807429981979404,
      "learning_rate": 9.946978255817121e-06,
      "loss": 0.0701,
      "step": 449
    },
    {
      "epoch": 0.4267425320056899,
      "grad_norm": 0.5315623424461096,
      "learning_rate": 9.946173302995382e-06,
      "loss": 0.0574,
      "step": 450
    },
    {
      "epoch": 0.42769084874348035,
      "grad_norm": 0.8562951763201797,
      "learning_rate": 9.94536231896311e-06,
      "loss": 0.0951,
      "step": 451
    },
    {
      "epoch": 0.42863916548127073,
      "grad_norm": 1.1965590998104225,
      "learning_rate": 9.944545304709192e-06,
      "loss": 0.0877,
      "step": 452
    },
    {
      "epoch": 0.4295874822190612,
      "grad_norm": 1.2735339749816497,
      "learning_rate": 9.943722261229872e-06,
      "loss": 0.0768,
      "step": 453
    },
    {
      "epoch": 0.43053579895685157,
      "grad_norm": 0.9370658659046329,
      "learning_rate": 9.942893189528743e-06,
      "loss": 0.0782,
      "step": 454
    },
    {
      "epoch": 0.431484115694642,
      "grad_norm": 1.5520551397042521,
      "learning_rate": 9.942058090616748e-06,
      "loss": 0.1039,
      "step": 455
    },
    {
      "epoch": 0.43243243243243246,
      "grad_norm": 1.3529615602541014,
      "learning_rate": 9.941216965512183e-06,
      "loss": 0.0867,
      "step": 456
    },
    {
      "epoch": 0.43338074917022285,
      "grad_norm": 1.192234505990805,
      "learning_rate": 9.940369815240688e-06,
      "loss": 0.0809,
      "step": 457
    },
    {
      "epoch": 0.4343290659080133,
      "grad_norm": 0.9763205758532367,
      "learning_rate": 9.939516640835254e-06,
      "loss": 0.0652,
      "step": 458
    },
    {
      "epoch": 0.4352773826458037,
      "grad_norm": 1.3415645605638937,
      "learning_rate": 9.938657443336212e-06,
      "loss": 0.109,
      "step": 459
    },
    {
      "epoch": 0.4362256993835941,
      "grad_norm": 1.1595154129634277,
      "learning_rate": 9.937792223791244e-06,
      "loss": 0.1002,
      "step": 460
    },
    {
      "epoch": 0.43717401612138457,
      "grad_norm": 1.33436975844217,
      "learning_rate": 9.936920983255372e-06,
      "loss": 0.114,
      "step": 461
    },
    {
      "epoch": 0.43812233285917496,
      "grad_norm": 1.0009653043703806,
      "learning_rate": 9.936043722790956e-06,
      "loss": 0.0827,
      "step": 462
    },
    {
      "epoch": 0.4390706495969654,
      "grad_norm": 1.1900315382859075,
      "learning_rate": 9.935160443467704e-06,
      "loss": 0.0991,
      "step": 463
    },
    {
      "epoch": 0.4400189663347558,
      "grad_norm": 0.7796648666540394,
      "learning_rate": 9.934271146362658e-06,
      "loss": 0.0729,
      "step": 464
    },
    {
      "epoch": 0.44096728307254623,
      "grad_norm": 0.7692033539386839,
      "learning_rate": 9.933375832560199e-06,
      "loss": 0.0752,
      "step": 465
    },
    {
      "epoch": 0.4419155998103367,
      "grad_norm": 0.7898679053377281,
      "learning_rate": 9.932474503152047e-06,
      "loss": 0.0557,
      "step": 466
    },
    {
      "epoch": 0.44286391654812707,
      "grad_norm": 1.308054442070126,
      "learning_rate": 9.931567159237252e-06,
      "loss": 0.1,
      "step": 467
    },
    {
      "epoch": 0.4438122332859175,
      "grad_norm": 0.8281027248286734,
      "learning_rate": 9.930653801922205e-06,
      "loss": 0.1066,
      "step": 468
    },
    {
      "epoch": 0.4447605500237079,
      "grad_norm": 0.6589498594732086,
      "learning_rate": 9.929734432320621e-06,
      "loss": 0.061,
      "step": 469
    },
    {
      "epoch": 0.44570886676149835,
      "grad_norm": 1.0105820136512023,
      "learning_rate": 9.928809051553554e-06,
      "loss": 0.0771,
      "step": 470
    },
    {
      "epoch": 0.4466571834992888,
      "grad_norm": 1.174475732403723,
      "learning_rate": 9.927877660749385e-06,
      "loss": 0.1029,
      "step": 471
    },
    {
      "epoch": 0.4476055002370792,
      "grad_norm": 0.7007588523937572,
      "learning_rate": 9.92694026104382e-06,
      "loss": 0.0548,
      "step": 472
    },
    {
      "epoch": 0.4485538169748696,
      "grad_norm": 0.7548622992450297,
      "learning_rate": 9.925996853579897e-06,
      "loss": 0.071,
      "step": 473
    },
    {
      "epoch": 0.44950213371266,
      "grad_norm": 0.9151211373906433,
      "learning_rate": 9.92504743950798e-06,
      "loss": 0.0728,
      "step": 474
    },
    {
      "epoch": 0.45045045045045046,
      "grad_norm": 1.3188113799099948,
      "learning_rate": 9.924092019985751e-06,
      "loss": 0.071,
      "step": 475
    },
    {
      "epoch": 0.45139876718824085,
      "grad_norm": 0.834826643366671,
      "learning_rate": 9.923130596178221e-06,
      "loss": 0.0827,
      "step": 476
    },
    {
      "epoch": 0.4523470839260313,
      "grad_norm": 0.8853088211117691,
      "learning_rate": 9.922163169257722e-06,
      "loss": 0.0714,
      "step": 477
    },
    {
      "epoch": 0.45329540066382173,
      "grad_norm": 0.9773650061711494,
      "learning_rate": 9.921189740403902e-06,
      "loss": 0.0902,
      "step": 478
    },
    {
      "epoch": 0.4542437174016121,
      "grad_norm": 0.8530429782086267,
      "learning_rate": 9.92021031080373e-06,
      "loss": 0.0896,
      "step": 479
    },
    {
      "epoch": 0.45519203413940257,
      "grad_norm": 0.6841245724165017,
      "learning_rate": 9.919224881651494e-06,
      "loss": 0.0574,
      "step": 480
    },
    {
      "epoch": 0.45614035087719296,
      "grad_norm": 0.8751901827667304,
      "learning_rate": 9.918233454148795e-06,
      "loss": 0.0712,
      "step": 481
    },
    {
      "epoch": 0.4570886676149834,
      "grad_norm": 0.8605318101074332,
      "learning_rate": 9.917236029504549e-06,
      "loss": 0.0758,
      "step": 482
    },
    {
      "epoch": 0.45803698435277385,
      "grad_norm": 0.6297402738230038,
      "learning_rate": 9.916232608934982e-06,
      "loss": 0.0835,
      "step": 483
    },
    {
      "epoch": 0.45898530109056423,
      "grad_norm": 1.2633792305334934,
      "learning_rate": 9.915223193663639e-06,
      "loss": 0.097,
      "step": 484
    },
    {
      "epoch": 0.4599336178283547,
      "grad_norm": 0.9453282561376489,
      "learning_rate": 9.914207784921366e-06,
      "loss": 0.0813,
      "step": 485
    },
    {
      "epoch": 0.46088193456614507,
      "grad_norm": 1.0981998450683066,
      "learning_rate": 9.913186383946322e-06,
      "loss": 0.0831,
      "step": 486
    },
    {
      "epoch": 0.4618302513039355,
      "grad_norm": 0.9453607555522517,
      "learning_rate": 9.91215899198397e-06,
      "loss": 0.0668,
      "step": 487
    },
    {
      "epoch": 0.46277856804172596,
      "grad_norm": 0.8480655824160724,
      "learning_rate": 9.911125610287085e-06,
      "loss": 0.0803,
      "step": 488
    },
    {
      "epoch": 0.46372688477951635,
      "grad_norm": 0.7365032755805906,
      "learning_rate": 9.910086240115738e-06,
      "loss": 0.0503,
      "step": 489
    },
    {
      "epoch": 0.4646752015173068,
      "grad_norm": 0.9926545138390478,
      "learning_rate": 9.909040882737301e-06,
      "loss": 0.0785,
      "step": 490
    },
    {
      "epoch": 0.4656235182550972,
      "grad_norm": 1.078153469225969,
      "learning_rate": 9.907989539426455e-06,
      "loss": 0.0942,
      "step": 491
    },
    {
      "epoch": 0.4665718349928876,
      "grad_norm": 0.891582918999742,
      "learning_rate": 9.906932211465173e-06,
      "loss": 0.0713,
      "step": 492
    },
    {
      "epoch": 0.46752015173067807,
      "grad_norm": 0.8352029023952229,
      "learning_rate": 9.90586890014273e-06,
      "loss": 0.0871,
      "step": 493
    },
    {
      "epoch": 0.46846846846846846,
      "grad_norm": 1.4543230270611818,
      "learning_rate": 9.904799606755695e-06,
      "loss": 0.1049,
      "step": 494
    },
    {
      "epoch": 0.4694167852062589,
      "grad_norm": 0.9571877161884975,
      "learning_rate": 9.90372433260793e-06,
      "loss": 0.0856,
      "step": 495
    },
    {
      "epoch": 0.4703651019440493,
      "grad_norm": 0.6657483404024113,
      "learning_rate": 9.90264307901059e-06,
      "loss": 0.0631,
      "step": 496
    },
    {
      "epoch": 0.47131341868183974,
      "grad_norm": 1.2493973473928695,
      "learning_rate": 9.901555847282123e-06,
      "loss": 0.0973,
      "step": 497
    },
    {
      "epoch": 0.4722617354196302,
      "grad_norm": 0.6689914382563446,
      "learning_rate": 9.900462638748266e-06,
      "loss": 0.0582,
      "step": 498
    },
    {
      "epoch": 0.47321005215742057,
      "grad_norm": 0.8246501895880392,
      "learning_rate": 9.899363454742044e-06,
      "loss": 0.0727,
      "step": 499
    },
    {
      "epoch": 0.474158368895211,
      "grad_norm": 1.442170890658491,
      "learning_rate": 9.898258296603769e-06,
      "loss": 0.0931,
      "step": 500
    },
    {
      "epoch": 0.4751066856330014,
      "grad_norm": 0.7582565389247256,
      "learning_rate": 9.897147165681034e-06,
      "loss": 0.0722,
      "step": 501
    },
    {
      "epoch": 0.47605500237079185,
      "grad_norm": 0.627525129279453,
      "learning_rate": 9.896030063328718e-06,
      "loss": 0.0597,
      "step": 502
    },
    {
      "epoch": 0.4770033191085823,
      "grad_norm": 0.6342149242840518,
      "learning_rate": 9.894906990908982e-06,
      "loss": 0.0725,
      "step": 503
    },
    {
      "epoch": 0.4779516358463727,
      "grad_norm": 0.8212079234115165,
      "learning_rate": 9.893777949791266e-06,
      "loss": 0.0649,
      "step": 504
    },
    {
      "epoch": 0.4788999525841631,
      "grad_norm": 0.8923951454231676,
      "learning_rate": 9.89264294135229e-06,
      "loss": 0.0595,
      "step": 505
    },
    {
      "epoch": 0.4798482693219535,
      "grad_norm": 1.0318440665130484,
      "learning_rate": 9.891501966976041e-06,
      "loss": 0.0842,
      "step": 506
    },
    {
      "epoch": 0.48079658605974396,
      "grad_norm": 0.6944537972828242,
      "learning_rate": 9.890355028053793e-06,
      "loss": 0.0752,
      "step": 507
    },
    {
      "epoch": 0.4817449027975344,
      "grad_norm": 1.0705584030604105,
      "learning_rate": 9.889202125984088e-06,
      "loss": 0.0647,
      "step": 508
    },
    {
      "epoch": 0.4826932195353248,
      "grad_norm": 0.9754252622446561,
      "learning_rate": 9.88804326217274e-06,
      "loss": 0.0687,
      "step": 509
    },
    {
      "epoch": 0.48364153627311524,
      "grad_norm": 0.9660762094606946,
      "learning_rate": 9.886878438032828e-06,
      "loss": 0.0789,
      "step": 510
    },
    {
      "epoch": 0.4845898530109056,
      "grad_norm": 0.5832722133461282,
      "learning_rate": 9.885707654984703e-06,
      "loss": 0.0636,
      "step": 511
    },
    {
      "epoch": 0.48553816974869607,
      "grad_norm": 0.7052006552554221,
      "learning_rate": 9.884530914455984e-06,
      "loss": 0.0586,
      "step": 512
    },
    {
      "epoch": 0.4864864864864865,
      "grad_norm": 0.9822072228951928,
      "learning_rate": 9.88334821788155e-06,
      "loss": 0.0645,
      "step": 513
    },
    {
      "epoch": 0.4874348032242769,
      "grad_norm": 0.9641946540266126,
      "learning_rate": 9.882159566703547e-06,
      "loss": 0.0885,
      "step": 514
    },
    {
      "epoch": 0.48838311996206735,
      "grad_norm": 0.6403136140606015,
      "learning_rate": 9.880964962371378e-06,
      "loss": 0.0678,
      "step": 515
    },
    {
      "epoch": 0.48933143669985774,
      "grad_norm": 0.7486541793123711,
      "learning_rate": 9.879764406341705e-06,
      "loss": 0.0741,
      "step": 516
    },
    {
      "epoch": 0.4902797534376482,
      "grad_norm": 0.5779229700891555,
      "learning_rate": 9.87855790007845e-06,
      "loss": 0.0646,
      "step": 517
    },
    {
      "epoch": 0.49122807017543857,
      "grad_norm": 0.7611283230447122,
      "learning_rate": 9.87734544505279e-06,
      "loss": 0.0768,
      "step": 518
    },
    {
      "epoch": 0.492176386913229,
      "grad_norm": 0.5823535883100547,
      "learning_rate": 9.876127042743155e-06,
      "loss": 0.0703,
      "step": 519
    },
    {
      "epoch": 0.49312470365101946,
      "grad_norm": 0.6827829977739827,
      "learning_rate": 9.874902694635226e-06,
      "loss": 0.0772,
      "step": 520
    },
    {
      "epoch": 0.49407302038880985,
      "grad_norm": 0.7254200544564426,
      "learning_rate": 9.873672402221937e-06,
      "loss": 0.0634,
      "step": 521
    },
    {
      "epoch": 0.4950213371266003,
      "grad_norm": 0.6425214796651868,
      "learning_rate": 9.872436167003468e-06,
      "loss": 0.064,
      "step": 522
    },
    {
      "epoch": 0.4959696538643907,
      "grad_norm": 0.623192525545158,
      "learning_rate": 9.871193990487242e-06,
      "loss": 0.077,
      "step": 523
    },
    {
      "epoch": 0.4969179706021811,
      "grad_norm": 0.7225947749173619,
      "learning_rate": 9.869945874187936e-06,
      "loss": 0.075,
      "step": 524
    },
    {
      "epoch": 0.49786628733997157,
      "grad_norm": 2.0516616577595435,
      "learning_rate": 9.868691819627462e-06,
      "loss": 0.0867,
      "step": 525
    },
    {
      "epoch": 0.49881460407776196,
      "grad_norm": 1.0257158284306434,
      "learning_rate": 9.867431828334974e-06,
      "loss": 0.0588,
      "step": 526
    },
    {
      "epoch": 0.4997629208155524,
      "grad_norm": 0.8403229438927825,
      "learning_rate": 9.86616590184687e-06,
      "loss": 0.0823,
      "step": 527
    },
    {
      "epoch": 0.5007112375533428,
      "grad_norm": 0.6449240492145598,
      "learning_rate": 9.864894041706779e-06,
      "loss": 0.0567,
      "step": 528
    },
    {
      "epoch": 0.5016595542911333,
      "grad_norm": 0.8789018684523284,
      "learning_rate": 9.863616249465567e-06,
      "loss": 0.0713,
      "step": 529
    },
    {
      "epoch": 0.5026078710289237,
      "grad_norm": 0.9524887983478211,
      "learning_rate": 9.862332526681336e-06,
      "loss": 0.0835,
      "step": 530
    },
    {
      "epoch": 0.5035561877667141,
      "grad_norm": 0.6422268170348604,
      "learning_rate": 9.861042874919417e-06,
      "loss": 0.0606,
      "step": 531
    },
    {
      "epoch": 0.5045045045045045,
      "grad_norm": 0.9032374038451735,
      "learning_rate": 9.859747295752374e-06,
      "loss": 0.0773,
      "step": 532
    },
    {
      "epoch": 0.505452821242295,
      "grad_norm": 0.9269404822199643,
      "learning_rate": 9.858445790759992e-06,
      "loss": 0.0822,
      "step": 533
    },
    {
      "epoch": 0.5064011379800853,
      "grad_norm": 0.7043514434980399,
      "learning_rate": 9.857138361529288e-06,
      "loss": 0.0688,
      "step": 534
    },
    {
      "epoch": 0.5073494547178757,
      "grad_norm": 0.8239211698855243,
      "learning_rate": 9.8558250096545e-06,
      "loss": 0.0542,
      "step": 535
    },
    {
      "epoch": 0.5082977714556662,
      "grad_norm": 0.8633975590563754,
      "learning_rate": 9.85450573673709e-06,
      "loss": 0.0744,
      "step": 536
    },
    {
      "epoch": 0.5092460881934566,
      "grad_norm": 0.6985004021466871,
      "learning_rate": 9.853180544385737e-06,
      "loss": 0.047,
      "step": 537
    },
    {
      "epoch": 0.510194404931247,
      "grad_norm": 0.5889042803503781,
      "learning_rate": 9.851849434216338e-06,
      "loss": 0.0557,
      "step": 538
    },
    {
      "epoch": 0.5111427216690374,
      "grad_norm": 0.7765705663935071,
      "learning_rate": 9.850512407852012e-06,
      "loss": 0.0669,
      "step": 539
    },
    {
      "epoch": 0.5120910384068279,
      "grad_norm": 0.8204550382112847,
      "learning_rate": 9.849169466923086e-06,
      "loss": 0.0685,
      "step": 540
    },
    {
      "epoch": 0.5130393551446183,
      "grad_norm": 0.5256883407913393,
      "learning_rate": 9.847820613067098e-06,
      "loss": 0.0537,
      "step": 541
    },
    {
      "epoch": 0.5139876718824087,
      "grad_norm": 0.6838576750776693,
      "learning_rate": 9.8464658479288e-06,
      "loss": 0.0704,
      "step": 542
    },
    {
      "epoch": 0.5149359886201992,
      "grad_norm": 0.8974806559813661,
      "learning_rate": 9.845105173160152e-06,
      "loss": 0.0899,
      "step": 543
    },
    {
      "epoch": 0.5158843053579896,
      "grad_norm": 0.7219053990698988,
      "learning_rate": 9.843738590420317e-06,
      "loss": 0.0468,
      "step": 544
    },
    {
      "epoch": 0.51683262209578,
      "grad_norm": 1.032987889739876,
      "learning_rate": 9.842366101375664e-06,
      "loss": 0.0562,
      "step": 545
    },
    {
      "epoch": 0.5177809388335705,
      "grad_norm": 0.7651951768284668,
      "learning_rate": 9.840987707699765e-06,
      "loss": 0.0669,
      "step": 546
    },
    {
      "epoch": 0.5187292555713608,
      "grad_norm": 0.6813496832389402,
      "learning_rate": 9.839603411073388e-06,
      "loss": 0.0706,
      "step": 547
    },
    {
      "epoch": 0.5196775723091512,
      "grad_norm": 0.7229692269198181,
      "learning_rate": 9.838213213184505e-06,
      "loss": 0.0771,
      "step": 548
    },
    {
      "epoch": 0.5206258890469416,
      "grad_norm": 1.157471128375012,
      "learning_rate": 9.836817115728277e-06,
      "loss": 0.0932,
      "step": 549
    },
    {
      "epoch": 0.5215742057847321,
      "grad_norm": 0.8058138449457062,
      "learning_rate": 9.835415120407063e-06,
      "loss": 0.0539,
      "step": 550
    },
    {
      "epoch": 0.5225225225225225,
      "grad_norm": 0.6915528599019737,
      "learning_rate": 9.834007228930414e-06,
      "loss": 0.0688,
      "step": 551
    },
    {
      "epoch": 0.5234708392603129,
      "grad_norm": 0.8835152385091712,
      "learning_rate": 9.832593443015068e-06,
      "loss": 0.0605,
      "step": 552
    },
    {
      "epoch": 0.5244191559981034,
      "grad_norm": 0.6896706794263241,
      "learning_rate": 9.83117376438495e-06,
      "loss": 0.0668,
      "step": 553
    },
    {
      "epoch": 0.5253674727358938,
      "grad_norm": 0.7651857964351815,
      "learning_rate": 9.829748194771175e-06,
      "loss": 0.064,
      "step": 554
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 0.6216741056003758,
      "learning_rate": 9.828316735912037e-06,
      "loss": 0.0541,
      "step": 555
    },
    {
      "epoch": 0.5272641062114747,
      "grad_norm": 0.6813673301708452,
      "learning_rate": 9.826879389553014e-06,
      "loss": 0.0574,
      "step": 556
    },
    {
      "epoch": 0.5282124229492651,
      "grad_norm": 0.7147998418504048,
      "learning_rate": 9.825436157446761e-06,
      "loss": 0.0576,
      "step": 557
    },
    {
      "epoch": 0.5291607396870555,
      "grad_norm": 0.6352148290105686,
      "learning_rate": 9.82398704135311e-06,
      "loss": 0.066,
      "step": 558
    },
    {
      "epoch": 0.5301090564248458,
      "grad_norm": 0.8511240887028577,
      "learning_rate": 9.822532043039068e-06,
      "loss": 0.0687,
      "step": 559
    },
    {
      "epoch": 0.5310573731626363,
      "grad_norm": 0.6876408977841421,
      "learning_rate": 9.821071164278815e-06,
      "loss": 0.0838,
      "step": 560
    },
    {
      "epoch": 0.5320056899004267,
      "grad_norm": 0.7354217835184531,
      "learning_rate": 9.819604406853703e-06,
      "loss": 0.0552,
      "step": 561
    },
    {
      "epoch": 0.5329540066382171,
      "grad_norm": 0.9572067784227991,
      "learning_rate": 9.818131772552249e-06,
      "loss": 0.1099,
      "step": 562
    },
    {
      "epoch": 0.5339023233760076,
      "grad_norm": 0.7931127239607592,
      "learning_rate": 9.816653263170137e-06,
      "loss": 0.0706,
      "step": 563
    },
    {
      "epoch": 0.534850640113798,
      "grad_norm": 0.8242420526129728,
      "learning_rate": 9.815168880510218e-06,
      "loss": 0.0946,
      "step": 564
    },
    {
      "epoch": 0.5357989568515884,
      "grad_norm": 1.0330372476146157,
      "learning_rate": 9.8136786263825e-06,
      "loss": 0.0951,
      "step": 565
    },
    {
      "epoch": 0.5367472735893789,
      "grad_norm": 0.7553297432270302,
      "learning_rate": 9.812182502604151e-06,
      "loss": 0.0663,
      "step": 566
    },
    {
      "epoch": 0.5376955903271693,
      "grad_norm": 0.8446853429895546,
      "learning_rate": 9.810680510999505e-06,
      "loss": 0.0728,
      "step": 567
    },
    {
      "epoch": 0.5386439070649597,
      "grad_norm": 0.5089680701907852,
      "learning_rate": 9.809172653400036e-06,
      "loss": 0.0501,
      "step": 568
    },
    {
      "epoch": 0.5395922238027501,
      "grad_norm": 0.7258180066288827,
      "learning_rate": 9.807658931644382e-06,
      "loss": 0.0752,
      "step": 569
    },
    {
      "epoch": 0.5405405405405406,
      "grad_norm": 0.7028402619162881,
      "learning_rate": 9.806139347578331e-06,
      "loss": 0.059,
      "step": 570
    },
    {
      "epoch": 0.541488857278331,
      "grad_norm": 0.7248854010393692,
      "learning_rate": 9.804613903054813e-06,
      "loss": 0.0851,
      "step": 571
    },
    {
      "epoch": 0.5424371740161213,
      "grad_norm": 0.7176555652391681,
      "learning_rate": 9.803082599933911e-06,
      "loss": 0.0697,
      "step": 572
    },
    {
      "epoch": 0.5433854907539118,
      "grad_norm": 0.4808404612456389,
      "learning_rate": 9.801545440082845e-06,
      "loss": 0.0569,
      "step": 573
    },
    {
      "epoch": 0.5443338074917022,
      "grad_norm": 0.8731137568130377,
      "learning_rate": 9.800002425375984e-06,
      "loss": 0.0657,
      "step": 574
    },
    {
      "epoch": 0.5452821242294926,
      "grad_norm": 0.7816194292982013,
      "learning_rate": 9.798453557694828e-06,
      "loss": 0.0724,
      "step": 575
    },
    {
      "epoch": 0.5462304409672831,
      "grad_norm": 0.9042436959378762,
      "learning_rate": 9.796898838928022e-06,
      "loss": 0.0784,
      "step": 576
    },
    {
      "epoch": 0.5471787577050735,
      "grad_norm": 1.0293154765529384,
      "learning_rate": 9.79533827097134e-06,
      "loss": 0.098,
      "step": 577
    },
    {
      "epoch": 0.5481270744428639,
      "grad_norm": 0.8678391414260259,
      "learning_rate": 9.793771855727691e-06,
      "loss": 0.0635,
      "step": 578
    },
    {
      "epoch": 0.5490753911806543,
      "grad_norm": 0.6041409950077287,
      "learning_rate": 9.792199595107115e-06,
      "loss": 0.0524,
      "step": 579
    },
    {
      "epoch": 0.5500237079184448,
      "grad_norm": 1.0292476772898875,
      "learning_rate": 9.790621491026773e-06,
      "loss": 0.0829,
      "step": 580
    },
    {
      "epoch": 0.5509720246562352,
      "grad_norm": 0.7074515600768486,
      "learning_rate": 9.78903754541096e-06,
      "loss": 0.0704,
      "step": 581
    },
    {
      "epoch": 0.5519203413940256,
      "grad_norm": 0.7603340975922476,
      "learning_rate": 9.787447760191092e-06,
      "loss": 0.0788,
      "step": 582
    },
    {
      "epoch": 0.5528686581318161,
      "grad_norm": 1.0766706695954442,
      "learning_rate": 9.785852137305699e-06,
      "loss": 0.079,
      "step": 583
    },
    {
      "epoch": 0.5538169748696065,
      "grad_norm": 0.7555731931730972,
      "learning_rate": 9.784250678700435e-06,
      "loss": 0.0705,
      "step": 584
    },
    {
      "epoch": 0.5547652916073968,
      "grad_norm": 0.7010961175305198,
      "learning_rate": 9.782643386328073e-06,
      "loss": 0.0713,
      "step": 585
    },
    {
      "epoch": 0.5557136083451872,
      "grad_norm": 1.0580272254821363,
      "learning_rate": 9.781030262148492e-06,
      "loss": 0.0671,
      "step": 586
    },
    {
      "epoch": 0.5566619250829777,
      "grad_norm": 0.6594876081209583,
      "learning_rate": 9.779411308128685e-06,
      "loss": 0.0867,
      "step": 587
    },
    {
      "epoch": 0.5576102418207681,
      "grad_norm": 1.3649847896410103,
      "learning_rate": 9.777786526242759e-06,
      "loss": 0.0847,
      "step": 588
    },
    {
      "epoch": 0.5585585585585585,
      "grad_norm": 0.6223880228627037,
      "learning_rate": 9.776155918471916e-06,
      "loss": 0.0579,
      "step": 589
    },
    {
      "epoch": 0.559506875296349,
      "grad_norm": 0.6862572922646061,
      "learning_rate": 9.774519486804476e-06,
      "loss": 0.053,
      "step": 590
    },
    {
      "epoch": 0.5604551920341394,
      "grad_norm": 0.6562455064809456,
      "learning_rate": 9.772877233235848e-06,
      "loss": 0.0651,
      "step": 591
    },
    {
      "epoch": 0.5614035087719298,
      "grad_norm": 0.7150505236504866,
      "learning_rate": 9.771229159768547e-06,
      "loss": 0.0697,
      "step": 592
    },
    {
      "epoch": 0.5623518255097203,
      "grad_norm": 0.7505406859172821,
      "learning_rate": 9.769575268412182e-06,
      "loss": 0.0691,
      "step": 593
    },
    {
      "epoch": 0.5633001422475107,
      "grad_norm": 0.7340490905887499,
      "learning_rate": 9.767915561183456e-06,
      "loss": 0.0748,
      "step": 594
    },
    {
      "epoch": 0.5642484589853011,
      "grad_norm": 0.7987611706335997,
      "learning_rate": 9.766250040106166e-06,
      "loss": 0.0682,
      "step": 595
    },
    {
      "epoch": 0.5651967757230915,
      "grad_norm": 1.2974449597341617,
      "learning_rate": 9.764578707211199e-06,
      "loss": 0.0751,
      "step": 596
    },
    {
      "epoch": 0.566145092460882,
      "grad_norm": 0.6191420122018653,
      "learning_rate": 9.762901564536523e-06,
      "loss": 0.0667,
      "step": 597
    },
    {
      "epoch": 0.5670934091986723,
      "grad_norm": 0.6903639931399153,
      "learning_rate": 9.761218614127193e-06,
      "loss": 0.0653,
      "step": 598
    },
    {
      "epoch": 0.5680417259364627,
      "grad_norm": 0.7974449669867185,
      "learning_rate": 9.759529858035351e-06,
      "loss": 0.0662,
      "step": 599
    },
    {
      "epoch": 0.5689900426742532,
      "grad_norm": 1.6445977802603875,
      "learning_rate": 9.75783529832021e-06,
      "loss": 0.0781,
      "step": 600
    },
    {
      "epoch": 0.5699383594120436,
      "grad_norm": 0.7682344601188886,
      "learning_rate": 9.756134937048066e-06,
      "loss": 0.0516,
      "step": 601
    },
    {
      "epoch": 0.570886676149834,
      "grad_norm": 0.6505039594954853,
      "learning_rate": 9.754428776292287e-06,
      "loss": 0.0522,
      "step": 602
    },
    {
      "epoch": 0.5718349928876245,
      "grad_norm": 1.0748139183671632,
      "learning_rate": 9.752716818133309e-06,
      "loss": 0.0787,
      "step": 603
    },
    {
      "epoch": 0.5727833096254149,
      "grad_norm": 0.7575374337239762,
      "learning_rate": 9.750999064658644e-06,
      "loss": 0.0618,
      "step": 604
    },
    {
      "epoch": 0.5737316263632053,
      "grad_norm": 0.5005741056916544,
      "learning_rate": 9.749275517962868e-06,
      "loss": 0.0579,
      "step": 605
    },
    {
      "epoch": 0.5746799431009957,
      "grad_norm": 0.9747236186565804,
      "learning_rate": 9.747546180147618e-06,
      "loss": 0.1137,
      "step": 606
    },
    {
      "epoch": 0.5756282598387862,
      "grad_norm": 0.5945741852680105,
      "learning_rate": 9.745811053321597e-06,
      "loss": 0.0528,
      "step": 607
    },
    {
      "epoch": 0.5765765765765766,
      "grad_norm": 0.8767385416979725,
      "learning_rate": 9.744070139600564e-06,
      "loss": 0.0756,
      "step": 608
    },
    {
      "epoch": 0.577524893314367,
      "grad_norm": 0.805183732938404,
      "learning_rate": 9.742323441107335e-06,
      "loss": 0.0796,
      "step": 609
    },
    {
      "epoch": 0.5784732100521575,
      "grad_norm": 0.4622182813428181,
      "learning_rate": 9.74057095997178e-06,
      "loss": 0.0466,
      "step": 610
    },
    {
      "epoch": 0.5794215267899478,
      "grad_norm": 1.323185570736391,
      "learning_rate": 9.738812698330821e-06,
      "loss": 0.0803,
      "step": 611
    },
    {
      "epoch": 0.5803698435277382,
      "grad_norm": 0.6017510939556475,
      "learning_rate": 9.737048658328428e-06,
      "loss": 0.0473,
      "step": 612
    },
    {
      "epoch": 0.5813181602655287,
      "grad_norm": 0.9340483579893749,
      "learning_rate": 9.735278842115616e-06,
      "loss": 0.0726,
      "step": 613
    },
    {
      "epoch": 0.5822664770033191,
      "grad_norm": 0.8017302866486061,
      "learning_rate": 9.733503251850443e-06,
      "loss": 0.0508,
      "step": 614
    },
    {
      "epoch": 0.5832147937411095,
      "grad_norm": 0.4915103436956615,
      "learning_rate": 9.73172188969801e-06,
      "loss": 0.0511,
      "step": 615
    },
    {
      "epoch": 0.5841631104788999,
      "grad_norm": 0.5454251857464146,
      "learning_rate": 9.729934757830455e-06,
      "loss": 0.043,
      "step": 616
    },
    {
      "epoch": 0.5851114272166904,
      "grad_norm": 0.45382702737394764,
      "learning_rate": 9.728141858426953e-06,
      "loss": 0.046,
      "step": 617
    },
    {
      "epoch": 0.5860597439544808,
      "grad_norm": 0.5609546349379012,
      "learning_rate": 9.726343193673707e-06,
      "loss": 0.0528,
      "step": 618
    },
    {
      "epoch": 0.5870080606922712,
      "grad_norm": 0.600673482298699,
      "learning_rate": 9.724538765763953e-06,
      "loss": 0.0539,
      "step": 619
    },
    {
      "epoch": 0.5879563774300617,
      "grad_norm": 0.9417089865736203,
      "learning_rate": 9.722728576897956e-06,
      "loss": 0.0583,
      "step": 620
    },
    {
      "epoch": 0.5889046941678521,
      "grad_norm": 0.4653439643190733,
      "learning_rate": 9.720912629283004e-06,
      "loss": 0.05,
      "step": 621
    },
    {
      "epoch": 0.5898530109056425,
      "grad_norm": 1.026549188147293,
      "learning_rate": 9.719090925133408e-06,
      "loss": 0.0643,
      "step": 622
    },
    {
      "epoch": 0.590801327643433,
      "grad_norm": 0.7947545630855374,
      "learning_rate": 9.717263466670496e-06,
      "loss": 0.0827,
      "step": 623
    },
    {
      "epoch": 0.5917496443812233,
      "grad_norm": 0.5505357789361721,
      "learning_rate": 9.715430256122616e-06,
      "loss": 0.057,
      "step": 624
    },
    {
      "epoch": 0.5926979611190137,
      "grad_norm": 0.6227650085275758,
      "learning_rate": 9.713591295725126e-06,
      "loss": 0.0613,
      "step": 625
    },
    {
      "epoch": 0.5936462778568041,
      "grad_norm": 0.8089764410308476,
      "learning_rate": 9.711746587720398e-06,
      "loss": 0.0575,
      "step": 626
    },
    {
      "epoch": 0.5945945945945946,
      "grad_norm": 0.8681782262186932,
      "learning_rate": 9.709896134357815e-06,
      "loss": 0.0664,
      "step": 627
    },
    {
      "epoch": 0.595542911332385,
      "grad_norm": 0.682165737662686,
      "learning_rate": 9.708039937893759e-06,
      "loss": 0.0558,
      "step": 628
    },
    {
      "epoch": 0.5964912280701754,
      "grad_norm": 0.6331915650172267,
      "learning_rate": 9.706178000591617e-06,
      "loss": 0.0628,
      "step": 629
    },
    {
      "epoch": 0.5974395448079659,
      "grad_norm": 0.5712611189361939,
      "learning_rate": 9.704310324721782e-06,
      "loss": 0.0741,
      "step": 630
    },
    {
      "epoch": 0.5983878615457563,
      "grad_norm": 0.4974903145873453,
      "learning_rate": 9.70243691256164e-06,
      "loss": 0.0569,
      "step": 631
    },
    {
      "epoch": 0.5993361782835467,
      "grad_norm": 0.8755421451427193,
      "learning_rate": 9.700557766395567e-06,
      "loss": 0.0884,
      "step": 632
    },
    {
      "epoch": 0.6002844950213371,
      "grad_norm": 0.5236784076286586,
      "learning_rate": 9.698672888514938e-06,
      "loss": 0.0493,
      "step": 633
    },
    {
      "epoch": 0.6012328117591276,
      "grad_norm": 0.6525012362182552,
      "learning_rate": 9.696782281218117e-06,
      "loss": 0.0683,
      "step": 634
    },
    {
      "epoch": 0.602181128496918,
      "grad_norm": 0.5119217968942416,
      "learning_rate": 9.69488594681045e-06,
      "loss": 0.0449,
      "step": 635
    },
    {
      "epoch": 0.6031294452347084,
      "grad_norm": 0.6576021927278618,
      "learning_rate": 9.692983887604269e-06,
      "loss": 0.0674,
      "step": 636
    },
    {
      "epoch": 0.6040777619724989,
      "grad_norm": 0.7157400695119305,
      "learning_rate": 9.691076105918885e-06,
      "loss": 0.0692,
      "step": 637
    },
    {
      "epoch": 0.6050260787102892,
      "grad_norm": 0.873028935018846,
      "learning_rate": 9.689162604080589e-06,
      "loss": 0.0999,
      "step": 638
    },
    {
      "epoch": 0.6059743954480796,
      "grad_norm": 0.8384167589559871,
      "learning_rate": 9.687243384422646e-06,
      "loss": 0.0771,
      "step": 639
    },
    {
      "epoch": 0.6069227121858701,
      "grad_norm": 0.5020655439555515,
      "learning_rate": 9.685318449285292e-06,
      "loss": 0.0512,
      "step": 640
    },
    {
      "epoch": 0.6078710289236605,
      "grad_norm": 0.36608001502573706,
      "learning_rate": 9.683387801015733e-06,
      "loss": 0.0377,
      "step": 641
    },
    {
      "epoch": 0.6088193456614509,
      "grad_norm": 0.7919506442179929,
      "learning_rate": 9.681451441968144e-06,
      "loss": 0.0775,
      "step": 642
    },
    {
      "epoch": 0.6097676623992413,
      "grad_norm": 0.6274619623629013,
      "learning_rate": 9.67950937450366e-06,
      "loss": 0.0645,
      "step": 643
    },
    {
      "epoch": 0.6107159791370318,
      "grad_norm": 0.5896565427831529,
      "learning_rate": 9.677561600990378e-06,
      "loss": 0.0595,
      "step": 644
    },
    {
      "epoch": 0.6116642958748222,
      "grad_norm": 0.5142338666265971,
      "learning_rate": 9.67560812380335e-06,
      "loss": 0.0597,
      "step": 645
    },
    {
      "epoch": 0.6126126126126126,
      "grad_norm": 0.6109668570207277,
      "learning_rate": 9.67364894532459e-06,
      "loss": 0.07,
      "step": 646
    },
    {
      "epoch": 0.6135609293504031,
      "grad_norm": 0.6756478515313759,
      "learning_rate": 9.671684067943056e-06,
      "loss": 0.0612,
      "step": 647
    },
    {
      "epoch": 0.6145092460881935,
      "grad_norm": 0.6142876685386528,
      "learning_rate": 9.669713494054662e-06,
      "loss": 0.06,
      "step": 648
    },
    {
      "epoch": 0.6154575628259839,
      "grad_norm": 0.8252522199066464,
      "learning_rate": 9.667737226062262e-06,
      "loss": 0.118,
      "step": 649
    },
    {
      "epoch": 0.6164058795637744,
      "grad_norm": 0.48924053020562824,
      "learning_rate": 9.665755266375657e-06,
      "loss": 0.0542,
      "step": 650
    },
    {
      "epoch": 0.6173541963015647,
      "grad_norm": 0.9087121397095356,
      "learning_rate": 9.663767617411587e-06,
      "loss": 0.0611,
      "step": 651
    },
    {
      "epoch": 0.6183025130393551,
      "grad_norm": 0.7764764902550111,
      "learning_rate": 9.66177428159373e-06,
      "loss": 0.0676,
      "step": 652
    },
    {
      "epoch": 0.6192508297771455,
      "grad_norm": 0.44918893065172116,
      "learning_rate": 9.659775261352697e-06,
      "loss": 0.0474,
      "step": 653
    },
    {
      "epoch": 0.620199146514936,
      "grad_norm": 0.9162652994629981,
      "learning_rate": 9.657770559126034e-06,
      "loss": 0.0981,
      "step": 654
    },
    {
      "epoch": 0.6211474632527264,
      "grad_norm": 0.6543823860401999,
      "learning_rate": 9.655760177358208e-06,
      "loss": 0.0744,
      "step": 655
    },
    {
      "epoch": 0.6220957799905168,
      "grad_norm": 0.44085186666179094,
      "learning_rate": 9.653744118500623e-06,
      "loss": 0.0532,
      "step": 656
    },
    {
      "epoch": 0.6230440967283073,
      "grad_norm": 0.7980175435844092,
      "learning_rate": 9.651722385011592e-06,
      "loss": 0.0807,
      "step": 657
    },
    {
      "epoch": 0.6239924134660977,
      "grad_norm": 0.4853866988799319,
      "learning_rate": 9.649694979356358e-06,
      "loss": 0.0454,
      "step": 658
    },
    {
      "epoch": 0.6249407302038881,
      "grad_norm": 0.5662361885259662,
      "learning_rate": 9.647661904007076e-06,
      "loss": 0.0621,
      "step": 659
    },
    {
      "epoch": 0.6258890469416786,
      "grad_norm": 0.8127269026146419,
      "learning_rate": 9.645623161442814e-06,
      "loss": 0.0773,
      "step": 660
    },
    {
      "epoch": 0.626837363679469,
      "grad_norm": 0.6294162739235921,
      "learning_rate": 9.643578754149552e-06,
      "loss": 0.0599,
      "step": 661
    },
    {
      "epoch": 0.6277856804172594,
      "grad_norm": 0.6965237350859914,
      "learning_rate": 9.641528684620179e-06,
      "loss": 0.0542,
      "step": 662
    },
    {
      "epoch": 0.6287339971550497,
      "grad_norm": 0.5265921422928361,
      "learning_rate": 9.639472955354483e-06,
      "loss": 0.0496,
      "step": 663
    },
    {
      "epoch": 0.6296823138928402,
      "grad_norm": 0.8663040094375097,
      "learning_rate": 9.63741156885916e-06,
      "loss": 0.0733,
      "step": 664
    },
    {
      "epoch": 0.6306306306306306,
      "grad_norm": 0.7508837936313448,
      "learning_rate": 9.635344527647798e-06,
      "loss": 0.08,
      "step": 665
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 0.6827540936282853,
      "learning_rate": 9.633271834240885e-06,
      "loss": 0.0732,
      "step": 666
    },
    {
      "epoch": 0.6325272641062115,
      "grad_norm": 0.7441700461651841,
      "learning_rate": 9.631193491165798e-06,
      "loss": 0.0555,
      "step": 667
    },
    {
      "epoch": 0.6334755808440019,
      "grad_norm": 0.8313881844290032,
      "learning_rate": 9.629109500956803e-06,
      "loss": 0.0782,
      "step": 668
    },
    {
      "epoch": 0.6344238975817923,
      "grad_norm": 0.47754915650781987,
      "learning_rate": 9.627019866155056e-06,
      "loss": 0.0547,
      "step": 669
    },
    {
      "epoch": 0.6353722143195828,
      "grad_norm": 0.6618532396312571,
      "learning_rate": 9.624924589308591e-06,
      "loss": 0.0515,
      "step": 670
    },
    {
      "epoch": 0.6363205310573732,
      "grad_norm": 1.147117197534475,
      "learning_rate": 9.622823672972323e-06,
      "loss": 0.0882,
      "step": 671
    },
    {
      "epoch": 0.6372688477951636,
      "grad_norm": 0.5779383814129484,
      "learning_rate": 9.620717119708047e-06,
      "loss": 0.0659,
      "step": 672
    },
    {
      "epoch": 0.638217164532954,
      "grad_norm": 0.5799389859663083,
      "learning_rate": 9.618604932084427e-06,
      "loss": 0.0606,
      "step": 673
    },
    {
      "epoch": 0.6391654812707445,
      "grad_norm": 6.608545253943764,
      "learning_rate": 9.616487112677e-06,
      "loss": 0.066,
      "step": 674
    },
    {
      "epoch": 0.6401137980085349,
      "grad_norm": 0.7235578117181891,
      "learning_rate": 9.614363664068168e-06,
      "loss": 0.0628,
      "step": 675
    },
    {
      "epoch": 0.6410621147463252,
      "grad_norm": 0.6994528460712487,
      "learning_rate": 9.6122345888472e-06,
      "loss": 0.0628,
      "step": 676
    },
    {
      "epoch": 0.6420104314841157,
      "grad_norm": 0.6208663188504899,
      "learning_rate": 9.610099889610224e-06,
      "loss": 0.0554,
      "step": 677
    },
    {
      "epoch": 0.6429587482219061,
      "grad_norm": 0.6345977149189366,
      "learning_rate": 9.607959568960226e-06,
      "loss": 0.0632,
      "step": 678
    },
    {
      "epoch": 0.6439070649596965,
      "grad_norm": 0.8061055021711904,
      "learning_rate": 9.605813629507046e-06,
      "loss": 0.0684,
      "step": 679
    },
    {
      "epoch": 0.6448553816974869,
      "grad_norm": 0.6913423639588181,
      "learning_rate": 9.603662073867375e-06,
      "loss": 0.0673,
      "step": 680
    },
    {
      "epoch": 0.6458036984352774,
      "grad_norm": 0.7586179752230898,
      "learning_rate": 9.601504904664758e-06,
      "loss": 0.0702,
      "step": 681
    },
    {
      "epoch": 0.6467520151730678,
      "grad_norm": 0.5215807067369997,
      "learning_rate": 9.599342124529576e-06,
      "loss": 0.0484,
      "step": 682
    },
    {
      "epoch": 0.6477003319108582,
      "grad_norm": 0.4193899811291156,
      "learning_rate": 9.597173736099056e-06,
      "loss": 0.0455,
      "step": 683
    },
    {
      "epoch": 0.6486486486486487,
      "grad_norm": 1.0231627903377674,
      "learning_rate": 9.594999742017267e-06,
      "loss": 0.0755,
      "step": 684
    },
    {
      "epoch": 0.6495969653864391,
      "grad_norm": 0.5818860445113369,
      "learning_rate": 9.592820144935107e-06,
      "loss": 0.0457,
      "step": 685
    },
    {
      "epoch": 0.6505452821242295,
      "grad_norm": 0.8523614115619248,
      "learning_rate": 9.590634947510312e-06,
      "loss": 0.0666,
      "step": 686
    },
    {
      "epoch": 0.65149359886202,
      "grad_norm": 0.6819462318103672,
      "learning_rate": 9.588444152407441e-06,
      "loss": 0.0621,
      "step": 687
    },
    {
      "epoch": 0.6524419155998104,
      "grad_norm": 0.7350860734842137,
      "learning_rate": 9.586247762297882e-06,
      "loss": 0.0616,
      "step": 688
    },
    {
      "epoch": 0.6533902323376007,
      "grad_norm": 0.6877200427996193,
      "learning_rate": 9.584045779859848e-06,
      "loss": 0.0691,
      "step": 689
    },
    {
      "epoch": 0.6543385490753911,
      "grad_norm": 0.7777410132259543,
      "learning_rate": 9.581838207778367e-06,
      "loss": 0.0672,
      "step": 690
    },
    {
      "epoch": 0.6552868658131816,
      "grad_norm": 1.0340407583447775,
      "learning_rate": 9.579625048745281e-06,
      "loss": 0.0692,
      "step": 691
    },
    {
      "epoch": 0.656235182550972,
      "grad_norm": 0.6061769180463831,
      "learning_rate": 9.577406305459251e-06,
      "loss": 0.0519,
      "step": 692
    },
    {
      "epoch": 0.6571834992887624,
      "grad_norm": 0.7287017758175208,
      "learning_rate": 9.575181980625743e-06,
      "loss": 0.0626,
      "step": 693
    },
    {
      "epoch": 0.6581318160265529,
      "grad_norm": 0.6923184185544935,
      "learning_rate": 9.57295207695703e-06,
      "loss": 0.0602,
      "step": 694
    },
    {
      "epoch": 0.6590801327643433,
      "grad_norm": 0.7441802004305137,
      "learning_rate": 9.570716597172187e-06,
      "loss": 0.0785,
      "step": 695
    },
    {
      "epoch": 0.6600284495021337,
      "grad_norm": 0.5600328414907927,
      "learning_rate": 9.568475543997088e-06,
      "loss": 0.0525,
      "step": 696
    },
    {
      "epoch": 0.6609767662399242,
      "grad_norm": 0.6179093672887623,
      "learning_rate": 9.566228920164405e-06,
      "loss": 0.0498,
      "step": 697
    },
    {
      "epoch": 0.6619250829777146,
      "grad_norm": 1.0001632318997007,
      "learning_rate": 9.563976728413602e-06,
      "loss": 0.1065,
      "step": 698
    },
    {
      "epoch": 0.662873399715505,
      "grad_norm": 0.6197443639375237,
      "learning_rate": 9.56171897149093e-06,
      "loss": 0.0429,
      "step": 699
    },
    {
      "epoch": 0.6638217164532954,
      "grad_norm": 0.7426532648337794,
      "learning_rate": 9.55945565214943e-06,
      "loss": 0.0603,
      "step": 700
    },
    {
      "epoch": 0.6647700331910859,
      "grad_norm": 0.9809220324323352,
      "learning_rate": 9.557186773148922e-06,
      "loss": 0.0844,
      "step": 701
    },
    {
      "epoch": 0.6657183499288762,
      "grad_norm": 0.6596268576375636,
      "learning_rate": 9.554912337256007e-06,
      "loss": 0.0627,
      "step": 702
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.6445430375796782,
      "learning_rate": 9.552632347244062e-06,
      "loss": 0.0621,
      "step": 703
    },
    {
      "epoch": 0.6676149834044571,
      "grad_norm": 0.6854389668990125,
      "learning_rate": 9.550346805893236e-06,
      "loss": 0.0709,
      "step": 704
    },
    {
      "epoch": 0.6685633001422475,
      "grad_norm": 0.9157472924094435,
      "learning_rate": 9.548055715990448e-06,
      "loss": 0.0669,
      "step": 705
    },
    {
      "epoch": 0.6695116168800379,
      "grad_norm": 0.6210182918721243,
      "learning_rate": 9.545759080329381e-06,
      "loss": 0.0642,
      "step": 706
    },
    {
      "epoch": 0.6704599336178284,
      "grad_norm": 0.5811606762164421,
      "learning_rate": 9.543456901710483e-06,
      "loss": 0.0734,
      "step": 707
    },
    {
      "epoch": 0.6714082503556188,
      "grad_norm": 0.6797271720519124,
      "learning_rate": 9.541149182940958e-06,
      "loss": 0.0543,
      "step": 708
    },
    {
      "epoch": 0.6723565670934092,
      "grad_norm": 0.5126068611905316,
      "learning_rate": 9.538835926834766e-06,
      "loss": 0.0504,
      "step": 709
    },
    {
      "epoch": 0.6733048838311996,
      "grad_norm": 0.6464058845065579,
      "learning_rate": 9.536517136212623e-06,
      "loss": 0.0596,
      "step": 710
    },
    {
      "epoch": 0.6742532005689901,
      "grad_norm": 0.5987248394746172,
      "learning_rate": 9.534192813901986e-06,
      "loss": 0.0561,
      "step": 711
    },
    {
      "epoch": 0.6752015173067805,
      "grad_norm": 0.5757268664620699,
      "learning_rate": 9.531862962737065e-06,
      "loss": 0.0662,
      "step": 712
    },
    {
      "epoch": 0.6761498340445709,
      "grad_norm": 0.6884820373956889,
      "learning_rate": 9.529527585558806e-06,
      "loss": 0.0734,
      "step": 713
    },
    {
      "epoch": 0.6770981507823614,
      "grad_norm": 0.5599551362853026,
      "learning_rate": 9.5271866852149e-06,
      "loss": 0.0497,
      "step": 714
    },
    {
      "epoch": 0.6780464675201517,
      "grad_norm": 1.2727013612767513,
      "learning_rate": 9.524840264559762e-06,
      "loss": 0.0806,
      "step": 715
    },
    {
      "epoch": 0.6789947842579421,
      "grad_norm": 0.5125594480614294,
      "learning_rate": 9.522488326454551e-06,
      "loss": 0.0464,
      "step": 716
    },
    {
      "epoch": 0.6799431009957326,
      "grad_norm": 0.9279881234599379,
      "learning_rate": 9.520130873767141e-06,
      "loss": 0.0466,
      "step": 717
    },
    {
      "epoch": 0.680891417733523,
      "grad_norm": 0.5884738866592291,
      "learning_rate": 9.517767909372143e-06,
      "loss": 0.0463,
      "step": 718
    },
    {
      "epoch": 0.6818397344713134,
      "grad_norm": 0.6405987798189022,
      "learning_rate": 9.515399436150879e-06,
      "loss": 0.0646,
      "step": 719
    },
    {
      "epoch": 0.6827880512091038,
      "grad_norm": 0.6141893191288851,
      "learning_rate": 9.513025456991394e-06,
      "loss": 0.0713,
      "step": 720
    },
    {
      "epoch": 0.6837363679468943,
      "grad_norm": 0.5294631004623913,
      "learning_rate": 9.510645974788441e-06,
      "loss": 0.0533,
      "step": 721
    },
    {
      "epoch": 0.6846846846846847,
      "grad_norm": 0.5983803884552171,
      "learning_rate": 9.508260992443492e-06,
      "loss": 0.0574,
      "step": 722
    },
    {
      "epoch": 0.6856330014224751,
      "grad_norm": 0.7168015362345571,
      "learning_rate": 9.505870512864715e-06,
      "loss": 0.0622,
      "step": 723
    },
    {
      "epoch": 0.6865813181602656,
      "grad_norm": 0.8061703745318712,
      "learning_rate": 9.503474538966992e-06,
      "loss": 0.072,
      "step": 724
    },
    {
      "epoch": 0.687529634898056,
      "grad_norm": 0.6410612258118752,
      "learning_rate": 9.501073073671896e-06,
      "loss": 0.0454,
      "step": 725
    },
    {
      "epoch": 0.6884779516358464,
      "grad_norm": 0.790215058142473,
      "learning_rate": 9.498666119907701e-06,
      "loss": 0.0677,
      "step": 726
    },
    {
      "epoch": 0.6894262683736367,
      "grad_norm": 0.6299133472058956,
      "learning_rate": 9.496253680609371e-06,
      "loss": 0.0585,
      "step": 727
    },
    {
      "epoch": 0.6903745851114272,
      "grad_norm": 1.0623017139889208,
      "learning_rate": 9.493835758718561e-06,
      "loss": 0.069,
      "step": 728
    },
    {
      "epoch": 0.6913229018492176,
      "grad_norm": 0.5536012592608316,
      "learning_rate": 9.491412357183607e-06,
      "loss": 0.0686,
      "step": 729
    },
    {
      "epoch": 0.692271218587008,
      "grad_norm": 0.6038206755461478,
      "learning_rate": 9.488983478959534e-06,
      "loss": 0.0706,
      "step": 730
    },
    {
      "epoch": 0.6932195353247985,
      "grad_norm": 0.6342419868913964,
      "learning_rate": 9.486549127008037e-06,
      "loss": 0.0496,
      "step": 731
    },
    {
      "epoch": 0.6941678520625889,
      "grad_norm": 1.1555208683238716,
      "learning_rate": 9.484109304297493e-06,
      "loss": 0.0834,
      "step": 732
    },
    {
      "epoch": 0.6951161688003793,
      "grad_norm": 0.8509380581545992,
      "learning_rate": 9.481664013802943e-06,
      "loss": 0.0794,
      "step": 733
    },
    {
      "epoch": 0.6960644855381698,
      "grad_norm": 0.8224046322343856,
      "learning_rate": 9.479213258506102e-06,
      "loss": 0.0869,
      "step": 734
    },
    {
      "epoch": 0.6970128022759602,
      "grad_norm": 0.6505920471844966,
      "learning_rate": 9.476757041395342e-06,
      "loss": 0.0642,
      "step": 735
    },
    {
      "epoch": 0.6979611190137506,
      "grad_norm": 0.5162948092375159,
      "learning_rate": 9.474295365465697e-06,
      "loss": 0.0539,
      "step": 736
    },
    {
      "epoch": 0.698909435751541,
      "grad_norm": 0.7194486779836317,
      "learning_rate": 9.471828233718863e-06,
      "loss": 0.0585,
      "step": 737
    },
    {
      "epoch": 0.6998577524893315,
      "grad_norm": 0.9014549238602243,
      "learning_rate": 9.46935564916318e-06,
      "loss": 0.0874,
      "step": 738
    },
    {
      "epoch": 0.7008060692271219,
      "grad_norm": 0.7378312572460828,
      "learning_rate": 9.466877614813645e-06,
      "loss": 0.0657,
      "step": 739
    },
    {
      "epoch": 0.7017543859649122,
      "grad_norm": 0.815800968244944,
      "learning_rate": 9.464394133691891e-06,
      "loss": 0.0538,
      "step": 740
    },
    {
      "epoch": 0.7027027027027027,
      "grad_norm": 0.5271528573688194,
      "learning_rate": 9.461905208826202e-06,
      "loss": 0.0619,
      "step": 741
    },
    {
      "epoch": 0.7036510194404931,
      "grad_norm": 0.9062594050922635,
      "learning_rate": 9.459410843251496e-06,
      "loss": 0.0659,
      "step": 742
    },
    {
      "epoch": 0.7045993361782835,
      "grad_norm": 0.6578698656781865,
      "learning_rate": 9.456911040009323e-06,
      "loss": 0.0577,
      "step": 743
    },
    {
      "epoch": 0.705547652916074,
      "grad_norm": 0.6791351680766123,
      "learning_rate": 9.454405802147864e-06,
      "loss": 0.0669,
      "step": 744
    },
    {
      "epoch": 0.7064959696538644,
      "grad_norm": 0.7662019136887008,
      "learning_rate": 9.451895132721933e-06,
      "loss": 0.0692,
      "step": 745
    },
    {
      "epoch": 0.7074442863916548,
      "grad_norm": 0.6997379483885225,
      "learning_rate": 9.449379034792961e-06,
      "loss": 0.0609,
      "step": 746
    },
    {
      "epoch": 0.7083926031294452,
      "grad_norm": 0.6231531262832446,
      "learning_rate": 9.446857511429e-06,
      "loss": 0.0568,
      "step": 747
    },
    {
      "epoch": 0.7093409198672357,
      "grad_norm": 0.638618143024491,
      "learning_rate": 9.444330565704715e-06,
      "loss": 0.0391,
      "step": 748
    },
    {
      "epoch": 0.7102892366050261,
      "grad_norm": 0.6101709327712237,
      "learning_rate": 9.441798200701388e-06,
      "loss": 0.0692,
      "step": 749
    },
    {
      "epoch": 0.7112375533428165,
      "grad_norm": 0.7771396965466206,
      "learning_rate": 9.439260419506906e-06,
      "loss": 0.0616,
      "step": 750
    },
    {
      "epoch": 0.712185870080607,
      "grad_norm": 0.663533581873393,
      "learning_rate": 9.436717225215761e-06,
      "loss": 0.0706,
      "step": 751
    },
    {
      "epoch": 0.7131341868183974,
      "grad_norm": 0.7406791150442034,
      "learning_rate": 9.434168620929045e-06,
      "loss": 0.0759,
      "step": 752
    },
    {
      "epoch": 0.7140825035561877,
      "grad_norm": 0.6589932311994989,
      "learning_rate": 9.431614609754446e-06,
      "loss": 0.0676,
      "step": 753
    },
    {
      "epoch": 0.7150308202939782,
      "grad_norm": 0.7873737037891946,
      "learning_rate": 9.429055194806247e-06,
      "loss": 0.0661,
      "step": 754
    },
    {
      "epoch": 0.7159791370317686,
      "grad_norm": 0.6588547169267579,
      "learning_rate": 9.42649037920532e-06,
      "loss": 0.068,
      "step": 755
    },
    {
      "epoch": 0.716927453769559,
      "grad_norm": 0.8208102856389554,
      "learning_rate": 9.423920166079122e-06,
      "loss": 0.0829,
      "step": 756
    },
    {
      "epoch": 0.7178757705073494,
      "grad_norm": 0.5652492127213,
      "learning_rate": 9.421344558561689e-06,
      "loss": 0.0754,
      "step": 757
    },
    {
      "epoch": 0.7188240872451399,
      "grad_norm": 2.03543668980321,
      "learning_rate": 9.418763559793639e-06,
      "loss": 0.0469,
      "step": 758
    },
    {
      "epoch": 0.7197724039829303,
      "grad_norm": 0.7132600676949169,
      "learning_rate": 9.41617717292216e-06,
      "loss": 0.058,
      "step": 759
    },
    {
      "epoch": 0.7207207207207207,
      "grad_norm": 0.5814418519545377,
      "learning_rate": 9.413585401101014e-06,
      "loss": 0.0676,
      "step": 760
    },
    {
      "epoch": 0.7216690374585112,
      "grad_norm": 0.778087468578043,
      "learning_rate": 9.410988247490527e-06,
      "loss": 0.0565,
      "step": 761
    },
    {
      "epoch": 0.7226173541963016,
      "grad_norm": 0.5978506887698309,
      "learning_rate": 9.408385715257589e-06,
      "loss": 0.0526,
      "step": 762
    },
    {
      "epoch": 0.723565670934092,
      "grad_norm": 0.7345386180038043,
      "learning_rate": 9.405777807575643e-06,
      "loss": 0.0779,
      "step": 763
    },
    {
      "epoch": 0.7245139876718824,
      "grad_norm": 0.6765882263629432,
      "learning_rate": 9.403164527624695e-06,
      "loss": 0.0739,
      "step": 764
    },
    {
      "epoch": 0.7254623044096729,
      "grad_norm": 0.6200059319183251,
      "learning_rate": 9.400545878591297e-06,
      "loss": 0.0425,
      "step": 765
    },
    {
      "epoch": 0.7264106211474632,
      "grad_norm": 0.5764913642807622,
      "learning_rate": 9.397921863668545e-06,
      "loss": 0.0525,
      "step": 766
    },
    {
      "epoch": 0.7273589378852536,
      "grad_norm": 0.5072870053545583,
      "learning_rate": 9.395292486056087e-06,
      "loss": 0.0466,
      "step": 767
    },
    {
      "epoch": 0.7283072546230441,
      "grad_norm": 0.6266493674563252,
      "learning_rate": 9.3926577489601e-06,
      "loss": 0.0564,
      "step": 768
    },
    {
      "epoch": 0.7292555713608345,
      "grad_norm": 0.6781903020718192,
      "learning_rate": 9.390017655593303e-06,
      "loss": 0.0625,
      "step": 769
    },
    {
      "epoch": 0.7302038880986249,
      "grad_norm": 0.6970906328583575,
      "learning_rate": 9.387372209174943e-06,
      "loss": 0.0499,
      "step": 770
    },
    {
      "epoch": 0.7311522048364154,
      "grad_norm": 0.4830643779006922,
      "learning_rate": 9.384721412930797e-06,
      "loss": 0.0522,
      "step": 771
    },
    {
      "epoch": 0.7321005215742058,
      "grad_norm": 0.5981146539751457,
      "learning_rate": 9.382065270093164e-06,
      "loss": 0.0503,
      "step": 772
    },
    {
      "epoch": 0.7330488383119962,
      "grad_norm": 0.6288690777841561,
      "learning_rate": 9.37940378390086e-06,
      "loss": 0.0505,
      "step": 773
    },
    {
      "epoch": 0.7339971550497866,
      "grad_norm": 0.6043657243192845,
      "learning_rate": 9.376736957599219e-06,
      "loss": 0.048,
      "step": 774
    },
    {
      "epoch": 0.7349454717875771,
      "grad_norm": 1.3199303132586044,
      "learning_rate": 9.37406479444009e-06,
      "loss": 0.0787,
      "step": 775
    },
    {
      "epoch": 0.7358937885253675,
      "grad_norm": 0.9970354985082576,
      "learning_rate": 9.37138729768182e-06,
      "loss": 0.0593,
      "step": 776
    },
    {
      "epoch": 0.7368421052631579,
      "grad_norm": 0.6154243426982743,
      "learning_rate": 9.36870447058927e-06,
      "loss": 0.0552,
      "step": 777
    },
    {
      "epoch": 0.7377904220009484,
      "grad_norm": 0.688917247579616,
      "learning_rate": 9.366016316433796e-06,
      "loss": 0.0688,
      "step": 778
    },
    {
      "epoch": 0.7387387387387387,
      "grad_norm": 0.8890574424533809,
      "learning_rate": 9.363322838493252e-06,
      "loss": 0.0616,
      "step": 779
    },
    {
      "epoch": 0.7396870554765291,
      "grad_norm": 0.5256518464793154,
      "learning_rate": 9.360624040051975e-06,
      "loss": 0.0449,
      "step": 780
    },
    {
      "epoch": 0.7406353722143196,
      "grad_norm": 0.7015686604630017,
      "learning_rate": 9.357919924400802e-06,
      "loss": 0.0744,
      "step": 781
    },
    {
      "epoch": 0.74158368895211,
      "grad_norm": 0.5444389461448026,
      "learning_rate": 9.355210494837046e-06,
      "loss": 0.058,
      "step": 782
    },
    {
      "epoch": 0.7425320056899004,
      "grad_norm": 0.8635005280396899,
      "learning_rate": 9.352495754664501e-06,
      "loss": 0.0817,
      "step": 783
    },
    {
      "epoch": 0.7434803224276908,
      "grad_norm": 0.3975227023619501,
      "learning_rate": 9.349775707193439e-06,
      "loss": 0.0325,
      "step": 784
    },
    {
      "epoch": 0.7444286391654813,
      "grad_norm": 0.9671794171858287,
      "learning_rate": 9.347050355740598e-06,
      "loss": 0.0942,
      "step": 785
    },
    {
      "epoch": 0.7453769559032717,
      "grad_norm": 0.8627076848581986,
      "learning_rate": 9.34431970362919e-06,
      "loss": 0.0603,
      "step": 786
    },
    {
      "epoch": 0.7463252726410621,
      "grad_norm": 0.676971569472859,
      "learning_rate": 9.341583754188887e-06,
      "loss": 0.0609,
      "step": 787
    },
    {
      "epoch": 0.7472735893788526,
      "grad_norm": 0.6234019106033082,
      "learning_rate": 9.338842510755822e-06,
      "loss": 0.0527,
      "step": 788
    },
    {
      "epoch": 0.748221906116643,
      "grad_norm": 0.5688808355503273,
      "learning_rate": 9.336095976672578e-06,
      "loss": 0.0746,
      "step": 789
    },
    {
      "epoch": 0.7491702228544334,
      "grad_norm": 0.8927220033190019,
      "learning_rate": 9.3333441552882e-06,
      "loss": 0.0663,
      "step": 790
    },
    {
      "epoch": 0.7501185395922239,
      "grad_norm": 0.6760705893906477,
      "learning_rate": 9.33058704995817e-06,
