{
  "best_global_step": 12260,
  "best_metric": 0.14923527836799622,
  "best_model_checkpoint": "saves/lntuning/llama-3-8b-instruct/train_multirc_1753094164/checkpoint-12260",
  "epoch": 10.0,
  "eval_steps": 3065,
  "global_step": 61300,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0008156606851549756,
      "grad_norm": 3.3118317127227783,
      "learning_rate": 3.262642740619902e-08,
      "loss": 1.8213,
      "num_input_tokens_seen": 8928,
      "step": 5
    },
    {
      "epoch": 0.0016313213703099511,
      "grad_norm": 18.186201095581055,
      "learning_rate": 7.34094616639478e-08,
      "loss": 1.1136,
      "num_input_tokens_seen": 20448,
      "step": 10
    },
    {
      "epoch": 0.0024469820554649264,
      "grad_norm": 7.66547155380249,
      "learning_rate": 1.1419249592169658e-07,
      "loss": 0.4922,
      "num_input_tokens_seen": 31072,
      "step": 15
    },
    {
      "epoch": 0.0032626427406199023,
      "grad_norm": 26.939163208007812,
      "learning_rate": 1.5497553017944535e-07,
      "loss": 1.0222,
      "num_input_tokens_seen": 42528,
      "step": 20
    },
    {
      "epoch": 0.004078303425774877,
      "grad_norm": 10.637971878051758,
      "learning_rate": 1.9575856443719413e-07,
      "loss": 1.2831,
      "num_input_tokens_seen": 52576,
      "step": 25
    },
    {
      "epoch": 0.004893964110929853,
      "grad_norm": 16.892108917236328,
      "learning_rate": 2.365415986949429e-07,
      "loss": 0.7375,
      "num_input_tokens_seen": 62944,
      "step": 30
    },
    {
      "epoch": 0.005709624796084829,
      "grad_norm": 5.437825679779053,
      "learning_rate": 2.773246329526917e-07,
      "loss": 0.7678,
      "num_input_tokens_seen": 73504,
      "step": 35
    },
    {
      "epoch": 0.0065252854812398045,
      "grad_norm": 17.547048568725586,
      "learning_rate": 3.1810766721044045e-07,
      "loss": 0.994,
      "num_input_tokens_seen": 84640,
      "step": 40
    },
    {
      "epoch": 0.00734094616639478,
      "grad_norm": 4.541830062866211,
      "learning_rate": 3.5889070146818926e-07,
      "loss": 0.7076,
      "num_input_tokens_seen": 96288,
      "step": 45
    },
    {
      "epoch": 0.008156606851549755,
      "grad_norm": 13.801979064941406,
      "learning_rate": 3.99673735725938e-07,
      "loss": 0.9782,
      "num_input_tokens_seen": 107360,
      "step": 50
    },
    {
      "epoch": 0.00897226753670473,
      "grad_norm": 14.718707084655762,
      "learning_rate": 4.4045676998368683e-07,
      "loss": 0.83,
      "num_input_tokens_seen": 118432,
      "step": 55
    },
    {
      "epoch": 0.009787928221859706,
      "grad_norm": 5.254812717437744,
      "learning_rate": 4.812398042414356e-07,
      "loss": 0.9871,
      "num_input_tokens_seen": 128416,
      "step": 60
    },
    {
      "epoch": 0.010603588907014683,
      "grad_norm": 4.80765962600708,
      "learning_rate": 5.220228384991843e-07,
      "loss": 0.4167,
      "num_input_tokens_seen": 138432,
      "step": 65
    },
    {
      "epoch": 0.011419249592169658,
      "grad_norm": 18.233137130737305,
      "learning_rate": 5.628058727569332e-07,
      "loss": 0.6041,
      "num_input_tokens_seen": 149504,
      "step": 70
    },
    {
      "epoch": 0.012234910277324634,
      "grad_norm": 9.67054557800293,
      "learning_rate": 6.03588907014682e-07,
      "loss": 0.6387,
      "num_input_tokens_seen": 160192,
      "step": 75
    },
    {
      "epoch": 0.013050570962479609,
      "grad_norm": 5.773282051086426,
      "learning_rate": 6.443719412724307e-07,
      "loss": 0.7717,
      "num_input_tokens_seen": 170432,
      "step": 80
    },
    {
      "epoch": 0.013866231647634585,
      "grad_norm": 5.768527030944824,
      "learning_rate": 6.851549755301795e-07,
      "loss": 0.6047,
      "num_input_tokens_seen": 180672,
      "step": 85
    },
    {
      "epoch": 0.01468189233278956,
      "grad_norm": 8.126620292663574,
      "learning_rate": 7.259380097879283e-07,
      "loss": 0.7343,
      "num_input_tokens_seen": 192352,
      "step": 90
    },
    {
      "epoch": 0.015497553017944535,
      "grad_norm": 23.64755630493164,
      "learning_rate": 7.66721044045677e-07,
      "loss": 0.7276,
      "num_input_tokens_seen": 203040,
      "step": 95
    },
    {
      "epoch": 0.01631321370309951,
      "grad_norm": 13.55368709564209,
      "learning_rate": 8.075040783034258e-07,
      "loss": 0.9729,
      "num_input_tokens_seen": 214272,
      "step": 100
    },
    {
      "epoch": 0.017128874388254486,
      "grad_norm": 10.714468955993652,
      "learning_rate": 8.482871125611746e-07,
      "loss": 0.6276,
      "num_input_tokens_seen": 225664,
      "step": 105
    },
    {
      "epoch": 0.01794453507340946,
      "grad_norm": 7.814089775085449,
      "learning_rate": 8.890701468189233e-07,
      "loss": 1.608,
      "num_input_tokens_seen": 237216,
      "step": 110
    },
    {
      "epoch": 0.018760195758564437,
      "grad_norm": 18.210569381713867,
      "learning_rate": 9.298531810766722e-07,
      "loss": 0.6829,
      "num_input_tokens_seen": 246560,
      "step": 115
    },
    {
      "epoch": 0.01957585644371941,
      "grad_norm": 22.19220733642578,
      "learning_rate": 9.70636215334421e-07,
      "loss": 1.3417,
      "num_input_tokens_seen": 257184,
      "step": 120
    },
    {
      "epoch": 0.020391517128874388,
      "grad_norm": 16.29233169555664,
      "learning_rate": 1.0114192495921699e-06,
      "loss": 0.6808,
      "num_input_tokens_seen": 268384,
      "step": 125
    },
    {
      "epoch": 0.021207177814029365,
      "grad_norm": 6.1285319328308105,
      "learning_rate": 1.0522022838499183e-06,
      "loss": 0.5364,
      "num_input_tokens_seen": 279680,
      "step": 130
    },
    {
      "epoch": 0.02202283849918434,
      "grad_norm": 14.859761238098145,
      "learning_rate": 1.0929853181076673e-06,
      "loss": 1.211,
      "num_input_tokens_seen": 290688,
      "step": 135
    },
    {
      "epoch": 0.022838499184339316,
      "grad_norm": 17.23931312561035,
      "learning_rate": 1.133768352365416e-06,
      "loss": 0.8324,
      "num_input_tokens_seen": 301408,
      "step": 140
    },
    {
      "epoch": 0.02365415986949429,
      "grad_norm": 5.048580646514893,
      "learning_rate": 1.1745513866231649e-06,
      "loss": 1.0377,
      "num_input_tokens_seen": 311968,
      "step": 145
    },
    {
      "epoch": 0.024469820554649267,
      "grad_norm": 13.512179374694824,
      "learning_rate": 1.2153344208809136e-06,
      "loss": 1.3006,
      "num_input_tokens_seen": 322080,
      "step": 150
    },
    {
      "epoch": 0.02528548123980424,
      "grad_norm": 12.56611442565918,
      "learning_rate": 1.2561174551386625e-06,
      "loss": 1.224,
      "num_input_tokens_seen": 333824,
      "step": 155
    },
    {
      "epoch": 0.026101141924959218,
      "grad_norm": 6.651486873626709,
      "learning_rate": 1.296900489396411e-06,
      "loss": 1.5912,
      "num_input_tokens_seen": 344896,
      "step": 160
    },
    {
      "epoch": 0.026916802610114192,
      "grad_norm": 5.611935138702393,
      "learning_rate": 1.33768352365416e-06,
      "loss": 0.8486,
      "num_input_tokens_seen": 356192,
      "step": 165
    },
    {
      "epoch": 0.02773246329526917,
      "grad_norm": 6.0018630027771,
      "learning_rate": 1.3784665579119086e-06,
      "loss": 0.5632,
      "num_input_tokens_seen": 367136,
      "step": 170
    },
    {
      "epoch": 0.028548123980424143,
      "grad_norm": 3.167198419570923,
      "learning_rate": 1.4192495921696575e-06,
      "loss": 0.6989,
      "num_input_tokens_seen": 376832,
      "step": 175
    },
    {
      "epoch": 0.02936378466557912,
      "grad_norm": 4.618821144104004,
      "learning_rate": 1.4600326264274062e-06,
      "loss": 0.4143,
      "num_input_tokens_seen": 387552,
      "step": 180
    },
    {
      "epoch": 0.030179445350734094,
      "grad_norm": 23.31751251220703,
      "learning_rate": 1.5008156606851552e-06,
      "loss": 0.9319,
      "num_input_tokens_seen": 398528,
      "step": 185
    },
    {
      "epoch": 0.03099510603588907,
      "grad_norm": 17.619123458862305,
      "learning_rate": 1.5415986949429036e-06,
      "loss": 0.8091,
      "num_input_tokens_seen": 408896,
      "step": 190
    },
    {
      "epoch": 0.03181076672104405,
      "grad_norm": 3.5539722442626953,
      "learning_rate": 1.5823817292006523e-06,
      "loss": 0.6515,
      "num_input_tokens_seen": 420416,
      "step": 195
    },
    {
      "epoch": 0.03262642740619902,
      "grad_norm": 18.352197647094727,
      "learning_rate": 1.6231647634584013e-06,
      "loss": 0.4857,
      "num_input_tokens_seen": 430912,
      "step": 200
    },
    {
      "epoch": 0.033442088091353996,
      "grad_norm": 26.030839920043945,
      "learning_rate": 1.6639477977161502e-06,
      "loss": 0.8654,
      "num_input_tokens_seen": 442176,
      "step": 205
    },
    {
      "epoch": 0.03425774877650897,
      "grad_norm": 11.81512451171875,
      "learning_rate": 1.704730831973899e-06,
      "loss": 0.5263,
      "num_input_tokens_seen": 454496,
      "step": 210
    },
    {
      "epoch": 0.03507340946166395,
      "grad_norm": 12.172980308532715,
      "learning_rate": 1.7455138662316478e-06,
      "loss": 0.3995,
      "num_input_tokens_seen": 464896,
      "step": 215
    },
    {
      "epoch": 0.03588907014681892,
      "grad_norm": 9.735648155212402,
      "learning_rate": 1.7862969004893963e-06,
      "loss": 1.0462,
      "num_input_tokens_seen": 477024,
      "step": 220
    },
    {
      "epoch": 0.0367047308319739,
      "grad_norm": 16.229707717895508,
      "learning_rate": 1.8270799347471452e-06,
      "loss": 0.5669,
      "num_input_tokens_seen": 487776,
      "step": 225
    },
    {
      "epoch": 0.037520391517128875,
      "grad_norm": 1.8517504930496216,
      "learning_rate": 1.8678629690048941e-06,
      "loss": 0.6044,
      "num_input_tokens_seen": 498528,
      "step": 230
    },
    {
      "epoch": 0.03833605220228385,
      "grad_norm": 5.680665493011475,
      "learning_rate": 1.908646003262643e-06,
      "loss": 0.2991,
      "num_input_tokens_seen": 509184,
      "step": 235
    },
    {
      "epoch": 0.03915171288743882,
      "grad_norm": 4.058143615722656,
      "learning_rate": 1.9494290375203913e-06,
      "loss": 0.5362,
      "num_input_tokens_seen": 520224,
      "step": 240
    },
    {
      "epoch": 0.0399673735725938,
      "grad_norm": 19.857254028320312,
      "learning_rate": 1.9902120717781402e-06,
      "loss": 1.5284,
      "num_input_tokens_seen": 530464,
      "step": 245
    },
    {
      "epoch": 0.040783034257748776,
      "grad_norm": 15.53018569946289,
      "learning_rate": 2.030995106035889e-06,
      "loss": 1.1205,
      "num_input_tokens_seen": 541952,
      "step": 250
    },
    {
      "epoch": 0.041598694942903754,
      "grad_norm": 3.2097983360290527,
      "learning_rate": 2.071778140293638e-06,
      "loss": 0.3677,
      "num_input_tokens_seen": 551424,
      "step": 255
    },
    {
      "epoch": 0.04241435562805873,
      "grad_norm": 10.602266311645508,
      "learning_rate": 2.1125611745513866e-06,
      "loss": 1.2435,
      "num_input_tokens_seen": 561440,
      "step": 260
    },
    {
      "epoch": 0.0432300163132137,
      "grad_norm": 15.465632438659668,
      "learning_rate": 2.1533442088091355e-06,
      "loss": 0.9176,
      "num_input_tokens_seen": 573440,
      "step": 265
    },
    {
      "epoch": 0.04404567699836868,
      "grad_norm": 6.000158309936523,
      "learning_rate": 2.1941272430668844e-06,
      "loss": 0.3343,
      "num_input_tokens_seen": 584032,
      "step": 270
    },
    {
      "epoch": 0.044861337683523655,
      "grad_norm": 22.78145980834961,
      "learning_rate": 2.2349102773246333e-06,
      "loss": 0.6821,
      "num_input_tokens_seen": 595328,
      "step": 275
    },
    {
      "epoch": 0.04567699836867863,
      "grad_norm": 4.328393936157227,
      "learning_rate": 2.275693311582382e-06,
      "loss": 1.0264,
      "num_input_tokens_seen": 605760,
      "step": 280
    },
    {
      "epoch": 0.0464926590538336,
      "grad_norm": 8.35729694366455,
      "learning_rate": 2.3164763458401307e-06,
      "loss": 0.5896,
      "num_input_tokens_seen": 616384,
      "step": 285
    },
    {
      "epoch": 0.04730831973898858,
      "grad_norm": 7.698673248291016,
      "learning_rate": 2.357259380097879e-06,
      "loss": 0.6132,
      "num_input_tokens_seen": 626880,
      "step": 290
    },
    {
      "epoch": 0.04812398042414356,
      "grad_norm": 1.6100306510925293,
      "learning_rate": 2.398042414355628e-06,
      "loss": 0.7909,
      "num_input_tokens_seen": 637536,
      "step": 295
    },
    {
      "epoch": 0.048939641109298535,
      "grad_norm": 18.588539123535156,
      "learning_rate": 2.4388254486133766e-06,
      "loss": 0.6532,
      "num_input_tokens_seen": 648000,
      "step": 300
    },
    {
      "epoch": 0.049755301794453505,
      "grad_norm": 8.66322135925293,
      "learning_rate": 2.4796084828711255e-06,
      "loss": 1.227,
      "num_input_tokens_seen": 659648,
      "step": 305
    },
    {
      "epoch": 0.05057096247960848,
      "grad_norm": 9.428548812866211,
      "learning_rate": 2.5203915171288745e-06,
      "loss": 0.5282,
      "num_input_tokens_seen": 669920,
      "step": 310
    },
    {
      "epoch": 0.05138662316476346,
      "grad_norm": 4.988115310668945,
      "learning_rate": 2.5611745513866234e-06,
      "loss": 0.7282,
      "num_input_tokens_seen": 681216,
      "step": 315
    },
    {
      "epoch": 0.052202283849918436,
      "grad_norm": 8.257129669189453,
      "learning_rate": 2.6019575856443723e-06,
      "loss": 1.0687,
      "num_input_tokens_seen": 691872,
      "step": 320
    },
    {
      "epoch": 0.05301794453507341,
      "grad_norm": 15.32027816772461,
      "learning_rate": 2.6427406199021208e-06,
      "loss": 0.3635,
      "num_input_tokens_seen": 702208,
      "step": 325
    },
    {
      "epoch": 0.053833605220228384,
      "grad_norm": 13.268648147583008,
      "learning_rate": 2.6835236541598697e-06,
      "loss": 0.6009,
      "num_input_tokens_seen": 712288,
      "step": 330
    },
    {
      "epoch": 0.05464926590538336,
      "grad_norm": 11.272858619689941,
      "learning_rate": 2.7243066884176186e-06,
      "loss": 1.0324,
      "num_input_tokens_seen": 722976,
      "step": 335
    },
    {
      "epoch": 0.05546492659053834,
      "grad_norm": 2.6093666553497314,
      "learning_rate": 2.7650897226753675e-06,
      "loss": 0.5323,
      "num_input_tokens_seen": 733760,
      "step": 340
    },
    {
      "epoch": 0.05628058727569331,
      "grad_norm": 2.8714146614074707,
      "learning_rate": 2.805872756933116e-06,
      "loss": 0.7561,
      "num_input_tokens_seen": 743520,
      "step": 345
    },
    {
      "epoch": 0.057096247960848286,
      "grad_norm": 3.42722225189209,
      "learning_rate": 2.8466557911908645e-06,
      "loss": 0.0811,
      "num_input_tokens_seen": 753984,
      "step": 350
    },
    {
      "epoch": 0.05791190864600326,
      "grad_norm": 2.2392630577087402,
      "learning_rate": 2.8874388254486134e-06,
      "loss": 0.3957,
      "num_input_tokens_seen": 765376,
      "step": 355
    },
    {
      "epoch": 0.05872756933115824,
      "grad_norm": 4.543656349182129,
      "learning_rate": 2.9282218597063623e-06,
      "loss": 0.7068,
      "num_input_tokens_seen": 775904,
      "step": 360
    },
    {
      "epoch": 0.05954323001631321,
      "grad_norm": 1.3153177499771118,
      "learning_rate": 2.969004893964111e-06,
      "loss": 0.7697,
      "num_input_tokens_seen": 788000,
      "step": 365
    },
    {
      "epoch": 0.06035889070146819,
      "grad_norm": 13.002714157104492,
      "learning_rate": 3.0097879282218597e-06,
      "loss": 0.9799,
      "num_input_tokens_seen": 798816,
      "step": 370
    },
    {
      "epoch": 0.061174551386623165,
      "grad_norm": 1.001260757446289,
      "learning_rate": 3.0505709624796087e-06,
      "loss": 0.365,
      "num_input_tokens_seen": 809856,
      "step": 375
    },
    {
      "epoch": 0.06199021207177814,
      "grad_norm": 1.4286184310913086,
      "learning_rate": 3.0913539967373576e-06,
      "loss": 0.523,
      "num_input_tokens_seen": 820448,
      "step": 380
    },
    {
      "epoch": 0.06280587275693311,
      "grad_norm": 18.706470489501953,
      "learning_rate": 3.132137030995106e-06,
      "loss": 1.088,
      "num_input_tokens_seen": 830624,
      "step": 385
    },
    {
      "epoch": 0.0636215334420881,
      "grad_norm": 1.0630874633789062,
      "learning_rate": 3.1729200652528554e-06,
      "loss": 0.5391,
      "num_input_tokens_seen": 840224,
      "step": 390
    },
    {
      "epoch": 0.06443719412724307,
      "grad_norm": 17.3529109954834,
      "learning_rate": 3.213703099510604e-06,
      "loss": 0.5071,
      "num_input_tokens_seen": 850176,
      "step": 395
    },
    {
      "epoch": 0.06525285481239804,
      "grad_norm": 3.985969066619873,
      "learning_rate": 3.2544861337683524e-06,
      "loss": 0.3553,
      "num_input_tokens_seen": 861056,
      "step": 400
    },
    {
      "epoch": 0.06606851549755302,
      "grad_norm": 8.156689643859863,
      "learning_rate": 3.2952691680261013e-06,
      "loss": 0.8092,
      "num_input_tokens_seen": 871360,
      "step": 405
    },
    {
      "epoch": 0.06688417618270799,
      "grad_norm": 31.989702224731445,
      "learning_rate": 3.33605220228385e-06,
      "loss": 1.3592,
      "num_input_tokens_seen": 882656,
      "step": 410
    },
    {
      "epoch": 0.06769983686786298,
      "grad_norm": 11.47252368927002,
      "learning_rate": 3.3768352365415987e-06,
      "loss": 0.5707,
      "num_input_tokens_seen": 893952,
      "step": 415
    },
    {
      "epoch": 0.06851549755301795,
      "grad_norm": 16.895809173583984,
      "learning_rate": 3.4176182707993476e-06,
      "loss": 0.3123,
      "num_input_tokens_seen": 905472,
      "step": 420
    },
    {
      "epoch": 0.06933115823817292,
      "grad_norm": 0.7314866781234741,
      "learning_rate": 3.458401305057096e-06,
      "loss": 0.5118,
      "num_input_tokens_seen": 916000,
      "step": 425
    },
    {
      "epoch": 0.0701468189233279,
      "grad_norm": 7.430657863616943,
      "learning_rate": 3.4991843393148455e-06,
      "loss": 0.594,
      "num_input_tokens_seen": 926944,
      "step": 430
    },
    {
      "epoch": 0.07096247960848287,
      "grad_norm": 2.6516807079315186,
      "learning_rate": 3.539967373572594e-06,
      "loss": 0.6643,
      "num_input_tokens_seen": 938496,
      "step": 435
    },
    {
      "epoch": 0.07177814029363784,
      "grad_norm": 0.5453417301177979,
      "learning_rate": 3.5807504078303425e-06,
      "loss": 0.7201,
      "num_input_tokens_seen": 949472,
      "step": 440
    },
    {
      "epoch": 0.07259380097879282,
      "grad_norm": 17.167049407958984,
      "learning_rate": 3.621533442088092e-06,
      "loss": 0.8084,
      "num_input_tokens_seen": 959712,
      "step": 445
    },
    {
      "epoch": 0.0734094616639478,
      "grad_norm": 19.318767547607422,
      "learning_rate": 3.6623164763458403e-06,
      "loss": 0.963,
      "num_input_tokens_seen": 969888,
      "step": 450
    },
    {
      "epoch": 0.07422512234910278,
      "grad_norm": 1.3418043851852417,
      "learning_rate": 3.7030995106035896e-06,
      "loss": 0.7009,
      "num_input_tokens_seen": 981024,
      "step": 455
    },
    {
      "epoch": 0.07504078303425775,
      "grad_norm": 13.274270057678223,
      "learning_rate": 3.743882544861338e-06,
      "loss": 0.3469,
      "num_input_tokens_seen": 991680,
      "step": 460
    },
    {
      "epoch": 0.07585644371941272,
      "grad_norm": 1.2422374486923218,
      "learning_rate": 3.7846655791190866e-06,
      "loss": 0.7731,
      "num_input_tokens_seen": 1003008,
      "step": 465
    },
    {
      "epoch": 0.0766721044045677,
      "grad_norm": 3.4855055809020996,
      "learning_rate": 3.8254486133768355e-06,
      "loss": 0.2859,
      "num_input_tokens_seen": 1014336,
      "step": 470
    },
    {
      "epoch": 0.07748776508972267,
      "grad_norm": 12.034229278564453,
      "learning_rate": 3.866231647634584e-06,
      "loss": 0.7338,
      "num_input_tokens_seen": 1024960,
      "step": 475
    },
    {
      "epoch": 0.07830342577487764,
      "grad_norm": 0.8161830306053162,
      "learning_rate": 3.9070146818923325e-06,
      "loss": 0.2315,
      "num_input_tokens_seen": 1034880,
      "step": 480
    },
    {
      "epoch": 0.07911908646003263,
      "grad_norm": 7.478862285614014,
      "learning_rate": 3.947797716150082e-06,
      "loss": 0.8651,
      "num_input_tokens_seen": 1046112,
      "step": 485
    },
    {
      "epoch": 0.0799347471451876,
      "grad_norm": 22.26531219482422,
      "learning_rate": 3.98858075040783e-06,
      "loss": 0.8098,
      "num_input_tokens_seen": 1056864,
      "step": 490
    },
    {
      "epoch": 0.08075040783034258,
      "grad_norm": 0.40392860770225525,
      "learning_rate": 4.02936378466558e-06,
      "loss": 0.5404,
      "num_input_tokens_seen": 1067072,
      "step": 495
    },
    {
      "epoch": 0.08156606851549755,
      "grad_norm": 12.599085807800293,
      "learning_rate": 4.070146818923328e-06,
      "loss": 0.498,
      "num_input_tokens_seen": 1079136,
      "step": 500
    },
    {
      "epoch": 0.08238172920065252,
      "grad_norm": 0.784478485584259,
      "learning_rate": 4.110929853181077e-06,
      "loss": 0.9097,
      "num_input_tokens_seen": 1089408,
      "step": 505
    },
    {
      "epoch": 0.08319738988580751,
      "grad_norm": 10.010594367980957,
      "learning_rate": 4.151712887438826e-06,
      "loss": 0.2794,
      "num_input_tokens_seen": 1099488,
      "step": 510
    },
    {
      "epoch": 0.08401305057096248,
      "grad_norm": 4.048063278198242,
      "learning_rate": 4.1924959216965745e-06,
      "loss": 0.675,
      "num_input_tokens_seen": 1110464,
      "step": 515
    },
    {
      "epoch": 0.08482871125611746,
      "grad_norm": 19.317440032958984,
      "learning_rate": 4.233278955954323e-06,
      "loss": 0.3273,
      "num_input_tokens_seen": 1121216,
      "step": 520
    },
    {
      "epoch": 0.08564437194127243,
      "grad_norm": 18.039342880249023,
      "learning_rate": 4.274061990212072e-06,
      "loss": 0.493,
      "num_input_tokens_seen": 1132192,
      "step": 525
    },
    {
      "epoch": 0.0864600326264274,
      "grad_norm": 11.444881439208984,
      "learning_rate": 4.314845024469821e-06,
      "loss": 0.6106,
      "num_input_tokens_seen": 1142944,
      "step": 530
    },
    {
      "epoch": 0.08727569331158239,
      "grad_norm": 9.746691703796387,
      "learning_rate": 4.35562805872757e-06,
      "loss": 0.6221,
      "num_input_tokens_seen": 1154080,
      "step": 535
    },
    {
      "epoch": 0.08809135399673736,
      "grad_norm": 13.027948379516602,
      "learning_rate": 4.396411092985319e-06,
      "loss": 0.488,
      "num_input_tokens_seen": 1165120,
      "step": 540
    },
    {
      "epoch": 0.08890701468189233,
      "grad_norm": 3.0586557388305664,
      "learning_rate": 4.437194127243067e-06,
      "loss": 0.2867,
      "num_input_tokens_seen": 1176352,
      "step": 545
    },
    {
      "epoch": 0.08972267536704731,
      "grad_norm": 8.584217071533203,
      "learning_rate": 4.477977161500816e-06,
      "loss": 0.2178,
      "num_input_tokens_seen": 1188288,
      "step": 550
    },
    {
      "epoch": 0.09053833605220228,
      "grad_norm": 0.5211539268493652,
      "learning_rate": 4.518760195758565e-06,
      "loss": 0.4756,
      "num_input_tokens_seen": 1200800,
      "step": 555
    },
    {
      "epoch": 0.09135399673735727,
      "grad_norm": 11.939558029174805,
      "learning_rate": 4.5595432300163135e-06,
      "loss": 0.5071,
      "num_input_tokens_seen": 1211360,
      "step": 560
    },
    {
      "epoch": 0.09216965742251224,
      "grad_norm": 5.799455165863037,
      "learning_rate": 4.600326264274062e-06,
      "loss": 0.3327,
      "num_input_tokens_seen": 1221184,
      "step": 565
    },
    {
      "epoch": 0.0929853181076672,
      "grad_norm": 4.998054504394531,
      "learning_rate": 4.6411092985318105e-06,
      "loss": 0.651,
      "num_input_tokens_seen": 1231808,
      "step": 570
    },
    {
      "epoch": 0.09380097879282219,
      "grad_norm": 12.160737037658691,
      "learning_rate": 4.68189233278956e-06,
      "loss": 0.3156,
      "num_input_tokens_seen": 1242560,
      "step": 575
    },
    {
      "epoch": 0.09461663947797716,
      "grad_norm": 3.426234245300293,
      "learning_rate": 4.722675367047308e-06,
      "loss": 0.183,
      "num_input_tokens_seen": 1253440,
      "step": 580
    },
    {
      "epoch": 0.09543230016313213,
      "grad_norm": 17.633731842041016,
      "learning_rate": 4.763458401305057e-06,
      "loss": 0.8472,
      "num_input_tokens_seen": 1263616,
      "step": 585
    },
    {
      "epoch": 0.09624796084828711,
      "grad_norm": 14.605329513549805,
      "learning_rate": 4.804241435562806e-06,
      "loss": 0.244,
      "num_input_tokens_seen": 1275488,
      "step": 590
    },
    {
      "epoch": 0.09706362153344208,
      "grad_norm": 0.626120388507843,
      "learning_rate": 4.845024469820555e-06,
      "loss": 0.2679,
      "num_input_tokens_seen": 1287200,
      "step": 595
    },
    {
      "epoch": 0.09787928221859707,
      "grad_norm": 7.383834362030029,
      "learning_rate": 4.885807504078304e-06,
      "loss": 0.7493,
      "num_input_tokens_seen": 1298336,
      "step": 600
    },
    {
      "epoch": 0.09869494290375204,
      "grad_norm": 4.6556620597839355,
      "learning_rate": 4.9265905383360524e-06,
      "loss": 0.2895,
      "num_input_tokens_seen": 1309056,
      "step": 605
    },
    {
      "epoch": 0.09951060358890701,
      "grad_norm": 1.5424631834030151,
      "learning_rate": 4.967373572593801e-06,
      "loss": 0.333,
      "num_input_tokens_seen": 1320864,
      "step": 610
    },
    {
      "epoch": 0.100326264274062,
      "grad_norm": 10.238938331604004,
      "learning_rate": 5.00815660685155e-06,
      "loss": 0.7882,
      "num_input_tokens_seen": 1332448,
      "step": 615
    },
    {
      "epoch": 0.10114192495921696,
      "grad_norm": 9.649983406066895,
      "learning_rate": 5.048939641109299e-06,
      "loss": 0.5933,
      "num_input_tokens_seen": 1342368,
      "step": 620
    },
    {
      "epoch": 0.10195758564437195,
      "grad_norm": 14.703304290771484,
      "learning_rate": 5.089722675367047e-06,
      "loss": 0.3742,
      "num_input_tokens_seen": 1353024,
      "step": 625
    },
    {
      "epoch": 0.10277324632952692,
      "grad_norm": 12.713678359985352,
      "learning_rate": 5.130505709624797e-06,
      "loss": 0.2437,
      "num_input_tokens_seen": 1364000,
      "step": 630
    },
    {
      "epoch": 0.10358890701468189,
      "grad_norm": 6.326504230499268,
      "learning_rate": 5.171288743882545e-06,
      "loss": 0.9262,
      "num_input_tokens_seen": 1374784,
      "step": 635
    },
    {
      "epoch": 0.10440456769983687,
      "grad_norm": 0.16197632253170013,
      "learning_rate": 5.2120717781402944e-06,
      "loss": 0.4605,
      "num_input_tokens_seen": 1384192,
      "step": 640
    },
    {
      "epoch": 0.10522022838499184,
      "grad_norm": 0.30914539098739624,
      "learning_rate": 5.252854812398043e-06,
      "loss": 0.1618,
      "num_input_tokens_seen": 1395392,
      "step": 645
    },
    {
      "epoch": 0.10603588907014681,
      "grad_norm": 6.694571018218994,
      "learning_rate": 5.293637846655791e-06,
      "loss": 0.29,
      "num_input_tokens_seen": 1406208,
      "step": 650
    },
    {
      "epoch": 0.1068515497553018,
      "grad_norm": 0.6397295594215393,
      "learning_rate": 5.334420880913541e-06,
      "loss": 0.2999,
      "num_input_tokens_seen": 1417024,
      "step": 655
    },
    {
      "epoch": 0.10766721044045677,
      "grad_norm": 0.3513702154159546,
      "learning_rate": 5.375203915171289e-06,
      "loss": 0.5852,
      "num_input_tokens_seen": 1427808,
      "step": 660
    },
    {
      "epoch": 0.10848287112561175,
      "grad_norm": 16.48516273498535,
      "learning_rate": 5.415986949429038e-06,
      "loss": 0.1626,
      "num_input_tokens_seen": 1437344,
      "step": 665
    },
    {
      "epoch": 0.10929853181076672,
      "grad_norm": 0.9608420729637146,
      "learning_rate": 5.456769983686786e-06,
      "loss": 0.3224,
      "num_input_tokens_seen": 1448352,
      "step": 670
    },
    {
      "epoch": 0.11011419249592169,
      "grad_norm": 2.4646878242492676,
      "learning_rate": 5.4975530179445356e-06,
      "loss": 0.1999,
      "num_input_tokens_seen": 1459520,
      "step": 675
    },
    {
      "epoch": 0.11092985318107668,
      "grad_norm": 1.0080846548080444,
      "learning_rate": 5.538336052202284e-06,
      "loss": 0.3119,
      "num_input_tokens_seen": 1470432,
      "step": 680
    },
    {
      "epoch": 0.11174551386623165,
      "grad_norm": 7.099843978881836,
      "learning_rate": 5.5791190864600326e-06,
      "loss": 0.3477,
      "num_input_tokens_seen": 1480544,
      "step": 685
    },
    {
      "epoch": 0.11256117455138662,
      "grad_norm": 9.918968200683594,
      "learning_rate": 5.619902120717781e-06,
      "loss": 0.3175,
      "num_input_tokens_seen": 1490752,
      "step": 690
    },
    {
      "epoch": 0.1133768352365416,
      "grad_norm": 13.043498039245605,
      "learning_rate": 5.66068515497553e-06,
      "loss": 0.2591,
      "num_input_tokens_seen": 1502240,
      "step": 695
    },
    {
      "epoch": 0.11419249592169657,
      "grad_norm": 9.64283275604248,
      "learning_rate": 5.701468189233279e-06,
      "loss": 0.4058,
      "num_input_tokens_seen": 1512256,
      "step": 700
    },
    {
      "epoch": 0.11500815660685156,
      "grad_norm": 0.20353975892066956,
      "learning_rate": 5.742251223491028e-06,
      "loss": 0.4454,
      "num_input_tokens_seen": 1523168,
      "step": 705
    },
    {
      "epoch": 0.11582381729200653,
      "grad_norm": 5.264739990234375,
      "learning_rate": 5.783034257748777e-06,
      "loss": 0.1964,
      "num_input_tokens_seen": 1534976,
      "step": 710
    },
    {
      "epoch": 0.1166394779771615,
      "grad_norm": 4.052374839782715,
      "learning_rate": 5.823817292006525e-06,
      "loss": 0.4494,
      "num_input_tokens_seen": 1545856,
      "step": 715
    },
    {
      "epoch": 0.11745513866231648,
      "grad_norm": 2.4262053966522217,
      "learning_rate": 5.8646003262642745e-06,
      "loss": 0.0953,
      "num_input_tokens_seen": 1555680,
      "step": 720
    },
    {
      "epoch": 0.11827079934747145,
      "grad_norm": 0.06375903636217117,
      "learning_rate": 5.905383360522023e-06,
      "loss": 0.0273,
      "num_input_tokens_seen": 1567328,
      "step": 725
    },
    {
      "epoch": 0.11908646003262642,
      "grad_norm": 11.60741901397705,
      "learning_rate": 5.9461663947797715e-06,
      "loss": 0.3622,
      "num_input_tokens_seen": 1577184,
      "step": 730
    },
    {
      "epoch": 0.1199021207177814,
      "grad_norm": 14.03055477142334,
      "learning_rate": 5.986949429037521e-06,
      "loss": 0.5287,
      "num_input_tokens_seen": 1588544,
      "step": 735
    },
    {
      "epoch": 0.12071778140293637,
      "grad_norm": 7.811862945556641,
      "learning_rate": 6.027732463295269e-06,
      "loss": 0.3598,
      "num_input_tokens_seen": 1599648,
      "step": 740
    },
    {
      "epoch": 0.12153344208809136,
      "grad_norm": 4.788828372955322,
      "learning_rate": 6.068515497553019e-06,
      "loss": 0.5577,
      "num_input_tokens_seen": 1612096,
      "step": 745
    },
    {
      "epoch": 0.12234910277324633,
      "grad_norm": 1.1032073497772217,
      "learning_rate": 6.109298531810767e-06,
      "loss": 0.0652,
      "num_input_tokens_seen": 1623584,
      "step": 750
    },
    {
      "epoch": 0.1231647634584013,
      "grad_norm": 11.049327850341797,
      "learning_rate": 6.150081566068516e-06,
      "loss": 0.2363,
      "num_input_tokens_seen": 1634688,
      "step": 755
    },
    {
      "epoch": 0.12398042414355628,
      "grad_norm": 10.284933090209961,
      "learning_rate": 6.190864600326265e-06,
      "loss": 0.517,
      "num_input_tokens_seen": 1645600,
      "step": 760
    },
    {
      "epoch": 0.12479608482871125,
      "grad_norm": 1.0161199569702148,
      "learning_rate": 6.2316476345840135e-06,
      "loss": 0.1868,
      "num_input_tokens_seen": 1655968,
      "step": 765
    },
    {
      "epoch": 0.12561174551386622,
      "grad_norm": 2.1625585556030273,
      "learning_rate": 6.272430668841763e-06,
      "loss": 0.3482,
      "num_input_tokens_seen": 1666848,
      "step": 770
    },
    {
      "epoch": 0.1264274061990212,
      "grad_norm": 13.45174789428711,
      "learning_rate": 6.3132137030995105e-06,
      "loss": 0.3038,
      "num_input_tokens_seen": 1676512,
      "step": 775
    },
    {
      "epoch": 0.1272430668841762,
      "grad_norm": 0.13592009246349335,
      "learning_rate": 6.35399673735726e-06,
      "loss": 0.614,
      "num_input_tokens_seen": 1687488,
      "step": 780
    },
    {
      "epoch": 0.12805872756933115,
      "grad_norm": 0.35469624400138855,
      "learning_rate": 6.394779771615008e-06,
      "loss": 0.5397,
      "num_input_tokens_seen": 1697312,
      "step": 785
    },
    {
      "epoch": 0.12887438825448613,
      "grad_norm": 10.634716033935547,
      "learning_rate": 6.435562805872757e-06,
      "loss": 0.5566,
      "num_input_tokens_seen": 1708096,
      "step": 790
    },
    {
      "epoch": 0.12969004893964112,
      "grad_norm": 0.12184014916419983,
      "learning_rate": 6.476345840130506e-06,
      "loss": 0.211,
      "num_input_tokens_seen": 1717824,
      "step": 795
    },
    {
      "epoch": 0.13050570962479607,
      "grad_norm": 5.945420265197754,
      "learning_rate": 6.517128874388255e-06,
      "loss": 0.7566,
      "num_input_tokens_seen": 1728608,
      "step": 800
    },
    {
      "epoch": 0.13132137030995106,
      "grad_norm": 4.466879367828369,
      "learning_rate": 6.557911908646004e-06,
      "loss": 0.1479,
      "num_input_tokens_seen": 1739456,
      "step": 805
    },
    {
      "epoch": 0.13213703099510604,
      "grad_norm": 7.0077948570251465,
      "learning_rate": 6.598694942903752e-06,
      "loss": 0.2857,
      "num_input_tokens_seen": 1750304,
      "step": 810
    },
    {
      "epoch": 0.132952691680261,
      "grad_norm": 6.266101837158203,
      "learning_rate": 6.639477977161501e-06,
      "loss": 0.0794,
      "num_input_tokens_seen": 1762496,
      "step": 815
    },
    {
      "epoch": 0.13376835236541598,
      "grad_norm": 7.936112880706787,
      "learning_rate": 6.68026101141925e-06,
      "loss": 0.2907,
      "num_input_tokens_seen": 1772640,
      "step": 820
    },
    {
      "epoch": 0.13458401305057097,
      "grad_norm": 0.10595320165157318,
      "learning_rate": 6.721044045676998e-06,
      "loss": 0.3284,
      "num_input_tokens_seen": 1782272,
      "step": 825
    },
    {
      "epoch": 0.13539967373572595,
      "grad_norm": 0.06858377903699875,
      "learning_rate": 6.761827079934747e-06,
      "loss": 0.2335,
      "num_input_tokens_seen": 1793120,
      "step": 830
    },
    {
      "epoch": 0.1362153344208809,
      "grad_norm": 10.44363784790039,
      "learning_rate": 6.802610114192497e-06,
      "loss": 0.4125,
      "num_input_tokens_seen": 1803264,
      "step": 835
    },
    {
      "epoch": 0.1370309951060359,
      "grad_norm": 3.860588312149048,
      "learning_rate": 6.843393148450244e-06,
      "loss": 0.137,
      "num_input_tokens_seen": 1813088,
      "step": 840
    },
    {
      "epoch": 0.13784665579119088,
      "grad_norm": 0.650309681892395,
      "learning_rate": 6.884176182707994e-06,
      "loss": 0.4879,
      "num_input_tokens_seen": 1824032,
      "step": 845
    },
    {
      "epoch": 0.13866231647634583,
      "grad_norm": 0.4944702982902527,
      "learning_rate": 6.924959216965743e-06,
      "loss": 0.3511,
      "num_input_tokens_seen": 1836512,
      "step": 850
    },
    {
      "epoch": 0.13947797716150082,
      "grad_norm": 0.20982003211975098,
      "learning_rate": 6.965742251223491e-06,
      "loss": 0.1203,
      "num_input_tokens_seen": 1848256,
      "step": 855
    },
    {
      "epoch": 0.1402936378466558,
      "grad_norm": 4.680217742919922,
      "learning_rate": 7.00652528548124e-06,
      "loss": 0.2072,
      "num_input_tokens_seen": 1859040,
      "step": 860
    },
    {
      "epoch": 0.14110929853181076,
      "grad_norm": 7.44214391708374,
      "learning_rate": 7.047308319738989e-06,
      "loss": 0.2866,
      "num_input_tokens_seen": 1869248,
      "step": 865
    },
    {
      "epoch": 0.14192495921696574,
      "grad_norm": 7.6567487716674805,
      "learning_rate": 7.088091353996739e-06,
      "loss": 0.5006,
      "num_input_tokens_seen": 1881184,
      "step": 870
    },
    {
      "epoch": 0.14274061990212072,
      "grad_norm": 0.6429446935653687,
      "learning_rate": 7.128874388254486e-06,
      "loss": 0.291,
      "num_input_tokens_seen": 1893760,
      "step": 875
    },
    {
      "epoch": 0.14355628058727568,
      "grad_norm": 4.556582927703857,
      "learning_rate": 7.169657422512236e-06,
      "loss": 0.1467,
      "num_input_tokens_seen": 1904800,
      "step": 880
    },
    {
      "epoch": 0.14437194127243066,
      "grad_norm": 0.15669970214366913,
      "learning_rate": 7.210440456769985e-06,
      "loss": 0.3282,
      "num_input_tokens_seen": 1915296,
      "step": 885
    },
    {
      "epoch": 0.14518760195758565,
      "grad_norm": 14.141194343566895,
      "learning_rate": 7.251223491027733e-06,
      "loss": 0.6234,
      "num_input_tokens_seen": 1925152,
      "step": 890
    },
    {
      "epoch": 0.14600326264274063,
      "grad_norm": 1.2011229991912842,
      "learning_rate": 7.292006525285482e-06,
      "loss": 0.0766,
      "num_input_tokens_seen": 1935200,
      "step": 895
    },
    {
      "epoch": 0.1468189233278956,
      "grad_norm": 0.25357726216316223,
      "learning_rate": 7.3327895595432304e-06,
      "loss": 0.4431,
      "num_input_tokens_seen": 1945632,
      "step": 900
    },
    {
      "epoch": 0.14763458401305057,
      "grad_norm": 11.212824821472168,
      "learning_rate": 7.373572593800979e-06,
      "loss": 0.2857,
      "num_input_tokens_seen": 1956992,
      "step": 905
    },
    {
      "epoch": 0.14845024469820556,
      "grad_norm": 0.09775493294000626,
      "learning_rate": 7.414355628058728e-06,
      "loss": 0.4232,
      "num_input_tokens_seen": 1968800,
      "step": 910
    },
    {
      "epoch": 0.14926590538336051,
      "grad_norm": 5.193148136138916,
      "learning_rate": 7.455138662316477e-06,
      "loss": 0.1417,
      "num_input_tokens_seen": 1979424,
      "step": 915
    },
    {
      "epoch": 0.1500815660685155,
      "grad_norm": 0.10857293009757996,
      "learning_rate": 7.495921696574225e-06,
      "loss": 0.4734,
      "num_input_tokens_seen": 1990976,
      "step": 920
    },
    {
      "epoch": 0.15089722675367048,
      "grad_norm": 9.459717750549316,
      "learning_rate": 7.536704730831974e-06,
      "loss": 0.1337,
      "num_input_tokens_seen": 2002432,
      "step": 925
    },
    {
      "epoch": 0.15171288743882544,
      "grad_norm": 0.15402376651763916,
      "learning_rate": 7.577487765089723e-06,
      "loss": 0.026,
      "num_input_tokens_seen": 2013632,
      "step": 930
    },
    {
      "epoch": 0.15252854812398042,
      "grad_norm": 12.633416175842285,
      "learning_rate": 7.6182707993474724e-06,
      "loss": 0.5885,
      "num_input_tokens_seen": 2024960,
      "step": 935
    },
    {
      "epoch": 0.1533442088091354,
      "grad_norm": 7.057833671569824,
      "learning_rate": 7.659053833605221e-06,
      "loss": 0.1533,
      "num_input_tokens_seen": 2035712,
      "step": 940
    },
    {
      "epoch": 0.15415986949429036,
      "grad_norm": 4.4467034339904785,
      "learning_rate": 7.69983686786297e-06,
      "loss": 0.1709,
      "num_input_tokens_seen": 2046656,
      "step": 945
    },
    {
      "epoch": 0.15497553017944535,
      "grad_norm": 5.337577819824219,
      "learning_rate": 7.740619902120718e-06,
      "loss": 0.2743,
      "num_input_tokens_seen": 2058112,
      "step": 950
    },
    {
      "epoch": 0.15579119086460033,
      "grad_norm": 10.638042449951172,
      "learning_rate": 7.781402936378467e-06,
      "loss": 0.1755,
      "num_input_tokens_seen": 2068768,
      "step": 955
    },
    {
      "epoch": 0.1566068515497553,
      "grad_norm": 5.158874034881592,
      "learning_rate": 7.822185970636217e-06,
      "loss": 0.1715,
      "num_input_tokens_seen": 2080928,
      "step": 960
    },
    {
      "epoch": 0.15742251223491027,
      "grad_norm": 0.08856307715177536,
      "learning_rate": 7.862969004893964e-06,
      "loss": 0.132,
      "num_input_tokens_seen": 2090784,
      "step": 965
    },
    {
      "epoch": 0.15823817292006526,
      "grad_norm": 2.6662380695343018,
      "learning_rate": 7.903752039151714e-06,
      "loss": 0.1575,
      "num_input_tokens_seen": 2100416,
      "step": 970
    },
    {
      "epoch": 0.15905383360522024,
      "grad_norm": 0.36138996481895447,
      "learning_rate": 7.944535073409461e-06,
      "loss": 0.0974,
      "num_input_tokens_seen": 2111424,
      "step": 975
    },
    {
      "epoch": 0.1598694942903752,
      "grad_norm": 4.972231864929199,
      "learning_rate": 7.98531810766721e-06,
      "loss": 0.178,
      "num_input_tokens_seen": 2122176,
      "step": 980
    },
    {
      "epoch": 0.16068515497553018,
      "grad_norm": 0.049096524715423584,
      "learning_rate": 8.026101141924958e-06,
      "loss": 0.1116,
      "num_input_tokens_seen": 2132928,
      "step": 985
    },
    {
      "epoch": 0.16150081566068517,
      "grad_norm": 6.180870532989502,
      "learning_rate": 8.066884176182708e-06,
      "loss": 0.1025,
      "num_input_tokens_seen": 2144480,
      "step": 990
    },
    {
      "epoch": 0.16231647634584012,
      "grad_norm": 9.751792907714844,
      "learning_rate": 8.107667210440457e-06,
      "loss": 0.172,
      "num_input_tokens_seen": 2154272,
      "step": 995
    },
    {
      "epoch": 0.1631321370309951,
      "grad_norm": 0.19507698714733124,
      "learning_rate": 8.148450244698205e-06,
      "loss": 0.008,
      "num_input_tokens_seen": 2163936,
      "step": 1000
    },
    {
      "epoch": 0.1639477977161501,
      "grad_norm": 0.43871402740478516,
      "learning_rate": 8.189233278955954e-06,
      "loss": 0.289,
      "num_input_tokens_seen": 2175392,
      "step": 1005
    },
    {
      "epoch": 0.16476345840130505,
      "grad_norm": 0.06107112392783165,
      "learning_rate": 8.230016313213703e-06,
      "loss": 0.2199,
      "num_input_tokens_seen": 2185120,
      "step": 1010
    },
    {
      "epoch": 0.16557911908646003,
      "grad_norm": 9.946898460388184,
      "learning_rate": 8.270799347471453e-06,
      "loss": 0.4255,
      "num_input_tokens_seen": 2196864,
      "step": 1015
    },
    {
      "epoch": 0.16639477977161501,
      "grad_norm": 1.8367321491241455,
      "learning_rate": 8.3115823817292e-06,
      "loss": 0.2934,
      "num_input_tokens_seen": 2207744,
      "step": 1020
    },
    {
      "epoch": 0.16721044045676997,
      "grad_norm": 8.737253189086914,
      "learning_rate": 8.35236541598695e-06,
      "loss": 0.3628,
      "num_input_tokens_seen": 2218528,
      "step": 1025
    },
    {
      "epoch": 0.16802610114192496,
      "grad_norm": 3.1235134601593018,
      "learning_rate": 8.393148450244699e-06,
      "loss": 0.0845,
      "num_input_tokens_seen": 2228320,
      "step": 1030
    },
    {
      "epoch": 0.16884176182707994,
      "grad_norm": 9.189127922058105,
      "learning_rate": 8.433931484502447e-06,
      "loss": 0.2927,
      "num_input_tokens_seen": 2238464,
      "step": 1035
    },
    {
      "epoch": 0.16965742251223492,
      "grad_norm": 0.46165475249290466,
      "learning_rate": 8.474714518760196e-06,
      "loss": 0.1265,
      "num_input_tokens_seen": 2249440,
      "step": 1040
    },
    {
      "epoch": 0.17047308319738988,
      "grad_norm": 0.327440083026886,
      "learning_rate": 8.515497553017945e-06,
      "loss": 0.2192,
      "num_input_tokens_seen": 2260576,
      "step": 1045
    },
    {
      "epoch": 0.17128874388254486,
      "grad_norm": 0.4591752886772156,
      "learning_rate": 8.556280587275693e-06,
      "loss": 0.1878,
      "num_input_tokens_seen": 2271904,
      "step": 1050
    },
    {
      "epoch": 0.17210440456769985,
      "grad_norm": 5.08633279800415,
      "learning_rate": 8.597063621533442e-06,
      "loss": 0.0593,
      "num_input_tokens_seen": 2281568,
      "step": 1055
    },
    {
      "epoch": 0.1729200652528548,
      "grad_norm": 2.9506537914276123,
      "learning_rate": 8.637846655791192e-06,
      "loss": 0.2285,
      "num_input_tokens_seen": 2292704,
      "step": 1060
    },
    {
      "epoch": 0.1737357259380098,
      "grad_norm": 0.40769487619400024,
      "learning_rate": 8.67862969004894e-06,
      "loss": 0.4531,
      "num_input_tokens_seen": 2303904,
      "step": 1065
    },
    {
      "epoch": 0.17455138662316477,
      "grad_norm": 10.724939346313477,
      "learning_rate": 8.719412724306688e-06,
      "loss": 0.1291,
      "num_input_tokens_seen": 2314336,
      "step": 1070
    },
    {
      "epoch": 0.17536704730831973,
      "grad_norm": 0.10260462760925293,
      "learning_rate": 8.760195758564438e-06,
      "loss": 0.0487,
      "num_input_tokens_seen": 2325024,
      "step": 1075
    },
    {
      "epoch": 0.1761827079934747,
      "grad_norm": 1.2693135738372803,
      "learning_rate": 8.800978792822187e-06,
      "loss": 0.3189,
      "num_input_tokens_seen": 2336224,
      "step": 1080
    },
    {
      "epoch": 0.1769983686786297,
      "grad_norm": 0.11829324811697006,
      "learning_rate": 8.841761827079935e-06,
      "loss": 0.0737,
      "num_input_tokens_seen": 2346880,
      "step": 1085
    },
    {
      "epoch": 0.17781402936378465,
      "grad_norm": 0.0629054456949234,
      "learning_rate": 8.882544861337684e-06,
      "loss": 0.1405,
      "num_input_tokens_seen": 2357632,
      "step": 1090
    },
    {
      "epoch": 0.17862969004893964,
      "grad_norm": 4.260016441345215,
      "learning_rate": 8.923327895595434e-06,
      "loss": 0.1017,
      "num_input_tokens_seen": 2370592,
      "step": 1095
    },
    {
      "epoch": 0.17944535073409462,
      "grad_norm": 7.259707450866699,
      "learning_rate": 8.964110929853181e-06,
      "loss": 0.4943,
      "num_input_tokens_seen": 2381024,
      "step": 1100
    },
    {
      "epoch": 0.1802610114192496,
      "grad_norm": 0.5672919154167175,
      "learning_rate": 9.00489396411093e-06,
      "loss": 0.0259,
      "num_input_tokens_seen": 2392160,
      "step": 1105
    },
    {
      "epoch": 0.18107667210440456,
      "grad_norm": 6.852181434631348,
      "learning_rate": 9.04567699836868e-06,
      "loss": 0.6329,
      "num_input_tokens_seen": 2403616,
      "step": 1110
    },
    {
      "epoch": 0.18189233278955955,
      "grad_norm": 5.7968573570251465,
      "learning_rate": 9.086460032626427e-06,
      "loss": 0.5964,
      "num_input_tokens_seen": 2416032,
      "step": 1115
    },
    {
      "epoch": 0.18270799347471453,
      "grad_norm": 0.06541386991739273,
      "learning_rate": 9.127243066884177e-06,
      "loss": 0.1879,
      "num_input_tokens_seen": 2426112,
      "step": 1120
    },
    {
      "epoch": 0.1835236541598695,
      "grad_norm": 6.544291973114014,
      "learning_rate": 9.168026101141926e-06,
      "loss": 0.2227,
      "num_input_tokens_seen": 2437344,
      "step": 1125
    },
    {
      "epoch": 0.18433931484502447,
      "grad_norm": 0.17347821593284607,
      "learning_rate": 9.208809135399674e-06,
      "loss": 0.2871,
      "num_input_tokens_seen": 2448672,
      "step": 1130
    },
    {
      "epoch": 0.18515497553017946,
      "grad_norm": 0.08097555488348007,
      "learning_rate": 9.249592169657423e-06,
      "loss": 0.1802,
      "num_input_tokens_seen": 2459552,
      "step": 1135
    },
    {
      "epoch": 0.1859706362153344,
      "grad_norm": 5.536982536315918,
      "learning_rate": 9.290375203915172e-06,
      "loss": 0.0671,
      "num_input_tokens_seen": 2469184,
      "step": 1140
    },
    {
      "epoch": 0.1867862969004894,
      "grad_norm": 2.1592328548431396,
      "learning_rate": 9.33115823817292e-06,
      "loss": 0.0956,
      "num_input_tokens_seen": 2481120,
      "step": 1145
    },
    {
      "epoch": 0.18760195758564438,
      "grad_norm": 7.730838775634766,
      "learning_rate": 9.37194127243067e-06,
      "loss": 0.1967,
      "num_input_tokens_seen": 2490976,
      "step": 1150
    },
    {
      "epoch": 0.18841761827079934,
      "grad_norm": 8.524580001831055,
      "learning_rate": 9.412724306688419e-06,
      "loss": 0.2055,
      "num_input_tokens_seen": 2501024,
      "step": 1155
    },
    {
      "epoch": 0.18923327895595432,
      "grad_norm": 8.385376930236816,
      "learning_rate": 9.453507340946168e-06,
      "loss": 0.4198,
      "num_input_tokens_seen": 2511264,
      "step": 1160
    },
    {
      "epoch": 0.1900489396411093,
      "grad_norm": 0.6890153288841248,
      "learning_rate": 9.494290375203916e-06,
      "loss": 0.1255,
      "num_input_tokens_seen": 2521920,
      "step": 1165
    },
    {
      "epoch": 0.19086460032626426,
      "grad_norm": 4.921489715576172,
      "learning_rate": 9.535073409461665e-06,
      "loss": 0.4664,
      "num_input_tokens_seen": 2533056,
      "step": 1170
    },
    {
      "epoch": 0.19168026101141925,
      "grad_norm": 8.706521034240723,
      "learning_rate": 9.575856443719414e-06,
      "loss": 0.5346,
      "num_input_tokens_seen": 2544320,
      "step": 1175
    },
    {
      "epoch": 0.19249592169657423,
      "grad_norm": 3.1115901470184326,
      "learning_rate": 9.616639477977162e-06,
      "loss": 0.1906,
      "num_input_tokens_seen": 2555168,
      "step": 1180
    },
    {
      "epoch": 0.1933115823817292,
      "grad_norm": 0.234554722905159,
      "learning_rate": 9.657422512234911e-06,
      "loss": 0.0873,
      "num_input_tokens_seen": 2566304,
      "step": 1185
    },
    {
      "epoch": 0.19412724306688417,
      "grad_norm": 0.4108467698097229,
      "learning_rate": 9.69820554649266e-06,
      "loss": 0.3393,
      "num_input_tokens_seen": 2575936,
      "step": 1190
    },
    {
      "epoch": 0.19494290375203915,
      "grad_norm": 9.332695007324219,
      "learning_rate": 9.738988580750408e-06,
      "loss": 0.2337,
      "num_input_tokens_seen": 2586528,
      "step": 1195
    },
    {
      "epoch": 0.19575856443719414,
      "grad_norm": 3.612908124923706,
      "learning_rate": 9.779771615008158e-06,
      "loss": 0.4979,
      "num_input_tokens_seen": 2596064,
      "step": 1200
    },
    {
      "epoch": 0.1965742251223491,
      "grad_norm": 0.4483790099620819,
      "learning_rate": 9.820554649265905e-06,
      "loss": 0.4875,
      "num_input_tokens_seen": 2606464,
      "step": 1205
    },
    {
      "epoch": 0.19738988580750408,
      "grad_norm": 3.2798922061920166,
      "learning_rate": 9.861337683523655e-06,
      "loss": 0.2635,
      "num_input_tokens_seen": 2617568,
      "step": 1210
    },
    {
      "epoch": 0.19820554649265906,
      "grad_norm": 0.25610870122909546,
      "learning_rate": 9.902120717781402e-06,
      "loss": 0.0868,
      "num_input_tokens_seen": 2628928,
      "step": 1215
    },
    {
      "epoch": 0.19902120717781402,
      "grad_norm": 7.029544830322266,
      "learning_rate": 9.942903752039152e-06,
      "loss": 0.2243,
      "num_input_tokens_seen": 2641344,
      "step": 1220
    },
    {
      "epoch": 0.199836867862969,
      "grad_norm": 14.858476638793945,
      "learning_rate": 9.983686786296901e-06,
      "loss": 0.214,
      "num_input_tokens_seen": 2651584,
      "step": 1225
    },
    {
      "epoch": 0.200652528548124,
      "grad_norm": 6.908661365509033,
      "learning_rate": 1.0024469820554649e-05,
      "loss": 0.2425,
      "num_input_tokens_seen": 2663360,
      "step": 1230
    },
    {
      "epoch": 0.20146818923327894,
      "grad_norm": 4.568848133087158,
      "learning_rate": 1.0065252854812398e-05,
      "loss": 0.256,
      "num_input_tokens_seen": 2675168,
      "step": 1235
    },
    {
      "epoch": 0.20228384991843393,
      "grad_norm": 9.485286712646484,
      "learning_rate": 1.0106035889070147e-05,
      "loss": 0.5909,
      "num_input_tokens_seen": 2685952,
      "step": 1240
    },
    {
      "epoch": 0.2030995106035889,
      "grad_norm": 6.773355960845947,
      "learning_rate": 1.0146818923327895e-05,
      "loss": 0.1391,
      "num_input_tokens_seen": 2695488,
      "step": 1245
    },
    {
      "epoch": 0.2039151712887439,
      "grad_norm": 5.654582977294922,
      "learning_rate": 1.0187601957585644e-05,
      "loss": 0.2623,
      "num_input_tokens_seen": 2706880,
      "step": 1250
    },
    {
      "epoch": 0.20473083197389885,
      "grad_norm": 5.590231418609619,
      "learning_rate": 1.0228384991843394e-05,
      "loss": 0.5785,
      "num_input_tokens_seen": 2717184,
      "step": 1255
    },
    {
      "epoch": 0.20554649265905384,
      "grad_norm": 0.3414728343486786,
      "learning_rate": 1.0269168026101141e-05,
      "loss": 0.1314,
      "num_input_tokens_seen": 2729184,
      "step": 1260
    },
    {
      "epoch": 0.20636215334420882,
      "grad_norm": 4.966456890106201,
      "learning_rate": 1.030995106035889e-05,
      "loss": 0.2237,
      "num_input_tokens_seen": 2740768,
      "step": 1265
    },
    {
      "epoch": 0.20717781402936378,
      "grad_norm": 3.5390045642852783,
      "learning_rate": 1.035073409461664e-05,
      "loss": 0.3007,
      "num_input_tokens_seen": 2751328,
      "step": 1270
    },
    {
      "epoch": 0.20799347471451876,
      "grad_norm": 0.3213955760002136,
      "learning_rate": 1.0391517128874388e-05,
      "loss": 0.3432,
      "num_input_tokens_seen": 2762464,
      "step": 1275
    },
    {
      "epoch": 0.20880913539967375,
      "grad_norm": 1.0880823135375977,
      "learning_rate": 1.0432300163132137e-05,
      "loss": 0.0912,
      "num_input_tokens_seen": 2773824,
      "step": 1280
    },
    {
      "epoch": 0.2096247960848287,
      "grad_norm": 5.550802230834961,
      "learning_rate": 1.0473083197389886e-05,
      "loss": 0.3858,
      "num_input_tokens_seen": 2784544,
      "step": 1285
    },
    {
      "epoch": 0.21044045676998369,
      "grad_norm": 0.036135949194431305,
      "learning_rate": 1.0513866231647634e-05,
      "loss": 0.1073,
      "num_input_tokens_seen": 2796032,
      "step": 1290
    },
    {
      "epoch": 0.21125611745513867,
      "grad_norm": 0.06197676435112953,
      "learning_rate": 1.0554649265905383e-05,
      "loss": 0.2201,
      "num_input_tokens_seen": 2808096,
      "step": 1295
    },
    {
      "epoch": 0.21207177814029363,
      "grad_norm": 6.072765350341797,
      "learning_rate": 1.0595432300163133e-05,
      "loss": 0.1024,
      "num_input_tokens_seen": 2818080,
      "step": 1300
    },
    {
      "epoch": 0.2128874388254486,
      "grad_norm": 1.302298665046692,
      "learning_rate": 1.0636215334420882e-05,
      "loss": 0.2044,
      "num_input_tokens_seen": 2829248,
      "step": 1305
    },
    {
      "epoch": 0.2137030995106036,
      "grad_norm": 0.19654478132724762,
      "learning_rate": 1.067699836867863e-05,
      "loss": 0.1302,
      "num_input_tokens_seen": 2840288,
      "step": 1310
    },
    {
      "epoch": 0.21451876019575855,
      "grad_norm": 5.693123817443848,
      "learning_rate": 1.0717781402936379e-05,
      "loss": 0.3089,
      "num_input_tokens_seen": 2849440,
      "step": 1315
    },
    {
      "epoch": 0.21533442088091354,
      "grad_norm": 0.02976483479142189,
      "learning_rate": 1.0758564437194128e-05,
      "loss": 0.1977,
      "num_input_tokens_seen": 2861056,
      "step": 1320
    },
    {
      "epoch": 0.21615008156606852,
      "grad_norm": 2.9492530822753906,
      "learning_rate": 1.0799347471451876e-05,
      "loss": 0.1258,
      "num_input_tokens_seen": 2872832,
      "step": 1325
    },
    {
      "epoch": 0.2169657422512235,
      "grad_norm": 8.189480781555176,
      "learning_rate": 1.0840130505709625e-05,
      "loss": 0.5312,
      "num_input_tokens_seen": 2883968,
      "step": 1330
    },
    {
      "epoch": 0.21778140293637846,
      "grad_norm": 0.1490117758512497,
      "learning_rate": 1.0880913539967375e-05,
      "loss": 0.1234,
      "num_input_tokens_seen": 2894880,
      "step": 1335
    },
    {
      "epoch": 0.21859706362153344,
      "grad_norm": 6.5888671875,
      "learning_rate": 1.0921696574225122e-05,
      "loss": 0.4302,
      "num_input_tokens_seen": 2905216,
      "step": 1340
    },
    {
      "epoch": 0.21941272430668843,
      "grad_norm": 8.062369346618652,
      "learning_rate": 1.0962479608482872e-05,
      "loss": 0.1458,
      "num_input_tokens_seen": 2917152,
      "step": 1345
    },
    {
      "epoch": 0.22022838499184338,
      "grad_norm": 5.160257339477539,
      "learning_rate": 1.1003262642740621e-05,
      "loss": 0.1481,
      "num_input_tokens_seen": 2928448,
      "step": 1350
    },
    {
      "epoch": 0.22104404567699837,
      "grad_norm": 2.4276535511016846,
      "learning_rate": 1.1044045676998369e-05,
      "loss": 0.4326,
      "num_input_tokens_seen": 2938336,
      "step": 1355
    },
    {
      "epoch": 0.22185970636215335,
      "grad_norm": 7.487486362457275,
      "learning_rate": 1.1084828711256118e-05,
      "loss": 0.1281,
      "num_input_tokens_seen": 2950496,
      "step": 1360
    },
    {
      "epoch": 0.2226753670473083,
      "grad_norm": 11.373209953308105,
      "learning_rate": 1.1125611745513867e-05,
      "loss": 0.2648,
      "num_input_tokens_seen": 2961024,
      "step": 1365
    },
    {
      "epoch": 0.2234910277324633,
      "grad_norm": 7.982120037078857,
      "learning_rate": 1.1166394779771617e-05,
      "loss": 0.1207,
      "num_input_tokens_seen": 2971744,
      "step": 1370
    },
    {
      "epoch": 0.22430668841761828,
      "grad_norm": 5.5169758796691895,
      "learning_rate": 1.1207177814029364e-05,
      "loss": 0.1653,
      "num_input_tokens_seen": 2984192,
      "step": 1375
    },
    {
      "epoch": 0.22512234910277323,
      "grad_norm": 6.303249835968018,
      "learning_rate": 1.1247960848287114e-05,
      "loss": 0.2739,
      "num_input_tokens_seen": 2994816,
      "step": 1380
    },
    {
      "epoch": 0.22593800978792822,
      "grad_norm": 0.6037583947181702,
      "learning_rate": 1.1288743882544863e-05,
      "loss": 0.178,
      "num_input_tokens_seen": 3005568,
      "step": 1385
    },
    {
      "epoch": 0.2267536704730832,
      "grad_norm": 0.6912223696708679,
      "learning_rate": 1.132952691680261e-05,
      "loss": 0.3771,
      "num_input_tokens_seen": 3015680,
      "step": 1390
    },
    {
      "epoch": 0.2275693311582382,
      "grad_norm": 1.4500184059143066,
      "learning_rate": 1.137030995106036e-05,
      "loss": 0.1616,
      "num_input_tokens_seen": 3026752,
      "step": 1395
    },
    {
      "epoch": 0.22838499184339314,
      "grad_norm": 0.18793430924415588,
      "learning_rate": 1.141109298531811e-05,
      "loss": 0.1681,
      "num_input_tokens_seen": 3036704,
      "step": 1400
    },
    {
      "epoch": 0.22920065252854813,
      "grad_norm": 5.307665824890137,
      "learning_rate": 1.1451876019575857e-05,
      "loss": 0.3579,
      "num_input_tokens_seen": 3048960,
      "step": 1405
    },
    {
      "epoch": 0.2300163132137031,
      "grad_norm": 6.192534446716309,
      "learning_rate": 1.1492659053833606e-05,
      "loss": 0.4311,
      "num_input_tokens_seen": 3059456,
      "step": 1410
    },
    {
      "epoch": 0.23083197389885807,
      "grad_norm": 2.5860745906829834,
      "learning_rate": 1.1533442088091356e-05,
      "loss": 0.141,
      "num_input_tokens_seen": 3070112,
      "step": 1415
    },
    {
      "epoch": 0.23164763458401305,
      "grad_norm": 0.14588327705860138,
      "learning_rate": 1.1574225122349103e-05,
      "loss": 0.1593,
      "num_input_tokens_seen": 3080224,
      "step": 1420
    },
    {
      "epoch": 0.23246329526916804,
      "grad_norm": 4.942193984985352,
      "learning_rate": 1.1615008156606853e-05,
      "loss": 0.461,
      "num_input_tokens_seen": 3090304,
      "step": 1425
    },
    {
      "epoch": 0.233278955954323,
      "grad_norm": 7.982557773590088,
      "learning_rate": 1.1655791190864602e-05,
      "loss": 0.283,
      "num_input_tokens_seen": 3099232,
      "step": 1430
    },
    {
      "epoch": 0.23409461663947798,
      "grad_norm": 6.3187947273254395,
      "learning_rate": 1.169657422512235e-05,
      "loss": 0.4601,
      "num_input_tokens_seen": 3109696,
      "step": 1435
    },
    {
      "epoch": 0.23491027732463296,
      "grad_norm": 6.894016265869141,
      "learning_rate": 1.1737357259380099e-05,
      "loss": 0.1709,
      "num_input_tokens_seen": 3120160,
      "step": 1440
    },
    {
      "epoch": 0.23572593800978792,
      "grad_norm": 4.156856060028076,
      "learning_rate": 1.1778140293637847e-05,
      "loss": 0.1095,
      "num_input_tokens_seen": 3130784,
      "step": 1445
    },
    {
      "epoch": 0.2365415986949429,
      "grad_norm": 2.375866413116455,
      "learning_rate": 1.1818923327895596e-05,
      "loss": 0.3756,
      "num_input_tokens_seen": 3141536,
      "step": 1450
    },
    {
      "epoch": 0.23735725938009788,
      "grad_norm": 0.34968551993370056,
      "learning_rate": 1.1859706362153344e-05,
      "loss": 0.2573,
      "num_input_tokens_seen": 3152160,
      "step": 1455
    },
    {
      "epoch": 0.23817292006525284,
      "grad_norm": 1.3204271793365479,
      "learning_rate": 1.1900489396411093e-05,
      "loss": 0.116,
      "num_input_tokens_seen": 3162560,
      "step": 1460
    },
    {
      "epoch": 0.23898858075040783,
      "grad_norm": 1.7406924962997437,
      "learning_rate": 1.1941272430668842e-05,
      "loss": 0.1799,
      "num_input_tokens_seen": 3175360,
      "step": 1465
    },
    {
      "epoch": 0.2398042414355628,
      "grad_norm": 3.952451705932617,
      "learning_rate": 1.198205546492659e-05,
      "loss": 0.1108,
      "num_input_tokens_seen": 3187936,
      "step": 1470
    },
    {
      "epoch": 0.2406199021207178,
      "grad_norm": 8.951556205749512,
      "learning_rate": 1.202283849918434e-05,
      "loss": 0.508,
      "num_input_tokens_seen": 3199424,
      "step": 1475
    },
    {
      "epoch": 0.24143556280587275,
      "grad_norm": 0.3572978079319,
      "learning_rate": 1.2063621533442089e-05,
      "loss": 0.3565,
      "num_input_tokens_seen": 3210272,
      "step": 1480
    },
    {
      "epoch": 0.24225122349102773,
      "grad_norm": 4.634958744049072,
      "learning_rate": 1.2104404567699836e-05,
      "loss": 0.1031,
      "num_input_tokens_seen": 3219680,
      "step": 1485
    },
    {
      "epoch": 0.24306688417618272,
      "grad_norm": 0.9866248369216919,
      "learning_rate": 1.2145187601957586e-05,
      "loss": 0.149,
      "num_input_tokens_seen": 3229376,
      "step": 1490
    },
    {
      "epoch": 0.24388254486133767,
      "grad_norm": 1.492843747138977,
      "learning_rate": 1.2185970636215335e-05,
      "loss": 0.1961,
      "num_input_tokens_seen": 3240896,
      "step": 1495
    },
    {
      "epoch": 0.24469820554649266,
      "grad_norm": 4.87559700012207,
      "learning_rate": 1.2226753670473083e-05,
      "loss": 0.4337,
      "num_input_tokens_seen": 3251808,
      "step": 1500
    },
    {
      "epoch": 0.24551386623164764,
      "grad_norm": 1.7679897546768188,
      "learning_rate": 1.2267536704730832e-05,
      "loss": 0.1422,
      "num_input_tokens_seen": 3262688,
      "step": 1505
    },
    {
      "epoch": 0.2463295269168026,
      "grad_norm": 3.7211174964904785,
      "learning_rate": 1.2308319738988581e-05,
      "loss": 0.0588,
      "num_input_tokens_seen": 3273344,
      "step": 1510
    },
    {
      "epoch": 0.24714518760195758,
      "grad_norm": 5.718593120574951,
      "learning_rate": 1.234910277324633e-05,
      "loss": 0.2599,
      "num_input_tokens_seen": 3283520,
      "step": 1515
    },
    {
      "epoch": 0.24796084828711257,
      "grad_norm": 0.31124433875083923,
      "learning_rate": 1.2389885807504078e-05,
      "loss": 0.3094,
      "num_input_tokens_seen": 3294816,
      "step": 1520
    },
    {
      "epoch": 0.24877650897226752,
      "grad_norm": 3.2396998405456543,
      "learning_rate": 1.2430668841761828e-05,
      "loss": 0.3018,
      "num_input_tokens_seen": 3305152,
      "step": 1525
    },
    {
      "epoch": 0.2495921696574225,
      "grad_norm": 0.19118860363960266,
      "learning_rate": 1.2471451876019577e-05,
      "loss": 0.2159,
      "num_input_tokens_seen": 3315264,
      "step": 1530
    },
    {
      "epoch": 0.25040783034257746,
      "grad_norm": 3.2154130935668945,
      "learning_rate": 1.2512234910277326e-05,
      "loss": 0.304,
      "num_input_tokens_seen": 3326208,
      "step": 1535
    },
    {
      "epoch": 0.25122349102773245,
      "grad_norm": 5.489169597625732,
      "learning_rate": 1.2553017944535072e-05,
      "loss": 0.3482,
      "num_input_tokens_seen": 3337472,
      "step": 1540
    },
    {
      "epoch": 0.25203915171288743,
      "grad_norm": 1.1349680423736572,
      "learning_rate": 1.2593800978792822e-05,
      "loss": 0.1839,
      "num_input_tokens_seen": 3348576,
      "step": 1545
    },
    {
      "epoch": 0.2528548123980424,
      "grad_norm": 9.128485679626465,
      "learning_rate": 1.2634584013050571e-05,
      "loss": 0.3182,
      "num_input_tokens_seen": 3359200,
      "step": 1550
    },
    {
      "epoch": 0.2536704730831974,
      "grad_norm": 0.4159963130950928,
      "learning_rate": 1.267536704730832e-05,
      "loss": 0.1003,
      "num_input_tokens_seen": 3369696,
      "step": 1555
    },
    {
      "epoch": 0.2544861337683524,
      "grad_norm": 5.541415214538574,
      "learning_rate": 1.271615008156607e-05,
      "loss": 0.2741,
      "num_input_tokens_seen": 3381088,
      "step": 1560
    },
    {
      "epoch": 0.2553017944535073,
      "grad_norm": 5.9848432540893555,
      "learning_rate": 1.2756933115823819e-05,
      "loss": 0.1997,
      "num_input_tokens_seen": 3392704,
      "step": 1565
    },
    {
      "epoch": 0.2561174551386623,
      "grad_norm": 1.0845085382461548,
      "learning_rate": 1.2797716150081568e-05,
      "loss": 0.447,
      "num_input_tokens_seen": 3403136,
      "step": 1570
    },
    {
      "epoch": 0.2569331158238173,
      "grad_norm": 0.04691677913069725,
      "learning_rate": 1.2838499184339314e-05,
      "loss": 0.1089,
      "num_input_tokens_seen": 3411968,
      "step": 1575
    },
    {
      "epoch": 0.25774877650897227,
      "grad_norm": 0.24104487895965576,
      "learning_rate": 1.2879282218597064e-05,
      "loss": 0.1357,
      "num_input_tokens_seen": 3421216,
      "step": 1580
    },
    {
      "epoch": 0.25856443719412725,
      "grad_norm": 6.295036792755127,
      "learning_rate": 1.2920065252854813e-05,
      "loss": 0.3222,
      "num_input_tokens_seen": 3433120,
      "step": 1585
    },
    {
      "epoch": 0.25938009787928223,
      "grad_norm": 4.436316967010498,
      "learning_rate": 1.2960848287112562e-05,
      "loss": 0.2229,
      "num_input_tokens_seen": 3444960,
      "step": 1590
    },
    {
      "epoch": 0.2601957585644372,
      "grad_norm": 1.5245418548583984,
      "learning_rate": 1.3001631321370312e-05,
      "loss": 0.088,
      "num_input_tokens_seen": 3456512,
      "step": 1595
    },
    {
      "epoch": 0.26101141924959215,
      "grad_norm": 3.516801595687866,
      "learning_rate": 1.3042414355628061e-05,
      "loss": 0.1196,
      "num_input_tokens_seen": 3466880,
      "step": 1600
    },
    {
      "epoch": 0.26182707993474713,
      "grad_norm": 0.6709426641464233,
      "learning_rate": 1.3083197389885807e-05,
      "loss": 0.1292,
      "num_input_tokens_seen": 3478528,
      "step": 1605
    },
    {
      "epoch": 0.2626427406199021,
      "grad_norm": 0.7419503331184387,
      "learning_rate": 1.3123980424143556e-05,
      "loss": 0.184,
      "num_input_tokens_seen": 3490240,
      "step": 1610
    },
    {
      "epoch": 0.2634584013050571,
      "grad_norm": 4.619818210601807,
      "learning_rate": 1.3164763458401306e-05,
      "loss": 0.2783,
      "num_input_tokens_seen": 3501152,
      "step": 1615
    },
    {
      "epoch": 0.2642740619902121,
      "grad_norm": 0.15130358934402466,
      "learning_rate": 1.3205546492659055e-05,
      "loss": 0.2706,
      "num_input_tokens_seen": 3511360,
      "step": 1620
    },
    {
      "epoch": 0.26508972267536707,
      "grad_norm": 3.8601553440093994,
      "learning_rate": 1.3246329526916804e-05,
      "loss": 0.1139,
      "num_input_tokens_seen": 3522880,
      "step": 1625
    },
    {
      "epoch": 0.265905383360522,
      "grad_norm": 0.13603436946868896,
      "learning_rate": 1.3287112561174554e-05,
      "loss": 0.1081,
      "num_input_tokens_seen": 3534720,
      "step": 1630
    },
    {
      "epoch": 0.266721044045677,
      "grad_norm": 2.212265729904175,
      "learning_rate": 1.3327895595432303e-05,
      "loss": 0.1012,
      "num_input_tokens_seen": 3544128,
      "step": 1635
    },
    {
      "epoch": 0.26753670473083196,
      "grad_norm": 4.783210754394531,
      "learning_rate": 1.3368678629690049e-05,
      "loss": 0.164,
      "num_input_tokens_seen": 3555552,
      "step": 1640
    },
    {
      "epoch": 0.26835236541598695,
      "grad_norm": 0.38099732995033264,
      "learning_rate": 1.3409461663947798e-05,
      "loss": 0.106,
      "num_input_tokens_seen": 3565824,
      "step": 1645
    },
    {
      "epoch": 0.26916802610114193,
      "grad_norm": 4.968090057373047,
      "learning_rate": 1.3450244698205548e-05,
      "loss": 0.1144,
      "num_input_tokens_seen": 3576864,
      "step": 1650
    },
    {
      "epoch": 0.2699836867862969,
      "grad_norm": 6.693702220916748,
      "learning_rate": 1.3491027732463297e-05,
      "loss": 0.109,
      "num_input_tokens_seen": 3588032,
      "step": 1655
    },
    {
      "epoch": 0.2707993474714519,
      "grad_norm": 6.759777069091797,
      "learning_rate": 1.3531810766721044e-05,
      "loss": 0.0746,
      "num_input_tokens_seen": 3599488,
      "step": 1660
    },
    {
      "epoch": 0.27161500815660683,
      "grad_norm": 0.8844873905181885,
      "learning_rate": 1.3572593800978794e-05,
      "loss": 0.0663,
      "num_input_tokens_seen": 3610368,
      "step": 1665
    },
    {
      "epoch": 0.2724306688417618,
      "grad_norm": 6.606276988983154,
      "learning_rate": 1.3613376835236541e-05,
      "loss": 0.2757,
      "num_input_tokens_seen": 3620128,
      "step": 1670
    },
    {
      "epoch": 0.2732463295269168,
      "grad_norm": 2.6373190879821777,
      "learning_rate": 1.365415986949429e-05,
      "loss": 0.1603,
      "num_input_tokens_seen": 3631872,
      "step": 1675
    },
    {
      "epoch": 0.2740619902120718,
      "grad_norm": 12.086174011230469,
      "learning_rate": 1.369494290375204e-05,
      "loss": 0.3898,
      "num_input_tokens_seen": 3643296,
      "step": 1680
    },
    {
      "epoch": 0.27487765089722677,
      "grad_norm": 2.731724500656128,
      "learning_rate": 1.3735725938009788e-05,
      "loss": 0.2679,
      "num_input_tokens_seen": 3654592,
      "step": 1685
    },
    {
      "epoch": 0.27569331158238175,
      "grad_norm": 10.402414321899414,
      "learning_rate": 1.3776508972267537e-05,
      "loss": 0.5287,
      "num_input_tokens_seen": 3665312,
      "step": 1690
    },
    {
      "epoch": 0.2765089722675367,
      "grad_norm": 5.5846428871154785,
      "learning_rate": 1.3817292006525286e-05,
      "loss": 0.1602,
      "num_input_tokens_seen": 3677024,
      "step": 1695
    },
    {
      "epoch": 0.27732463295269166,
      "grad_norm": 1.3808804750442505,
      "learning_rate": 1.3858075040783036e-05,
      "loss": 0.371,
      "num_input_tokens_seen": 3687904,
      "step": 1700
    },
    {
      "epoch": 0.27814029363784665,
      "grad_norm": 2.9026827812194824,
      "learning_rate": 1.3898858075040783e-05,
      "loss": 0.3399,
      "num_input_tokens_seen": 3698848,
      "step": 1705
    },
    {
      "epoch": 0.27895595432300163,
      "grad_norm": 6.748539447784424,
      "learning_rate": 1.3939641109298531e-05,
      "loss": 0.1927,
      "num_input_tokens_seen": 3709856,
      "step": 1710
    },
    {
      "epoch": 0.2797716150081566,
      "grad_norm": 0.5773642063140869,
      "learning_rate": 1.398042414355628e-05,
      "loss": 0.1874,
      "num_input_tokens_seen": 3720736,
      "step": 1715
    },
    {
      "epoch": 0.2805872756933116,
      "grad_norm": 3.2085487842559814,
      "learning_rate": 1.402120717781403e-05,
      "loss": 0.3319,
      "num_input_tokens_seen": 3731264,
      "step": 1720
    },
    {
      "epoch": 0.2814029363784666,
      "grad_norm": 5.882781028747559,
      "learning_rate": 1.4061990212071779e-05,
      "loss": 0.2759,
      "num_input_tokens_seen": 3743072,
      "step": 1725
    },
    {
      "epoch": 0.2822185970636215,
      "grad_norm": 2.6723382472991943,
      "learning_rate": 1.4102773246329528e-05,
      "loss": 0.1888,
      "num_input_tokens_seen": 3754656,
      "step": 1730
    },
    {
      "epoch": 0.2830342577487765,
      "grad_norm": 4.209661960601807,
      "learning_rate": 1.4143556280587274e-05,
      "loss": 0.1748,
      "num_input_tokens_seen": 3765248,
      "step": 1735
    },
    {
      "epoch": 0.2838499184339315,
      "grad_norm": 8.234864234924316,
      "learning_rate": 1.4184339314845024e-05,
      "loss": 0.4846,
      "num_input_tokens_seen": 3774464,
      "step": 1740
    },
    {
      "epoch": 0.28466557911908646,
      "grad_norm": 7.3376054763793945,
      "learning_rate": 1.4225122349102773e-05,
      "loss": 0.3614,
      "num_input_tokens_seen": 3784608,
      "step": 1745
    },
    {
      "epoch": 0.28548123980424145,
      "grad_norm": 2.9826624393463135,
      "learning_rate": 1.4265905383360522e-05,
      "loss": 0.2459,
      "num_input_tokens_seen": 3796672,
      "step": 1750
    },
    {
      "epoch": 0.28629690048939643,
      "grad_norm": 1.7873778343200684,
      "learning_rate": 1.4306688417618272e-05,
      "loss": 0.1198,
      "num_input_tokens_seen": 3807712,
      "step": 1755
    },
    {
      "epoch": 0.28711256117455136,
      "grad_norm": 3.915682315826416,
      "learning_rate": 1.4347471451876021e-05,
      "loss": 0.1048,
      "num_input_tokens_seen": 3817984,
      "step": 1760
    },
    {
      "epoch": 0.28792822185970635,
      "grad_norm": 2.6385345458984375,
      "learning_rate": 1.438825448613377e-05,
      "loss": 0.0767,
      "num_input_tokens_seen": 3829152,
      "step": 1765
    },
    {
      "epoch": 0.28874388254486133,
      "grad_norm": 6.35852575302124,
      "learning_rate": 1.4429037520391516e-05,
      "loss": 0.4122,
      "num_input_tokens_seen": 3840352,
      "step": 1770
    },
    {
      "epoch": 0.2895595432300163,
      "grad_norm": 0.30161672830581665,
      "learning_rate": 1.4469820554649266e-05,
      "loss": 0.2518,
      "num_input_tokens_seen": 3851456,
      "step": 1775
    },
    {
      "epoch": 0.2903752039151713,
      "grad_norm": 1.889992594718933,
      "learning_rate": 1.4510603588907015e-05,
      "loss": 0.0475,
      "num_input_tokens_seen": 3863040,
      "step": 1780
    },
    {
      "epoch": 0.2911908646003263,
      "grad_norm": 2.7470626831054688,
      "learning_rate": 1.4551386623164764e-05,
      "loss": 0.0746,
      "num_input_tokens_seen": 3874176,
      "step": 1785
    },
    {
      "epoch": 0.29200652528548127,
      "grad_norm": 0.05625578388571739,
      "learning_rate": 1.4592169657422514e-05,
      "loss": 0.2891,
      "num_input_tokens_seen": 3884736,
      "step": 1790
    },
    {
      "epoch": 0.2928221859706362,
      "grad_norm": 5.239809513092041,
      "learning_rate": 1.4632952691680263e-05,
      "loss": 0.2012,
      "num_input_tokens_seen": 3895584,
      "step": 1795
    },
    {
      "epoch": 0.2936378466557912,
      "grad_norm": 0.3846181333065033,
      "learning_rate": 1.4673735725938009e-05,
      "loss": 0.1749,
      "num_input_tokens_seen": 3907584,
      "step": 1800
    },
    {
      "epoch": 0.29445350734094616,
      "grad_norm": 2.8441622257232666,
      "learning_rate": 1.4714518760195758e-05,
      "loss": 0.1626,
      "num_input_tokens_seen": 3919424,
      "step": 1805
    },
    {
      "epoch": 0.29526916802610115,
      "grad_norm": 4.441000461578369,
      "learning_rate": 1.4755301794453508e-05,
      "loss": 0.1205,
      "num_input_tokens_seen": 3929152,
      "step": 1810
    },
    {
      "epoch": 0.29608482871125613,
      "grad_norm": 3.806591749191284,
      "learning_rate": 1.4796084828711257e-05,
      "loss": 0.1792,
      "num_input_tokens_seen": 3940928,
      "step": 1815
    },
    {
      "epoch": 0.2969004893964111,
      "grad_norm": 4.7424774169921875,
      "learning_rate": 1.4836867862969006e-05,
      "loss": 0.1372,
      "num_input_tokens_seen": 3949568,
      "step": 1820
    },
    {
      "epoch": 0.29771615008156604,
      "grad_norm": 1.578133225440979,
      "learning_rate": 1.4877650897226756e-05,
      "loss": 0.0935,
      "num_input_tokens_seen": 3961088,
      "step": 1825
    },
    {
      "epoch": 0.29853181076672103,
      "grad_norm": 0.3075026571750641,
      "learning_rate": 1.4918433931484505e-05,
      "loss": 0.3372,
      "num_input_tokens_seen": 3973088,
      "step": 1830
    },
    {
      "epoch": 0.299347471451876,
      "grad_norm": 0.20886602997779846,
      "learning_rate": 1.4959216965742251e-05,
      "loss": 0.1815,
      "num_input_tokens_seen": 3982976,
      "step": 1835
    },
    {
      "epoch": 0.300163132137031,
      "grad_norm": 5.134726524353027,
      "learning_rate": 1.5e-05,
      "loss": 0.3164,
      "num_input_tokens_seen": 3993440,
      "step": 1840
    },
    {
      "epoch": 0.300978792822186,
      "grad_norm": 0.7109973430633545,
      "learning_rate": 1.504078303425775e-05,
      "loss": 0.0632,
      "num_input_tokens_seen": 4002720,
      "step": 1845
    },
    {
      "epoch": 0.30179445350734097,
      "grad_norm": 2.32041072845459,
      "learning_rate": 1.5081566068515499e-05,
      "loss": 0.4135,
      "num_input_tokens_seen": 4012896,
      "step": 1850
    },
    {
      "epoch": 0.30261011419249595,
      "grad_norm": 2.2132656574249268,
      "learning_rate": 1.5122349102773248e-05,
      "loss": 0.1208,
      "num_input_tokens_seen": 4023680,
      "step": 1855
    },
    {
      "epoch": 0.3034257748776509,
      "grad_norm": 0.8771421909332275,
      "learning_rate": 1.5163132137030998e-05,
      "loss": 0.0557,
      "num_input_tokens_seen": 4033760,
      "step": 1860
    },
    {
      "epoch": 0.30424143556280586,
      "grad_norm": 5.692502975463867,
      "learning_rate": 1.5203915171288744e-05,
      "loss": 0.1944,
      "num_input_tokens_seen": 4044192,
      "step": 1865
    },
    {
      "epoch": 0.30505709624796085,
      "grad_norm": 3.3692212104797363,
      "learning_rate": 1.5244698205546493e-05,
      "loss": 0.2023,
      "num_input_tokens_seen": 4055520,
      "step": 1870
    },
    {
      "epoch": 0.30587275693311583,
      "grad_norm": 0.5607348084449768,
      "learning_rate": 1.5285481239804242e-05,
      "loss": 0.1193,
      "num_input_tokens_seen": 4067264,
      "step": 1875
    },
    {
      "epoch": 0.3066884176182708,
      "grad_norm": 5.53071403503418,
      "learning_rate": 1.532626427406199e-05,
      "loss": 0.2618,
      "num_input_tokens_seen": 4078016,
      "step": 1880
    },
    {
      "epoch": 0.3075040783034258,
      "grad_norm": 0.22230537235736847,
      "learning_rate": 1.536704730831974e-05,
      "loss": 0.2164,
      "num_input_tokens_seen": 4088352,
      "step": 1885
    },
    {
      "epoch": 0.3083197389885807,
      "grad_norm": 6.9999566078186035,
      "learning_rate": 1.540783034257749e-05,
      "loss": 0.1746,
      "num_input_tokens_seen": 4100416,
      "step": 1890
    },
    {
      "epoch": 0.3091353996737357,
      "grad_norm": 3.544222354888916,
      "learning_rate": 1.5448613376835236e-05,
      "loss": 0.2752,
      "num_input_tokens_seen": 4110080,
      "step": 1895
    },
    {
      "epoch": 0.3099510603588907,
      "grad_norm": 5.308750629425049,
      "learning_rate": 1.5489396411092984e-05,
      "loss": 0.4527,
      "num_input_tokens_seen": 4120736,
      "step": 1900
    },
    {
      "epoch": 0.3107667210440457,
      "grad_norm": 0.9366431832313538,
      "learning_rate": 1.5530179445350735e-05,
      "loss": 0.16,
      "num_input_tokens_seen": 4132736,
      "step": 1905
    },
    {
      "epoch": 0.31158238172920066,
      "grad_norm": 6.9811224937438965,
      "learning_rate": 1.5570962479608483e-05,
      "loss": 0.2702,
      "num_input_tokens_seen": 4143808,
      "step": 1910
    },
    {
      "epoch": 0.31239804241435565,
      "grad_norm": 4.50189733505249,
      "learning_rate": 1.5611745513866234e-05,
      "loss": 0.2213,
      "num_input_tokens_seen": 4154976,
      "step": 1915
    },
    {
      "epoch": 0.3132137030995106,
      "grad_norm": 2.1178994178771973,
      "learning_rate": 1.565252854812398e-05,
      "loss": 0.3245,
      "num_input_tokens_seen": 4165504,
      "step": 1920
    },
    {
      "epoch": 0.31402936378466556,
      "grad_norm": 9.411482810974121,
      "learning_rate": 1.5693311582381732e-05,
      "loss": 0.2616,
      "num_input_tokens_seen": 4177184,
      "step": 1925
    },
    {
      "epoch": 0.31484502446982054,
      "grad_norm": 6.287456512451172,
      "learning_rate": 1.5734094616639477e-05,
      "loss": 0.1775,
      "num_input_tokens_seen": 4189088,
      "step": 1930
    },
    {
      "epoch": 0.31566068515497553,
      "grad_norm": 0.42288240790367126,
      "learning_rate": 1.5774877650897228e-05,
      "loss": 0.0773,
      "num_input_tokens_seen": 4200064,
      "step": 1935
    },
    {
      "epoch": 0.3164763458401305,
      "grad_norm": 0.4060116708278656,
      "learning_rate": 1.5815660685154975e-05,
      "loss": 0.2332,
      "num_input_tokens_seen": 4211936,
      "step": 1940
    },
    {
      "epoch": 0.3172920065252855,
      "grad_norm": 1.5796598196029663,
      "learning_rate": 1.5856443719412726e-05,
      "loss": 0.1302,
      "num_input_tokens_seen": 4221952,
      "step": 1945
    },
    {
      "epoch": 0.3181076672104405,
      "grad_norm": 0.36358773708343506,
      "learning_rate": 1.5897226753670474e-05,
      "loss": 0.3447,
      "num_input_tokens_seen": 4233024,
      "step": 1950
    },
    {
      "epoch": 0.3189233278955954,
      "grad_norm": 2.5228257179260254,
      "learning_rate": 1.5938009787928225e-05,
      "loss": 0.0938,
      "num_input_tokens_seen": 4244480,
      "step": 1955
    },
    {
      "epoch": 0.3197389885807504,
      "grad_norm": 0.09291154146194458,
      "learning_rate": 1.597879282218597e-05,
      "loss": 0.1656,
      "num_input_tokens_seen": 4255904,
      "step": 1960
    },
    {
      "epoch": 0.3205546492659054,
      "grad_norm": 3.384127616882324,
      "learning_rate": 1.601957585644372e-05,
      "loss": 0.4983,
      "num_input_tokens_seen": 4265792,
      "step": 1965
    },
    {
      "epoch": 0.32137030995106036,
      "grad_norm": 0.04204917326569557,
      "learning_rate": 1.6060358890701468e-05,
      "loss": 0.1385,
      "num_input_tokens_seen": 4276160,
      "step": 1970
    },
    {
      "epoch": 0.32218597063621535,
      "grad_norm": 4.716171741485596,
      "learning_rate": 1.610114192495922e-05,
      "loss": 0.1854,
      "num_input_tokens_seen": 4286624,
      "step": 1975
    },
    {
      "epoch": 0.32300163132137033,
      "grad_norm": 1.6673164367675781,
      "learning_rate": 1.6141924959216967e-05,
      "loss": 0.0894,
      "num_input_tokens_seen": 4297600,
      "step": 1980
    },
    {
      "epoch": 0.32381729200652526,
      "grad_norm": 7.22198486328125,
      "learning_rate": 1.6182707993474718e-05,
      "loss": 0.2282,
      "num_input_tokens_seen": 4308352,
      "step": 1985
    },
    {
      "epoch": 0.32463295269168024,
      "grad_norm": 5.862957000732422,
      "learning_rate": 1.6223491027732465e-05,
      "loss": 0.2771,
      "num_input_tokens_seen": 4319360,
      "step": 1990
    },
    {
      "epoch": 0.3254486133768352,
      "grad_norm": 3.0516316890716553,
      "learning_rate": 1.6264274061990213e-05,
      "loss": 0.0749,
      "num_input_tokens_seen": 4331552,
      "step": 1995
    },
    {
      "epoch": 0.3262642740619902,
      "grad_norm": 0.3817313015460968,
      "learning_rate": 1.630505709624796e-05,
      "loss": 0.3125,
      "num_input_tokens_seen": 4343744,
      "step": 2000
    },
    {
      "epoch": 0.3270799347471452,
      "grad_norm": 0.12780296802520752,
      "learning_rate": 1.634584013050571e-05,
      "loss": 0.2445,
      "num_input_tokens_seen": 4354304,
      "step": 2005
    },
    {
      "epoch": 0.3278955954323002,
      "grad_norm": 4.317539691925049,
      "learning_rate": 1.638662316476346e-05,
      "loss": 0.1835,
      "num_input_tokens_seen": 4365696,
      "step": 2010
    },
    {
      "epoch": 0.32871125611745516,
      "grad_norm": 15.590194702148438,
      "learning_rate": 1.6427406199021207e-05,
      "loss": 0.3673,
      "num_input_tokens_seen": 4376736,
      "step": 2015
    },
    {
      "epoch": 0.3295269168026101,
      "grad_norm": 0.751114547252655,
      "learning_rate": 1.6468189233278958e-05,
      "loss": 0.2193,
      "num_input_tokens_seen": 4388064,
      "step": 2020
    },
    {
      "epoch": 0.3303425774877651,
      "grad_norm": 9.185699462890625,
      "learning_rate": 1.6508972267536706e-05,
      "loss": 0.2041,
      "num_input_tokens_seen": 4398912,
      "step": 2025
    },
    {
      "epoch": 0.33115823817292006,
      "grad_norm": 7.339178085327148,
      "learning_rate": 1.6549755301794453e-05,
      "loss": 0.2251,
      "num_input_tokens_seen": 4409856,
      "step": 2030
    },
    {
      "epoch": 0.33197389885807504,
      "grad_norm": 2.1822028160095215,
      "learning_rate": 1.6590538336052204e-05,
      "loss": 0.1465,
      "num_input_tokens_seen": 4419552,
      "step": 2035
    },
    {
      "epoch": 0.33278955954323003,
      "grad_norm": 0.5807727575302124,
      "learning_rate": 1.6631321370309952e-05,
      "loss": 0.156,
      "num_input_tokens_seen": 4430080,
      "step": 2040
    },
    {
      "epoch": 0.333605220228385,
      "grad_norm": 6.257948875427246,
      "learning_rate": 1.66721044045677e-05,