      "loss": 0.0607,
      "step": 791
    },
    {
      "epoch": 0.7510668563300142,
      "grad_norm": 0.6421619908578323,
      "learning_rate": 9.327824664044418e-06,
      "loss": 0.0601,
      "step": 792
    },
    {
      "epoch": 0.7520151730678046,
      "grad_norm": 0.7064042205046658,
      "learning_rate": 9.32505700091531e-06,
      "loss": 0.0656,
      "step": 793
    },
    {
      "epoch": 0.752963489805595,
      "grad_norm": 0.6789456621715841,
      "learning_rate": 9.322284063945651e-06,
      "loss": 0.0754,
      "step": 794
    },
    {
      "epoch": 0.7539118065433855,
      "grad_norm": 0.6349001762224292,
      "learning_rate": 9.319505856516674e-06,
      "loss": 0.055,
      "step": 795
    },
    {
      "epoch": 0.7548601232811759,
      "grad_norm": 0.7970733715082516,
      "learning_rate": 9.316722382016037e-06,
      "loss": 0.0606,
      "step": 796
    },
    {
      "epoch": 0.7558084400189663,
      "grad_norm": 0.6989262918440643,
      "learning_rate": 9.313933643837825e-06,
      "loss": 0.0419,
      "step": 797
    },
    {
      "epoch": 0.7567567567567568,
      "grad_norm": 0.45444386596780545,
      "learning_rate": 9.311139645382539e-06,
      "loss": 0.0499,
      "step": 798
    },
    {
      "epoch": 0.7577050734945472,
      "grad_norm": 0.7340919059070612,
      "learning_rate": 9.308340390057091e-06,
      "loss": 0.0605,
      "step": 799
    },
    {
      "epoch": 0.7586533902323376,
      "grad_norm": 0.49624254277855845,
      "learning_rate": 9.305535881274812e-06,
      "loss": 0.038,
      "step": 800
    },
    {
      "epoch": 0.7596017069701281,
      "grad_norm": 0.5558026345234958,
      "learning_rate": 9.302726122455425e-06,
      "loss": 0.0477,
      "step": 801
    },
    {
      "epoch": 0.7605500237079185,
      "grad_norm": 0.7377034794768454,
      "learning_rate": 9.299911117025071e-06,
      "loss": 0.0798,
      "step": 802
    },
    {
      "epoch": 0.7614983404457089,
      "grad_norm": 0.6642309640857783,
      "learning_rate": 9.297090868416276e-06,
      "loss": 0.0578,
      "step": 803
    },
    {
      "epoch": 0.7624466571834992,
      "grad_norm": 0.4901567357915514,
      "learning_rate": 9.294265380067965e-06,
      "loss": 0.0546,
      "step": 804
    },
    {
      "epoch": 0.7633949739212897,
      "grad_norm": 0.5504015183910195,
      "learning_rate": 9.291434655425452e-06,
      "loss": 0.0476,
      "step": 805
    },
    {
      "epoch": 0.7643432906590801,
      "grad_norm": 0.7880325712467479,
      "learning_rate": 9.288598697940433e-06,
      "loss": 0.0967,
      "step": 806
    },
    {
      "epoch": 0.7652916073968705,
      "grad_norm": 1.0094413699993006,
      "learning_rate": 9.285757511070987e-06,
      "loss": 0.0547,
      "step": 807
    },
    {
      "epoch": 0.766239924134661,
      "grad_norm": 0.5462824953438216,
      "learning_rate": 9.28291109828157e-06,
      "loss": 0.0622,
      "step": 808
    },
    {
      "epoch": 0.7671882408724514,
      "grad_norm": 0.6095693174069973,
      "learning_rate": 9.28005946304301e-06,
      "loss": 0.054,
      "step": 809
    },
    {
      "epoch": 0.7681365576102418,
      "grad_norm": 0.5522598480936777,
      "learning_rate": 9.277202608832502e-06,
      "loss": 0.0608,
      "step": 810
    },
    {
      "epoch": 0.7690848743480322,
      "grad_norm": 0.8887551561479244,
      "learning_rate": 9.274340539133604e-06,
      "loss": 0.0733,
      "step": 811
    },
    {
      "epoch": 0.7700331910858227,
      "grad_norm": 0.6536519593388536,
      "learning_rate": 9.271473257436239e-06,
      "loss": 0.0704,
      "step": 812
    },
    {
      "epoch": 0.7709815078236131,
      "grad_norm": 0.6903014054311826,
      "learning_rate": 9.268600767236677e-06,
      "loss": 0.0839,
      "step": 813
    },
    {
      "epoch": 0.7719298245614035,
      "grad_norm": 0.5929159416904847,
      "learning_rate": 9.265723072037546e-06,
      "loss": 0.0592,
      "step": 814
    },
    {
      "epoch": 0.772878141299194,
      "grad_norm": 0.7439638317959937,
      "learning_rate": 9.26284017534782e-06,
      "loss": 0.0568,
      "step": 815
    },
    {
      "epoch": 0.7738264580369844,
      "grad_norm": 0.5860050856048022,
      "learning_rate": 9.259952080682812e-06,
      "loss": 0.0667,
      "step": 816
    },
    {
      "epoch": 0.7747747747747747,
      "grad_norm": 0.4842910654706692,
      "learning_rate": 9.257058791564175e-06,
      "loss": 0.0513,
      "step": 817
    },
    {
      "epoch": 0.7757230915125652,
      "grad_norm": 0.789038697553299,
      "learning_rate": 9.254160311519896e-06,
      "loss": 0.0557,
      "step": 818
    },
    {
      "epoch": 0.7766714082503556,
      "grad_norm": 0.5387139258318481,
      "learning_rate": 9.251256644084292e-06,
      "loss": 0.0558,
      "step": 819
    },
    {
      "epoch": 0.777619724988146,
      "grad_norm": 0.8887946106511906,
      "learning_rate": 9.248347792798006e-06,
      "loss": 0.0776,
      "step": 820
    },
    {
      "epoch": 0.7785680417259364,
      "grad_norm": 0.7477907494684204,
      "learning_rate": 9.245433761208e-06,
      "loss": 0.0706,
      "step": 821
    },
    {
      "epoch": 0.7795163584637269,
      "grad_norm": 0.8176178183928178,
      "learning_rate": 9.242514552867556e-06,
      "loss": 0.0806,
      "step": 822
    },
    {
      "epoch": 0.7804646752015173,
      "grad_norm": 0.5104409829727489,
      "learning_rate": 9.239590171336262e-06,
      "loss": 0.0427,
      "step": 823
    },
    {
      "epoch": 0.7814129919393077,
      "grad_norm": 0.5922185838285359,
      "learning_rate": 9.236660620180024e-06,
      "loss": 0.0553,
      "step": 824
    },
    {
      "epoch": 0.7823613086770982,
      "grad_norm": 0.9414341871189567,
      "learning_rate": 9.23372590297104e-06,
      "loss": 0.0678,
      "step": 825
    },
    {
      "epoch": 0.7833096254148886,
      "grad_norm": 0.49939628701466243,
      "learning_rate": 9.230786023287819e-06,
      "loss": 0.0437,
      "step": 826
    },
    {
      "epoch": 0.784257942152679,
      "grad_norm": 0.519425273825053,
      "learning_rate": 9.227840984715154e-06,
      "loss": 0.0497,
      "step": 827
    },
    {
      "epoch": 0.7852062588904695,
      "grad_norm": 0.5443123255099412,
      "learning_rate": 9.224890790844137e-06,
      "loss": 0.0612,
      "step": 828
    },
    {
      "epoch": 0.7861545756282599,
      "grad_norm": 0.511905527310258,
      "learning_rate": 9.221935445272144e-06,
      "loss": 0.0449,
      "step": 829
    },
    {
      "epoch": 0.7871028923660502,
      "grad_norm": 0.6705781452415145,
      "learning_rate": 9.218974951602829e-06,
      "loss": 0.063,
      "step": 830
    },
    {
      "epoch": 0.7880512091038406,
      "grad_norm": 0.47754646141190604,
      "learning_rate": 9.216009313446125e-06,
      "loss": 0.0688,
      "step": 831
    },
    {
      "epoch": 0.7889995258416311,
      "grad_norm": 0.5705276893342319,
      "learning_rate": 9.213038534418244e-06,
      "loss": 0.0686,
      "step": 832
    },
    {
      "epoch": 0.7899478425794215,
      "grad_norm": 0.4253509537520698,
      "learning_rate": 9.21006261814166e-06,
      "loss": 0.0427,
      "step": 833
    },
    {
      "epoch": 0.7908961593172119,
      "grad_norm": 0.533220697742502,
      "learning_rate": 9.207081568245112e-06,
      "loss": 0.0394,
      "step": 834
    },
    {
      "epoch": 0.7918444760550024,
      "grad_norm": 0.5786737951816707,
      "learning_rate": 9.2040953883636e-06,
      "loss": 0.0556,
      "step": 835
    },
    {
      "epoch": 0.7927927927927928,
      "grad_norm": 1.05765776588404,
      "learning_rate": 9.20110408213838e-06,
      "loss": 0.0388,
      "step": 836
    },
    {
      "epoch": 0.7937411095305832,
      "grad_norm": 0.809530041430475,
      "learning_rate": 9.19810765321696e-06,
      "loss": 0.1042,
      "step": 837
    },
    {
      "epoch": 0.7946894262683737,
      "grad_norm": 0.4767483114016521,
      "learning_rate": 9.19510610525309e-06,
      "loss": 0.0586,
      "step": 838
    },
    {
      "epoch": 0.7956377430061641,
      "grad_norm": 0.6212000890855088,
      "learning_rate": 9.192099441906765e-06,
      "loss": 0.063,
      "step": 839
    },
    {
      "epoch": 0.7965860597439545,
      "grad_norm": 0.5793471462839893,
      "learning_rate": 9.189087666844219e-06,
      "loss": 0.0599,
      "step": 840
    },
    {
      "epoch": 0.7975343764817449,
      "grad_norm": 0.6109133021965912,
      "learning_rate": 9.186070783737915e-06,
      "loss": 0.0655,
      "step": 841
    },
    {
      "epoch": 0.7984826932195354,
      "grad_norm": 1.7579309929430755,
      "learning_rate": 9.183048796266547e-06,
      "loss": 0.0531,
      "step": 842
    },
    {
      "epoch": 0.7994310099573257,
      "grad_norm": 0.6305893305402994,
      "learning_rate": 9.180021708115034e-06,
      "loss": 0.069,
      "step": 843
    },
    {
      "epoch": 0.8003793266951161,
      "grad_norm": 0.5799218206040034,
      "learning_rate": 9.176989522974512e-06,
      "loss": 0.0548,
      "step": 844
    },
    {
      "epoch": 0.8013276434329066,
      "grad_norm": 0.5205329821796497,
      "learning_rate": 9.173952244542335e-06,
      "loss": 0.0551,
      "step": 845
    },
    {
      "epoch": 0.802275960170697,
      "grad_norm": 0.6401356176971456,
      "learning_rate": 9.170909876522067e-06,
      "loss": 0.0613,
      "step": 846
    },
    {
      "epoch": 0.8032242769084874,
      "grad_norm": 0.6283553782308525,
      "learning_rate": 9.167862422623474e-06,
      "loss": 0.0681,
      "step": 847
    },
    {
      "epoch": 0.8041725936462779,
      "grad_norm": 0.5291087716357314,
      "learning_rate": 9.164809886562532e-06,
      "loss": 0.0428,
      "step": 848
    },
    {
      "epoch": 0.8051209103840683,
      "grad_norm": 0.6176212098121372,
      "learning_rate": 9.161752272061405e-06,
      "loss": 0.0607,
      "step": 849
    },
    {
      "epoch": 0.8060692271218587,
      "grad_norm": 0.5258734780929885,
      "learning_rate": 9.158689582848454e-06,
      "loss": 0.0555,
      "step": 850
    },
    {
      "epoch": 0.8070175438596491,
      "grad_norm": 0.5473102285657928,
      "learning_rate": 9.155621822658229e-06,
      "loss": 0.0461,
      "step": 851
    },
    {
      "epoch": 0.8079658605974396,
      "grad_norm": 0.7147069989389465,
      "learning_rate": 9.15254899523146e-06,
      "loss": 0.0699,
      "step": 852
    },
    {
      "epoch": 0.80891417733523,
      "grad_norm": 0.5116476113725856,
      "learning_rate": 9.14947110431506e-06,
      "loss": 0.0593,
      "step": 853
    },
    {
      "epoch": 0.8098624940730204,
      "grad_norm": 0.599625799358922,
      "learning_rate": 9.146388153662109e-06,
      "loss": 0.0719,
      "step": 854
    },
    {
      "epoch": 0.8108108108108109,
      "grad_norm": 0.5657265833927722,
      "learning_rate": 9.143300147031866e-06,
      "loss": 0.0539,
      "step": 855
    },
    {
      "epoch": 0.8117591275486012,
      "grad_norm": 0.490285928003467,
      "learning_rate": 9.14020708818975e-06,
      "loss": 0.0551,
      "step": 856
    },
    {
      "epoch": 0.8127074442863916,
      "grad_norm": 0.5667257690255696,
      "learning_rate": 9.137108980907341e-06,
      "loss": 0.0485,
      "step": 857
    },
    {
      "epoch": 0.813655761024182,
      "grad_norm": 0.7288808283591064,
      "learning_rate": 9.134005828962373e-06,
      "loss": 0.0464,
      "step": 858
    },
    {
      "epoch": 0.8146040777619725,
      "grad_norm": 0.6578159612053353,
      "learning_rate": 9.130897636138736e-06,
      "loss": 0.0458,
      "step": 859
    },
    {
      "epoch": 0.8155523944997629,
      "grad_norm": 0.6699312438910595,
      "learning_rate": 9.127784406226462e-06,
      "loss": 0.0484,
      "step": 860
    },
    {
      "epoch": 0.8165007112375533,
      "grad_norm": 0.7577555099867996,
      "learning_rate": 9.124666143021728e-06,
      "loss": 0.0552,
      "step": 861
    },
    {
      "epoch": 0.8174490279753438,
      "grad_norm": 0.6571718231580975,
      "learning_rate": 9.121542850326849e-06,
      "loss": 0.0418,
      "step": 862
    },
    {
      "epoch": 0.8183973447131342,
      "grad_norm": 0.7375729716381728,
      "learning_rate": 9.118414531950268e-06,
      "loss": 0.0586,
      "step": 863
    },
    {
      "epoch": 0.8193456614509246,
      "grad_norm": 0.7778186396499422,
      "learning_rate": 9.115281191706563e-06,
      "loss": 0.0638,
      "step": 864
    },
    {
      "epoch": 0.8202939781887151,
      "grad_norm": 1.1753642296648885,
      "learning_rate": 9.11214283341643e-06,
      "loss": 0.0914,
      "step": 865
    },
    {
      "epoch": 0.8212422949265055,
      "grad_norm": 0.6221136965708439,
      "learning_rate": 9.108999460906687e-06,
      "loss": 0.0513,
      "step": 866
    },
    {
      "epoch": 0.8221906116642959,
      "grad_norm": 1.0126156537474953,
      "learning_rate": 9.105851078010265e-06,
      "loss": 0.0511,
      "step": 867
    },
    {
      "epoch": 0.8231389284020862,
      "grad_norm": 0.609505398312846,
      "learning_rate": 9.102697688566204e-06,
      "loss": 0.0607,
      "step": 868
    },
    {
      "epoch": 0.8240872451398767,
      "grad_norm": 0.6781545775462046,
      "learning_rate": 9.09953929641965e-06,
      "loss": 0.0537,
      "step": 869
    },
    {
      "epoch": 0.8250355618776671,
      "grad_norm": 0.6162914997785193,
      "learning_rate": 9.096375905421849e-06,
      "loss": 0.0514,
      "step": 870
    },
    {
      "epoch": 0.8259838786154575,
      "grad_norm": 0.9380195573648793,
      "learning_rate": 9.093207519430138e-06,
      "loss": 0.0592,
      "step": 871
    },
    {
      "epoch": 0.826932195353248,
      "grad_norm": 0.6891518456384623,
      "learning_rate": 9.090034142307955e-06,
      "loss": 0.0611,
      "step": 872
    },
    {
      "epoch": 0.8278805120910384,
      "grad_norm": 0.6860355795137043,
      "learning_rate": 9.086855777924813e-06,
      "loss": 0.0651,
      "step": 873
    },
    {
      "epoch": 0.8288288288288288,
      "grad_norm": 0.5941193542193252,
      "learning_rate": 9.083672430156313e-06,
      "loss": 0.0561,
      "step": 874
    },
    {
      "epoch": 0.8297771455666193,
      "grad_norm": 0.9859763647912905,
      "learning_rate": 9.080484102884132e-06,
      "loss": 0.0558,
      "step": 875
    },
    {
      "epoch": 0.8307254623044097,
      "grad_norm": 0.6607364577205248,
      "learning_rate": 9.077290799996015e-06,
      "loss": 0.0445,
      "step": 876
    },
    {
      "epoch": 0.8316737790422001,
      "grad_norm": 0.4579344621348973,
      "learning_rate": 9.074092525385777e-06,
      "loss": 0.0532,
      "step": 877
    },
    {
      "epoch": 0.8326220957799905,
      "grad_norm": 0.44020290978074095,
      "learning_rate": 9.070889282953297e-06,
      "loss": 0.0432,
      "step": 878
    },
    {
      "epoch": 0.833570412517781,
      "grad_norm": 0.7817453278171299,
      "learning_rate": 9.067681076604507e-06,
      "loss": 0.0622,
      "step": 879
    },
    {
      "epoch": 0.8345187292555714,
      "grad_norm": 0.6106825636941368,
      "learning_rate": 9.064467910251396e-06,
      "loss": 0.0499,
      "step": 880
    },
    {
      "epoch": 0.8354670459933617,
      "grad_norm": 0.5733918003298187,
      "learning_rate": 9.061249787812e-06,
      "loss": 0.058,
      "step": 881
    },
    {
      "epoch": 0.8364153627311522,
      "grad_norm": 0.734104839469145,
      "learning_rate": 9.058026713210396e-06,
      "loss": 0.0603,
      "step": 882
    },
    {
      "epoch": 0.8373636794689426,
      "grad_norm": 0.5863205921902287,
      "learning_rate": 9.054798690376702e-06,
      "loss": 0.0542,
      "step": 883
    },
    {
      "epoch": 0.838311996206733,
      "grad_norm": 0.6529541400114963,
      "learning_rate": 9.051565723247072e-06,
      "loss": 0.0546,
      "step": 884
    },
    {
      "epoch": 0.8392603129445235,
      "grad_norm": 0.8496840763418192,
      "learning_rate": 9.048327815763682e-06,
      "loss": 0.0499,
      "step": 885
    },
    {
      "epoch": 0.8402086296823139,
      "grad_norm": 0.4879463969986272,
      "learning_rate": 9.045084971874738e-06,
      "loss": 0.0404,
      "step": 886
    },
    {
      "epoch": 0.8411569464201043,
      "grad_norm": 0.48366631890428774,
      "learning_rate": 9.041837195534462e-06,
      "loss": 0.0438,
      "step": 887
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 0.5668428745474414,
      "learning_rate": 9.038584490703095e-06,
      "loss": 0.0577,
      "step": 888
    },
    {
      "epoch": 0.8430535798956852,
      "grad_norm": 0.6139669391301276,
      "learning_rate": 9.03532686134688e-06,
      "loss": 0.0699,
      "step": 889
    },
    {
      "epoch": 0.8440018966334756,
      "grad_norm": 0.6191388281587789,
      "learning_rate": 9.032064311438073e-06,
      "loss": 0.0588,
      "step": 890
    },
    {
      "epoch": 0.844950213371266,
      "grad_norm": 0.6152385003685913,
      "learning_rate": 9.028796844954924e-06,
      "loss": 0.0537,
      "step": 891
    },
    {
      "epoch": 0.8458985301090565,
      "grad_norm": 0.755005932732524,
      "learning_rate": 9.025524465881683e-06,
      "loss": 0.0649,
      "step": 892
    },
    {
      "epoch": 0.8468468468468469,
      "grad_norm": 0.7676513218085563,
      "learning_rate": 9.022247178208585e-06,
      "loss": 0.0635,
      "step": 893
    },
    {
      "epoch": 0.8477951635846372,
      "grad_norm": 0.5920047067355723,
      "learning_rate": 9.018964985931856e-06,
      "loss": 0.06,
      "step": 894
    },
    {
      "epoch": 0.8487434803224277,
      "grad_norm": 0.5807083572715754,
      "learning_rate": 9.015677893053695e-06,
      "loss": 0.0505,
      "step": 895
    },
    {
      "epoch": 0.8496917970602181,
      "grad_norm": 0.7897487160161104,
      "learning_rate": 9.012385903582286e-06,
      "loss": 0.0714,
      "step": 896
    },
    {
      "epoch": 0.8506401137980085,
      "grad_norm": 0.5382652341176712,
      "learning_rate": 9.009089021531777e-06,
      "loss": 0.0512,
      "step": 897
    },
    {
      "epoch": 0.8515884305357989,
      "grad_norm": 0.8441756486986386,
      "learning_rate": 9.005787250922285e-06,
      "loss": 0.0766,
      "step": 898
    },
    {
      "epoch": 0.8525367472735894,
      "grad_norm": 0.42966299233294036,
      "learning_rate": 9.002480595779883e-06,
      "loss": 0.0469,
      "step": 899
    },
    {
      "epoch": 0.8534850640113798,
      "grad_norm": 0.5779848432711783,
      "learning_rate": 8.999169060136609e-06,
      "loss": 0.0549,
      "step": 900
    },
    {
      "epoch": 0.8544333807491702,
      "grad_norm": 0.49828391414464324,
      "learning_rate": 8.995852648030444e-06,
      "loss": 0.0513,
      "step": 901
    },
    {
      "epoch": 0.8553816974869607,
      "grad_norm": 0.5712972033755797,
      "learning_rate": 8.99253136350532e-06,
      "loss": 0.0642,
      "step": 902
    },
    {
      "epoch": 0.8563300142247511,
      "grad_norm": 0.7463859566833713,
      "learning_rate": 8.989205210611106e-06,
      "loss": 0.0669,
      "step": 903
    },
    {
      "epoch": 0.8572783309625415,
      "grad_norm": 0.6015754760898006,
      "learning_rate": 8.98587419340361e-06,
      "loss": 0.0518,
      "step": 904
    },
    {
      "epoch": 0.8582266477003319,
      "grad_norm": 0.7279488477743896,
      "learning_rate": 8.982538315944573e-06,
      "loss": 0.0603,
      "step": 905
    },
    {
      "epoch": 0.8591749644381224,
      "grad_norm": 0.41210687518386613,
      "learning_rate": 8.979197582301662e-06,
      "loss": 0.0508,
      "step": 906
    },
    {
      "epoch": 0.8601232811759127,
      "grad_norm": 1.4900739335277513,
      "learning_rate": 8.97585199654846e-06,
      "loss": 0.072,
      "step": 907
    },
    {
      "epoch": 0.8610715979137031,
      "grad_norm": 0.5450963951689192,
      "learning_rate": 8.972501562764476e-06,
      "loss": 0.0566,
      "step": 908
    },
    {
      "epoch": 0.8620199146514936,
      "grad_norm": 0.5356916948533633,
      "learning_rate": 8.969146285035119e-06,
      "loss": 0.0471,
      "step": 909
    },
    {
      "epoch": 0.862968231389284,
      "grad_norm": 0.6064958608566305,
      "learning_rate": 8.965786167451713e-06,
      "loss": 0.0586,
      "step": 910
    },
    {
      "epoch": 0.8639165481270744,
      "grad_norm": 0.6550030676781202,
      "learning_rate": 8.962421214111486e-06,
      "loss": 0.0622,
      "step": 911
    },
    {
      "epoch": 0.8648648648648649,
      "grad_norm": 0.5789487697080219,
      "learning_rate": 8.959051429117551e-06,
      "loss": 0.0587,
      "step": 912
    },
    {
      "epoch": 0.8658131816026553,
      "grad_norm": 0.6480466907010984,
      "learning_rate": 8.955676816578922e-06,
      "loss": 0.0596,
      "step": 913
    },
    {
      "epoch": 0.8667614983404457,
      "grad_norm": 0.703037972481164,
      "learning_rate": 8.9522973806105e-06,
      "loss": 0.0836,
      "step": 914
    },
    {
      "epoch": 0.8677098150782361,
      "grad_norm": 0.49499510899266297,
      "learning_rate": 8.94891312533306e-06,
      "loss": 0.0493,
      "step": 915
    },
    {
      "epoch": 0.8686581318160266,
      "grad_norm": 0.4679737716122778,
      "learning_rate": 8.945524054873261e-06,
      "loss": 0.0473,
      "step": 916
    },
    {
      "epoch": 0.869606448553817,
      "grad_norm": 0.4868047238192127,
      "learning_rate": 8.942130173363628e-06,
      "loss": 0.0617,
      "step": 917
    },
    {
      "epoch": 0.8705547652916074,
      "grad_norm": 0.48143223119722567,
      "learning_rate": 8.938731484942557e-06,
      "loss": 0.0459,
      "step": 918
    },
    {
      "epoch": 0.8715030820293979,
      "grad_norm": 0.5109365563225756,
      "learning_rate": 8.935327993754307e-06,
      "loss": 0.0603,
      "step": 919
    },
    {
      "epoch": 0.8724513987671882,
      "grad_norm": 0.5946328530954544,
      "learning_rate": 8.931919703948981e-06,
      "loss": 0.0663,
      "step": 920
    },
    {
      "epoch": 0.8733997155049786,
      "grad_norm": 0.6675396299202498,
      "learning_rate": 8.928506619682549e-06,
      "loss": 0.0522,
      "step": 921
    },
    {
      "epoch": 0.8743480322427691,
      "grad_norm": 0.5242785281728278,
      "learning_rate": 8.925088745116817e-06,
      "loss": 0.0477,
      "step": 922
    },
    {
      "epoch": 0.8752963489805595,
      "grad_norm": 0.4607255100157249,
      "learning_rate": 8.921666084419435e-06,
      "loss": 0.0444,
      "step": 923
    },
    {
      "epoch": 0.8762446657183499,
      "grad_norm": 0.6127086410246447,
      "learning_rate": 8.918238641763894e-06,
      "loss": 0.0505,
      "step": 924
    },
    {
      "epoch": 0.8771929824561403,
      "grad_norm": 0.7108664485212953,
      "learning_rate": 8.914806421329505e-06,
      "loss": 0.0372,
      "step": 925
    },
    {
      "epoch": 0.8781412991939308,
      "grad_norm": 0.48171514690034495,
      "learning_rate": 8.911369427301418e-06,
      "loss": 0.0467,
      "step": 926
    },
    {
      "epoch": 0.8790896159317212,
      "grad_norm": 0.5032020795283936,
      "learning_rate": 8.907927663870592e-06,
      "loss": 0.0383,
      "step": 927
    },
    {
      "epoch": 0.8800379326695116,
      "grad_norm": 0.6490864569323296,
      "learning_rate": 8.90448113523381e-06,
      "loss": 0.0703,
      "step": 928
    },
    {
      "epoch": 0.8809862494073021,
      "grad_norm": 0.5274849878368799,
      "learning_rate": 8.901029845593658e-06,
      "loss": 0.0497,
      "step": 929
    },
    {
      "epoch": 0.8819345661450925,
      "grad_norm": 0.7209898569229573,
      "learning_rate": 8.897573799158534e-06,
      "loss": 0.0845,
      "step": 930
    },
    {
      "epoch": 0.8828828828828829,
      "grad_norm": 0.653701403062353,
      "learning_rate": 8.894113000142636e-06,
      "loss": 0.0528,
      "step": 931
    },
    {
      "epoch": 0.8838311996206734,
      "grad_norm": 0.5252034559155617,
      "learning_rate": 8.890647452765954e-06,
      "loss": 0.054,
      "step": 932
    },
    {
      "epoch": 0.8847795163584637,
      "grad_norm": 0.6597062824750437,
      "learning_rate": 8.887177161254267e-06,
      "loss": 0.0508,
      "step": 933
    },
    {
      "epoch": 0.8857278330962541,
      "grad_norm": 0.9841434864966624,
      "learning_rate": 8.883702129839144e-06,
      "loss": 0.06,
      "step": 934
    },
    {
      "epoch": 0.8866761498340445,
      "grad_norm": 0.4716559195813748,
      "learning_rate": 8.880222362757928e-06,
      "loss": 0.0484,
      "step": 935
    },
    {
      "epoch": 0.887624466571835,
      "grad_norm": 0.6275887169553205,
      "learning_rate": 8.87673786425374e-06,
      "loss": 0.055,
      "step": 936
    },
    {
      "epoch": 0.8885727833096254,
      "grad_norm": 0.5480616561224483,
      "learning_rate": 8.87324863857547e-06,
      "loss": 0.0512,
      "step": 937
    },
    {
      "epoch": 0.8895211000474158,
      "grad_norm": 0.5716073816122306,
      "learning_rate": 8.869754689977774e-06,
      "loss": 0.0575,
      "step": 938
    },
    {
      "epoch": 0.8904694167852063,
      "grad_norm": 0.8761043849726794,
      "learning_rate": 8.866256022721062e-06,
      "loss": 0.0508,
      "step": 939
    },
    {
      "epoch": 0.8914177335229967,
      "grad_norm": 0.7017157731117182,
      "learning_rate": 8.862752641071499e-06,
      "loss": 0.0546,
      "step": 940
    },
    {
      "epoch": 0.8923660502607871,
      "grad_norm": 1.5138916151321196,
      "learning_rate": 8.859244549301005e-06,
      "loss": 0.0658,
      "step": 941
    },
    {
      "epoch": 0.8933143669985776,
      "grad_norm": 0.8433261605133346,
      "learning_rate": 8.855731751687233e-06,
      "loss": 0.0553,
      "step": 942
    },
    {
      "epoch": 0.894262683736368,
      "grad_norm": 0.5494966721887847,
      "learning_rate": 8.852214252513582e-06,
      "loss": 0.0494,
      "step": 943
    },
    {
      "epoch": 0.8952110004741584,
      "grad_norm": 0.6006177701179363,
      "learning_rate": 8.848692056069184e-06,
      "loss": 0.0612,
      "step": 944
    },
    {
      "epoch": 0.8961593172119487,
      "grad_norm": 0.6876171031491582,
      "learning_rate": 8.84516516664889e-06,
      "loss": 0.0609,
      "step": 945
    },
    {
      "epoch": 0.8971076339497392,
      "grad_norm": 0.846588378426009,
      "learning_rate": 8.841633588553287e-06,
      "loss": 0.0593,
      "step": 946
    },
    {
      "epoch": 0.8980559506875296,
      "grad_norm": 1.175631640532978,
      "learning_rate": 8.838097326088667e-06,
      "loss": 0.0767,
      "step": 947
    },
    {
      "epoch": 0.89900426742532,
      "grad_norm": 0.7010270158444133,
      "learning_rate": 8.834556383567042e-06,
      "loss": 0.0637,
      "step": 948
    },
    {
      "epoch": 0.8999525841631105,
      "grad_norm": 0.7103962193756044,
      "learning_rate": 8.831010765306124e-06,
      "loss": 0.047,
      "step": 949
    },
    {
      "epoch": 0.9009009009009009,
      "grad_norm": 0.9919713077792982,
      "learning_rate": 8.827460475629334e-06,
      "loss": 0.0699,
      "step": 950
    },
    {
      "epoch": 0.9018492176386913,
      "grad_norm": 0.9438936607800321,
      "learning_rate": 8.823905518865782e-06,
      "loss": 0.0962,
      "step": 951
    },
    {
      "epoch": 0.9027975343764817,
      "grad_norm": 0.41357107371942303,
      "learning_rate": 8.820345899350275e-06,
      "loss": 0.0393,
      "step": 952
    },
    {
      "epoch": 0.9037458511142722,
      "grad_norm": 0.6094306471098007,
      "learning_rate": 8.8167816214233e-06,
      "loss": 0.0547,
      "step": 953
    },
    {
      "epoch": 0.9046941678520626,
      "grad_norm": 0.45434395748515616,
      "learning_rate": 8.81321268943103e-06,
      "loss": 0.0458,
      "step": 954
    },
    {
      "epoch": 0.905642484589853,
      "grad_norm": 0.584662000585842,
      "learning_rate": 8.809639107725308e-06,
      "loss": 0.0684,
      "step": 955
    },
    {
      "epoch": 0.9065908013276435,
      "grad_norm": 0.6281479664499341,
      "learning_rate": 8.80606088066365e-06,
      "loss": 0.0485,
      "step": 956
    },
    {
      "epoch": 0.9075391180654339,
      "grad_norm": 0.5220137398785665,
      "learning_rate": 8.802478012609235e-06,
      "loss": 0.0478,
      "step": 957
    },
    {
      "epoch": 0.9084874348032242,
      "grad_norm": 0.7613507347001472,
      "learning_rate": 8.798890507930899e-06,
      "loss": 0.0534,
      "step": 958
    },
    {
      "epoch": 0.9094357515410147,
      "grad_norm": 0.5338153539509801,
      "learning_rate": 8.795298371003138e-06,
      "loss": 0.0467,
      "step": 959
    },
    {
      "epoch": 0.9103840682788051,
      "grad_norm": 0.508435320780577,
      "learning_rate": 8.791701606206092e-06,
      "loss": 0.05,
      "step": 960
    },
    {
      "epoch": 0.9113323850165955,
      "grad_norm": 0.6801979027503147,
      "learning_rate": 8.788100217925541e-06,
      "loss": 0.0654,
      "step": 961
    },
    {
      "epoch": 0.9122807017543859,
      "grad_norm": 0.5472159955708181,
      "learning_rate": 8.78449421055291e-06,
      "loss": 0.0566,
      "step": 962
    },
    {
      "epoch": 0.9132290184921764,
      "grad_norm": 0.5546852372370231,
      "learning_rate": 8.78088358848525e-06,
      "loss": 0.0544,
      "step": 963
    },
    {
      "epoch": 0.9141773352299668,
      "grad_norm": 0.7376086419870055,
      "learning_rate": 8.777268356125244e-06,
      "loss": 0.0618,
      "step": 964
    },
    {
      "epoch": 0.9151256519677572,
      "grad_norm": 0.461174714622349,
      "learning_rate": 8.773648517881194e-06,
      "loss": 0.0527,
      "step": 965
    },
    {
      "epoch": 0.9160739687055477,
      "grad_norm": 1.100649311314461,
      "learning_rate": 8.770024078167017e-06,
      "loss": 0.075,
      "step": 966
    },
    {
      "epoch": 0.9170222854433381,
      "grad_norm": 0.5385193734337945,
      "learning_rate": 8.766395041402245e-06,
      "loss": 0.056,
      "step": 967
    },
    {
      "epoch": 0.9179706021811285,
      "grad_norm": 0.4215583451342763,
      "learning_rate": 8.762761412012011e-06,
      "loss": 0.045,
      "step": 968
    },
    {
      "epoch": 0.918918918918919,
      "grad_norm": 0.5690890175604749,
      "learning_rate": 8.75912319442705e-06,
      "loss": 0.0568,
      "step": 969
    },
    {
      "epoch": 0.9198672356567094,
      "grad_norm": 0.5598668678593514,
      "learning_rate": 8.755480393083694e-06,
      "loss": 0.0629,
      "step": 970
    },
    {
      "epoch": 0.9208155523944997,
      "grad_norm": 0.4230299561301444,
      "learning_rate": 8.751833012423861e-06,
      "loss": 0.0402,
      "step": 971
    },
    {
      "epoch": 0.9217638691322901,
      "grad_norm": 0.8504416588391118,
      "learning_rate": 8.74818105689505e-06,
      "loss": 0.0521,
      "step": 972
    },
    {
      "epoch": 0.9227121858700806,
      "grad_norm": 0.461086821346764,
      "learning_rate": 8.744524530950351e-06,
      "loss": 0.0426,
      "step": 973
    },
    {
      "epoch": 0.923660502607871,
      "grad_norm": 0.5086789755859074,
      "learning_rate": 8.740863439048412e-06,
      "loss": 0.0487,
      "step": 974
    },
    {
      "epoch": 0.9246088193456614,
      "grad_norm": 0.45915883182777006,
      "learning_rate": 8.737197785653457e-06,
      "loss": 0.0444,
      "step": 975
    },
    {
      "epoch": 0.9255571360834519,
      "grad_norm": 0.6701095989032753,
      "learning_rate": 8.73352757523527e-06,
      "loss": 0.0707,
      "step": 976
    },
    {
      "epoch": 0.9265054528212423,
      "grad_norm": 0.460793794881083,
      "learning_rate": 8.729852812269192e-06,
      "loss": 0.0462,
      "step": 977
    },
    {
      "epoch": 0.9274537695590327,
      "grad_norm": 0.42146552351647865,
      "learning_rate": 8.726173501236115e-06,
      "loss": 0.0413,
      "step": 978
    },
    {
      "epoch": 0.9284020862968232,
      "grad_norm": 0.4515670497285217,
      "learning_rate": 8.722489646622477e-06,
      "loss": 0.0486,
      "step": 979
    },
    {
      "epoch": 0.9293504030346136,
      "grad_norm": 0.7298661971153528,
      "learning_rate": 8.718801252920257e-06,
      "loss": 0.0728,
      "step": 980
    },
    {
      "epoch": 0.930298719772404,
      "grad_norm": 0.6123325398467794,
      "learning_rate": 8.715108324626967e-06,
      "loss": 0.0528,
      "step": 981
    },
    {
      "epoch": 0.9312470365101944,
      "grad_norm": 0.5334963078534037,
      "learning_rate": 8.711410866245648e-06,
      "loss": 0.0409,
      "step": 982
    },
    {
      "epoch": 0.9321953532479849,
      "grad_norm": 0.44851971952458897,
      "learning_rate": 8.70770888228487e-06,
      "loss": 0.0509,
      "step": 983
    },
    {
      "epoch": 0.9331436699857752,
      "grad_norm": 0.9770313333004932,
      "learning_rate": 8.704002377258714e-06,
      "loss": 0.0463,
      "step": 984
    },
    {
      "epoch": 0.9340919867235656,
      "grad_norm": 0.7370636377202378,
      "learning_rate": 8.700291355686779e-06,
      "loss": 0.0637,
      "step": 985
    },
    {
      "epoch": 0.9350403034613561,
      "grad_norm": 0.6070776528057518,
      "learning_rate": 8.69657582209417e-06,
      "loss": 0.0488,
      "step": 986
    },
    {
      "epoch": 0.9359886201991465,
      "grad_norm": 0.7278417266877663,
      "learning_rate": 8.692855781011494e-06,
      "loss": 0.0501,
      "step": 987
    },
    {
      "epoch": 0.9369369369369369,
      "grad_norm": 0.4731052806759658,
      "learning_rate": 8.689131236974853e-06,
      "loss": 0.0417,
      "step": 988
    },
    {
      "epoch": 0.9378852536747273,
      "grad_norm": 0.45598792555472306,
      "learning_rate": 8.68540219452584e-06,
      "loss": 0.0396,
      "step": 989
    },
    {
      "epoch": 0.9388335704125178,
      "grad_norm": 0.5661429908370399,
      "learning_rate": 8.681668658211535e-06,
      "loss": 0.0577,
      "step": 990
    },
    {
      "epoch": 0.9397818871503082,
      "grad_norm": 0.41955875165931145,
      "learning_rate": 8.677930632584496e-06,
      "loss": 0.0432,
      "step": 991
    },
    {
      "epoch": 0.9407302038880986,
      "grad_norm": 0.4107826749470781,
      "learning_rate": 8.674188122202756e-06,
      "loss": 0.0535,
      "step": 992
    },
    {
      "epoch": 0.9416785206258891,
      "grad_norm": 0.47653411892607034,
      "learning_rate": 8.670441131629816e-06,
      "loss": 0.0586,
      "step": 993
    },
    {
      "epoch": 0.9426268373636795,
      "grad_norm": 0.53171021829938,
      "learning_rate": 8.66668966543464e-06,
      "loss": 0.0518,
      "step": 994
    },
    {
      "epoch": 0.9435751541014699,
      "grad_norm": 0.43148473645836083,
      "learning_rate": 8.662933728191651e-06,
      "loss": 0.0431,
      "step": 995
    },
    {
      "epoch": 0.9445234708392604,
      "grad_norm": 0.4471351558402442,
      "learning_rate": 8.659173324480722e-06,
      "loss": 0.0438,
      "step": 996
    },
    {
      "epoch": 0.9454717875770507,
      "grad_norm": 0.5782265716940447,
      "learning_rate": 8.65540845888717e-06,
      "loss": 0.0719,
      "step": 997
    },
    {
      "epoch": 0.9464201043148411,
      "grad_norm": 0.4141433604011682,
      "learning_rate": 8.651639136001762e-06,
      "loss": 0.0469,
      "step": 998
    },
    {
      "epoch": 0.9473684210526315,
      "grad_norm": 0.6061219180547935,
      "learning_rate": 8.647865360420686e-06,
      "loss": 0.0489,
      "step": 999
    },
    {
      "epoch": 0.948316737790422,
      "grad_norm": 0.48916698447496854,
      "learning_rate": 8.644087136745572e-06,
      "loss": 0.0513,
      "step": 1000
    },
    {
      "epoch": 0.9492650545282124,
      "grad_norm": 0.37441669864478105,
      "learning_rate": 8.640304469583469e-06,
      "loss": 0.0412,
      "step": 1001
    },
    {
      "epoch": 0.9502133712660028,
      "grad_norm": 1.0623152293680482,
      "learning_rate": 8.636517363546838e-06,
      "loss": 0.0655,
      "step": 1002
    },
    {
      "epoch": 0.9511616880037933,
      "grad_norm": 0.7061581986197312,
      "learning_rate": 8.63272582325357e-06,
      "loss": 0.0499,
      "step": 1003
    },
    {
      "epoch": 0.9521100047415837,
      "grad_norm": 0.5399127227606683,
      "learning_rate": 8.62892985332694e-06,
      "loss": 0.0518,
      "step": 1004
    },
    {
      "epoch": 0.9530583214793741,
      "grad_norm": 0.4559892605058489,
      "learning_rate": 8.625129458395643e-06,
      "loss": 0.0459,
      "step": 1005
    },
    {
      "epoch": 0.9540066382171646,
      "grad_norm": 0.485355373272851,
      "learning_rate": 8.621324643093762e-06,
      "loss": 0.0454,
      "step": 1006
    },
    {
      "epoch": 0.954954954954955,
      "grad_norm": 0.7459047370537332,
      "learning_rate": 8.617515412060771e-06,
      "loss": 0.06,
      "step": 1007
    },
    {
      "epoch": 0.9559032716927454,
      "grad_norm": 0.7374476556281685,
      "learning_rate": 8.613701769941526e-06,
      "loss": 0.0677,
      "step": 1008
    },
    {
      "epoch": 0.9568515884305357,
      "grad_norm": 0.5640575902917073,
      "learning_rate": 8.609883721386266e-06,
      "loss": 0.0464,
      "step": 1009
    },
    {
      "epoch": 0.9577999051683262,
      "grad_norm": 0.5329518829334081,
      "learning_rate": 8.606061271050601e-06,
      "loss": 0.0422,
      "step": 1010
    },
    {
      "epoch": 0.9587482219061166,
      "grad_norm": 0.5672285885118362,
      "learning_rate": 8.602234423595509e-06,
      "loss": 0.0432,
      "step": 1011
    },
    {
      "epoch": 0.959696538643907,
      "grad_norm": 0.49279890911522445,
      "learning_rate": 8.598403183687328e-06,
      "loss": 0.0411,
      "step": 1012
    },
    {
      "epoch": 0.9606448553816975,
      "grad_norm": 0.5195118583178678,
      "learning_rate": 8.594567555997755e-06,
      "loss": 0.0575,
      "step": 1013
    },
    {
      "epoch": 0.9615931721194879,
      "grad_norm": 1.372925234445775,
      "learning_rate": 8.590727545203833e-06,
      "loss": 0.0615,
      "step": 1014
    },
    {
      "epoch": 0.9625414888572783,
      "grad_norm": 0.7147315054833345,
      "learning_rate": 8.586883155987955e-06,
      "loss": 0.0712,
      "step": 1015
    },
    {
      "epoch": 0.9634898055950688,
      "grad_norm": 0.5802509696174448,
      "learning_rate": 8.583034393037848e-06,
      "loss": 0.0552,
      "step": 1016
    },
    {
      "epoch": 0.9644381223328592,
      "grad_norm": 0.49007583048635933,
      "learning_rate": 8.579181261046576e-06,
      "loss": 0.0449,
      "step": 1017
    },
    {
      "epoch": 0.9653864390706496,
      "grad_norm": 0.48751614831454176,
      "learning_rate": 8.57532376471253e-06,
      "loss": 0.0475,
      "step": 1018
    },
    {
      "epoch": 0.96633475580844,
      "grad_norm": 0.6496160692100631,
      "learning_rate": 8.571461908739415e-06,
      "loss": 0.0523,
      "step": 1019
    },
    {
      "epoch": 0.9672830725462305,
      "grad_norm": 0.481345745516473,
      "learning_rate": 8.567595697836266e-06,
      "loss": 0.0515,
      "step": 1020
    },
    {
      "epoch": 0.9682313892840209,
      "grad_norm": 0.5247818144993567,
      "learning_rate": 8.563725136717419e-06,
      "loss": 0.0494,
      "step": 1021
    },
    {
      "epoch": 0.9691797060218112,
      "grad_norm": 0.8474516614825078,
      "learning_rate": 8.559850230102513e-06,
      "loss": 0.0578,
      "step": 1022
    },
    {
      "epoch": 0.9701280227596017,
      "grad_norm": 0.7494686751693889,
      "learning_rate": 8.555970982716492e-06,
      "loss": 0.0613,
      "step": 1023
    },
    {
      "epoch": 0.9710763394973921,
      "grad_norm": 0.528161959351856,
      "learning_rate": 8.55208739928959e-06,
      "loss": 0.0446,
      "step": 1024
    },
    {
      "epoch": 0.9720246562351825,
      "grad_norm": 0.7556057248494816,
      "learning_rate": 8.54819948455733e-06,
      "loss": 0.0611,
      "step": 1025
    },
    {
      "epoch": 0.972972972972973,
      "grad_norm": 0.4857201457975449,
      "learning_rate": 8.54430724326051e-06,
      "loss": 0.0396,
      "step": 1026
    },
    {
      "epoch": 0.9739212897107634,
      "grad_norm": 0.4633933638270801,
      "learning_rate": 8.540410680145213e-06,
      "loss": 0.045,
      "step": 1027
    },
    {
      "epoch": 0.9748696064485538,
      "grad_norm": 0.5215732727679809,
      "learning_rate": 8.536509799962784e-06,
      "loss": 0.047,
      "step": 1028
    },
    {
      "epoch": 0.9758179231863442,
      "grad_norm": 1.5449712519877792,
      "learning_rate": 8.532604607469839e-06,
      "loss": 0.0717,
      "step": 1029
    },
    {
      "epoch": 0.9767662399241347,
      "grad_norm": 0.46693259860172376,
      "learning_rate": 8.528695107428247e-06,
      "loss": 0.0458,
      "step": 1030
    },
    {
      "epoch": 0.9777145566619251,
      "grad_norm": 0.5388054089062692,
      "learning_rate": 8.52478130460513e-06,
      "loss": 0.047,
      "step": 1031
    },
    {
      "epoch": 0.9786628733997155,
      "grad_norm": 0.5283181708144433,
      "learning_rate": 8.520863203772858e-06,
      "loss": 0.0496,
      "step": 1032
    },
    {
      "epoch": 0.979611190137506,
      "grad_norm": 0.5890035811704775,
      "learning_rate": 8.516940809709044e-06,
      "loss": 0.0437,
      "step": 1033
    },
    {
      "epoch": 0.9805595068752964,
      "grad_norm": 0.446739345865473,
      "learning_rate": 8.513014127196533e-06,
      "loss": 0.042,
      "step": 1034
    },
    {
      "epoch": 0.9815078236130867,
      "grad_norm": 0.49851759898580866,
      "learning_rate": 8.509083161023399e-06,
      "loss": 0.0553,
      "step": 1035
    },
    {
      "epoch": 0.9824561403508771,
      "grad_norm": 0.8986990099986447,
      "learning_rate": 8.505147915982943e-06,
      "loss": 0.0491,
      "step": 1036
    },
    {
      "epoch": 0.9834044570886676,
      "grad_norm": 0.4813313700157437,
      "learning_rate": 8.501208396873677e-06,
      "loss": 0.0524,
      "step": 1037
    },
    {
      "epoch": 0.984352773826458,
      "grad_norm": 0.7823009578163489,
      "learning_rate": 8.497264608499332e-06,
      "loss": 0.0542,
      "step": 1038
    },
    {
      "epoch": 0.9853010905642484,
      "grad_norm": 0.5256393060960738,
      "learning_rate": 8.49331655566884e-06,
      "loss": 0.0545,
      "step": 1039
    },
    {
      "epoch": 0.9862494073020389,
      "grad_norm": 0.5400471979930811,
      "learning_rate": 8.489364243196334e-06,
      "loss": 0.0495,
      "step": 1040
    },
    {
      "epoch": 0.9871977240398293,
      "grad_norm": 0.5862041954662611,
      "learning_rate": 8.485407675901142e-06,
      "loss": 0.0442,
      "step": 1041
    },
    {
      "epoch": 0.9881460407776197,
      "grad_norm": 0.6834922008296388,
      "learning_rate": 8.48144685860778e-06,
      "loss": 0.064,
      "step": 1042
    },
    {
      "epoch": 0.9890943575154102,
      "grad_norm": 0.8002369541010694,
      "learning_rate": 8.477481796145945e-06,
      "loss": 0.0464,
      "step": 1043
    },
    {
      "epoch": 0.9900426742532006,
      "grad_norm": 0.47393154077930216,
      "learning_rate": 8.47351249335051e-06,
      "loss": 0.0485,
      "step": 1044
    },
    {
      "epoch": 0.990990990990991,
      "grad_norm": 0.4987272807246751,
      "learning_rate": 8.469538955061525e-06,
      "loss": 0.0478,
      "step": 1045
    },
    {
      "epoch": 0.9919393077287814,
      "grad_norm": 0.6406968710094035,
      "learning_rate": 8.465561186124193e-06,
      "loss": 0.0494,
      "step": 1046
    },
    {
      "epoch": 0.9928876244665719,
      "grad_norm": 0.5319476049591959,
      "learning_rate": 8.46157919138889e-06,
      "loss": 0.038,
      "step": 1047
    },
    {
      "epoch": 0.9938359412043622,
      "grad_norm": 0.5377926003236448,
      "learning_rate": 8.457592975711128e-06,
      "loss": 0.0415,
      "step": 1048
    },
    {
      "epoch": 0.9947842579421526,
      "grad_norm": 0.5054973123174826,
      "learning_rate": 8.45360254395158e-06,
      "loss": 0.0509,
      "step": 1049
    },
    {
      "epoch": 0.9957325746799431,
      "grad_norm": 0.6511826899131821,
      "learning_rate": 8.449607900976056e-06,
      "loss": 0.0496,
      "step": 1050
    },
    {
      "epoch": 0.9966808914177335,
      "grad_norm": 0.34335574918053036,
      "learning_rate": 8.445609051655497e-06,
      "loss": 0.0322,
      "step": 1051
    },
    {
      "epoch": 0.9976292081555239,
      "grad_norm": 0.5324023086103392,
      "learning_rate": 8.441606000865978e-06,
      "loss": 0.0465,
      "step": 1052
    },
    {
      "epoch": 0.9985775248933144,
      "grad_norm": 0.3971741987281817,
      "learning_rate": 8.437598753488693e-06,
      "loss": 0.0316,
      "step": 1053
    },
    {
      "epoch": 0.9995258416311048,
      "grad_norm": 0.4702644191912913,
      "learning_rate": 8.43358731440996e-06,
      "loss": 0.0424,
      "step": 1054
    },
    {
      "epoch": 0.9995258416311048,
      "eval_loss": 0.05579984560608864,
      "eval_runtime": 205.6016,
      "eval_samples_per_second": 34.547,
      "eval_steps_per_second": 1.08,
      "step": 1054
    },
    {
      "epoch": 1.0004741583688952,
      "grad_norm": 0.4866459767505418,
      "learning_rate": 8.429571688521196e-06,
      "loss": 0.0508,
      "step": 1055
    },
    {
      "epoch": 1.0014224751066856,
      "grad_norm": 0.43980501175306935,
      "learning_rate": 8.425551880718938e-06,
      "loss": 0.0348,
      "step": 1056
    },
    {
      "epoch": 1.002370791844476,
      "grad_norm": 0.5671234502453784,
      "learning_rate": 8.421527895904811e-06,
      "loss": 0.0407,
      "step": 1057
    },
    {
      "epoch": 1.0033191085822666,
      "grad_norm": 0.47452207722793455,
      "learning_rate": 8.417499738985539e-06,
      "loss": 0.0481,
      "step": 1058
    },
    {
      "epoch": 1.004267425320057,
      "grad_norm": 0.5013037212797574,
      "learning_rate": 8.413467414872934e-06,
      "loss": 0.0433,
      "step": 1059
    },
    {
      "epoch": 1.0052157420578474,
      "grad_norm": 0.6889982998548566,
      "learning_rate": 8.409430928483881e-06,
      "loss": 0.0465,
      "step": 1060
    },
    {
      "epoch": 1.0061640587956378,
      "grad_norm": 0.6860518531915715,
      "learning_rate": 8.40539028474035e-06,
      "loss": 0.0463,
      "step": 1061
    },
    {
      "epoch": 1.0071123755334281,
      "grad_norm": 0.883560729451268,
      "learning_rate": 8.40134548856938e-06,
      "loss": 0.0503,
      "step": 1062
    },
    {
      "epoch": 1.0080606922712185,
      "grad_norm": 0.535065661745393,
      "learning_rate": 8.397296544903067e-06,
      "loss": 0.0516,
      "step": 1063
    },
    {
      "epoch": 1.009009009009009,
      "grad_norm": 0.5038131455825761,
      "learning_rate": 8.393243458678565e-06,
      "loss": 0.0446,
      "step": 1064
    },
    {
      "epoch": 1.0099573257467995,
      "grad_norm": 0.5572240876317798,
      "learning_rate": 8.389186234838086e-06,
      "loss": 0.0501,
      "step": 1065
    },
    {
      "epoch": 1.01090564248459,
      "grad_norm": 0.5298425351016989,
      "learning_rate": 8.385124878328881e-06,
      "loss": 0.049,
      "step": 1066
    },
    {
      "epoch": 1.0118539592223803,
      "grad_norm": 0.6691906659193966,
      "learning_rate": 8.381059394103244e-06,
      "loss": 0.054,
      "step": 1067
    },
    {
      "epoch": 1.0128022759601707,
      "grad_norm": 0.43516390524601356,
      "learning_rate": 8.376989787118498e-06,
      "loss": 0.0409,
      "step": 1068
    },
    {
      "epoch": 1.013750592697961,
      "grad_norm": 0.5385522154452346,
      "learning_rate": 8.372916062336998e-06,
      "loss": 0.0424,
      "step": 1069
    },
    {
      "epoch": 1.0146989094357515,
      "grad_norm": 0.48362376282426833,
      "learning_rate": 8.368838224726117e-06,
      "loss": 0.0555,
      "step": 1070
    },
    {
      "epoch": 1.0156472261735419,
      "grad_norm": 0.41768940684559036,
      "learning_rate": 8.364756279258245e-06,
      "loss": 0.0379,
      "step": 1071
    },
    {
      "epoch": 1.0165955429113325,
      "grad_norm": 0.5803726520999999,
      "learning_rate": 8.360670230910777e-06,
      "loss": 0.0434,
      "step": 1072
    },
    {
      "epoch": 1.0175438596491229,
      "grad_norm": 0.4357479407163672,
      "learning_rate": 8.356580084666114e-06,
      "loss": 0.0411,
      "step": 1073
    },
    {
      "epoch": 1.0184921763869133,
      "grad_norm": 0.5709121778249012,
      "learning_rate": 8.352485845511658e-06,
      "loss": 0.0409,
      "step": 1074
    },
    {
      "epoch": 1.0194404931247036,
      "grad_norm": 0.5523633969751309,
      "learning_rate": 8.348387518439794e-06,
      "loss": 0.0489,
      "step": 1075
    },
    {
      "epoch": 1.020388809862494,
      "grad_norm": 0.6742232849407267,
      "learning_rate": 8.344285108447896e-06,
      "loss": 0.0493,
      "step": 1076
    },
    {
      "epoch": 1.0213371266002844,
      "grad_norm": 0.3560104483717237,
      "learning_rate": 8.340178620538316e-06,
      "loss": 0.0367,
      "step": 1077
    },
    {
      "epoch": 1.0222854433380748,
      "grad_norm": 0.5324136816558409,
      "learning_rate": 8.336068059718375e-06,
      "loss": 0.0468,
      "step": 1078
    },
    {
      "epoch": 1.0232337600758654,
      "grad_norm": 0.506743337131701,
      "learning_rate": 8.331953431000368e-06,
      "loss": 0.0712,
      "step": 1079
    },
    {
      "epoch": 1.0241820768136558,
      "grad_norm": 0.415891887112626,
      "learning_rate": 8.327834739401543e-06,
      "loss": 0.0431,
      "step": 1080
    },
    {
      "epoch": 1.0251303935514462,
      "grad_norm": 0.62544359839528,
      "learning_rate": 8.323711989944107e-06,
      "loss": 0.0418,
      "step": 1081
    },
    {
      "epoch": 1.0260787102892366,
      "grad_norm": 0.5632716938219344,
      "learning_rate": 8.319585187655211e-06,
      "loss": 0.0547,
      "step": 1082
    },
    {
      "epoch": 1.027027027027027,
      "grad_norm": 0.4086615551677634,
      "learning_rate": 8.315454337566952e-06,
      "loss": 0.0427,
      "step": 1083
    },
    {
      "epoch": 1.0279753437648174,
      "grad_norm": 0.5161406978477902,
      "learning_rate": 8.311319444716358e-06,
      "loss": 0.06,
      "step": 1084
    },
    {
      "epoch": 1.028923660502608,
      "grad_norm": 0.604906094268659,
      "learning_rate": 8.307180514145392e-06,
      "loss": 0.052,
      "step": 1085
    },
    {
      "epoch": 1.0298719772403984,
      "grad_norm": 0.5968839553881752,
      "learning_rate": 8.303037550900935e-06,
      "loss": 0.0529,
      "step": 1086
    },
    {
      "epoch": 1.0308202939781888,
      "grad_norm": 0.5598009215343084,
      "learning_rate": 8.298890560034792e-06,
      "loss": 0.0397,
      "step": 1087
    },
    {
      "epoch": 1.0317686107159791,
      "grad_norm": 0.6572448390948811,
      "learning_rate": 8.29473954660367e-06,
      "loss": 0.0678,
      "step": 1088
    },
    {
      "epoch": 1.0327169274537695,
      "grad_norm": 0.6576638795258621,
      "learning_rate": 8.29058451566919e-06,
      "loss": 0.0598,
      "step": 1089
    },
    {
      "epoch": 1.03366524419156,
      "grad_norm": 0.7400846520221458,
      "learning_rate": 8.286425472297868e-06,
      "loss": 0.0415,
      "step": 1090
    },
    {
      "epoch": 1.0346135609293503,
      "grad_norm": 0.5570271824791166,
      "learning_rate": 8.282262421561111e-06,
      "loss": 0.0398,
      "step": 1091
    },
    {
      "epoch": 1.035561877667141,
      "grad_norm": 0.46875414138844956,
      "learning_rate": 8.278095368535215e-06,
      "loss": 0.0494,
      "step": 1092
    },
    {
      "epoch": 1.0365101944049313,
      "grad_norm": 0.5969883635346438,
      "learning_rate": 8.273924318301354e-06,
      "loss": 0.0747,
      "step": 1093
    },
    {
      "epoch": 1.0374585111427217,
      "grad_norm": 0.4752850264908405,
      "learning_rate": 8.26974927594558e-06,
      "loss": 0.0425,
      "step": 1094
    },
    {
      "epoch": 1.038406827880512,
      "grad_norm": 0.44930878997000867,
      "learning_rate": 8.26557024655881e-06,
      "loss": 0.0433,
      "step": 1095
    },
    {
      "epoch": 1.0393551446183025,
      "grad_norm": 0.6013056895336477,
      "learning_rate": 8.261387235236821e-06,
      "loss": 0.0329,
      "step": 1096
    },
    {
      "epoch": 1.0403034613560929,
      "grad_norm": 0.6657503155051709,
      "learning_rate": 8.257200247080249e-06,
      "loss": 0.0601,
      "step": 1097
    },
    {
      "epoch": 1.0412517780938833,
      "grad_norm": 0.6832978810435474,
      "learning_rate": 8.253009287194576e-06,
      "loss": 0.0466,
      "step": 1098
    },
    {
      "epoch": 1.0422000948316739,
      "grad_norm": 0.6455854633128209,
      "learning_rate": 8.248814360690129e-06,
      "loss": 0.0399,
      "step": 1099
    },
    {
      "epoch": 1.0431484115694643,
      "grad_norm": 0.4721829163859669,
      "learning_rate": 8.244615472682074e-06,
      "loss": 0.0442,
      "step": 1100
    },
    {
      "epoch": 1.0440967283072546,
      "grad_norm": 0.6488059712530941,
      "learning_rate": 8.240412628290405e-06,
      "loss": 0.0553,
      "step": 1101
    },
    {
      "epoch": 1.045045045045045,
      "grad_norm": 0.4881875750545383,
      "learning_rate": 8.236205832639936e-06,
      "loss": 0.0394,
      "step": 1102
    },
    {
      "epoch": 1.0459933617828354,
      "grad_norm": 0.5543410049132131,
      "learning_rate": 8.231995090860306e-06,
      "loss": 0.0496,
      "step": 1103
    },
    {
      "epoch": 1.0469416785206258,
      "grad_norm": 0.4693748617137878,
      "learning_rate": 8.227780408085964e-06,
      "loss": 0.0455,
      "step": 1104
    },
    {
      "epoch": 1.0478899952584164,
      "grad_norm": 0.43672702099083865,
      "learning_rate": 8.22356178945616e-06,
      "loss": 0.033,
      "step": 1105
    },
    {
      "epoch": 1.0488383119962068,
      "grad_norm": 0.53698914158417,
      "learning_rate": 8.219339240114955e-06,
      "loss": 0.0428,
      "step": 1106
    },
    {
      "epoch": 1.0497866287339972,
      "grad_norm": 0.4824835633653137,
      "learning_rate": 8.215112765211186e-06,
      "loss": 0.0454,
      "step": 1107
    },
    {
      "epoch": 1.0507349454717876,
      "grad_norm": 0.3929806462421986,
      "learning_rate": 8.210882369898487e-06,
      "loss": 0.0331,
      "step": 1108
    },
    {
      "epoch": 1.051683262209578,
      "grad_norm": 0.4717562604419889,
      "learning_rate": 8.206648059335276e-06,
      "loss": 0.0371,
      "step": 1109
    },
    {
      "epoch": 1.0526315789473684,
      "grad_norm": 0.48523700167872613,
      "learning_rate": 8.202409838684737e-06,
      "loss": 0.044,
      "step": 1110
    },
    {
      "epoch": 1.0535798956851588,
      "grad_norm": 0.5711347825035666,
      "learning_rate": 8.198167713114824e-06,
      "loss": 0.0482,
      "step": 1111
    },
    {
      "epoch": 1.0545282124229494,
      "grad_norm": 0.6497519216289137,
      "learning_rate": 8.193921687798257e-06,
      "loss": 0.0583,
      "step": 1112
    },
    {
      "epoch": 1.0554765291607398,
      "grad_norm": 0.41025741095361523,
      "learning_rate": 8.189671767912502e-06,
      "loss": 0.0391,
      "step": 1113
    },
    {
      "epoch": 1.0564248458985301,
      "grad_norm": 0.42565607076822326,
      "learning_rate": 8.185417958639787e-06,
      "loss": 0.0511,
      "step": 1114
    },
    {
      "epoch": 1.0573731626363205,
      "grad_norm": 0.6066779731441627,
      "learning_rate": 8.18116026516707e-06,
      "loss": 0.0425,
      "step": 1115
    },
    {
      "epoch": 1.058321479374111,
      "grad_norm": 0.614194232019259,
      "learning_rate": 8.17689869268605e-06,
      "loss": 0.0399,
      "step": 1116
    },
    {
      "epoch": 1.0592697961119013,
      "grad_norm": 0.654835728000199,
      "learning_rate": 8.17263324639316e-06,
      "loss": 0.0509,
      "step": 1117
    },
    {
      "epoch": 1.0602181128496917,
      "grad_norm": 2.390822654668922,
      "learning_rate": 8.168363931489554e-06,
      "loss": 0.0538,
      "step": 1118
    },
    {
      "epoch": 1.0611664295874823,
      "grad_norm": 0.5309786733261611,
      "learning_rate": 8.164090753181097e-06,
      "loss": 0.0438,
      "step": 1119
    },
    {
      "epoch": 1.0621147463252727,
      "grad_norm": 0.5765552296723528,
      "learning_rate": 8.159813716678375e-06,
      "loss": 0.0573,
      "step": 1120
    },
    {
      "epoch": 1.063063063063063,
      "grad_norm": 0.702317664551232,
      "learning_rate": 8.15553282719667e-06,
      "loss": 0.0498,
      "step": 1121
    },
    {
      "epoch": 1.0640113798008535,
      "grad_norm": 0.7582330479572109,
      "learning_rate": 8.15124808995597e-06,
      "loss": 0.0686,
      "step": 1122
    },
    {
      "epoch": 1.0649596965386439,
      "grad_norm": 2.8142295533632353,
      "learning_rate": 8.146959510180947e-06,
      "loss": 0.0448,
      "step": 1123
    },
    {
      "epoch": 1.0659080132764343,
      "grad_norm": 0.8554647159912601,
      "learning_rate": 8.142667093100966e-06,
      "loss": 0.0636,
      "step": 1124
    },
    {
      "epoch": 1.0668563300142249,
      "grad_norm": 0.4583216098085599,
      "learning_rate": 8.138370843950067e-06,
      "loss": 0.0407,
      "step": 1125
    },
    {
      "epoch": 1.0678046467520153,
      "grad_norm": 0.5879158711527235,
      "learning_rate": 8.13407076796696e-06,
      "loss": 0.0779,
      "step": 1126
    },
    {
      "epoch": 1.0687529634898056,
      "grad_norm": 0.39347630183739596,
      "learning_rate": 8.129766870395027e-06,
      "loss": 0.0364,
      "step": 1127
    },
    {
      "epoch": 1.069701280227596,
      "grad_norm": 0.42771084872916015,
      "learning_rate": 8.125459156482307e-06,
      "loss": 0.0413,
      "step": 1128
    },
    {
      "epoch": 1.0706495969653864,
      "grad_norm": 0.5517158326408007,
      "learning_rate": 8.121147631481491e-06,
      "loss": 0.037,
      "step": 1129
    },
    {
      "epoch": 1.0715979137031768,
      "grad_norm": 0.49889308342240923,
      "learning_rate": 8.116832300649924e-06,
      "loss": 0.0464,
      "step": 1130
    },
    {
      "epoch": 1.0725462304409672,
      "grad_norm": 0.5993137897003766,
      "learning_rate": 8.112513169249583e-06,
      "loss": 0.0609,
      "step": 1131
    },
    {
      "epoch": 1.0734945471787578,
      "grad_norm": 0.543205478981709,
      "learning_rate": 8.108190242547082e-06,
      "loss": 0.0469,
      "step": 1132
    },
    {
      "epoch": 1.0744428639165482,
      "grad_norm": 0.7091841906617682,
      "learning_rate": 8.103863525813667e-06,
      "loss": 0.0414,
      "step": 1133
    },
    {
      "epoch": 1.0753911806543386,
      "grad_norm": 0.4370481397065285,
      "learning_rate": 8.0995330243252e-06,
      "loss": 0.0458,
      "step": 1134
    },
    {
      "epoch": 1.076339497392129,
      "grad_norm": 0.4502863591933898,
      "learning_rate": 8.095198743362162e-06,
      "loss": 0.0473,
      "step": 1135
    },
    {
      "epoch": 1.0772878141299194,
      "grad_norm": 0.5013748238976868,
      "learning_rate": 8.090860688209641e-06,
      "loss": 0.0466,
      "step": 1136
    },
    {
      "epoch": 1.0782361308677098,
      "grad_norm": 0.6748865559678139,
      "learning_rate": 8.086518864157325e-06,
      "loss": 0.0553,
      "step": 1137
    },
    {
      "epoch": 1.0791844476055001,
      "grad_norm": 0.6409644237971479,
      "learning_rate": 8.0821732764995e-06,
      "loss": 0.049,
      "step": 1138
    },
    {
      "epoch": 1.0801327643432908,
      "grad_norm": 0.6146233989253884,
      "learning_rate": 8.077823930535045e-06,
      "loss": 0.0472,
      "step": 1139
    },
    {
      "epoch": 1.0810810810810811,
      "grad_norm": 0.4066821137097016,
      "learning_rate": 8.073470831567414e-06,
      "loss": 0.0326,
      "step": 1140
    },
    {
      "epoch": 1.0820293978188715,
      "grad_norm": 0.8437034804113782,
      "learning_rate": 8.069113984904642e-06,
      "loss": 0.054,
      "step": 1141
    },
    {
      "epoch": 1.082977714556662,
      "grad_norm": 0.5161472652823138,
      "learning_rate": 8.064753395859333e-06,
      "loss": 0.0436,
      "step": 1142
    },
    {
      "epoch": 1.0839260312944523,
      "grad_norm": 0.48759533161402335,
      "learning_rate": 8.060389069748653e-06,
      "loss": 0.0462,
      "step": 1143
    },
    {
      "epoch": 1.0848743480322427,
      "grad_norm": 0.4172865968707472,
      "learning_rate": 8.05602101189433e-06,
      "loss": 0.0376,
      "step": 1144
    },
    {
      "epoch": 1.085822664770033,
      "grad_norm": 0.6834848316978862,
      "learning_rate": 8.051649227622634e-06,
      "loss": 0.0474,
      "step": 1145
    },
    {
      "epoch": 1.0867709815078237,
      "grad_norm": 0.6135237968103057,
      "learning_rate": 8.047273722264384e-06,
      "loss": 0.0377,
      "step": 1146
    },
    {
      "epoch": 1.087719298245614,
      "grad_norm": 0.5735481827911786,
      "learning_rate": 8.042894501154937e-06,
      "loss": 0.0577,
      "step": 1147
    },
    {
      "epoch": 1.0886676149834045,
      "grad_norm": 0.5126797458774718,
      "learning_rate": 8.03851156963418e-06,
      "loss": 0.0342,
      "step": 1148
    },
    {
      "epoch": 1.0896159317211949,
      "grad_norm": 0.5394680925590052,
      "learning_rate": 8.034124933046523e-06,
      "loss": 0.0427,
      "step": 1149
    },
    {
      "epoch": 1.0905642484589853,
      "grad_norm": 1.0619045413666424,
      "learning_rate": 8.029734596740895e-06,
      "loss": 0.0347,
      "step": 1150
    },
    {
      "epoch": 1.0915125651967756,
      "grad_norm": 0.5219216100491317,
      "learning_rate": 8.025340566070737e-06,
      "loss": 0.0392,
      "step": 1151
    },
    {
      "epoch": 1.0924608819345663,
      "grad_norm": 0.47125200466301426,
      "learning_rate": 8.020942846393993e-06,
      "loss": 0.049,
      "step": 1152
    },
    {
      "epoch": 1.0934091986723566,
      "grad_norm": 0.4310790509560298,
      "learning_rate": 8.016541443073105e-06,
      "loss": 0.038,
      "step": 1153
    },
    {
      "epoch": 1.094357515410147,
      "grad_norm": 0.7534317067913003,
      "learning_rate": 8.01213636147501e-06,
      "loss": 0.0536,
      "step": 1154
    },
    {
      "epoch": 1.0953058321479374,
      "grad_norm": 0.6207928110119649,
      "learning_rate": 8.007727606971127e-06,
      "loss": 0.0434,
      "step": 1155
    },
    {
      "epoch": 1.0962541488857278,
      "grad_norm": 0.57563137722865,
      "learning_rate": 8.003315184937355e-06,
      "loss": 0.0364,
      "step": 1156
    },
    {
      "epoch": 1.0972024656235182,
      "grad_norm": 0.5509639511135459,
      "learning_rate": 7.998899100754065e-06,
      "loss": 0.0517,
      "step": 1157
    },
    {
      "epoch": 1.0981507823613086,
      "grad_norm": 0.4267295425494574,
      "learning_rate": 7.994479359806091e-06,
      "loss": 0.0405,
      "step": 1158
    },
    {
      "epoch": 1.0990990990990992,
      "grad_norm": 0.49807239389073216,
      "learning_rate": 7.990055967482733e-06,
      "loss": 0.0352,
      "step": 1159
    },
    {
      "epoch": 1.1000474158368896,
      "grad_norm": 0.7895890014997967,
      "learning_rate": 7.985628929177733e-06,
      "loss": 0.0373,
      "step": 1160
    },
    {
      "epoch": 1.10099573257468,
      "grad_norm": 0.9249656986473295,
      "learning_rate": 7.98119825028929e-06,
      "loss": 0.0457,
      "step": 1161
    },
    {
      "epoch": 1.1019440493124704,
      "grad_norm": 0.627936744238416,
      "learning_rate": 7.976763936220031e-06,
      "loss": 0.047,
      "step": 1162
    },
    {
      "epoch": 1.1028923660502608,
      "grad_norm": 0.6238528107028002,
      "learning_rate": 7.972325992377026e-06,
      "loss": 0.0488,
      "step": 1163
    },
    {
      "epoch": 1.1038406827880511,
      "grad_norm": 0.4657651602214921,
      "learning_rate": 7.967884424171764e-06,
      "loss": 0.0327,
      "step": 1164
    },
    {
      "epoch": 1.1047889995258415,
      "grad_norm": 0.5850593930568389,
      "learning_rate": 7.963439237020157e-06,
      "loss": 0.0521,
      "step": 1165
    },
    {
      "epoch": 1.1057373162636321,
      "grad_norm": 0.7614458412566218,
      "learning_rate": 7.958990436342525e-06,
      "loss": 0.0708,
      "step": 1166
    },
    {
      "epoch": 1.1066856330014225,
      "grad_norm": 0.5375594263604339,
      "learning_rate": 7.954538027563601e-06,
      "loss": 0.0427,
      "step": 1167
    },
    {
      "epoch": 1.107633949739213,
      "grad_norm": 0.6028510385361613,
      "learning_rate": 7.950082016112514e-06,
      "loss": 0.0461,
      "step": 1168
    },
    {
      "epoch": 1.1085822664770033,
      "grad_norm": 0.7339781460019883,
      "learning_rate": 7.945622407422787e-06,
      "loss": 0.0449,
      "step": 1169
    },
    {
      "epoch": 1.1095305832147937,
      "grad_norm": 0.5248693288710427,
      "learning_rate": 7.941159206932327e-06,
      "loss": 0.0451,
      "step": 1170
    },
    {
      "epoch": 1.110478899952584,
      "grad_norm": 0.500024623512,
      "learning_rate": 7.936692420083423e-06,
      "loss": 0.0438,
      "step": 1171
    },
    {
      "epoch": 1.1114272166903745,
      "grad_norm": 0.46620520422481104,
      "learning_rate": 7.932222052322736e-06,
      "loss": 0.0469,
      "step": 1172
    },
    {
      "epoch": 1.112375533428165,
      "grad_norm": 0.7140775122015546,
      "learning_rate": 7.927748109101296e-06,
      "loss": 0.0492,
      "step": 1173
    },
    {
      "epoch": 1.1133238501659555,
      "grad_norm": 0.5204616685358204,
      "learning_rate": 7.923270595874489e-06,
      "loss": 0.0516,
      "step": 1174
    },
    {
      "epoch": 1.1142721669037459,
      "grad_norm": 0.41118460218118136,
      "learning_rate": 7.918789518102057e-06,
      "loss": 0.0374,
      "step": 1175
    },
    {
      "epoch": 1.1152204836415363,
      "grad_norm": 0.5459512894558529,
      "learning_rate": 7.914304881248085e-06,
      "loss": 0.0506,
      "step": 1176
    },
    {
      "epoch": 1.1161688003793266,
      "grad_norm": 0.5574376078006338,
      "learning_rate": 7.909816690781005e-06,
      "loss": 0.0515,
      "step": 1177
    },
    {
      "epoch": 1.117117117117117,
      "grad_norm": 0.2933208036522205,
      "learning_rate": 7.905324952173573e-06,
      "loss": 0.0345,
      "step": 1178
    },
    {
      "epoch": 1.1180654338549076,
      "grad_norm": 0.4993016043563598,
      "learning_rate": 7.900829670902876e-06,
      "loss": 0.0487,
      "step": 1179
    },
    {
      "epoch": 1.119013750592698,
      "grad_norm": 0.4981848296785552,
      "learning_rate": 7.896330852450324e-06,
      "loss": 0.049,
      "step": 1180
    },
    {
      "epoch": 1.1199620673304884,
      "grad_norm": 0.5492037552465713,
      "learning_rate": 7.891828502301637e-06,
      "loss": 0.0626,
      "step": 1181
    },
    {
      "epoch": 1.1209103840682788,
      "grad_norm": 0.5945243855558169,
      "learning_rate": 7.887322625946836e-06,
      "loss": 0.0566,
      "step": 1182
    },
    {
      "epoch": 1.1218587008060692,
      "grad_norm": 0.5008044550443231,
      "learning_rate": 7.882813228880253e-06,
      "loss": 0.0365,
      "step": 1183
    },
    {
      "epoch": 1.1228070175438596,
      "grad_norm": 0.4974033363856913,
      "learning_rate": 7.878300316600504e-06,
      "loss": 0.0508,
      "step": 1184
    },
    {
      "epoch": 1.12375533428165,
      "grad_norm": 1.1163525573958883,
      "learning_rate": 7.873783894610496e-06,
      "loss": 0.0591,
      "step": 1185
    },
    {
      "epoch": 1.1247036510194406,
      "grad_norm": 0.529056835529291,
      "learning_rate": 7.869263968417412e-06,
      "loss": 0.041,
      "step": 1186
    },
    {
      "epoch": 1.125651967757231,
      "grad_norm": 0.39561323424421474,
      "learning_rate": 7.864740543532711e-06,
      "loss": 0.043,
      "step": 1187
    },
    {
      "epoch": 1.1266002844950214,
      "grad_norm": 0.5719661025649166,
      "learning_rate": 7.860213625472119e-06,
      "loss": 0.0394,
      "step": 1188
    },
    {
      "epoch": 1.1275486012328118,
      "grad_norm": 0.683758567602495,
      "learning_rate": 7.855683219755617e-06,
      "loss": 0.0596,
      "step": 1189
    },
    {
      "epoch": 1.1284969179706021,
      "grad_norm": 0.43615448430346,
      "learning_rate": 7.851149331907442e-06,
      "loss": 0.0458,
      "step": 1190
    },
    {
      "epoch": 1.1294452347083925,
      "grad_norm": 0.5455285097679818,
      "learning_rate": 7.84661196745608e-06,
      "loss": 0.0458,
      "step": 1191
    },
    {
      "epoch": 1.1303935514461831,
      "grad_norm": 0.46823267018975095,
      "learning_rate": 7.842071131934246e-06,
      "loss": 0.0395,
      "step": 1192
    },
    {
      "epoch": 1.1313418681839735,
      "grad_norm": 0.6410339405297693,
      "learning_rate": 7.837526830878901e-06,
      "loss": 0.0528,
      "step": 1193
    },
    {
      "epoch": 1.132290184921764,
      "grad_norm": 0.6329347246677137,
      "learning_rate": 7.83297906983122e-06,
      "loss": 0.0694,
      "step": 1194
    },
    {
      "epoch": 1.1332385016595543,
      "grad_norm": 1.1861963543383596,
      "learning_rate": 7.828427854336604e-06,
      "loss": 0.0748,
      "step": 1195
    },
    {
      "epoch": 1.1341868183973447,
      "grad_norm": 0.4102764477580429,
      "learning_rate": 7.823873189944664e-06,
      "loss": 0.0393,
      "step": 1196
    },
    {
      "epoch": 1.135135135135135,
      "grad_norm": 0.4452453730186256,
      "learning_rate": 7.819315082209217e-06,
      "loss": 0.0354,
      "step": 1197
    },
    {
      "epoch": 1.1360834518729255,
      "grad_norm": 0.45033548160883075,
      "learning_rate": 7.814753536688278e-06,
      "loss": 0.0366,
      "step": 1198
    },
    {
      "epoch": 1.1370317686107159,
      "grad_norm": 0.5479341929657664,
      "learning_rate": 7.810188558944054e-06,
      "loss": 0.0606,
      "step": 1199
    },
    {
      "epoch": 1.1379800853485065,
      "grad_norm": 0.5550674838469363,
      "learning_rate": 7.805620154542938e-06,
      "loss": 0.0406,
      "step": 1200
    },
    {
      "epoch": 1.1389284020862969,
      "grad_norm": 0.6478804122717543,
      "learning_rate": 7.801048329055502e-06,
      "loss": 0.0644,
      "step": 1201
    },
    {
      "epoch": 1.1398767188240873,
      "grad_norm": 0.40041584333311264,
      "learning_rate": 7.796473088056487e-06,
      "loss": 0.0378,
      "step": 1202
    },
    {
      "epoch": 1.1408250355618776,
      "grad_norm": 0.4223598793738602,
      "learning_rate": 7.7918944371248e-06,
      "loss": 0.039,
      "step": 1203
    },
    {
      "epoch": 1.141773352299668,
      "grad_norm": 0.4961808103718134,
      "learning_rate": 7.787312381843503e-06,
      "loss": 0.042,
      "step": 1204
    },
    {
      "epoch": 1.1427216690374584,
      "grad_norm": 0.4738964503477037,
      "learning_rate": 7.78272692779982e-06,
      "loss": 0.036,
      "step": 1205
    },
    {
      "epoch": 1.143669985775249,
      "grad_norm": 0.48670332456917464,
      "learning_rate": 7.778138080585107e-06,
      "loss": 0.0298,
      "step": 1206
    },
    {
      "epoch": 1.1446183025130394,
      "grad_norm": 0.5855118813585286,
      "learning_rate": 7.77354584579486e-06,
      "loss": 0.0421,
      "step": 1207
    },
    {
      "epoch": 1.1455666192508298,
      "grad_norm": 0.4306606042859337,
      "learning_rate": 7.768950229028713e-06,
      "loss": 0.0417,
      "step": 1208
    },
    {
      "epoch": 1.1465149359886202,
      "grad_norm": 0.4725287807287063,
      "learning_rate": 7.764351235890416e-06,
      "loss": 0.0386,
      "step": 1209
    },
    {
      "epoch": 1.1474632527264106,
      "grad_norm": 0.6431359997140053,
      "learning_rate": 7.759748871987838e-06,
      "loss": 0.0759,
      "step": 1210
    },
    {
      "epoch": 1.148411569464201,
      "grad_norm": 0.42470773245514043,
      "learning_rate": 7.75514314293296e-06,
      "loss": 0.0393,
      "step": 1211
    },
    {
      "epoch": 1.1493598862019914,
      "grad_norm": 0.4346642115827222,
      "learning_rate": 7.750534054341866e-06,
      "loss": 0.0381,
      "step": 1212
    },
    {
      "epoch": 1.150308202939782,
      "grad_norm": 0.5312065161872603,
      "learning_rate": 7.745921611834734e-06,
      "loss": 0.0397,
      "step": 1213
    },
    {
      "epoch": 1.1512565196775724,
      "grad_norm": 0.5385091421622926,
      "learning_rate": 7.741305821035836e-06,
      "loss": 0.0582,
      "step": 1214
    },
    {
      "epoch": 1.1522048364153628,
      "grad_norm": 0.4778078646025971,
      "learning_rate": 7.736686687573523e-06,
      "loss": 0.0437,
      "step": 1215
    },
    {
      "epoch": 1.1531531531531531,
      "grad_norm": 0.709371140831048,
      "learning_rate": 7.732064217080224e-06,
      "loss": 0.0492,
      "step": 1216
    },
    {
      "epoch": 1.1541014698909435,
      "grad_norm": 0.4183114831368168,
      "learning_rate": 7.727438415192434e-06,
      "loss": 0.0393,
      "step": 1217
    },
    {
      "epoch": 1.155049786628734,
      "grad_norm": 0.541221858338076,
      "learning_rate": 7.722809287550716e-06,
      "loss": 0.0459,
      "step": 1218
    },
    {
      "epoch": 1.1559981033665245,
      "grad_norm": 0.4067212457533803,
      "learning_rate": 7.718176839799682e-06,
      "loss": 0.0362,
      "step": 1219
    },
    {
      "epoch": 1.156946420104315,
      "grad_norm": 0.4383704109007449,
      "learning_rate": 7.713541077587996e-06,
      "loss": 0.0339,
      "step": 1220
    },
    {
      "epoch": 1.1578947368421053,
      "grad_norm": 0.4650740040984162,
      "learning_rate": 7.708902006568365e-06,
      "loss": 0.036,
      "step": 1221
    },
    {
      "epoch": 1.1588430535798957,
      "grad_norm": 0.3918848259293643,
      "learning_rate": 7.704259632397525e-06,
      "loss": 0.0307,
      "step": 1222
    },
    {
      "epoch": 1.159791370317686,
      "grad_norm": 0.49823361822736345,
      "learning_rate": 7.699613960736247e-06,
      "loss": 0.0474,
      "step": 1223
    },
    {
      "epoch": 1.1607396870554765,
      "grad_norm": 0.4868259723121106,
      "learning_rate": 7.694964997249317e-06,
      "loss": 0.0463,
      "step": 1224
    },
    {
      "epoch": 1.1616880037932669,
      "grad_norm": 0.48069332380581936,
      "learning_rate": 7.690312747605536e-06,
      "loss": 0.0426,
      "step": 1225
    },
    {
      "epoch": 1.1626363205310573,
      "grad_norm": 0.491998777057983,
      "learning_rate": 7.685657217477716e-06,
      "loss": 0.0389,
      "step": 1226
    },
    {
      "epoch": 1.1635846372688479,
      "grad_norm": 0.35813001299038266,
      "learning_rate": 7.680998412542665e-06,
      "loss": 0.0331,
      "step": 1227
    },
    {
      "epoch": 1.1645329540066383,
      "grad_norm": 0.5968215449565825,
      "learning_rate": 7.676336338481182e-06,
      "loss": 0.0406,
      "step": 1228
    },
    {
      "epoch": 1.1654812707444286,
      "grad_norm": 0.5559719660043877,
      "learning_rate": 7.671671000978063e-06,
      "loss": 0.0712,
      "step": 1229
    },
    {
      "epoch": 1.166429587482219,
      "grad_norm": 0.42335427343139226,
      "learning_rate": 7.66700240572207e-06,
      "loss": 0.0376,
      "step": 1230
    },
    {
      "epoch": 1.1673779042200094,
      "grad_norm": 0.49655937772975667,
      "learning_rate": 7.662330558405943e-06,
      "loss": 0.0414,
      "step": 1231
    },
    {
      "epoch": 1.1683262209577998,
      "grad_norm": 0.3994453444748161,
      "learning_rate": 7.657655464726395e-06,
      "loss": 0.0404,
      "step": 1232
    },
    {
      "epoch": 1.1692745376955904,
      "grad_norm": 0.41285430890496555,
      "learning_rate": 7.652977130384083e-06,
      "loss": 0.0486,
      "step": 1233
    },
    {
      "epoch": 1.1702228544333808,
      "grad_norm": 0.6071610144149513,
      "learning_rate": 7.648295561083627e-06,
      "loss": 0.0633,
      "step": 1234
    },
    {
      "epoch": 1.1711711711711712,
      "grad_norm": 0.5259863664992493,
      "learning_rate": 7.643610762533584e-06,
      "loss": 0.0585,
      "step": 1235
    },
    {
      "epoch": 1.1721194879089616,
      "grad_norm": 0.3946045435612188,
      "learning_rate": 7.638922740446457e-06,
      "loss": 0.0355,
      "step": 1236
    },
    {
      "epoch": 1.173067804646752,
      "grad_norm": 0.3568622458812664,
      "learning_rate": 7.63423150053867e-06,
      "loss": 0.0295,
      "step": 1237
    },
    {
      "epoch": 1.1740161213845424,
      "grad_norm": 0.5469312105534027,
      "learning_rate": 7.629537048530583e-06,
      "loss": 0.0403,
      "step": 1238
    },
    {
      "epoch": 1.1749644381223328,
      "grad_norm": 0.5387702508086157,
      "learning_rate": 7.6248393901464564e-06,
      "loss": 0.0483,
      "step": 1239
    },
    {
      "epoch": 1.1759127548601234,
      "grad_norm": 0.5406781757294887,
      "learning_rate": 7.620138531114476e-06,
      "loss": 0.0419,
      "step": 1240
    },
    {
      "epoch": 1.1768610715979138,
      "grad_norm": 0.48066173819586494,
      "learning_rate": 7.61543447716672e-06,
      "loss": 0.0471,
      "step": 1241
    },
    {
      "epoch": 1.1778093883357041,
      "grad_norm": 0.5132766758180038,
      "learning_rate": 7.610727234039168e-06,
      "loss": 0.0373,
      "step": 1242
    },
    {
      "epoch": 1.1787577050734945,
      "grad_norm": 0.4440200211246783,
      "learning_rate": 7.606016807471686e-06,
      "loss": 0.0303,
      "step": 1243
    },
    {
      "epoch": 1.179706021811285,
      "grad_norm": 0.5120784198135685,
      "learning_rate": 7.601303203208021e-06,
      "loss": 0.0415,
      "step": 1244
    },
    {
      "epoch": 1.1806543385490753,
      "grad_norm": 0.5119968299555113,
      "learning_rate": 7.596586426995798e-06,
      "loss": 0.033,
      "step": 1245
    },
    {
      "epoch": 1.181602655286866,
      "grad_norm": 0.649285479277398,
      "learning_rate": 7.591866484586505e-06,
      "loss": 0.0412,
      "step": 1246
    },
    {
      "epoch": 1.1825509720246563,
      "grad_norm": 0.8899853131901044,
      "learning_rate": 7.587143381735498e-06,
      "loss": 0.0495,
      "step": 1247
    },
    {
      "epoch": 1.1834992887624467,
      "grad_norm": 0.6010967444971024,
      "learning_rate": 7.5824171242019796e-06,
      "loss": 0.0412,
      "step": 1248
    },
    {
      "epoch": 1.184447605500237,
      "grad_norm": 0.5257812163804602,
      "learning_rate": 7.5776877177490004e-06,
      "loss": 0.041,
      "step": 1249
    },
    {
      "epoch": 1.1853959222380275,
      "grad_norm": 0.5059417156750543,
      "learning_rate": 7.572955168143456e-06,
      "loss": 0.0564,
      "step": 1250
    },
    {
      "epoch": 1.1863442389758179,
      "grad_norm": 0.5287163894503856,
      "learning_rate": 7.568219481156067e-06,
      "loss": 0.0443,
      "step": 1251
    },
    {
      "epoch": 1.1872925557136083,
      "grad_norm": 0.5764185328336728,
      "learning_rate": 7.563480662561386e-06,
      "loss": 0.0408,
      "step": 1252
    },
    {
      "epoch": 1.1882408724513986,
      "grad_norm": 0.4606910371892103,
      "learning_rate": 7.55873871813778e-06,
      "loss": 0.0486,
      "step": 1253
    },
    {
      "epoch": 1.1891891891891893,
      "grad_norm": 0.4307521432754428,
      "learning_rate": 7.553993653667432e-06,
      "loss": 0.0311,
      "step": 1254
    },
    {
      "epoch": 1.1901375059269796,
      "grad_norm": 0.9134386020304202,
      "learning_rate": 7.549245474936324e-06,
      "loss": 0.0461,
      "step": 1255
    },
    {
      "epoch": 1.19108582266477,
      "grad_norm": 0.4558303355166927,
      "learning_rate": 7.544494187734237e-06,
      "loss": 0.0499,
      "step": 1256
    },
    {
      "epoch": 1.1920341394025604,
      "grad_norm": 0.3989843552269527,
      "learning_rate": 7.539739797854746e-06,
      "loss": 0.0349,
      "step": 1257
    },
    {
      "epoch": 1.1929824561403508,
      "grad_norm": 0.3906291841756068,
      "learning_rate": 7.5349823110952046e-06,
      "loss": 0.0281,
      "step": 1258
    },
    {
      "epoch": 1.1939307728781414,
      "grad_norm": 0.41315135278835374,
      "learning_rate": 7.530221733256749e-06,
      "loss": 0.0436,
      "step": 1259
    },
    {
      "epoch": 1.1948790896159318,
      "grad_norm": 1.0487429774583699,
      "learning_rate": 7.525458070144276e-06,
      "loss": 0.0456,
      "step": 1260
    },
    {
      "epoch": 1.1958274063537222,
      "grad_norm": 0.7492477244429876,
      "learning_rate": 7.520691327566449e-06,
      "loss": 0.061,
      "step": 1261
    },
    {
      "epoch": 1.1967757230915126,
      "grad_norm": 0.6635915757614768,
      "learning_rate": 7.515921511335689e-06,
      "loss": 0.0686,
      "step": 1262
    },
    {
      "epoch": 1.197724039829303,
      "grad_norm": 0.5712864519563702,
      "learning_rate": 7.511148627268161e-06,
      "loss": 0.0487,
      "step": 1263
    },
    {
      "epoch": 1.1986723565670934,
      "grad_norm": 0.603343671959781,
      "learning_rate": 7.5063726811837716e-06,
      "loss": 0.052,
      "step": 1264
    },
    {
      "epoch": 1.1996206733048838,
      "grad_norm": 0.4406684429170475,
      "learning_rate": 7.501593678906161e-06,
      "loss": 0.0414,
      "step": 1265
    },
    {
      "epoch": 1.2005689900426741,
      "grad_norm": 0.6442697808186765,
      "learning_rate": 7.496811626262699e-06,
      "loss": 0.0423,
      "step": 1266
    },
    {
      "epoch": 1.2015173067804648,
      "grad_norm": 0.5559558578993309,
      "learning_rate": 7.492026529084468e-06,
      "loss": 0.0402,
      "step": 1267
    },
    {
      "epoch": 1.2024656235182551,
      "grad_norm": 0.5454144102950605,
      "learning_rate": 7.487238393206271e-06,
      "loss": 0.0367,
      "step": 1268
    },
    {
      "epoch": 1.2034139402560455,
      "grad_norm": 0.5693008724632117,
      "learning_rate": 7.4824472244666134e-06,
      "loss": 0.043,
      "step": 1269
    },
    {
      "epoch": 1.204362256993836,
      "grad_norm": 0.6776730947602304,
      "learning_rate": 7.477653028707694e-06,
      "loss": 0.0455,
      "step": 1270
    },
    {
      "epoch": 1.2053105737316263,
      "grad_norm": 0.5055727203867033,
      "learning_rate": 7.472855811775411e-06,
      "loss": 0.0398,
      "step": 1271
    },
    {
      "epoch": 1.2062588904694167,
      "grad_norm": 0.6863655340816167,
      "learning_rate": 7.468055579519338e-06,
      "loss": 0.0338,
      "step": 1272
    },
    {
      "epoch": 1.2072072072072073,
      "grad_norm": 0.5081101945117912,
      "learning_rate": 7.4632523377927335e-06,
      "loss": 0.0495,
      "step": 1273
    },
    {
      "epoch": 1.2081555239449977,
      "grad_norm": 0.2936628736656289,
      "learning_rate": 7.458446092452518e-06,
      "loss": 0.0271,
      "step": 1274
    },
    {
      "epoch": 1.209103840682788,
      "grad_norm": 0.36318292005655967,
      "learning_rate": 7.453636849359281e-06,
      "loss": 0.0291,
      "step": 1275
    },
    {
      "epoch": 1.2100521574205785,
      "grad_norm": 0.6126900566455881,
      "learning_rate": 7.448824614377264e-06,
      "loss": 0.0486,
      "step": 1276
    },
    {
      "epoch": 1.2110004741583689,
      "grad_norm": 0.5964527538723606,
      "learning_rate": 7.444009393374356e-06,
      "loss": 0.0459,
      "step": 1277
    },
    {
      "epoch": 1.2119487908961593,
      "grad_norm": 0.473196708998212,
      "learning_rate": 7.43919119222209e-06,
      "loss": 0.0323,
      "step": 1278
    },
    {
      "epoch": 1.2128971076339496,
      "grad_norm": 0.4144321881388711,
      "learning_rate": 7.434370016795629e-06,
      "loss": 0.0393,
      "step": 1279
    },
    {
      "epoch": 1.2138454243717403,
      "grad_norm": 0.5122361225802895,
      "learning_rate": 7.429545872973765e-06,
      "loss": 0.0405,
      "step": 1280
    },
    {
      "epoch": 1.2147937411095306,
      "grad_norm": 0.5224599737121133,
      "learning_rate": 7.424718766638915e-06,
      "loss": 0.0468,
      "step": 1281
    },
    {
      "epoch": 1.215742057847321,
      "grad_norm": 0.3874552596536454,
      "learning_rate": 7.419888703677097e-06,
      "loss": 0.0352,
      "step": 1282
    },
    {
      "epoch": 1.2166903745851114,
      "grad_norm": 0.5431714072384897,
      "learning_rate": 7.415055689977943e-06,
      "loss": 0.0573,
      "step": 1283
    },
    {
      "epoch": 1.2176386913229018,
      "grad_norm": 0.6074067354337574,
      "learning_rate": 7.4102197314346765e-06,
      "loss": 0.0569,
      "step": 1284
    },
    {
      "epoch": 1.2185870080606922,
      "grad_norm": 0.5384133252576224,
      "learning_rate": 7.40538083394412e-06,
      "loss": 0.0386,
      "step": 1285
    },
    {
      "epoch": 1.2195353247984828,
      "grad_norm": 0.6817357798912372,
      "learning_rate": 7.400539003406675e-06,
      "loss": 0.0514,
      "step": 1286
    },
    {
      "epoch": 1.2204836415362732,
      "grad_norm": 0.576076706598001,
      "learning_rate": 7.3956942457263184e-06,
      "loss": 0.0475,
      "step": 1287
    },
    {
      "epoch": 1.2214319582740636,
      "grad_norm": 0.3761588683293121,
      "learning_rate": 7.3908465668105955e-06,
      "loss": 0.0342,
      "step": 1288
    },
    {
      "epoch": 1.222380275011854,
      "grad_norm": 0.5622394777117021,
      "learning_rate": 7.3859959725706185e-06,
      "loss": 0.0513,
      "step": 1289
    },
    {
      "epoch": 1.2233285917496444,
      "grad_norm": 0.4516184849695626,
      "learning_rate": 7.381142468921052e-06,
      "loss": 0.0388,
      "step": 1290
    },
    {
      "epoch": 1.2242769084874348,
      "grad_norm": 0.3617313806173479,
      "learning_rate": 7.376286061780108e-06,
      "loss": 0.0279,
      "step": 1291
    },
    {
      "epoch": 1.2252252252252251,
      "grad_norm": 0.6044099547624433,
      "learning_rate": 7.371426757069538e-06,
      "loss": 0.0772,
      "step": 1292
    },
    {
      "epoch": 1.2261735419630155,
      "grad_norm": 0.46194460498572665,
      "learning_rate": 7.3665645607146266e-06,
      "loss": 0.0332,
      "step": 1293
    },
    {
      "epoch": 1.2271218587008061,
      "grad_norm": 0.6412897652910509,
      "learning_rate": 7.3616994786441865e-06,
      "loss": 0.0655,
      "step": 1294
    },
    {
      "epoch": 1.2280701754385965,
      "grad_norm": 0.4834273536592659,
      "learning_rate": 7.356831516790549e-06,
      "loss": 0.0424,
      "step": 1295
    },
    {
      "epoch": 1.229018492176387,
      "grad_norm": 0.4916473080153912,
      "learning_rate": 7.351960681089555e-06,
      "loss": 0.0374,
      "step": 1296
    },
    {
      "epoch": 1.2299668089141773,
      "grad_norm": 0.5082915641168106,
      "learning_rate": 7.347086977480552e-06,
      "loss": 0.0545,
      "step": 1297
    },
    {
      "epoch": 1.2309151256519677,
      "grad_norm": 0.47685406428324695,
      "learning_rate": 7.3422104119063815e-06,
      "loss": 0.0647,
      "step": 1298
    },
    {
      "epoch": 1.231863442389758,
      "grad_norm": 0.40154465693360425,
      "learning_rate": 7.337330990313377e-06,
      "loss": 0.0404,
      "step": 1299
    },
    {
      "epoch": 1.2328117591275487,
      "grad_norm": 0.535708958789697,
      "learning_rate": 7.332448718651355e-06,
      "loss": 0.0365,
      "step": 1300
    },
    {
      "epoch": 1.233760075865339,
      "grad_norm": 0.7669741889240554,
      "learning_rate": 7.327563602873604e-06,
      "loss": 0.0563,
      "step": 1301
    },
    {
      "epoch": 1.2347083926031295,
      "grad_norm": 0.4935695147587443,
      "learning_rate": 7.322675648936887e-06,
      "loss": 0.042,
      "step": 1302
    },
    {
      "epoch": 1.2356567093409199,
      "grad_norm": 0.5649346481989781,
      "learning_rate": 7.31778486280142e-06,
      "loss": 0.0596,
      "step": 1303
    },
    {
      "epoch": 1.2366050260787103,
      "grad_norm": 0.5681237194541117,
      "learning_rate": 7.312891250430878e-06,
      "loss": 0.0582,
      "step": 1304
    },
    {
      "epoch": 1.2375533428165006,
      "grad_norm": 0.3890664055979468,
      "learning_rate": 7.3079948177923785e-06,
      "loss": 0.0363,
      "step": 1305
    },
    {
      "epoch": 1.238501659554291,
      "grad_norm": 0.7279319600890861,
      "learning_rate": 7.303095570856482e-06,
      "loss": 0.0403,
      "step": 1306
    },
    {
      "epoch": 1.2394499762920816,
      "grad_norm": 0.6266894196950298,
      "learning_rate": 7.298193515597177e-06,
      "loss": 0.0481,
      "step": 1307
    },
    {
      "epoch": 1.240398293029872,
      "grad_norm": 0.5020074442444163,
      "learning_rate": 7.29328865799188e-06,
      "loss": 0.0325,
      "step": 1308
    },
    {
      "epoch": 1.2413466097676624,
      "grad_norm": 0.4375223150765817,
      "learning_rate": 7.288381004021419e-06,
      "loss": 0.0512,
      "step": 1309
    },
    {
      "epoch": 1.2422949265054528,
      "grad_norm": 0.38682876731404114,
      "learning_rate": 7.283470559670037e-06,
      "loss": 0.0458,
      "step": 1310
    },
    {
      "epoch": 1.2432432432432432,
      "grad_norm": 0.821002820077065,
      "learning_rate": 7.278557330925378e-06,
      "loss": 0.0441,
      "step": 1311
    },
    {
      "epoch": 1.2441915599810336,
      "grad_norm": 0.36531044461502105,
      "learning_rate": 7.273641323778482e-06,
      "loss": 0.0321,
      "step": 1312
    },
    {
      "epoch": 1.2451398767188242,
      "grad_norm": 0.8754259648216565,
      "learning_rate": 7.268722544223773e-06,
      "loss": 0.0551,
      "step": 1313
    },
    {
      "epoch": 1.2460881934566146,
      "grad_norm": 0.7875132084608685,
      "learning_rate": 7.263800998259061e-06,
      "loss": 0.0576,
      "step": 1314
    },
    {
      "epoch": 1.247036510194405,
      "grad_norm": 0.7147209757326294,
      "learning_rate": 7.258876691885526e-06,
      "loss": 0.0596,
      "step": 1315
    },
    {
      "epoch": 1.2479848269321954,
      "grad_norm": 0.6384875688620932,
      "learning_rate": 7.253949631107713e-06,
      "loss": 0.0539,
      "step": 1316
    },
    {
      "epoch": 1.2489331436699858,
      "grad_norm": 0.4048643826220915,
      "learning_rate": 7.24901982193353e-06,
      "loss": 0.0353,
      "step": 1317
    },
    {
      "epoch": 1.2498814604077761,
      "grad_norm": 0.5060955434368241,
      "learning_rate": 7.2440872703742314e-06,
      "loss": 0.0375,
      "step": 1318
    },
    {
      "epoch": 1.2508297771455665,
      "grad_norm": 0.3576520845581809,
      "learning_rate": 7.239151982444421e-06,
      "loss": 0.0311,
      "step": 1319
    },
    {
      "epoch": 1.251778093883357,
      "grad_norm": 0.6201923777708439,
      "learning_rate": 7.234213964162033e-06,
      "loss": 0.0461,
      "step": 1320
    },
    {
      "epoch": 1.2527264106211475,
      "grad_norm": 0.48992618453543096,
      "learning_rate": 7.2292732215483316e-06,
      "loss": 0.0425,
      "step": 1321
    },
    {
      "epoch": 1.253674727358938,
      "grad_norm": 0.5956731057701343,
      "learning_rate": 7.2243297606279114e-06,
      "loss": 0.0525,
      "step": 1322
    },
    {
      "epoch": 1.2546230440967283,
      "grad_norm": 0.5525693608729175,
      "learning_rate": 7.219383587428673e-06,
      "loss": 0.0423,
      "step": 1323
    },
    {
      "epoch": 1.2555713608345187,
      "grad_norm": 0.4690246094368384,
      "learning_rate": 7.214434707981825e-06,
      "loss": 0.0454,
      "step": 1324
    },
    {
      "epoch": 1.256519677572309,
      "grad_norm": 0.44673878692961855,
      "learning_rate": 7.209483128321881e-06,
      "loss": 0.0249,
      "step": 1325
    },
    {
      "epoch": 1.2574679943100997,
      "grad_norm": 0.43586039667176457,
      "learning_rate": 7.204528854486641e-06,
      "loss": 0.0479,
      "step": 1326
    },
    {
      "epoch": 1.25841631104789,
      "grad_norm": 0.5118218171714337,
      "learning_rate": 7.199571892517194e-06,
      "loss": 0.0567,
      "step": 1327
    },
    {
      "epoch": 1.2593646277856805,
      "grad_norm": 0.5965669555284678,
      "learning_rate": 7.194612248457907e-06,
      "loss": 0.044,
      "step": 1328
    },
    {
      "epoch": 1.2603129445234709,
      "grad_norm": 0.8010455651119536,
      "learning_rate": 7.189649928356413e-06,
      "loss": 0.0685,
      "step": 1329
    },
    {
      "epoch": 1.2612612612612613,
      "grad_norm": 0.4947586841353111,
      "learning_rate": 7.184684938263617e-06,
      "loss": 0.048,
      "step": 1330
    },
    {
      "epoch": 1.2622095779990516,
      "grad_norm": 0.43519037882369493,
      "learning_rate": 7.179717284233671e-06,
      "loss": 0.04,
      "step": 1331
    },
    {
      "epoch": 1.263157894736842,
      "grad_norm": 0.7723051121174525,
      "learning_rate": 7.174746972323976e-06,
      "loss": 0.0613,
      "step": 1332
    },
    {
      "epoch": 1.2641062114746324,
      "grad_norm": 0.670904737479652,
      "learning_rate": 7.169774008595178e-06,
      "loss": 0.042,
      "step": 1333
    },
    {
      "epoch": 1.2650545282124228,
      "grad_norm": 0.43695736935248614,
      "learning_rate": 7.164798399111159e-06,
      "loss": 0.0448,
      "step": 1334
    },
    {
      "epoch": 1.2660028449502134,
      "grad_norm": 0.45307761051009254,
      "learning_rate": 7.159820149939019e-06,
      "loss": 0.0448,
      "step": 1335
    },
    {
      "epoch": 1.2669511616880038,
      "grad_norm": 0.41666909675783154,
      "learning_rate": 7.154839267149082e-06,
      "loss": 0.042,
      "step": 1336
    },
    {
      "epoch": 1.2678994784257942,
      "grad_norm": 0.3919081120060574,
      "learning_rate": 7.1498557568148795e-06,
      "loss": 0.0452,
      "step": 1337
    },
    {
      "epoch": 1.2688477951635846,
      "grad_norm": 0.7267472981372689,
      "learning_rate": 7.144869625013155e-06,
      "loss": 0.0409,
      "step": 1338
    },
    {
      "epoch": 1.269796111901375,
      "grad_norm": 0.40985430024156805,
      "learning_rate": 7.1398808778238395e-06,
      "loss": 0.0418,
      "step": 1339
    },
    {
      "epoch": 1.2707444286391656,
      "grad_norm": 0.47624354455999457,
      "learning_rate": 7.134889521330056e-06,
      "loss": 0.0452,
      "step": 1340
    },
    {
      "epoch": 1.271692745376956,
      "grad_norm": 0.45691791490421313,
      "learning_rate": 7.129895561618113e-06,
      "loss": 0.0421,
      "step": 1341
    },
    {
      "epoch": 1.2726410621147464,
      "grad_norm": 0.997741074324212,
      "learning_rate": 7.124899004777489e-06,
      "loss": 0.0586,
      "step": 1342
    },
    {
      "epoch": 1.2735893788525368,
      "grad_norm": 0.4702689373485785,
      "learning_rate": 7.119899856900831e-06,
      "loss": 0.0381,
      "step": 1343
    },
    {
      "epoch": 1.2745376955903271,
      "grad_norm": 0.454291173319732,
      "learning_rate": 7.114898124083944e-06,
      "loss": 0.0289,
      "step": 1344
    },
    {
      "epoch": 1.2754860123281175,
      "grad_norm": 0.4439776694870836,
      "learning_rate": 7.109893812425785e-06,
      "loss": 0.0357,
      "step": 1345
    },
    {
      "epoch": 1.276434329065908,
      "grad_norm": 0.747360067085319,
      "learning_rate": 7.104886928028462e-06,
      "loss": 0.0537,
      "step": 1346
    },
    {
      "epoch": 1.2773826458036983,
      "grad_norm": 0.5685529445018764,
      "learning_rate": 7.09987747699721e-06,
      "loss": 0.0527,
      "step": 1347
    },
    {
      "epoch": 1.278330962541489,
      "grad_norm": 0.5663642475796429,
      "learning_rate": 7.0948654654403996e-06,
      "loss": 0.043,
      "step": 1348
    },
    {
      "epoch": 1.2792792792792793,
      "grad_norm": 0.4203561592323753,
      "learning_rate": 7.08985089946952e-06,
      "loss": 0.0442,
      "step": 1349
    },
    {
      "epoch": 1.2802275960170697,
      "grad_norm": 0.8806133526242714,
      "learning_rate": 7.0848337851991836e-06,
      "loss": 0.0683,
      "step": 1350
    },
    {
      "epoch": 1.28117591275486,
      "grad_norm": 0.4621171462388856,
      "learning_rate": 7.0798141287470974e-06,
      "loss": 0.0441,
      "step": 1351
    },
    {
      "epoch": 1.2821242294926505,
      "grad_norm": 0.4303832781558267,
      "learning_rate": 7.074791936234083e-06,
      "loss": 0.0432,
      "step": 1352
    },
    {
      "epoch": 1.283072546230441,
      "grad_norm": 0.593851630131963,
      "learning_rate": 7.069767213784037e-06,
      "loss": 0.0583,
      "step": 1353
    },
    {
      "epoch": 1.2840208629682315,
      "grad_norm": 0.45030313612782086,
      "learning_rate": 7.064739967523957e-06,
      "loss": 0.0367,
      "step": 1354
    },
    {
      "epoch": 1.2849691797060219,
      "grad_norm": 0.5237382802642779,
      "learning_rate": 7.059710203583908e-06,
      "loss": 0.0335,
      "step": 1355
    },
    {
      "epoch": 1.2859174964438123,
      "grad_norm": 0.46802621728751537,
      "learning_rate": 7.054677928097031e-06,
      "loss": 0.0362,
      "step": 1356
    },
    {
      "epoch": 1.2868658131816026,
      "grad_norm": 0.7397145007928059,
      "learning_rate": 7.0496431471995255e-06,
      "loss": 0.0352,
      "step": 1357
    },
    {
      "epoch": 1.287814129919393,
      "grad_norm": 0.44236361469981716,
      "learning_rate": 7.044605867030647e-06,
      "loss": 0.0367,
      "step": 1358
    },
    {
      "epoch": 1.2887624466571834,
      "grad_norm": 0.45828570870759805,
      "learning_rate": 7.039566093732701e-06,
      "loss": 0.0323,
      "step": 1359
    },
    {
      "epoch": 1.2897107633949738,
      "grad_norm": 0.5487427414966448,
      "learning_rate": 7.034523833451028e-06,
      "loss": 0.0399,
      "step": 1360
    },
    {
      "epoch": 1.2906590801327644,
      "grad_norm": 0.4529932412914496,
      "learning_rate": 7.029479092334005e-06,
      "loss": 0.0366,
      "step": 1361
    },
    {
      "epoch": 1.2916073968705548,
      "grad_norm": 0.49013485919505534,
      "learning_rate": 7.024431876533035e-06,
      "loss": 0.0296,
      "step": 1362
    },
    {
      "epoch": 1.2925557136083452,
      "grad_norm": 0.37179582929309557,
      "learning_rate": 7.019382192202535e-06,
      "loss": 0.033,
      "step": 1363
    },
    {
      "epoch": 1.2935040303461356,
      "grad_norm": 0.743064802050604,
      "learning_rate": 7.014330045499933e-06,
      "loss": 0.0559,
      "step": 1364
    },
    {
      "epoch": 1.294452347083926,
      "grad_norm": 0.7804393891223528,
      "learning_rate": 7.009275442585661e-06,
      "loss": 0.0695,
      "step": 1365
    },
    {
      "epoch": 1.2954006638217166,
      "grad_norm": 0.6357982901839727,
      "learning_rate": 7.004218389623144e-06,
      "loss": 0.0534,
      "step": 1366
    },
    {
      "epoch": 1.296348980559507,
      "grad_norm": 0.7619430095143129,
      "learning_rate": 6.999158892778799e-06,
      "loss": 0.0879,
      "step": 1367
    },
    {
      "epoch": 1.2972972972972974,
      "grad_norm": 0.45770470276857766,
      "learning_rate": 6.994096958222018e-06,
      "loss": 0.0367,
      "step": 1368
    },
    {
      "epoch": 1.2982456140350878,
      "grad_norm": 0.410119514751568,
      "learning_rate": 6.989032592125167e-06,
      "loss": 0.0365,
      "step": 1369
    },
    {
      "epoch": 1.2991939307728781,
      "grad_norm": 0.6850761718031776,
      "learning_rate": 6.983965800663575e-06,
      "loss": 0.0463,
      "step": 1370
    },
    {
      "epoch": 1.3001422475106685,
      "grad_norm": 0.46513153341224145,
      "learning_rate": 6.978896590015534e-06,
      "loss": 0.0337,
      "step": 1371
    },
    {
      "epoch": 1.301090564248459,
      "grad_norm": 0.34172001512071726,
      "learning_rate": 6.973824966362281e-06,
      "loss": 0.0335,
      "step": 1372
    },
    {
      "epoch": 1.3020388809862493,
      "grad_norm": 0.6484578028683804,
      "learning_rate": 6.968750935887998e-06,
      "loss": 0.051,
      "step": 1373
    },
    {
      "epoch": 1.3029871977240397,
      "grad_norm": 0.5873634371396844,
      "learning_rate": 6.963674504779799e-06,
      "loss": 0.0549,
      "step": 1374
    },
    {
      "epoch": 1.3039355144618303,
      "grad_norm": 0.3675784223470725,
      "learning_rate": 6.958595679227726e-06,
      "loss": 0.0391,
      "step": 1375
    },
    {
      "epoch": 1.3048838311996207,
      "grad_norm": 0.7235773354505487,
      "learning_rate": 6.9535144654247445e-06,
      "loss": 0.0413,
      "step": 1376
    },
    {
      "epoch": 1.305832147937411,
      "grad_norm": 0.39024326330660314,
      "learning_rate": 6.948430869566728e-06,
      "loss": 0.0327,
      "step": 1377
    },
    {
      "epoch": 1.3067804646752015,
      "grad_norm": 0.516621026121203,
      "learning_rate": 6.943344897852455e-06,
      "loss": 0.0573,
      "step": 1378
    },
    {
      "epoch": 1.3077287814129919,
      "grad_norm": 0.6477258197173674,
      "learning_rate": 6.938256556483604e-06,
      "loss": 0.0471,
      "step": 1379
    },
    {
      "epoch": 1.3086770981507825,
      "grad_norm": 0.6102880930185496,
      "learning_rate": 6.933165851664739e-06,
      "loss": 0.0632,
      "step": 1380
    },
    {
      "epoch": 1.3096254148885729,
      "grad_norm": 0.5470569354066884,
      "learning_rate": 6.928072789603306e-06,
      "loss": 0.0555,
      "step": 1381
    },
    {
      "epoch": 1.3105737316263633,
      "grad_norm": 0.3806001449449141,
      "learning_rate": 6.92297737650963e-06,
      "loss": 0.0376,
      "step": 1382
    },
    {
      "epoch": 1.3115220483641536,
      "grad_norm": 0.3967286117945718,
      "learning_rate": 6.9178796185969e-06,
      "loss": 0.032,
      "step": 1383
    },
    {
      "epoch": 1.312470365101944,
      "grad_norm": 0.4801789565687751,
      "learning_rate": 6.912779522081164e-06,
      "loss": 0.0419,
      "step": 1384
    },
    {
      "epoch": 1.3134186818397344,
      "grad_norm": 0.574663042478777,
      "learning_rate": 6.90767709318132e-06,
      "loss": 0.0407,
      "step": 1385
    },
    {
      "epoch": 1.3143669985775248,
      "grad_norm": 0.5147733905582667,
      "learning_rate": 6.902572338119112e-06,
      "loss": 0.0426,
      "step": 1386
    },
    {
      "epoch": 1.3153153153153152,
      "grad_norm": 0.5809011274647721,
      "learning_rate": 6.897465263119123e-06,
      "loss": 0.0324,
      "step": 1387
    },
    {
      "epoch": 1.3162636320531058,
      "grad_norm": 0.5966890883725477,
      "learning_rate": 6.8923558744087596e-06,
      "loss": 0.0382,
      "step": 1388
    },
    {
      "epoch": 1.3172119487908962,
      "grad_norm": 0.3584098045942277,
      "learning_rate": 6.887244178218252e-06,
      "loss": 0.0312,
      "step": 1389
    },
    {
      "epoch": 1.3181602655286866,
      "grad_norm": 0.42233674326286247,
      "learning_rate": 6.882130180780645e-06,
      "loss": 0.0316,
      "step": 1390
    },
    {
      "epoch": 1.319108582266477,
      "grad_norm": 0.6092760045208329,
      "learning_rate": 6.8770138883317895e-06,
      "loss": 0.0424,
      "step": 1391
    },
    {
      "epoch": 1.3200568990042674,
      "grad_norm": 0.4390762035044672,
      "learning_rate": 6.871895307110332e-06,
      "loss": 0.0498,
      "step": 1392
    },
    {
      "epoch": 1.321005215742058,
      "grad_norm": 0.45130718481148246,
      "learning_rate": 6.866774443357714e-06,
      "loss": 0.0408,
      "step": 1393
    },
    {
      "epoch": 1.3219535324798484,
      "grad_norm": 0.8863760397108307,
      "learning_rate": 6.861651303318156e-06,
      "loss": 0.0308,
      "step": 1394
    },
    {
      "epoch": 1.3229018492176388,
      "grad_norm": 0.37381232566509554,
      "learning_rate": 6.856525893238659e-06,
      "loss": 0.0335,
      "step": 1395
    },
    {
      "epoch": 1.3238501659554291,
      "grad_norm": 2.173476279498807,
      "learning_rate": 6.851398219368987e-06,
      "loss": 0.0652,
      "step": 1396
    },
    {
      "epoch": 1.3247984826932195,
      "grad_norm": 0.5112102465327201,
      "learning_rate": 6.846268287961667e-06,
      "loss": 0.0518,
      "step": 1397
    },
    {
      "epoch": 1.32574679943101,
      "grad_norm": 0.44499703462617834,
      "learning_rate": 6.841136105271979e-06,
      "loss": 0.0359,
      "step": 1398
    },
    {
      "epoch": 1.3266951161688003,
      "grad_norm": 0.4541899725717781,
      "learning_rate": 6.8360016775579486e-06,
      "loss": 0.0433,
      "step": 1399
    },
    {
      "epoch": 1.3276434329065907,
      "grad_norm": 0.7733390965492225,
      "learning_rate": 6.830865011080336e-06,
      "loss": 0.0683,
      "step": 1400
    },
    {
      "epoch": 1.328591749644381,
      "grad_norm": 0.4772059867844145,
      "learning_rate": 6.825726112102635e-06,
      "loss": 0.0459,
      "step": 1401
    },
    {
      "epoch": 1.3295400663821717,
      "grad_norm": 0.6433805256237654,
      "learning_rate": 6.820584986891058e-06,
      "loss": 0.0303,
      "step": 1402
    },
    {
      "epoch": 1.330488383119962,
      "grad_norm": 0.34738184879986195,
      "learning_rate": 6.8154416417145356e-06,
      "loss": 0.0283,
      "step": 1403
    },
    {
      "epoch": 1.3314366998577525,
      "grad_norm": 0.8994234854688081,
      "learning_rate": 6.8102960828447026e-06,
      "loss": 0.0495,
      "step": 1404
    },
    {
      "epoch": 1.3323850165955429,
      "grad_norm": 0.745413740002341,
      "learning_rate": 6.805148316555894e-06,
      "loss": 0.0532,
      "step": 1405
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.3870360377631428,
      "learning_rate": 6.799998349125137e-06,
      "loss": 0.0398,
      "step": 1406
    },
    {
      "epoch": 1.3342816500711239,
      "grad_norm": 0.5424025684088033,
      "learning_rate": 6.794846186832144e-06,
      "loss": 0.0464,
      "step": 1407
    },
    {
      "epoch": 1.3352299668089143,
      "grad_norm": 0.39099563297098944,
      "learning_rate": 6.789691835959299e-06,
      "loss": 0.0319,
      "step": 1408
    },
    {
      "epoch": 1.3361782835467046,
      "grad_norm": 0.5899481823910665,
      "learning_rate": 6.784535302791659e-06,
      "loss": 0.0488,
      "step": 1409
    },
    {
      "epoch": 1.337126600284495,
      "grad_norm": 0.504351897768241,
      "learning_rate": 6.779376593616941e-06,
      "loss": 0.0333,
      "step": 1410
    },
    {
      "epoch": 1.3380749170222854,
      "grad_norm": 0.7237802044557335,
      "learning_rate": 6.774215714725516e-06,
      "loss": 0.0592,
      "step": 1411
    },
    {
      "epoch": 1.3390232337600758,
      "grad_norm": 0.4534589627135591,
      "learning_rate": 6.769052672410399e-06,
      "loss": 0.0452,
      "step": 1412
    },
    {
      "epoch": 1.3399715504978662,
      "grad_norm": 0.6267058247818036,
      "learning_rate": 6.763887472967245e-06,
      "loss": 0.0711,
      "step": 1413
    },
    {
      "epoch": 1.3409198672356566,
      "grad_norm": 0.4858351172760667,
      "learning_rate": 6.758720122694336e-06,
      "loss": 0.0425,
      "step": 1414
    },
    {
      "epoch": 1.3418681839734472,
      "grad_norm": 0.432602033472308,
      "learning_rate": 6.75355062789258e-06,
      "loss": 0.0496,
      "step": 1415
    },
    {
      "epoch": 1.3428165007112376,
      "grad_norm": 0.6727456282074611,
      "learning_rate": 6.7483789948654986e-06,
      "loss": 0.074,
      "step": 1416
    },
    {
      "epoch": 1.343764817449028,
      "grad_norm": 0.39807817067818035,
      "learning_rate": 6.743205229919224e-06,
      "loss": 0.0318,
      "step": 1417
    },
    {
      "epoch": 1.3447131341868184,
      "grad_norm": 0.7372478449414445,
      "learning_rate": 6.73802933936248e-06,
      "loss": 0.0593,
      "step": 1418
    },
    {
      "epoch": 1.3456614509246088,
      "grad_norm": 0.3985144055506702,
      "learning_rate": 6.73285132950659e-06,
      "loss": 0.0358,
      "step": 1419
    },
    {
      "epoch": 1.3466097676623994,
      "grad_norm": 0.5951692184417887,
      "learning_rate": 6.727671206665458e-06,
      "loss": 0.0464,
      "step": 1420
    },
    {
      "epoch": 1.3475580844001898,
      "grad_norm": 0.38420866497778394,
      "learning_rate": 6.722488977155567e-06,
      "loss": 0.0284,
      "step": 1421
    },
    {
      "epoch": 1.3485064011379801,
      "grad_norm": 0.7464916944500526,
      "learning_rate": 6.717304647295965e-06,
      "loss": 0.0574,
      "step": 1422
    },
    {
      "epoch": 1.3494547178757705,
      "grad_norm": 0.5693062216202691,
      "learning_rate": 6.712118223408264e-06,
      "loss": 0.0392,
      "step": 1423
    },
    {
      "epoch": 1.350403034613561,
      "grad_norm": 0.4592788558938253,
      "learning_rate": 6.7069297118166295e-06,
      "loss": 0.0347,
      "step": 1424
    },
    {
      "epoch": 1.3513513513513513,
      "grad_norm": 0.5175822786697956,
      "learning_rate": 6.701739118847771e-06,
      "loss": 0.0536,
      "step": 1425
    },
    {
      "epoch": 1.3522996680891417,
      "grad_norm": 0.6912802037999823,
      "learning_rate": 6.696546450830937e-06,
      "loss": 0.0536,
      "step": 1426
    },
    {
      "epoch": 1.353247984826932,
      "grad_norm": 0.40537336423409387,
      "learning_rate": 6.691351714097906e-06,
      "loss": 0.0365,
      "step": 1427
    },
    {
      "epoch": 1.3541963015647225,
      "grad_norm": 0.532506588444211,
      "learning_rate": 6.6861549149829785e-06,
      "loss": 0.0463,
      "step": 1428
    },
    {
      "epoch": 1.355144618302513,
      "grad_norm": 0.5449727239357885,
      "learning_rate": 6.680956059822971e-06,
      "loss": 0.0445,
      "step": 1429
    },
    {
      "epoch": 1.3560929350403035,
      "grad_norm": 0.35162216761507775,
      "learning_rate": 6.675755154957208e-06,
      "loss": 0.03,
      "step": 1430
    },
    {
      "epoch": 1.3570412517780939,
      "grad_norm": 0.5723964216096473,
      "learning_rate": 6.670552206727507e-06,
      "loss": 0.0458,
      "step": 1431
    },
    {
      "epoch": 1.3579895685158843,
      "grad_norm": 0.6308418866530405,
      "learning_rate": 6.665347221478188e-06,
      "loss": 0.057,
      "step": 1432
    },
    {
      "epoch": 1.3589378852536746,
      "grad_norm": 0.498253296219615,
      "learning_rate": 6.660140205556046e-06,
      "loss": 0.0383,
      "step": 1433
    },
    {
      "epoch": 1.3598862019914653,
      "grad_norm": 0.40110848050156933,
      "learning_rate": 6.6549311653103544e-06,
      "loss": 0.0462,
      "step": 1434
    },
    {
      "epoch": 1.3608345187292556,
      "grad_norm": 0.3808306857794104,
      "learning_rate": 6.649720107092854e-06,
      "loss": 0.0368,
      "step": 1435
    },
    {
      "epoch": 1.361782835467046,
      "grad_norm": 0.5026216712918046,
      "learning_rate": 6.644507037257753e-06,
      "loss": 0.0394,
      "step": 1436
    },
    {
      "epoch": 1.3627311522048364,
      "grad_norm": 0.5063922606607929,
      "learning_rate": 6.639291962161702e-06,
      "loss": 0.0354,
      "step": 1437
    },
    {
      "epoch": 1.3636794689426268,
      "grad_norm": 0.34810512153075385,
      "learning_rate": 6.634074888163805e-06,
      "loss": 0.0253,
      "step": 1438
    },
    {
      "epoch": 1.3646277856804172,
      "grad_norm": 0.8525521272225376,
      "learning_rate": 6.628855821625601e-06,
      "loss": 0.0662,
      "step": 1439
    },
    {
      "epoch": 1.3655761024182076,
      "grad_norm": 0.4145939665119975,
      "learning_rate": 6.6236347689110546e-06,
      "loss": 0.0296,
      "step": 1440
    },
    {
      "epoch": 1.366524419155998,
      "grad_norm": 0.4775660065145518,
      "learning_rate": 6.618411736386558e-06,
      "loss": 0.0523,
      "step": 1441
    },
    {
      "epoch": 1.3674727358937886,
      "grad_norm": 0.37369507508155264,
      "learning_rate": 6.613186730420917e-06,
      "loss": 0.0365,
      "step": 1442
    },
    {
      "epoch": 1.368421052631579,
      "grad_norm": 0.40495701137032075,
      "learning_rate": 6.607959757385338e-06,
      "loss": 0.0362,
      "step": 1443
    },
    {
      "epoch": 1.3693693693693694,
      "grad_norm": 0.36575125377415557,
      "learning_rate": 6.602730823653436e-06,
      "loss": 0.0338,
      "step": 1444
    },
    {
      "epoch": 1.3703176861071598,
      "grad_norm": 0.6249163504364997,
      "learning_rate": 6.597499935601204e-06,
      "loss": 0.051,
      "step": 1445
    },
    {
      "epoch": 1.3712660028449501,
      "grad_norm": 0.582172833981279,
      "learning_rate": 6.592267099607029e-06,
      "loss": 0.0438,
      "step": 1446
    },
    {
      "epoch": 1.3722143195827408,
      "grad_norm": 0.3237159800921359,
      "learning_rate": 6.587032322051667e-06,
      "loss": 0.0241,
      "step": 1447
    },
    {
      "epoch": 1.3731626363205311,
      "grad_norm": 0.39835957752966755,
      "learning_rate": 6.581795609318247e-06,
      "loss": 0.036,
      "step": 1448
    },
    {
      "epoch": 1.3741109530583215,
      "grad_norm": 0.5547572067644376,
      "learning_rate": 6.57655696779225e-06,
      "loss": 0.0371,
      "step": 1449
    },
    {
      "epoch": 1.375059269796112,
      "grad_norm": 0.5290408595276358,
      "learning_rate": 6.571316403861518e-06,
      "loss": 0.066,
      "step": 1450
    },
    {
      "epoch": 1.3760075865339023,
      "grad_norm": 0.5333270731702193,
      "learning_rate": 6.566073923916226e-06,
      "loss": 0.0483,
      "step": 1451
    },
    {
      "epoch": 1.3769559032716927,
      "grad_norm": 0.47040112700666353,
      "learning_rate": 6.5608295343488985e-06,
      "loss": 0.0424,
      "step": 1452
    },
    {
      "epoch": 1.377904220009483,
      "grad_norm": 0.38901166465860154,
      "learning_rate": 6.555583241554376e-06,
      "loss": 0.0432,
      "step": 1453
    },
    {
      "epoch": 1.3788525367472735,
      "grad_norm": 0.49414149390465306,
      "learning_rate": 6.55033505192983e-06,
      "loss": 0.0407,
      "step": 1454
    },
    {
      "epoch": 1.379800853485064,
      "grad_norm": 0.45253883891283475,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.0388,
      "step": 1455
    },
    {
      "epoch": 1.3807491702228545,
      "grad_norm": 0.5302487098048316,
      "learning_rate": 6.539833007790885e-06,
      "loss": 0.0295,
      "step": 1456
    },
    {
      "epoch": 1.3816974869606449,
      "grad_norm": 0.37787261318074306,
      "learning_rate": 6.534579166082355e-06,
      "loss": 0.0286,
      "step": 1457
    },
    {
      "epoch": 1.3826458036984353,
      "grad_norm": 0.4747311054007047,
      "learning_rate": 6.529323453155516e-06,
      "loss": 0.0338,
      "step": 1458
    },
    {
      "epoch": 1.3835941204362256,
      "grad_norm": 1.0261775992168287,
      "learning_rate": 6.524065875419025e-06,
      "loss": 0.043,
      "step": 1459
    },
    {
      "epoch": 1.384542437174016,
      "grad_norm": 0.5317475686310853,
      "learning_rate": 6.5188064392838114e-06,
      "loss": 0.0533,
      "step": 1460
    },
    {
      "epoch": 1.3854907539118066,
      "grad_norm": 0.5479284515608536,
      "learning_rate": 6.5135451511630636e-06,
      "loss": 0.058,
      "step": 1461
    },
    {
      "epoch": 1.386439070649597,
      "grad_norm": 0.6932046763335258,
      "learning_rate": 6.5082820174722364e-06,
      "loss": 0.0506,
      "step": 1462
    },
    {
      "epoch": 1.3873873873873874,
      "grad_norm": 0.3763860463333079,
      "learning_rate": 6.503017044629029e-06,
      "loss": 0.0291,
      "step": 1463
    },
    {
      "epoch": 1.3883357041251778,
      "grad_norm": 0.5382489774529615,
      "learning_rate": 6.4977502390533905e-06,
      "loss": 0.0394,
      "step": 1464
    },
    {
      "epoch": 1.3892840208629682,
      "grad_norm": 0.6889518317604814,
      "learning_rate": 6.4924816071675e-06,
      "loss": 0.0698,
      "step": 1465
    },
    {
      "epoch": 1.3902323376007586,
      "grad_norm": 0.6495562092445605,
      "learning_rate": 6.487211155395762e-06,
      "loss": 0.0516,
      "step": 1466
    },
    {
      "epoch": 1.391180654338549,
      "grad_norm": 0.6308551102529311,
      "learning_rate": 6.4819388901648e-06,
      "loss": 0.0491,
      "step": 1467
    },
    {
      "epoch": 1.3921289710763394,
      "grad_norm": 0.5361583528081098,
      "learning_rate": 6.476664817903457e-06,
      "loss": 0.0467,
      "step": 1468
    },
    {
      "epoch": 1.39307728781413,
      "grad_norm": 0.36452515652974693,
      "learning_rate": 6.471388945042769e-06,
      "loss": 0.0337,
      "step": 1469
    },
    {
      "epoch": 1.3940256045519204,
      "grad_norm": 0.5709516666140196,
      "learning_rate": 6.466111278015973e-06,
      "loss": 0.0356,
      "step": 1470
    },
    {
      "epoch": 1.3949739212897108,
      "grad_norm": 0.6635702342274311,
      "learning_rate": 6.460831823258492e-06,
      "loss": 0.052,