      "loss": 0.2496,
      "num_input_tokens_seen": 4441568,
      "step": 2045
    },
    {
      "epoch": 0.33442088091353994,
      "grad_norm": 6.7260050773620605,
      "learning_rate": 1.671288743882545e-05,
      "loss": 0.2738,
      "num_input_tokens_seen": 4451200,
      "step": 2050
    },
    {
      "epoch": 0.3352365415986949,
      "grad_norm": 1.4698982238769531,
      "learning_rate": 1.6753670473083198e-05,
      "loss": 0.2373,
      "num_input_tokens_seen": 4461888,
      "step": 2055
    },
    {
      "epoch": 0.3360522022838499,
      "grad_norm": 3.716040849685669,
      "learning_rate": 1.6794453507340946e-05,
      "loss": 0.1157,
      "num_input_tokens_seen": 4471392,
      "step": 2060
    },
    {
      "epoch": 0.3368678629690049,
      "grad_norm": 2.1151514053344727,
      "learning_rate": 1.6835236541598694e-05,
      "loss": 0.1184,
      "num_input_tokens_seen": 4482656,
      "step": 2065
    },
    {
      "epoch": 0.3376835236541599,
      "grad_norm": 4.903298854827881,
      "learning_rate": 1.6876019575856445e-05,
      "loss": 0.1514,
      "num_input_tokens_seen": 4494336,
      "step": 2070
    },
    {
      "epoch": 0.33849918433931486,
      "grad_norm": 5.810368537902832,
      "learning_rate": 1.6916802610114192e-05,
      "loss": 0.3192,
      "num_input_tokens_seen": 4505760,
      "step": 2075
    },
    {
      "epoch": 0.33931484502446985,
      "grad_norm": 5.632075786590576,
      "learning_rate": 1.6957585644371943e-05,
      "loss": 0.2564,
      "num_input_tokens_seen": 4514976,
      "step": 2080
    },
    {
      "epoch": 0.3401305057096248,
      "grad_norm": 0.09141606837511063,
      "learning_rate": 1.699836867862969e-05,
      "loss": 0.1058,
      "num_input_tokens_seen": 4524928,
      "step": 2085
    },
    {
      "epoch": 0.34094616639477976,
      "grad_norm": 4.968491077423096,
      "learning_rate": 1.703915171288744e-05,
      "loss": 0.2507,
      "num_input_tokens_seen": 4536192,
      "step": 2090
    },
    {
      "epoch": 0.34176182707993474,
      "grad_norm": 0.6799984574317932,
      "learning_rate": 1.7079934747145186e-05,
      "loss": 0.0741,
      "num_input_tokens_seen": 4548160,
      "step": 2095
    },
    {
      "epoch": 0.3425774877650897,
      "grad_norm": 2.8747897148132324,
      "learning_rate": 1.7120717781402937e-05,
      "loss": 0.2407,
      "num_input_tokens_seen": 4558560,
      "step": 2100
    },
    {
      "epoch": 0.3433931484502447,
      "grad_norm": 6.427094459533691,
      "learning_rate": 1.7161500815660685e-05,
      "loss": 0.2703,
      "num_input_tokens_seen": 4570272,
      "step": 2105
    },
    {
      "epoch": 0.3442088091353997,
      "grad_norm": 1.3053245544433594,
      "learning_rate": 1.7202283849918436e-05,
      "loss": 0.2167,
      "num_input_tokens_seen": 4580608,
      "step": 2110
    },
    {
      "epoch": 0.3450244698205546,
      "grad_norm": 1.7092195749282837,
      "learning_rate": 1.7243066884176184e-05,
      "loss": 0.0917,
      "num_input_tokens_seen": 4591072,
      "step": 2115
    },
    {
      "epoch": 0.3458401305057096,
      "grad_norm": 1.0552736520767212,
      "learning_rate": 1.7283849918433935e-05,
      "loss": 0.1147,
      "num_input_tokens_seen": 4602368,
      "step": 2120
    },
    {
      "epoch": 0.3466557911908646,
      "grad_norm": 4.107524394989014,
      "learning_rate": 1.732463295269168e-05,
      "loss": 0.156,
      "num_input_tokens_seen": 4613600,
      "step": 2125
    },
    {
      "epoch": 0.3474714518760196,
      "grad_norm": 0.09314478933811188,
      "learning_rate": 1.736541598694943e-05,
      "loss": 0.1241,
      "num_input_tokens_seen": 4623872,
      "step": 2130
    },
    {
      "epoch": 0.34828711256117456,
      "grad_norm": 5.320088863372803,
      "learning_rate": 1.7406199021207178e-05,
      "loss": 0.3021,
      "num_input_tokens_seen": 4635136,
      "step": 2135
    },
    {
      "epoch": 0.34910277324632955,
      "grad_norm": 0.19724801182746887,
      "learning_rate": 1.744698205546493e-05,
      "loss": 0.0691,
      "num_input_tokens_seen": 4645152,
      "step": 2140
    },
    {
      "epoch": 0.34991843393148453,
      "grad_norm": 4.815561294555664,
      "learning_rate": 1.7487765089722676e-05,
      "loss": 0.1682,
      "num_input_tokens_seen": 4654240,
      "step": 2145
    },
    {
      "epoch": 0.35073409461663946,
      "grad_norm": 1.5736823081970215,
      "learning_rate": 1.7528548123980427e-05,
      "loss": 0.1974,
      "num_input_tokens_seen": 4666240,
      "step": 2150
    },
    {
      "epoch": 0.35154975530179444,
      "grad_norm": 1.9475501775741577,
      "learning_rate": 1.756933115823817e-05,
      "loss": 0.1333,
      "num_input_tokens_seen": 4678528,
      "step": 2155
    },
    {
      "epoch": 0.3523654159869494,
      "grad_norm": 1.0774749517440796,
      "learning_rate": 1.7610114192495923e-05,
      "loss": 0.2698,
      "num_input_tokens_seen": 4688768,
      "step": 2160
    },
    {
      "epoch": 0.3531810766721044,
      "grad_norm": 2.190877676010132,
      "learning_rate": 1.765089722675367e-05,
      "loss": 0.1058,
      "num_input_tokens_seen": 4698496,
      "step": 2165
    },
    {
      "epoch": 0.3539967373572594,
      "grad_norm": 0.6246117949485779,
      "learning_rate": 1.769168026101142e-05,
      "loss": 0.0693,
      "num_input_tokens_seen": 4710208,
      "step": 2170
    },
    {
      "epoch": 0.3548123980424144,
      "grad_norm": 2.2666268348693848,
      "learning_rate": 1.773246329526917e-05,
      "loss": 0.2784,
      "num_input_tokens_seen": 4720480,
      "step": 2175
    },
    {
      "epoch": 0.3556280587275693,
      "grad_norm": 0.6565126776695251,
      "learning_rate": 1.777324632952692e-05,
      "loss": 0.3135,
      "num_input_tokens_seen": 4730560,
      "step": 2180
    },
    {
      "epoch": 0.3564437194127243,
      "grad_norm": 1.3437923192977905,
      "learning_rate": 1.7814029363784668e-05,
      "loss": 0.0311,
      "num_input_tokens_seen": 4740768,
      "step": 2185
    },
    {
      "epoch": 0.3572593800978793,
      "grad_norm": 2.0712358951568604,
      "learning_rate": 1.7854812398042415e-05,
      "loss": 0.5265,
      "num_input_tokens_seen": 4751040,
      "step": 2190
    },
    {
      "epoch": 0.35807504078303426,
      "grad_norm": 2.5632283687591553,
      "learning_rate": 1.7895595432300163e-05,
      "loss": 0.0912,
      "num_input_tokens_seen": 4762016,
      "step": 2195
    },
    {
      "epoch": 0.35889070146818924,
      "grad_norm": 2.5864036083221436,
      "learning_rate": 1.7936378466557914e-05,
      "loss": 0.1039,
      "num_input_tokens_seen": 4771648,
      "step": 2200
    },
    {
      "epoch": 0.35970636215334423,
      "grad_norm": 6.183559894561768,
      "learning_rate": 1.797716150081566e-05,
      "loss": 0.1572,
      "num_input_tokens_seen": 4781440,
      "step": 2205
    },
    {
      "epoch": 0.3605220228384992,
      "grad_norm": 1.8455876111984253,
      "learning_rate": 1.8017944535073413e-05,
      "loss": 0.1543,
      "num_input_tokens_seen": 4793312,
      "step": 2210
    },
    {
      "epoch": 0.36133768352365414,
      "grad_norm": 7.173401832580566,
      "learning_rate": 1.805872756933116e-05,
      "loss": 0.1579,
      "num_input_tokens_seen": 4804288,
      "step": 2215
    },
    {
      "epoch": 0.3621533442088091,
      "grad_norm": 9.608040809631348,
      "learning_rate": 1.8099510603588908e-05,
      "loss": 0.4598,
      "num_input_tokens_seen": 4814784,
      "step": 2220
    },
    {
      "epoch": 0.3629690048939641,
      "grad_norm": 0.30811965465545654,
      "learning_rate": 1.8140293637846655e-05,
      "loss": 0.4575,
      "num_input_tokens_seen": 4825152,
      "step": 2225
    },
    {
      "epoch": 0.3637846655791191,
      "grad_norm": 1.2179059982299805,
      "learning_rate": 1.8181076672104407e-05,
      "loss": 0.2849,
      "num_input_tokens_seen": 4835776,
      "step": 2230
    },
    {
      "epoch": 0.3646003262642741,
      "grad_norm": 1.9670007228851318,
      "learning_rate": 1.8221859706362154e-05,
      "loss": 0.1188,
      "num_input_tokens_seen": 4846848,
      "step": 2235
    },
    {
      "epoch": 0.36541598694942906,
      "grad_norm": 0.5373203158378601,
      "learning_rate": 1.8262642740619905e-05,
      "loss": 0.0217,
      "num_input_tokens_seen": 4857664,
      "step": 2240
    },
    {
      "epoch": 0.366231647634584,
      "grad_norm": 8.48862361907959,
      "learning_rate": 1.8303425774877653e-05,
      "loss": 0.3324,
      "num_input_tokens_seen": 4868800,
      "step": 2245
    },
    {
      "epoch": 0.367047308319739,
      "grad_norm": 0.15726402401924133,
      "learning_rate": 1.83442088091354e-05,
      "loss": 0.2802,
      "num_input_tokens_seen": 4879776,
      "step": 2250
    },
    {
      "epoch": 0.36786296900489396,
      "grad_norm": 0.5811845660209656,
      "learning_rate": 1.8384991843393148e-05,
      "loss": 0.2547,
      "num_input_tokens_seen": 4890816,
      "step": 2255
    },
    {
      "epoch": 0.36867862969004894,
      "grad_norm": 5.363942623138428,
      "learning_rate": 1.84257748776509e-05,
      "loss": 0.2685,
      "num_input_tokens_seen": 4899136,
      "step": 2260
    },
    {
      "epoch": 0.3694942903752039,
      "grad_norm": 0.11295244842767715,
      "learning_rate": 1.8466557911908647e-05,
      "loss": 0.144,
      "num_input_tokens_seen": 4910112,
      "step": 2265
    },
    {
      "epoch": 0.3703099510603589,
      "grad_norm": 3.5267744064331055,
      "learning_rate": 1.8507340946166394e-05,
      "loss": 0.2599,
      "num_input_tokens_seen": 4920032,
      "step": 2270
    },
    {
      "epoch": 0.37112561174551384,
      "grad_norm": 4.316346645355225,
      "learning_rate": 1.8548123980424145e-05,
      "loss": 0.161,
      "num_input_tokens_seen": 4930752,
      "step": 2275
    },
    {
      "epoch": 0.3719412724306688,
      "grad_norm": 0.11088819056749344,
      "learning_rate": 1.8588907014681893e-05,
      "loss": 0.1533,
      "num_input_tokens_seen": 4940960,
      "step": 2280
    },
    {
      "epoch": 0.3727569331158238,
      "grad_norm": 3.6862525939941406,
      "learning_rate": 1.862969004893964e-05,
      "loss": 0.2039,
      "num_input_tokens_seen": 4952128,
      "step": 2285
    },
    {
      "epoch": 0.3735725938009788,
      "grad_norm": 1.4289133548736572,
      "learning_rate": 1.867047308319739e-05,
      "loss": 0.3226,
      "num_input_tokens_seen": 4962848,
      "step": 2290
    },
    {
      "epoch": 0.3743882544861338,
      "grad_norm": 2.987999200820923,
      "learning_rate": 1.871125611745514e-05,
      "loss": 0.0703,
      "num_input_tokens_seen": 4972448,
      "step": 2295
    },
    {
      "epoch": 0.37520391517128876,
      "grad_norm": 0.7900047898292542,
      "learning_rate": 1.8752039151712887e-05,
      "loss": 0.184,
      "num_input_tokens_seen": 4982432,
      "step": 2300
    },
    {
      "epoch": 0.37601957585644374,
      "grad_norm": 0.06719323247671127,
      "learning_rate": 1.8792822185970638e-05,
      "loss": 0.0839,
      "num_input_tokens_seen": 4992800,
      "step": 2305
    },
    {
      "epoch": 0.3768352365415987,
      "grad_norm": 1.453745722770691,
      "learning_rate": 1.8833605220228386e-05,
      "loss": 0.0334,
      "num_input_tokens_seen": 5004352,
      "step": 2310
    },
    {
      "epoch": 0.37765089722675366,
      "grad_norm": 0.4515019357204437,
      "learning_rate": 1.8874388254486133e-05,
      "loss": 0.1465,
      "num_input_tokens_seen": 5015968,
      "step": 2315
    },
    {
      "epoch": 0.37846655791190864,
      "grad_norm": 4.4329304695129395,
      "learning_rate": 1.891517128874388e-05,
      "loss": 0.0731,
      "num_input_tokens_seen": 5026624,
      "step": 2320
    },
    {
      "epoch": 0.3792822185970636,
      "grad_norm": 6.45882511138916,
      "learning_rate": 1.8955954323001632e-05,
      "loss": 0.1383,
      "num_input_tokens_seen": 5037312,
      "step": 2325
    },
    {
      "epoch": 0.3800978792822186,
      "grad_norm": 1.0029453039169312,
      "learning_rate": 1.899673735725938e-05,
      "loss": 0.1604,
      "num_input_tokens_seen": 5048544,
      "step": 2330
    },
    {
      "epoch": 0.3809135399673736,
      "grad_norm": 1.8722848892211914,
      "learning_rate": 1.903752039151713e-05,
      "loss": 0.1606,
      "num_input_tokens_seen": 5059584,
      "step": 2335
    },
    {
      "epoch": 0.3817292006525285,
      "grad_norm": 2.014350175857544,
      "learning_rate": 1.907830342577488e-05,
      "loss": 0.2206,
      "num_input_tokens_seen": 5072736,
      "step": 2340
    },
    {
      "epoch": 0.3825448613376835,
      "grad_norm": 3.0931100845336914,
      "learning_rate": 1.911908646003263e-05,
      "loss": 0.5229,
      "num_input_tokens_seen": 5083456,
      "step": 2345
    },
    {
      "epoch": 0.3833605220228385,
      "grad_norm": 1.8392821550369263,
      "learning_rate": 1.9159869494290374e-05,
      "loss": 0.0961,
      "num_input_tokens_seen": 5094240,
      "step": 2350
    },
    {
      "epoch": 0.3841761827079935,
      "grad_norm": 6.834700107574463,
      "learning_rate": 1.9200652528548125e-05,
      "loss": 0.2011,
      "num_input_tokens_seen": 5103744,
      "step": 2355
    },
    {
      "epoch": 0.38499184339314846,
      "grad_norm": 5.237668037414551,
      "learning_rate": 1.9241435562805872e-05,
      "loss": 0.2381,
      "num_input_tokens_seen": 5114048,
      "step": 2360
    },
    {
      "epoch": 0.38580750407830344,
      "grad_norm": 6.375014305114746,
      "learning_rate": 1.9282218597063623e-05,
      "loss": 0.3084,
      "num_input_tokens_seen": 5123744,
      "step": 2365
    },
    {
      "epoch": 0.3866231647634584,
      "grad_norm": 6.529644966125488,
      "learning_rate": 1.932300163132137e-05,
      "loss": 0.2536,
      "num_input_tokens_seen": 5133888,
      "step": 2370
    },
    {
      "epoch": 0.38743882544861336,
      "grad_norm": 0.03799702227115631,
      "learning_rate": 1.9363784665579122e-05,
      "loss": 0.157,
      "num_input_tokens_seen": 5145024,
      "step": 2375
    },
    {
      "epoch": 0.38825448613376834,
      "grad_norm": 2.366711139678955,
      "learning_rate": 1.9404567699836866e-05,
      "loss": 0.0423,
      "num_input_tokens_seen": 5156544,
      "step": 2380
    },
    {
      "epoch": 0.3890701468189233,
      "grad_norm": 4.6945600509643555,
      "learning_rate": 1.9445350734094617e-05,
      "loss": 0.2764,
      "num_input_tokens_seen": 5168192,
      "step": 2385
    },
    {
      "epoch": 0.3898858075040783,
      "grad_norm": 2.4937102794647217,
      "learning_rate": 1.9486133768352365e-05,
      "loss": 0.1952,
      "num_input_tokens_seen": 5178240,
      "step": 2390
    },
    {
      "epoch": 0.3907014681892333,
      "grad_norm": 0.42788416147232056,
      "learning_rate": 1.9526916802610116e-05,
      "loss": 0.0542,
      "num_input_tokens_seen": 5189248,
      "step": 2395
    },
    {
      "epoch": 0.3915171288743883,
      "grad_norm": 1.277098298072815,
      "learning_rate": 1.9567699836867864e-05,
      "loss": 0.0796,
      "num_input_tokens_seen": 5199392,
      "step": 2400
    },
    {
      "epoch": 0.3923327895595432,
      "grad_norm": 5.205860614776611,
      "learning_rate": 1.9608482871125615e-05,
      "loss": 0.2279,
      "num_input_tokens_seen": 5210240,
      "step": 2405
    },
    {
      "epoch": 0.3931484502446982,
      "grad_norm": 8.374488830566406,
      "learning_rate": 1.9649265905383362e-05,
      "loss": 0.2954,
      "num_input_tokens_seen": 5220000,
      "step": 2410
    },
    {
      "epoch": 0.3939641109298532,
      "grad_norm": 3.9300363063812256,
      "learning_rate": 1.969004893964111e-05,
      "loss": 0.0821,
      "num_input_tokens_seen": 5231520,
      "step": 2415
    },
    {
      "epoch": 0.39477977161500816,
      "grad_norm": 4.3988776206970215,
      "learning_rate": 1.9730831973898858e-05,
      "loss": 0.3133,
      "num_input_tokens_seen": 5243200,
      "step": 2420
    },
    {
      "epoch": 0.39559543230016314,
      "grad_norm": 3.794332504272461,
      "learning_rate": 1.977161500815661e-05,
      "loss": 0.139,
      "num_input_tokens_seen": 5253632,
      "step": 2425
    },
    {
      "epoch": 0.3964110929853181,
      "grad_norm": 0.4077603816986084,
      "learning_rate": 1.9812398042414356e-05,
      "loss": 0.1638,
      "num_input_tokens_seen": 5264736,
      "step": 2430
    },
    {
      "epoch": 0.3972267536704731,
      "grad_norm": 1.15377938747406,
      "learning_rate": 1.9853181076672107e-05,
      "loss": 0.193,
      "num_input_tokens_seen": 5276192,
      "step": 2435
    },
    {
      "epoch": 0.39804241435562804,
      "grad_norm": 0.22548221051692963,
      "learning_rate": 1.9893964110929855e-05,
      "loss": 0.1289,
      "num_input_tokens_seen": 5287296,
      "step": 2440
    },
    {
      "epoch": 0.398858075040783,
      "grad_norm": 0.8492670655250549,
      "learning_rate": 1.9934747145187603e-05,
      "loss": 0.1942,
      "num_input_tokens_seen": 5298336,
      "step": 2445
    },
    {
      "epoch": 0.399673735725938,
      "grad_norm": 0.9038564562797546,
      "learning_rate": 1.997553017944535e-05,
      "loss": 0.0822,
      "num_input_tokens_seen": 5309248,
      "step": 2450
    },
    {
      "epoch": 0.400489396411093,
      "grad_norm": 0.5144060254096985,
      "learning_rate": 2.00163132137031e-05,
      "loss": 0.3752,
      "num_input_tokens_seen": 5320384,
      "step": 2455
    },
    {
      "epoch": 0.401305057096248,
      "grad_norm": 0.5081189274787903,
      "learning_rate": 2.005709624796085e-05,
      "loss": 0.0487,
      "num_input_tokens_seen": 5331904,
      "step": 2460
    },
    {
      "epoch": 0.40212071778140296,
      "grad_norm": 0.2553884983062744,
      "learning_rate": 2.00978792822186e-05,
      "loss": 0.1636,
      "num_input_tokens_seen": 5343904,
      "step": 2465
    },
    {
      "epoch": 0.4029363784665579,
      "grad_norm": 1.0112682580947876,
      "learning_rate": 2.0138662316476348e-05,
      "loss": 0.0555,
      "num_input_tokens_seen": 5354336,
      "step": 2470
    },
    {
      "epoch": 0.40375203915171287,
      "grad_norm": 0.5315742492675781,
      "learning_rate": 2.0179445350734095e-05,
      "loss": 0.2583,
      "num_input_tokens_seen": 5366144,
      "step": 2475
    },
    {
      "epoch": 0.40456769983686786,
      "grad_norm": 4.658558368682861,
      "learning_rate": 2.0220228384991843e-05,
      "loss": 0.1024,
      "num_input_tokens_seen": 5377344,
      "step": 2480
    },
    {
      "epoch": 0.40538336052202284,
      "grad_norm": 4.860203266143799,
      "learning_rate": 2.0261011419249594e-05,
      "loss": 0.0958,
      "num_input_tokens_seen": 5386752,
      "step": 2485
    },
    {
      "epoch": 0.4061990212071778,
      "grad_norm": 0.3144293427467346,
      "learning_rate": 2.030179445350734e-05,
      "loss": 0.0502,
      "num_input_tokens_seen": 5396384,
      "step": 2490
    },
    {
      "epoch": 0.4070146818923328,
      "grad_norm": 5.124461650848389,
      "learning_rate": 2.034257748776509e-05,
      "loss": 0.2127,
      "num_input_tokens_seen": 5408256,
      "step": 2495
    },
    {
      "epoch": 0.4078303425774878,
      "grad_norm": 0.35638633370399475,
      "learning_rate": 2.038336052202284e-05,
      "loss": 0.2208,
      "num_input_tokens_seen": 5418752,
      "step": 2500
    },
    {
      "epoch": 0.4086460032626427,
      "grad_norm": 0.1471187174320221,
      "learning_rate": 2.0424143556280588e-05,
      "loss": 0.1479,
      "num_input_tokens_seen": 5429792,
      "step": 2505
    },
    {
      "epoch": 0.4094616639477977,
      "grad_norm": 6.38442850112915,
      "learning_rate": 2.0464926590538336e-05,
      "loss": 0.3057,
      "num_input_tokens_seen": 5440608,
      "step": 2510
    },
    {
      "epoch": 0.4102773246329527,
      "grad_norm": 2.963207483291626,
      "learning_rate": 2.0505709624796087e-05,
      "loss": 0.2826,
      "num_input_tokens_seen": 5450016,
      "step": 2515
    },
    {
      "epoch": 0.4110929853181077,
      "grad_norm": 12.700345039367676,
      "learning_rate": 2.0546492659053834e-05,
      "loss": 0.5221,
      "num_input_tokens_seen": 5461120,
      "step": 2520
    },
    {
      "epoch": 0.41190864600326266,
      "grad_norm": 8.157638549804688,
      "learning_rate": 2.0587275693311582e-05,
      "loss": 0.0987,
      "num_input_tokens_seen": 5471648,
      "step": 2525
    },
    {
      "epoch": 0.41272430668841764,
      "grad_norm": 0.29782620072364807,
      "learning_rate": 2.0628058727569333e-05,
      "loss": 0.2887,
      "num_input_tokens_seen": 5482656,
      "step": 2530
    },
    {
      "epoch": 0.41353996737357257,
      "grad_norm": 2.9385223388671875,
      "learning_rate": 2.066884176182708e-05,
      "loss": 0.0505,
      "num_input_tokens_seen": 5491168,
      "step": 2535
    },
    {
      "epoch": 0.41435562805872755,
      "grad_norm": 1.7700849771499634,
      "learning_rate": 2.070962479608483e-05,
      "loss": 0.2606,
      "num_input_tokens_seen": 5501792,
      "step": 2540
    },
    {
      "epoch": 0.41517128874388254,
      "grad_norm": 5.325303077697754,
      "learning_rate": 2.0750407830342576e-05,
      "loss": 0.3087,
      "num_input_tokens_seen": 5513152,
      "step": 2545
    },
    {
      "epoch": 0.4159869494290375,
      "grad_norm": 0.27992990612983704,
      "learning_rate": 2.0791190864600327e-05,
      "loss": 0.031,
      "num_input_tokens_seen": 5524480,
      "step": 2550
    },
    {
      "epoch": 0.4168026101141925,
      "grad_norm": 0.08975686132907867,
      "learning_rate": 2.0831973898858075e-05,
      "loss": 0.1271,
      "num_input_tokens_seen": 5535008,
      "step": 2555
    },
    {
      "epoch": 0.4176182707993475,
      "grad_norm": 0.9344058036804199,
      "learning_rate": 2.0872756933115826e-05,
      "loss": 0.0887,
      "num_input_tokens_seen": 5546048,
      "step": 2560
    },
    {
      "epoch": 0.4184339314845024,
      "grad_norm": 7.105401039123535,
      "learning_rate": 2.0913539967373573e-05,
      "loss": 0.3133,
      "num_input_tokens_seen": 5557760,
      "step": 2565
    },
    {
      "epoch": 0.4192495921696574,
      "grad_norm": 0.6101921200752258,
      "learning_rate": 2.0954323001631324e-05,
      "loss": 0.2166,
      "num_input_tokens_seen": 5567488,
      "step": 2570
    },
    {
      "epoch": 0.4200652528548124,
      "grad_norm": 4.419879913330078,
      "learning_rate": 2.099510603588907e-05,
      "loss": 0.2825,
      "num_input_tokens_seen": 5578592,
      "step": 2575
    },
    {
      "epoch": 0.42088091353996737,
      "grad_norm": 3.2440834045410156,
      "learning_rate": 2.103588907014682e-05,
      "loss": 0.0648,
      "num_input_tokens_seen": 5588576,
      "step": 2580
    },
    {
      "epoch": 0.42169657422512236,
      "grad_norm": 1.5354163646697998,
      "learning_rate": 2.1076672104404567e-05,
      "loss": 0.078,
      "num_input_tokens_seen": 5599264,
      "step": 2585
    },
    {
      "epoch": 0.42251223491027734,
      "grad_norm": 0.42529988288879395,
      "learning_rate": 2.1117455138662318e-05,
      "loss": 0.1078,
      "num_input_tokens_seen": 5609504,
      "step": 2590
    },
    {
      "epoch": 0.4233278955954323,
      "grad_norm": 10.483762741088867,
      "learning_rate": 2.1158238172920066e-05,
      "loss": 0.2525,
      "num_input_tokens_seen": 5620256,
      "step": 2595
    },
    {
      "epoch": 0.42414355628058725,
      "grad_norm": 5.660029888153076,
      "learning_rate": 2.1199021207177817e-05,
      "loss": 0.5248,
      "num_input_tokens_seen": 5630592,
      "step": 2600
    },
    {
      "epoch": 0.42495921696574224,
      "grad_norm": 6.411313056945801,
      "learning_rate": 2.123980424143556e-05,
      "loss": 0.3363,
      "num_input_tokens_seen": 5641536,
      "step": 2605
    },
    {
      "epoch": 0.4257748776508972,
      "grad_norm": 0.7426936626434326,
      "learning_rate": 2.1280587275693312e-05,
      "loss": 0.0497,
      "num_input_tokens_seen": 5652416,
      "step": 2610
    },
    {
      "epoch": 0.4265905383360522,
      "grad_norm": 5.806601047515869,
      "learning_rate": 2.132137030995106e-05,
      "loss": 0.2456,
      "num_input_tokens_seen": 5662528,
      "step": 2615
    },
    {
      "epoch": 0.4274061990212072,
      "grad_norm": 3.508190631866455,
      "learning_rate": 2.136215334420881e-05,
      "loss": 0.1547,
      "num_input_tokens_seen": 5672288,
      "step": 2620
    },
    {
      "epoch": 0.4282218597063622,
      "grad_norm": 1.4045436382293701,
      "learning_rate": 2.140293637846656e-05,
      "loss": 0.1169,
      "num_input_tokens_seen": 5682976,
      "step": 2625
    },
    {
      "epoch": 0.4290375203915171,
      "grad_norm": 0.7463783025741577,
      "learning_rate": 2.144371941272431e-05,
      "loss": 0.082,
      "num_input_tokens_seen": 5694240,
      "step": 2630
    },
    {
      "epoch": 0.4298531810766721,
      "grad_norm": 0.21157114207744598,
      "learning_rate": 2.1484502446982057e-05,
      "loss": 0.1706,
      "num_input_tokens_seen": 5704192,
      "step": 2635
    },
    {
      "epoch": 0.43066884176182707,
      "grad_norm": 0.380730003118515,
      "learning_rate": 2.1525285481239805e-05,
      "loss": 0.0482,
      "num_input_tokens_seen": 5715968,
      "step": 2640
    },
    {
      "epoch": 0.43148450244698205,
      "grad_norm": 2.2118470668792725,
      "learning_rate": 2.1566068515497553e-05,
      "loss": 0.1042,
      "num_input_tokens_seen": 5726784,
      "step": 2645
    },
    {
      "epoch": 0.43230016313213704,
      "grad_norm": 9.747725486755371,
      "learning_rate": 2.1606851549755304e-05,
      "loss": 0.5689,
      "num_input_tokens_seen": 5737952,
      "step": 2650
    },
    {
      "epoch": 0.433115823817292,
      "grad_norm": 1.891772985458374,
      "learning_rate": 2.164763458401305e-05,
      "loss": 0.1042,
      "num_input_tokens_seen": 5748512,
      "step": 2655
    },
    {
      "epoch": 0.433931484502447,
      "grad_norm": 0.3178868591785431,
      "learning_rate": 2.1688417618270802e-05,
      "loss": 0.2221,
      "num_input_tokens_seen": 5758944,
      "step": 2660
    },
    {
      "epoch": 0.43474714518760194,
      "grad_norm": 1.9077990055084229,
      "learning_rate": 2.172920065252855e-05,
      "loss": 0.1318,
      "num_input_tokens_seen": 5769632,
      "step": 2665
    },
    {
      "epoch": 0.4355628058727569,
      "grad_norm": 3.1999354362487793,
      "learning_rate": 2.1769983686786298e-05,
      "loss": 0.0573,
      "num_input_tokens_seen": 5781248,
      "step": 2670
    },
    {
      "epoch": 0.4363784665579119,
      "grad_norm": 1.6499738693237305,
      "learning_rate": 2.1810766721044045e-05,
      "loss": 0.2059,
      "num_input_tokens_seen": 5792192,
      "step": 2675
    },
    {
      "epoch": 0.4371941272430669,
      "grad_norm": 3.8542420864105225,
      "learning_rate": 2.1851549755301796e-05,
      "loss": 0.0722,
      "num_input_tokens_seen": 5803264,
      "step": 2680
    },
    {
      "epoch": 0.43800978792822187,
      "grad_norm": 0.095824234187603,
      "learning_rate": 2.1892332789559544e-05,
      "loss": 0.0717,
      "num_input_tokens_seen": 5815072,
      "step": 2685
    },
    {
      "epoch": 0.43882544861337686,
      "grad_norm": 0.39370542764663696,
      "learning_rate": 2.1933115823817295e-05,
      "loss": 0.2932,
      "num_input_tokens_seen": 5825248,
      "step": 2690
    },
    {
      "epoch": 0.4396411092985318,
      "grad_norm": 7.779138088226318,
      "learning_rate": 2.1973898858075043e-05,
      "loss": 0.2817,
      "num_input_tokens_seen": 5835904,
      "step": 2695
    },
    {
      "epoch": 0.44045676998368677,
      "grad_norm": 2.862678050994873,
      "learning_rate": 2.201468189233279e-05,
      "loss": 0.1971,
      "num_input_tokens_seen": 5844736,
      "step": 2700
    },
    {
      "epoch": 0.44127243066884175,
      "grad_norm": 1.2629541158676147,
      "learning_rate": 2.2055464926590538e-05,
      "loss": 0.3501,
      "num_input_tokens_seen": 5856672,
      "step": 2705
    },
    {
      "epoch": 0.44208809135399674,
      "grad_norm": 0.3131316602230072,
      "learning_rate": 2.209624796084829e-05,
      "loss": 0.1294,
      "num_input_tokens_seen": 5866848,
      "step": 2710
    },
    {
      "epoch": 0.4429037520391517,
      "grad_norm": 0.14001265168190002,
      "learning_rate": 2.2137030995106037e-05,
      "loss": 0.0316,
      "num_input_tokens_seen": 5877856,
      "step": 2715
    },
    {
      "epoch": 0.4437194127243067,
      "grad_norm": 0.40793848037719727,
      "learning_rate": 2.2177814029363788e-05,
      "loss": 0.2193,
      "num_input_tokens_seen": 5888896,
      "step": 2720
    },
    {
      "epoch": 0.4445350734094617,
      "grad_norm": 0.7146620154380798,
      "learning_rate": 2.2218597063621535e-05,
      "loss": 0.2333,
      "num_input_tokens_seen": 5899872,
      "step": 2725
    },
    {
      "epoch": 0.4453507340946166,
      "grad_norm": 6.070364475250244,
      "learning_rate": 2.2259380097879283e-05,
      "loss": 0.1109,
      "num_input_tokens_seen": 5910272,
      "step": 2730
    },
    {
      "epoch": 0.4461663947797716,
      "grad_norm": 6.409831523895264,
      "learning_rate": 2.230016313213703e-05,
      "loss": 0.2467,
      "num_input_tokens_seen": 5921312,
      "step": 2735
    },
    {
      "epoch": 0.4469820554649266,
      "grad_norm": 1.4456238746643066,
      "learning_rate": 2.234094616639478e-05,
      "loss": 0.0908,
      "num_input_tokens_seen": 5932032,
      "step": 2740
    },
    {
      "epoch": 0.44779771615008157,
      "grad_norm": 0.36780887842178345,
      "learning_rate": 2.238172920065253e-05,
      "loss": 0.0621,
      "num_input_tokens_seen": 5942176,
      "step": 2745
    },
    {
      "epoch": 0.44861337683523655,
      "grad_norm": 7.774724006652832,
      "learning_rate": 2.2422512234910277e-05,
      "loss": 0.4973,
      "num_input_tokens_seen": 5954336,
      "step": 2750
    },
    {
      "epoch": 0.44942903752039154,
      "grad_norm": 8.789905548095703,
      "learning_rate": 2.2463295269168028e-05,
      "loss": 0.37,
      "num_input_tokens_seen": 5965152,
      "step": 2755
    },
    {
      "epoch": 0.45024469820554647,
      "grad_norm": 2.8583152294158936,
      "learning_rate": 2.2504078303425776e-05,
      "loss": 0.3133,
      "num_input_tokens_seen": 5977216,
      "step": 2760
    },
    {
      "epoch": 0.45106035889070145,
      "grad_norm": 0.6330432891845703,
      "learning_rate": 2.2544861337683527e-05,
      "loss": 0.201,
      "num_input_tokens_seen": 5988000,
      "step": 2765
    },
    {
      "epoch": 0.45187601957585644,
      "grad_norm": 5.926965713500977,
      "learning_rate": 2.258564437194127e-05,
      "loss": 0.1667,
      "num_input_tokens_seen": 5998400,
      "step": 2770
    },
    {
      "epoch": 0.4526916802610114,
      "grad_norm": 4.193154811859131,
      "learning_rate": 2.2626427406199022e-05,
      "loss": 0.2632,
      "num_input_tokens_seen": 6010176,
      "step": 2775
    },
    {
      "epoch": 0.4535073409461664,
      "grad_norm": 5.64955997467041,
      "learning_rate": 2.266721044045677e-05,
      "loss": 0.178,
      "num_input_tokens_seen": 6021312,
      "step": 2780
    },
    {
      "epoch": 0.4543230016313214,
      "grad_norm": 5.58404016494751,
      "learning_rate": 2.270799347471452e-05,
      "loss": 0.1576,
      "num_input_tokens_seen": 6033024,
      "step": 2785
    },
    {
      "epoch": 0.4551386623164764,
      "grad_norm": 0.048357389867305756,
      "learning_rate": 2.2748776508972268e-05,
      "loss": 0.0812,
      "num_input_tokens_seen": 6042432,
      "step": 2790
    },
    {
      "epoch": 0.4559543230016313,
      "grad_norm": 3.361707925796509,
      "learning_rate": 2.278955954323002e-05,
      "loss": 0.205,
      "num_input_tokens_seen": 6054304,
      "step": 2795
    },
    {
      "epoch": 0.4567699836867863,
      "grad_norm": 2.470402956008911,
      "learning_rate": 2.2830342577487763e-05,
      "loss": 0.2135,
      "num_input_tokens_seen": 6065312,
      "step": 2800
    },
    {
      "epoch": 0.45758564437194127,
      "grad_norm": 4.321540832519531,
      "learning_rate": 2.2871125611745514e-05,
      "loss": 0.4454,
      "num_input_tokens_seen": 6075072,
      "step": 2805
    },
    {
      "epoch": 0.45840130505709625,
      "grad_norm": 6.536118507385254,
      "learning_rate": 2.2911908646003262e-05,
      "loss": 0.0626,
      "num_input_tokens_seen": 6084992,
      "step": 2810
    },
    {
      "epoch": 0.45921696574225124,
      "grad_norm": 8.955811500549316,
      "learning_rate": 2.2952691680261013e-05,
      "loss": 0.0782,
      "num_input_tokens_seen": 6095424,
      "step": 2815
    },
    {
      "epoch": 0.4600326264274062,
      "grad_norm": 3.7599077224731445,
      "learning_rate": 2.299347471451876e-05,
      "loss": 0.0874,
      "num_input_tokens_seen": 6107328,
      "step": 2820
    },
    {
      "epoch": 0.46084828711256115,
      "grad_norm": 2.6868855953216553,
      "learning_rate": 2.3034257748776512e-05,
      "loss": 0.0272,
      "num_input_tokens_seen": 6118048,
      "step": 2825
    },
    {
      "epoch": 0.46166394779771613,
      "grad_norm": 5.307155132293701,
      "learning_rate": 2.307504078303426e-05,
      "loss": 0.4782,
      "num_input_tokens_seen": 6128512,
      "step": 2830
    },
    {
      "epoch": 0.4624796084828711,
      "grad_norm": 0.8071622848510742,
      "learning_rate": 2.3115823817292007e-05,
      "loss": 0.0383,
      "num_input_tokens_seen": 6138720,
      "step": 2835
    },
    {
      "epoch": 0.4632952691680261,
      "grad_norm": 4.2207417488098145,
      "learning_rate": 2.3156606851549755e-05,
      "loss": 0.2491,
      "num_input_tokens_seen": 6149600,
      "step": 2840
    },
    {
      "epoch": 0.4641109298531811,
      "grad_norm": 8.52420425415039,
      "learning_rate": 2.3197389885807506e-05,
      "loss": 0.1157,
      "num_input_tokens_seen": 6160608,
      "step": 2845
    },
    {
      "epoch": 0.46492659053833607,
      "grad_norm": 3.6866848468780518,
      "learning_rate": 2.3238172920065253e-05,
      "loss": 0.2482,
      "num_input_tokens_seen": 6171968,
      "step": 2850
    },
    {
      "epoch": 0.46574225122349105,
      "grad_norm": 4.755484104156494,
      "learning_rate": 2.3278955954323004e-05,
      "loss": 0.2233,
      "num_input_tokens_seen": 6183360,
      "step": 2855
    },
    {
      "epoch": 0.466557911908646,
      "grad_norm": 2.6169004440307617,
      "learning_rate": 2.3319738988580752e-05,
      "loss": 0.1574,
      "num_input_tokens_seen": 6193664,
      "step": 2860
    },
    {
      "epoch": 0.46737357259380097,
      "grad_norm": 0.6334134340286255,
      "learning_rate": 2.33605220228385e-05,
      "loss": 0.0197,
      "num_input_tokens_seen": 6204320,
      "step": 2865
    },
    {
      "epoch": 0.46818923327895595,
      "grad_norm": 6.50516939163208,
      "learning_rate": 2.3401305057096247e-05,
      "loss": 0.4034,
      "num_input_tokens_seen": 6214560,
      "step": 2870
    },
    {
      "epoch": 0.46900489396411094,
      "grad_norm": 2.0981626510620117,
      "learning_rate": 2.3442088091354e-05,
      "loss": 0.3495,
      "num_input_tokens_seen": 6226176,
      "step": 2875
    },
    {
      "epoch": 0.4698205546492659,
      "grad_norm": 2.2720494270324707,
      "learning_rate": 2.3482871125611746e-05,
      "loss": 0.5628,
      "num_input_tokens_seen": 6236768,
      "step": 2880
    },
    {
      "epoch": 0.4706362153344209,
      "grad_norm": 7.052456855773926,
      "learning_rate": 2.3523654159869497e-05,
      "loss": 0.1432,
      "num_input_tokens_seen": 6247424,
      "step": 2885
    },
    {
      "epoch": 0.47145187601957583,
      "grad_norm": 0.3121151924133301,
      "learning_rate": 2.3564437194127245e-05,
      "loss": 0.0287,
      "num_input_tokens_seen": 6257280,
      "step": 2890
    },
    {
      "epoch": 0.4722675367047308,
      "grad_norm": 2.9958417415618896,
      "learning_rate": 2.3605220228384996e-05,
      "loss": 0.1357,
      "num_input_tokens_seen": 6268160,
      "step": 2895
    },
    {
      "epoch": 0.4730831973898858,
      "grad_norm": 0.3367376923561096,
      "learning_rate": 2.364600326264274e-05,
      "loss": 0.0691,
      "num_input_tokens_seen": 6278976,
      "step": 2900
    },
    {
      "epoch": 0.4738988580750408,
      "grad_norm": 0.07608085870742798,
      "learning_rate": 2.368678629690049e-05,
      "loss": 0.1561,
      "num_input_tokens_seen": 6290144,
      "step": 2905
    },
    {
      "epoch": 0.47471451876019577,
      "grad_norm": 2.0899765491485596,
      "learning_rate": 2.372756933115824e-05,
      "loss": 0.1741,
      "num_input_tokens_seen": 6301760,
      "step": 2910
    },
    {
      "epoch": 0.47553017944535075,
      "grad_norm": 0.1436070054769516,
      "learning_rate": 2.376835236541599e-05,
      "loss": 0.4465,
      "num_input_tokens_seen": 6312416,
      "step": 2915
    },
    {
      "epoch": 0.4763458401305057,
      "grad_norm": 0.6537477970123291,
      "learning_rate": 2.3809135399673737e-05,
      "loss": 0.0616,
      "num_input_tokens_seen": 6323488,
      "step": 2920
    },
    {
      "epoch": 0.47716150081566067,
      "grad_norm": 5.123737335205078,
      "learning_rate": 2.384991843393149e-05,
      "loss": 0.2836,
      "num_input_tokens_seen": 6334496,
      "step": 2925
    },
    {
      "epoch": 0.47797716150081565,
      "grad_norm": 2.81748104095459,
      "learning_rate": 2.3890701468189233e-05,
      "loss": 0.2396,
      "num_input_tokens_seen": 6344320,
      "step": 2930
    },
    {
      "epoch": 0.47879282218597063,
      "grad_norm": 5.903765678405762,
      "learning_rate": 2.3931484502446984e-05,
      "loss": 0.2596,
      "num_input_tokens_seen": 6355904,
      "step": 2935
    },
    {
      "epoch": 0.4796084828711256,
      "grad_norm": 4.702868461608887,
      "learning_rate": 2.397226753670473e-05,
      "loss": 0.3952,
      "num_input_tokens_seen": 6366016,
      "step": 2940
    },
    {
      "epoch": 0.4804241435562806,
      "grad_norm": 3.574524402618408,
      "learning_rate": 2.4013050570962482e-05,
      "loss": 0.0749,
      "num_input_tokens_seen": 6375712,
      "step": 2945
    },
    {
      "epoch": 0.4812398042414356,
      "grad_norm": 0.4243795871734619,
      "learning_rate": 2.405383360522023e-05,
      "loss": 0.1481,
      "num_input_tokens_seen": 6386720,
      "step": 2950
    },
    {
      "epoch": 0.4820554649265905,
      "grad_norm": 0.09564966708421707,
      "learning_rate": 2.4094616639477978e-05,
      "loss": 0.0398,
      "num_input_tokens_seen": 6397536,
      "step": 2955
    },
    {
      "epoch": 0.4828711256117455,
      "grad_norm": 4.892819404602051,
      "learning_rate": 2.4135399673735725e-05,
      "loss": 0.2196,
      "num_input_tokens_seen": 6408384,
      "step": 2960
    },
    {
      "epoch": 0.4836867862969005,
      "grad_norm": 3.464731454849243,
      "learning_rate": 2.4176182707993476e-05,
      "loss": 0.0864,
      "num_input_tokens_seen": 6420640,
      "step": 2965
    },
    {
      "epoch": 0.48450244698205547,
      "grad_norm": 0.6103954315185547,
      "learning_rate": 2.4216965742251224e-05,
      "loss": 0.1325,
      "num_input_tokens_seen": 6431680,
      "step": 2970
    },
    {
      "epoch": 0.48531810766721045,
      "grad_norm": 0.8500288724899292,
      "learning_rate": 2.425774877650897e-05,
      "loss": 0.1004,
      "num_input_tokens_seen": 6442368,
      "step": 2975
    },
    {
      "epoch": 0.48613376835236544,
      "grad_norm": 0.02255425974726677,
      "learning_rate": 2.4298531810766723e-05,
      "loss": 0.1682,
      "num_input_tokens_seen": 6453856,
      "step": 2980
    },
    {
      "epoch": 0.48694942903752036,
      "grad_norm": 2.5796749591827393,
      "learning_rate": 2.433931484502447e-05,
      "loss": 0.3037,
      "num_input_tokens_seen": 6464640,
      "step": 2985
    },
    {
      "epoch": 0.48776508972267535,
      "grad_norm": 11.368684768676758,
      "learning_rate": 2.438009787928222e-05,
      "loss": 0.1652,
      "num_input_tokens_seen": 6475392,
      "step": 2990
    },
    {
      "epoch": 0.48858075040783033,
      "grad_norm": 3.247875690460205,
      "learning_rate": 2.442088091353997e-05,
      "loss": 0.1504,
      "num_input_tokens_seen": 6486880,
      "step": 2995
    },
    {
      "epoch": 0.4893964110929853,
      "grad_norm": 0.5905230045318604,
      "learning_rate": 2.4461663947797717e-05,
      "loss": 0.2028,
      "num_input_tokens_seen": 6497216,
      "step": 3000
    },
    {
      "epoch": 0.4902120717781403,
      "grad_norm": 5.586061954498291,
      "learning_rate": 2.4502446982055464e-05,
      "loss": 0.248,
      "num_input_tokens_seen": 6508320,
      "step": 3005
    },
    {
      "epoch": 0.4910277324632953,
      "grad_norm": 5.63607120513916,
      "learning_rate": 2.4543230016313215e-05,
      "loss": 0.1885,
      "num_input_tokens_seen": 6519520,
      "step": 3010
    },
    {
      "epoch": 0.49184339314845027,
      "grad_norm": 3.0965116024017334,
      "learning_rate": 2.4584013050570963e-05,
      "loss": 0.2348,
      "num_input_tokens_seen": 6530240,
      "step": 3015
    },
    {
      "epoch": 0.4926590538336052,
      "grad_norm": 4.9088664054870605,
      "learning_rate": 2.4624796084828714e-05,
      "loss": 0.3563,
      "num_input_tokens_seen": 6540192,
      "step": 3020
    },
    {
      "epoch": 0.4934747145187602,
      "grad_norm": 0.8939807415008545,
      "learning_rate": 2.466557911908646e-05,
      "loss": 0.1468,
      "num_input_tokens_seen": 6550016,
      "step": 3025
    },
    {
      "epoch": 0.49429037520391517,
      "grad_norm": 0.1896563619375229,
      "learning_rate": 2.470636215334421e-05,
      "loss": 0.1385,
      "num_input_tokens_seen": 6560672,
      "step": 3030
    },
    {
      "epoch": 0.49510603588907015,
      "grad_norm": 0.6451339721679688,
      "learning_rate": 2.4747145187601957e-05,
      "loss": 0.086,
      "num_input_tokens_seen": 6571808,
      "step": 3035
    },
    {
      "epoch": 0.49592169657422513,
      "grad_norm": 2.1462085247039795,
      "learning_rate": 2.4787928221859708e-05,
      "loss": 0.292,
      "num_input_tokens_seen": 6582624,
      "step": 3040
    },
    {
      "epoch": 0.4967373572593801,
      "grad_norm": 0.92559814453125,
      "learning_rate": 2.4828711256117456e-05,
      "loss": 0.2414,
      "num_input_tokens_seen": 6595008,
      "step": 3045
    },
    {
      "epoch": 0.49755301794453505,
      "grad_norm": 3.88887882232666,
      "learning_rate": 2.4869494290375207e-05,
      "loss": 0.1398,
      "num_input_tokens_seen": 6606976,
      "step": 3050
    },
    {
      "epoch": 0.49836867862969003,
      "grad_norm": 4.913876056671143,
      "learning_rate": 2.4910277324632954e-05,
      "loss": 0.1024,
      "num_input_tokens_seen": 6617760,
      "step": 3055
    },
    {
      "epoch": 0.499184339314845,
      "grad_norm": 6.368566513061523,
      "learning_rate": 2.4951060358890702e-05,
      "loss": 0.1389,
      "num_input_tokens_seen": 6628896,
      "step": 3060
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.820760250091553,
      "learning_rate": 2.499184339314845e-05,
      "loss": 0.2959,
      "num_input_tokens_seen": 6639424,
      "step": 3065
    },
    {
      "epoch": 0.5,
      "eval_loss": 0.176969513297081,
      "eval_runtime": 133.1919,
      "eval_samples_per_second": 20.459,
      "eval_steps_per_second": 5.12,
      "num_input_tokens_seen": 6639424,
      "step": 3065
    },
    {
      "epoch": 0.5008156606851549,
      "grad_norm": 2.821260690689087,
      "learning_rate": 2.5032626427406204e-05,
      "loss": 0.2044,
      "num_input_tokens_seen": 6649280,
      "step": 3070
    },
    {
      "epoch": 0.50163132137031,
      "grad_norm": 0.5129310488700867,
      "learning_rate": 2.507340946166395e-05,
      "loss": 0.0932,
      "num_input_tokens_seen": 6659008,
      "step": 3075
    },
    {
      "epoch": 0.5024469820554649,
      "grad_norm": 5.1510539054870605,
      "learning_rate": 2.5114192495921696e-05,
      "loss": 0.1632,
      "num_input_tokens_seen": 6669312,
      "step": 3080
    },
    {
      "epoch": 0.5032626427406199,
      "grad_norm": 4.602128028869629,
      "learning_rate": 2.5154975530179447e-05,
      "loss": 0.303,
      "num_input_tokens_seen": 6679968,
      "step": 3085
    },
    {
      "epoch": 0.5040783034257749,
      "grad_norm": 5.982812404632568,
      "learning_rate": 2.5195758564437195e-05,
      "loss": 0.2011,
      "num_input_tokens_seen": 6689728,
      "step": 3090
    },
    {
      "epoch": 0.5048939641109299,
      "grad_norm": 4.853014945983887,
      "learning_rate": 2.5236541598694946e-05,
      "loss": 0.5461,
      "num_input_tokens_seen": 6700448,
      "step": 3095
    },
    {
      "epoch": 0.5057096247960848,
      "grad_norm": 3.3327314853668213,
      "learning_rate": 2.5277324632952693e-05,
      "loss": 0.1175,
      "num_input_tokens_seen": 6711424,
      "step": 3100
    },
    {
      "epoch": 0.5065252854812398,
      "grad_norm": 2.074833869934082,
      "learning_rate": 2.5318107667210438e-05,
      "loss": 0.0559,
      "num_input_tokens_seen": 6722752,
      "step": 3105
    },
    {
      "epoch": 0.5073409461663948,
      "grad_norm": 1.116888403892517,
      "learning_rate": 2.5358890701468192e-05,
      "loss": 0.3728,
      "num_input_tokens_seen": 6734080,
      "step": 3110
    },
    {
      "epoch": 0.5081566068515497,
      "grad_norm": 0.8630831241607666,
      "learning_rate": 2.5399673735725936e-05,
      "loss": 0.0812,
      "num_input_tokens_seen": 6744672,
      "step": 3115
    },
    {
      "epoch": 0.5089722675367048,
      "grad_norm": 5.993602275848389,
      "learning_rate": 2.544045676998369e-05,
      "loss": 0.2033,
      "num_input_tokens_seen": 6755104,
      "step": 3120
    },
    {
      "epoch": 0.5097879282218597,
      "grad_norm": 4.010820388793945,
      "learning_rate": 2.5481239804241435e-05,
      "loss": 0.1895,
      "num_input_tokens_seen": 6767168,
      "step": 3125
    },
    {
      "epoch": 0.5106035889070146,
      "grad_norm": 0.18873050808906555,
      "learning_rate": 2.552202283849919e-05,
      "loss": 0.2547,
      "num_input_tokens_seen": 6777472,
      "step": 3130
    },
    {
      "epoch": 0.5114192495921697,
      "grad_norm": 9.538300514221191,
      "learning_rate": 2.5562805872756934e-05,
      "loss": 0.2986,
      "num_input_tokens_seen": 6787776,
      "step": 3135
    },
    {
      "epoch": 0.5122349102773246,
      "grad_norm": 3.935617208480835,
      "learning_rate": 2.560358890701468e-05,
      "loss": 0.3441,
      "num_input_tokens_seen": 6799008,
      "step": 3140
    },
    {
      "epoch": 0.5130505709624796,
      "grad_norm": 0.1453946828842163,
      "learning_rate": 2.5644371941272432e-05,
      "loss": 0.0793,
      "num_input_tokens_seen": 6809440,
      "step": 3145
    },
    {
      "epoch": 0.5138662316476346,
      "grad_norm": 0.26558801531791687,
      "learning_rate": 2.568515497553018e-05,
      "loss": 0.0611,
      "num_input_tokens_seen": 6819456,
      "step": 3150
    },
    {
      "epoch": 0.5146818923327896,
      "grad_norm": 6.354449272155762,
      "learning_rate": 2.572593800978793e-05,
      "loss": 0.1466,
      "num_input_tokens_seen": 6829312,
      "step": 3155
    },
    {
      "epoch": 0.5154975530179445,
      "grad_norm": 3.53004789352417,
      "learning_rate": 2.576672104404568e-05,
      "loss": 0.223,
      "num_input_tokens_seen": 6839456,
      "step": 3160
    },
    {
      "epoch": 0.5163132137030995,
      "grad_norm": 0.7306339740753174,
      "learning_rate": 2.580750407830343e-05,
      "loss": 0.2472,
      "num_input_tokens_seen": 6850752,
      "step": 3165
    },
    {
      "epoch": 0.5171288743882545,
      "grad_norm": 1.7089464664459229,
      "learning_rate": 2.5848287112561177e-05,
      "loss": 0.1606,
      "num_input_tokens_seen": 6862144,
      "step": 3170
    },
    {
      "epoch": 0.5179445350734094,
      "grad_norm": 6.444469451904297,
      "learning_rate": 2.588907014681892e-05,
      "loss": 0.2617,
      "num_input_tokens_seen": 6872704,
      "step": 3175
    },
    {
      "epoch": 0.5187601957585645,
      "grad_norm": 7.05857515335083,
      "learning_rate": 2.5929853181076673e-05,
      "loss": 0.1505,
      "num_input_tokens_seen": 6883232,
      "step": 3180
    },
    {
      "epoch": 0.5195758564437194,
      "grad_norm": 5.446662425994873,
      "learning_rate": 2.597063621533442e-05,
      "loss": 0.3952,
      "num_input_tokens_seen": 6894368,
      "step": 3185
    },
    {
      "epoch": 0.5203915171288744,
      "grad_norm": 4.710870265960693,
      "learning_rate": 2.601141924959217e-05,
      "loss": 0.2189,
      "num_input_tokens_seen": 6906336,
      "step": 3190
    },
    {
      "epoch": 0.5212071778140294,
      "grad_norm": 0.2568054497241974,
      "learning_rate": 2.605220228384992e-05,
      "loss": 0.0819,
      "num_input_tokens_seen": 6917440,
      "step": 3195
    },
    {
      "epoch": 0.5220228384991843,
      "grad_norm": 4.117044448852539,
      "learning_rate": 2.609298531810767e-05,
      "loss": 0.4198,
      "num_input_tokens_seen": 6928736,
      "step": 3200
    },
    {
      "epoch": 0.5228384991843393,
      "grad_norm": 3.110626697540283,
      "learning_rate": 2.6133768352365418e-05,
      "loss": 0.1727,
      "num_input_tokens_seen": 6939488,
      "step": 3205
    },
    {
      "epoch": 0.5236541598694943,
      "grad_norm": 0.6773119568824768,
      "learning_rate": 2.6174551386623165e-05,
      "loss": 0.0335,
      "num_input_tokens_seen": 6950016,
      "step": 3210
    },
    {
      "epoch": 0.5244698205546493,
      "grad_norm": 5.615257263183594,
      "learning_rate": 2.6215334420880916e-05,
      "loss": 0.2907,
      "num_input_tokens_seen": 6961280,
      "step": 3215
    },
    {
      "epoch": 0.5252854812398042,
      "grad_norm": 0.9965311288833618,
      "learning_rate": 2.6256117455138664e-05,
      "loss": 0.0736,
      "num_input_tokens_seen": 6971360,
      "step": 3220
    },
    {
      "epoch": 0.5261011419249593,
      "grad_norm": 4.916421413421631,
      "learning_rate": 2.6296900489396415e-05,
      "loss": 0.4214,
      "num_input_tokens_seen": 6982528,
      "step": 3225
    },
    {
      "epoch": 0.5269168026101142,
      "grad_norm": 1.3702571392059326,
      "learning_rate": 2.633768352365416e-05,
      "loss": 0.063,
      "num_input_tokens_seen": 6992704,
      "step": 3230
    },
    {
      "epoch": 0.5277324632952691,
      "grad_norm": 4.162288665771484,
      "learning_rate": 2.6378466557911907e-05,
      "loss": 0.2509,
      "num_input_tokens_seen": 7003072,
      "step": 3235
    },
    {
      "epoch": 0.5285481239804242,
      "grad_norm": 2.886589527130127,
      "learning_rate": 2.6419249592169658e-05,
      "loss": 0.1931,
      "num_input_tokens_seen": 7013728,
      "step": 3240
    },
    {
      "epoch": 0.5293637846655791,
      "grad_norm": 2.574683904647827,
      "learning_rate": 2.6460032626427406e-05,
      "loss": 0.0575,
      "num_input_tokens_seen": 7024256,
      "step": 3245
    },
    {
      "epoch": 0.5301794453507341,
      "grad_norm": 3.243692636489868,
      "learning_rate": 2.6500815660685157e-05,
      "loss": 0.4423,
      "num_input_tokens_seen": 7034688,
      "step": 3250
    },
    {
      "epoch": 0.5309951060358891,
      "grad_norm": 4.145303249359131,
      "learning_rate": 2.6541598694942904e-05,
      "loss": 0.1589,
      "num_input_tokens_seen": 7046080,
      "step": 3255
    },
    {
      "epoch": 0.531810766721044,
      "grad_norm": 5.3809027671813965,
      "learning_rate": 2.6582381729200655e-05,
      "loss": 0.2613,
      "num_input_tokens_seen": 7056032,
      "step": 3260
    },
    {
      "epoch": 0.532626427406199,
      "grad_norm": 3.7870163917541504,
      "learning_rate": 2.6623164763458403e-05,
      "loss": 0.2471,
      "num_input_tokens_seen": 7066880,
      "step": 3265
    },
    {
      "epoch": 0.533442088091354,
      "grad_norm": 5.299981594085693,
      "learning_rate": 2.666394779771615e-05,
      "loss": 0.3057,
      "num_input_tokens_seen": 7077664,
      "step": 3270
    },
    {
      "epoch": 0.534257748776509,
      "grad_norm": 6.679792404174805,
      "learning_rate": 2.67047308319739e-05,
      "loss": 0.1684,
      "num_input_tokens_seen": 7088928,
      "step": 3275
    },
    {
      "epoch": 0.5350734094616639,
      "grad_norm": 0.7039870023727417,
      "learning_rate": 2.6745513866231646e-05,
      "loss": 0.2108,
      "num_input_tokens_seen": 7100800,
      "step": 3280
    },
    {
      "epoch": 0.535889070146819,
      "grad_norm": 2.9606547355651855,
      "learning_rate": 2.67862969004894e-05,
      "loss": 0.0965,
      "num_input_tokens_seen": 7112160,
      "step": 3285
    },
    {
      "epoch": 0.5367047308319739,
      "grad_norm": 1.262087106704712,
      "learning_rate": 2.6827079934747145e-05,
      "loss": 0.1558,
      "num_input_tokens_seen": 7122848,
      "step": 3290
    },
    {
      "epoch": 0.5375203915171288,
      "grad_norm": 3.0593302249908447,
      "learning_rate": 2.68678629690049e-05,
      "loss": 0.1694,
      "num_input_tokens_seen": 7132800,
      "step": 3295
    },
    {
      "epoch": 0.5383360522022839,
      "grad_norm": 0.8122572898864746,
      "learning_rate": 2.6908646003262643e-05,
      "loss": 0.1244,
      "num_input_tokens_seen": 7144064,
      "step": 3300
    },
    {
      "epoch": 0.5391517128874388,
      "grad_norm": 1.4844942092895508,
      "learning_rate": 2.694942903752039e-05,
      "loss": 0.1656,
      "num_input_tokens_seen": 7155744,
      "step": 3305
    },
    {
      "epoch": 0.5399673735725938,
      "grad_norm": 0.482521653175354,
      "learning_rate": 2.6990212071778142e-05,
      "loss": 0.0457,
      "num_input_tokens_seen": 7166336,
      "step": 3310
    },
    {
      "epoch": 0.5407830342577488,
      "grad_norm": 0.0839756652712822,
      "learning_rate": 2.703099510603589e-05,
      "loss": 0.1977,
      "num_input_tokens_seen": 7176832,
      "step": 3315
    },
    {
      "epoch": 0.5415986949429038,
      "grad_norm": 0.9602549076080322,
      "learning_rate": 2.707177814029364e-05,
      "loss": 0.1921,
      "num_input_tokens_seen": 7187616,
      "step": 3320
    },
    {
      "epoch": 0.5424143556280587,
      "grad_norm": 0.056319817900657654,
      "learning_rate": 2.7112561174551388e-05,
      "loss": 0.1743,
      "num_input_tokens_seen": 7198144,
      "step": 3325
    },
    {
      "epoch": 0.5432300163132137,
      "grad_norm": 2.8947079181671143,
      "learning_rate": 2.715334420880914e-05,
      "loss": 0.0901,
      "num_input_tokens_seen": 7208960,
      "step": 3330
    },
    {
      "epoch": 0.5440456769983687,
      "grad_norm": 0.40086036920547485,
      "learning_rate": 2.7194127243066887e-05,
      "loss": 0.2511,
      "num_input_tokens_seen": 7220064,
      "step": 3335
    },
    {
      "epoch": 0.5448613376835236,
      "grad_norm": 0.3649408519268036,
      "learning_rate": 2.723491027732463e-05,
      "loss": 0.1671,
      "num_input_tokens_seen": 7231008,
      "step": 3340
    },
    {
      "epoch": 0.5456769983686787,
      "grad_norm": 2.579338550567627,
      "learning_rate": 2.7275693311582386e-05,
      "loss": 0.3123,
      "num_input_tokens_seen": 7242144,
      "step": 3345
    },
    {
      "epoch": 0.5464926590538336,
      "grad_norm": 4.868326663970947,
      "learning_rate": 2.731647634584013e-05,
      "loss": 0.2258,
      "num_input_tokens_seen": 7252448,
      "step": 3350
    },
    {
      "epoch": 0.5473083197389886,
      "grad_norm": 0.2523765563964844,
      "learning_rate": 2.7357259380097884e-05,
      "loss": 0.1712,
      "num_input_tokens_seen": 7261856,
      "step": 3355
    },
    {
      "epoch": 0.5481239804241436,
      "grad_norm": 0.5494678020477295,
      "learning_rate": 2.739804241435563e-05,
      "loss": 0.1164,
      "num_input_tokens_seen": 7272800,
      "step": 3360
    },
    {
      "epoch": 0.5489396411092985,
      "grad_norm": 4.799815654754639,
      "learning_rate": 2.7438825448613376e-05,
      "loss": 0.2332,
      "num_input_tokens_seen": 7283808,
      "step": 3365
    },
    {
      "epoch": 0.5497553017944535,
      "grad_norm": 0.9906727075576782,
      "learning_rate": 2.7479608482871127e-05,
      "loss": 0.0323,
      "num_input_tokens_seen": 7294976,
      "step": 3370
    },
    {
      "epoch": 0.5505709624796085,
      "grad_norm": 2.8852052688598633,
      "learning_rate": 2.7520391517128875e-05,
      "loss": 0.1403,
      "num_input_tokens_seen": 7306560,
      "step": 3375
    },
    {
      "epoch": 0.5513866231647635,
      "grad_norm": 0.6081190705299377,
      "learning_rate": 2.7561174551386626e-05,
      "loss": 0.0219,
      "num_input_tokens_seen": 7317600,
      "step": 3380
    },
    {
      "epoch": 0.5522022838499184,
      "grad_norm": 3.466200590133667,
      "learning_rate": 2.7601957585644373e-05,
      "loss": 0.0817,
      "num_input_tokens_seen": 7330144,
      "step": 3385
    },
    {
      "epoch": 0.5530179445350734,
      "grad_norm": 8.591158866882324,
      "learning_rate": 2.7642740619902125e-05,
      "loss": 0.2721,
      "num_input_tokens_seen": 7341248,
      "step": 3390
    },
    {
      "epoch": 0.5538336052202284,
      "grad_norm": 4.575144290924072,
      "learning_rate": 2.7683523654159872e-05,
      "loss": 0.2259,
      "num_input_tokens_seen": 7351200,
      "step": 3395
    },
    {
      "epoch": 0.5546492659053833,
      "grad_norm": 0.4166700839996338,
      "learning_rate": 2.7724306688417616e-05,
      "loss": 0.0269,
      "num_input_tokens_seen": 7361792,
      "step": 3400
    },
    {
      "epoch": 0.5554649265905384,
      "grad_norm": 4.073925018310547,
      "learning_rate": 2.776508972267537e-05,
      "loss": 0.2099,
      "num_input_tokens_seen": 7372160,
      "step": 3405
    },
    {
      "epoch": 0.5562805872756933,
      "grad_norm": 0.2443922460079193,
      "learning_rate": 2.7805872756933115e-05,
      "loss": 0.06,
      "num_input_tokens_seen": 7381408,
      "step": 3410
    },
    {
      "epoch": 0.5570962479608483,
      "grad_norm": 7.976419925689697,
      "learning_rate": 2.7846655791190866e-05,
      "loss": 0.2754,
      "num_input_tokens_seen": 7392512,
      "step": 3415
    },
    {
      "epoch": 0.5579119086460033,
      "grad_norm": 4.312510967254639,
      "learning_rate": 2.7887438825448614e-05,
      "loss": 0.3246,
      "num_input_tokens_seen": 7404192,
      "step": 3420
    },
    {
      "epoch": 0.5587275693311582,
      "grad_norm": 0.02727222442626953,
      "learning_rate": 2.7928221859706365e-05,
      "loss": 0.1545,
      "num_input_tokens_seen": 7415008,
      "step": 3425
    },
    {
      "epoch": 0.5595432300163132,
      "grad_norm": 2.2358570098876953,
      "learning_rate": 2.7969004893964112e-05,
      "loss": 0.0869,
      "num_input_tokens_seen": 7425088,
      "step": 3430
    },
    {
      "epoch": 0.5603588907014682,
      "grad_norm": 3.2919957637786865,
      "learning_rate": 2.800978792822186e-05,
      "loss": 0.3297,
      "num_input_tokens_seen": 7436608,
      "step": 3435
    },
    {
      "epoch": 0.5611745513866232,
      "grad_norm": 0.15448737144470215,
      "learning_rate": 2.805057096247961e-05,
      "loss": 0.1976,
      "num_input_tokens_seen": 7446784,
      "step": 3440
    },
    {
      "epoch": 0.5619902120717781,
      "grad_norm": 0.3430439233779907,
      "learning_rate": 2.809135399673736e-05,
      "loss": 0.1111,
      "num_input_tokens_seen": 7456640,
      "step": 3445
    },
    {
      "epoch": 0.5628058727569332,
      "grad_norm": 0.5385096669197083,
      "learning_rate": 2.813213703099511e-05,
      "loss": 0.1322,
      "num_input_tokens_seen": 7468096,
      "step": 3450
    },
    {
      "epoch": 0.5636215334420881,
      "grad_norm": 3.0548486709594727,
      "learning_rate": 2.8172920065252857e-05,
      "loss": 0.2489,
      "num_input_tokens_seen": 7480096,
      "step": 3455
    },
    {
      "epoch": 0.564437194127243,
      "grad_norm": 0.20929400622844696,
      "learning_rate": 2.8213703099510602e-05,
      "loss": 0.1348,
      "num_input_tokens_seen": 7491136,
      "step": 3460
    },
    {
      "epoch": 0.5652528548123981,
      "grad_norm": 0.22163662314414978,
      "learning_rate": 2.8254486133768353e-05,
      "loss": 0.1038,
      "num_input_tokens_seen": 7502304,
      "step": 3465
    },
    {
      "epoch": 0.566068515497553,
      "grad_norm": 0.5673784017562866,
      "learning_rate": 2.82952691680261e-05,
      "loss": 0.2097,
      "num_input_tokens_seen": 7512928,
      "step": 3470
    },
    {
      "epoch": 0.566884176182708,
      "grad_norm": 0.227246955037117,
      "learning_rate": 2.833605220228385e-05,
      "loss": 0.2233,
      "num_input_tokens_seen": 7523904,
      "step": 3475
    },
    {
      "epoch": 0.567699836867863,
      "grad_norm": 5.121859073638916,
      "learning_rate": 2.83768352365416e-05,
      "loss": 0.4761,
      "num_input_tokens_seen": 7534496,
      "step": 3480
    },
    {
      "epoch": 0.5685154975530179,
      "grad_norm": 5.115076065063477,
      "learning_rate": 2.841761827079935e-05,
      "loss": 0.4245,
      "num_input_tokens_seen": 7545120,
      "step": 3485
    },
    {
      "epoch": 0.5693311582381729,
      "grad_norm": 3.4532887935638428,
      "learning_rate": 2.8458401305057098e-05,
      "loss": 0.0717,
      "num_input_tokens_seen": 7555712,
      "step": 3490
    },
    {
      "epoch": 0.5701468189233279,
      "grad_norm": 0.19635164737701416,
      "learning_rate": 2.8499184339314845e-05,
      "loss": 0.0957,
      "num_input_tokens_seen": 7567200,
      "step": 3495
    },
    {
      "epoch": 0.5709624796084829,
      "grad_norm": 5.896024227142334,
      "learning_rate": 2.8539967373572596e-05,
      "loss": 0.3407,
      "num_input_tokens_seen": 7578464,
      "step": 3500
    },
    {
      "epoch": 0.5717781402936378,
      "grad_norm": 1.1312222480773926,
      "learning_rate": 2.858075040783034e-05,
      "loss": 0.1151,
      "num_input_tokens_seen": 7588256,
      "step": 3505
    },
    {
      "epoch": 0.5725938009787929,
      "grad_norm": 1.881733775138855,
      "learning_rate": 2.8621533442088095e-05,
      "loss": 0.302,
      "num_input_tokens_seen": 7598848,
      "step": 3510
    },
    {
      "epoch": 0.5734094616639478,
      "grad_norm": 1.0623258352279663,
      "learning_rate": 2.866231647634584e-05,
      "loss": 0.0913,
      "num_input_tokens_seen": 7609696,
      "step": 3515
    },
    {
      "epoch": 0.5742251223491027,
      "grad_norm": 1.1519025564193726,
      "learning_rate": 2.8703099510603594e-05,
      "loss": 0.0698,
      "num_input_tokens_seen": 7620864,
      "step": 3520
    },
    {
      "epoch": 0.5750407830342578,
      "grad_norm": 4.33655309677124,
      "learning_rate": 2.8743882544861338e-05,
      "loss": 0.0705,
      "num_input_tokens_seen": 7631392,
      "step": 3525
    },
    {
      "epoch": 0.5758564437194127,
      "grad_norm": 0.47238653898239136,
      "learning_rate": 2.8784665579119086e-05,
      "loss": 0.0643,
      "num_input_tokens_seen": 7643232,
      "step": 3530
    },
    {
      "epoch": 0.5766721044045677,
      "grad_norm": 0.4578634798526764,
      "learning_rate": 2.8825448613376837e-05,
      "loss": 0.1905,
      "num_input_tokens_seen": 7654944,
      "step": 3535
    },
    {
      "epoch": 0.5774877650897227,
      "grad_norm": 4.807013988494873,
      "learning_rate": 2.8866231647634584e-05,
      "loss": 0.1149,
      "num_input_tokens_seen": 7665760,
      "step": 3540
    },
    {
      "epoch": 0.5783034257748777,
      "grad_norm": 0.3435484766960144,
      "learning_rate": 2.8907014681892335e-05,
      "loss": 0.041,
      "num_input_tokens_seen": 7676576,
      "step": 3545
    },
    {
      "epoch": 0.5791190864600326,
      "grad_norm": 1.3135573863983154,
      "learning_rate": 2.8947797716150083e-05,
      "loss": 0.2718,
      "num_input_tokens_seen": 7687104,
      "step": 3550
    },
    {
      "epoch": 0.5799347471451876,
      "grad_norm": 5.1496195793151855,
      "learning_rate": 2.8988580750407834e-05,
      "loss": 0.0551,
      "num_input_tokens_seen": 7698080,
      "step": 3555
    },
    {
      "epoch": 0.5807504078303426,
      "grad_norm": 3.7905707359313965,
      "learning_rate": 2.9029363784665582e-05,
      "loss": 0.2317,
      "num_input_tokens_seen": 7709632,
      "step": 3560
    },
    {
      "epoch": 0.5815660685154975,
      "grad_norm": 0.5441896915435791,
      "learning_rate": 2.9070146818923326e-05,
      "loss": 0.0329,
      "num_input_tokens_seen": 7719936,
      "step": 3565
    },
    {
      "epoch": 0.5823817292006526,
      "grad_norm": 0.4525549113750458,
      "learning_rate": 2.911092985318108e-05,
      "loss": 0.0733,
      "num_input_tokens_seen": 7730080,
      "step": 3570
    },
    {
      "epoch": 0.5831973898858075,
      "grad_norm": 4.433751583099365,
      "learning_rate": 2.9151712887438825e-05,
      "loss": 0.2059,
      "num_input_tokens_seen": 7740576,
      "step": 3575
    },
    {
      "epoch": 0.5840130505709625,
      "grad_norm": 0.03426478058099747,
      "learning_rate": 2.919249592169658e-05,
      "loss": 0.1512,
      "num_input_tokens_seen": 7751040,
      "step": 3580
    },
    {
      "epoch": 0.5848287112561175,
      "grad_norm": 0.027612321078777313,
      "learning_rate": 2.9233278955954323e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 7760768,
      "step": 3585
    },
    {
      "epoch": 0.5856443719412724,
      "grad_norm": 0.19678345322608948,
      "learning_rate": 2.927406199021207e-05,
      "loss": 0.0735,
      "num_input_tokens_seen": 7771680,
      "step": 3590
    },
    {
      "epoch": 0.5864600326264274,
      "grad_norm": 0.29991891980171204,
      "learning_rate": 2.9314845024469822e-05,
      "loss": 0.1521,
      "num_input_tokens_seen": 7783072,
      "step": 3595
    },
    {
      "epoch": 0.5872756933115824,
      "grad_norm": 0.11637599021196365,
      "learning_rate": 2.935562805872757e-05,
      "loss": 0.082,
      "num_input_tokens_seen": 7792608,
      "step": 3600
    },
    {
      "epoch": 0.5880913539967374,
      "grad_norm": 0.23875682055950165,
      "learning_rate": 2.939641109298532e-05,
      "loss": 0.3333,
      "num_input_tokens_seen": 7803744,
      "step": 3605
    },
    {
      "epoch": 0.5889070146818923,
      "grad_norm": 0.7318027019500732,
      "learning_rate": 2.943719412724307e-05,
      "loss": 0.1406,
      "num_input_tokens_seen": 7815008,
      "step": 3610
    },
    {
      "epoch": 0.5897226753670473,
      "grad_norm": 3.634417772293091,
      "learning_rate": 2.947797716150082e-05,
      "loss": 0.2403,
      "num_input_tokens_seen": 7825536,
      "step": 3615
    },
    {
      "epoch": 0.5905383360522023,
      "grad_norm": 0.05685891583561897,
      "learning_rate": 2.9518760195758567e-05,
      "loss": 0.2194,
      "num_input_tokens_seen": 7836160,
      "step": 3620
    },
    {
      "epoch": 0.5913539967373572,
      "grad_norm": 1.4172544479370117,
      "learning_rate": 2.955954323001631e-05,
      "loss": 0.0216,
      "num_input_tokens_seen": 7848864,
      "step": 3625
    },
    {
      "epoch": 0.5921696574225123,
      "grad_norm": 8.229720115661621,
      "learning_rate": 2.9600326264274066e-05,
      "loss": 0.2237,
      "num_input_tokens_seen": 7859488,
      "step": 3630
    },
    {
      "epoch": 0.5929853181076672,
      "grad_norm": 2.7573013305664062,
      "learning_rate": 2.964110929853181e-05,
      "loss": 0.1209,
      "num_input_tokens_seen": 7869600,
      "step": 3635
    },
    {
      "epoch": 0.5938009787928222,
      "grad_norm": 4.573496341705322,
      "learning_rate": 2.968189233278956e-05,
      "loss": 0.2684,
      "num_input_tokens_seen": 7879840,
      "step": 3640
    },
    {
      "epoch": 0.5946166394779772,
      "grad_norm": 0.07046195864677429,
      "learning_rate": 2.972267536704731e-05,
      "loss": 0.0752,
      "num_input_tokens_seen": 7890272,
      "step": 3645
    },
    {
      "epoch": 0.5954323001631321,
      "grad_norm": 9.139192581176758,
      "learning_rate": 2.976345840130506e-05,
      "loss": 0.3903,
      "num_input_tokens_seen": 7901120,
      "step": 3650
    },
    {
      "epoch": 0.5962479608482871,
      "grad_norm": 2.8443400859832764,
      "learning_rate": 2.9804241435562807e-05,
      "loss": 0.0952,
      "num_input_tokens_seen": 7912608,
      "step": 3655
    },
    {
      "epoch": 0.5970636215334421,
      "grad_norm": 2.2369651794433594,
      "learning_rate": 2.9845024469820555e-05,
      "loss": 0.2448,
      "num_input_tokens_seen": 7923936,
      "step": 3660
    },
    {
      "epoch": 0.5978792822185971,
      "grad_norm": 4.761455535888672,
      "learning_rate": 2.9885807504078306e-05,
      "loss": 0.352,
      "num_input_tokens_seen": 7934496,
      "step": 3665
    },
    {
      "epoch": 0.598694942903752,
      "grad_norm": 3.5061333179473877,
      "learning_rate": 2.9926590538336054e-05,
      "loss": 0.1885,
      "num_input_tokens_seen": 7945728,
      "step": 3670
    },
    {
      "epoch": 0.5995106035889071,
      "grad_norm": 0.19609789550304413,
      "learning_rate": 2.9967373572593805e-05,
      "loss": 0.0977,
      "num_input_tokens_seen": 7956960,
      "step": 3675
    },
    {
      "epoch": 0.600326264274062,
      "grad_norm": 5.6819915771484375,
      "learning_rate": 3.0008156606851552e-05,
      "loss": 0.1786,
      "num_input_tokens_seen": 7969440,
      "step": 3680
    },
    {
      "epoch": 0.6011419249592169,
      "grad_norm": 4.5360798835754395,
      "learning_rate": 3.0048939641109303e-05,
      "loss": 0.1381,
      "num_input_tokens_seen": 7978560,
      "step": 3685
    },
    {
      "epoch": 0.601957585644372,
      "grad_norm": 2.839482307434082,
      "learning_rate": 3.0089722675367048e-05,
      "loss": 0.2044,
      "num_input_tokens_seen": 7990400,
      "step": 3690
    },
    {
      "epoch": 0.6027732463295269,
      "grad_norm": 5.692098617553711,
      "learning_rate": 3.0130505709624795e-05,
      "loss": 0.1635,
      "num_input_tokens_seen": 8000128,
      "step": 3695
    },
    {
      "epoch": 0.6035889070146819,
      "grad_norm": 0.029796523973345757,
      "learning_rate": 3.0171288743882546e-05,
      "loss": 0.1286,
      "num_input_tokens_seen": 8010592,
      "step": 3700
    },
    {
      "epoch": 0.6044045676998369,
      "grad_norm": 1.200098991394043,
      "learning_rate": 3.0212071778140294e-05,
      "loss": 0.1984,
      "num_input_tokens_seen": 8022240,
      "step": 3705
    },
    {
      "epoch": 0.6052202283849919,
      "grad_norm": 0.27159032225608826,
      "learning_rate": 3.0252854812398045e-05,
      "loss": 0.1891,
      "num_input_tokens_seen": 8032096,
      "step": 3710
    },
    {
      "epoch": 0.6060358890701468,
      "grad_norm": 7.265265941619873,
      "learning_rate": 3.0293637846655793e-05,
      "loss": 0.2965,
      "num_input_tokens_seen": 8042400,
      "step": 3715
    },
    {
      "epoch": 0.6068515497553018,
      "grad_norm": 0.26513463258743286,
      "learning_rate": 3.033442088091354e-05,
      "loss": 0.085,
      "num_input_tokens_seen": 8052672,
      "step": 3720
    },
    {
      "epoch": 0.6076672104404568,
      "grad_norm": 2.2898662090301514,
      "learning_rate": 3.037520391517129e-05,
      "loss": 0.1487,
      "num_input_tokens_seen": 8063712,
      "step": 3725
    },
    {
      "epoch": 0.6084828711256117,
      "grad_norm": 3.571632146835327,
      "learning_rate": 3.041598694942904e-05,
      "loss": 0.1305,
      "num_input_tokens_seen": 8073600,
      "step": 3730
    },
    {
      "epoch": 0.6092985318107668,
      "grad_norm": 0.5955801606178284,
      "learning_rate": 3.045676998368679e-05,
      "loss": 0.2931,
      "num_input_tokens_seen": 8085664,
      "step": 3735
    },
    {
      "epoch": 0.6101141924959217,
      "grad_norm": 4.0143351554870605,
      "learning_rate": 3.0497553017944534e-05,
      "loss": 0.161,
      "num_input_tokens_seen": 8097152,
      "step": 3740
    },
    {
      "epoch": 0.6109298531810766,
      "grad_norm": 3.4268391132354736,
      "learning_rate": 3.0538336052202285e-05,
      "loss": 0.0641,
      "num_input_tokens_seen": 8107648,
      "step": 3745
    },
    {
      "epoch": 0.6117455138662317,
      "grad_norm": 4.198644638061523,
      "learning_rate": 3.057911908646003e-05,
      "loss": 0.3436,
      "num_input_tokens_seen": 8118272,
      "step": 3750
    },
    {
      "epoch": 0.6125611745513866,
      "grad_norm": 4.588216304779053,
      "learning_rate": 3.061990212071778e-05,
      "loss": 0.1209,
      "num_input_tokens_seen": 8129120,
      "step": 3755
    },
    {
      "epoch": 0.6133768352365416,
      "grad_norm": 0.33906522393226624,
      "learning_rate": 3.0660685154975535e-05,
      "loss": 0.1412,
      "num_input_tokens_seen": 8139200,
      "step": 3760
    },
    {
      "epoch": 0.6141924959216966,
      "grad_norm": 0.1643534004688263,
      "learning_rate": 3.0701468189233276e-05,
      "loss": 0.0695,
      "num_input_tokens_seen": 8150272,
      "step": 3765
    },
    {
      "epoch": 0.6150081566068516,
      "grad_norm": 5.936880588531494,
      "learning_rate": 3.074225122349103e-05,
      "loss": 0.3042,
      "num_input_tokens_seen": 8162496,
      "step": 3770
    },
    {
      "epoch": 0.6158238172920065,
      "grad_norm": 1.3060961961746216,
      "learning_rate": 3.078303425774878e-05,
      "loss": 0.1619,
      "num_input_tokens_seen": 8173664,
      "step": 3775
    },
    {
      "epoch": 0.6166394779771615,
      "grad_norm": 0.3549754321575165,
      "learning_rate": 3.082381729200653e-05,
      "loss": 0.2705,
      "num_input_tokens_seen": 8184512,
      "step": 3780
    },
    {
      "epoch": 0.6174551386623165,
      "grad_norm": 5.725491523742676,
      "learning_rate": 3.086460032626427e-05,
      "loss": 0.2719,
      "num_input_tokens_seen": 8194432,
      "step": 3785
    },
    {
      "epoch": 0.6182707993474714,
      "grad_norm": 5.004452705383301,
      "learning_rate": 3.090538336052202e-05,
      "loss": 0.095,
      "num_input_tokens_seen": 8205152,
      "step": 3790
    },
    {
      "epoch": 0.6190864600326265,
      "grad_norm": 0.28644075989723206,
      "learning_rate": 3.0946166394779775e-05,
      "loss": 0.1491,
      "num_input_tokens_seen": 8215360,
      "step": 3795
    },
    {
      "epoch": 0.6199021207177814,
      "grad_norm": 0.6890418529510498,
      "learning_rate": 3.098694942903752e-05,
      "loss": 0.0998,
      "num_input_tokens_seen": 8225216,
      "step": 3800
    },
    {
      "epoch": 0.6207177814029364,
      "grad_norm": 6.75236701965332,
      "learning_rate": 3.102773246329527e-05,
      "loss": 0.1457,
      "num_input_tokens_seen": 8235488,
      "step": 3805
    },
    {
      "epoch": 0.6215334420880914,
      "grad_norm": 1.7343748807907104,
      "learning_rate": 3.106851549755302e-05,
      "loss": 0.1449,
      "num_input_tokens_seen": 8246208,
      "step": 3810
    },
    {
      "epoch": 0.6223491027732463,
      "grad_norm": 2.9255847930908203,
      "learning_rate": 3.1109298531810766e-05,
      "loss": 0.1294,
      "num_input_tokens_seen": 8258048,
      "step": 3815
    },
    {
      "epoch": 0.6231647634584013,
      "grad_norm": 5.216608047485352,
      "learning_rate": 3.115008156606852e-05,
      "loss": 0.0629,
      "num_input_tokens_seen": 8269472,
      "step": 3820
    },
    {
      "epoch": 0.6239804241435563,
      "grad_norm": 7.773916721343994,
      "learning_rate": 3.119086460032626e-05,
      "loss": 0.4178,
      "num_input_tokens_seen": 8280512,
      "step": 3825
    },
    {
      "epoch": 0.6247960848287113,
      "grad_norm": 0.6229372620582581,
      "learning_rate": 3.1231647634584016e-05,
      "loss": 0.1706,
      "num_input_tokens_seen": 8291360,
      "step": 3830
    },
    {
      "epoch": 0.6256117455138662,
      "grad_norm": 0.06130954995751381,
      "learning_rate": 3.127243066884176e-05,
      "loss": 0.0877,
      "num_input_tokens_seen": 8301440,
      "step": 3835
    },
    {
      "epoch": 0.6264274061990212,
      "grad_norm": 5.62656831741333,
      "learning_rate": 3.131321370309952e-05,
      "loss": 0.3183,
      "num_input_tokens_seen": 8313472,
      "step": 3840
    },
    {
      "epoch": 0.6272430668841762,
      "grad_norm": 0.0833306834101677,
      "learning_rate": 3.135399673735726e-05,
      "loss": 0.0129,
      "num_input_tokens_seen": 8323488,
      "step": 3845
    },
    {
      "epoch": 0.6280587275693311,
      "grad_norm": 0.41827625036239624,
      "learning_rate": 3.1394779771615006e-05,
      "loss": 0.0189,
      "num_input_tokens_seen": 8336096,
      "step": 3850
    },
    {
      "epoch": 0.6288743882544862,
      "grad_norm": 6.341390132904053,
      "learning_rate": 3.143556280587276e-05,
      "loss": 0.2367,
      "num_input_tokens_seen": 8344736,
      "step": 3855
    },
    {
      "epoch": 0.6296900489396411,
      "grad_norm": 0.191243514418602,
      "learning_rate": 3.147634584013051e-05,
      "loss": 0.0972,
      "num_input_tokens_seen": 8356416,
      "step": 3860
    },
    {
      "epoch": 0.6305057096247961,
      "grad_norm": 0.08787572383880615,
      "learning_rate": 3.1517128874388256e-05,
      "loss": 0.1867,
      "num_input_tokens_seen": 8367712,
      "step": 3865
    },
    {
      "epoch": 0.6313213703099511,
      "grad_norm": 0.05379727482795715,
      "learning_rate": 3.1557911908646004e-05,
      "loss": 0.2884,
      "num_input_tokens_seen": 8378752,
      "step": 3870
    },
    {
      "epoch": 0.632137030995106,
      "grad_norm": 2.846280097961426,
      "learning_rate": 3.159869494290376e-05,
      "loss": 0.2594,
      "num_input_tokens_seen": 8389472,
      "step": 3875
    },
    {
      "epoch": 0.632952691680261,
      "grad_norm": 0.4946746528148651,
      "learning_rate": 3.1639477977161506e-05,
      "loss": 0.1907,
      "num_input_tokens_seen": 8400064,
      "step": 3880
    },
    {
      "epoch": 0.633768352365416,
      "grad_norm": 1.0731377601623535,
      "learning_rate": 3.1680261011419246e-05,
      "loss": 0.1228,
      "num_input_tokens_seen": 8410176,
      "step": 3885
    },
    {
      "epoch": 0.634584013050571,
      "grad_norm": 2.3902738094329834,
      "learning_rate": 3.1721044045677e-05,
      "loss": 0.4067,
      "num_input_tokens_seen": 8420256,
      "step": 3890
    },
    {
      "epoch": 0.6353996737357259,
      "grad_norm": 0.821153461933136,
      "learning_rate": 3.176182707993475e-05,
      "loss": 0.1574,
      "num_input_tokens_seen": 8431808,
      "step": 3895
    },
    {
      "epoch": 0.636215334420881,
      "grad_norm": 1.441624641418457,
      "learning_rate": 3.1802610114192496e-05,
      "loss": 0.0372,
      "num_input_tokens_seen": 8444064,
      "step": 3900
    },
    {
      "epoch": 0.6370309951060359,
      "grad_norm": 3.8657689094543457,
      "learning_rate": 3.1843393148450244e-05,
      "loss": 0.2786,
      "num_input_tokens_seen": 8454944,
      "step": 3905
    },
    {
      "epoch": 0.6378466557911908,
      "grad_norm": 0.2954348921775818,
      "learning_rate": 3.1884176182708e-05,
      "loss": 0.0424,
      "num_input_tokens_seen": 8466208,
      "step": 3910
    },
    {
      "epoch": 0.6386623164763459,
      "grad_norm": 0.3389508128166199,
      "learning_rate": 3.1924959216965746e-05,
      "loss": 0.1062,
      "num_input_tokens_seen": 8476480,
      "step": 3915
    },
    {
      "epoch": 0.6394779771615008,
      "grad_norm": 4.907018661499023,
      "learning_rate": 3.1965742251223494e-05,
      "loss": 0.1585,
      "num_input_tokens_seen": 8486688,
      "step": 3920
    },
    {
      "epoch": 0.6402936378466558,
      "grad_norm": 0.17697042226791382,
      "learning_rate": 3.200652528548124e-05,
      "loss": 0.1982,
      "num_input_tokens_seen": 8499200,
      "step": 3925
    },
    {
      "epoch": 0.6411092985318108,
      "grad_norm": 1.9519888162612915,
      "learning_rate": 3.204730831973899e-05,
      "loss": 0.1257,
      "num_input_tokens_seen": 8509728,
      "step": 3930
    },
    {
      "epoch": 0.6419249592169658,
      "grad_norm": 7.020340442657471,
      "learning_rate": 3.208809135399674e-05,
      "loss": 0.1737,
      "num_input_tokens_seen": 8520928,
      "step": 3935
    },
    {
      "epoch": 0.6427406199021207,
      "grad_norm": 0.11750974506139755,
      "learning_rate": 3.2128874388254484e-05,
      "loss": 0.1452,
      "num_input_tokens_seen": 8532448,
      "step": 3940
    },
    {
      "epoch": 0.6435562805872757,
      "grad_norm": 0.19156263768672943,
      "learning_rate": 3.216965742251223e-05,
      "loss": 0.1432,
      "num_input_tokens_seen": 8543360,
      "step": 3945
    },
    {
      "epoch": 0.6443719412724307,
      "grad_norm": 1.1783024072647095,
      "learning_rate": 3.2210440456769986e-05,
      "loss": 0.3555,
      "num_input_tokens_seen": 8554176,
      "step": 3950
    },
    {
      "epoch": 0.6451876019575856,
      "grad_norm": 3.0158650875091553,
      "learning_rate": 3.2251223491027734e-05,
      "loss": 0.1813,
      "num_input_tokens_seen": 8565376,
      "step": 3955
    },
    {
      "epoch": 0.6460032626427407,
      "grad_norm": 0.21972206234931946,
      "learning_rate": 3.229200652528548e-05,
      "loss": 0.333,
      "num_input_tokens_seen": 8576480,
      "step": 3960
    },
    {
      "epoch": 0.6468189233278956,
      "grad_norm": 0.046773120760917664,
      "learning_rate": 3.233278955954323e-05,
      "loss": 0.1075,
      "num_input_tokens_seen": 8586944,
      "step": 3965
    },
    {
      "epoch": 0.6476345840130505,
      "grad_norm": 0.25712352991104126,
      "learning_rate": 3.2373572593800984e-05,
      "loss": 0.1852,
      "num_input_tokens_seen": 8596640,
      "step": 3970
    },
    {
      "epoch": 0.6484502446982056,
      "grad_norm": 0.24655215442180634,
      "learning_rate": 3.241435562805873e-05,
      "loss": 0.0594,
      "num_input_tokens_seen": 8608256,
      "step": 3975
    },
    {
      "epoch": 0.6492659053833605,
      "grad_norm": 1.8580498695373535,
      "learning_rate": 3.245513866231648e-05,
      "loss": 0.182,
      "num_input_tokens_seen": 8618752,
      "step": 3980
    },
    {
      "epoch": 0.6500815660685155,
      "grad_norm": 0.043817225843667984,
      "learning_rate": 3.2495921696574226e-05,
      "loss": 0.2053,
      "num_input_tokens_seen": 8629824,
      "step": 3985
    },
    {
      "epoch": 0.6508972267536705,
      "grad_norm": 0.09651307761669159,
      "learning_rate": 3.2536704730831974e-05,
      "loss": 0.2543,
      "num_input_tokens_seen": 8640832,
      "step": 3990
    },
    {
      "epoch": 0.6517128874388255,
      "grad_norm": 0.22825945913791656,
      "learning_rate": 3.257748776508973e-05,
      "loss": 0.25,
      "num_input_tokens_seen": 8651104,
      "step": 3995
    },
    {
      "epoch": 0.6525285481239804,
      "grad_norm": 6.691067218780518,
      "learning_rate": 3.261827079934747e-05,
      "loss": 0.2699,
      "num_input_tokens_seen": 8662112,
      "step": 4000
    },
    {
      "epoch": 0.6533442088091354,
      "grad_norm": 3.487319231033325,
      "learning_rate": 3.2659053833605224e-05,
      "loss": 0.0776,
      "num_input_tokens_seen": 8672960,
      "step": 4005
    },
    {
      "epoch": 0.6541598694942904,
      "grad_norm": 0.07446221262216568,
      "learning_rate": 3.269983686786297e-05,
      "loss": 0.0553,
      "num_input_tokens_seen": 8683968,
      "step": 4010
    },
    {
      "epoch": 0.6549755301794453,
      "grad_norm": 5.4846696853637695,
      "learning_rate": 3.274061990212072e-05,
      "loss": 0.1836,
      "num_input_tokens_seen": 8695200,
      "step": 4015
    },
    {
      "epoch": 0.6557911908646004,
      "grad_norm": 1.5700244903564453,
      "learning_rate": 3.278140293637847e-05,
      "loss": 0.2142,
      "num_input_tokens_seen": 8705408,
      "step": 4020
    },
    {
      "epoch": 0.6566068515497553,
      "grad_norm": 6.220165729522705,
      "learning_rate": 3.2822185970636214e-05,
      "loss": 0.175,
      "num_input_tokens_seen": 8716128,
      "step": 4025
    },
    {
      "epoch": 0.6574225122349103,
      "grad_norm": 1.2835204601287842,
      "learning_rate": 3.286296900489397e-05,
      "loss": 0.3228,
      "num_input_tokens_seen": 8727744,
      "step": 4030
    },
    {
      "epoch": 0.6582381729200653,
      "grad_norm": 1.2002179622650146,
      "learning_rate": 3.2903752039151716e-05,
      "loss": 0.221,
      "num_input_tokens_seen": 8740128,
      "step": 4035
    },
    {
      "epoch": 0.6590538336052202,
      "grad_norm": 9.06164836883545,
      "learning_rate": 3.2944535073409464e-05,
      "loss": 0.2626,
      "num_input_tokens_seen": 8752320,
      "step": 4040
    },
    {
      "epoch": 0.6598694942903752,
      "grad_norm": 0.5982420444488525,
      "learning_rate": 3.298531810766721e-05,
      "loss": 0.2252,
      "num_input_tokens_seen": 8762784,
      "step": 4045
    },
    {
      "epoch": 0.6606851549755302,
      "grad_norm": 0.3971848785877228,
      "learning_rate": 3.302610114192496e-05,
      "loss": 0.1462,
      "num_input_tokens_seen": 8773312,
      "step": 4050
    },
    {
      "epoch": 0.6615008156606852,
      "grad_norm": 2.4643943309783936,
      "learning_rate": 3.3066884176182714e-05,
      "loss": 0.161,
      "num_input_tokens_seen": 8783744,
      "step": 4055
    },
    {
      "epoch": 0.6623164763458401,
      "grad_norm": 0.3730051517486572,
      "learning_rate": 3.3107667210440455e-05,
      "loss": 0.036,
      "num_input_tokens_seen": 8795104,
      "step": 4060
    },
    {
      "epoch": 0.6631321370309952,
      "grad_norm": 0.6188569068908691,
      "learning_rate": 3.314845024469821e-05,
      "loss": 0.1303,
      "num_input_tokens_seen": 8805760,
      "step": 4065
    },
    {
      "epoch": 0.6639477977161501,
      "grad_norm": 5.813721656799316,
      "learning_rate": 3.318923327895596e-05,
      "loss": 0.1185,
      "num_input_tokens_seen": 8817536,
      "step": 4070
    },
    {
      "epoch": 0.664763458401305,
      "grad_norm": 2.0859711170196533,
      "learning_rate": 3.3230016313213704e-05,
      "loss": 0.0693,
      "num_input_tokens_seen": 8829184,
      "step": 4075
    },
    {
      "epoch": 0.6655791190864601,
      "grad_norm": 0.031587276607751846,
      "learning_rate": 3.327079934747145e-05,
      "loss": 0.0916,
      "num_input_tokens_seen": 8839328,
      "step": 4080
    },
    {
      "epoch": 0.666394779771615,
      "grad_norm": 0.9004303216934204,
      "learning_rate": 3.33115823817292e-05,
      "loss": 0.2124,