      "step": 1471
    },
    {
      "epoch": 1.3959222380275011,
      "grad_norm": 0.6089384298241021,
      "learning_rate": 6.455550587207931e-06,
      "loss": 0.0464,
      "step": 1472
    },
    {
      "epoch": 1.3968705547652915,
      "grad_norm": 0.5865814694576227,
      "learning_rate": 6.450267576304065e-06,
      "loss": 0.0493,
      "step": 1473
    },
    {
      "epoch": 1.3978188715030821,
      "grad_norm": 0.4169326925237183,
      "learning_rate": 6.444982796988835e-06,
      "loss": 0.0309,
      "step": 1474
    },
    {
      "epoch": 1.3987671882408725,
      "grad_norm": 0.4907017898205256,
      "learning_rate": 6.439696255706334e-06,
      "loss": 0.0267,
      "step": 1475
    },
    {
      "epoch": 1.399715504978663,
      "grad_norm": 0.6149659771951538,
      "learning_rate": 6.434407958902809e-06,
      "loss": 0.0461,
      "step": 1476
    },
    {
      "epoch": 1.4006638217164533,
      "grad_norm": 0.4695061046804246,
      "learning_rate": 6.429117913026647e-06,
      "loss": 0.0511,
      "step": 1477
    },
    {
      "epoch": 1.4016121384542437,
      "grad_norm": 0.5455301875944718,
      "learning_rate": 6.423826124528363e-06,
      "loss": 0.0413,
      "step": 1478
    },
    {
      "epoch": 1.402560455192034,
      "grad_norm": 0.40051594288158987,
      "learning_rate": 6.418532599860602e-06,
      "loss": 0.0382,
      "step": 1479
    },
    {
      "epoch": 1.4035087719298245,
      "grad_norm": 0.3913732454226365,
      "learning_rate": 6.413237345478121e-06,
      "loss": 0.0424,
      "step": 1480
    },
    {
      "epoch": 1.4044570886676149,
      "grad_norm": 0.42177204727179535,
      "learning_rate": 6.407940367837794e-06,
      "loss": 0.0341,
      "step": 1481
    },
    {
      "epoch": 1.4054054054054055,
      "grad_norm": 0.5927648854025513,
      "learning_rate": 6.402641673398589e-06,
      "loss": 0.0457,
      "step": 1482
    },
    {
      "epoch": 1.4063537221431959,
      "grad_norm": 0.6649752510649648,
      "learning_rate": 6.39734126862157e-06,
      "loss": 0.0399,
      "step": 1483
    },
    {
      "epoch": 1.4073020388809863,
      "grad_norm": 0.6602312989482694,
      "learning_rate": 6.392039159969884e-06,
      "loss": 0.0389,
      "step": 1484
    },
    {
      "epoch": 1.4082503556187767,
      "grad_norm": 0.4538930406178064,
      "learning_rate": 6.386735353908762e-06,
      "loss": 0.0295,
      "step": 1485
    },
    {
      "epoch": 1.409198672356567,
      "grad_norm": 0.5026530505058707,
      "learning_rate": 6.381429856905499e-06,
      "loss": 0.0391,
      "step": 1486
    },
    {
      "epoch": 1.4101469890943577,
      "grad_norm": 0.5396295314825587,
      "learning_rate": 6.3761226754294525e-06,
      "loss": 0.0524,
      "step": 1487
    },
    {
      "epoch": 1.411095305832148,
      "grad_norm": 0.700524696737488,
      "learning_rate": 6.370813815952039e-06,
      "loss": 0.0456,
      "step": 1488
    },
    {
      "epoch": 1.4120436225699384,
      "grad_norm": 0.5570830545575999,
      "learning_rate": 6.365503284946713e-06,
      "loss": 0.0376,
      "step": 1489
    },
    {
      "epoch": 1.4129919393077288,
      "grad_norm": 0.5029568178955325,
      "learning_rate": 6.360191088888975e-06,
      "loss": 0.0441,
      "step": 1490
    },
    {
      "epoch": 1.4139402560455192,
      "grad_norm": 0.39143841524057965,
      "learning_rate": 6.35487723425635e-06,
      "loss": 0.0322,
      "step": 1491
    },
    {
      "epoch": 1.4148885727833096,
      "grad_norm": 0.5179251130538431,
      "learning_rate": 6.349561727528388e-06,
      "loss": 0.039,
      "step": 1492
    },
    {
      "epoch": 1.4158368895211,
      "grad_norm": 0.4971363492885207,
      "learning_rate": 6.344244575186655e-06,
      "loss": 0.0389,
      "step": 1493
    },
    {
      "epoch": 1.4167852062588904,
      "grad_norm": 0.4650371165745209,
      "learning_rate": 6.338925783714721e-06,
      "loss": 0.0354,
      "step": 1494
    },
    {
      "epoch": 1.4177335229966808,
      "grad_norm": 0.5910877479719807,
      "learning_rate": 6.333605359598154e-06,
      "loss": 0.0458,
      "step": 1495
    },
    {
      "epoch": 1.4186818397344714,
      "grad_norm": 0.4704211557886303,
      "learning_rate": 6.328283309324516e-06,
      "loss": 0.0375,
      "step": 1496
    },
    {
      "epoch": 1.4196301564722618,
      "grad_norm": 0.5026935559926589,
      "learning_rate": 6.32295963938335e-06,
      "loss": 0.0368,
      "step": 1497
    },
    {
      "epoch": 1.4205784732100522,
      "grad_norm": 0.7197271409887576,
      "learning_rate": 6.317634356266175e-06,
      "loss": 0.0625,
      "step": 1498
    },
    {
      "epoch": 1.4215267899478425,
      "grad_norm": 0.5369735581060212,
      "learning_rate": 6.312307466466477e-06,
      "loss": 0.035,
      "step": 1499
    },
    {
      "epoch": 1.422475106685633,
      "grad_norm": 0.396321033510949,
      "learning_rate": 6.306978976479695e-06,
      "loss": 0.0294,
      "step": 1500
    },
    {
      "epoch": 1.4234234234234235,
      "grad_norm": 0.3409142303173711,
      "learning_rate": 6.30164889280323e-06,
      "loss": 0.035,
      "step": 1501
    },
    {
      "epoch": 1.424371740161214,
      "grad_norm": 0.49779800032590527,
      "learning_rate": 6.296317221936421e-06,
      "loss": 0.0344,
      "step": 1502
    },
    {
      "epoch": 1.4253200568990043,
      "grad_norm": 0.398334192892684,
      "learning_rate": 6.290983970380539e-06,
      "loss": 0.0298,
      "step": 1503
    },
    {
      "epoch": 1.4262683736367947,
      "grad_norm": 2.7922441337396253,
      "learning_rate": 6.28564914463879e-06,
      "loss": 0.0517,
      "step": 1504
    },
    {
      "epoch": 1.427216690374585,
      "grad_norm": 0.4840208519427491,
      "learning_rate": 6.280312751216291e-06,
      "loss": 0.0337,
      "step": 1505
    },
    {
      "epoch": 1.4281650071123755,
      "grad_norm": 0.5619043194162205,
      "learning_rate": 6.274974796620078e-06,
      "loss": 0.0537,
      "step": 1506
    },
    {
      "epoch": 1.4291133238501659,
      "grad_norm": 0.5122598193956247,
      "learning_rate": 6.269635287359086e-06,
      "loss": 0.0424,
      "step": 1507
    },
    {
      "epoch": 1.4300616405879563,
      "grad_norm": 0.9600984611912299,
      "learning_rate": 6.264294229944146e-06,
      "loss": 0.0589,
      "step": 1508
    },
    {
      "epoch": 1.4310099573257469,
      "grad_norm": 0.8407809045144361,
      "learning_rate": 6.258951630887982e-06,
      "loss": 0.0372,
      "step": 1509
    },
    {
      "epoch": 1.4319582740635373,
      "grad_norm": 0.7744939127361691,
      "learning_rate": 6.253607496705191e-06,
      "loss": 0.0444,
      "step": 1510
    },
    {
      "epoch": 1.4329065908013277,
      "grad_norm": 0.7703533886302798,
      "learning_rate": 6.248261833912243e-06,
      "loss": 0.0491,
      "step": 1511
    },
    {
      "epoch": 1.433854907539118,
      "grad_norm": 0.5182968502177208,
      "learning_rate": 6.242914649027476e-06,
      "loss": 0.0349,
      "step": 1512
    },
    {
      "epoch": 1.4348032242769084,
      "grad_norm": 0.5575121095897854,
      "learning_rate": 6.237565948571082e-06,
      "loss": 0.0436,
      "step": 1513
    },
    {
      "epoch": 1.435751541014699,
      "grad_norm": 0.5187081289232499,
      "learning_rate": 6.232215739065102e-06,
      "loss": 0.0414,
      "step": 1514
    },
    {
      "epoch": 1.4366998577524894,
      "grad_norm": 0.43186737340559983,
      "learning_rate": 6.226864027033413e-06,
      "loss": 0.0278,
      "step": 1515
    },
    {
      "epoch": 1.4376481744902798,
      "grad_norm": 0.44892769483591366,
      "learning_rate": 6.221510819001725e-06,
      "loss": 0.0297,
      "step": 1516
    },
    {
      "epoch": 1.4385964912280702,
      "grad_norm": 0.5269062893506762,
      "learning_rate": 6.216156121497579e-06,
      "loss": 0.0514,
      "step": 1517
    },
    {
      "epoch": 1.4395448079658606,
      "grad_norm": 0.6625936787851716,
      "learning_rate": 6.2107999410503235e-06,
      "loss": 0.0419,
      "step": 1518
    },
    {
      "epoch": 1.440493124703651,
      "grad_norm": 0.5592530362043953,
      "learning_rate": 6.205442284191122e-06,
      "loss": 0.0432,
      "step": 1519
    },
    {
      "epoch": 1.4414414414414414,
      "grad_norm": 0.47852425989622077,
      "learning_rate": 6.200083157452934e-06,
      "loss": 0.031,
      "step": 1520
    },
    {
      "epoch": 1.4423897581792318,
      "grad_norm": 0.5811912176295393,
      "learning_rate": 6.194722567370511e-06,
      "loss": 0.057,
      "step": 1521
    },
    {
      "epoch": 1.4433380749170222,
      "grad_norm": 0.564889130546722,
      "learning_rate": 6.189360520480394e-06,
      "loss": 0.0423,
      "step": 1522
    },
    {
      "epoch": 1.4442863916548128,
      "grad_norm": 0.7421168415388308,
      "learning_rate": 6.183997023320894e-06,
      "loss": 0.058,
      "step": 1523
    },
    {
      "epoch": 1.4452347083926032,
      "grad_norm": 0.4925903110099146,
      "learning_rate": 6.178632082432093e-06,
      "loss": 0.043,
      "step": 1524
    },
    {
      "epoch": 1.4461830251303935,
      "grad_norm": 0.4022649394814297,
      "learning_rate": 6.173265704355838e-06,
      "loss": 0.0319,
      "step": 1525
    },
    {
      "epoch": 1.447131341868184,
      "grad_norm": 0.46419553742343383,
      "learning_rate": 6.16789789563572e-06,
      "loss": 0.0511,
      "step": 1526
    },
    {
      "epoch": 1.4480796586059743,
      "grad_norm": 0.686004067534123,
      "learning_rate": 6.16252866281708e-06,
      "loss": 0.0353,
      "step": 1527
    },
    {
      "epoch": 1.449027975343765,
      "grad_norm": 0.5485413747656448,
      "learning_rate": 6.1571580124469935e-06,
      "loss": 0.0387,
      "step": 1528
    },
    {
      "epoch": 1.4499762920815553,
      "grad_norm": 1.0798253002470137,
      "learning_rate": 6.151785951074266e-06,
      "loss": 0.0378,
      "step": 1529
    },
    {
      "epoch": 1.4509246088193457,
      "grad_norm": 0.6205350073393234,
      "learning_rate": 6.146412485249424e-06,
      "loss": 0.0441,
      "step": 1530
    },
    {
      "epoch": 1.451872925557136,
      "grad_norm": 0.5359792529386427,
      "learning_rate": 6.141037621524704e-06,
      "loss": 0.0457,
      "step": 1531
    },
    {
      "epoch": 1.4528212422949265,
      "grad_norm": 0.6496519633454682,
      "learning_rate": 6.1356613664540455e-06,
      "loss": 0.0322,
      "step": 1532
    },
    {
      "epoch": 1.4537695590327169,
      "grad_norm": 0.500829293631121,
      "learning_rate": 6.1302837265930925e-06,
      "loss": 0.0318,
      "step": 1533
    },
    {
      "epoch": 1.4547178757705073,
      "grad_norm": 0.6387522771798803,
      "learning_rate": 6.124904708499169e-06,
      "loss": 0.0375,
      "step": 1534
    },
    {
      "epoch": 1.4556661925082977,
      "grad_norm": 0.6065572240886973,
      "learning_rate": 6.119524318731286e-06,
      "loss": 0.0578,
      "step": 1535
    },
    {
      "epoch": 1.4566145092460883,
      "grad_norm": 0.4804845250123355,
      "learning_rate": 6.114142563850122e-06,
      "loss": 0.0464,
      "step": 1536
    },
    {
      "epoch": 1.4575628259838787,
      "grad_norm": 0.5386846320310618,
      "learning_rate": 6.1087594504180226e-06,
      "loss": 0.0354,
      "step": 1537
    },
    {
      "epoch": 1.458511142721669,
      "grad_norm": 0.5814430131886619,
      "learning_rate": 6.1033749849989896e-06,
      "loss": 0.0528,
      "step": 1538
    },
    {
      "epoch": 1.4594594594594594,
      "grad_norm": 0.41608042979081394,
      "learning_rate": 6.097989174158674e-06,
      "loss": 0.0371,
      "step": 1539
    },
    {
      "epoch": 1.4604077761972498,
      "grad_norm": 0.42647526096873056,
      "learning_rate": 6.092602024464364e-06,
      "loss": 0.041,
      "step": 1540
    },
    {
      "epoch": 1.4613560929350404,
      "grad_norm": 0.7880765483408534,
      "learning_rate": 6.087213542484987e-06,
      "loss": 0.0285,
      "step": 1541
    },
    {
      "epoch": 1.4623044096728308,
      "grad_norm": 0.3794551189159985,
      "learning_rate": 6.0818237347910905e-06,
      "loss": 0.0348,
      "step": 1542
    },
    {
      "epoch": 1.4632527264106212,
      "grad_norm": 0.5824672410897602,
      "learning_rate": 6.076432607954837e-06,
      "loss": 0.0473,
      "step": 1543
    },
    {
      "epoch": 1.4642010431484116,
      "grad_norm": 0.7642250674769905,
      "learning_rate": 6.071040168549999e-06,
      "loss": 0.055,
      "step": 1544
    },
    {
      "epoch": 1.465149359886202,
      "grad_norm": 0.4383923165951326,
      "learning_rate": 6.0656464231519505e-06,
      "loss": 0.0435,
      "step": 1545
    },
    {
      "epoch": 1.4660976766239924,
      "grad_norm": 0.5766848636422996,
      "learning_rate": 6.060251378337662e-06,
      "loss": 0.0472,
      "step": 1546
    },
    {
      "epoch": 1.4670459933617828,
      "grad_norm": 0.4669671289773233,
      "learning_rate": 6.05485504068568e-06,
      "loss": 0.0414,
      "step": 1547
    },
    {
      "epoch": 1.4679943100995732,
      "grad_norm": 0.5655404897789605,
      "learning_rate": 6.049457416776131e-06,
      "loss": 0.04,
      "step": 1548
    },
    {
      "epoch": 1.4689426268373635,
      "grad_norm": 0.5365501006997421,
      "learning_rate": 6.044058513190711e-06,
      "loss": 0.0383,
      "step": 1549
    },
    {
      "epoch": 1.4698909435751542,
      "grad_norm": 0.48431123525495323,
      "learning_rate": 6.038658336512677e-06,
      "loss": 0.0548,
      "step": 1550
    },
    {
      "epoch": 1.4708392603129445,
      "grad_norm": 0.5554437642357153,
      "learning_rate": 6.0332568933268375e-06,
      "loss": 0.0417,
      "step": 1551
    },
    {
      "epoch": 1.471787577050735,
      "grad_norm": 1.7635981147318227,
      "learning_rate": 6.0278541902195446e-06,
      "loss": 0.0515,
      "step": 1552
    },
    {
      "epoch": 1.4727358937885253,
      "grad_norm": 0.7232935188306358,
      "learning_rate": 6.022450233778685e-06,
      "loss": 0.0657,
      "step": 1553
    },
    {
      "epoch": 1.4736842105263157,
      "grad_norm": 0.5072524556149093,
      "learning_rate": 6.017045030593679e-06,
      "loss": 0.0438,
      "step": 1554
    },
    {
      "epoch": 1.4746325272641063,
      "grad_norm": 0.4631383186501888,
      "learning_rate": 6.011638587255461e-06,
      "loss": 0.0329,
      "step": 1555
    },
    {
      "epoch": 1.4755808440018967,
      "grad_norm": 0.45366075659990895,
      "learning_rate": 6.006230910356483e-06,
      "loss": 0.0352,
      "step": 1556
    },
    {
      "epoch": 1.476529160739687,
      "grad_norm": 0.5767111918678757,
      "learning_rate": 6.000822006490696e-06,
      "loss": 0.0375,
      "step": 1557
    },
    {
      "epoch": 1.4774774774774775,
      "grad_norm": 0.4071796252251615,
      "learning_rate": 5.995411882253554e-06,
      "loss": 0.0331,
      "step": 1558
    },
    {
      "epoch": 1.4784257942152679,
      "grad_norm": 0.43238641694345853,
      "learning_rate": 5.990000544241993e-06,
      "loss": 0.0463,
      "step": 1559
    },
    {
      "epoch": 1.4793741109530583,
      "grad_norm": 0.6798403742092483,
      "learning_rate": 5.984587999054428e-06,
      "loss": 0.0365,
      "step": 1560
    },
    {
      "epoch": 1.4803224276908487,
      "grad_norm": 0.5374744532934378,
      "learning_rate": 5.979174253290752e-06,
      "loss": 0.0445,
      "step": 1561
    },
    {
      "epoch": 1.481270744428639,
      "grad_norm": 0.36196300041262275,
      "learning_rate": 5.973759313552318e-06,
      "loss": 0.0303,
      "step": 1562
    },
    {
      "epoch": 1.4822190611664297,
      "grad_norm": 0.5642959313665694,
      "learning_rate": 5.9683431864419384e-06,
      "loss": 0.0449,
      "step": 1563
    },
    {
      "epoch": 1.48316737790422,
      "grad_norm": 0.5061601710793207,
      "learning_rate": 5.9629258785638675e-06,
      "loss": 0.0327,
      "step": 1564
    },
    {
      "epoch": 1.4841156946420104,
      "grad_norm": 0.44567373677114336,
      "learning_rate": 5.957507396523803e-06,
      "loss": 0.0299,
      "step": 1565
    },
    {
      "epoch": 1.4850640113798008,
      "grad_norm": 1.7557947841187413,
      "learning_rate": 5.952087746928875e-06,
      "loss": 0.041,
      "step": 1566
    },
    {
      "epoch": 1.4860123281175912,
      "grad_norm": 0.4668457994449059,
      "learning_rate": 5.9466669363876375e-06,
      "loss": 0.0344,
      "step": 1567
    },
    {
      "epoch": 1.4869606448553818,
      "grad_norm": 0.4261254948879091,
      "learning_rate": 5.9412449715100576e-06,
      "loss": 0.0303,
      "step": 1568
    },
    {
      "epoch": 1.4879089615931722,
      "grad_norm": 0.42754498865715357,
      "learning_rate": 5.93582185890751e-06,
      "loss": 0.0261,
      "step": 1569
    },
    {
      "epoch": 1.4888572783309626,
      "grad_norm": 0.4906574698227752,
      "learning_rate": 5.930397605192773e-06,
      "loss": 0.0444,
      "step": 1570
    },
    {
      "epoch": 1.489805595068753,
      "grad_norm": 0.7067845216764639,
      "learning_rate": 5.924972216980013e-06,
      "loss": 0.042,
      "step": 1571
    },
    {
      "epoch": 1.4907539118065434,
      "grad_norm": 0.48961667245950796,
      "learning_rate": 5.919545700884779e-06,
      "loss": 0.0425,
      "step": 1572
    },
    {
      "epoch": 1.4917022285443338,
      "grad_norm": 1.0609050385233834,
      "learning_rate": 5.914118063523996e-06,
      "loss": 0.0449,
      "step": 1573
    },
    {
      "epoch": 1.4926505452821242,
      "grad_norm": 1.0807634420097412,
      "learning_rate": 5.9086893115159605e-06,
      "loss": 0.0502,
      "step": 1574
    },
    {
      "epoch": 1.4935988620199145,
      "grad_norm": 0.6416643508613592,
      "learning_rate": 5.903259451480321e-06,
      "loss": 0.0441,
      "step": 1575
    },
    {
      "epoch": 1.4945471787577052,
      "grad_norm": 0.6277434629874132,
      "learning_rate": 5.897828490038082e-06,
      "loss": 0.0713,
      "step": 1576
    },
    {
      "epoch": 1.4954954954954955,
      "grad_norm": 0.33767932825829644,
      "learning_rate": 5.8923964338115895e-06,
      "loss": 0.0347,
      "step": 1577
    },
    {
      "epoch": 1.496443812233286,
      "grad_norm": 0.4477567855475462,
      "learning_rate": 5.886963289424524e-06,
      "loss": 0.0364,
      "step": 1578
    },
    {
      "epoch": 1.4973921289710763,
      "grad_norm": 0.41594278967554993,
      "learning_rate": 5.881529063501896e-06,
      "loss": 0.0338,
      "step": 1579
    },
    {
      "epoch": 1.4983404457088667,
      "grad_norm": 0.35754135982690183,
      "learning_rate": 5.8760937626700306e-06,
      "loss": 0.0287,
      "step": 1580
    },
    {
      "epoch": 1.4992887624466573,
      "grad_norm": 0.4123079201310612,
      "learning_rate": 5.8706573935565645e-06,
      "loss": 0.0347,
      "step": 1581
    },
    {
      "epoch": 1.5002370791844477,
      "grad_norm": 0.6029428312058318,
      "learning_rate": 5.865219962790438e-06,
      "loss": 0.051,
      "step": 1582
    },
    {
      "epoch": 1.501185395922238,
      "grad_norm": 0.5215535313309333,
      "learning_rate": 5.859781477001887e-06,
      "loss": 0.0502,
      "step": 1583
    },
    {
      "epoch": 1.5021337126600285,
      "grad_norm": 0.5200577257819287,
      "learning_rate": 5.8543419428224325e-06,
      "loss": 0.04,
      "step": 1584
    },
    {
      "epoch": 1.5030820293978189,
      "grad_norm": 0.3586196172148145,
      "learning_rate": 5.8489013668848726e-06,
      "loss": 0.0217,
      "step": 1585
    },
    {
      "epoch": 1.5040303461356093,
      "grad_norm": 0.6761618716488011,
      "learning_rate": 5.843459755823279e-06,
      "loss": 0.0628,
      "step": 1586
    },
    {
      "epoch": 1.5049786628733997,
      "grad_norm": 0.49941615403305933,
      "learning_rate": 5.8380171162729814e-06,
      "loss": 0.0376,
      "step": 1587
    },
    {
      "epoch": 1.50592697961119,
      "grad_norm": 0.7013305412374657,
      "learning_rate": 5.832573454870567e-06,
      "loss": 0.0519,
      "step": 1588
    },
    {
      "epoch": 1.5068752963489804,
      "grad_norm": 0.36446558452178207,
      "learning_rate": 5.827128778253867e-06,
      "loss": 0.0315,
      "step": 1589
    },
    {
      "epoch": 1.5078236130867708,
      "grad_norm": 0.6059271506224598,
      "learning_rate": 5.821683093061956e-06,
      "loss": 0.0588,
      "step": 1590
    },
    {
      "epoch": 1.5087719298245614,
      "grad_norm": 0.5681480190393359,
      "learning_rate": 5.81623640593513e-06,
      "loss": 0.0471,
      "step": 1591
    },
    {
      "epoch": 1.5097202465623518,
      "grad_norm": 0.5550547812099469,
      "learning_rate": 5.810788723514908e-06,
      "loss": 0.0535,
      "step": 1592
    },
    {
      "epoch": 1.5106685633001422,
      "grad_norm": 0.49253764363091057,
      "learning_rate": 5.805340052444028e-06,
      "loss": 0.0428,
      "step": 1593
    },
    {
      "epoch": 1.5116168800379328,
      "grad_norm": 0.5766442929322906,
      "learning_rate": 5.799890399366434e-06,
      "loss": 0.0399,
      "step": 1594
    },
    {
      "epoch": 1.5125651967757232,
      "grad_norm": 0.34949487700262954,
      "learning_rate": 5.79443977092726e-06,
      "loss": 0.0278,
      "step": 1595
    },
    {
      "epoch": 1.5135135135135136,
      "grad_norm": 0.659958933278198,
      "learning_rate": 5.788988173772835e-06,
      "loss": 0.047,
      "step": 1596
    },
    {
      "epoch": 1.514461830251304,
      "grad_norm": 0.5597846667352361,
      "learning_rate": 5.783535614550666e-06,
      "loss": 0.0404,
      "step": 1597
    },
    {
      "epoch": 1.5154101469890944,
      "grad_norm": 0.7866842188486388,
      "learning_rate": 5.778082099909436e-06,
      "loss": 0.0541,
      "step": 1598
    },
    {
      "epoch": 1.5163584637268848,
      "grad_norm": 0.3147662026332043,
      "learning_rate": 5.772627636498992e-06,
      "loss": 0.0308,
      "step": 1599
    },
    {
      "epoch": 1.5173067804646752,
      "grad_norm": 0.5584398566971551,
      "learning_rate": 5.76717223097034e-06,
      "loss": 0.0395,
      "step": 1600
    },
    {
      "epoch": 1.5182550972024655,
      "grad_norm": 0.4132888410051328,
      "learning_rate": 5.7617158899756276e-06,
      "loss": 0.034,
      "step": 1601
    },
    {
      "epoch": 1.519203413940256,
      "grad_norm": 0.37539181864065124,
      "learning_rate": 5.756258620168152e-06,
      "loss": 0.0416,
      "step": 1602
    },
    {
      "epoch": 1.5201517306780463,
      "grad_norm": 0.7346309696726602,
      "learning_rate": 5.750800428202338e-06,
      "loss": 0.0373,
      "step": 1603
    },
    {
      "epoch": 1.521100047415837,
      "grad_norm": 0.620382314193856,
      "learning_rate": 5.745341320733735e-06,
      "loss": 0.0331,
      "step": 1604
    },
    {
      "epoch": 1.5220483641536273,
      "grad_norm": 0.5684312369471517,
      "learning_rate": 5.739881304419009e-06,
      "loss": 0.0331,
      "step": 1605
    },
    {
      "epoch": 1.5229966808914177,
      "grad_norm": 0.9180566528310432,
      "learning_rate": 5.734420385915937e-06,
      "loss": 0.0396,
      "step": 1606
    },
    {
      "epoch": 1.5239449976292083,
      "grad_norm": 0.4119447176442743,
      "learning_rate": 5.728958571883393e-06,
      "loss": 0.0347,
      "step": 1607
    },
    {
      "epoch": 1.5248933143669987,
      "grad_norm": 0.46293074284568964,
      "learning_rate": 5.723495868981344e-06,
      "loss": 0.0297,
      "step": 1608
    },
    {
      "epoch": 1.525841631104789,
      "grad_norm": 0.4707389836290285,
      "learning_rate": 5.718032283870839e-06,
      "loss": 0.0354,
      "step": 1609
    },
    {
      "epoch": 1.5267899478425795,
      "grad_norm": 0.4178418025887583,
      "learning_rate": 5.7125678232140054e-06,
      "loss": 0.0286,
      "step": 1610
    },
    {
      "epoch": 1.5277382645803699,
      "grad_norm": 0.43658626940138434,
      "learning_rate": 5.70710249367404e-06,
      "loss": 0.0387,
      "step": 1611
    },
    {
      "epoch": 1.5286865813181603,
      "grad_norm": 0.45394473647054495,
      "learning_rate": 5.701636301915192e-06,
      "loss": 0.0306,
      "step": 1612
    },
    {
      "epoch": 1.5296348980559507,
      "grad_norm": 1.333103451681259,
      "learning_rate": 5.69616925460277e-06,
      "loss": 0.0481,
      "step": 1613
    },
    {
      "epoch": 1.530583214793741,
      "grad_norm": 0.6256616011256059,
      "learning_rate": 5.690701358403119e-06,
      "loss": 0.0442,
      "step": 1614
    },
    {
      "epoch": 1.5315315315315314,
      "grad_norm": 0.32612870761505114,
      "learning_rate": 5.685232619983625e-06,
      "loss": 0.026,
      "step": 1615
    },
    {
      "epoch": 1.5324798482693218,
      "grad_norm": 0.3891832320687818,
      "learning_rate": 5.679763046012697e-06,
      "loss": 0.0255,
      "step": 1616
    },
    {
      "epoch": 1.5334281650071122,
      "grad_norm": 0.46453499563730116,
      "learning_rate": 5.6742926431597645e-06,
      "loss": 0.0427,
      "step": 1617
    },
    {
      "epoch": 1.5343764817449028,
      "grad_norm": 0.47783783736140073,
      "learning_rate": 5.668821418095266e-06,
      "loss": 0.0421,
      "step": 1618
    },
    {
      "epoch": 1.5353247984826932,
      "grad_norm": 0.6743992447584107,
      "learning_rate": 5.663349377490646e-06,
      "loss": 0.0358,
      "step": 1619
    },
    {
      "epoch": 1.5362731152204836,
      "grad_norm": 0.4107463403974195,
      "learning_rate": 5.65787652801834e-06,
      "loss": 0.0357,
      "step": 1620
    },
    {
      "epoch": 1.5372214319582742,
      "grad_norm": 0.5248494282764801,
      "learning_rate": 5.65240287635177e-06,
      "loss": 0.0344,
      "step": 1621
    },
    {
      "epoch": 1.5381697486960646,
      "grad_norm": 0.606570796875963,
      "learning_rate": 5.646928429165339e-06,
      "loss": 0.0281,
      "step": 1622
    },
    {
      "epoch": 1.539118065433855,
      "grad_norm": 0.42829215552732597,
      "learning_rate": 5.641453193134421e-06,
      "loss": 0.0447,
      "step": 1623
    },
    {
      "epoch": 1.5400663821716454,
      "grad_norm": 0.5164772119258303,
      "learning_rate": 5.635977174935343e-06,
      "loss": 0.029,
      "step": 1624
    },
    {
      "epoch": 1.5410146989094358,
      "grad_norm": 0.37297442039349055,
      "learning_rate": 5.630500381245397e-06,
      "loss": 0.0358,
      "step": 1625
    },
    {
      "epoch": 1.5419630156472262,
      "grad_norm": 0.3902979566469329,
      "learning_rate": 5.625022818742812e-06,
      "loss": 0.0463,
      "step": 1626
    },
    {
      "epoch": 1.5429113323850165,
      "grad_norm": 0.6496769657487748,
      "learning_rate": 5.619544494106761e-06,
      "loss": 0.0474,
      "step": 1627
    },
    {
      "epoch": 1.543859649122807,
      "grad_norm": 0.4223958965983336,
      "learning_rate": 5.6140654140173424e-06,
      "loss": 0.0364,
      "step": 1628
    },
    {
      "epoch": 1.5448079658605973,
      "grad_norm": 0.43884925035729194,
      "learning_rate": 5.6085855851555784e-06,
      "loss": 0.033,
      "step": 1629
    },
    {
      "epoch": 1.5457562825983877,
      "grad_norm": 0.530824449652907,
      "learning_rate": 5.603105014203398e-06,
      "loss": 0.0483,
      "step": 1630
    },
    {
      "epoch": 1.5467045993361783,
      "grad_norm": 0.4284263908585556,
      "learning_rate": 5.597623707843643e-06,
      "loss": 0.0508,
      "step": 1631
    },
    {
      "epoch": 1.5476529160739687,
      "grad_norm": 0.4535017203711632,
      "learning_rate": 5.59214167276005e-06,
      "loss": 0.0377,
      "step": 1632
    },
    {
      "epoch": 1.548601232811759,
      "grad_norm": 0.45970117181525466,
      "learning_rate": 5.586658915637239e-06,
      "loss": 0.0271,
      "step": 1633
    },
    {
      "epoch": 1.5495495495495497,
      "grad_norm": 0.48042151558541607,
      "learning_rate": 5.581175443160718e-06,
      "loss": 0.053,
      "step": 1634
    },
    {
      "epoch": 1.55049786628734,
      "grad_norm": 0.38033052193121797,
      "learning_rate": 5.57569126201686e-06,
      "loss": 0.0413,
      "step": 1635
    },
    {
      "epoch": 1.5514461830251305,
      "grad_norm": 0.3695377907477157,
      "learning_rate": 5.570206378892909e-06,
      "loss": 0.0291,
      "step": 1636
    },
    {
      "epoch": 1.5523944997629209,
      "grad_norm": 0.354032936247863,
      "learning_rate": 5.564720800476958e-06,
      "loss": 0.0422,
      "step": 1637
    },
    {
      "epoch": 1.5533428165007113,
      "grad_norm": 1.0102411364289114,
      "learning_rate": 5.559234533457953e-06,
      "loss": 0.042,
      "step": 1638
    },
    {
      "epoch": 1.5542911332385017,
      "grad_norm": 0.738120928557875,
      "learning_rate": 5.553747584525682e-06,
      "loss": 0.0389,
      "step": 1639
    },
    {
      "epoch": 1.555239449976292,
      "grad_norm": 0.799209724109593,
      "learning_rate": 5.548259960370754e-06,
      "loss": 0.0395,
      "step": 1640
    },
    {
      "epoch": 1.5561877667140824,
      "grad_norm": 0.4062247593100686,
      "learning_rate": 5.542771667684612e-06,
      "loss": 0.0315,
      "step": 1641
    },
    {
      "epoch": 1.5571360834518728,
      "grad_norm": 0.3308586428988418,
      "learning_rate": 5.537282713159507e-06,
      "loss": 0.031,
      "step": 1642
    },
    {
      "epoch": 1.5580844001896632,
      "grad_norm": 0.6086645010113465,
      "learning_rate": 5.5317931034885044e-06,
      "loss": 0.0463,
      "step": 1643
    },
    {
      "epoch": 1.5590327169274538,
      "grad_norm": 0.4762141993084212,
      "learning_rate": 5.526302845365461e-06,
      "loss": 0.0351,
      "step": 1644
    },
    {
      "epoch": 1.5599810336652442,
      "grad_norm": 0.7153367461924497,
      "learning_rate": 5.520811945485031e-06,
      "loss": 0.0489,
      "step": 1645
    },
    {
      "epoch": 1.5609293504030346,
      "grad_norm": 0.47448305807931646,
      "learning_rate": 5.515320410542642e-06,
      "loss": 0.0568,
      "step": 1646
    },
    {
      "epoch": 1.561877667140825,
      "grad_norm": 0.5897241369310743,
      "learning_rate": 5.509828247234505e-06,
      "loss": 0.0489,
      "step": 1647
    },
    {
      "epoch": 1.5628259838786156,
      "grad_norm": 0.43836735991299125,
      "learning_rate": 5.5043354622575955e-06,
      "loss": 0.0276,
      "step": 1648
    },
    {
      "epoch": 1.563774300616406,
      "grad_norm": 0.37950941420274353,
      "learning_rate": 5.498842062309643e-06,
      "loss": 0.0265,
      "step": 1649
    },
    {
      "epoch": 1.5647226173541964,
      "grad_norm": 0.5616670816349663,
      "learning_rate": 5.4933480540891295e-06,
      "loss": 0.0577,
      "step": 1650
    },
    {
      "epoch": 1.5656709340919868,
      "grad_norm": 0.6568262974602079,
      "learning_rate": 5.487853444295278e-06,
      "loss": 0.0466,
      "step": 1651
    },
    {
      "epoch": 1.5666192508297772,
      "grad_norm": 0.7263693663581391,
      "learning_rate": 5.482358239628047e-06,
      "loss": 0.0411,
      "step": 1652
    },
    {
      "epoch": 1.5675675675675675,
      "grad_norm": 0.5207396419110628,
      "learning_rate": 5.476862446788118e-06,
      "loss": 0.0517,
      "step": 1653
    },
    {
      "epoch": 1.568515884305358,
      "grad_norm": 0.5188917964277431,
      "learning_rate": 5.471366072476891e-06,
      "loss": 0.0493,
      "step": 1654
    },
    {
      "epoch": 1.5694642010431483,
      "grad_norm": 0.4000030591629918,
      "learning_rate": 5.465869123396474e-06,
      "loss": 0.0304,
      "step": 1655
    },
    {
      "epoch": 1.5704125177809387,
      "grad_norm": 0.42697872862878716,
      "learning_rate": 5.460371606249677e-06,
      "loss": 0.0338,
      "step": 1656
    },
    {
      "epoch": 1.571360834518729,
      "grad_norm": 0.4896290919012838,
      "learning_rate": 5.454873527740002e-06,
      "loss": 0.0489,
      "step": 1657
    },
    {
      "epoch": 1.5723091512565197,
      "grad_norm": 0.7554443088523393,
      "learning_rate": 5.449374894571635e-06,
      "loss": 0.0391,
      "step": 1658
    },
    {
      "epoch": 1.57325746799431,
      "grad_norm": 0.5802368408383383,
      "learning_rate": 5.443875713449439e-06,
      "loss": 0.0411,
      "step": 1659
    },
    {
      "epoch": 1.5742057847321005,
      "grad_norm": 0.5018529796661364,
      "learning_rate": 5.438375991078946e-06,
      "loss": 0.0416,
      "step": 1660
    },
    {
      "epoch": 1.575154101469891,
      "grad_norm": 0.3183205147770903,
      "learning_rate": 5.4328757341663475e-06,
      "loss": 0.0349,
      "step": 1661
    },
    {
      "epoch": 1.5761024182076815,
      "grad_norm": 1.011391019460894,
      "learning_rate": 5.427374949418487e-06,
      "loss": 0.0337,
      "step": 1662
    },
    {
      "epoch": 1.5770507349454719,
      "grad_norm": 0.3191949723530378,
      "learning_rate": 5.421873643542846e-06,
      "loss": 0.0305,
      "step": 1663
    },
    {
      "epoch": 1.5779990516832623,
      "grad_norm": 0.634640230734434,
      "learning_rate": 5.4163718232475525e-06,
      "loss": 0.0508,
      "step": 1664
    },
    {
      "epoch": 1.5789473684210527,
      "grad_norm": 0.5852720760242924,
      "learning_rate": 5.4108694952413546e-06,
      "loss": 0.0443,
      "step": 1665
    },
    {
      "epoch": 1.579895685158843,
      "grad_norm": 0.5855319152782336,
      "learning_rate": 5.4053666662336176e-06,
      "loss": 0.0327,
      "step": 1666
    },
    {
      "epoch": 1.5808440018966334,
      "grad_norm": 0.3586298744998486,
      "learning_rate": 5.399863342934324e-06,
      "loss": 0.0406,
      "step": 1667
    },
    {
      "epoch": 1.5817923186344238,
      "grad_norm": 0.49399058191728706,
      "learning_rate": 5.394359532054054e-06,
      "loss": 0.0301,
      "step": 1668
    },
    {
      "epoch": 1.5827406353722142,
      "grad_norm": 1.2375067802560742,
      "learning_rate": 5.388855240303985e-06,
      "loss": 0.0701,
      "step": 1669
    },
    {
      "epoch": 1.5836889521100046,
      "grad_norm": 0.6195169804808017,
      "learning_rate": 5.38335047439588e-06,
      "loss": 0.0536,
      "step": 1670
    },
    {
      "epoch": 1.5846372688477952,
      "grad_norm": 0.4760457301339429,
      "learning_rate": 5.377845241042077e-06,
      "loss": 0.0459,
      "step": 1671
    },
    {
      "epoch": 1.5855855855855856,
      "grad_norm": 0.42237056822248414,
      "learning_rate": 5.372339546955493e-06,
      "loss": 0.041,
      "step": 1672
    },
    {
      "epoch": 1.586533902323376,
      "grad_norm": 0.7588679568473088,
      "learning_rate": 5.3668333988495955e-06,
      "loss": 0.0617,
      "step": 1673
    },
    {
      "epoch": 1.5874822190611664,
      "grad_norm": 0.5129085018035644,
      "learning_rate": 5.361326803438414e-06,
      "loss": 0.0466,
      "step": 1674
    },
    {
      "epoch": 1.588430535798957,
      "grad_norm": 0.35504034518206484,
      "learning_rate": 5.3558197674365174e-06,
      "loss": 0.0319,
      "step": 1675
    },
    {
      "epoch": 1.5893788525367474,
      "grad_norm": 0.4719261955609522,
      "learning_rate": 5.350312297559018e-06,
      "loss": 0.0434,
      "step": 1676
    },
    {
      "epoch": 1.5903271692745378,
      "grad_norm": 0.3948218019688561,
      "learning_rate": 5.344804400521554e-06,
      "loss": 0.034,
      "step": 1677
    },
    {
      "epoch": 1.5912754860123282,
      "grad_norm": 0.3895494909416238,
      "learning_rate": 5.3392960830402825e-06,
      "loss": 0.0297,
      "step": 1678
    },
    {
      "epoch": 1.5922238027501185,
      "grad_norm": 0.4137215707959619,
      "learning_rate": 5.333787351831875e-06,
      "loss": 0.0362,
      "step": 1679
    },
    {
      "epoch": 1.593172119487909,
      "grad_norm": 0.44517521813349287,
      "learning_rate": 5.328278213613509e-06,
      "loss": 0.0419,
      "step": 1680
    },
    {
      "epoch": 1.5941204362256993,
      "grad_norm": 0.46221054971476766,
      "learning_rate": 5.322768675102857e-06,
      "loss": 0.0426,
      "step": 1681
    },
    {
      "epoch": 1.5950687529634897,
      "grad_norm": 0.49217226245635887,
      "learning_rate": 5.31725874301808e-06,
      "loss": 0.044,
      "step": 1682
    },
    {
      "epoch": 1.59601706970128,
      "grad_norm": 0.5613029052207704,
      "learning_rate": 5.3117484240778176e-06,
      "loss": 0.0452,
      "step": 1683
    },
    {
      "epoch": 1.5969653864390705,
      "grad_norm": 0.38744508130227856,
      "learning_rate": 5.306237725001181e-06,
      "loss": 0.0346,
      "step": 1684
    },
    {
      "epoch": 1.597913703176861,
      "grad_norm": 0.6033140558546168,
      "learning_rate": 5.3007266525077484e-06,
      "loss": 0.0519,
      "step": 1685
    },
    {
      "epoch": 1.5988620199146515,
      "grad_norm": 0.3844932046652491,
      "learning_rate": 5.295215213317549e-06,
      "loss": 0.0339,
      "step": 1686
    },
    {
      "epoch": 1.5998103366524419,
      "grad_norm": 0.35948437819579393,
      "learning_rate": 5.289703414151062e-06,
      "loss": 0.0324,
      "step": 1687
    },
    {
      "epoch": 1.6007586533902325,
      "grad_norm": 0.49403545880259037,
      "learning_rate": 5.284191261729206e-06,
      "loss": 0.0377,
      "step": 1688
    },
    {
      "epoch": 1.6017069701280229,
      "grad_norm": 0.3950827081822717,
      "learning_rate": 5.278678762773326e-06,
      "loss": 0.0348,
      "step": 1689
    },
    {
      "epoch": 1.6026552868658133,
      "grad_norm": 0.46679652631135615,
      "learning_rate": 5.273165924005195e-06,
      "loss": 0.0354,
      "step": 1690
    },
    {
      "epoch": 1.6036036036036037,
      "grad_norm": 0.4187949082045406,
      "learning_rate": 5.267652752146994e-06,
      "loss": 0.0358,
      "step": 1691
    },
    {
      "epoch": 1.604551920341394,
      "grad_norm": 0.5390502663154565,
      "learning_rate": 5.2621392539213185e-06,
      "loss": 0.0415,
      "step": 1692
    },
    {
      "epoch": 1.6055002370791844,
      "grad_norm": 0.3749414556008541,
      "learning_rate": 5.256625436051156e-06,
      "loss": 0.0297,
      "step": 1693
    },
    {
      "epoch": 1.6064485538169748,
      "grad_norm": 0.43791392697589476,
      "learning_rate": 5.251111305259886e-06,
      "loss": 0.0427,
      "step": 1694
    },
    {
      "epoch": 1.6073968705547652,
      "grad_norm": 0.4147327877599569,
      "learning_rate": 5.245596868271265e-06,
      "loss": 0.0339,
      "step": 1695
    },
    {
      "epoch": 1.6083451872925556,
      "grad_norm": 0.4367673755602324,
      "learning_rate": 5.240082131809431e-06,
      "loss": 0.041,
      "step": 1696
    },
    {
      "epoch": 1.609293504030346,
      "grad_norm": 0.3920759685283346,
      "learning_rate": 5.234567102598881e-06,
      "loss": 0.0393,
      "step": 1697
    },
    {
      "epoch": 1.6102418207681366,
      "grad_norm": 0.38612730122526456,
      "learning_rate": 5.229051787364471e-06,
      "loss": 0.0385,
      "step": 1698
    },
    {
      "epoch": 1.611190137505927,
      "grad_norm": 0.3810633326641417,
      "learning_rate": 5.223536192831405e-06,
      "loss": 0.0413,
      "step": 1699
    },
    {
      "epoch": 1.6121384542437174,
      "grad_norm": 0.4008625923351243,
      "learning_rate": 5.21802032572523e-06,
      "loss": 0.0359,
      "step": 1700
    },
    {
      "epoch": 1.613086770981508,
      "grad_norm": 0.4980860436464716,
      "learning_rate": 5.212504192771822e-06,
      "loss": 0.0443,
      "step": 1701
    },
    {
      "epoch": 1.6140350877192984,
      "grad_norm": 0.41197267376046887,
      "learning_rate": 5.206987800697383e-06,
      "loss": 0.0337,
      "step": 1702
    },
    {
      "epoch": 1.6149834044570888,
      "grad_norm": 0.4783120794734666,
      "learning_rate": 5.20147115622843e-06,
      "loss": 0.0533,
      "step": 1703
    },
    {
      "epoch": 1.6159317211948792,
      "grad_norm": 0.4038354701384072,
      "learning_rate": 5.19595426609179e-06,
      "loss": 0.0414,
      "step": 1704
    },
    {
      "epoch": 1.6168800379326695,
      "grad_norm": 0.5116662661624387,
      "learning_rate": 5.1904371370145866e-06,
      "loss": 0.0433,
      "step": 1705
    },
    {
      "epoch": 1.61782835467046,
      "grad_norm": 0.3616855724287448,
      "learning_rate": 5.184919775724236e-06,
      "loss": 0.0344,
      "step": 1706
    },
    {
      "epoch": 1.6187766714082503,
      "grad_norm": 0.498168738572277,
      "learning_rate": 5.179402188948438e-06,
      "loss": 0.044,
      "step": 1707
    },
    {
      "epoch": 1.6197249881460407,
      "grad_norm": 0.5241305541833553,
      "learning_rate": 5.173884383415168e-06,
      "loss": 0.0496,
      "step": 1708
    },
    {
      "epoch": 1.620673304883831,
      "grad_norm": 0.6574907730267259,
      "learning_rate": 5.168366365852666e-06,
      "loss": 0.045,
      "step": 1709
    },
    {
      "epoch": 1.6216216216216215,
      "grad_norm": 0.40685958969794783,
      "learning_rate": 5.162848142989434e-06,
      "loss": 0.0298,
      "step": 1710
    },
    {
      "epoch": 1.6225699383594119,
      "grad_norm": 0.3934908581350256,
      "learning_rate": 5.157329721554218e-06,
      "loss": 0.0385,
      "step": 1711
    },
    {
      "epoch": 1.6235182550972025,
      "grad_norm": 0.5479655834153282,
      "learning_rate": 5.151811108276011e-06,
      "loss": 0.0414,
      "step": 1712
    },
    {
      "epoch": 1.6244665718349929,
      "grad_norm": 0.4485734151651903,
      "learning_rate": 5.146292309884043e-06,
      "loss": 0.034,
      "step": 1713
    },
    {
      "epoch": 1.6254148885727833,
      "grad_norm": 0.38005314977033383,
      "learning_rate": 5.140773333107763e-06,
      "loss": 0.0319,
      "step": 1714
    },
    {
      "epoch": 1.6263632053105739,
      "grad_norm": 0.49033527821168377,
      "learning_rate": 5.13525418467684e-06,
      "loss": 0.0347,
      "step": 1715
    },
    {
      "epoch": 1.6273115220483643,
      "grad_norm": 0.4376750362085485,
      "learning_rate": 5.129734871321153e-06,
      "loss": 0.0277,
      "step": 1716
    },
    {
      "epoch": 1.6282598387861547,
      "grad_norm": 0.4508191180466739,
      "learning_rate": 5.124215399770782e-06,
      "loss": 0.0298,
      "step": 1717
    },
    {
      "epoch": 1.629208155523945,
      "grad_norm": 0.5458986876368422,
      "learning_rate": 5.118695776756001e-06,
      "loss": 0.045,
      "step": 1718
    },
    {
      "epoch": 1.6301564722617354,
      "grad_norm": 0.48550938492428214,
      "learning_rate": 5.113176009007264e-06,
      "loss": 0.0353,
      "step": 1719
    },
    {
      "epoch": 1.6311047889995258,
      "grad_norm": 0.5214878006549084,
      "learning_rate": 5.1076561032552076e-06,
      "loss": 0.0492,
      "step": 1720
    },
    {
      "epoch": 1.6320531057373162,
      "grad_norm": 0.3575931662216187,
      "learning_rate": 5.102136066230634e-06,
      "loss": 0.0296,
      "step": 1721
    },
    {
      "epoch": 1.6330014224751066,
      "grad_norm": 0.38118575881201,
      "learning_rate": 5.096615904664505e-06,
      "loss": 0.0301,
      "step": 1722
    },
    {
      "epoch": 1.633949739212897,
      "grad_norm": 0.5525837485507148,
      "learning_rate": 5.091095625287933e-06,
      "loss": 0.0476,
      "step": 1723
    },
    {
      "epoch": 1.6348980559506874,
      "grad_norm": 0.45798268615313587,
      "learning_rate": 5.085575234832177e-06,
      "loss": 0.0375,
      "step": 1724
    },
    {
      "epoch": 1.635846372688478,
      "grad_norm": 0.32084838714691044,
      "learning_rate": 5.08005474002863e-06,
      "loss": 0.0288,
      "step": 1725
    },
    {
      "epoch": 1.6367946894262684,
      "grad_norm": 0.48894734808026186,
      "learning_rate": 5.074534147608813e-06,
      "loss": 0.037,
      "step": 1726
    },
    {
      "epoch": 1.6377430061640588,
      "grad_norm": 0.5863056845718708,
      "learning_rate": 5.069013464304365e-06,
      "loss": 0.0467,
      "step": 1727
    },
    {
      "epoch": 1.6386913229018494,
      "grad_norm": 0.5276793927642295,
      "learning_rate": 5.063492696847035e-06,
      "loss": 0.048,
      "step": 1728
    },
    {
      "epoch": 1.6396396396396398,
      "grad_norm": 0.4500869018064269,
      "learning_rate": 5.057971851968678e-06,
      "loss": 0.0346,
      "step": 1729
    },
    {
      "epoch": 1.6405879563774302,
      "grad_norm": 0.4974245711828883,
      "learning_rate": 5.05245093640124e-06,
      "loss": 0.0401,
      "step": 1730
    },
    {
      "epoch": 1.6415362731152205,
      "grad_norm": 0.5070745888081949,
      "learning_rate": 5.046929956876755e-06,
      "loss": 0.0376,
      "step": 1731
    },
    {
      "epoch": 1.642484589853011,
      "grad_norm": 0.6592711837920248,
      "learning_rate": 5.0414089201273335e-06,
      "loss": 0.0337,
      "step": 1732
    },
    {
      "epoch": 1.6434329065908013,
      "grad_norm": 0.6797373207354449,
      "learning_rate": 5.035887832885158e-06,
      "loss": 0.0403,
      "step": 1733
    },
    {
      "epoch": 1.6443812233285917,
      "grad_norm": 0.4710691221262725,
      "learning_rate": 5.03036670188247e-06,
      "loss": 0.0411,
      "step": 1734
    },
    {
      "epoch": 1.645329540066382,
      "grad_norm": 0.5376307614124892,
      "learning_rate": 5.024845533851567e-06,
      "loss": 0.0505,
      "step": 1735
    },
    {
      "epoch": 1.6462778568041725,
      "grad_norm": 0.5486265059532447,
      "learning_rate": 5.019324335524787e-06,
      "loss": 0.0447,
      "step": 1736
    },
    {
      "epoch": 1.6472261735419629,
      "grad_norm": 0.40758845237104635,
      "learning_rate": 5.013803113634513e-06,
      "loss": 0.034,
      "step": 1737
    },
    {
      "epoch": 1.6481744902797535,
      "grad_norm": 0.6104444776515453,
      "learning_rate": 5.00828187491315e-06,
      "loss": 0.0398,
      "step": 1738
    },
    {
      "epoch": 1.6491228070175439,
      "grad_norm": 0.4307141085331387,
      "learning_rate": 5.002760626093125e-06,
      "loss": 0.0366,
      "step": 1739
    },
    {
      "epoch": 1.6500711237553343,
      "grad_norm": 0.6440031916908964,
      "learning_rate": 4.997239373906877e-06,
      "loss": 0.0425,
      "step": 1740
    },
    {
      "epoch": 1.6510194404931247,
      "grad_norm": 0.5807419766923566,
      "learning_rate": 4.991718125086851e-06,
      "loss": 0.0434,
      "step": 1741
    },
    {
      "epoch": 1.6519677572309153,
      "grad_norm": 0.3942692224681114,
      "learning_rate": 4.986196886365488e-06,
      "loss": 0.031,
      "step": 1742
    },
    {
      "epoch": 1.6529160739687057,
      "grad_norm": 0.4128003706084824,
      "learning_rate": 4.9806756644752145e-06,
      "loss": 0.0383,
      "step": 1743
    },
    {
      "epoch": 1.653864390706496,
      "grad_norm": 0.4085578606989102,
      "learning_rate": 4.975154466148435e-06,
      "loss": 0.0261,
      "step": 1744
    },
    {
      "epoch": 1.6548127074442864,
      "grad_norm": 0.4837103927049282,
      "learning_rate": 4.969633298117533e-06,
      "loss": 0.0351,
      "step": 1745
    },
    {
      "epoch": 1.6557610241820768,
      "grad_norm": 0.7395875455109805,
      "learning_rate": 4.964112167114844e-06,
      "loss": 0.039,
      "step": 1746
    },
    {
      "epoch": 1.6567093409198672,
      "grad_norm": 0.4854107983390489,
      "learning_rate": 4.958591079872667e-06,
      "loss": 0.0357,
      "step": 1747
    },
    {
      "epoch": 1.6576576576576576,
      "grad_norm": 0.3483061645891692,
      "learning_rate": 4.953070043123247e-06,
      "loss": 0.035,
      "step": 1748
    },
    {
      "epoch": 1.658605974395448,
      "grad_norm": 0.37143969392815496,
      "learning_rate": 4.947549063598761e-06,
      "loss": 0.0299,
      "step": 1749
    },
    {
      "epoch": 1.6595542911332384,
      "grad_norm": 0.3749415284322426,
      "learning_rate": 4.942028148031322e-06,
      "loss": 0.0389,
      "step": 1750
    },
    {
      "epoch": 1.6605026078710288,
      "grad_norm": 0.4991282731845152,
      "learning_rate": 4.936507303152966e-06,
      "loss": 0.0262,
      "step": 1751
    },
    {
      "epoch": 1.6614509246088194,
      "grad_norm": 0.3863346846624415,
      "learning_rate": 4.930986535695636e-06,
      "loss": 0.0338,
      "step": 1752
    },
    {
      "epoch": 1.6623992413466098,
      "grad_norm": 0.3863586588308062,
      "learning_rate": 4.9254658523911885e-06,
      "loss": 0.0288,
      "step": 1753
    },
    {
      "epoch": 1.6633475580844002,
      "grad_norm": 0.5587025668116927,
      "learning_rate": 4.9199452599713715e-06,
      "loss": 0.0461,
      "step": 1754
    },
    {
      "epoch": 1.6642958748221908,
      "grad_norm": 0.4497105307702213,
      "learning_rate": 4.914424765167824e-06,
      "loss": 0.0385,
      "step": 1755
    },
    {
      "epoch": 1.6652441915599812,
      "grad_norm": 0.42629532853182683,
      "learning_rate": 4.908904374712069e-06,
      "loss": 0.0357,
      "step": 1756
    },
    {
      "epoch": 1.6661925082977715,
      "grad_norm": 0.3850936464294085,
      "learning_rate": 4.903384095335497e-06,
      "loss": 0.0261,
      "step": 1757
    },
    {
      "epoch": 1.667140825035562,
      "grad_norm": 0.5328143876026115,
      "learning_rate": 4.8978639337693665e-06,
      "loss": 0.0318,
      "step": 1758
    },
    {
      "epoch": 1.6680891417733523,
      "grad_norm": 0.3369899929014371,
      "learning_rate": 4.892343896744794e-06,
      "loss": 0.0246,
      "step": 1759
    },
    {
      "epoch": 1.6690374585111427,
      "grad_norm": 0.6423782785289791,
      "learning_rate": 4.8868239909927365e-06,
      "loss": 0.034,
      "step": 1760
    },
    {
      "epoch": 1.669985775248933,
      "grad_norm": 0.3956672279589752,
      "learning_rate": 4.881304223244002e-06,
      "loss": 0.0363,
      "step": 1761
    },
    {
      "epoch": 1.6709340919867235,
      "grad_norm": 0.3549262283376078,
      "learning_rate": 4.875784600229219e-06,
      "loss": 0.0274,
      "step": 1762
    },
    {
      "epoch": 1.6718824087245139,
      "grad_norm": 0.44033629831242393,
      "learning_rate": 4.870265128678848e-06,
      "loss": 0.036,
      "step": 1763
    },
    {
      "epoch": 1.6728307254623043,
      "grad_norm": 0.38102544990314435,
      "learning_rate": 4.864745815323162e-06,
      "loss": 0.0354,
      "step": 1764
    },
    {
      "epoch": 1.6737790422000949,
      "grad_norm": 0.7336679820913383,
      "learning_rate": 4.859226666892239e-06,
      "loss": 0.0372,
      "step": 1765
    },
    {
      "epoch": 1.6747273589378853,
      "grad_norm": 0.4783768456867206,
      "learning_rate": 4.8537076901159575e-06,
      "loss": 0.0319,
      "step": 1766
    },
    {
      "epoch": 1.6756756756756757,
      "grad_norm": 0.535219503295994,
      "learning_rate": 4.84818889172399e-06,
      "loss": 0.0452,
      "step": 1767
    },
    {
      "epoch": 1.676623992413466,
      "grad_norm": 0.2982755405751966,
      "learning_rate": 4.8426702784457835e-06,
      "loss": 0.0235,
      "step": 1768
    },
    {
      "epoch": 1.6775723091512567,
      "grad_norm": 0.45187225376000945,
      "learning_rate": 4.8371518570105695e-06,
      "loss": 0.0332,
      "step": 1769
    },
    {
      "epoch": 1.678520625889047,
      "grad_norm": 0.5137536616757206,
      "learning_rate": 4.831633634147335e-06,
      "loss": 0.0516,
      "step": 1770
    },
    {
      "epoch": 1.6794689426268374,
      "grad_norm": 0.45838875310802574,
      "learning_rate": 4.826115616584832e-06,
      "loss": 0.0401,
      "step": 1771
    },
    {
      "epoch": 1.6804172593646278,
      "grad_norm": 0.430174715041985,
      "learning_rate": 4.820597811051563e-06,
      "loss": 0.0297,
      "step": 1772
    },
    {
      "epoch": 1.6813655761024182,
      "grad_norm": 0.2962896754783431,
      "learning_rate": 4.815080224275765e-06,
      "loss": 0.0252,
      "step": 1773
    },
    {
      "epoch": 1.6823138928402086,
      "grad_norm": 0.37585418826664413,
      "learning_rate": 4.809562862985416e-06,
      "loss": 0.0321,
      "step": 1774
    },
    {
      "epoch": 1.683262209577999,
      "grad_norm": 0.6008863761423583,
      "learning_rate": 4.8040457339082115e-06,
      "loss": 0.0313,
      "step": 1775
    },
    {
      "epoch": 1.6842105263157894,
      "grad_norm": 0.6656061062299986,
      "learning_rate": 4.79852884377157e-06,
      "loss": 0.0642,
      "step": 1776
    },
    {
      "epoch": 1.6851588430535798,
      "grad_norm": 0.3957049880883251,
      "learning_rate": 4.793012199302619e-06,
      "loss": 0.0352,
      "step": 1777
    },
    {
      "epoch": 1.6861071597913702,
      "grad_norm": 0.5470753034512885,
      "learning_rate": 4.787495807228179e-06,
      "loss": 0.0417,
      "step": 1778
    },
    {
      "epoch": 1.6870554765291608,
      "grad_norm": 0.2999392604451325,
      "learning_rate": 4.78197967427477e-06,
      "loss": 0.0246,
      "step": 1779
    },
    {
      "epoch": 1.6880037932669512,
      "grad_norm": 0.39413640160909796,
      "learning_rate": 4.776463807168596e-06,
      "loss": 0.0348,
      "step": 1780
    },
    {
      "epoch": 1.6889521100047415,
      "grad_norm": 0.4398686088172201,
      "learning_rate": 4.770948212635531e-06,
      "loss": 0.0316,
      "step": 1781
    },
    {
      "epoch": 1.6899004267425322,
      "grad_norm": 0.43305122714845934,
      "learning_rate": 4.765432897401121e-06,
      "loss": 0.0317,
      "step": 1782
    },
    {
      "epoch": 1.6908487434803225,
      "grad_norm": 0.4517409472016299,
      "learning_rate": 4.7599178681905705e-06,
      "loss": 0.0413,
      "step": 1783
    },
    {
      "epoch": 1.691797060218113,
      "grad_norm": 0.5616680977844734,
      "learning_rate": 4.754403131728736e-06,
      "loss": 0.043,
      "step": 1784
    },
    {
      "epoch": 1.6927453769559033,
      "grad_norm": 0.37478799993443473,
      "learning_rate": 4.748888694740117e-06,
      "loss": 0.033,
      "step": 1785
    },
    {
      "epoch": 1.6936936936936937,
      "grad_norm": 0.5182765249765543,
      "learning_rate": 4.743374563948846e-06,
      "loss": 0.0412,
      "step": 1786
    },
    {
      "epoch": 1.694642010431484,
      "grad_norm": 0.5349561090586026,
      "learning_rate": 4.737860746078682e-06,
      "loss": 0.036,
      "step": 1787
    },
    {
      "epoch": 1.6955903271692745,
      "grad_norm": 0.8610126674967102,
      "learning_rate": 4.7323472478530075e-06,
      "loss": 0.0615,
      "step": 1788
    },
    {
      "epoch": 1.6965386439070649,
      "grad_norm": 0.5757179501499256,
      "learning_rate": 4.726834075994807e-06,
      "loss": 0.0471,
      "step": 1789
    },
    {
      "epoch": 1.6974869606448553,
      "grad_norm": 0.5279814169362466,
      "learning_rate": 4.721321237226676e-06,
      "loss": 0.0457,
      "step": 1790
    },
    {
      "epoch": 1.6984352773826457,
      "grad_norm": 0.3926629636024604,
      "learning_rate": 4.7158087382707955e-06,
      "loss": 0.045,
      "step": 1791
    },
    {
      "epoch": 1.6993835941204363,
      "grad_norm": 0.453224341946254,
      "learning_rate": 4.710296585848938e-06,
      "loss": 0.037,
      "step": 1792
    },
    {
      "epoch": 1.7003319108582267,
      "grad_norm": 0.44388438579404543,
      "learning_rate": 4.704784786682453e-06,
      "loss": 0.0343,
      "step": 1793
    },
    {
      "epoch": 1.701280227596017,
      "grad_norm": 0.38919604350005726,
      "learning_rate": 4.699273347492253e-06,
      "loss": 0.0235,
      "step": 1794
    },
    {
      "epoch": 1.7022285443338074,
      "grad_norm": 0.4603473957874724,
      "learning_rate": 4.693762274998819e-06,
      "loss": 0.0338,
      "step": 1795
    },
    {
      "epoch": 1.703176861071598,
      "grad_norm": 0.5169626767035945,
      "learning_rate": 4.688251575922185e-06,
      "loss": 0.0414,
      "step": 1796
    },
    {
      "epoch": 1.7041251778093884,
      "grad_norm": 0.3372120183155561,
      "learning_rate": 4.682741256981922e-06,
      "loss": 0.0263,
      "step": 1797
    },
    {
      "epoch": 1.7050734945471788,
      "grad_norm": 0.42042614991062877,
      "learning_rate": 4.6772313248971455e-06,
      "loss": 0.0381,
      "step": 1798
    },
    {
      "epoch": 1.7060218112849692,
      "grad_norm": 0.3793277130075472,
      "learning_rate": 4.671721786386492e-06,
      "loss": 0.0364,
      "step": 1799
    },
    {
      "epoch": 1.7069701280227596,
      "grad_norm": 0.36173441445946797,
      "learning_rate": 4.6662126481681255e-06,
      "loss": 0.027,
      "step": 1800
    },
    {
      "epoch": 1.70791844476055,
      "grad_norm": 0.42820923235128877,
      "learning_rate": 4.660703916959719e-06,
      "loss": 0.0318,
      "step": 1801
    },
    {
      "epoch": 1.7088667614983404,
      "grad_norm": 0.4465434244504414,
      "learning_rate": 4.655195599478448e-06,
      "loss": 0.0313,
      "step": 1802
    },
    {
      "epoch": 1.7098150782361308,
      "grad_norm": 0.46894391958023623,
      "learning_rate": 4.649687702440982e-06,
      "loss": 0.0446,
      "step": 1803
    },
    {
      "epoch": 1.7107633949739212,
      "grad_norm": 0.37867650726134006,
      "learning_rate": 4.644180232563484e-06,
      "loss": 0.0311,
      "step": 1804
    },
    {
      "epoch": 1.7117117117117115,
      "grad_norm": 0.3702374369015596,
      "learning_rate": 4.638673196561587e-06,
      "loss": 0.0289,
      "step": 1805
    },
    {
      "epoch": 1.7126600284495022,
      "grad_norm": 0.36563753129172316,
      "learning_rate": 4.633166601150407e-06,
      "loss": 0.033,
      "step": 1806
    },
    {
      "epoch": 1.7136083451872925,
      "grad_norm": 0.4200759775460671,
      "learning_rate": 4.627660453044508e-06,
      "loss": 0.0315,
      "step": 1807
    },
    {
      "epoch": 1.714556661925083,
      "grad_norm": 0.4364188122631288,
      "learning_rate": 4.622154758957923e-06,
      "loss": 0.0344,
      "step": 1808
    },
    {
      "epoch": 1.7155049786628735,
      "grad_norm": 0.36606066056529396,
      "learning_rate": 4.616649525604123e-06,
      "loss": 0.0285,
      "step": 1809
    },
    {
      "epoch": 1.716453295400664,
      "grad_norm": 0.5148294395120504,
      "learning_rate": 4.611144759696016e-06,
      "loss": 0.0587,
      "step": 1810
    },
    {
      "epoch": 1.7174016121384543,
      "grad_norm": 0.6881685438316215,
      "learning_rate": 4.605640467945946e-06,
      "loss": 0.0598,
      "step": 1811
    },
    {
      "epoch": 1.7183499288762447,
      "grad_norm": 0.5735474647239276,
      "learning_rate": 4.600136657065678e-06,
      "loss": 0.0399,
      "step": 1812
    },
    {
      "epoch": 1.719298245614035,
      "grad_norm": 0.36219250240066375,
      "learning_rate": 4.594633333766383e-06,
      "loss": 0.0343,
      "step": 1813
    },
    {
      "epoch": 1.7202465623518255,
      "grad_norm": 0.40670988595909385,
      "learning_rate": 4.589130504758648e-06,
      "loss": 0.0281,
      "step": 1814
    },
    {
      "epoch": 1.7211948790896159,
      "grad_norm": 0.43024128435525394,
      "learning_rate": 4.583628176752448e-06,
      "loss": 0.0342,
      "step": 1815
    },
    {
      "epoch": 1.7221431958274063,
      "grad_norm": 0.35903986981532177,
      "learning_rate": 4.578126356457154e-06,
      "loss": 0.0267,
      "step": 1816
    },
    {
      "epoch": 1.7230915125651967,
      "grad_norm": 0.3783152391951721,
      "learning_rate": 4.572625050581516e-06,
      "loss": 0.0268,
      "step": 1817
    },
    {
      "epoch": 1.724039829302987,
      "grad_norm": 0.4612759243512315,
      "learning_rate": 4.567124265833654e-06,
      "loss": 0.0314,
      "step": 1818
    },
    {
      "epoch": 1.7249881460407777,
      "grad_norm": 0.36167166767777,
      "learning_rate": 4.561624008921054e-06,
      "loss": 0.0313,
      "step": 1819
    },
    {
      "epoch": 1.725936462778568,
      "grad_norm": 0.5603215159296359,
      "learning_rate": 4.556124286550563e-06,
      "loss": 0.0536,
      "step": 1820
    },
    {
      "epoch": 1.7268847795163584,
      "grad_norm": 0.3181070277595216,
      "learning_rate": 4.550625105428367e-06,
      "loss": 0.0227,
      "step": 1821
    },
    {
      "epoch": 1.727833096254149,
      "grad_norm": 0.4081175989973651,
      "learning_rate": 4.545126472260001e-06,
      "loss": 0.0347,
      "step": 1822
    },
    {
      "epoch": 1.7287814129919394,
      "grad_norm": 0.4338531761499665,
      "learning_rate": 4.539628393750324e-06,
      "loss": 0.0349,
      "step": 1823
    },
    {
      "epoch": 1.7297297297297298,
      "grad_norm": 0.5070890405906837,
      "learning_rate": 4.534130876603527e-06,
      "loss": 0.0352,
      "step": 1824
    },
    {
      "epoch": 1.7306780464675202,
      "grad_norm": 0.3871814529225249,
      "learning_rate": 4.5286339275231115e-06,
      "loss": 0.0285,
      "step": 1825
    },
    {
      "epoch": 1.7316263632053106,
      "grad_norm": 0.40989108008019104,
      "learning_rate": 4.523137553211883e-06,
      "loss": 0.0325,
      "step": 1826
    },
    {
      "epoch": 1.732574679943101,
      "grad_norm": 0.4366308345846694,
      "learning_rate": 4.5176417603719555e-06,
      "loss": 0.0404,
      "step": 1827
    },
    {
      "epoch": 1.7335229966808914,
      "grad_norm": 0.3764252415741553,
      "learning_rate": 4.512146555704723e-06,
      "loss": 0.0333,
      "step": 1828
    },
    {
      "epoch": 1.7344713134186818,
      "grad_norm": 0.3711934688107721,
      "learning_rate": 4.506651945910872e-06,
      "loss": 0.0253,
      "step": 1829
    },
    {
      "epoch": 1.7354196301564722,
      "grad_norm": 1.0453131397548512,
      "learning_rate": 4.501157937690359e-06,
      "loss": 0.02,
      "step": 1830
    },
    {
      "epoch": 1.7363679468942625,
      "grad_norm": 0.4301932604454958,
      "learning_rate": 4.495664537742405e-06,
      "loss": 0.0383,
      "step": 1831
    },
    {
      "epoch": 1.7373162636320532,
      "grad_norm": 0.35893295448394297,
      "learning_rate": 4.490171752765494e-06,
      "loss": 0.0327,
      "step": 1832
    },
    {
      "epoch": 1.7382645803698435,
      "grad_norm": 0.5133813007167333,
      "learning_rate": 4.48467958945736e-06,
      "loss": 0.0296,
      "step": 1833
    },
    {
      "epoch": 1.739212897107634,
      "grad_norm": 0.43884167593440265,
      "learning_rate": 4.479188054514971e-06,
      "loss": 0.031,
      "step": 1834
    },
    {
      "epoch": 1.7401612138454243,
      "grad_norm": 0.3979319557061671,
      "learning_rate": 4.47369715463454e-06,
      "loss": 0.0278,
      "step": 1835
    },
    {
      "epoch": 1.741109530583215,
      "grad_norm": 0.3781187564948932,
      "learning_rate": 4.468206896511497e-06,
      "loss": 0.0273,
      "step": 1836
    },
    {
      "epoch": 1.7420578473210053,
      "grad_norm": 0.5240765391753767,
      "learning_rate": 4.462717286840493e-06,
      "loss": 0.0352,
      "step": 1837
    },
    {
      "epoch": 1.7430061640587957,
      "grad_norm": 0.3344508789434431,
      "learning_rate": 4.457228332315391e-06,
      "loss": 0.0295,
      "step": 1838
    },
    {
      "epoch": 1.743954480796586,
      "grad_norm": 0.3765289513807636,
      "learning_rate": 4.451740039629247e-06,
      "loss": 0.0321,
      "step": 1839
    },
    {
      "epoch": 1.7449027975343765,
      "grad_norm": 0.605396755278652,
      "learning_rate": 4.446252415474321e-06,
      "loss": 0.0618,
      "step": 1840
    },
    {
      "epoch": 1.7458511142721669,
      "grad_norm": 0.37145283802650364,
      "learning_rate": 4.440765466542048e-06,
      "loss": 0.0324,
      "step": 1841
    },
    {
      "epoch": 1.7467994310099573,
      "grad_norm": 0.34847655213343387,
      "learning_rate": 4.435279199523043e-06,
      "loss": 0.0331,
      "step": 1842
    },
    {
      "epoch": 1.7477477477477477,
      "grad_norm": 0.550675362545679,
      "learning_rate": 4.429793621107094e-06,
      "loss": 0.0381,
      "step": 1843
    },
    {
      "epoch": 1.748696064485538,
      "grad_norm": 0.43671841183507826,
      "learning_rate": 4.424308737983141e-06,
      "loss": 0.0427,
      "step": 1844
    },
    {
      "epoch": 1.7496443812233284,
      "grad_norm": 0.399738947003064,
      "learning_rate": 4.418824556839284e-06,
      "loss": 0.0259,
      "step": 1845
    },
    {
      "epoch": 1.750592697961119,
      "grad_norm": 0.4045286330954736,
      "learning_rate": 4.413341084362762e-06,
      "loss": 0.0283,
      "step": 1846
    },
    {
      "epoch": 1.7515410146989094,
      "grad_norm": 0.596321484570831,
      "learning_rate": 4.407858327239952e-06,
      "loss": 0.0416,
      "step": 1847
    },
    {
      "epoch": 1.7524893314366998,
      "grad_norm": 0.3962890329473834,
      "learning_rate": 4.402376292156357e-06,
      "loss": 0.0225,
      "step": 1848
    },
    {
      "epoch": 1.7534376481744904,
      "grad_norm": 0.6108314208398574,
      "learning_rate": 4.396894985796603e-06,
      "loss": 0.0549,
      "step": 1849
    },
    {
      "epoch": 1.7543859649122808,
      "grad_norm": 0.4926118481779225,
      "learning_rate": 4.391414414844423e-06,
      "loss": 0.0352,
      "step": 1850
    },
    {
      "epoch": 1.7553342816500712,
      "grad_norm": 0.44216328985669257,
      "learning_rate": 4.385934585982658e-06,
      "loss": 0.0359,
      "step": 1851
    },
    {
      "epoch": 1.7562825983878616,
      "grad_norm": 0.5108809743875607,
      "learning_rate": 4.38045550589324e-06,
      "loss": 0.0409,
      "step": 1852
    },
    {
      "epoch": 1.757230915125652,
      "grad_norm": 0.3407129811866031,
      "learning_rate": 4.374977181257189e-06,
      "loss": 0.0281,
      "step": 1853
    },
    {
      "epoch": 1.7581792318634424,
      "grad_norm": 0.5348515455628596,
      "learning_rate": 4.369499618754606e-06,
      "loss": 0.0342,
      "step": 1854
    },
    {
      "epoch": 1.7591275486012328,
      "grad_norm": 0.5279301635319789,
      "learning_rate": 4.364022825064658e-06,
      "loss": 0.045,
      "step": 1855
    },
    {
      "epoch": 1.7600758653390232,
      "grad_norm": 0.5050771561563465,
      "learning_rate": 4.358546806865581e-06,
      "loss": 0.0416,
      "step": 1856
    },
    {
      "epoch": 1.7610241820768135,
      "grad_norm": 0.43174863952207065,
      "learning_rate": 4.353071570834662e-06,
      "loss": 0.0317,
      "step": 1857
    },
    {
      "epoch": 1.761972498814604,
      "grad_norm": 0.41940789267051237,
      "learning_rate": 4.3475971236482304e-06,
      "loss": 0.0277,
      "step": 1858
    },
    {
      "epoch": 1.7629208155523945,
      "grad_norm": 0.44670846590238744,
      "learning_rate": 4.342123471981663e-06,
      "loss": 0.0496,
      "step": 1859
    },
    {
      "epoch": 1.763869132290185,
      "grad_norm": 0.4642364288838703,
      "learning_rate": 4.336650622509356e-06,
      "loss": 0.0402,
      "step": 1860
    },
    {
      "epoch": 1.7648174490279753,
      "grad_norm": 0.37073060927337753,
      "learning_rate": 4.331178581904735e-06,
      "loss": 0.0308,
      "step": 1861
    },
    {
      "epoch": 1.7657657657657657,
      "grad_norm": 0.4435398096512773,
      "learning_rate": 4.325707356840237e-06,
      "loss": 0.034,
      "step": 1862
    },
    {
      "epoch": 1.7667140825035563,
      "grad_norm": 0.4369429326386228,
      "learning_rate": 4.320236953987304e-06,
      "loss": 0.0339,
      "step": 1863
    },
    {
      "epoch": 1.7676623992413467,
      "grad_norm": 0.37836536445740354,
      "learning_rate": 4.3147673800163744e-06,
      "loss": 0.0281,
      "step": 1864
    },
    {
      "epoch": 1.768610715979137,
      "grad_norm": 0.3265793848210341,
      "learning_rate": 4.309298641596882e-06,
      "loss": 0.0316,
      "step": 1865
    },
    {
      "epoch": 1.7695590327169275,
      "grad_norm": 0.32105094423844494,
      "learning_rate": 4.303830745397231e-06,
      "loss": 0.0297,
      "step": 1866
    },
    {
      "epoch": 1.7705073494547179,
      "grad_norm": 0.9731314448776828,
      "learning_rate": 4.29836369808481e-06,
      "loss": 0.0496,
      "step": 1867
    },
    {
      "epoch": 1.7714556661925083,
      "grad_norm": 0.41064791082286095,
      "learning_rate": 4.292897506325962e-06,
      "loss": 0.0314,
      "step": 1868
    },
    {
      "epoch": 1.7724039829302987,
      "grad_norm": 0.41521808764811713,
      "learning_rate": 4.2874321767859945e-06,
      "loss": 0.0394,
      "step": 1869
    },
    {
      "epoch": 1.773352299668089,
      "grad_norm": 0.5429748722960639,
      "learning_rate": 4.281967716129163e-06,
      "loss": 0.0409,
      "step": 1870
    },
    {
      "epoch": 1.7743006164058794,
      "grad_norm": 0.4896073053023119,
      "learning_rate": 4.276504131018658e-06,
      "loss": 0.0434,
      "step": 1871
    },
    {
      "epoch": 1.7752489331436698,
      "grad_norm": 0.45935466151081317,
      "learning_rate": 4.271041428116608e-06,
      "loss": 0.0429,
      "step": 1872
    },
    {
      "epoch": 1.7761972498814604,
      "grad_norm": 0.4608816010569723,
      "learning_rate": 4.265579614084065e-06,
      "loss": 0.0329,
      "step": 1873
    },
    {
      "epoch": 1.7771455666192508,
      "grad_norm": 0.4675738747142554,
      "learning_rate": 4.260118695580992e-06,
      "loss": 0.0412,
      "step": 1874
    },
    {
      "epoch": 1.7780938833570412,
      "grad_norm": 0.5351511581707835,
      "learning_rate": 4.254658679266268e-06,
      "loss": 0.0342,
      "step": 1875
    },
    {
      "epoch": 1.7790422000948318,
      "grad_norm": 0.47923179366593766,
      "learning_rate": 4.249199571797664e-06,
      "loss": 0.0354,
      "step": 1876
    },
    {
      "epoch": 1.7799905168326222,
      "grad_norm": 0.4751740142576013,
      "learning_rate": 4.243741379831848e-06,
      "loss": 0.048,
      "step": 1877
    },
    {
      "epoch": 1.7809388335704126,
      "grad_norm": 0.43079624768664576,
      "learning_rate": 4.238284110024374e-06,
      "loss": 0.0383,
      "step": 1878
    },
    {
      "epoch": 1.781887150308203,
      "grad_norm": 0.4308505818689027,
      "learning_rate": 4.232827769029663e-06,
      "loss": 0.0306,
      "step": 1879
    },
    {
      "epoch": 1.7828354670459934,
      "grad_norm": 0.34173757368848234,
      "learning_rate": 4.2273723635010075e-06,
      "loss": 0.0279,
      "step": 1880
    },
    {
      "epoch": 1.7837837837837838,
      "grad_norm": 0.2974816565902399,
      "learning_rate": 4.221917900090566e-06,
      "loss": 0.0245,
      "step": 1881
    },
    {
      "epoch": 1.7847321005215742,
      "grad_norm": 0.429383794979146,
      "learning_rate": 4.216464385449335e-06,
      "loss": 0.0452,
      "step": 1882
    },
    {
      "epoch": 1.7856804172593645,
      "grad_norm": 0.37510369228807144,
      "learning_rate": 4.211011826227168e-06,
      "loss": 0.0313,
      "step": 1883
    },
    {
      "epoch": 1.786628733997155,
      "grad_norm": 0.5164577689856673,
      "learning_rate": 4.205560229072742e-06,
      "loss": 0.0406,
      "step": 1884
    },
    {
      "epoch": 1.7875770507349453,
      "grad_norm": 0.396145933769621,
      "learning_rate": 4.200109600633567e-06,
      "loss": 0.0442,
      "step": 1885
    },
    {
      "epoch": 1.788525367472736,
      "grad_norm": 0.4024500110736181,
      "learning_rate": 4.1946599475559724e-06,
      "loss": 0.0395,
      "step": 1886
    },
    {
      "epoch": 1.7894736842105263,
      "grad_norm": 0.4541562531386443,
      "learning_rate": 4.189211276485093e-06,
      "loss": 0.0343,
      "step": 1887
    },
    {
      "epoch": 1.7904220009483167,
      "grad_norm": 0.33885170624784317,
      "learning_rate": 4.183763594064874e-06,
      "loss": 0.027,
      "step": 1888
    },
    {
      "epoch": 1.791370317686107,
      "grad_norm": 0.7885135242857828,
      "learning_rate": 4.1783169069380445e-06,
      "loss": 0.0529,
      "step": 1889
    },
    {
      "epoch": 1.7923186344238977,
      "grad_norm": 1.5910177901330982,
      "learning_rate": 4.172871221746132e-06,
      "loss": 0.0693,
      "step": 1890
    },
    {
      "epoch": 1.793266951161688,
      "grad_norm": 0.3613904128403713,
      "learning_rate": 4.167426545129435e-06,
      "loss": 0.0317,
      "step": 1891
    },
    {
      "epoch": 1.7942152678994785,
      "grad_norm": 0.4949823076890155,
      "learning_rate": 4.16198288372702e-06,
      "loss": 0.0385,
      "step": 1892
    },
    {
      "epoch": 1.7951635846372689,
      "grad_norm": 0.5873559373413337,
      "learning_rate": 4.156540244176722e-06,
      "loss": 0.0434,
      "step": 1893
    },
    {
      "epoch": 1.7961119013750593,
      "grad_norm": 0.7461574029871659,
      "learning_rate": 4.151098633115129e-06,
      "loss": 0.0556,
      "step": 1894
    },
    {
      "epoch": 1.7970602181128497,
      "grad_norm": 0.4822043358893843,
      "learning_rate": 4.145658057177569e-06,
      "loss": 0.0323,
      "step": 1895
    },
    {
      "epoch": 1.79800853485064,
      "grad_norm": 0.8738249467892109,
      "learning_rate": 4.1402185229981155e-06,
      "loss": 0.0428,
      "step": 1896
    },
    {
      "epoch": 1.7989568515884304,
      "grad_norm": 0.5198052374533522,
      "learning_rate": 4.134780037209563e-06,
      "loss": 0.0399,
      "step": 1897
    },
    {
      "epoch": 1.7999051683262208,
      "grad_norm": 0.6244484104917516,
      "learning_rate": 4.129342606443436e-06,
      "loss": 0.0396,
      "step": 1898
    },
    {
      "epoch": 1.8008534850640112,
      "grad_norm": 0.6966474793382613,
      "learning_rate": 4.123906237329971e-06,
      "loss": 0.0383,
      "step": 1899
    },
    {
      "epoch": 1.8018018018018018,
      "grad_norm": 0.8286265100115936,
      "learning_rate": 4.118470936498105e-06,
      "loss": 0.0503,
      "step": 1900
    },
    {
      "epoch": 1.8027501185395922,
      "grad_norm": 0.6974574681974006,
      "learning_rate": 4.113036710575476e-06,
      "loss": 0.0395,
      "step": 1901
    },
    {
      "epoch": 1.8036984352773826,
      "grad_norm": 0.448930049036162,
      "learning_rate": 4.107603566188412e-06,
      "loss": 0.0377,
      "step": 1902
    },
    {
      "epoch": 1.8046467520151732,
      "grad_norm": 0.4293294488172685,
      "learning_rate": 4.102171509961919e-06,
      "loss": 0.0366,
      "step": 1903
    },
    {
      "epoch": 1.8055950687529636,
      "grad_norm": 0.5127412423653616,
      "learning_rate": 4.096740548519681e-06,
      "loss": 0.0486,
      "step": 1904
    },
    {
      "epoch": 1.806543385490754,
      "grad_norm": 0.5189142966976821,
      "learning_rate": 4.091310688484041e-06,
      "loss": 0.0385,
      "step": 1905
    },
    {
      "epoch": 1.8074917022285444,
      "grad_norm": 0.41364900033387003,
      "learning_rate": 4.085881936476005e-06,
      "loss": 0.0288,
      "step": 1906
    },
    {
      "epoch": 1.8084400189663348,
      "grad_norm": 0.687884255688505,
      "learning_rate": 4.080454299115224e-06,
      "loss": 0.0319,
      "step": 1907
    },
    {
      "epoch": 1.8093883357041252,
      "grad_norm": 0.5197620472106413,
      "learning_rate": 4.075027783019989e-06,
      "loss": 0.0352,
      "step": 1908
    },
    {
      "epoch": 1.8103366524419156,
      "grad_norm": 0.42525429976878526,
      "learning_rate": 4.0696023948072274e-06,
      "loss": 0.0408,
      "step": 1909
    },
    {
      "epoch": 1.811284969179706,
      "grad_norm": 0.5828270165546796,
      "learning_rate": 4.064178141092491e-06,
      "loss": 0.0335,
      "step": 1910
    },
    {
      "epoch": 1.8122332859174963,
      "grad_norm": 0.4343714216988695,
      "learning_rate": 4.058755028489945e-06,
      "loss": 0.0257,
      "step": 1911
    },
    {
      "epoch": 1.8131816026552867,
      "grad_norm": 0.5099005671458844,
      "learning_rate": 4.053333063612365e-06,
      "loss": 0.0382,
      "step": 1912
    },
    {
      "epoch": 1.8141299193930773,
      "grad_norm": 0.47310374354497153,
      "learning_rate": 4.0479122530711255e-06,
      "loss": 0.0386,
      "step": 1913
    },
    {
      "epoch": 1.8150782361308677,
      "grad_norm": 0.3880372722401674,
      "learning_rate": 4.042492603476197e-06,
      "loss": 0.0318,
      "step": 1914
    },
    {
      "epoch": 1.816026552868658,
      "grad_norm": 0.5630916853522246,
      "learning_rate": 4.037074121436135e-06,
      "loss": 0.0421,
      "step": 1915
    },
    {
      "epoch": 1.8169748696064487,
      "grad_norm": 0.4050061779858133,
      "learning_rate": 4.031656813558063e-06,
      "loss": 0.0218,
      "step": 1916
    },
    {
      "epoch": 1.817923186344239,
      "grad_norm": 0.5487831061447199,
      "learning_rate": 4.0262406864476816e-06,
      "loss": 0.0414,
      "step": 1917
    },
    {
      "epoch": 1.8188715030820295,
      "grad_norm": 0.4118953389358272,
      "learning_rate": 4.020825746709249e-06,
      "loss": 0.0275,
      "step": 1918
    },
    {
      "epoch": 1.8198198198198199,
      "grad_norm": 0.3646881365149475,
      "learning_rate": 4.015412000945573e-06,
      "loss": 0.0306,
      "step": 1919
    },
    {
      "epoch": 1.8207681365576103,
      "grad_norm": 0.36407786342534326,
      "learning_rate": 4.009999455758011e-06,
      "loss": 0.032,
      "step": 1920
    },
    {
      "epoch": 1.8217164532954007,
      "grad_norm": 0.6322503860720039,
      "learning_rate": 4.004588117746447e-06,
      "loss": 0.0305,
      "step": 1921
    },
    {
      "epoch": 1.822664770033191,
      "grad_norm": 0.35424208882329783,
      "learning_rate": 3.999177993509303e-06,
      "loss": 0.0296,
      "step": 1922
    },
    {
      "epoch": 1.8236130867709814,
      "grad_norm": 0.5037021770720644,
      "learning_rate": 3.9937690896435195e-06,
      "loss": 0.0456,
      "step": 1923
    },
    {
      "epoch": 1.8245614035087718,
      "grad_norm": 0.28275182943356875,
      "learning_rate": 3.98836141274454e-06,
      "loss": 0.0254,
      "step": 1924
    },
    {
      "epoch": 1.8255097202465622,
      "grad_norm": 1.0874197920409632,
      "learning_rate": 3.982954969406322e-06,
      "loss": 0.0378,
      "step": 1925
    },
    {
      "epoch": 1.8264580369843528,
      "grad_norm": 0.39797960479767225,
      "learning_rate": 3.977549766221316e-06,
      "loss": 0.0329,
      "step": 1926
    },
    {
      "epoch": 1.8274063537221432,
      "grad_norm": 0.4589297830996474,
      "learning_rate": 3.972145809780457e-06,
      "loss": 0.0341,
      "step": 1927
    },
    {
      "epoch": 1.8283546704599336,
      "grad_norm": 0.6342219517636322,
      "learning_rate": 3.966743106673165e-06,
      "loss": 0.0287,
      "step": 1928
    },
    {
      "epoch": 1.829302987197724,
      "grad_norm": 0.5386132100170457,
      "learning_rate": 3.961341663487324e-06,
      "loss": 0.0427,
      "step": 1929
    },
    {
      "epoch": 1.8302513039355146,
      "grad_norm": 0.5080057913810255,
      "learning_rate": 3.95594148680929e-06,
      "loss": 0.0355,
      "step": 1930
    },
    {
      "epoch": 1.831199620673305,
      "grad_norm": 0.3669532593099998,
      "learning_rate": 3.950542583223871e-06,
      "loss": 0.0336,
      "step": 1931
    },
    {
      "epoch": 1.8321479374110954,
      "grad_norm": 0.4557500825827911,
      "learning_rate": 3.945144959314322e-06,
      "loss": 0.0449,
      "step": 1932
    },
    {
      "epoch": 1.8330962541488858,
      "grad_norm": 0.41433118107831046,
      "learning_rate": 3.939748621662339e-06,
      "loss": 0.0394,
      "step": 1933
    },
    {
      "epoch": 1.8340445708866762,
      "grad_norm": 0.36284629747376757,
      "learning_rate": 3.93435357684805e-06,
      "loss": 0.0324,
      "step": 1934
    },
    {
      "epoch": 1.8349928876244666,
      "grad_norm": 0.6488793981008372,
      "learning_rate": 3.928959831450003e-06,
      "loss": 0.036,
      "step": 1935
    },
    {
      "epoch": 1.835941204362257,
      "grad_norm": 0.4072871196865869,
      "learning_rate": 3.923567392045167e-06,
      "loss": 0.0361,
      "step": 1936
    },
    {
      "epoch": 1.8368895211000473,
      "grad_norm": 0.5995330166879924,
      "learning_rate": 3.918176265208912e-06,
      "loss": 0.047,
      "step": 1937
    },
    {
      "epoch": 1.8378378378378377,
      "grad_norm": 0.5742511704396418,
      "learning_rate": 3.912786457515013e-06,
      "loss": 0.0368,
      "step": 1938
    },
    {
      "epoch": 1.838786154575628,
      "grad_norm": 0.37173427951758653,
      "learning_rate": 3.907397975535637e-06,
      "loss": 0.0295,
      "step": 1939
    },
    {
      "epoch": 1.8397344713134187,
      "grad_norm": 0.4283387112035294,
      "learning_rate": 3.902010825841328e-06,
      "loss": 0.0418,
      "step": 1940
    },
    {
      "epoch": 1.840682788051209,
      "grad_norm": 0.42658758128230867,
      "learning_rate": 3.896625015001011e-06,
      "loss": 0.0307,
      "step": 1941
    },
    {
      "epoch": 1.8416311047889995,
      "grad_norm": 0.4103737401087204,
      "learning_rate": 3.891240549581979e-06,
      "loss": 0.0466,
      "step": 1942
    },
    {
      "epoch": 1.84257942152679,
      "grad_norm": 0.5179813798652112,
      "learning_rate": 3.885857436149879e-06,
      "loss": 0.032,
      "step": 1943
    },
    {
      "epoch": 1.8435277382645805,
      "grad_norm": 0.36142149152719033,
      "learning_rate": 3.880475681268716e-06,
      "loss": 0.025,
      "step": 1944
    },
    {
      "epoch": 1.8444760550023709,
      "grad_norm": 0.38988766477546516,
      "learning_rate": 3.8750952915008315e-06,
      "loss": 0.0266,
      "step": 1945
    },
    {
      "epoch": 1.8454243717401613,
      "grad_norm": 0.4047589561085592,
      "learning_rate": 3.869716273406908e-06,
      "loss": 0.0291,
      "step": 1946
    },
    {
      "epoch": 1.8463726884779517,
      "grad_norm": 0.5591559055271785,
      "learning_rate": 3.864338633545956e-06,
      "loss": 0.0281,
      "step": 1947
    },
    {
      "epoch": 1.847321005215742,
      "grad_norm": 0.44542403886367465,
      "learning_rate": 3.8589623784752985e-06,
      "loss": 0.0407,
      "step": 1948
    },
    {
      "epoch": 1.8482693219535324,
      "grad_norm": 0.45876612433553804,
      "learning_rate": 3.853587514750579e-06,
      "loss": 0.0315,
      "step": 1949
    },
    {
      "epoch": 1.8492176386913228,
      "grad_norm": 0.3111868531988744,
      "learning_rate": 3.848214048925736e-06,
      "loss": 0.0277,
      "step": 1950
    },
    {
      "epoch": 1.8501659554291132,
      "grad_norm": 0.4353007357835014,
      "learning_rate": 3.842841987553007e-06,
      "loss": 0.0319,
      "step": 1951
    },
    {
      "epoch": 1.8511142721669036,
      "grad_norm": 0.5289123449785142,
      "learning_rate": 3.837471337182923e-06,
      "loss": 0.0318,
      "step": 1952
    },
    {
      "epoch": 1.8520625889046942,
      "grad_norm": 0.5868521892114408,
      "learning_rate": 3.832102104364281e-06,
      "loss": 0.0344,
      "step": 1953
    },
    {
      "epoch": 1.8530109056424846,
      "grad_norm": 0.5617353335767225,
      "learning_rate": 3.826734295644163e-06,
      "loss": 0.0352,
      "step": 1954
    },
    {
      "epoch": 1.853959222380275,
      "grad_norm": 0.5789507471320139,
      "learning_rate": 3.821367917567908e-06,
      "loss": 0.0529,
      "step": 1955
    },
    {
      "epoch": 1.8549075391180654,
      "grad_norm": 0.4921075321136551,
      "learning_rate": 3.816002976679107e-06,
      "loss": 0.0279,
      "step": 1956
    },
    {
      "epoch": 1.855855855855856,
      "grad_norm": 0.4600403890394494,
      "learning_rate": 3.8106394795196087e-06,
      "loss": 0.0317,
      "step": 1957
    },
    {
      "epoch": 1.8568041725936464,
      "grad_norm": 0.6494450216853208,
      "learning_rate": 3.80527743262949e-06,
      "loss": 0.0501,
      "step": 1958
    },
    {
      "epoch": 1.8577524893314368,
      "grad_norm": 0.3991594068949773,
      "learning_rate": 3.799916842547068e-06,
      "loss": 0.0337,
      "step": 1959
    },
    {
      "epoch": 1.8587008060692272,
      "grad_norm": 0.6181340333450788,
      "learning_rate": 3.7945577158088793e-06,
      "loss": 0.0364,
      "step": 1960
    },
    {
      "epoch": 1.8596491228070176,
      "grad_norm": 0.37075530743846147,
      "learning_rate": 3.7892000589496773e-06,
      "loss": 0.0246,
      "step": 1961
    },
    {
      "epoch": 1.860597439544808,
      "grad_norm": 0.5054091368468414,
      "learning_rate": 3.7838438785024216e-06,
      "loss": 0.0379,
      "step": 1962
    },
    {
      "epoch": 1.8615457562825983,
      "grad_norm": 0.44895407969120477,
      "learning_rate": 3.7784891809982767e-06,
      "loss": 0.0352,
      "step": 1963
    },
    {
      "epoch": 1.8624940730203887,
      "grad_norm": 0.35070271043852363,
      "learning_rate": 3.773135972966589e-06,
      "loss": 0.032,
      "step": 1964
    },
    {
      "epoch": 1.863442389758179,
      "grad_norm": 0.9903843261764405,
      "learning_rate": 3.7677842609349e-06,
      "loss": 0.0365,
      "step": 1965
    },
    {
      "epoch": 1.8643907064959695,
      "grad_norm": 0.48477176290481,
      "learning_rate": 3.762434051428918e-06,
      "loss": 0.0323,
      "step": 1966
    },
    {
      "epoch": 1.86533902323376,
      "grad_norm": 0.39374562548602104,
      "learning_rate": 3.7570853509725234e-06,
      "loss": 0.0302,
      "step": 1967
    },
    {
      "epoch": 1.8662873399715505,
      "grad_norm": 0.47578607994744204,
      "learning_rate": 3.7517381660877587e-06,
      "loss": 0.0431,
      "step": 1968
    },
    {
      "epoch": 1.8672356567093409,
      "grad_norm": 0.5424376735264428,
      "learning_rate": 3.7463925032948114e-06,
      "loss": 0.0487,
      "step": 1969
    },
    {
      "epoch": 1.8681839734471315,
      "grad_norm": 0.44176826832585336,
      "learning_rate": 3.741048369112019e-06,
      "loss": 0.0389,
      "step": 1970
    },
    {
      "epoch": 1.8691322901849219,
      "grad_norm": 0.4390592738444192,
      "learning_rate": 3.735705770055855e-06,
      "loss": 0.0412,
      "step": 1971
    },
    {
      "epoch": 1.8700806069227123,
      "grad_norm": 0.4183210896360201,
      "learning_rate": 3.7303647126409153e-06,
      "loss": 0.0366,
      "step": 1972
    },
    {
      "epoch": 1.8710289236605027,
      "grad_norm": 0.6065333984388325,
      "learning_rate": 3.7250252033799243e-06,
      "loss": 0.0467,
      "step": 1973
    },
    {
      "epoch": 1.871977240398293,
      "grad_norm": 0.518349410639526,
      "learning_rate": 3.71968724878371e-06,
      "loss": 0.041,
      "step": 1974
    },
    {
      "epoch": 1.8729255571360834,
      "grad_norm": 0.43597774812646534,
      "learning_rate": 3.714350855361212e-06,
      "loss": 0.0381,
      "step": 1975
    },
    {
      "epoch": 1.8738738738738738,
      "grad_norm": 0.5067159246242335,
      "learning_rate": 3.709016029619461e-06,
      "loss": 0.0482,
      "step": 1976
    },
    {
      "epoch": 1.8748221906116642,
      "grad_norm": 0.34210931197001854,
      "learning_rate": 3.703682778063581e-06,
      "loss": 0.0303,
      "step": 1977
    },
    {
      "epoch": 1.8757705073494546,
      "grad_norm": 0.42520698759519004,
      "learning_rate": 3.69835110719677e-06,
      "loss": 0.0445,
      "step": 1978
    },
    {
      "epoch": 1.876718824087245,
      "grad_norm": 0.4230721150319016,
      "learning_rate": 3.6930210235203067e-06,
      "loss": 0.0283,
      "step": 1979
    },
    {
      "epoch": 1.8776671408250356,
      "grad_norm": 0.42411343718392436,
      "learning_rate": 3.6876925335335255e-06,
      "loss": 0.03,
      "step": 1980
    },
    {
      "epoch": 1.878615457562826,
      "grad_norm": 0.616820313190297,
      "learning_rate": 3.6823656437338267e-06,
      "loss": 0.0374,
      "step": 1981
    },
    {
      "epoch": 1.8795637743006164,
      "grad_norm": 0.41781910861393473,
      "learning_rate": 3.6770403606166514e-06,
      "loss": 0.0304,
      "step": 1982
    },
    {
      "epoch": 1.8805120910384068,
      "grad_norm": 0.5860766950716109,
      "learning_rate": 3.6717166906754843e-06,
      "loss": 0.0404,
      "step": 1983
    },
    {
      "epoch": 1.8814604077761974,
      "grad_norm": 0.7252531827073903,
      "learning_rate": 3.666394640401848e-06,
      "loss": 0.0312,
      "step": 1984
    },
    {
      "epoch": 1.8824087245139878,
      "grad_norm": 0.553908681165879,
      "learning_rate": 3.6610742162852807e-06,
      "loss": 0.0452,
      "step": 1985
    },
    {
      "epoch": 1.8833570412517782,
      "grad_norm": 0.30665698348068665,
      "learning_rate": 3.655755424813346e-06,
      "loss": 0.0229,
      "step": 1986
    },
    {
      "epoch": 1.8843053579895686,
      "grad_norm": 0.6570412463785198,
      "learning_rate": 3.6504382724716137e-06,
      "loss": 0.0316,
      "step": 1987
    },
    {
      "epoch": 1.885253674727359,
      "grad_norm": 0.5144282052416823,
      "learning_rate": 3.6451227657436517e-06,
      "loss": 0.0349,
      "step": 1988
    },
    {
      "epoch": 1.8862019914651493,
      "grad_norm": 0.4071279928061887,
      "learning_rate": 3.639808911111028e-06,
      "loss": 0.0285,
      "step": 1989
    },
    {
      "epoch": 1.8871503082029397,
      "grad_norm": 0.5358728007249505,
      "learning_rate": 3.634496715053288e-06,
      "loss": 0.0351,
      "step": 1990
    },
    {
      "epoch": 1.88809862494073,
      "grad_norm": 0.368200617097554,
      "learning_rate": 3.6291861840479613e-06,
      "loss": 0.0298,
      "step": 1991
    },
    {
      "epoch": 1.8890469416785205,
      "grad_norm": 0.4507922059461737,
      "learning_rate": 3.6238773245705483e-06,
      "loss": 0.0375,
      "step": 1992
    },
    {
      "epoch": 1.8899952584163109,
      "grad_norm": 0.4081128435721486,
      "learning_rate": 3.6185701430945032e-06,
      "loss": 0.0383,
      "step": 1993
    },
    {
      "epoch": 1.8909435751541015,
      "grad_norm": 0.33297226361322074,
      "learning_rate": 3.613264646091239e-06,
      "loss": 0.0286,
      "step": 1994
    },
    {
      "epoch": 1.8918918918918919,
      "grad_norm": 0.53979762982992,
      "learning_rate": 3.6079608400301182e-06,
      "loss": 0.0358,
      "step": 1995
    },
    {
      "epoch": 1.8928402086296823,
      "grad_norm": 0.4090634897905571,
      "learning_rate": 3.6026587313784325e-06,
      "loss": 0.0319,
      "step": 1996
    },
    {
      "epoch": 1.8937885253674729,
      "grad_norm": 0.46149770151838015,
      "learning_rate": 3.597358326601413e-06,
      "loss": 0.0451,
      "step": 1997
    },
    {
      "epoch": 1.8947368421052633,
      "grad_norm": 0.3639527520811153,
      "learning_rate": 3.592059632162207e-06,
      "loss": 0.0334,
      "step": 1998
    },
    {
      "epoch": 1.8956851588430537,
      "grad_norm": 0.3891264328147345,
      "learning_rate": 3.5867626545218786e-06,
      "loss": 0.0352,
      "step": 1999
    },
    {
      "epoch": 1.896633475580844,
      "grad_norm": 0.5597673879815559,
      "learning_rate": 3.5814674001394007e-06,
      "loss": 0.0363,
      "step": 2000
    },
    {
      "epoch": 1.8975817923186344,
      "grad_norm": 0.3685996313252056,
      "learning_rate": 3.576173875471638e-06,
      "loss": 0.0326,
      "step": 2001
    },
    {
      "epoch": 1.8985301090564248,
      "grad_norm": 0.787312091428569,
      "learning_rate": 3.5708820869733552e-06,
      "loss": 0.0419,
      "step": 2002
    },
    {
      "epoch": 1.8994784257942152,
      "grad_norm": 0.3830970434240292,
      "learning_rate": 3.565592041097191e-06,
      "loss": 0.0284,
      "step": 2003
    },
    {
      "epoch": 1.9004267425320056,
      "grad_norm": 0.33426646878917393,
      "learning_rate": 3.5603037442936672e-06,
      "loss": 0.0297,
      "step": 2004
    },
    {
      "epoch": 1.901375059269796,
      "grad_norm": 0.33746177626737933,
      "learning_rate": 3.5550172030111687e-06,
      "loss": 0.0223,
      "step": 2005
    },
    {
      "epoch": 1.9023233760075864,
      "grad_norm": 0.44722672220400045,
      "learning_rate": 3.5497324236959363e-06,
      "loss": 0.0321,
      "step": 2006
    },
    {
      "epoch": 1.903271692745377,
      "grad_norm": 0.4834586259648445,
      "learning_rate": 3.5444494127920694e-06,
      "loss": 0.0335,
      "step": 2007
    },
    {
      "epoch": 1.9042200094831674,
      "grad_norm": 0.38059120675515146,
      "learning_rate": 3.5391681767415093e-06,
      "loss": 0.0287,
      "step": 2008
    },
    {
      "epoch": 1.9051683262209578,
      "grad_norm": 0.3075997620618822,
      "learning_rate": 3.533888721984029e-06,
      "loss": 0.0278,
      "step": 2009
    },
    {
      "epoch": 1.9061166429587484,
      "grad_norm": 0.40917618286352053,
      "learning_rate": 3.5286110549572337e-06,
      "loss": 0.0357,
      "step": 2010
    },
    {
      "epoch": 1.9070649596965388,
      "grad_norm": 0.42430230207581826,
      "learning_rate": 3.5233351820965445e-06,
      "loss": 0.0502,
      "step": 2011
    },
    {
      "epoch": 1.9080132764343292,
      "grad_norm": 0.3515137768052769,
      "learning_rate": 3.518061109835199e-06,
      "loss": 0.0315,
      "step": 2012
    },
    {
      "epoch": 1.9089615931721196,
      "grad_norm": 0.5095722782417215,
      "learning_rate": 3.51278884460424e-06,
      "loss": 0.0354,
      "step": 2013
    },
    {
      "epoch": 1.90990990990991,
      "grad_norm": 0.4270657425513164,
      "learning_rate": 3.5075183928325018e-06,
      "loss": 0.0368,
      "step": 2014
    },
    {
      "epoch": 1.9108582266477003,
      "grad_norm": 0.3808378401797508,
      "learning_rate": 3.5022497609466086e-06,
      "loss": 0.0291,
      "step": 2015
    },
    {
      "epoch": 1.9118065433854907,
      "grad_norm": 0.4100158806633352,
      "learning_rate": 3.4969829553709715e-06,
      "loss": 0.0349,
      "step": 2016
    },
    {
      "epoch": 1.912754860123281,
      "grad_norm": 0.5809676625893341,
      "learning_rate": 3.4917179825277652e-06,
      "loss": 0.04,
      "step": 2017
    },
    {
      "epoch": 1.9137031768610715,
      "grad_norm": 0.3906675617466994,
      "learning_rate": 3.4864548488369385e-06,
      "loss": 0.0355,
      "step": 2018
    },
    {
      "epoch": 1.9146514935988619,
      "grad_norm": 0.4223285103973374,
      "learning_rate": 3.4811935607161907e-06,
      "loss": 0.0271,
      "step": 2019
    },
    {
      "epoch": 1.9155998103366523,
      "grad_norm": 0.37651968514215806,
      "learning_rate": 3.4759341245809754e-06,
      "loss": 0.0261,
      "step": 2020
    },
    {
      "epoch": 1.9165481270744429,
      "grad_norm": 0.35511518960123745,
      "learning_rate": 3.470676546844486e-06,
      "loss": 0.0268,
      "step": 2021
    },
    {
      "epoch": 1.9174964438122333,
      "grad_norm": 0.6409970088958556,
      "learning_rate": 3.4654208339176475e-06,
      "loss": 0.04,
      "step": 2022
    },
    {
      "epoch": 1.9184447605500237,
      "grad_norm": 0.4635735551631786,
      "learning_rate": 3.460166992209115e-06,
      "loss": 0.0292,
      "step": 2023
    },
    {
      "epoch": 1.9193930772878143,
      "grad_norm": 0.38062200204638524,
      "learning_rate": 3.4549150281252635e-06,
      "loss": 0.0295,
      "step": 2024
    },
    {
      "epoch": 1.9203413940256047,
      "grad_norm": 0.6007637897320102,
      "learning_rate": 3.4496649480701717e-06,
      "loss": 0.0416,
      "step": 2025
    },
    {
      "epoch": 1.921289710763395,
      "grad_norm": 0.36191314067497893,
      "learning_rate": 3.4444167584456257e-06,
      "loss": 0.0315,
      "step": 2026
    },
    {
      "epoch": 1.9222380275011854,
      "grad_norm": 0.3231541173926272,
      "learning_rate": 3.439170465651104e-06,
      "loss": 0.0312,
      "step": 2027
    },
    {
      "epoch": 1.9231863442389758,
      "grad_norm": 0.38583478675562105,
      "learning_rate": 3.433926076083774e-06,
      "loss": 0.0385,
      "step": 2028
    },
    {
      "epoch": 1.9241346609767662,
      "grad_norm": 0.36409102702107604,
      "learning_rate": 3.4286835961384853e-06,
      "loss": 0.0315,
      "step": 2029
    },
    {
      "epoch": 1.9250829777145566,
      "grad_norm": 0.36312525560443826,
      "learning_rate": 3.4234430322077517e-06,
      "loss": 0.0253,
      "step": 2030
    },
    {
      "epoch": 1.926031294452347,
      "grad_norm": 0.33289279097003616,
      "learning_rate": 3.418204390681754e-06,
      "loss": 0.0256,
      "step": 2031
    },
    {
      "epoch": 1.9269796111901374,
      "grad_norm": 0.5320939559829481,
      "learning_rate": 3.4129676779483344e-06,
      "loss": 0.0281,
      "step": 2032
    },
    {
      "epoch": 1.9279279279279278,
      "grad_norm": 0.45005546438627286,
      "learning_rate": 3.407732900392973e-06,
      "loss": 0.0416,
      "step": 2033
    },
    {
      "epoch": 1.9288762446657184,
      "grad_norm": 0.409851104412246,
      "learning_rate": 3.402500064398798e-06,
      "loss": 0.0315,
      "step": 2034
    },
    {
      "epoch": 1.9298245614035088,
      "grad_norm": 0.3697105553298173,
      "learning_rate": 3.397269176346566e-06,
      "loss": 0.0235,
      "step": 2035
    },
    {
      "epoch": 1.9307728781412992,
      "grad_norm": 0.3938602684566997,
      "learning_rate": 3.3920402426146613e-06,
      "loss": 0.0307,
      "step": 2036
    },
    {
      "epoch": 1.9317211948790898,
      "grad_norm": 0.45578802286334,
      "learning_rate": 3.3868132695790856e-06,
      "loss": 0.0374,
      "step": 2037
    },
    {
      "epoch": 1.9326695116168802,
      "grad_norm": 0.36256121198136504,
      "learning_rate": 3.3815882636134423e-06,
      "loss": 0.033,
      "step": 2038
    },
    {
      "epoch": 1.9336178283546706,
      "grad_norm": 0.9568492163330367,
      "learning_rate": 3.3763652310889454e-06,
      "loss": 0.0425,
      "step": 2039
    },
    {
      "epoch": 1.934566145092461,
      "grad_norm": 0.3815862735223731,
      "learning_rate": 3.3711441783744014e-06,
      "loss": 0.0258,
      "step": 2040
    },
    {
      "epoch": 1.9355144618302513,
      "grad_norm": 0.5667383697238997,
      "learning_rate": 3.3659251118361955e-06,
      "loss": 0.0303,
      "step": 2041
    },
    {
      "epoch": 1.9364627785680417,
      "grad_norm": 0.6867359992131764,
      "learning_rate": 3.3607080378383006e-06,
      "loss": 0.0478,
      "step": 2042
    },
    {
      "epoch": 1.937411095305832,
      "grad_norm": 0.4296690617166379,
      "learning_rate": 3.3554929627422494e-06,
      "loss": 0.0333,
      "step": 2043
    },
    {
      "epoch": 1.9383594120436225,
      "grad_norm": 0.5162976443262678,
      "learning_rate": 3.3502798929071454e-06,
      "loss": 0.0554,
      "step": 2044
    },
    {
      "epoch": 1.9393077287814129,
      "grad_norm": 0.3368621100557327,
      "learning_rate": 3.345068834689649e-06,
      "loss": 0.0324,
      "step": 2045
    },
    {
      "epoch": 1.9402560455192033,
      "grad_norm": 0.37365269979182164,
      "learning_rate": 3.3398597944439554e-06,
      "loss": 0.0338,
      "step": 2046
    },
    {
      "epoch": 1.9412043622569939,
      "grad_norm": 0.5681446998189141,
      "learning_rate": 3.334652778521813e-06,
      "loss": 0.0411,
      "step": 2047
    },
    {
      "epoch": 1.9421526789947843,
      "grad_norm": 0.4740696492249626,
      "learning_rate": 3.3294477932724946e-06,
      "loss": 0.045,
      "step": 2048
    },
    {
      "epoch": 1.9431009957325747,
      "grad_norm": 0.3608937786548174,
      "learning_rate": 3.3242448450427945e-06,
      "loss": 0.0332,
      "step": 2049
    },
    {
      "epoch": 1.944049312470365,
      "grad_norm": 0.38506053572775895,
      "learning_rate": 3.319043940177031e-06,
      "loss": 0.028,
      "step": 2050
    },
    {
      "epoch": 1.9449976292081557,
      "grad_norm": 0.5698067483625857,
      "learning_rate": 3.3138450850170227e-06,
      "loss": 0.041,
      "step": 2051
    },
    {
      "epoch": 1.945945945945946,
      "grad_norm": 0.4420338241665105,
      "learning_rate": 3.3086482859020957e-06,
      "loss": 0.0373,
      "step": 2052
    },
    {
      "epoch": 1.9468942626837364,
      "grad_norm": 0.4953262443173467,
      "learning_rate": 3.3034535491690654e-06,
      "loss": 0.051,
      "step": 2053
    },
    {
      "epoch": 1.9478425794215268,
      "grad_norm": 0.4852410535068405,
      "learning_rate": 3.2982608811522306e-06,
      "loss": 0.0274,
      "step": 2054
    },
    {
      "epoch": 1.9487908961593172,
      "grad_norm": 0.6029040593040903,
      "learning_rate": 3.293070288183371e-06,
      "loss": 0.0532,
      "step": 2055
    },
    {
      "epoch": 1.9497392128971076,
      "grad_norm": 0.45973934841106084,
      "learning_rate": 3.287881776591737e-06,
      "loss": 0.0405,
      "step": 2056
    },
    {
      "epoch": 1.950687529634898,
      "grad_norm": 0.7309917431565905,
      "learning_rate": 3.282695352704036e-06,
      "loss": 0.0426,
      "step": 2057
    },
    {
      "epoch": 1.9516358463726884,
      "grad_norm": 0.33625680044079603,
      "learning_rate": 3.277511022844436e-06,
      "loss": 0.028,
      "step": 2058
    },
    {
      "epoch": 1.9525841631104788,
      "grad_norm": 0.8671444099283062,
      "learning_rate": 3.2723287933345426e-06,
      "loss": 0.0417,
      "step": 2059
    },
    {
      "epoch": 1.9535324798482692,
      "grad_norm": 0.4342735760080419,
      "learning_rate": 3.267148670493411e-06,
      "loss": 0.0348,
      "step": 2060
    },
    {
      "epoch": 1.9544807965860598,
      "grad_norm": 0.35268469905572575,
      "learning_rate": 3.2619706606375216e-06,
      "loss": 0.027,
      "step": 2061
    },
    {
      "epoch": 1.9554291133238502,
      "grad_norm": 0.3845496094181825,
      "learning_rate": 3.256794770080778e-06,
      "loss": 0.0318,
      "step": 2062
    },
    {
      "epoch": 1.9563774300616406,
      "grad_norm": 0.38361703791786117,
      "learning_rate": 3.2516210051345023e-06,
      "loss": 0.0371,
      "step": 2063
    },
    {
      "epoch": 1.9573257467994312,
      "grad_norm": 0.563671322887721,
      "learning_rate": 3.246449372107422e-06,
      "loss": 0.0362,
      "step": 2064
    },
    {
      "epoch": 1.9582740635372216,
      "grad_norm": 0.42306475584040243,
      "learning_rate": 3.2412798773056653e-06,
      "loss": 0.0339,
      "step": 2065
    },
    {
      "epoch": 1.959222380275012,
      "grad_norm": 0.40482383621032814,
      "learning_rate": 3.2361125270327578e-06,
      "loss": 0.0347,
      "step": 2066
    },
    {
      "epoch": 1.9601706970128023,
      "grad_norm": 0.44088167927957383,
      "learning_rate": 3.230947327589602e-06,
      "loss": 0.0384,
      "step": 2067
    },
    {
      "epoch": 1.9611190137505927,
      "grad_norm": 0.37961080268463626,
      "learning_rate": 3.2257842852744846e-06,
      "loss": 0.0228,
      "step": 2068
    },
    {
      "epoch": 1.962067330488383,
      "grad_norm": 0.4444789473749266,
      "learning_rate": 3.220623406383061e-06,
      "loss": 0.0449,
      "step": 2069
    },
    {
      "epoch": 1.9630156472261735,
      "grad_norm": 0.6015703335784637,
      "learning_rate": 3.2154646972083427e-06,
      "loss": 0.0595,
      "step": 2070
    },
    {
      "epoch": 1.9639639639639639,
      "grad_norm": 0.4963402770212196,
      "learning_rate": 3.210308164040704e-06,
      "loss": 0.0436,
      "step": 2071
    },
    {
      "epoch": 1.9649122807017543,
      "grad_norm": 0.3403181961401529,
      "learning_rate": 3.2051538131678585e-06,
      "loss": 0.036,
      "step": 2072
    },
    {
      "epoch": 1.9658605974395447,
      "grad_norm": 0.5178795816060555,
      "learning_rate": 3.200001650874863e-06,
      "loss": 0.0287,
      "step": 2073
    },
    {
      "epoch": 1.9668089141773353,
      "grad_norm": 0.2657038221760043,
      "learning_rate": 3.194851683444107e-06,
      "loss": 0.023,
      "step": 2074
    },
    {
      "epoch": 1.9677572309151257,
      "grad_norm": 0.36414575441011054,
      "learning_rate": 3.189703917155299e-06,
      "loss": 0.0256,
      "step": 2075
    },
    {
      "epoch": 1.968705547652916,
      "grad_norm": 0.399014676654186,
      "learning_rate": 3.184558358285465e-06,
      "loss": 0.0341,
      "step": 2076
    },
    {
      "epoch": 1.9696538643907064,
      "grad_norm": 0.48424044481722756,
      "learning_rate": 3.1794150131089434e-06,
      "loss": 0.0322,
      "step": 2077
    },
    {
      "epoch": 1.970602181128497,
      "grad_norm": 0.36430947108225803,
      "learning_rate": 3.1742738878973654e-06,
      "loss": 0.0355,
      "step": 2078
    },
    {
      "epoch": 1.9715504978662874,
      "grad_norm": 0.44863405179866295,
      "learning_rate": 3.1691349889196653e-06,
      "loss": 0.0438,
      "step": 2079
    },
    {
      "epoch": 1.9724988146040778,
      "grad_norm": 0.7130335206259649,
      "learning_rate": 3.1639983224420535e-06,
      "loss": 0.0378,
      "step": 2080
    },
    {
      "epoch": 1.9734471313418682,
      "grad_norm": 0.4430950820041455,
      "learning_rate": 3.1588638947280215e-06,
      "loss": 0.0365,
      "step": 2081
    },
    {
      "epoch": 1.9743954480796586,
      "grad_norm": 0.32869156151115164,
      "learning_rate": 3.153731712038335e-06,
      "loss": 0.0211,
      "step": 2082
    },
    {
      "epoch": 1.975343764817449,
      "grad_norm": 0.4257317909820218,
      "learning_rate": 3.1486017806310145e-06,
      "loss": 0.0386,
      "step": 2083
    },
    {
      "epoch": 1.9762920815552394,
      "grad_norm": 1.440703206043199,
      "learning_rate": 3.143474106761343e-06,
      "loss": 0.028,
      "step": 2084
    },
    {
      "epoch": 1.9772403982930298,
      "grad_norm": 0.39647997595721773,
      "learning_rate": 3.138348696681846e-06,
      "loss": 0.031,
      "step": 2085
    },
    {
      "epoch": 1.9781887150308202,
      "grad_norm": 0.394821466886043,
      "learning_rate": 3.1332255566422876e-06,
      "loss": 0.0324,
      "step": 2086
    },
    {
      "epoch": 1.9791370317686106,
      "grad_norm": 0.3544381757947608,
      "learning_rate": 3.1281046928896703e-06,
      "loss": 0.0265,
      "step": 2087
    },
    {
      "epoch": 1.9800853485064012,
      "grad_norm": 0.4768128542585291,
      "learning_rate": 3.1229861116682126e-06,
      "loss": 0.0366,
      "step": 2088
    },
    {
      "epoch": 1.9810336652441916,
      "grad_norm": 0.7243948999698084,
      "learning_rate": 3.1178698192193563e-06,
      "loss": 0.0479,
      "step": 2089
    },
    {
      "epoch": 1.981981981981982,
      "grad_norm": 0.4524011909213368,
      "learning_rate": 3.11275582178175e-06,
      "loss": 0.0313,
      "step": 2090
    },
    {
      "epoch": 1.9829302987197726,
      "grad_norm": 0.5179468508049,
      "learning_rate": 3.1076441255912425e-06,
      "loss": 0.0332,
      "step": 2091
    },
    {
      "epoch": 1.983878615457563,
      "grad_norm": 0.6765627732053985,
      "learning_rate": 3.102534736880878e-06,
      "loss": 0.0425,
      "step": 2092
    },
    {
      "epoch": 1.9848269321953533,
      "grad_norm": 1.035454737611202,
      "learning_rate": 3.097427661880889e-06,
      "loss": 0.0288,
      "step": 2093
    },
    {
      "epoch": 1.9857752489331437,
      "grad_norm": 0.4638694830728162,
      "learning_rate": 3.092322906818681e-06,
      "loss": 0.0379,
      "step": 2094
    },
    {
      "epoch": 1.986723565670934,
      "grad_norm": 0.4542609073123639,
      "learning_rate": 3.087220477918839e-06,
      "loss": 0.0433,
      "step": 2095
    },
    {
      "epoch": 1.9876718824087245,
      "grad_norm": 0.3059593385789022,
      "learning_rate": 3.082120381403102e-06,
      "loss": 0.0263,
      "step": 2096
    },
    {
      "epoch": 1.9886201991465149,
      "grad_norm": 0.8343736689898676,
      "learning_rate": 3.077022623490371e-06,
      "loss": 0.0445,
      "step": 2097
    },
    {
      "epoch": 1.9895685158843053,
      "grad_norm": 0.3930889133833989,
      "learning_rate": 3.0719272103966964e-06,
      "loss": 0.0316,
      "step": 2098
    },
    {
      "epoch": 1.9905168326220957,
      "grad_norm": 0.6283499649646375,
      "learning_rate": 3.066834148335264e-06,
      "loss": 0.043,
      "step": 2099
    },
    {
      "epoch": 1.991465149359886,
      "grad_norm": 0.3334337011671591,
      "learning_rate": 3.0617434435163983e-06,
      "loss": 0.0221,
      "step": 2100
    },
    {
      "epoch": 1.9924134660976767,
      "grad_norm": 0.44480723283000395,
      "learning_rate": 3.0566551021475467e-06,
      "loss": 0.0247,
      "step": 2101
    },
    {
      "epoch": 1.993361782835467,
      "grad_norm": 0.42335568280839386,
      "learning_rate": 3.0515691304332728e-06,
      "loss": 0.0342,
      "step": 2102
    },
    {
      "epoch": 1.9943100995732574,
      "grad_norm": 0.39275472992791943,
      "learning_rate": 3.0464855345752576e-06,
      "loss": 0.0258,
      "step": 2103
    },
    {
      "epoch": 1.995258416311048,
      "grad_norm": 0.4252459704357132,
      "learning_rate": 3.041404320772275e-06,
      "loss": 0.0395,
      "step": 2104
    },
    {
      "epoch": 1.9962067330488384,
      "grad_norm": 0.44204282810063367,
      "learning_rate": 3.036325495220202e-06,
      "loss": 0.03,
      "step": 2105
    },
    {
      "epoch": 1.9971550497866288,
      "grad_norm": 0.5267895944658703,
      "learning_rate": 3.031249064112004e-06,
      "loss": 0.0336,
      "step": 2106
    },
    {
      "epoch": 1.9981033665244192,
      "grad_norm": 0.7817180674735154,
      "learning_rate": 3.0261750336377204e-06,
      "loss": 0.0433,
      "step": 2107
    },
    {
      "epoch": 1.9990516832622096,
      "grad_norm": 0.39050551405758693,
      "learning_rate": 3.0211034099844667e-06,
      "loss": 0.0286,
      "step": 2108
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.411241525627785,
      "learning_rate": 3.0160341993364267e-06,
      "loss": 0.0272,
      "step": 2109
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.03899507224559784,
      "eval_runtime": 200.9629,
      "eval_samples_per_second": 35.345,
      "eval_steps_per_second": 1.105,
      "step": 2109
    },
    {
      "epoch": 2.0009483167377904,
      "grad_norm": 0.42867573482523097,
      "learning_rate": 3.010967407874835e-06,
      "loss": 0.0305,
      "step": 2110
    },
    {
      "epoch": 2.0018966334755808,
      "grad_norm": 0.7670970988426332,
      "learning_rate": 3.0059030417779843e-06,
      "loss": 0.0431,
      "step": 2111
    },
    {
      "epoch": 2.002844950213371,
      "grad_norm": 0.2967349239255996,
      "learning_rate": 3.000841107221203e-06,
      "loss": 0.0174,
      "step": 2112
    },
    {
      "epoch": 2.0037932669511616,
      "grad_norm": 0.35929335388522343,
      "learning_rate": 2.995781610376856e-06,
      "loss": 0.0271,
      "step": 2113
    },
    {
      "epoch": 2.004741583688952,
      "grad_norm": 0.31468404701781494,
      "learning_rate": 2.990724557414342e-06,
      "loss": 0.0239,
      "step": 2114
    },
    {
      "epoch": 2.0056899004267423,
      "grad_norm": 0.4799589378363981,
      "learning_rate": 2.985669954500069e-06,
      "loss": 0.0253,
      "step": 2115
    },
    {
      "epoch": 2.006638217164533,
      "grad_norm": 0.3826498293765311,
      "learning_rate": 2.980617807797467e-06,
      "loss": 0.0258,
      "step": 2116
    },
    {
      "epoch": 2.0075865339023236,
      "grad_norm": 0.40695416013123065,
      "learning_rate": 2.9755681234669663e-06,
      "loss": 0.0317,
      "step": 2117
    },
    {
      "epoch": 2.008534850640114,
      "grad_norm": 0.7967953274247074,
      "learning_rate": 2.9705209076659957e-06,
      "loss": 0.0221,
      "step": 2118
    },
    {
      "epoch": 2.0094831673779043,
      "grad_norm": 0.4552358236810522,
      "learning_rate": 2.965476166548974e-06,
      "loss": 0.0358,
      "step": 2119
    },
    {
      "epoch": 2.0104314841156947,
      "grad_norm": 0.28631994213003603,
      "learning_rate": 2.9604339062673003e-06,
      "loss": 0.0175,
      "step": 2120
    },
    {
      "epoch": 2.011379800853485,
      "grad_norm": 0.32057860078911904,
      "learning_rate": 2.9553941329693526e-06,
      "loss": 0.0178,
      "step": 2121
    },
    {
      "epoch": 2.0123281175912755,
      "grad_norm": 0.3624106623635382,
      "learning_rate": 2.9503568528004757e-06,
      "loss": 0.0201,
      "step": 2122
    },
    {
      "epoch": 2.013276434329066,
      "grad_norm": 0.4636739823079543,
      "learning_rate": 2.9453220719029694e-06,
      "loss": 0.0387,
      "step": 2123
    },
    {
      "epoch": 2.0142247510668563,
      "grad_norm": 0.48994564694002685,
      "learning_rate": 2.940289796416094e-06,
      "loss": 0.0295,
      "step": 2124
    },
    {
      "epoch": 2.0151730678046467,
      "grad_norm": 0.4386644278170872,
      "learning_rate": 2.935260032476045e-06,
      "loss": 0.0268,
      "step": 2125
    },
    {
      "epoch": 2.016121384542437,
      "grad_norm": 0.3758526320713401,
      "learning_rate": 2.9302327862159634e-06,
      "loss": 0.0175,
      "step": 2126
    },
    {
      "epoch": 2.0170697012802274,
      "grad_norm": 0.6491230889815814,
      "learning_rate": 2.9252080637659203e-06,
      "loss": 0.0248,
      "step": 2127
    },
    {
      "epoch": 2.018018018018018,
      "grad_norm": 0.39148598510057125,
      "learning_rate": 2.920185871252903e-06,
      "loss": 0.0191,
      "step": 2128
    },
    {
      "epoch": 2.0189663347558082,
      "grad_norm": 0.38730619737964056,
      "learning_rate": 2.9151662148008177e-06,
      "loss": 0.0206,
      "step": 2129
    },
    {
      "epoch": 2.019914651493599,
      "grad_norm": 0.2739978249686544,
      "learning_rate": 2.9101491005304803e-06,
      "loss": 0.0208,
      "step": 2130
    },
    {
      "epoch": 2.0208629682313894,
      "grad_norm": 0.2492924713448619,
      "learning_rate": 2.905134534559602e-06,
      "loss": 0.0182,
      "step": 2131
    },
    {
      "epoch": 2.02181128496918,
      "grad_norm": 0.3670158300028282,
      "learning_rate": 2.9001225230027923e-06,
      "loss": 0.0258,
      "step": 2132
    },
    {
      "epoch": 2.0227596017069702,
      "grad_norm": 0.5687471196705789,
      "learning_rate": 2.8951130719715393e-06,
      "loss": 0.0283,
      "step": 2133
    },
    {
      "epoch": 2.0237079184447606,
      "grad_norm": 0.4512446574071027,
      "learning_rate": 2.890106187574213e-06,
      "loss": 0.0369,
      "step": 2134
    },
    {
      "epoch": 2.024656235182551,
      "grad_norm": 0.302480559027787,
      "learning_rate": 2.8851018759160575e-06,
      "loss": 0.0188,
      "step": 2135
    },
    {
      "epoch": 2.0256045519203414,
      "grad_norm": 0.37270010872170284,
      "learning_rate": 2.8801001430991714e-06,
      "loss": 0.0231,
      "step": 2136
    },
    {
      "epoch": 2.026552868658132,
      "grad_norm": 0.37835817327163923,
      "learning_rate": 2.875100995222512e-06,
      "loss": 0.024,
      "step": 2137
    },
    {
      "epoch": 2.027501185395922,
      "grad_norm": 0.33368312776065706,
      "learning_rate": 2.8701044383818887e-06,
      "loss": 0.0151,
      "step": 2138
    },
    {
      "epoch": 2.0284495021337126,
      "grad_norm": 0.3635689040002093,
      "learning_rate": 2.8651104786699446e-06,
      "loss": 0.0207,
      "step": 2139
    },
    {
      "epoch": 2.029397818871503,
      "grad_norm": 0.33610953009892625,
      "learning_rate": 2.860119122176164e-06,
      "loss": 0.0165,
      "step": 2140
    },
    {
      "epoch": 2.0303461356092933,
      "grad_norm": 0.4536480112083883,
      "learning_rate": 2.855130374986847e-06,
      "loss": 0.0301,
      "step": 2141
    },
    {
      "epoch": 2.0312944523470837,
      "grad_norm": 0.4862579959409953,
      "learning_rate": 2.85014424318512e-06,
      "loss": 0.025,
      "step": 2142
    },
    {
      "epoch": 2.0322427690848746,
      "grad_norm": 0.37646189979934114,
      "learning_rate": 2.8451607328509206e-06,
      "loss": 0.019,
      "step": 2143
    },
    {
      "epoch": 2.033191085822665,
      "grad_norm": 0.4658791421599927,
      "learning_rate": 2.8401798500609825e-06,
      "loss": 0.0242,
      "step": 2144
    },
    {
      "epoch": 2.0341394025604553,
      "grad_norm": 0.3898784281727516,
      "learning_rate": 2.8352016008888407e-06,
      "loss": 0.0257,
      "step": 2145
    },
    {
      "epoch": 2.0350877192982457,
      "grad_norm": 0.3643245441073441,
      "learning_rate": 2.830225991404821e-06,
      "loss": 0.0187,
      "step": 2146
    },
    {
      "epoch": 2.036036036036036,
      "grad_norm": 0.5491866993620294,
      "learning_rate": 2.825253027676026e-06,
      "loss": 0.0337,
      "step": 2147
    },
    {
      "epoch": 2.0369843527738265,
      "grad_norm": 0.35627319428709453,
      "learning_rate": 2.8202827157663336e-06,
      "loss": 0.0238,
      "step": 2148
    },
    {
      "epoch": 2.037932669511617,