      "num_input_tokens_seen": 8850336,
      "step": 4085
    },
    {
      "epoch": 0.66721044045677,
      "grad_norm": 0.3769245147705078,
      "learning_rate": 3.3352365415986954e-05,
      "loss": 0.1446,
      "num_input_tokens_seen": 8859392,
      "step": 4090
    },
    {
      "epoch": 0.668026101141925,
      "grad_norm": 0.039969563484191895,
      "learning_rate": 3.33931484502447e-05,
      "loss": 0.0521,
      "num_input_tokens_seen": 8871008,
      "step": 4095
    },
    {
      "epoch": 0.6688417618270799,
      "grad_norm": 0.047942303121089935,
      "learning_rate": 3.343393148450245e-05,
      "loss": 0.1959,
      "num_input_tokens_seen": 8881760,
      "step": 4100
    },
    {
      "epoch": 0.6696574225122349,
      "grad_norm": 3.715879201889038,
      "learning_rate": 3.34747145187602e-05,
      "loss": 0.3242,
      "num_input_tokens_seen": 8892544,
      "step": 4105
    },
    {
      "epoch": 0.6704730831973899,
      "grad_norm": 8.826482772827148,
      "learning_rate": 3.3515497553017945e-05,
      "loss": 0.4686,
      "num_input_tokens_seen": 8903936,
      "step": 4110
    },
    {
      "epoch": 0.6712887438825449,
      "grad_norm": 0.09291672706604004,
      "learning_rate": 3.35562805872757e-05,
      "loss": 0.227,
      "num_input_tokens_seen": 8913952,
      "step": 4115
    },
    {
      "epoch": 0.6721044045676998,
      "grad_norm": 3.907045602798462,
      "learning_rate": 3.359706362153344e-05,
      "loss": 0.1729,
      "num_input_tokens_seen": 8925120,
      "step": 4120
    },
    {
      "epoch": 0.6729200652528549,
      "grad_norm": 0.5516533255577087,
      "learning_rate": 3.3637846655791194e-05,
      "loss": 0.199,
      "num_input_tokens_seen": 8935872,
      "step": 4125
    },
    {
      "epoch": 0.6737357259380098,
      "grad_norm": 0.08874432742595673,
      "learning_rate": 3.367862969004894e-05,
      "loss": 0.1666,
      "num_input_tokens_seen": 8946624,
      "step": 4130
    },
    {
      "epoch": 0.6745513866231647,
      "grad_norm": 0.21990741789340973,
      "learning_rate": 3.371941272430669e-05,
      "loss": 0.1323,
      "num_input_tokens_seen": 8957568,
      "step": 4135
    },
    {
      "epoch": 0.6753670473083198,
      "grad_norm": 4.526270389556885,
      "learning_rate": 3.376019575856444e-05,
      "loss": 0.2651,
      "num_input_tokens_seen": 8968160,
      "step": 4140
    },
    {
      "epoch": 0.6761827079934747,
      "grad_norm": 0.9621414542198181,
      "learning_rate": 3.3800978792822185e-05,
      "loss": 0.1044,
      "num_input_tokens_seen": 8978560,
      "step": 4145
    },
    {
      "epoch": 0.6769983686786297,
      "grad_norm": 8.423593521118164,
      "learning_rate": 3.384176182707994e-05,
      "loss": 0.2068,
      "num_input_tokens_seen": 8988896,
      "step": 4150
    },
    {
      "epoch": 0.6778140293637847,
      "grad_norm": 6.477504730224609,
      "learning_rate": 3.388254486133769e-05,
      "loss": 0.2755,
      "num_input_tokens_seen": 9000928,
      "step": 4155
    },
    {
      "epoch": 0.6786296900489397,
      "grad_norm": 0.6715103983879089,
      "learning_rate": 3.3923327895595435e-05,
      "loss": 0.0682,
      "num_input_tokens_seen": 9011712,
      "step": 4160
    },
    {
      "epoch": 0.6794453507340946,
      "grad_norm": 6.027068614959717,
      "learning_rate": 3.396411092985318e-05,
      "loss": 0.1917,
      "num_input_tokens_seen": 9022624,
      "step": 4165
    },
    {
      "epoch": 0.6802610114192496,
      "grad_norm": 0.04681265354156494,
      "learning_rate": 3.400489396411093e-05,
      "loss": 0.2185,
      "num_input_tokens_seen": 9033088,
      "step": 4170
    },
    {
      "epoch": 0.6810766721044046,
      "grad_norm": 8.601092338562012,
      "learning_rate": 3.404567699836868e-05,
      "loss": 0.3408,
      "num_input_tokens_seen": 9043680,
      "step": 4175
    },
    {
      "epoch": 0.6818923327895595,
      "grad_norm": 2.7318994998931885,
      "learning_rate": 3.4086460032626425e-05,
      "loss": 0.1675,
      "num_input_tokens_seen": 9055424,
      "step": 4180
    },
    {
      "epoch": 0.6827079934747146,
      "grad_norm": 4.463717937469482,
      "learning_rate": 3.412724306688418e-05,
      "loss": 0.0983,
      "num_input_tokens_seen": 9066816,
      "step": 4185
    },
    {
      "epoch": 0.6835236541598695,
      "grad_norm": 0.1789654791355133,
      "learning_rate": 3.416802610114193e-05,
      "loss": 0.2268,
      "num_input_tokens_seen": 9077792,
      "step": 4190
    },
    {
      "epoch": 0.6843393148450244,
      "grad_norm": 0.3223755359649658,
      "learning_rate": 3.4208809135399675e-05,
      "loss": 0.3277,
      "num_input_tokens_seen": 9088160,
      "step": 4195
    },
    {
      "epoch": 0.6851549755301795,
      "grad_norm": 0.13708998262882233,
      "learning_rate": 3.424959216965742e-05,
      "loss": 0.0267,
      "num_input_tokens_seen": 9099264,
      "step": 4200
    },
    {
      "epoch": 0.6859706362153344,
      "grad_norm": 4.07029914855957,
      "learning_rate": 3.429037520391517e-05,
      "loss": 0.2568,
      "num_input_tokens_seen": 9110336,
      "step": 4205
    },
    {
      "epoch": 0.6867862969004894,
      "grad_norm": 7.171268939971924,
      "learning_rate": 3.4331158238172925e-05,
      "loss": 0.0953,
      "num_input_tokens_seen": 9121248,
      "step": 4210
    },
    {
      "epoch": 0.6876019575856444,
      "grad_norm": 5.684694290161133,
      "learning_rate": 3.4371941272430666e-05,
      "loss": 0.0785,
      "num_input_tokens_seen": 9132896,
      "step": 4215
    },
    {
      "epoch": 0.6884176182707994,
      "grad_norm": 0.05936404690146446,
      "learning_rate": 3.441272430668842e-05,
      "loss": 0.0886,
      "num_input_tokens_seen": 9143968,
      "step": 4220
    },
    {
      "epoch": 0.6892332789559543,
      "grad_norm": 0.088518887758255,
      "learning_rate": 3.445350734094617e-05,
      "loss": 0.1126,
      "num_input_tokens_seen": 9154848,
      "step": 4225
    },
    {
      "epoch": 0.6900489396411092,
      "grad_norm": 0.05748211219906807,
      "learning_rate": 3.449429037520392e-05,
      "loss": 0.1716,
      "num_input_tokens_seen": 9167264,
      "step": 4230
    },
    {
      "epoch": 0.6908646003262643,
      "grad_norm": 0.12895634770393372,
      "learning_rate": 3.453507340946166e-05,
      "loss": 0.0454,
      "num_input_tokens_seen": 9178016,
      "step": 4235
    },
    {
      "epoch": 0.6916802610114192,
      "grad_norm": 0.31009072065353394,
      "learning_rate": 3.457585644371941e-05,
      "loss": 0.1187,
      "num_input_tokens_seen": 9188224,
      "step": 4240
    },
    {
      "epoch": 0.6924959216965743,
      "grad_norm": 0.01949448138475418,
      "learning_rate": 3.4616639477977165e-05,
      "loss": 0.0353,
      "num_input_tokens_seen": 9198208,
      "step": 4245
    },
    {
      "epoch": 0.6933115823817292,
      "grad_norm": 0.3380926251411438,
      "learning_rate": 3.465742251223491e-05,
      "loss": 0.1166,
      "num_input_tokens_seen": 9208128,
      "step": 4250
    },
    {
      "epoch": 0.6941272430668842,
      "grad_norm": 0.0255452711135149,
      "learning_rate": 3.469820554649266e-05,
      "loss": 0.0684,
      "num_input_tokens_seen": 9218080,
      "step": 4255
    },
    {
      "epoch": 0.6949429037520392,
      "grad_norm": 0.8182644844055176,
      "learning_rate": 3.473898858075041e-05,
      "loss": 0.0126,
      "num_input_tokens_seen": 9228416,
      "step": 4260
    },
    {
      "epoch": 0.6957585644371941,
      "grad_norm": 5.6626715660095215,
      "learning_rate": 3.477977161500816e-05,
      "loss": 0.3426,
      "num_input_tokens_seen": 9239296,
      "step": 4265
    },
    {
      "epoch": 0.6965742251223491,
      "grad_norm": 0.26666080951690674,
      "learning_rate": 3.482055464926591e-05,
      "loss": 0.3668,
      "num_input_tokens_seen": 9249472,
      "step": 4270
    },
    {
      "epoch": 0.697389885807504,
      "grad_norm": 4.0731682777404785,
      "learning_rate": 3.486133768352365e-05,
      "loss": 0.1809,
      "num_input_tokens_seen": 9259616,
      "step": 4275
    },
    {
      "epoch": 0.6982055464926591,
      "grad_norm": 0.7779442071914673,
      "learning_rate": 3.4902120717781405e-05,
      "loss": 0.0951,
      "num_input_tokens_seen": 9271360,
      "step": 4280
    },
    {
      "epoch": 0.699021207177814,
      "grad_norm": 0.9017668962478638,
      "learning_rate": 3.494290375203915e-05,
      "loss": 0.1988,
      "num_input_tokens_seen": 9281952,
      "step": 4285
    },
    {
      "epoch": 0.6998368678629691,
      "grad_norm": 8.216322898864746,
      "learning_rate": 3.498368678629691e-05,
      "loss": 0.2337,
      "num_input_tokens_seen": 9292480,
      "step": 4290
    },
    {
      "epoch": 0.700652528548124,
      "grad_norm": 0.034616608172655106,
      "learning_rate": 3.502446982055465e-05,
      "loss": 0.422,
      "num_input_tokens_seen": 9301696,
      "step": 4295
    },
    {
      "epoch": 0.7014681892332789,
      "grad_norm": 1.7564533948898315,
      "learning_rate": 3.5065252854812396e-05,
      "loss": 0.063,
      "num_input_tokens_seen": 9311360,
      "step": 4300
    },
    {
      "epoch": 0.702283849918434,
      "grad_norm": 5.38205623626709,
      "learning_rate": 3.510603588907015e-05,
      "loss": 0.2506,
      "num_input_tokens_seen": 9321504,
      "step": 4305
    },
    {
      "epoch": 0.7030995106035889,
      "grad_norm": 1.9961726665496826,
      "learning_rate": 3.51468189233279e-05,
      "loss": 0.2344,
      "num_input_tokens_seen": 9333024,
      "step": 4310
    },
    {
      "epoch": 0.7039151712887439,
      "grad_norm": 4.119628429412842,
      "learning_rate": 3.5187601957585646e-05,
      "loss": 0.1086,
      "num_input_tokens_seen": 9343424,
      "step": 4315
    },
    {
      "epoch": 0.7047308319738989,
      "grad_norm": 0.02915322594344616,
      "learning_rate": 3.522838499184339e-05,
      "loss": 0.0216,
      "num_input_tokens_seen": 9354016,
      "step": 4320
    },
    {
      "epoch": 0.7055464926590538,
      "grad_norm": 0.06092199683189392,
      "learning_rate": 3.526916802610115e-05,
      "loss": 0.1912,
      "num_input_tokens_seen": 9365600,
      "step": 4325
    },
    {
      "epoch": 0.7063621533442088,
      "grad_norm": 0.24240359663963318,
      "learning_rate": 3.5309951060358895e-05,
      "loss": 0.2048,
      "num_input_tokens_seen": 9377216,
      "step": 4330
    },
    {
      "epoch": 0.7071778140293637,
      "grad_norm": 1.4081376791000366,
      "learning_rate": 3.5350734094616636e-05,
      "loss": 0.0669,
      "num_input_tokens_seen": 9388000,
      "step": 4335
    },
    {
      "epoch": 0.7079934747145188,
      "grad_norm": 4.129550933837891,
      "learning_rate": 3.539151712887439e-05,
      "loss": 0.2263,
      "num_input_tokens_seen": 9399616,
      "step": 4340
    },
    {
      "epoch": 0.7088091353996737,
      "grad_norm": 3.214871883392334,
      "learning_rate": 3.543230016313214e-05,
      "loss": 0.1523,
      "num_input_tokens_seen": 9411872,
      "step": 4345
    },
    {
      "epoch": 0.7096247960848288,
      "grad_norm": 1.9875812530517578,
      "learning_rate": 3.5473083197389886e-05,
      "loss": 0.2107,
      "num_input_tokens_seen": 9422912,
      "step": 4350
    },
    {
      "epoch": 0.7104404567699837,
      "grad_norm": 2.3759799003601074,
      "learning_rate": 3.5513866231647634e-05,
      "loss": 0.0238,
      "num_input_tokens_seen": 9434176,
      "step": 4355
    },
    {
      "epoch": 0.7112561174551386,
      "grad_norm": 1.6661633253097534,
      "learning_rate": 3.555464926590539e-05,
      "loss": 0.1262,
      "num_input_tokens_seen": 9443936,
      "step": 4360
    },
    {
      "epoch": 0.7120717781402937,
      "grad_norm": 0.21446746587753296,
      "learning_rate": 3.5595432300163136e-05,
      "loss": 0.1062,
      "num_input_tokens_seen": 9453984,
      "step": 4365
    },
    {
      "epoch": 0.7128874388254486,
      "grad_norm": 0.376142680644989,
      "learning_rate": 3.563621533442088e-05,
      "loss": 0.2228,
      "num_input_tokens_seen": 9465408,
      "step": 4370
    },
    {
      "epoch": 0.7137030995106036,
      "grad_norm": 0.8069067001342773,
      "learning_rate": 3.567699836867863e-05,
      "loss": 0.175,
      "num_input_tokens_seen": 9475392,
      "step": 4375
    },
    {
      "epoch": 0.7145187601957586,
      "grad_norm": 4.707369327545166,
      "learning_rate": 3.571778140293638e-05,
      "loss": 0.2439,
      "num_input_tokens_seen": 9485280,
      "step": 4380
    },
    {
      "epoch": 0.7153344208809136,
      "grad_norm": 1.5703744888305664,
      "learning_rate": 3.575856443719413e-05,
      "loss": 0.0592,
      "num_input_tokens_seen": 9496224,
      "step": 4385
    },
    {
      "epoch": 0.7161500815660685,
      "grad_norm": 1.8300573825836182,
      "learning_rate": 3.579934747145188e-05,
      "loss": 0.1236,
      "num_input_tokens_seen": 9507456,
      "step": 4390
    },
    {
      "epoch": 0.7169657422512234,
      "grad_norm": 6.576985836029053,
      "learning_rate": 3.584013050570963e-05,
      "loss": 0.3141,
      "num_input_tokens_seen": 9517952,
      "step": 4395
    },
    {
      "epoch": 0.7177814029363785,
      "grad_norm": 4.5055718421936035,
      "learning_rate": 3.5880913539967376e-05,
      "loss": 0.1647,
      "num_input_tokens_seen": 9528672,
      "step": 4400
    },
    {
      "epoch": 0.7185970636215334,
      "grad_norm": 0.5434504151344299,
      "learning_rate": 3.5921696574225124e-05,
      "loss": 0.0497,
      "num_input_tokens_seen": 9539552,
      "step": 4405
    },
    {
      "epoch": 0.7194127243066885,
      "grad_norm": 3.5970358848571777,
      "learning_rate": 3.596247960848287e-05,
      "loss": 0.3283,
      "num_input_tokens_seen": 9550112,
      "step": 4410
    },
    {
      "epoch": 0.7202283849918434,
      "grad_norm": 0.43845248222351074,
      "learning_rate": 3.600326264274062e-05,
      "loss": 0.2055,
      "num_input_tokens_seen": 9560544,
      "step": 4415
    },
    {
      "epoch": 0.7210440456769984,
      "grad_norm": 9.78809928894043,
      "learning_rate": 3.604404567699837e-05,
      "loss": 0.3818,
      "num_input_tokens_seen": 9571232,
      "step": 4420
    },
    {
      "epoch": 0.7218597063621534,
      "grad_norm": 0.12565630674362183,
      "learning_rate": 3.608482871125612e-05,
      "loss": 0.0643,
      "num_input_tokens_seen": 9581920,
      "step": 4425
    },
    {
      "epoch": 0.7226753670473083,
      "grad_norm": 0.7135125994682312,
      "learning_rate": 3.612561174551387e-05,
      "loss": 0.0689,
      "num_input_tokens_seen": 9592448,
      "step": 4430
    },
    {
      "epoch": 0.7234910277324633,
      "grad_norm": 0.40397709608078003,
      "learning_rate": 3.6166394779771616e-05,
      "loss": 0.1868,
      "num_input_tokens_seen": 9603360,
      "step": 4435
    },
    {
      "epoch": 0.7243066884176182,
      "grad_norm": 2.0303561687469482,
      "learning_rate": 3.6207177814029364e-05,
      "loss": 0.1742,
      "num_input_tokens_seen": 9614336,
      "step": 4440
    },
    {
      "epoch": 0.7251223491027733,
      "grad_norm": 0.05959249660372734,
      "learning_rate": 3.624796084828712e-05,
      "loss": 0.065,
      "num_input_tokens_seen": 9624512,
      "step": 4445
    },
    {
      "epoch": 0.7259380097879282,
      "grad_norm": 0.4840633273124695,
      "learning_rate": 3.628874388254486e-05,
      "loss": 0.1762,
      "num_input_tokens_seen": 9635168,
      "step": 4450
    },
    {
      "epoch": 0.7267536704730831,
      "grad_norm": 4.319450855255127,
      "learning_rate": 3.6329526916802614e-05,
      "loss": 0.2741,
      "num_input_tokens_seen": 9645280,
      "step": 4455
    },
    {
      "epoch": 0.7275693311582382,
      "grad_norm": 3.6525118350982666,
      "learning_rate": 3.637030995106036e-05,
      "loss": 0.0815,
      "num_input_tokens_seen": 9656032,
      "step": 4460
    },
    {
      "epoch": 0.7283849918433931,
      "grad_norm": 0.30015629529953003,
      "learning_rate": 3.641109298531811e-05,
      "loss": 0.2921,
      "num_input_tokens_seen": 9667968,
      "step": 4465
    },
    {
      "epoch": 0.7292006525285482,
      "grad_norm": 1.6846249103546143,
      "learning_rate": 3.6451876019575856e-05,
      "loss": 0.0801,
      "num_input_tokens_seen": 9680032,
      "step": 4470
    },
    {
      "epoch": 0.7300163132137031,
      "grad_norm": 4.6199116706848145,
      "learning_rate": 3.6492659053833604e-05,
      "loss": 0.2839,
      "num_input_tokens_seen": 9691424,
      "step": 4475
    },
    {
      "epoch": 0.7308319738988581,
      "grad_norm": 4.570083141326904,
      "learning_rate": 3.653344208809136e-05,
      "loss": 0.2675,
      "num_input_tokens_seen": 9702400,
      "step": 4480
    },
    {
      "epoch": 0.731647634584013,
      "grad_norm": 0.24840693175792694,
      "learning_rate": 3.6574225122349106e-05,
      "loss": 0.0725,
      "num_input_tokens_seen": 9712928,
      "step": 4485
    },
    {
      "epoch": 0.732463295269168,
      "grad_norm": 5.6841630935668945,
      "learning_rate": 3.6615008156606854e-05,
      "loss": 0.5455,
      "num_input_tokens_seen": 9723360,
      "step": 4490
    },
    {
      "epoch": 0.733278955954323,
      "grad_norm": 1.1254421472549438,
      "learning_rate": 3.66557911908646e-05,
      "loss": 0.0408,
      "num_input_tokens_seen": 9734496,
      "step": 4495
    },
    {
      "epoch": 0.734094616639478,
      "grad_norm": 5.724114418029785,
      "learning_rate": 3.669657422512235e-05,
      "loss": 0.2773,
      "num_input_tokens_seen": 9744288,
      "step": 4500
    },
    {
      "epoch": 0.734910277324633,
      "grad_norm": 5.264193534851074,
      "learning_rate": 3.6737357259380104e-05,
      "loss": 0.2471,
      "num_input_tokens_seen": 9755072,
      "step": 4505
    },
    {
      "epoch": 0.7357259380097879,
      "grad_norm": 0.7798420786857605,
      "learning_rate": 3.6778140293637844e-05,
      "loss": 0.101,
      "num_input_tokens_seen": 9764192,
      "step": 4510
    },
    {
      "epoch": 0.736541598694943,
      "grad_norm": 6.110246658325195,
      "learning_rate": 3.68189233278956e-05,
      "loss": 0.3526,
      "num_input_tokens_seen": 9775392,
      "step": 4515
    },
    {
      "epoch": 0.7373572593800979,
      "grad_norm": 1.7233092784881592,
      "learning_rate": 3.6859706362153346e-05,
      "loss": 0.0899,
      "num_input_tokens_seen": 9784864,
      "step": 4520
    },
    {
      "epoch": 0.7381729200652528,
      "grad_norm": 1.033625602722168,
      "learning_rate": 3.6900489396411094e-05,
      "loss": 0.067,
      "num_input_tokens_seen": 9796512,
      "step": 4525
    },
    {
      "epoch": 0.7389885807504079,
      "grad_norm": 1.708029866218567,
      "learning_rate": 3.694127243066884e-05,
      "loss": 0.0855,
      "num_input_tokens_seen": 9808416,
      "step": 4530
    },
    {
      "epoch": 0.7398042414355628,
      "grad_norm": 5.670646667480469,
      "learning_rate": 3.698205546492659e-05,
      "loss": 0.1154,
      "num_input_tokens_seen": 9819552,
      "step": 4535
    },
    {
      "epoch": 0.7406199021207178,
      "grad_norm": 0.8500028252601624,
      "learning_rate": 3.7022838499184344e-05,
      "loss": 0.0779,
      "num_input_tokens_seen": 9831168,
      "step": 4540
    },
    {
      "epoch": 0.7414355628058727,
      "grad_norm": 5.93217134475708,
      "learning_rate": 3.706362153344209e-05,
      "loss": 0.0827,
      "num_input_tokens_seen": 9841280,
      "step": 4545
    },
    {
      "epoch": 0.7422512234910277,
      "grad_norm": 1.6935220956802368,
      "learning_rate": 3.710440456769984e-05,
      "loss": 0.166,
      "num_input_tokens_seen": 9852224,
      "step": 4550
    },
    {
      "epoch": 0.7430668841761827,
      "grad_norm": 0.8768059015274048,
      "learning_rate": 3.714518760195759e-05,
      "loss": 0.3636,
      "num_input_tokens_seen": 9862656,
      "step": 4555
    },
    {
      "epoch": 0.7438825448613376,
      "grad_norm": 0.26617231965065,
      "learning_rate": 3.7185970636215334e-05,
      "loss": 0.0367,
      "num_input_tokens_seen": 9874304,
      "step": 4560
    },
    {
      "epoch": 0.7446982055464927,
      "grad_norm": 0.44966986775398254,
      "learning_rate": 3.722675367047309e-05,
      "loss": 0.0693,
      "num_input_tokens_seen": 9884128,
      "step": 4565
    },
    {
      "epoch": 0.7455138662316476,
      "grad_norm": 7.610336780548096,
      "learning_rate": 3.726753670473083e-05,
      "loss": 0.173,
      "num_input_tokens_seen": 9894208,
      "step": 4570
    },
    {
      "epoch": 0.7463295269168027,
      "grad_norm": 4.594542026519775,
      "learning_rate": 3.7308319738988584e-05,
      "loss": 0.1771,
      "num_input_tokens_seen": 9904864,
      "step": 4575
    },
    {
      "epoch": 0.7471451876019576,
      "grad_norm": 0.7763423323631287,
      "learning_rate": 3.734910277324633e-05,
      "loss": 0.17,
      "num_input_tokens_seen": 9915904,
      "step": 4580
    },
    {
      "epoch": 0.7479608482871125,
      "grad_norm": 0.045461248606443405,
      "learning_rate": 3.738988580750408e-05,
      "loss": 0.2818,
      "num_input_tokens_seen": 9926048,
      "step": 4585
    },
    {
      "epoch": 0.7487765089722676,
      "grad_norm": 0.1534917950630188,
      "learning_rate": 3.743066884176183e-05,
      "loss": 0.0633,
      "num_input_tokens_seen": 9936736,
      "step": 4590
    },
    {
      "epoch": 0.7495921696574225,
      "grad_norm": 3.768371820449829,
      "learning_rate": 3.7471451876019575e-05,
      "loss": 0.1892,
      "num_input_tokens_seen": 9949408,
      "step": 4595
    },
    {
      "epoch": 0.7504078303425775,
      "grad_norm": 3.5366828441619873,
      "learning_rate": 3.751223491027733e-05,
      "loss": 0.1483,
      "num_input_tokens_seen": 9960160,
      "step": 4600
    },
    {
      "epoch": 0.7512234910277324,
      "grad_norm": 5.063387393951416,
      "learning_rate": 3.755301794453508e-05,
      "loss": 0.2159,
      "num_input_tokens_seen": 9970624,
      "step": 4605
    },
    {
      "epoch": 0.7520391517128875,
      "grad_norm": 1.4474318027496338,
      "learning_rate": 3.7593800978792824e-05,
      "loss": 0.0649,
      "num_input_tokens_seen": 9982080,
      "step": 4610
    },
    {
      "epoch": 0.7528548123980424,
      "grad_norm": 0.2901652157306671,
      "learning_rate": 3.763458401305057e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 9992480,
      "step": 4615
    },
    {
      "epoch": 0.7536704730831973,
      "grad_norm": 0.274383544921875,
      "learning_rate": 3.7675367047308326e-05,
      "loss": 0.1726,
      "num_input_tokens_seen": 10002496,
      "step": 4620
    },
    {
      "epoch": 0.7544861337683524,
      "grad_norm": 6.004702091217041,
      "learning_rate": 3.771615008156607e-05,
      "loss": 0.3612,
      "num_input_tokens_seen": 10013408,
      "step": 4625
    },
    {
      "epoch": 0.7553017944535073,
      "grad_norm": 1.2919461727142334,
      "learning_rate": 3.7756933115823815e-05,
      "loss": 0.1936,
      "num_input_tokens_seen": 10022880,
      "step": 4630
    },
    {
      "epoch": 0.7561174551386624,
      "grad_norm": 4.892871379852295,
      "learning_rate": 3.779771615008157e-05,
      "loss": 0.1224,
      "num_input_tokens_seen": 10034208,
      "step": 4635
    },
    {
      "epoch": 0.7569331158238173,
      "grad_norm": 2.343761682510376,
      "learning_rate": 3.783849918433932e-05,
      "loss": 0.1395,
      "num_input_tokens_seen": 10044832,
      "step": 4640
    },
    {
      "epoch": 0.7577487765089723,
      "grad_norm": 0.14088532328605652,
      "learning_rate": 3.7879282218597065e-05,
      "loss": 0.2024,
      "num_input_tokens_seen": 10055712,
      "step": 4645
    },
    {
      "epoch": 0.7585644371941273,
      "grad_norm": 3.9574947357177734,
      "learning_rate": 3.792006525285481e-05,
      "loss": 0.2083,
      "num_input_tokens_seen": 10066976,
      "step": 4650
    },
    {
      "epoch": 0.7593800978792822,
      "grad_norm": 2.6159284114837646,
      "learning_rate": 3.796084828711256e-05,
      "loss": 0.2117,
      "num_input_tokens_seen": 10078304,
      "step": 4655
    },
    {
      "epoch": 0.7601957585644372,
      "grad_norm": 9.170280456542969,
      "learning_rate": 3.8001631321370314e-05,
      "loss": 0.4091,
      "num_input_tokens_seen": 10089056,
      "step": 4660
    },
    {
      "epoch": 0.7610114192495921,
      "grad_norm": 0.13579274713993073,
      "learning_rate": 3.804241435562806e-05,
      "loss": 0.2127,
      "num_input_tokens_seen": 10099904,
      "step": 4665
    },
    {
      "epoch": 0.7618270799347472,
      "grad_norm": 2.140343189239502,
      "learning_rate": 3.808319738988581e-05,
      "loss": 0.0377,
      "num_input_tokens_seen": 10111552,
      "step": 4670
    },
    {
      "epoch": 0.7626427406199021,
      "grad_norm": 4.381788730621338,
      "learning_rate": 3.812398042414356e-05,
      "loss": 0.331,
      "num_input_tokens_seen": 10121568,
      "step": 4675
    },
    {
      "epoch": 0.763458401305057,
      "grad_norm": 1.919498085975647,
      "learning_rate": 3.816476345840131e-05,
      "loss": 0.2048,
      "num_input_tokens_seen": 10132000,
      "step": 4680
    },
    {
      "epoch": 0.7642740619902121,
      "grad_norm": 4.956511497497559,
      "learning_rate": 3.820554649265905e-05,
      "loss": 0.3375,
      "num_input_tokens_seen": 10141952,
      "step": 4685
    },
    {
      "epoch": 0.765089722675367,
      "grad_norm": 0.4133996069431305,
      "learning_rate": 3.82463295269168e-05,
      "loss": 0.1261,
      "num_input_tokens_seen": 10151776,
      "step": 4690
    },
    {
      "epoch": 0.765905383360522,
      "grad_norm": 1.9626580476760864,
      "learning_rate": 3.8287112561174555e-05,
      "loss": 0.1644,
      "num_input_tokens_seen": 10162688,
      "step": 4695
    },
    {
      "epoch": 0.766721044045677,
      "grad_norm": 0.1816297173500061,
      "learning_rate": 3.83278955954323e-05,
      "loss": 0.0105,
      "num_input_tokens_seen": 10174624,
      "step": 4700
    },
    {
      "epoch": 0.767536704730832,
      "grad_norm": 0.3867439031600952,
      "learning_rate": 3.836867862969005e-05,
      "loss": 0.0274,
      "num_input_tokens_seen": 10185248,
      "step": 4705
    },
    {
      "epoch": 0.768352365415987,
      "grad_norm": 0.06129593402147293,
      "learning_rate": 3.84094616639478e-05,
      "loss": 0.0324,
      "num_input_tokens_seen": 10196768,
      "step": 4710
    },
    {
      "epoch": 0.7691680261011419,
      "grad_norm": 0.291681170463562,
      "learning_rate": 3.845024469820555e-05,
      "loss": 0.0494,
      "num_input_tokens_seen": 10206816,
      "step": 4715
    },
    {
      "epoch": 0.7699836867862969,
      "grad_norm": 0.4279186725616455,
      "learning_rate": 3.84910277324633e-05,
      "loss": 0.061,
      "num_input_tokens_seen": 10217600,
      "step": 4720
    },
    {
      "epoch": 0.7707993474714518,
      "grad_norm": 0.767796516418457,
      "learning_rate": 3.853181076672104e-05,
      "loss": 0.2075,
      "num_input_tokens_seen": 10228544,
      "step": 4725
    },
    {
      "epoch": 0.7716150081566069,
      "grad_norm": 0.10908225923776627,
      "learning_rate": 3.8572593800978795e-05,
      "loss": 0.102,
      "num_input_tokens_seen": 10239168,
      "step": 4730
    },
    {
      "epoch": 0.7724306688417618,
      "grad_norm": 1.484430193901062,
      "learning_rate": 3.861337683523654e-05,
      "loss": 0.1301,
      "num_input_tokens_seen": 10250784,
      "step": 4735
    },
    {
      "epoch": 0.7732463295269169,
      "grad_norm": 4.065375328063965,
      "learning_rate": 3.86541598694943e-05,
      "loss": 0.1236,
      "num_input_tokens_seen": 10260480,
      "step": 4740
    },
    {
      "epoch": 0.7740619902120718,
      "grad_norm": 4.384725570678711,
      "learning_rate": 3.869494290375204e-05,
      "loss": 0.2311,
      "num_input_tokens_seen": 10272064,
      "step": 4745
    },
    {
      "epoch": 0.7748776508972267,
      "grad_norm": 0.06715980172157288,
      "learning_rate": 3.873572593800979e-05,
      "loss": 0.0408,
      "num_input_tokens_seen": 10281472,
      "step": 4750
    },
    {
      "epoch": 0.7756933115823818,
      "grad_norm": 1.4746594429016113,
      "learning_rate": 3.877650897226754e-05,
      "loss": 0.1715,
      "num_input_tokens_seen": 10293536,
      "step": 4755
    },
    {
      "epoch": 0.7765089722675367,
      "grad_norm": 0.1626254916191101,
      "learning_rate": 3.881729200652529e-05,
      "loss": 0.1153,
      "num_input_tokens_seen": 10303968,
      "step": 4760
    },
    {
      "epoch": 0.7773246329526917,
      "grad_norm": 0.15786142647266388,
      "learning_rate": 3.8858075040783035e-05,
      "loss": 0.1055,
      "num_input_tokens_seen": 10316064,
      "step": 4765
    },
    {
      "epoch": 0.7781402936378466,
      "grad_norm": 0.27948567271232605,
      "learning_rate": 3.889885807504078e-05,
      "loss": 0.1204,
      "num_input_tokens_seen": 10327328,
      "step": 4770
    },
    {
      "epoch": 0.7789559543230016,
      "grad_norm": 6.373471736907959,
      "learning_rate": 3.893964110929854e-05,
      "loss": 0.3351,
      "num_input_tokens_seen": 10337088,
      "step": 4775
    },
    {
      "epoch": 0.7797716150081566,
      "grad_norm": 5.559701442718506,
      "learning_rate": 3.8980424143556285e-05,
      "loss": 0.311,
      "num_input_tokens_seen": 10347488,
      "step": 4780
    },
    {
      "epoch": 0.7805872756933115,
      "grad_norm": 5.775311470031738,
      "learning_rate": 3.9021207177814026e-05,
      "loss": 0.2612,
      "num_input_tokens_seen": 10358656,
      "step": 4785
    },
    {
      "epoch": 0.7814029363784666,
      "grad_norm": 2.5803449153900146,
      "learning_rate": 3.906199021207178e-05,
      "loss": 0.2102,
      "num_input_tokens_seen": 10370816,
      "step": 4790
    },
    {
      "epoch": 0.7822185970636215,
      "grad_norm": 4.448340892791748,
      "learning_rate": 3.910277324632953e-05,
      "loss": 0.0847,
      "num_input_tokens_seen": 10381856,
      "step": 4795
    },
    {
      "epoch": 0.7830342577487766,
      "grad_norm": 5.02500581741333,
      "learning_rate": 3.914355628058728e-05,
      "loss": 0.2397,
      "num_input_tokens_seen": 10392320,
      "step": 4800
    },
    {
      "epoch": 0.7838499184339315,
      "grad_norm": 4.225478172302246,
      "learning_rate": 3.918433931484502e-05,
      "loss": 0.3016,
      "num_input_tokens_seen": 10401568,
      "step": 4805
    },
    {
      "epoch": 0.7846655791190864,
      "grad_norm": 1.1861470937728882,
      "learning_rate": 3.922512234910278e-05,
      "loss": 0.2616,
      "num_input_tokens_seen": 10413248,
      "step": 4810
    },
    {
      "epoch": 0.7854812398042414,
      "grad_norm": 5.154033660888672,
      "learning_rate": 3.9265905383360525e-05,
      "loss": 0.0759,
      "num_input_tokens_seen": 10423168,
      "step": 4815
    },
    {
      "epoch": 0.7862969004893964,
      "grad_norm": 0.817079484462738,
      "learning_rate": 3.930668841761827e-05,
      "loss": 0.1227,
      "num_input_tokens_seen": 10433792,
      "step": 4820
    },
    {
      "epoch": 0.7871125611745514,
      "grad_norm": 3.3402276039123535,
      "learning_rate": 3.934747145187602e-05,
      "loss": 0.1032,
      "num_input_tokens_seen": 10445120,
      "step": 4825
    },
    {
      "epoch": 0.7879282218597063,
      "grad_norm": 2.4098806381225586,
      "learning_rate": 3.938825448613377e-05,
      "loss": 0.1869,
      "num_input_tokens_seen": 10455584,
      "step": 4830
    },
    {
      "epoch": 0.7887438825448614,
      "grad_norm": 0.1025315448641777,
      "learning_rate": 3.942903752039152e-05,
      "loss": 0.0962,
      "num_input_tokens_seen": 10466752,
      "step": 4835
    },
    {
      "epoch": 0.7895595432300163,
      "grad_norm": 3.8288652896881104,
      "learning_rate": 3.946982055464927e-05,
      "loss": 0.1866,
      "num_input_tokens_seen": 10477376,
      "step": 4840
    },
    {
      "epoch": 0.7903752039151712,
      "grad_norm": 6.3380231857299805,
      "learning_rate": 3.951060358890702e-05,
      "loss": 0.1946,
      "num_input_tokens_seen": 10487392,
      "step": 4845
    },
    {
      "epoch": 0.7911908646003263,
      "grad_norm": 3.5893802642822266,
      "learning_rate": 3.9551386623164766e-05,
      "loss": 0.3604,
      "num_input_tokens_seen": 10498208,
      "step": 4850
    },
    {
      "epoch": 0.7920065252854812,
      "grad_norm": 1.9882179498672485,
      "learning_rate": 3.959216965742251e-05,
      "loss": 0.1252,
      "num_input_tokens_seen": 10507200,
      "step": 4855
    },
    {
      "epoch": 0.7928221859706363,
      "grad_norm": 0.5306270122528076,
      "learning_rate": 3.963295269168026e-05,
      "loss": 0.2172,
      "num_input_tokens_seen": 10516416,
      "step": 4860
    },
    {
      "epoch": 0.7936378466557912,
      "grad_norm": 3.275817632675171,
      "learning_rate": 3.967373572593801e-05,
      "loss": 0.3197,
      "num_input_tokens_seen": 10527648,
      "step": 4865
    },
    {
      "epoch": 0.7944535073409462,
      "grad_norm": 1.6505345106124878,
      "learning_rate": 3.971451876019576e-05,
      "loss": 0.0284,
      "num_input_tokens_seen": 10537952,
      "step": 4870
    },
    {
      "epoch": 0.7952691680261011,
      "grad_norm": 3.008704662322998,
      "learning_rate": 3.975530179445351e-05,
      "loss": 0.1331,
      "num_input_tokens_seen": 10548320,
      "step": 4875
    },
    {
      "epoch": 0.7960848287112561,
      "grad_norm": 0.39359229803085327,
      "learning_rate": 3.979608482871126e-05,
      "loss": 0.0615,
      "num_input_tokens_seen": 10557824,
      "step": 4880
    },
    {
      "epoch": 0.7969004893964111,
      "grad_norm": 0.1988229602575302,
      "learning_rate": 3.9836867862969006e-05,
      "loss": 0.0253,
      "num_input_tokens_seen": 10568896,
      "step": 4885
    },
    {
      "epoch": 0.797716150081566,
      "grad_norm": 3.483168601989746,
      "learning_rate": 3.9877650897226754e-05,
      "loss": 0.3658,
      "num_input_tokens_seen": 10580064,
      "step": 4890
    },
    {
      "epoch": 0.7985318107667211,
      "grad_norm": 0.05558078736066818,
      "learning_rate": 3.991843393148451e-05,
      "loss": 0.158,
      "num_input_tokens_seen": 10590976,
      "step": 4895
    },
    {
      "epoch": 0.799347471451876,
      "grad_norm": 4.738973140716553,
      "learning_rate": 3.995921696574225e-05,
      "loss": 0.1316,
      "num_input_tokens_seen": 10600448,
      "step": 4900
    },
    {
      "epoch": 0.8001631321370309,
      "grad_norm": 2.7792470455169678,
      "learning_rate": 4e-05,
      "loss": 0.1654,
      "num_input_tokens_seen": 10611904,
      "step": 4905
    },
    {
      "epoch": 0.800978792822186,
      "grad_norm": 1.0185917615890503,
      "learning_rate": 4.004078303425775e-05,
      "loss": 0.0977,
      "num_input_tokens_seen": 10622240,
      "step": 4910
    },
    {
      "epoch": 0.8017944535073409,
      "grad_norm": 4.466019153594971,
      "learning_rate": 4.00815660685155e-05,
      "loss": 0.2974,
      "num_input_tokens_seen": 10632864,
      "step": 4915
    },
    {
      "epoch": 0.802610114192496,
      "grad_norm": 0.8997367024421692,
      "learning_rate": 4.0122349102773246e-05,
      "loss": 0.0422,
      "num_input_tokens_seen": 10642464,
      "step": 4920
    },
    {
      "epoch": 0.8034257748776509,
      "grad_norm": 1.2520920038223267,
      "learning_rate": 4.0163132137030994e-05,
      "loss": 0.1862,
      "num_input_tokens_seen": 10653024,
      "step": 4925
    },
    {
      "epoch": 0.8042414355628059,
      "grad_norm": 1.0238492488861084,
      "learning_rate": 4.020391517128875e-05,
      "loss": 0.1163,
      "num_input_tokens_seen": 10664384,
      "step": 4930
    },
    {
      "epoch": 0.8050570962479608,
      "grad_norm": 5.637166976928711,
      "learning_rate": 4.0244698205546496e-05,
      "loss": 0.3622,
      "num_input_tokens_seen": 10675360,
      "step": 4935
    },
    {
      "epoch": 0.8058727569331158,
      "grad_norm": 0.8354208469390869,
      "learning_rate": 4.0285481239804244e-05,
      "loss": 0.1535,
      "num_input_tokens_seen": 10685824,
      "step": 4940
    },
    {
      "epoch": 0.8066884176182708,
      "grad_norm": 4.137977600097656,
      "learning_rate": 4.032626427406199e-05,
      "loss": 0.2299,
      "num_input_tokens_seen": 10697536,
      "step": 4945
    },
    {
      "epoch": 0.8075040783034257,
      "grad_norm": 1.5421990156173706,
      "learning_rate": 4.036704730831974e-05,
      "loss": 0.0498,
      "num_input_tokens_seen": 10709888,
      "step": 4950
    },
    {
      "epoch": 0.8083197389885808,
      "grad_norm": 0.06120558828115463,
      "learning_rate": 4.040783034257749e-05,
      "loss": 0.0596,
      "num_input_tokens_seen": 10721216,
      "step": 4955
    },
    {
      "epoch": 0.8091353996737357,
      "grad_norm": 0.17592903971672058,
      "learning_rate": 4.0448613376835234e-05,
      "loss": 0.102,
      "num_input_tokens_seen": 10729952,
      "step": 4960
    },
    {
      "epoch": 0.8099510603588908,
      "grad_norm": 4.301577568054199,
      "learning_rate": 4.048939641109299e-05,
      "loss": 0.1266,
      "num_input_tokens_seen": 10741216,
      "step": 4965
    },
    {
      "epoch": 0.8107667210440457,
      "grad_norm": 0.1696639209985733,
      "learning_rate": 4.0530179445350736e-05,
      "loss": 0.1888,
      "num_input_tokens_seen": 10752544,
      "step": 4970
    },
    {
      "epoch": 0.8115823817292006,
      "grad_norm": 0.501167893409729,
      "learning_rate": 4.057096247960849e-05,
      "loss": 0.2189,
      "num_input_tokens_seen": 10764960,
      "step": 4975
    },
    {
      "epoch": 0.8123980424143556,
      "grad_norm": 1.5304832458496094,
      "learning_rate": 4.061174551386623e-05,
      "loss": 0.0462,
      "num_input_tokens_seen": 10775296,
      "step": 4980
    },
    {
      "epoch": 0.8132137030995106,
      "grad_norm": 3.5793774127960205,
      "learning_rate": 4.065252854812398e-05,
      "loss": 0.1042,
      "num_input_tokens_seen": 10786112,
      "step": 4985
    },
    {
      "epoch": 0.8140293637846656,
      "grad_norm": 0.9823907017707825,
      "learning_rate": 4.0693311582381734e-05,
      "loss": 0.1694,
      "num_input_tokens_seen": 10797152,
      "step": 4990
    },
    {
      "epoch": 0.8148450244698205,
      "grad_norm": 5.591163635253906,
      "learning_rate": 4.073409461663948e-05,
      "loss": 0.1854,
      "num_input_tokens_seen": 10808480,
      "step": 4995
    },
    {
      "epoch": 0.8156606851549756,
      "grad_norm": 4.029349327087402,
      "learning_rate": 4.077487765089723e-05,
      "loss": 0.0428,
      "num_input_tokens_seen": 10819424,
      "step": 5000
    },
    {
      "epoch": 0.8164763458401305,
      "grad_norm": 0.7126393914222717,
      "learning_rate": 4.0815660685154977e-05,
      "loss": 0.1138,
      "num_input_tokens_seen": 10830048,
      "step": 5005
    },
    {
      "epoch": 0.8172920065252854,
      "grad_norm": 1.9477592706680298,
      "learning_rate": 4.0856443719412724e-05,
      "loss": 0.2361,
      "num_input_tokens_seen": 10840768,
      "step": 5010
    },
    {
      "epoch": 0.8181076672104405,
      "grad_norm": 0.12592485547065735,
      "learning_rate": 4.089722675367048e-05,
      "loss": 0.2728,
      "num_input_tokens_seen": 10849792,
      "step": 5015
    },
    {
      "epoch": 0.8189233278955954,
      "grad_norm": 0.0797116607427597,
      "learning_rate": 4.093800978792822e-05,
      "loss": 0.0368,
      "num_input_tokens_seen": 10861472,
      "step": 5020
    },
    {
      "epoch": 0.8197389885807504,
      "grad_norm": 2.3102195262908936,
      "learning_rate": 4.0978792822185974e-05,
      "loss": 0.0623,
      "num_input_tokens_seen": 10871968,
      "step": 5025
    },
    {
      "epoch": 0.8205546492659054,
      "grad_norm": 3.467373847961426,
      "learning_rate": 4.101957585644372e-05,
      "loss": 0.1971,
      "num_input_tokens_seen": 10882560,
      "step": 5030
    },
    {
      "epoch": 0.8213703099510603,
      "grad_norm": 7.26324462890625,
      "learning_rate": 4.106035889070147e-05,
      "loss": 0.2299,
      "num_input_tokens_seen": 10894048,
      "step": 5035
    },
    {
      "epoch": 0.8221859706362153,
      "grad_norm": 6.0892438888549805,
      "learning_rate": 4.110114192495922e-05,
      "loss": 0.3053,
      "num_input_tokens_seen": 10905408,
      "step": 5040
    },
    {
      "epoch": 0.8230016313213703,
      "grad_norm": 3.208545684814453,
      "learning_rate": 4.1141924959216964e-05,
      "loss": 0.0953,
      "num_input_tokens_seen": 10915648,
      "step": 5045
    },
    {
      "epoch": 0.8238172920065253,
      "grad_norm": 2.9109947681427,
      "learning_rate": 4.118270799347472e-05,
      "loss": 0.1152,
      "num_input_tokens_seen": 10926880,
      "step": 5050
    },
    {
      "epoch": 0.8246329526916802,
      "grad_norm": 0.8327919840812683,
      "learning_rate": 4.1223491027732467e-05,
      "loss": 0.2608,
      "num_input_tokens_seen": 10938432,
      "step": 5055
    },
    {
      "epoch": 0.8254486133768353,
      "grad_norm": 1.2478437423706055,
      "learning_rate": 4.1264274061990214e-05,
      "loss": 0.2698,
      "num_input_tokens_seen": 10948992,
      "step": 5060
    },
    {
      "epoch": 0.8262642740619902,
      "grad_norm": 8.861740112304688,
      "learning_rate": 4.130505709624796e-05,
      "loss": 0.3661,
      "num_input_tokens_seen": 10958272,
      "step": 5065
    },
    {
      "epoch": 0.8270799347471451,
      "grad_norm": 0.43016546964645386,
      "learning_rate": 4.1345840130505716e-05,
      "loss": 0.1065,
      "num_input_tokens_seen": 10968512,
      "step": 5070
    },
    {
      "epoch": 0.8278955954323002,
      "grad_norm": 0.11255605518817902,
      "learning_rate": 4.1386623164763464e-05,
      "loss": 0.0838,
      "num_input_tokens_seen": 10979072,
      "step": 5075
    },
    {
      "epoch": 0.8287112561174551,
      "grad_norm": 0.8912560343742371,
      "learning_rate": 4.1427406199021205e-05,
      "loss": 0.1326,
      "num_input_tokens_seen": 10988736,
      "step": 5080
    },
    {
      "epoch": 0.8295269168026101,
      "grad_norm": 4.577517986297607,
      "learning_rate": 4.146818923327896e-05,
      "loss": 0.1894,
      "num_input_tokens_seen": 11000512,
      "step": 5085
    },
    {
      "epoch": 0.8303425774877651,
      "grad_norm": 1.794396162033081,
      "learning_rate": 4.150897226753671e-05,
      "loss": 0.2606,
      "num_input_tokens_seen": 11010240,
      "step": 5090
    },
    {
      "epoch": 0.8311582381729201,
      "grad_norm": 0.0777493417263031,
      "learning_rate": 4.1549755301794454e-05,
      "loss": 0.169,
      "num_input_tokens_seen": 11020672,
      "step": 5095
    },
    {
      "epoch": 0.831973898858075,
      "grad_norm": 0.19618628919124603,
      "learning_rate": 4.15905383360522e-05,
      "loss": 0.0844,
      "num_input_tokens_seen": 11031808,
      "step": 5100
    },
    {
      "epoch": 0.83278955954323,
      "grad_norm": 6.898108959197998,
      "learning_rate": 4.1631321370309957e-05,
      "loss": 0.2086,
      "num_input_tokens_seen": 11042144,
      "step": 5105
    },
    {
      "epoch": 0.833605220228385,
      "grad_norm": 0.18968331813812256,
      "learning_rate": 4.1672104404567704e-05,
      "loss": 0.1027,
      "num_input_tokens_seen": 11053728,
      "step": 5110
    },
    {
      "epoch": 0.8344208809135399,
      "grad_norm": 2.2396914958953857,
      "learning_rate": 4.171288743882545e-05,
      "loss": 0.1606,
      "num_input_tokens_seen": 11064032,
      "step": 5115
    },
    {
      "epoch": 0.835236541598695,
      "grad_norm": 4.880401134490967,
      "learning_rate": 4.17536704730832e-05,
      "loss": 0.2587,
      "num_input_tokens_seen": 11073792,
      "step": 5120
    },
    {
      "epoch": 0.8360522022838499,
      "grad_norm": 0.6538611650466919,
      "learning_rate": 4.179445350734095e-05,
      "loss": 0.0975,
      "num_input_tokens_seen": 11085216,
      "step": 5125
    },
    {
      "epoch": 0.8368678629690048,
      "grad_norm": 0.5744665861129761,
      "learning_rate": 4.18352365415987e-05,
      "loss": 0.2902,
      "num_input_tokens_seen": 11096288,
      "step": 5130
    },
    {
      "epoch": 0.8376835236541599,
      "grad_norm": 0.2094990760087967,
      "learning_rate": 4.187601957585644e-05,
      "loss": 0.127,
      "num_input_tokens_seen": 11107584,
      "step": 5135
    },
    {
      "epoch": 0.8384991843393148,
      "grad_norm": 0.8373374342918396,
      "learning_rate": 4.191680261011419e-05,
      "loss": 0.0899,
      "num_input_tokens_seen": 11118752,
      "step": 5140
    },
    {
      "epoch": 0.8393148450244698,
      "grad_norm": 2.5657176971435547,
      "learning_rate": 4.1957585644371944e-05,
      "loss": 0.1069,
      "num_input_tokens_seen": 11127840,
      "step": 5145
    },
    {
      "epoch": 0.8401305057096248,
      "grad_norm": 0.07352052628993988,
      "learning_rate": 4.199836867862969e-05,
      "loss": 0.1223,
      "num_input_tokens_seen": 11137952,
      "step": 5150
    },
    {
      "epoch": 0.8409461663947798,
      "grad_norm": 0.14405055344104767,
      "learning_rate": 4.203915171288744e-05,
      "loss": 0.2233,
      "num_input_tokens_seen": 11147008,
      "step": 5155
    },
    {
      "epoch": 0.8417618270799347,
      "grad_norm": 5.144182205200195,
      "learning_rate": 4.207993474714519e-05,
      "loss": 0.0808,
      "num_input_tokens_seen": 11158912,
      "step": 5160
    },
    {
      "epoch": 0.8425774877650897,
      "grad_norm": 1.5371413230895996,
      "learning_rate": 4.212071778140294e-05,
      "loss": 0.0364,
      "num_input_tokens_seen": 11169440,
      "step": 5165
    },
    {
      "epoch": 0.8433931484502447,
      "grad_norm": 2.367783784866333,
      "learning_rate": 4.216150081566069e-05,
      "loss": 0.1658,
      "num_input_tokens_seen": 11179488,
      "step": 5170
    },
    {
      "epoch": 0.8442088091353996,
      "grad_norm": 0.971149742603302,
      "learning_rate": 4.220228384991843e-05,
      "loss": 0.2325,
      "num_input_tokens_seen": 11190912,
      "step": 5175
    },
    {
      "epoch": 0.8450244698205547,
      "grad_norm": 2.1906208992004395,
      "learning_rate": 4.2243066884176185e-05,
      "loss": 0.1955,
      "num_input_tokens_seen": 11202208,
      "step": 5180
    },
    {
      "epoch": 0.8458401305057096,
      "grad_norm": 5.115178108215332,
      "learning_rate": 4.228384991843393e-05,
      "loss": 0.3306,
      "num_input_tokens_seen": 11212448,
      "step": 5185
    },
    {
      "epoch": 0.8466557911908646,
      "grad_norm": 1.3455501794815063,
      "learning_rate": 4.232463295269169e-05,
      "loss": 0.1725,
      "num_input_tokens_seen": 11223648,
      "step": 5190
    },
    {
      "epoch": 0.8474714518760196,
      "grad_norm": 0.13641172647476196,
      "learning_rate": 4.236541598694943e-05,
      "loss": 0.2347,
      "num_input_tokens_seen": 11234880,
      "step": 5195
    },
    {
      "epoch": 0.8482871125611745,
      "grad_norm": 1.0995267629623413,
      "learning_rate": 4.240619902120718e-05,
      "loss": 0.0977,
      "num_input_tokens_seen": 11245184,
      "step": 5200
    },
    {
      "epoch": 0.8491027732463295,
      "grad_norm": 3.476732015609741,
      "learning_rate": 4.244698205546493e-05,
      "loss": 0.066,
      "num_input_tokens_seen": 11256288,
      "step": 5205
    },
    {
      "epoch": 0.8499184339314845,
      "grad_norm": 0.5097530484199524,
      "learning_rate": 4.248776508972268e-05,
      "loss": 0.1736,
      "num_input_tokens_seen": 11265728,
      "step": 5210
    },
    {
      "epoch": 0.8507340946166395,
      "grad_norm": 0.16053344309329987,
      "learning_rate": 4.2528548123980425e-05,
      "loss": 0.0833,
      "num_input_tokens_seen": 11275712,
      "step": 5215
    },
    {
      "epoch": 0.8515497553017944,
      "grad_norm": 0.5644147992134094,
      "learning_rate": 4.256933115823817e-05,
      "loss": 0.1389,
      "num_input_tokens_seen": 11287776,
      "step": 5220
    },
    {
      "epoch": 0.8523654159869495,
      "grad_norm": 4.689136505126953,
      "learning_rate": 4.261011419249593e-05,
      "loss": 0.2571,
      "num_input_tokens_seen": 11298528,
      "step": 5225
    },
    {
      "epoch": 0.8531810766721044,
      "grad_norm": 1.5279276371002197,
      "learning_rate": 4.2650897226753675e-05,
      "loss": 0.2784,
      "num_input_tokens_seen": 11309536,
      "step": 5230
    },
    {
      "epoch": 0.8539967373572593,
      "grad_norm": 1.378790020942688,
      "learning_rate": 4.2691680261011416e-05,
      "loss": 0.3113,
      "num_input_tokens_seen": 11319200,
      "step": 5235
    },
    {
      "epoch": 0.8548123980424144,
      "grad_norm": 0.20116767287254333,
      "learning_rate": 4.273246329526917e-05,
      "loss": 0.0569,
      "num_input_tokens_seen": 11329120,
      "step": 5240
    },
    {
      "epoch": 0.8556280587275693,
      "grad_norm": 0.06263948231935501,
      "learning_rate": 4.277324632952692e-05,
      "loss": 0.1385,
      "num_input_tokens_seen": 11339360,
      "step": 5245
    },
    {
      "epoch": 0.8564437194127243,
      "grad_norm": 6.8799214363098145,
      "learning_rate": 4.281402936378467e-05,
      "loss": 0.2868,
      "num_input_tokens_seen": 11350848,
      "step": 5250
    },
    {
      "epoch": 0.8572593800978793,
      "grad_norm": 0.3296712040901184,
      "learning_rate": 4.285481239804241e-05,
      "loss": 0.1441,
      "num_input_tokens_seen": 11360704,
      "step": 5255
    },
    {
      "epoch": 0.8580750407830342,
      "grad_norm": 1.6185336112976074,
      "learning_rate": 4.289559543230017e-05,
      "loss": 0.0794,
      "num_input_tokens_seen": 11370816,
      "step": 5260
    },
    {
      "epoch": 0.8588907014681892,
      "grad_norm": 3.606036901473999,
      "learning_rate": 4.2936378466557915e-05,
      "loss": 0.1085,
      "num_input_tokens_seen": 11382080,
      "step": 5265
    },
    {
      "epoch": 0.8597063621533442,
      "grad_norm": 0.3723049461841583,
      "learning_rate": 4.297716150081566e-05,
      "loss": 0.2821,
      "num_input_tokens_seen": 11392736,
      "step": 5270
    },
    {
      "epoch": 0.8605220228384992,
      "grad_norm": 0.08519739657640457,
      "learning_rate": 4.301794453507341e-05,
      "loss": 0.1957,
      "num_input_tokens_seen": 11402720,
      "step": 5275
    },
    {
      "epoch": 0.8613376835236541,
      "grad_norm": 0.9225186109542847,
      "learning_rate": 4.305872756933116e-05,
      "loss": 0.0355,
      "num_input_tokens_seen": 11413984,
      "step": 5280
    },
    {
      "epoch": 0.8621533442088092,
      "grad_norm": 5.518332004547119,
      "learning_rate": 4.309951060358891e-05,
      "loss": 0.2303,
      "num_input_tokens_seen": 11424832,
      "step": 5285
    },
    {
      "epoch": 0.8629690048939641,
      "grad_norm": 2.6699180603027344,
      "learning_rate": 4.314029363784666e-05,
      "loss": 0.0447,
      "num_input_tokens_seen": 11436128,
      "step": 5290
    },
    {
      "epoch": 0.863784665579119,
      "grad_norm": 0.0447678305208683,
      "learning_rate": 4.318107667210441e-05,
      "loss": 0.2048,
      "num_input_tokens_seen": 11448448,
      "step": 5295
    },
    {
      "epoch": 0.8646003262642741,
      "grad_norm": 1.1556092500686646,
      "learning_rate": 4.3221859706362155e-05,
      "loss": 0.1352,
      "num_input_tokens_seen": 11458176,
      "step": 5300
    },
    {
      "epoch": 0.865415986949429,
      "grad_norm": 1.437555193901062,
      "learning_rate": 4.32626427406199e-05,
      "loss": 0.2217,
      "num_input_tokens_seen": 11468480,
      "step": 5305
    },
    {
      "epoch": 0.866231647634584,
      "grad_norm": 4.650927543640137,
      "learning_rate": 4.330342577487765e-05,
      "loss": 0.1743,
      "num_input_tokens_seen": 11478880,
      "step": 5310
    },
    {
      "epoch": 0.867047308319739,
      "grad_norm": 5.042733669281006,
      "learning_rate": 4.33442088091354e-05,
      "loss": 0.1615,
      "num_input_tokens_seen": 11490176,
      "step": 5315
    },
    {
      "epoch": 0.867862969004894,
      "grad_norm": 0.22132863104343414,
      "learning_rate": 4.338499184339315e-05,
      "loss": 0.1746,
      "num_input_tokens_seen": 11500992,
      "step": 5320
    },
    {
      "epoch": 0.8686786296900489,
      "grad_norm": 0.5395056009292603,
      "learning_rate": 4.34257748776509e-05,
      "loss": 0.0294,
      "num_input_tokens_seen": 11512288,
      "step": 5325
    },
    {
      "epoch": 0.8694942903752039,
      "grad_norm": 4.689180850982666,
      "learning_rate": 4.346655791190865e-05,
      "loss": 0.163,
      "num_input_tokens_seen": 11521664,
      "step": 5330
    },
    {
      "epoch": 0.8703099510603589,
      "grad_norm": 2.488198757171631,
      "learning_rate": 4.3507340946166396e-05,
      "loss": 0.0325,
      "num_input_tokens_seen": 11532608,
      "step": 5335
    },
    {
      "epoch": 0.8711256117455138,
      "grad_norm": 7.0070037841796875,
      "learning_rate": 4.354812398042414e-05,
      "loss": 0.2382,
      "num_input_tokens_seen": 11543072,
      "step": 5340
    },
    {
      "epoch": 0.8719412724306689,
      "grad_norm": 4.199175834655762,
      "learning_rate": 4.35889070146819e-05,
      "loss": 0.1242,
      "num_input_tokens_seen": 11554144,
      "step": 5345
    },
    {
      "epoch": 0.8727569331158238,
      "grad_norm": 0.19933462142944336,
      "learning_rate": 4.3629690048939645e-05,
      "loss": 0.1015,
      "num_input_tokens_seen": 11565376,
      "step": 5350
    },
    {
      "epoch": 0.8735725938009788,
      "grad_norm": 4.612356185913086,
      "learning_rate": 4.367047308319739e-05,
      "loss": 0.2884,
      "num_input_tokens_seen": 11576128,
      "step": 5355
    },
    {
      "epoch": 0.8743882544861338,
      "grad_norm": 0.09427787363529205,
      "learning_rate": 4.371125611745514e-05,
      "loss": 0.0866,
      "num_input_tokens_seen": 11588000,
      "step": 5360
    },
    {
      "epoch": 0.8752039151712887,
      "grad_norm": 0.5477355718612671,
      "learning_rate": 4.375203915171289e-05,
      "loss": 0.071,
      "num_input_tokens_seen": 11599520,
      "step": 5365
    },
    {
      "epoch": 0.8760195758564437,
      "grad_norm": 0.9199375510215759,
      "learning_rate": 4.3792822185970636e-05,
      "loss": 0.1351,
      "num_input_tokens_seen": 11611168,
      "step": 5370
    },
    {
      "epoch": 0.8768352365415987,
      "grad_norm": 0.3509751260280609,
      "learning_rate": 4.3833605220228384e-05,
      "loss": 0.0722,
      "num_input_tokens_seen": 11621632,
      "step": 5375
    },
    {
      "epoch": 0.8776508972267537,
      "grad_norm": 4.951541900634766,
      "learning_rate": 4.387438825448614e-05,
      "loss": 0.4848,
      "num_input_tokens_seen": 11632160,
      "step": 5380
    },
    {
      "epoch": 0.8784665579119086,
      "grad_norm": 0.33898553252220154,
      "learning_rate": 4.3915171288743886e-05,
      "loss": 0.1484,
      "num_input_tokens_seen": 11641344,
      "step": 5385
    },
    {
      "epoch": 0.8792822185970636,
      "grad_norm": 0.5188796520233154,
      "learning_rate": 4.395595432300163e-05,
      "loss": 0.1587,
      "num_input_tokens_seen": 11652128,
      "step": 5390
    },
    {
      "epoch": 0.8800978792822186,
      "grad_norm": 0.25006893277168274,
      "learning_rate": 4.399673735725938e-05,
      "loss": 0.0804,
      "num_input_tokens_seen": 11662528,
      "step": 5395
    },
    {
      "epoch": 0.8809135399673735,
      "grad_norm": 0.10641758143901825,
      "learning_rate": 4.403752039151713e-05,
      "loss": 0.1148,
      "num_input_tokens_seen": 11672544,
      "step": 5400
    },
    {
      "epoch": 0.8817292006525286,
      "grad_norm": 4.050087928771973,
      "learning_rate": 4.407830342577488e-05,
      "loss": 0.1352,
      "num_input_tokens_seen": 11683552,
      "step": 5405
    },
    {
      "epoch": 0.8825448613376835,
      "grad_norm": 2.528367042541504,
      "learning_rate": 4.4119086460032624e-05,
      "loss": 0.0982,
      "num_input_tokens_seen": 11693408,
      "step": 5410
    },
    {
      "epoch": 0.8833605220228385,
      "grad_norm": 0.33357101678848267,
      "learning_rate": 4.415986949429038e-05,
      "loss": 0.1551,
      "num_input_tokens_seen": 11704160,
      "step": 5415
    },
    {
      "epoch": 0.8841761827079935,
      "grad_norm": 0.32154837250709534,
      "learning_rate": 4.4200652528548126e-05,
      "loss": 0.1043,
      "num_input_tokens_seen": 11714752,
      "step": 5420
    },
    {
      "epoch": 0.8849918433931484,
      "grad_norm": 0.09626504778862,
      "learning_rate": 4.424143556280588e-05,
      "loss": 0.024,
      "num_input_tokens_seen": 11725408,
      "step": 5425
    },
    {
      "epoch": 0.8858075040783034,
      "grad_norm": 0.7662316560745239,
      "learning_rate": 4.428221859706362e-05,
      "loss": 0.1123,
      "num_input_tokens_seen": 11736640,
      "step": 5430
    },
    {
      "epoch": 0.8866231647634584,
      "grad_norm": 2.163783311843872,
      "learning_rate": 4.432300163132137e-05,
      "loss": 0.1954,
      "num_input_tokens_seen": 11747008,
      "step": 5435
    },
    {
      "epoch": 0.8874388254486134,
      "grad_norm": 3.4351084232330322,
      "learning_rate": 4.436378466557912e-05,
      "loss": 0.1193,
      "num_input_tokens_seen": 11757728,
      "step": 5440
    },
    {
      "epoch": 0.8882544861337683,
      "grad_norm": 0.2823643386363983,
      "learning_rate": 4.440456769983687e-05,
      "loss": 0.2107,
      "num_input_tokens_seen": 11768704,
      "step": 5445
    },
    {
      "epoch": 0.8890701468189234,
      "grad_norm": 11.116686820983887,
      "learning_rate": 4.444535073409462e-05,
      "loss": 0.1374,
      "num_input_tokens_seen": 11780288,
      "step": 5450
    },
    {
      "epoch": 0.8898858075040783,
      "grad_norm": 7.736528396606445,
      "learning_rate": 4.4486133768352366e-05,
      "loss": 0.5311,
      "num_input_tokens_seen": 11790080,
      "step": 5455
    },
    {
      "epoch": 0.8907014681892332,
      "grad_norm": 0.37017178535461426,
      "learning_rate": 4.4526916802610114e-05,
      "loss": 0.4017,
      "num_input_tokens_seen": 11801952,
      "step": 5460
    },
    {
      "epoch": 0.8915171288743883,
      "grad_norm": 0.2874157726764679,
      "learning_rate": 4.456769983686787e-05,
      "loss": 0.0954,
      "num_input_tokens_seen": 11811520,
      "step": 5465
    },
    {
      "epoch": 0.8923327895595432,
      "grad_norm": 0.4490336775779724,
      "learning_rate": 4.460848287112561e-05,
      "loss": 0.1741,
      "num_input_tokens_seen": 11822240,
      "step": 5470
    },
    {
      "epoch": 0.8931484502446982,
      "grad_norm": 5.030572414398193,
      "learning_rate": 4.4649265905383364e-05,
      "loss": 0.1663,
      "num_input_tokens_seen": 11833632,
      "step": 5475
    },
    {
      "epoch": 0.8939641109298532,
      "grad_norm": 0.028783371672034264,
      "learning_rate": 4.469004893964111e-05,
      "loss": 0.042,
      "num_input_tokens_seen": 11844640,
      "step": 5480
    },
    {
      "epoch": 0.8947797716150081,
      "grad_norm": 4.203812122344971,
      "learning_rate": 4.4730831973898866e-05,
      "loss": 0.1365,
      "num_input_tokens_seen": 11855776,
      "step": 5485
    },
    {
      "epoch": 0.8955954323001631,
      "grad_norm": 1.6102211475372314,
      "learning_rate": 4.4771615008156607e-05,
      "loss": 0.3933,
      "num_input_tokens_seen": 11865856,
      "step": 5490
    },
    {
      "epoch": 0.8964110929853181,
      "grad_norm": 2.1297247409820557,
      "learning_rate": 4.4812398042414354e-05,
      "loss": 0.0667,
      "num_input_tokens_seen": 11877504,
      "step": 5495
    },
    {
      "epoch": 0.8972267536704731,
      "grad_norm": 0.13318589329719543,
      "learning_rate": 4.485318107667211e-05,
      "loss": 0.2619,
      "num_input_tokens_seen": 11888320,
      "step": 5500
    },
    {
      "epoch": 0.898042414355628,
      "grad_norm": 0.3416684567928314,
      "learning_rate": 4.4893964110929856e-05,
      "loss": 0.1459,
      "num_input_tokens_seen": 11899456,
      "step": 5505
    },
    {
      "epoch": 0.8988580750407831,
      "grad_norm": 6.452507019042969,
      "learning_rate": 4.4934747145187604e-05,
      "loss": 0.1818,
      "num_input_tokens_seen": 11910240,
      "step": 5510
    },
    {
      "epoch": 0.899673735725938,
      "grad_norm": 0.9621750712394714,
      "learning_rate": 4.497553017944535e-05,
      "loss": 0.0758,
      "num_input_tokens_seen": 11920512,
      "step": 5515
    },
    {
      "epoch": 0.9004893964110929,
      "grad_norm": 0.2789660096168518,
      "learning_rate": 4.5016313213703106e-05,
      "loss": 0.1328,
      "num_input_tokens_seen": 11930592,
      "step": 5520
    },
    {
      "epoch": 0.901305057096248,
      "grad_norm": 2.8978145122528076,
      "learning_rate": 4.5057096247960854e-05,
      "loss": 0.371,
      "num_input_tokens_seen": 11941088,
      "step": 5525
    },
    {
      "epoch": 0.9021207177814029,
      "grad_norm": 1.2749956846237183,
      "learning_rate": 4.5097879282218594e-05,
      "loss": 0.1228,
      "num_input_tokens_seen": 11952320,
      "step": 5530
    },
    {
      "epoch": 0.9029363784665579,
      "grad_norm": 1.1564136743545532,
      "learning_rate": 4.513866231647635e-05,
      "loss": 0.1583,
      "num_input_tokens_seen": 11962880,
      "step": 5535
    },
    {
      "epoch": 0.9037520391517129,
      "grad_norm": 0.5101920962333679,
      "learning_rate": 4.5179445350734097e-05,
      "loss": 0.0189,
      "num_input_tokens_seen": 11973472,
      "step": 5540
    },
    {
      "epoch": 0.9045676998368679,
      "grad_norm": 4.70657205581665,
      "learning_rate": 4.5220228384991844e-05,
      "loss": 0.4949,
      "num_input_tokens_seen": 11984096,
      "step": 5545
    },
    {
      "epoch": 0.9053833605220228,
      "grad_norm": 3.420771360397339,
      "learning_rate": 4.526101141924959e-05,
      "loss": 0.1192,
      "num_input_tokens_seen": 11995488,
      "step": 5550
    },
    {
      "epoch": 0.9061990212071778,
      "grad_norm": 8.305743217468262,
      "learning_rate": 4.5301794453507346e-05,
      "loss": 0.1186,
      "num_input_tokens_seen": 12007104,
      "step": 5555
    },
    {
      "epoch": 0.9070146818923328,
      "grad_norm": 4.708660125732422,
      "learning_rate": 4.5342577487765094e-05,
      "loss": 0.195,
      "num_input_tokens_seen": 12018688,
      "step": 5560
    },
    {
      "epoch": 0.9078303425774877,
      "grad_norm": 0.42921167612075806,
      "learning_rate": 4.538336052202284e-05,
      "loss": 0.0933,
      "num_input_tokens_seen": 12028096,
      "step": 5565
    },
    {
      "epoch": 0.9086460032626428,
      "grad_norm": 3.1014652252197266,
      "learning_rate": 4.542414355628059e-05,
      "loss": 0.1672,
      "num_input_tokens_seen": 12038976,
      "step": 5570
    },
    {
      "epoch": 0.9094616639477977,
      "grad_norm": 2.6501834392547607,
      "learning_rate": 4.546492659053834e-05,
      "loss": 0.1629,
      "num_input_tokens_seen": 12050816,
      "step": 5575
    },
    {
      "epoch": 0.9102773246329527,
      "grad_norm": 0.2537744641304016,
      "learning_rate": 4.550570962479609e-05,
      "loss": 0.2324,
      "num_input_tokens_seen": 12062336,
      "step": 5580
    },
    {
      "epoch": 0.9110929853181077,
      "grad_norm": 6.535691261291504,
      "learning_rate": 4.554649265905383e-05,
      "loss": 0.2013,
      "num_input_tokens_seen": 12072096,
      "step": 5585
    },
    {
      "epoch": 0.9119086460032626,
      "grad_norm": 0.7936147451400757,
      "learning_rate": 4.558727569331158e-05,
      "loss": 0.0583,
      "num_input_tokens_seen": 12083136,
      "step": 5590
    },
    {
      "epoch": 0.9127243066884176,
      "grad_norm": 0.3955764174461365,
      "learning_rate": 4.5628058727569334e-05,
      "loss": 0.2451,
      "num_input_tokens_seen": 12094272,
      "step": 5595
    },
    {
      "epoch": 0.9135399673735726,
      "grad_norm": 0.2736872136592865,
      "learning_rate": 4.566884176182708e-05,
      "loss": 0.261,
      "num_input_tokens_seen": 12105280,
      "step": 5600
    },
    {
      "epoch": 0.9143556280587276,
      "grad_norm": 0.49092239141464233,
      "learning_rate": 4.570962479608483e-05,
      "loss": 0.1982,
      "num_input_tokens_seen": 12115968,
      "step": 5605
    },
    {
      "epoch": 0.9151712887438825,
      "grad_norm": 3.235626697540283,
      "learning_rate": 4.575040783034258e-05,
      "loss": 0.1825,
      "num_input_tokens_seen": 12127136,
      "step": 5610
    },
    {
      "epoch": 0.9159869494290375,
      "grad_norm": 5.939905166625977,
      "learning_rate": 4.579119086460033e-05,
      "loss": 0.0918,
      "num_input_tokens_seen": 12136704,
      "step": 5615
    },
    {
      "epoch": 0.9168026101141925,
      "grad_norm": 0.28519207239151,
      "learning_rate": 4.583197389885808e-05,
      "loss": 0.0813,
      "num_input_tokens_seen": 12146784,
      "step": 5620
    },
    {
      "epoch": 0.9176182707993474,
      "grad_norm": 2.29233980178833,
      "learning_rate": 4.587275693311583e-05,
      "loss": 0.0455,
      "num_input_tokens_seen": 12157248,
      "step": 5625
    },
    {
      "epoch": 0.9184339314845025,
      "grad_norm": 1.8638032674789429,
      "learning_rate": 4.5913539967373574e-05,
      "loss": 0.2188,
      "num_input_tokens_seen": 12168256,
      "step": 5630
    },
    {
      "epoch": 0.9192495921696574,
      "grad_norm": 3.1932172775268555,
      "learning_rate": 4.595432300163132e-05,
      "loss": 0.0907,
      "num_input_tokens_seen": 12177760,
      "step": 5635
    },
    {
      "epoch": 0.9200652528548124,
      "grad_norm": 2.8444621562957764,
      "learning_rate": 4.5995106035889077e-05,
      "loss": 0.0658,
      "num_input_tokens_seen": 12188192,
      "step": 5640
    },
    {
      "epoch": 0.9208809135399674,
      "grad_norm": 4.4717278480529785,
      "learning_rate": 4.603588907014682e-05,
      "loss": 0.2657,
      "num_input_tokens_seen": 12198880,
      "step": 5645
    },
    {
      "epoch": 0.9216965742251223,
      "grad_norm": 1.8809369802474976,
      "learning_rate": 4.607667210440457e-05,
      "loss": 0.2194,
      "num_input_tokens_seen": 12208800,
      "step": 5650
    },
    {
      "epoch": 0.9225122349102773,
      "grad_norm": 0.24707771837711334,
      "learning_rate": 4.611745513866232e-05,
      "loss": 0.1215,
      "num_input_tokens_seen": 12219360,
      "step": 5655
    },
    {
      "epoch": 0.9233278955954323,
      "grad_norm": 3.440296173095703,
      "learning_rate": 4.615823817292007e-05,
      "loss": 0.1499,
      "num_input_tokens_seen": 12229376,
      "step": 5660
    },
    {
      "epoch": 0.9241435562805873,
      "grad_norm": 4.585781574249268,
      "learning_rate": 4.6199021207177815e-05,
      "loss": 0.2547,
      "num_input_tokens_seen": 12241152,
      "step": 5665
    },
    {
      "epoch": 0.9249592169657422,
      "grad_norm": 0.6023845076560974,
      "learning_rate": 4.623980424143556e-05,
      "loss": 0.2379,
      "num_input_tokens_seen": 12252192,
      "step": 5670
    },
    {
      "epoch": 0.9257748776508973,
      "grad_norm": 0.4795178472995758,
      "learning_rate": 4.628058727569332e-05,
      "loss": 0.2099,
      "num_input_tokens_seen": 12262848,
      "step": 5675
    },
    {
      "epoch": 0.9265905383360522,
      "grad_norm": 3.0352981090545654,
      "learning_rate": 4.6321370309951064e-05,
      "loss": 0.1927,
      "num_input_tokens_seen": 12273728,
      "step": 5680
    },
    {
      "epoch": 0.9274061990212071,
      "grad_norm": 2.9030098915100098,
      "learning_rate": 4.636215334420881e-05,
      "loss": 0.1262,
      "num_input_tokens_seen": 12284512,
      "step": 5685
    },
    {
      "epoch": 0.9282218597063622,
      "grad_norm": 5.807152271270752,
      "learning_rate": 4.640293637846656e-05,
      "loss": 0.2835,
      "num_input_tokens_seen": 12295968,
      "step": 5690
    },
    {
      "epoch": 0.9290375203915171,
      "grad_norm": 0.8265318274497986,
      "learning_rate": 4.644371941272431e-05,
      "loss": 0.1558,
      "num_input_tokens_seen": 12307616,
      "step": 5695
    },
    {
      "epoch": 0.9298531810766721,
      "grad_norm": 3.8171863555908203,
      "learning_rate": 4.648450244698206e-05,
      "loss": 0.1775,
      "num_input_tokens_seen": 12318400,
      "step": 5700
    },
    {
      "epoch": 0.9306688417618271,
      "grad_norm": 0.7892698049545288,
      "learning_rate": 4.65252854812398e-05,
      "loss": 0.1465,
      "num_input_tokens_seen": 12329472,
      "step": 5705
    },
    {
      "epoch": 0.9314845024469821,
      "grad_norm": 2.0436034202575684,
      "learning_rate": 4.656606851549756e-05,
      "loss": 0.0662,
      "num_input_tokens_seen": 12340032,
      "step": 5710
    },
    {
      "epoch": 0.932300163132137,
      "grad_norm": 0.7949777841567993,
      "learning_rate": 4.6606851549755305e-05,
      "loss": 0.0445,
      "num_input_tokens_seen": 12351584,
      "step": 5715
    },
    {
      "epoch": 0.933115823817292,
      "grad_norm": 4.621256351470947,
      "learning_rate": 4.664763458401305e-05,
      "loss": 0.3786,
      "num_input_tokens_seen": 12362752,
      "step": 5720
    },
    {
      "epoch": 0.933931484502447,
      "grad_norm": 2.6562387943267822,
      "learning_rate": 4.66884176182708e-05,
      "loss": 0.2519,
      "num_input_tokens_seen": 12374720,
      "step": 5725
    },
    {
      "epoch": 0.9347471451876019,
      "grad_norm": 0.8015565276145935,
      "learning_rate": 4.672920065252855e-05,
      "loss": 0.1531,
      "num_input_tokens_seen": 12384896,
      "step": 5730
    },
    {
      "epoch": 0.935562805872757,
      "grad_norm": 1.8839294910430908,
      "learning_rate": 4.67699836867863e-05,
      "loss": 0.099,
      "num_input_tokens_seen": 12395808,
      "step": 5735
    },
    {
      "epoch": 0.9363784665579119,
      "grad_norm": 2.145501136779785,
      "learning_rate": 4.681076672104405e-05,
      "loss": 0.1181,
      "num_input_tokens_seen": 12406880,
      "step": 5740
    },
    {
      "epoch": 0.9371941272430668,
      "grad_norm": 1.8614461421966553,
      "learning_rate": 4.68515497553018e-05,
      "loss": 0.2624,
      "num_input_tokens_seen": 12418240,
      "step": 5745
    },
    {
      "epoch": 0.9380097879282219,
      "grad_norm": 4.2035346031188965,
      "learning_rate": 4.6892332789559545e-05,
      "loss": 0.2653,
      "num_input_tokens_seen": 12428896,
      "step": 5750
    },
    {
      "epoch": 0.9388254486133768,
      "grad_norm": 0.3168288767337799,
      "learning_rate": 4.693311582381729e-05,
      "loss": 0.0554,
      "num_input_tokens_seen": 12439872,
      "step": 5755
    },
    {
      "epoch": 0.9396411092985318,
      "grad_norm": 0.09010463953018188,
      "learning_rate": 4.697389885807505e-05,
      "loss": 0.0137,
      "num_input_tokens_seen": 12451840,
      "step": 5760
    },
    {
      "epoch": 0.9404567699836868,
      "grad_norm": 0.5660171508789062,
      "learning_rate": 4.701468189233279e-05,
      "loss": 0.1133,
      "num_input_tokens_seen": 12463328,
      "step": 5765
    },
    {
      "epoch": 0.9412724306688418,
      "grad_norm": 0.30117490887641907,
      "learning_rate": 4.705546492659054e-05,
      "loss": 0.2971,
      "num_input_tokens_seen": 12474080,
      "step": 5770
    },
    {
      "epoch": 0.9420880913539967,
      "grad_norm": 0.11573529243469238,
      "learning_rate": 4.709624796084829e-05,
      "loss": 0.4065,
      "num_input_tokens_seen": 12484960,
      "step": 5775
    },
    {
      "epoch": 0.9429037520391517,
      "grad_norm": 0.47013425827026367,
      "learning_rate": 4.713703099510604e-05,
      "loss": 0.0553,
      "num_input_tokens_seen": 12495584,
      "step": 5780
    },
    {
      "epoch": 0.9437194127243067,
      "grad_norm": 0.387765109539032,
      "learning_rate": 4.7177814029363785e-05,
      "loss": 0.157,
      "num_input_tokens_seen": 12505568,
      "step": 5785
    },
    {
      "epoch": 0.9445350734094616,
      "grad_norm": 0.29751497507095337,
      "learning_rate": 4.721859706362153e-05,
      "loss": 0.3058,
      "num_input_tokens_seen": 12516480,
      "step": 5790
    },
    {
      "epoch": 0.9453507340946167,
      "grad_norm": 5.351414203643799,
      "learning_rate": 4.725938009787929e-05,
      "loss": 0.2212,
      "num_input_tokens_seen": 12527136,
      "step": 5795
    },
    {
      "epoch": 0.9461663947797716,
      "grad_norm": 0.13309234380722046,
      "learning_rate": 4.7300163132137035e-05,
      "loss": 0.3603,
      "num_input_tokens_seen": 12537568,
      "step": 5800
    },
    {
      "epoch": 0.9469820554649266,
      "grad_norm": 0.9149518609046936,
      "learning_rate": 4.734094616639478e-05,
      "loss": 0.0854,
      "num_input_tokens_seen": 12548736,
      "step": 5805
    },
    {
      "epoch": 0.9477977161500816,
      "grad_norm": 0.08186270296573639,
      "learning_rate": 4.738172920065253e-05,
      "loss": 0.1006,
      "num_input_tokens_seen": 12560192,
      "step": 5810
    },
    {
      "epoch": 0.9486133768352365,
      "grad_norm": 0.45442628860473633,
      "learning_rate": 4.742251223491028e-05,
      "loss": 0.0944,
      "num_input_tokens_seen": 12570816,
      "step": 5815
    },
    {
      "epoch": 0.9494290375203915,
      "grad_norm": 8.15549087524414,
      "learning_rate": 4.7463295269168026e-05,
      "loss": 0.3167,
      "num_input_tokens_seen": 12582272,
      "step": 5820
    },
    {
      "epoch": 0.9502446982055465,
      "grad_norm": 0.2367493063211441,
      "learning_rate": 4.750407830342577e-05,
      "loss": 0.2528,
      "num_input_tokens_seen": 12594432,
      "step": 5825
    },
    {
      "epoch": 0.9510603588907015,
      "grad_norm": 0.2754059135913849,
      "learning_rate": 4.754486133768353e-05,
      "loss": 0.0117,
      "num_input_tokens_seen": 12606048,
      "step": 5830
    },
    {
      "epoch": 0.9518760195758564,
      "grad_norm": 0.514473557472229,
      "learning_rate": 4.7585644371941275e-05,
      "loss": 0.2696,
      "num_input_tokens_seen": 12617088,
      "step": 5835
    },
    {
      "epoch": 0.9526916802610114,
      "grad_norm": 4.245872974395752,
      "learning_rate": 4.762642740619902e-05,
      "loss": 0.2253,
      "num_input_tokens_seen": 12627808,
      "step": 5840
    },
    {
      "epoch": 0.9535073409461664,
      "grad_norm": 0.13982197642326355,
      "learning_rate": 4.766721044045677e-05,
      "loss": 0.0519,
      "num_input_tokens_seen": 12638720,
      "step": 5845
    },
    {
      "epoch": 0.9543230016313213,
      "grad_norm": 2.4407217502593994,
      "learning_rate": 4.770799347471452e-05,
      "loss": 0.1421,
      "num_input_tokens_seen": 12648608,
      "step": 5850
    },
    {
      "epoch": 0.9551386623164764,
      "grad_norm": 1.4845774173736572,
      "learning_rate": 4.774877650897227e-05,
      "loss": 0.0421,
      "num_input_tokens_seen": 12658944,
      "step": 5855
    },
    {
      "epoch": 0.9559543230016313,
      "grad_norm": 5.853550434112549,
      "learning_rate": 4.7789559543230014e-05,
      "loss": 0.165,
      "num_input_tokens_seen": 12669856,
      "step": 5860
    },
    {
      "epoch": 0.9567699836867863,
      "grad_norm": 1.4343255758285522,
      "learning_rate": 4.783034257748777e-05,
      "loss": 0.198,
      "num_input_tokens_seen": 12680096,
      "step": 5865
    },
    {
      "epoch": 0.9575856443719413,
      "grad_norm": 0.08989223092794418,
      "learning_rate": 4.7871125611745516e-05,
      "loss": 0.0122,
      "num_input_tokens_seen": 12691040,
      "step": 5870
    },
    {
      "epoch": 0.9584013050570962,
      "grad_norm": 0.16980361938476562,
      "learning_rate": 4.791190864600327e-05,
      "loss": 0.1666,
      "num_input_tokens_seen": 12701728,
      "step": 5875
    },
    {
      "epoch": 0.9592169657422512,
      "grad_norm": 0.14723515510559082,
      "learning_rate": 4.795269168026101e-05,
      "loss": 0.0835,
      "num_input_tokens_seen": 12712224,
      "step": 5880
    },
    {
      "epoch": 0.9600326264274062,
      "grad_norm": 2.7446348667144775,
      "learning_rate": 4.799347471451876e-05,
      "loss": 0.215,
      "num_input_tokens_seen": 12723264,
      "step": 5885
    },
    {
      "epoch": 0.9608482871125612,
      "grad_norm": 0.6583641171455383,
      "learning_rate": 4.803425774877651e-05,
      "loss": 0.2323,
      "num_input_tokens_seen": 12734400,
      "step": 5890
    },
    {
      "epoch": 0.9616639477977161,
      "grad_norm": 0.018399178981781006,
      "learning_rate": 4.807504078303426e-05,
      "loss": 0.1507,
      "num_input_tokens_seen": 12745504,
      "step": 5895
    },
    {
      "epoch": 0.9624796084828712,
      "grad_norm": 4.040611743927002,
      "learning_rate": 4.811582381729201e-05,
      "loss": 0.2672,
      "num_input_tokens_seen": 12755520,
      "step": 5900
    },
    {
      "epoch": 0.9632952691680261,
      "grad_norm": 0.750511884689331,
      "learning_rate": 4.8156606851549756e-05,
      "loss": 0.0282,
      "num_input_tokens_seen": 12766336,
      "step": 5905
    },
    {
      "epoch": 0.964110929853181,
      "grad_norm": 3.4301741123199463,
      "learning_rate": 4.819738988580751e-05,
      "loss": 0.1245,
      "num_input_tokens_seen": 12776704,
      "step": 5910
    },
    {
      "epoch": 0.9649265905383361,
      "grad_norm": 0.17444053292274475,
      "learning_rate": 4.823817292006526e-05,
      "loss": 0.296,
      "num_input_tokens_seen": 12787872,
      "step": 5915
    },
    {
      "epoch": 0.965742251223491,
      "grad_norm": 0.4860982298851013,
      "learning_rate": 4.8278955954323e-05,
      "loss": 0.0428,
      "num_input_tokens_seen": 12798880,
      "step": 5920
    },
    {
      "epoch": 0.966557911908646,
      "grad_norm": 5.006972789764404,
      "learning_rate": 4.831973898858075e-05,
      "loss": 0.283,
      "num_input_tokens_seen": 12809696,
      "step": 5925
    },
    {
      "epoch": 0.967373572593801,
      "grad_norm": 4.0059709548950195,
      "learning_rate": 4.83605220228385e-05,
      "loss": 0.1566,
      "num_input_tokens_seen": 12820192,
      "step": 5930
    },
    {
      "epoch": 0.968189233278956,
      "grad_norm": 5.839953422546387,
      "learning_rate": 4.8401305057096255e-05,
      "loss": 0.0971,
      "num_input_tokens_seen": 12832384,
      "step": 5935
    },
    {
      "epoch": 0.9690048939641109,
      "grad_norm": 0.6276607513427734,
      "learning_rate": 4.8442088091353996e-05,
      "loss": 0.076,
      "num_input_tokens_seen": 12843936,
      "step": 5940
    },
    {
      "epoch": 0.9698205546492659,
      "grad_norm": 0.4621135890483856,
      "learning_rate": 4.8482871125611744e-05,
      "loss": 0.1962,
      "num_input_tokens_seen": 12854752,
      "step": 5945
    },
    {
      "epoch": 0.9706362153344209,
      "grad_norm": 2.2090275287628174,
      "learning_rate": 4.85236541598695e-05,
      "loss": 0.0434,
      "num_input_tokens_seen": 12865248,
      "step": 5950
    },
    {
      "epoch": 0.9714518760195758,
      "grad_norm": 5.447160720825195,
      "learning_rate": 4.8564437194127246e-05,
      "loss": 0.1976,
      "num_input_tokens_seen": 12876128,
      "step": 5955
    },
    {
      "epoch": 0.9722675367047309,
      "grad_norm": 2.3920841217041016,
      "learning_rate": 4.8605220228384994e-05,
      "loss": 0.2233,
      "num_input_tokens_seen": 12887008,
      "step": 5960
    },
    {
      "epoch": 0.9730831973898858,
      "grad_norm": 0.03698760271072388,
      "learning_rate": 4.864600326264274e-05,
      "loss": 0.0925,
      "num_input_tokens_seen": 12898624,
      "step": 5965
    },
    {
      "epoch": 0.9738988580750407,
      "grad_norm": 0.3625604212284088,
      "learning_rate": 4.8686786296900496e-05,
      "loss": 0.0934,
      "num_input_tokens_seen": 12909888,
      "step": 5970
    },
    {
      "epoch": 0.9747145187601958,
      "grad_norm": 10.285679817199707,
      "learning_rate": 4.872756933115824e-05,
      "loss": 0.4727,
      "num_input_tokens_seen": 12920032,
      "step": 5975
    },
    {
      "epoch": 0.9755301794453507,
      "grad_norm": 7.591075897216797,
      "learning_rate": 4.8768352365415984e-05,
      "loss": 0.4998,
      "num_input_tokens_seen": 12931392,
      "step": 5980
    },
    {
      "epoch": 0.9763458401305057,
      "grad_norm": 4.060094833374023,
      "learning_rate": 4.880913539967374e-05,
      "loss": 0.2346,
      "num_input_tokens_seen": 12941472,
      "step": 5985
    },
    {
      "epoch": 0.9771615008156607,
      "grad_norm": 2.5794951915740967,
      "learning_rate": 4.8849918433931486e-05,
      "loss": 0.1454,
      "num_input_tokens_seen": 12952192,
      "step": 5990
    },
    {
      "epoch": 0.9779771615008157,
      "grad_norm": 3.3665897846221924,
      "learning_rate": 4.8890701468189234e-05,
      "loss": 0.2409,
      "num_input_tokens_seen": 12963200,
      "step": 5995
    },
    {
      "epoch": 0.9787928221859706,
      "grad_norm": 0.7922720909118652,
      "learning_rate": 4.893148450244698e-05,
      "loss": 0.2427,
      "num_input_tokens_seen": 12972768,
      "step": 6000
    },
    {
      "epoch": 0.9796084828711256,
      "grad_norm": 2.926109790802002,
      "learning_rate": 4.8972267536704736e-05,
      "loss": 0.0478,
      "num_input_tokens_seen": 12984064,
      "step": 6005
    },
    {
      "epoch": 0.9804241435562806,
      "grad_norm": 0.7572949528694153,
      "learning_rate": 4.9013050570962484e-05,
      "loss": 0.1912,
      "num_input_tokens_seen": 12994336,
      "step": 6010
    },
    {
      "epoch": 0.9812398042414355,
      "grad_norm": 0.14401741325855255,
      "learning_rate": 4.905383360522023e-05,
      "loss": 0.0152,
      "num_input_tokens_seen": 13004096,
      "step": 6015
    },
    {
      "epoch": 0.9820554649265906,
      "grad_norm": 0.6650245189666748,
      "learning_rate": 4.909461663947798e-05,
      "loss": 0.0563,
      "num_input_tokens_seen": 13015168,
      "step": 6020
    },
    {
      "epoch": 0.9828711256117455,
      "grad_norm": 0.5241230130195618,
      "learning_rate": 4.9135399673735727e-05,
      "loss": 0.0966,
      "num_input_tokens_seen": 13026080,
      "step": 6025
    },
    {
      "epoch": 0.9836867862969005,
      "grad_norm": 5.424735069274902,
      "learning_rate": 4.917618270799348e-05,
      "loss": 0.3274,
      "num_input_tokens_seen": 13037888,
      "step": 6030
    },
    {
      "epoch": 0.9845024469820555,
      "grad_norm": 6.97016716003418,
      "learning_rate": 4.921696574225123e-05,
      "loss": 0.4111,
      "num_input_tokens_seen": 13049440,
      "step": 6035
    },
    {
      "epoch": 0.9853181076672104,
      "grad_norm": 0.3491359055042267,
      "learning_rate": 4.9257748776508976e-05,
      "loss": 0.0414,
      "num_input_tokens_seen": 13060480,
      "step": 6040
    },
    {
      "epoch": 0.9861337683523654,
      "grad_norm": 0.2605774402618408,
      "learning_rate": 4.9298531810766724e-05,
      "loss": 0.0565,
      "num_input_tokens_seen": 13071584,
      "step": 6045
    },
    {
      "epoch": 0.9869494290375204,
      "grad_norm": 3.8022637367248535,
      "learning_rate": 4.933931484502447e-05,
      "loss": 0.1964,
      "num_input_tokens_seen": 13081376,
      "step": 6050
    },
    {
      "epoch": 0.9877650897226754,
      "grad_norm": 0.08020448684692383,
      "learning_rate": 4.938009787928222e-05,
      "loss": 0.0841,
      "num_input_tokens_seen": 13092768,
      "step": 6055
    },
    {
      "epoch": 0.9885807504078303,
      "grad_norm": 0.8128753900527954,
      "learning_rate": 4.942088091353997e-05,
      "loss": 0.2018,
      "num_input_tokens_seen": 13104096,
      "step": 6060
    },
    {
      "epoch": 0.9893964110929854,
      "grad_norm": 0.16580741107463837,
      "learning_rate": 4.946166394779772e-05,
      "loss": 0.1163,
      "num_input_tokens_seen": 13115488,
      "step": 6065
    },
    {
      "epoch": 0.9902120717781403,
      "grad_norm": 0.20980337262153625,
      "learning_rate": 4.950244698205547e-05,
      "loss": 0.0744,
      "num_input_tokens_seen": 13126752,
      "step": 6070
    },
    {
      "epoch": 0.9910277324632952,
      "grad_norm": 0.9464369416236877,
      "learning_rate": 4.9543230016313217e-05,
      "loss": 0.1535,
      "num_input_tokens_seen": 13137920,
      "step": 6075
    },
    {
      "epoch": 0.9918433931484503,
      "grad_norm": 3.200751781463623,
      "learning_rate": 4.9584013050570964e-05,
      "loss": 0.2442,
      "num_input_tokens_seen": 13148416,
      "step": 6080
    },
    {
      "epoch": 0.9926590538336052,
      "grad_norm": 5.150273323059082,
      "learning_rate": 4.962479608482871e-05,
      "loss": 0.1137,
      "num_input_tokens_seen": 13157920,
      "step": 6085
    },
    {
      "epoch": 0.9934747145187602,
      "grad_norm": 0.9907219409942627,
      "learning_rate": 4.9665579119086466e-05,
      "loss": 0.0343,
      "num_input_tokens_seen": 13167968,
      "step": 6090
    },
    {
      "epoch": 0.9942903752039152,
      "grad_norm": 5.316732883453369,
      "learning_rate": 4.970636215334421e-05,
      "loss": 0.2873,
      "num_input_tokens_seen": 13179648,
      "step": 6095
    },
    {
      "epoch": 0.9951060358890701,
      "grad_norm": 0.18861132860183716,
      "learning_rate": 4.974714518760196e-05,
      "loss": 0.1942,
      "num_input_tokens_seen": 13190656,
      "step": 6100
    },
    {
      "epoch": 0.9959216965742251,
      "grad_norm": 0.1188916265964508,
      "learning_rate": 4.978792822185971e-05,
      "loss": 0.1605,
      "num_input_tokens_seen": 13201120,
      "step": 6105
    },
    {
      "epoch": 0.9967373572593801,
      "grad_norm": 1.468907117843628,
      "learning_rate": 4.982871125611746e-05,
      "loss": 0.102,
      "num_input_tokens_seen": 13211744,
      "step": 6110
    },
    {
      "epoch": 0.9975530179445351,
      "grad_norm": 0.04991580545902252,
      "learning_rate": 4.9869494290375205e-05,
      "loss": 0.1987,
      "num_input_tokens_seen": 13222752,
      "step": 6115
    },
    {
      "epoch": 0.99836867862969,
      "grad_norm": 0.611285924911499,
      "learning_rate": 4.991027732463295e-05,
      "loss": 0.2471,
      "num_input_tokens_seen": 13233696,
      "step": 6120
    },
    {
      "epoch": 0.9991843393148451,
      "grad_norm": 0.6825065612792969,
      "learning_rate": 4.9951060358890707e-05,
      "loss": 0.1463,
      "num_input_tokens_seen": 13245504,
      "step": 6125
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.14032478630542755,