      "grad_norm": 0.40610530442283266,
      "learning_rate": 2.815315061736385e-06,
      "loss": 0.0294,
      "step": 2149
    },
    {
      "epoch": 2.0388809862494073,
      "grad_norm": 0.7297441686133572,
      "learning_rate": 2.8103500716435867e-06,
      "loss": 0.0519,
      "step": 2150
    },
    {
      "epoch": 2.0398293029871977,
      "grad_norm": 0.4458875962553178,
      "learning_rate": 2.805387751542095e-06,
      "loss": 0.0315,
      "step": 2151
    },
    {
      "epoch": 2.040777619724988,
      "grad_norm": 0.41848406140264266,
      "learning_rate": 2.8004281074828066e-06,
      "loss": 0.0215,
      "step": 2152
    },
    {
      "epoch": 2.0417259364627784,
      "grad_norm": 0.5998930027439111,
      "learning_rate": 2.7954711455133587e-06,
      "loss": 0.0261,
      "step": 2153
    },
    {
      "epoch": 2.042674253200569,
      "grad_norm": 0.3643955841072442,
      "learning_rate": 2.7905168716781207e-06,
      "loss": 0.0189,
      "step": 2154
    },
    {
      "epoch": 2.0436225699383592,
      "grad_norm": 0.3851753580513905,
      "learning_rate": 2.7855652920181743e-06,
      "loss": 0.0263,
      "step": 2155
    },
    {
      "epoch": 2.0445708866761496,
      "grad_norm": 0.35889957914574705,
      "learning_rate": 2.7806164125713287e-06,
      "loss": 0.0249,
      "step": 2156
    },
    {
      "epoch": 2.0455192034139404,
      "grad_norm": 0.45717380785394435,
      "learning_rate": 2.775670239372088e-06,
      "loss": 0.0231,
      "step": 2157
    },
    {
      "epoch": 2.046467520151731,
      "grad_norm": 0.45034784091600044,
      "learning_rate": 2.770726778451669e-06,
      "loss": 0.0238,
      "step": 2158
    },
    {
      "epoch": 2.0474158368895212,
      "grad_norm": 0.40079641133062344,
      "learning_rate": 2.765786035837971e-06,
      "loss": 0.0222,
      "step": 2159
    },
    {
      "epoch": 2.0483641536273116,
      "grad_norm": 0.35961652919739295,
      "learning_rate": 2.760848017555582e-06,
      "loss": 0.0248,
      "step": 2160
    },
    {
      "epoch": 2.049312470365102,
      "grad_norm": 0.29591792109816933,
      "learning_rate": 2.7559127296257694e-06,
      "loss": 0.0151,
      "step": 2161
    },
    {
      "epoch": 2.0502607871028924,
      "grad_norm": 0.34278967001786625,
      "learning_rate": 2.7509801780664725e-06,
      "loss": 0.017,
      "step": 2162
    },
    {
      "epoch": 2.051209103840683,
      "grad_norm": 0.6265169417466546,
      "learning_rate": 2.746050368892288e-06,
      "loss": 0.037,
      "step": 2163
    },
    {
      "epoch": 2.052157420578473,
      "grad_norm": 0.4978111695622906,
      "learning_rate": 2.741123308114477e-06,
      "loss": 0.0269,
      "step": 2164
    },
    {
      "epoch": 2.0531057373162636,
      "grad_norm": 0.3482001539800016,
      "learning_rate": 2.7361990017409406e-06,
      "loss": 0.0193,
      "step": 2165
    },
    {
      "epoch": 2.054054054054054,
      "grad_norm": 0.5690545161609503,
      "learning_rate": 2.7312774557762274e-06,
      "loss": 0.0371,
      "step": 2166
    },
    {
      "epoch": 2.0550023707918443,
      "grad_norm": 0.40646429616659996,
      "learning_rate": 2.7263586762215197e-06,
      "loss": 0.0205,
      "step": 2167
    },
    {
      "epoch": 2.0559506875296347,
      "grad_norm": 0.33861620939002895,
      "learning_rate": 2.721442669074622e-06,
      "loss": 0.0178,
      "step": 2168
    },
    {
      "epoch": 2.056899004267425,
      "grad_norm": 0.4037345819140745,
      "learning_rate": 2.716529440329965e-06,
      "loss": 0.0289,
      "step": 2169
    },
    {
      "epoch": 2.057847321005216,
      "grad_norm": 0.482627797833161,
      "learning_rate": 2.711618995978584e-06,
      "loss": 0.0352,
      "step": 2170
    },
    {
      "epoch": 2.0587956377430063,
      "grad_norm": 0.5082804569426425,
      "learning_rate": 2.7067113420081227e-06,
      "loss": 0.0349,
      "step": 2171
    },
    {
      "epoch": 2.0597439544807967,
      "grad_norm": 0.38145913486228483,
      "learning_rate": 2.7018064844028247e-06,
      "loss": 0.0251,
      "step": 2172
    },
    {
      "epoch": 2.060692271218587,
      "grad_norm": 0.5841097247038757,
      "learning_rate": 2.6969044291435194e-06,
      "loss": 0.0331,
      "step": 2173
    },
    {
      "epoch": 2.0616405879563775,
      "grad_norm": 0.41296435259903375,
      "learning_rate": 2.6920051822076215e-06,
      "loss": 0.0244,
      "step": 2174
    },
    {
      "epoch": 2.062588904694168,
      "grad_norm": 0.4727942470216596,
      "learning_rate": 2.687108749569124e-06,
      "loss": 0.0251,
      "step": 2175
    },
    {
      "epoch": 2.0635372214319583,
      "grad_norm": 0.43281565858200033,
      "learning_rate": 2.6822151371985806e-06,
      "loss": 0.0362,
      "step": 2176
    },
    {
      "epoch": 2.0644855381697487,
      "grad_norm": 0.7073861164110138,
      "learning_rate": 2.6773243510631147e-06,
      "loss": 0.0283,
      "step": 2177
    },
    {
      "epoch": 2.065433854907539,
      "grad_norm": 0.3892763301096195,
      "learning_rate": 2.6724363971263958e-06,
      "loss": 0.0277,
      "step": 2178
    },
    {
      "epoch": 2.0663821716453294,
      "grad_norm": 0.6146711615193001,
      "learning_rate": 2.667551281348647e-06,
      "loss": 0.0278,
      "step": 2179
    },
    {
      "epoch": 2.06733048838312,
      "grad_norm": 0.6424481373001304,
      "learning_rate": 2.6626690096866236e-06,
      "loss": 0.0226,
      "step": 2180
    },
    {
      "epoch": 2.0682788051209102,
      "grad_norm": 0.5062917952323074,
      "learning_rate": 2.657789588093621e-06,
      "loss": 0.0277,
      "step": 2181
    },
    {
      "epoch": 2.0692271218587006,
      "grad_norm": 0.4031910900078797,
      "learning_rate": 2.6529130225194494e-06,
      "loss": 0.0236,
      "step": 2182
    },
    {
      "epoch": 2.0701754385964914,
      "grad_norm": 0.3696551665775792,
      "learning_rate": 2.648039318910447e-06,
      "loss": 0.0241,
      "step": 2183
    },
    {
      "epoch": 2.071123755334282,
      "grad_norm": 0.4994149517651512,
      "learning_rate": 2.643168483209452e-06,
      "loss": 0.0373,
      "step": 2184
    },
    {
      "epoch": 2.0720720720720722,
      "grad_norm": 0.3435127049249243,
      "learning_rate": 2.6383005213558156e-06,
      "loss": 0.0202,
      "step": 2185
    },
    {
      "epoch": 2.0730203888098626,
      "grad_norm": 0.3550421787961135,
      "learning_rate": 2.6334354392853755e-06,
      "loss": 0.0307,
      "step": 2186
    },
    {
      "epoch": 2.073968705547653,
      "grad_norm": 0.5367065919243836,
      "learning_rate": 2.6285732429304634e-06,
      "loss": 0.0325,
      "step": 2187
    },
    {
      "epoch": 2.0749170222854434,
      "grad_norm": 0.41551554257616574,
      "learning_rate": 2.623713938219894e-06,
      "loss": 0.0194,
      "step": 2188
    },
    {
      "epoch": 2.075865339023234,
      "grad_norm": 0.42065530424138786,
      "learning_rate": 2.6188575310789475e-06,
      "loss": 0.0268,
      "step": 2189
    },
    {
      "epoch": 2.076813655761024,
      "grad_norm": 0.6462822911603555,
      "learning_rate": 2.614004027429382e-06,
      "loss": 0.0385,
      "step": 2190
    },
    {
      "epoch": 2.0777619724988146,
      "grad_norm": 0.36529894157157955,
      "learning_rate": 2.609153433189406e-06,
      "loss": 0.0169,
      "step": 2191
    },
    {
      "epoch": 2.078710289236605,
      "grad_norm": 0.4317237150788793,
      "learning_rate": 2.6043057542736837e-06,
      "loss": 0.0268,
      "step": 2192
    },
    {
      "epoch": 2.0796586059743953,
      "grad_norm": 0.42998002758472326,
      "learning_rate": 2.599460996593327e-06,
      "loss": 0.0308,
      "step": 2193
    },
    {
      "epoch": 2.0806069227121857,
      "grad_norm": 0.3558311679324967,
      "learning_rate": 2.5946191660558804e-06,
      "loss": 0.0214,
      "step": 2194
    },
    {
      "epoch": 2.081555239449976,
      "grad_norm": 0.38571064268038,
      "learning_rate": 2.5897802685653235e-06,
      "loss": 0.0226,
      "step": 2195
    },
    {
      "epoch": 2.0825035561877665,
      "grad_norm": 0.38651538906208016,
      "learning_rate": 2.58494431002206e-06,
      "loss": 0.0212,
      "step": 2196
    },
    {
      "epoch": 2.0834518729255573,
      "grad_norm": 0.4181252743146899,
      "learning_rate": 2.580111296322904e-06,
      "loss": 0.0226,
      "step": 2197
    },
    {
      "epoch": 2.0844001896633477,
      "grad_norm": 0.5094856971487342,
      "learning_rate": 2.5752812333610856e-06,
      "loss": 0.0296,
      "step": 2198
    },
    {
      "epoch": 2.085348506401138,
      "grad_norm": 0.40859568824515613,
      "learning_rate": 2.5704541270262347e-06,
      "loss": 0.0241,
      "step": 2199
    },
    {
      "epoch": 2.0862968231389285,
      "grad_norm": 0.4517293982946693,
      "learning_rate": 2.5656299832043718e-06,
      "loss": 0.0251,
      "step": 2200
    },
    {
      "epoch": 2.087245139876719,
      "grad_norm": 0.4318042476611322,
      "learning_rate": 2.5608088077779125e-06,
      "loss": 0.0227,
      "step": 2201
    },
    {
      "epoch": 2.0881934566145093,
      "grad_norm": 0.5689691059283708,
      "learning_rate": 2.5559906066256467e-06,
      "loss": 0.0311,
      "step": 2202
    },
    {
      "epoch": 2.0891417733522997,
      "grad_norm": 0.31693729502341195,
      "learning_rate": 2.551175385622737e-06,
      "loss": 0.0163,
      "step": 2203
    },
    {
      "epoch": 2.09009009009009,
      "grad_norm": 0.28795871774991333,
      "learning_rate": 2.546363150640721e-06,
      "loss": 0.0139,
      "step": 2204
    },
    {
      "epoch": 2.0910384068278804,
      "grad_norm": 0.6960867511827779,
      "learning_rate": 2.541553907547483e-06,
      "loss": 0.0435,
      "step": 2205
    },
    {
      "epoch": 2.091986723565671,
      "grad_norm": 0.31057178211753483,
      "learning_rate": 2.5367476622072674e-06,
      "loss": 0.0184,
      "step": 2206
    },
    {
      "epoch": 2.0929350403034612,
      "grad_norm": 0.39246559073720605,
      "learning_rate": 2.5319444204806624e-06,
      "loss": 0.0239,
      "step": 2207
    },
    {
      "epoch": 2.0938833570412516,
      "grad_norm": 0.3606580233864144,
      "learning_rate": 2.5271441882245896e-06,
      "loss": 0.0211,
      "step": 2208
    },
    {
      "epoch": 2.094831673779042,
      "grad_norm": 0.5082639846967826,
      "learning_rate": 2.5223469712923066e-06,
      "loss": 0.0301,
      "step": 2209
    },
    {
      "epoch": 2.095779990516833,
      "grad_norm": 0.3855278523669711,
      "learning_rate": 2.5175527755333874e-06,
      "loss": 0.0197,
      "step": 2210
    },
    {
      "epoch": 2.0967283072546232,
      "grad_norm": 0.433480401949784,
      "learning_rate": 2.5127616067937276e-06,
      "loss": 0.0243,
      "step": 2211
    },
    {
      "epoch": 2.0976766239924136,
      "grad_norm": 0.32045799775532713,
      "learning_rate": 2.507973470915532e-06,
      "loss": 0.0172,
      "step": 2212
    },
    {
      "epoch": 2.098624940730204,
      "grad_norm": 0.5026090192485358,
      "learning_rate": 2.503188373737304e-06,
      "loss": 0.0313,
      "step": 2213
    },
    {
      "epoch": 2.0995732574679944,
      "grad_norm": 0.5640464541480521,
      "learning_rate": 2.49840632109384e-06,
      "loss": 0.0303,
      "step": 2214
    },
    {
      "epoch": 2.100521574205785,
      "grad_norm": 0.41866789277310873,
      "learning_rate": 2.493627318816231e-06,
      "loss": 0.0188,
      "step": 2215
    },
    {
      "epoch": 2.101469890943575,
      "grad_norm": 0.6027998069884065,
      "learning_rate": 2.4888513727318405e-06,
      "loss": 0.0241,
      "step": 2216
    },
    {
      "epoch": 2.1024182076813656,
      "grad_norm": 0.4611166482064652,
      "learning_rate": 2.484078488664313e-06,
      "loss": 0.0218,
      "step": 2217
    },
    {
      "epoch": 2.103366524419156,
      "grad_norm": 0.3931582590198883,
      "learning_rate": 2.479308672433552e-06,
      "loss": 0.0203,
      "step": 2218
    },
    {
      "epoch": 2.1043148411569463,
      "grad_norm": 0.37466971883437067,
      "learning_rate": 2.474541929855725e-06,
      "loss": 0.0174,
      "step": 2219
    },
    {
      "epoch": 2.1052631578947367,
      "grad_norm": 0.45633595162427804,
      "learning_rate": 2.469778266743253e-06,
      "loss": 0.0314,
      "step": 2220
    },
    {
      "epoch": 2.106211474632527,
      "grad_norm": 0.3903053469088167,
      "learning_rate": 2.465017688904795e-06,
      "loss": 0.0176,
      "step": 2221
    },
    {
      "epoch": 2.1071597913703175,
      "grad_norm": 0.8643363208938867,
      "learning_rate": 2.460260202145256e-06,
      "loss": 0.0431,
      "step": 2222
    },
    {
      "epoch": 2.108108108108108,
      "grad_norm": 0.3287931874943448,
      "learning_rate": 2.4555058122657637e-06,
      "loss": 0.0174,
      "step": 2223
    },
    {
      "epoch": 2.1090564248458987,
      "grad_norm": 0.36189985329961605,
      "learning_rate": 2.4507545250636787e-06,
      "loss": 0.0203,
      "step": 2224
    },
    {
      "epoch": 2.110004741583689,
      "grad_norm": 0.37032833947211646,
      "learning_rate": 2.4460063463325713e-06,
      "loss": 0.0257,
      "step": 2225
    },
    {
      "epoch": 2.1109530583214795,
      "grad_norm": 0.3687022423175376,
      "learning_rate": 2.441261281862221e-06,
      "loss": 0.0195,
      "step": 2226
    },
    {
      "epoch": 2.11190137505927,
      "grad_norm": 0.43305034123331726,
      "learning_rate": 2.436519337438615e-06,
      "loss": 0.0336,
      "step": 2227
    },
    {
      "epoch": 2.1128496917970603,
      "grad_norm": 0.39966610064341257,
      "learning_rate": 2.431780518843935e-06,
      "loss": 0.0243,
      "step": 2228
    },
    {
      "epoch": 2.1137980085348507,
      "grad_norm": 0.31938860162562066,
      "learning_rate": 2.4270448318565455e-06,
      "loss": 0.0178,
      "step": 2229
    },
    {
      "epoch": 2.114746325272641,
      "grad_norm": 0.3541847991387582,
      "learning_rate": 2.422312282250999e-06,
      "loss": 0.0204,
      "step": 2230
    },
    {
      "epoch": 2.1156946420104314,
      "grad_norm": 0.629139027756549,
      "learning_rate": 2.4175828757980225e-06,
      "loss": 0.0329,
      "step": 2231
    },
    {
      "epoch": 2.116642958748222,
      "grad_norm": 0.44488333984325906,
      "learning_rate": 2.4128566182645023e-06,
      "loss": 0.0218,
      "step": 2232
    },
    {
      "epoch": 2.1175912754860122,
      "grad_norm": 0.44422645256345245,
      "learning_rate": 2.4081335154134956e-06,
      "loss": 0.023,
      "step": 2233
    },
    {
      "epoch": 2.1185395922238026,
      "grad_norm": 0.3902960872474805,
      "learning_rate": 2.4034135730042052e-06,
      "loss": 0.0275,
      "step": 2234
    },
    {
      "epoch": 2.119487908961593,
      "grad_norm": 0.5253829874658488,
      "learning_rate": 2.3986967967919807e-06,
      "loss": 0.0317,
      "step": 2235
    },
    {
      "epoch": 2.1204362256993834,
      "grad_norm": 0.43919978311796937,
      "learning_rate": 2.3939831925283176e-06,
      "loss": 0.0255,
      "step": 2236
    },
    {
      "epoch": 2.1213845424371742,
      "grad_norm": 0.42626314322476566,
      "learning_rate": 2.389272765960834e-06,
      "loss": 0.0248,
      "step": 2237
    },
    {
      "epoch": 2.1223328591749646,
      "grad_norm": 0.43952254077722236,
      "learning_rate": 2.384565522833282e-06,
      "loss": 0.0272,
      "step": 2238
    },
    {
      "epoch": 2.123281175912755,
      "grad_norm": 0.4951260444039173,
      "learning_rate": 2.3798614688855254e-06,
      "loss": 0.03,
      "step": 2239
    },
    {
      "epoch": 2.1242294926505454,
      "grad_norm": 0.3756698841381475,
      "learning_rate": 2.3751606098535435e-06,
      "loss": 0.024,
      "step": 2240
    },
    {
      "epoch": 2.125177809388336,
      "grad_norm": 0.2861348407988811,
      "learning_rate": 2.370462951469419e-06,
      "loss": 0.0155,
      "step": 2241
    },
    {
      "epoch": 2.126126126126126,
      "grad_norm": 0.4876169224219029,
      "learning_rate": 2.3657684994613285e-06,
      "loss": 0.0275,
      "step": 2242
    },
    {
      "epoch": 2.1270744428639166,
      "grad_norm": 0.4319771129151901,
      "learning_rate": 2.3610772595535423e-06,
      "loss": 0.0193,
      "step": 2243
    },
    {
      "epoch": 2.128022759601707,
      "grad_norm": 0.40162005659288424,
      "learning_rate": 2.356389237466416e-06,
      "loss": 0.0184,
      "step": 2244
    },
    {
      "epoch": 2.1289710763394973,
      "grad_norm": 0.5033492435726082,
      "learning_rate": 2.351704438916376e-06,
      "loss": 0.0256,
      "step": 2245
    },
    {
      "epoch": 2.1299193930772877,
      "grad_norm": 0.39876138394841576,
      "learning_rate": 2.3470228696159196e-06,
      "loss": 0.0209,
      "step": 2246
    },
    {
      "epoch": 2.130867709815078,
      "grad_norm": 0.5019158990595434,
      "learning_rate": 2.342344535273608e-06,
      "loss": 0.0274,
      "step": 2247
    },
    {
      "epoch": 2.1318160265528685,
      "grad_norm": 0.35799980319498204,
      "learning_rate": 2.3376694415940565e-06,
      "loss": 0.0226,
      "step": 2248
    },
    {
      "epoch": 2.132764343290659,
      "grad_norm": 0.45079273257538477,
      "learning_rate": 2.332997594277933e-06,
      "loss": 0.0301,
      "step": 2249
    },
    {
      "epoch": 2.1337126600284497,
      "grad_norm": 0.32674403307718536,
      "learning_rate": 2.3283289990219395e-06,
      "loss": 0.0212,
      "step": 2250
    },
    {
      "epoch": 2.13466097676624,
      "grad_norm": 0.5296184962800978,
      "learning_rate": 2.3236636615188175e-06,
      "loss": 0.0317,
      "step": 2251
    },
    {
      "epoch": 2.1356092935040305,
      "grad_norm": 0.3754232671858479,
      "learning_rate": 2.3190015874573373e-06,
      "loss": 0.0229,
      "step": 2252
    },
    {
      "epoch": 2.136557610241821,
      "grad_norm": 0.3656558614584108,
      "learning_rate": 2.3143427825222847e-06,
      "loss": 0.0219,
      "step": 2253
    },
    {
      "epoch": 2.1375059269796113,
      "grad_norm": 0.5213276424505731,
      "learning_rate": 2.3096872523944654e-06,
      "loss": 0.0343,
      "step": 2254
    },
    {
      "epoch": 2.1384542437174017,
      "grad_norm": 0.34243403378994786,
      "learning_rate": 2.305035002750684e-06,
      "loss": 0.0179,
      "step": 2255
    },
    {
      "epoch": 2.139402560455192,
      "grad_norm": 0.3976411991318559,
      "learning_rate": 2.3003860392637545e-06,
      "loss": 0.0184,
      "step": 2256
    },
    {
      "epoch": 2.1403508771929824,
      "grad_norm": 0.4602750365294984,
      "learning_rate": 2.2957403676024766e-06,
      "loss": 0.0224,
      "step": 2257
    },
    {
      "epoch": 2.141299193930773,
      "grad_norm": 0.9716168085000075,
      "learning_rate": 2.2910979934316368e-06,
      "loss": 0.0309,
      "step": 2258
    },
    {
      "epoch": 2.1422475106685632,
      "grad_norm": 0.3944634875673138,
      "learning_rate": 2.286458922412004e-06,
      "loss": 0.0248,
      "step": 2259
    },
    {
      "epoch": 2.1431958274063536,
      "grad_norm": 0.5135149676803258,
      "learning_rate": 2.28182316020032e-06,
      "loss": 0.0252,
      "step": 2260
    },
    {
      "epoch": 2.144144144144144,
      "grad_norm": 0.5864406162589035,
      "learning_rate": 2.2771907124492855e-06,
      "loss": 0.031,
      "step": 2261
    },
    {
      "epoch": 2.1450924608819344,
      "grad_norm": 0.3595593256803766,
      "learning_rate": 2.272561584807567e-06,
      "loss": 0.0211,
      "step": 2262
    },
    {
      "epoch": 2.146040777619725,
      "grad_norm": 0.5644326980878621,
      "learning_rate": 2.2679357829197773e-06,
      "loss": 0.025,
      "step": 2263
    },
    {
      "epoch": 2.1469890943575156,
      "grad_norm": 0.40039567554067224,
      "learning_rate": 2.263313312426477e-06,
      "loss": 0.0251,
      "step": 2264
    },
    {
      "epoch": 2.147937411095306,
      "grad_norm": 0.3494192171240132,
      "learning_rate": 2.2586941789641644e-06,
      "loss": 0.0279,
      "step": 2265
    },
    {
      "epoch": 2.1488857278330964,
      "grad_norm": 0.543967374973067,
      "learning_rate": 2.2540783881652672e-06,
      "loss": 0.0466,
      "step": 2266
    },
    {
      "epoch": 2.149834044570887,
      "grad_norm": 0.3348355579550652,
      "learning_rate": 2.249465945658135e-06,
      "loss": 0.0157,
      "step": 2267
    },
    {
      "epoch": 2.150782361308677,
      "grad_norm": 0.5992344414940606,
      "learning_rate": 2.2448568570670422e-06,
      "loss": 0.0201,
      "step": 2268
    },
    {
      "epoch": 2.1517306780464676,
      "grad_norm": 0.27638109684371753,
      "learning_rate": 2.2402511280121635e-06,
      "loss": 0.0139,
      "step": 2269
    },
    {
      "epoch": 2.152678994784258,
      "grad_norm": 0.4864039946795088,
      "learning_rate": 2.235648764109587e-06,
      "loss": 0.0232,
      "step": 2270
    },
    {
      "epoch": 2.1536273115220483,
      "grad_norm": 0.4375793326410557,
      "learning_rate": 2.2310497709712883e-06,
      "loss": 0.0239,
      "step": 2271
    },
    {
      "epoch": 2.1545756282598387,
      "grad_norm": 0.4049296761357532,
      "learning_rate": 2.22645415420514e-06,
      "loss": 0.0198,
      "step": 2272
    },
    {
      "epoch": 2.155523944997629,
      "grad_norm": 0.3080411672189438,
      "learning_rate": 2.2218619194148957e-06,
      "loss": 0.0192,
      "step": 2273
    },
    {
      "epoch": 2.1564722617354195,
      "grad_norm": 0.4256264536412931,
      "learning_rate": 2.2172730722001806e-06,
      "loss": 0.0223,
      "step": 2274
    },
    {
      "epoch": 2.15742057847321,
      "grad_norm": 0.27136420910261355,
      "learning_rate": 2.2126876181564955e-06,
      "loss": 0.0109,
      "step": 2275
    },
    {
      "epoch": 2.1583688952110003,
      "grad_norm": 0.4115704050998557,
      "learning_rate": 2.208105562875203e-06,
      "loss": 0.0293,
      "step": 2276
    },
    {
      "epoch": 2.1593172119487907,
      "grad_norm": 0.3998489255300917,
      "learning_rate": 2.2035269119435164e-06,
      "loss": 0.0213,
      "step": 2277
    },
    {
      "epoch": 2.1602655286865815,
      "grad_norm": 0.44243900795322916,
      "learning_rate": 2.1989516709445013e-06,
      "loss": 0.0224,
      "step": 2278
    },
    {
      "epoch": 2.161213845424372,
      "grad_norm": 0.44256484344324315,
      "learning_rate": 2.194379845457064e-06,
      "loss": 0.0297,
      "step": 2279
    },
    {
      "epoch": 2.1621621621621623,
      "grad_norm": 0.3483421838923968,
      "learning_rate": 2.1898114410559475e-06,
      "loss": 0.0228,
      "step": 2280
    },
    {
      "epoch": 2.1631104788999527,
      "grad_norm": 0.49993876242502255,
      "learning_rate": 2.185246463311725e-06,
      "loss": 0.0332,
      "step": 2281
    },
    {
      "epoch": 2.164058795637743,
      "grad_norm": 0.42001008552894303,
      "learning_rate": 2.1806849177907844e-06,
      "loss": 0.0264,
      "step": 2282
    },
    {
      "epoch": 2.1650071123755334,
      "grad_norm": 0.41695424744520754,
      "learning_rate": 2.176126810055337e-06,
      "loss": 0.0303,
      "step": 2283
    },
    {
      "epoch": 2.165955429113324,
      "grad_norm": 0.3240800567288141,
      "learning_rate": 2.171572145663398e-06,
      "loss": 0.0183,
      "step": 2284
    },
    {
      "epoch": 2.1669037458511142,
      "grad_norm": 0.39669442654415515,
      "learning_rate": 2.167020930168781e-06,
      "loss": 0.0229,
      "step": 2285
    },
    {
      "epoch": 2.1678520625889046,
      "grad_norm": 0.39546239248455106,
      "learning_rate": 2.1624731691211016e-06,
      "loss": 0.0231,
      "step": 2286
    },
    {
      "epoch": 2.168800379326695,
      "grad_norm": 0.4087436587688708,
      "learning_rate": 2.157928868065754e-06,
      "loss": 0.0277,
      "step": 2287
    },
    {
      "epoch": 2.1697486960644854,
      "grad_norm": 0.4530786653379887,
      "learning_rate": 2.153388032543923e-06,
      "loss": 0.0194,
      "step": 2288
    },
    {
      "epoch": 2.170697012802276,
      "grad_norm": 0.43939543462713626,
      "learning_rate": 2.1488506680925596e-06,
      "loss": 0.0258,
      "step": 2289
    },
    {
      "epoch": 2.171645329540066,
      "grad_norm": 0.4522241596008493,
      "learning_rate": 2.144316780244385e-06,
      "loss": 0.0184,
      "step": 2290
    },
    {
      "epoch": 2.172593646277857,
      "grad_norm": 1.1516269930596483,
      "learning_rate": 2.1397863745278825e-06,
      "loss": 0.0224,
      "step": 2291
    },
    {
      "epoch": 2.1735419630156474,
      "grad_norm": 0.3736163797314226,
      "learning_rate": 2.1352594564672907e-06,
      "loss": 0.0252,
      "step": 2292
    },
    {
      "epoch": 2.174490279753438,
      "grad_norm": 0.390543627117664,
      "learning_rate": 2.1307360315825894e-06,
      "loss": 0.0189,
      "step": 2293
    },
    {
      "epoch": 2.175438596491228,
      "grad_norm": 0.41425912587622055,
      "learning_rate": 2.1262161053895068e-06,
      "loss": 0.0214,
      "step": 2294
    },
    {
      "epoch": 2.1763869132290186,
      "grad_norm": 0.4604953079593103,
      "learning_rate": 2.121699683399497e-06,
      "loss": 0.0215,
      "step": 2295
    },
    {
      "epoch": 2.177335229966809,
      "grad_norm": 0.4114732763858242,
      "learning_rate": 2.1171867711197474e-06,
      "loss": 0.0214,
      "step": 2296
    },
    {
      "epoch": 2.1782835467045993,
      "grad_norm": 0.4056111357292128,
      "learning_rate": 2.112677374053164e-06,
      "loss": 0.0233,
      "step": 2297
    },
    {
      "epoch": 2.1792318634423897,
      "grad_norm": 0.32625279399687396,
      "learning_rate": 2.108171497698364e-06,
      "loss": 0.017,
      "step": 2298
    },
    {
      "epoch": 2.18018018018018,
      "grad_norm": 0.39967265560405546,
      "learning_rate": 2.1036691475496764e-06,
      "loss": 0.0182,
      "step": 2299
    },
    {
      "epoch": 2.1811284969179705,
      "grad_norm": 0.3725151692726603,
      "learning_rate": 2.0991703290971255e-06,
      "loss": 0.0287,
      "step": 2300
    },
    {
      "epoch": 2.182076813655761,
      "grad_norm": 0.3970538993097859,
      "learning_rate": 2.0946750478264287e-06,
      "loss": 0.0255,
      "step": 2301
    },
    {
      "epoch": 2.1830251303935513,
      "grad_norm": 0.3311384010233447,
      "learning_rate": 2.090183309218998e-06,
      "loss": 0.0156,
      "step": 2302
    },
    {
      "epoch": 2.1839734471313417,
      "grad_norm": 0.7722545177624061,
      "learning_rate": 2.085695118751916e-06,
      "loss": 0.0276,
      "step": 2303
    },
    {
      "epoch": 2.1849217638691325,
      "grad_norm": 0.6058568443970096,
      "learning_rate": 2.0812104818979437e-06,
      "loss": 0.022,
      "step": 2304
    },
    {
      "epoch": 2.185870080606923,
      "grad_norm": 0.37429742353382056,
      "learning_rate": 2.076729404125513e-06,
      "loss": 0.019,
      "step": 2305
    },
    {
      "epoch": 2.1868183973447133,
      "grad_norm": 0.5299104777253576,
      "learning_rate": 2.0722518908987043e-06,
      "loss": 0.0257,
      "step": 2306
    },
    {
      "epoch": 2.1877667140825037,
      "grad_norm": 0.3915958586603079,
      "learning_rate": 2.067777947677265e-06,
      "loss": 0.024,
      "step": 2307
    },
    {
      "epoch": 2.188715030820294,
      "grad_norm": 0.397731251360786,
      "learning_rate": 2.063307579916578e-06,
      "loss": 0.0195,
      "step": 2308
    },
    {
      "epoch": 2.1896633475580844,
      "grad_norm": 0.507467765523117,
      "learning_rate": 2.058840793067675e-06,
      "loss": 0.0318,
      "step": 2309
    },
    {
      "epoch": 2.190611664295875,
      "grad_norm": 0.38413895508990836,
      "learning_rate": 2.054377592577214e-06,
      "loss": 0.0202,
      "step": 2310
    },
    {
      "epoch": 2.1915599810336652,
      "grad_norm": 0.45976607540693787,
      "learning_rate": 2.049917983887487e-06,
      "loss": 0.0229,
      "step": 2311
    },
    {
      "epoch": 2.1925082977714556,
      "grad_norm": 0.361658805053007,
      "learning_rate": 2.0454619724363993e-06,
      "loss": 0.0227,
      "step": 2312
    },
    {
      "epoch": 2.193456614509246,
      "grad_norm": 0.3619090798648459,
      "learning_rate": 2.041009563657477e-06,
      "loss": 0.0199,
      "step": 2313
    },
    {
      "epoch": 2.1944049312470364,
      "grad_norm": 0.3771128928199703,
      "learning_rate": 2.036560762979845e-06,
      "loss": 0.0189,
      "step": 2314
    },
    {
      "epoch": 2.195353247984827,
      "grad_norm": 0.3715021611413161,
      "learning_rate": 2.032115575828238e-06,
      "loss": 0.0259,
      "step": 2315
    },
    {
      "epoch": 2.196301564722617,
      "grad_norm": 0.43350725720240935,
      "learning_rate": 2.027674007622975e-06,
      "loss": 0.0232,
      "step": 2316
    },
    {
      "epoch": 2.197249881460408,
      "grad_norm": 0.35582869608226325,
      "learning_rate": 2.0232360637799687e-06,
      "loss": 0.0189,
      "step": 2317
    },
    {
      "epoch": 2.1981981981981984,
      "grad_norm": 0.46958694231370757,
      "learning_rate": 2.0188017497107114e-06,
      "loss": 0.0221,
      "step": 2318
    },
    {
      "epoch": 2.199146514935989,
      "grad_norm": 0.3646835051350837,
      "learning_rate": 2.014371070822266e-06,
      "loss": 0.0243,
      "step": 2319
    },
    {
      "epoch": 2.200094831673779,
      "grad_norm": 0.3901856309198319,
      "learning_rate": 2.009944032517269e-06,
      "loss": 0.0191,
      "step": 2320
    },
    {
      "epoch": 2.2010431484115696,
      "grad_norm": 0.47290921322501533,
      "learning_rate": 2.005520640193911e-06,
      "loss": 0.0245,
      "step": 2321
    },
    {
      "epoch": 2.20199146514936,
      "grad_norm": 0.38202716048927027,
      "learning_rate": 2.0011008992459375e-06,
      "loss": 0.015,
      "step": 2322
    },
    {
      "epoch": 2.2029397818871503,
      "grad_norm": 0.40684472495286483,
      "learning_rate": 1.9966848150626478e-06,
      "loss": 0.0187,
      "step": 2323
    },
    {
      "epoch": 2.2038880986249407,
      "grad_norm": 0.41394820108088604,
      "learning_rate": 1.992272393028875e-06,
      "loss": 0.021,
      "step": 2324
    },
    {
      "epoch": 2.204836415362731,
      "grad_norm": 0.35052141592281283,
      "learning_rate": 1.987863638524991e-06,
      "loss": 0.0184,
      "step": 2325
    },
    {
      "epoch": 2.2057847321005215,
      "grad_norm": 0.5224256513759739,
      "learning_rate": 1.9834585569268973e-06,
      "loss": 0.031,
      "step": 2326
    },
    {
      "epoch": 2.206733048838312,
      "grad_norm": 0.4306504185840522,
      "learning_rate": 1.979057153606009e-06,
      "loss": 0.0252,
      "step": 2327
    },
    {
      "epoch": 2.2076813655761023,
      "grad_norm": 0.34500352219686414,
      "learning_rate": 1.9746594339292634e-06,
      "loss": 0.0139,
      "step": 2328
    },
    {
      "epoch": 2.2086296823138927,
      "grad_norm": 0.47428095884091725,
      "learning_rate": 1.9702654032591057e-06,
      "loss": 0.0299,
      "step": 2329
    },
    {
      "epoch": 2.209577999051683,
      "grad_norm": 0.31715247322701795,
      "learning_rate": 1.965875066953477e-06,
      "loss": 0.0187,
      "step": 2330
    },
    {
      "epoch": 2.2105263157894735,
      "grad_norm": 0.35097104725381645,
      "learning_rate": 1.961488430365821e-06,
      "loss": 0.018,
      "step": 2331
    },
    {
      "epoch": 2.2114746325272643,
      "grad_norm": 0.43904077797503993,
      "learning_rate": 1.957105498845065e-06,
      "loss": 0.0273,
      "step": 2332
    },
    {
      "epoch": 2.2124229492650547,
      "grad_norm": 0.37433258153809074,
      "learning_rate": 1.9527262777356175e-06,
      "loss": 0.0189,
      "step": 2333
    },
    {
      "epoch": 2.213371266002845,
      "grad_norm": 0.452382015943325,
      "learning_rate": 1.9483507723773693e-06,
      "loss": 0.02,
      "step": 2334
    },
    {
      "epoch": 2.2143195827406355,
      "grad_norm": 0.3844273691974591,
      "learning_rate": 1.9439789881056724e-06,
      "loss": 0.0296,
      "step": 2335
    },
    {
      "epoch": 2.215267899478426,
      "grad_norm": 0.3251514659777261,
      "learning_rate": 1.9396109302513465e-06,
      "loss": 0.0241,
      "step": 2336
    },
    {
      "epoch": 2.2162162162162162,
      "grad_norm": 0.3675273639039549,
      "learning_rate": 1.9352466041406687e-06,
      "loss": 0.0134,
      "step": 2337
    },
    {
      "epoch": 2.2171645329540066,
      "grad_norm": 0.3983214976925876,
      "learning_rate": 1.9308860150953583e-06,
      "loss": 0.0232,
      "step": 2338
    },
    {
      "epoch": 2.218112849691797,
      "grad_norm": 0.7144984519278534,
      "learning_rate": 1.926529168432587e-06,
      "loss": 0.036,
      "step": 2339
    },
    {
      "epoch": 2.2190611664295874,
      "grad_norm": 0.48637705204645265,
      "learning_rate": 1.9221760694649553e-06,
      "loss": 0.0339,
      "step": 2340
    },
    {
      "epoch": 2.220009483167378,
      "grad_norm": 0.7318106591805768,
      "learning_rate": 1.9178267235004984e-06,
      "loss": 0.0206,
      "step": 2341
    },
    {
      "epoch": 2.220957799905168,
      "grad_norm": 0.4343031634314887,
      "learning_rate": 1.9134811358426756e-06,
      "loss": 0.0293,
      "step": 2342
    },
    {
      "epoch": 2.2219061166429586,
      "grad_norm": 0.5411094163596777,
      "learning_rate": 1.909139311790362e-06,
      "loss": 0.0513,
      "step": 2343
    },
    {
      "epoch": 2.222854433380749,
      "grad_norm": 0.4262502588360842,
      "learning_rate": 1.9048012566378387e-06,
      "loss": 0.0273,
      "step": 2344
    },
    {
      "epoch": 2.22380275011854,
      "grad_norm": 0.38390892408447685,
      "learning_rate": 1.9004669756748017e-06,
      "loss": 0.0237,
      "step": 2345
    },
    {
      "epoch": 2.22475106685633,
      "grad_norm": 0.3107308166605013,
      "learning_rate": 1.8961364741863342e-06,
      "loss": 0.0184,
      "step": 2346
    },
    {
      "epoch": 2.2256993835941206,
      "grad_norm": 0.32935486433027755,
      "learning_rate": 1.8918097574529193e-06,
      "loss": 0.0168,
      "step": 2347
    },
    {
      "epoch": 2.226647700331911,
      "grad_norm": 0.3322923262299374,
      "learning_rate": 1.8874868307504185e-06,
      "loss": 0.0179,
      "step": 2348
    },
    {
      "epoch": 2.2275960170697013,
      "grad_norm": 0.36089333207950935,
      "learning_rate": 1.8831676993500758e-06,
      "loss": 0.0224,
      "step": 2349
    },
    {
      "epoch": 2.2285443338074917,
      "grad_norm": 0.41395718666883813,
      "learning_rate": 1.8788523685185084e-06,
      "loss": 0.021,
      "step": 2350
    },
    {
      "epoch": 2.229492650545282,
      "grad_norm": 0.4439323647263712,
      "learning_rate": 1.8745408435176932e-06,
      "loss": 0.0268,
      "step": 2351
    },
    {
      "epoch": 2.2304409672830725,
      "grad_norm": 0.427895772910043,
      "learning_rate": 1.8702331296049741e-06,
      "loss": 0.021,
      "step": 2352
    },
    {
      "epoch": 2.231389284020863,
      "grad_norm": 0.37486703243376646,
      "learning_rate": 1.8659292320330408e-06,
      "loss": 0.0177,
      "step": 2353
    },
    {
      "epoch": 2.2323376007586533,
      "grad_norm": 0.3841975396897219,
      "learning_rate": 1.861629156049935e-06,
      "loss": 0.0237,
      "step": 2354
    },
    {
      "epoch": 2.2332859174964437,
      "grad_norm": 0.3907334477612354,
      "learning_rate": 1.8573329068990358e-06,
      "loss": 0.0209,
      "step": 2355
    },
    {
      "epoch": 2.234234234234234,
      "grad_norm": 0.43207097544527573,
      "learning_rate": 1.8530404898190534e-06,
      "loss": 0.015,
      "step": 2356
    },
    {
      "epoch": 2.2351825509720245,
      "grad_norm": 0.40641958306313414,
      "learning_rate": 1.8487519100440316e-06,
      "loss": 0.0212,
      "step": 2357
    },
    {
      "epoch": 2.2361308677098153,
      "grad_norm": 0.45601425060223455,
      "learning_rate": 1.8444671728033314e-06,
      "loss": 0.025,
      "step": 2358
    },
    {
      "epoch": 2.2370791844476057,
      "grad_norm": 0.40452585916751155,
      "learning_rate": 1.8401862833216272e-06,
      "loss": 0.0221,
      "step": 2359
    },
    {
      "epoch": 2.238027501185396,
      "grad_norm": 0.32206562407761236,
      "learning_rate": 1.8359092468189048e-06,
      "loss": 0.0166,
      "step": 2360
    },
    {
      "epoch": 2.2389758179231865,
      "grad_norm": 0.4665259635354582,
      "learning_rate": 1.8316360685104478e-06,
      "loss": 0.0276,
      "step": 2361
    },
    {
      "epoch": 2.239924134660977,
      "grad_norm": 0.6336444720182508,
      "learning_rate": 1.827366753606839e-06,
      "loss": 0.0308,
      "step": 2362
    },
    {
      "epoch": 2.2408724513987672,
      "grad_norm": 0.4423139744797903,
      "learning_rate": 1.8231013073139504e-06,
      "loss": 0.0279,
      "step": 2363
    },
    {
      "epoch": 2.2418207681365576,
      "grad_norm": 0.4376868624329782,
      "learning_rate": 1.8188397348329328e-06,
      "loss": 0.0203,
      "step": 2364
    },
    {
      "epoch": 2.242769084874348,
      "grad_norm": 0.36625273715371953,
      "learning_rate": 1.814582041360215e-06,
      "loss": 0.018,
      "step": 2365
    },
    {
      "epoch": 2.2437174016121384,
      "grad_norm": 0.5777502114143614,
      "learning_rate": 1.8103282320874987e-06,
      "loss": 0.0245,
      "step": 2366
    },
    {
      "epoch": 2.244665718349929,
      "grad_norm": 0.45081701594898027,
      "learning_rate": 1.806078312201745e-06,
      "loss": 0.0302,
      "step": 2367
    },
    {
      "epoch": 2.245614035087719,
      "grad_norm": 0.4140190228187624,
      "learning_rate": 1.8018322868851779e-06,
      "loss": 0.0184,
      "step": 2368
    },
    {
      "epoch": 2.2465623518255096,
      "grad_norm": 0.457871291883172,
      "learning_rate": 1.7975901613152645e-06,
      "loss": 0.0181,
      "step": 2369
    },
    {
      "epoch": 2.2475106685633,
      "grad_norm": 0.36063582182993614,
      "learning_rate": 1.7933519406647243e-06,
      "loss": 0.0188,
      "step": 2370
    },
    {
      "epoch": 2.248458985301091,
      "grad_norm": 0.4098690379894618,
      "learning_rate": 1.7891176301015135e-06,
      "loss": 0.0301,
      "step": 2371
    },
    {
      "epoch": 2.249407302038881,
      "grad_norm": 0.39708742138503444,
      "learning_rate": 1.7848872347888163e-06,
      "loss": 0.0247,
      "step": 2372
    },
    {
      "epoch": 2.2503556187766716,
      "grad_norm": 0.41417175900835923,
      "learning_rate": 1.7806607598850467e-06,
      "loss": 0.018,
      "step": 2373
    },
    {
      "epoch": 2.251303935514462,
      "grad_norm": 0.4754670959345918,
      "learning_rate": 1.7764382105438394e-06,
      "loss": 0.0253,
      "step": 2374
    },
    {
      "epoch": 2.2522522522522523,
      "grad_norm": 0.8568347200994995,
      "learning_rate": 1.7722195919140383e-06,
      "loss": 0.0263,
      "step": 2375
    },
    {
      "epoch": 2.2532005689900427,
      "grad_norm": 0.38552968326653064,
      "learning_rate": 1.7680049091396967e-06,
      "loss": 0.022,
      "step": 2376
    },
    {
      "epoch": 2.254148885727833,
      "grad_norm": 0.5324795698459304,
      "learning_rate": 1.7637941673600668e-06,
      "loss": 0.0248,
      "step": 2377
    },
    {
      "epoch": 2.2550972024656235,
      "grad_norm": 0.5170883292235453,
      "learning_rate": 1.7595873717095973e-06,
      "loss": 0.0277,
      "step": 2378
    },
    {
      "epoch": 2.256045519203414,
      "grad_norm": 0.5100513984737847,
      "learning_rate": 1.7553845273179277e-06,
      "loss": 0.0274,
      "step": 2379
    },
    {
      "epoch": 2.2569938359412043,
      "grad_norm": 0.3704538407484803,
      "learning_rate": 1.7511856393098715e-06,
      "loss": 0.0213,
      "step": 2380
    },
    {
      "epoch": 2.2579421526789947,
      "grad_norm": 0.48491216247522145,
      "learning_rate": 1.7469907128054247e-06,
      "loss": 0.026,
      "step": 2381
    },
    {
      "epoch": 2.258890469416785,
      "grad_norm": 0.649876149364862,
      "learning_rate": 1.7427997529197533e-06,
      "loss": 0.0173,
      "step": 2382
    },
    {
      "epoch": 2.2598387861545755,
      "grad_norm": 0.48045437554087517,
      "learning_rate": 1.7386127647631802e-06,
      "loss": 0.0402,
      "step": 2383
    },
    {
      "epoch": 2.2607871028923663,
      "grad_norm": 0.38709200279463996,
      "learning_rate": 1.7344297534411918e-06,
      "loss": 0.0228,
      "step": 2384
    },
    {
      "epoch": 2.2617354196301562,
      "grad_norm": 0.39911202838197224,
      "learning_rate": 1.7302507240544197e-06,
      "loss": 0.0263,
      "step": 2385
    },
    {
      "epoch": 2.262683736367947,
      "grad_norm": 0.402703895487539,
      "learning_rate": 1.7260756816986468e-06,
      "loss": 0.0199,
      "step": 2386
    },
    {
      "epoch": 2.2636320531057375,
      "grad_norm": 0.4327109923756727,
      "learning_rate": 1.7219046314647875e-06,
      "loss": 0.0293,
      "step": 2387
    },
    {
      "epoch": 2.264580369843528,
      "grad_norm": 0.46823736554030093,
      "learning_rate": 1.7177375784388906e-06,
      "loss": 0.0202,
      "step": 2388
    },
    {
      "epoch": 2.2655286865813182,
      "grad_norm": 0.37427697435461244,
      "learning_rate": 1.7135745277021332e-06,
      "loss": 0.0188,
      "step": 2389
    },
    {
      "epoch": 2.2664770033191086,
      "grad_norm": 0.3760762544416307,
      "learning_rate": 1.7094154843308114e-06,
      "loss": 0.018,
      "step": 2390
    },
    {
      "epoch": 2.267425320056899,
      "grad_norm": 0.3185328153435644,
      "learning_rate": 1.7052604533963308e-06,
      "loss": 0.0168,
      "step": 2391
    },
    {
      "epoch": 2.2683736367946894,
      "grad_norm": 0.39366485747456337,
      "learning_rate": 1.7011094399652107e-06,
      "loss": 0.0179,
      "step": 2392
    },
    {
      "epoch": 2.26932195353248,
      "grad_norm": 0.4374421147215151,
      "learning_rate": 1.6969624490990654e-06,
      "loss": 0.0195,
      "step": 2393
    },
    {
      "epoch": 2.27027027027027,
      "grad_norm": 0.40108389408149603,
      "learning_rate": 1.6928194858546088e-06,
      "loss": 0.0175,
      "step": 2394
    },
    {
      "epoch": 2.2712185870080606,
      "grad_norm": 0.30247705105026723,
      "learning_rate": 1.6886805552836433e-06,
      "loss": 0.0159,
      "step": 2395
    },
    {
      "epoch": 2.272166903745851,
      "grad_norm": 0.4862095152615184,
      "learning_rate": 1.6845456624330492e-06,
      "loss": 0.0227,
      "step": 2396
    },
    {
      "epoch": 2.273115220483642,
      "grad_norm": 0.3745563544160728,
      "learning_rate": 1.68041481234479e-06,
      "loss": 0.018,
      "step": 2397
    },
    {
      "epoch": 2.2740635372214317,
      "grad_norm": 0.36704972477579006,
      "learning_rate": 1.6762880100558954e-06,
      "loss": 0.0236,
      "step": 2398
    },
    {
      "epoch": 2.2750118539592226,
      "grad_norm": 0.4747742150229455,
      "learning_rate": 1.6721652605984585e-06,
      "loss": 0.0219,
      "step": 2399
    },
    {
      "epoch": 2.275960170697013,
      "grad_norm": 0.3953571044047917,
      "learning_rate": 1.6680465689996345e-06,
      "loss": 0.0198,
      "step": 2400
    },
    {
      "epoch": 2.2769084874348033,
      "grad_norm": 0.43525927498232264,
      "learning_rate": 1.6639319402816263e-06,
      "loss": 0.0312,
      "step": 2401
    },
    {
      "epoch": 2.2778568041725937,
      "grad_norm": 0.3436816679628124,
      "learning_rate": 1.6598213794616862e-06,
      "loss": 0.0147,
      "step": 2402
    },
    {
      "epoch": 2.278805120910384,
      "grad_norm": 0.44877364797530483,
      "learning_rate": 1.655714891552106e-06,
      "loss": 0.0201,
      "step": 2403
    },
    {
      "epoch": 2.2797534376481745,
      "grad_norm": 0.3834821514632032,
      "learning_rate": 1.651612481560207e-06,
      "loss": 0.0209,
      "step": 2404
    },
    {
      "epoch": 2.280701754385965,
      "grad_norm": 0.472080313698594,
      "learning_rate": 1.6475141544883416e-06,
      "loss": 0.0251,
      "step": 2405
    },
    {
      "epoch": 2.2816500711237553,
      "grad_norm": 0.4625744377816762,
      "learning_rate": 1.643419915333886e-06,
      "loss": 0.0344,
      "step": 2406
    },
    {
      "epoch": 2.2825983878615457,
      "grad_norm": 0.3367193925166736,
      "learning_rate": 1.6393297690892256e-06,
      "loss": 0.0166,
      "step": 2407
    },
    {
      "epoch": 2.283546704599336,
      "grad_norm": 0.40850259795694177,
      "learning_rate": 1.6352437207417571e-06,
      "loss": 0.0196,
      "step": 2408
    },
    {
      "epoch": 2.2844950213371265,
      "grad_norm": 0.38565301084020687,
      "learning_rate": 1.631161775273885e-06,
      "loss": 0.0162,
      "step": 2409
    },
    {
      "epoch": 2.285443338074917,
      "grad_norm": 0.3851049737678369,
      "learning_rate": 1.6270839376630033e-06,
      "loss": 0.0194,
      "step": 2410
    },
    {
      "epoch": 2.2863916548127072,
      "grad_norm": 0.5577488269441532,
      "learning_rate": 1.6230102128815033e-06,
      "loss": 0.0367,
      "step": 2411
    },
    {
      "epoch": 2.287339971550498,
      "grad_norm": 0.5016112831857718,
      "learning_rate": 1.6189406058967577e-06,
      "loss": 0.0189,
      "step": 2412
    },
    {
      "epoch": 2.2882882882882885,
      "grad_norm": 0.5585880833977839,
      "learning_rate": 1.6148751216711206e-06,
      "loss": 0.026,
      "step": 2413
    },
    {
      "epoch": 2.289236605026079,
      "grad_norm": 0.4195620847627191,
      "learning_rate": 1.6108137651619154e-06,
      "loss": 0.0267,
      "step": 2414
    },
    {
      "epoch": 2.2901849217638692,
      "grad_norm": 0.5490900492941025,
      "learning_rate": 1.6067565413214353e-06,
      "loss": 0.0221,
      "step": 2415
    },
    {
      "epoch": 2.2911332385016596,
      "grad_norm": 0.5073374779777672,
      "learning_rate": 1.6027034550969356e-06,
      "loss": 0.0232,
      "step": 2416
    },
    {
      "epoch": 2.29208155523945,
      "grad_norm": 0.3707053895150809,
      "learning_rate": 1.5986545114306202e-06,
      "loss": 0.0216,
      "step": 2417
    },
    {
      "epoch": 2.2930298719772404,
      "grad_norm": 0.4036642305925845,
      "learning_rate": 1.5946097152596496e-06,
      "loss": 0.0198,
      "step": 2418
    },
    {
      "epoch": 2.293978188715031,
      "grad_norm": 0.2837666309597713,
      "learning_rate": 1.5905690715161209e-06,
      "loss": 0.0164,
      "step": 2419
    },
    {
      "epoch": 2.294926505452821,
      "grad_norm": 0.5566663426849048,
      "learning_rate": 1.586532585127069e-06,
      "loss": 0.0257,
      "step": 2420
    },
    {
      "epoch": 2.2958748221906116,
      "grad_norm": 0.6345053044837282,
      "learning_rate": 1.5825002610144623e-06,
      "loss": 0.0257,
      "step": 2421
    },
    {
      "epoch": 2.296823138928402,
      "grad_norm": 0.35480991236789217,
      "learning_rate": 1.57847210409519e-06,
      "loss": 0.0182,
      "step": 2422
    },
    {
      "epoch": 2.2977714556661923,
      "grad_norm": 0.5404075847950341,
      "learning_rate": 1.574448119281063e-06,
      "loss": 0.0213,
      "step": 2423
    },
    {
      "epoch": 2.2987197724039827,
      "grad_norm": 0.3390560549791458,
      "learning_rate": 1.570428311478805e-06,
      "loss": 0.0162,
      "step": 2424
    },
    {
      "epoch": 2.2996680891417736,
      "grad_norm": 0.7426039658923921,
      "learning_rate": 1.5664126855900424e-06,
      "loss": 0.0238,
      "step": 2425
    },
    {
      "epoch": 2.300616405879564,
      "grad_norm": 0.4266296618803969,
      "learning_rate": 1.5624012465113065e-06,
      "loss": 0.02,
      "step": 2426
    },
    {
      "epoch": 2.3015647226173543,
      "grad_norm": 0.3065546446187386,
      "learning_rate": 1.558393999134023e-06,
      "loss": 0.0156,
      "step": 2427
    },
    {
      "epoch": 2.3025130393551447,
      "grad_norm": 0.4512682712027585,
      "learning_rate": 1.554390948344503e-06,
      "loss": 0.0304,
      "step": 2428
    },
    {
      "epoch": 2.303461356092935,
      "grad_norm": 0.3422412628357766,
      "learning_rate": 1.5503920990239452e-06,
      "loss": 0.0226,
      "step": 2429
    },
    {
      "epoch": 2.3044096728307255,
      "grad_norm": 0.31396493988976687,
      "learning_rate": 1.5463974560484213e-06,
      "loss": 0.0205,
      "step": 2430
    },
    {
      "epoch": 2.305357989568516,
      "grad_norm": 0.35208787261922914,
      "learning_rate": 1.5424070242888733e-06,
      "loss": 0.0208,
      "step": 2431
    },
    {
      "epoch": 2.3063063063063063,
      "grad_norm": 0.6959424315073545,
      "learning_rate": 1.538420808611114e-06,
      "loss": 0.0144,
      "step": 2432
    },
    {
      "epoch": 2.3072546230440967,
      "grad_norm": 0.35449272063606063,
      "learning_rate": 1.534438813875807e-06,
      "loss": 0.0242,
      "step": 2433
    },
    {
      "epoch": 2.308202939781887,
      "grad_norm": 0.3970212613465219,
      "learning_rate": 1.530461044938476e-06,
      "loss": 0.0282,
      "step": 2434
    },
    {
      "epoch": 2.3091512565196775,
      "grad_norm": 0.3486460962177107,
      "learning_rate": 1.52648750664949e-06,
      "loss": 0.0168,
      "step": 2435
    },
    {
      "epoch": 2.310099573257468,
      "grad_norm": 0.33438665576098,
      "learning_rate": 1.522518203854056e-06,
      "loss": 0.0167,
      "step": 2436
    },
    {
      "epoch": 2.3110478899952582,
      "grad_norm": 0.4974236473530219,
      "learning_rate": 1.5185531413922217e-06,
      "loss": 0.0238,
      "step": 2437
    },
    {
      "epoch": 2.311996206733049,
      "grad_norm": 0.4821267448581767,
      "learning_rate": 1.5145923240988587e-06,
      "loss": 0.0212,
      "step": 2438
    },
    {
      "epoch": 2.3129445234708395,
      "grad_norm": 1.245911380396821,
      "learning_rate": 1.5106357568036662e-06,
      "loss": 0.0258,
      "step": 2439
    },
    {
      "epoch": 2.31389284020863,
      "grad_norm": 0.4281277216603392,
      "learning_rate": 1.5066834443311613e-06,
      "loss": 0.0278,
      "step": 2440
    },
    {
      "epoch": 2.3148411569464202,
      "grad_norm": 0.43810859496573806,
      "learning_rate": 1.50273539150067e-06,
      "loss": 0.0229,
      "step": 2441
    },
    {
      "epoch": 2.3157894736842106,
      "grad_norm": 0.33235104756376627,
      "learning_rate": 1.4987916031263234e-06,
      "loss": 0.0136,
      "step": 2442
    },
    {
      "epoch": 2.316737790422001,
      "grad_norm": 0.3297892350058022,
      "learning_rate": 1.4948520840170594e-06,
      "loss": 0.0261,
      "step": 2443
    },
    {
      "epoch": 2.3176861071597914,
      "grad_norm": 0.42008924032009215,
      "learning_rate": 1.4909168389766015e-06,
      "loss": 0.0312,
      "step": 2444
    },
    {
      "epoch": 2.318634423897582,
      "grad_norm": 0.3389579383262707,
      "learning_rate": 1.486985872803469e-06,
      "loss": 0.0176,
      "step": 2445
    },
    {
      "epoch": 2.319582740635372,
      "grad_norm": 0.5487799107230558,
      "learning_rate": 1.483059190290957e-06,
      "loss": 0.026,
      "step": 2446
    },
    {
      "epoch": 2.3205310573731626,
      "grad_norm": 0.4890771192858219,
      "learning_rate": 1.4791367962271425e-06,
      "loss": 0.0266,
      "step": 2447
    },
    {
      "epoch": 2.321479374110953,
      "grad_norm": 0.4438394146588021,
      "learning_rate": 1.4752186953948728e-06,
      "loss": 0.0196,
      "step": 2448
    },
    {
      "epoch": 2.3224276908487433,
      "grad_norm": 0.3662367949751032,
      "learning_rate": 1.4713048925717548e-06,
      "loss": 0.0175,
      "step": 2449
    },
    {
      "epoch": 2.3233760075865337,
      "grad_norm": 0.43993413507805795,
      "learning_rate": 1.4673953925301625e-06,
      "loss": 0.0219,
      "step": 2450
    },
    {
      "epoch": 2.3243243243243246,
      "grad_norm": 0.32060744843760514,
      "learning_rate": 1.463490200037216e-06,
      "loss": 0.0183,
      "step": 2451
    },
    {
      "epoch": 2.3252726410621145,
      "grad_norm": 0.3084422018115681,
      "learning_rate": 1.4595893198547889e-06,
      "loss": 0.0165,
      "step": 2452
    },
    {
      "epoch": 2.3262209577999053,
      "grad_norm": 0.512124514376596,
      "learning_rate": 1.4556927567394918e-06,
      "loss": 0.0256,
      "step": 2453
    },
    {
      "epoch": 2.3271692745376957,
      "grad_norm": 0.37175509555253117,
      "learning_rate": 1.4518005154426722e-06,
      "loss": 0.019,
      "step": 2454
    },
    {
      "epoch": 2.328117591275486,
      "grad_norm": 0.3864056129285522,
      "learning_rate": 1.44791260071041e-06,
      "loss": 0.0138,
      "step": 2455
    },
    {
      "epoch": 2.3290659080132765,
      "grad_norm": 1.0059816314469427,
      "learning_rate": 1.4440290172835087e-06,
      "loss": 0.0268,
      "step": 2456
    },
    {
      "epoch": 2.330014224751067,
      "grad_norm": 0.41813610021235836,
      "learning_rate": 1.4401497698974875e-06,
      "loss": 0.028,
      "step": 2457
    },
    {
      "epoch": 2.3309625414888573,
      "grad_norm": 0.5405977205603913,
      "learning_rate": 1.4362748632825824e-06,
      "loss": 0.0432,
      "step": 2458
    },
    {
      "epoch": 2.3319108582266477,
      "grad_norm": 0.4101040764363063,
      "learning_rate": 1.4324043021637346e-06,
      "loss": 0.0215,
      "step": 2459
    },
    {
      "epoch": 2.332859174964438,
      "grad_norm": 0.4178607760421023,
      "learning_rate": 1.4285380912605846e-06,
      "loss": 0.0238,
      "step": 2460
    },
    {
      "epoch": 2.3338074917022285,
      "grad_norm": 0.34664363111136054,
      "learning_rate": 1.4246762352874732e-06,
      "loss": 0.014,
      "step": 2461
    },
    {
      "epoch": 2.334755808440019,
      "grad_norm": 0.35675426926357756,
      "learning_rate": 1.4208187389534256e-06,
      "loss": 0.0236,
      "step": 2462
    },
    {
      "epoch": 2.3357041251778092,
      "grad_norm": 0.5405249144315799,
      "learning_rate": 1.4169656069621529e-06,
      "loss": 0.0204,
      "step": 2463
    },
    {
      "epoch": 2.3366524419155996,
      "grad_norm": 0.4288922042641505,
      "learning_rate": 1.4131168440120473e-06,
      "loss": 0.0252,
      "step": 2464
    },
    {
      "epoch": 2.33760075865339,
      "grad_norm": 0.3140019976226754,
      "learning_rate": 1.4092724547961678e-06,
      "loss": 0.0131,
      "step": 2465
    },
    {
      "epoch": 2.338549075391181,
      "grad_norm": 0.29742347585209866,
      "learning_rate": 1.4054324440022477e-06,
      "loss": 0.0157,
      "step": 2466
    },
    {
      "epoch": 2.3394973921289712,
      "grad_norm": 0.4063964441515196,
      "learning_rate": 1.401596816312673e-06,
      "loss": 0.028,
      "step": 2467
    },
    {
      "epoch": 2.3404457088667616,
      "grad_norm": 0.3493753945743093,
      "learning_rate": 1.3977655764044917e-06,
      "loss": 0.0183,
      "step": 2468
    },
    {
      "epoch": 2.341394025604552,
      "grad_norm": 0.3833665376816688,
      "learning_rate": 1.3939387289494e-06,
      "loss": 0.0197,
      "step": 2469
    },
    {
      "epoch": 2.3423423423423424,
      "grad_norm": 0.4778883575723333,
      "learning_rate": 1.3901162786137345e-06,
      "loss": 0.0215,
      "step": 2470
    },
    {
      "epoch": 2.343290659080133,
      "grad_norm": 0.4491332932865572,
      "learning_rate": 1.3862982300584738e-06,
      "loss": 0.0298,
      "step": 2471
    },
    {
      "epoch": 2.344238975817923,
      "grad_norm": 0.4417496299388204,
      "learning_rate": 1.3824845879392301e-06,
      "loss": 0.019,
      "step": 2472
    },
    {
      "epoch": 2.3451872925557136,
      "grad_norm": 0.6247482118776613,
      "learning_rate": 1.3786753569062389e-06,
      "loss": 0.0177,
      "step": 2473
    },
    {
      "epoch": 2.346135609293504,
      "grad_norm": 0.33182737994484557,
      "learning_rate": 1.3748705416043584e-06,
      "loss": 0.0196,
      "step": 2474
    },
    {
      "epoch": 2.3470839260312943,
      "grad_norm": 0.33528750198268,
      "learning_rate": 1.3710701466730613e-06,
      "loss": 0.0208,
      "step": 2475
    },
    {
      "epoch": 2.3480322427690847,
      "grad_norm": 0.3574549420742016,
      "learning_rate": 1.3672741767464327e-06,
      "loss": 0.0194,
      "step": 2476
    },
    {
      "epoch": 2.348980559506875,
      "grad_norm": 0.27124601377608953,
      "learning_rate": 1.3634826364531616e-06,
      "loss": 0.0104,
      "step": 2477
    },
    {
      "epoch": 2.3499288762446655,
      "grad_norm": 0.30245014567972933,
      "learning_rate": 1.3596955304165333e-06,
      "loss": 0.0172,
      "step": 2478
    },
    {
      "epoch": 2.3508771929824563,
      "grad_norm": 0.3867837433390609,
      "learning_rate": 1.3559128632544277e-06,
      "loss": 0.0352,
      "step": 2479
    },
    {
      "epoch": 2.3518255097202467,
      "grad_norm": 0.3628249738004892,
      "learning_rate": 1.3521346395793145e-06,
      "loss": 0.0194,
      "step": 2480
    },
    {
      "epoch": 2.352773826458037,
      "grad_norm": 0.4847287370973444,
      "learning_rate": 1.3483608639982388e-06,
      "loss": 0.0304,
      "step": 2481
    },
    {
      "epoch": 2.3537221431958275,
      "grad_norm": 0.386773109078386,
      "learning_rate": 1.3445915411128296e-06,
      "loss": 0.0244,
      "step": 2482
    },
    {
      "epoch": 2.354670459933618,
      "grad_norm": 0.40717680421180374,
      "learning_rate": 1.3408266755192785e-06,
      "loss": 0.0152,
      "step": 2483
    },
    {
      "epoch": 2.3556187766714083,
      "grad_norm": 0.45152313917690334,
      "learning_rate": 1.3370662718083498e-06,
      "loss": 0.0278,
      "step": 2484
    },
    {
      "epoch": 2.3565670934091987,
      "grad_norm": 0.381547289327534,
      "learning_rate": 1.3333103345653615e-06,
      "loss": 0.0214,
      "step": 2485
    },
    {
      "epoch": 2.357515410146989,
      "grad_norm": 0.394442820160816,
      "learning_rate": 1.3295588683701854e-06,
      "loss": 0.022,
      "step": 2486
    },
    {
      "epoch": 2.3584637268847795,
      "grad_norm": 0.3030069689069135,
      "learning_rate": 1.325811877797245e-06,
      "loss": 0.0174,
      "step": 2487
    },
    {
      "epoch": 2.35941204362257,
      "grad_norm": 0.4007388320851803,
      "learning_rate": 1.3220693674155054e-06,
      "loss": 0.0284,
      "step": 2488
    },
    {
      "epoch": 2.3603603603603602,
      "grad_norm": 0.299749300305697,
      "learning_rate": 1.3183313417884664e-06,
      "loss": 0.0144,
      "step": 2489
    },
    {
      "epoch": 2.3613086770981506,
      "grad_norm": 0.36912145130254453,
      "learning_rate": 1.3145978054741621e-06,
      "loss": 0.0243,
      "step": 2490
    },
    {
      "epoch": 2.362256993835941,
      "grad_norm": 0.3261697534340822,
      "learning_rate": 1.3108687630251487e-06,
      "loss": 0.0164,
      "step": 2491
    },
    {
      "epoch": 2.363205310573732,
      "grad_norm": 0.4265509212089778,
      "learning_rate": 1.307144218988507e-06,
      "loss": 0.0359,
      "step": 2492
    },
    {
      "epoch": 2.3641536273115222,
      "grad_norm": 0.3568021639376206,
      "learning_rate": 1.303424177905831e-06,
      "loss": 0.0258,
      "step": 2493
    },
    {
      "epoch": 2.3651019440493126,
      "grad_norm": 0.42325489445166964,
      "learning_rate": 1.2997086443132218e-06,
      "loss": 0.0202,
      "step": 2494
    },
    {
      "epoch": 2.366050260787103,
      "grad_norm": 0.3616531140656992,
      "learning_rate": 1.2959976227412879e-06,
      "loss": 0.0162,
      "step": 2495
    },
    {
      "epoch": 2.3669985775248934,
      "grad_norm": 0.46009969632898884,
      "learning_rate": 1.2922911177151332e-06,
      "loss": 0.0269,
      "step": 2496
    },
    {
      "epoch": 2.367946894262684,
      "grad_norm": 0.45119455082431575,
      "learning_rate": 1.2885891337543539e-06,
      "loss": 0.0254,
      "step": 2497
    },
    {
      "epoch": 2.368895211000474,
      "grad_norm": 0.42641075850683124,
      "learning_rate": 1.2848916753730366e-06,
      "loss": 0.0202,
      "step": 2498
    },
    {
      "epoch": 2.3698435277382646,
      "grad_norm": 0.43843407405722246,
      "learning_rate": 1.2811987470797455e-06,
      "loss": 0.0228,
      "step": 2499
    },
    {
      "epoch": 2.370791844476055,
      "grad_norm": 0.3639166296679119,
      "learning_rate": 1.277510353377524e-06,
      "loss": 0.0235,
      "step": 2500
    },
    {
      "epoch": 2.3717401612138453,
      "grad_norm": 0.6231047022819872,
      "learning_rate": 1.2738264987638865e-06,
      "loss": 0.0219,
      "step": 2501
    },
    {
      "epoch": 2.3726884779516357,
      "grad_norm": 0.39227134362244037,
      "learning_rate": 1.2701471877308091e-06,
      "loss": 0.0223,
      "step": 2502
    },
    {
      "epoch": 2.373636794689426,
      "grad_norm": 0.33779641698396323,
      "learning_rate": 1.2664724247647303e-06,
      "loss": 0.0187,
      "step": 2503
    },
    {
      "epoch": 2.3745851114272165,
      "grad_norm": 0.3165810683945163,
      "learning_rate": 1.262802214346544e-06,
      "loss": 0.0156,
      "step": 2504
    },
    {
      "epoch": 2.3755334281650073,
      "grad_norm": 0.3233124265456635,
      "learning_rate": 1.2591365609515892e-06,
      "loss": 0.0181,
      "step": 2505
    },
    {
      "epoch": 2.3764817449027973,
      "grad_norm": 0.4607946489288941,
      "learning_rate": 1.2554754690496496e-06,
      "loss": 0.0176,
      "step": 2506
    },
    {
      "epoch": 2.377430061640588,
      "grad_norm": 0.5043713488293808,
      "learning_rate": 1.25181894310495e-06,
      "loss": 0.0324,
      "step": 2507
    },
    {
      "epoch": 2.3783783783783785,
      "grad_norm": 0.31629505288698206,
      "learning_rate": 1.248166987576141e-06,
      "loss": 0.0157,
      "step": 2508
    },
    {
      "epoch": 2.379326695116169,
      "grad_norm": 0.3803164685028412,
      "learning_rate": 1.2445196069163078e-06,
      "loss": 0.0218,
      "step": 2509
    },
    {
      "epoch": 2.3802750118539593,
      "grad_norm": 0.43944861644470423,
      "learning_rate": 1.240876805572951e-06,
      "loss": 0.0192,
      "step": 2510
    },
    {
      "epoch": 2.3812233285917497,
      "grad_norm": 0.36603580264615937,
      "learning_rate": 1.23723858798799e-06,
      "loss": 0.0211,
      "step": 2511
    },
    {
      "epoch": 2.38217164532954,
      "grad_norm": 0.39441769552699,
      "learning_rate": 1.2336049585977566e-06,
      "loss": 0.0206,
      "step": 2512
    },
    {
      "epoch": 2.3831199620673305,
      "grad_norm": 0.5104756079026775,
      "learning_rate": 1.2299759218329833e-06,
      "loss": 0.0182,
      "step": 2513
    },
    {
      "epoch": 2.384068278805121,
      "grad_norm": 0.4036452102453263,
      "learning_rate": 1.2263514821188078e-06,
      "loss": 0.0298,
      "step": 2514
    },
    {
      "epoch": 2.3850165955429112,
      "grad_norm": 0.34363638723562545,
      "learning_rate": 1.2227316438747566e-06,
      "loss": 0.0193,
      "step": 2515
    },
    {
      "epoch": 2.3859649122807016,
      "grad_norm": 0.36067365522749045,
      "learning_rate": 1.2191164115147513e-06,
      "loss": 0.016,
      "step": 2516
    },
    {
      "epoch": 2.386913229018492,
      "grad_norm": 0.33927962718163435,
      "learning_rate": 1.2155057894470928e-06,
      "loss": 0.0136,
      "step": 2517
    },
    {
      "epoch": 2.387861545756283,
      "grad_norm": 0.42725471109065416,
      "learning_rate": 1.211899782074461e-06,
      "loss": 0.0251,
      "step": 2518
    },
    {
      "epoch": 2.388809862494073,
      "grad_norm": 0.33443915124320717,
      "learning_rate": 1.2082983937939101e-06,
      "loss": 0.0137,
      "step": 2519
    },
    {
      "epoch": 2.3897581792318636,
      "grad_norm": 0.48954597652416665,
      "learning_rate": 1.2047016289968632e-06,
      "loss": 0.034,
      "step": 2520
    },
    {
      "epoch": 2.390706495969654,
      "grad_norm": 0.3683504845385266,
      "learning_rate": 1.201109492069102e-06,
      "loss": 0.0137,
      "step": 2521
    },
    {
      "epoch": 2.3916548127074444,
      "grad_norm": 0.4613644539571518,
      "learning_rate": 1.1975219873907679e-06,
      "loss": 0.0227,
      "step": 2522
    },
    {
      "epoch": 2.392603129445235,
      "grad_norm": 0.3541642000129723,
      "learning_rate": 1.193939119336352e-06,
      "loss": 0.0304,
      "step": 2523
    },
    {
      "epoch": 2.393551446183025,
      "grad_norm": 0.4363097081176452,
      "learning_rate": 1.1903608922746929e-06,
      "loss": 0.019,
      "step": 2524
    },
    {
      "epoch": 2.3944997629208156,
      "grad_norm": 0.48307067062916964,
      "learning_rate": 1.1867873105689714e-06,
      "loss": 0.0246,
      "step": 2525
    },
    {
      "epoch": 2.395448079658606,
      "grad_norm": 0.3539294327572431,
      "learning_rate": 1.1832183785767004e-06,
      "loss": 0.0198,
      "step": 2526
    },
    {
      "epoch": 2.3963963963963963,
      "grad_norm": 0.3538424483292842,
      "learning_rate": 1.179654100649727e-06,
      "loss": 0.017,
      "step": 2527
    },
    {
      "epoch": 2.3973447131341867,
      "grad_norm": 0.4009227484286087,
      "learning_rate": 1.17609448113422e-06,
      "loss": 0.0226,
      "step": 2528
    },
    {
      "epoch": 2.398293029871977,
      "grad_norm": 0.383841849833669,
      "learning_rate": 1.1725395243706678e-06,
      "loss": 0.0171,
      "step": 2529
    },
    {
      "epoch": 2.3992413466097675,
      "grad_norm": 0.4068400535839265,
      "learning_rate": 1.168989234693878e-06,
      "loss": 0.02,
      "step": 2530
    },
    {
      "epoch": 2.400189663347558,
      "grad_norm": 0.35983771900013944,
      "learning_rate": 1.1654436164329602e-06,
      "loss": 0.0166,
      "step": 2531
    },
    {
      "epoch": 2.4011379800853483,
      "grad_norm": 0.2995912937081424,
      "learning_rate": 1.1619026739113332e-06,
      "loss": 0.0143,
      "step": 2532
    },
    {
      "epoch": 2.402086296823139,
      "grad_norm": 0.3237682151824729,
      "learning_rate": 1.1583664114467146e-06,
      "loss": 0.0196,
      "step": 2533
    },
    {
      "epoch": 2.4030346135609295,
      "grad_norm": 0.3277525209186715,
      "learning_rate": 1.1548348333511095e-06,
      "loss": 0.0247,
      "step": 2534
    },
    {
      "epoch": 2.40398293029872,
      "grad_norm": 0.3855701948617413,
      "learning_rate": 1.1513079439308183e-06,
      "loss": 0.0187,
      "step": 2535
    },
    {
      "epoch": 2.4049312470365103,
      "grad_norm": 0.4217863375770685,
      "learning_rate": 1.147785747486418e-06,
      "loss": 0.0254,
      "step": 2536
    },
    {
      "epoch": 2.4058795637743007,
      "grad_norm": 0.40959180694037955,
      "learning_rate": 1.1442682483127686e-06,
      "loss": 0.0251,
      "step": 2537
    },
    {
      "epoch": 2.406827880512091,
      "grad_norm": 0.4370468363653208,
      "learning_rate": 1.1407554506989965e-06,
      "loss": 0.0223,
      "step": 2538
    },
    {
      "epoch": 2.4077761972498815,
      "grad_norm": 0.48429407932806134,
      "learning_rate": 1.1372473589285016e-06,
      "loss": 0.025,
      "step": 2539
    },
    {
      "epoch": 2.408724513987672,
      "grad_norm": 0.33025993789585734,
      "learning_rate": 1.1337439772789388e-06,
      "loss": 0.0157,
      "step": 2540
    },
    {
      "epoch": 2.4096728307254622,
      "grad_norm": 0.3487731113190881,
      "learning_rate": 1.1302453100222272e-06,
      "loss": 0.0202,
      "step": 2541
    },
    {
      "epoch": 2.4106211474632526,
      "grad_norm": 0.31353805106431143,
      "learning_rate": 1.1267513614245291e-06,
      "loss": 0.0154,
      "step": 2542
    },
    {
      "epoch": 2.411569464201043,
      "grad_norm": 0.3389412149212434,
      "learning_rate": 1.1232621357462609e-06,
      "loss": 0.0205,
      "step": 2543
    },
    {
      "epoch": 2.4125177809388334,
      "grad_norm": 0.38136622760182587,
      "learning_rate": 1.1197776372420733e-06,
      "loss": 0.0219,
      "step": 2544
    },
    {
      "epoch": 2.413466097676624,
      "grad_norm": 0.27921173877988736,
      "learning_rate": 1.1162978701608574e-06,
      "loss": 0.0153,
      "step": 2545
    },
    {
      "epoch": 2.4144144144144146,
      "grad_norm": 0.40968982802283244,
      "learning_rate": 1.112822838745734e-06,
      "loss": 0.0285,
      "step": 2546
    },
    {
      "epoch": 2.415362731152205,
      "grad_norm": 0.4062864100779731,
      "learning_rate": 1.1093525472340471e-06,
      "loss": 0.0239,
      "step": 2547
    },
    {
      "epoch": 2.4163110478899954,
      "grad_norm": 0.3736524597587387,
      "learning_rate": 1.105886999857365e-06,
      "loss": 0.0225,
      "step": 2548
    },
    {
      "epoch": 2.417259364627786,
      "grad_norm": 0.3515641886656497,
      "learning_rate": 1.1024262008414677e-06,
      "loss": 0.028,
      "step": 2549
    },
    {
      "epoch": 2.418207681365576,
      "grad_norm": 0.6471429652164479,
      "learning_rate": 1.098970154406344e-06,
      "loss": 0.0262,
      "step": 2550
    },
    {
      "epoch": 2.4191559981033666,
      "grad_norm": 0.3650914464417128,
      "learning_rate": 1.095518864766194e-06,
      "loss": 0.0182,
      "step": 2551
    },
    {
      "epoch": 2.420104314841157,
      "grad_norm": 0.4376020269032313,
      "learning_rate": 1.09207233612941e-06,
      "loss": 0.0153,
      "step": 2552
    },
    {
      "epoch": 2.4210526315789473,
      "grad_norm": 0.34862166575759534,
      "learning_rate": 1.088630572698583e-06,
      "loss": 0.0162,
      "step": 2553
    },
    {
      "epoch": 2.4220009483167377,
      "grad_norm": 0.5082871266947292,
      "learning_rate": 1.0851935786704954e-06,
      "loss": 0.0272,
      "step": 2554
    },
    {
      "epoch": 2.422949265054528,
      "grad_norm": 0.41158101422665216,
      "learning_rate": 1.0817613582361074e-06,
      "loss": 0.0229,
      "step": 2555
    },
    {
      "epoch": 2.4238975817923185,
      "grad_norm": 0.43603252981364155,
      "learning_rate": 1.0783339155805644e-06,
      "loss": 0.027,
      "step": 2556
    },
    {
      "epoch": 2.424845898530109,
      "grad_norm": 0.49024524037096656,
      "learning_rate": 1.0749112548831848e-06,
      "loss": 0.0254,
      "step": 2557
    },
    {
      "epoch": 2.4257942152678993,
      "grad_norm": 0.3814270330954172,
      "learning_rate": 1.0714933803174526e-06,
      "loss": 0.0213,
      "step": 2558
    },
    {
      "epoch": 2.42674253200569,
      "grad_norm": 0.3976340849529511,
      "learning_rate": 1.0680802960510207e-06,
      "loss": 0.0227,
      "step": 2559
    },
    {
      "epoch": 2.4276908487434805,
      "grad_norm": 0.35166175212564016,
      "learning_rate": 1.0646720062456968e-06,
      "loss": 0.0201,
      "step": 2560
    },
    {
      "epoch": 2.428639165481271,
      "grad_norm": 0.5056479105854602,
      "learning_rate": 1.0612685150574432e-06,
      "loss": 0.0251,
      "step": 2561
    },
    {
      "epoch": 2.4295874822190613,
      "grad_norm": 0.49161815028338257,
      "learning_rate": 1.0578698266363734e-06,
      "loss": 0.0198,
      "step": 2562
    },
    {
      "epoch": 2.4305357989568517,
      "grad_norm": 0.41629650229173676,
      "learning_rate": 1.054475945126741e-06,
      "loss": 0.0216,
      "step": 2563
    },
    {
      "epoch": 2.431484115694642,
      "grad_norm": 0.4745068679798503,
      "learning_rate": 1.051086874666941e-06,
      "loss": 0.027,
      "step": 2564
    },
    {
      "epoch": 2.4324324324324325,
      "grad_norm": 0.5231051341052511,
      "learning_rate": 1.0477026193895024e-06,
      "loss": 0.0225,
      "step": 2565
    },
    {
      "epoch": 2.433380749170223,
      "grad_norm": 0.5996082247716173,
      "learning_rate": 1.0443231834210781e-06,
      "loss": 0.0343,
      "step": 2566
    },
    {
      "epoch": 2.4343290659080132,
      "grad_norm": 0.3254194676503433,
      "learning_rate": 1.0409485708824506e-06,
      "loss": 0.0172,
      "step": 2567
    },
    {
      "epoch": 2.4352773826458036,
      "grad_norm": 0.4487239432643308,
      "learning_rate": 1.0375787858885162e-06,
      "loss": 0.0294,
      "step": 2568
    },
    {
      "epoch": 2.436225699383594,
      "grad_norm": 0.46358117621037215,
      "learning_rate": 1.0342138325482859e-06,
      "loss": 0.023,
      "step": 2569
    },
    {
      "epoch": 2.4371740161213844,
      "grad_norm": 0.36382789199564586,
      "learning_rate": 1.0308537149648823e-06,
      "loss": 0.0166,
      "step": 2570
    },
    {
      "epoch": 2.438122332859175,
      "grad_norm": 0.5348352187185549,
      "learning_rate": 1.0274984372355273e-06,
      "loss": 0.0288,
      "step": 2571
    },
    {
      "epoch": 2.4390706495969656,
      "grad_norm": 0.3585580345192679,
      "learning_rate": 1.0241480034515406e-06,
      "loss": 0.023,
      "step": 2572
    },
    {
      "epoch": 2.4400189663347556,
      "grad_norm": 0.3805010713917909,
      "learning_rate": 1.02080241769834e-06,
      "loss": 0.0205,
      "step": 2573
    },
    {
      "epoch": 2.4409672830725464,
      "grad_norm": 0.3113821678519327,
      "learning_rate": 1.0174616840554274e-06,
      "loss": 0.0182,
      "step": 2574
    },
    {
      "epoch": 2.441915599810337,
      "grad_norm": 0.3576257333753975,
      "learning_rate": 1.0141258065963916e-06,
      "loss": 0.0182,
      "step": 2575
    },
    {
      "epoch": 2.442863916548127,
      "grad_norm": 0.4104979653412175,
      "learning_rate": 1.0107947893888965e-06,
      "loss": 0.0207,
      "step": 2576
    },
    {
      "epoch": 2.4438122332859176,
      "grad_norm": 0.521831837691463,
      "learning_rate": 1.0074686364946823e-06,
      "loss": 0.0288,
      "step": 2577
    },
    {
      "epoch": 2.444760550023708,
      "grad_norm": 0.44124587907058904,
      "learning_rate": 1.0041473519695577e-06,
      "loss": 0.0206,
      "step": 2578
    },
    {
      "epoch": 2.4457088667614983,
      "grad_norm": 0.435513491122848,
      "learning_rate": 1.000830939863392e-06,
      "loss": 0.0222,
      "step": 2579
    },
    {
      "epoch": 2.4466571834992887,
      "grad_norm": 0.5135158628816288,
      "learning_rate": 9.97519404220118e-07,
      "loss": 0.0534,
      "step": 2580
    },
    {
      "epoch": 2.447605500237079,
      "grad_norm": 0.40646237085276543,
      "learning_rate": 9.94212749077717e-07,
      "loss": 0.027,
      "step": 2581
    },
    {
      "epoch": 2.4485538169748695,
      "grad_norm": 0.7325890802201012,
      "learning_rate": 9.90910978468224e-07,
      "loss": 0.0369,
      "step": 2582
    },
    {
      "epoch": 2.44950213371266,
      "grad_norm": 0.4343579637391866,
      "learning_rate": 9.876140964177162e-07,
      "loss": 0.0218,
      "step": 2583
    },
    {
      "epoch": 2.4504504504504503,
      "grad_norm": 0.5766706870308571,
      "learning_rate": 9.84322106946306e-07,
      "loss": 0.0237,
      "step": 2584
    },
    {
      "epoch": 2.4513987671882407,
      "grad_norm": 0.39849795851452613,
      "learning_rate": 9.810350140681457e-07,
      "loss": 0.0182,
      "step": 2585
    },
    {
      "epoch": 2.452347083926031,
      "grad_norm": 0.4783557947198164,
      "learning_rate": 9.777528217914162e-07,
      "loss": 0.0294,
      "step": 2586
    },
    {
      "epoch": 2.453295400663822,
      "grad_norm": 0.42456242270260697,
      "learning_rate": 9.744755341183181e-07,
      "loss": 0.0168,
      "step": 2587
    },
    {
      "epoch": 2.4542437174016123,
      "grad_norm": 0.39678351340169365,
      "learning_rate": 9.712031550450774e-07,
      "loss": 0.0174,
      "step": 2588
    },
    {
      "epoch": 2.4551920341394027,
      "grad_norm": 0.47949218889264555,
      "learning_rate": 9.679356885619284e-07,
      "loss": 0.0262,
      "step": 2589
    },
    {
      "epoch": 2.456140350877193,
      "grad_norm": 0.3236340258110289,
      "learning_rate": 9.646731386531204e-07,
      "loss": 0.0166,
      "step": 2590
    },
    {
      "epoch": 2.4570886676149835,
      "grad_norm": 0.4597376524730793,
      "learning_rate": 9.61415509296907e-07,
      "loss": 0.0259,
      "step": 2591
    },
    {
      "epoch": 2.458036984352774,
      "grad_norm": 0.37074395951240846,
      "learning_rate": 9.581628044655394e-07,
      "loss": 0.0224,
      "step": 2592
    },
    {
      "epoch": 2.4589853010905642,
      "grad_norm": 0.37957880087331625,
      "learning_rate": 9.549150281252633e-07,
      "loss": 0.0251,
      "step": 2593
    },
    {
      "epoch": 2.4599336178283546,
      "grad_norm": 0.29271615580129845,
      "learning_rate": 9.516721842363197e-07,
      "loss": 0.0146,
      "step": 2594
    },
    {
      "epoch": 2.460881934566145,
      "grad_norm": 0.35582409975735413,
      "learning_rate": 9.484342767529292e-07,
      "loss": 0.0187,
      "step": 2595
    },
    {
      "epoch": 2.4618302513039354,
      "grad_norm": 0.3500315779102406,
      "learning_rate": 9.452013096232981e-07,
      "loss": 0.0208,
      "step": 2596
    },
    {
      "epoch": 2.462778568041726,
      "grad_norm": 0.3382420903851306,
      "learning_rate": 9.419732867896048e-07,
      "loss": 0.0255,
      "step": 2597
    },
    {
      "epoch": 2.463726884779516,
      "grad_norm": 0.578123966367445,
      "learning_rate": 9.38750212188001e-07,
      "loss": 0.0407,
      "step": 2598
    },
    {
      "epoch": 2.4646752015173066,
      "grad_norm": 0.43085532284231304,
      "learning_rate": 9.355320897486053e-07,
      "loss": 0.02,
      "step": 2599
    },
    {
      "epoch": 2.4656235182550974,
      "grad_norm": 0.35880891064401854,
      "learning_rate": 9.323189233954937e-07,
      "loss": 0.0205,
      "step": 2600
    },
    {
      "epoch": 2.466571834992888,
      "grad_norm": 0.40312804249813705,
      "learning_rate": 9.291107170467034e-07,
      "loss": 0.0221,
      "step": 2601
    },
    {
      "epoch": 2.467520151730678,
      "grad_norm": 0.6304307314634919,
      "learning_rate": 9.259074746142238e-07,
      "loss": 0.0199,
      "step": 2602
    },
    {
      "epoch": 2.4684684684684686,
      "grad_norm": 0.40697982149629996,
      "learning_rate": 9.227092000039867e-07,
      "loss": 0.0338,
      "step": 2603
    },
    {
      "epoch": 2.469416785206259,
      "grad_norm": 0.36786978507872714,
      "learning_rate": 9.195158971158702e-07,
      "loss": 0.0184,
      "step": 2604
    },
    {
      "epoch": 2.4703651019440493,
      "grad_norm": 0.6005718767374743,
      "learning_rate": 9.16327569843688e-07,
      "loss": 0.0295,
      "step": 2605
    },
    {
      "epoch": 2.4713134186818397,
      "grad_norm": 0.565930456400567,
      "learning_rate": 9.131442220751874e-07,
      "loss": 0.031,
      "step": 2606
    },
    {
      "epoch": 2.47226173541963,
      "grad_norm": 0.42111140772170896,
      "learning_rate": 9.099658576920467e-07,
      "loss": 0.0233,
      "step": 2607
    },
    {
      "epoch": 2.4732100521574205,
      "grad_norm": 0.47019365608336894,
      "learning_rate": 9.067924805698619e-07,
      "loss": 0.0263,
      "step": 2608
    },
    {
      "epoch": 2.474158368895211,
      "grad_norm": 0.4078998491617388,
      "learning_rate": 9.036240945781527e-07,
      "loss": 0.0185,
      "step": 2609
    },
    {
      "epoch": 2.4751066856330013,
      "grad_norm": 0.40269682929382367,
      "learning_rate": 9.004607035803508e-07,
      "loss": 0.0216,
      "step": 2610
    },
    {
      "epoch": 2.4760550023707917,
      "grad_norm": 0.2937728488786148,
      "learning_rate": 8.973023114337958e-07,
      "loss": 0.0126,
      "step": 2611
    },
    {
      "epoch": 2.477003319108582,
      "grad_norm": 0.33063640858727344,
      "learning_rate": 8.941489219897354e-07,
      "loss": 0.017,
      "step": 2612
    },
    {
      "epoch": 2.477951635846373,
      "grad_norm": 0.38049725788517497,
      "learning_rate": 8.910005390933124e-07,
      "loss": 0.0194,
      "step": 2613
    },
    {
      "epoch": 2.4788999525841633,
      "grad_norm": 0.5046033844115824,
      "learning_rate": 8.8785716658357e-07,
      "loss": 0.0261,
      "step": 2614
    },
    {
      "epoch": 2.4798482693219537,
      "grad_norm": 0.5299027510329987,
      "learning_rate": 8.847188082934383e-07,
      "loss": 0.0282,
      "step": 2615
    },
    {
      "epoch": 2.480796586059744,
      "grad_norm": 0.39419036803568774,
      "learning_rate": 8.815854680497327e-07,
      "loss": 0.0218,
      "step": 2616
    },
    {
      "epoch": 2.4817449027975345,
      "grad_norm": 0.3955758545159766,
      "learning_rate": 8.78457149673152e-07,
      "loss": 0.0228,
      "step": 2617
    },
    {
      "epoch": 2.482693219535325,
      "grad_norm": 0.47313816441818424,
      "learning_rate": 8.753338569782732e-07,
      "loss": 0.0221,
      "step": 2618
    },
    {
      "epoch": 2.4836415362731152,
      "grad_norm": 0.271371398644424,
      "learning_rate": 8.722155937735394e-07,
      "loss": 0.01,
      "step": 2619
    },
    {
      "epoch": 2.4845898530109056,
      "grad_norm": 0.3398216687335157,
      "learning_rate": 8.691023638612661e-07,
      "loss": 0.0163,
      "step": 2620
    },
    {
      "epoch": 2.485538169748696,
      "grad_norm": 0.6048741244957424,
      "learning_rate": 8.659941710376279e-07,
      "loss": 0.0258,
      "step": 2621
    },
    {
      "epoch": 2.4864864864864864,
      "grad_norm": 0.3436798347323189,
      "learning_rate": 8.628910190926599e-07,
      "loss": 0.0181,
      "step": 2622
    },
    {
      "epoch": 2.487434803224277,
      "grad_norm": 0.4156429457098121,
      "learning_rate": 8.59792911810251e-07,
      "loss": 0.0237,
      "step": 2623
    },
    {
      "epoch": 2.488383119962067,
      "grad_norm": 0.2954143857319232,
      "learning_rate": 8.566998529681337e-07,
      "loss": 0.0132,
      "step": 2624
    },
    {
      "epoch": 2.4893314366998576,
      "grad_norm": 0.39413478905762644,
      "learning_rate": 8.536118463378918e-07,
      "loss": 0.0162,
      "step": 2625
    },
    {
      "epoch": 2.4902797534376484,
      "grad_norm": 0.3892273346253101,
      "learning_rate": 8.505288956849428e-07,
      "loss": 0.0153,
      "step": 2626
    },
    {
      "epoch": 2.4912280701754383,
      "grad_norm": 0.6174085428601906,
      "learning_rate": 8.474510047685408e-07,
      "loss": 0.0311,
      "step": 2627
    },
    {
      "epoch": 2.492176386913229,
      "grad_norm": 0.4357016874336171,
      "learning_rate": 8.443781773417726e-07,
      "loss": 0.029,
      "step": 2628
    },
    {
      "epoch": 2.4931247036510196,
      "grad_norm": 0.34487830360142213,
      "learning_rate": 8.413104171515462e-07,
      "loss": 0.0146,
      "step": 2629
    },
    {
      "epoch": 2.49407302038881,
      "grad_norm": 0.36125275627714193,
      "learning_rate": 8.382477279385953e-07,
      "loss": 0.0226,
      "step": 2630
    },
    {
      "epoch": 2.4950213371266003,
      "grad_norm": 0.42552335088399396,
      "learning_rate": 8.35190113437469e-07,
      "loss": 0.0264,
      "step": 2631
    },
    {
      "epoch": 2.4959696538643907,
      "grad_norm": 0.34791091511753236,
      "learning_rate": 8.321375773765256e-07,
      "loss": 0.0175,
      "step": 2632
    },
    {
      "epoch": 2.496917970602181,
      "grad_norm": 0.3954500689347747,
      "learning_rate": 8.290901234779336e-07,
      "loss": 0.0188,
      "step": 2633
    },
    {
      "epoch": 2.4978662873399715,
      "grad_norm": 0.3663693629694859,
      "learning_rate": 8.260477554576657e-07,
      "loss": 0.0171,
      "step": 2634
    },
    {
      "epoch": 2.498814604077762,
      "grad_norm": 0.5494442209985945,
      "learning_rate": 8.230104770254899e-07,
      "loss": 0.0367,
      "step": 2635
    },
    {
      "epoch": 2.4997629208155523,
      "grad_norm": 0.5528576688557691,
      "learning_rate": 8.199782918849681e-07,
      "loss": 0.0285,
      "step": 2636
    },
    {
      "epoch": 2.5007112375533427,
      "grad_norm": 0.6404166113112415,
      "learning_rate": 8.169512037334553e-07,
      "loss": 0.0261,
      "step": 2637
    },
    {
      "epoch": 2.501659554291133,
      "grad_norm": 0.4286300123661097,
      "learning_rate": 8.139292162620866e-07,
      "loss": 0.0217,
      "step": 2638
    },
    {
      "epoch": 2.502607871028924,
      "grad_norm": 0.42741045341288486,
      "learning_rate": 8.109123331557828e-07,
      "loss": 0.021,
      "step": 2639
    },
    {
      "epoch": 2.503556187766714,
      "grad_norm": 0.4501781665468482,
      "learning_rate": 8.079005580932353e-07,
      "loss": 0.0245,
      "step": 2640
    },
    {
      "epoch": 2.5045045045045047,
      "grad_norm": 0.5105968099593348,
      "learning_rate": 8.048938947469109e-07,
      "loss": 0.0274,
      "step": 2641
    },
    {
      "epoch": 2.505452821242295,
      "grad_norm": 0.41552893159895754,
      "learning_rate": 8.018923467830403e-07,
      "loss": 0.0211,
      "step": 2642
    },
    {
      "epoch": 2.5064011379800855,
      "grad_norm": 0.4791684882671791,
      "learning_rate": 7.988959178616184e-07,
      "loss": 0.022,
      "step": 2643
    },
    {
      "epoch": 2.507349454717876,
      "grad_norm": 0.3439927281902292,
      "learning_rate": 7.959046116364e-07,
      "loss": 0.0147,
      "step": 2644
    },
    {
      "epoch": 2.5082977714556662,
      "grad_norm": 0.3146069088165673,
      "learning_rate": 7.929184317548888e-07,
      "loss": 0.014,
      "step": 2645
    },
    {
      "epoch": 2.5092460881934566,
      "grad_norm": 0.3109625989087509,
      "learning_rate": 7.899373818583417e-07,
      "loss": 0.0146,
      "step": 2646
    },
    {
      "epoch": 2.510194404931247,
      "grad_norm": 0.5903370862543935,
      "learning_rate": 7.869614655817576e-07,
      "loss": 0.0266,
      "step": 2647
    },
    {
      "epoch": 2.5111427216690374,
      "grad_norm": 0.42346350090954193,
      "learning_rate": 7.839906865538754e-07,
      "loss": 0.0228,
      "step": 2648
    },
    {
      "epoch": 2.512091038406828,
      "grad_norm": 0.5644352189518651,
      "learning_rate": 7.810250483971743e-07,
      "loss": 0.0177,
      "step": 2649
    },
    {
      "epoch": 2.513039355144618,
      "grad_norm": 0.6800732019791883,
      "learning_rate": 7.780645547278587e-07,
      "loss": 0.0388,
      "step": 2650
    },
    {
      "epoch": 2.5139876718824086,
      "grad_norm": 0.3672619635328786,
      "learning_rate": 7.751092091558632e-07,
      "loss": 0.0175,
      "step": 2651
    },
    {
      "epoch": 2.5149359886201994,
      "grad_norm": 0.36697647870892186,
      "learning_rate": 7.721590152848474e-07,
      "loss": 0.0186,
      "step": 2652
    },
    {
      "epoch": 2.5158843053579893,
      "grad_norm": 0.3949457428777839,
      "learning_rate": 7.692139767121826e-07,
      "loss": 0.013,
      "step": 2653
    },
    {
      "epoch": 2.51683262209578,
      "grad_norm": 0.36894744299049603,
      "learning_rate": 7.662740970289595e-07,
      "loss": 0.0206,
      "step": 2654
    },
    {
      "epoch": 2.5177809388335706,
      "grad_norm": 0.3004729811076744,
      "learning_rate": 7.633393798199778e-07,
      "loss": 0.0145,
      "step": 2655
    },
    {
      "epoch": 2.518729255571361,
      "grad_norm": 0.30582539821490057,
      "learning_rate": 7.604098286637379e-07,
      "loss": 0.0151,
      "step": 2656
    },
    {
      "epoch": 2.5196775723091513,
      "grad_norm": 0.41815777077524247,
      "learning_rate": 7.574854471324461e-07,
      "loss": 0.0253,
      "step": 2657
    },
    {
      "epoch": 2.5206258890469417,
      "grad_norm": 0.31145601353840013,
      "learning_rate": 7.545662387920016e-07,
      "loss": 0.0165,
      "step": 2658
    },
    {
      "epoch": 2.521574205784732,
      "grad_norm": 0.3562269510719666,
      "learning_rate": 7.516522072019955e-07,
      "loss": 0.0158,
      "step": 2659
    },
    {
      "epoch": 2.5225225225225225,
      "grad_norm": 0.4008350356129895,
      "learning_rate": 7.487433559157098e-07,
      "loss": 0.0171,
      "step": 2660
    },
    {
      "epoch": 2.523470839260313,
      "grad_norm": 0.6137367883176392,
      "learning_rate": 7.458396884801056e-07,
      "loss": 0.0244,
      "step": 2661
    },
    {
      "epoch": 2.5244191559981033,
      "grad_norm": 0.4813680993375203,
      "learning_rate": 7.429412084358262e-07,
      "loss": 0.0269,
      "step": 2662
    },
    {
      "epoch": 2.5253674727358937,
      "grad_norm": 0.46215577248037415,
      "learning_rate": 7.400479193171895e-07,
      "loss": 0.0266,
      "step": 2663
    },
    {
      "epoch": 2.526315789473684,
      "grad_norm": 0.37793669989539486,
      "learning_rate": 7.371598246521805e-07,
      "loss": 0.0158,
      "step": 2664
    },
    {
      "epoch": 2.527264106211475,
      "grad_norm": 0.41136067655717384,
      "learning_rate": 7.34276927962454e-07,
      "loss": 0.0238,
      "step": 2665
    },
    {
      "epoch": 2.528212422949265,
      "grad_norm": 0.33259267222332123,
      "learning_rate": 7.313992327633235e-07,
      "loss": 0.0209,
      "step": 2666
    },
    {
      "epoch": 2.5291607396870557,
      "grad_norm": 0.5819528717109627,
      "learning_rate": 7.285267425637621e-07,
      "loss": 0.0172,
      "step": 2667
    },
    {
      "epoch": 2.5301090564248456,
      "grad_norm": 0.30059814565316323,
      "learning_rate": 7.256594608663963e-07,
      "loss": 0.0185,
      "step": 2668
    },
    {
      "epoch": 2.5310573731626365,
      "grad_norm": 0.3712777470405421,
      "learning_rate": 7.227973911675002e-07,
      "loss": 0.0179,
      "step": 2669
    },
    {
      "epoch": 2.532005689900427,
      "grad_norm": 0.29855322296258097,
      "learning_rate": 7.199405369569911e-07,
      "loss": 0.0149,
      "step": 2670
    },
    {
      "epoch": 2.5329540066382172,
      "grad_norm": 0.40829074659712467,
      "learning_rate": 7.170889017184313e-07,
      "loss": 0.0187,
      "step": 2671
    },
    {
      "epoch": 2.5339023233760076,
      "grad_norm": 0.4407705399218079,
      "learning_rate": 7.142424889290139e-07,
      "loss": 0.0197,
      "step": 2672
    },
    {
      "epoch": 2.534850640113798,
      "grad_norm": 0.4866101217898875,
      "learning_rate": 7.114013020595684e-07,
      "loss": 0.0251,
      "step": 2673
    },
    {
      "epoch": 2.5357989568515884,
      "grad_norm": 0.40018962733942437,
      "learning_rate": 7.085653445745488e-07,
      "loss": 0.0233,
      "step": 2674
    },
    {
      "epoch": 2.536747273589379,
      "grad_norm": 0.383701601440475,
      "learning_rate": 7.057346199320342e-07,
      "loss": 0.0275,
      "step": 2675
    },
    {
      "epoch": 2.537695590327169,
      "grad_norm": 0.5349755091825953,
      "learning_rate": 7.029091315837245e-07,
      "loss": 0.0217,
      "step": 2676
    },
    {
      "epoch": 2.5386439070649596,
      "grad_norm": 0.3679483002865713,
      "learning_rate": 7.000888829749292e-07,
      "loss": 0.0158,
      "step": 2677
    },
    {
      "epoch": 2.53959222380275,
      "grad_norm": 0.35330307265879174,
      "learning_rate": 6.972738775445747e-07,
      "loss": 0.0191,
      "step": 2678
    },
    {
      "epoch": 2.5405405405405403,
      "grad_norm": 0.38659927985192755,
      "learning_rate": 6.9446411872519e-07,
      "loss": 0.0211,
      "step": 2679
    },
    {
      "epoch": 2.541488857278331,
      "grad_norm": 0.4064147988484457,
      "learning_rate": 6.916596099429096e-07,
      "loss": 0.0233,
      "step": 2680
    },
    {
      "epoch": 2.542437174016121,
      "grad_norm": 0.6580118862505251,
      "learning_rate": 6.888603546174638e-07,
      "loss": 0.0245,
      "step": 2681
    },
    {
      "epoch": 2.543385490753912,
      "grad_norm": 0.3944629105173342,
      "learning_rate": 6.860663561621767e-07,
      "loss": 0.0258,
      "step": 2682
    },
    {
      "epoch": 2.5443338074917023,
      "grad_norm": 0.33272900655389953,
      "learning_rate": 6.83277617983964e-07,
      "loss": 0.0179,
      "step": 2683
    },
    {
      "epoch": 2.5452821242294927,
      "grad_norm": 0.4167482558841912,
      "learning_rate": 6.804941434833285e-07,
      "loss": 0.0177,
      "step": 2684
    },
    {
      "epoch": 2.546230440967283,
      "grad_norm": 0.3992502938558958,
      "learning_rate": 6.777159360543501e-07,
      "loss": 0.0274,
      "step": 2685
    },
    {
      "epoch": 2.5471787577050735,
      "grad_norm": 0.40094207085691724,
      "learning_rate": 6.749429990846901e-07,
      "loss": 0.0274,
      "step": 2686
    },
    {
      "epoch": 2.548127074442864,
      "grad_norm": 0.3792531400747548,
      "learning_rate": 6.721753359555833e-07,
      "loss": 0.0241,
      "step": 2687
    },
    {
      "epoch": 2.5490753911806543,
      "grad_norm": 0.37741471606051397,
      "learning_rate": 6.6941295004183e-07,
      "loss": 0.0195,
      "step": 2688
    },
    {
      "epoch": 2.5500237079184447,
      "grad_norm": 0.32470565833329973,
      "learning_rate": 6.666558447118005e-07,
      "loss": 0.015,
      "step": 2689
    },
    {
      "epoch": 2.550972024656235,
      "grad_norm": 0.4293192559707278,
      "learning_rate": 6.639040233274225e-07,
      "loss": 0.0217,
      "step": 2690
    },
    {
      "epoch": 2.5519203413940255,
      "grad_norm": 0.3929905923840855,
      "learning_rate": 6.611574892441802e-07,
      "loss": 0.0196,
      "step": 2691
    },
    {
      "epoch": 2.552868658131816,
      "grad_norm": 0.3561960449808376,
      "learning_rate": 6.584162458111148e-07,
      "loss": 0.0226,
      "step": 2692
    },
    {
      "epoch": 2.5538169748696067,
      "grad_norm": 0.37662449677307064,
      "learning_rate": 6.556802963708115e-07,
      "loss": 0.0276,
      "step": 2693
    },
    {
      "epoch": 2.5547652916073966,
      "grad_norm": 0.5129824661775971,
      "learning_rate": 6.52949644259403e-07,
      "loss": 0.0362,
      "step": 2694
    },
    {
      "epoch": 2.5557136083451875,
      "grad_norm": 0.6692404460687522,
      "learning_rate": 6.502242928065633e-07,
      "loss": 0.0332,
      "step": 2695
    },
    {
      "epoch": 2.556661925082978,
      "grad_norm": 0.393927786278907,
      "learning_rate": 6.475042453354996e-07,
      "loss": 0.0206,
      "step": 2696
    },
    {
      "epoch": 2.5576102418207682,
      "grad_norm": 0.3577618677497919,
      "learning_rate": 6.44789505162955e-07,
      "loss": 0.0175,
      "step": 2697
    },
    {
      "epoch": 2.5585585585585586,
      "grad_norm": 0.3661222623781511,
      "learning_rate": 6.42080075599198e-07,
      "loss": 0.0117,
      "step": 2698
    },
    {
      "epoch": 2.559506875296349,
      "grad_norm": 0.30688738276260336,
      "learning_rate": 6.393759599480243e-07,
      "loss": 0.0161,
      "step": 2699
    },
    {
      "epoch": 2.5604551920341394,
      "grad_norm": 0.41630163331920195,
      "learning_rate": 6.366771615067497e-07,
      "loss": 0.0189,
      "step": 2700
    },
    {
      "epoch": 2.56140350877193,
      "grad_norm": 0.5340246040237304,
      "learning_rate": 6.339836835662039e-07,
      "loss": 0.0275,
      "step": 2701
    },
    {
      "epoch": 2.56235182550972,
      "grad_norm": 0.35423728423688666,
      "learning_rate": 6.312955294107304e-07,
      "loss": 0.0158,
      "step": 2702
    },
    {
      "epoch": 2.5633001422475106,
      "grad_norm": 0.47028395732329115,
      "learning_rate": 6.28612702318181e-07,
      "loss": 0.0289,
      "step": 2703
    },
    {
      "epoch": 2.564248458985301,
      "grad_norm": 0.4005227388788351,
      "learning_rate": 6.259352055599127e-07,
      "loss": 0.032,
      "step": 2704
    },
    {
      "epoch": 2.5651967757230913,
      "grad_norm": 0.467516363616176,
      "learning_rate": 6.23263042400783e-07,
      "loss": 0.0212,
      "step": 2705
    },
    {
      "epoch": 2.566145092460882,
      "grad_norm": 0.31564569013460975,
      "learning_rate": 6.205962160991424e-07,
      "loss": 0.0175,
      "step": 2706
    },
    {
      "epoch": 2.567093409198672,
      "grad_norm": 0.3373546761437316,
      "learning_rate": 6.179347299068377e-07,
      "loss": 0.0142,
      "step": 2707
    },
    {
      "epoch": 2.568041725936463,
      "grad_norm": 0.47280981766734664,
      "learning_rate": 6.152785870692041e-07,
      "loss": 0.0223,
      "step": 2708
    },
    {
      "epoch": 2.5689900426742533,
      "grad_norm": 0.36467322171181704,
      "learning_rate": 6.126277908250572e-07,
      "loss": 0.0187,
      "step": 2709
    },
    {
      "epoch": 2.5699383594120437,
      "grad_norm": 0.4297930835831616,
      "learning_rate": 6.099823444066982e-07,
      "loss": 0.0292,
      "step": 2710
    },
    {
      "epoch": 2.570886676149834,
      "grad_norm": 0.41906846780451784,
      "learning_rate": 6.073422510399007e-07,
      "loss": 0.0252,
      "step": 2711
    },
    {
      "epoch": 2.5718349928876245,
      "grad_norm": 0.38650910584228265,
      "learning_rate": 6.047075139439151e-07,
      "loss": 0.0226,
      "step": 2712
    },
    {
      "epoch": 2.572783309625415,
      "grad_norm": 0.5084741884777527,
      "learning_rate": 6.02078136331456e-07,
      "loss": 0.0195,
      "step": 2713
    },
    {
      "epoch": 2.5737316263632053,
      "grad_norm": 0.39278830005678883,
      "learning_rate": 5.994541214087052e-07,
      "loss": 0.021,
      "step": 2714
    },
    {
      "epoch": 2.5746799431009957,
      "grad_norm": 0.2952909468393027,
      "learning_rate": 5.968354723753056e-07,
      "loss": 0.0151,
      "step": 2715
    },
    {
      "epoch": 2.575628259838786,
      "grad_norm": 0.37045235908101837,
      "learning_rate": 5.94222192424358e-07,
      "loss": 0.0161,
      "step": 2716
    },
    {
      "epoch": 2.5765765765765765,
      "grad_norm": 0.3281172799358503,
      "learning_rate": 5.916142847424127e-07,
      "loss": 0.0148,
      "step": 2717
    },
    {
      "epoch": 2.577524893314367,
      "grad_norm": 0.4924919864359691,
      "learning_rate": 5.890117525094741e-07,
      "loss": 0.0285,
      "step": 2718
    },
    {
      "epoch": 2.5784732100521577,
      "grad_norm": 0.4481794894764316,
      "learning_rate": 5.864145988989867e-07,
      "loss": 0.0194,
      "step": 2719
    },
    {
      "epoch": 2.5794215267899476,
      "grad_norm": 0.5321347244112973,
      "learning_rate": 5.838228270778407e-07,
      "loss": 0.0204,
      "step": 2720
    },
    {
      "epoch": 2.5803698435277385,
      "grad_norm": 0.31872189522321803,
      "learning_rate": 5.812364402063636e-07,
      "loss": 0.0169,
      "step": 2721
    },
    {
      "epoch": 2.581318160265529,
      "grad_norm": 0.41681981250054423,
      "learning_rate": 5.786554414383128e-07,
      "loss": 0.0201,
      "step": 2722
    },
    {
      "epoch": 2.5822664770033192,
      "grad_norm": 0.5067386260406679,
      "learning_rate": 5.760798339208807e-07,
      "loss": 0.0234,
      "step": 2723
    },
    {
      "epoch": 2.5832147937411096,
      "grad_norm": 0.33542737410974843,
      "learning_rate": 5.735096207946822e-07,
      "loss": 0.0151,
      "step": 2724
    },
    {
      "epoch": 2.5841631104789,
      "grad_norm": 0.6467429650160105,
      "learning_rate": 5.709448051937544e-07,
      "loss": 0.0204,
      "step": 2725
    },
    {
      "epoch": 2.5851114272166904,
      "grad_norm": 0.38155909066198,
      "learning_rate": 5.683853902455561e-07,
      "loss": 0.0217,
      "step": 2726
    },
    {
      "epoch": 2.586059743954481,
      "grad_norm": 0.28662641689056056,
      "learning_rate": 5.658313790709569e-07,
      "loss": 0.0119,
      "step": 2727
    },
    {
      "epoch": 2.587008060692271,
      "grad_norm": 0.3990322895513413,
      "learning_rate": 5.632827747842395e-07,
      "loss": 0.023,
      "step": 2728
    },
    {
      "epoch": 2.5879563774300616,
      "grad_norm": 0.40716429826152295,
      "learning_rate": 5.607395804930943e-07,
      "loss": 0.0207,
      "step": 2729
    },
    {
      "epoch": 2.588904694167852,
      "grad_norm": 0.35974137294779707,
      "learning_rate": 5.582017992986122e-07,
      "loss": 0.0249,
      "step": 2730
    },
    {
      "epoch": 2.5898530109056423,
      "grad_norm": 0.38646462128999254,
      "learning_rate": 5.556694342952845e-07,
      "loss": 0.0186,
      "step": 2731
    },
    {
      "epoch": 2.590801327643433,
      "grad_norm": 0.37653004258963946,
      "learning_rate": 5.531424885710012e-07,
      "loss": 0.0268,
      "step": 2732
    },
    {
      "epoch": 2.591749644381223,
      "grad_norm": 0.4299208034382125,
      "learning_rate": 5.506209652070399e-07,
      "loss": 0.0349,
      "step": 2733
    },
    {
      "epoch": 2.592697961119014,
      "grad_norm": 0.38861976736094617,
      "learning_rate": 5.48104867278067e-07,
      "loss": 0.018,
      "step": 2734
    },
    {
      "epoch": 2.593646277856804,
      "grad_norm": 0.4175379738102784,
      "learning_rate": 5.455941978521368e-07,
      "loss": 0.0168,
      "step": 2735
    },
    {
      "epoch": 2.5945945945945947,
      "grad_norm": 0.33966670471790067,
      "learning_rate": 5.430889599906792e-07,
      "loss": 0.0151,
      "step": 2736
    },
    {
      "epoch": 2.595542911332385,
      "grad_norm": 0.3835190152377659,
      "learning_rate": 5.405891567485066e-07,
      "loss": 0.0213,
      "step": 2737
    },
    {
      "epoch": 2.5964912280701755,
      "grad_norm": 0.4113531869569931,
      "learning_rate": 5.380947911737988e-07,
      "loss": 0.0314,
      "step": 2738
    },
    {
      "epoch": 2.597439544807966,
      "grad_norm": 0.40520964068776627,
      "learning_rate": 5.356058663081093e-07,
      "loss": 0.0264,
      "step": 2739
    },
    {
      "epoch": 2.5983878615457563,
      "grad_norm": 0.7837946356297472,
      "learning_rate": 5.33122385186357e-07,
      "loss": 0.0217,
      "step": 2740
    },
    {
      "epoch": 2.5993361782835467,
      "grad_norm": 0.8806203038130418,
      "learning_rate": 5.306443508368197e-07,
      "loss": 0.0387,
      "step": 2741
    },
    {
      "epoch": 2.600284495021337,
      "grad_norm": 0.34074270045547633,
      "learning_rate": 5.281717662811381e-07,
      "loss": 0.0149,
      "step": 2742
    },
    {
      "epoch": 2.6012328117591275,
      "grad_norm": 0.4477584975020037,
      "learning_rate": 5.257046345343031e-07,
      "loss": 0.0246,
      "step": 2743
    },
    {
      "epoch": 2.602181128496918,
      "grad_norm": 0.5551990298380998,
      "learning_rate": 5.2324295860466e-07,
      "loss": 0.0254,
      "step": 2744
    },
    {
      "epoch": 2.6031294452347082,
      "grad_norm": 0.45009786762895765,
      "learning_rate": 5.207867414939005e-07,
      "loss": 0.024,
      "step": 2745
    },
    {
      "epoch": 2.6040777619724986,
      "grad_norm": 0.4832226736950542,
      "learning_rate": 5.183359861970572e-07,
      "loss": 0.0297,
      "step": 2746
    },
    {
      "epoch": 2.6050260787102895,
      "grad_norm": 0.3779678120598059,
      "learning_rate": 5.158906957025079e-07,
      "loss": 0.0137,
      "step": 2747
    },
    {
      "epoch": 2.6059743954480794,
      "grad_norm": 0.35142073959413767,
      "learning_rate": 5.134508729919635e-07,
      "loss": 0.0179,
      "step": 2748
    },
    {
      "epoch": 2.6069227121858702,
      "grad_norm": 0.40042406085470267,
      "learning_rate": 5.110165210404672e-07,
      "loss": 0.0225,
      "step": 2749
    },
    {
      "epoch": 2.6078710289236606,
      "grad_norm": 0.33663608338691814,
      "learning_rate": 5.085876428163938e-07,
      "loss": 0.0294,
      "step": 2750
    },
    {
      "epoch": 2.608819345661451,
      "grad_norm": 0.433579962152764,
      "learning_rate": 5.061642412814405e-07,
      "loss": 0.0188,
      "step": 2751
    },
    {
      "epoch": 2.6097676623992414,
      "grad_norm": 0.31497843130176856,
      "learning_rate": 5.037463193906295e-07,
      "loss": 0.018,
      "step": 2752
    },
    {
      "epoch": 2.610715979137032,
      "grad_norm": 0.4392939445838396,
      "learning_rate": 5.013338800923001e-07,
      "loss": 0.0217,
      "step": 2753
    },
    {
      "epoch": 2.611664295874822,
      "grad_norm": 0.38382293150188085,
      "learning_rate": 4.989269263281044e-07,
      "loss": 0.0216,
      "step": 2754
    },
    {
      "epoch": 2.6126126126126126,
      "grad_norm": 0.4649233732841508,
      "learning_rate": 4.965254610330089e-07,
      "loss": 0.032,
      "step": 2755
    },
    {
      "epoch": 2.613560929350403,
      "grad_norm": 0.37494406277381903,
      "learning_rate": 4.941294871352859e-07,
      "loss": 0.0223,
      "step": 2756
    },
    {
      "epoch": 2.6145092460881934,
      "grad_norm": 0.42108122053404773,
      "learning_rate": 4.917390075565098e-07,
      "loss": 0.0231,
      "step": 2757
    },
    {
      "epoch": 2.6154575628259837,
      "grad_norm": 0.47265684925500273,
      "learning_rate": 4.893540252115597e-07,
      "loss": 0.0209,
      "step": 2758
    },
    {
      "epoch": 2.616405879563774,
      "grad_norm": 0.3344680664465188,
      "learning_rate": 4.869745430086076e-07,
      "loss": 0.0143,
      "step": 2759
    },
    {
      "epoch": 2.617354196301565,
      "grad_norm": 0.38459266477883736,
      "learning_rate": 4.846005638491213e-07,
      "loss": 0.0203,
      "step": 2760
    },
    {
      "epoch": 2.618302513039355,
      "grad_norm": 0.3412043515396964,
      "learning_rate": 4.822320906278577e-07,
      "loss": 0.0139,
      "step": 2761
    },
    {
      "epoch": 2.6192508297771457,
      "grad_norm": 0.5208903375959671,
      "learning_rate": 4.798691262328586e-07,
      "loss": 0.0243,
      "step": 2762
    },
    {
      "epoch": 2.620199146514936,
      "grad_norm": 0.39653893800416085,
      "learning_rate": 4.775116735454511e-07,
      "loss": 0.0253,
      "step": 2763
    },
    {
      "epoch": 2.6211474632527265,
      "grad_norm": 0.2892258222331325,
      "learning_rate": 4.751597354402382e-07,
      "loss": 0.0145,
      "step": 2764
    },
    {
      "epoch": 2.622095779990517,
      "grad_norm": 0.40845586230105796,
      "learning_rate": 4.728133147851011e-07,
      "loss": 0.0234,
      "step": 2765
    },
    {
      "epoch": 2.6230440967283073,
      "grad_norm": 0.30239125683950585,
      "learning_rate": 4.7047241444119376e-07,
      "loss": 0.0143,
      "step": 2766
    },
    {
      "epoch": 2.6239924134660977,
      "grad_norm": 0.4211431383939641,
      "learning_rate": 4.681370372629368e-07,
      "loss": 0.027,
      "step": 2767
    },
    {
      "epoch": 2.624940730203888,
      "grad_norm": 0.5032346373585926,
      "learning_rate": 4.6580718609801513e-07,
      "loss": 0.0185,
      "step": 2768
    },
    {
      "epoch": 2.6258890469416785,
      "grad_norm": 0.3655746633014726,
      "learning_rate": 4.634828637873795e-07,
      "loss": 0.0223,
      "step": 2769
    },
    {
      "epoch": 2.626837363679469,
      "grad_norm": 0.3074549708482703,
      "learning_rate": 4.6116407316523446e-07,
      "loss": 0.018,
      "step": 2770
    },
    {
      "epoch": 2.6277856804172592,
      "grad_norm": 0.46834675787128,
      "learning_rate": 4.5885081705904334e-07,
      "loss": 0.0255,
      "step": 2771
    },
    {
      "epoch": 2.6287339971550496,
      "grad_norm": 0.37778189298548,
      "learning_rate": 4.565430982895175e-07,
      "loss": 0.0229,
      "step": 2772
    },
    {
      "epoch": 2.6296823138928405,
      "grad_norm": 0.46883682144212735,
      "learning_rate": 4.5424091967061877e-07,
      "loss": 0.0157,
      "step": 2773
    },
    {
      "epoch": 2.6306306306306304,
      "grad_norm": 0.40055616739928074,
      "learning_rate": 4.5194428400955224e-07,
      "loss": 0.0208,
      "step": 2774
    },
    {
      "epoch": 2.6315789473684212,
      "grad_norm": 0.5249183820066241,
      "learning_rate": 4.496531941067639e-07,
      "loss": 0.0247,
      "step": 2775
    },
    {
      "epoch": 2.6325272641062116,
      "grad_norm": 0.34712677696632804,
      "learning_rate": 4.4736765275593897e-07,
      "loss": 0.0216,
      "step": 2776
    },
    {
      "epoch": 2.633475580844002,
      "grad_norm": 0.3929916373129789,
      "learning_rate": 4.4508766274399426e-07,
      "loss": 0.0142,
      "step": 2777
    },
    {
      "epoch": 2.6344238975817924,
      "grad_norm": 0.4849560252243947,
      "learning_rate": 4.428132268510793e-07,
      "loss": 0.0219,
      "step": 2778
    },
    {
      "epoch": 2.635372214319583,
      "grad_norm": 0.36600196074929786,
      "learning_rate": 4.4054434785057163e-07,
      "loss": 0.0164,
      "step": 2779
    },
    {
      "epoch": 2.636320531057373,
      "grad_norm": 0.5093185802679411,
      "learning_rate": 4.3828102850907104e-07,
      "loss": 0.0236,
      "step": 2780
    },
    {
      "epoch": 2.6372688477951636,
      "grad_norm": 0.3862188682640724,
      "learning_rate": 4.360232715863988e-07,
      "loss": 0.0264,
      "step": 2781
    },
    {
      "epoch": 2.638217164532954,
      "grad_norm": 0.3574239021740862,
      "learning_rate": 4.33771079835596e-07,
      "loss": 0.0162,
      "step": 2782
    },
    {
      "epoch": 2.6391654812707444,
      "grad_norm": 0.4706427605900885,
      "learning_rate": 4.3152445600291306e-07,
      "loss": 0.0289,
      "step": 2783
    },
    {
      "epoch": 2.6401137980085347,
      "grad_norm": 0.3667202133626864,
      "learning_rate": 4.2928340282781366e-07,
      "loss": 0.0204,
      "step": 2784
    },
    {
      "epoch": 2.641062114746325,
      "grad_norm": 0.413719793835401,
      "learning_rate": 4.270479230429708e-07,
      "loss": 0.0235,
      "step": 2785
    },
    {
      "epoch": 2.642010431484116,
      "grad_norm": 0.44950116717007177,
      "learning_rate": 4.2481801937425735e-07,
      "loss": 0.0223,
      "step": 2786
    },
    {
      "epoch": 2.642958748221906,
      "grad_norm": 0.35663470393335955,
      "learning_rate": 4.225936945407494e-07,
      "loss": 0.014,
      "step": 2787
    },
    {
      "epoch": 2.6439070649596967,
      "grad_norm": 0.39202001192001895,
      "learning_rate": 4.203749512547201e-07,
      "loss": 0.0196,
      "step": 2788
    },
    {
      "epoch": 2.6448553816974867,
      "grad_norm": 0.32486780874552906,
      "learning_rate": 4.1816179222163525e-07,
      "loss": 0.0215,
      "step": 2789
    },
    {
      "epoch": 2.6458036984352775,
      "grad_norm": 0.3723048747718762,
      "learning_rate": 4.1595422014015343e-07,
      "loss": 0.0224,
      "step": 2790
    },
    {
      "epoch": 2.646752015173068,
      "grad_norm": 0.357232800265432,
      "learning_rate": 4.137522377021186e-07,
      "loss": 0.0171,
      "step": 2791
    },
    {
      "epoch": 2.6477003319108583,
      "grad_norm": 0.4698743666062284,
      "learning_rate": 4.115558475925602e-07,
      "loss": 0.0254,
      "step": 2792
    },
    {
      "epoch": 2.6486486486486487,
      "grad_norm": 0.4382212268806192,
      "learning_rate": 4.093650524896897e-07,
      "loss": 0.0275,
      "step": 2793
    },
    {
      "epoch": 2.649596965386439,
      "grad_norm": 0.4043616799679715,
      "learning_rate": 4.0717985506489366e-07,
      "loss": 0.0235,
      "step": 2794
    },
    {
      "epoch": 2.6505452821242295,
      "grad_norm": 0.3392698836350075,
      "learning_rate": 4.050002579827345e-07,
      "loss": 0.0146,
      "step": 2795
    },
    {
      "epoch": 2.65149359886202,
      "grad_norm": 0.3722272428398198,
      "learning_rate": 4.028262639009445e-07,
      "loss": 0.0164,
      "step": 2796
    },
    {
      "epoch": 2.6524419155998102,
      "grad_norm": 0.47666670751210216,
      "learning_rate": 4.0065787547042543e-07,
      "loss": 0.0247,
      "step": 2797
    },
    {
      "epoch": 2.6533902323376006,
      "grad_norm": 0.35724907861541844,
      "learning_rate": 3.9849509533524343e-07,
      "loss": 0.0162,
      "step": 2798
    },
    {
      "epoch": 2.654338549075391,
      "grad_norm": 0.6380750662875044,
      "learning_rate": 3.9633792613262557e-07,
      "loss": 0.0328,
      "step": 2799
    },
    {
      "epoch": 2.6552868658131814,
      "grad_norm": 0.4876904102585979,
      "learning_rate": 3.941863704929555e-07,
      "loss": 0.0269,
      "step": 2800
    },
    {
      "epoch": 2.6562351825509722,
      "grad_norm": 0.3549869940847376,
      "learning_rate": 3.920404310397763e-07,
      "loss": 0.0202,
      "step": 2801
    },
    {
      "epoch": 2.657183499288762,
      "grad_norm": 0.5769104527453145,
      "learning_rate": 3.89900110389777e-07,
      "loss": 0.023,
      "step": 2802
    },
    {
      "epoch": 2.658131816026553,
      "grad_norm": 0.5050863547101014,
      "learning_rate": 3.877654111528012e-07,
      "loss": 0.027,
      "step": 2803
    },
    {
      "epoch": 2.6590801327643434,
      "grad_norm": 0.35737691730875915,
      "learning_rate": 3.8563633593183217e-07,
      "loss": 0.0162,
      "step": 2804
    },
    {
      "epoch": 2.660028449502134,
      "grad_norm": 0.37236396191380444,
      "learning_rate": 3.835128873230004e-07,
      "loss": 0.0209,
      "step": 2805
    },
    {
      "epoch": 2.660976766239924,
      "grad_norm": 0.32787817974246386,
      "learning_rate": 3.813950679155731e-07,
      "loss": 0.0174,
      "step": 2806
    },
    {
      "epoch": 2.6619250829777146,
      "grad_norm": 0.3052178572544167,
      "learning_rate": 3.7928288029195294e-07,
      "loss": 0.0141,
      "step": 2807
    },
    {
      "epoch": 2.662873399715505,
      "grad_norm": 0.5978262864524307,
      "learning_rate": 3.771763270276768e-07,
      "loss": 0.0313,
      "step": 2808
    },
    {
      "epoch": 2.6638217164532954,
      "grad_norm": 0.5112733834365627,
      "learning_rate": 3.750754106914101e-07,
      "loss": 0.0221,
      "step": 2809
    },
    {
      "epoch": 2.6647700331910857,
      "grad_norm": 0.39109717270185107,
      "learning_rate": 3.729801338449451e-07,
      "loss": 0.0211,
      "step": 2810
    },
    {
      "epoch": 2.665718349928876,
      "grad_norm": 0.34630995301406425,
      "learning_rate": 3.708904990431983e-07,
      "loss": 0.0245,
      "step": 2811
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 0.3894835061084856,
      "learning_rate": 3.688065088342041e-07,
      "loss": 0.0135,
      "step": 2812
    },
    {
      "epoch": 2.667614983404457,
      "grad_norm": 0.413966885464809,
      "learning_rate": 3.6672816575911597e-07,
      "loss": 0.0286,
      "step": 2813
    },
    {
      "epoch": 2.6685633001422477,
      "grad_norm": 0.3333453955002542,
      "learning_rate": 3.646554723522028e-07,
      "loss": 0.0156,
      "step": 2814
    },
    {
      "epoch": 2.6695116168800377,
      "grad_norm": 0.5143420713686682,
      "learning_rate": 3.6258843114084066e-07,
      "loss": 0.0197,
      "step": 2815
    },
    {
      "epoch": 2.6704599336178285,
      "grad_norm": 0.3937058493387821,
      "learning_rate": 3.6052704464551723e-07,
      "loss": 0.0269,
      "step": 2816
    },
    {
      "epoch": 2.671408250355619,
      "grad_norm": 0.34107596225123676,
      "learning_rate": 3.584713153798214e-07,
      "loss": 0.0163,
      "step": 2817
    },
    {
      "epoch": 2.6723565670934093,
      "grad_norm": 0.4433497311363595,
      "learning_rate": 3.5642124585044736e-07,
      "loss": 0.0264,
      "step": 2818
    },
    {
      "epoch": 2.6733048838311997,
      "grad_norm": 0.47719708455229654,
      "learning_rate": 3.5437683855718663e-07,
      "loss": 0.0249,
      "step": 2819
    },
    {
      "epoch": 2.67425320056899,
      "grad_norm": 0.37082517315713176,
      "learning_rate": 3.5233809599292565e-07,
      "loss": 0.0216,
      "step": 2820
    },
    {
      "epoch": 2.6752015173067805,
      "grad_norm": 0.4449107865815135,
      "learning_rate": 3.5030502064364314e-07,
      "loss": 0.0242,
      "step": 2821
    },
    {
      "epoch": 2.676149834044571,
      "grad_norm": 0.7583947129307697,
      "learning_rate": 3.482776149884093e-07,
      "loss": 0.0206,
      "step": 2822
    },
    {
      "epoch": 2.6770981507823612,
      "grad_norm": 0.47130851405721447,
      "learning_rate": 3.4625588149937883e-07,
      "loss": 0.0246,
      "step": 2823
    },
    {
      "epoch": 2.6780464675201516,
      "grad_norm": 0.40574178102286834,
      "learning_rate": 3.44239822641792e-07,
      "loss": 0.0211,
      "step": 2824
    },
    {
      "epoch": 2.678994784257942,
      "grad_norm": 0.39016069764421896,
      "learning_rate": 3.4222944087396737e-07,
      "loss": 0.02,
      "step": 2825
    },
    {
      "epoch": 2.6799431009957324,
      "grad_norm": 0.4193631351806872,
      "learning_rate": 3.402247386473029e-07,
      "loss": 0.0262,
      "step": 2826
    },
    {
      "epoch": 2.6808914177335232,
      "grad_norm": 0.46922685051431373,