      "learning_rate": 4.9991843393148454e-05,
      "loss": 0.1984,
      "num_input_tokens_seen": 13255424,
      "step": 6130
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.15720832347869873,
      "eval_runtime": 133.1334,
      "eval_samples_per_second": 20.468,
      "eval_steps_per_second": 5.123,
      "num_input_tokens_seen": 13255424,
      "step": 6130
    },
    {
      "epoch": 1.000815660685155,
      "grad_norm": 3.067218780517578,
      "learning_rate": 4.999999935147941e-05,
      "loss": 0.1373,
      "num_input_tokens_seen": 13265952,
      "step": 6135
    },
    {
      "epoch": 1.0016313213703099,
      "grad_norm": 0.8992579579353333,
      "learning_rate": 4.999999671686456e-05,
      "loss": 0.0564,
      "num_input_tokens_seen": 13276768,
      "step": 6140
    },
    {
      "epoch": 1.002446982055465,
      "grad_norm": 1.2270188331604004,
      "learning_rate": 4.999999205562312e-05,
      "loss": 0.1712,
      "num_input_tokens_seen": 13287776,
      "step": 6145
    },
    {
      "epoch": 1.00326264274062,
      "grad_norm": 0.5810588598251343,
      "learning_rate": 4.999998536775549e-05,
      "loss": 0.3012,
      "num_input_tokens_seen": 13299488,
      "step": 6150
    },
    {
      "epoch": 1.004078303425775,
      "grad_norm": 2.3066442012786865,
      "learning_rate": 4.9999976653262184e-05,
      "loss": 0.1016,
      "num_input_tokens_seen": 13310208,
      "step": 6155
    },
    {
      "epoch": 1.0048939641109298,
      "grad_norm": 0.11014427244663239,
      "learning_rate": 4.999996591214392e-05,
      "loss": 0.0257,
      "num_input_tokens_seen": 13319904,
      "step": 6160
    },
    {
      "epoch": 1.0057096247960848,
      "grad_norm": 0.6082314252853394,
      "learning_rate": 4.999995314440158e-05,
      "loss": 0.1063,
      "num_input_tokens_seen": 13330528,
      "step": 6165
    },
    {
      "epoch": 1.0065252854812399,
      "grad_norm": 0.16786378622055054,
      "learning_rate": 4.999993835003618e-05,
      "loss": 0.1076,
      "num_input_tokens_seen": 13341760,
      "step": 6170
    },
    {
      "epoch": 1.0073409461663947,
      "grad_norm": 0.052277714014053345,
      "learning_rate": 4.9999921529048945e-05,
      "loss": 0.2745,
      "num_input_tokens_seen": 13352448,
      "step": 6175
    },
    {
      "epoch": 1.0081566068515497,
      "grad_norm": 3.4511992931365967,
      "learning_rate": 4.9999902681441205e-05,
      "loss": 0.2549,
      "num_input_tokens_seen": 13363296,
      "step": 6180
    },
    {
      "epoch": 1.0089722675367048,
      "grad_norm": 1.0652965307235718,
      "learning_rate": 4.9999881807214515e-05,
      "loss": 0.0328,
      "num_input_tokens_seen": 13374368,
      "step": 6185
    },
    {
      "epoch": 1.0097879282218598,
      "grad_norm": 3.025557518005371,
      "learning_rate": 4.9999858906370553e-05,
      "loss": 0.0905,
      "num_input_tokens_seen": 13385184,
      "step": 6190
    },
    {
      "epoch": 1.0106035889070146,
      "grad_norm": 3.7201688289642334,
      "learning_rate": 4.9999833978911185e-05,
      "loss": 0.2842,
      "num_input_tokens_seen": 13395104,
      "step": 6195
    },
    {
      "epoch": 1.0114192495921697,
      "grad_norm": 0.9407456517219543,
      "learning_rate": 4.999980702483842e-05,
      "loss": 0.0479,
      "num_input_tokens_seen": 13406080,
      "step": 6200
    },
    {
      "epoch": 1.0122349102773247,
      "grad_norm": 1.9295414686203003,
      "learning_rate": 4.999977804415446e-05,
      "loss": 0.2064,
      "num_input_tokens_seen": 13417952,
      "step": 6205
    },
    {
      "epoch": 1.0130505709624795,
      "grad_norm": 5.365414619445801,
      "learning_rate": 4.999974703686164e-05,
      "loss": 0.1248,
      "num_input_tokens_seen": 13428512,
      "step": 6210
    },
    {
      "epoch": 1.0138662316476346,
      "grad_norm": 0.3215469419956207,
      "learning_rate": 4.9999714002962474e-05,
      "loss": 0.0202,
      "num_input_tokens_seen": 13439648,
      "step": 6215
    },
    {
      "epoch": 1.0146818923327896,
      "grad_norm": 0.15902908146381378,
      "learning_rate": 4.999967894245965e-05,
      "loss": 0.1377,
      "num_input_tokens_seen": 13451520,
      "step": 6220
    },
    {
      "epoch": 1.0154975530179446,
      "grad_norm": 3.213712215423584,
      "learning_rate": 4.9999641855355995e-05,
      "loss": 0.1352,
      "num_input_tokens_seen": 13460480,
      "step": 6225
    },
    {
      "epoch": 1.0163132137030995,
      "grad_norm": 2.431142568588257,
      "learning_rate": 4.999960274165453e-05,
      "loss": 0.0963,
      "num_input_tokens_seen": 13471328,
      "step": 6230
    },
    {
      "epoch": 1.0171288743882545,
      "grad_norm": 7.266343593597412,
      "learning_rate": 4.999956160135842e-05,
      "loss": 0.2119,
      "num_input_tokens_seen": 13482048,
      "step": 6235
    },
    {
      "epoch": 1.0179445350734095,
      "grad_norm": 1.8660143613815308,
      "learning_rate": 4.999951843447099e-05,
      "loss": 0.1378,
      "num_input_tokens_seen": 13492512,
      "step": 6240
    },
    {
      "epoch": 1.0187601957585644,
      "grad_norm": 3.638964891433716,
      "learning_rate": 4.999947324099576e-05,
      "loss": 0.1171,
      "num_input_tokens_seen": 13504000,
      "step": 6245
    },
    {
      "epoch": 1.0195758564437194,
      "grad_norm": 0.7262020111083984,
      "learning_rate": 4.999942602093638e-05,
      "loss": 0.1194,
      "num_input_tokens_seen": 13514784,
      "step": 6250
    },
    {
      "epoch": 1.0203915171288744,
      "grad_norm": 0.6040750741958618,
      "learning_rate": 4.999937677429669e-05,
      "loss": 0.1189,
      "num_input_tokens_seen": 13525376,
      "step": 6255
    },
    {
      "epoch": 1.0212071778140293,
      "grad_norm": 0.39444753527641296,
      "learning_rate": 4.9999325501080666e-05,
      "loss": 0.2655,
      "num_input_tokens_seen": 13535872,
      "step": 6260
    },
    {
      "epoch": 1.0220228384991843,
      "grad_norm": 1.2012454271316528,
      "learning_rate": 4.999927220129247e-05,
      "loss": 0.1076,
      "num_input_tokens_seen": 13546464,
      "step": 6265
    },
    {
      "epoch": 1.0228384991843393,
      "grad_norm": 0.8436634540557861,
      "learning_rate": 4.9999216874936426e-05,
      "loss": 0.2635,
      "num_input_tokens_seen": 13555936,
      "step": 6270
    },
    {
      "epoch": 1.0236541598694944,
      "grad_norm": 0.17298643290996552,
      "learning_rate": 4.9999159522017015e-05,
      "loss": 0.0112,
      "num_input_tokens_seen": 13566528,
      "step": 6275
    },
    {
      "epoch": 1.0244698205546492,
      "grad_norm": 0.07080504298210144,
      "learning_rate": 4.999910014253889e-05,
      "loss": 0.2112,
      "num_input_tokens_seen": 13577152,
      "step": 6280
    },
    {
      "epoch": 1.0252854812398042,
      "grad_norm": 4.6885528564453125,
      "learning_rate": 4.999903873650687e-05,
      "loss": 0.1048,
      "num_input_tokens_seen": 13588704,
      "step": 6285
    },
    {
      "epoch": 1.0261011419249593,
      "grad_norm": 0.9719001650810242,
      "learning_rate": 4.999897530392591e-05,
      "loss": 0.0153,
      "num_input_tokens_seen": 13599552,
      "step": 6290
    },
    {
      "epoch": 1.026916802610114,
      "grad_norm": 0.26410770416259766,
      "learning_rate": 4.9998909844801176e-05,
      "loss": 0.0158,
      "num_input_tokens_seen": 13611264,
      "step": 6295
    },
    {
      "epoch": 1.0277324632952691,
      "grad_norm": 3.8483128547668457,
      "learning_rate": 4.999884235913797e-05,
      "loss": 0.2487,
      "num_input_tokens_seen": 13621760,
      "step": 6300
    },
    {
      "epoch": 1.0285481239804242,
      "grad_norm": 3.1497063636779785,
      "learning_rate": 4.999877284694177e-05,
      "loss": 0.1964,
      "num_input_tokens_seen": 13632672,
      "step": 6305
    },
    {
      "epoch": 1.0293637846655792,
      "grad_norm": 0.11703000217676163,
      "learning_rate": 4.999870130821818e-05,
      "loss": 0.0095,
      "num_input_tokens_seen": 13643584,
      "step": 6310
    },
    {
      "epoch": 1.030179445350734,
      "grad_norm": 0.10111640393733978,
      "learning_rate": 4.9998627742973025e-05,
      "loss": 0.3219,
      "num_input_tokens_seen": 13655488,
      "step": 6315
    },
    {
      "epoch": 1.030995106035889,
      "grad_norm": 2.653560161590576,
      "learning_rate": 4.9998552151212276e-05,
      "loss": 0.1615,
      "num_input_tokens_seen": 13666368,
      "step": 6320
    },
    {
      "epoch": 1.031810766721044,
      "grad_norm": 1.1174497604370117,
      "learning_rate": 4.999847453294204e-05,
      "loss": 0.2452,
      "num_input_tokens_seen": 13678048,
      "step": 6325
    },
    {
      "epoch": 1.032626427406199,
      "grad_norm": 2.7836174964904785,
      "learning_rate": 4.999839488816861e-05,
      "loss": 0.0305,
      "num_input_tokens_seen": 13689024,
      "step": 6330
    },
    {
      "epoch": 1.033442088091354,
      "grad_norm": 0.1570463925600052,
      "learning_rate": 4.9998313216898454e-05,
      "loss": 0.1406,
      "num_input_tokens_seen": 13699744,
      "step": 6335
    },
    {
      "epoch": 1.034257748776509,
      "grad_norm": 0.2226400524377823,
      "learning_rate": 4.99982295191382e-05,
      "loss": 0.0638,
      "num_input_tokens_seen": 13710016,
      "step": 6340
    },
    {
      "epoch": 1.035073409461664,
      "grad_norm": 2.299910306930542,
      "learning_rate": 4.99981437948946e-05,
      "loss": 0.1252,
      "num_input_tokens_seen": 13721664,
      "step": 6345
    },
    {
      "epoch": 1.0358890701468189,
      "grad_norm": 3.804659605026245,
      "learning_rate": 4.999805604417464e-05,
      "loss": 0.0408,
      "num_input_tokens_seen": 13734752,
      "step": 6350
    },
    {
      "epoch": 1.036704730831974,
      "grad_norm": 0.15337339043617249,
      "learning_rate": 4.999796626698542e-05,
      "loss": 0.0818,
      "num_input_tokens_seen": 13745504,
      "step": 6355
    },
    {
      "epoch": 1.037520391517129,
      "grad_norm": 6.057324409484863,
      "learning_rate": 4.999787446333421e-05,
      "loss": 0.1707,
      "num_input_tokens_seen": 13755968,
      "step": 6360
    },
    {
      "epoch": 1.0383360522022838,
      "grad_norm": 3.1725149154663086,
      "learning_rate": 4.999778063322846e-05,
      "loss": 0.193,
      "num_input_tokens_seen": 13767296,
      "step": 6365
    },
    {
      "epoch": 1.0391517128874388,
      "grad_norm": 0.1678391546010971,
      "learning_rate": 4.9997684776675775e-05,
      "loss": 0.2717,
      "num_input_tokens_seen": 13778656,
      "step": 6370
    },
    {
      "epoch": 1.0399673735725938,
      "grad_norm": 6.550046920776367,
      "learning_rate": 4.999758689368392e-05,
      "loss": 0.1537,
      "num_input_tokens_seen": 13789216,
      "step": 6375
    },
    {
      "epoch": 1.0407830342577489,
      "grad_norm": 4.02315616607666,
      "learning_rate": 4.999748698426084e-05,
      "loss": 0.0584,
      "num_input_tokens_seen": 13799680,
      "step": 6380
    },
    {
      "epoch": 1.0415986949429037,
      "grad_norm": 0.07721056789159775,
      "learning_rate": 4.9997385048414624e-05,
      "loss": 0.1017,
      "num_input_tokens_seen": 13810720,
      "step": 6385
    },
    {
      "epoch": 1.0424143556280587,
      "grad_norm": 0.5864396691322327,
      "learning_rate": 4.999728108615355e-05,
      "loss": 0.1158,
      "num_input_tokens_seen": 13821856,
      "step": 6390
    },
    {
      "epoch": 1.0432300163132138,
      "grad_norm": 0.8936943411827087,
      "learning_rate": 4.9997175097486026e-05,
      "loss": 0.2071,
      "num_input_tokens_seen": 13833984,
      "step": 6395
    },
    {
      "epoch": 1.0440456769983686,
      "grad_norm": 3.2895848751068115,
      "learning_rate": 4.9997067082420655e-05,
      "loss": 0.3046,
      "num_input_tokens_seen": 13844928,
      "step": 6400
    },
    {
      "epoch": 1.0448613376835236,
      "grad_norm": 7.387818813323975,
      "learning_rate": 4.9996957040966205e-05,
      "loss": 0.1894,
      "num_input_tokens_seen": 13856384,
      "step": 6405
    },
    {
      "epoch": 1.0456769983686787,
      "grad_norm": 0.20572587847709656,
      "learning_rate": 4.999684497313157e-05,
      "loss": 0.1028,
      "num_input_tokens_seen": 13867456,
      "step": 6410
    },
    {
      "epoch": 1.0464926590538337,
      "grad_norm": 1.095292329788208,
      "learning_rate": 4.9996730878925856e-05,
      "loss": 0.0209,
      "num_input_tokens_seen": 13877216,
      "step": 6415
    },
    {
      "epoch": 1.0473083197389885,
      "grad_norm": 4.3961181640625,
      "learning_rate": 4.99966147583583e-05,
      "loss": 0.3864,
      "num_input_tokens_seen": 13888192,
      "step": 6420
    },
    {
      "epoch": 1.0481239804241436,
      "grad_norm": 0.32123053073883057,
      "learning_rate": 4.9996496611438326e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 13899072,
      "step": 6425
    },
    {
      "epoch": 1.0489396411092986,
      "grad_norm": 7.540583610534668,
      "learning_rate": 4.99963764381755e-05,
      "loss": 0.2563,
      "num_input_tokens_seen": 13909792,
      "step": 6430
    },
    {
      "epoch": 1.0497553017944534,
      "grad_norm": 0.44093719124794006,
      "learning_rate": 4.999625423857958e-05,
      "loss": 0.0607,
      "num_input_tokens_seen": 13920800,
      "step": 6435
    },
    {
      "epoch": 1.0505709624796085,
      "grad_norm": 1.7876123189926147,
      "learning_rate": 4.999613001266045e-05,
      "loss": 0.0533,
      "num_input_tokens_seen": 13930976,
      "step": 6440
    },
    {
      "epoch": 1.0513866231647635,
      "grad_norm": 0.38826173543930054,
      "learning_rate": 4.999600376042819e-05,
      "loss": 0.0381,
      "num_input_tokens_seen": 13942080,
      "step": 6445
    },
    {
      "epoch": 1.0522022838499185,
      "grad_norm": 5.167582035064697,
      "learning_rate": 4.999587548189305e-05,
      "loss": 0.0599,
      "num_input_tokens_seen": 13952736,
      "step": 6450
    },
    {
      "epoch": 1.0530179445350734,
      "grad_norm": 3.2204172611236572,
      "learning_rate": 4.99957451770654e-05,
      "loss": 0.1036,
      "num_input_tokens_seen": 13962400,
      "step": 6455
    },
    {
      "epoch": 1.0538336052202284,
      "grad_norm": 0.04041092470288277,
      "learning_rate": 4.999561284595583e-05,
      "loss": 0.0162,
      "num_input_tokens_seen": 13974304,
      "step": 6460
    },
    {
      "epoch": 1.0546492659053834,
      "grad_norm": 3.1175785064697266,
      "learning_rate": 4.9995478488575054e-05,
      "loss": 0.1,
      "num_input_tokens_seen": 13985248,
      "step": 6465
    },
    {
      "epoch": 1.0554649265905383,
      "grad_norm": 5.50757360458374,
      "learning_rate": 4.999534210493396e-05,
      "loss": 0.0569,
      "num_input_tokens_seen": 13995840,
      "step": 6470
    },
    {
      "epoch": 1.0562805872756933,
      "grad_norm": 0.15792645514011383,
      "learning_rate": 4.999520369504362e-05,
      "loss": 0.1343,
      "num_input_tokens_seen": 14007104,
      "step": 6475
    },
    {
      "epoch": 1.0570962479608483,
      "grad_norm": 4.697031497955322,
      "learning_rate": 4.9995063258915235e-05,
      "loss": 0.0593,
      "num_input_tokens_seen": 14018144,
      "step": 6480
    },
    {
      "epoch": 1.0579119086460032,
      "grad_norm": 1.4745863676071167,
      "learning_rate": 4.9994920796560205e-05,
      "loss": 0.0781,
      "num_input_tokens_seen": 14029024,
      "step": 6485
    },
    {
      "epoch": 1.0587275693311582,
      "grad_norm": 0.05364038795232773,
      "learning_rate": 4.999477630799007e-05,
      "loss": 0.1515,
      "num_input_tokens_seen": 14040480,
      "step": 6490
    },
    {
      "epoch": 1.0595432300163132,
      "grad_norm": 0.04595933482050896,
      "learning_rate": 4.999462979321654e-05,
      "loss": 0.0023,
      "num_input_tokens_seen": 14050976,
      "step": 6495
    },
    {
      "epoch": 1.0603588907014683,
      "grad_norm": 1.792112112045288,
      "learning_rate": 4.9994481252251506e-05,
      "loss": 0.0109,
      "num_input_tokens_seen": 14060352,
      "step": 6500
    },
    {
      "epoch": 1.061174551386623,
      "grad_norm": 19.34042739868164,
      "learning_rate": 4.999433068510699e-05,
      "loss": 0.1274,
      "num_input_tokens_seen": 14071680,
      "step": 6505
    },
    {
      "epoch": 1.0619902120717781,
      "grad_norm": 2.285905599594116,
      "learning_rate": 4.999417809179523e-05,
      "loss": 0.0847,
      "num_input_tokens_seen": 14083296,
      "step": 6510
    },
    {
      "epoch": 1.0628058727569332,
      "grad_norm": 5.264870643615723,
      "learning_rate": 4.9994023472328555e-05,
      "loss": 0.1433,
      "num_input_tokens_seen": 14094752,
      "step": 6515
    },
    {
      "epoch": 1.0636215334420882,
      "grad_norm": 7.430415630340576,
      "learning_rate": 4.999386682671953e-05,
      "loss": 0.372,
      "num_input_tokens_seen": 14105088,
      "step": 6520
    },
    {
      "epoch": 1.064437194127243,
      "grad_norm": 5.810910224914551,
      "learning_rate": 4.9993708154980836e-05,
      "loss": 0.1906,
      "num_input_tokens_seen": 14114976,
      "step": 6525
    },
    {
      "epoch": 1.065252854812398,
      "grad_norm": 4.199947834014893,
      "learning_rate": 4.999354745712534e-05,
      "loss": 0.2266,
      "num_input_tokens_seen": 14126592,
      "step": 6530
    },
    {
      "epoch": 1.066068515497553,
      "grad_norm": 0.7832189798355103,
      "learning_rate": 4.999338473316607e-05,
      "loss": 0.1949,
      "num_input_tokens_seen": 14138048,
      "step": 6535
    },
    {
      "epoch": 1.066884176182708,
      "grad_norm": 0.1015256941318512,
      "learning_rate": 4.9993219983116223e-05,
      "loss": 0.0962,
      "num_input_tokens_seen": 14148128,
      "step": 6540
    },
    {
      "epoch": 1.067699836867863,
      "grad_norm": 3.800353765487671,
      "learning_rate": 4.999305320698915e-05,
      "loss": 0.1037,
      "num_input_tokens_seen": 14160096,
      "step": 6545
    },
    {
      "epoch": 1.068515497553018,
      "grad_norm": 0.3930117189884186,
      "learning_rate": 4.999288440479837e-05,
      "loss": 0.0725,
      "num_input_tokens_seen": 14170752,
      "step": 6550
    },
    {
      "epoch": 1.0693311582381728,
      "grad_norm": 0.09467582404613495,
      "learning_rate": 4.999271357655757e-05,
      "loss": 0.3594,
      "num_input_tokens_seen": 14181952,
      "step": 6555
    },
    {
      "epoch": 1.0701468189233279,
      "grad_norm": 0.18013601005077362,
      "learning_rate": 4.999254072228059e-05,
      "loss": 0.1734,
      "num_input_tokens_seen": 14192384,
      "step": 6560
    },
    {
      "epoch": 1.070962479608483,
      "grad_norm": 0.04747767001390457,
      "learning_rate": 4.9992365841981456e-05,
      "loss": 0.0124,
      "num_input_tokens_seen": 14204384,
      "step": 6565
    },
    {
      "epoch": 1.071778140293638,
      "grad_norm": 0.08544167876243591,
      "learning_rate": 4.9992188935674335e-05,
      "loss": 0.1268,
      "num_input_tokens_seen": 14215776,
      "step": 6570
    },
    {
      "epoch": 1.0725938009787928,
      "grad_norm": 0.8479664921760559,
      "learning_rate": 4.999201000337356e-05,
      "loss": 0.1506,
      "num_input_tokens_seen": 14227520,
      "step": 6575
    },
    {
      "epoch": 1.0734094616639478,
      "grad_norm": 0.21483114361763,
      "learning_rate": 4.999182904509366e-05,
      "loss": 0.2186,
      "num_input_tokens_seen": 14238976,
      "step": 6580
    },
    {
      "epoch": 1.0742251223491028,
      "grad_norm": 0.05625781789422035,
      "learning_rate": 4.9991646060849285e-05,
      "loss": 0.1296,
      "num_input_tokens_seen": 14249344,
      "step": 6585
    },
    {
      "epoch": 1.0750407830342577,
      "grad_norm": 0.15453799068927765,
      "learning_rate": 4.9991461050655264e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 14259840,
      "step": 6590
    },
    {
      "epoch": 1.0758564437194127,
      "grad_norm": 0.21485309302806854,
      "learning_rate": 4.999127401452662e-05,
      "loss": 0.2197,
      "num_input_tokens_seen": 14271456,
      "step": 6595
    },
    {
      "epoch": 1.0766721044045677,
      "grad_norm": 4.267156600952148,
      "learning_rate": 4.999108495247849e-05,
      "loss": 0.1725,
      "num_input_tokens_seen": 14282688,
      "step": 6600
    },
    {
      "epoch": 1.0774877650897228,
      "grad_norm": 4.1854681968688965,
      "learning_rate": 4.9990893864526214e-05,
      "loss": 0.1068,
      "num_input_tokens_seen": 14293920,
      "step": 6605
    },
    {
      "epoch": 1.0783034257748776,
      "grad_norm": 2.9711430072784424,
      "learning_rate": 4.9990700750685276e-05,
      "loss": 0.0176,
      "num_input_tokens_seen": 14304512,
      "step": 6610
    },
    {
      "epoch": 1.0791190864600326,
      "grad_norm": 1.481642246246338,
      "learning_rate": 4.999050561097134e-05,
      "loss": 0.0216,
      "num_input_tokens_seen": 14315616,
      "step": 6615
    },
    {
      "epoch": 1.0799347471451877,
      "grad_norm": 1.1094454526901245,
      "learning_rate": 4.999030844540021e-05,
      "loss": 0.0354,
      "num_input_tokens_seen": 14326400,
      "step": 6620
    },
    {
      "epoch": 1.0807504078303425,
      "grad_norm": 0.04047788307070732,
      "learning_rate": 4.999010925398788e-05,
      "loss": 0.2067,
      "num_input_tokens_seen": 14336288,
      "step": 6625
    },
    {
      "epoch": 1.0815660685154975,
      "grad_norm": 0.31624218821525574,
      "learning_rate": 4.99899080367505e-05,
      "loss": 0.1622,
      "num_input_tokens_seen": 14347616,
      "step": 6630
    },
    {
      "epoch": 1.0823817292006526,
      "grad_norm": 0.08543675392866135,
      "learning_rate": 4.9989704793704374e-05,
      "loss": 0.1119,
      "num_input_tokens_seen": 14357664,
      "step": 6635
    },
    {
      "epoch": 1.0831973898858076,
      "grad_norm": 5.039618968963623,
      "learning_rate": 4.998949952486598e-05,
      "loss": 0.4289,
      "num_input_tokens_seen": 14367232,
      "step": 6640
    },
    {
      "epoch": 1.0840130505709624,
      "grad_norm": 0.1791653037071228,
      "learning_rate": 4.998929223025196e-05,
      "loss": 0.1644,
      "num_input_tokens_seen": 14378048,
      "step": 6645
    },
    {
      "epoch": 1.0848287112561175,
      "grad_norm": 1.6573693752288818,
      "learning_rate": 4.998908290987913e-05,
      "loss": 0.1897,
      "num_input_tokens_seen": 14389248,
      "step": 6650
    },
    {
      "epoch": 1.0856443719412725,
      "grad_norm": 0.6262145042419434,
      "learning_rate": 4.998887156376443e-05,
      "loss": 0.0761,
      "num_input_tokens_seen": 14399616,
      "step": 6655
    },
    {
      "epoch": 1.0864600326264273,
      "grad_norm": 7.90919828414917,
      "learning_rate": 4.998865819192501e-05,
      "loss": 0.3444,
      "num_input_tokens_seen": 14412000,
      "step": 6660
    },
    {
      "epoch": 1.0872756933115824,
      "grad_norm": 0.22059565782546997,
      "learning_rate": 4.9988442794378166e-05,
      "loss": 0.2168,
      "num_input_tokens_seen": 14423392,
      "step": 6665
    },
    {
      "epoch": 1.0880913539967374,
      "grad_norm": 2.1804144382476807,
      "learning_rate": 4.998822537114136e-05,
      "loss": 0.2435,
      "num_input_tokens_seen": 14433792,
      "step": 6670
    },
    {
      "epoch": 1.0889070146818924,
      "grad_norm": 0.1316068172454834,
      "learning_rate": 4.998800592223222e-05,
      "loss": 0.0807,
      "num_input_tokens_seen": 14445152,
      "step": 6675
    },
    {
      "epoch": 1.0897226753670473,
      "grad_norm": 0.2838009297847748,
      "learning_rate": 4.9987784447668526e-05,
      "loss": 0.125,
      "num_input_tokens_seen": 14455968,
      "step": 6680
    },
    {
      "epoch": 1.0905383360522023,
      "grad_norm": 0.4855414032936096,
      "learning_rate": 4.9987560947468245e-05,
      "loss": 0.047,
      "num_input_tokens_seen": 14465824,
      "step": 6685
    },
    {
      "epoch": 1.0913539967373573,
      "grad_norm": 6.95814847946167,
      "learning_rate": 4.998733542164948e-05,
      "loss": 0.2047,
      "num_input_tokens_seen": 14476480,
      "step": 6690
    },
    {
      "epoch": 1.0921696574225122,
      "grad_norm": 0.47371330857276917,
      "learning_rate": 4.998710787023053e-05,
      "loss": 0.1393,
      "num_input_tokens_seen": 14486144,
      "step": 6695
    },
    {
      "epoch": 1.0929853181076672,
      "grad_norm": 1.3137682676315308,
      "learning_rate": 4.998687829322983e-05,
      "loss": 0.2674,
      "num_input_tokens_seen": 14497664,
      "step": 6700
    },
    {
      "epoch": 1.0938009787928222,
      "grad_norm": 4.352360725402832,
      "learning_rate": 4.9986646690665996e-05,
      "loss": 0.1401,
      "num_input_tokens_seen": 14509888,
      "step": 6705
    },
    {
      "epoch": 1.094616639477977,
      "grad_norm": 1.3058325052261353,
      "learning_rate": 4.998641306255779e-05,
      "loss": 0.1986,
      "num_input_tokens_seen": 14521664,
      "step": 6710
    },
    {
      "epoch": 1.095432300163132,
      "grad_norm": 1.6888068914413452,
      "learning_rate": 4.998617740892417e-05,
      "loss": 0.1728,
      "num_input_tokens_seen": 14532768,
      "step": 6715
    },
    {
      "epoch": 1.0962479608482871,
      "grad_norm": 1.2989392280578613,
      "learning_rate": 4.998593972978423e-05,
      "loss": 0.1047,
      "num_input_tokens_seen": 14544160,
      "step": 6720
    },
    {
      "epoch": 1.0970636215334422,
      "grad_norm": 0.5585448145866394,
      "learning_rate": 4.9985700025157236e-05,
      "loss": 0.0322,
      "num_input_tokens_seen": 14554112,
      "step": 6725
    },
    {
      "epoch": 1.097879282218597,
      "grad_norm": 1.6414457559585571,
      "learning_rate": 4.998545829506263e-05,
      "loss": 0.0519,
      "num_input_tokens_seen": 14564096,
      "step": 6730
    },
    {
      "epoch": 1.098694942903752,
      "grad_norm": 0.15382681787014008,
      "learning_rate": 4.998521453951999e-05,
      "loss": 0.1204,
      "num_input_tokens_seen": 14574592,
      "step": 6735
    },
    {
      "epoch": 1.099510603588907,
      "grad_norm": 0.043996699154376984,
      "learning_rate": 4.998496875854908e-05,
      "loss": 0.015,
      "num_input_tokens_seen": 14585728,
      "step": 6740
    },
    {
      "epoch": 1.100326264274062,
      "grad_norm": 2.8113901615142822,
      "learning_rate": 4.998472095216984e-05,
      "loss": 0.2483,
      "num_input_tokens_seen": 14594464,
      "step": 6745
    },
    {
      "epoch": 1.101141924959217,
      "grad_norm": 3.563047409057617,
      "learning_rate": 4.998447112040235e-05,
      "loss": 0.1717,
      "num_input_tokens_seen": 14604480,
      "step": 6750
    },
    {
      "epoch": 1.101957585644372,
      "grad_norm": 3.7106142044067383,
      "learning_rate": 4.998421926326685e-05,
      "loss": 0.0621,
      "num_input_tokens_seen": 14616128,
      "step": 6755
    },
    {
      "epoch": 1.102773246329527,
      "grad_norm": 0.08062312752008438,
      "learning_rate": 4.998396538078378e-05,
      "loss": 0.0147,
      "num_input_tokens_seen": 14627008,
      "step": 6760
    },
    {
      "epoch": 1.1035889070146818,
      "grad_norm": 1.1190317869186401,
      "learning_rate": 4.99837094729737e-05,
      "loss": 0.2473,
      "num_input_tokens_seen": 14637280,
      "step": 6765
    },
    {
      "epoch": 1.1044045676998369,
      "grad_norm": 5.4377264976501465,
      "learning_rate": 4.998345153985738e-05,
      "loss": 0.1444,
      "num_input_tokens_seen": 14646976,
      "step": 6770
    },
    {
      "epoch": 1.105220228384992,
      "grad_norm": 19.340038299560547,
      "learning_rate": 4.998319158145569e-05,
      "loss": 0.0803,
      "num_input_tokens_seen": 14657152,
      "step": 6775
    },
    {
      "epoch": 1.1060358890701467,
      "grad_norm": 0.9310303926467896,
      "learning_rate": 4.998292959778974e-05,
      "loss": 0.3306,
      "num_input_tokens_seen": 14667488,
      "step": 6780
    },
    {
      "epoch": 1.1068515497553018,
      "grad_norm": 7.0268168449401855,
      "learning_rate": 4.9982665588880753e-05,
      "loss": 0.2986,
      "num_input_tokens_seen": 14678112,
      "step": 6785
    },
    {
      "epoch": 1.1076672104404568,
      "grad_norm": 0.04533948376774788,
      "learning_rate": 4.9982399554750136e-05,
      "loss": 0.0561,
      "num_input_tokens_seen": 14689312,
      "step": 6790
    },
    {
      "epoch": 1.1084828711256118,
      "grad_norm": 0.22335876524448395,
      "learning_rate": 4.998213149541945e-05,
      "loss": 0.3764,
      "num_input_tokens_seen": 14700032,
      "step": 6795
    },
    {
      "epoch": 1.1092985318107667,
      "grad_norm": 2.8416543006896973,
      "learning_rate": 4.9981861410910424e-05,
      "loss": 0.2424,
      "num_input_tokens_seen": 14711008,
      "step": 6800
    },
    {
      "epoch": 1.1101141924959217,
      "grad_norm": 0.06697509437799454,
      "learning_rate": 4.9981589301244956e-05,
      "loss": 0.0252,
      "num_input_tokens_seen": 14720448,
      "step": 6805
    },
    {
      "epoch": 1.1109298531810767,
      "grad_norm": 0.13166627287864685,
      "learning_rate": 4.99813151664451e-05,
      "loss": 0.0869,
      "num_input_tokens_seen": 14732128,
      "step": 6810
    },
    {
      "epoch": 1.1117455138662315,
      "grad_norm": 0.06577765941619873,
      "learning_rate": 4.998103900653309e-05,
      "loss": 0.1219,
      "num_input_tokens_seen": 14743200,
      "step": 6815
    },
    {
      "epoch": 1.1125611745513866,
      "grad_norm": 4.53542947769165,
      "learning_rate": 4.9980760821531304e-05,
      "loss": 0.3091,
      "num_input_tokens_seen": 14754464,
      "step": 6820
    },
    {
      "epoch": 1.1133768352365416,
      "grad_norm": 3.502213716506958,
      "learning_rate": 4.998048061146229e-05,
      "loss": 0.2777,
      "num_input_tokens_seen": 14763840,
      "step": 6825
    },
    {
      "epoch": 1.1141924959216967,
      "grad_norm": 0.13414250314235687,
      "learning_rate": 4.9980198376348774e-05,
      "loss": 0.0437,
      "num_input_tokens_seen": 14774304,
      "step": 6830
    },
    {
      "epoch": 1.1150081566068515,
      "grad_norm": 6.36775541305542,
      "learning_rate": 4.997991411621362e-05,
      "loss": 0.0415,
      "num_input_tokens_seen": 14783456,
      "step": 6835
    },
    {
      "epoch": 1.1158238172920065,
      "grad_norm": 0.055438391864299774,
      "learning_rate": 4.9979627831079894e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 14793664,
      "step": 6840
    },
    {
      "epoch": 1.1166394779771616,
      "grad_norm": 2.260453462600708,
      "learning_rate": 4.997933952097078e-05,
      "loss": 0.0955,
      "num_input_tokens_seen": 14804256,
      "step": 6845
    },
    {
      "epoch": 1.1174551386623164,
      "grad_norm": 0.5315571427345276,
      "learning_rate": 4.997904918590966e-05,
      "loss": 0.2423,
      "num_input_tokens_seen": 14815744,
      "step": 6850
    },
    {
      "epoch": 1.1182707993474714,
      "grad_norm": 0.12336158007383347,
      "learning_rate": 4.997875682592008e-05,
      "loss": 0.0575,
      "num_input_tokens_seen": 14826304,
      "step": 6855
    },
    {
      "epoch": 1.1190864600326265,
      "grad_norm": 0.6235188841819763,
      "learning_rate": 4.997846244102573e-05,
      "loss": 0.2058,
      "num_input_tokens_seen": 14838016,
      "step": 6860
    },
    {
      "epoch": 1.1199021207177815,
      "grad_norm": 2.3068888187408447,
      "learning_rate": 4.997816603125047e-05,
      "loss": 0.0434,
      "num_input_tokens_seen": 14849344,
      "step": 6865
    },
    {
      "epoch": 1.1207177814029363,
      "grad_norm": 3.278022527694702,
      "learning_rate": 4.9977867596618333e-05,
      "loss": 0.2902,
      "num_input_tokens_seen": 14860608,
      "step": 6870
    },
    {
      "epoch": 1.1215334420880914,
      "grad_norm": 5.233441352844238,
      "learning_rate": 4.997756713715352e-05,
      "loss": 0.0885,
      "num_input_tokens_seen": 14870976,
      "step": 6875
    },
    {
      "epoch": 1.1223491027732464,
      "grad_norm": 0.9062994718551636,
      "learning_rate": 4.997726465288037e-05,
      "loss": 0.0712,
      "num_input_tokens_seen": 14880960,
      "step": 6880
    },
    {
      "epoch": 1.1231647634584012,
      "grad_norm": 0.045212745666503906,
      "learning_rate": 4.997696014382341e-05,
      "loss": 0.1472,
      "num_input_tokens_seen": 14892256,
      "step": 6885
    },
    {
      "epoch": 1.1239804241435563,
      "grad_norm": 1.470868468284607,
      "learning_rate": 4.997665361000735e-05,
      "loss": 0.3008,
      "num_input_tokens_seen": 14902304,
      "step": 6890
    },
    {
      "epoch": 1.1247960848287113,
      "grad_norm": 0.13788996636867523,
      "learning_rate": 4.9976345051456995e-05,
      "loss": 0.132,
      "num_input_tokens_seen": 14913536,
      "step": 6895
    },
    {
      "epoch": 1.1256117455138663,
      "grad_norm": 5.891739368438721,
      "learning_rate": 4.99760344681974e-05,
      "loss": 0.2338,
      "num_input_tokens_seen": 14924480,
      "step": 6900
    },
    {
      "epoch": 1.1264274061990212,
      "grad_norm": 2.3570706844329834,
      "learning_rate": 4.997572186025371e-05,
      "loss": 0.1666,
      "num_input_tokens_seen": 14935264,
      "step": 6905
    },
    {
      "epoch": 1.1272430668841762,
      "grad_norm": 5.320467948913574,
      "learning_rate": 4.997540722765128e-05,
      "loss": 0.1228,
      "num_input_tokens_seen": 14947072,
      "step": 6910
    },
    {
      "epoch": 1.1280587275693312,
      "grad_norm": 2.584219217300415,
      "learning_rate": 4.997509057041563e-05,
      "loss": 0.1896,
      "num_input_tokens_seen": 14957888,
      "step": 6915
    },
    {
      "epoch": 1.128874388254486,
      "grad_norm": 2.8470618724823,
      "learning_rate": 4.99747718885724e-05,
      "loss": 0.2429,
      "num_input_tokens_seen": 14967424,
      "step": 6920
    },
    {
      "epoch": 1.129690048939641,
      "grad_norm": 2.975456953048706,
      "learning_rate": 4.9974451182147456e-05,
      "loss": 0.1532,
      "num_input_tokens_seen": 14977536,
      "step": 6925
    },
    {
      "epoch": 1.1305057096247961,
      "grad_norm": 0.3898094594478607,
      "learning_rate": 4.997412845116677e-05,
      "loss": 0.0593,
      "num_input_tokens_seen": 14987008,
      "step": 6930
    },
    {
      "epoch": 1.131321370309951,
      "grad_norm": 0.1392330378293991,
      "learning_rate": 4.997380369565652e-05,
      "loss": 0.39,
      "num_input_tokens_seen": 14998272,
      "step": 6935
    },
    {
      "epoch": 1.132137030995106,
      "grad_norm": 4.5014495849609375,
      "learning_rate": 4.9973476915643015e-05,
      "loss": 0.2747,
      "num_input_tokens_seen": 15007392,
      "step": 6940
    },
    {
      "epoch": 1.132952691680261,
      "grad_norm": 3.5496082305908203,
      "learning_rate": 4.997314811115277e-05,
      "loss": 0.312,
      "num_input_tokens_seen": 15016608,
      "step": 6945
    },
    {
      "epoch": 1.133768352365416,
      "grad_norm": 2.735983371734619,
      "learning_rate": 4.997281728221242e-05,
      "loss": 0.1643,
      "num_input_tokens_seen": 15027712,
      "step": 6950
    },
    {
      "epoch": 1.1345840130505709,
      "grad_norm": 1.7396841049194336,
      "learning_rate": 4.997248442884879e-05,
      "loss": 0.1043,
      "num_input_tokens_seen": 15038080,
      "step": 6955
    },
    {
      "epoch": 1.135399673735726,
      "grad_norm": 0.6531755328178406,
      "learning_rate": 4.997214955108887e-05,
      "loss": 0.1422,
      "num_input_tokens_seen": 15050432,
      "step": 6960
    },
    {
      "epoch": 1.136215334420881,
      "grad_norm": 0.45595988631248474,
      "learning_rate": 4.9971812648959796e-05,
      "loss": 0.1733,
      "num_input_tokens_seen": 15060992,
      "step": 6965
    },
    {
      "epoch": 1.137030995106036,
      "grad_norm": 1.885270595550537,
      "learning_rate": 4.997147372248887e-05,
      "loss": 0.0791,
      "num_input_tokens_seen": 15070816,
      "step": 6970
    },
    {
      "epoch": 1.1378466557911908,
      "grad_norm": 0.7401260137557983,
      "learning_rate": 4.99711327717036e-05,
      "loss": 0.1806,
      "num_input_tokens_seen": 15080128,
      "step": 6975
    },
    {
      "epoch": 1.1386623164763459,
      "grad_norm": 0.2424100637435913,
      "learning_rate": 4.997078979663159e-05,
      "loss": 0.1129,
      "num_input_tokens_seen": 15090464,
      "step": 6980
    },
    {
      "epoch": 1.139477977161501,
      "grad_norm": 3.2046449184417725,
      "learning_rate": 4.997044479730067e-05,
      "loss": 0.1906,
      "num_input_tokens_seen": 15101472,
      "step": 6985
    },
    {
      "epoch": 1.1402936378466557,
      "grad_norm": 3.129856824874878,
      "learning_rate": 4.997009777373879e-05,
      "loss": 0.1128,
      "num_input_tokens_seen": 15111616,
      "step": 6990
    },
    {
      "epoch": 1.1411092985318108,
      "grad_norm": 6.59491491317749,
      "learning_rate": 4.9969748725974085e-05,
      "loss": 0.2529,
      "num_input_tokens_seen": 15122496,
      "step": 6995
    },
    {
      "epoch": 1.1419249592169658,
      "grad_norm": 0.2636934816837311,
      "learning_rate": 4.996939765403486e-05,
      "loss": 0.1272,
      "num_input_tokens_seen": 15134112,
      "step": 7000
    },
    {
      "epoch": 1.1427406199021206,
      "grad_norm": 5.670815467834473,
      "learning_rate": 4.996904455794956e-05,
      "loss": 0.0784,
      "num_input_tokens_seen": 15144352,
      "step": 7005
    },
    {
      "epoch": 1.1435562805872757,
      "grad_norm": 0.15952295064926147,
      "learning_rate": 4.996868943774683e-05,
      "loss": 0.1361,
      "num_input_tokens_seen": 15154304,
      "step": 7010
    },
    {
      "epoch": 1.1443719412724307,
      "grad_norm": 1.9397414922714233,
      "learning_rate": 4.9968332293455433e-05,
      "loss": 0.0549,
      "num_input_tokens_seen": 15165024,
      "step": 7015
    },
    {
      "epoch": 1.1451876019575857,
      "grad_norm": 2.2744736671447754,
      "learning_rate": 4.996797312510433e-05,
      "loss": 0.1571,
      "num_input_tokens_seen": 15176224,
      "step": 7020
    },
    {
      "epoch": 1.1460032626427405,
      "grad_norm": 5.310516834259033,
      "learning_rate": 4.9967611932722645e-05,
      "loss": 0.1318,
      "num_input_tokens_seen": 15186464,
      "step": 7025
    },
    {
      "epoch": 1.1468189233278956,
      "grad_norm": 4.836538314819336,
      "learning_rate": 4.9967248716339656e-05,
      "loss": 0.2183,
      "num_input_tokens_seen": 15197408,
      "step": 7030
    },
    {
      "epoch": 1.1476345840130506,
      "grad_norm": 0.4939217269420624,
      "learning_rate": 4.9966883475984796e-05,
      "loss": 0.0798,
      "num_input_tokens_seen": 15209504,
      "step": 7035
    },
    {
      "epoch": 1.1484502446982057,
      "grad_norm": 3.923380136489868,
      "learning_rate": 4.996651621168768e-05,
      "loss": 0.2964,
      "num_input_tokens_seen": 15220576,
      "step": 7040
    },
    {
      "epoch": 1.1492659053833605,
      "grad_norm": 0.5404402017593384,
      "learning_rate": 4.9966146923478086e-05,
      "loss": 0.1547,
      "num_input_tokens_seen": 15230752,
      "step": 7045
    },
    {
      "epoch": 1.1500815660685155,
      "grad_norm": 6.2645440101623535,
      "learning_rate": 4.996577561138594e-05,
      "loss": 0.0831,
      "num_input_tokens_seen": 15241312,
      "step": 7050
    },
    {
      "epoch": 1.1508972267536706,
      "grad_norm": 4.551450729370117,
      "learning_rate": 4.996540227544136e-05,
      "loss": 0.1611,
      "num_input_tokens_seen": 15252512,
      "step": 7055
    },
    {
      "epoch": 1.1517128874388254,
      "grad_norm": 0.6982122659683228,
      "learning_rate": 4.9965026915674584e-05,
      "loss": 0.0937,
      "num_input_tokens_seen": 15262400,
      "step": 7060
    },
    {
      "epoch": 1.1525285481239804,
      "grad_norm": 5.603863716125488,
      "learning_rate": 4.9964649532116065e-05,
      "loss": 0.2171,
      "num_input_tokens_seen": 15274816,
      "step": 7065
    },
    {
      "epoch": 1.1533442088091355,
      "grad_norm": 0.09071218222379684,
      "learning_rate": 4.996427012479638e-05,
      "loss": 0.0906,
      "num_input_tokens_seen": 15285248,
      "step": 7070
    },
    {
      "epoch": 1.1541598694942903,
      "grad_norm": 1.8232356309890747,
      "learning_rate": 4.9963888693746294e-05,
      "loss": 0.018,
      "num_input_tokens_seen": 15295552,
      "step": 7075
    },
    {
      "epoch": 1.1549755301794453,
      "grad_norm": 0.33773085474967957,
      "learning_rate": 4.996350523899672e-05,
      "loss": 0.0267,
      "num_input_tokens_seen": 15305248,
      "step": 7080
    },
    {
      "epoch": 1.1557911908646004,
      "grad_norm": 0.06964617222547531,
      "learning_rate": 4.9963119760578756e-05,
      "loss": 0.0268,
      "num_input_tokens_seen": 15316576,
      "step": 7085
    },
    {
      "epoch": 1.1566068515497552,
      "grad_norm": 0.03703215718269348,
      "learning_rate": 4.996273225852364e-05,
      "loss": 0.1948,
      "num_input_tokens_seen": 15328256,
      "step": 7090
    },
    {
      "epoch": 1.1574225122349102,
      "grad_norm": 3.359741449356079,
      "learning_rate": 4.996234273286278e-05,
      "loss": 0.2215,
      "num_input_tokens_seen": 15338752,
      "step": 7095
    },
    {
      "epoch": 1.1582381729200653,
      "grad_norm": 0.04670816659927368,
      "learning_rate": 4.996195118362777e-05,
      "loss": 0.0109,
      "num_input_tokens_seen": 15348224,
      "step": 7100
    },
    {
      "epoch": 1.1590538336052203,
      "grad_norm": 0.11876621842384338,
      "learning_rate": 4.996155761085034e-05,
      "loss": 0.0895,
      "num_input_tokens_seen": 15359008,
      "step": 7105
    },
    {
      "epoch": 1.1598694942903751,
      "grad_norm": 0.1864207237958908,
      "learning_rate": 4.996116201456239e-05,
      "loss": 0.0529,
      "num_input_tokens_seen": 15369408,
      "step": 7110
    },
    {
      "epoch": 1.1606851549755302,
      "grad_norm": 0.23362979292869568,
      "learning_rate": 4.9960764394796e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 15379392,
      "step": 7115
    },
    {
      "epoch": 1.1615008156606852,
      "grad_norm": 0.9414898157119751,
      "learning_rate": 4.99603647515834e-05,
      "loss": 0.0729,
      "num_input_tokens_seen": 15389920,
      "step": 7120
    },
    {
      "epoch": 1.1623164763458402,
      "grad_norm": 0.13409867882728577,
      "learning_rate": 4.9959963084956986e-05,
      "loss": 0.2052,
      "num_input_tokens_seen": 15400896,
      "step": 7125
    },
    {
      "epoch": 1.163132137030995,
      "grad_norm": 0.18150252103805542,
      "learning_rate": 4.9959559394949315e-05,
      "loss": 0.0223,
      "num_input_tokens_seen": 15412320,
      "step": 7130
    },
    {
      "epoch": 1.16394779771615,
      "grad_norm": 0.06716419756412506,
      "learning_rate": 4.9959153681593114e-05,
      "loss": 0.102,
      "num_input_tokens_seen": 15423680,
      "step": 7135
    },
    {
      "epoch": 1.1647634584013051,
      "grad_norm": 0.2635498046875,
      "learning_rate": 4.9958745944921275e-05,
      "loss": 0.0494,
      "num_input_tokens_seen": 15433952,
      "step": 7140
    },
    {
      "epoch": 1.16557911908646,
      "grad_norm": 0.7040302753448486,
      "learning_rate": 4.995833618496685e-05,
      "loss": 0.1906,
      "num_input_tokens_seen": 15444896,
      "step": 7145
    },
    {
      "epoch": 1.166394779771615,
      "grad_norm": 0.14256872236728668,
      "learning_rate": 4.9957924401763065e-05,
      "loss": 0.0646,
      "num_input_tokens_seen": 15456192,
      "step": 7150
    },
    {
      "epoch": 1.16721044045677,
      "grad_norm": 3.789982557296753,
      "learning_rate": 4.9957510595343285e-05,
      "loss": 0.2344,
      "num_input_tokens_seen": 15465312,
      "step": 7155
    },
    {
      "epoch": 1.1680261011419248,
      "grad_norm": 0.07367774099111557,
      "learning_rate": 4.995709476574106e-05,
      "loss": 0.1398,
      "num_input_tokens_seen": 15474496,
      "step": 7160
    },
    {
      "epoch": 1.1688417618270799,
      "grad_norm": 5.558032989501953,
      "learning_rate": 4.9956676912990105e-05,
      "loss": 0.1262,
      "num_input_tokens_seen": 15484992,
      "step": 7165
    },
    {
      "epoch": 1.169657422512235,
      "grad_norm": 0.6465906500816345,
      "learning_rate": 4.99562570371243e-05,
      "loss": 0.0429,
      "num_input_tokens_seen": 15495232,
      "step": 7170
    },
    {
      "epoch": 1.17047308319739,
      "grad_norm": 0.06183154881000519,
      "learning_rate": 4.9955835138177667e-05,
      "loss": 0.1305,
      "num_input_tokens_seen": 15506304,
      "step": 7175
    },
    {
      "epoch": 1.1712887438825448,
      "grad_norm": 0.1421712338924408,
      "learning_rate": 4.9955411216184414e-05,
      "loss": 0.2294,
      "num_input_tokens_seen": 15517248,
      "step": 7180
    },
    {
      "epoch": 1.1721044045676998,
      "grad_norm": 0.3086666464805603,
      "learning_rate": 4.9954985271178903e-05,
      "loss": 0.0355,
      "num_input_tokens_seen": 15527040,
      "step": 7185
    },
    {
      "epoch": 1.1729200652528549,
      "grad_norm": 6.889306545257568,
      "learning_rate": 4.995455730319566e-05,
      "loss": 0.3019,
      "num_input_tokens_seen": 15538368,
      "step": 7190
    },
    {
      "epoch": 1.17373572593801,
      "grad_norm": 0.12419155240058899,
      "learning_rate": 4.9954127312269386e-05,
      "loss": 0.0041,
      "num_input_tokens_seen": 15548608,
      "step": 7195
    },
    {
      "epoch": 1.1745513866231647,
      "grad_norm": 0.6525631546974182,
      "learning_rate": 4.9953695298434944e-05,
      "loss": 0.2912,
      "num_input_tokens_seen": 15560032,
      "step": 7200
    },
    {
      "epoch": 1.1753670473083198,
      "grad_norm": 1.1164226531982422,
      "learning_rate": 4.9953261261727334e-05,
      "loss": 0.188,
      "num_input_tokens_seen": 15571616,
      "step": 7205
    },
    {
      "epoch": 1.1761827079934748,
      "grad_norm": 0.09184513241052628,
      "learning_rate": 4.9952825202181766e-05,
      "loss": 0.1342,
      "num_input_tokens_seen": 15583392,
      "step": 7210
    },
    {
      "epoch": 1.1769983686786296,
      "grad_norm": 0.11937636137008667,
      "learning_rate": 4.995238711983358e-05,
      "loss": 0.2329,
      "num_input_tokens_seen": 15593504,
      "step": 7215
    },
    {
      "epoch": 1.1778140293637847,
      "grad_norm": 2.529798746109009,
      "learning_rate": 4.995194701471828e-05,
      "loss": 0.0952,
      "num_input_tokens_seen": 15604928,
      "step": 7220
    },
    {
      "epoch": 1.1786296900489397,
      "grad_norm": 2.5741701126098633,
      "learning_rate": 4.9951504886871545e-05,
      "loss": 0.1982,
      "num_input_tokens_seen": 15615200,
      "step": 7225
    },
    {
      "epoch": 1.1794453507340945,
      "grad_norm": 1.6471775770187378,
      "learning_rate": 4.995106073632924e-05,
      "loss": 0.0458,
      "num_input_tokens_seen": 15626176,
      "step": 7230
    },
    {
      "epoch": 1.1802610114192496,
      "grad_norm": 2.7898664474487305,
      "learning_rate": 4.995061456312733e-05,
      "loss": 0.1147,
      "num_input_tokens_seen": 15636768,
      "step": 7235
    },
    {
      "epoch": 1.1810766721044046,
      "grad_norm": 0.9005841612815857,
      "learning_rate": 4.995016636730202e-05,
      "loss": 0.195,
      "num_input_tokens_seen": 15648512,
      "step": 7240
    },
    {
      "epoch": 1.1818923327895596,
      "grad_norm": 6.270269393920898,
      "learning_rate": 4.994971614888962e-05,
      "loss": 0.097,
      "num_input_tokens_seen": 15659328,
      "step": 7245
    },
    {
      "epoch": 1.1827079934747144,
      "grad_norm": 0.20680290460586548,
      "learning_rate": 4.994926390792664e-05,
      "loss": 0.0193,
      "num_input_tokens_seen": 15669920,
      "step": 7250
    },
    {
      "epoch": 1.1835236541598695,
      "grad_norm": 3.018476724624634,
      "learning_rate": 4.9948809644449734e-05,
      "loss": 0.1868,
      "num_input_tokens_seen": 15681152,
      "step": 7255
    },
    {
      "epoch": 1.1843393148450245,
      "grad_norm": 4.8543853759765625,
      "learning_rate": 4.994835335849573e-05,
      "loss": 0.1502,
      "num_input_tokens_seen": 15690912,
      "step": 7260
    },
    {
      "epoch": 1.1851549755301796,
      "grad_norm": 3.1689562797546387,
      "learning_rate": 4.994789505010161e-05,
      "loss": 0.1375,
      "num_input_tokens_seen": 15701216,
      "step": 7265
    },
    {
      "epoch": 1.1859706362153344,
      "grad_norm": 2.3359832763671875,
      "learning_rate": 4.994743471930454e-05,
      "loss": 0.1836,
      "num_input_tokens_seen": 15712000,
      "step": 7270
    },
    {
      "epoch": 1.1867862969004894,
      "grad_norm": 3.6418848037719727,
      "learning_rate": 4.994697236614183e-05,
      "loss": 0.2498,
      "num_input_tokens_seen": 15722848,
      "step": 7275
    },
    {
      "epoch": 1.1876019575856445,
      "grad_norm": 1.657184362411499,
      "learning_rate": 4.994650799065096e-05,
      "loss": 0.2006,
      "num_input_tokens_seen": 15734816,
      "step": 7280
    },
    {
      "epoch": 1.1884176182707993,
      "grad_norm": 0.7946965098381042,
      "learning_rate": 4.9946041592869576e-05,
      "loss": 0.3594,
      "num_input_tokens_seen": 15745664,
      "step": 7285
    },
    {
      "epoch": 1.1892332789559543,
      "grad_norm": 8.222297668457031,
      "learning_rate": 4.994557317283548e-05,
      "loss": 0.2051,
      "num_input_tokens_seen": 15756896,
      "step": 7290
    },
    {
      "epoch": 1.1900489396411094,
      "grad_norm": 0.49199816584587097,
      "learning_rate": 4.9945102730586655e-05,
      "loss": 0.1951,
      "num_input_tokens_seen": 15766656,
      "step": 7295
    },
    {
      "epoch": 1.1908646003262642,
      "grad_norm": 3.079603672027588,
      "learning_rate": 4.994463026616123e-05,
      "loss": 0.3055,
      "num_input_tokens_seen": 15777792,
      "step": 7300
    },
    {
      "epoch": 1.1916802610114192,
      "grad_norm": 1.5396080017089844,
      "learning_rate": 4.994415577959751e-05,
      "loss": 0.2226,
      "num_input_tokens_seen": 15787968,
      "step": 7305
    },
    {
      "epoch": 1.1924959216965743,
      "grad_norm": 2.956415891647339,
      "learning_rate": 4.9943679270933954e-05,
      "loss": 0.3819,
      "num_input_tokens_seen": 15797920,
      "step": 7310
    },
    {
      "epoch": 1.1933115823817293,
      "grad_norm": 0.6605135798454285,
      "learning_rate": 4.99432007402092e-05,
      "loss": 0.148,
      "num_input_tokens_seen": 15809376,
      "step": 7315
    },
    {
      "epoch": 1.1941272430668841,
      "grad_norm": 6.03127384185791,
      "learning_rate": 4.9942720187462025e-05,
      "loss": 0.1243,
      "num_input_tokens_seen": 15821408,
      "step": 7320
    },
    {
      "epoch": 1.1949429037520392,
      "grad_norm": 0.9249451160430908,
      "learning_rate": 4.9942237612731395e-05,
      "loss": 0.1065,
      "num_input_tokens_seen": 15830816,
      "step": 7325
    },
    {
      "epoch": 1.1957585644371942,
      "grad_norm": 0.7879658341407776,
      "learning_rate": 4.994175301605644e-05,
      "loss": 0.0357,
      "num_input_tokens_seen": 15841792,
      "step": 7330
    },
    {
      "epoch": 1.196574225122349,
      "grad_norm": 0.23040802776813507,
      "learning_rate": 4.9941266397476414e-05,
      "loss": 0.2713,
      "num_input_tokens_seen": 15852768,
      "step": 7335
    },
    {
      "epoch": 1.197389885807504,
      "grad_norm": 0.11664178222417831,
      "learning_rate": 4.9940777757030796e-05,
      "loss": 0.055,
      "num_input_tokens_seen": 15863680,
      "step": 7340
    },
    {
      "epoch": 1.198205546492659,
      "grad_norm": 5.903824806213379,
      "learning_rate": 4.994028709475917e-05,
      "loss": 0.1763,
      "num_input_tokens_seen": 15874976,
      "step": 7345
    },
    {
      "epoch": 1.1990212071778141,
      "grad_norm": 0.06786380708217621,
      "learning_rate": 4.993979441070135e-05,
      "loss": 0.0218,
      "num_input_tokens_seen": 15885408,
      "step": 7350
    },
    {
      "epoch": 1.199836867862969,
      "grad_norm": 6.23204231262207,
      "learning_rate": 4.9939299704897236e-05,
      "loss": 0.2324,
      "num_input_tokens_seen": 15895168,
      "step": 7355
    },
    {
      "epoch": 1.200652528548124,
      "grad_norm": 3.4383184909820557,
      "learning_rate": 4.993880297738694e-05,
      "loss": 0.1526,
      "num_input_tokens_seen": 15906240,
      "step": 7360
    },
    {
      "epoch": 1.201468189233279,
      "grad_norm": 4.5740580558776855,
      "learning_rate": 4.9938304228210754e-05,
      "loss": 0.1911,
      "num_input_tokens_seen": 15917024,
      "step": 7365
    },
    {
      "epoch": 1.2022838499184338,
      "grad_norm": 6.418374538421631,
      "learning_rate": 4.9937803457409084e-05,
      "loss": 0.4276,
      "num_input_tokens_seen": 15926496,
      "step": 7370
    },
    {
      "epoch": 1.2030995106035889,
      "grad_norm": 4.022270679473877,
      "learning_rate": 4.9937300665022535e-05,
      "loss": 0.1046,
      "num_input_tokens_seen": 15937600,
      "step": 7375
    },
    {
      "epoch": 1.203915171288744,
      "grad_norm": 3.2668960094451904,
      "learning_rate": 4.9936795851091854e-05,
      "loss": 0.1221,
      "num_input_tokens_seen": 15948064,
      "step": 7380
    },
    {
      "epoch": 1.2047308319738987,
      "grad_norm": 0.07322722673416138,
      "learning_rate": 4.993628901565799e-05,
      "loss": 0.1286,
      "num_input_tokens_seen": 15958752,
      "step": 7385
    },
    {
      "epoch": 1.2055464926590538,
      "grad_norm": 6.813504695892334,
      "learning_rate": 4.9935780158762e-05,
      "loss": 0.2462,
      "num_input_tokens_seen": 15970080,
      "step": 7390
    },
    {
      "epoch": 1.2063621533442088,
      "grad_norm": 1.1422133445739746,
      "learning_rate": 4.993526928044515e-05,
      "loss": 0.0705,
      "num_input_tokens_seen": 15982080,
      "step": 7395
    },
    {
      "epoch": 1.2071778140293639,
      "grad_norm": 0.5997249484062195,
      "learning_rate": 4.9934756380748846e-05,
      "loss": 0.0465,
      "num_input_tokens_seen": 15993056,
      "step": 7400
    },
    {
      "epoch": 1.2079934747145187,
      "grad_norm": 3.280303478240967,
      "learning_rate": 4.993424145971468e-05,
      "loss": 0.1448,
      "num_input_tokens_seen": 16003584,
      "step": 7405
    },
    {
      "epoch": 1.2088091353996737,
      "grad_norm": 2.7594776153564453,
      "learning_rate": 4.993372451738439e-05,
      "loss": 0.2384,
      "num_input_tokens_seen": 16014912,
      "step": 7410
    },
    {
      "epoch": 1.2096247960848288,
      "grad_norm": 0.6693934202194214,
      "learning_rate": 4.993320555379987e-05,
      "loss": 0.0899,
      "num_input_tokens_seen": 16026304,
      "step": 7415
    },
    {
      "epoch": 1.2104404567699838,
      "grad_norm": 1.054826021194458,
      "learning_rate": 4.9932684569003205e-05,
      "loss": 0.0551,
      "num_input_tokens_seen": 16038464,
      "step": 7420
    },
    {
      "epoch": 1.2112561174551386,
      "grad_norm": 5.335202217102051,
      "learning_rate": 4.993216156303662e-05,
      "loss": 0.1442,
      "num_input_tokens_seen": 16049728,
      "step": 7425
    },
    {
      "epoch": 1.2120717781402937,
      "grad_norm": 3.579210042953491,
      "learning_rate": 4.9931636535942506e-05,
      "loss": 0.3986,
      "num_input_tokens_seen": 16060032,
      "step": 7430
    },
    {
      "epoch": 1.2128874388254487,
      "grad_norm": 4.611842632293701,
      "learning_rate": 4.993110948776344e-05,
      "loss": 0.1536,
      "num_input_tokens_seen": 16072064,
      "step": 7435
    },
    {
      "epoch": 1.2137030995106035,
      "grad_norm": 0.4529060125350952,
      "learning_rate": 4.993058041854214e-05,
      "loss": 0.1385,
      "num_input_tokens_seen": 16082112,
      "step": 7440
    },
    {
      "epoch": 1.2145187601957586,
      "grad_norm": 0.110162153840065,
      "learning_rate": 4.9930049328321495e-05,
      "loss": 0.2334,
      "num_input_tokens_seen": 16093280,
      "step": 7445
    },
    {
      "epoch": 1.2153344208809136,
      "grad_norm": 4.0999369621276855,
      "learning_rate": 4.9929516217144554e-05,
      "loss": 0.2506,
      "num_input_tokens_seen": 16104352,
      "step": 7450
    },
    {
      "epoch": 1.2161500815660684,
      "grad_norm": 0.1871415227651596,
      "learning_rate": 4.992898108505454e-05,
      "loss": 0.1421,
      "num_input_tokens_seen": 16114784,
      "step": 7455
    },
    {
      "epoch": 1.2169657422512234,
      "grad_norm": 1.4177812337875366,
      "learning_rate": 4.992844393209483e-05,
      "loss": 0.0688,
      "num_input_tokens_seen": 16126080,
      "step": 7460
    },
    {
      "epoch": 1.2177814029363785,
      "grad_norm": 1.9091269969940186,
      "learning_rate": 4.992790475830896e-05,
      "loss": 0.2458,
      "num_input_tokens_seen": 16136480,
      "step": 7465
    },
    {
      "epoch": 1.2185970636215335,
      "grad_norm": 2.707463026046753,
      "learning_rate": 4.992736356374066e-05,
      "loss": 0.3477,
      "num_input_tokens_seen": 16147424,
      "step": 7470
    },
    {
      "epoch": 1.2194127243066883,
      "grad_norm": 0.41100895404815674,
      "learning_rate": 4.992682034843379e-05,
      "loss": 0.2539,
      "num_input_tokens_seen": 16157280,
      "step": 7475
    },
    {
      "epoch": 1.2202283849918434,
      "grad_norm": 4.53908634185791,
      "learning_rate": 4.992627511243238e-05,
      "loss": 0.1339,
      "num_input_tokens_seen": 16168320,
      "step": 7480
    },
    {
      "epoch": 1.2210440456769984,
      "grad_norm": 0.13700874149799347,
      "learning_rate": 4.992572785578063e-05,
      "loss": 0.076,
      "num_input_tokens_seen": 16178976,
      "step": 7485
    },
    {
      "epoch": 1.2218597063621535,
      "grad_norm": 0.1897386610507965,
      "learning_rate": 4.9925178578522914e-05,
      "loss": 0.1635,
      "num_input_tokens_seen": 16190208,
      "step": 7490
    },
    {
      "epoch": 1.2226753670473083,
      "grad_norm": 1.5856431722640991,
      "learning_rate": 4.992462728070375e-05,
      "loss": 0.2558,
      "num_input_tokens_seen": 16199488,
      "step": 7495
    },
    {
      "epoch": 1.2234910277324633,
      "grad_norm": 0.24407242238521576,
      "learning_rate": 4.992407396236784e-05,
      "loss": 0.2119,
      "num_input_tokens_seen": 16210432,
      "step": 7500
    },
    {
      "epoch": 1.2243066884176184,
      "grad_norm": 0.5603540539741516,
      "learning_rate": 4.992351862356003e-05,
      "loss": 0.0716,
      "num_input_tokens_seen": 16221664,
      "step": 7505
    },
    {
      "epoch": 1.2251223491027732,
      "grad_norm": 0.23095354437828064,
      "learning_rate": 4.992296126432533e-05,
      "loss": 0.1511,
      "num_input_tokens_seen": 16232064,
      "step": 7510
    },
    {
      "epoch": 1.2259380097879282,
      "grad_norm": 0.13533274829387665,
      "learning_rate": 4.992240188470894e-05,
      "loss": 0.0716,
      "num_input_tokens_seen": 16242976,
      "step": 7515
    },
    {
      "epoch": 1.2267536704730833,
      "grad_norm": 0.14358443021774292,
      "learning_rate": 4.99218404847562e-05,
      "loss": 0.1667,
      "num_input_tokens_seen": 16253440,
      "step": 7520
    },
    {
      "epoch": 1.227569331158238,
      "grad_norm": 1.9528138637542725,
      "learning_rate": 4.9921277064512614e-05,
      "loss": 0.0796,
      "num_input_tokens_seen": 16263680,
      "step": 7525
    },
    {
      "epoch": 1.2283849918433931,
      "grad_norm": 0.19445714354515076,
      "learning_rate": 4.992071162402386e-05,
      "loss": 0.2385,
      "num_input_tokens_seen": 16274944,
      "step": 7530
    },
    {
      "epoch": 1.2292006525285482,
      "grad_norm": 1.89347505569458,
      "learning_rate": 4.992014416333577e-05,
      "loss": 0.0619,
      "num_input_tokens_seen": 16285376,
      "step": 7535
    },
    {
      "epoch": 1.2300163132137032,
      "grad_norm": 2.8959848880767822,
      "learning_rate": 4.991957468249436e-05,
      "loss": 0.0882,
      "num_input_tokens_seen": 16296544,
      "step": 7540
    },
    {
      "epoch": 1.230831973898858,
      "grad_norm": 0.16721372306346893,
      "learning_rate": 4.991900318154578e-05,
      "loss": 0.2256,
      "num_input_tokens_seen": 16306752,
      "step": 7545
    },
    {
      "epoch": 1.231647634584013,
      "grad_norm": 4.97653341293335,
      "learning_rate": 4.991842966053637e-05,
      "loss": 0.1995,
      "num_input_tokens_seen": 16317536,
      "step": 7550
    },
    {
      "epoch": 1.232463295269168,
      "grad_norm": 7.182518482208252,
      "learning_rate": 4.991785411951261e-05,
      "loss": 0.17,
      "num_input_tokens_seen": 16328640,
      "step": 7555
    },
    {
      "epoch": 1.233278955954323,
      "grad_norm": 3.598841428756714,
      "learning_rate": 4.9917276558521164e-05,
      "loss": 0.3151,
      "num_input_tokens_seen": 16340032,
      "step": 7560
    },
    {
      "epoch": 1.234094616639478,
      "grad_norm": 3.606041193008423,
      "learning_rate": 4.9916696977608855e-05,
      "loss": 0.2259,
      "num_input_tokens_seen": 16351328,
      "step": 7565
    },
    {
      "epoch": 1.234910277324633,
      "grad_norm": 0.5559537410736084,
      "learning_rate": 4.991611537682266e-05,
      "loss": 0.1566,
      "num_input_tokens_seen": 16360832,
      "step": 7570
    },
    {
      "epoch": 1.235725938009788,
      "grad_norm": 0.22433821856975555,
      "learning_rate": 4.991553175620973e-05,
      "loss": 0.0131,
      "num_input_tokens_seen": 16372640,
      "step": 7575
    },
    {
      "epoch": 1.2365415986949428,
      "grad_norm": 0.6391193270683289,
      "learning_rate": 4.991494611581738e-05,
      "loss": 0.0338,
      "num_input_tokens_seen": 16383360,
      "step": 7580
    },
    {
      "epoch": 1.2373572593800979,
      "grad_norm": 0.726963222026825,
      "learning_rate": 4.9914358455693076e-05,
      "loss": 0.1841,
      "num_input_tokens_seen": 16393728,
      "step": 7585
    },
    {
      "epoch": 1.238172920065253,
      "grad_norm": 4.856930255889893,
      "learning_rate": 4.991376877588446e-05,
      "loss": 0.269,
      "num_input_tokens_seen": 16405440,
      "step": 7590
    },
    {
      "epoch": 1.2389885807504077,
      "grad_norm": 3.5513672828674316,
      "learning_rate": 4.991317707643934e-05,
      "loss": 0.22,
      "num_input_tokens_seen": 16416672,
      "step": 7595
    },
    {
      "epoch": 1.2398042414355628,
      "grad_norm": 3.371920585632324,
      "learning_rate": 4.991258335740568e-05,
      "loss": 0.2127,
      "num_input_tokens_seen": 16427136,
      "step": 7600
    },
    {
      "epoch": 1.2406199021207178,
      "grad_norm": 1.6889550685882568,
      "learning_rate": 4.99119876188316e-05,
      "loss": 0.109,
      "num_input_tokens_seen": 16437984,
      "step": 7605
    },
    {
      "epoch": 1.2414355628058726,
      "grad_norm": 1.0969815254211426,
      "learning_rate": 4.9911389860765406e-05,
      "loss": 0.1772,
      "num_input_tokens_seen": 16448640,
      "step": 7610
    },
    {
      "epoch": 1.2422512234910277,
      "grad_norm": 3.5157084465026855,
      "learning_rate": 4.9910790083255555e-05,
      "loss": 0.3096,
      "num_input_tokens_seen": 16457856,
      "step": 7615
    },
    {
      "epoch": 1.2430668841761827,
      "grad_norm": 0.1653311550617218,
      "learning_rate": 4.991018828635066e-05,
      "loss": 0.059,
      "num_input_tokens_seen": 16469056,
      "step": 7620
    },
    {
      "epoch": 1.2438825448613378,
      "grad_norm": 0.17270918190479279,
      "learning_rate": 4.99095844700995e-05,
      "loss": 0.1477,
      "num_input_tokens_seen": 16479616,
      "step": 7625
    },
    {
      "epoch": 1.2446982055464926,
      "grad_norm": 0.31358715891838074,
      "learning_rate": 4.9908978634551045e-05,
      "loss": 0.0557,
      "num_input_tokens_seen": 16489952,
      "step": 7630
    },
    {
      "epoch": 1.2455138662316476,
      "grad_norm": 0.26833394169807434,
      "learning_rate": 4.990837077975439e-05,
      "loss": 0.0935,
      "num_input_tokens_seen": 16501152,
      "step": 7635
    },
    {
      "epoch": 1.2463295269168027,
      "grad_norm": 0.24005572497844696,
      "learning_rate": 4.990776090575881e-05,
      "loss": 0.0116,
      "num_input_tokens_seen": 16512608,
      "step": 7640
    },
    {
      "epoch": 1.2471451876019577,
      "grad_norm": 5.5196533203125,
      "learning_rate": 4.990714901261376e-05,
      "loss": 0.1512,
      "num_input_tokens_seen": 16524064,
      "step": 7645
    },
    {
      "epoch": 1.2479608482871125,
      "grad_norm": 3.4011459350585938,
      "learning_rate": 4.990653510036883e-05,
      "loss": 0.1,
      "num_input_tokens_seen": 16534688,
      "step": 7650
    },
    {
      "epoch": 1.2487765089722676,
      "grad_norm": 3.6264588832855225,
      "learning_rate": 4.99059191690738e-05,
      "loss": 0.0947,
      "num_input_tokens_seen": 16544928,
      "step": 7655
    },
    {
      "epoch": 1.2495921696574226,
      "grad_norm": 3.0403060913085938,
      "learning_rate": 4.9905301218778575e-05,
      "loss": 0.2576,
      "num_input_tokens_seen": 16555200,
      "step": 7660
    },
    {
      "epoch": 1.2504078303425774,
      "grad_norm": 5.403607368469238,
      "learning_rate": 4.990468124953328e-05,
      "loss": 0.2939,
      "num_input_tokens_seen": 16566304,
      "step": 7665
    },
    {
      "epoch": 1.2512234910277324,
      "grad_norm": 3.9963810443878174,
      "learning_rate": 4.990405926138815e-05,
      "loss": 0.3114,
      "num_input_tokens_seen": 16578560,
      "step": 7670
    },
    {
      "epoch": 1.2520391517128875,
      "grad_norm": 1.2811282873153687,
      "learning_rate": 4.9903435254393616e-05,
      "loss": 0.1692,
      "num_input_tokens_seen": 16588704,
      "step": 7675
    },
    {
      "epoch": 1.2528548123980423,
      "grad_norm": 3.366016387939453,
      "learning_rate": 4.990280922860026e-05,
      "loss": 0.05,
      "num_input_tokens_seen": 16599776,
      "step": 7680
    },
    {
      "epoch": 1.2536704730831973,
      "grad_norm": 0.13389497995376587,
      "learning_rate": 4.990218118405883e-05,
      "loss": 0.1158,
      "num_input_tokens_seen": 16609824,
      "step": 7685
    },
    {
      "epoch": 1.2544861337683524,
      "grad_norm": 4.384880542755127,
      "learning_rate": 4.990155112082024e-05,
      "loss": 0.2521,
      "num_input_tokens_seen": 16621280,
      "step": 7690
    },
    {
      "epoch": 1.2553017944535072,
      "grad_norm": 0.6415671110153198,
      "learning_rate": 4.9900919038935564e-05,
      "loss": 0.107,
      "num_input_tokens_seen": 16631776,
      "step": 7695
    },
    {
      "epoch": 1.2561174551386622,
      "grad_norm": 6.010123252868652,
      "learning_rate": 4.9900284938456056e-05,
      "loss": 0.1302,
      "num_input_tokens_seen": 16643584,
      "step": 7700
    },
    {
      "epoch": 1.2569331158238173,
      "grad_norm": 0.11299262195825577,
      "learning_rate": 4.98996488194331e-05,
      "loss": 0.0749,
      "num_input_tokens_seen": 16654432,
      "step": 7705
    },
    {
      "epoch": 1.2577487765089723,
      "grad_norm": 0.10881415754556656,
      "learning_rate": 4.989901068191828e-05,
      "loss": 0.074,
      "num_input_tokens_seen": 16664064,
      "step": 7710
    },
    {
      "epoch": 1.2585644371941274,
      "grad_norm": 0.6600273251533508,
      "learning_rate": 4.9898370525963314e-05,
      "loss": 0.1275,
      "num_input_tokens_seen": 16674432,
      "step": 7715
    },
    {
      "epoch": 1.2593800978792822,
      "grad_norm": 2.8763315677642822,
      "learning_rate": 4.9897728351620085e-05,
      "loss": 0.0988,
      "num_input_tokens_seen": 16685920,
      "step": 7720
    },
    {
      "epoch": 1.2601957585644372,
      "grad_norm": 3.1279098987579346,
      "learning_rate": 4.989708415894069e-05,
      "loss": 0.1397,
      "num_input_tokens_seen": 16696320,
      "step": 7725
    },
    {
      "epoch": 1.2610114192495923,
      "grad_norm": 0.3222804665565491,
      "learning_rate": 4.9896437947977306e-05,
      "loss": 0.0757,
      "num_input_tokens_seen": 16707040,
      "step": 7730
    },
    {
      "epoch": 1.261827079934747,
      "grad_norm": 3.6984167098999023,
      "learning_rate": 4.989578971878235e-05,
      "loss": 0.3237,
      "num_input_tokens_seen": 16718208,
      "step": 7735
    },
    {
      "epoch": 1.2626427406199021,
      "grad_norm": 3.510631561279297,
      "learning_rate": 4.9895139471408356e-05,
      "loss": 0.0439,
      "num_input_tokens_seen": 16728480,
      "step": 7740
    },
    {
      "epoch": 1.2634584013050572,
      "grad_norm": 0.07595990598201752,
      "learning_rate": 4.9894487205908044e-05,
      "loss": 0.1572,
      "num_input_tokens_seen": 16740384,
      "step": 7745
    },
    {
      "epoch": 1.264274061990212,
      "grad_norm": 3.165645122528076,
      "learning_rate": 4.9893832922334285e-05,
      "loss": 0.282,
      "num_input_tokens_seen": 16749888,
      "step": 7750
    },
    {
      "epoch": 1.265089722675367,
      "grad_norm": 5.2043633460998535,
      "learning_rate": 4.989317662074011e-05,
      "loss": 0.2996,
      "num_input_tokens_seen": 16761024,
      "step": 7755
    },
    {
      "epoch": 1.265905383360522,
      "grad_norm": 0.07771420478820801,
      "learning_rate": 4.989251830117874e-05,
      "loss": 0.0128,
      "num_input_tokens_seen": 16771296,
      "step": 7760
    },
    {
      "epoch": 1.2667210440456769,
      "grad_norm": 2.6050169467926025,
      "learning_rate": 4.9891857963703535e-05,
      "loss": 0.1125,
      "num_input_tokens_seen": 16781984,
      "step": 7765
    },
    {
      "epoch": 1.267536704730832,
      "grad_norm": 4.233067989349365,
      "learning_rate": 4.989119560836802e-05,
      "loss": 0.1228,
      "num_input_tokens_seen": 16793056,
      "step": 7770
    },
    {
      "epoch": 1.268352365415987,
      "grad_norm": 0.8741741180419922,
      "learning_rate": 4.989053123522589e-05,
      "loss": 0.061,
      "num_input_tokens_seen": 16805568,
      "step": 7775
    },
    {
      "epoch": 1.269168026101142,
      "grad_norm": 0.4571402072906494,
      "learning_rate": 4.988986484433101e-05,
      "loss": 0.0753,
      "num_input_tokens_seen": 16815808,
      "step": 7780
    },
    {
      "epoch": 1.269983686786297,
      "grad_norm": 0.1190493032336235,
      "learning_rate": 4.988919643573739e-05,
      "loss": 0.1496,
      "num_input_tokens_seen": 16826272,
      "step": 7785
    },
    {
      "epoch": 1.2707993474714518,
      "grad_norm": 12.23763370513916,
      "learning_rate": 4.9888526009499223e-05,
      "loss": 0.2057,
      "num_input_tokens_seen": 16836576,
      "step": 7790
    },
    {
      "epoch": 1.2716150081566069,
      "grad_norm": 0.23459982872009277,
      "learning_rate": 4.9887853565670854e-05,
      "loss": 0.0345,
      "num_input_tokens_seen": 16846720,
      "step": 7795
    },
    {
      "epoch": 1.272430668841762,
      "grad_norm": 0.4631684124469757,
      "learning_rate": 4.9887179104306796e-05,
      "loss": 0.104,
      "num_input_tokens_seen": 16856960,
      "step": 7800
    },
    {
      "epoch": 1.2732463295269167,
      "grad_norm": 3.170642614364624,
      "learning_rate": 4.988650262546173e-05,
      "loss": 0.2514,
      "num_input_tokens_seen": 16868512,
      "step": 7805
    },
    {
      "epoch": 1.2740619902120718,
      "grad_norm": 0.30952563881874084,
      "learning_rate": 4.9885824129190476e-05,
      "loss": 0.1314,
      "num_input_tokens_seen": 16879744,
      "step": 7810
    },
    {
      "epoch": 1.2748776508972268,
      "grad_norm": 2.387211799621582,
      "learning_rate": 4.988514361554806e-05,
      "loss": 0.1511,
      "num_input_tokens_seen": 16889440,
      "step": 7815
    },
    {
      "epoch": 1.2756933115823816,
      "grad_norm": 1.0796942710876465,
      "learning_rate": 4.988446108458963e-05,
      "loss": 0.1575,
      "num_input_tokens_seen": 16900768,
      "step": 7820
    },
    {
      "epoch": 1.2765089722675367,
      "grad_norm": 1.151346206665039,
      "learning_rate": 4.988377653637052e-05,
      "loss": 0.1886,
      "num_input_tokens_seen": 16910240,
      "step": 7825
    },
    {
      "epoch": 1.2773246329526917,
      "grad_norm": 0.6544666290283203,
      "learning_rate": 4.988308997094623e-05,
      "loss": 0.0358,
      "num_input_tokens_seen": 16921600,
      "step": 7830
    },
    {
      "epoch": 1.2781402936378465,
      "grad_norm": 6.242870807647705,
      "learning_rate": 4.988240138837241e-05,
      "loss": 0.1791,
      "num_input_tokens_seen": 16933344,
      "step": 7835
    },
    {
      "epoch": 1.2789559543230016,
      "grad_norm": 4.08312463760376,
      "learning_rate": 4.988171078870488e-05,
      "loss": 0.2494,
      "num_input_tokens_seen": 16944416,
      "step": 7840
    },
    {
      "epoch": 1.2797716150081566,
      "grad_norm": 5.498296737670898,
      "learning_rate": 4.988101817199963e-05,
      "loss": 0.2828,
      "num_input_tokens_seen": 16954816,
      "step": 7845
    },
    {
      "epoch": 1.2805872756933117,
      "grad_norm": 4.145997524261475,
      "learning_rate": 4.988032353831279e-05,
      "loss": 0.1137,
      "num_input_tokens_seen": 16964992,
      "step": 7850
    },
    {
      "epoch": 1.2814029363784667,
      "grad_norm": 0.0852891206741333,
      "learning_rate": 4.9879626887700694e-05,
      "loss": 0.0351,
      "num_input_tokens_seen": 16975200,
      "step": 7855
    },
    {
      "epoch": 1.2822185970636215,
      "grad_norm": 2.0257375240325928,
      "learning_rate": 4.98789282202198e-05,
      "loss": 0.0681,
      "num_input_tokens_seen": 16985984,
      "step": 7860
    },
    {
      "epoch": 1.2830342577487766,
      "grad_norm": 0.12425827234983444,
      "learning_rate": 4.9878227535926745e-05,
      "loss": 0.0915,
      "num_input_tokens_seen": 16997504,
      "step": 7865
    },
    {
      "epoch": 1.2838499184339316,
      "grad_norm": 0.049235470592975616,
      "learning_rate": 4.987752483487834e-05,
      "loss": 0.0288,
      "num_input_tokens_seen": 17007296,
      "step": 7870
    },
    {
      "epoch": 1.2846655791190864,
      "grad_norm": 3.61763072013855,
      "learning_rate": 4.987682011713155e-05,
      "loss": 0.22,
      "num_input_tokens_seen": 17017792,
      "step": 7875
    },
    {
      "epoch": 1.2854812398042414,
      "grad_norm": 0.8922204375267029,
      "learning_rate": 4.9876113382743496e-05,
      "loss": 0.1027,
      "num_input_tokens_seen": 17029472,
      "step": 7880
    },
    {
      "epoch": 1.2862969004893965,
      "grad_norm": 0.139808788895607,
      "learning_rate": 4.987540463177147e-05,
      "loss": 0.0355,
      "num_input_tokens_seen": 17039808,
      "step": 7885
    },
    {
      "epoch": 1.2871125611745513,
      "grad_norm": 0.26891791820526123,
      "learning_rate": 4.987469386427292e-05,
      "loss": 0.0768,
      "num_input_tokens_seen": 17050720,
      "step": 7890
    },
    {
      "epoch": 1.2879282218597063,
      "grad_norm": 0.3657711148262024,
      "learning_rate": 4.987398108030548e-05,
      "loss": 0.2023,
      "num_input_tokens_seen": 17061792,
      "step": 7895
    },
    {
      "epoch": 1.2887438825448614,
      "grad_norm": 0.6014899611473083,
      "learning_rate": 4.987326627992692e-05,
      "loss": 0.1014,
      "num_input_tokens_seen": 17072512,
      "step": 7900
    },
    {
      "epoch": 1.2895595432300162,
      "grad_norm": 0.4137114882469177,
      "learning_rate": 4.98725494631952e-05,
      "loss": 0.1236,
      "num_input_tokens_seen": 17084288,
      "step": 7905
    },
    {
      "epoch": 1.2903752039151712,
      "grad_norm": 6.638491630554199,
      "learning_rate": 4.9871830630168404e-05,
      "loss": 0.4086,
      "num_input_tokens_seen": 17095648,
      "step": 7910
    },
    {
      "epoch": 1.2911908646003263,
      "grad_norm": 3.11419677734375,
      "learning_rate": 4.987110978090482e-05,
      "loss": 0.1669,
      "num_input_tokens_seen": 17106976,
      "step": 7915
    },
    {
      "epoch": 1.2920065252854813,
      "grad_norm": 5.048390865325928,
      "learning_rate": 4.9870386915462894e-05,
      "loss": 0.0826,
      "num_input_tokens_seen": 17116640,
      "step": 7920
    },
    {
      "epoch": 1.2928221859706361,
      "grad_norm": 0.12708722054958344,
      "learning_rate": 4.986966203390121e-05,
      "loss": 0.2733,
      "num_input_tokens_seen": 17127584,
      "step": 7925
    },
    {
      "epoch": 1.2936378466557912,
      "grad_norm": 0.03831891342997551,
      "learning_rate": 4.986893513627853e-05,
      "loss": 0.0283,
      "num_input_tokens_seen": 17137760,
      "step": 7930
    },
    {
      "epoch": 1.2944535073409462,
      "grad_norm": 3.6218161582946777,
      "learning_rate": 4.9868206222653785e-05,
      "loss": 0.2629,
      "num_input_tokens_seen": 17149056,
      "step": 7935
    },
    {
      "epoch": 1.2952691680261013,
      "grad_norm": 0.09029592573642731,
      "learning_rate": 4.9867475293086066e-05,
      "loss": 0.1806,
      "num_input_tokens_seen": 17160096,
      "step": 7940
    },
    {
      "epoch": 1.296084828711256,
      "grad_norm": 4.290560722351074,
      "learning_rate": 4.9866742347634624e-05,
      "loss": 0.1409,
      "num_input_tokens_seen": 17170592,
      "step": 7945
    },
    {
      "epoch": 1.2969004893964111,
      "grad_norm": 3.9079580307006836,
      "learning_rate": 4.986600738635887e-05,
      "loss": 0.3408,
      "num_input_tokens_seen": 17181760,
      "step": 7950
    },
    {
      "epoch": 1.2977161500815662,
      "grad_norm": 3.26395320892334,
      "learning_rate": 4.986527040931839e-05,
      "loss": 0.2637,
      "num_input_tokens_seen": 17192416,
      "step": 7955
    },
    {
      "epoch": 1.298531810766721,
      "grad_norm": 4.412578105926514,
      "learning_rate": 4.9864531416572926e-05,
      "loss": 0.0548,
      "num_input_tokens_seen": 17204256,
      "step": 7960
    },
    {
      "epoch": 1.299347471451876,
      "grad_norm": 3.5438807010650635,
      "learning_rate": 4.986379040818239e-05,
      "loss": 0.2506,
      "num_input_tokens_seen": 17215520,
      "step": 7965
    },
    {
      "epoch": 1.300163132137031,
      "grad_norm": 4.044760227203369,
      "learning_rate": 4.9863047384206835e-05,
      "loss": 0.2148,
      "num_input_tokens_seen": 17226560,
      "step": 7970
    },
    {
      "epoch": 1.3009787928221859,
      "grad_norm": 0.22716321051120758,
      "learning_rate": 4.986230234470651e-05,
      "loss": 0.0901,
      "num_input_tokens_seen": 17236864,
      "step": 7975
    },
    {
      "epoch": 1.301794453507341,
      "grad_norm": 2.31801700592041,
      "learning_rate": 4.986155528974181e-05,
      "loss": 0.1649,
      "num_input_tokens_seen": 17247424,
      "step": 7980
    },
    {
      "epoch": 1.302610114192496,
      "grad_norm": 3.5804312229156494,
      "learning_rate": 4.986080621937329e-05,
      "loss": 0.06,
      "num_input_tokens_seen": 17258208,
      "step": 7985
    },
    {
      "epoch": 1.3034257748776508,
      "grad_norm": 3.015270233154297,
      "learning_rate": 4.9860055133661675e-05,
      "loss": 0.0771,
      "num_input_tokens_seen": 17269280,
      "step": 7990
    },
    {
      "epoch": 1.3042414355628058,
      "grad_norm": 0.5983902215957642,
      "learning_rate": 4.985930203266785e-05,
      "loss": 0.059,
      "num_input_tokens_seen": 17278816,
      "step": 7995
    },
    {
      "epoch": 1.3050570962479608,
      "grad_norm": 2.452056884765625,
      "learning_rate": 4.985854691645287e-05,
      "loss": 0.0612,
      "num_input_tokens_seen": 17289824,
      "step": 8000
    },
    {
      "epoch": 1.3058727569331159,
      "grad_norm": 0.5834968090057373,
      "learning_rate": 4.985778978507795e-05,
      "loss": 0.0479,
      "num_input_tokens_seen": 17300448,
      "step": 8005
    },
    {
      "epoch": 1.306688417618271,
      "grad_norm": 2.934300184249878,
      "learning_rate": 4.9857030638604454e-05,
      "loss": 0.2972,
      "num_input_tokens_seen": 17310592,
      "step": 8010
    },
    {
      "epoch": 1.3075040783034257,
      "grad_norm": 0.10457460582256317,
      "learning_rate": 4.985626947709393e-05,
      "loss": 0.1711,
      "num_input_tokens_seen": 17321632,
      "step": 8015
    },
    {
      "epoch": 1.3083197389885808,
      "grad_norm": 1.1078598499298096,
      "learning_rate": 4.985550630060809e-05,
      "loss": 0.1833,
      "num_input_tokens_seen": 17332960,
      "step": 8020
    },
    {
      "epoch": 1.3091353996737358,
      "grad_norm": 5.162299156188965,
      "learning_rate": 4.985474110920879e-05,
      "loss": 0.3693,
      "num_input_tokens_seen": 17343808,
      "step": 8025
    },
    {
      "epoch": 1.3099510603588906,
      "grad_norm": 0.14242039620876312,
      "learning_rate": 4.985397390295807e-05,
      "loss": 0.2391,
      "num_input_tokens_seen": 17353696,
      "step": 8030
    },
    {
      "epoch": 1.3107667210440457,
      "grad_norm": 1.9484858512878418,
      "learning_rate": 4.985320468191811e-05,
      "loss": 0.1691,
      "num_input_tokens_seen": 17364000,
      "step": 8035
    },
    {
      "epoch": 1.3115823817292007,
      "grad_norm": 1.8402628898620605,
      "learning_rate": 4.985243344615128e-05,
      "loss": 0.1106,
      "num_input_tokens_seen": 17371936,
      "step": 8040
    },
    {
      "epoch": 1.3123980424143555,
      "grad_norm": 3.3673973083496094,
      "learning_rate": 4.9851660195720095e-05,
      "loss": 0.1843,
      "num_input_tokens_seen": 17382304,
      "step": 8045
    },
    {
      "epoch": 1.3132137030995106,
      "grad_norm": 0.8606354594230652,
      "learning_rate": 4.985088493068724e-05,
      "loss": 0.0396,
      "num_input_tokens_seen": 17392000,
      "step": 8050
    },
    {
      "epoch": 1.3140293637846656,
      "grad_norm": 0.5301800966262817,
      "learning_rate": 4.985010765111555e-05,
      "loss": 0.1012,
      "num_input_tokens_seen": 17402528,
      "step": 8055
    },
    {
      "epoch": 1.3148450244698204,
      "grad_norm": 0.20276932418346405,
      "learning_rate": 4.984932835706805e-05,
      "loss": 0.24,
      "num_input_tokens_seen": 17412480,
      "step": 8060
    },
    {
      "epoch": 1.3156606851549755,
      "grad_norm": 0.7233949303627014,
      "learning_rate": 4.984854704860791e-05,
      "loss": 0.0774,
      "num_input_tokens_seen": 17423520,
      "step": 8065
    },
    {
      "epoch": 1.3164763458401305,
      "grad_norm": 2.157273530960083,
      "learning_rate": 4.984776372579847e-05,
      "loss": 0.0993,
      "num_input_tokens_seen": 17434176,
      "step": 8070
    },
    {
      "epoch": 1.3172920065252856,
      "grad_norm": 3.237471580505371,
      "learning_rate": 4.984697838870322e-05,
      "loss": 0.1383,
      "num_input_tokens_seen": 17445504,
      "step": 8075
    },
    {
      "epoch": 1.3181076672104406,
      "grad_norm": 1.3257803916931152,
      "learning_rate": 4.984619103738584e-05,
      "loss": 0.0582,
      "num_input_tokens_seen": 17456544,
      "step": 8080
    },
    {
      "epoch": 1.3189233278955954,
      "grad_norm": 1.4488815069198608,
      "learning_rate": 4.984540167191014e-05,
      "loss": 0.0891,
      "num_input_tokens_seen": 17466432,
      "step": 8085
    },
    {
      "epoch": 1.3197389885807504,
      "grad_norm": 4.252408504486084,
      "learning_rate": 4.984461029234011e-05,
      "loss": 0.0422,
      "num_input_tokens_seen": 17477472,
      "step": 8090
    },
    {
      "epoch": 1.3205546492659055,
      "grad_norm": 0.4625602662563324,
      "learning_rate": 4.9843816898739913e-05,
      "loss": 0.0391,
      "num_input_tokens_seen": 17487360,
      "step": 8095
    },
    {
      "epoch": 1.3213703099510603,
      "grad_norm": 0.09763139486312866,
      "learning_rate": 4.984302149117387e-05,
      "loss": 0.0207,
      "num_input_tokens_seen": 17497568,
      "step": 8100
    },
    {
      "epoch": 1.3221859706362153,
      "grad_norm": 1.8961961269378662,
      "learning_rate": 4.984222406970644e-05,
      "loss": 0.0359,
      "num_input_tokens_seen": 17508768,
      "step": 8105
    },
    {
      "epoch": 1.3230016313213704,
      "grad_norm": 0.11448470503091812,
      "learning_rate": 4.984142463440229e-05,
      "loss": 0.1468,
      "num_input_tokens_seen": 17518368,
      "step": 8110
    },
    {
      "epoch": 1.3238172920065252,
      "grad_norm": 8.590022087097168,
      "learning_rate": 4.984062318532621e-05,
      "loss": 0.4463,
      "num_input_tokens_seen": 17528416,
      "step": 8115
    },
    {
      "epoch": 1.3246329526916802,
      "grad_norm": 0.28208616375923157,
      "learning_rate": 4.983981972254317e-05,
      "loss": 0.3348,
      "num_input_tokens_seen": 17537600,
      "step": 8120
    },
    {
      "epoch": 1.3254486133768353,
      "grad_norm": 0.3865681290626526,
      "learning_rate": 4.983901424611832e-05,
      "loss": 0.0758,
      "num_input_tokens_seen": 17548736,
      "step": 8125
    },
    {
      "epoch": 1.32626427406199,
      "grad_norm": 3.612586259841919,
      "learning_rate": 4.9838206756116926e-05,
      "loss": 0.3078,
      "num_input_tokens_seen": 17558336,
      "step": 8130
    },
    {
      "epoch": 1.3270799347471451,
      "grad_norm": 2.4190359115600586,
      "learning_rate": 4.983739725260448e-05,
      "loss": 0.0441,
      "num_input_tokens_seen": 17570272,
      "step": 8135
    },
    {
      "epoch": 1.3278955954323002,
      "grad_norm": 0.7069585919380188,
      "learning_rate": 4.983658573564658e-05,
      "loss": 0.0208,
      "num_input_tokens_seen": 17580960,
      "step": 8140
    },
    {
      "epoch": 1.3287112561174552,
      "grad_norm": 0.09782170504331589,
      "learning_rate": 4.983577220530902e-05,
      "loss": 0.1728,
      "num_input_tokens_seen": 17590528,
      "step": 8145
    },
    {
      "epoch": 1.32952691680261,
      "grad_norm": 3.0552656650543213,
      "learning_rate": 4.983495666165775e-05,
      "loss": 0.1243,
      "num_input_tokens_seen": 17600672,
      "step": 8150
    },
    {
      "epoch": 1.330342577487765,
      "grad_norm": 1.2478984594345093,
      "learning_rate": 4.983413910475889e-05,
      "loss": 0.3207,
      "num_input_tokens_seen": 17611648,
      "step": 8155
    },
    {
      "epoch": 1.3311582381729201,
      "grad_norm": 1.831872582435608,
      "learning_rate": 4.98333195346787e-05,
      "loss": 0.2454,
      "num_input_tokens_seen": 17622656,
      "step": 8160
    },
    {
      "epoch": 1.3319738988580752,
      "grad_norm": 0.2922876179218292,
      "learning_rate": 4.983249795148363e-05,
      "loss": 0.0801,
      "num_input_tokens_seen": 17632864,
      "step": 8165
    },
    {
      "epoch": 1.33278955954323,
      "grad_norm": 0.2000967562198639,
      "learning_rate": 4.983167435524027e-05,
      "loss": 0.1674,
      "num_input_tokens_seen": 17643968,
      "step": 8170
    },
    {
      "epoch": 1.333605220228385,
      "grad_norm": 0.5970892310142517,
      "learning_rate": 4.98308487460154e-05,
      "loss": 0.1559,
      "num_input_tokens_seen": 17654272,
      "step": 8175
    },
    {
      "epoch": 1.33442088091354,
      "grad_norm": 0.967530369758606,
      "learning_rate": 4.983002112387594e-05,
      "loss": 0.0755,
      "num_input_tokens_seen": 17665728,
      "step": 8180
    },
    {
      "epoch": 1.3352365415986949,
      "grad_norm": 0.3429315686225891,
      "learning_rate": 4.982919148888897e-05,
      "loss": 0.196,
      "num_input_tokens_seen": 17677856,
      "step": 8185
    },
    {
      "epoch": 1.33605220228385,
      "grad_norm": 0.09351611882448196,
      "learning_rate": 4.982835984112177e-05,
      "loss": 0.1281,
      "num_input_tokens_seen": 17688672,
      "step": 8190
    },
    {
      "epoch": 1.336867862969005,
      "grad_norm": 0.23307408392429352,
      "learning_rate": 4.982752618064174e-05,
      "loss": 0.0638,
      "num_input_tokens_seen": 17699456,
      "step": 8195
    },
    {
      "epoch": 1.3376835236541598,
      "grad_norm": 6.512901782989502,
      "learning_rate": 4.982669050751646e-05,
      "loss": 0.1444,
      "num_input_tokens_seen": 17710976,
      "step": 8200
    },
    {
      "epoch": 1.3384991843393148,
      "grad_norm": 1.9145153760910034,
      "learning_rate": 4.982585282181368e-05,
      "loss": 0.099,
      "num_input_tokens_seen": 17721152,
      "step": 8205
    },
    {
      "epoch": 1.3393148450244698,
      "grad_norm": 2.3821253776550293,
      "learning_rate": 4.9825013123601305e-05,
      "loss": 0.1493,
      "num_input_tokens_seen": 17731200,
      "step": 8210
    },
    {
      "epoch": 1.3401305057096247,
      "grad_norm": 5.011270523071289,
      "learning_rate": 4.9824171412947404e-05,
      "loss": 0.2082,
      "num_input_tokens_seen": 17741760,
      "step": 8215
    },
    {
      "epoch": 1.3409461663947797,
      "grad_norm": 4.1288604736328125,
      "learning_rate": 4.982332768992021e-05,
      "loss": 0.1214,
      "num_input_tokens_seen": 17754368,
      "step": 8220
    },
    {
      "epoch": 1.3417618270799347,
      "grad_norm": 2.3313276767730713,
      "learning_rate": 4.982248195458812e-05,
      "loss": 0.2546,
      "num_input_tokens_seen": 17764096,
      "step": 8225
    },
    {
      "epoch": 1.3425774877650898,
      "grad_norm": 2.1330060958862305,
      "learning_rate": 4.98216342070197e-05,
      "loss": 0.1471,
      "num_input_tokens_seen": 17774784,
      "step": 8230
    },
    {
      "epoch": 1.3433931484502448,
      "grad_norm": 0.05763570964336395,
      "learning_rate": 4.982078444728367e-05,
      "loss": 0.0151,
      "num_input_tokens_seen": 17786112,
      "step": 8235
    },
    {
      "epoch": 1.3442088091353996,
      "grad_norm": 5.828211307525635,
      "learning_rate": 4.981993267544891e-05,
      "loss": 0.1202,
      "num_input_tokens_seen": 17797120,
      "step": 8240
    },
    {
      "epoch": 1.3450244698205547,
      "grad_norm": 3.2131574153900146,
      "learning_rate": 4.9819078891584467e-05,
      "loss": 0.1813,
      "num_input_tokens_seen": 17808032,
      "step": 8245
    },
    {
      "epoch": 1.3458401305057097,
      "grad_norm": 0.8757337927818298,
      "learning_rate": 4.981822309575956e-05,
      "loss": 0.0207,
      "num_input_tokens_seen": 17818144,
      "step": 8250
    },
    {
      "epoch": 1.3466557911908645,
      "grad_norm": 0.15001745522022247,
      "learning_rate": 4.981736528804357e-05,
      "loss": 0.0431,
      "num_input_tokens_seen": 17828032,
      "step": 8255
    },
    {
      "epoch": 1.3474714518760196,
      "grad_norm": 1.268739938735962,
      "learning_rate": 4.9816505468506026e-05,
      "loss": 0.0701,
      "num_input_tokens_seen": 17839232,
      "step": 8260
    },
    {
      "epoch": 1.3482871125611746,
      "grad_norm": 2.76004958152771,
      "learning_rate": 4.981564363721663e-05,
      "loss": 0.0458,
      "num_input_tokens_seen": 17849088,
      "step": 8265
    },
    {
      "epoch": 1.3491027732463294,
      "grad_norm": 0.21174126863479614,
      "learning_rate": 4.981477979424524e-05,
      "loss": 0.0495,
      "num_input_tokens_seen": 17860512,
      "step": 8270
    },
    {
      "epoch": 1.3499184339314845,
      "grad_norm": 0.1321340948343277,
      "learning_rate": 4.98139139396619e-05,
      "loss": 0.3906,
      "num_input_tokens_seen": 17872000,
      "step": 8275
    },
    {
      "epoch": 1.3507340946166395,
      "grad_norm": 0.21824656426906586,
      "learning_rate": 4.981304607353678e-05,
      "loss": 0.1646,
      "num_input_tokens_seen": 17883712,
      "step": 8280
    },
    {
      "epoch": 1.3515497553017943,
      "grad_norm": 5.227387428283691,
      "learning_rate": 4.981217619594026e-05,
      "loss": 0.1417,
      "num_input_tokens_seen": 17894240,
      "step": 8285
    },
    {
      "epoch": 1.3523654159869494,
      "grad_norm": 0.08399225771427155,
      "learning_rate": 4.981130430694283e-05,
      "loss": 0.2309,
      "num_input_tokens_seen": 17903904,
      "step": 8290
    },
    {
      "epoch": 1.3531810766721044,
      "grad_norm": 0.37167224287986755,
      "learning_rate": 4.9810430406615194e-05,
      "loss": 0.0779,
      "num_input_tokens_seen": 17914208,
      "step": 8295
    },
    {
      "epoch": 1.3539967373572595,
      "grad_norm": 0.03977183252573013,
      "learning_rate": 4.980955449502818e-05,
      "loss": 0.0061,
      "num_input_tokens_seen": 17924640,
      "step": 8300
    },
    {
      "epoch": 1.3548123980424145,
      "grad_norm": 0.07032673805952072,
      "learning_rate": 4.980867657225279e-05,
      "loss": 0.0751,
      "num_input_tokens_seen": 17935520,
      "step": 8305
    },
    {
      "epoch": 1.3556280587275693,
      "grad_norm": 0.7361053824424744,
      "learning_rate": 4.980779663836019e-05,
      "loss": 0.1407,
      "num_input_tokens_seen": 17946848,
      "step": 8310
    },
    {
      "epoch": 1.3564437194127243,
      "grad_norm": 1.1821690797805786,
      "learning_rate": 4.980691469342174e-05,
      "loss": 0.0675,
      "num_input_tokens_seen": 17957568,
      "step": 8315
    },
    {
      "epoch": 1.3572593800978794,
      "grad_norm": 4.284687519073486,
      "learning_rate": 4.98060307375089e-05,
      "loss": 0.0545,
      "num_input_tokens_seen": 17967776,
      "step": 8320
    },
    {
      "epoch": 1.3580750407830342,
      "grad_norm": 4.312143802642822,
      "learning_rate": 4.980514477069336e-05,
      "loss": 0.1381,
      "num_input_tokens_seen": 17978688,
      "step": 8325
    },
    {
      "epoch": 1.3588907014681892,
      "grad_norm": 0.1006174311041832,
      "learning_rate": 4.980425679304691e-05,
      "loss": 0.1376,
      "num_input_tokens_seen": 17990432,
      "step": 8330
    },
    {
      "epoch": 1.3597063621533443,
      "grad_norm": 2.875581979751587,
      "learning_rate": 4.9803366804641556e-05,
      "loss": 0.1652,
      "num_input_tokens_seen": 18001888,
      "step": 8335
    },
    {
      "epoch": 1.360522022838499,
      "grad_norm": 0.1079099252820015,
      "learning_rate": 4.980247480554944e-05,
      "loss": 0.0801,
      "num_input_tokens_seen": 18013536,
      "step": 8340
    },
    {
      "epoch": 1.3613376835236541,
      "grad_norm": 5.862875461578369,
      "learning_rate": 4.980158079584286e-05,
      "loss": 0.4018,
      "num_input_tokens_seen": 18024864,
      "step": 8345
    },
    {
      "epoch": 1.3621533442088092,
      "grad_norm": 0.4484040439128876,
      "learning_rate": 4.9800684775594306e-05,
      "loss": 0.0091,
      "num_input_tokens_seen": 18036192,
      "step": 8350
    },
    {
      "epoch": 1.362969004893964,
      "grad_norm": 5.443084239959717,
      "learning_rate": 4.979978674487641e-05,
      "loss": 0.2277,
      "num_input_tokens_seen": 18044864,
      "step": 8355
    },
    {
      "epoch": 1.363784665579119,
      "grad_norm": 0.09923567622900009,
      "learning_rate": 4.979888670376196e-05,
      "loss": 0.2332,
      "num_input_tokens_seen": 18056192,
      "step": 8360
    },
    {
      "epoch": 1.364600326264274,
      "grad_norm": 0.11489620059728622,
      "learning_rate": 4.979798465232393e-05,
      "loss": 0.0727,
      "num_input_tokens_seen": 18066592,
      "step": 8365
    },
    {
      "epoch": 1.3654159869494291,
      "grad_norm": 4.0995683670043945,
      "learning_rate": 4.9797080590635434e-05,
      "loss": 0.1438,
      "num_input_tokens_seen": 18077408,
      "step": 8370
    },
    {
      "epoch": 1.366231647634584,
      "grad_norm": 2.340662956237793,
      "learning_rate": 4.979617451876978e-05,
      "loss": 0.0526,
      "num_input_tokens_seen": 18088544,
      "step": 8375
    },
    {
      "epoch": 1.367047308319739,
      "grad_norm": 4.116942882537842,
      "learning_rate": 4.979526643680039e-05,
      "loss": 0.262,
      "num_input_tokens_seen": 18100192,
      "step": 8380
    },
    {
      "epoch": 1.367862969004894,
      "grad_norm": 0.5059496164321899,
      "learning_rate": 4.9794356344800894e-05,
      "loss": 0.2898,
      "num_input_tokens_seen": 18110432,
      "step": 8385
    },
    {
      "epoch": 1.368678629690049,
      "grad_norm": 0.1756363958120346,
      "learning_rate": 4.9793444242845075e-05,
      "loss": 0.197,
      "num_input_tokens_seen": 18120928,
      "step": 8390
    },
    {
      "epoch": 1.3694942903752039,
      "grad_norm": 0.18464797735214233,
      "learning_rate": 4.979253013100686e-05,
      "loss": 0.1876,
      "num_input_tokens_seen": 18131104,
      "step": 8395
    },
    {
      "epoch": 1.370309951060359,
      "grad_norm": 0.05068230628967285,
      "learning_rate": 4.979161400936036e-05,
      "loss": 0.0163,
      "num_input_tokens_seen": 18142720,
      "step": 8400
    },
    {
      "epoch": 1.371125611745514,
      "grad_norm": 0.08261064440011978,
      "learning_rate": 4.979069587797984e-05,
      "loss": 0.1064,
      "num_input_tokens_seen": 18153664,
      "step": 8405
    },
    {
      "epoch": 1.3719412724306688,
      "grad_norm": 2.9787333011627197,
      "learning_rate": 4.978977573693972e-05,
      "loss": 0.5077,
      "num_input_tokens_seen": 18164512,
      "step": 8410
    },
    {
      "epoch": 1.3727569331158238,
      "grad_norm": 0.3857441544532776,
      "learning_rate": 4.97888535863146e-05,
      "loss": 0.3214,
      "num_input_tokens_seen": 18174368,
      "step": 8415
    },
    {
      "epoch": 1.3735725938009788,
      "grad_norm": 3.607125759124756,
      "learning_rate": 4.9787929426179224e-05,
      "loss": 0.1154,
      "num_input_tokens_seen": 18185216,
      "step": 8420
    },
    {
      "epoch": 1.3743882544861337,
      "grad_norm": 0.5385351777076721,
      "learning_rate": 4.978700325660852e-05,
      "loss": 0.1098,
      "num_input_tokens_seen": 18196256,
      "step": 8425
    },
    {
      "epoch": 1.3752039151712887,
      "grad_norm": 3.186066150665283,
      "learning_rate": 4.978607507767757e-05,
      "loss": 0.0432,
      "num_input_tokens_seen": 18207264,
      "step": 8430
    },
    {
      "epoch": 1.3760195758564437,
      "grad_norm": 3.8766305446624756,
      "learning_rate": 4.9785144889461606e-05,
      "loss": 0.3158,
      "num_input_tokens_seen": 18217760,
      "step": 8435
    },
    {
      "epoch": 1.3768352365415986,
      "grad_norm": 0.1176619753241539,
      "learning_rate": 4.978421269203604e-05,
      "loss": 0.125,
      "num_input_tokens_seen": 18228608,
      "step": 8440
    },
    {
      "epoch": 1.3776508972267536,
      "grad_norm": 5.708771705627441,
      "learning_rate": 4.9783278485476434e-05,
      "loss": 0.2388,
      "num_input_tokens_seen": 18239200,
      "step": 8445
    },
    {
      "epoch": 1.3784665579119086,
      "grad_norm": 4.8118367195129395,
      "learning_rate": 4.978234226985853e-05,
      "loss": 0.2544,
      "num_input_tokens_seen": 18251200,
      "step": 8450
    },
    {
      "epoch": 1.3792822185970637,
      "grad_norm": 1.8131279945373535,
      "learning_rate": 4.978140404525822e-05,
      "loss": 0.0307,
      "num_input_tokens_seen": 18261856,
      "step": 8455
    },
    {
      "epoch": 1.3800978792822187,
      "grad_norm": 0.2119222730398178,
      "learning_rate": 4.978046381175155e-05,
      "loss": 0.0128,
      "num_input_tokens_seen": 18273152,
      "step": 8460
    },
    {
      "epoch": 1.3809135399673735,
      "grad_norm": 0.21933406591415405,
      "learning_rate": 4.977952156941476e-05,
      "loss": 0.0756,
      "num_input_tokens_seen": 18284896,
      "step": 8465
    },
    {
      "epoch": 1.3817292006525286,
      "grad_norm": 0.1431412696838379,
      "learning_rate": 4.977857731832421e-05,
      "loss": 0.1027,
      "num_input_tokens_seen": 18294912,
      "step": 8470
    },
    {
      "epoch": 1.3825448613376836,
      "grad_norm": 0.31807583570480347,
      "learning_rate": 4.977763105855646e-05,
      "loss": 0.0976,
      "num_input_tokens_seen": 18304672,
      "step": 8475
    },
    {
      "epoch": 1.3833605220228384,
      "grad_norm": 0.24125567078590393,
      "learning_rate": 4.9776682790188225e-05,
      "loss": 0.1622,
      "num_input_tokens_seen": 18316256,
      "step": 8480
    },
    {
      "epoch": 1.3841761827079935,
      "grad_norm": 0.12713797390460968,
      "learning_rate": 4.977573251329636e-05,
      "loss": 0.2192,
      "num_input_tokens_seen": 18327904,
      "step": 8485
    },
    {
      "epoch": 1.3849918433931485,
      "grad_norm": 0.1462138593196869,
      "learning_rate": 4.97747802279579e-05,
      "loss": 0.0199,
      "num_input_tokens_seen": 18338752,
      "step": 8490
    },
    {
      "epoch": 1.3858075040783033,
      "grad_norm": 0.19010719656944275,
      "learning_rate": 4.9773825934250056e-05,
      "loss": 0.0993,
      "num_input_tokens_seen": 18350464,
      "step": 8495
    },
    {
      "epoch": 1.3866231647634584,
      "grad_norm": 2.17090106010437,
      "learning_rate": 4.977286963225018e-05,
      "loss": 0.2162,
      "num_input_tokens_seen": 18360832,
      "step": 8500
    },
    {
      "epoch": 1.3874388254486134,
      "grad_norm": 4.011931896209717,
      "learning_rate": 4.9771911322035794e-05,
      "loss": 0.1839,
      "num_input_tokens_seen": 18370976,
      "step": 8505
    },
    {
      "epoch": 1.3882544861337682,
      "grad_norm": 0.12120926380157471,
      "learning_rate": 4.977095100368459e-05,
      "loss": 0.0996,
      "num_input_tokens_seen": 18381920,
      "step": 8510
    },
    {
      "epoch": 1.3890701468189233,
      "grad_norm": 1.0785771608352661,
      "learning_rate": 4.9769988677274405e-05,
      "loss": 0.1777,
      "num_input_tokens_seen": 18393152,
      "step": 8515
    },
    {
      "epoch": 1.3898858075040783,
      "grad_norm": 0.7257936596870422,
      "learning_rate": 4.976902434288326e-05,
      "loss": 0.0733,
      "num_input_tokens_seen": 18403168,
      "step": 8520
    },
    {
      "epoch": 1.3907014681892333,
      "grad_norm": 0.11688599735498428,
      "learning_rate": 4.9768058000589325e-05,
      "loss": 0.1568,
      "num_input_tokens_seen": 18413792,
      "step": 8525
    },
    {
      "epoch": 1.3915171288743884,
      "grad_norm": 4.304495334625244,
      "learning_rate": 4.976708965047093e-05,
      "loss": 0.1298,
      "num_input_tokens_seen": 18424864,
      "step": 8530
    },
    {
      "epoch": 1.3923327895595432,
      "grad_norm": 3.4943485260009766,
      "learning_rate": 4.976611929260659e-05,
      "loss": 0.0414,
      "num_input_tokens_seen": 18434144,
      "step": 8535
    },
    {
      "epoch": 1.3931484502446982,
      "grad_norm": 1.9871363639831543,
      "learning_rate": 4.976514692707496e-05,
      "loss": 0.1636,
      "num_input_tokens_seen": 18444576,
      "step": 8540
    },
    {
      "epoch": 1.3939641109298533,
      "grad_norm": 0.3116692900657654,
      "learning_rate": 4.9764172553954855e-05,
      "loss": 0.2656,
      "num_input_tokens_seen": 18455808,
      "step": 8545
    },
    {
      "epoch": 1.394779771615008,
      "grad_norm": 2.6198909282684326,
      "learning_rate": 4.976319617332527e-05,
      "loss": 0.0423,
      "num_input_tokens_seen": 18466848,
      "step": 8550
    },
    {
      "epoch": 1.3955954323001631,
      "grad_norm": 0.45830538868904114,
      "learning_rate": 4.9762217785265356e-05,
      "loss": 0.2964,
      "num_input_tokens_seen": 18479040,
      "step": 8555
    },
    {
      "epoch": 1.3964110929853182,
      "grad_norm": 3.4091854095458984,
      "learning_rate": 4.976123738985443e-05,
      "loss": 0.1384,
      "num_input_tokens_seen": 18490080,
      "step": 8560
    },
    {
      "epoch": 1.397226753670473,
      "grad_norm": 5.418155193328857,
      "learning_rate": 4.976025498717196e-05,
      "loss": 0.3336,
      "num_input_tokens_seen": 18501632,
      "step": 8565
    },
    {
      "epoch": 1.398042414355628,
      "grad_norm": 0.10266997665166855,
      "learning_rate": 4.9759270577297603e-05,
      "loss": 0.1686,
      "num_input_tokens_seen": 18513056,
      "step": 8570
    },
    {
      "epoch": 1.398858075040783,
      "grad_norm": 0.42906904220581055,
      "learning_rate": 4.975828416031113e-05,
      "loss": 0.009,
      "num_input_tokens_seen": 18523616,
      "step": 8575
    },
    {
      "epoch": 1.399673735725938,
      "grad_norm": 0.7605011463165283,
      "learning_rate": 4.975729573629252e-05,
      "loss": 0.237,
      "num_input_tokens_seen": 18534688,
      "step": 8580
    },
    {
      "epoch": 1.400489396411093,
      "grad_norm": 0.226216658949852,
      "learning_rate": 4.9756305305321906e-05,
      "loss": 0.2454,
      "num_input_tokens_seen": 18545120,
      "step": 8585
    },
    {
      "epoch": 1.401305057096248,
      "grad_norm": 0.17236967384815216,
      "learning_rate": 4.975531286747958e-05,
      "loss": 0.0725,
      "num_input_tokens_seen": 18556224,
      "step": 8590
    },
    {
      "epoch": 1.402120717781403,
      "grad_norm": 3.434866428375244,
      "learning_rate": 4.975431842284597e-05,
      "loss": 0.4768,
      "num_input_tokens_seen": 18566272,
      "step": 8595
    },
    {
      "epoch": 1.4029363784665578,
      "grad_norm": 0.6368977427482605,
      "learning_rate": 4.975332197150171e-05,
      "loss": 0.1642,
      "num_input_tokens_seen": 18577216,
      "step": 8600
    },
    {
      "epoch": 1.4037520391517129,
      "grad_norm": 0.1678442656993866,
      "learning_rate": 4.975232351352758e-05,
      "loss": 0.3473,
      "num_input_tokens_seen": 18587072,
      "step": 8605
    },
    {
      "epoch": 1.404567699836868,
      "grad_norm": 4.314147472381592,
      "learning_rate": 4.975132304900451e-05,
      "loss": 0.1323,
      "num_input_tokens_seen": 18598112,
      "step": 8610
    },
    {
      "epoch": 1.405383360522023,
      "grad_norm": 0.5971012711524963,
      "learning_rate": 4.975032057801361e-05,
      "loss": 0.1713,
      "num_input_tokens_seen": 18608384,
      "step": 8615
    },
    {
      "epoch": 1.4061990212071778,
      "grad_norm": 4.2186713218688965,
      "learning_rate": 4.974931610063613e-05,
      "loss": 0.0658,
      "num_input_tokens_seen": 18618944,
      "step": 8620
    },
    {
      "epoch": 1.4070146818923328,
      "grad_norm": 0.15565381944179535,
      "learning_rate": 4.974830961695353e-05,
      "loss": 0.1091,
      "num_input_tokens_seen": 18629248,
      "step": 8625
    },
    {
      "epoch": 1.4078303425774878,
      "grad_norm": 0.4324505627155304,
      "learning_rate": 4.9747301127047366e-05,
      "loss": 0.032,
      "num_input_tokens_seen": 18641248,
      "step": 8630
    },
    {
      "epoch": 1.4086460032626427,
      "grad_norm": 0.2560870945453644,
      "learning_rate": 4.974629063099942e-05,
      "loss": 0.0226,
      "num_input_tokens_seen": 18651520,
      "step": 8635
    },
    {
      "epoch": 1.4094616639477977,
      "grad_norm": 1.649951457977295,
      "learning_rate": 4.974527812889158e-05,
      "loss": 0.1574,
      "num_input_tokens_seen": 18662304,
      "step": 8640
    },
    {
      "epoch": 1.4102773246329527,
      "grad_norm": 0.06403063982725143,
      "learning_rate": 4.974426362080594e-05,
      "loss": 0.1435,
      "num_input_tokens_seen": 18672416,
      "step": 8645
    },
    {
      "epoch": 1.4110929853181076,
      "grad_norm": 3.4695379734039307,
      "learning_rate": 4.974324710682474e-05,
      "loss": 0.0544,
      "num_input_tokens_seen": 18683136,
      "step": 8650
    },
    {
      "epoch": 1.4119086460032626,
      "grad_norm": 0.19298793375492096,
      "learning_rate": 4.974222858703039e-05,
      "loss": 0.2589,
      "num_input_tokens_seen": 18693472,
      "step": 8655
    },
    {
      "epoch": 1.4127243066884176,
      "grad_norm": 0.20529305934906006,
      "learning_rate": 4.9741208061505454e-05,
      "loss": 0.0915,
      "num_input_tokens_seen": 18704128,
      "step": 8660
    },
    {
      "epoch": 1.4135399673735725,
      "grad_norm": 0.9639427661895752,
      "learning_rate": 4.974018553033264e-05,
      "loss": 0.0374,
      "num_input_tokens_seen": 18715456,
      "step": 8665
    },
    {
      "epoch": 1.4143556280587275,
      "grad_norm": 0.4283669590950012,
      "learning_rate": 4.973916099359487e-05,
      "loss": 0.2163,
      "num_input_tokens_seen": 18726240,
      "step": 8670
    },
    {
      "epoch": 1.4151712887438825,
      "grad_norm": 0.44654086232185364,
      "learning_rate": 4.973813445137518e-05,
      "loss": 0.0779,
      "num_input_tokens_seen": 18736992,
      "step": 8675
    },
    {
      "epoch": 1.4159869494290376,
      "grad_norm": 0.32161450386047363,
      "learning_rate": 4.9737105903756794e-05,
      "loss": 0.0227,
      "num_input_tokens_seen": 18749216,
      "step": 8680
    },
    {
      "epoch": 1.4168026101141926,
      "grad_norm": 0.15897774696350098,
      "learning_rate": 4.973607535082309e-05,
      "loss": 0.0631,
      "num_input_tokens_seen": 18760352,
      "step": 8685
    },
    {
      "epoch": 1.4176182707993474,
      "grad_norm": 0.13181594014167786,
      "learning_rate": 4.97350427926576e-05,
      "loss": 0.0557,
      "num_input_tokens_seen": 18769792,
      "step": 8690
    },
    {
      "epoch": 1.4184339314845025,
      "grad_norm": 0.13483324646949768,
      "learning_rate": 4.973400822934404e-05,
      "loss": 0.2174,
      "num_input_tokens_seen": 18781696,
      "step": 8695
    },
    {
      "epoch": 1.4192495921696575,
      "grad_norm": 1.089842438697815,
      "learning_rate": 4.973297166096628e-05,
      "loss": 0.2138,
      "num_input_tokens_seen": 18790912,
      "step": 8700
    },
    {
      "epoch": 1.4200652528548123,
      "grad_norm": 4.54611349105835,
      "learning_rate": 4.9731933087608334e-05,
      "loss": 0.2273,
      "num_input_tokens_seen": 18800672,
      "step": 8705
    },
    {
      "epoch": 1.4208809135399674,
      "grad_norm": 4.534762382507324,
      "learning_rate": 4.973089250935441e-05,
      "loss": 0.1754,
      "num_input_tokens_seen": 18812032,
      "step": 8710
    },
    {
      "epoch": 1.4216965742251224,
      "grad_norm": 4.2552361488342285,
      "learning_rate": 4.972984992628885e-05,
      "loss": 0.3013,
      "num_input_tokens_seen": 18823328,
      "step": 8715
    },
    {
      "epoch": 1.4225122349102772,
      "grad_norm": 4.679597854614258,
      "learning_rate": 4.972880533849619e-05,
      "loss": 0.3067,
      "num_input_tokens_seen": 18834048,
      "step": 8720
    },
    {
      "epoch": 1.4233278955954323,
      "grad_norm": 1.3409359455108643,
      "learning_rate": 4.9727758746061084e-05,
      "loss": 0.1296,
      "num_input_tokens_seen": 18845216,
      "step": 8725
    },
    {
      "epoch": 1.4241435562805873,
      "grad_norm": 0.1467946171760559,
      "learning_rate": 4.972671014906839e-05,
      "loss": 0.1887,
      "num_input_tokens_seen": 18854848,
      "step": 8730
    },
    {
      "epoch": 1.4249592169657421,
      "grad_norm": 1.7204428911209106,
      "learning_rate": 4.972565954760311e-05,
      "loss": 0.1263,
      "num_input_tokens_seen": 18866368,
      "step": 8735
    },
    {
      "epoch": 1.4257748776508972,
      "grad_norm": 3.4446494579315186,
      "learning_rate": 4.9724606941750406e-05,
      "loss": 0.3014,
      "num_input_tokens_seen": 18878464,
      "step": 8740
    },
    {
      "epoch": 1.4265905383360522,
      "grad_norm": 0.11606528609991074,
      "learning_rate": 4.972355233159562e-05,
      "loss": 0.144,
      "num_input_tokens_seen": 18889856,
      "step": 8745
    },
    {
      "epoch": 1.4274061990212072,
      "grad_norm": 0.30330905318260193,
      "learning_rate": 4.972249571722423e-05,
      "loss": 0.188,
      "num_input_tokens_seen": 18900352,
      "step": 8750
    },
    {
      "epoch": 1.4282218597063623,
      "grad_norm": 0.20018945634365082,
      "learning_rate": 4.97214370987219e-05,
      "loss": 0.1408,
      "num_input_tokens_seen": 18911936,
      "step": 8755
    },
    {
      "epoch": 1.429037520391517,
      "grad_norm": 0.1850886046886444,
      "learning_rate": 4.972037647617444e-05,
      "loss": 0.0465,
      "num_input_tokens_seen": 18922048,
      "step": 8760
    },
    {
      "epoch": 1.4298531810766721,
      "grad_norm": 3.108037233352661,
      "learning_rate": 4.9719313849667835e-05,
      "loss": 0.1357,
      "num_input_tokens_seen": 18932288,
      "step": 8765
    },
    {
      "epoch": 1.4306688417618272,
      "grad_norm": 1.8283677101135254,
      "learning_rate": 4.9718249219288226e-05,
      "loss": 0.1841,
      "num_input_tokens_seen": 18943008,
      "step": 8770
    },
    {
      "epoch": 1.431484502446982,
      "grad_norm": 6.500990867614746,
      "learning_rate": 4.971718258512191e-05,
      "loss": 0.1459,
      "num_input_tokens_seen": 18953376,
      "step": 8775
    },
    {
      "epoch": 1.432300163132137,
      "grad_norm": 1.0945346355438232,
      "learning_rate": 4.971611394725537e-05,
      "loss": 0.0602,
      "num_input_tokens_seen": 18965600,
      "step": 8780
    },
    {
      "epoch": 1.433115823817292,
      "grad_norm": 0.162503182888031,
      "learning_rate": 4.971504330577521e-05,
      "loss": 0.0703,
      "num_input_tokens_seen": 18975936,
      "step": 8785
    },
    {
      "epoch": 1.433931484502447,
      "grad_norm": 0.03908466547727585,
      "learning_rate": 4.971397066076825e-05,
      "loss": 0.0595,
      "num_input_tokens_seen": 18986048,
      "step": 8790
    },
    {
      "epoch": 1.434747145187602,
      "grad_norm": 0.5271322727203369,
      "learning_rate": 4.971289601232143e-05,
      "loss": 0.1402,
      "num_input_tokens_seen": 18997056,
      "step": 8795
    },
    {
      "epoch": 1.435562805872757,
      "grad_norm": 0.08785668015480042,
      "learning_rate": 4.971181936052186e-05,
      "loss": 0.0255,
      "num_input_tokens_seen": 19007744,
      "step": 8800
    },
    {
      "epoch": 1.4363784665579118,
      "grad_norm": 4.008883953094482,
      "learning_rate": 4.971074070545684e-05,
      "loss": 0.1603,
      "num_input_tokens_seen": 19018784,
      "step": 8805
    },
    {
      "epoch": 1.4371941272430668,
      "grad_norm": 3.2489356994628906,
      "learning_rate": 4.970966004721378e-05,
      "loss": 0.112,
      "num_input_tokens_seen": 19030176,
      "step": 8810
    },
    {
      "epoch": 1.4380097879282219,
      "grad_norm": 0.3383443057537079,
      "learning_rate": 4.970857738588031e-05,
      "loss": 0.0913,
      "num_input_tokens_seen": 19041088,
      "step": 8815
    },
    {
      "epoch": 1.438825448613377,
      "grad_norm": 0.21210889518260956,
      "learning_rate": 4.9707492721544185e-05,
      "loss": 0.0082,
      "num_input_tokens_seen": 19052352,
      "step": 8820
    },
    {
      "epoch": 1.4396411092985317,
      "grad_norm": 1.9470065832138062,
      "learning_rate": 4.970640605429334e-05,
      "loss": 0.0698,
      "num_input_tokens_seen": 19062688,
      "step": 8825
    },
    {
      "epoch": 1.4404567699836868,
      "grad_norm": 6.853029251098633,
      "learning_rate": 4.970531738421585e-05,
      "loss": 0.1794,
      "num_input_tokens_seen": 19073504,
      "step": 8830
    },
    {
      "epoch": 1.4412724306688418,
      "grad_norm": 3.3483693599700928,
      "learning_rate": 4.970422671139999e-05,
      "loss": 0.3258,
      "num_input_tokens_seen": 19085792,
      "step": 8835
    },
    {
      "epoch": 1.4420880913539968,
      "grad_norm": 0.28674644231796265,
      "learning_rate": 4.970313403593416e-05,
      "loss": 0.1313,
      "num_input_tokens_seen": 19096512,
      "step": 8840
    },
    {
      "epoch": 1.4429037520391517,
      "grad_norm": 4.106522560119629,
      "learning_rate": 4.970203935790695e-05,
      "loss": 0.2026,
      "num_input_tokens_seen": 19106336,
      "step": 8845
    },
    {
      "epoch": 1.4437194127243067,
      "grad_norm": 4.873129367828369,
      "learning_rate": 4.970094267740708e-05,
      "loss": 0.1281,
      "num_input_tokens_seen": 19116800,
      "step": 8850
    },
    {
      "epoch": 1.4445350734094617,
      "grad_norm": 5.79468297958374,
      "learning_rate": 4.969984399452347e-05,
      "loss": 0.2079,
      "num_input_tokens_seen": 19127488,
      "step": 8855
    },
    {
      "epoch": 1.4453507340946166,
      "grad_norm": 0.08337918668985367,
      "learning_rate": 4.9698743309345184e-05,
      "loss": 0.2117,
      "num_input_tokens_seen": 19135264,
      "step": 8860
    },
    {
      "epoch": 1.4461663947797716,
      "grad_norm": 5.34727668762207,
      "learning_rate": 4.969764062196145e-05,
      "loss": 0.179,
      "num_input_tokens_seen": 19146656,
      "step": 8865
    },
    {
      "epoch": 1.4469820554649266,
      "grad_norm": 0.050858285278081894,
      "learning_rate": 4.969653593246164e-05,
      "loss": 0.0158,
      "num_input_tokens_seen": 19158880,
      "step": 8870
    },
    {
      "epoch": 1.4477977161500815,
      "grad_norm": 2.2910377979278564,
      "learning_rate": 4.9695429240935335e-05,
      "loss": 0.1317,
      "num_input_tokens_seen": 19170240,
      "step": 8875
    },
    {
      "epoch": 1.4486133768352365,
      "grad_norm": 0.17886285483837128,
      "learning_rate": 4.9694320547472215e-05,
      "loss": 0.0451,
      "num_input_tokens_seen": 19180960,
      "step": 8880
    },
    {
      "epoch": 1.4494290375203915,
      "grad_norm": 2.9980437755584717,
      "learning_rate": 4.9693209852162184e-05,
      "loss": 0.4063,
      "num_input_tokens_seen": 19192576,
      "step": 8885
    },
    {
      "epoch": 1.4502446982055464,
      "grad_norm": 1.1452898979187012,
      "learning_rate": 4.969209715509526e-05,
      "loss": 0.0696,
      "num_input_tokens_seen": 19202752,
      "step": 8890
    },
    {
      "epoch": 1.4510603588907014,
      "grad_norm": 3.7562623023986816,
      "learning_rate": 4.969098245636167e-05,
      "loss": 0.0542,
      "num_input_tokens_seen": 19214048,
      "step": 8895
    },
    {
      "epoch": 1.4518760195758564,
      "grad_norm": 6.766478061676025,
      "learning_rate": 4.968986575605175e-05,
      "loss": 0.2298,
      "num_input_tokens_seen": 19225056,
      "step": 8900
    },
    {
      "epoch": 1.4526916802610115,
      "grad_norm": 12.138822555541992,
      "learning_rate": 4.968874705425604e-05,
      "loss": 0.2153,
      "num_input_tokens_seen": 19236480,
      "step": 8905
    },
    {
      "epoch": 1.4535073409461665,
      "grad_norm": 5.017162799835205,
      "learning_rate": 4.968762635106522e-05,
      "loss": 0.24,
      "num_input_tokens_seen": 19248192,
      "step": 8910
    },
    {
      "epoch": 1.4543230016313213,
      "grad_norm": 0.35115402936935425,
      "learning_rate": 4.9686503646570146e-05,
      "loss": 0.1287,
      "num_input_tokens_seen": 19258368,
      "step": 8915
    },
    {
      "epoch": 1.4551386623164764,
      "grad_norm": 0.2280079871416092,
      "learning_rate": 4.9685378940861826e-05,
      "loss": 0.0872,
      "num_input_tokens_seen": 19269088,
      "step": 8920
    },
    {
      "epoch": 1.4559543230016314,
      "grad_norm": 2.5575084686279297,
      "learning_rate": 4.9684252234031446e-05,
      "loss": 0.1059,
      "num_input_tokens_seen": 19280384,
      "step": 8925
    },
    {
      "epoch": 1.4567699836867862,
      "grad_norm": 0.19328927993774414,
      "learning_rate": 4.968312352617033e-05,
      "loss": 0.2862,
      "num_input_tokens_seen": 19292192,
      "step": 8930
    },
    {
      "epoch": 1.4575856443719413,
      "grad_norm": 0.379935085773468,
      "learning_rate": 4.968199281736997e-05,
      "loss": 0.1058,
      "num_input_tokens_seen": 19303904,
      "step": 8935
    },
    {
      "epoch": 1.4584013050570963,
      "grad_norm": 0.6336538195610046,
      "learning_rate": 4.968086010772205e-05,
      "loss": 0.0835,
      "num_input_tokens_seen": 19315712,
      "step": 8940
    },
    {
      "epoch": 1.4592169657422511,
      "grad_norm": 0.13827356696128845,
      "learning_rate": 4.9679725397318375e-05,
      "loss": 0.2034,
      "num_input_tokens_seen": 19326848,
      "step": 8945
    },
    {
      "epoch": 1.4600326264274062,
      "grad_norm": 8.300313949584961,
      "learning_rate": 4.967858868625094e-05,
      "loss": 0.3573,
      "num_input_tokens_seen": 19338496,
      "step": 8950
    },
    {
      "epoch": 1.4608482871125612,
      "grad_norm": 6.544249057769775,
      "learning_rate": 4.967744997461188e-05,
      "loss": 0.3305,
      "num_input_tokens_seen": 19348992,
      "step": 8955
    },
    {
      "epoch": 1.461663947797716,
      "grad_norm": 2.2257773876190186,
      "learning_rate": 4.9676309262493513e-05,
      "loss": 0.1591,
      "num_input_tokens_seen": 19359680,
      "step": 8960
    },
    {
      "epoch": 1.462479608482871,
      "grad_norm": 3.6891024112701416,
      "learning_rate": 4.9675166549988314e-05,
      "loss": 0.076,
      "num_input_tokens_seen": 19370848,
      "step": 8965
    },
    {
      "epoch": 1.463295269168026,
      "grad_norm": 0.443911612033844,
      "learning_rate": 4.9674021837188917e-05,
      "loss": 0.1519,
      "num_input_tokens_seen": 19381888,
      "step": 8970
    },
    {
      "epoch": 1.4641109298531811,
      "grad_norm": 0.08385307341814041,
      "learning_rate": 4.967287512418811e-05,
      "loss": 0.0991,
      "num_input_tokens_seen": 19392768,
      "step": 8975
    },
    {
      "epoch": 1.4649265905383362,
      "grad_norm": 0.12367914617061615,
      "learning_rate": 4.9671726411078864e-05,
      "loss": 0.0911,
      "num_input_tokens_seen": 19403488,
      "step": 8980
    },
    {
      "epoch": 1.465742251223491,
      "grad_norm": 0.9014115333557129,
      "learning_rate": 4.967057569795428e-05,
      "loss": 0.2221,
      "num_input_tokens_seen": 19414656,
      "step": 8985
    },
    {
      "epoch": 1.466557911908646,
      "grad_norm": 0.43794381618499756,
      "learning_rate": 4.966942298490767e-05,
      "loss": 0.2248,
      "num_input_tokens_seen": 19424960,
      "step": 8990
    },
    {
      "epoch": 1.467373572593801,
      "grad_norm": 5.037932872772217,
      "learning_rate": 4.966826827203245e-05,
      "loss": 0.0941,
      "num_input_tokens_seen": 19436512,
      "step": 8995
    },
    {
      "epoch": 1.468189233278956,
      "grad_norm": 4.7633209228515625,
      "learning_rate": 4.966711155942223e-05,
      "loss": 0.1554,
      "num_input_tokens_seen": 19447808,
      "step": 9000
    },
    {
      "epoch": 1.469004893964111,
      "grad_norm": 0.4131033718585968,
      "learning_rate": 4.966595284717081e-05,
      "loss": 0.0731,
      "num_input_tokens_seen": 19457984,
      "step": 9005
    },
    {
      "epoch": 1.469820554649266,
      "grad_norm": 0.31178730726242065,
      "learning_rate": 4.966479213537207e-05,
      "loss": 0.0987,
      "num_input_tokens_seen": 19467680,
      "step": 9010
    },
    {
      "epoch": 1.4706362153344208,
      "grad_norm": 0.5420904159545898,
      "learning_rate": 4.966362942412015e-05,
      "loss": 0.16,
      "num_input_tokens_seen": 19475968,
      "step": 9015
    },
    {
      "epoch": 1.4714518760195758,
      "grad_norm": 3.456433057785034,
      "learning_rate": 4.9662464713509285e-05,
      "loss": 0.2759,
      "num_input_tokens_seen": 19487328,
      "step": 9020
    },
    {
      "epoch": 1.4722675367047309,
      "grad_norm": 0.2179572433233261,
      "learning_rate": 4.966129800363389e-05,
      "loss": 0.0207,
      "num_input_tokens_seen": 19497280,
      "step": 9025
    },
    {
      "epoch": 1.4730831973898857,
      "grad_norm": 0.4606347680091858,
      "learning_rate": 4.9660129294588554e-05,
      "loss": 0.1661,
      "num_input_tokens_seen": 19507264,
      "step": 9030
    },
    {
      "epoch": 1.4738988580750407,
      "grad_norm": 1.0950851440429688,
      "learning_rate": 4.965895858646801e-05,
      "loss": 0.1189,
      "num_input_tokens_seen": 19518080,
      "step": 9035
    },
    {
      "epoch": 1.4747145187601958,
      "grad_norm": 3.8063807487487793,
      "learning_rate": 4.9657785879367166e-05,
      "loss": 0.0902,
      "num_input_tokens_seen": 19528288,
      "step": 9040
    },
    {
      "epoch": 1.4755301794453508,
      "grad_norm": 7.3399786949157715,
      "learning_rate": 4.965661117338108e-05,
      "loss": 0.2115,
      "num_input_tokens_seen": 19539072,
      "step": 9045
    },
    {
      "epoch": 1.4763458401305056,
      "grad_norm": 3.9826085567474365,
      "learning_rate": 4.9655434468605e-05,
      "loss": 0.256,
      "num_input_tokens_seen": 19550496,
      "step": 9050
    },
    {
      "epoch": 1.4771615008156607,
      "grad_norm": 0.15139615535736084,
      "learning_rate": 4.9654255765134294e-05,
      "loss": 0.0415,
      "num_input_tokens_seen": 19562752,
      "step": 9055
    },
    {
      "epoch": 1.4779771615008157,
      "grad_norm": 0.25424888730049133,
      "learning_rate": 4.965307506306452e-05,
      "loss": 0.2528,
      "num_input_tokens_seen": 19572512,
      "step": 9060
    },
    {
      "epoch": 1.4787928221859707,
      "grad_norm": 0.16260232031345367,
      "learning_rate": 4.965189236249139e-05,
      "loss": 0.1952,
      "num_input_tokens_seen": 19582752,
      "step": 9065
    },
    {
      "epoch": 1.4796084828711256,
      "grad_norm": 4.861161231994629,
      "learning_rate": 4.9650707663510785e-05,
      "loss": 0.1756,
      "num_input_tokens_seen": 19593184,
      "step": 9070
    },
    {
      "epoch": 1.4804241435562806,
      "grad_norm": 1.71065092086792,
      "learning_rate": 4.9649520966218744e-05,
      "loss": 0.0373,
      "num_input_tokens_seen": 19604768,
      "step": 9075
    },
    {
      "epoch": 1.4812398042414356,
      "grad_norm": 0.21857862174510956,
      "learning_rate": 4.9648332270711463e-05,
      "loss": 0.1268,
      "num_input_tokens_seen": 19616576,
      "step": 9080
    },
    {
      "epoch": 1.4820554649265905,
      "grad_norm": 0.23377980291843414,
      "learning_rate": 4.96471415770853e-05,
      "loss": 0.0381,
      "num_input_tokens_seen": 19628384,
      "step": 9085
    },
    {
      "epoch": 1.4828711256117455,
      "grad_norm": 2.2004427909851074,
      "learning_rate": 4.964594888543678e-05,
      "loss": 0.1432,
      "num_input_tokens_seen": 19639136,
      "step": 9090
    },
    {
      "epoch": 1.4836867862969005,
      "grad_norm": 4.055203437805176,
      "learning_rate": 4.9644754195862597e-05,
      "loss": 0.1213,
      "num_input_tokens_seen": 19649856,
      "step": 9095
    },
    {
      "epoch": 1.4845024469820554,
      "grad_norm": 0.4328365623950958,
      "learning_rate": 4.964355750845959e-05,
      "loss": 0.1327,
      "num_input_tokens_seen": 19661120,
      "step": 9100
    },
    {
      "epoch": 1.4853181076672104,
      "grad_norm": 4.144128322601318,
      "learning_rate": 4.9642358823324776e-05,
      "loss": 0.1021,
      "num_input_tokens_seen": 19671648,
      "step": 9105
    },
    {
      "epoch": 1.4861337683523654,
      "grad_norm": 0.16398490965366364,
      "learning_rate": 4.964115814055531e-05,
      "loss": 0.0928,
      "num_input_tokens_seen": 19683584,
      "step": 9110
    },
    {
      "epoch": 1.4869494290375203,
      "grad_norm": 0.2402666062116623,
      "learning_rate": 4.963995546024854e-05,
      "loss": 0.292,
      "num_input_tokens_seen": 19694080,
      "step": 9115
    },
    {
      "epoch": 1.4877650897226753,
      "grad_norm": 3.6252024173736572,
      "learning_rate": 4.963875078250197e-05,
      "loss": 0.3565,
      "num_input_tokens_seen": 19705824,
      "step": 9120
    },
    {
      "epoch": 1.4885807504078303,
      "grad_norm": 3.403533697128296,
      "learning_rate": 4.963754410741324e-05,
      "loss": 0.3943,
      "num_input_tokens_seen": 19717056,
      "step": 9125
    },
    {
      "epoch": 1.4893964110929854,
      "grad_norm": 1.4570884704589844,
      "learning_rate": 4.9636335435080174e-05,
      "loss": 0.0448,
      "num_input_tokens_seen": 19728512,
      "step": 9130
    },
    {
      "epoch": 1.4902120717781404,
      "grad_norm": 4.06797981262207,
      "learning_rate": 4.963512476560075e-05,
      "loss": 0.0556,
      "num_input_tokens_seen": 19740768,
      "step": 9135
    },
    {
      "epoch": 1.4910277324632952,
      "grad_norm": 5.106830596923828,
      "learning_rate": 4.963391209907312e-05,
      "loss": 0.1786,
      "num_input_tokens_seen": 19752576,
      "step": 9140
    },
    {
      "epoch": 1.4918433931484503,
      "grad_norm": 0.10509225726127625,
      "learning_rate": 4.9632697435595585e-05,
      "loss": 0.015,
      "num_input_tokens_seen": 19764064,
      "step": 9145
    },
    {
      "epoch": 1.4926590538336053,
      "grad_norm": 0.033262308686971664,
      "learning_rate": 4.96314807752666e-05,
      "loss": 0.0454,
      "num_input_tokens_seen": 19776608,
      "step": 9150
    },
    {
      "epoch": 1.4934747145187601,
      "grad_norm": 0.31555286049842834,
      "learning_rate": 4.963026211818482e-05,
      "loss": 0.232,
      "num_input_tokens_seen": 19787488,
      "step": 9155
    },
    {
      "epoch": 1.4942903752039152,
      "grad_norm": 0.533806324005127,
      "learning_rate": 4.962904146444901e-05,
      "loss": 0.178,
      "num_input_tokens_seen": 19798240,
      "step": 9160
    },
    {
      "epoch": 1.4951060358890702,
      "grad_norm": 0.05178459733724594,
      "learning_rate": 4.962781881415814e-05,
      "loss": 0.4078,
      "num_input_tokens_seen": 19808384,
      "step": 9165
    },
    {
      "epoch": 1.495921696574225,
      "grad_norm": 1.5524178743362427,
      "learning_rate": 4.962659416741131e-05,
      "loss": 0.1784,
      "num_input_tokens_seen": 19817696,
      "step": 9170
    },
    {
      "epoch": 1.49673735725938,
      "grad_norm": 0.8013507723808289,
      "learning_rate": 4.962536752430781e-05,
      "loss": 0.1411,
      "num_input_tokens_seen": 19828384,
      "step": 9175
    },
    {
      "epoch": 1.497553017944535,
      "grad_norm": 0.268471360206604,
      "learning_rate": 4.962413888494706e-05,
      "loss": 0.0817,
      "num_input_tokens_seen": 19840704,
      "step": 9180
    },
    {
      "epoch": 1.49836867862969,
      "grad_norm": 1.3799679279327393,
      "learning_rate": 4.9622908249428676e-05,
      "loss": 0.1637,
      "num_input_tokens_seen": 19851648,
      "step": 9185
    },
    {
      "epoch": 1.499184339314845,
      "grad_norm": 4.304414749145508,
      "learning_rate": 4.962167561785241e-05,
      "loss": 0.1251,
      "num_input_tokens_seen": 19861568,
      "step": 9190
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.08585814386606216,
      "learning_rate": 4.962044099031819e-05,
      "loss": 0.0889,
      "num_input_tokens_seen": 19871232,
      "step": 9195
    },
    {
      "epoch": 1.5,
      "eval_loss": 0.1519882082939148,
      "eval_runtime": 133.2806,
      "eval_samples_per_second": 20.446,
      "eval_steps_per_second": 5.117,
      "num_input_tokens_seen": 19871232,
      "step": 9195
    },
    {
      "epoch": 1.5008156606851548,
      "grad_norm": 1.3533514738082886,
      "learning_rate": 4.9619204366926106e-05,
      "loss": 0.0204,
      "num_input_tokens_seen": 19882400,
      "step": 9200
    },
    {
      "epoch": 1.50163132137031,
      "grad_norm": 0.6818300485610962,
      "learning_rate": 4.96179657477764e-05,
      "loss": 0.1166,
      "num_input_tokens_seen": 19893824,
      "step": 9205
    },
    {
      "epoch": 1.502446982055465,
      "grad_norm": 0.19116778671741486,
      "learning_rate": 4.961672513296948e-05,
      "loss": 0.1069,
      "num_input_tokens_seen": 19905280,
      "step": 9210
    },
    {
      "epoch": 1.50326264274062,
      "grad_norm": 0.24817152321338654,
      "learning_rate": 4.9615482522605915e-05,
      "loss": 0.0113,
      "num_input_tokens_seen": 19916032,
      "step": 9215
    },
    {
      "epoch": 1.504078303425775,
      "grad_norm": 0.788390576839447,
      "learning_rate": 4.9614237916786434e-05,
      "loss": 0.1029,
      "num_input_tokens_seen": 19927232,
      "step": 9220
    },
    {
      "epoch": 1.5048939641109298,
      "grad_norm": 0.06866984814405441,
      "learning_rate": 4.961299131561194e-05,
      "loss": 0.0033,
      "num_input_tokens_seen": 19936992,
      "step": 9225
    },
    {
      "epoch": 1.5057096247960848,
      "grad_norm": 2.1778368949890137,
      "learning_rate": 4.961174271918349e-05,
      "loss": 0.158,
      "num_input_tokens_seen": 19948256,
      "step": 9230
    },
    {
      "epoch": 1.5065252854812399,
      "grad_norm": 1.7241088151931763,
      "learning_rate": 4.961049212760229e-05,
      "loss": 0.2009,
      "num_input_tokens_seen": 19960416,
      "step": 9235
    },
    {
      "epoch": 1.5073409461663947,
      "grad_norm": 5.672590255737305,
      "learning_rate": 4.960923954096972e-05,
      "loss": 0.1006,
      "num_input_tokens_seen": 19970656,
      "step": 9240
    },
    {
      "epoch": 1.5081566068515497,
      "grad_norm": 0.04307292029261589,
      "learning_rate": 4.960798495938734e-05,
      "loss": 0.1079,
      "num_input_tokens_seen": 19982176,
      "step": 9245
    },
    {
      "epoch": 1.5089722675367048,
      "grad_norm": 4.042102336883545,
      "learning_rate": 4.960672838295683e-05,
      "loss": 0.2614,
      "num_input_tokens_seen": 19991968,
      "step": 9250
    },
    {
      "epoch": 1.5097879282218596,
      "grad_norm": 2.628999948501587,
      "learning_rate": 4.960546981178007e-05,
      "loss": 0.2371,
      "num_input_tokens_seen": 20003968,
      "step": 9255
    },
    {
      "epoch": 1.5106035889070146,
      "grad_norm": 3.945488929748535,
      "learning_rate": 4.9604209245959076e-05,
      "loss": 0.1066,
      "num_input_tokens_seen": 20014784,
      "step": 9260
    },
    {
      "epoch": 1.5114192495921697,
      "grad_norm": 1.1156165599822998,
      "learning_rate": 4.960294668559604e-05,
      "loss": 0.289,
      "num_input_tokens_seen": 20024192,
      "step": 9265
    },
    {
      "epoch": 1.5122349102773245,
      "grad_norm": 0.1157669648528099,
      "learning_rate": 4.960168213079331e-05,
      "loss": 0.0745,
      "num_input_tokens_seen": 20034368,
      "step": 9270
    },
    {
      "epoch": 1.5130505709624797,
      "grad_norm": 8.24000072479248,
      "learning_rate": 4.9600415581653406e-05,
      "loss": 0.2667,
      "num_input_tokens_seen": 20045216,
      "step": 9275
    },
    {
      "epoch": 1.5138662316476346,
      "grad_norm": 0.4580290913581848,
      "learning_rate": 4.9599147038278984e-05,
      "loss": 0.0723,
      "num_input_tokens_seen": 20056928,
      "step": 9280
    },
    {
      "epoch": 1.5146818923327896,
      "grad_norm": 4.009837627410889,
      "learning_rate": 4.95978765007729e-05,
      "loss": 0.2275,
      "num_input_tokens_seen": 20067520,
      "step": 9285
    },
    {
      "epoch": 1.5154975530179446,
      "grad_norm": 0.13445672392845154,
      "learning_rate": 4.959660396923813e-05,
      "loss": 0.0468,
      "num_input_tokens_seen": 20077568,
      "step": 9290
    },
    {
      "epoch": 1.5163132137030995,
      "grad_norm": 2.3956339359283447,
      "learning_rate": 4.9595329443777836e-05,
      "loss": 0.3213,
      "num_input_tokens_seen": 20087584,
      "step": 9295
    },
    {
      "epoch": 1.5171288743882545,
      "grad_norm": 1.0057824850082397,
      "learning_rate": 4.959405292449535e-05,
      "loss": 0.1141,
      "num_input_tokens_seen": 20098880,
      "step": 9300
    },
    {
      "epoch": 1.5179445350734095,
      "grad_norm": 7.3586907386779785,
      "learning_rate": 4.959277441149415e-05,
      "loss": 0.2152,
      "num_input_tokens_seen": 20109568,
      "step": 9305
    },
    {
      "epoch": 1.5187601957585644,
      "grad_norm": 0.20948190987110138,
      "learning_rate": 4.959149390487786e-05,
      "loss": 0.0594,
      "num_input_tokens_seen": 20121184,
      "step": 9310
    },
    {
      "epoch": 1.5195758564437194,
      "grad_norm": 0.2613030970096588,
      "learning_rate": 4.959021140475031e-05,
      "loss": 0.1212,
      "num_input_tokens_seen": 20132608,
      "step": 9315
    },
    {
      "epoch": 1.5203915171288744,
      "grad_norm": 0.5939708948135376,
      "learning_rate": 4.958892691121545e-05,
      "loss": 0.0263,
      "num_input_tokens_seen": 20143200,
      "step": 9320
    },
    {
      "epoch": 1.5212071778140293,
      "grad_norm": 0.059133075177669525,
      "learning_rate": 4.958764042437741e-05,
      "loss": 0.1101,
      "num_input_tokens_seen": 20153344,
      "step": 9325
    },
    {
      "epoch": 1.5220228384991843,
      "grad_norm": 0.09682247042655945,
      "learning_rate": 4.958635194434048e-05,
      "loss": 0.1299,
      "num_input_tokens_seen": 20164288,
      "step": 9330
    },
    {
      "epoch": 1.5228384991843393,
      "grad_norm": 0.4423978924751282,
      "learning_rate": 4.958506147120912e-05,
      "loss": 0.0717,
      "num_input_tokens_seen": 20174880,
      "step": 9335
    },
    {
      "epoch": 1.5236541598694942,
      "grad_norm": 0.1162685751914978,
      "learning_rate": 4.958376900508792e-05,
      "loss": 0.1498,
      "num_input_tokens_seen": 20184960,
      "step": 9340
    },
    {
      "epoch": 1.5244698205546494,
      "grad_norm": 0.1906140297651291,
      "learning_rate": 4.958247454608167e-05,
      "loss": 0.0792,
      "num_input_tokens_seen": 20194464,
      "step": 9345
    },
    {
      "epoch": 1.5252854812398042,
      "grad_norm": 0.6163058876991272,
      "learning_rate": 4.958117809429531e-05,
      "loss": 0.4561,
      "num_input_tokens_seen": 20204352,
      "step": 9350
    },
    {
      "epoch": 1.5261011419249593,
      "grad_norm": 0.1939542442560196,
      "learning_rate": 4.9579879649833925e-05,
      "loss": 0.0988,
      "num_input_tokens_seen": 20214816,
      "step": 9355
    },
    {
      "epoch": 1.5269168026101143,
      "grad_norm": 3.6777944564819336,
      "learning_rate": 4.957857921280279e-05,
      "loss": 0.288,
      "num_input_tokens_seen": 20225664,
      "step": 9360
    },
    {
      "epoch": 1.5277324632952691,
      "grad_norm": 4.889492034912109,
      "learning_rate": 4.9577276783307296e-05,
      "loss": 0.1006,
      "num_input_tokens_seen": 20236672,
      "step": 9365
    },
    {
      "epoch": 1.5285481239804242,
      "grad_norm": 4.427472114562988,
      "learning_rate": 4.9575972361453046e-05,
      "loss": 0.2351,
      "num_input_tokens_seen": 20246784,
      "step": 9370
    },
    {
      "epoch": 1.5293637846655792,
      "grad_norm": 0.09633055329322815,
      "learning_rate": 4.957466594734579e-05,
      "loss": 0.0995,
      "num_input_tokens_seen": 20257088,
      "step": 9375
    },
    {
      "epoch": 1.530179445350734,
      "grad_norm": 0.11631061881780624,
      "learning_rate": 4.9573357541091414e-05,
      "loss": 0.0408,
      "num_input_tokens_seen": 20267520,
      "step": 9380
    },
    {
      "epoch": 1.530995106035889,
      "grad_norm": 0.1659749150276184,
      "learning_rate": 4.957204714279599e-05,
      "loss": 0.0962,
      "num_input_tokens_seen": 20277568,
      "step": 9385
    },
    {
      "epoch": 1.531810766721044,
      "grad_norm": 3.294208526611328,
      "learning_rate": 4.957073475256575e-05,
      "loss": 0.0976,
      "num_input_tokens_seen": 20288768,
      "step": 9390
    },
    {
      "epoch": 1.532626427406199,
      "grad_norm": 3.342176675796509,
      "learning_rate": 4.9569420370507087e-05,
      "loss": 0.0922,
      "num_input_tokens_seen": 20299968,
      "step": 9395
    },
    {
      "epoch": 1.533442088091354,
      "grad_norm": 2.932504892349243,
      "learning_rate": 4.956810399672653e-05,
      "loss": 0.3596,
      "num_input_tokens_seen": 20310592,
      "step": 9400
    },
    {
      "epoch": 1.534257748776509,
      "grad_norm": 4.083658218383789,
      "learning_rate": 4.956678563133082e-05,
      "loss": 0.4003,
      "num_input_tokens_seen": 20321088,
      "step": 9405
    },
    {
      "epoch": 1.5350734094616638,
      "grad_norm": 0.15010380744934082,
      "learning_rate": 4.956546527442681e-05,
      "loss": 0.1766,
      "num_input_tokens_seen": 20331136,
      "step": 9410
    },
    {
      "epoch": 1.535889070146819,
      "grad_norm": 2.304518222808838,
      "learning_rate": 4.956414292612154e-05,
      "loss": 0.1801,
      "num_input_tokens_seen": 20343136,
      "step": 9415
    },
    {
      "epoch": 1.536704730831974,
      "grad_norm": 0.20615914463996887,
      "learning_rate": 4.9562818586522206e-05,
      "loss": 0.1304,
      "num_input_tokens_seen": 20354656,
      "step": 9420
    },
    {
      "epoch": 1.5375203915171287,
      "grad_norm": 0.20106583833694458,
      "learning_rate": 4.9561492255736175e-05,
      "loss": 0.0132,
      "num_input_tokens_seen": 20365856,
      "step": 9425
    },
    {
      "epoch": 1.538336052202284,
      "grad_norm": 3.1261301040649414,
      "learning_rate": 4.9560163933870954e-05,
      "loss": 0.2497,
      "num_input_tokens_seen": 20376064,
      "step": 9430
    },
    {
      "epoch": 1.5391517128874388,
      "grad_norm": 1.1069012880325317,
      "learning_rate": 4.9558833621034224e-05,
      "loss": 0.1004,
      "num_input_tokens_seen": 20387104,
      "step": 9435
    },
    {
      "epoch": 1.5399673735725938,
      "grad_norm": 4.373126029968262,
      "learning_rate": 4.955750131733383e-05,
      "loss": 0.1297,
      "num_input_tokens_seen": 20398112,
      "step": 9440
    },
    {
      "epoch": 1.5407830342577489,
      "grad_norm": 0.38516080379486084,
      "learning_rate": 4.955616702287778e-05,
      "loss": 0.0138,
      "num_input_tokens_seen": 20409152,
      "step": 9445
    },
    {
      "epoch": 1.5415986949429037,
      "grad_norm": 1.4748661518096924,
      "learning_rate": 4.9554830737774226e-05,
      "loss": 0.086,
      "num_input_tokens_seen": 20420704,
      "step": 9450
    },
    {
      "epoch": 1.5424143556280587,
      "grad_norm": 0.3413572907447815,
      "learning_rate": 4.955349246213151e-05,
      "loss": 0.1354,
      "num_input_tokens_seen": 20431456,
      "step": 9455
    },
    {
      "epoch": 1.5432300163132138,
      "grad_norm": 3.291419744491577,
      "learning_rate": 4.9552152196058114e-05,
      "loss": 0.1367,
      "num_input_tokens_seen": 20442880,
      "step": 9460
    },
    {
      "epoch": 1.5440456769983686,
      "grad_norm": 1.0140867233276367,
      "learning_rate": 4.955080993966268e-05,
      "loss": 0.0152,
      "num_input_tokens_seen": 20454624,
      "step": 9465
    },
    {
      "epoch": 1.5448613376835236,
      "grad_norm": 1.005288004875183,
      "learning_rate": 4.954946569305402e-05,
      "loss": 0.1254,
      "num_input_tokens_seen": 20465824,
      "step": 9470
    },
    {
      "epoch": 1.5456769983686787,
      "grad_norm": 0.6372134685516357,
      "learning_rate": 4.9548119456341114e-05,
      "loss": 0.1936,
      "num_input_tokens_seen": 20477088,
      "step": 9475
    },
    {
      "epoch": 1.5464926590538335,
      "grad_norm": 0.03720410540699959,
      "learning_rate": 4.954677122963309e-05,
      "loss": 0.0782,
      "num_input_tokens_seen": 20487296,
      "step": 9480
    },
    {
      "epoch": 1.5473083197389887,
      "grad_norm": 0.13508373498916626,
      "learning_rate": 4.954542101303924e-05,
      "loss": 0.2571,
      "num_input_tokens_seen": 20498080,
      "step": 9485
    },
    {
      "epoch": 1.5481239804241436,
      "grad_norm": 3.5957140922546387,
      "learning_rate": 4.954406880666902e-05,
      "loss": 0.1838,
      "num_input_tokens_seen": 20509824,
      "step": 9490
    },
    {
      "epoch": 1.5489396411092984,
      "grad_norm": 1.9207464456558228,
      "learning_rate": 4.954271461063204e-05,
      "loss": 0.2569,
      "num_input_tokens_seen": 20520928,
      "step": 9495
    },
    {
      "epoch": 1.5497553017944536,
      "grad_norm": 1.5969135761260986,
      "learning_rate": 4.9541358425038095e-05,
      "loss": 0.2794,
      "num_input_tokens_seen": 20532288,
      "step": 9500
    },
    {
      "epoch": 1.5505709624796085,
      "grad_norm": 2.585832118988037,
      "learning_rate": 4.954000024999711e-05,
      "loss": 0.2777,
      "num_input_tokens_seen": 20543456,
      "step": 9505
    },
    {
      "epoch": 1.5513866231647635,
      "grad_norm": 1.9454797506332397,
      "learning_rate": 4.9538640085619184e-05,
      "loss": 0.0811,
      "num_input_tokens_seen": 20554592,
      "step": 9510
    },
    {
      "epoch": 1.5522022838499185,
      "grad_norm": 2.7306134700775146,
      "learning_rate": 4.953727793201459e-05,
      "loss": 0.0692,
      "num_input_tokens_seen": 20565504,
      "step": 9515
    },
    {
      "epoch": 1.5530179445350734,
      "grad_norm": 5.1417236328125,
      "learning_rate": 4.953591378929375e-05,
      "loss": 0.2984,
      "num_input_tokens_seen": 20576896,
      "step": 9520
    },
    {
      "epoch": 1.5538336052202284,
      "grad_norm": 1.9761908054351807,
      "learning_rate": 4.953454765756724e-05,
      "loss": 0.0693,
      "num_input_tokens_seen": 20586720,
      "step": 9525
    },
    {
      "epoch": 1.5546492659053834,
      "grad_norm": 3.7507734298706055,
      "learning_rate": 4.953317953694582e-05,
      "loss": 0.1395,
      "num_input_tokens_seen": 20597376,
      "step": 9530
    },
    {
      "epoch": 1.5554649265905383,
      "grad_norm": 0.5698208212852478,
      "learning_rate": 4.953180942754037e-05,
      "loss": 0.0368,
      "num_input_tokens_seen": 20608608,
      "step": 9535
    },
    {
      "epoch": 1.5562805872756933,
      "grad_norm": 2.6742403507232666,
      "learning_rate": 4.9530437329461987e-05,
      "loss": 0.0987,
      "num_input_tokens_seen": 20620128,
      "step": 9540
    },
    {
      "epoch": 1.5570962479608483,
      "grad_norm": 0.09604479372501373,
      "learning_rate": 4.952906324282188e-05,
      "loss": 0.0658,
      "num_input_tokens_seen": 20631200,
      "step": 9545
    },
    {
      "epoch": 1.5579119086460032,
      "grad_norm": 0.10088922083377838,
      "learning_rate": 4.952768716773145e-05,
      "loss": 0.0292,
      "num_input_tokens_seen": 20640352,
      "step": 9550
    },
    {
      "epoch": 1.5587275693311582,
      "grad_norm": 2.8340446949005127,
      "learning_rate": 4.9526309104302246e-05,
      "loss": 0.2607,
      "num_input_tokens_seen": 20652832,
      "step": 9555
    },
    {
      "epoch": 1.5595432300163132,
      "grad_norm": 0.8983645439147949,
      "learning_rate": 4.952492905264599e-05,
      "loss": 0.0425,
      "num_input_tokens_seen": 20663488,
      "step": 9560
    },
    {
      "epoch": 1.560358890701468,
      "grad_norm": 0.3439129590988159,
      "learning_rate": 4.9523547012874524e-05,
      "loss": 0.0462,
      "num_input_tokens_seen": 20674464,
      "step": 9565
    },
    {
      "epoch": 1.5611745513866233,
      "grad_norm": 1.29847252368927,
      "learning_rate": 4.952216298509993e-05,
      "loss": 0.1046,
      "num_input_tokens_seen": 20685760,
      "step": 9570
    },
    {
      "epoch": 1.5619902120717781,
      "grad_norm": 6.125744342803955,
      "learning_rate": 4.952077696943437e-05,
      "loss": 0.2323,
      "num_input_tokens_seen": 20697280,
      "step": 9575
    },
    {
      "epoch": 1.5628058727569332,
      "grad_norm": 0.07258804887533188,
      "learning_rate": 4.951938896599021e-05,
      "loss": 0.2327,
      "num_input_tokens_seen": 20708000,
      "step": 9580
    },
    {
      "epoch": 1.5636215334420882,
      "grad_norm": 0.2027614712715149,
      "learning_rate": 4.951799897487997e-05,
      "loss": 0.1106,
      "num_input_tokens_seen": 20719424,
      "step": 9585
    },
    {
      "epoch": 1.564437194127243,
      "grad_norm": 2.6210386753082275,
      "learning_rate": 4.951660699621633e-05,
      "loss": 0.1683,
      "num_input_tokens_seen": 20728096,
      "step": 9590
    },
    {
      "epoch": 1.565252854812398,
      "grad_norm": 3.2500882148742676,
      "learning_rate": 4.9515213030112135e-05,
      "loss": 0.1551,
      "num_input_tokens_seen": 20738816,
      "step": 9595
    },
    {
      "epoch": 1.566068515497553,
      "grad_norm": 0.1480618715286255,
      "learning_rate": 4.951381707668038e-05,
      "loss": 0.1608,
      "num_input_tokens_seen": 20749408,
      "step": 9600
    },
    {
      "epoch": 1.566884176182708,
      "grad_norm": 0.06155373901128769,
      "learning_rate": 4.951241913603423e-05,
      "loss": 0.0466,
      "num_input_tokens_seen": 20759200,
      "step": 9605
    },
    {
      "epoch": 1.567699836867863,
      "grad_norm": 1.1635830402374268,
      "learning_rate": 4.9511019208287014e-05,
      "loss": 0.0979,
      "num_input_tokens_seen": 20770560,
      "step": 9610
    },
    {
      "epoch": 1.568515497553018,
      "grad_norm": 2.791290521621704,
      "learning_rate": 4.9509617293552215e-05,
      "loss": 0.1495,
      "num_input_tokens_seen": 20781056,
      "step": 9615
    },
    {
      "epoch": 1.5693311582381728,
      "grad_norm": 1.9688714742660522,
      "learning_rate": 4.9508213391943467e-05,
      "loss": 0.0452,
      "num_input_tokens_seen": 20791648,
      "step": 9620
    },
    {
      "epoch": 1.5701468189233279,
      "grad_norm": 4.5324320793151855,
      "learning_rate": 4.950680750357459e-05,
      "loss": 0.1624,
      "num_input_tokens_seen": 20804128,
      "step": 9625
    },
    {
      "epoch": 1.570962479608483,
      "grad_norm": 7.816247463226318,
      "learning_rate": 4.950539962855956e-05,
      "loss": 0.0982,
      "num_input_tokens_seen": 20814752,
      "step": 9630
    },
    {
      "epoch": 1.5717781402936377,
      "grad_norm": 0.8073441982269287,
      "learning_rate": 4.9503989767012493e-05,
      "loss": 0.168,
      "num_input_tokens_seen": 20826720,
      "step": 9635
    },
    {
      "epoch": 1.572593800978793,
      "grad_norm": 4.1045379638671875,
      "learning_rate": 4.950257791904768e-05,
      "loss": 0.09,
      "num_input_tokens_seen": 20837536,
      "step": 9640
    },
    {
      "epoch": 1.5734094616639478,
      "grad_norm": 6.973872184753418,
      "learning_rate": 4.950116408477958e-05,
      "loss": 0.2424,
      "num_input_tokens_seen": 20847744,
      "step": 9645
    },
    {
      "epoch": 1.5742251223491026,
      "grad_norm": 4.554234504699707,
      "learning_rate": 4.94997482643228e-05,
      "loss": 0.3488,
      "num_input_tokens_seen": 20859136,
      "step": 9650
    },
    {
      "epoch": 1.5750407830342579,
      "grad_norm": 6.059474468231201,
      "learning_rate": 4.949833045779212e-05,
      "loss": 0.1507,
      "num_input_tokens_seen": 20868832,
      "step": 9655
    },
    {
      "epoch": 1.5758564437194127,
      "grad_norm": 0.741489589214325,
      "learning_rate": 4.9496910665302467e-05,
      "loss": 0.0825,
      "num_input_tokens_seen": 20878080,
      "step": 9660
    },
    {
      "epoch": 1.5766721044045677,
      "grad_norm": 1.6913024187088013,
      "learning_rate": 4.949548888696893e-05,
      "loss": 0.1598,
      "num_input_tokens_seen": 20889600,
      "step": 9665
    },
    {
      "epoch": 1.5774877650897228,
      "grad_norm": 5.594682216644287,
      "learning_rate": 4.9494065122906787e-05,
      "loss": 0.2739,
      "num_input_tokens_seen": 20899392,
      "step": 9670
    },
    {
      "epoch": 1.5783034257748776,
      "grad_norm": 0.30009037256240845,
      "learning_rate": 4.9492639373231436e-05,
      "loss": 0.2161,
      "num_input_tokens_seen": 20909888,
      "step": 9675
    },
    {
      "epoch": 1.5791190864600326,
      "grad_norm": 0.20817957818508148,
      "learning_rate": 4.949121163805847e-05,
      "loss": 0.0206,
      "num_input_tokens_seen": 20920032,
      "step": 9680
    },
    {
      "epoch": 1.5799347471451877,
      "grad_norm": 0.060204293578863144,
      "learning_rate": 4.948978191750362e-05,
      "loss": 0.1828,
      "num_input_tokens_seen": 20931136,
      "step": 9685
    },
    {
      "epoch": 1.5807504078303425,
      "grad_norm": 0.053723905235528946,
      "learning_rate": 4.948835021168278e-05,
      "loss": 0.0399,
      "num_input_tokens_seen": 20941792,
      "step": 9690
    },
    {
      "epoch": 1.5815660685154975,
      "grad_norm": 0.6115943193435669,
      "learning_rate": 4.9486916520712026e-05,
      "loss": 0.1158,
      "num_input_tokens_seen": 20952672,
      "step": 9695
    },
    {
      "epoch": 1.5823817292006526,
      "grad_norm": 0.3317805528640747,
      "learning_rate": 4.948548084470757e-05,
      "loss": 0.0291,
      "num_input_tokens_seen": 20964480,
      "step": 9700
    },
    {
      "epoch": 1.5831973898858074,
      "grad_norm": 0.36165598034858704,
      "learning_rate": 4.94840431837858e-05,
      "loss": 0.0418,
      "num_input_tokens_seen": 20975392,
      "step": 9705
    },
    {
      "epoch": 1.5840130505709626,
      "grad_norm": 0.12291105091571808,
      "learning_rate": 4.948260353806326e-05,
      "loss": 0.2428,
      "num_input_tokens_seen": 20985344,
      "step": 9710
    },
    {
      "epoch": 1.5848287112561175,
      "grad_norm": 2.6017065048217773,
      "learning_rate": 4.948116190765665e-05,
      "loss": 0.1741,
      "num_input_tokens_seen": 20995456,
      "step": 9715
    },
    {
      "epoch": 1.5856443719412723,
      "grad_norm": 0.10842294991016388,
      "learning_rate": 4.9479718292682846e-05,
      "loss": 0.0578,
      "num_input_tokens_seen": 21006816,
      "step": 9720
    },
    {
      "epoch": 1.5864600326264275,
      "grad_norm": 3.0721845626831055,
      "learning_rate": 4.9478272693258866e-05,
      "loss": 0.1575,
      "num_input_tokens_seen": 21017344,
      "step": 9725
    },
    {
      "epoch": 1.5872756933115824,
      "grad_norm": 4.885010242462158,
      "learning_rate": 4.94768251095019e-05,
      "loss": 0.2057,
      "num_input_tokens_seen": 21027040,
      "step": 9730
    },
    {
      "epoch": 1.5880913539967374,
      "grad_norm": 1.0548969507217407,
      "learning_rate": 4.9475375541529294e-05,
      "loss": 0.0189,
      "num_input_tokens_seen": 21037472,
      "step": 9735
    },
    {
      "epoch": 1.5889070146818924,
      "grad_norm": 7.931169509887695,
      "learning_rate": 4.947392398945856e-05,
      "loss": 0.2874,
      "num_input_tokens_seen": 21048800,
      "step": 9740
    },
    {
      "epoch": 1.5897226753670473,
      "grad_norm": 2.7945945262908936,
      "learning_rate": 4.9472470453407374e-05,
      "loss": 0.3119,
      "num_input_tokens_seen": 21060672,
      "step": 9745
    },
    {
      "epoch": 1.5905383360522023,
      "grad_norm": 0.10913848876953125,
      "learning_rate": 4.947101493349355e-05,
      "loss": 0.274,
      "num_input_tokens_seen": 21070880,
      "step": 9750
    },
    {
      "epoch": 1.5913539967373573,
      "grad_norm": 1.3329194784164429,
      "learning_rate": 4.94695574298351e-05,
      "loss": 0.323,
      "num_input_tokens_seen": 21081792,
      "step": 9755
    },
    {
      "epoch": 1.5921696574225122,
      "grad_norm": 0.2509414851665497,
      "learning_rate": 4.946809794255016e-05,
      "loss": 0.0835,
      "num_input_tokens_seen": 21093280,
      "step": 9760
    },
    {
      "epoch": 1.5929853181076672,
      "grad_norm": 2.833569049835205,
      "learning_rate": 4.946663647175706e-05,
      "loss": 0.0485,
      "num_input_tokens_seen": 21103264,
      "step": 9765
    },
    {
      "epoch": 1.5938009787928222,
      "grad_norm": 0.09655069559812546,
      "learning_rate": 4.946517301757426e-05,
      "loss": 0.1494,
      "num_input_tokens_seen": 21113632,
      "step": 9770
    },
    {
      "epoch": 1.594616639477977,
      "grad_norm": 2.8146326541900635,
      "learning_rate": 4.94637075801204e-05,
      "loss": 0.1869,
      "num_input_tokens_seen": 21124448,
      "step": 9775
    },
    {
      "epoch": 1.595432300163132,
      "grad_norm": 8.07801342010498,
      "learning_rate": 4.946224015951427e-05,
      "loss": 0.1695,
      "num_input_tokens_seen": 21135776,
      "step": 9780
    },
    {
      "epoch": 1.5962479608482871,
      "grad_norm": 0.2656671702861786,
      "learning_rate": 4.946077075587484e-05,
      "loss": 0.2517,
      "num_input_tokens_seen": 21147072,
      "step": 9785
    },
    {
      "epoch": 1.597063621533442,
      "grad_norm": 0.2358125001192093,
      "learning_rate": 4.945929936932122e-05,
      "loss": 0.1235,
      "num_input_tokens_seen": 21158848,
      "step": 9790
    },
    {
      "epoch": 1.5978792822185972,
      "grad_norm": 2.6165988445281982,
      "learning_rate": 4.945782599997269e-05,
      "loss": 0.1093,
      "num_input_tokens_seen": 21168448,
      "step": 9795
    },
    {
      "epoch": 1.598694942903752,
      "grad_norm": 5.36940860748291,
      "learning_rate": 4.945635064794869e-05,
      "loss": 0.2194,
      "num_input_tokens_seen": 21179744,
      "step": 9800
    },
    {
      "epoch": 1.599510603588907,
      "grad_norm": 3.525139570236206,
      "learning_rate": 4.94548733133688e-05,
      "loss": 0.1811,
      "num_input_tokens_seen": 21191904,
      "step": 9805
    },
    {
      "epoch": 1.600326264274062,
      "grad_norm": 0.09005577862262726,
      "learning_rate": 4.945339399635281e-05,
      "loss": 0.124,
      "num_input_tokens_seen": 21202848,
      "step": 9810
    },
    {
      "epoch": 1.601141924959217,
      "grad_norm": 0.1309748888015747,
      "learning_rate": 4.945191269702062e-05,
      "loss": 0.0408,
      "num_input_tokens_seen": 21214304,
      "step": 9815
    },
    {
      "epoch": 1.601957585644372,
      "grad_norm": 2.700035810470581,
      "learning_rate": 4.945042941549233e-05,
      "loss": 0.0976,
      "num_input_tokens_seen": 21224448,
      "step": 9820
    },
    {
      "epoch": 1.602773246329527,
      "grad_norm": 4.2569260597229,
      "learning_rate": 4.944894415188815e-05,
      "loss": 0.2857,
      "num_input_tokens_seen": 21235392,
      "step": 9825
    },
    {
      "epoch": 1.6035889070146818,
      "grad_norm": 4.408572196960449,
      "learning_rate": 4.944745690632852e-05,
      "loss": 0.1675,
      "num_input_tokens_seen": 21246656,
      "step": 9830
    },
    {
      "epoch": 1.6044045676998369,
      "grad_norm": 5.793549060821533,
      "learning_rate": 4.944596767893399e-05,
      "loss": 0.1217,
      "num_input_tokens_seen": 21257376,
      "step": 9835
    },
    {
      "epoch": 1.605220228384992,
      "grad_norm": 0.5074617862701416,
      "learning_rate": 4.944447646982529e-05,
      "loss": 0.134,
      "num_input_tokens_seen": 21267456,
      "step": 9840
    },
    {
      "epoch": 1.6060358890701467,
      "grad_norm": 3.0906171798706055,
      "learning_rate": 4.9442983279123276e-05,
      "loss": 0.2018,
      "num_input_tokens_seen": 21279104,
      "step": 9845
    },
    {
      "epoch": 1.6068515497553018,
      "grad_norm": 0.09473452717065811,
      "learning_rate": 4.944148810694903e-05,
      "loss": 0.2745,
      "num_input_tokens_seen": 21291328,
      "step": 9850
    },
    {
      "epoch": 1.6076672104404568,
      "grad_norm": 3.595895290374756,
      "learning_rate": 4.9439990953423735e-05,
      "loss": 0.0922,
      "num_input_tokens_seen": 21302016,
      "step": 9855
    },
    {
      "epoch": 1.6084828711256116,
      "grad_norm": 1.9205716848373413,
      "learning_rate": 4.943849181866876e-05,
      "loss": 0.0972,
      "num_input_tokens_seen": 21312448,
      "step": 9860
    },
    {
      "epoch": 1.6092985318107669,
      "grad_norm": 0.738785982131958,
      "learning_rate": 4.943699070280565e-05,
      "loss": 0.0264,
      "num_input_tokens_seen": 21323136,
      "step": 9865
    },
    {
      "epoch": 1.6101141924959217,
      "grad_norm": 0.8316954970359802,
      "learning_rate": 4.9435487605956084e-05,
      "loss": 0.0611,
      "num_input_tokens_seen": 21334368,
      "step": 9870
    },
    {
      "epoch": 1.6109298531810765,
      "grad_norm": 1.3046718835830688,
      "learning_rate": 4.94339825282419e-05,
      "loss": 0.0991,
      "num_input_tokens_seen": 21344992,
      "step": 9875
    },
    {
      "epoch": 1.6117455138662318,
      "grad_norm": 0.14773312211036682,
      "learning_rate": 4.943247546978512e-05,
      "loss": 0.0802,
      "num_input_tokens_seen": 21354688,
      "step": 9880
    },
    {
      "epoch": 1.6125611745513866,
      "grad_norm": 3.677513837814331,
      "learning_rate": 4.943096643070791e-05,
      "loss": 0.281,
      "num_input_tokens_seen": 21365696,
      "step": 9885
    },
    {
      "epoch": 1.6133768352365416,
      "grad_norm": 2.2047548294067383,
      "learning_rate": 4.9429455411132596e-05,
      "loss": 0.2397,
      "num_input_tokens_seen": 21378240,
      "step": 9890
    },
    {
      "epoch": 1.6141924959216967,
      "grad_norm": 7.5614013671875,
      "learning_rate": 4.942794241118167e-05,
      "loss": 0.4745,
      "num_input_tokens_seen": 21388992,
      "step": 9895
    },
    {
      "epoch": 1.6150081566068515,
      "grad_norm": 1.1326520442962646,
      "learning_rate": 4.9426427430977796e-05,
      "loss": 0.1331,
      "num_input_tokens_seen": 21399808,
      "step": 9900
    },
    {
      "epoch": 1.6158238172920065,
      "grad_norm": 1.2015715837478638,
      "learning_rate": 4.942491047064377e-05,
      "loss": 0.0292,
      "num_input_tokens_seen": 21409696,
      "step": 9905
    },
    {
      "epoch": 1.6166394779771616,
      "grad_norm": 0.10749071836471558,
      "learning_rate": 4.942339153030257e-05,
      "loss": 0.0778,
      "num_input_tokens_seen": 21420416,
      "step": 9910
    },
    {
      "epoch": 1.6174551386623164,
      "grad_norm": 3.0217397212982178,
      "learning_rate": 4.942187061007732e-05,
      "loss": 0.3587,
      "num_input_tokens_seen": 21431648,
      "step": 9915
    },
    {
      "epoch": 1.6182707993474714,
      "grad_norm": 1.1433262825012207,
      "learning_rate": 4.942034771009134e-05,
      "loss": 0.2492,
      "num_input_tokens_seen": 21441856,
      "step": 9920
    },
    {
      "epoch": 1.6190864600326265,
      "grad_norm": 2.3705992698669434,
      "learning_rate": 4.941882283046806e-05,
      "loss": 0.1498,
      "num_input_tokens_seen": 21452736,
      "step": 9925
    },
    {
      "epoch": 1.6199021207177813,
      "grad_norm": 0.23972788453102112,
      "learning_rate": 4.94172959713311e-05,
      "loss": 0.1306,
      "num_input_tokens_seen": 21463552,
      "step": 9930
    },
    {
      "epoch": 1.6207177814029365,
      "grad_norm": 2.9236345291137695,
      "learning_rate": 4.941576713280424e-05,
      "loss": 0.2271,
      "num_input_tokens_seen": 21475136,
      "step": 9935
    },
    {
      "epoch": 1.6215334420880914,
      "grad_norm": 0.05357769504189491,
      "learning_rate": 4.941423631501141e-05,
      "loss": 0.1754,
      "num_input_tokens_seen": 21485536,
      "step": 9940
    },
    {
      "epoch": 1.6223491027732462,
      "grad_norm": 0.1600756049156189,
      "learning_rate": 4.941270351807671e-05,
      "loss": 0.058,
      "num_input_tokens_seen": 21496320,
      "step": 9945
    },
    {
      "epoch": 1.6231647634584014,
      "grad_norm": 3.389401435852051,
      "learning_rate": 4.941116874212439e-05,
      "loss": 0.1327,
      "num_input_tokens_seen": 21506688,
      "step": 9950
    },
    {
      "epoch": 1.6239804241435563,
      "grad_norm": 5.759918689727783,
      "learning_rate": 4.940963198727887e-05,
      "loss": 0.3432,
      "num_input_tokens_seen": 21517184,
      "step": 9955
    },
    {
      "epoch": 1.6247960848287113,
      "grad_norm": 0.1930082142353058,
      "learning_rate": 4.940809325366473e-05,
      "loss": 0.0287,
      "num_input_tokens_seen": 21527104,
      "step": 9960
    },
    {
      "epoch": 1.6256117455138663,
      "grad_norm": 6.6077470779418945,
      "learning_rate": 4.9406552541406707e-05,
      "loss": 0.2845,
      "num_input_tokens_seen": 21537024,
      "step": 9965
    },
    {
      "epoch": 1.6264274061990212,
      "grad_norm": 4.8132004737854,
      "learning_rate": 4.94050098506297e-05,
      "loss": 0.112,
      "num_input_tokens_seen": 21547648,
      "step": 9970
    },
    {
      "epoch": 1.6272430668841762,
      "grad_norm": 5.695581912994385,
      "learning_rate": 4.940346518145876e-05,
      "loss": 0.1869,
      "num_input_tokens_seen": 21558944,
      "step": 9975
    },
    {
      "epoch": 1.6280587275693312,
      "grad_norm": 6.937102317810059,
      "learning_rate": 4.940191853401911e-05,
      "loss": 0.3996,
      "num_input_tokens_seen": 21569408,
      "step": 9980
    },
    {
      "epoch": 1.628874388254486,
      "grad_norm": 2.9301321506500244,
      "learning_rate": 4.940036990843613e-05,
      "loss": 0.1656,
      "num_input_tokens_seen": 21580384,
      "step": 9985
    },
    {
      "epoch": 1.629690048939641,
      "grad_norm": 2.733616590499878,
      "learning_rate": 4.9398819304835364e-05,
      "loss": 0.0979,
      "num_input_tokens_seen": 21591328,
      "step": 9990
    },
    {
      "epoch": 1.6305057096247961,
      "grad_norm": 1.712278962135315,
      "learning_rate": 4.93972667233425e-05,
      "loss": 0.0576,
      "num_input_tokens_seen": 21603456,
      "step": 9995
    },
    {
      "epoch": 1.631321370309951,
      "grad_norm": 1.8469611406326294,
      "learning_rate": 4.9395712164083406e-05,
      "loss": 0.162,
      "num_input_tokens_seen": 21613248,
      "step": 10000
    },
    {
      "epoch": 1.632137030995106,
      "grad_norm": 0.2925925552845001,
      "learning_rate": 4.93941556271841e-05,
      "loss": 0.0475,
      "num_input_tokens_seen": 21624256,
      "step": 10005
    },
    {
      "epoch": 1.632952691680261,
      "grad_norm": 0.38624608516693115,
      "learning_rate": 4.9392597112770765e-05,
      "loss": 0.1058,
      "num_input_tokens_seen": 21636032,
      "step": 10010
    },
    {
      "epoch": 1.6337683523654158,
      "grad_norm": 3.0619940757751465,
      "learning_rate": 4.939103662096974e-05,
      "loss": 0.1101,
      "num_input_tokens_seen": 21647264,
      "step": 10015
    },
    {
      "epoch": 1.634584013050571,
      "grad_norm": 4.746956825256348,
      "learning_rate": 4.938947415190754e-05,
      "loss": 0.0825,
      "num_input_tokens_seen": 21658496,
      "step": 10020
    },
    {
      "epoch": 1.635399673735726,
      "grad_norm": 1.4105510711669922,
      "learning_rate": 4.93879097057108e-05,
      "loss": 0.1926,
      "num_input_tokens_seen": 21669248,
      "step": 10025
    },
    {
      "epoch": 1.636215334420881,
      "grad_norm": 2.56160044670105,
      "learning_rate": 4.938634328250636e-05,
      "loss": 0.0852,
      "num_input_tokens_seen": 21679840,
      "step": 10030
    },
    {
      "epoch": 1.637030995106036,
      "grad_norm": 0.12285134196281433,
      "learning_rate": 4.93847748824212e-05,
      "loss": 0.1782,
      "num_input_tokens_seen": 21691904,
      "step": 10035
    },
    {
      "epoch": 1.6378466557911908,
      "grad_norm": 0.17458020150661469,
      "learning_rate": 4.938320450558246e-05,
      "loss": 0.0126,
      "num_input_tokens_seen": 21701408,
      "step": 10040
    },
    {
      "epoch": 1.6386623164763459,
      "grad_norm": 0.1861964613199234,
      "learning_rate": 4.938163215211745e-05,
      "loss": 0.0405,
      "num_input_tokens_seen": 21712256,
      "step": 10045
    },
    {
      "epoch": 1.639477977161501,
      "grad_norm": 0.37605103850364685,
      "learning_rate": 4.938005782215362e-05,
      "loss": 0.0657,
      "num_input_tokens_seen": 21722304,
      "step": 10050
    },
    {
      "epoch": 1.6402936378466557,
      "grad_norm": 0.7383480072021484,
      "learning_rate": 4.93784815158186e-05,
      "loss": 0.0866,
      "num_input_tokens_seen": 21733152,
      "step": 10055
    },
    {
      "epoch": 1.6411092985318108,
      "grad_norm": 2.417325973510742,
      "learning_rate": 4.937690323324017e-05,
      "loss": 0.157,
      "num_input_tokens_seen": 21744160,
      "step": 10060
    },
    {
      "epoch": 1.6419249592169658,
      "grad_norm": 3.986969232559204,
      "learning_rate": 4.9375322974546285e-05,
      "loss": 0.0965,
      "num_input_tokens_seen": 21754432,
      "step": 10065
    },
    {
      "epoch": 1.6427406199021206,
      "grad_norm": 0.06366409361362457,
      "learning_rate": 4.937374073986504e-05,
      "loss": 0.0182,
      "num_input_tokens_seen": 21764928,
      "step": 10070
    },
    {
      "epoch": 1.6435562805872757,
      "grad_norm": 2.6650710105895996,
      "learning_rate": 4.937215652932469e-05,
      "loss": 0.2547,
      "num_input_tokens_seen": 21775008,
      "step": 10075
    },
    {
      "epoch": 1.6443719412724307,
      "grad_norm": 4.918404579162598,
      "learning_rate": 4.937057034305368e-05,
      "loss": 0.1259,
      "num_input_tokens_seen": 21785792,
      "step": 10080
    },
    {
      "epoch": 1.6451876019575855,
      "grad_norm": 0.11018738150596619,
      "learning_rate": 4.9368982181180576e-05,
      "loss": 0.1627,
      "num_input_tokens_seen": 21796704,
      "step": 10085
    },
    {
      "epoch": 1.6460032626427408,
      "grad_norm": 0.10577882081270218,
      "learning_rate": 4.936739204383413e-05,
      "loss": 0.3467,
      "num_input_tokens_seen": 21807840,
      "step": 10090
    },
    {
      "epoch": 1.6468189233278956,
      "grad_norm": 3.6516923904418945,
      "learning_rate": 4.936579993114324e-05,
      "loss": 0.1327,
      "num_input_tokens_seen": 21818528,
      "step": 10095
    },
    {
      "epoch": 1.6476345840130504,
      "grad_norm": 0.052826497703790665,
      "learning_rate": 4.936420584323699e-05,
      "loss": 0.0168,
      "num_input_tokens_seen": 21829472,
      "step": 10100
    },
    {
      "epoch": 1.6484502446982057,
      "grad_norm": 5.709690093994141,
      "learning_rate": 4.936260978024458e-05,
      "loss": 0.5152,
      "num_input_tokens_seen": 21841024,
      "step": 10105
    },
    {
      "epoch": 1.6492659053833605,
      "grad_norm": 1.8346400260925293,
      "learning_rate": 4.936101174229541e-05,
      "loss": 0.4441,
      "num_input_tokens_seen": 21852096,
      "step": 10110
    },
    {
      "epoch": 1.6500815660685155,
      "grad_norm": 4.051974773406982,
      "learning_rate": 4.935941172951902e-05,
      "loss": 0.1744,
      "num_input_tokens_seen": 21863904,
      "step": 10115
    },
    {
      "epoch": 1.6508972267536706,
      "grad_norm": 1.8480687141418457,
      "learning_rate": 4.9357809742045126e-05,
      "loss": 0.1047,
      "num_input_tokens_seen": 21874048,
      "step": 10120
    },
    {
      "epoch": 1.6517128874388254,
      "grad_norm": 4.244266986846924,
      "learning_rate": 4.935620578000358e-05,
      "loss": 0.1778,
      "num_input_tokens_seen": 21885632,
      "step": 10125
    },
    {
      "epoch": 1.6525285481239804,
      "grad_norm": 0.2436949461698532,
      "learning_rate": 4.935459984352441e-05,
      "loss": 0.2817,
      "num_input_tokens_seen": 21896864,
      "step": 10130
    },
    {
      "epoch": 1.6533442088091355,
      "grad_norm": 2.2461154460906982,
      "learning_rate": 4.93529919327378e-05,
      "loss": 0.0961,
      "num_input_tokens_seen": 21907712,
      "step": 10135
    },
    {
      "epoch": 1.6541598694942903,
      "grad_norm": 0.9836466908454895,
      "learning_rate": 4.9351382047774095e-05,
      "loss": 0.1588,
      "num_input_tokens_seen": 21918464,
      "step": 10140
    },
    {
      "epoch": 1.6549755301794453,
      "grad_norm": 0.1777764856815338,
      "learning_rate": 4.934977018876381e-05,
      "loss": 0.2104,
      "num_input_tokens_seen": 21929664,
      "step": 10145
    },
    {
      "epoch": 1.6557911908646004,
      "grad_norm": 0.8241901397705078,
      "learning_rate": 4.93481563558376e-05,
      "loss": 0.1703,
      "num_input_tokens_seen": 21940288,
      "step": 10150
    },
    {
      "epoch": 1.6566068515497552,
      "grad_norm": 3.3961079120635986,
      "learning_rate": 4.9346540549126305e-05,
      "loss": 0.044,
      "num_input_tokens_seen": 21950464,
      "step": 10155
    },
    {
      "epoch": 1.6574225122349104,
      "grad_norm": 1.6570359468460083,
      "learning_rate": 4.934492276876089e-05,
      "loss": 0.1409,
      "num_input_tokens_seen": 21961696,
      "step": 10160
    },
    {
      "epoch": 1.6582381729200653,
      "grad_norm": 0.07733599096536636,
      "learning_rate": 4.934330301487251e-05,
      "loss": 0.1334,
      "num_input_tokens_seen": 21972192,
      "step": 10165
    },
    {
      "epoch": 1.65905383360522,
      "grad_norm": 0.19854383170604706,
      "learning_rate": 4.934168128759248e-05,
      "loss": 0.2047,
      "num_input_tokens_seen": 21982112,
      "step": 10170
    },
    {
      "epoch": 1.6598694942903753,
      "grad_norm": 4.915229320526123,
      "learning_rate": 4.9340057587052245e-05,
      "loss": 0.1606,
      "num_input_tokens_seen": 21993568,
      "step": 10175
    },
    {
      "epoch": 1.6606851549755302,
      "grad_norm": 1.7136536836624146,
      "learning_rate": 4.9338431913383444e-05,
      "loss": 0.0769,
      "num_input_tokens_seen": 22003296,
      "step": 10180
    },
    {
      "epoch": 1.6615008156606852,
      "grad_norm": 0.07920566201210022,
      "learning_rate": 4.9336804266717864e-05,
      "loss": 0.062,
      "num_input_tokens_seen": 22013344,
      "step": 10185
    },
    {
      "epoch": 1.6623164763458402,
      "grad_norm": 0.15559551119804382,
      "learning_rate": 4.933517464718744e-05,
      "loss": 0.2693,
      "num_input_tokens_seen": 22022112,
      "step": 10190
    },
    {
      "epoch": 1.663132137030995,
      "grad_norm": 0.16380707919597626,
      "learning_rate": 4.933354305492429e-05,
      "loss": 0.1376,
      "num_input_tokens_seen": 22032704,
      "step": 10195
    },
    {
      "epoch": 1.66394779771615,
      "grad_norm": 0.12119174748659134,