      "learning_rate": 3.3822571840627095e-07,
      "loss": 0.0356,
      "step": 2827
    },
    {
      "epoch": 2.681839734471313,
      "grad_norm": 0.3653043259783921,
      "learning_rate": 3.362323825884134e-07,
      "loss": 0.0208,
      "step": 2828
    },
    {
      "epoch": 2.682788051209104,
      "grad_norm": 0.3426904459665249,
      "learning_rate": 3.342447336243437e-07,
      "loss": 0.0153,
      "step": 2829
    },
    {
      "epoch": 2.6837363679468944,
      "grad_norm": 0.36215857651916467,
      "learning_rate": 3.322627739377388e-07,
      "loss": 0.018,
      "step": 2830
    },
    {
      "epoch": 2.684684684684685,
      "grad_norm": 0.3802416370751486,
      "learning_rate": 3.3028650594533975e-07,
      "loss": 0.0268,
      "step": 2831
    },
    {
      "epoch": 2.685633001422475,
      "grad_norm": 0.36281885778655093,
      "learning_rate": 3.283159320569451e-07,
      "loss": 0.0195,
      "step": 2832
    },
    {
      "epoch": 2.6865813181602656,
      "grad_norm": 0.39875935883241936,
      "learning_rate": 3.263510546754117e-07,
      "loss": 0.0175,
      "step": 2833
    },
    {
      "epoch": 2.687529634898056,
      "grad_norm": 0.4168338240621692,
      "learning_rate": 3.243918761966508e-07,
      "loss": 0.0244,
      "step": 2834
    },
    {
      "epoch": 2.6884779516358464,
      "grad_norm": 0.47057878935135905,
      "learning_rate": 3.224383990096247e-07,
      "loss": 0.0229,
      "step": 2835
    },
    {
      "epoch": 2.6894262683736367,
      "grad_norm": 0.38441852189692455,
      "learning_rate": 3.2049062549634137e-07,
      "loss": 0.0249,
      "step": 2836
    },
    {
      "epoch": 2.690374585111427,
      "grad_norm": 0.5637198135695414,
      "learning_rate": 3.1854855803185645e-07,
      "loss": 0.0212,
      "step": 2837
    },
    {
      "epoch": 2.6913229018492175,
      "grad_norm": 0.37781951972907957,
      "learning_rate": 3.1661219898426743e-07,
      "loss": 0.0145,
      "step": 2838
    },
    {
      "epoch": 2.692271218587008,
      "grad_norm": 0.35734390206818134,
      "learning_rate": 3.1468155071470885e-07,
      "loss": 0.0192,
      "step": 2839
    },
    {
      "epoch": 2.6932195353247987,
      "grad_norm": 0.4077467536996209,
      "learning_rate": 3.127566155773554e-07,
      "loss": 0.0201,
      "step": 2840
    },
    {
      "epoch": 2.6941678520625887,
      "grad_norm": 0.5028083771887946,
      "learning_rate": 3.1083739591941167e-07,
      "loss": 0.0383,
      "step": 2841
    },
    {
      "epoch": 2.6951161688003795,
      "grad_norm": 0.501841795908295,
      "learning_rate": 3.0892389408111625e-07,
      "loss": 0.0235,
      "step": 2842
    },
    {
      "epoch": 2.69606448553817,
      "grad_norm": 0.38475074034060736,
      "learning_rate": 3.070161123957327e-07,
      "loss": 0.0221,
      "step": 2843
    },
    {
      "epoch": 2.6970128022759603,
      "grad_norm": 0.4020329048390891,
      "learning_rate": 3.0511405318955136e-07,
      "loss": 0.0259,
      "step": 2844
    },
    {
      "epoch": 2.6979611190137507,
      "grad_norm": 0.48369634537926337,
      "learning_rate": 3.0321771878188353e-07,
      "loss": 0.019,
      "step": 2845
    },
    {
      "epoch": 2.698909435751541,
      "grad_norm": 1.0740629753507862,
      "learning_rate": 3.0132711148506243e-07,
      "loss": 0.02,
      "step": 2846
    },
    {
      "epoch": 2.6998577524893315,
      "grad_norm": 0.3695371754345027,
      "learning_rate": 2.994422336044345e-07,
      "loss": 0.0185,
      "step": 2847
    },
    {
      "epoch": 2.700806069227122,
      "grad_norm": 0.5038388629520099,
      "learning_rate": 2.9756308743836284e-07,
      "loss": 0.0208,
      "step": 2848
    },
    {
      "epoch": 2.7017543859649122,
      "grad_norm": 0.7640080852891192,
      "learning_rate": 2.9568967527821846e-07,
      "loss": 0.0249,
      "step": 2849
    },
    {
      "epoch": 2.7027027027027026,
      "grad_norm": 0.46768948100137725,
      "learning_rate": 2.93821999408383e-07,
      "loss": 0.0204,
      "step": 2850
    },
    {
      "epoch": 2.703651019440493,
      "grad_norm": 0.4319992020883265,
      "learning_rate": 2.9196006210624296e-07,
      "loss": 0.0328,
      "step": 2851
    },
    {
      "epoch": 2.7045993361782834,
      "grad_norm": 0.41315615466424443,
      "learning_rate": 2.9010386564218616e-07,
      "loss": 0.0153,
      "step": 2852
    },
    {
      "epoch": 2.7055476529160742,
      "grad_norm": 0.3744469122321566,
      "learning_rate": 2.882534122796016e-07,
      "loss": 0.0209,
      "step": 2853
    },
    {
      "epoch": 2.706495969653864,
      "grad_norm": 0.42369368397512175,
      "learning_rate": 2.864087042748753e-07,
      "loss": 0.0218,
      "step": 2854
    },
    {
      "epoch": 2.707444286391655,
      "grad_norm": 0.5051296882817911,
      "learning_rate": 2.8456974387738535e-07,
      "loss": 0.0215,
      "step": 2855
    },
    {
      "epoch": 2.708392603129445,
      "grad_norm": 0.41590105364038504,
      "learning_rate": 2.827365333295051e-07,
      "loss": 0.0213,
      "step": 2856
    },
    {
      "epoch": 2.709340919867236,
      "grad_norm": 0.587547148927632,
      "learning_rate": 2.8090907486659324e-07,
      "loss": 0.0334,
      "step": 2857
    },
    {
      "epoch": 2.710289236605026,
      "grad_norm": 0.34887434178067056,
      "learning_rate": 2.790873707169961e-07,
      "loss": 0.017,
      "step": 2858
    },
    {
      "epoch": 2.7112375533428166,
      "grad_norm": 0.39187346235873055,
      "learning_rate": 2.7727142310204515e-07,
      "loss": 0.0203,
      "step": 2859
    },
    {
      "epoch": 2.712185870080607,
      "grad_norm": 0.41720690104657004,
      "learning_rate": 2.75461234236048e-07,
      "loss": 0.0167,
      "step": 2860
    },
    {
      "epoch": 2.7131341868183974,
      "grad_norm": 0.31518371371061915,
      "learning_rate": 2.7365680632629455e-07,
      "loss": 0.0144,
      "step": 2861
    },
    {
      "epoch": 2.7140825035561877,
      "grad_norm": 0.2855027969557929,
      "learning_rate": 2.718581415730481e-07,
      "loss": 0.0137,
      "step": 2862
    },
    {
      "epoch": 2.715030820293978,
      "grad_norm": 0.3870774282480667,
      "learning_rate": 2.7006524216954543e-07,
      "loss": 0.0253,
      "step": 2863
    },
    {
      "epoch": 2.7159791370317685,
      "grad_norm": 0.38164386495554165,
      "learning_rate": 2.682781103019905e-07,
      "loss": 0.0179,
      "step": 2864
    },
    {
      "epoch": 2.716927453769559,
      "grad_norm": 0.4248661479785991,
      "learning_rate": 2.664967481495584e-07,
      "loss": 0.0326,
      "step": 2865
    },
    {
      "epoch": 2.7178757705073493,
      "grad_norm": 0.5433988511920199,
      "learning_rate": 2.647211578843856e-07,
      "loss": 0.0247,
      "step": 2866
    },
    {
      "epoch": 2.7188240872451397,
      "grad_norm": 0.39724442593767356,
      "learning_rate": 2.629513416715734e-07,
      "loss": 0.0188,
      "step": 2867
    },
    {
      "epoch": 2.7197724039829305,
      "grad_norm": 0.33528171893218883,
      "learning_rate": 2.611873016691796e-07,
      "loss": 0.0202,
      "step": 2868
    },
    {
      "epoch": 2.7207207207207205,
      "grad_norm": 0.3965938814255718,
      "learning_rate": 2.594290400282201e-07,
      "loss": 0.0295,
      "step": 2869
    },
    {
      "epoch": 2.7216690374585113,
      "grad_norm": 0.30564880056172694,
      "learning_rate": 2.576765588926661e-07,
      "loss": 0.0153,
      "step": 2870
    },
    {
      "epoch": 2.7226173541963017,
      "grad_norm": 0.3137314915113477,
      "learning_rate": 2.5592986039943703e-07,
      "loss": 0.0141,
      "step": 2871
    },
    {
      "epoch": 2.723565670934092,
      "grad_norm": 0.49092832578220685,
      "learning_rate": 2.5418894667840363e-07,
      "loss": 0.0277,
      "step": 2872
    },
    {
      "epoch": 2.7245139876718825,
      "grad_norm": 0.4015470991385358,
      "learning_rate": 2.524538198523824e-07,
      "loss": 0.0188,
      "step": 2873
    },
    {
      "epoch": 2.725462304409673,
      "grad_norm": 0.41058168996354416,
      "learning_rate": 2.5072448203713294e-07,
      "loss": 0.0191,
      "step": 2874
    },
    {
      "epoch": 2.7264106211474632,
      "grad_norm": 0.5285881611083112,
      "learning_rate": 2.4900093534135695e-07,
      "loss": 0.0299,
      "step": 2875
    },
    {
      "epoch": 2.7273589378852536,
      "grad_norm": 0.40022195777865227,
      "learning_rate": 2.472831818666921e-07,
      "loss": 0.0209,
      "step": 2876
    },
    {
      "epoch": 2.728307254623044,
      "grad_norm": 0.46750753879370927,
      "learning_rate": 2.455712237077157e-07,
      "loss": 0.0266,
      "step": 2877
    },
    {
      "epoch": 2.7292555713608344,
      "grad_norm": 0.3959499198022272,
      "learning_rate": 2.438650629519351e-07,
      "loss": 0.0284,
      "step": 2878
    },
    {
      "epoch": 2.730203888098625,
      "grad_norm": 0.43008633569157745,
      "learning_rate": 2.421647016797907e-07,
      "loss": 0.0176,
      "step": 2879
    },
    {
      "epoch": 2.731152204836415,
      "grad_norm": 0.37556553750404836,
      "learning_rate": 2.404701419646505e-07,
      "loss": 0.019,
      "step": 2880
    },
    {
      "epoch": 2.732100521574206,
      "grad_norm": 0.5032204383672553,
      "learning_rate": 2.3878138587280717e-07,
      "loss": 0.0323,
      "step": 2881
    },
    {
      "epoch": 2.733048838311996,
      "grad_norm": 0.3505093748403086,
      "learning_rate": 2.3709843546347865e-07,
      "loss": 0.0236,
      "step": 2882
    },
    {
      "epoch": 2.733997155049787,
      "grad_norm": 0.2970624200348729,
      "learning_rate": 2.354212927888022e-07,
      "loss": 0.0135,
      "step": 2883
    },
    {
      "epoch": 2.734945471787577,
      "grad_norm": 0.465166372609731,
      "learning_rate": 2.3374995989383366e-07,
      "loss": 0.0181,
      "step": 2884
    },
    {
      "epoch": 2.7358937885253676,
      "grad_norm": 0.44064948151774963,
      "learning_rate": 2.3208443881654442e-07,
      "loss": 0.0105,
      "step": 2885
    },
    {
      "epoch": 2.736842105263158,
      "grad_norm": 0.41355456746628383,
      "learning_rate": 2.3042473158781998e-07,
      "loss": 0.0198,
      "step": 2886
    },
    {
      "epoch": 2.7377904220009484,
      "grad_norm": 0.35065718913864835,
      "learning_rate": 2.2877084023145425e-07,
      "loss": 0.0167,
      "step": 2887
    },
    {
      "epoch": 2.7387387387387387,
      "grad_norm": 0.39535153479567114,
      "learning_rate": 2.2712276676415346e-07,
      "loss": 0.0266,
      "step": 2888
    },
    {
      "epoch": 2.739687055476529,
      "grad_norm": 0.4217230635090741,
      "learning_rate": 2.254805131955251e-07,
      "loss": 0.0202,
      "step": 2889
    },
    {
      "epoch": 2.7406353722143195,
      "grad_norm": 0.384980173712967,
      "learning_rate": 2.238440815280829e-07,
      "loss": 0.0196,
      "step": 2890
    },
    {
      "epoch": 2.74158368895211,
      "grad_norm": 0.4678666261730461,
      "learning_rate": 2.222134737572429e-07,
      "loss": 0.041,
      "step": 2891
    },
    {
      "epoch": 2.7425320056899003,
      "grad_norm": 0.3519750639388043,
      "learning_rate": 2.2058869187131514e-07,
      "loss": 0.0211,
      "step": 2892
    },
    {
      "epoch": 2.7434803224276907,
      "grad_norm": 0.3512574692443541,
      "learning_rate": 2.1896973785150987e-07,
      "loss": 0.0153,
      "step": 2893
    },
    {
      "epoch": 2.7444286391654815,
      "grad_norm": 0.6136655914796003,
      "learning_rate": 2.173566136719285e-07,
      "loss": 0.0256,
      "step": 2894
    },
    {
      "epoch": 2.7453769559032715,
      "grad_norm": 0.40007142163498555,
      "learning_rate": 2.1574932129956483e-07,
      "loss": 0.0149,
      "step": 2895
    },
    {
      "epoch": 2.7463252726410623,
      "grad_norm": 0.42316670484695895,
      "learning_rate": 2.1414786269430276e-07,
      "loss": 0.0254,
      "step": 2896
    },
    {
      "epoch": 2.7472735893788527,
      "grad_norm": 0.42725270284330796,
      "learning_rate": 2.1255223980891027e-07,
      "loss": 0.023,
      "step": 2897
    },
    {
      "epoch": 2.748221906116643,
      "grad_norm": 0.515462917264128,
      "learning_rate": 2.1096245458903985e-07,
      "loss": 0.0315,
      "step": 2898
    },
    {
      "epoch": 2.7491702228544335,
      "grad_norm": 0.3452667601112083,
      "learning_rate": 2.0937850897322754e-07,
      "loss": 0.0181,
      "step": 2899
    },
    {
      "epoch": 2.750118539592224,
      "grad_norm": 0.44912762956346985,
      "learning_rate": 2.0780040489288666e-07,
      "loss": 0.0233,
      "step": 2900
    },
    {
      "epoch": 2.7510668563300142,
      "grad_norm": 0.3191155443482015,
      "learning_rate": 2.0622814427230908e-07,
      "loss": 0.0128,
      "step": 2901
    },
    {
      "epoch": 2.7520151730678046,
      "grad_norm": 0.37094981215906553,
      "learning_rate": 2.046617290286601e-07,
      "loss": 0.0167,
      "step": 2902
    },
    {
      "epoch": 2.752963489805595,
      "grad_norm": 0.4763003959675575,
      "learning_rate": 2.0310116107197852e-07,
      "loss": 0.0158,
      "step": 2903
    },
    {
      "epoch": 2.7539118065433854,
      "grad_norm": 0.4329212628954346,
      "learning_rate": 2.0154644230517272e-07,
      "loss": 0.0325,
      "step": 2904
    },
    {
      "epoch": 2.754860123281176,
      "grad_norm": 0.4514024321208407,
      "learning_rate": 1.999975746240179e-07,
      "loss": 0.0256,
      "step": 2905
    },
    {
      "epoch": 2.755808440018966,
      "grad_norm": 0.376259827428853,
      "learning_rate": 1.9845455991715613e-07,
      "loss": 0.0263,
      "step": 2906
    },
    {
      "epoch": 2.756756756756757,
      "grad_norm": 0.33740568444255076,
      "learning_rate": 1.969174000660906e-07,
      "loss": 0.0176,
      "step": 2907
    },
    {
      "epoch": 2.757705073494547,
      "grad_norm": 0.41875612207034385,
      "learning_rate": 1.9538609694518763e-07,
      "loss": 0.0218,
      "step": 2908
    },
    {
      "epoch": 2.758653390232338,
      "grad_norm": 0.41857690137005205,
      "learning_rate": 1.9386065242167074e-07,
      "loss": 0.0245,
      "step": 2909
    },
    {
      "epoch": 2.759601706970128,
      "grad_norm": 0.4052596751702686,
      "learning_rate": 1.9234106835561873e-07,
      "loss": 0.0196,
      "step": 2910
    },
    {
      "epoch": 2.7605500237079186,
      "grad_norm": 0.5686641821576979,
      "learning_rate": 1.9082734659996548e-07,
      "loss": 0.0182,
      "step": 2911
    },
    {
      "epoch": 2.761498340445709,
      "grad_norm": 0.3255702279516522,
      "learning_rate": 1.893194890004979e-07,
      "loss": 0.0143,
      "step": 2912
    },
    {
      "epoch": 2.7624466571834994,
      "grad_norm": 0.36251640680639674,
      "learning_rate": 1.878174973958491e-07,
      "loss": 0.0289,
      "step": 2913
    },
    {
      "epoch": 2.7633949739212897,
      "grad_norm": 0.4490481069822958,
      "learning_rate": 1.863213736175018e-07,
      "loss": 0.0226,
      "step": 2914
    },
    {
      "epoch": 2.76434329065908,
      "grad_norm": 0.4277937689582375,
      "learning_rate": 1.8483111948978394e-07,
      "loss": 0.0167,
      "step": 2915
    },
    {
      "epoch": 2.7652916073968705,
      "grad_norm": 0.42002641898344123,
      "learning_rate": 1.833467368298636e-07,
      "loss": 0.0243,
      "step": 2916
    },
    {
      "epoch": 2.766239924134661,
      "grad_norm": 0.4499658388831021,
      "learning_rate": 1.8186822744775234e-07,
      "loss": 0.0268,
      "step": 2917
    },
    {
      "epoch": 2.7671882408724513,
      "grad_norm": 0.4274669486577644,
      "learning_rate": 1.8039559314629808e-07,
      "loss": 0.0163,
      "step": 2918
    },
    {
      "epoch": 2.7681365576102417,
      "grad_norm": 0.4400359676580664,
      "learning_rate": 1.789288357211849e-07,
      "loss": 0.0277,
      "step": 2919
    },
    {
      "epoch": 2.769084874348032,
      "grad_norm": 0.3837018238158169,
      "learning_rate": 1.7746795696093276e-07,
      "loss": 0.019,
      "step": 2920
    },
    {
      "epoch": 2.7700331910858225,
      "grad_norm": 0.350819880840381,
      "learning_rate": 1.760129586468906e-07,
      "loss": 0.0222,
      "step": 2921
    },
    {
      "epoch": 2.7709815078236133,
      "grad_norm": 0.4005678605328111,
      "learning_rate": 1.745638425532392e-07,
      "loss": 0.0151,
      "step": 2922
    },
    {
      "epoch": 2.7719298245614032,
      "grad_norm": 0.3792448928276872,
      "learning_rate": 1.7312061044698569e-07,
      "loss": 0.0287,
      "step": 2923
    },
    {
      "epoch": 2.772878141299194,
      "grad_norm": 0.6223382836502508,
      "learning_rate": 1.7168326408796233e-07,
      "loss": 0.0196,
      "step": 2924
    },
    {
      "epoch": 2.7738264580369845,
      "grad_norm": 0.5267243071815664,
      "learning_rate": 1.7025180522882546e-07,
      "loss": 0.0189,
      "step": 2925
    },
    {
      "epoch": 2.774774774774775,
      "grad_norm": 0.3763913747813696,
      "learning_rate": 1.688262356150505e-07,
      "loss": 0.0198,
      "step": 2926
    },
    {
      "epoch": 2.7757230915125652,
      "grad_norm": 0.6370991606268415,
      "learning_rate": 1.6740655698493313e-07,
      "loss": 0.0309,
      "step": 2927
    },
    {
      "epoch": 2.7766714082503556,
      "grad_norm": 0.42771643414825106,
      "learning_rate": 1.659927710695869e-07,
      "loss": 0.0336,
      "step": 2928
    },
    {
      "epoch": 2.777619724988146,
      "grad_norm": 0.4352455322886255,
      "learning_rate": 1.6458487959293722e-07,
      "loss": 0.0232,
      "step": 2929
    },
    {
      "epoch": 2.7785680417259364,
      "grad_norm": 0.46779414418617427,
      "learning_rate": 1.6318288427172424e-07,
      "loss": 0.0183,
      "step": 2930
    },
    {
      "epoch": 2.779516358463727,
      "grad_norm": 0.37840061172320577,
      "learning_rate": 1.61786786815496e-07,
      "loss": 0.0208,
      "step": 2931
    },
    {
      "epoch": 2.780464675201517,
      "grad_norm": 0.3181428862321126,
      "learning_rate": 1.6039658892661182e-07,
      "loss": 0.0177,
      "step": 2932
    },
    {
      "epoch": 2.7814129919393076,
      "grad_norm": 0.4433106803305223,
      "learning_rate": 1.590122923002363e-07,
      "loss": 0.0223,
      "step": 2933
    },
    {
      "epoch": 2.782361308677098,
      "grad_norm": 0.2888850895455348,
      "learning_rate": 1.576338986243364e-07,
      "loss": 0.0136,
      "step": 2934
    },
    {
      "epoch": 2.783309625414889,
      "grad_norm": 0.4252472222588191,
      "learning_rate": 1.5626140957968383e-07,
      "loss": 0.0266,
      "step": 2935
    },
    {
      "epoch": 2.7842579421526787,
      "grad_norm": 0.39718864763140277,
      "learning_rate": 1.5489482683984925e-07,
      "loss": 0.0191,
      "step": 2936
    },
    {
      "epoch": 2.7852062588904696,
      "grad_norm": 0.4490343074595942,
      "learning_rate": 1.5353415207120082e-07,
      "loss": 0.0263,
      "step": 2937
    },
    {
      "epoch": 2.78615457562826,
      "grad_norm": 0.5490222154334174,
      "learning_rate": 1.5217938693290359e-07,
      "loss": 0.0316,
      "step": 2938
    },
    {
      "epoch": 2.7871028923660504,
      "grad_norm": 0.4469751880236888,
      "learning_rate": 1.5083053307691608e-07,
      "loss": 0.0284,
      "step": 2939
    },
    {
      "epoch": 2.7880512091038407,
      "grad_norm": 0.4212606273209142,
      "learning_rate": 1.494875921479888e-07,
      "loss": 0.0225,
      "step": 2940
    },
    {
      "epoch": 2.788999525841631,
      "grad_norm": 0.572352413209743,
      "learning_rate": 1.4815056578366294e-07,
      "loss": 0.029,
      "step": 2941
    },
    {
      "epoch": 2.7899478425794215,
      "grad_norm": 0.5479657138204983,
      "learning_rate": 1.4681945561426548e-07,
      "loss": 0.0275,
      "step": 2942
    },
    {
      "epoch": 2.790896159317212,
      "grad_norm": 0.3807981714509639,
      "learning_rate": 1.4549426326291193e-07,
      "loss": 0.0204,
      "step": 2943
    },
    {
      "epoch": 2.7918444760550023,
      "grad_norm": 0.4491797751181635,
      "learning_rate": 1.4417499034550143e-07,
      "loss": 0.0251,
      "step": 2944
    },
    {
      "epoch": 2.7927927927927927,
      "grad_norm": 0.3987956661004981,
      "learning_rate": 1.4286163847071377e-07,
      "loss": 0.0274,
      "step": 2945
    },
    {
      "epoch": 2.793741109530583,
      "grad_norm": 0.47220214159729007,
      "learning_rate": 1.4155420924000963e-07,
      "loss": 0.0328,
      "step": 2946
    },
    {
      "epoch": 2.7946894262683735,
      "grad_norm": 0.6378932543205542,
      "learning_rate": 1.402527042476276e-07,
      "loss": 0.0231,
      "step": 2947
    },
    {
      "epoch": 2.7956377430061643,
      "grad_norm": 0.3691335230452688,
      "learning_rate": 1.3895712508058269e-07,
      "loss": 0.0217,
      "step": 2948
    },
    {
      "epoch": 2.7965860597439542,
      "grad_norm": 0.4517905770163704,
      "learning_rate": 1.3766747331866447e-07,
      "loss": 0.0281,
      "step": 2949
    },
    {
      "epoch": 2.797534376481745,
      "grad_norm": 0.3558619798708121,
      "learning_rate": 1.3638375053443343e-07,
      "loss": 0.0172,
      "step": 2950
    },
    {
      "epoch": 2.7984826932195355,
      "grad_norm": 0.3057791004881369,
      "learning_rate": 1.3510595829322237e-07,
      "loss": 0.0132,
      "step": 2951
    },
    {
      "epoch": 2.799431009957326,
      "grad_norm": 0.32216168713007337,
      "learning_rate": 1.3383409815313108e-07,
      "loss": 0.0181,
      "step": 2952
    },
    {
      "epoch": 2.8003793266951162,
      "grad_norm": 0.4740758300640553,
      "learning_rate": 1.3256817166502567e-07,
      "loss": 0.0311,
      "step": 2953
    },
    {
      "epoch": 2.8013276434329066,
      "grad_norm": 0.3702293823546692,
      "learning_rate": 1.3130818037253967e-07,
      "loss": 0.0239,
      "step": 2954
    },
    {
      "epoch": 2.802275960170697,
      "grad_norm": 0.6478107471426148,
      "learning_rate": 1.3005412581206521e-07,
      "loss": 0.0306,
      "step": 2955
    },
    {
      "epoch": 2.8032242769084874,
      "grad_norm": 0.3593783860531484,
      "learning_rate": 1.2880600951275857e-07,
      "loss": 0.0214,
      "step": 2956
    },
    {
      "epoch": 2.804172593646278,
      "grad_norm": 0.38273343270548155,
      "learning_rate": 1.2756383299653452e-07,
      "loss": 0.0231,
      "step": 2957
    },
    {
      "epoch": 2.805120910384068,
      "grad_norm": 0.35039846503549316,
      "learning_rate": 1.263275977780637e-07,
      "loss": 0.0245,
      "step": 2958
    },
    {
      "epoch": 2.8060692271218586,
      "grad_norm": 0.4375014938283871,
      "learning_rate": 1.250973053647736e-07,
      "loss": 0.0227,
      "step": 2959
    },
    {
      "epoch": 2.807017543859649,
      "grad_norm": 0.42091207467959957,
      "learning_rate": 1.2387295725684534e-07,
      "loss": 0.0237,
      "step": 2960
    },
    {
      "epoch": 2.80796586059744,
      "grad_norm": 0.46361943079914314,
      "learning_rate": 1.2265455494721024e-07,
      "loss": 0.0269,
      "step": 2961
    },
    {
      "epoch": 2.8089141773352297,
      "grad_norm": 0.4287848018482948,
      "learning_rate": 1.2144209992155043e-07,
      "loss": 0.0171,
      "step": 2962
    },
    {
      "epoch": 2.8098624940730206,
      "grad_norm": 0.4400952401156122,
      "learning_rate": 1.2023559365829606e-07,
      "loss": 0.0215,
      "step": 2963
    },
    {
      "epoch": 2.810810810810811,
      "grad_norm": 0.3500703922509506,
      "learning_rate": 1.1903503762862311e-07,
      "loss": 0.0172,
      "step": 2964
    },
    {
      "epoch": 2.8117591275486014,
      "grad_norm": 0.31141988161587025,
      "learning_rate": 1.1784043329645389e-07,
      "loss": 0.0111,
      "step": 2965
    },
    {
      "epoch": 2.8127074442863917,
      "grad_norm": 0.3598348161337118,
      "learning_rate": 1.1665178211844985e-07,
      "loss": 0.0186,
      "step": 2966
    },
    {
      "epoch": 2.813655761024182,
      "grad_norm": 0.4603355314327043,
      "learning_rate": 1.1546908554401659e-07,
      "loss": 0.032,
      "step": 2967
    },
    {
      "epoch": 2.8146040777619725,
      "grad_norm": 0.3641648570177723,
      "learning_rate": 1.1429234501529773e-07,
      "loss": 0.0182,
      "step": 2968
    },
    {
      "epoch": 2.815552394499763,
      "grad_norm": 0.5217885429345293,
      "learning_rate": 1.1312156196717383e-07,
      "loss": 0.0264,
      "step": 2969
    },
    {
      "epoch": 2.8165007112375533,
      "grad_norm": 0.48416094699379036,
      "learning_rate": 1.1195673782726235e-07,
      "loss": 0.0213,
      "step": 2970
    },
    {
      "epoch": 2.8174490279753437,
      "grad_norm": 0.45587516725174587,
      "learning_rate": 1.1079787401591213e-07,
      "loss": 0.0255,
      "step": 2971
    },
    {
      "epoch": 2.818397344713134,
      "grad_norm": 0.3105673525304809,
      "learning_rate": 1.0964497194620727e-07,
      "loss": 0.0199,
      "step": 2972
    },
    {
      "epoch": 2.8193456614509245,
      "grad_norm": 0.33311835087454994,
      "learning_rate": 1.08498033023961e-07,
      "loss": 0.0177,
      "step": 2973
    },
    {
      "epoch": 2.8202939781887153,
      "grad_norm": 0.36202184705860024,
      "learning_rate": 1.0735705864771351e-07,
      "loss": 0.0233,
      "step": 2974
    },
    {
      "epoch": 2.8212422949265052,
      "grad_norm": 0.3721122938385976,
      "learning_rate": 1.0622205020873467e-07,
      "loss": 0.0192,
      "step": 2975
    },
    {
      "epoch": 2.822190611664296,
      "grad_norm": 0.41898867839074483,
      "learning_rate": 1.0509300909101904e-07,
      "loss": 0.0327,
      "step": 2976
    },
    {
      "epoch": 2.823138928402086,
      "grad_norm": 0.3496114095941287,
      "learning_rate": 1.0396993667128318e-07,
      "loss": 0.0197,
      "step": 2977
    },
    {
      "epoch": 2.824087245139877,
      "grad_norm": 0.3870480111236857,
      "learning_rate": 1.028528343189683e-07,
      "loss": 0.0191,
      "step": 2978
    },
    {
      "epoch": 2.8250355618776672,
      "grad_norm": 0.44439329160244306,
      "learning_rate": 1.0174170339623313e-07,
      "loss": 0.0209,
      "step": 2979
    },
    {
      "epoch": 2.8259838786154576,
      "grad_norm": 0.2985339684820892,
      "learning_rate": 1.0063654525795663e-07,
      "loss": 0.016,
      "step": 2980
    },
    {
      "epoch": 2.826932195353248,
      "grad_norm": 0.5609480662641474,
      "learning_rate": 9.953736125173474e-08,
      "loss": 0.0216,
      "step": 2981
    },
    {
      "epoch": 2.8278805120910384,
      "grad_norm": 0.3391707432744039,
      "learning_rate": 9.84441527178781e-08,
      "loss": 0.0175,
      "step": 2982
    },
    {
      "epoch": 2.828828828828829,
      "grad_norm": 0.34240861985112864,
      "learning_rate": 9.735692098941207e-08,
      "loss": 0.0159,
      "step": 2983
    },
    {
      "epoch": 2.829777145566619,
      "grad_norm": 0.7564810284696015,
      "learning_rate": 9.627566739207229e-08,
      "loss": 0.0238,
      "step": 2984
    },
    {
      "epoch": 2.8307254623044096,
      "grad_norm": 0.35861289080670644,
      "learning_rate": 9.52003932443063e-08,
      "loss": 0.0184,
      "step": 2985
    },
    {
      "epoch": 2.8316737790422,
      "grad_norm": 0.3168867052467738,
      "learning_rate": 9.413109985727032e-08,
      "loss": 0.0147,
      "step": 2986
    },
    {
      "epoch": 2.8326220957799904,
      "grad_norm": 0.30328581611262106,
      "learning_rate": 9.306778853482745e-08,
      "loss": 0.0135,
      "step": 2987
    },
    {
      "epoch": 2.8335704125177807,
      "grad_norm": 0.39852807663419376,
      "learning_rate": 9.201046057354613e-08,
      "loss": 0.02,
      "step": 2988
    },
    {
      "epoch": 2.8345187292555716,
      "grad_norm": 0.42062156640088805,
      "learning_rate": 9.095911726270057e-08,
      "loss": 0.0214,
      "step": 2989
    },
    {
      "epoch": 2.8354670459933615,
      "grad_norm": 0.40669783911572394,
      "learning_rate": 8.991375988426421e-08,
      "loss": 0.0191,
      "step": 2990
    },
    {
      "epoch": 2.8364153627311524,
      "grad_norm": 0.38716601394773986,
      "learning_rate": 8.887438971291574e-08,
      "loss": 0.0217,
      "step": 2991
    },
    {
      "epoch": 2.8373636794689427,
      "grad_norm": 0.47316775873816014,
      "learning_rate": 8.784100801602913e-08,
      "loss": 0.0269,
      "step": 2992
    },
    {
      "epoch": 2.838311996206733,
      "grad_norm": 0.3893105038874972,
      "learning_rate": 8.681361605367921e-08,
      "loss": 0.0246,
      "step": 2993
    },
    {
      "epoch": 2.8392603129445235,
      "grad_norm": 0.44212071533465297,
      "learning_rate": 8.579221507863555e-08,
      "loss": 0.0162,
      "step": 2994
    },
    {
      "epoch": 2.840208629682314,
      "grad_norm": 0.295444897835934,
      "learning_rate": 8.477680633636298e-08,
      "loss": 0.0141,
      "step": 2995
    },
    {
      "epoch": 2.8411569464201043,
      "grad_norm": 0.5950424381452033,
      "learning_rate": 8.376739106501886e-08,
      "loss": 0.0347,
      "step": 2996
    },
    {
      "epoch": 2.8421052631578947,
      "grad_norm": 0.4003901243922404,
      "learning_rate": 8.276397049545359e-08,
      "loss": 0.0226,
      "step": 2997
    },
    {
      "epoch": 2.843053579895685,
      "grad_norm": 0.3217709626825981,
      "learning_rate": 8.176654585120625e-08,
      "loss": 0.0146,
      "step": 2998
    },
    {
      "epoch": 2.8440018966334755,
      "grad_norm": 0.32999877056945826,
      "learning_rate": 8.077511834850727e-08,
      "loss": 0.0191,
      "step": 2999
    },
    {
      "epoch": 2.844950213371266,
      "grad_norm": 0.41667581680322996,
      "learning_rate": 7.978968919627073e-08,
      "loss": 0.0292,
      "step": 3000
    },
    {
      "epoch": 2.8458985301090562,
      "grad_norm": 0.4081492608814963,
      "learning_rate": 7.881025959609933e-08,
      "loss": 0.0203,
      "step": 3001
    },
    {
      "epoch": 2.846846846846847,
      "grad_norm": 0.391926194225634,
      "learning_rate": 7.783683074227943e-08,
      "loss": 0.0176,
      "step": 3002
    },
    {
      "epoch": 2.847795163584637,
      "grad_norm": 0.3969469040185239,
      "learning_rate": 7.686940382177933e-08,
      "loss": 0.0224,
      "step": 3003
    },
    {
      "epoch": 2.848743480322428,
      "grad_norm": 0.458127293518202,
      "learning_rate": 7.590798001425038e-08,
      "loss": 0.0266,
      "step": 3004
    },
    {
      "epoch": 2.8496917970602182,
      "grad_norm": 0.33047498969579775,
      "learning_rate": 7.495256049202148e-08,
      "loss": 0.014,
      "step": 3005
    },
    {
      "epoch": 2.8506401137980086,
      "grad_norm": 0.3768004564754747,
      "learning_rate": 7.400314642010353e-08,
      "loss": 0.0188,
      "step": 3006
    },
    {
      "epoch": 2.851588430535799,
      "grad_norm": 0.4032452837194265,
      "learning_rate": 7.305973895618157e-08,
      "loss": 0.0194,
      "step": 3007
    },
    {
      "epoch": 2.8525367472735894,
      "grad_norm": 0.2862916882723256,
      "learning_rate": 7.212233925061707e-08,
      "loss": 0.0146,
      "step": 3008
    },
    {
      "epoch": 2.85348506401138,
      "grad_norm": 0.45519088532977087,
      "learning_rate": 7.119094844644681e-08,
      "loss": 0.028,
      "step": 3009
    },
    {
      "epoch": 2.85443338074917,
      "grad_norm": 0.5934720177918184,
      "learning_rate": 7.026556767938009e-08,
      "loss": 0.0313,
      "step": 3010
    },
    {
      "epoch": 2.8553816974869606,
      "grad_norm": 0.32292834230201356,
      "learning_rate": 6.934619807779652e-08,
      "loss": 0.0176,
      "step": 3011
    },
    {
      "epoch": 2.856330014224751,
      "grad_norm": 0.357283774394047,
      "learning_rate": 6.843284076274769e-08,
      "loss": 0.0254,
      "step": 3012
    },
    {
      "epoch": 2.8572783309625414,
      "grad_norm": 0.3461413190649742,
      "learning_rate": 6.752549684795385e-08,
      "loss": 0.0158,
      "step": 3013
    },
    {
      "epoch": 2.8582266477003317,
      "grad_norm": 0.33139564022160245,
      "learning_rate": 6.662416743980105e-08,
      "loss": 0.0245,
      "step": 3014
    },
    {
      "epoch": 2.8591749644381226,
      "grad_norm": 0.5110634604202572,
      "learning_rate": 6.572885363734294e-08,
      "loss": 0.0222,
      "step": 3015
    },
    {
      "epoch": 2.8601232811759125,
      "grad_norm": 0.4720638779321911,
      "learning_rate": 6.483955653229735e-08,
      "loss": 0.0205,
      "step": 3016
    },
    {
      "epoch": 2.8610715979137034,
      "grad_norm": 0.3523152512602496,
      "learning_rate": 6.395627720904518e-08,
      "loss": 0.0172,
      "step": 3017
    },
    {
      "epoch": 2.8620199146514937,
      "grad_norm": 0.4762763264950715,
      "learning_rate": 6.307901674463046e-08,
      "loss": 0.0157,
      "step": 3018
    },
    {
      "epoch": 2.862968231389284,
      "grad_norm": 0.38026270453910344,
      "learning_rate": 6.220777620875695e-08,
      "loss": 0.0186,
      "step": 3019
    },
    {
      "epoch": 2.8639165481270745,
      "grad_norm": 0.42457200354403224,
      "learning_rate": 6.134255666378874e-08,
      "loss": 0.0197,
      "step": 3020
    },
    {
      "epoch": 2.864864864864865,
      "grad_norm": 0.39614457102266576,
      "learning_rate": 6.0483359164748e-08,
      "loss": 0.0257,
      "step": 3021
    },
    {
      "epoch": 2.8658131816026553,
      "grad_norm": 0.3617668641258794,
      "learning_rate": 5.963018475931282e-08,
      "loss": 0.0228,
      "step": 3022
    },
    {
      "epoch": 2.8667614983404457,
      "grad_norm": 0.43808543708423153,
      "learning_rate": 5.8783034487818194e-08,
      "loss": 0.0245,
      "step": 3023
    },
    {
      "epoch": 2.867709815078236,
      "grad_norm": 0.43175991173965117,
      "learning_rate": 5.794190938325284e-08,
      "loss": 0.0264,
      "step": 3024
    },
    {
      "epoch": 2.8686581318160265,
      "grad_norm": 0.4848482461129806,
      "learning_rate": 5.710681047125799e-08,
      "loss": 0.0225,
      "step": 3025
    },
    {
      "epoch": 2.869606448553817,
      "grad_norm": 0.41841626879899474,
      "learning_rate": 5.627773877012854e-08,
      "loss": 0.0227,
      "step": 3026
    },
    {
      "epoch": 2.8705547652916072,
      "grad_norm": 0.37759148273500365,
      "learning_rate": 5.54546952908086e-08,
      "loss": 0.0169,
      "step": 3027
    },
    {
      "epoch": 2.871503082029398,
      "grad_norm": 0.37983258288815297,
      "learning_rate": 5.4637681036890935e-08,
      "loss": 0.0243,
      "step": 3028
    },
    {
      "epoch": 2.872451398767188,
      "grad_norm": 0.5681184040431467,
      "learning_rate": 5.382669700461862e-08,
      "loss": 0.0202,
      "step": 3029
    },
    {
      "epoch": 2.873399715504979,
      "grad_norm": 0.5891881317634622,
      "learning_rate": 5.302174418287953e-08,
      "loss": 0.0195,
      "step": 3030
    },
    {
      "epoch": 2.8743480322427692,
      "grad_norm": 0.4508271499743202,
      "learning_rate": 5.222282355320851e-08,
      "loss": 0.0164,
      "step": 3031
    },
    {
      "epoch": 2.8752963489805596,
      "grad_norm": 0.4435781831406531,
      "learning_rate": 5.142993608978519e-08,
      "loss": 0.0316,
      "step": 3032
    },
    {
      "epoch": 2.87624466571835,
      "grad_norm": 0.34441181405039545,
      "learning_rate": 5.064308275943064e-08,
      "loss": 0.0176,
      "step": 3033
    },
    {
      "epoch": 2.8771929824561404,
      "grad_norm": 0.3795476447951702,
      "learning_rate": 4.9862264521611245e-08,
      "loss": 0.0265,
      "step": 3034
    },
    {
      "epoch": 2.878141299193931,
      "grad_norm": 0.3827141067101136,
      "learning_rate": 4.9087482328430967e-08,
      "loss": 0.0202,
      "step": 3035
    },
    {
      "epoch": 2.879089615931721,
      "grad_norm": 0.43056600166627135,
      "learning_rate": 4.831873712463631e-08,
      "loss": 0.023,
      "step": 3036
    },
    {
      "epoch": 2.8800379326695116,
      "grad_norm": 0.3400748622374083,
      "learning_rate": 4.755602984761132e-08,
      "loss": 0.0203,
      "step": 3037
    },
    {
      "epoch": 2.880986249407302,
      "grad_norm": 0.3358342329948678,
      "learning_rate": 4.679936142737651e-08,
      "loss": 0.0182,
      "step": 3038
    },
    {
      "epoch": 2.8819345661450924,
      "grad_norm": 0.3518260649894719,
      "learning_rate": 4.6048732786591057e-08,
      "loss": 0.0216,
      "step": 3039
    },
    {
      "epoch": 2.8828828828828827,
      "grad_norm": 0.373396098412495,
      "learning_rate": 4.530414484054779e-08,
      "loss": 0.0181,
      "step": 3040
    },
    {
      "epoch": 2.8838311996206736,
      "grad_norm": 0.45648381228955215,
      "learning_rate": 4.456559849717435e-08,
      "loss": 0.0248,
      "step": 3041
    },
    {
      "epoch": 2.8847795163584635,
      "grad_norm": 0.5723077765956877,
      "learning_rate": 4.383309465703145e-08,
      "loss": 0.0292,
      "step": 3042
    },
    {
      "epoch": 2.8857278330962544,
      "grad_norm": 0.3588197970323949,
      "learning_rate": 4.310663421331074e-08,
      "loss": 0.0153,
      "step": 3043
    },
    {
      "epoch": 2.8866761498340443,
      "grad_norm": 0.3503812732911283,
      "learning_rate": 4.23862180518364e-08,
      "loss": 0.02,
      "step": 3044
    },
    {
      "epoch": 2.887624466571835,
      "grad_norm": 0.3506207303029663,
      "learning_rate": 4.167184705106131e-08,
      "loss": 0.0164,
      "step": 3045
    },
    {
      "epoch": 2.8885727833096255,
      "grad_norm": 0.42034903702104165,
      "learning_rate": 4.0963522082067555e-08,
      "loss": 0.0237,
      "step": 3046
    },
    {
      "epoch": 2.889521100047416,
      "grad_norm": 0.4494851053535652,
      "learning_rate": 4.026124400856479e-08,
      "loss": 0.035,
      "step": 3047
    },
    {
      "epoch": 2.8904694167852063,
      "grad_norm": 0.3197898192752823,
      "learning_rate": 3.956501368688859e-08,
      "loss": 0.0158,
      "step": 3048
    },
    {
      "epoch": 2.8914177335229967,
      "grad_norm": 0.4563666271619231,
      "learning_rate": 3.8874831966000946e-08,
      "loss": 0.0208,
      "step": 3049
    },
    {
      "epoch": 2.892366050260787,
      "grad_norm": 0.40272960757734055,
      "learning_rate": 3.819069968748812e-08,
      "loss": 0.0195,
      "step": 3050
    },
    {
      "epoch": 2.8933143669985775,
      "grad_norm": 0.46058356280405144,
      "learning_rate": 3.751261768555947e-08,
      "loss": 0.0191,
      "step": 3051
    },
    {
      "epoch": 2.894262683736368,
      "grad_norm": 0.42553257610569817,
      "learning_rate": 3.684058678704805e-08,
      "loss": 0.0209,
      "step": 3052
    },
    {
      "epoch": 2.8952110004741582,
      "grad_norm": 0.33487787758500354,
      "learning_rate": 3.6174607811406134e-08,
      "loss": 0.0163,
      "step": 3053
    },
    {
      "epoch": 2.8961593172119486,
      "grad_norm": 0.3625318971935175,
      "learning_rate": 3.551468157070914e-08,
      "loss": 0.0173,
      "step": 3054
    },
    {
      "epoch": 2.897107633949739,
      "grad_norm": 0.5370407138568671,
      "learning_rate": 3.4860808869650595e-08,
      "loss": 0.0313,
      "step": 3055
    },
    {
      "epoch": 2.89805595068753,
      "grad_norm": 0.44545436437243463,
      "learning_rate": 3.421299050554161e-08,
      "loss": 0.0292,
      "step": 3056
    },
    {
      "epoch": 2.89900426742532,
      "grad_norm": 0.34087790333913004,
      "learning_rate": 3.357122726831252e-08,
      "loss": 0.0157,
      "step": 3057
    },
    {
      "epoch": 2.8999525841631106,
      "grad_norm": 0.3832172647070527,
      "learning_rate": 3.293551994051014e-08,
      "loss": 0.0226,
      "step": 3058
    },
    {
      "epoch": 2.900900900900901,
      "grad_norm": 0.5081414251987051,
      "learning_rate": 3.230586929729496e-08,
      "loss": 0.0324,
      "step": 3059
    },
    {
      "epoch": 2.9018492176386914,
      "grad_norm": 0.4636198191740265,
      "learning_rate": 3.1682276106444474e-08,
      "loss": 0.0189,
      "step": 3060
    },
    {
      "epoch": 2.902797534376482,
      "grad_norm": 0.37250398138400564,
      "learning_rate": 3.1064741128348786e-08,
      "loss": 0.0159,
      "step": 3061
    },
    {
      "epoch": 2.903745851114272,
      "grad_norm": 0.38602339870373964,
      "learning_rate": 3.0453265116009986e-08,
      "loss": 0.0133,
      "step": 3062
    },
    {
      "epoch": 2.9046941678520626,
      "grad_norm": 0.46169156998385363,
      "learning_rate": 2.984784881504443e-08,
      "loss": 0.0181,
      "step": 3063
    },
    {
      "epoch": 2.905642484589853,
      "grad_norm": 0.4028285252992143,
      "learning_rate": 2.9248492963677156e-08,
      "loss": 0.0244,
      "step": 3064
    },
    {
      "epoch": 2.9065908013276434,
      "grad_norm": 0.3732446124156518,
      "learning_rate": 2.8655198292744104e-08,
      "loss": 0.02,
      "step": 3065
    },
    {
      "epoch": 2.9075391180654337,
      "grad_norm": 0.7546073925694232,
      "learning_rate": 2.8067965525690467e-08,
      "loss": 0.0329,
      "step": 3066
    },
    {
      "epoch": 2.908487434803224,
      "grad_norm": 0.3667730204837699,
      "learning_rate": 2.748679537857013e-08,
      "loss": 0.0158,
      "step": 3067
    },
    {
      "epoch": 2.9094357515410145,
      "grad_norm": 0.3803667665108587,
      "learning_rate": 2.6911688560043447e-08,
      "loss": 0.0225,
      "step": 3068
    },
    {
      "epoch": 2.9103840682788054,
      "grad_norm": 0.4424178102463165,
      "learning_rate": 2.634264577137835e-08,
      "loss": 0.0209,
      "step": 3069
    },
    {
      "epoch": 2.9113323850165953,
      "grad_norm": 0.36197907514861327,
      "learning_rate": 2.5779667706447577e-08,
      "loss": 0.0199,
      "step": 3070
    },
    {
      "epoch": 2.912280701754386,
      "grad_norm": 0.354290871645171,
      "learning_rate": 2.522275505172922e-08,
      "loss": 0.0208,
      "step": 3071
    },
    {
      "epoch": 2.9132290184921765,
      "grad_norm": 0.47534916673431526,
      "learning_rate": 2.4671908486305075e-08,
      "loss": 0.0276,
      "step": 3072
    },
    {
      "epoch": 2.914177335229967,
      "grad_norm": 0.4430709376730269,
      "learning_rate": 2.4127128681861178e-08,
      "loss": 0.025,
      "step": 3073
    },
    {
      "epoch": 2.9151256519677573,
      "grad_norm": 0.35749048896901775,
      "learning_rate": 2.358841630268449e-08,
      "loss": 0.0198,
      "step": 3074
    },
    {
      "epoch": 2.9160739687055477,
      "grad_norm": 0.3016708451573105,
      "learning_rate": 2.3055772005664558e-08,
      "loss": 0.0182,
      "step": 3075
    },
    {
      "epoch": 2.917022285443338,
      "grad_norm": 0.476466900640434,
      "learning_rate": 2.2529196440290723e-08,
      "loss": 0.0329,
      "step": 3076
    },
    {
      "epoch": 2.9179706021811285,
      "grad_norm": 0.34737568586667705,
      "learning_rate": 2.2008690248653264e-08,
      "loss": 0.018,
      "step": 3077
    },
    {
      "epoch": 2.918918918918919,
      "grad_norm": 0.45746335820706724,
      "learning_rate": 2.14942540654417e-08,
      "loss": 0.0282,
      "step": 3078
    },
    {
      "epoch": 2.9198672356567092,
      "grad_norm": 0.44008704548623256,
      "learning_rate": 2.0985888517943697e-08,
      "loss": 0.0191,
      "step": 3079
    },
    {
      "epoch": 2.9208155523944996,
      "grad_norm": 0.4198768929738493,
      "learning_rate": 2.04835942260434e-08,
      "loss": 0.0149,
      "step": 3080
    },
    {
      "epoch": 2.92176386913229,
      "grad_norm": 0.46200695328290603,
      "learning_rate": 1.9987371802223655e-08,
      "loss": 0.042,
      "step": 3081
    },
    {
      "epoch": 2.922712185870081,
      "grad_norm": 0.30343363208397217,
      "learning_rate": 1.9497221851562665e-08,
      "loss": 0.0173,
      "step": 3082
    },
    {
      "epoch": 2.923660502607871,
      "grad_norm": 0.558923165292205,
      "learning_rate": 1.901314497173401e-08,
      "loss": 0.0345,
      "step": 3083
    },
    {
      "epoch": 2.9246088193456616,
      "grad_norm": 0.5277396892934476,
      "learning_rate": 1.8535141753006634e-08,
      "loss": 0.0244,
      "step": 3084
    },
    {
      "epoch": 2.925557136083452,
      "grad_norm": 0.3301572904249269,
      "learning_rate": 1.8063212778241523e-08,
      "loss": 0.0161,
      "step": 3085
    },
    {
      "epoch": 2.9265054528212424,
      "grad_norm": 0.39537910641276297,
      "learning_rate": 1.7597358622895023e-08,
      "loss": 0.0252,
      "step": 3086
    },
    {
      "epoch": 2.927453769559033,
      "grad_norm": 0.5693156857346456,
      "learning_rate": 1.7137579855016073e-08,
      "loss": 0.0695,
      "step": 3087
    },
    {
      "epoch": 2.928402086296823,
      "grad_norm": 0.3875441600495483,
      "learning_rate": 1.668387703524288e-08,
      "loss": 0.0192,
      "step": 3088
    },
    {
      "epoch": 2.9293504030346136,
      "grad_norm": 0.38815951485681266,
      "learning_rate": 1.62362507168079e-08,
      "loss": 0.0194,
      "step": 3089
    },
    {
      "epoch": 2.930298719772404,
      "grad_norm": 0.3849521596062896,
      "learning_rate": 1.5794701445532856e-08,
      "loss": 0.0205,
      "step": 3090
    },
    {
      "epoch": 2.9312470365101944,
      "grad_norm": 0.371593598213233,
      "learning_rate": 1.535922975982873e-08,
      "loss": 0.0201,
      "step": 3091
    },
    {
      "epoch": 2.9321953532479847,
      "grad_norm": 0.32985046282671887,
      "learning_rate": 1.4929836190696322e-08,
      "loss": 0.0259,
      "step": 3092
    },
    {
      "epoch": 2.933143669985775,
      "grad_norm": 0.5164324189253016,
      "learning_rate": 1.4506521261725137e-08,
      "loss": 0.0267,
      "step": 3093
    },
    {
      "epoch": 2.9340919867235655,
      "grad_norm": 0.4789431073781785,
      "learning_rate": 1.4089285489091719e-08,
      "loss": 0.0201,
      "step": 3094
    },
    {
      "epoch": 2.9350403034613564,
      "grad_norm": 0.8320517563012525,
      "learning_rate": 1.3678129381560768e-08,
      "loss": 0.0215,
      "step": 3095
    },
    {
      "epoch": 2.9359886201991463,
      "grad_norm": 0.3177776879644222,
      "learning_rate": 1.3273053440483463e-08,
      "loss": 0.0145,
      "step": 3096
    },
    {
      "epoch": 2.936936936936937,
      "grad_norm": 0.4448838819863309,
      "learning_rate": 1.2874058159796366e-08,
      "loss": 0.0253,
      "step": 3097
    },
    {
      "epoch": 2.937885253674727,
      "grad_norm": 0.4731371445018261,
      "learning_rate": 1.2481144026022519e-08,
      "loss": 0.0206,
      "step": 3098
    },
    {
      "epoch": 2.938833570412518,
      "grad_norm": 0.3975126226697693,
      "learning_rate": 1.209431151826923e-08,
      "loss": 0.023,
      "step": 3099
    },
    {
      "epoch": 2.9397818871503083,
      "grad_norm": 0.3839730934184733,
      "learning_rate": 1.1713561108228077e-08,
      "loss": 0.0173,
      "step": 3100
    },
    {
      "epoch": 2.9407302038880987,
      "grad_norm": 0.3128709085052025,
      "learning_rate": 1.1338893260173234e-08,
      "loss": 0.0173,
      "step": 3101
    },
    {
      "epoch": 2.941678520625889,
      "grad_norm": 0.43719703769605983,
      "learning_rate": 1.0970308430964804e-08,
      "loss": 0.0222,
      "step": 3102
    },
    {
      "epoch": 2.9426268373636795,
      "grad_norm": 0.3316575913593028,
      "learning_rate": 1.0607807070042719e-08,
      "loss": 0.0135,
      "step": 3103
    },
    {
      "epoch": 2.94357515410147,
      "grad_norm": 0.4809128170286833,
      "learning_rate": 1.0251389619430063e-08,
      "loss": 0.0227,
      "step": 3104
    },
    {
      "epoch": 2.9445234708392602,
      "grad_norm": 0.3149863969572906,
      "learning_rate": 9.901056513730856e-09,
      "loss": 0.0157,
      "step": 3105
    },
    {
      "epoch": 2.9454717875770506,
      "grad_norm": 0.3504971113027487,
      "learning_rate": 9.55680818013116e-09,
      "loss": 0.0137,
      "step": 3106
    },
    {
      "epoch": 2.946420104314841,
      "grad_norm": 0.3871851442267693,
      "learning_rate": 9.218645038396312e-09,
      "loss": 0.0247,
      "step": 3107
    },
    {
      "epoch": 2.9473684210526314,
      "grad_norm": 0.3216947869349185,
      "learning_rate": 8.886567500872024e-09,
      "loss": 0.0159,
      "step": 3108
    },
    {
      "epoch": 2.948316737790422,
      "grad_norm": 0.4069675609461263,
      "learning_rate": 8.560575972483276e-09,
      "loss": 0.0216,
      "step": 3109
    },
    {
      "epoch": 2.9492650545282126,
      "grad_norm": 0.46983515261566355,
      "learning_rate": 8.240670850734322e-09,
      "loss": 0.027,
      "step": 3110
    },
    {
      "epoch": 2.9502133712660026,
      "grad_norm": 0.3483044312681159,
      "learning_rate": 7.926852525707574e-09,
      "loss": 0.0214,
      "step": 3111
    },
    {
      "epoch": 2.9511616880037934,
      "grad_norm": 0.38354468067188363,
      "learning_rate": 7.619121380063044e-09,
      "loss": 0.0211,
      "step": 3112
    },
    {
      "epoch": 2.952110004741584,
      "grad_norm": 0.3452999961426753,
      "learning_rate": 7.317477789038352e-09,
      "loss": 0.0126,
      "step": 3113
    },
    {
      "epoch": 2.953058321479374,
      "grad_norm": 0.9052240783328028,
      "learning_rate": 7.021922120449276e-09,
      "loss": 0.0355,
      "step": 3114
    },
    {
      "epoch": 2.9540066382171646,
      "grad_norm": 0.376692331880163,
      "learning_rate": 6.732454734686422e-09,
      "loss": 0.0194,
      "step": 3115
    },
    {
      "epoch": 2.954954954954955,
      "grad_norm": 0.2888073338536502,
      "learning_rate": 6.449075984717446e-09,
      "loss": 0.0151,
      "step": 3116
    },
    {
      "epoch": 2.9559032716927454,
      "grad_norm": 0.6205803084744425,
      "learning_rate": 6.17178621608594e-09,
      "loss": 0.0387,
      "step": 3117
    },
    {
      "epoch": 2.9568515884305357,
      "grad_norm": 0.6830407506925207,
      "learning_rate": 5.9005857669103276e-09,
      "loss": 0.0204,
      "step": 3118
    },
    {
      "epoch": 2.957799905168326,
      "grad_norm": 0.3239139372515465,
      "learning_rate": 5.635474967883858e-09,
      "loss": 0.0172,
      "step": 3119
    },
    {
      "epoch": 2.9587482219061165,
      "grad_norm": 0.4204983758314779,
      "learning_rate": 5.376454142274612e-09,
      "loss": 0.0194,
      "step": 3120
    },
    {
      "epoch": 2.959696538643907,
      "grad_norm": 0.33796542007720826,
      "learning_rate": 5.1235236059249405e-09,
      "loss": 0.015,
      "step": 3121
    },
    {
      "epoch": 2.9606448553816973,
      "grad_norm": 0.3853177543515329,
      "learning_rate": 4.876683667249804e-09,
      "loss": 0.0175,
      "step": 3122
    },
    {
      "epoch": 2.961593172119488,
      "grad_norm": 0.3750767973380027,
      "learning_rate": 4.635934627238436e-09,
      "loss": 0.0224,
      "step": 3123
    },
    {
      "epoch": 2.962541488857278,
      "grad_norm": 0.3595739902347792,
      "learning_rate": 4.401276779452679e-09,
      "loss": 0.0185,
      "step": 3124
    },
    {
      "epoch": 2.963489805595069,
      "grad_norm": 0.34064999802722745,
      "learning_rate": 4.1727104100275366e-09,
      "loss": 0.0142,
      "step": 3125
    },
    {
      "epoch": 2.9644381223328593,
      "grad_norm": 0.4298659095790116,
      "learning_rate": 3.950235797668955e-09,
      "loss": 0.0183,
      "step": 3126
    },
    {
      "epoch": 2.9653864390706497,
      "grad_norm": 0.3154960617212208,
      "learning_rate": 3.733853213656046e-09,
      "loss": 0.016,
      "step": 3127
    },
    {
      "epoch": 2.96633475580844,
      "grad_norm": 0.43428561564507495,
      "learning_rate": 3.5235629218394184e-09,
      "loss": 0.0245,
      "step": 3128
    },
    {
      "epoch": 2.9672830725462305,
      "grad_norm": 0.5129124909596331,
      "learning_rate": 3.3193651786395107e-09,
      "loss": 0.0298,
      "step": 3129
    },
    {
      "epoch": 2.968231389284021,
      "grad_norm": 0.3653304436659778,
      "learning_rate": 3.1212602330499274e-09,
      "loss": 0.0188,
      "step": 3130
    },
    {
      "epoch": 2.9691797060218112,
      "grad_norm": 0.38747167610148703,
      "learning_rate": 2.929248326632994e-09,
      "loss": 0.0328,
      "step": 3131
    },
    {
      "epoch": 2.9701280227596016,
      "grad_norm": 0.37075949158686394,
      "learning_rate": 2.743329693521979e-09,
      "loss": 0.0162,
      "step": 3132
    },
    {
      "epoch": 2.971076339497392,
      "grad_norm": 0.32816513761541277,
      "learning_rate": 2.563504560420538e-09,
      "loss": 0.0175,
      "step": 3133
    },
    {
      "epoch": 2.9720246562351824,
      "grad_norm": 0.3176724510277183,
      "learning_rate": 2.389773146602159e-09,
      "loss": 0.0141,
      "step": 3134
    },
    {
      "epoch": 2.972972972972973,
      "grad_norm": 0.3777928889224251,
      "learning_rate": 2.222135663909053e-09,
      "loss": 0.0206,
      "step": 3135
    },
    {
      "epoch": 2.9739212897107636,
      "grad_norm": 0.34874205105798695,
      "learning_rate": 2.0605923167532625e-09,
      "loss": 0.0195,
      "step": 3136
    },
    {
      "epoch": 2.9748696064485536,
      "grad_norm": 0.37751192563561503,
      "learning_rate": 1.905143302116108e-09,
      "loss": 0.0186,
      "step": 3137
    },
    {
      "epoch": 2.9758179231863444,
      "grad_norm": 0.413833078919697,
      "learning_rate": 1.7557888095465215e-09,
      "loss": 0.0285,
      "step": 3138
    },
    {
      "epoch": 2.976766239924135,
      "grad_norm": 0.3495742580635406,
      "learning_rate": 1.6125290211638222e-09,
      "loss": 0.0184,
      "step": 3139
    },
    {
      "epoch": 2.977714556661925,
      "grad_norm": 0.3947867296797152,
      "learning_rate": 1.475364111653832e-09,
      "loss": 0.0275,
      "step": 3140
    },
    {
      "epoch": 2.9786628733997156,
      "grad_norm": 0.30521605231534926,
      "learning_rate": 1.3442942482710941e-09,
      "loss": 0.0176,
      "step": 3141
    },
    {
      "epoch": 2.979611190137506,
      "grad_norm": 0.3974193523935978,
      "learning_rate": 1.2193195908388744e-09,
      "loss": 0.0174,
      "step": 3142
    },
    {
      "epoch": 2.9805595068752964,
      "grad_norm": 0.3321585434506631,
      "learning_rate": 1.1004402917469403e-09,
      "loss": 0.0156,
      "step": 3143
    },
    {
      "epoch": 2.9815078236130867,
      "grad_norm": 0.4374292249333356,
      "learning_rate": 9.876564959526712e-10,
      "loss": 0.0265,
      "step": 3144
    },
    {
      "epoch": 2.982456140350877,
      "grad_norm": 0.36532880710618687,
      "learning_rate": 8.809683409816138e-10,
      "loss": 0.0178,
      "step": 3145
    },
    {
      "epoch": 2.9834044570886675,
      "grad_norm": 0.39027337727641087,
      "learning_rate": 7.803759569258163e-10,
      "loss": 0.0243,
      "step": 3146
    },
    {
      "epoch": 2.984352773826458,
      "grad_norm": 0.37419423922853146,
      "learning_rate": 6.858794664449386e-10,
      "loss": 0.0176,
      "step": 3147
    },
    {
      "epoch": 2.9853010905642483,
      "grad_norm": 0.6397596966792142,
      "learning_rate": 5.974789847640328e-10,
      "loss": 0.0235,
      "step": 3148
    },
    {
      "epoch": 2.986249407302039,
      "grad_norm": 0.3484579240241185,
      "learning_rate": 5.151746196774277e-10,
      "loss": 0.0203,
      "step": 3149
    },
    {
      "epoch": 2.987197724039829,
      "grad_norm": 0.417063201267999,
      "learning_rate": 4.389664715431785e-10,
      "loss": 0.02,
      "step": 3150
    },
    {
      "epoch": 2.98814604077762,
      "grad_norm": 0.2971137707448454,
      "learning_rate": 3.688546332875076e-10,
      "loss": 0.0154,
      "step": 3151
    },
    {
      "epoch": 2.9890943575154103,
      "grad_norm": 0.4175389094327218,
      "learning_rate": 3.048391904031389e-10,
      "loss": 0.0171,
      "step": 3152
    },
    {
      "epoch": 2.9900426742532007,
      "grad_norm": 0.3809367324704737,
      "learning_rate": 2.469202209476329e-10,
      "loss": 0.0218,
      "step": 3153
    },
    {
      "epoch": 2.990990990990991,
      "grad_norm": 0.4043099039616555,
      "learning_rate": 1.9509779554671704e-10,
      "loss": 0.0252,
      "step": 3154
    },
    {
      "epoch": 2.9919393077287815,
      "grad_norm": 0.3951864633474192,
      "learning_rate": 1.49371977389845e-10,
      "loss": 0.0149,
      "step": 3155
    },
    {
      "epoch": 2.992887624466572,
      "grad_norm": 0.4273760381840025,
      "learning_rate": 1.097428222346375e-10,
      "loss": 0.0196,
      "step": 3156
    },
    {
      "epoch": 2.9938359412043622,
      "grad_norm": 0.40712842132273613,
      "learning_rate": 7.62103784029966e-11,
      "loss": 0.026,
      "step": 3157
    },
    {
      "epoch": 2.9947842579421526,
      "grad_norm": 0.49552854470553426,
      "learning_rate": 4.87746867838812e-11,
      "loss": 0.022,
      "step": 3158
    },
    {
      "epoch": 2.995732574679943,
      "grad_norm": 0.3752732337103295,
      "learning_rate": 2.7435780831086557e-11,
      "loss": 0.0185,
      "step": 3159
    },
    {
      "epoch": 2.9966808914177334,
      "grad_norm": 0.46297687373006957,
      "learning_rate": 1.21936865643546e-11,
      "loss": 0.0211,
      "step": 3160
    },
    {
      "epoch": 2.997629208155524,
      "grad_norm": 0.3278922229087332,
      "learning_rate": 3.0484225704841e-12,
      "loss": 0.0157,
      "step": 3161
    },
    {
      "epoch": 2.9985775248933146,
      "grad_norm": 0.4130595870331912,
      "learning_rate": 0.0,
      "loss": 0.0283,
      "step": 3162
    },
    {
      "epoch": 2.9985775248933146,
      "eval_loss": 0.03720796853303909,
      "eval_runtime": 200.101,
      "eval_samples_per_second": 35.497,
      "eval_steps_per_second": 1.109,
      "step": 3162
    },
    {
      "epoch": 2.9985775248933146,
      "step": 3162,
      "total_flos": 1.1417019656175288e+18,
      "train_loss": 0.048461434403982234,
      "train_runtime": 39585.3646,
      "train_samples_per_second": 10.226,
      "train_steps_per_second": 0.08
    }
  ],
  "logging_steps": 1,
  "max_steps": 3162,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.1417019656175288e+18,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}