      "learning_rate": 4.933190949006068e-05,
      "loss": 0.0976,
      "num_input_tokens_seen": 22044320,
      "step": 10200
    },
    {
      "epoch": 1.6647634584013051,
      "grad_norm": 0.12529213726520538,
      "learning_rate": 4.933027395272901e-05,
      "loss": 0.0086,
      "num_input_tokens_seen": 22055008,
      "step": 10205
    },
    {
      "epoch": 1.66557911908646,
      "grad_norm": 0.555073618888855,
      "learning_rate": 4.9328636443061894e-05,
      "loss": 0.133,
      "num_input_tokens_seen": 22065920,
      "step": 10210
    },
    {
      "epoch": 1.666394779771615,
      "grad_norm": 0.13457554578781128,
      "learning_rate": 4.932699696119207e-05,
      "loss": 0.0583,
      "num_input_tokens_seen": 22077024,
      "step": 10215
    },
    {
      "epoch": 1.66721044045677,
      "grad_norm": 3.605102777481079,
      "learning_rate": 4.932535550725243e-05,
      "loss": 0.3744,
      "num_input_tokens_seen": 22088320,
      "step": 10220
    },
    {
      "epoch": 1.6680261011419248,
      "grad_norm": 0.05995646119117737,
      "learning_rate": 4.932371208137605e-05,
      "loss": 0.0263,
      "num_input_tokens_seen": 22098528,
      "step": 10225
    },
    {
      "epoch": 1.6688417618270799,
      "grad_norm": 2.473446846008301,
      "learning_rate": 4.932206668369615e-05,
      "loss": 0.157,
      "num_input_tokens_seen": 22109792,
      "step": 10230
    },
    {
      "epoch": 1.669657422512235,
      "grad_norm": 0.20219111442565918,
      "learning_rate": 4.932041931434611e-05,
      "loss": 0.2797,
      "num_input_tokens_seen": 22120064,
      "step": 10235
    },
    {
      "epoch": 1.6704730831973897,
      "grad_norm": 0.13476024568080902,
      "learning_rate": 4.931876997345949e-05,
      "loss": 0.0701,
      "num_input_tokens_seen": 22131200,
      "step": 10240
    },
    {
      "epoch": 1.671288743882545,
      "grad_norm": 0.13390684127807617,
      "learning_rate": 4.931711866116998e-05,
      "loss": 0.1255,
      "num_input_tokens_seen": 22142880,
      "step": 10245
    },
    {
      "epoch": 1.6721044045676998,
      "grad_norm": 0.10579252988100052,
      "learning_rate": 4.9315465377611445e-05,
      "loss": 0.2971,
      "num_input_tokens_seen": 22151744,
      "step": 10250
    },
    {
      "epoch": 1.6729200652528549,
      "grad_norm": 5.694382190704346,
      "learning_rate": 4.9313810122917914e-05,
      "loss": 0.1125,
      "num_input_tokens_seen": 22162432,
      "step": 10255
    },
    {
      "epoch": 1.67373572593801,
      "grad_norm": 2.870382785797119,
      "learning_rate": 4.931215289722357e-05,
      "loss": 0.067,
      "num_input_tokens_seen": 22174432,
      "step": 10260
    },
    {
      "epoch": 1.6745513866231647,
      "grad_norm": 2.902642250061035,
      "learning_rate": 4.931049370066275e-05,
      "loss": 0.2172,
      "num_input_tokens_seen": 22185184,
      "step": 10265
    },
    {
      "epoch": 1.6753670473083198,
      "grad_norm": 0.22015023231506348,
      "learning_rate": 4.930883253336996e-05,
      "loss": 0.1038,
      "num_input_tokens_seen": 22196480,
      "step": 10270
    },
    {
      "epoch": 1.6761827079934748,
      "grad_norm": 0.15334255993366241,
      "learning_rate": 4.930716939547986e-05,
      "loss": 0.1083,
      "num_input_tokens_seen": 22207392,
      "step": 10275
    },
    {
      "epoch": 1.6769983686786296,
      "grad_norm": 4.0195231437683105,
      "learning_rate": 4.930550428712728e-05,
      "loss": 0.1289,
      "num_input_tokens_seen": 22218880,
      "step": 10280
    },
    {
      "epoch": 1.6778140293637847,
      "grad_norm": 0.15493622422218323,
      "learning_rate": 4.93038372084472e-05,
      "loss": 0.2316,
      "num_input_tokens_seen": 22229184,
      "step": 10285
    },
    {
      "epoch": 1.6786296900489397,
      "grad_norm": 5.129248142242432,
      "learning_rate": 4.9302168159574756e-05,
      "loss": 0.2169,
      "num_input_tokens_seen": 22240096,
      "step": 10290
    },
    {
      "epoch": 1.6794453507340945,
      "grad_norm": 0.1276528239250183,
      "learning_rate": 4.930049714064525e-05,
      "loss": 0.3012,
      "num_input_tokens_seen": 22251616,
      "step": 10295
    },
    {
      "epoch": 1.6802610114192496,
      "grad_norm": 1.9392294883728027,
      "learning_rate": 4.9298824151794154e-05,
      "loss": 0.1245,
      "num_input_tokens_seen": 22262464,
      "step": 10300
    },
    {
      "epoch": 1.6810766721044046,
      "grad_norm": 3.2291712760925293,
      "learning_rate": 4.9297149193157075e-05,
      "loss": 0.1465,
      "num_input_tokens_seen": 22272640,
      "step": 10305
    },
    {
      "epoch": 1.6818923327895594,
      "grad_norm": 5.780815124511719,
      "learning_rate": 4.9295472264869804e-05,
      "loss": 0.1781,
      "num_input_tokens_seen": 22283264,
      "step": 10310
    },
    {
      "epoch": 1.6827079934747147,
      "grad_norm": 0.17058567702770233,
      "learning_rate": 4.929379336706827e-05,
      "loss": 0.1163,
      "num_input_tokens_seen": 22294112,
      "step": 10315
    },
    {
      "epoch": 1.6835236541598695,
      "grad_norm": 0.38892847299575806,
      "learning_rate": 4.9292112499888584e-05,
      "loss": 0.177,
      "num_input_tokens_seen": 22305184,
      "step": 10320
    },
    {
      "epoch": 1.6843393148450243,
      "grad_norm": 0.9305853843688965,
      "learning_rate": 4.929042966346701e-05,
      "loss": 0.123,
      "num_input_tokens_seen": 22315840,
      "step": 10325
    },
    {
      "epoch": 1.6851549755301796,
      "grad_norm": 2.4178287982940674,
      "learning_rate": 4.928874485793995e-05,
      "loss": 0.1348,
      "num_input_tokens_seen": 22325728,
      "step": 10330
    },
    {
      "epoch": 1.6859706362153344,
      "grad_norm": 3.133922815322876,
      "learning_rate": 4.928705808344399e-05,
      "loss": 0.1868,
      "num_input_tokens_seen": 22335904,
      "step": 10335
    },
    {
      "epoch": 1.6867862969004894,
      "grad_norm": 0.4173702597618103,
      "learning_rate": 4.928536934011587e-05,
      "loss": 0.2283,
      "num_input_tokens_seen": 22345184,
      "step": 10340
    },
    {
      "epoch": 1.6876019575856445,
      "grad_norm": 9.072600364685059,
      "learning_rate": 4.92836786280925e-05,
      "loss": 0.2349,
      "num_input_tokens_seen": 22356736,
      "step": 10345
    },
    {
      "epoch": 1.6884176182707993,
      "grad_norm": 1.4616029262542725,
      "learning_rate": 4.9281985947510915e-05,
      "loss": 0.1325,
      "num_input_tokens_seen": 22368704,
      "step": 10350
    },
    {
      "epoch": 1.6892332789559543,
      "grad_norm": 0.3134356439113617,
      "learning_rate": 4.9280291298508355e-05,
      "loss": 0.2812,
      "num_input_tokens_seen": 22379904,
      "step": 10355
    },
    {
      "epoch": 1.6900489396411094,
      "grad_norm": 5.039915561676025,
      "learning_rate": 4.927859468122217e-05,
      "loss": 0.164,
      "num_input_tokens_seen": 22391776,
      "step": 10360
    },
    {
      "epoch": 1.6908646003262642,
      "grad_norm": 2.0959596633911133,
      "learning_rate": 4.9276896095789924e-05,
      "loss": 0.0899,
      "num_input_tokens_seen": 22401792,
      "step": 10365
    },
    {
      "epoch": 1.6916802610114192,
      "grad_norm": 2.153296947479248,
      "learning_rate": 4.927519554234929e-05,
      "loss": 0.1166,
      "num_input_tokens_seen": 22411968,
      "step": 10370
    },
    {
      "epoch": 1.6924959216965743,
      "grad_norm": 3.7540318965911865,
      "learning_rate": 4.9273493021038146e-05,
      "loss": 0.1372,
      "num_input_tokens_seen": 22424000,
      "step": 10375
    },
    {
      "epoch": 1.693311582381729,
      "grad_norm": 3.609611749649048,
      "learning_rate": 4.927178853199449e-05,
      "loss": 0.1268,
      "num_input_tokens_seen": 22434048,
      "step": 10380
    },
    {
      "epoch": 1.6941272430668843,
      "grad_norm": 0.1421138048171997,
      "learning_rate": 4.927008207535651e-05,
      "loss": 0.0819,
      "num_input_tokens_seen": 22446176,
      "step": 10385
    },
    {
      "epoch": 1.6949429037520392,
      "grad_norm": 0.2003869116306305,
      "learning_rate": 4.9268373651262515e-05,
      "loss": 0.0414,
      "num_input_tokens_seen": 22457600,
      "step": 10390
    },
    {
      "epoch": 1.695758564437194,
      "grad_norm": 0.43225133419036865,
      "learning_rate": 4.9266663259851025e-05,
      "loss": 0.0819,
      "num_input_tokens_seen": 22468768,
      "step": 10395
    },
    {
      "epoch": 1.6965742251223492,
      "grad_norm": 5.00507926940918,
      "learning_rate": 4.926495090126068e-05,
      "loss": 0.2094,
      "num_input_tokens_seen": 22480448,
      "step": 10400
    },
    {
      "epoch": 1.697389885807504,
      "grad_norm": 2.335838556289673,
      "learning_rate": 4.92632365756303e-05,
      "loss": 0.0683,
      "num_input_tokens_seen": 22490880,
      "step": 10405
    },
    {
      "epoch": 1.698205546492659,
      "grad_norm": 0.15659040212631226,
      "learning_rate": 4.926152028309885e-05,
      "loss": 0.3744,
      "num_input_tokens_seen": 22500704,
      "step": 10410
    },
    {
      "epoch": 1.6990212071778141,
      "grad_norm": 0.08425473421812057,
      "learning_rate": 4.9259802023805466e-05,
      "loss": 0.0573,
      "num_input_tokens_seen": 22511040,
      "step": 10415
    },
    {
      "epoch": 1.699836867862969,
      "grad_norm": 0.16577233374118805,
      "learning_rate": 4.9258081797889434e-05,
      "loss": 0.1264,
      "num_input_tokens_seen": 22521664,
      "step": 10420
    },
    {
      "epoch": 1.700652528548124,
      "grad_norm": 0.4766976833343506,
      "learning_rate": 4.925635960549021e-05,
      "loss": 0.1119,
      "num_input_tokens_seen": 22532192,
      "step": 10425
    },
    {
      "epoch": 1.701468189233279,
      "grad_norm": 0.28510862588882446,
      "learning_rate": 4.92546354467474e-05,
      "loss": 0.1663,
      "num_input_tokens_seen": 22543712,
      "step": 10430
    },
    {
      "epoch": 1.7022838499184338,
      "grad_norm": 0.7656769156455994,
      "learning_rate": 4.9252909321800775e-05,
      "loss": 0.2079,
      "num_input_tokens_seen": 22555232,
      "step": 10435
    },
    {
      "epoch": 1.7030995106035889,
      "grad_norm": 0.7301608920097351,
      "learning_rate": 4.925118123079026e-05,
      "loss": 0.0141,
      "num_input_tokens_seen": 22566720,
      "step": 10440
    },
    {
      "epoch": 1.703915171288744,
      "grad_norm": 0.6367702484130859,
      "learning_rate": 4.924945117385594e-05,
      "loss": 0.0552,
      "num_input_tokens_seen": 22577376,
      "step": 10445
    },
    {
      "epoch": 1.7047308319738987,
      "grad_norm": 3.5113401412963867,
      "learning_rate": 4.9247719151138086e-05,
      "loss": 0.2347,
      "num_input_tokens_seen": 22589568,
      "step": 10450
    },
    {
      "epoch": 1.7055464926590538,
      "grad_norm": 1.2844570875167847,
      "learning_rate": 4.924598516277707e-05,
      "loss": 0.1636,
      "num_input_tokens_seen": 22600672,
      "step": 10455
    },
    {
      "epoch": 1.7063621533442088,
      "grad_norm": 0.269253671169281,
      "learning_rate": 4.924424920891347e-05,
      "loss": 0.28,
      "num_input_tokens_seen": 22611200,
      "step": 10460
    },
    {
      "epoch": 1.7071778140293636,
      "grad_norm": 0.36347147822380066,
      "learning_rate": 4.9242511289688024e-05,
      "loss": 0.1163,
      "num_input_tokens_seen": 22620608,
      "step": 10465
    },
    {
      "epoch": 1.707993474714519,
      "grad_norm": 0.0721467137336731,
      "learning_rate": 4.924077140524161e-05,
      "loss": 0.1647,
      "num_input_tokens_seen": 22631168,
      "step": 10470
    },
    {
      "epoch": 1.7088091353996737,
      "grad_norm": 0.4332468807697296,
      "learning_rate": 4.9239029555715264e-05,
      "loss": 0.2585,
      "num_input_tokens_seen": 22640640,
      "step": 10475
    },
    {
      "epoch": 1.7096247960848288,
      "grad_norm": 0.36143365502357483,
      "learning_rate": 4.92372857412502e-05,
      "loss": 0.1593,
      "num_input_tokens_seen": 22650976,
      "step": 10480
    },
    {
      "epoch": 1.7104404567699838,
      "grad_norm": 1.6781527996063232,
      "learning_rate": 4.9235539961987766e-05,
      "loss": 0.2733,
      "num_input_tokens_seen": 22661472,
      "step": 10485
    },
    {
      "epoch": 1.7112561174551386,
      "grad_norm": 0.6840170621871948,
      "learning_rate": 4.9233792218069494e-05,
      "loss": 0.1775,
      "num_input_tokens_seen": 22673024,
      "step": 10490
    },
    {
      "epoch": 1.7120717781402937,
      "grad_norm": 0.07148187607526779,
      "learning_rate": 4.923204250963707e-05,
      "loss": 0.1668,
      "num_input_tokens_seen": 22684416,
      "step": 10495
    },
    {
      "epoch": 1.7128874388254487,
      "grad_norm": 0.15426300466060638,
      "learning_rate": 4.923029083683233e-05,
      "loss": 0.0727,
      "num_input_tokens_seen": 22695136,
      "step": 10500
    },
    {
      "epoch": 1.7137030995106035,
      "grad_norm": 0.618942379951477,
      "learning_rate": 4.9228537199797263e-05,
      "loss": 0.204,
      "num_input_tokens_seen": 22705408,
      "step": 10505
    },
    {
      "epoch": 1.7145187601957586,
      "grad_norm": 4.171415328979492,
      "learning_rate": 4.9226781598674047e-05,
      "loss": 0.1214,
      "num_input_tokens_seen": 22716000,
      "step": 10510
    },
    {
      "epoch": 1.7153344208809136,
      "grad_norm": 4.400952339172363,
      "learning_rate": 4.922502403360498e-05,
      "loss": 0.2735,
      "num_input_tokens_seen": 22727936,
      "step": 10515
    },
    {
      "epoch": 1.7161500815660684,
      "grad_norm": 6.595067024230957,
      "learning_rate": 4.922326450473255e-05,
      "loss": 0.3739,
      "num_input_tokens_seen": 22737632,
      "step": 10520
    },
    {
      "epoch": 1.7169657422512234,
      "grad_norm": 4.229296684265137,
      "learning_rate": 4.9221503012199386e-05,
      "loss": 0.1762,
      "num_input_tokens_seen": 22748288,
      "step": 10525
    },
    {
      "epoch": 1.7177814029363785,
      "grad_norm": 2.971266508102417,
      "learning_rate": 4.92197395561483e-05,
      "loss": 0.234,
      "num_input_tokens_seen": 22760288,
      "step": 10530
    },
    {
      "epoch": 1.7185970636215333,
      "grad_norm": 0.2944638133049011,
      "learning_rate": 4.9217974136722235e-05,
      "loss": 0.1311,
      "num_input_tokens_seen": 22772032,
      "step": 10535
    },
    {
      "epoch": 1.7194127243066886,
      "grad_norm": 1.3731074333190918,
      "learning_rate": 4.92162067540643e-05,
      "loss": 0.0977,
      "num_input_tokens_seen": 22783392,
      "step": 10540
    },
    {
      "epoch": 1.7202283849918434,
      "grad_norm": 0.07093972712755203,
      "learning_rate": 4.921443740831778e-05,
      "loss": 0.072,
      "num_input_tokens_seen": 22795808,
      "step": 10545
    },
    {
      "epoch": 1.7210440456769984,
      "grad_norm": 0.4293394982814789,
      "learning_rate": 4.9212666099626095e-05,
      "loss": 0.298,
      "num_input_tokens_seen": 22806400,
      "step": 10550
    },
    {
      "epoch": 1.7218597063621535,
      "grad_norm": 2.244494915008545,
      "learning_rate": 4.9210892828132835e-05,
      "loss": 0.1399,
      "num_input_tokens_seen": 22816256,
      "step": 10555
    },
    {
      "epoch": 1.7226753670473083,
      "grad_norm": 0.49821212887763977,
      "learning_rate": 4.920911759398177e-05,
      "loss": 0.0197,
      "num_input_tokens_seen": 22827200,
      "step": 10560
    },
    {
      "epoch": 1.7234910277324633,
      "grad_norm": 1.3760017156600952,
      "learning_rate": 4.920734039731679e-05,
      "loss": 0.1363,
      "num_input_tokens_seen": 22837248,
      "step": 10565
    },
    {
      "epoch": 1.7243066884176184,
      "grad_norm": 0.7770042419433594,
      "learning_rate": 4.9205561238281985e-05,
      "loss": 0.076,
      "num_input_tokens_seen": 22846048,
      "step": 10570
    },
    {
      "epoch": 1.7251223491027732,
      "grad_norm": 4.274723529815674,
      "learning_rate": 4.920378011702155e-05,
      "loss": 0.1818,
      "num_input_tokens_seen": 22856672,
      "step": 10575
    },
    {
      "epoch": 1.7259380097879282,
      "grad_norm": 2.931492328643799,
      "learning_rate": 4.92019970336799e-05,
      "loss": 0.1957,
      "num_input_tokens_seen": 22868224,
      "step": 10580
    },
    {
      "epoch": 1.7267536704730833,
      "grad_norm": 4.527055740356445,
      "learning_rate": 4.920021198840157e-05,
      "loss": 0.3171,
      "num_input_tokens_seen": 22879584,
      "step": 10585
    },
    {
      "epoch": 1.727569331158238,
      "grad_norm": 0.10872957110404968,
      "learning_rate": 4.919842498133126e-05,
      "loss": 0.0381,
      "num_input_tokens_seen": 22889696,
      "step": 10590
    },
    {
      "epoch": 1.7283849918433931,
      "grad_norm": 0.2734326124191284,
      "learning_rate": 4.919663601261384e-05,
      "loss": 0.0757,
      "num_input_tokens_seen": 22900224,
      "step": 10595
    },
    {
      "epoch": 1.7292006525285482,
      "grad_norm": 2.8099300861358643,
      "learning_rate": 4.919484508239434e-05,
      "loss": 0.2513,
      "num_input_tokens_seen": 22911136,
      "step": 10600
    },
    {
      "epoch": 1.730016313213703,
      "grad_norm": 0.2676504850387573,
      "learning_rate": 4.9193052190817926e-05,
      "loss": 0.2708,
      "num_input_tokens_seen": 22922400,
      "step": 10605
    },
    {
      "epoch": 1.7308319738988582,
      "grad_norm": 0.5753697752952576,
      "learning_rate": 4.919125733802995e-05,
      "loss": 0.1411,
      "num_input_tokens_seen": 22933664,
      "step": 10610
    },
    {
      "epoch": 1.731647634584013,
      "grad_norm": 2.312779426574707,
      "learning_rate": 4.9189460524175915e-05,
      "loss": 0.16,
      "num_input_tokens_seen": 22943104,
      "step": 10615
    },
    {
      "epoch": 1.7324632952691679,
      "grad_norm": 4.1412129402160645,
      "learning_rate": 4.918766174940146e-05,
      "loss": 0.1877,
      "num_input_tokens_seen": 22953344,
      "step": 10620
    },
    {
      "epoch": 1.7332789559543231,
      "grad_norm": 0.7794988751411438,
      "learning_rate": 4.918586101385243e-05,
      "loss": 0.1442,
      "num_input_tokens_seen": 22964928,
      "step": 10625
    },
    {
      "epoch": 1.734094616639478,
      "grad_norm": 0.28300607204437256,
      "learning_rate": 4.918405831767478e-05,
      "loss": 0.1773,
      "num_input_tokens_seen": 22977152,
      "step": 10630
    },
    {
      "epoch": 1.734910277324633,
      "grad_norm": 3.008607864379883,
      "learning_rate": 4.9182253661014656e-05,
      "loss": 0.1388,
      "num_input_tokens_seen": 22988448,
      "step": 10635
    },
    {
      "epoch": 1.735725938009788,
      "grad_norm": 0.1705198734998703,
      "learning_rate": 4.9180447044018354e-05,
      "loss": 0.0846,
      "num_input_tokens_seen": 22998592,
      "step": 10640
    },
    {
      "epoch": 1.7365415986949428,
      "grad_norm": 1.8257312774658203,
      "learning_rate": 4.917863846683232e-05,
      "loss": 0.0785,
      "num_input_tokens_seen": 23009344,
      "step": 10645
    },
    {
      "epoch": 1.7373572593800979,
      "grad_norm": 4.587921142578125,
      "learning_rate": 4.9176827929603176e-05,
      "loss": 0.3389,
      "num_input_tokens_seen": 23019584,
      "step": 10650
    },
    {
      "epoch": 1.738172920065253,
      "grad_norm": 3.5429532527923584,
      "learning_rate": 4.9175015432477686e-05,
      "loss": 0.1734,
      "num_input_tokens_seen": 23030336,
      "step": 10655
    },
    {
      "epoch": 1.7389885807504077,
      "grad_norm": 0.692238450050354,
      "learning_rate": 4.9173200975602776e-05,
      "loss": 0.1656,
      "num_input_tokens_seen": 23041760,
      "step": 10660
    },
    {
      "epoch": 1.7398042414355628,
      "grad_norm": 3.643800735473633,
      "learning_rate": 4.917138455912555e-05,
      "loss": 0.1636,
      "num_input_tokens_seen": 23051840,
      "step": 10665
    },
    {
      "epoch": 1.7406199021207178,
      "grad_norm": 4.496054649353027,
      "learning_rate": 4.916956618319324e-05,
      "loss": 0.0523,
      "num_input_tokens_seen": 23061632,
      "step": 10670
    },
    {
      "epoch": 1.7414355628058726,
      "grad_norm": 4.332545757293701,
      "learning_rate": 4.916774584795327e-05,
      "loss": 0.1602,
      "num_input_tokens_seen": 23071520,
      "step": 10675
    },
    {
      "epoch": 1.7422512234910277,
      "grad_norm": 0.1590498834848404,
      "learning_rate": 4.916592355355318e-05,
      "loss": 0.2678,
      "num_input_tokens_seen": 23082304,
      "step": 10680
    },
    {
      "epoch": 1.7430668841761827,
      "grad_norm": 4.96671724319458,
      "learning_rate": 4.916409930014073e-05,
      "loss": 0.2867,
      "num_input_tokens_seen": 23092416,
      "step": 10685
    },
    {
      "epoch": 1.7438825448613375,
      "grad_norm": 3.715205192565918,
      "learning_rate": 4.916227308786377e-05,
      "loss": 0.2121,
      "num_input_tokens_seen": 23102720,
      "step": 10690
    },
    {
      "epoch": 1.7446982055464928,
      "grad_norm": 0.7393945455551147,
      "learning_rate": 4.916044491687036e-05,
      "loss": 0.0452,
      "num_input_tokens_seen": 23114016,
      "step": 10695
    },
    {
      "epoch": 1.7455138662316476,
      "grad_norm": 0.9213941693305969,
      "learning_rate": 4.915861478730869e-05,
      "loss": 0.0591,
      "num_input_tokens_seen": 23125888,
      "step": 10700
    },
    {
      "epoch": 1.7463295269168027,
      "grad_norm": 3.8318026065826416,
      "learning_rate": 4.915678269932713e-05,
      "loss": 0.0616,
      "num_input_tokens_seen": 23137248,
      "step": 10705
    },
    {
      "epoch": 1.7471451876019577,
      "grad_norm": 0.2188013345003128,
      "learning_rate": 4.91549486530742e-05,
      "loss": 0.1431,
      "num_input_tokens_seen": 23148064,
      "step": 10710
    },
    {
      "epoch": 1.7479608482871125,
      "grad_norm": 0.37179723381996155,
      "learning_rate": 4.9153112648698565e-05,
      "loss": 0.0091,
      "num_input_tokens_seen": 23158208,
      "step": 10715
    },
    {
      "epoch": 1.7487765089722676,
      "grad_norm": 2.3651340007781982,
      "learning_rate": 4.915127468634906e-05,
      "loss": 0.129,
      "num_input_tokens_seen": 23170144,
      "step": 10720
    },
    {
      "epoch": 1.7495921696574226,
      "grad_norm": 0.7593448758125305,
      "learning_rate": 4.9149434766174695e-05,
      "loss": 0.1706,
      "num_input_tokens_seen": 23180768,
      "step": 10725
    },
    {
      "epoch": 1.7504078303425774,
      "grad_norm": 4.086053371429443,
      "learning_rate": 4.914759288832462e-05,
      "loss": 0.2413,
      "num_input_tokens_seen": 23191840,
      "step": 10730
    },
    {
      "epoch": 1.7512234910277324,
      "grad_norm": 0.35563644766807556,
      "learning_rate": 4.914574905294813e-05,
      "loss": 0.1839,
      "num_input_tokens_seen": 23201760,
      "step": 10735
    },
    {
      "epoch": 1.7520391517128875,
      "grad_norm": 0.1594395488500595,
      "learning_rate": 4.9143903260194715e-05,
      "loss": 0.0933,
      "num_input_tokens_seen": 23213024,
      "step": 10740
    },
    {
      "epoch": 1.7528548123980423,
      "grad_norm": 0.2561168968677521,
      "learning_rate": 4.914205551021399e-05,
      "loss": 0.1451,
      "num_input_tokens_seen": 23224512,
      "step": 10745
    },
    {
      "epoch": 1.7536704730831973,
      "grad_norm": 0.6068389415740967,
      "learning_rate": 4.914020580315576e-05,
      "loss": 0.0916,
      "num_input_tokens_seen": 23235040,
      "step": 10750
    },
    {
      "epoch": 1.7544861337683524,
      "grad_norm": 0.6194460391998291,
      "learning_rate": 4.913835413916996e-05,
      "loss": 0.1701,
      "num_input_tokens_seen": 23245056,
      "step": 10755
    },
    {
      "epoch": 1.7553017944535072,
      "grad_norm": 0.8164099454879761,
      "learning_rate": 4.9136500518406694e-05,
      "loss": 0.1467,
      "num_input_tokens_seen": 23257824,
      "step": 10760
    },
    {
      "epoch": 1.7561174551386625,
      "grad_norm": 1.7459872961044312,
      "learning_rate": 4.913464494101622e-05,
      "loss": 0.1432,
      "num_input_tokens_seen": 23269376,
      "step": 10765
    },
    {
      "epoch": 1.7569331158238173,
      "grad_norm": 3.7063353061676025,
      "learning_rate": 4.913278740714898e-05,
      "loss": 0.1261,
      "num_input_tokens_seen": 23282560,
      "step": 10770
    },
    {
      "epoch": 1.7577487765089723,
      "grad_norm": 0.7812284231185913,
      "learning_rate": 4.913092791695554e-05,
      "loss": 0.074,
      "num_input_tokens_seen": 23292896,
      "step": 10775
    },
    {
      "epoch": 1.7585644371941274,
      "grad_norm": 1.1562386751174927,
      "learning_rate": 4.912906647058664e-05,
      "loss": 0.0276,
      "num_input_tokens_seen": 23303168,
      "step": 10780
    },
    {
      "epoch": 1.7593800978792822,
      "grad_norm": 0.2978338301181793,
      "learning_rate": 4.912720306819319e-05,
      "loss": 0.1808,
      "num_input_tokens_seen": 23314784,
      "step": 10785
    },
    {
      "epoch": 1.7601957585644372,
      "grad_norm": 0.9384825825691223,
      "learning_rate": 4.9125337709926235e-05,
      "loss": 0.1236,
      "num_input_tokens_seen": 23325696,
      "step": 10790
    },
    {
      "epoch": 1.7610114192495923,
      "grad_norm": 2.842236042022705,
      "learning_rate": 4.9123470395937e-05,
      "loss": 0.2587,
      "num_input_tokens_seen": 23337120,
      "step": 10795
    },
    {
      "epoch": 1.761827079934747,
      "grad_norm": 1.5775529146194458,
      "learning_rate": 4.9121601126376845e-05,
      "loss": 0.1221,
      "num_input_tokens_seen": 23347584,
      "step": 10800
    },
    {
      "epoch": 1.7626427406199021,
      "grad_norm": 2.2849128246307373,
      "learning_rate": 4.9119729901397313e-05,
      "loss": 0.3481,
      "num_input_tokens_seen": 23357504,
      "step": 10805
    },
    {
      "epoch": 1.7634584013050572,
      "grad_norm": 0.4386286735534668,
      "learning_rate": 4.9117856721150095e-05,
      "loss": 0.1882,
      "num_input_tokens_seen": 23367456,
      "step": 10810
    },
    {
      "epoch": 1.764274061990212,
      "grad_norm": 0.11932945996522903,
      "learning_rate": 4.911598158578704e-05,
      "loss": 0.0874,
      "num_input_tokens_seen": 23377984,
      "step": 10815
    },
    {
      "epoch": 1.765089722675367,
      "grad_norm": 1.375817060470581,
      "learning_rate": 4.9114104495460154e-05,
      "loss": 0.2791,
      "num_input_tokens_seen": 23390272,
      "step": 10820
    },
    {
      "epoch": 1.765905383360522,
      "grad_norm": 0.24430467188358307,
      "learning_rate": 4.9112225450321606e-05,
      "loss": 0.0177,
      "num_input_tokens_seen": 23400416,
      "step": 10825
    },
    {
      "epoch": 1.7667210440456769,
      "grad_norm": 3.3636837005615234,
      "learning_rate": 4.911034445052371e-05,
      "loss": 0.1068,
      "num_input_tokens_seen": 23412704,
      "step": 10830
    },
    {
      "epoch": 1.7675367047308321,
      "grad_norm": 0.5361234545707703,
      "learning_rate": 4.910846149621896e-05,
      "loss": 0.148,
      "num_input_tokens_seen": 23421952,
      "step": 10835
    },
    {
      "epoch": 1.768352365415987,
      "grad_norm": 0.1068536564707756,
      "learning_rate": 4.9106576587560006e-05,
      "loss": 0.1008,
      "num_input_tokens_seen": 23431136,
      "step": 10840
    },
    {
      "epoch": 1.7691680261011418,
      "grad_norm": 0.2754809558391571,
      "learning_rate": 4.9104689724699625e-05,
      "loss": 0.1675,
      "num_input_tokens_seen": 23442880,
      "step": 10845
    },
    {
      "epoch": 1.769983686786297,
      "grad_norm": 4.434019565582275,
      "learning_rate": 4.91028009077908e-05,
      "loss": 0.17,
      "num_input_tokens_seen": 23453120,
      "step": 10850
    },
    {
      "epoch": 1.7707993474714518,
      "grad_norm": 0.20431602001190186,
      "learning_rate": 4.910091013698663e-05,
      "loss": 0.2072,
      "num_input_tokens_seen": 23464192,
      "step": 10855
    },
    {
      "epoch": 1.7716150081566069,
      "grad_norm": 0.8551867008209229,
      "learning_rate": 4.90990174124404e-05,
      "loss": 0.2168,
      "num_input_tokens_seen": 23475264,
      "step": 10860
    },
    {
      "epoch": 1.772430668841762,
      "grad_norm": 1.3833519220352173,
      "learning_rate": 4.909712273430554e-05,
      "loss": 0.073,
      "num_input_tokens_seen": 23485440,
      "step": 10865
    },
    {
      "epoch": 1.7732463295269167,
      "grad_norm": 0.8069531917572021,
      "learning_rate": 4.9095226102735645e-05,
      "loss": 0.1187,
      "num_input_tokens_seen": 23496768,
      "step": 10870
    },
    {
      "epoch": 1.7740619902120718,
      "grad_norm": 0.052002016454935074,
      "learning_rate": 4.909332751788447e-05,
      "loss": 0.1645,
      "num_input_tokens_seen": 23507328,
      "step": 10875
    },
    {
      "epoch": 1.7748776508972268,
      "grad_norm": 1.4189658164978027,
      "learning_rate": 4.909142697990591e-05,
      "loss": 0.0986,
      "num_input_tokens_seen": 23518976,
      "step": 10880
    },
    {
      "epoch": 1.7756933115823816,
      "grad_norm": 0.11173592507839203,
      "learning_rate": 4.908952448895404e-05,
      "loss": 0.1,
      "num_input_tokens_seen": 23530720,
      "step": 10885
    },
    {
      "epoch": 1.7765089722675367,
      "grad_norm": 1.56269109249115,
      "learning_rate": 4.908762004518309e-05,
      "loss": 0.0558,
      "num_input_tokens_seen": 23541312,
      "step": 10890
    },
    {
      "epoch": 1.7773246329526917,
      "grad_norm": 0.5744493007659912,
      "learning_rate": 4.908571364874743e-05,
      "loss": 0.103,
      "num_input_tokens_seen": 23552064,
      "step": 10895
    },
    {
      "epoch": 1.7781402936378465,
      "grad_norm": 0.17743384838104248,
      "learning_rate": 4.9083805299801626e-05,
      "loss": 0.1079,
      "num_input_tokens_seen": 23563104,
      "step": 10900
    },
    {
      "epoch": 1.7789559543230016,
      "grad_norm": 3.0502898693084717,
      "learning_rate": 4.908189499850036e-05,
      "loss": 0.136,
      "num_input_tokens_seen": 23572992,
      "step": 10905
    },
    {
      "epoch": 1.7797716150081566,
      "grad_norm": 3.5304319858551025,
      "learning_rate": 4.907998274499849e-05,
      "loss": 0.2232,
      "num_input_tokens_seen": 23583328,
      "step": 10910
    },
    {
      "epoch": 1.7805872756933114,
      "grad_norm": 0.8347854614257812,
      "learning_rate": 4.9078068539451045e-05,
      "loss": 0.0981,
      "num_input_tokens_seen": 23593888,
      "step": 10915
    },
    {
      "epoch": 1.7814029363784667,
      "grad_norm": 4.794600486755371,
      "learning_rate": 4.907615238201319e-05,
      "loss": 0.1206,
      "num_input_tokens_seen": 23604544,
      "step": 10920
    },
    {
      "epoch": 1.7822185970636215,
      "grad_norm": 0.2617413103580475,
      "learning_rate": 4.907423427284026e-05,
      "loss": 0.2593,
      "num_input_tokens_seen": 23615584,
      "step": 10925
    },
    {
      "epoch": 1.7830342577487766,
      "grad_norm": 0.05737384408712387,
      "learning_rate": 4.907231421208775e-05,
      "loss": 0.0256,
      "num_input_tokens_seen": 23625568,
      "step": 10930
    },
    {
      "epoch": 1.7838499184339316,
      "grad_norm": 0.07537528872489929,
      "learning_rate": 4.907039219991131e-05,
      "loss": 0.0192,
      "num_input_tokens_seen": 23636320,
      "step": 10935
    },
    {
      "epoch": 1.7846655791190864,
      "grad_norm": 2.5834670066833496,
      "learning_rate": 4.906846823646675e-05,
      "loss": 0.2319,
      "num_input_tokens_seen": 23646912,
      "step": 10940
    },
    {
      "epoch": 1.7854812398042414,
      "grad_norm": 0.09616564214229584,
      "learning_rate": 4.906654232191002e-05,
      "loss": 0.1385,
      "num_input_tokens_seen": 23657120,
      "step": 10945
    },
    {
      "epoch": 1.7862969004893965,
      "grad_norm": 3.8253488540649414,
      "learning_rate": 4.906461445639726e-05,
      "loss": 0.208,
      "num_input_tokens_seen": 23667712,
      "step": 10950
    },
    {
      "epoch": 1.7871125611745513,
      "grad_norm": 0.83757483959198,
      "learning_rate": 4.906268464008476e-05,
      "loss": 0.106,
      "num_input_tokens_seen": 23678720,
      "step": 10955
    },
    {
      "epoch": 1.7879282218597063,
      "grad_norm": 2.588719129562378,
      "learning_rate": 4.9060752873128946e-05,
      "loss": 0.0493,
      "num_input_tokens_seen": 23688160,
      "step": 10960
    },
    {
      "epoch": 1.7887438825448614,
      "grad_norm": 4.905638217926025,
      "learning_rate": 4.905881915568642e-05,
      "loss": 0.2047,
      "num_input_tokens_seen": 23699680,
      "step": 10965
    },
    {
      "epoch": 1.7895595432300162,
      "grad_norm": 2.158893585205078,
      "learning_rate": 4.905688348791394e-05,
      "loss": 0.0967,
      "num_input_tokens_seen": 23710880,
      "step": 10970
    },
    {
      "epoch": 1.7903752039151712,
      "grad_norm": 0.4053361713886261,
      "learning_rate": 4.905494586996842e-05,
      "loss": 0.0522,
      "num_input_tokens_seen": 23722304,
      "step": 10975
    },
    {
      "epoch": 1.7911908646003263,
      "grad_norm": 4.744560241699219,
      "learning_rate": 4.905300630200693e-05,
      "loss": 0.2627,
      "num_input_tokens_seen": 23734048,
      "step": 10980
    },
    {
      "epoch": 1.792006525285481,
      "grad_norm": 0.09347156435251236,
      "learning_rate": 4.9051064784186704e-05,
      "loss": 0.0866,
      "num_input_tokens_seen": 23745600,
      "step": 10985
    },
    {
      "epoch": 1.7928221859706364,
      "grad_norm": 0.11766652762889862,
      "learning_rate": 4.9049121316665146e-05,
      "loss": 0.0331,
      "num_input_tokens_seen": 23756800,
      "step": 10990
    },
    {
      "epoch": 1.7936378466557912,
      "grad_norm": 2.3665807247161865,
      "learning_rate": 4.904717589959978e-05,
      "loss": 0.282,
      "num_input_tokens_seen": 23766880,
      "step": 10995
    },
    {
      "epoch": 1.7944535073409462,
      "grad_norm": 0.056640349328517914,
      "learning_rate": 4.904522853314833e-05,
      "loss": 0.1131,
      "num_input_tokens_seen": 23777472,
      "step": 11000
    },
    {
      "epoch": 1.7952691680261013,
      "grad_norm": 2.2606041431427,
      "learning_rate": 4.904327921746864e-05,
      "loss": 0.1389,
      "num_input_tokens_seen": 23786336,
      "step": 11005
    },
    {
      "epoch": 1.796084828711256,
      "grad_norm": 3.8277926445007324,
      "learning_rate": 4.904132795271875e-05,
      "loss": 0.2275,
      "num_input_tokens_seen": 23797184,
      "step": 11010
    },
    {
      "epoch": 1.7969004893964111,
      "grad_norm": 0.5380455255508423,
      "learning_rate": 4.9039374739056825e-05,
      "loss": 0.0943,
      "num_input_tokens_seen": 23808288,
      "step": 11015
    },
    {
      "epoch": 1.7977161500815662,
      "grad_norm": 1.1800470352172852,
      "learning_rate": 4.903741957664121e-05,
      "loss": 0.2547,
      "num_input_tokens_seen": 23818176,
      "step": 11020
    },
    {
      "epoch": 1.798531810766721,
      "grad_norm": 0.8688233494758606,
      "learning_rate": 4.903546246563041e-05,
      "loss": 0.1267,
      "num_input_tokens_seen": 23829344,
      "step": 11025
    },
    {
      "epoch": 1.799347471451876,
      "grad_norm": 2.287400484085083,
      "learning_rate": 4.9033503406183055e-05,
      "loss": 0.1267,
      "num_input_tokens_seen": 23839936,
      "step": 11030
    },
    {
      "epoch": 1.800163132137031,
      "grad_norm": 3.783482074737549,
      "learning_rate": 4.9031542398457974e-05,
      "loss": 0.1061,
      "num_input_tokens_seen": 23852736,
      "step": 11035
    },
    {
      "epoch": 1.8009787928221859,
      "grad_norm": 0.21201197803020477,
      "learning_rate": 4.902957944261413e-05,
      "loss": 0.1609,
      "num_input_tokens_seen": 23864736,
      "step": 11040
    },
    {
      "epoch": 1.801794453507341,
      "grad_norm": 2.7451019287109375,
      "learning_rate": 4.902761453881065e-05,
      "loss": 0.1805,
      "num_input_tokens_seen": 23874432,
      "step": 11045
    },
    {
      "epoch": 1.802610114192496,
      "grad_norm": 1.6343696117401123,
      "learning_rate": 4.9025647687206824e-05,
      "loss": 0.0406,
      "num_input_tokens_seen": 23885056,
      "step": 11050
    },
    {
      "epoch": 1.8034257748776508,
      "grad_norm": 0.5249418020248413,
      "learning_rate": 4.90236788879621e-05,
      "loss": 0.1174,
      "num_input_tokens_seen": 23896096,
      "step": 11055
    },
    {
      "epoch": 1.804241435562806,
      "grad_norm": 0.22775468230247498,
      "learning_rate": 4.9021708141236056e-05,
      "loss": 0.0624,
      "num_input_tokens_seen": 23907200,
      "step": 11060
    },
    {
      "epoch": 1.8050570962479608,
      "grad_norm": 0.21686673164367676,
      "learning_rate": 4.901973544718847e-05,
      "loss": 0.1234,
      "num_input_tokens_seen": 23916576,
      "step": 11065
    },
    {
      "epoch": 1.8058727569331157,
      "grad_norm": 0.07676563411951065,
      "learning_rate": 4.901776080597926e-05,
      "loss": 0.0371,
      "num_input_tokens_seen": 23926784,
      "step": 11070
    },
    {
      "epoch": 1.806688417618271,
      "grad_norm": 1.3494975566864014,
      "learning_rate": 4.9015784217768487e-05,
      "loss": 0.2541,
      "num_input_tokens_seen": 23937632,
      "step": 11075
    },
    {
      "epoch": 1.8075040783034257,
      "grad_norm": 0.22478091716766357,
      "learning_rate": 4.901380568271639e-05,
      "loss": 0.0092,
      "num_input_tokens_seen": 23949024,
      "step": 11080
    },
    {
      "epoch": 1.8083197389885808,
      "grad_norm": 2.6358680725097656,
      "learning_rate": 4.901182520098336e-05,
      "loss": 0.144,
      "num_input_tokens_seen": 23960192,
      "step": 11085
    },
    {
      "epoch": 1.8091353996737358,
      "grad_norm": 0.964083194732666,
      "learning_rate": 4.9009842772729944e-05,
      "loss": 0.2404,
      "num_input_tokens_seen": 23970304,
      "step": 11090
    },
    {
      "epoch": 1.8099510603588906,
      "grad_norm": 0.11122696846723557,
      "learning_rate": 4.9007858398116856e-05,
      "loss": 0.1999,
      "num_input_tokens_seen": 23980544,
      "step": 11095
    },
    {
      "epoch": 1.8107667210440457,
      "grad_norm": 5.403952598571777,
      "learning_rate": 4.9005872077304944e-05,
      "loss": 0.0819,
      "num_input_tokens_seen": 23991680,
      "step": 11100
    },
    {
      "epoch": 1.8115823817292007,
      "grad_norm": 0.19622884690761566,
      "learning_rate": 4.900388381045524e-05,
      "loss": 0.282,
      "num_input_tokens_seen": 24000672,
      "step": 11105
    },
    {
      "epoch": 1.8123980424143555,
      "grad_norm": 0.861206591129303,
      "learning_rate": 4.9001893597728915e-05,
      "loss": 0.1903,
      "num_input_tokens_seen": 24010496,
      "step": 11110
    },
    {
      "epoch": 1.8132137030995106,
      "grad_norm": 0.15993455052375793,
      "learning_rate": 4.899990143928731e-05,
      "loss": 0.13,
      "num_input_tokens_seen": 24021056,
      "step": 11115
    },
    {
      "epoch": 1.8140293637846656,
      "grad_norm": 0.10411029309034348,
      "learning_rate": 4.899790733529193e-05,
      "loss": 0.0775,
      "num_input_tokens_seen": 24031136,
      "step": 11120
    },
    {
      "epoch": 1.8148450244698204,
      "grad_norm": 0.17643879354000092,
      "learning_rate": 4.8995911285904404e-05,
      "loss": 0.0145,
      "num_input_tokens_seen": 24040864,
      "step": 11125
    },
    {
      "epoch": 1.8156606851549757,
      "grad_norm": 0.20683497190475464,
      "learning_rate": 4.899391329128656e-05,
      "loss": 0.2032,
      "num_input_tokens_seen": 24051168,
      "step": 11130
    },
    {
      "epoch": 1.8164763458401305,
      "grad_norm": 2.6556005477905273,
      "learning_rate": 4.899191335160037e-05,
      "loss": 0.159,
      "num_input_tokens_seen": 24060448,
      "step": 11135
    },
    {
      "epoch": 1.8172920065252853,
      "grad_norm": 0.18561014533042908,
      "learning_rate": 4.898991146700794e-05,
      "loss": 0.4073,
      "num_input_tokens_seen": 24070208,
      "step": 11140
    },
    {
      "epoch": 1.8181076672104406,
      "grad_norm": 0.1081450805068016,
      "learning_rate": 4.898790763767157e-05,
      "loss": 0.0513,
      "num_input_tokens_seen": 24080000,
      "step": 11145
    },
    {
      "epoch": 1.8189233278955954,
      "grad_norm": 0.46138057112693787,
      "learning_rate": 4.8985901863753694e-05,
      "loss": 0.0705,
      "num_input_tokens_seen": 24089120,
      "step": 11150
    },
    {
      "epoch": 1.8197389885807504,
      "grad_norm": 0.14180487394332886,
      "learning_rate": 4.8983894145416896e-05,
      "loss": 0.0435,
      "num_input_tokens_seen": 24099968,
      "step": 11155
    },
    {
      "epoch": 1.8205546492659055,
      "grad_norm": 1.645262360572815,
      "learning_rate": 4.898188448282396e-05,
      "loss": 0.0534,
      "num_input_tokens_seen": 24110560,
      "step": 11160
    },
    {
      "epoch": 1.8213703099510603,
      "grad_norm": 0.09216941893100739,
      "learning_rate": 4.897987287613778e-05,
      "loss": 0.1165,
      "num_input_tokens_seen": 24121984,
      "step": 11165
    },
    {
      "epoch": 1.8221859706362153,
      "grad_norm": 3.4673421382904053,
      "learning_rate": 4.897785932552143e-05,
      "loss": 0.2744,
      "num_input_tokens_seen": 24132544,
      "step": 11170
    },
    {
      "epoch": 1.8230016313213704,
      "grad_norm": 3.1680285930633545,
      "learning_rate": 4.897584383113814e-05,
      "loss": 0.1616,
      "num_input_tokens_seen": 24143808,
      "step": 11175
    },
    {
      "epoch": 1.8238172920065252,
      "grad_norm": 2.7808051109313965,
      "learning_rate": 4.89738263931513e-05,
      "loss": 0.3092,
      "num_input_tokens_seen": 24154496,
      "step": 11180
    },
    {
      "epoch": 1.8246329526916802,
      "grad_norm": 2.7151877880096436,
      "learning_rate": 4.8971807011724444e-05,
      "loss": 0.1983,
      "num_input_tokens_seen": 24165472,
      "step": 11185
    },
    {
      "epoch": 1.8254486133768353,
      "grad_norm": 0.22334140539169312,
      "learning_rate": 4.8969785687021294e-05,
      "loss": 0.0279,
      "num_input_tokens_seen": 24177088,
      "step": 11190
    },
    {
      "epoch": 1.82626427406199,
      "grad_norm": 0.19176222383975983,
      "learning_rate": 4.8967762419205684e-05,
      "loss": 0.1226,
      "num_input_tokens_seen": 24188768,
      "step": 11195
    },
    {
      "epoch": 1.8270799347471451,
      "grad_norm": 0.1524587869644165,
      "learning_rate": 4.896573720844164e-05,
      "loss": 0.2339,
      "num_input_tokens_seen": 24199712,
      "step": 11200
    },
    {
      "epoch": 1.8278955954323002,
      "grad_norm": 1.981576919555664,
      "learning_rate": 4.896371005489334e-05,
      "loss": 0.164,
      "num_input_tokens_seen": 24210912,
      "step": 11205
    },
    {
      "epoch": 1.828711256117455,
      "grad_norm": 5.648102760314941,
      "learning_rate": 4.896168095872511e-05,
      "loss": 0.2033,
      "num_input_tokens_seen": 24221824,
      "step": 11210
    },
    {
      "epoch": 1.8295269168026103,
      "grad_norm": 0.17025180160999298,
      "learning_rate": 4.895964992010145e-05,
      "loss": 0.0773,
      "num_input_tokens_seen": 24231168,
      "step": 11215
    },
    {
      "epoch": 1.830342577487765,
      "grad_norm": 0.08324454724788666,
      "learning_rate": 4.895761693918699e-05,
      "loss": 0.0206,
      "num_input_tokens_seen": 24241408,
      "step": 11220
    },
    {
      "epoch": 1.8311582381729201,
      "grad_norm": 1.4679114818572998,
      "learning_rate": 4.895558201614654e-05,
      "loss": 0.1481,
      "num_input_tokens_seen": 24251520,
      "step": 11225
    },
    {
      "epoch": 1.8319738988580752,
      "grad_norm": 0.139203280210495,
      "learning_rate": 4.895354515114506e-05,
      "loss": 0.2452,
      "num_input_tokens_seen": 24262144,
      "step": 11230
    },
    {
      "epoch": 1.83278955954323,
      "grad_norm": 1.3829185962677002,
      "learning_rate": 4.895150634434769e-05,
      "loss": 0.1032,
      "num_input_tokens_seen": 24273440,
      "step": 11235
    },
    {
      "epoch": 1.833605220228385,
      "grad_norm": 2.407167673110962,
      "learning_rate": 4.894946559591966e-05,
      "loss": 0.0307,
      "num_input_tokens_seen": 24284832,
      "step": 11240
    },
    {
      "epoch": 1.83442088091354,
      "grad_norm": 0.31610408425331116,
      "learning_rate": 4.8947422906026446e-05,
      "loss": 0.1837,
      "num_input_tokens_seen": 24295488,
      "step": 11245
    },
    {
      "epoch": 1.8352365415986949,
      "grad_norm": 1.0342751741409302,
      "learning_rate": 4.894537827483362e-05,
      "loss": 0.0547,
      "num_input_tokens_seen": 24306624,
      "step": 11250
    },
    {
      "epoch": 1.83605220228385,
      "grad_norm": 0.16724175214767456,
      "learning_rate": 4.8943331702506935e-05,
      "loss": 0.0728,
      "num_input_tokens_seen": 24318176,
      "step": 11255
    },
    {
      "epoch": 1.836867862969005,
      "grad_norm": 1.1085926294326782,
      "learning_rate": 4.894128318921229e-05,
      "loss": 0.0768,
      "num_input_tokens_seen": 24329024,
      "step": 11260
    },
    {
      "epoch": 1.8376835236541598,
      "grad_norm": 0.05294278636574745,
      "learning_rate": 4.893923273511576e-05,
      "loss": 0.0307,
      "num_input_tokens_seen": 24339680,
      "step": 11265
    },
    {
      "epoch": 1.8384991843393148,
      "grad_norm": 0.09843534976243973,
      "learning_rate": 4.893718034038355e-05,
      "loss": 0.2451,
      "num_input_tokens_seen": 24350368,
      "step": 11270
    },
    {
      "epoch": 1.8393148450244698,
      "grad_norm": 0.2138974964618683,
      "learning_rate": 4.8935126005182056e-05,
      "loss": 0.1078,
      "num_input_tokens_seen": 24361536,
      "step": 11275
    },
    {
      "epoch": 1.8401305057096247,
      "grad_norm": 0.11413642764091492,
      "learning_rate": 4.8933069729677795e-05,
      "loss": 0.2499,
      "num_input_tokens_seen": 24372768,
      "step": 11280
    },
    {
      "epoch": 1.84094616639478,
      "grad_norm": 5.159852504730225,
      "learning_rate": 4.893101151403747e-05,
      "loss": 0.189,
      "num_input_tokens_seen": 24382592,
      "step": 11285
    },
    {
      "epoch": 1.8417618270799347,
      "grad_norm": 2.8542661666870117,
      "learning_rate": 4.892895135842792e-05,
      "loss": 0.084,
      "num_input_tokens_seen": 24393248,
      "step": 11290
    },
    {
      "epoch": 1.8425774877650896,
      "grad_norm": 0.7388946413993835,
      "learning_rate": 4.892688926301616e-05,
      "loss": 0.1437,
      "num_input_tokens_seen": 24403232,
      "step": 11295
    },
    {
      "epoch": 1.8433931484502448,
      "grad_norm": 1.2321661710739136,
      "learning_rate": 4.892482522796936e-05,
      "loss": 0.0948,
      "num_input_tokens_seen": 24413792,
      "step": 11300
    },
    {
      "epoch": 1.8442088091353996,
      "grad_norm": 0.13754427433013916,
      "learning_rate": 4.892275925345483e-05,
      "loss": 0.0725,
      "num_input_tokens_seen": 24423328,
      "step": 11305
    },
    {
      "epoch": 1.8450244698205547,
      "grad_norm": 1.5106384754180908,
      "learning_rate": 4.8920691339640055e-05,
      "loss": 0.3124,
      "num_input_tokens_seen": 24434464,
      "step": 11310
    },
    {
      "epoch": 1.8458401305057097,
      "grad_norm": 0.12568306922912598,
      "learning_rate": 4.8918621486692663e-05,
      "loss": 0.1007,
      "num_input_tokens_seen": 24444224,
      "step": 11315
    },
    {
      "epoch": 1.8466557911908645,
      "grad_norm": 0.09501215815544128,
      "learning_rate": 4.8916549694780455e-05,
      "loss": 0.1191,
      "num_input_tokens_seen": 24454912,
      "step": 11320
    },
    {
      "epoch": 1.8474714518760196,
      "grad_norm": 0.7249631881713867,
      "learning_rate": 4.891447596407137e-05,
      "loss": 0.0572,
      "num_input_tokens_seen": 24466304,
      "step": 11325
    },
    {
      "epoch": 1.8482871125611746,
      "grad_norm": 2.0175819396972656,
      "learning_rate": 4.8912400294733526e-05,
      "loss": 0.1571,
      "num_input_tokens_seen": 24476960,
      "step": 11330
    },
    {
      "epoch": 1.8491027732463294,
      "grad_norm": 1.7725948095321655,
      "learning_rate": 4.891032268693519e-05,
      "loss": 0.2005,
      "num_input_tokens_seen": 24486784,
      "step": 11335
    },
    {
      "epoch": 1.8499184339314845,
      "grad_norm": 0.12097588926553726,
      "learning_rate": 4.8908243140844765e-05,
      "loss": 0.0358,
      "num_input_tokens_seen": 24497120,
      "step": 11340
    },
    {
      "epoch": 1.8507340946166395,
      "grad_norm": 1.4836633205413818,
      "learning_rate": 4.890616165663085e-05,
      "loss": 0.1303,
      "num_input_tokens_seen": 24507200,
      "step": 11345
    },
    {
      "epoch": 1.8515497553017943,
      "grad_norm": 3.1321067810058594,
      "learning_rate": 4.890407823446218e-05,
      "loss": 0.2664,
      "num_input_tokens_seen": 24517920,
      "step": 11350
    },
    {
      "epoch": 1.8523654159869496,
      "grad_norm": 0.13268068432807922,
      "learning_rate": 4.890199287450763e-05,
      "loss": 0.0902,
      "num_input_tokens_seen": 24528640,
      "step": 11355
    },
    {
      "epoch": 1.8531810766721044,
      "grad_norm": 4.3858537673950195,
      "learning_rate": 4.889990557693626e-05,
      "loss": 0.2153,
      "num_input_tokens_seen": 24540352,
      "step": 11360
    },
    {
      "epoch": 1.8539967373572592,
      "grad_norm": 2.9111762046813965,
      "learning_rate": 4.889781634191728e-05,
      "loss": 0.0478,
      "num_input_tokens_seen": 24551584,
      "step": 11365
    },
    {
      "epoch": 1.8548123980424145,
      "grad_norm": 0.45477044582366943,
      "learning_rate": 4.889572516962006e-05,
      "loss": 0.0905,
      "num_input_tokens_seen": 24562400,
      "step": 11370
    },
    {
      "epoch": 1.8556280587275693,
      "grad_norm": 0.1414952278137207,
      "learning_rate": 4.889363206021409e-05,
      "loss": 0.1127,
      "num_input_tokens_seen": 24572736,
      "step": 11375
    },
    {
      "epoch": 1.8564437194127243,
      "grad_norm": 2.483764171600342,
      "learning_rate": 4.889153701386908e-05,
      "loss": 0.201,
      "num_input_tokens_seen": 24584416,
      "step": 11380
    },
    {
      "epoch": 1.8572593800978794,
      "grad_norm": 3.6788392066955566,
      "learning_rate": 4.888944003075486e-05,
      "loss": 0.2004,
      "num_input_tokens_seen": 24593472,
      "step": 11385
    },
    {
      "epoch": 1.8580750407830342,
      "grad_norm": 0.10282847285270691,
      "learning_rate": 4.888734111104142e-05,
      "loss": 0.4256,
      "num_input_tokens_seen": 24604544,
      "step": 11390
    },
    {
      "epoch": 1.8588907014681892,
      "grad_norm": 0.22144851088523865,
      "learning_rate": 4.88852402548989e-05,
      "loss": 0.1469,
      "num_input_tokens_seen": 24614784,
      "step": 11395
    },
    {
      "epoch": 1.8597063621533443,
      "grad_norm": 2.5628433227539062,
      "learning_rate": 4.8883137462497615e-05,
      "loss": 0.0804,
      "num_input_tokens_seen": 24625504,
      "step": 11400
    },
    {
      "epoch": 1.860522022838499,
      "grad_norm": 2.815721035003662,
      "learning_rate": 4.8881032734008024e-05,
      "loss": 0.0992,
      "num_input_tokens_seen": 24636192,
      "step": 11405
    },
    {
      "epoch": 1.8613376835236541,
      "grad_norm": 2.4981741905212402,
      "learning_rate": 4.887892606960075e-05,
      "loss": 0.1739,
      "num_input_tokens_seen": 24647200,
      "step": 11410
    },
    {
      "epoch": 1.8621533442088092,
      "grad_norm": 0.11964482814073563,
      "learning_rate": 4.887681746944657e-05,
      "loss": 0.0335,
      "num_input_tokens_seen": 24656640,
      "step": 11415
    },
    {
      "epoch": 1.862969004893964,
      "grad_norm": 1.2380677461624146,
      "learning_rate": 4.8874706933716406e-05,
      "loss": 0.0191,
      "num_input_tokens_seen": 24665664,
      "step": 11420
    },
    {
      "epoch": 1.863784665579119,
      "grad_norm": 0.6048164963722229,
      "learning_rate": 4.887259446258137e-05,
      "loss": 0.1681,
      "num_input_tokens_seen": 24676896,
      "step": 11425
    },
    {
      "epoch": 1.864600326264274,
      "grad_norm": 0.26376572251319885,
      "learning_rate": 4.887048005621269e-05,
      "loss": 0.0464,
      "num_input_tokens_seen": 24688960,
      "step": 11430
    },
    {
      "epoch": 1.865415986949429,
      "grad_norm": 2.958484172821045,
      "learning_rate": 4.886836371478178e-05,
      "loss": 0.0831,
      "num_input_tokens_seen": 24699200,
      "step": 11435
    },
    {
      "epoch": 1.8662316476345842,
      "grad_norm": 3.382356882095337,
      "learning_rate": 4.8866245438460215e-05,
      "loss": 0.0957,
      "num_input_tokens_seen": 24709600,
      "step": 11440
    },
    {
      "epoch": 1.867047308319739,
      "grad_norm": 6.393452167510986,
      "learning_rate": 4.886412522741968e-05,
      "loss": 0.2363,
      "num_input_tokens_seen": 24719968,
      "step": 11445
    },
    {
      "epoch": 1.867862969004894,
      "grad_norm": 0.09577061980962753,
      "learning_rate": 4.886200308183207e-05,
      "loss": 0.0147,
      "num_input_tokens_seen": 24731520,
      "step": 11450
    },
    {
      "epoch": 1.868678629690049,
      "grad_norm": 1.72732412815094,
      "learning_rate": 4.885987900186943e-05,
      "loss": 0.0535,
      "num_input_tokens_seen": 24742880,
      "step": 11455
    },
    {
      "epoch": 1.8694942903752039,
      "grad_norm": 0.040071967989206314,
      "learning_rate": 4.8857752987703924e-05,
      "loss": 0.153,
      "num_input_tokens_seen": 24753600,
      "step": 11460
    },
    {
      "epoch": 1.870309951060359,
      "grad_norm": 0.05324089527130127,
      "learning_rate": 4.8855625039507916e-05,
      "loss": 0.2349,
      "num_input_tokens_seen": 24764608,
      "step": 11465
    },
    {
      "epoch": 1.871125611745514,
      "grad_norm": 5.415706634521484,
      "learning_rate": 4.8853495157453886e-05,
      "loss": 0.2286,
      "num_input_tokens_seen": 24775872,
      "step": 11470
    },
    {
      "epoch": 1.8719412724306688,
      "grad_norm": 3.6094143390655518,
      "learning_rate": 4.885136334171452e-05,
      "loss": 0.1601,
      "num_input_tokens_seen": 24786208,
      "step": 11475
    },
    {
      "epoch": 1.8727569331158238,
      "grad_norm": 3.190077304840088,
      "learning_rate": 4.8849229592462615e-05,
      "loss": 0.3495,
      "num_input_tokens_seen": 24795296,
      "step": 11480
    },
    {
      "epoch": 1.8735725938009788,
      "grad_norm": 0.18625961244106293,
      "learning_rate": 4.884709390987115e-05,
      "loss": 0.0388,
      "num_input_tokens_seen": 24805632,
      "step": 11485
    },
    {
      "epoch": 1.8743882544861337,
      "grad_norm": 3.2550137042999268,
      "learning_rate": 4.8844956294113255e-05,
      "loss": 0.1069,
      "num_input_tokens_seen": 24815552,
      "step": 11490
    },
    {
      "epoch": 1.8752039151712887,
      "grad_norm": 0.11549729853868484,
      "learning_rate": 4.884281674536221e-05,
      "loss": 0.2082,
      "num_input_tokens_seen": 24827264,
      "step": 11495
    },
    {
      "epoch": 1.8760195758564437,
      "grad_norm": 0.13546086847782135,
      "learning_rate": 4.884067526379147e-05,
      "loss": 0.1353,
      "num_input_tokens_seen": 24839168,
      "step": 11500
    },
    {
      "epoch": 1.8768352365415986,
      "grad_norm": 3.3355772495269775,
      "learning_rate": 4.8838531849574624e-05,
      "loss": 0.1486,
      "num_input_tokens_seen": 24850336,
      "step": 11505
    },
    {
      "epoch": 1.8776508972267538,
      "grad_norm": 0.2108527272939682,
      "learning_rate": 4.8836386502885426e-05,
      "loss": 0.0319,
      "num_input_tokens_seen": 24861984,
      "step": 11510
    },
    {
      "epoch": 1.8784665579119086,
      "grad_norm": 1.1564456224441528,
      "learning_rate": 4.88342392238978e-05,
      "loss": 0.0966,
      "num_input_tokens_seen": 24872160,
      "step": 11515
    },
    {
      "epoch": 1.8792822185970635,
      "grad_norm": 2.4284708499908447,
      "learning_rate": 4.88320900127858e-05,
      "loss": 0.2325,
      "num_input_tokens_seen": 24882272,
      "step": 11520
    },
    {
      "epoch": 1.8800978792822187,
      "grad_norm": 3.3414292335510254,
      "learning_rate": 4.882993886972367e-05,
      "loss": 0.2882,
      "num_input_tokens_seen": 24892448,
      "step": 11525
    },
    {
      "epoch": 1.8809135399673735,
      "grad_norm": 0.647790789604187,
      "learning_rate": 4.882778579488578e-05,
      "loss": 0.0531,
      "num_input_tokens_seen": 24902016,
      "step": 11530
    },
    {
      "epoch": 1.8817292006525286,
      "grad_norm": 4.748507499694824,
      "learning_rate": 4.882563078844668e-05,
      "loss": 0.2505,
      "num_input_tokens_seen": 24913344,
      "step": 11535
    },
    {
      "epoch": 1.8825448613376836,
      "grad_norm": 0.508827269077301,
      "learning_rate": 4.882347385058105e-05,
      "loss": 0.2493,
      "num_input_tokens_seen": 24923840,
      "step": 11540
    },
    {
      "epoch": 1.8833605220228384,
      "grad_norm": 0.22358126938343048,
      "learning_rate": 4.882131498146375e-05,
      "loss": 0.1672,
      "num_input_tokens_seen": 24933568,
      "step": 11545
    },
    {
      "epoch": 1.8841761827079935,
      "grad_norm": 0.16932672262191772,
      "learning_rate": 4.881915418126979e-05,
      "loss": 0.0433,
      "num_input_tokens_seen": 24943520,
      "step": 11550
    },
    {
      "epoch": 1.8849918433931485,
      "grad_norm": 4.0258870124816895,
      "learning_rate": 4.8816991450174334e-05,
      "loss": 0.1707,
      "num_input_tokens_seen": 24954976,
      "step": 11555
    },
    {
      "epoch": 1.8858075040783033,
      "grad_norm": 2.1621270179748535,
      "learning_rate": 4.881482678835271e-05,
      "loss": 0.0654,
      "num_input_tokens_seen": 24964384,
      "step": 11560
    },
    {
      "epoch": 1.8866231647634584,
      "grad_norm": 0.20863114297389984,
      "learning_rate": 4.881266019598039e-05,
      "loss": 0.0321,
      "num_input_tokens_seen": 24975584,
      "step": 11565
    },
    {
      "epoch": 1.8874388254486134,
      "grad_norm": 5.4353790283203125,
      "learning_rate": 4.8810491673233006e-05,
      "loss": 0.1001,
      "num_input_tokens_seen": 24987520,
      "step": 11570
    },
    {
      "epoch": 1.8882544861337682,
      "grad_norm": 0.11817409843206406,
      "learning_rate": 4.880832122028635e-05,
      "loss": 0.1111,
      "num_input_tokens_seen": 24999424,
      "step": 11575
    },
    {
      "epoch": 1.8890701468189235,
      "grad_norm": 4.3622517585754395,
      "learning_rate": 4.880614883731638e-05,
      "loss": 0.3256,
      "num_input_tokens_seen": 25010624,
      "step": 11580
    },
    {
      "epoch": 1.8898858075040783,
      "grad_norm": 6.816831111907959,
      "learning_rate": 4.88039745244992e-05,
      "loss": 0.3157,
      "num_input_tokens_seen": 25021696,
      "step": 11585
    },
    {
      "epoch": 1.8907014681892331,
      "grad_norm": 1.683463215827942,
      "learning_rate": 4.880179828201106e-05,
      "loss": 0.1249,
      "num_input_tokens_seen": 25032640,
      "step": 11590
    },
    {
      "epoch": 1.8915171288743884,
      "grad_norm": 0.07192590832710266,
      "learning_rate": 4.8799620110028375e-05,
      "loss": 0.2552,
      "num_input_tokens_seen": 25043744,
      "step": 11595
    },
    {
      "epoch": 1.8923327895595432,
      "grad_norm": 3.1122612953186035,
      "learning_rate": 4.879744000872774e-05,
      "loss": 0.3787,
      "num_input_tokens_seen": 25055136,
      "step": 11600
    },
    {
      "epoch": 1.8931484502446982,
      "grad_norm": 0.08800233155488968,
      "learning_rate": 4.879525797828585e-05,
      "loss": 0.126,
      "num_input_tokens_seen": 25066528,
      "step": 11605
    },
    {
      "epoch": 1.8939641109298533,
      "grad_norm": 0.6428315043449402,
      "learning_rate": 4.879307401887963e-05,
      "loss": 0.0265,
      "num_input_tokens_seen": 25077280,
      "step": 11610
    },
    {
      "epoch": 1.894779771615008,
      "grad_norm": 3.7041077613830566,
      "learning_rate": 4.87908881306861e-05,
      "loss": 0.1289,
      "num_input_tokens_seen": 25088544,
      "step": 11615
    },
    {
      "epoch": 1.8955954323001631,
      "grad_norm": 0.13210909068584442,
      "learning_rate": 4.878870031388246e-05,
      "loss": 0.0186,
      "num_input_tokens_seen": 25097344,
      "step": 11620
    },
    {
      "epoch": 1.8964110929853182,
      "grad_norm": 3.7125773429870605,
      "learning_rate": 4.8786510568646074e-05,
      "loss": 0.2127,
      "num_input_tokens_seen": 25108736,
      "step": 11625
    },
    {
      "epoch": 1.897226753670473,
      "grad_norm": 4.024951934814453,
      "learning_rate": 4.878431889515445e-05,
      "loss": 0.1576,
      "num_input_tokens_seen": 25119648,
      "step": 11630
    },
    {
      "epoch": 1.898042414355628,
      "grad_norm": 0.28142425417900085,
      "learning_rate": 4.8782125293585255e-05,
      "loss": 0.0438,
      "num_input_tokens_seen": 25130048,
      "step": 11635
    },
    {
      "epoch": 1.898858075040783,
      "grad_norm": 2.7388367652893066,
      "learning_rate": 4.877992976411632e-05,
      "loss": 0.1148,
      "num_input_tokens_seen": 25140032,
      "step": 11640
    },
    {
      "epoch": 1.899673735725938,
      "grad_norm": 0.16469921171665192,
      "learning_rate": 4.8777732306925614e-05,
      "loss": 0.2104,
      "num_input_tokens_seen": 25151136,
      "step": 11645
    },
    {
      "epoch": 1.900489396411093,
      "grad_norm": 0.21354708075523376,
      "learning_rate": 4.877553292219128e-05,
      "loss": 0.1238,
      "num_input_tokens_seen": 25161408,
      "step": 11650
    },
    {
      "epoch": 1.901305057096248,
      "grad_norm": 2.321537494659424,
      "learning_rate": 4.877333161009161e-05,
      "loss": 0.1364,
      "num_input_tokens_seen": 25172224,
      "step": 11655
    },
    {
      "epoch": 1.9021207177814028,
      "grad_norm": 1.2828636169433594,
      "learning_rate": 4.8771128370805066e-05,
      "loss": 0.2614,
      "num_input_tokens_seen": 25183424,
      "step": 11660
    },
    {
      "epoch": 1.902936378466558,
      "grad_norm": 1.7964338064193726,
      "learning_rate": 4.876892320451023e-05,
      "loss": 0.1097,
      "num_input_tokens_seen": 25194944,
      "step": 11665
    },
    {
      "epoch": 1.9037520391517129,
      "grad_norm": 1.1999586820602417,
      "learning_rate": 4.876671611138588e-05,
      "loss": 0.1334,
      "num_input_tokens_seen": 25205728,
      "step": 11670
    },
    {
      "epoch": 1.904567699836868,
      "grad_norm": 0.12653008103370667,
      "learning_rate": 4.876450709161093e-05,
      "loss": 0.1627,
      "num_input_tokens_seen": 25216576,
      "step": 11675
    },
    {
      "epoch": 1.905383360522023,
      "grad_norm": 0.07031231373548508,
      "learning_rate": 4.876229614536446e-05,
      "loss": 0.0095,
      "num_input_tokens_seen": 25227616,
      "step": 11680
    },
    {
      "epoch": 1.9061990212071778,
      "grad_norm": 0.8870084285736084,
      "learning_rate": 4.8760083272825695e-05,
      "loss": 0.1272,
      "num_input_tokens_seen": 25237856,
      "step": 11685
    },
    {
      "epoch": 1.9070146818923328,
      "grad_norm": 7.386667251586914,
      "learning_rate": 4.875786847417402e-05,
      "loss": 0.2315,
      "num_input_tokens_seen": 25248256,
      "step": 11690
    },
    {
      "epoch": 1.9078303425774878,
      "grad_norm": 0.2217559665441513,
      "learning_rate": 4.875565174958898e-05,
      "loss": 0.0843,
      "num_input_tokens_seen": 25258944,
      "step": 11695
    },
    {
      "epoch": 1.9086460032626427,
      "grad_norm": 0.147831529378891,
      "learning_rate": 4.8753433099250276e-05,
      "loss": 0.1228,
      "num_input_tokens_seen": 25269120,
      "step": 11700
    },
    {
      "epoch": 1.9094616639477977,
      "grad_norm": 1.9237598180770874,
      "learning_rate": 4.875121252333776e-05,
      "loss": 0.1464,
      "num_input_tokens_seen": 25280160,
      "step": 11705
    },
    {
      "epoch": 1.9102773246329527,
      "grad_norm": 4.616643905639648,
      "learning_rate": 4.874899002203145e-05,
      "loss": 0.2244,
      "num_input_tokens_seen": 25290112,
      "step": 11710
    },
    {
      "epoch": 1.9110929853181076,
      "grad_norm": 0.8603242039680481,
      "learning_rate": 4.8746765595511504e-05,
      "loss": 0.0974,
      "num_input_tokens_seen": 25301664,
      "step": 11715
    },
    {
      "epoch": 1.9119086460032626,
      "grad_norm": 2.7926738262176514,
      "learning_rate": 4.874453924395824e-05,
      "loss": 0.2612,
      "num_input_tokens_seen": 25311264,
      "step": 11720
    },
    {
      "epoch": 1.9127243066884176,
      "grad_norm": 0.6210606694221497,
      "learning_rate": 4.874231096755216e-05,
      "loss": 0.0201,
      "num_input_tokens_seen": 25321984,
      "step": 11725
    },
    {
      "epoch": 1.9135399673735725,
      "grad_norm": 0.3121470510959625,
      "learning_rate": 4.8740080766473876e-05,
      "loss": 0.0963,
      "num_input_tokens_seen": 25333664,
      "step": 11730
    },
    {
      "epoch": 1.9143556280587277,
      "grad_norm": 0.15780290961265564,
      "learning_rate": 4.873784864090419e-05,
      "loss": 0.1531,
      "num_input_tokens_seen": 25342944,
      "step": 11735
    },
    {
      "epoch": 1.9151712887438825,
      "grad_norm": 2.160059928894043,
      "learning_rate": 4.873561459102406e-05,
      "loss": 0.1624,
      "num_input_tokens_seen": 25353568,
      "step": 11740
    },
    {
      "epoch": 1.9159869494290374,
      "grad_norm": 0.15781857073307037,
      "learning_rate": 4.873337861701456e-05,
      "loss": 0.1003,
      "num_input_tokens_seen": 25364736,
      "step": 11745
    },
    {
      "epoch": 1.9168026101141926,
      "grad_norm": 0.0535578653216362,
      "learning_rate": 4.8731140719056977e-05,
      "loss": 0.0627,
      "num_input_tokens_seen": 25374528,
      "step": 11750
    },
    {
      "epoch": 1.9176182707993474,
      "grad_norm": 0.15755146741867065,
      "learning_rate": 4.872890089733272e-05,
      "loss": 0.1495,
      "num_input_tokens_seen": 25385440,
      "step": 11755
    },
    {
      "epoch": 1.9184339314845025,
      "grad_norm": 9.134349822998047,
      "learning_rate": 4.8726659152023356e-05,
      "loss": 0.3055,
      "num_input_tokens_seen": 25397344,
      "step": 11760
    },
    {
      "epoch": 1.9192495921696575,
      "grad_norm": 0.09816361218690872,
      "learning_rate": 4.87244154833106e-05,
      "loss": 0.0622,
      "num_input_tokens_seen": 25408448,
      "step": 11765
    },
    {
      "epoch": 1.9200652528548123,
      "grad_norm": 1.739254117012024,
      "learning_rate": 4.872216989137637e-05,
      "loss": 0.2179,
      "num_input_tokens_seen": 25418336,
      "step": 11770
    },
    {
      "epoch": 1.9208809135399674,
      "grad_norm": 0.8327562808990479,
      "learning_rate": 4.871992237640267e-05,
      "loss": 0.1653,
      "num_input_tokens_seen": 25429824,
      "step": 11775
    },
    {
      "epoch": 1.9216965742251224,
      "grad_norm": 0.17584045231342316,
      "learning_rate": 4.871767293857171e-05,
      "loss": 0.2071,
      "num_input_tokens_seen": 25439584,
      "step": 11780
    },
    {
      "epoch": 1.9225122349102772,
      "grad_norm": 3.230736017227173,
      "learning_rate": 4.871542157806584e-05,
      "loss": 0.1719,
      "num_input_tokens_seen": 25451744,
      "step": 11785
    },
    {
      "epoch": 1.9233278955954323,
      "grad_norm": 1.1818323135375977,
      "learning_rate": 4.871316829506757e-05,
      "loss": 0.0281,
      "num_input_tokens_seen": 25461856,
      "step": 11790
    },
    {
      "epoch": 1.9241435562805873,
      "grad_norm": 5.720760345458984,
      "learning_rate": 4.871091308975955e-05,
      "loss": 0.3042,
      "num_input_tokens_seen": 25473120,
      "step": 11795
    },
    {
      "epoch": 1.9249592169657421,
      "grad_norm": 3.1483027935028076,
      "learning_rate": 4.8708655962324615e-05,
      "loss": 0.0345,
      "num_input_tokens_seen": 25485632,
      "step": 11800
    },
    {
      "epoch": 1.9257748776508974,
      "grad_norm": 0.8307350873947144,
      "learning_rate": 4.870639691294573e-05,
      "loss": 0.0248,
      "num_input_tokens_seen": 25496160,
      "step": 11805
    },
    {
      "epoch": 1.9265905383360522,
      "grad_norm": 2.910285711288452,
      "learning_rate": 4.8704135941806016e-05,
      "loss": 0.135,
      "num_input_tokens_seen": 25507712,
      "step": 11810
    },
    {
      "epoch": 1.927406199021207,
      "grad_norm": 7.951888561248779,
      "learning_rate": 4.870187304908878e-05,
      "loss": 0.3368,
      "num_input_tokens_seen": 25519328,
      "step": 11815
    },
    {
      "epoch": 1.9282218597063623,
      "grad_norm": 0.47642260789871216,
      "learning_rate": 4.869960823497745e-05,
      "loss": 0.019,
      "num_input_tokens_seen": 25528992,
      "step": 11820
    },
    {
      "epoch": 1.929037520391517,
      "grad_norm": 0.1696988046169281,
      "learning_rate": 4.8697341499655626e-05,
      "loss": 0.1483,
      "num_input_tokens_seen": 25540032,
      "step": 11825
    },
    {
      "epoch": 1.9298531810766721,
      "grad_norm": 9.308799743652344,
      "learning_rate": 4.8695072843307064e-05,
      "loss": 0.1182,
      "num_input_tokens_seen": 25551392,
      "step": 11830
    },
    {
      "epoch": 1.9306688417618272,
      "grad_norm": 5.08847713470459,
      "learning_rate": 4.869280226611567e-05,
      "loss": 0.1406,
      "num_input_tokens_seen": 25561856,
      "step": 11835
    },
    {
      "epoch": 1.931484502446982,
      "grad_norm": 4.5817718505859375,
      "learning_rate": 4.86905297682655e-05,
      "loss": 0.2429,
      "num_input_tokens_seen": 25571712,
      "step": 11840
    },
    {
      "epoch": 1.932300163132137,
      "grad_norm": 0.09621615707874298,
      "learning_rate": 4.868825534994078e-05,
      "loss": 0.0559,
      "num_input_tokens_seen": 25581088,
      "step": 11845
    },
    {
      "epoch": 1.933115823817292,
      "grad_norm": 0.06166497990489006,
      "learning_rate": 4.86859790113259e-05,
      "loss": 0.0089,
      "num_input_tokens_seen": 25593376,
      "step": 11850
    },
    {
      "epoch": 1.933931484502447,
      "grad_norm": 4.5079169273376465,
      "learning_rate": 4.868370075260538e-05,
      "loss": 0.2731,
      "num_input_tokens_seen": 25603488,
      "step": 11855
    },
    {
      "epoch": 1.934747145187602,
      "grad_norm": 3.5476584434509277,
      "learning_rate": 4.86814205739639e-05,
      "loss": 0.3042,
      "num_input_tokens_seen": 25613216,
      "step": 11860
    },
    {
      "epoch": 1.935562805872757,
      "grad_norm": 1.5228770971298218,
      "learning_rate": 4.86791384755863e-05,
      "loss": 0.0359,
      "num_input_tokens_seen": 25624096,
      "step": 11865
    },
    {
      "epoch": 1.9363784665579118,
      "grad_norm": 0.13201339542865753,
      "learning_rate": 4.86768544576576e-05,
      "loss": 0.1756,
      "num_input_tokens_seen": 25633760,
      "step": 11870
    },
    {
      "epoch": 1.9371941272430668,
      "grad_norm": 0.7958288788795471,
      "learning_rate": 4.867456852036295e-05,
      "loss": 0.1408,
      "num_input_tokens_seen": 25645216,
      "step": 11875
    },
    {
      "epoch": 1.9380097879282219,
      "grad_norm": 0.15591788291931152,
      "learning_rate": 4.867228066388765e-05,
      "loss": 0.2832,
      "num_input_tokens_seen": 25656288,
      "step": 11880
    },
    {
      "epoch": 1.9388254486133767,
      "grad_norm": 0.15843915939331055,
      "learning_rate": 4.866999088841716e-05,
      "loss": 0.2027,
      "num_input_tokens_seen": 25667488,
      "step": 11885
    },
    {
      "epoch": 1.939641109298532,
      "grad_norm": 3.267289400100708,
      "learning_rate": 4.866769919413711e-05,
      "loss": 0.1275,
      "num_input_tokens_seen": 25678848,
      "step": 11890
    },
    {
      "epoch": 1.9404567699836868,
      "grad_norm": 0.848806619644165,
      "learning_rate": 4.866540558123328e-05,
      "loss": 0.0825,
      "num_input_tokens_seen": 25689280,
      "step": 11895
    },
    {
      "epoch": 1.9412724306688418,
      "grad_norm": 3.1739892959594727,
      "learning_rate": 4.8663110049891595e-05,
      "loss": 0.1432,
      "num_input_tokens_seen": 25699552,
      "step": 11900
    },
    {
      "epoch": 1.9420880913539968,
      "grad_norm": 0.08247596770524979,
      "learning_rate": 4.866081260029813e-05,
      "loss": 0.1974,
      "num_input_tokens_seen": 25709920,
      "step": 11905
    },
    {
      "epoch": 1.9429037520391517,
      "grad_norm": 4.789215564727783,
      "learning_rate": 4.8658513232639155e-05,
      "loss": 0.2647,
      "num_input_tokens_seen": 25722592,
      "step": 11910
    },
    {
      "epoch": 1.9437194127243067,
      "grad_norm": 0.2024151086807251,
      "learning_rate": 4.8656211947101054e-05,
      "loss": 0.0144,
      "num_input_tokens_seen": 25733696,
      "step": 11915
    },
    {
      "epoch": 1.9445350734094617,
      "grad_norm": 2.598015546798706,
      "learning_rate": 4.865390874387038e-05,
      "loss": 0.1755,
      "num_input_tokens_seen": 25743712,
      "step": 11920
    },
    {
      "epoch": 1.9453507340946166,
      "grad_norm": 0.09131888300180435,
      "learning_rate": 4.865160362313384e-05,
      "loss": 0.2475,
      "num_input_tokens_seen": 25753952,
      "step": 11925
    },
    {
      "epoch": 1.9461663947797716,
      "grad_norm": 0.42168277502059937,
      "learning_rate": 4.8649296585078316e-05,
      "loss": 0.1209,
      "num_input_tokens_seen": 25764896,
      "step": 11930
    },
    {
      "epoch": 1.9469820554649266,
      "grad_norm": 1.9647661447525024,
      "learning_rate": 4.864698762989081e-05,
      "loss": 0.12,
      "num_input_tokens_seen": 25774944,
      "step": 11935
    },
    {
      "epoch": 1.9477977161500815,
      "grad_norm": 0.47306153178215027,
      "learning_rate": 4.86446767577585e-05,
      "loss": 0.0901,
      "num_input_tokens_seen": 25785472,
      "step": 11940
    },
    {
      "epoch": 1.9486133768352365,
      "grad_norm": 1.2947392463684082,
      "learning_rate": 4.864236396886872e-05,
      "loss": 0.1879,
      "num_input_tokens_seen": 25796448,
      "step": 11945
    },
    {
      "epoch": 1.9494290375203915,
      "grad_norm": 0.14454756677150726,
      "learning_rate": 4.864004926340896e-05,
      "loss": 0.061,
      "num_input_tokens_seen": 25807168,
      "step": 11950
    },
    {
      "epoch": 1.9502446982055464,
      "grad_norm": 7.566388130187988,
      "learning_rate": 4.8637732641566855e-05,
      "loss": 0.1213,
      "num_input_tokens_seen": 25817792,
      "step": 11955
    },
    {
      "epoch": 1.9510603588907016,
      "grad_norm": 1.5212756395339966,
      "learning_rate": 4.8635414103530205e-05,
      "loss": 0.228,
      "num_input_tokens_seen": 25827392,
      "step": 11960
    },
    {
      "epoch": 1.9518760195758564,
      "grad_norm": 0.12417714297771454,
      "learning_rate": 4.863309364948697e-05,
      "loss": 0.0169,
      "num_input_tokens_seen": 25838720,
      "step": 11965
    },
    {
      "epoch": 1.9526916802610113,
      "grad_norm": 2.240032911300659,
      "learning_rate": 4.863077127962524e-05,
      "loss": 0.0409,
      "num_input_tokens_seen": 25850496,
      "step": 11970
    },
    {
      "epoch": 1.9535073409461665,
      "grad_norm": 0.1251329779624939,
      "learning_rate": 4.8628446994133306e-05,
      "loss": 0.1351,
      "num_input_tokens_seen": 25862240,
      "step": 11975
    },
    {
      "epoch": 1.9543230016313213,
      "grad_norm": 1.1251899003982544,
      "learning_rate": 4.8626120793199545e-05,
      "loss": 0.2288,
      "num_input_tokens_seen": 25873280,
      "step": 11980
    },
    {
      "epoch": 1.9551386623164764,
      "grad_norm": 0.14849115908145905,
      "learning_rate": 4.862379267701257e-05,
      "loss": 0.0192,
      "num_input_tokens_seen": 25883872,
      "step": 11985
    },
    {
      "epoch": 1.9559543230016314,
      "grad_norm": 0.17326806485652924,
      "learning_rate": 4.86214626457611e-05,
      "loss": 0.0124,
      "num_input_tokens_seen": 25895424,
      "step": 11990
    },
    {
      "epoch": 1.9567699836867862,
      "grad_norm": 5.171052932739258,
      "learning_rate": 4.8619130699633994e-05,
      "loss": 0.3317,
      "num_input_tokens_seen": 25906944,
      "step": 11995
    },
    {
      "epoch": 1.9575856443719413,
      "grad_norm": 3.372026205062866,
      "learning_rate": 4.861679683882033e-05,
      "loss": 0.1096,
      "num_input_tokens_seen": 25918720,
      "step": 12000
    },
    {
      "epoch": 1.9584013050570963,
      "grad_norm": 3.6195554733276367,
      "learning_rate": 4.861446106350928e-05,
      "loss": 0.068,
      "num_input_tokens_seen": 25929248,
      "step": 12005
    },
    {
      "epoch": 1.9592169657422511,
      "grad_norm": 5.8719635009765625,
      "learning_rate": 4.861212337389019e-05,
      "loss": 0.259,
      "num_input_tokens_seen": 25939040,
      "step": 12010
    },
    {
      "epoch": 1.9600326264274062,
      "grad_norm": 0.049886275082826614,
      "learning_rate": 4.8609783770152575e-05,
      "loss": 0.2155,
      "num_input_tokens_seen": 25949024,
      "step": 12015
    },
    {
      "epoch": 1.9608482871125612,
      "grad_norm": 0.07218191027641296,
      "learning_rate": 4.8607442252486095e-05,
      "loss": 0.2368,
      "num_input_tokens_seen": 25959264,
      "step": 12020
    },
    {
      "epoch": 1.961663947797716,
      "grad_norm": 0.17256250977516174,
      "learning_rate": 4.8605098821080564e-05,
      "loss": 0.0527,
      "num_input_tokens_seen": 25970432,
      "step": 12025
    },
    {
      "epoch": 1.9624796084828713,
      "grad_norm": 12.043556213378906,
      "learning_rate": 4.8602753476125954e-05,
      "loss": 0.2184,
      "num_input_tokens_seen": 25981856,
      "step": 12030
    },
    {
      "epoch": 1.963295269168026,
      "grad_norm": 0.8876869678497314,
      "learning_rate": 4.860040621781238e-05,
      "loss": 0.2989,
      "num_input_tokens_seen": 25993696,
      "step": 12035
    },
    {
      "epoch": 1.964110929853181,
      "grad_norm": 3.2724945545196533,
      "learning_rate": 4.8598057046330135e-05,
      "loss": 0.1858,
      "num_input_tokens_seen": 26004736,
      "step": 12040
    },
    {
      "epoch": 1.9649265905383362,
      "grad_norm": 0.14167501032352448,
      "learning_rate": 4.8595705961869656e-05,
      "loss": 0.1371,
      "num_input_tokens_seen": 26016480,
      "step": 12045
    },
    {
      "epoch": 1.965742251223491,
      "grad_norm": 0.7142931818962097,
      "learning_rate": 4.859335296462152e-05,
      "loss": 0.0118,
      "num_input_tokens_seen": 26027072,
      "step": 12050
    },
    {
      "epoch": 1.966557911908646,
      "grad_norm": 1.256037712097168,
      "learning_rate": 4.859099805477648e-05,
      "loss": 0.1166,
      "num_input_tokens_seen": 26036096,
      "step": 12055
    },
    {
      "epoch": 1.967373572593801,
      "grad_norm": 0.32416561245918274,
      "learning_rate": 4.858864123252544e-05,
      "loss": 0.1108,
      "num_input_tokens_seen": 26047264,
      "step": 12060
    },
    {
      "epoch": 1.968189233278956,
      "grad_norm": 0.6069583892822266,
      "learning_rate": 4.8586282498059456e-05,
      "loss": 0.1216,
      "num_input_tokens_seen": 26058176,
      "step": 12065
    },
    {
      "epoch": 1.969004893964111,
      "grad_norm": 0.22854197025299072,
      "learning_rate": 4.8583921851569735e-05,
      "loss": 0.0635,
      "num_input_tokens_seen": 26069664,
      "step": 12070
    },
    {
      "epoch": 1.969820554649266,
      "grad_norm": 0.08224233984947205,
      "learning_rate": 4.8581559293247655e-05,
      "loss": 0.1001,
      "num_input_tokens_seen": 26080640,
      "step": 12075
    },
    {
      "epoch": 1.9706362153344208,
      "grad_norm": 0.8420477509498596,
      "learning_rate": 4.857919482328471e-05,
      "loss": 0.0428,
      "num_input_tokens_seen": 26091360,
      "step": 12080
    },
    {
      "epoch": 1.9714518760195758,
      "grad_norm": 0.20236536860466003,
      "learning_rate": 4.857682844187261e-05,
      "loss": 0.0353,
      "num_input_tokens_seen": 26102080,
      "step": 12085
    },
    {
      "epoch": 1.9722675367047309,
      "grad_norm": 8.648172378540039,
      "learning_rate": 4.857446014920316e-05,
      "loss": 0.3215,
      "num_input_tokens_seen": 26112192,
      "step": 12090
    },
    {
      "epoch": 1.9730831973898857,
      "grad_norm": 4.223199844360352,
      "learning_rate": 4.857208994546836e-05,
      "loss": 0.18,
      "num_input_tokens_seen": 26122496,
      "step": 12095
    },
    {
      "epoch": 1.9738988580750407,
      "grad_norm": 6.034249782562256,
      "learning_rate": 4.856971783086034e-05,
      "loss": 0.086,
      "num_input_tokens_seen": 26134048,
      "step": 12100
    },
    {
      "epoch": 1.9747145187601958,
      "grad_norm": 0.1975557953119278,
      "learning_rate": 4.85673438055714e-05,
      "loss": 0.2566,
      "num_input_tokens_seen": 26144192,
      "step": 12105
    },
    {
      "epoch": 1.9755301794453506,
      "grad_norm": 0.14109517633914948,
      "learning_rate": 4.856496786979399e-05,
      "loss": 0.1572,
      "num_input_tokens_seen": 26155328,
      "step": 12110
    },
    {
      "epoch": 1.9763458401305058,
      "grad_norm": 0.08842913806438446,
      "learning_rate": 4.8562590023720725e-05,
      "loss": 0.1905,
      "num_input_tokens_seen": 26166272,
      "step": 12115
    },
    {
      "epoch": 1.9771615008156607,
      "grad_norm": 2.3355515003204346,
      "learning_rate": 4.8560210267544345e-05,
      "loss": 0.2171,
      "num_input_tokens_seen": 26175520,
      "step": 12120
    },
    {
      "epoch": 1.9779771615008157,
      "grad_norm": 4.464560508728027,
      "learning_rate": 4.855782860145779e-05,
      "loss": 0.4179,
      "num_input_tokens_seen": 26187008,
      "step": 12125
    },
    {
      "epoch": 1.9787928221859707,
      "grad_norm": 3.1640465259552,
      "learning_rate": 4.8555445025654116e-05,
      "loss": 0.132,
      "num_input_tokens_seen": 26198016,
      "step": 12130
    },
    {
      "epoch": 1.9796084828711256,
      "grad_norm": 0.2288096845149994,
      "learning_rate": 4.855305954032655e-05,
      "loss": 0.199,
      "num_input_tokens_seen": 26209440,
      "step": 12135
    },
    {
      "epoch": 1.9804241435562806,
      "grad_norm": 0.7268104553222656,
      "learning_rate": 4.855067214566846e-05,
      "loss": 0.0869,
      "num_input_tokens_seen": 26220736,
      "step": 12140
    },
    {
      "epoch": 1.9812398042414356,
      "grad_norm": 2.697340250015259,
      "learning_rate": 4.85482828418734e-05,
      "loss": 0.0684,
      "num_input_tokens_seen": 26231456,
      "step": 12145
    },
    {
      "epoch": 1.9820554649265905,
      "grad_norm": 0.1778566837310791,
      "learning_rate": 4.854589162913505e-05,
      "loss": 0.0792,
      "num_input_tokens_seen": 26241568,
      "step": 12150
    },
    {
      "epoch": 1.9828711256117455,
      "grad_norm": 2.7179176807403564,
      "learning_rate": 4.854349850764725e-05,
      "loss": 0.2748,
      "num_input_tokens_seen": 26252480,
      "step": 12155
    },
    {
      "epoch": 1.9836867862969005,
      "grad_norm": 0.2689560651779175,
      "learning_rate": 4.8541103477604e-05,
      "loss": 0.0481,
      "num_input_tokens_seen": 26263264,
      "step": 12160
    },
    {
      "epoch": 1.9845024469820554,
      "grad_norm": 4.350931167602539,
      "learning_rate": 4.853870653919946e-05,
      "loss": 0.1086,
      "num_input_tokens_seen": 26274496,
      "step": 12165
    },
    {
      "epoch": 1.9853181076672104,
      "grad_norm": 4.802256107330322,
      "learning_rate": 4.853630769262794e-05,
      "loss": 0.1675,
      "num_input_tokens_seen": 26285152,
      "step": 12170
    },
    {
      "epoch": 1.9861337683523654,
      "grad_norm": 3.5757229328155518,
      "learning_rate": 4.853390693808388e-05,
      "loss": 0.2827,
      "num_input_tokens_seen": 26296064,
      "step": 12175
    },
    {
      "epoch": 1.9869494290375203,
      "grad_norm": 3.319054365158081,
      "learning_rate": 4.853150427576193e-05,
      "loss": 0.1783,
      "num_input_tokens_seen": 26305984,
      "step": 12180
    },
    {
      "epoch": 1.9877650897226755,
      "grad_norm": 0.572554886341095,
      "learning_rate": 4.852909970585684e-05,
      "loss": 0.1024,
      "num_input_tokens_seen": 26316896,
      "step": 12185
    },
    {
      "epoch": 1.9885807504078303,
      "grad_norm": 0.4888591468334198,
      "learning_rate": 4.852669322856354e-05,
      "loss": 0.1044,
      "num_input_tokens_seen": 26326976,
      "step": 12190
    },
    {
      "epoch": 1.9893964110929854,
      "grad_norm": 2.590435266494751,
      "learning_rate": 4.8524284844077116e-05,
      "loss": 0.2301,
      "num_input_tokens_seen": 26337504,
      "step": 12195
    },
    {
      "epoch": 1.9902120717781404,
      "grad_norm": 2.9432034492492676,
      "learning_rate": 4.8521874552592805e-05,
      "loss": 0.1644,
      "num_input_tokens_seen": 26348608,
      "step": 12200
    },
    {
      "epoch": 1.9910277324632952,
      "grad_norm": 6.0449066162109375,
      "learning_rate": 4.851946235430599e-05,
      "loss": 0.2743,
      "num_input_tokens_seen": 26358240,
      "step": 12205
    },
    {
      "epoch": 1.9918433931484503,
      "grad_norm": 0.1594952940940857,
      "learning_rate": 4.851704824941222e-05,
      "loss": 0.0722,
      "num_input_tokens_seen": 26369568,
      "step": 12210
    },
    {
      "epoch": 1.9926590538336053,
      "grad_norm": 2.65429949760437,
      "learning_rate": 4.8514632238107194e-05,
      "loss": 0.1383,
      "num_input_tokens_seen": 26379936,
      "step": 12215
    },
    {
      "epoch": 1.9934747145187601,
      "grad_norm": 1.7988312244415283,
      "learning_rate": 4.851221432058677e-05,
      "loss": 0.124,
      "num_input_tokens_seen": 26390624,
      "step": 12220
    },
    {
      "epoch": 1.9942903752039152,
      "grad_norm": 0.48065802454948425,
      "learning_rate": 4.850979449704695e-05,
      "loss": 0.1274,
      "num_input_tokens_seen": 26399616,
      "step": 12225
    },
    {
      "epoch": 1.9951060358890702,
      "grad_norm": 0.7311342358589172,
      "learning_rate": 4.85073727676839e-05,
      "loss": 0.0981,
      "num_input_tokens_seen": 26410208,
      "step": 12230
    },
    {
      "epoch": 1.995921696574225,
      "grad_norm": 0.2876925468444824,
      "learning_rate": 4.8504949132693936e-05,
      "loss": 0.0782,
      "num_input_tokens_seen": 26420160,
      "step": 12235
    },
    {
      "epoch": 1.99673735725938,
      "grad_norm": 0.24604609608650208,
      "learning_rate": 4.850252359227353e-05,
      "loss": 0.0747,
      "num_input_tokens_seen": 26430432,
      "step": 12240
    },
    {
      "epoch": 1.997553017944535,
      "grad_norm": 0.2904493510723114,
      "learning_rate": 4.8500096146619325e-05,
      "loss": 0.1003,
      "num_input_tokens_seen": 26440128,
      "step": 12245
    },
    {
      "epoch": 1.99836867862969,
      "grad_norm": 4.259380340576172,
      "learning_rate": 4.849766679592808e-05,
      "loss": 0.094,
      "num_input_tokens_seen": 26450496,
      "step": 12250
    },
    {
      "epoch": 1.9991843393148452,
      "grad_norm": 2.26529598236084,
      "learning_rate": 4.849523554039673e-05,
      "loss": 0.126,
      "num_input_tokens_seen": 26461824,
      "step": 12255
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0837936401367188,
      "learning_rate": 4.8492802380222393e-05,
      "loss": 0.1146,
      "num_input_tokens_seen": 26471216,
      "step": 12260
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.14923527836799622,
      "eval_runtime": 132.9831,
      "eval_samples_per_second": 20.491,
      "eval_steps_per_second": 5.128,
      "num_input_tokens_seen": 26471216,
      "step": 12260
    },
    {
      "epoch": 2.000815660685155,
      "grad_norm": 3.045311689376831,
      "learning_rate": 4.849036731560228e-05,
      "loss": 0.1005,
      "num_input_tokens_seen": 26481904,
      "step": 12265
    },
    {
      "epoch": 2.00163132137031,
      "grad_norm": 3.7052078247070312,
      "learning_rate": 4.84879303467338e-05,
      "loss": 0.2073,
      "num_input_tokens_seen": 26491344,
      "step": 12270
    },
    {
      "epoch": 2.002446982055465,
      "grad_norm": 0.19312560558319092,
      "learning_rate": 4.8485491473814514e-05,
      "loss": 0.0304,
      "num_input_tokens_seen": 26502160,
      "step": 12275
    },
    {
      "epoch": 2.0032626427406197,
      "grad_norm": 2.924866199493408,
      "learning_rate": 4.8483050697042135e-05,
      "loss": 0.076,
      "num_input_tokens_seen": 26512816,
      "step": 12280
    },
    {
      "epoch": 2.004078303425775,
      "grad_norm": 0.06528328359127045,
      "learning_rate": 4.8480608016614504e-05,
      "loss": 0.0106,
      "num_input_tokens_seen": 26522640,
      "step": 12285
    },
    {
      "epoch": 2.00489396411093,
      "grad_norm": 0.07696110010147095,
      "learning_rate": 4.847816343272965e-05,
      "loss": 0.2894,
      "num_input_tokens_seen": 26533072,
      "step": 12290
    },
    {
      "epoch": 2.0057096247960846,
      "grad_norm": 0.037772562354803085,
      "learning_rate": 4.847571694558574e-05,
      "loss": 0.1201,
      "num_input_tokens_seen": 26543408,
      "step": 12295
    },
    {
      "epoch": 2.00652528548124,
      "grad_norm": 0.08915374428033829,
      "learning_rate": 4.84732685553811e-05,
      "loss": 0.2462,
      "num_input_tokens_seen": 26554320,
      "step": 12300
    },
    {
      "epoch": 2.0073409461663947,
      "grad_norm": 0.136813685297966,
      "learning_rate": 4.847081826231421e-05,
      "loss": 0.1828,
      "num_input_tokens_seen": 26565008,
      "step": 12305
    },
    {
      "epoch": 2.00815660685155,
      "grad_norm": 0.06767254322767258,
      "learning_rate": 4.846836606658371e-05,
      "loss": 0.0326,
      "num_input_tokens_seen": 26576912,
      "step": 12310
    },
    {
      "epoch": 2.0089722675367048,
      "grad_norm": 3.284564971923828,
      "learning_rate": 4.8465911968388364e-05,
      "loss": 0.4687,
      "num_input_tokens_seen": 26586800,
      "step": 12315
    },
    {
      "epoch": 2.0097879282218596,
      "grad_norm": 1.2693392038345337,
      "learning_rate": 4.846345596792713e-05,
      "loss": 0.0776,
      "num_input_tokens_seen": 26598160,
      "step": 12320
    },
    {
      "epoch": 2.010603588907015,
      "grad_norm": 2.8209829330444336,
      "learning_rate": 4.846099806539911e-05,
      "loss": 0.1851,
      "num_input_tokens_seen": 26609136,
      "step": 12325
    },
    {
      "epoch": 2.0114192495921697,
      "grad_norm": 0.29326483607292175,
      "learning_rate": 4.845853826100355e-05,
      "loss": 0.0905,
      "num_input_tokens_seen": 26620752,
      "step": 12330
    },
    {
      "epoch": 2.0122349102773245,
      "grad_norm": 5.118554592132568,
      "learning_rate": 4.845607655493984e-05,
      "loss": 0.2262,
      "num_input_tokens_seen": 26630256,
      "step": 12335
    },
    {
      "epoch": 2.0130505709624797,
      "grad_norm": 3.600283145904541,
      "learning_rate": 4.8453612947407564e-05,
      "loss": 0.3009,
      "num_input_tokens_seen": 26641136,
      "step": 12340
    },
    {
      "epoch": 2.0138662316476346,
      "grad_norm": 0.3470727205276489,
      "learning_rate": 4.8451147438606416e-05,
      "loss": 0.0599,
      "num_input_tokens_seen": 26651824,
      "step": 12345
    },
    {
      "epoch": 2.0146818923327894,
      "grad_norm": 6.615456581115723,
      "learning_rate": 4.844868002873626e-05,
      "loss": 0.2983,
      "num_input_tokens_seen": 26662064,
      "step": 12350
    },
    {
      "epoch": 2.0154975530179446,
      "grad_norm": 0.5639845132827759,
      "learning_rate": 4.844621071799712e-05,
      "loss": 0.1157,
      "num_input_tokens_seen": 26673104,
      "step": 12355
    },
    {
      "epoch": 2.0163132137030995,
      "grad_norm": 2.5894696712493896,
      "learning_rate": 4.844373950658918e-05,
      "loss": 0.2547,
      "num_input_tokens_seen": 26682736,
      "step": 12360
    },
    {
      "epoch": 2.0171288743882543,
      "grad_norm": 0.23206761479377747,
      "learning_rate": 4.844126639471277e-05,
      "loss": 0.1492,
      "num_input_tokens_seen": 26694224,
      "step": 12365
    },
    {
      "epoch": 2.0179445350734095,
      "grad_norm": 0.5956500172615051,
      "learning_rate": 4.843879138256836e-05,
      "loss": 0.0564,
      "num_input_tokens_seen": 26704880,
      "step": 12370
    },
    {
      "epoch": 2.0187601957585644,
      "grad_norm": 1.3979803323745728,
      "learning_rate": 4.843631447035659e-05,
      "loss": 0.1814,
      "num_input_tokens_seen": 26713648,
      "step": 12375
    },
    {
      "epoch": 2.0195758564437196,
      "grad_norm": 0.1042853444814682,
      "learning_rate": 4.843383565827826e-05,
      "loss": 0.0313,
      "num_input_tokens_seen": 26724560,
      "step": 12380
    },
    {
      "epoch": 2.0203915171288744,
      "grad_norm": 4.647457599639893,
      "learning_rate": 4.843135494653431e-05,
      "loss": 0.1446,
      "num_input_tokens_seen": 26735344,
      "step": 12385
    },
    {
      "epoch": 2.0212071778140293,
      "grad_norm": 2.719658851623535,
      "learning_rate": 4.842887233532584e-05,
      "loss": 0.1123,
      "num_input_tokens_seen": 26746288,
      "step": 12390
    },
    {
      "epoch": 2.0220228384991845,
      "grad_norm": 2.9892961978912354,
      "learning_rate": 4.842638782485409e-05,
      "loss": 0.1384,
      "num_input_tokens_seen": 26757936,
      "step": 12395
    },
    {
      "epoch": 2.0228384991843393,
      "grad_norm": 2.41135835647583,
      "learning_rate": 4.8423901415320486e-05,
      "loss": 0.1465,
      "num_input_tokens_seen": 26770256,
      "step": 12400
    },
    {
      "epoch": 2.023654159869494,
      "grad_norm": 3.05769681930542,
      "learning_rate": 4.8421413106926586e-05,
      "loss": 0.0795,
      "num_input_tokens_seen": 26780240,
      "step": 12405
    },
    {
      "epoch": 2.0244698205546494,
      "grad_norm": 0.4137382209300995,
      "learning_rate": 4.841892289987409e-05,
      "loss": 0.1214,
      "num_input_tokens_seen": 26791824,
      "step": 12410
    },
    {
      "epoch": 2.0252854812398042,
      "grad_norm": 0.10253965109586716,
      "learning_rate": 4.841643079436489e-05,
      "loss": 0.1238,
      "num_input_tokens_seen": 26802576,
      "step": 12415
    },
    {
      "epoch": 2.026101141924959,
      "grad_norm": 0.252849817276001,
      "learning_rate": 4.841393679060099e-05,
      "loss": 0.1483,
      "num_input_tokens_seen": 26814256,
      "step": 12420
    },
    {
      "epoch": 2.0269168026101143,
      "grad_norm": 6.598630428314209,
      "learning_rate": 4.841144088878457e-05,
      "loss": 0.3621,
      "num_input_tokens_seen": 26824528,
      "step": 12425
    },
    {
      "epoch": 2.027732463295269,
      "grad_norm": 3.748215913772583,
      "learning_rate": 4.8408943089117964e-05,
      "loss": 0.0528,
      "num_input_tokens_seen": 26836816,
      "step": 12430
    },
    {
      "epoch": 2.028548123980424,
      "grad_norm": 1.9309725761413574,
      "learning_rate": 4.840644339180366e-05,
      "loss": 0.1367,
      "num_input_tokens_seen": 26846800,
      "step": 12435
    },
    {
      "epoch": 2.029363784665579,
      "grad_norm": 1.057390570640564,
      "learning_rate": 4.8403941797044286e-05,
      "loss": 0.1011,
      "num_input_tokens_seen": 26858288,
      "step": 12440
    },
    {
      "epoch": 2.030179445350734,
      "grad_norm": 0.6322423219680786,
      "learning_rate": 4.840143830504264e-05,
      "loss": 0.0339,
      "num_input_tokens_seen": 26868624,
      "step": 12445
    },
    {
      "epoch": 2.0309951060358893,
      "grad_norm": 0.3212203085422516,
      "learning_rate": 4.839893291600167e-05,
      "loss": 0.1978,
      "num_input_tokens_seen": 26879536,
      "step": 12450
    },
    {
      "epoch": 2.031810766721044,
      "grad_norm": 1.5423448085784912,
      "learning_rate": 4.839642563012447e-05,
      "loss": 0.4648,
      "num_input_tokens_seen": 26890320,
      "step": 12455
    },
    {
      "epoch": 2.032626427406199,
      "grad_norm": 0.2827596366405487,
      "learning_rate": 4.83939164476143e-05,
      "loss": 0.1585,
      "num_input_tokens_seen": 26900784,
      "step": 12460
    },
    {
      "epoch": 2.033442088091354,
      "grad_norm": 0.22221170365810394,
      "learning_rate": 4.839140536867456e-05,
      "loss": 0.0696,
      "num_input_tokens_seen": 26912944,
      "step": 12465
    },
    {
      "epoch": 2.034257748776509,
      "grad_norm": 0.13560988008975983,
      "learning_rate": 4.838889239350881e-05,
      "loss": 0.1641,
      "num_input_tokens_seen": 26923792,
      "step": 12470
    },
    {
      "epoch": 2.035073409461664,
      "grad_norm": 0.320088267326355,
      "learning_rate": 4.838637752232078e-05,
      "loss": 0.0278,
      "num_input_tokens_seen": 26934736,
      "step": 12475
    },
    {
      "epoch": 2.035889070146819,
      "grad_norm": 0.11676841229200363,
      "learning_rate": 4.838386075531432e-05,
      "loss": 0.1107,
      "num_input_tokens_seen": 26945840,
      "step": 12480
    },
    {
      "epoch": 2.036704730831974,
      "grad_norm": 0.10097681730985641,
      "learning_rate": 4.8381342092693464e-05,
      "loss": 0.0379,
      "num_input_tokens_seen": 26956304,
      "step": 12485
    },
    {
      "epoch": 2.0375203915171287,
      "grad_norm": 1.9858624935150146,
      "learning_rate": 4.837882153466237e-05,
      "loss": 0.0379,
      "num_input_tokens_seen": 26967632,
      "step": 12490
    },
    {
      "epoch": 2.038336052202284,
      "grad_norm": 0.4636116623878479,
      "learning_rate": 4.837629908142539e-05,
      "loss": 0.0771,
      "num_input_tokens_seen": 26978160,
      "step": 12495
    },
    {
      "epoch": 2.039151712887439,
      "grad_norm": 3.140413284301758,
      "learning_rate": 4.837377473318699e-05,
      "loss": 0.2424,
      "num_input_tokens_seen": 26987568,
      "step": 12500
    },
    {
      "epoch": 2.0399673735725936,
      "grad_norm": 0.1420295536518097,
      "learning_rate": 4.837124849015182e-05,
      "loss": 0.0991,
      "num_input_tokens_seen": 26999760,
      "step": 12505
    },
    {
      "epoch": 2.040783034257749,
      "grad_norm": 3.2932164669036865,
      "learning_rate": 4.8368720352524655e-05,
      "loss": 0.0692,
      "num_input_tokens_seen": 27011696,
      "step": 12510
    },
    {
      "epoch": 2.0415986949429037,
      "grad_norm": 1.278024673461914,
      "learning_rate": 4.8366190320510454e-05,
      "loss": 0.054,
      "num_input_tokens_seen": 27022608,
      "step": 12515
    },
    {
      "epoch": 2.0424143556280585,
      "grad_norm": 2.2411954402923584,
      "learning_rate": 4.83636583943143e-05,
      "loss": 0.1548,
      "num_input_tokens_seen": 27033680,
      "step": 12520
    },
    {
      "epoch": 2.0432300163132138,
      "grad_norm": 3.933164119720459,
      "learning_rate": 4.8361124574141455e-05,
      "loss": 0.0331,
      "num_input_tokens_seen": 27045136,
      "step": 12525
    },
    {
      "epoch": 2.0440456769983686,
      "grad_norm": 0.5014989972114563,
      "learning_rate": 4.835858886019732e-05,
      "loss": 0.1856,
      "num_input_tokens_seen": 27055760,
      "step": 12530
    },
    {
      "epoch": 2.044861337683524,
      "grad_norm": 0.11863043904304504,
      "learning_rate": 4.835605125268745e-05,
      "loss": 0.0668,
      "num_input_tokens_seen": 27066352,
      "step": 12535
    },
    {
      "epoch": 2.0456769983686787,
      "grad_norm": 0.6212408542633057,
      "learning_rate": 4.835351175181755e-05,
      "loss": 0.1345,
      "num_input_tokens_seen": 27077872,
      "step": 12540
    },
    {
      "epoch": 2.0464926590538335,
      "grad_norm": 5.539122104644775,
      "learning_rate": 4.83509703577935e-05,
      "loss": 0.1086,
      "num_input_tokens_seen": 27088912,
      "step": 12545
    },
    {
      "epoch": 2.0473083197389887,
      "grad_norm": 0.050495702773332596,
      "learning_rate": 4.834842707082131e-05,
      "loss": 0.0102,
      "num_input_tokens_seen": 27099216,
      "step": 12550
    },
    {
      "epoch": 2.0481239804241436,
      "grad_norm": 0.1347498893737793,
      "learning_rate": 4.834588189110716e-05,
      "loss": 0.2236,
      "num_input_tokens_seen": 27109712,
      "step": 12555
    },
    {
      "epoch": 2.0489396411092984,
      "grad_norm": 0.47515809535980225,
      "learning_rate": 4.834333481885735e-05,
      "loss": 0.0167,
      "num_input_tokens_seen": 27120304,
      "step": 12560
    },
    {
      "epoch": 2.0497553017944536,
      "grad_norm": 0.0680442750453949,
      "learning_rate": 4.8340785854278395e-05,
      "loss": 0.1594,
      "num_input_tokens_seen": 27130992,
      "step": 12565
    },
    {
      "epoch": 2.0505709624796085,
      "grad_norm": 2.6305956840515137,
      "learning_rate": 4.83382349975769e-05,
      "loss": 0.1084,
      "num_input_tokens_seen": 27141136,
      "step": 12570
    },
    {
      "epoch": 2.0513866231647633,
      "grad_norm": 0.8694716095924377,
      "learning_rate": 4.833568224895967e-05,
      "loss": 0.1113,
      "num_input_tokens_seen": 27152208,
      "step": 12575
    },
    {
      "epoch": 2.0522022838499185,
      "grad_norm": 0.0830124020576477,
      "learning_rate": 4.833312760863362e-05,
      "loss": 0.1575,
      "num_input_tokens_seen": 27164016,
      "step": 12580
    },
    {
      "epoch": 2.0530179445350734,
      "grad_norm": 0.39163777232170105,
      "learning_rate": 4.833057107680586e-05,
      "loss": 0.1352,
      "num_input_tokens_seen": 27174800,
      "step": 12585
    },
    {
      "epoch": 2.053833605220228,
      "grad_norm": 0.5324012041091919,
      "learning_rate": 4.832801265368363e-05,
      "loss": 0.0634,
      "num_input_tokens_seen": 27184784,
      "step": 12590
    },
    {
      "epoch": 2.0546492659053834,
      "grad_norm": 0.484078973531723,
      "learning_rate": 4.832545233947433e-05,
      "loss": 0.0428,
      "num_input_tokens_seen": 27195600,
      "step": 12595
    },
    {
      "epoch": 2.0554649265905383,
      "grad_norm": 0.09587056934833527,
      "learning_rate": 4.832289013438551e-05,
      "loss": 0.0715,
      "num_input_tokens_seen": 27206416,
      "step": 12600
    },
    {
      "epoch": 2.0562805872756935,
      "grad_norm": 1.2562344074249268,
      "learning_rate": 4.8320326038624875e-05,
      "loss": 0.1726,
      "num_input_tokens_seen": 27217360,
      "step": 12605
    },
    {
      "epoch": 2.0570962479608483,
      "grad_norm": 3.0299909114837646,
      "learning_rate": 4.831776005240029e-05,
      "loss": 0.2203,
      "num_input_tokens_seen": 27228304,
      "step": 12610
    },
    {
      "epoch": 2.057911908646003,
      "grad_norm": 0.25369992852211,
      "learning_rate": 4.831519217591976e-05,
      "loss": 0.157,
      "num_input_tokens_seen": 27240208,
      "step": 12615
    },
    {
      "epoch": 2.0587275693311584,
      "grad_norm": 0.15049006044864655,
      "learning_rate": 4.831262240939144e-05,
      "loss": 0.044,
      "num_input_tokens_seen": 27250256,
      "step": 12620
    },
    {
      "epoch": 2.0595432300163132,
      "grad_norm": 6.327296733856201,
      "learning_rate": 4.8310050753023674e-05,
      "loss": 0.2034,
      "num_input_tokens_seen": 27260080,
      "step": 12625
    },
    {
      "epoch": 2.060358890701468,
      "grad_norm": 2.6136813163757324,
      "learning_rate": 4.8307477207024923e-05,
      "loss": 0.0326,
      "num_input_tokens_seen": 27272112,
      "step": 12630
    },
    {
      "epoch": 2.0611745513866233,
      "grad_norm": 0.1514524519443512,
      "learning_rate": 4.83049017716038e-05,
      "loss": 0.1483,
      "num_input_tokens_seen": 27282672,
      "step": 12635
    },
    {
      "epoch": 2.061990212071778,
      "grad_norm": 2.3813018798828125,
      "learning_rate": 4.8302324446969094e-05,
      "loss": 0.0663,
      "num_input_tokens_seen": 27294704,
      "step": 12640
    },
    {
      "epoch": 2.062805872756933,
      "grad_norm": 0.9157320857048035,
      "learning_rate": 4.829974523332973e-05,
      "loss": 0.1115,
      "num_input_tokens_seen": 27306320,
      "step": 12645
    },
    {
      "epoch": 2.063621533442088,
      "grad_norm": 0.3651241064071655,
      "learning_rate": 4.8297164130894804e-05,
      "loss": 0.1239,
      "num_input_tokens_seen": 27316720,
      "step": 12650
    },
    {
      "epoch": 2.064437194127243,
      "grad_norm": 0.14332373440265656,
      "learning_rate": 4.8294581139873544e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 27327408,
      "step": 12655
    },
    {
      "epoch": 2.065252854812398,
      "grad_norm": 2.2300896644592285,
      "learning_rate": 4.829199626047534e-05,
      "loss": 0.1645,
      "num_input_tokens_seen": 27338672,
      "step": 12660
    },
    {
      "epoch": 2.066068515497553,
      "grad_norm": 3.436211109161377,
      "learning_rate": 4.8289409492909726e-05,
      "loss": 0.2541,
      "num_input_tokens_seen": 27350352,
      "step": 12665
    },
    {
      "epoch": 2.066884176182708,
      "grad_norm": 0.06419938057661057,
      "learning_rate": 4.8286820837386416e-05,
      "loss": 0.0668,
      "num_input_tokens_seen": 27360432,
      "step": 12670
    },
    {
      "epoch": 2.067699836867863,
      "grad_norm": 0.6670409440994263,
      "learning_rate": 4.828423029411526e-05,
      "loss": 0.1714,
      "num_input_tokens_seen": 27369840,
      "step": 12675
    },
    {
      "epoch": 2.068515497553018,
      "grad_norm": 0.11572045087814331,
      "learning_rate": 4.828163786330624e-05,
      "loss": 0.0127,
      "num_input_tokens_seen": 27380016,
      "step": 12680
    },
    {
      "epoch": 2.069331158238173,
      "grad_norm": 3.221712827682495,
      "learning_rate": 4.8279043545169535e-05,
      "loss": 0.1313,
      "num_input_tokens_seen": 27391504,
      "step": 12685
    },
    {
      "epoch": 2.070146818923328,
      "grad_norm": 0.22601261734962463,
      "learning_rate": 4.8276447339915446e-05,
      "loss": 0.0173,
      "num_input_tokens_seen": 27402064,
      "step": 12690
    },
    {
      "epoch": 2.070962479608483,
      "grad_norm": 0.08553045243024826,
      "learning_rate": 4.827384924775442e-05,
      "loss": 0.0924,
      "num_input_tokens_seen": 27413808,
      "step": 12695
    },
    {
      "epoch": 2.0717781402936377,
      "grad_norm": 0.33960750699043274,
      "learning_rate": 4.8271249268897094e-05,
      "loss": 0.2201,
      "num_input_tokens_seen": 27425520,
      "step": 12700
    },
    {
      "epoch": 2.072593800978793,
      "grad_norm": 3.99654221534729,
      "learning_rate": 4.826864740355422e-05,
      "loss": 0.2413,
      "num_input_tokens_seen": 27436624,
      "step": 12705
    },
    {
      "epoch": 2.073409461663948,
      "grad_norm": 0.1754871904850006,
      "learning_rate": 4.826604365193673e-05,
      "loss": 0.15,
      "num_input_tokens_seen": 27448176,
      "step": 12710
    },
    {
      "epoch": 2.0742251223491026,
      "grad_norm": 0.36933010816574097,
      "learning_rate": 4.8263438014255687e-05,
      "loss": 0.0558,
      "num_input_tokens_seen": 27459568,
      "step": 12715
    },
    {
      "epoch": 2.075040783034258,
      "grad_norm": 6.893112659454346,
      "learning_rate": 4.8260830490722317e-05,
      "loss": 0.1323,
      "num_input_tokens_seen": 27470800,
      "step": 12720
    },
    {
      "epoch": 2.0758564437194127,
      "grad_norm": 0.9920274019241333,
      "learning_rate": 4.8258221081548004e-05,
      "loss": 0.1025,
      "num_input_tokens_seen": 27481168,
      "step": 12725
    },
    {
      "epoch": 2.0766721044045675,
      "grad_norm": 0.5149859189987183,
      "learning_rate": 4.825560978694429e-05,
      "loss": 0.0987,
      "num_input_tokens_seen": 27492528,
      "step": 12730
    },
    {
      "epoch": 2.0774877650897228,
      "grad_norm": 0.38302597403526306,
      "learning_rate": 4.8252996607122835e-05,
      "loss": 0.1309,
      "num_input_tokens_seen": 27503152,
      "step": 12735
    },
    {
      "epoch": 2.0783034257748776,
      "grad_norm": 0.5170636773109436,
      "learning_rate": 4.82503815422955e-05,
      "loss": 0.0333,
      "num_input_tokens_seen": 27513904,
      "step": 12740
    },
    {
      "epoch": 2.0791190864600324,
      "grad_norm": 2.662625312805176,
      "learning_rate": 4.824776459267426e-05,
      "loss": 0.2391,
      "num_input_tokens_seen": 27525648,
      "step": 12745
    },
    {
      "epoch": 2.0799347471451877,
      "grad_norm": 4.804661750793457,
      "learning_rate": 4.824514575847127e-05,
      "loss": 0.1396,
      "num_input_tokens_seen": 27537264,
      "step": 12750
    },
    {
      "epoch": 2.0807504078303425,
      "grad_norm": 1.9787460565567017,
      "learning_rate": 4.824252503989881e-05,
      "loss": 0.0634,
      "num_input_tokens_seen": 27548144,
      "step": 12755
    },
    {
      "epoch": 2.0815660685154977,
      "grad_norm": 0.3444620966911316,
      "learning_rate": 4.823990243716935e-05,
      "loss": 0.0351,
      "num_input_tokens_seen": 27559568,
      "step": 12760
    },
    {
      "epoch": 2.0823817292006526,
      "grad_norm": 4.222772598266602,
      "learning_rate": 4.823727795049548e-05,
      "loss": 0.2073,
      "num_input_tokens_seen": 27570544,
      "step": 12765
    },
    {
      "epoch": 2.0831973898858074,
      "grad_norm": 0.6607770919799805,
      "learning_rate": 4.8234651580089945e-05,
      "loss": 0.1915,
      "num_input_tokens_seen": 27581616,
      "step": 12770
    },
    {
      "epoch": 2.0840130505709626,
      "grad_norm": 0.049387168139219284,
      "learning_rate": 4.823202332616567e-05,
      "loss": 0.2267,
      "num_input_tokens_seen": 27592112,
      "step": 12775
    },
    {
      "epoch": 2.0848287112561175,
      "grad_norm": 2.145779609680176,
      "learning_rate": 4.8229393188935703e-05,
      "loss": 0.2072,
      "num_input_tokens_seen": 27603056,
      "step": 12780
    },
    {
      "epoch": 2.0856443719412723,
      "grad_norm": 1.4277530908584595,
      "learning_rate": 4.8226761168613255e-05,
      "loss": 0.1471,
      "num_input_tokens_seen": 27613616,
      "step": 12785
    },
    {
      "epoch": 2.0864600326264275,
      "grad_norm": 1.1328977346420288,
      "learning_rate": 4.82241272654117e-05,
      "loss": 0.1536,
      "num_input_tokens_seen": 27624784,
      "step": 12790
    },
    {
      "epoch": 2.0872756933115824,
      "grad_norm": 0.17303496599197388,
      "learning_rate": 4.822149147954455e-05,
      "loss": 0.0318,
      "num_input_tokens_seen": 27636048,
      "step": 12795
    },
    {
      "epoch": 2.088091353996737,
      "grad_norm": 1.8548846244812012,
      "learning_rate": 4.8218853811225475e-05,
      "loss": 0.1574,
      "num_input_tokens_seen": 27645328,
      "step": 12800
    },
    {
      "epoch": 2.0889070146818924,
      "grad_norm": 0.09046615660190582,
      "learning_rate": 4.8216214260668304e-05,
      "loss": 0.1622,
      "num_input_tokens_seen": 27655024,
      "step": 12805
    },
    {
      "epoch": 2.0897226753670473,
      "grad_norm": 4.659290313720703,
      "learning_rate": 4.8213572828087e-05,
      "loss": 0.1731,
      "num_input_tokens_seen": 27666224,
      "step": 12810
    },
    {
      "epoch": 2.090538336052202,
      "grad_norm": 0.06531661748886108,
      "learning_rate": 4.82109295136957e-05,
      "loss": 0.0062,
      "num_input_tokens_seen": 27677904,
      "step": 12815
    },
    {
      "epoch": 2.0913539967373573,
      "grad_norm": 3.016369104385376,
      "learning_rate": 4.820828431770868e-05,
      "loss": 0.1218,
      "num_input_tokens_seen": 27687088,
      "step": 12820
    },
    {
      "epoch": 2.092169657422512,
      "grad_norm": 3.686668872833252,
      "learning_rate": 4.820563724034039e-05,
      "loss": 0.2888,
      "num_input_tokens_seen": 27697008,
      "step": 12825
    },
    {
      "epoch": 2.0929853181076674,
      "grad_norm": 4.606642246246338,
      "learning_rate": 4.820298828180538e-05,
      "loss": 0.0425,
      "num_input_tokens_seen": 27708208,
      "step": 12830
    },
    {
      "epoch": 2.0938009787928222,
      "grad_norm": 0.16520646214485168,
      "learning_rate": 4.8200337442318424e-05,
      "loss": 0.316,
      "num_input_tokens_seen": 27718704,
      "step": 12835
    },
    {
      "epoch": 2.094616639477977,
      "grad_norm": 0.08105045557022095,
      "learning_rate": 4.819768472209439e-05,
      "loss": 0.0807,
      "num_input_tokens_seen": 27728816,
      "step": 12840
    },
    {
      "epoch": 2.0954323001631323,
      "grad_norm": 3.2872564792633057,
      "learning_rate": 4.8195030121348336e-05,
      "loss": 0.0702,
      "num_input_tokens_seen": 27738992,
      "step": 12845
    },
    {
      "epoch": 2.096247960848287,
      "grad_norm": 0.22190487384796143,
      "learning_rate": 4.819237364029544e-05,
      "loss": 0.1307,
      "num_input_tokens_seen": 27750384,
      "step": 12850
    },
    {
      "epoch": 2.097063621533442,
      "grad_norm": 0.9946272969245911,
      "learning_rate": 4.818971527915107e-05,
      "loss": 0.0873,
      "num_input_tokens_seen": 27760784,
      "step": 12855
    },
    {
      "epoch": 2.097879282218597,
      "grad_norm": 0.13623514771461487,
      "learning_rate": 4.818705503813071e-05,
      "loss": 0.0473,
      "num_input_tokens_seen": 27771952,
      "step": 12860
    },
    {
      "epoch": 2.098694942903752,
      "grad_norm": 0.6727923154830933,
      "learning_rate": 4.818439291745002e-05,
      "loss": 0.0806,
      "num_input_tokens_seen": 27781744,
      "step": 12865
    },
    {
      "epoch": 2.099510603588907,
      "grad_norm": 0.07763024419546127,
      "learning_rate": 4.81817289173248e-05,
      "loss": 0.0227,
      "num_input_tokens_seen": 27792944,
      "step": 12870
    },
    {
      "epoch": 2.100326264274062,
      "grad_norm": 0.552696704864502,
      "learning_rate": 4.8179063037971016e-05,
      "loss": 0.0718,
      "num_input_tokens_seen": 27804048,
      "step": 12875
    },
    {
      "epoch": 2.101141924959217,
      "grad_norm": 2.295488119125366,
      "learning_rate": 4.817639527960477e-05,
      "loss": 0.1117,
      "num_input_tokens_seen": 27814960,
      "step": 12880
    },
    {
      "epoch": 2.1019575856443717,
      "grad_norm": 0.11163581907749176,
      "learning_rate": 4.817372564244233e-05,
      "loss": 0.0343,
      "num_input_tokens_seen": 27827280,
      "step": 12885
    },
    {
      "epoch": 2.102773246329527,
      "grad_norm": 1.5921348333358765,
      "learning_rate": 4.817105412670011e-05,
      "loss": 0.0979,
      "num_input_tokens_seen": 27837520,
      "step": 12890
    },
    {
      "epoch": 2.103588907014682,
      "grad_norm": 5.123642444610596,
      "learning_rate": 4.8168380732594666e-05,
      "loss": 0.0743,
      "num_input_tokens_seen": 27848912,
      "step": 12895
    },
    {
      "epoch": 2.104404567699837,
      "grad_norm": 0.34312704205513,
      "learning_rate": 4.816570546034273e-05,
      "loss": 0.0867,
      "num_input_tokens_seen": 27859504,
      "step": 12900
    },
    {
      "epoch": 2.105220228384992,
      "grad_norm": 0.45883709192276,
      "learning_rate": 4.816302831016116e-05,
      "loss": 0.0312,
      "num_input_tokens_seen": 27870736,
      "step": 12905
    },
    {
      "epoch": 2.1060358890701467,
      "grad_norm": 4.113556385040283,
      "learning_rate": 4.8160349282266995e-05,
      "loss": 0.1945,
      "num_input_tokens_seen": 27880720,
      "step": 12910
    },
    {
      "epoch": 2.106851549755302,
      "grad_norm": 4.943325519561768,
      "learning_rate": 4.81576683768774e-05,
      "loss": 0.3171,
      "num_input_tokens_seen": 27892048,
      "step": 12915
    },
    {
      "epoch": 2.107667210440457,
      "grad_norm": 0.42248857021331787,
      "learning_rate": 4.81549855942097e-05,
      "loss": 0.0241,
      "num_input_tokens_seen": 27903760,
      "step": 12920
    },
    {
      "epoch": 2.1084828711256116,
      "grad_norm": 0.08590734750032425,
      "learning_rate": 4.8152300934481384e-05,
      "loss": 0.1001,
      "num_input_tokens_seen": 27914192,
      "step": 12925
    },
    {
      "epoch": 2.109298531810767,
      "grad_norm": 0.6834118366241455,
      "learning_rate": 4.8149614397910094e-05,
      "loss": 0.0145,
      "num_input_tokens_seen": 27925936,
      "step": 12930
    },
    {
      "epoch": 2.1101141924959217,
      "grad_norm": 1.6590182781219482,
      "learning_rate": 4.8146925984713585e-05,
      "loss": 0.29,
      "num_input_tokens_seen": 27937392,
      "step": 12935
    },
    {
      "epoch": 2.1109298531810765,
      "grad_norm": 0.3167704939842224,
      "learning_rate": 4.814423569510981e-05,
      "loss": 0.0985,
      "num_input_tokens_seen": 27948304,
      "step": 12940
    },
    {
      "epoch": 2.1117455138662318,
      "grad_norm": 3.9980692863464355,
      "learning_rate": 4.814154352931687e-05,
      "loss": 0.0544,
      "num_input_tokens_seen": 27959344,
      "step": 12945
    },
    {
      "epoch": 2.1125611745513866,
      "grad_norm": 3.7122998237609863,
      "learning_rate": 4.813884948755298e-05,
      "loss": 0.3088,
      "num_input_tokens_seen": 27970032,
      "step": 12950
    },
    {
      "epoch": 2.1133768352365414,
      "grad_norm": 0.3176769018173218,
      "learning_rate": 4.8136153570036544e-05,
      "loss": 0.0613,
      "num_input_tokens_seen": 27979984,
      "step": 12955
    },
    {
      "epoch": 2.1141924959216967,
      "grad_norm": 0.08330029249191284,
      "learning_rate": 4.8133455776986114e-05,
      "loss": 0.1169,
      "num_input_tokens_seen": 27989200,
      "step": 12960
    },
    {
      "epoch": 2.1150081566068515,
      "grad_norm": 0.30361437797546387,
      "learning_rate": 4.813075610862038e-05,
      "loss": 0.1103,
      "num_input_tokens_seen": 28000432,
      "step": 12965
    },
    {
      "epoch": 2.1158238172920063,
      "grad_norm": 0.9845252633094788,
      "learning_rate": 4.8128054565158196e-05,
      "loss": 0.0299,
      "num_input_tokens_seen": 28010064,
      "step": 12970
    },
    {
      "epoch": 2.1166394779771616,
      "grad_norm": 2.610048532485962,
      "learning_rate": 4.8125351146818556e-05,
      "loss": 0.1835,
      "num_input_tokens_seen": 28021776,
      "step": 12975
    },
    {
      "epoch": 2.1174551386623164,
      "grad_norm": 0.10616115480661392,
      "learning_rate": 4.8122645853820604e-05,
      "loss": 0.2588,
      "num_input_tokens_seen": 28032656,
      "step": 12980
    },
    {
      "epoch": 2.1182707993474716,
      "grad_norm": 0.028080087155103683,
      "learning_rate": 4.811993868638367e-05,
      "loss": 0.0717,
      "num_input_tokens_seen": 28043728,
      "step": 12985
    },
    {
      "epoch": 2.1190864600326265,
      "grad_norm": 0.18192705512046814,
      "learning_rate": 4.811722964472719e-05,
      "loss": 0.2153,
      "num_input_tokens_seen": 28055120,
      "step": 12990
    },
    {
      "epoch": 2.1199021207177813,
      "grad_norm": 6.684364318847656,
      "learning_rate": 4.811451872907078e-05,
      "loss": 0.1343,
      "num_input_tokens_seen": 28065936,
      "step": 12995
    },
    {
      "epoch": 2.1207177814029365,
      "grad_norm": 5.173242092132568,
      "learning_rate": 4.8111805939634204e-05,
      "loss": 0.1583,
      "num_input_tokens_seen": 28077040,
      "step": 13000
    },
    {
      "epoch": 2.1215334420880914,
      "grad_norm": 1.8242324590682983,
      "learning_rate": 4.810909127663736e-05,
      "loss": 0.0703,
      "num_input_tokens_seen": 28088624,
      "step": 13005
    },
    {
      "epoch": 2.122349102773246,
      "grad_norm": 3.3235926628112793,
      "learning_rate": 4.810637474030033e-05,
      "loss": 0.215,
      "num_input_tokens_seen": 28097520,
      "step": 13010
    },
    {
      "epoch": 2.1231647634584014,
      "grad_norm": 0.15635597705841064,
      "learning_rate": 4.810365633084333e-05,
      "loss": 0.3075,
      "num_input_tokens_seen": 28107504,
      "step": 13015
    },
    {
      "epoch": 2.1239804241435563,
      "grad_norm": 4.994592189788818,
      "learning_rate": 4.810093604848671e-05,
      "loss": 0.2109,
      "num_input_tokens_seen": 28118352,
      "step": 13020
    },
    {
      "epoch": 2.124796084828711,
      "grad_norm": 1.1246777772903442,
      "learning_rate": 4.8098213893451005e-05,
      "loss": 0.1026,
      "num_input_tokens_seen": 28129936,
      "step": 13025
    },
    {
      "epoch": 2.1256117455138663,
      "grad_norm": 2.986356019973755,
      "learning_rate": 4.809548986595688e-05,
      "loss": 0.1595,
      "num_input_tokens_seen": 28140496,
      "step": 13030
    },
    {
      "epoch": 2.126427406199021,
      "grad_norm": 4.179571628570557,
      "learning_rate": 4.809276396622516e-05,
      "loss": 0.1303,
      "num_input_tokens_seen": 28151440,
      "step": 13035
    },
    {
      "epoch": 2.1272430668841764,
      "grad_norm": 1.1820942163467407,
      "learning_rate": 4.809003619447683e-05,
      "loss": 0.1196,
      "num_input_tokens_seen": 28161200,
      "step": 13040
    },
    {
      "epoch": 2.1280587275693312,
      "grad_norm": 4.96096134185791,
      "learning_rate": 4.8087306550932996e-05,
      "loss": 0.058,
      "num_input_tokens_seen": 28171568,
      "step": 13045
    },
    {
      "epoch": 2.128874388254486,
      "grad_norm": 3.9760048389434814,
      "learning_rate": 4.808457503581496e-05,
      "loss": 0.1363,
      "num_input_tokens_seen": 28183344,
      "step": 13050
    },
    {
      "epoch": 2.1296900489396413,
      "grad_norm": 0.23450660705566406,
      "learning_rate": 4.808184164934414e-05,
      "loss": 0.0445,
      "num_input_tokens_seen": 28193552,
      "step": 13055
    },
    {
      "epoch": 2.130505709624796,
      "grad_norm": 0.19968116283416748,
      "learning_rate": 4.8079106391742115e-05,
      "loss": 0.1831,
      "num_input_tokens_seen": 28203600,
      "step": 13060
    },
    {
      "epoch": 2.131321370309951,
      "grad_norm": 1.3951623439788818,
      "learning_rate": 4.807636926323063e-05,
      "loss": 0.4211,
      "num_input_tokens_seen": 28213584,
      "step": 13065
    },
    {
      "epoch": 2.132137030995106,
      "grad_norm": 0.18396364152431488,
      "learning_rate": 4.8073630264031556e-05,
      "loss": 0.0669,
      "num_input_tokens_seen": 28224016,
      "step": 13070
    },
    {
      "epoch": 2.132952691680261,
      "grad_norm": 0.12926486134529114,
      "learning_rate": 4.807088939436695e-05,
      "loss": 0.1624,
      "num_input_tokens_seen": 28235760,
      "step": 13075
    },
    {
      "epoch": 2.133768352365416,
      "grad_norm": 0.3938828110694885,
      "learning_rate": 4.806814665445898e-05,
      "loss": 0.162,
      "num_input_tokens_seen": 28244656,
      "step": 13080
    },
    {
      "epoch": 2.134584013050571,
      "grad_norm": 1.4899837970733643,
      "learning_rate": 4.8065402044529994e-05,
      "loss": 0.022,
      "num_input_tokens_seen": 28254096,
      "step": 13085
    },
    {
      "epoch": 2.135399673735726,
      "grad_norm": 2.1409873962402344,
      "learning_rate": 4.806265556480249e-05,
      "loss": 0.0767,
      "num_input_tokens_seen": 28264880,
      "step": 13090
    },
    {
      "epoch": 2.1362153344208807,
      "grad_norm": 3.346529722213745,
      "learning_rate": 4.80599072154991e-05,
      "loss": 0.1405,
      "num_input_tokens_seen": 28274448,
      "step": 13095
    },
    {
      "epoch": 2.137030995106036,
      "grad_norm": 0.546467661857605,
      "learning_rate": 4.805715699684264e-05,
      "loss": 0.0342,
      "num_input_tokens_seen": 28285776,
      "step": 13100
    },
    {
      "epoch": 2.137846655791191,
      "grad_norm": 5.533677577972412,
      "learning_rate": 4.8054404909056036e-05,
      "loss": 0.0847,
      "num_input_tokens_seen": 28296016,
      "step": 13105
    },
    {
      "epoch": 2.1386623164763456,
      "grad_norm": 3.5788450241088867,
      "learning_rate": 4.805165095236239e-05,
      "loss": 0.2059,
      "num_input_tokens_seen": 28307952,
      "step": 13110
    },
    {
      "epoch": 2.139477977161501,
      "grad_norm": 1.0270748138427734,
      "learning_rate": 4.804889512698496e-05,
      "loss": 0.1529,
      "num_input_tokens_seen": 28318960,
      "step": 13115
    },
    {
      "epoch": 2.1402936378466557,
      "grad_norm": 0.12226854264736176,
      "learning_rate": 4.804613743314714e-05,
      "loss": 0.0078,
      "num_input_tokens_seen": 28329136,
      "step": 13120
    },
    {
      "epoch": 2.141109298531811,
      "grad_norm": 0.07593917101621628,
      "learning_rate": 4.804337787107248e-05,
      "loss": 0.0218,
      "num_input_tokens_seen": 28339696,
      "step": 13125
    },
    {
      "epoch": 2.141924959216966,
      "grad_norm": 0.5602729916572571,
      "learning_rate": 4.80406164409847e-05,
      "loss": 0.1575,
      "num_input_tokens_seen": 28351440,
      "step": 13130
    },
    {
      "epoch": 2.1427406199021206,
      "grad_norm": 0.35029932856559753,
      "learning_rate": 4.8037853143107634e-05,
      "loss": 0.1944,
      "num_input_tokens_seen": 28361232,
      "step": 13135
    },
    {
      "epoch": 2.143556280587276,
      "grad_norm": 0.10239975154399872,
      "learning_rate": 4.8035087977665304e-05,
      "loss": 0.3515,
      "num_input_tokens_seen": 28370864,
      "step": 13140
    },
    {
      "epoch": 2.1443719412724307,
      "grad_norm": 0.37655508518218994,
      "learning_rate": 4.803232094488186e-05,
      "loss": 0.0357,
      "num_input_tokens_seen": 28381456,
      "step": 13145
    },
    {
      "epoch": 2.1451876019575855,
      "grad_norm": 0.0940922424197197,
      "learning_rate": 4.802955204498162e-05,
      "loss": 0.0229,
      "num_input_tokens_seen": 28391600,
      "step": 13150
    },
    {
      "epoch": 2.1460032626427408,
      "grad_norm": 1.0168960094451904,
      "learning_rate": 4.802678127818904e-05,
      "loss": 0.1323,
      "num_input_tokens_seen": 28402640,
      "step": 13155
    },
    {
      "epoch": 2.1468189233278956,
      "grad_norm": 3.733787775039673,
      "learning_rate": 4.802400864472873e-05,
      "loss": 0.3108,
      "num_input_tokens_seen": 28412272,
      "step": 13160
    },
    {
      "epoch": 2.1476345840130504,
      "grad_norm": 0.9021793007850647,
      "learning_rate": 4.8021234144825456e-05,
      "loss": 0.1442,
      "num_input_tokens_seen": 28423312,
      "step": 13165
    },
    {
      "epoch": 2.1484502446982057,
      "grad_norm": 1.0865229368209839,
      "learning_rate": 4.801845777870414e-05,
      "loss": 0.2975,
      "num_input_tokens_seen": 28433424,
      "step": 13170
    },
    {
      "epoch": 2.1492659053833605,
      "grad_norm": 3.830258846282959,
      "learning_rate": 4.801567954658984e-05,
      "loss": 0.2252,
      "num_input_tokens_seen": 28443984,
      "step": 13175
    },
    {
      "epoch": 2.1500815660685153,
      "grad_norm": 0.3991985619068146,
      "learning_rate": 4.801289944870777e-05,
      "loss": 0.1898,
      "num_input_tokens_seen": 28454832,
      "step": 13180
    },
    {
      "epoch": 2.1508972267536706,
      "grad_norm": 0.0991239994764328,
      "learning_rate": 4.8010117485283305e-05,
      "loss": 0.1274,
      "num_input_tokens_seen": 28467280,
      "step": 13185
    },
    {
      "epoch": 2.1517128874388254,
      "grad_norm": 1.4642499685287476,
      "learning_rate": 4.800733365654197e-05,
      "loss": 0.217,
      "num_input_tokens_seen": 28478128,
      "step": 13190
    },
    {
      "epoch": 2.15252854812398,
      "grad_norm": 2.121992588043213,
      "learning_rate": 4.8004547962709424e-05,
      "loss": 0.1124,
      "num_input_tokens_seen": 28489200,
      "step": 13195
    },
    {
      "epoch": 2.1533442088091355,
      "grad_norm": 0.3353828191757202,
      "learning_rate": 4.80017604040115e-05,
      "loss": 0.0315,
      "num_input_tokens_seen": 28500240,
      "step": 13200
    },
    {
      "epoch": 2.1541598694942903,
      "grad_norm": 2.9096145629882812,
      "learning_rate": 4.799897098067417e-05,
      "loss": 0.0922,
      "num_input_tokens_seen": 28511824,
      "step": 13205
    },
    {
      "epoch": 2.1549755301794455,
      "grad_norm": 0.10797697305679321,
      "learning_rate": 4.799617969292355e-05,
      "loss": 0.095,
      "num_input_tokens_seen": 28521456,
      "step": 13210
    },
    {
      "epoch": 2.1557911908646004,
      "grad_norm": 3.181422710418701,
      "learning_rate": 4.799338654098593e-05,
      "loss": 0.1553,
      "num_input_tokens_seen": 28531888,
      "step": 13215
    },
    {
      "epoch": 2.156606851549755,
      "grad_norm": 0.06177400425076485,
      "learning_rate": 4.799059152508773e-05,
      "loss": 0.053,
      "num_input_tokens_seen": 28541616,
      "step": 13220
    },
    {
      "epoch": 2.1574225122349104,
      "grad_norm": 3.903703451156616,
      "learning_rate": 4.798779464545552e-05,
      "loss": 0.1561,
      "num_input_tokens_seen": 28553776,
      "step": 13225
    },
    {
      "epoch": 2.1582381729200653,
      "grad_norm": 3.8479654788970947,
      "learning_rate": 4.7984995902316045e-05,
      "loss": 0.1142,
      "num_input_tokens_seen": 28565680,
      "step": 13230
    },
    {
      "epoch": 2.15905383360522,
      "grad_norm": 5.2140278816223145,
      "learning_rate": 4.798219529589618e-05,
      "loss": 0.1837,
      "num_input_tokens_seen": 28576560,
      "step": 13235
    },
    {
      "epoch": 2.1598694942903753,
      "grad_norm": 0.19050484895706177,
      "learning_rate": 4.797939282642294e-05,
      "loss": 0.2851,
      "num_input_tokens_seen": 28586352,
      "step": 13240
    },
    {
      "epoch": 2.16068515497553,
      "grad_norm": 3.2972590923309326,
      "learning_rate": 4.797658849412353e-05,
      "loss": 0.3033,
      "num_input_tokens_seen": 28598416,
      "step": 13245
    },
    {
      "epoch": 2.161500815660685,
      "grad_norm": 0.08232977241277695,
      "learning_rate": 4.797378229922528e-05,
      "loss": 0.1517,
      "num_input_tokens_seen": 28609424,
      "step": 13250
    },
    {
      "epoch": 2.1623164763458402,
      "grad_norm": 1.3809925317764282,
      "learning_rate": 4.797097424195566e-05,
      "loss": 0.0568,
      "num_input_tokens_seen": 28619184,
      "step": 13255
    },
    {
      "epoch": 2.163132137030995,
      "grad_norm": 1.9804331064224243,
      "learning_rate": 4.796816432254232e-05,
      "loss": 0.1551,
      "num_input_tokens_seen": 28629232,
      "step": 13260
    },
    {
      "epoch": 2.1639477977161503,
      "grad_norm": 2.687870979309082,
      "learning_rate": 4.796535254121304e-05,
      "loss": 0.0535,
      "num_input_tokens_seen": 28639408,
      "step": 13265
    },
    {
      "epoch": 2.164763458401305,
      "grad_norm": 0.5376800894737244,
      "learning_rate": 4.7962538898195754e-05,
      "loss": 0.0974,
      "num_input_tokens_seen": 28650736,
      "step": 13270
    },
    {
      "epoch": 2.16557911908646,
      "grad_norm": 4.385888576507568,
      "learning_rate": 4.7959723393718556e-05,
      "loss": 0.0901,
      "num_input_tokens_seen": 28660784,
      "step": 13275
    },
    {
      "epoch": 2.166394779771615,
      "grad_norm": 3.2208805084228516,
      "learning_rate": 4.7956906028009683e-05,
      "loss": 0.1131,
      "num_input_tokens_seen": 28671248,
      "step": 13280
    },
    {
      "epoch": 2.16721044045677,
      "grad_norm": 3.118640422821045,
      "learning_rate": 4.795408680129753e-05,
      "loss": 0.1467,
      "num_input_tokens_seen": 28682000,
      "step": 13285
    },
    {
      "epoch": 2.168026101141925,
      "grad_norm": 2.172536611557007,
      "learning_rate": 4.795126571381062e-05,
      "loss": 0.2531,
      "num_input_tokens_seen": 28692912,
      "step": 13290
    },
    {
      "epoch": 2.16884176182708,
      "grad_norm": 5.100015640258789,
      "learning_rate": 4.794844276577767e-05,
      "loss": 0.2147,
      "num_input_tokens_seen": 28703504,
      "step": 13295
    },
    {
      "epoch": 2.169657422512235,
      "grad_norm": 0.16371570527553558,
      "learning_rate": 4.794561795742751e-05,
      "loss": 0.0312,
      "num_input_tokens_seen": 28714416,
      "step": 13300
    },
    {
      "epoch": 2.1704730831973897,
      "grad_norm": 3.3605289459228516,
      "learning_rate": 4.794279128898913e-05,
      "loss": 0.1119,
      "num_input_tokens_seen": 28725200,
      "step": 13305
    },
    {
      "epoch": 2.171288743882545,
      "grad_norm": 0.49802085757255554,
      "learning_rate": 4.7939962760691675e-05,
      "loss": 0.0897,
      "num_input_tokens_seen": 28736176,
      "step": 13310
    },
    {
      "epoch": 2.1721044045677,
      "grad_norm": 0.26157346367836,
      "learning_rate": 4.793713237276445e-05,
      "loss": 0.0249,
      "num_input_tokens_seen": 28746928,
      "step": 13315
    },
    {
      "epoch": 2.1729200652528546,
      "grad_norm": 3.901275873184204,
      "learning_rate": 4.7934300125436885e-05,
      "loss": 0.144,
      "num_input_tokens_seen": 28757584,
      "step": 13320
    },
    {
      "epoch": 2.17373572593801,
      "grad_norm": 2.934119939804077,
      "learning_rate": 4.7931466018938586e-05,
      "loss": 0.0525,
      "num_input_tokens_seen": 28766224,
      "step": 13325
    },
    {
      "epoch": 2.1745513866231647,
      "grad_norm": 0.10597917437553406,
      "learning_rate": 4.79286300534993e-05,
      "loss": 0.0311,
      "num_input_tokens_seen": 28775984,
      "step": 13330
    },
    {
      "epoch": 2.1753670473083195,
      "grad_norm": 0.04173220321536064,
      "learning_rate": 4.792579222934892e-05,
      "loss": 0.1027,
      "num_input_tokens_seen": 28787056,
      "step": 13335
    },
    {
      "epoch": 2.176182707993475,
      "grad_norm": 1.1457356214523315,
      "learning_rate": 4.79229525467175e-05,
      "loss": 0.0717,
      "num_input_tokens_seen": 28797648,
      "step": 13340
    },
    {
      "epoch": 2.1769983686786296,
      "grad_norm": 0.10137543827295303,
      "learning_rate": 4.792011100583524e-05,
      "loss": 0.0495,
      "num_input_tokens_seen": 28808656,
      "step": 13345
    },
    {
      "epoch": 2.177814029363785,
      "grad_norm": 4.8282999992370605,
      "learning_rate": 4.791726760693248e-05,
      "loss": 0.0331,
      "num_input_tokens_seen": 28818448,
      "step": 13350
    },
    {
      "epoch": 2.1786296900489397,
      "grad_norm": 6.721035480499268,
      "learning_rate": 4.791442235023974e-05,
      "loss": 0.1084,
      "num_input_tokens_seen": 28828752,
      "step": 13355
    },
    {
      "epoch": 2.1794453507340945,
      "grad_norm": 3.9628283977508545,
      "learning_rate": 4.7911575235987644e-05,
      "loss": 0.2347,
      "num_input_tokens_seen": 28838896,
      "step": 13360
    },
    {
      "epoch": 2.1802610114192498,
      "grad_norm": 0.14923158288002014,
      "learning_rate": 4.790872626440701e-05,
      "loss": 0.1262,
      "num_input_tokens_seen": 28848880,
      "step": 13365
    },
    {
      "epoch": 2.1810766721044046,
      "grad_norm": 5.65922737121582,
      "learning_rate": 4.790587543572879e-05,
      "loss": 0.4029,
      "num_input_tokens_seen": 28860592,
      "step": 13370
    },
    {
      "epoch": 2.1818923327895594,
      "grad_norm": 0.25351399183273315,
      "learning_rate": 4.790302275018408e-05,
      "loss": 0.1504,
      "num_input_tokens_seen": 28872432,
      "step": 13375
    },
    {
      "epoch": 2.1827079934747147,
      "grad_norm": 0.8721989393234253,
      "learning_rate": 4.790016820800414e-05,
      "loss": 0.0133,
      "num_input_tokens_seen": 28883856,
      "step": 13380
    },
    {
      "epoch": 2.1835236541598695,
      "grad_norm": 0.45588570833206177,
      "learning_rate": 4.789731180942037e-05,
      "loss": 0.3111,
      "num_input_tokens_seen": 28894384,
      "step": 13385
    },
    {
      "epoch": 2.1843393148450243,
      "grad_norm": 5.443912506103516,
      "learning_rate": 4.7894453554664325e-05,
      "loss": 0.1183,
      "num_input_tokens_seen": 28902288,
      "step": 13390
    },
    {
      "epoch": 2.1851549755301796,
      "grad_norm": 3.1598997116088867,
      "learning_rate": 4.7891593443967706e-05,
      "loss": 0.1033,
      "num_input_tokens_seen": 28913296,
      "step": 13395
    },
    {
      "epoch": 2.1859706362153344,
      "grad_norm": 0.06592902541160583,
      "learning_rate": 4.788873147756238e-05,
      "loss": 0.1292,
      "num_input_tokens_seen": 28923888,
      "step": 13400
    },
    {
      "epoch": 2.186786296900489,
      "grad_norm": 2.075955390930176,
      "learning_rate": 4.788586765568034e-05,
      "loss": 0.3123,
      "num_input_tokens_seen": 28934640,
      "step": 13405
    },
    {
      "epoch": 2.1876019575856445,
      "grad_norm": 0.06410079449415207,
      "learning_rate": 4.788300197855374e-05,
      "loss": 0.1714,
      "num_input_tokens_seen": 28945648,
      "step": 13410
    },
    {
      "epoch": 2.1884176182707993,
      "grad_norm": 5.264660835266113,
      "learning_rate": 4.788013444641491e-05,
      "loss": 0.0807,
      "num_input_tokens_seen": 28957392,
      "step": 13415
    },
    {
      "epoch": 2.189233278955954,
      "grad_norm": 0.2360692024230957,
      "learning_rate": 4.7877265059496266e-05,
      "loss": 0.0486,
      "num_input_tokens_seen": 28967056,
      "step": 13420
    },
    {
      "epoch": 2.1900489396411094,
      "grad_norm": 0.11466488987207413,
      "learning_rate": 4.7874393818030456e-05,
      "loss": 0.2175,
      "num_input_tokens_seen": 28977808,
      "step": 13425
    },
    {
      "epoch": 2.190864600326264,
      "grad_norm": 0.20892131328582764,
      "learning_rate": 4.7871520722250214e-05,
      "loss": 0.1056,
      "num_input_tokens_seen": 28987440,
      "step": 13430
    },
    {
      "epoch": 2.1916802610114194,
      "grad_norm": 7.450143814086914,
      "learning_rate": 4.786864577238845e-05,
      "loss": 0.0841,
      "num_input_tokens_seen": 28997392,
      "step": 13435
    },
    {
      "epoch": 2.1924959216965743,
      "grad_norm": 1.5859612226486206,
      "learning_rate": 4.7865768968678226e-05,
      "loss": 0.2124,
      "num_input_tokens_seen": 29007824,
      "step": 13440
    },
    {
      "epoch": 2.193311582381729,
      "grad_norm": 0.43595489859580994,
      "learning_rate": 4.786289031135275e-05,
      "loss": 0.13,
      "num_input_tokens_seen": 29019024,
      "step": 13445
    },
    {
      "epoch": 2.1941272430668843,
      "grad_norm": 5.910229682922363,
      "learning_rate": 4.786000980064538e-05,
      "loss": 0.1705,
      "num_input_tokens_seen": 29031600,
      "step": 13450
    },
    {
      "epoch": 2.194942903752039,
      "grad_norm": 5.994784355163574,
      "learning_rate": 4.785712743678963e-05,
      "loss": 0.1151,
      "num_input_tokens_seen": 29043056,
      "step": 13455
    },
    {
      "epoch": 2.195758564437194,
      "grad_norm": 2.1848740577697754,
      "learning_rate": 4.785424322001915e-05,
      "loss": 0.1138,
      "num_input_tokens_seen": 29053648,
      "step": 13460
    },
    {
      "epoch": 2.1965742251223492,
      "grad_norm": 2.133012533187866,
      "learning_rate": 4.785135715056775e-05,
      "loss": 0.043,
      "num_input_tokens_seen": 29064112,
      "step": 13465
    },
    {
      "epoch": 2.197389885807504,
      "grad_norm": 0.09406670928001404,
      "learning_rate": 4.78484692286694e-05,
      "loss": 0.0183,
      "num_input_tokens_seen": 29075216,
      "step": 13470
    },
    {
      "epoch": 2.198205546492659,
      "grad_norm": 4.004861354827881,
      "learning_rate": 4.7845579454558196e-05,
      "loss": 0.173,
      "num_input_tokens_seen": 29084368,
      "step": 13475
    },
    {
      "epoch": 2.199021207177814,
      "grad_norm": 0.1620786041021347,
      "learning_rate": 4.784268782846841e-05,
      "loss": 0.0272,
      "num_input_tokens_seen": 29096016,
      "step": 13480
    },
    {
      "epoch": 2.199836867862969,
      "grad_norm": 1.8775203227996826,
      "learning_rate": 4.783979435063445e-05,
      "loss": 0.1249,
      "num_input_tokens_seen": 29106768,
      "step": 13485
    },
    {
      "epoch": 2.200652528548124,
      "grad_norm": 5.59013557434082,
      "learning_rate": 4.783689902129086e-05,
      "loss": 0.2208,
      "num_input_tokens_seen": 29116560,
      "step": 13490
    },
    {
      "epoch": 2.201468189233279,
      "grad_norm": 1.0871704816818237,
      "learning_rate": 4.783400184067237e-05,
      "loss": 0.1266,
      "num_input_tokens_seen": 29127376,
      "step": 13495
    },
    {
      "epoch": 2.202283849918434,
      "grad_norm": 0.44361236691474915,
      "learning_rate": 4.783110280901383e-05,
      "loss": 0.1347,
      "num_input_tokens_seen": 29139088,
      "step": 13500
    },
    {
      "epoch": 2.203099510603589,
      "grad_norm": 0.11101436614990234,
      "learning_rate": 4.7828201926550245e-05,
      "loss": 0.0708,
      "num_input_tokens_seen": 29149488,
      "step": 13505
    },
    {
      "epoch": 2.203915171288744,
      "grad_norm": 0.06615414470434189,
      "learning_rate": 4.7825299193516794e-05,
      "loss": 0.0471,
      "num_input_tokens_seen": 29159728,
      "step": 13510
    },
    {
      "epoch": 2.2047308319738987,
      "grad_norm": 1.3377586603164673,
      "learning_rate": 4.782239461014877e-05,
      "loss": 0.1196,
      "num_input_tokens_seen": 29170928,
      "step": 13515
    },
    {
      "epoch": 2.205546492659054,
      "grad_norm": 4.223592281341553,
      "learning_rate": 4.781948817668164e-05,
      "loss": 0.2729,
      "num_input_tokens_seen": 29180912,
      "step": 13520
    },
    {
      "epoch": 2.206362153344209,
      "grad_norm": 0.12885291874408722,
      "learning_rate": 4.7816579893351014e-05,
      "loss": 0.1262,
      "num_input_tokens_seen": 29191216,
      "step": 13525
    },
    {
      "epoch": 2.2071778140293636,
      "grad_norm": 0.08625456690788269,
      "learning_rate": 4.781366976039265e-05,
      "loss": 0.0098,
      "num_input_tokens_seen": 29201424,
      "step": 13530
    },
    {
      "epoch": 2.207993474714519,
      "grad_norm": 0.11553726345300674,
      "learning_rate": 4.781075777804246e-05,
      "loss": 0.1708,
      "num_input_tokens_seen": 29211152,
      "step": 13535
    },
    {
      "epoch": 2.2088091353996737,
      "grad_norm": 3.9450199604034424,
      "learning_rate": 4.7807843946536514e-05,
      "loss": 0.2024,
      "num_input_tokens_seen": 29221744,
      "step": 13540
    },
    {
      "epoch": 2.2096247960848285,
      "grad_norm": 0.16982929408550262,
      "learning_rate": 4.7804928266110996e-05,
      "loss": 0.3198,
      "num_input_tokens_seen": 29231568,
      "step": 13545
    },
    {
      "epoch": 2.210440456769984,
      "grad_norm": 7.157938003540039,
      "learning_rate": 4.780201073700229e-05,
      "loss": 0.5554,
      "num_input_tokens_seen": 29242128,
      "step": 13550
    },
    {
      "epoch": 2.2112561174551386,
      "grad_norm": 0.3418239653110504,
      "learning_rate": 4.7799091359446905e-05,
      "loss": 0.1646,
      "num_input_tokens_seen": 29250992,
      "step": 13555
    },
    {
      "epoch": 2.2120717781402934,
      "grad_norm": 1.7065843343734741,
      "learning_rate": 4.779617013368148e-05,
      "loss": 0.0285,
      "num_input_tokens_seen": 29261072,
      "step": 13560
    },
    {
      "epoch": 2.2128874388254487,
      "grad_norm": 0.3221491873264313,
      "learning_rate": 4.7793247059942845e-05,
      "loss": 0.1075,
      "num_input_tokens_seen": 29271760,
      "step": 13565
    },
    {
      "epoch": 2.2137030995106035,
      "grad_norm": 0.3167019486427307,
      "learning_rate": 4.779032213846795e-05,
      "loss": 0.1779,
      "num_input_tokens_seen": 29282480,
      "step": 13570
    },
    {
      "epoch": 2.2145187601957588,
      "grad_norm": 1.490031361579895,
      "learning_rate": 4.77873953694939e-05,
      "loss": 0.0974,
      "num_input_tokens_seen": 29292912,
      "step": 13575
    },
    {
      "epoch": 2.2153344208809136,
      "grad_norm": 0.22250661253929138,
      "learning_rate": 4.778446675325796e-05,
      "loss": 0.1102,
      "num_input_tokens_seen": 29302800,
      "step": 13580
    },
    {
      "epoch": 2.2161500815660684,
      "grad_norm": 0.31745657324790955,
      "learning_rate": 4.778153628999754e-05,
      "loss": 0.055,
      "num_input_tokens_seen": 29313424,
      "step": 13585
    },
    {
      "epoch": 2.2169657422512237,
      "grad_norm": 2.743222236633301,
      "learning_rate": 4.7778603979950196e-05,
      "loss": 0.2179,
      "num_input_tokens_seen": 29323856,
      "step": 13590
    },
    {
      "epoch": 2.2177814029363785,
      "grad_norm": 0.0542568676173687,
      "learning_rate": 4.777566982335364e-05,
      "loss": 0.0383,
      "num_input_tokens_seen": 29334832,
      "step": 13595
    },
    {
      "epoch": 2.2185970636215333,
      "grad_norm": 0.03890717774629593,
      "learning_rate": 4.777273382044572e-05,
      "loss": 0.0706,
      "num_input_tokens_seen": 29345936,
      "step": 13600
    },
    {
      "epoch": 2.2194127243066886,
      "grad_norm": 3.6464173793792725,
      "learning_rate": 4.7769795971464456e-05,
      "loss": 0.1469,
      "num_input_tokens_seen": 29356560,
      "step": 13605
    },
    {
      "epoch": 2.2202283849918434,
      "grad_norm": 0.06402415037155151,
      "learning_rate": 4.7766856276647986e-05,
      "loss": 0.045,
      "num_input_tokens_seen": 29367024,
      "step": 13610
    },
    {
      "epoch": 2.221044045676998,
      "grad_norm": 2.0355734825134277,
      "learning_rate": 4.776391473623464e-05,
      "loss": 0.0548,
      "num_input_tokens_seen": 29376944,
      "step": 13615
    },
    {
      "epoch": 2.2218597063621535,
      "grad_norm": 4.244828224182129,
      "learning_rate": 4.7760971350462856e-05,
      "loss": 0.2537,
      "num_input_tokens_seen": 29389552,
      "step": 13620
    },
    {
      "epoch": 2.2226753670473083,
      "grad_norm": 0.0875852108001709,
      "learning_rate": 4.775802611957125e-05,
      "loss": 0.014,
      "num_input_tokens_seen": 29399600,
      "step": 13625
    },
    {
      "epoch": 2.223491027732463,
      "grad_norm": 3.1022586822509766,
      "learning_rate": 4.7755079043798565e-05,
      "loss": 0.0613,
      "num_input_tokens_seen": 29410896,
      "step": 13630
    },
    {
      "epoch": 2.2243066884176184,
      "grad_norm": 2.86226224899292,
      "learning_rate": 4.775213012338373e-05,
      "loss": 0.1702,
      "num_input_tokens_seen": 29421136,
      "step": 13635
    },
    {
      "epoch": 2.225122349102773,
      "grad_norm": 2.082360029220581,
      "learning_rate": 4.774917935856577e-05,
      "loss": 0.1033,
      "num_input_tokens_seen": 29431536,
      "step": 13640
    },
    {
      "epoch": 2.225938009787928,
      "grad_norm": 0.041097868233919144,
      "learning_rate": 4.774622674958391e-05,
      "loss": 0.1653,
      "num_input_tokens_seen": 29442416,
      "step": 13645
    },
    {
      "epoch": 2.2267536704730833,
      "grad_norm": 3.4875710010528564,
      "learning_rate": 4.7743272296677495e-05,
      "loss": 0.0571,
      "num_input_tokens_seen": 29453232,
      "step": 13650
    },
    {
      "epoch": 2.227569331158238,
      "grad_norm": 2.4293694496154785,
      "learning_rate": 4.774031600008603e-05,
      "loss": 0.1168,
      "num_input_tokens_seen": 29464976,
      "step": 13655
    },
    {
      "epoch": 2.2283849918433933,
      "grad_norm": 3.0849180221557617,
      "learning_rate": 4.7737357860049164e-05,
      "loss": 0.1924,
      "num_input_tokens_seen": 29475536,
      "step": 13660
    },
    {
      "epoch": 2.229200652528548,
      "grad_norm": 0.1027827262878418,
      "learning_rate": 4.7734397876806704e-05,
      "loss": 0.009,
      "num_input_tokens_seen": 29486256,
      "step": 13665
    },
    {
      "epoch": 2.230016313213703,
      "grad_norm": 7.251781463623047,
      "learning_rate": 4.77314360505986e-05,
      "loss": 0.2379,
      "num_input_tokens_seen": 29498512,
      "step": 13670
    },
    {
      "epoch": 2.2308319738988582,
      "grad_norm": 3.810894727706909,
      "learning_rate": 4.772847238166495e-05,
      "loss": 0.1145,
      "num_input_tokens_seen": 29509360,
      "step": 13675
    },
    {
      "epoch": 2.231647634584013,
      "grad_norm": 0.1406639963388443,
      "learning_rate": 4.7725506870246006e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 29520400,
      "step": 13680
    },
    {
      "epoch": 2.232463295269168,
      "grad_norm": 0.7439082264900208,
      "learning_rate": 4.772253951658217e-05,
      "loss": 0.1638,
      "num_input_tokens_seen": 29532016,
      "step": 13685
    },
    {
      "epoch": 2.233278955954323,
      "grad_norm": 0.1474129557609558,
      "learning_rate": 4.771957032091398e-05,
      "loss": 0.0163,
      "num_input_tokens_seen": 29543728,
      "step": 13690
    },
    {
      "epoch": 2.234094616639478,
      "grad_norm": 2.2892446517944336,
      "learning_rate": 4.771659928348214e-05,
      "loss": 0.5025,
      "num_input_tokens_seen": 29554352,
      "step": 13695
    },
    {
      "epoch": 2.2349102773246328,
      "grad_norm": 3.372269630432129,
      "learning_rate": 4.7713626404527514e-05,
      "loss": 0.1512,
      "num_input_tokens_seen": 29565168,
      "step": 13700
    },
    {
      "epoch": 2.235725938009788,
      "grad_norm": 3.662602186203003,
      "learning_rate": 4.7710651684291074e-05,
      "loss": 0.2718,
      "num_input_tokens_seen": 29575792,
      "step": 13705
    },
    {
      "epoch": 2.236541598694943,
      "grad_norm": 0.12891605496406555,
      "learning_rate": 4.770767512301398e-05,
      "loss": 0.0985,
      "num_input_tokens_seen": 29586384,
      "step": 13710
    },
    {
      "epoch": 2.237357259380098,
      "grad_norm": 0.0943826287984848,
      "learning_rate": 4.770469672093752e-05,
      "loss": 0.0976,
      "num_input_tokens_seen": 29597232,
      "step": 13715
    },
    {
      "epoch": 2.238172920065253,
      "grad_norm": 2.561107635498047,
      "learning_rate": 4.7701716478303135e-05,
      "loss": 0.0606,
      "num_input_tokens_seen": 29606896,
      "step": 13720
    },
    {
      "epoch": 2.2389885807504077,
      "grad_norm": 0.13817407190799713,
      "learning_rate": 4.769873439535244e-05,
      "loss": 0.0217,
      "num_input_tokens_seen": 29617648,
      "step": 13725
    },
    {
      "epoch": 2.239804241435563,
      "grad_norm": 2.578217029571533,
      "learning_rate": 4.769575047232715e-05,
      "loss": 0.3705,
      "num_input_tokens_seen": 29627664,
      "step": 13730
    },
    {
      "epoch": 2.240619902120718,
      "grad_norm": 0.2209499031305313,
      "learning_rate": 4.769276470946917e-05,
      "loss": 0.0574,
      "num_input_tokens_seen": 29639088,
      "step": 13735
    },
    {
      "epoch": 2.2414355628058726,
      "grad_norm": 3.821312427520752,
      "learning_rate": 4.768977710702055e-05,
      "loss": 0.1059,
      "num_input_tokens_seen": 29649648,
      "step": 13740
    },
    {
      "epoch": 2.242251223491028,
      "grad_norm": 1.264604091644287,
      "learning_rate": 4.768678766522347e-05,
      "loss": 0.1255,
      "num_input_tokens_seen": 29661360,
      "step": 13745
    },
    {
      "epoch": 2.2430668841761827,
      "grad_norm": 0.49606001377105713,
      "learning_rate": 4.768379638432026e-05,
      "loss": 0.0849,
      "num_input_tokens_seen": 29672336,
      "step": 13750
    },
    {
      "epoch": 2.2438825448613375,
      "grad_norm": 5.561700820922852,
      "learning_rate": 4.768080326455343e-05,
      "loss": 0.2971,
      "num_input_tokens_seen": 29683472,
      "step": 13755
    },
    {
      "epoch": 2.244698205546493,
      "grad_norm": 0.08782978355884552,
      "learning_rate": 4.7677808306165596e-05,
      "loss": 0.0086,
      "num_input_tokens_seen": 29695472,
      "step": 13760
    },
    {
      "epoch": 2.2455138662316476,
      "grad_norm": 2.264838457107544,
      "learning_rate": 4.767481150939956e-05,
      "loss": 0.2447,
      "num_input_tokens_seen": 29705520,
      "step": 13765
    },
    {
      "epoch": 2.2463295269168024,
      "grad_norm": 0.361931174993515,
      "learning_rate": 4.767181287449825e-05,
      "loss": 0.0115,
      "num_input_tokens_seen": 29715824,
      "step": 13770
    },
    {
      "epoch": 2.2471451876019577,
      "grad_norm": 0.4308825135231018,
      "learning_rate": 4.766881240170475e-05,
      "loss": 0.0165,
      "num_input_tokens_seen": 29726864,
      "step": 13775
    },
    {
      "epoch": 2.2479608482871125,
      "grad_norm": 0.065828338265419,
      "learning_rate": 4.7665810091262305e-05,
      "loss": 0.0096,
      "num_input_tokens_seen": 29738128,
      "step": 13780
    },
    {
      "epoch": 2.2487765089722673,
      "grad_norm": 0.14022226631641388,
      "learning_rate": 4.766280594341428e-05,
      "loss": 0.2112,
      "num_input_tokens_seen": 29747152,
      "step": 13785
    },
    {
      "epoch": 2.2495921696574226,
      "grad_norm": 3.939537286758423,
      "learning_rate": 4.7659799958404225e-05,
      "loss": 0.1179,
      "num_input_tokens_seen": 29756912,
      "step": 13790
    },
    {
      "epoch": 2.2504078303425774,
      "grad_norm": 2.971608877182007,
      "learning_rate": 4.7656792136475804e-05,
      "loss": 0.2464,
      "num_input_tokens_seen": 29766768,
      "step": 13795
    },
    {
      "epoch": 2.2512234910277327,
      "grad_norm": 5.468246936798096,
      "learning_rate": 4.765378247787285e-05,
      "loss": 0.1061,
      "num_input_tokens_seen": 29778032,
      "step": 13800
    },
    {
      "epoch": 2.2520391517128875,
      "grad_norm": 4.380643844604492,
      "learning_rate": 4.765077098283935e-05,
      "loss": 0.3154,
      "num_input_tokens_seen": 29787408,
      "step": 13805
    },
    {
      "epoch": 2.2528548123980423,
      "grad_norm": 0.9004188179969788,
      "learning_rate": 4.764775765161943e-05,
      "loss": 0.0235,
      "num_input_tokens_seen": 29797232,
      "step": 13810
    },
    {
      "epoch": 2.2536704730831976,
      "grad_norm": 5.672367095947266,
      "learning_rate": 4.764474248445735e-05,
      "loss": 0.2793,
      "num_input_tokens_seen": 29808016,
      "step": 13815
    },
    {
      "epoch": 2.2544861337683524,
      "grad_norm": 0.7587845325469971,
      "learning_rate": 4.764172548159755e-05,
      "loss": 0.0612,
      "num_input_tokens_seen": 29818640,
      "step": 13820
    },
    {
      "epoch": 2.255301794453507,
      "grad_norm": 0.09891112148761749,
      "learning_rate": 4.7638706643284605e-05,
      "loss": 0.1374,
      "num_input_tokens_seen": 29829776,
      "step": 13825
    },
    {
      "epoch": 2.2561174551386625,
      "grad_norm": 0.09488624334335327,
      "learning_rate": 4.7635685969763225e-05,
      "loss": 0.0576,
      "num_input_tokens_seen": 29840304,
      "step": 13830
    },
    {
      "epoch": 2.2569331158238173,
      "grad_norm": 0.2625058889389038,
      "learning_rate": 4.763266346127829e-05,
      "loss": 0.3236,
      "num_input_tokens_seen": 29852048,
      "step": 13835
    },
    {
      "epoch": 2.257748776508972,
      "grad_norm": 4.30433464050293,
      "learning_rate": 4.7629639118074816e-05,
      "loss": 0.1341,
      "num_input_tokens_seen": 29861264,
      "step": 13840
    },
    {
      "epoch": 2.2585644371941274,
      "grad_norm": 0.10462737083435059,
      "learning_rate": 4.7626612940397976e-05,
      "loss": 0.2599,
      "num_input_tokens_seen": 29871984,
      "step": 13845
    },
    {
      "epoch": 2.259380097879282,
      "grad_norm": 0.9041357040405273,
      "learning_rate": 4.762358492849308e-05,
      "loss": 0.2253,
      "num_input_tokens_seen": 29882256,
      "step": 13850
    },
    {
      "epoch": 2.2601957585644374,
      "grad_norm": 3.5074238777160645,
      "learning_rate": 4.762055508260561e-05,
      "loss": 0.1098,
      "num_input_tokens_seen": 29891696,
      "step": 13855
    },
    {
      "epoch": 2.2610114192495923,
      "grad_norm": 1.1502227783203125,
      "learning_rate": 4.7617523402981155e-05,
      "loss": 0.0542,
      "num_input_tokens_seen": 29903856,
      "step": 13860
    },
    {
      "epoch": 2.261827079934747,
      "grad_norm": 0.7846003770828247,
      "learning_rate": 4.7614489889865506e-05,
      "loss": 0.0743,
      "num_input_tokens_seen": 29913520,
      "step": 13865
    },
    {
      "epoch": 2.262642740619902,
      "grad_norm": 3.0706188678741455,
      "learning_rate": 4.761145454350455e-05,
      "loss": 0.0275,
      "num_input_tokens_seen": 29924016,
      "step": 13870
    },
    {
      "epoch": 2.263458401305057,
      "grad_norm": 3.6503069400787354,
      "learning_rate": 4.760841736414437e-05,
      "loss": 0.2495,
      "num_input_tokens_seen": 29935120,
      "step": 13875
    },
    {
      "epoch": 2.264274061990212,
      "grad_norm": 4.099191188812256,
      "learning_rate": 4.760537835203116e-05,
      "loss": 0.1509,
      "num_input_tokens_seen": 29946288,
      "step": 13880
    },
    {
      "epoch": 2.2650897226753672,
      "grad_norm": 7.713541507720947,
      "learning_rate": 4.760233750741128e-05,
      "loss": 0.1661,
      "num_input_tokens_seen": 29957008,
      "step": 13885
    },
    {
      "epoch": 2.265905383360522,
      "grad_norm": 0.08547984808683395,
      "learning_rate": 4.7599294830531235e-05,
      "loss": 0.0592,
      "num_input_tokens_seen": 29966896,
      "step": 13890
    },
    {
      "epoch": 2.266721044045677,
      "grad_norm": 5.586185932159424,
      "learning_rate": 4.759625032163769e-05,
      "loss": 0.2013,
      "num_input_tokens_seen": 29976528,
      "step": 13895
    },
    {
      "epoch": 2.267536704730832,
      "grad_norm": 0.24295584857463837,
      "learning_rate": 4.7593203980977444e-05,
      "loss": 0.1003,
      "num_input_tokens_seen": 29987120,
      "step": 13900
    },
    {
      "epoch": 2.268352365415987,
      "grad_norm": 1.3970733880996704,
      "learning_rate": 4.759015580879744e-05,
      "loss": 0.0388,
      "num_input_tokens_seen": 29997680,
      "step": 13905
    },
    {
      "epoch": 2.2691680261011418,
      "grad_norm": 0.16488811373710632,
      "learning_rate": 4.758710580534479e-05,
      "loss": 0.2564,
      "num_input_tokens_seen": 30008784,
      "step": 13910
    },
    {
      "epoch": 2.269983686786297,
      "grad_norm": 0.07985883206129074,
      "learning_rate": 4.758405397086674e-05,
      "loss": 0.3888,
      "num_input_tokens_seen": 30018416,
      "step": 13915
    },
    {
      "epoch": 2.270799347471452,
      "grad_norm": 2.3947901725769043,
      "learning_rate": 4.758100030561068e-05,
      "loss": 0.111,
      "num_input_tokens_seen": 30030352,
      "step": 13920
    },
    {
      "epoch": 2.2716150081566067,
      "grad_norm": 0.1784278154373169,
      "learning_rate": 4.757794480982416e-05,
      "loss": 0.0863,
      "num_input_tokens_seen": 30041392,
      "step": 13925
    },
    {
      "epoch": 2.272430668841762,
      "grad_norm": 0.3512839674949646,
      "learning_rate": 4.757488748375487e-05,
      "loss": 0.0106,
      "num_input_tokens_seen": 30052240,
      "step": 13930
    },
    {
      "epoch": 2.2732463295269167,
      "grad_norm": 0.2251429557800293,
      "learning_rate": 4.757182832765067e-05,
      "loss": 0.2198,
      "num_input_tokens_seen": 30062192,
      "step": 13935
    },
    {
      "epoch": 2.274061990212072,
      "grad_norm": 1.2161368131637573,
      "learning_rate": 4.7568767341759526e-05,
      "loss": 0.0991,
      "num_input_tokens_seen": 30073808,
      "step": 13940
    },
    {
      "epoch": 2.274877650897227,
      "grad_norm": 0.9560844898223877,
      "learning_rate": 4.756570452632959e-05,
      "loss": 0.1643,
      "num_input_tokens_seen": 30084016,
      "step": 13945
    },
    {
      "epoch": 2.2756933115823816,
      "grad_norm": 3.6224071979522705,
      "learning_rate": 4.756263988160915e-05,
      "loss": 0.1163,
      "num_input_tokens_seen": 30095600,
      "step": 13950
    },
    {
      "epoch": 2.2765089722675365,
      "grad_norm": 2.437493085861206,
      "learning_rate": 4.755957340784664e-05,
      "loss": 0.0388,
      "num_input_tokens_seen": 30106800,
      "step": 13955
    },
    {
      "epoch": 2.2773246329526917,
      "grad_norm": 0.09992051124572754,
      "learning_rate": 4.755650510529064e-05,
      "loss": 0.0214,
      "num_input_tokens_seen": 30118288,
      "step": 13960
    },
    {
      "epoch": 2.2781402936378465,
      "grad_norm": 0.2428797483444214,
      "learning_rate": 4.755343497418989e-05,
      "loss": 0.0804,
      "num_input_tokens_seen": 30129552,
      "step": 13965
    },
    {
      "epoch": 2.278955954323002,
      "grad_norm": 0.4452281594276428,
      "learning_rate": 4.7550363014793264e-05,
      "loss": 0.1458,
      "num_input_tokens_seen": 30140720,
      "step": 13970
    },
    {
      "epoch": 2.2797716150081566,
      "grad_norm": 1.0804243087768555,
      "learning_rate": 4.754728922734979e-05,
      "loss": 0.0942,
      "num_input_tokens_seen": 30150640,
      "step": 13975
    },
    {
      "epoch": 2.2805872756933114,
      "grad_norm": 2.807777166366577,
      "learning_rate": 4.754421361210865e-05,
      "loss": 0.0585,
      "num_input_tokens_seen": 30160976,
      "step": 13980
    },
    {
      "epoch": 2.2814029363784667,
      "grad_norm": 6.057126045227051,
      "learning_rate": 4.7541136169319165e-05,
      "loss": 0.318,
      "num_input_tokens_seen": 30170576,
      "step": 13985
    },
    {
      "epoch": 2.2822185970636215,
      "grad_norm": 4.7831807136535645,
      "learning_rate": 4.7538056899230815e-05,
      "loss": 0.1246,
      "num_input_tokens_seen": 30181904,
      "step": 13990
    },
    {
      "epoch": 2.2830342577487763,
      "grad_norm": 0.5198666453361511,
      "learning_rate": 4.753497580209321e-05,
      "loss": 0.0424,
      "num_input_tokens_seen": 30193392,
      "step": 13995
    },
    {
      "epoch": 2.2838499184339316,
      "grad_norm": 0.29861927032470703,
      "learning_rate": 4.7531892878156125e-05,
      "loss": 0.0781,
      "num_input_tokens_seen": 30204528,
      "step": 14000
    },
    {
      "epoch": 2.2846655791190864,
      "grad_norm": 2.2439777851104736,
      "learning_rate": 4.752880812766948e-05,
      "loss": 0.162,
      "num_input_tokens_seen": 30215760,
      "step": 14005
    },
    {
      "epoch": 2.2854812398042412,
      "grad_norm": 0.21506105363368988,
      "learning_rate": 4.752572155088334e-05,
      "loss": 0.0174,
      "num_input_tokens_seen": 30226096,
      "step": 14010
    },
    {
      "epoch": 2.2862969004893965,
      "grad_norm": 0.14795491099357605,
      "learning_rate": 4.752263314804791e-05,
      "loss": 0.0904,
      "num_input_tokens_seen": 30236784,
      "step": 14015
    },
    {
      "epoch": 2.2871125611745513,
      "grad_norm": 0.24999599158763885,
      "learning_rate": 4.7519542919413566e-05,
      "loss": 0.1317,
      "num_input_tokens_seen": 30248432,
      "step": 14020
    },
    {
      "epoch": 2.2879282218597066,
      "grad_norm": 0.5208101868629456,
      "learning_rate": 4.751645086523081e-05,
      "loss": 0.1923,
      "num_input_tokens_seen": 30258448,
      "step": 14025
    },
    {
      "epoch": 2.2887438825448614,
      "grad_norm": 0.5559170842170715,
      "learning_rate": 4.75133569857503e-05,
      "loss": 0.1787,
      "num_input_tokens_seen": 30269424,
      "step": 14030
    },
    {
      "epoch": 2.289559543230016,
      "grad_norm": 0.18382087349891663,
      "learning_rate": 4.751026128122283e-05,
      "loss": 0.0508,
      "num_input_tokens_seen": 30280144,
      "step": 14035
    },
    {
      "epoch": 2.2903752039151715,
      "grad_norm": 0.3146985173225403,
      "learning_rate": 4.7507163751899374e-05,
      "loss": 0.0464,
      "num_input_tokens_seen": 30290064,
      "step": 14040
    },
    {
      "epoch": 2.2911908646003263,
      "grad_norm": 0.35729309916496277,
      "learning_rate": 4.750406439803102e-05,
      "loss": 0.2322,
      "num_input_tokens_seen": 30299984,
      "step": 14045
    },
    {
      "epoch": 2.292006525285481,
      "grad_norm": 0.6355782151222229,
      "learning_rate": 4.750096321986902e-05,
      "loss": 0.0146,
      "num_input_tokens_seen": 30311248,
      "step": 14050
    },
    {
      "epoch": 2.2928221859706364,
      "grad_norm": 0.1297059953212738,
      "learning_rate": 4.749786021766478e-05,
      "loss": 0.1262,
      "num_input_tokens_seen": 30322160,
      "step": 14055
    },
    {
      "epoch": 2.293637846655791,
      "grad_norm": 5.0604939460754395,
      "learning_rate": 4.749475539166983e-05,
      "loss": 0.1377,
      "num_input_tokens_seen": 30333616,
      "step": 14060
    },
    {
      "epoch": 2.294453507340946,
      "grad_norm": 7.509324073791504,
      "learning_rate": 4.749164874213588e-05,
      "loss": 0.0514,
      "num_input_tokens_seen": 30344112,
      "step": 14065
    },
    {
      "epoch": 2.2952691680261013,
      "grad_norm": 2.4683310985565186,
      "learning_rate": 4.7488540269314756e-05,
      "loss": 0.0391,
      "num_input_tokens_seen": 30354192,
      "step": 14070
    },
    {
      "epoch": 2.296084828711256,
      "grad_norm": 7.585381507873535,
      "learning_rate": 4.748542997345845e-05,
      "loss": 0.0866,
      "num_input_tokens_seen": 30365168,
      "step": 14075
    },
    {
      "epoch": 2.2969004893964113,
      "grad_norm": 2.6453585624694824,
      "learning_rate": 4.74823178548191e-05,
      "loss": 0.111,
      "num_input_tokens_seen": 30376464,
      "step": 14080
    },
    {
      "epoch": 2.297716150081566,
      "grad_norm": 0.7056415677070618,
      "learning_rate": 4.7479203913649e-05,
      "loss": 0.0497,
      "num_input_tokens_seen": 30387472,
      "step": 14085
    },
    {
      "epoch": 2.298531810766721,
      "grad_norm": 9.016286849975586,
      "learning_rate": 4.747608815020056e-05,
      "loss": 0.3355,
      "num_input_tokens_seen": 30398224,
      "step": 14090
    },
    {
      "epoch": 2.299347471451876,
      "grad_norm": 5.035734176635742,
      "learning_rate": 4.747297056472638e-05,
      "loss": 0.2694,
      "num_input_tokens_seen": 30409872,
      "step": 14095
    },
    {
      "epoch": 2.300163132137031,
      "grad_norm": 7.550227642059326,
      "learning_rate": 4.7469851157479177e-05,
      "loss": 0.1785,
      "num_input_tokens_seen": 30420144,
      "step": 14100
    },
    {
      "epoch": 2.300978792822186,
      "grad_norm": 1.597720742225647,
      "learning_rate": 4.746672992871183e-05,
      "loss": 0.0707,
      "num_input_tokens_seen": 30430704,
      "step": 14105
    },
    {
      "epoch": 2.301794453507341,
      "grad_norm": 0.23507060110569,
      "learning_rate": 4.746360687867736e-05,
      "loss": 0.1525,
      "num_input_tokens_seen": 30441328,
      "step": 14110
    },
    {
      "epoch": 2.302610114192496,
      "grad_norm": 0.27064698934555054,
      "learning_rate": 4.746048200762893e-05,
      "loss": 0.1133,
      "num_input_tokens_seen": 30452208,
      "step": 14115
    },
    {
      "epoch": 2.3034257748776508,
      "grad_norm": 0.8912692666053772,
      "learning_rate": 4.7457355315819874e-05,
      "loss": 0.1297,
      "num_input_tokens_seen": 30462960,
      "step": 14120
    },
    {
      "epoch": 2.304241435562806,
      "grad_norm": 2.2029247283935547,
      "learning_rate": 4.745422680350364e-05,
      "loss": 0.2764,
      "num_input_tokens_seen": 30473168,
      "step": 14125
    },
    {
      "epoch": 2.305057096247961,
      "grad_norm": 3.689709186553955,
      "learning_rate": 4.745109647093385e-05,
      "loss": 0.1368,
      "num_input_tokens_seen": 30483824,
      "step": 14130
    },
    {
      "epoch": 2.3058727569331157,
      "grad_norm": 1.900779366493225,
      "learning_rate": 4.744796431836428e-05,
      "loss": 0.1906,
      "num_input_tokens_seen": 30495632,
      "step": 14135
    },
    {
      "epoch": 2.306688417618271,
      "grad_norm": 2.435326337814331,
      "learning_rate": 4.7444830346048804e-05,
      "loss": 0.1039,
      "num_input_tokens_seen": 30505808,
      "step": 14140
    },
    {
      "epoch": 2.3075040783034257,
      "grad_norm": 4.353400230407715,
      "learning_rate": 4.744169455424151e-05,
      "loss": 0.0595,
      "num_input_tokens_seen": 30516368,
      "step": 14145
    },
    {
      "epoch": 2.3083197389885806,
      "grad_norm": 0.09357459843158722,
      "learning_rate": 4.7438556943196574e-05,
      "loss": 0.0658,
      "num_input_tokens_seen": 30527856,
      "step": 14150
    },
    {
      "epoch": 2.309135399673736,
      "grad_norm": 6.035260200500488,
      "learning_rate": 4.743541751316837e-05,
      "loss": 0.3764,
      "num_input_tokens_seen": 30537936,
      "step": 14155
    },
    {
      "epoch": 2.3099510603588906,
      "grad_norm": 3.8033721446990967,
      "learning_rate": 4.743227626441139e-05,
      "loss": 0.0865,
      "num_input_tokens_seen": 30548304,
      "step": 14160
    },
    {
      "epoch": 2.310766721044046,
      "grad_norm": 1.0218576192855835,
      "learning_rate": 4.7429133197180264e-05,
      "loss": 0.0135,
      "num_input_tokens_seen": 30559728,
      "step": 14165
    },
    {
      "epoch": 2.3115823817292007,
      "grad_norm": 0.36538276076316833,
      "learning_rate": 4.7425988311729805e-05,
      "loss": 0.0562,
      "num_input_tokens_seen": 30569840,
      "step": 14170
    },
    {
      "epoch": 2.3123980424143555,
      "grad_norm": 0.20794284343719482,
      "learning_rate": 4.742284160831494e-05,
      "loss": 0.1339,
      "num_input_tokens_seen": 30580400,
      "step": 14175
    },
    {
      "epoch": 2.3132137030995104,
      "grad_norm": 4.326881408691406,
      "learning_rate": 4.741969308719076e-05,
      "loss": 0.2695,
      "num_input_tokens_seen": 30594544,
      "step": 14180
    },
    {
      "epoch": 2.3140293637846656,
      "grad_norm": 0.6262915134429932,
      "learning_rate": 4.741654274861251e-05,
      "loss": 0.0165,
      "num_input_tokens_seen": 30604912,
      "step": 14185
    },
    {
      "epoch": 2.3148450244698204,
      "grad_norm": 0.602825403213501,
      "learning_rate": 4.741339059283556e-05,
      "loss": 0.02,
      "num_input_tokens_seen": 30616400,
      "step": 14190
    },
    {
      "epoch": 2.3156606851549757,
      "grad_norm": 0.15345075726509094,
      "learning_rate": 4.7410236620115444e-05,
      "loss": 0.0954,
      "num_input_tokens_seen": 30627824,
      "step": 14195
    },
    {
      "epoch": 2.3164763458401305,
      "grad_norm": 0.13649988174438477,
      "learning_rate": 4.740708083070784e-05,
      "loss": 0.0336,
      "num_input_tokens_seen": 30638320,
      "step": 14200
    },
    {
      "epoch": 2.3172920065252853,
      "grad_norm": 0.13931918144226074,
      "learning_rate": 4.740392322486858e-05,
      "loss": 0.2585,
      "num_input_tokens_seen": 30649744,
      "step": 14205
    },
    {
      "epoch": 2.3181076672104406,
      "grad_norm": 4.44880485534668,
      "learning_rate": 4.740076380285361e-05,
      "loss": 0.3618,
      "num_input_tokens_seen": 30661136,
      "step": 14210
    },
    {
      "epoch": 2.3189233278955954,
      "grad_norm": 0.4982512891292572,
      "learning_rate": 4.739760256491908e-05,
      "loss": 0.074,
      "num_input_tokens_seen": 30672368,
      "step": 14215
    },
    {
      "epoch": 2.3197389885807502,
      "grad_norm": 6.6637983322143555,
      "learning_rate": 4.7394439511321225e-05,
      "loss": 0.1824,
      "num_input_tokens_seen": 30684784,
      "step": 14220
    },
    {
      "epoch": 2.3205546492659055,
      "grad_norm": 0.8021751046180725,
      "learning_rate": 4.7391274642316485e-05,
      "loss": 0.0226,
      "num_input_tokens_seen": 30696432,
      "step": 14225
    },
    {
      "epoch": 2.3213703099510603,
      "grad_norm": 2.5145702362060547,
      "learning_rate": 4.7388107958161414e-05,
      "loss": 0.2304,
      "num_input_tokens_seen": 30706256,
      "step": 14230
    },
    {
      "epoch": 2.322185970636215,
      "grad_norm": 7.98170280456543,
      "learning_rate": 4.738493945911271e-05,
      "loss": 0.0551,
      "num_input_tokens_seen": 30716976,
      "step": 14235
    },
    {
      "epoch": 2.3230016313213704,
      "grad_norm": 3.667553663253784,
      "learning_rate": 4.738176914542723e-05,
      "loss": 0.2516,
      "num_input_tokens_seen": 30726768,
      "step": 14240
    },
    {
      "epoch": 2.323817292006525,
      "grad_norm": 7.572120666503906,
      "learning_rate": 4.737859701736199e-05,
      "loss": 0.1267,
      "num_input_tokens_seen": 30737072,
      "step": 14245
    },
    {
      "epoch": 2.3246329526916805,
      "grad_norm": 2.4298205375671387,
      "learning_rate": 4.737542307517413e-05,
      "loss": 0.1191,
      "num_input_tokens_seen": 30748400,
      "step": 14250
    },
    {
      "epoch": 2.3254486133768353,
      "grad_norm": 6.543609619140625,
      "learning_rate": 4.737224731912093e-05,
      "loss": 0.1612,
      "num_input_tokens_seen": 30760240,
      "step": 14255
    },
    {
      "epoch": 2.32626427406199,
      "grad_norm": 0.08930646628141403,
      "learning_rate": 4.736906974945986e-05,
      "loss": 0.1548,
      "num_input_tokens_seen": 30770928,
      "step": 14260
    },
    {
      "epoch": 2.3270799347471454,
      "grad_norm": 3.269556999206543,
      "learning_rate": 4.736589036644848e-05,
      "loss": 0.0759,
      "num_input_tokens_seen": 30780240,
      "step": 14265
    },
    {
      "epoch": 2.3278955954323,
      "grad_norm": 0.06753698736429214,
      "learning_rate": 4.736270917034456e-05,
      "loss": 0.0633,
      "num_input_tokens_seen": 30792144,
      "step": 14270
    },
    {
      "epoch": 2.328711256117455,
      "grad_norm": 3.205472946166992,
      "learning_rate": 4.735952616140597e-05,
      "loss": 0.1823,
      "num_input_tokens_seen": 30802416,
      "step": 14275
    },
    {
      "epoch": 2.3295269168026103,
      "grad_norm": 4.39502477645874,
      "learning_rate": 4.735634133989072e-05,
      "loss": 0.1622,
      "num_input_tokens_seen": 30813872,
      "step": 14280
    },
    {
      "epoch": 2.330342577487765,
      "grad_norm": 1.3082256317138672,
      "learning_rate": 4.735315470605702e-05,
      "loss": 0.0281,
      "num_input_tokens_seen": 30825584,
      "step": 14285
    },
    {
      "epoch": 2.33115823817292,
      "grad_norm": 2.325822591781616,
      "learning_rate": 4.734996626016317e-05,
      "loss": 0.0367,
      "num_input_tokens_seen": 30836688,
      "step": 14290
    },
    {
      "epoch": 2.331973898858075,
      "grad_norm": 0.07380348443984985,
      "learning_rate": 4.7346776002467664e-05,
      "loss": 0.1007,
      "num_input_tokens_seen": 30848816,
      "step": 14295
    },
    {
      "epoch": 2.33278955954323,
      "grad_norm": 3.9989590644836426,
      "learning_rate": 4.73435839332291e-05,
      "loss": 0.1435,
      "num_input_tokens_seen": 30858608,
      "step": 14300
    },
    {
      "epoch": 2.3336052202283852,
      "grad_norm": 0.07307717204093933,
      "learning_rate": 4.734039005270625e-05,
      "loss": 0.0699,
      "num_input_tokens_seen": 30868560,
      "step": 14305
    },
    {
      "epoch": 2.33442088091354,
      "grad_norm": 0.0705452486872673,
      "learning_rate": 4.733719436115804e-05,
      "loss": 0.0139,
      "num_input_tokens_seen": 30878192,
      "step": 14310
    },
    {
      "epoch": 2.335236541598695,
      "grad_norm": 0.11348162591457367,
      "learning_rate": 4.733399685884351e-05,
      "loss": 0.0747,
      "num_input_tokens_seen": 30888432,
      "step": 14315
    },
    {
      "epoch": 2.3360522022838497,
      "grad_norm": 0.11116781830787659,
      "learning_rate": 4.7330797546021876e-05,
      "loss": 0.0686,
      "num_input_tokens_seen": 30898864,
      "step": 14320
    },
    {
      "epoch": 2.336867862969005,
      "grad_norm": 3.1997568607330322,
      "learning_rate": 4.732759642295248e-05,
      "loss": 0.2044,
      "num_input_tokens_seen": 30909424,
      "step": 14325
    },
    {
      "epoch": 2.3376835236541598,
      "grad_norm": 3.342360258102417,
      "learning_rate": 4.732439348989484e-05,
      "loss": 0.1344,
      "num_input_tokens_seen": 30919600,
      "step": 14330
    },
    {
      "epoch": 2.338499184339315,
      "grad_norm": 0.05780024081468582,
      "learning_rate": 4.732118874710858e-05,
      "loss": 0.0248,
      "num_input_tokens_seen": 30930416,
      "step": 14335
    },
    {
      "epoch": 2.33931484502447,
      "grad_norm": 0.0849500522017479,
      "learning_rate": 4.731798219485351e-05,
      "loss": 0.0346,
      "num_input_tokens_seen": 30940560,
      "step": 14340
    },
    {
      "epoch": 2.3401305057096247,
      "grad_norm": 2.2444562911987305,
      "learning_rate": 4.7314773833389567e-05,
      "loss": 0.2389,
      "num_input_tokens_seen": 30951440,
      "step": 14345
    },
    {
      "epoch": 2.34094616639478,
      "grad_norm": 0.053432710468769073,
      "learning_rate": 4.731156366297682e-05,
      "loss": 0.2715,
      "num_input_tokens_seen": 30961936,
      "step": 14350
    },
    {
      "epoch": 2.3417618270799347,
      "grad_norm": 3.597527503967285,
      "learning_rate": 4.730835168387553e-05,
      "loss": 0.183,
      "num_input_tokens_seen": 30972112,
      "step": 14355
    },
    {
      "epoch": 2.3425774877650896,
      "grad_norm": 4.307960033416748,
      "learning_rate": 4.730513789634605e-05,
      "loss": 0.1867,
      "num_input_tokens_seen": 30984272,
      "step": 14360
    },
    {
      "epoch": 2.343393148450245,
      "grad_norm": 0.33503457903862,
      "learning_rate": 4.7301922300648926e-05,
      "loss": 0.2667,
      "num_input_tokens_seen": 30996368,
      "step": 14365
    },
    {
      "epoch": 2.3442088091353996,
      "grad_norm": 1.6205902099609375,
      "learning_rate": 4.729870489704481e-05,
      "loss": 0.2137,
      "num_input_tokens_seen": 31007216,
      "step": 14370
    },
    {
      "epoch": 2.3450244698205545,
      "grad_norm": 0.1872188299894333,
      "learning_rate": 4.729548568579454e-05,
      "loss": 0.0181,
      "num_input_tokens_seen": 31018416,
      "step": 14375
    },
    {
      "epoch": 2.3458401305057097,
      "grad_norm": 0.05943664163351059,
      "learning_rate": 4.729226466715907e-05,
      "loss": 0.0146,
      "num_input_tokens_seen": 31030416,
      "step": 14380
    },
    {
      "epoch": 2.3466557911908645,
      "grad_norm": 1.1938236951828003,
      "learning_rate": 4.728904184139952e-05,
      "loss": 0.041,
      "num_input_tokens_seen": 31040752,
      "step": 14385
    },
    {
      "epoch": 2.34747145187602,
      "grad_norm": 0.23707178235054016,
      "learning_rate": 4.728581720877715e-05,
      "loss": 0.0882,
      "num_input_tokens_seen": 31052048,
      "step": 14390
    },
    {
      "epoch": 2.3482871125611746,
      "grad_norm": 10.16082763671875,
      "learning_rate": 4.7282590769553346e-05,
      "loss": 0.1714,
      "num_input_tokens_seen": 31062320,
      "step": 14395
    },
    {
      "epoch": 2.3491027732463294,
      "grad_norm": 9.198515892028809,
      "learning_rate": 4.727936252398969e-05,
      "loss": 0.113,
      "num_input_tokens_seen": 31071856,
      "step": 14400
    },
    {
      "epoch": 2.3499184339314847,
      "grad_norm": 0.029299002140760422,
      "learning_rate": 4.727613247234785e-05,
      "loss": 0.0055,
      "num_input_tokens_seen": 31082096,
      "step": 14405
    },
    {
      "epoch": 2.3507340946166395,
      "grad_norm": 5.021510124206543,
      "learning_rate": 4.727290061488969e-05,
      "loss": 0.2366,
      "num_input_tokens_seen": 31092400,
      "step": 14410
    },
    {
      "epoch": 2.3515497553017943,
      "grad_norm": 0.1946427822113037,
      "learning_rate": 4.726966695187719e-05,
      "loss": 0.1824,
      "num_input_tokens_seen": 31102768,
      "step": 14415
    },
    {
      "epoch": 2.3523654159869496,
      "grad_norm": 5.19529914855957,
      "learning_rate": 4.7266431483572495e-05,
      "loss": 0.0701,
      "num_input_tokens_seen": 31113936,
      "step": 14420
    },
    {
      "epoch": 2.3531810766721044,
      "grad_norm": 0.08390124142169952,
      "learning_rate": 4.726319421023789e-05,
      "loss": 0.0937,
      "num_input_tokens_seen": 31123920,
      "step": 14425
    },
    {
      "epoch": 2.3539967373572592,
      "grad_norm": 4.709795951843262,
      "learning_rate": 4.725995513213579e-05,
      "loss": 0.3095,
      "num_input_tokens_seen": 31135472,
      "step": 14430
    },
    {
      "epoch": 2.3548123980424145,
      "grad_norm": 0.028088008984923363,
      "learning_rate": 4.725671424952879e-05,
      "loss": 0.1675,
      "num_input_tokens_seen": 31145552,
      "step": 14435
    },
    {
      "epoch": 2.3556280587275693,
      "grad_norm": 5.511946201324463,
      "learning_rate": 4.7253471562679594e-05,
      "loss": 0.1055,
      "num_input_tokens_seen": 31157680,
      "step": 14440
    },
    {
      "epoch": 2.356443719412724,
      "grad_norm": 0.23965363204479218,
      "learning_rate": 4.725022707185109e-05,
      "loss": 0.0938,
      "num_input_tokens_seen": 31168720,
      "step": 14445
    },
    {
      "epoch": 2.3572593800978794,
      "grad_norm": 5.716320514678955,
      "learning_rate": 4.724698077730628e-05,
      "loss": 0.4099,
      "num_input_tokens_seen": 31178832,
      "step": 14450
    },
    {
      "epoch": 2.358075040783034,
      "grad_norm": 5.578467845916748,
      "learning_rate": 4.7243732679308325e-05,
      "loss": 0.0624,
      "num_input_tokens_seen": 31190288,
      "step": 14455
    },
    {
      "epoch": 2.358890701468189,
      "grad_norm": 0.20651887357234955,
      "learning_rate": 4.724048277812054e-05,
      "loss": 0.1101,
      "num_input_tokens_seen": 31202128,
      "step": 14460
    },
    {
      "epoch": 2.3597063621533443,
      "grad_norm": 0.09927832335233688,
      "learning_rate": 4.7237231074006374e-05,
      "loss": 0.0596,
      "num_input_tokens_seen": 31213232,
      "step": 14465
    },
    {
      "epoch": 2.360522022838499,
      "grad_norm": 5.533904552459717,
      "learning_rate": 4.723397756722942e-05,
      "loss": 0.1979,
      "num_input_tokens_seen": 31224272,
      "step": 14470
    },
    {
      "epoch": 2.3613376835236544,
      "grad_norm": 0.3006957471370697,
      "learning_rate": 4.7230722258053434e-05,
      "loss": 0.0408,
      "num_input_tokens_seen": 31234064,
      "step": 14475
    },
    {
      "epoch": 2.362153344208809,
      "grad_norm": 0.05534852668642998,
      "learning_rate": 4.7227465146742304e-05,
      "loss": 0.4224,
      "num_input_tokens_seen": 31245776,
      "step": 14480
    },
    {
      "epoch": 2.362969004893964,
      "grad_norm": 0.07525574415922165,
      "learning_rate": 4.722420623356007e-05,
      "loss": 0.1208,
      "num_input_tokens_seen": 31257392,
      "step": 14485
    },
    {
      "epoch": 2.3637846655791193,
      "grad_norm": 3.0173566341400146,
      "learning_rate": 4.722094551877091e-05,
      "loss": 0.064,
      "num_input_tokens_seen": 31267408,
      "step": 14490
    },
    {
      "epoch": 2.364600326264274,
      "grad_norm": 0.47848212718963623,
      "learning_rate": 4.7217683002639165e-05,
      "loss": 0.1229,
      "num_input_tokens_seen": 31278960,
      "step": 14495
    },
    {
      "epoch": 2.365415986949429,
      "grad_norm": 2.3555610179901123,
      "learning_rate": 4.7214418685429295e-05,
      "loss": 0.323,
      "num_input_tokens_seen": 31289264,
      "step": 14500
    },
    {
      "epoch": 2.366231647634584,
      "grad_norm": 7.209322452545166,
      "learning_rate": 4.721115256740594e-05,
      "loss": 0.0825,
      "num_input_tokens_seen": 31299152,
      "step": 14505
    },
    {
      "epoch": 2.367047308319739,
      "grad_norm": 0.688676655292511,
      "learning_rate": 4.720788464883385e-05,
      "loss": 0.1016,
      "num_input_tokens_seen": 31308944,
      "step": 14510
    },
    {
      "epoch": 2.367862969004894,
      "grad_norm": 0.032087117433547974,
      "learning_rate": 4.720461492997796e-05,
      "loss": 0.0918,
      "num_input_tokens_seen": 31319824,
      "step": 14515
    },
    {
      "epoch": 2.368678629690049,
      "grad_norm": 0.8098523020744324,
      "learning_rate": 4.720134341110332e-05,
      "loss": 0.1364,
      "num_input_tokens_seen": 31332016,
      "step": 14520
    },
    {
      "epoch": 2.369494290375204,
      "grad_norm": 2.5464463233947754,
      "learning_rate": 4.719807009247513e-05,
      "loss": 0.2921,
      "num_input_tokens_seen": 31342512,
      "step": 14525
    },
    {
      "epoch": 2.370309951060359,
      "grad_norm": 0.1941344439983368,
      "learning_rate": 4.7194794974358744e-05,
      "loss": 0.0212,
      "num_input_tokens_seen": 31352528,
      "step": 14530
    },
    {
      "epoch": 2.371125611745514,
      "grad_norm": 0.8727250695228577,
      "learning_rate": 4.719151805701966e-05,
      "loss": 0.0416,
      "num_input_tokens_seen": 31361520,
      "step": 14535
    },
    {
      "epoch": 2.3719412724306688,
      "grad_norm": 0.5036548376083374,
      "learning_rate": 4.7188239340723526e-05,
      "loss": 0.1059,
      "num_input_tokens_seen": 31372304,
      "step": 14540
    },
    {
      "epoch": 2.3727569331158236,
      "grad_norm": 1.291709065437317,
      "learning_rate": 4.7184958825736135e-05,
      "loss": 0.1785,
      "num_input_tokens_seen": 31382960,
      "step": 14545
    },
    {
      "epoch": 2.373572593800979,
      "grad_norm": 5.894839763641357,
      "learning_rate": 4.718167651232341e-05,
      "loss": 0.2981,
      "num_input_tokens_seen": 31394128,
      "step": 14550
    },
    {
      "epoch": 2.3743882544861337,
      "grad_norm": 2.329805612564087,
      "learning_rate": 4.7178392400751433e-05,
      "loss": 0.0469,
      "num_input_tokens_seen": 31405072,
      "step": 14555
    },
    {
      "epoch": 2.375203915171289,
      "grad_norm": 0.2343021035194397,
      "learning_rate": 4.7175106491286446e-05,
      "loss": 0.0138,
      "num_input_tokens_seen": 31415280,
      "step": 14560
    },
    {
      "epoch": 2.3760195758564437,
      "grad_norm": 0.8916614651679993,
      "learning_rate": 4.717181878419481e-05,
      "loss": 0.0206,
      "num_input_tokens_seen": 31426384,
      "step": 14565
    },
    {
      "epoch": 2.3768352365415986,
      "grad_norm": 1.5694172382354736,
      "learning_rate": 4.7168529279743046e-05,
      "loss": 0.0934,
      "num_input_tokens_seen": 31436304,
      "step": 14570
    },
    {
      "epoch": 2.377650897226754,
      "grad_norm": 0.12804140150547028,
      "learning_rate": 4.716523797819781e-05,
      "loss": 0.1344,
      "num_input_tokens_seen": 31448240,
      "step": 14575
    },
    {
      "epoch": 2.3784665579119086,
      "grad_norm": 0.0611204132437706,
      "learning_rate": 4.716194487982592e-05,
      "loss": 0.2789,
      "num_input_tokens_seen": 31458928,
      "step": 14580
    },
    {
      "epoch": 2.3792822185970635,
      "grad_norm": 1.1616871356964111,
      "learning_rate": 4.715864998489433e-05,
      "loss": 0.116,
      "num_input_tokens_seen": 31470480,
      "step": 14585
    },
    {
      "epoch": 2.3800978792822187,
      "grad_norm": 0.12439309060573578,
      "learning_rate": 4.715535329367014e-05,
      "loss": 0.0303,
      "num_input_tokens_seen": 31481968,
      "step": 14590
    },
    {
      "epoch": 2.3809135399673735,
      "grad_norm": 4.315296173095703,
      "learning_rate": 4.71520548064206e-05,
      "loss": 0.1764,
      "num_input_tokens_seen": 31493360,
      "step": 14595
    },
    {
      "epoch": 2.3817292006525284,
      "grad_norm": 4.915417194366455,
      "learning_rate": 4.71487545234131e-05,
      "loss": 0.1525,
      "num_input_tokens_seen": 31505552,
      "step": 14600
    },
    {
      "epoch": 2.3825448613376836,
      "grad_norm": 0.05003701522946358,
      "learning_rate": 4.7145452444915175e-05,
      "loss": 0.0541,
      "num_input_tokens_seen": 31516336,
      "step": 14605
    },
    {
      "epoch": 2.3833605220228384,
      "grad_norm": 0.053478702902793884,
      "learning_rate": 4.71421485711945e-05,
      "loss": 0.0606,
      "num_input_tokens_seen": 31527152,
      "step": 14610
    },
    {
      "epoch": 2.3841761827079937,
      "grad_norm": 0.049502789974212646,
      "learning_rate": 4.713884290251892e-05,
      "loss": 0.0105,
      "num_input_tokens_seen": 31536400,
      "step": 14615
    },
    {
      "epoch": 2.3849918433931485,
      "grad_norm": 5.173823356628418,
      "learning_rate": 4.713553543915641e-05,
      "loss": 0.0732,
      "num_input_tokens_seen": 31547056,
      "step": 14620
    },
    {
      "epoch": 2.3858075040783033,
      "grad_norm": 0.07904334366321564,
      "learning_rate": 4.713222618137508e-05,
      "loss": 0.1324,
      "num_input_tokens_seen": 31557840,
      "step": 14625
    },
    {
      "epoch": 2.3866231647634586,
      "grad_norm": 0.037357017397880554,
      "learning_rate": 4.71289151294432e-05,
      "loss": 0.0076,
      "num_input_tokens_seen": 31569424,
      "step": 14630
    },
    {
      "epoch": 2.3874388254486134,
      "grad_norm": 0.529985249042511,
      "learning_rate": 4.7125602283629166e-05,
      "loss": 0.2955,
      "num_input_tokens_seen": 31579504,
      "step": 14635
    },
    {
      "epoch": 2.3882544861337682,
      "grad_norm": 1.5207878351211548,
      "learning_rate": 4.7122287644201556e-05,
      "loss": 0.2426,
      "num_input_tokens_seen": 31591120,
      "step": 14640
    },
    {
      "epoch": 2.3890701468189235,
      "grad_norm": 1.2961686849594116,
      "learning_rate": 4.711897121142906e-05,
      "loss": 0.031,
      "num_input_tokens_seen": 31601968,
      "step": 14645
    },
    {
      "epoch": 2.3898858075040783,
      "grad_norm": 0.14282628893852234,
      "learning_rate": 4.711565298558053e-05,
      "loss": 0.1925,
      "num_input_tokens_seen": 31612944,
      "step": 14650
    },
    {
      "epoch": 2.390701468189233,
      "grad_norm": 5.399940013885498,
      "learning_rate": 4.711233296692495e-05,
      "loss": 0.0322,
      "num_input_tokens_seen": 31623472,
      "step": 14655
    },
    {
      "epoch": 2.3915171288743884,
      "grad_norm": 1.1830055713653564,
      "learning_rate": 4.7109011155731475e-05,
      "loss": 0.1171,
      "num_input_tokens_seen": 31635248,
      "step": 14660
    },
    {
      "epoch": 2.392332789559543,
      "grad_norm": 1.982262134552002,
      "learning_rate": 4.710568755226936e-05,
      "loss": 0.1877,
      "num_input_tokens_seen": 31646320,
      "step": 14665
    },
    {
      "epoch": 2.393148450244698,
      "grad_norm": 0.13083043694496155,
      "learning_rate": 4.710236215680806e-05,
      "loss": 0.0632,
      "num_input_tokens_seen": 31658608,
      "step": 14670
    },
    {
      "epoch": 2.3939641109298533,
      "grad_norm": 5.85105037689209,
      "learning_rate": 4.709903496961713e-05,
      "loss": 0.1228,
      "num_input_tokens_seen": 31669680,
      "step": 14675
    },
    {
      "epoch": 2.394779771615008,
      "grad_norm": 2.7455968856811523,
      "learning_rate": 4.7095705990966306e-05,
      "loss": 0.2834,
      "num_input_tokens_seen": 31679696,
      "step": 14680
    },
    {
      "epoch": 2.395595432300163,
      "grad_norm": 9.03864860534668,
      "learning_rate": 4.709237522112543e-05,
      "loss": 0.1886,
      "num_input_tokens_seen": 31691472,
      "step": 14685
    },
    {
      "epoch": 2.396411092985318,
      "grad_norm": 2.2991671562194824,
      "learning_rate": 4.708904266036453e-05,
      "loss": 0.1675,
      "num_input_tokens_seen": 31701968,
      "step": 14690
    },
    {
      "epoch": 2.397226753670473,
      "grad_norm": 0.08642101287841797,
      "learning_rate": 4.7085708308953754e-05,
      "loss": 0.2329,
      "num_input_tokens_seen": 31713104,
      "step": 14695
    },
    {
      "epoch": 2.3980424143556283,
      "grad_norm": 0.17132538557052612,
      "learning_rate": 4.7082372167163394e-05,
      "loss": 0.1597,
      "num_input_tokens_seen": 31724560,
      "step": 14700
    },
    {
      "epoch": 2.398858075040783,
      "grad_norm": 0.12105961889028549,
      "learning_rate": 4.707903423526391e-05,
      "loss": 0.0614,
      "num_input_tokens_seen": 31733776,
      "step": 14705
    },
    {
      "epoch": 2.399673735725938,
      "grad_norm": 0.23157398402690887,
      "learning_rate": 4.707569451352588e-05,
      "loss": 0.0887,
      "num_input_tokens_seen": 31745680,
      "step": 14710
    },
    {
      "epoch": 2.400489396411093,
      "grad_norm": 3.026545763015747,
      "learning_rate": 4.707235300222004e-05,
      "loss": 0.1202,
      "num_input_tokens_seen": 31756240,
      "step": 14715
    },
    {
      "epoch": 2.401305057096248,
      "grad_norm": 0.14984041452407837,
      "learning_rate": 4.706900970161727e-05,
      "loss": 0.0343,
      "num_input_tokens_seen": 31767056,
      "step": 14720
    },
    {
      "epoch": 2.402120717781403,
      "grad_norm": 0.2534407377243042,
      "learning_rate": 4.7065664611988596e-05,
      "loss": 0.0474,
      "num_input_tokens_seen": 31777968,
      "step": 14725
    },
    {
      "epoch": 2.402936378466558,
      "grad_norm": 0.39624321460723877,
      "learning_rate": 4.7062317733605185e-05,
      "loss": 0.1078,
      "num_input_tokens_seen": 31788976,
      "step": 14730
    },
    {
      "epoch": 2.403752039151713,
      "grad_norm": 0.4739513099193573,
      "learning_rate": 4.705896906673837e-05,
      "loss": 0.0786,
      "num_input_tokens_seen": 31799280,
      "step": 14735
    },
    {
      "epoch": 2.4045676998368677,
      "grad_norm": 0.2549099922180176,
      "learning_rate": 4.705561861165959e-05,
      "loss": 0.1446,
      "num_input_tokens_seen": 31810288,
      "step": 14740
    },
    {
      "epoch": 2.405383360522023,
      "grad_norm": 0.5927517414093018,
      "learning_rate": 4.705226636864045e-05,
      "loss": 0.0974,
      "num_input_tokens_seen": 31820880,
      "step": 14745
    },
    {
      "epoch": 2.4061990212071778,
      "grad_norm": 0.13888344168663025,
      "learning_rate": 4.704891233795271e-05,
      "loss": 0.0182,
      "num_input_tokens_seen": 31830704,
      "step": 14750
    },
    {
      "epoch": 2.407014681892333,
      "grad_norm": 0.20728279650211334,
      "learning_rate": 4.704555651986826e-05,
      "loss": 0.0819,
      "num_input_tokens_seen": 31842608,
      "step": 14755
    },
    {
      "epoch": 2.407830342577488,
      "grad_norm": 0.04445790499448776,
      "learning_rate": 4.704219891465914e-05,
      "loss": 0.1251,
      "num_input_tokens_seen": 31853456,
      "step": 14760
    },
    {
      "epoch": 2.4086460032626427,
      "grad_norm": 0.18355217576026917,
      "learning_rate": 4.703883952259754e-05,
      "loss": 0.2662,
      "num_input_tokens_seen": 31864240,
      "step": 14765
    },
    {
      "epoch": 2.4094616639477975,
      "grad_norm": 0.319480299949646,
      "learning_rate": 4.7035478343955774e-05,
      "loss": 0.1982,
      "num_input_tokens_seen": 31874896,
      "step": 14770
    },
    {
      "epoch": 2.4102773246329527,
      "grad_norm": 3.1677727699279785,
      "learning_rate": 4.7032115379006337e-05,
      "loss": 0.0879,
      "num_input_tokens_seen": 31887120,
      "step": 14775
    },
    {
      "epoch": 2.4110929853181076,
      "grad_norm": 7.563157081604004,
      "learning_rate": 4.7028750628021834e-05,
      "loss": 0.1887,
      "num_input_tokens_seen": 31899216,
      "step": 14780
    },
    {
      "epoch": 2.411908646003263,
      "grad_norm": 5.556710720062256,
      "learning_rate": 4.702538409127503e-05,
      "loss": 0.086,
      "num_input_tokens_seen": 31911248,
      "step": 14785
    },
    {
      "epoch": 2.4127243066884176,
      "grad_norm": 0.4326038658618927,
      "learning_rate": 4.7022015769038844e-05,
      "loss": 0.0516,
      "num_input_tokens_seen": 31921264,
      "step": 14790
    },
    {
      "epoch": 2.4135399673735725,
      "grad_norm": 1.8095623254776,
      "learning_rate": 4.701864566158631e-05,
      "loss": 0.0194,
      "num_input_tokens_seen": 31932208,
      "step": 14795
    },
    {
      "epoch": 2.4143556280587277,
      "grad_norm": 4.630144119262695,
      "learning_rate": 4.701527376919064e-05,
      "loss": 0.3303,
      "num_input_tokens_seen": 31943792,
      "step": 14800
    },
    {
      "epoch": 2.4151712887438825,
      "grad_norm": 0.05155915766954422,
      "learning_rate": 4.701190009212518e-05,
      "loss": 0.2663,
      "num_input_tokens_seen": 31954128,
      "step": 14805
    },
    {
      "epoch": 2.4159869494290374,
      "grad_norm": 3.377645969390869,
      "learning_rate": 4.700852463066341e-05,
      "loss": 0.0917,
      "num_input_tokens_seen": 31964176,
      "step": 14810
    },
    {
      "epoch": 2.4168026101141926,
      "grad_norm": 0.655836284160614,
      "learning_rate": 4.7005147385078956e-05,
      "loss": 0.0666,
      "num_input_tokens_seen": 31975312,
      "step": 14815
    },
    {
      "epoch": 2.4176182707993474,
      "grad_norm": 0.5649428963661194,
      "learning_rate": 4.700176835564561e-05,
      "loss": 0.0829,
      "num_input_tokens_seen": 31985520,
      "step": 14820
    },
    {
      "epoch": 2.4184339314845023,
      "grad_norm": 0.06093466654419899,
      "learning_rate": 4.699838754263728e-05,
      "loss": 0.0071,
      "num_input_tokens_seen": 31996656,
      "step": 14825
    },
    {
      "epoch": 2.4192495921696575,
      "grad_norm": 1.8105956315994263,
      "learning_rate": 4.6995004946328035e-05,
      "loss": 0.0355,
      "num_input_tokens_seen": 32005680,
      "step": 14830
    },
    {
      "epoch": 2.4200652528548123,
      "grad_norm": 0.07185716181993484,
      "learning_rate": 4.699162056699209e-05,
      "loss": 0.1257,
      "num_input_tokens_seen": 32017232,
      "step": 14835
    },
    {
      "epoch": 2.4208809135399676,
      "grad_norm": 0.10656130313873291,
      "learning_rate": 4.698823440490381e-05,
      "loss": 0.182,
      "num_input_tokens_seen": 32028112,
      "step": 14840
    },
    {
      "epoch": 2.4216965742251224,
      "grad_norm": 2.879488945007324,
      "learning_rate": 4.6984846460337664e-05,
      "loss": 0.0894,
      "num_input_tokens_seen": 32038608,
      "step": 14845
    },
    {
      "epoch": 2.4225122349102772,
      "grad_norm": 0.4780980944633484,
      "learning_rate": 4.698145673356832e-05,
      "loss": 0.0985,
      "num_input_tokens_seen": 32048880,
      "step": 14850
    },
    {
      "epoch": 2.4233278955954325,
      "grad_norm": 3.4437508583068848,
      "learning_rate": 4.697806522487056e-05,
      "loss": 0.2392,
      "num_input_tokens_seen": 32058864,
      "step": 14855
    },
    {
      "epoch": 2.4241435562805873,
      "grad_norm": 0.07802551984786987,
      "learning_rate": 4.697467193451932e-05,
      "loss": 0.0387,
      "num_input_tokens_seen": 32068816,
      "step": 14860
    },
    {
      "epoch": 2.424959216965742,
      "grad_norm": 1.7365397214889526,
      "learning_rate": 4.6971276862789674e-05,
      "loss": 0.1649,
      "num_input_tokens_seen": 32080848,
      "step": 14865
    },
    {
      "epoch": 2.4257748776508974,
      "grad_norm": 5.480813980102539,
      "learning_rate": 4.6967880009956845e-05,
      "loss": 0.043,
      "num_input_tokens_seen": 32091824,
      "step": 14870
    },
    {
      "epoch": 2.426590538336052,
      "grad_norm": 2.316845178604126,
      "learning_rate": 4.69644813762962e-05,
      "loss": 0.0888,
      "num_input_tokens_seen": 32102448,
      "step": 14875
    },
    {
      "epoch": 2.427406199021207,
      "grad_norm": 5.749186038970947,
      "learning_rate": 4.696108096208324e-05,
      "loss": 0.0514,
      "num_input_tokens_seen": 32114000,
      "step": 14880
    },
    {
      "epoch": 2.4282218597063623,
      "grad_norm": 0.1210162416100502,
      "learning_rate": 4.695767876759363e-05,
      "loss": 0.08,
      "num_input_tokens_seen": 32126192,
      "step": 14885
    },
    {
      "epoch": 2.429037520391517,
      "grad_norm": 1.1517932415008545,
      "learning_rate": 4.695427479310317e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 32136880,
      "step": 14890
    },
    {
      "epoch": 2.429853181076672,
      "grad_norm": 3.0411510467529297,
      "learning_rate": 4.6950869038887804e-05,
      "loss": 0.2664,
      "num_input_tokens_seen": 32147408,
      "step": 14895
    },
    {
      "epoch": 2.430668841761827,
      "grad_norm": 0.1486297845840454,
      "learning_rate": 4.6947461505223614e-05,
      "loss": 0.2751,
      "num_input_tokens_seen": 32158416,
      "step": 14900
    },
    {
      "epoch": 2.431484502446982,
      "grad_norm": 3.4809763431549072,
      "learning_rate": 4.6944052192386836e-05,
      "loss": 0.2939,
      "num_input_tokens_seen": 32169264,
      "step": 14905
    },
    {
      "epoch": 2.432300163132137,
      "grad_norm": 0.273034930229187,
      "learning_rate": 4.6940641100653834e-05,
      "loss": 0.2019,
      "num_input_tokens_seen": 32181872,
      "step": 14910
    },
    {
      "epoch": 2.433115823817292,
      "grad_norm": 0.11415091156959534,
      "learning_rate": 4.693722823030114e-05,
      "loss": 0.348,
      "num_input_tokens_seen": 32192816,
      "step": 14915
    },
    {
      "epoch": 2.433931484502447,
      "grad_norm": 1.0659358501434326,
      "learning_rate": 4.693381358160543e-05,
      "loss": 0.0275,
      "num_input_tokens_seen": 32203952,
      "step": 14920
    },
    {
      "epoch": 2.434747145187602,
      "grad_norm": 0.3799728453159332,
      "learning_rate": 4.693039715484349e-05,
      "loss": 0.1396,
      "num_input_tokens_seen": 32215088,
      "step": 14925
    },
    {
      "epoch": 2.435562805872757,
      "grad_norm": 0.10501531511545181,
      "learning_rate": 4.692697895029229e-05,
      "loss": 0.028,
      "num_input_tokens_seen": 32226256,
      "step": 14930
    },
    {
      "epoch": 2.436378466557912,
      "grad_norm": 2.2539470195770264,
      "learning_rate": 4.6923558968228906e-05,
      "loss": 0.1439,
      "num_input_tokens_seen": 32238096,
      "step": 14935
    },
    {
      "epoch": 2.437194127243067,
      "grad_norm": 0.1651032269001007,
      "learning_rate": 4.692013720893061e-05,
      "loss": 0.0736,
      "num_input_tokens_seen": 32247440,
      "step": 14940
    },
    {
      "epoch": 2.438009787928222,
      "grad_norm": 0.1005806028842926,
      "learning_rate": 4.691671367267476e-05,
      "loss": 0.0151,
      "num_input_tokens_seen": 32257488,
      "step": 14945
    },
    {
      "epoch": 2.4388254486133767,
      "grad_norm": 0.12197540700435638,
      "learning_rate": 4.6913288359738895e-05,
      "loss": 0.0385,
      "num_input_tokens_seen": 32268816,
      "step": 14950
    },
    {
      "epoch": 2.439641109298532,
      "grad_norm": 1.650969386100769,
      "learning_rate": 4.690986127040069e-05,
      "loss": 0.1656,
      "num_input_tokens_seen": 32280048,
      "step": 14955
    },
    {
      "epoch": 2.4404567699836868,
      "grad_norm": 3.752460479736328,
      "learning_rate": 4.690643240493797e-05,
      "loss": 0.0589,
      "num_input_tokens_seen": 32289680,
      "step": 14960
    },
    {
      "epoch": 2.4412724306688416,
      "grad_norm": 0.1766907274723053,
      "learning_rate": 4.690300176362867e-05,
      "loss": 0.023,
      "num_input_tokens_seen": 32300656,
      "step": 14965
    },
    {
      "epoch": 2.442088091353997,
      "grad_norm": 11.739470481872559,
      "learning_rate": 4.6899569346750924e-05,
      "loss": 0.1824,
      "num_input_tokens_seen": 32311888,
      "step": 14970
    },
    {
      "epoch": 2.4429037520391517,
      "grad_norm": 1.436651349067688,
      "learning_rate": 4.689613515458297e-05,
      "loss": 0.2738,
      "num_input_tokens_seen": 32323248,
      "step": 14975
    },
    {
      "epoch": 2.443719412724307,
      "grad_norm": 0.15432757139205933,
      "learning_rate": 4.68926991874032e-05,
      "loss": 0.0313,
      "num_input_tokens_seen": 32333808,
      "step": 14980
    },
    {
      "epoch": 2.4445350734094617,
      "grad_norm": 5.070051670074463,
      "learning_rate": 4.688926144549015e-05,
      "loss": 0.2821,
      "num_input_tokens_seen": 32345648,
      "step": 14985
    },
    {
      "epoch": 2.4453507340946166,
      "grad_norm": 0.10778035968542099,
      "learning_rate": 4.6885821929122497e-05,
      "loss": 0.0694,
      "num_input_tokens_seen": 32356272,
      "step": 14990
    },
    {
      "epoch": 2.4461663947797714,
      "grad_norm": 0.10623596608638763,
      "learning_rate": 4.688238063857908e-05,
      "loss": 0.2396,
      "num_input_tokens_seen": 32366800,
      "step": 14995
    },
    {
      "epoch": 2.4469820554649266,
      "grad_norm": 0.05677817016839981,
      "learning_rate": 4.687893757413885e-05,
      "loss": 0.2477,
      "num_input_tokens_seen": 32377072,
      "step": 15000
    },
    {
      "epoch": 2.4477977161500815,
      "grad_norm": 2.137606382369995,
      "learning_rate": 4.6875492736080935e-05,
      "loss": 0.1913,
      "num_input_tokens_seen": 32386640,
      "step": 15005
    },
    {
      "epoch": 2.4486133768352367,
      "grad_norm": 5.929174900054932,
      "learning_rate": 4.687204612468458e-05,
      "loss": 0.0861,
      "num_input_tokens_seen": 32396496,
      "step": 15010
    },
    {
      "epoch": 2.4494290375203915,
      "grad_norm": 2.540224075317383,
      "learning_rate": 4.6868597740229186e-05,
      "loss": 0.1164,
      "num_input_tokens_seen": 32407824,
      "step": 15015
    },
    {
      "epoch": 2.4502446982055464,
      "grad_norm": 0.1759481132030487,
      "learning_rate": 4.68651475829943e-05,
      "loss": 0.1133,
      "num_input_tokens_seen": 32417968,
      "step": 15020
    },
    {
      "epoch": 2.4510603588907016,
      "grad_norm": 0.07518824189901352,
      "learning_rate": 4.686169565325961e-05,
      "loss": 0.2448,
      "num_input_tokens_seen": 32428240,
      "step": 15025
    },
    {
      "epoch": 2.4518760195758564,
      "grad_norm": 1.3536690473556519,
      "learning_rate": 4.685824195130495e-05,
      "loss": 0.0332,
      "num_input_tokens_seen": 32438704,
      "step": 15030
    },
    {
      "epoch": 2.4526916802610113,
      "grad_norm": 3.277383804321289,
      "learning_rate": 4.6854786477410286e-05,
      "loss": 0.1731,
      "num_input_tokens_seen": 32449776,
      "step": 15035
    },
    {
      "epoch": 2.4535073409461665,
      "grad_norm": 0.16884741187095642,
      "learning_rate": 4.6851329231855736e-05,
      "loss": 0.0281,
      "num_input_tokens_seen": 32460240,
      "step": 15040
    },
    {
      "epoch": 2.4543230016313213,
      "grad_norm": 0.06038467586040497,
      "learning_rate": 4.6847870214921566e-05,
      "loss": 0.0622,
      "num_input_tokens_seen": 32469808,
      "step": 15045
    },
    {
      "epoch": 2.455138662316476,
      "grad_norm": 4.148122310638428,
      "learning_rate": 4.6844409426888186e-05,
      "loss": 0.0775,
      "num_input_tokens_seen": 32481232,
      "step": 15050
    },
    {
      "epoch": 2.4559543230016314,
      "grad_norm": 4.650274753570557,
      "learning_rate": 4.684094686803614e-05,
      "loss": 0.2861,
      "num_input_tokens_seen": 32492272,
      "step": 15055
    },
    {
      "epoch": 2.4567699836867862,
      "grad_norm": 0.22571337223052979,
      "learning_rate": 4.683748253864612e-05,
      "loss": 0.1318,
      "num_input_tokens_seen": 32503312,
      "step": 15060
    },
    {
      "epoch": 2.4575856443719415,
      "grad_norm": 2.9444875717163086,
      "learning_rate": 4.6834016438998965e-05,
      "loss": 0.1337,
      "num_input_tokens_seen": 32513424,
      "step": 15065
    },
    {
      "epoch": 2.4584013050570963,
      "grad_norm": 3.9076199531555176,
      "learning_rate": 4.6830548569375645e-05,
      "loss": 0.1281,
      "num_input_tokens_seen": 32525264,
      "step": 15070
    },
    {
      "epoch": 2.459216965742251,
      "grad_norm": 7.194041728973389,
      "learning_rate": 4.68270789300573e-05,
      "loss": 0.196,
      "num_input_tokens_seen": 32536592,
      "step": 15075
    },
    {
      "epoch": 2.4600326264274064,
      "grad_norm": 0.21722622215747833,
      "learning_rate": 4.682360752132518e-05,
      "loss": 0.0695,
      "num_input_tokens_seen": 32547344,
      "step": 15080
    },
    {
      "epoch": 2.460848287112561,
      "grad_norm": 0.585285484790802,
      "learning_rate": 4.682013434346071e-05,
      "loss": 0.0129,
      "num_input_tokens_seen": 32558800,
      "step": 15085
    },
    {
      "epoch": 2.461663947797716,
      "grad_norm": 1.7635830640792847,
      "learning_rate": 4.6816659396745424e-05,
      "loss": 0.0484,
      "num_input_tokens_seen": 32569392,
      "step": 15090
    },
    {
      "epoch": 2.4624796084828713,
      "grad_norm": 0.07490061968564987,
      "learning_rate": 4.6813182681461044e-05,
      "loss": 0.054,
      "num_input_tokens_seen": 32579632,
      "step": 15095
    },
    {
      "epoch": 2.463295269168026,
      "grad_norm": 0.08279639482498169,
      "learning_rate": 4.680970419788939e-05,
      "loss": 0.066,
      "num_input_tokens_seen": 32590128,
      "step": 15100
    },
    {
      "epoch": 2.464110929853181,
      "grad_norm": 0.08523435890674591,
      "learning_rate": 4.6806223946312455e-05,
      "loss": 0.0923,
      "num_input_tokens_seen": 32600656,
      "step": 15105
    },
    {
      "epoch": 2.464926590538336,
      "grad_norm": 0.3098160922527313,
      "learning_rate": 4.6802741927012363e-05,
      "loss": 0.0975,
      "num_input_tokens_seen": 32610928,
      "step": 15110
    },
    {
      "epoch": 2.465742251223491,
      "grad_norm": 0.06355145573616028,
      "learning_rate": 4.679925814027138e-05,
      "loss": 0.101,
      "num_input_tokens_seen": 32621488,
      "step": 15115
    },
    {
      "epoch": 2.466557911908646,
      "grad_norm": 0.5660656690597534,
      "learning_rate": 4.6795772586371934e-05,
      "loss": 0.1325,
      "num_input_tokens_seen": 32633712,
      "step": 15120
    },
    {
      "epoch": 2.467373572593801,
      "grad_norm": 0.025278862565755844,
      "learning_rate": 4.679228526559656e-05,
      "loss": 0.1558,
      "num_input_tokens_seen": 32644560,
      "step": 15125
    },
    {
      "epoch": 2.468189233278956,
      "grad_norm": 0.09633134305477142,
      "learning_rate": 4.678879617822798e-05,
      "loss": 0.078,
      "num_input_tokens_seen": 32655888,
      "step": 15130
    },
    {
      "epoch": 2.4690048939641107,
      "grad_norm": 0.09229657053947449,
      "learning_rate": 4.6785305324549016e-05,
      "loss": 0.1135,
      "num_input_tokens_seen": 32666832,
      "step": 15135
    },
    {
      "epoch": 2.469820554649266,
      "grad_norm": 1.2514058351516724,
      "learning_rate": 4.678181270484267e-05,
      "loss": 0.0573,
      "num_input_tokens_seen": 32677616,
      "step": 15140
    },
    {
      "epoch": 2.470636215334421,
      "grad_norm": 8.427809715270996,
      "learning_rate": 4.677831831939207e-05,
      "loss": 0.1057,
      "num_input_tokens_seen": 32687696,
      "step": 15145
    },
    {
      "epoch": 2.471451876019576,
      "grad_norm": 3.933966636657715,
      "learning_rate": 4.6774822168480476e-05,
      "loss": 0.2268,
      "num_input_tokens_seen": 32698096,
      "step": 15150
    },
    {
      "epoch": 2.472267536704731,
      "grad_norm": 0.10171698033809662,
      "learning_rate": 4.677132425239132e-05,
      "loss": 0.089,
      "num_input_tokens_seen": 32707920,
      "step": 15155
    },
    {
      "epoch": 2.4730831973898857,
      "grad_norm": 3.6427295207977295,
      "learning_rate": 4.676782457140815e-05,
      "loss": 0.1902,
      "num_input_tokens_seen": 32719472,
      "step": 15160
    },
    {
      "epoch": 2.473898858075041,
      "grad_norm": 4.919198989868164,
      "learning_rate": 4.676432312581467e-05,
      "loss": 0.1909,
      "num_input_tokens_seen": 32730064,
      "step": 15165
    },
    {
      "epoch": 2.4747145187601958,
      "grad_norm": 0.2900311350822449,
      "learning_rate": 4.676081991589473e-05,
      "loss": 0.1988,
      "num_input_tokens_seen": 32740944,
      "step": 15170
    },
    {
      "epoch": 2.4755301794453506,
      "grad_norm": 0.059633657336235046,
      "learning_rate": 4.6757314941932315e-05,
      "loss": 0.0829,
      "num_input_tokens_seen": 32751856,
      "step": 15175
    },
    {
      "epoch": 2.476345840130506,
      "grad_norm": 2.0496883392333984,
      "learning_rate": 4.6753808204211554e-05,
      "loss": 0.2245,
      "num_input_tokens_seen": 32762928,
      "step": 15180
    },
    {
      "epoch": 2.4771615008156607,
      "grad_norm": 2.0595855712890625,
      "learning_rate": 4.675029970301672e-05,
      "loss": 0.0569,
      "num_input_tokens_seen": 32773840,
      "step": 15185
    },
    {
      "epoch": 2.4779771615008155,
      "grad_norm": 2.525069236755371,
      "learning_rate": 4.674678943863223e-05,
      "loss": 0.2212,
      "num_input_tokens_seen": 32786064,
      "step": 15190
    },
    {
      "epoch": 2.4787928221859707,
      "grad_norm": 3.346924304962158,
      "learning_rate": 4.674327741134266e-05,
      "loss": 0.1739,
      "num_input_tokens_seen": 32796816,
      "step": 15195
    },
    {
      "epoch": 2.4796084828711256,
      "grad_norm": 1.2025363445281982,
      "learning_rate": 4.673976362143269e-05,
      "loss": 0.2415,
      "num_input_tokens_seen": 32807088,
      "step": 15200
    },
    {
      "epoch": 2.480424143556281,
      "grad_norm": 6.3952412605285645,
      "learning_rate": 4.673624806918717e-05,
      "loss": 0.2665,
      "num_input_tokens_seen": 32817424,
      "step": 15205
    },
    {
      "epoch": 2.4812398042414356,
      "grad_norm": 1.9188134670257568,
      "learning_rate": 4.673273075489109e-05,
      "loss": 0.182,
      "num_input_tokens_seen": 32827984,
      "step": 15210
    },
    {
      "epoch": 2.4820554649265905,
      "grad_norm": 1.138280987739563,
      "learning_rate": 4.6729211678829595e-05,
      "loss": 0.1766,
      "num_input_tokens_seen": 32839664,
      "step": 15215
    },
    {
      "epoch": 2.4828711256117453,
      "grad_norm": 3.514547109603882,
      "learning_rate": 4.672569084128794e-05,
      "loss": 0.2367,
      "num_input_tokens_seen": 32851280,
      "step": 15220
    },
    {
      "epoch": 2.4836867862969005,
      "grad_norm": 0.11544863879680634,
      "learning_rate": 4.6722168242551554e-05,
      "loss": 0.1579,
      "num_input_tokens_seen": 32862224,
      "step": 15225
    },
    {
      "epoch": 2.4845024469820554,
      "grad_norm": 0.05716465413570404,
      "learning_rate": 4.671864388290599e-05,
      "loss": 0.0094,
      "num_input_tokens_seen": 32872816,
      "step": 15230
    },
    {
      "epoch": 2.4853181076672106,
      "grad_norm": 1.00725519657135,
      "learning_rate": 4.671511776263696e-05,
      "loss": 0.1438,
      "num_input_tokens_seen": 32882896,
      "step": 15235
    },
    {
      "epoch": 2.4861337683523654,
      "grad_norm": 3.159532308578491,
      "learning_rate": 4.67115898820303e-05,
      "loss": 0.077,
      "num_input_tokens_seen": 32894256,
      "step": 15240
    },
    {
      "epoch": 2.4869494290375203,
      "grad_norm": 0.07962616533041,
      "learning_rate": 4.6708060241372e-05,
      "loss": 0.0425,
      "num_input_tokens_seen": 32905392,
      "step": 15245
    },
    {
      "epoch": 2.4877650897226755,
      "grad_norm": 0.1137455627322197,
      "learning_rate": 4.670452884094819e-05,
      "loss": 0.1247,
      "num_input_tokens_seen": 32916592,
      "step": 15250
    },
    {
      "epoch": 2.4885807504078303,
      "grad_norm": 0.1146974116563797,
      "learning_rate": 4.6700995681045144e-05,
      "loss": 0.0115,
      "num_input_tokens_seen": 32927216,
      "step": 15255
    },
    {
      "epoch": 2.489396411092985,
      "grad_norm": 2.8338823318481445,
      "learning_rate": 4.669746076194928e-05,
      "loss": 0.046,
      "num_input_tokens_seen": 32937648,
      "step": 15260
    },
    {
      "epoch": 2.4902120717781404,
      "grad_norm": 0.07691849768161774,
      "learning_rate": 4.669392408394716e-05,
      "loss": 0.1563,
      "num_input_tokens_seen": 32948496,
      "step": 15265
    },
    {
      "epoch": 2.4910277324632952,
      "grad_norm": 2.104459524154663,
      "learning_rate": 4.669038564732548e-05,
      "loss": 0.1469,
      "num_input_tokens_seen": 32958832,
      "step": 15270
    },
    {
      "epoch": 2.49184339314845,
      "grad_norm": 6.215854167938232,
      "learning_rate": 4.668684545237107e-05,
      "loss": 0.1873,
      "num_input_tokens_seen": 32969456,
      "step": 15275
    },
    {
      "epoch": 2.4926590538336053,
      "grad_norm": 0.09748450666666031,
      "learning_rate": 4.668330349937093e-05,
      "loss": 0.0697,
      "num_input_tokens_seen": 32980496,
      "step": 15280
    },
    {
      "epoch": 2.49347471451876,
      "grad_norm": 0.21889671683311462,
      "learning_rate": 4.6679759788612205e-05,
      "loss": 0.1825,
      "num_input_tokens_seen": 32991888,
      "step": 15285
    },
    {
      "epoch": 2.4942903752039154,
      "grad_norm": 3.8613903522491455,
      "learning_rate": 4.667621432038214e-05,
      "loss": 0.0933,
      "num_input_tokens_seen": 33002672,
      "step": 15290
    },
    {
      "epoch": 2.49510603588907,
      "grad_norm": 1.3740220069885254,
      "learning_rate": 4.6672667094968156e-05,
      "loss": 0.0575,
      "num_input_tokens_seen": 33012720,
      "step": 15295
    },
    {
      "epoch": 2.495921696574225,
      "grad_norm": 0.1635286808013916,
      "learning_rate": 4.6669118112657814e-05,
      "loss": 0.1084,
      "num_input_tokens_seen": 33023760,
      "step": 15300
    },
    {
      "epoch": 2.4967373572593803,
      "grad_norm": 1.9827061891555786,
      "learning_rate": 4.666556737373881e-05,
      "loss": 0.1727,
      "num_input_tokens_seen": 33034192,
      "step": 15305
    },
    {
      "epoch": 2.497553017944535,
      "grad_norm": 7.943353176116943,
      "learning_rate": 4.666201487849898e-05,
      "loss": 0.2828,
      "num_input_tokens_seen": 33044080,
      "step": 15310
    },
    {
      "epoch": 2.49836867862969,
      "grad_norm": 0.5402096509933472,
      "learning_rate": 4.665846062722632e-05,
      "loss": 0.0179,
      "num_input_tokens_seen": 33055184,
      "step": 15315
    },
    {
      "epoch": 2.499184339314845,
      "grad_norm": 1.8655792474746704,
      "learning_rate": 4.665490462020895e-05,
      "loss": 0.1486,
      "num_input_tokens_seen": 33065712,
      "step": 15320
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.123403787612915,
      "learning_rate": 4.665134685773513e-05,
      "loss": 0.1268,
      "num_input_tokens_seen": 33075856,
      "step": 15325
    },
    {
      "epoch": 2.5,
      "eval_loss": 0.15446700155735016,
      "eval_runtime": 132.912,
      "eval_samples_per_second": 20.502,
      "eval_steps_per_second": 5.131,
      "num_input_tokens_seen": 33075856,
      "step": 15325
    },
    {
      "epoch": 2.500815660685155,
      "grad_norm": 0.08432100713253021,
      "learning_rate": 4.664778734009327e-05,
      "loss": 0.0126,
      "num_input_tokens_seen": 33087312,
      "step": 15330
    },
    {
      "epoch": 2.50163132137031,
      "grad_norm": 0.07163143157958984,
      "learning_rate": 4.664422606757194e-05,
      "loss": 0.2113,
      "num_input_tokens_seen": 33098960,
      "step": 15335
    },
    {
      "epoch": 2.502446982055465,
      "grad_norm": 0.1372404843568802,
      "learning_rate": 4.664066304045982e-05,
      "loss": 0.1979,
      "num_input_tokens_seen": 33109712,
      "step": 15340
    },
    {
      "epoch": 2.50326264274062,
      "grad_norm": 2.33196759223938,
      "learning_rate": 4.6637098259045744e-05,
      "loss": 0.1085,
      "num_input_tokens_seen": 33121008,
      "step": 15345
    },
    {
      "epoch": 2.504078303425775,
      "grad_norm": 0.09625104069709778,
      "learning_rate": 4.66335317236187e-05,
      "loss": 0.1927,
      "num_input_tokens_seen": 33133360,
      "step": 15350
    },
    {
      "epoch": 2.50489396411093,
      "grad_norm": 0.8944844007492065,
      "learning_rate": 4.662996343446781e-05,
      "loss": 0.0355,
      "num_input_tokens_seen": 33144688,
      "step": 15355
    },
    {
      "epoch": 2.5057096247960846,
      "grad_norm": 0.08113139867782593,
      "learning_rate": 4.6626393391882326e-05,
      "loss": 0.1126,
      "num_input_tokens_seen": 33155024,
      "step": 15360
    },
    {
      "epoch": 2.50652528548124,
      "grad_norm": 0.12328080832958221,
      "learning_rate": 4.6622821596151676e-05,
      "loss": 0.3855,
      "num_input_tokens_seen": 33166576,
      "step": 15365
    },
    {
      "epoch": 2.5073409461663947,
      "grad_norm": 0.2539548873901367,
      "learning_rate": 4.6619248047565386e-05,
      "loss": 0.1121,
      "num_input_tokens_seen": 33176720,
      "step": 15370
    },
    {
      "epoch": 2.50815660685155,
      "grad_norm": 3.820850133895874,
      "learning_rate": 4.6615672746413156e-05,
      "loss": 0.1299,
      "num_input_tokens_seen": 33187824,
      "step": 15375
    },
    {
      "epoch": 2.5089722675367048,
      "grad_norm": 0.8509734869003296,
      "learning_rate": 4.661209569298482e-05,
      "loss": 0.0881,
      "num_input_tokens_seen": 33198800,
      "step": 15380
    },
    {
      "epoch": 2.5097879282218596,
      "grad_norm": 0.9104613661766052,
      "learning_rate": 4.660851688757034e-05,
      "loss": 0.1186,
      "num_input_tokens_seen": 33209232,
      "step": 15385
    },
    {
      "epoch": 2.5106035889070144,
      "grad_norm": 2.4266676902770996,
      "learning_rate": 4.6604936330459845e-05,
      "loss": 0.3417,
      "num_input_tokens_seen": 33219760,
      "step": 15390
    },
    {
      "epoch": 2.5114192495921697,
      "grad_norm": 0.26394614577293396,
      "learning_rate": 4.660135402194359e-05,
      "loss": 0.0792,
      "num_input_tokens_seen": 33229040,
      "step": 15395
    },
    {
      "epoch": 2.5122349102773245,
      "grad_norm": 0.27636078000068665,
      "learning_rate": 4.6597769962311975e-05,
      "loss": 0.0794,
      "num_input_tokens_seen": 33238960,
      "step": 15400
    },
    {
      "epoch": 2.5130505709624797,
      "grad_norm": 2.8311269283294678,
      "learning_rate": 4.6594184151855536e-05,
      "loss": 0.2758,
      "num_input_tokens_seen": 33247920,
      "step": 15405
    },
    {
      "epoch": 2.5138662316476346,
      "grad_norm": 0.316254198551178,
      "learning_rate": 4.6590596590864966e-05,
      "loss": 0.1256,
      "num_input_tokens_seen": 33259504,
      "step": 15410
    },
    {
      "epoch": 2.5146818923327894,
      "grad_norm": 0.11673296988010406,
      "learning_rate": 4.658700727963109e-05,
      "loss": 0.0366,
      "num_input_tokens_seen": 33269872,
      "step": 15415
    },
    {
      "epoch": 2.5154975530179446,
      "grad_norm": 0.36511024832725525,
      "learning_rate": 4.6583416218444866e-05,
      "loss": 0.0189,
      "num_input_tokens_seen": 33282192,
      "step": 15420
    },
    {
      "epoch": 2.5163132137030995,
      "grad_norm": 2.4341371059417725,
      "learning_rate": 4.657982340759741e-05,
      "loss": 0.0702,
      "num_input_tokens_seen": 33293648,
      "step": 15425
    },
    {
      "epoch": 2.5171288743882547,
      "grad_norm": 1.751753330230713,
      "learning_rate": 4.657622884737998e-05,
      "loss": 0.2745,
      "num_input_tokens_seen": 33304752,
      "step": 15430
    },
    {
      "epoch": 2.5179445350734095,
      "grad_norm": 2.4139862060546875,
      "learning_rate": 4.657263253808396e-05,
      "loss": 0.0779,
      "num_input_tokens_seen": 33315216,
      "step": 15435
    },
    {
      "epoch": 2.5187601957585644,
      "grad_norm": 6.911896228790283,
      "learning_rate": 4.6569034480000887e-05,
      "loss": 0.174,
      "num_input_tokens_seen": 33325392,
      "step": 15440
    },
    {
      "epoch": 2.519575856443719,
      "grad_norm": 0.10046017915010452,
      "learning_rate": 4.656543467342244e-05,
      "loss": 0.1558,
      "num_input_tokens_seen": 33336656,
      "step": 15445
    },
    {
      "epoch": 2.5203915171288744,
      "grad_norm": 0.28276556730270386,
      "learning_rate": 4.656183311864043e-05,
      "loss": 0.0927,
      "num_input_tokens_seen": 33347856,
      "step": 15450
    },
    {
      "epoch": 2.5212071778140293,
      "grad_norm": 1.1627421379089355,
      "learning_rate": 4.655822981594683e-05,
      "loss": 0.1814,
      "num_input_tokens_seen": 33358640,
      "step": 15455
    },
    {
      "epoch": 2.5220228384991845,
      "grad_norm": 0.24366597831249237,
      "learning_rate": 4.6554624765633734e-05,
      "loss": 0.1068,
      "num_input_tokens_seen": 33369200,
      "step": 15460
    },
    {
      "epoch": 2.5228384991843393,
      "grad_norm": 0.275344580411911,
      "learning_rate": 4.655101796799338e-05,
      "loss": 0.1087,
      "num_input_tokens_seen": 33378320,
      "step": 15465
    },
    {
      "epoch": 2.523654159869494,
      "grad_norm": 0.14054085314273834,
      "learning_rate": 4.654740942331818e-05,
      "loss": 0.1326,
      "num_input_tokens_seen": 33388976,
      "step": 15470
    },
    {
      "epoch": 2.5244698205546494,
      "grad_norm": 0.28822585940361023,
      "learning_rate": 4.6543799131900625e-05,
      "loss": 0.1237,
      "num_input_tokens_seen": 33399664,
      "step": 15475
    },
    {
      "epoch": 2.5252854812398042,
      "grad_norm": 4.053973197937012,
      "learning_rate": 4.6540187094033407e-05,
      "loss": 0.1205,
      "num_input_tokens_seen": 33409616,
      "step": 15480
    },
    {
      "epoch": 2.5261011419249595,
      "grad_norm": 0.2551969885826111,
      "learning_rate": 4.6536573310009326e-05,
      "loss": 0.014,
      "num_input_tokens_seen": 33420624,
      "step": 15485
    },
    {
      "epoch": 2.5269168026101143,
      "grad_norm": 2.315868616104126,
      "learning_rate": 4.653295778012134e-05,
      "loss": 0.111,
      "num_input_tokens_seen": 33430960,
      "step": 15490
    },
    {
      "epoch": 2.527732463295269,
      "grad_norm": 0.19548150897026062,
      "learning_rate": 4.652934050466254e-05,
      "loss": 0.0263,
      "num_input_tokens_seen": 33443344,
      "step": 15495
    },
    {
      "epoch": 2.528548123980424,
      "grad_norm": 0.32879120111465454,
      "learning_rate": 4.652572148392616e-05,
      "loss": 0.0985,
      "num_input_tokens_seen": 33454032,
      "step": 15500
    },
    {
      "epoch": 2.529363784665579,
      "grad_norm": 1.3231101036071777,
      "learning_rate": 4.652210071820557e-05,
      "loss": 0.09,
      "num_input_tokens_seen": 33465360,
      "step": 15505
    },
    {
      "epoch": 2.530179445350734,
      "grad_norm": 0.09188534319400787,
      "learning_rate": 4.6518478207794304e-05,
      "loss": 0.0437,
      "num_input_tokens_seen": 33476336,
      "step": 15510
    },
    {
      "epoch": 2.5309951060358893,
      "grad_norm": 0.04122030735015869,
      "learning_rate": 4.6514853952986e-05,
      "loss": 0.0115,
      "num_input_tokens_seen": 33486736,
      "step": 15515
    },
    {
      "epoch": 2.531810766721044,
      "grad_norm": 0.2291618436574936,
      "learning_rate": 4.6511227954074476e-05,
      "loss": 0.0937,
      "num_input_tokens_seen": 33497680,
      "step": 15520
    },
    {
      "epoch": 2.532626427406199,
      "grad_norm": 4.7056379318237305,
      "learning_rate": 4.650760021135366e-05,
      "loss": 0.1712,
      "num_input_tokens_seen": 33509456,
      "step": 15525
    },
    {
      "epoch": 2.5334420880913537,
      "grad_norm": 8.325273513793945,
      "learning_rate": 4.650397072511765e-05,
      "loss": 0.1599,
      "num_input_tokens_seen": 33521040,
      "step": 15530
    },
    {
      "epoch": 2.534257748776509,
      "grad_norm": 4.120744228363037,
      "learning_rate": 4.650033949566066e-05,
      "loss": 0.0692,
      "num_input_tokens_seen": 33531472,
      "step": 15535
    },
    {
      "epoch": 2.535073409461664,
      "grad_norm": 0.34146207571029663,
      "learning_rate": 4.6496706523277054e-05,
      "loss": 0.043,
      "num_input_tokens_seen": 33541552,
      "step": 15540
    },
    {
      "epoch": 2.535889070146819,
      "grad_norm": 0.21056298911571503,
      "learning_rate": 4.649307180826136e-05,
      "loss": 0.039,
      "num_input_tokens_seen": 33551920,
      "step": 15545
    },
    {
      "epoch": 2.536704730831974,
      "grad_norm": 1.2509381771087646,
      "learning_rate": 4.64894353509082e-05,
      "loss": 0.4235,
      "num_input_tokens_seen": 33562960,
      "step": 15550
    },
    {
      "epoch": 2.5375203915171287,
      "grad_norm": 0.04825669527053833,
      "learning_rate": 4.648579715151237e-05,
      "loss": 0.1027,
      "num_input_tokens_seen": 33572848,
      "step": 15555
    },
    {
      "epoch": 2.538336052202284,
      "grad_norm": 4.791048049926758,
      "learning_rate": 4.648215721036881e-05,
      "loss": 0.2568,
      "num_input_tokens_seen": 33582224,
      "step": 15560
    },
    {
      "epoch": 2.539151712887439,
      "grad_norm": 0.1789199411869049,
      "learning_rate": 4.647851552777258e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 33593680,
      "step": 15565
    },
    {
      "epoch": 2.539967373572594,
      "grad_norm": 0.9188748598098755,
      "learning_rate": 4.6474872104018907e-05,
      "loss": 0.1188,
      "num_input_tokens_seen": 33605616,
      "step": 15570
    },
    {
      "epoch": 2.540783034257749,
      "grad_norm": 0.8341172933578491,
      "learning_rate": 4.6471226939403145e-05,
      "loss": 0.2392,
      "num_input_tokens_seen": 33616752,
      "step": 15575
    },
    {
      "epoch": 2.5415986949429037,
      "grad_norm": 0.07152128964662552,
      "learning_rate": 4.646758003422077e-05,
      "loss": 0.1538,
      "num_input_tokens_seen": 33628048,
      "step": 15580
    },
    {
      "epoch": 2.5424143556280585,
      "grad_norm": 3.8207027912139893,
      "learning_rate": 4.646393138876745e-05,
      "loss": 0.0258,
      "num_input_tokens_seen": 33639280,
      "step": 15585
    },
    {
      "epoch": 2.5432300163132138,
      "grad_norm": 0.29015499353408813,
      "learning_rate": 4.6460281003338924e-05,
      "loss": 0.0836,
      "num_input_tokens_seen": 33649808,
      "step": 15590
    },
    {
      "epoch": 2.5440456769983686,
      "grad_norm": 0.10190565884113312,
      "learning_rate": 4.6456628878231144e-05,
      "loss": 0.298,
      "num_input_tokens_seen": 33658992,
      "step": 15595
    },
    {
      "epoch": 2.544861337683524,
      "grad_norm": 0.10488893836736679,
      "learning_rate": 4.645297501374015e-05,
      "loss": 0.0102,
      "num_input_tokens_seen": 33669904,
      "step": 15600
    },
    {
      "epoch": 2.5456769983686787,
      "grad_norm": 0.06030065566301346,
      "learning_rate": 4.644931941016216e-05,
      "loss": 0.1156,
      "num_input_tokens_seen": 33680080,
      "step": 15605
    },
    {
      "epoch": 2.5464926590538335,
      "grad_norm": 0.11721018701791763,
      "learning_rate": 4.644566206779349e-05,
      "loss": 0.1102,
      "num_input_tokens_seen": 33692208,
      "step": 15610
    },
    {
      "epoch": 2.5473083197389887,
      "grad_norm": 0.3074197471141815,
      "learning_rate": 4.6442002986930656e-05,
      "loss": 0.2263,
      "num_input_tokens_seen": 33703088,
      "step": 15615
    },
    {
      "epoch": 2.5481239804241436,
      "grad_norm": 0.06651334464550018,
      "learning_rate": 4.6438342167870255e-05,
      "loss": 0.1658,
      "num_input_tokens_seen": 33712944,
      "step": 15620
    },
    {
      "epoch": 2.5489396411092984,
      "grad_norm": 0.023425359278917313,
      "learning_rate": 4.643467961090906e-05,
      "loss": 0.0949,
      "num_input_tokens_seen": 33723408,
      "step": 15625
    },
    {
      "epoch": 2.5497553017944536,
      "grad_norm": 2.7293004989624023,
      "learning_rate": 4.643101531634399e-05,
      "loss": 0.1028,
      "num_input_tokens_seen": 33733488,
      "step": 15630
    },
    {
      "epoch": 2.5505709624796085,
      "grad_norm": 0.12186019122600555,
      "learning_rate": 4.642734928447207e-05,
      "loss": 0.1996,
      "num_input_tokens_seen": 33742640,
      "step": 15635
    },
    {
      "epoch": 2.5513866231647633,
      "grad_norm": 0.07401527464389801,
      "learning_rate": 4.642368151559049e-05,
      "loss": 0.2536,
      "num_input_tokens_seen": 33753584,
      "step": 15640
    },
    {
      "epoch": 2.5522022838499185,
      "grad_norm": 0.14932815730571747,
      "learning_rate": 4.642001200999659e-05,
      "loss": 0.0142,
      "num_input_tokens_seen": 33764656,
      "step": 15645
    },
    {
      "epoch": 2.5530179445350734,
      "grad_norm": 0.10367850959300995,
      "learning_rate": 4.6416340767987833e-05,
      "loss": 0.1579,
      "num_input_tokens_seen": 33775632,
      "step": 15650
    },
    {
      "epoch": 2.5538336052202286,
      "grad_norm": 0.04860873147845268,
      "learning_rate": 4.641266778986182e-05,
      "loss": 0.0717,
      "num_input_tokens_seen": 33787536,
      "step": 15655
    },
    {
      "epoch": 2.5546492659053834,
      "grad_norm": 2.193845272064209,
      "learning_rate": 4.640899307591632e-05,
      "loss": 0.1361,
      "num_input_tokens_seen": 33798736,
      "step": 15660
    },
    {
      "epoch": 2.5554649265905383,
      "grad_norm": 0.1617467850446701,
      "learning_rate": 4.64053166264492e-05,
      "loss": 0.0706,
      "num_input_tokens_seen": 33810256,
      "step": 15665
    },
    {
      "epoch": 2.556280587275693,
      "grad_norm": 0.3594755232334137,
      "learning_rate": 4.640163844175852e-05,
      "loss": 0.0497,
      "num_input_tokens_seen": 33820272,
      "step": 15670
    },
    {
      "epoch": 2.5570962479608483,
      "grad_norm": 0.4882911443710327,
      "learning_rate": 4.6397958522142426e-05,
      "loss": 0.1827,
      "num_input_tokens_seen": 33830448,
      "step": 15675
    },
    {
      "epoch": 2.557911908646003,
      "grad_norm": 3.7488725185394287,
      "learning_rate": 4.639427686789924e-05,
      "loss": 0.316,
      "num_input_tokens_seen": 33841808,
      "step": 15680
    },
    {
      "epoch": 2.5587275693311584,
      "grad_norm": 0.12173541635274887,
      "learning_rate": 4.6390593479327424e-05,
      "loss": 0.1763,
      "num_input_tokens_seen": 33851088,
      "step": 15685
    },
    {
      "epoch": 2.5595432300163132,
      "grad_norm": 1.2456576824188232,
      "learning_rate": 4.6386908356725564e-05,
      "loss": 0.0291,
      "num_input_tokens_seen": 33863600,
      "step": 15690
    },
    {
      "epoch": 2.560358890701468,
      "grad_norm": 0.0961676612496376,
      "learning_rate": 4.63832215003924e-05,
      "loss": 0.1181,
      "num_input_tokens_seen": 33873712,
      "step": 15695
    },
    {
      "epoch": 2.5611745513866233,
      "grad_norm": 1.2785987854003906,
      "learning_rate": 4.63795329106268e-05,
      "loss": 0.1444,
      "num_input_tokens_seen": 33884496,
      "step": 15700
    },
    {
      "epoch": 2.561990212071778,
      "grad_norm": 5.591950416564941,
      "learning_rate": 4.637584258772779e-05,
      "loss": 0.2803,
      "num_input_tokens_seen": 33896176,
      "step": 15705
    },
    {
      "epoch": 2.5628058727569334,
      "grad_norm": 0.16706594824790955,
      "learning_rate": 4.637215053199451e-05,
      "loss": 0.0251,
      "num_input_tokens_seen": 33907760,
      "step": 15710
    },
    {
      "epoch": 2.563621533442088,
      "grad_norm": 0.13217823207378387,
      "learning_rate": 4.6368456743726276e-05,
      "loss": 0.1172,
      "num_input_tokens_seen": 33918864,
      "step": 15715
    },
    {
      "epoch": 2.564437194127243,
      "grad_norm": 1.666069746017456,
      "learning_rate": 4.636476122322251e-05,
      "loss": 0.0872,
      "num_input_tokens_seen": 33929072,
      "step": 15720
    },
    {
      "epoch": 2.565252854812398,
      "grad_norm": 0.08260957896709442,
      "learning_rate": 4.636106397078279e-05,
      "loss": 0.1391,
      "num_input_tokens_seen": 33939440,
      "step": 15725
    },
    {
      "epoch": 2.566068515497553,
      "grad_norm": 4.464944839477539,
      "learning_rate": 4.635736498670685e-05,
      "loss": 0.0534,
      "num_input_tokens_seen": 33949168,
      "step": 15730
    },
    {
      "epoch": 2.566884176182708,
      "grad_norm": 0.10898555815219879,
      "learning_rate": 4.635366427129454e-05,
      "loss": 0.0321,
      "num_input_tokens_seen": 33958800,
      "step": 15735
    },
    {
      "epoch": 2.567699836867863,
      "grad_norm": 0.1415897011756897,
      "learning_rate": 4.634996182484584e-05,
      "loss": 0.1536,
      "num_input_tokens_seen": 33968144,
      "step": 15740
    },
    {
      "epoch": 2.568515497553018,
      "grad_norm": 0.38334110379219055,
      "learning_rate": 4.634625764766093e-05,
      "loss": 0.0839,
      "num_input_tokens_seen": 33979344,
      "step": 15745
    },
    {
      "epoch": 2.569331158238173,
      "grad_norm": 4.0980448722839355,
      "learning_rate": 4.6342551740040053e-05,
      "loss": 0.115,
      "num_input_tokens_seen": 33990160,
      "step": 15750
    },
    {
      "epoch": 2.5701468189233276,
      "grad_norm": 3.0779635906219482,
      "learning_rate": 4.633884410228364e-05,
      "loss": 0.1243,
      "num_input_tokens_seen": 34000336,
      "step": 15755
    },
    {
      "epoch": 2.570962479608483,
      "grad_norm": 0.6685082316398621,
      "learning_rate": 4.633513473469225e-05,
      "loss": 0.2121,
      "num_input_tokens_seen": 34011696,
      "step": 15760
    },
    {
      "epoch": 2.5717781402936377,
      "grad_norm": 0.2305363416671753,
      "learning_rate": 4.633142363756658e-05,
      "loss": 0.199,
      "num_input_tokens_seen": 34022256,
      "step": 15765
    },
    {
      "epoch": 2.572593800978793,
      "grad_norm": 3.24003529548645,
      "learning_rate": 4.6327710811207486e-05,
      "loss": 0.2613,
      "num_input_tokens_seen": 34032848,
      "step": 15770
    },
    {
      "epoch": 2.573409461663948,
      "grad_norm": 4.1800994873046875,
      "learning_rate": 4.6323996255915936e-05,
      "loss": 0.2036,
      "num_input_tokens_seen": 34043888,
      "step": 15775
    },
    {
      "epoch": 2.5742251223491026,
      "grad_norm": 2.6655170917510986,
      "learning_rate": 4.6320279971993055e-05,
      "loss": 0.2776,
      "num_input_tokens_seen": 34055760,
      "step": 15780
    },
    {
      "epoch": 2.575040783034258,
      "grad_norm": 0.7486165165901184,
      "learning_rate": 4.631656195974009e-05,
      "loss": 0.1037,
      "num_input_tokens_seen": 34067536,
      "step": 15785
    },
    {
      "epoch": 2.5758564437194127,
      "grad_norm": 0.48042935132980347,
      "learning_rate": 4.631284221945846e-05,
      "loss": 0.0574,
      "num_input_tokens_seen": 34078896,
      "step": 15790
    },
    {
      "epoch": 2.576672104404568,
      "grad_norm": 0.08358737826347351,
      "learning_rate": 4.6309120751449706e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 34089648,
      "step": 15795
    },
    {
      "epoch": 2.5774877650897228,
      "grad_norm": 0.076420359313488,
      "learning_rate": 4.63053975560155e-05,
      "loss": 0.0215,
      "num_input_tokens_seen": 34101392,
      "step": 15800
    },
    {
      "epoch": 2.5783034257748776,
      "grad_norm": 0.2541749179363251,
      "learning_rate": 4.630167263345766e-05,
      "loss": 0.0602,
      "num_input_tokens_seen": 34111184,
      "step": 15805
    },
    {
      "epoch": 2.5791190864600324,
      "grad_norm": 2.386906147003174,
      "learning_rate": 4.629794598407815e-05,
      "loss": 0.2211,
      "num_input_tokens_seen": 34121872,
      "step": 15810
    },
    {
      "epoch": 2.5799347471451877,
      "grad_norm": 0.09215260297060013,
      "learning_rate": 4.629421760817908e-05,
      "loss": 0.104,
      "num_input_tokens_seen": 34132912,
      "step": 15815
    },
    {
      "epoch": 2.5807504078303425,
      "grad_norm": 0.09753682464361191,
      "learning_rate": 4.6290487506062685e-05,
      "loss": 0.0566,
      "num_input_tokens_seen": 34143632,
      "step": 15820
    },
    {
      "epoch": 2.5815660685154977,
      "grad_norm": 0.12670865654945374,
      "learning_rate": 4.6286755678031344e-05,
      "loss": 0.0094,
      "num_input_tokens_seen": 34154928,
      "step": 15825
    },
    {
      "epoch": 2.5823817292006526,
      "grad_norm": 0.23903071880340576,
      "learning_rate": 4.628302212438758e-05,
      "loss": 0.0715,
      "num_input_tokens_seen": 34165168,
      "step": 15830
    },
    {
      "epoch": 2.5831973898858074,
      "grad_norm": 2.1843554973602295,
      "learning_rate": 4.627928684543406e-05,
      "loss": 0.135,
      "num_input_tokens_seen": 34176304,
      "step": 15835
    },
    {
      "epoch": 2.5840130505709626,
      "grad_norm": 0.27093032002449036,
      "learning_rate": 4.627554984147357e-05,
      "loss": 0.0337,
      "num_input_tokens_seen": 34186928,
      "step": 15840
    },
    {
      "epoch": 2.5848287112561175,
      "grad_norm": 5.873973369598389,
      "learning_rate": 4.627181111280906e-05,
      "loss": 0.1361,
      "num_input_tokens_seen": 34197712,
      "step": 15845
    },
    {
      "epoch": 2.5856443719412723,
      "grad_norm": 0.762215793132782,
      "learning_rate": 4.6268070659743605e-05,
      "loss": 0.0176,
      "num_input_tokens_seen": 34208144,
      "step": 15850
    },
    {
      "epoch": 2.5864600326264275,
      "grad_norm": 0.3678542375564575,
      "learning_rate": 4.626432848258044e-05,
      "loss": 0.1944,
      "num_input_tokens_seen": 34219504,
      "step": 15855
    },
    {
      "epoch": 2.5872756933115824,
      "grad_norm": 0.046675343066453934,
      "learning_rate": 4.62605845816229e-05,
      "loss": 0.2332,
      "num_input_tokens_seen": 34230032,
      "step": 15860
    },
    {
      "epoch": 2.588091353996737,
      "grad_norm": 2.1609106063842773,
      "learning_rate": 4.625683895717451e-05,
      "loss": 0.0379,
      "num_input_tokens_seen": 34240400,
      "step": 15865
    },
    {
      "epoch": 2.5889070146818924,
      "grad_norm": 1.1828417778015137,
      "learning_rate": 4.62530916095389e-05,
      "loss": 0.1004,
      "num_input_tokens_seen": 34252048,
      "step": 15870
    },
    {
      "epoch": 2.5897226753670473,
      "grad_norm": 0.25823017954826355,
      "learning_rate": 4.6249342539019844e-05,
      "loss": 0.0129,
      "num_input_tokens_seen": 34262832,
      "step": 15875
    },
    {
      "epoch": 2.5905383360522025,
      "grad_norm": 0.10200931131839752,
      "learning_rate": 4.6245591745921254e-05,
      "loss": 0.02,
      "num_input_tokens_seen": 34274512,
      "step": 15880
    },
    {
      "epoch": 2.5913539967373573,
      "grad_norm": 3.583845853805542,
      "learning_rate": 4.624183923054721e-05,
      "loss": 0.1516,
      "num_input_tokens_seen": 34284240,
      "step": 15885
    },
    {
      "epoch": 2.592169657422512,
      "grad_norm": 0.035426244139671326,
      "learning_rate": 4.623808499320189e-05,
      "loss": 0.1173,
      "num_input_tokens_seen": 34293744,
      "step": 15890
    },
    {
      "epoch": 2.592985318107667,
      "grad_norm": 2.031107187271118,
      "learning_rate": 4.623432903418965e-05,
      "loss": 0.0326,
      "num_input_tokens_seen": 34304144,
      "step": 15895
    },
    {
      "epoch": 2.5938009787928222,
      "grad_norm": 0.019137678667902946,
      "learning_rate": 4.6230571353814944e-05,
      "loss": 0.228,
      "num_input_tokens_seen": 34314992,
      "step": 15900
    },
    {
      "epoch": 2.594616639477977,
      "grad_norm": 10.572669982910156,
      "learning_rate": 4.622681195238241e-05,
      "loss": 0.202,
      "num_input_tokens_seen": 34326768,
      "step": 15905
    },
    {
      "epoch": 2.5954323001631323,
      "grad_norm": 0.7545855641365051,
      "learning_rate": 4.622305083019679e-05,
      "loss": 0.0891,
      "num_input_tokens_seen": 34336080,
      "step": 15910
    },
    {
      "epoch": 2.596247960848287,
      "grad_norm": 0.16171157360076904,
      "learning_rate": 4.621928798756299e-05,
      "loss": 0.0642,
      "num_input_tokens_seen": 34347568,
      "step": 15915
    },
    {
      "epoch": 2.597063621533442,
      "grad_norm": 5.608657360076904,
      "learning_rate": 4.621552342478604e-05,
      "loss": 0.3341,
      "num_input_tokens_seen": 34358864,
      "step": 15920
    },
    {
      "epoch": 2.597879282218597,
      "grad_norm": 0.5222229361534119,
      "learning_rate": 4.6211757142171105e-05,
      "loss": 0.0754,
      "num_input_tokens_seen": 34371024,
      "step": 15925
    },
    {
      "epoch": 2.598694942903752,
      "grad_norm": 0.0646638423204422,
      "learning_rate": 4.620798914002352e-05,
      "loss": 0.2143,
      "num_input_tokens_seen": 34381520,
      "step": 15930
    },
    {
      "epoch": 2.5995106035889073,
      "grad_norm": 0.09868259727954865,
      "learning_rate": 4.6204219418648724e-05,
      "loss": 0.0045,
      "num_input_tokens_seen": 34392880,
      "step": 15935
    },
    {
      "epoch": 2.600326264274062,
      "grad_norm": 6.142043590545654,
      "learning_rate": 4.6200447978352315e-05,
      "loss": 0.1834,
      "num_input_tokens_seen": 34404816,
      "step": 15940
    },
    {
      "epoch": 2.601141924959217,
      "grad_norm": 0.4567389190196991,
      "learning_rate": 4.6196674819440015e-05,
      "loss": 0.2938,
      "num_input_tokens_seen": 34414896,
      "step": 15945
    },
    {
      "epoch": 2.6019575856443717,
      "grad_norm": 6.175649166107178,
      "learning_rate": 4.619289994221771e-05,
      "loss": 0.2159,
      "num_input_tokens_seen": 34426480,
      "step": 15950
    },
    {
      "epoch": 2.602773246329527,
      "grad_norm": 0.21967223286628723,
      "learning_rate": 4.61891233469914e-05,
      "loss": 0.0323,
      "num_input_tokens_seen": 34437328,
      "step": 15955
    },
    {
      "epoch": 2.603588907014682,
      "grad_norm": 0.34079158306121826,
      "learning_rate": 4.618534503406724e-05,
      "loss": 0.1226,
      "num_input_tokens_seen": 34449584,
      "step": 15960
    },
    {
      "epoch": 2.604404567699837,
      "grad_norm": 1.7389894723892212,
      "learning_rate": 4.6181565003751525e-05,
      "loss": 0.1572,
      "num_input_tokens_seen": 34460592,
      "step": 15965
    },
    {
      "epoch": 2.605220228384992,
      "grad_norm": 0.2830871045589447,
      "learning_rate": 4.617778325635067e-05,
      "loss": 0.0254,
      "num_input_tokens_seen": 34472944,
      "step": 15970
    },
    {
      "epoch": 2.6060358890701467,
      "grad_norm": 0.07348176091909409,
      "learning_rate": 4.617399979217125e-05,
      "loss": 0.0096,
      "num_input_tokens_seen": 34483536,
      "step": 15975
    },
    {
      "epoch": 2.6068515497553015,
      "grad_norm": 0.17420241236686707,
      "learning_rate": 4.617021461151997e-05,
      "loss": 0.1893,
      "num_input_tokens_seen": 34495248,
      "step": 15980
    },
    {
      "epoch": 2.607667210440457,
      "grad_norm": 3.890345335006714,
      "learning_rate": 4.616642771470367e-05,
      "loss": 0.1241,
      "num_input_tokens_seen": 34506832,
      "step": 15985
    },
    {
      "epoch": 2.6084828711256116,
      "grad_norm": 0.051773801445961,
      "learning_rate": 4.616263910202936e-05,
      "loss": 0.0594,
      "num_input_tokens_seen": 34517168,
      "step": 15990
    },
    {
      "epoch": 2.609298531810767,
      "grad_norm": 0.1775064319372177,
      "learning_rate": 4.615884877380413e-05,
      "loss": 0.148,
      "num_input_tokens_seen": 34528272,
      "step": 15995
    },
    {
      "epoch": 2.6101141924959217,
      "grad_norm": 0.052289098501205444,
      "learning_rate": 4.6155056730335274e-05,
      "loss": 0.0061,
      "num_input_tokens_seen": 34537904,
      "step": 16000
    },
    {
      "epoch": 2.6109298531810765,
      "grad_norm": 0.22928167879581451,
      "learning_rate": 4.615126297193017e-05,
      "loss": 0.2099,
      "num_input_tokens_seen": 34548368,
      "step": 16005
    },
    {
      "epoch": 2.6117455138662318,
      "grad_norm": 0.6492549180984497,
      "learning_rate": 4.614746749889637e-05,
      "loss": 0.0688,
      "num_input_tokens_seen": 34559024,
      "step": 16010
    },
    {
      "epoch": 2.6125611745513866,
      "grad_norm": 0.8116114139556885,
      "learning_rate": 4.614367031154155e-05,
      "loss": 0.2135,
      "num_input_tokens_seen": 34570224,
      "step": 16015
    },
    {
      "epoch": 2.613376835236542,
      "grad_norm": 3.537290573120117,
      "learning_rate": 4.613987141017354e-05,
      "loss": 0.1358,
      "num_input_tokens_seen": 34581968,
      "step": 16020
    },
    {
      "epoch": 2.6141924959216967,
      "grad_norm": 0.11925935000181198,
      "learning_rate": 4.6136070795100285e-05,
      "loss": 0.1599,
      "num_input_tokens_seen": 34593392,
      "step": 16025
    },
    {
      "epoch": 2.6150081566068515,
      "grad_norm": 0.2562631070613861,
      "learning_rate": 4.613226846662989e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 34604176,
      "step": 16030
    },
    {
      "epoch": 2.6158238172920063,
      "grad_norm": 1.875343680381775,
      "learning_rate": 4.6128464425070595e-05,
      "loss": 0.0286,
      "num_input_tokens_seen": 34615024,
      "step": 16035
    },
    {
      "epoch": 2.6166394779771616,
      "grad_norm": 4.020516395568848,
      "learning_rate": 4.612465867073076e-05,
      "loss": 0.1692,
      "num_input_tokens_seen": 34626160,
      "step": 16040
    },
    {
      "epoch": 2.6174551386623164,
      "grad_norm": 0.5955748558044434,
      "learning_rate": 4.612085120391891e-05,
      "loss": 0.0345,
      "num_input_tokens_seen": 34635632,
      "step": 16045
    },
    {
      "epoch": 2.6182707993474716,
      "grad_norm": 3.323315143585205,
      "learning_rate": 4.61170420249437e-05,
      "loss": 0.0284,
      "num_input_tokens_seen": 34646288,
      "step": 16050
    },
    {
      "epoch": 2.6190864600326265,
      "grad_norm": 0.021180521696805954,
      "learning_rate": 4.611323113411391e-05,
      "loss": 0.3613,
      "num_input_tokens_seen": 34657232,
      "step": 16055
    },
    {
      "epoch": 2.6199021207177813,
      "grad_norm": 5.6233367919921875,
      "learning_rate": 4.610941853173848e-05,
      "loss": 0.2993,
      "num_input_tokens_seen": 34667408,
      "step": 16060
    },
    {
      "epoch": 2.6207177814029365,
      "grad_norm": 4.445132255554199,
      "learning_rate": 4.610560421812647e-05,
      "loss": 0.1756,
      "num_input_tokens_seen": 34677616,
      "step": 16065
    },
    {
      "epoch": 2.6215334420880914,
      "grad_norm": 0.05530526489019394,
      "learning_rate": 4.6101788193587103e-05,
      "loss": 0.0497,
      "num_input_tokens_seen": 34688688,
      "step": 16070
    },
    {
      "epoch": 2.622349102773246,
      "grad_norm": 0.7006406188011169,
      "learning_rate": 4.609797045842972e-05,
      "loss": 0.061,
      "num_input_tokens_seen": 34700272,
      "step": 16075
    },
    {
      "epoch": 2.6231647634584014,
      "grad_norm": 0.1208910420536995,
      "learning_rate": 4.609415101296379e-05,
      "loss": 0.3196,
      "num_input_tokens_seen": 34710832,
      "step": 16080
    },
    {
      "epoch": 2.6239804241435563,
      "grad_norm": 6.794556617736816,
      "learning_rate": 4.609032985749895e-05,
      "loss": 0.2261,
      "num_input_tokens_seen": 34722224,
      "step": 16085
    },
    {
      "epoch": 2.624796084828711,
      "grad_norm": 0.06990183889865875,
      "learning_rate": 4.6086506992344956e-05,
      "loss": 0.0761,
      "num_input_tokens_seen": 34733136,
      "step": 16090
    },
    {
      "epoch": 2.6256117455138663,
      "grad_norm": 0.11986415088176727,
      "learning_rate": 4.608268241781172e-05,
      "loss": 0.0538,
      "num_input_tokens_seen": 34743408,
      "step": 16095
    },
    {
      "epoch": 2.626427406199021,
      "grad_norm": 6.667908191680908,
      "learning_rate": 4.6078856134209284e-05,
      "loss": 0.3723,
      "num_input_tokens_seen": 34755184,
      "step": 16100
    },
    {
      "epoch": 2.6272430668841764,
      "grad_norm": 4.083271026611328,
      "learning_rate": 4.6075028141847795e-05,
      "loss": 0.0794,
      "num_input_tokens_seen": 34765040,
      "step": 16105
    },
    {
      "epoch": 2.6280587275693312,
      "grad_norm": 0.24231897294521332,
      "learning_rate": 4.607119844103761e-05,
      "loss": 0.0985,
      "num_input_tokens_seen": 34775312,
      "step": 16110
    },
    {
      "epoch": 2.628874388254486,
      "grad_norm": 1.1019506454467773,
      "learning_rate": 4.606736703208916e-05,
      "loss": 0.1174,
      "num_input_tokens_seen": 34784400,
      "step": 16115
    },
    {
      "epoch": 2.629690048939641,
      "grad_norm": 0.6285136342048645,
      "learning_rate": 4.606353391531304e-05,
      "loss": 0.0736,
      "num_input_tokens_seen": 34796080,
      "step": 16120
    },
    {
      "epoch": 2.630505709624796,
      "grad_norm": 1.0275976657867432,
      "learning_rate": 4.605969909101998e-05,
      "loss": 0.01,
      "num_input_tokens_seen": 34807280,
      "step": 16125
    },
    {
      "epoch": 2.631321370309951,
      "grad_norm": 0.47369199991226196,
      "learning_rate": 4.605586255952087e-05,
      "loss": 0.0611,
      "num_input_tokens_seen": 34818896,
      "step": 16130
    },
    {
      "epoch": 2.632137030995106,
      "grad_norm": 0.15059323608875275,
      "learning_rate": 4.6052024321126695e-05,
      "loss": 0.0192,
      "num_input_tokens_seen": 34830288,
      "step": 16135
    },
    {
      "epoch": 2.632952691680261,
      "grad_norm": 0.13787412643432617,
      "learning_rate": 4.6048184376148616e-05,
      "loss": 0.0845,
      "num_input_tokens_seen": 34840688,
      "step": 16140
    },
    {
      "epoch": 2.633768352365416,
      "grad_norm": 4.663808345794678,
      "learning_rate": 4.6044342724897915e-05,
      "loss": 0.0936,
      "num_input_tokens_seen": 34852848,
      "step": 16145
    },
    {
      "epoch": 2.634584013050571,
      "grad_norm": 3.8075475692749023,
      "learning_rate": 4.604049936768601e-05,
      "loss": 0.2947,
      "num_input_tokens_seen": 34863728,
      "step": 16150
    },
    {
      "epoch": 2.635399673735726,
      "grad_norm": 0.08273309469223022,
      "learning_rate": 4.6036654304824464e-05,
      "loss": 0.115,
      "num_input_tokens_seen": 34874672,
      "step": 16155
    },
    {
      "epoch": 2.636215334420881,
      "grad_norm": 1.2531658411026,
      "learning_rate": 4.603280753662499e-05,
      "loss": 0.0505,
      "num_input_tokens_seen": 34885904,
      "step": 16160
    },
    {
      "epoch": 2.637030995106036,
      "grad_norm": 2.630958080291748,
      "learning_rate": 4.602895906339941e-05,
      "loss": 0.1805,
      "num_input_tokens_seen": 34896144,
      "step": 16165
    },
    {
      "epoch": 2.637846655791191,
      "grad_norm": 3.709801197052002,
      "learning_rate": 4.6025108885459725e-05,
      "loss": 0.2926,
      "num_input_tokens_seen": 34906928,
      "step": 16170
    },
    {
      "epoch": 2.6386623164763456,
      "grad_norm": 0.6037567853927612,
      "learning_rate": 4.602125700311801e-05,
      "loss": 0.0163,
      "num_input_tokens_seen": 34917392,
      "step": 16175
    },
    {
      "epoch": 2.639477977161501,
      "grad_norm": 0.12990660965442657,
      "learning_rate": 4.6017403416686555e-05,
      "loss": 0.015,
      "num_input_tokens_seen": 34928240,
      "step": 16180
    },
    {
      "epoch": 2.6402936378466557,
      "grad_norm": 1.0022168159484863,
      "learning_rate": 4.601354812647774e-05,
      "loss": 0.0079,
      "num_input_tokens_seen": 34939824,
      "step": 16185
    },
    {
      "epoch": 2.641109298531811,
      "grad_norm": 0.272097110748291,
      "learning_rate": 4.600969113280409e-05,
      "loss": 0.0604,
      "num_input_tokens_seen": 34950704,
      "step": 16190
    },
    {
      "epoch": 2.641924959216966,
      "grad_norm": 0.08372607827186584,
      "learning_rate": 4.6005832435978266e-05,
      "loss": 0.1061,
      "num_input_tokens_seen": 34962224,
      "step": 16195
    },
    {
      "epoch": 2.6427406199021206,
      "grad_norm": 0.10326100140810013,
      "learning_rate": 4.600197203631309e-05,
      "loss": 0.0601,
      "num_input_tokens_seen": 34973936,
      "step": 16200
    },
    {
      "epoch": 2.6435562805872754,
      "grad_norm": 4.949693202972412,
      "learning_rate": 4.5998109934121494e-05,
      "loss": 0.0226,
      "num_input_tokens_seen": 34985296,
      "step": 16205
    },
    {
      "epoch": 2.6443719412724307,
      "grad_norm": 4.129422664642334,
      "learning_rate": 4.599424612971657e-05,
      "loss": 0.2191,
      "num_input_tokens_seen": 34996528,
      "step": 16210
    },
    {
      "epoch": 2.6451876019575855,
      "grad_norm": 3.720494508743286,
      "learning_rate": 4.599038062341153e-05,
      "loss": 0.1687,
      "num_input_tokens_seen": 35007056,
      "step": 16215
    },
    {
      "epoch": 2.6460032626427408,
      "grad_norm": 3.5929319858551025,
      "learning_rate": 4.598651341551973e-05,
      "loss": 0.066,
      "num_input_tokens_seen": 35017744,
      "step": 16220
    },
    {
      "epoch": 2.6468189233278956,
      "grad_norm": 3.253682851791382,
      "learning_rate": 4.5982644506354666e-05,
      "loss": 0.2127,
      "num_input_tokens_seen": 35029104,
      "step": 16225
    },
    {
      "epoch": 2.6476345840130504,
      "grad_norm": 0.07084818929433823,
      "learning_rate": 4.5978773896229977e-05,
      "loss": 0.1509,
      "num_input_tokens_seen": 35039632,
      "step": 16230
    },
    {
      "epoch": 2.6484502446982057,
      "grad_norm": 2.815587043762207,
      "learning_rate": 4.597490158545943e-05,
      "loss": 0.1168,
      "num_input_tokens_seen": 35050704,
      "step": 16235
    },
    {
      "epoch": 2.6492659053833605,
      "grad_norm": 0.06077788025140762,
      "learning_rate": 4.5971027574356926e-05,
      "loss": 0.0822,
      "num_input_tokens_seen": 35060688,
      "step": 16240
    },
    {
      "epoch": 2.6500815660685157,
      "grad_norm": 3.9549713134765625,
      "learning_rate": 4.5967151863236534e-05,
      "loss": 0.1066,
      "num_input_tokens_seen": 35071920,
      "step": 16245
    },
    {
      "epoch": 2.6508972267536706,
      "grad_norm": 0.061554085463285446,
      "learning_rate": 4.5963274452412416e-05,
      "loss": 0.0163,
      "num_input_tokens_seen": 35083056,
      "step": 16250
    },
    {
      "epoch": 2.6517128874388254,
      "grad_norm": 0.12308736890554428,
      "learning_rate": 4.595939534219891e-05,
      "loss": 0.0939,
      "num_input_tokens_seen": 35094320,
      "step": 16255
    },
    {
      "epoch": 2.65252854812398,
      "grad_norm": 0.061252955347299576,
      "learning_rate": 4.595551453291047e-05,
      "loss": 0.2009,
      "num_input_tokens_seen": 35104560,
      "step": 16260
    },
    {
      "epoch": 2.6533442088091355,
      "grad_norm": 5.05578088760376,
      "learning_rate": 4.5951632024861694e-05,
      "loss": 0.2439,
      "num_input_tokens_seen": 35115376,
      "step": 16265
    },
    {
      "epoch": 2.6541598694942903,
      "grad_norm": 0.19694742560386658,
      "learning_rate": 4.594774781836732e-05,
      "loss": 0.0649,
      "num_input_tokens_seen": 35126192,
      "step": 16270
    },
    {
      "epoch": 2.6549755301794455,
      "grad_norm": 0.9478245377540588,
      "learning_rate": 4.594386191374221e-05,
      "loss": 0.2022,
      "num_input_tokens_seen": 35137936,
      "step": 16275
    },
    {
      "epoch": 2.6557911908646004,
      "grad_norm": 0.3759821653366089,
      "learning_rate": 4.5939974311301406e-05,
      "loss": 0.1266,
      "num_input_tokens_seen": 35148560,
      "step": 16280
    },
    {
      "epoch": 2.656606851549755,
      "grad_norm": 0.5978142619132996,
      "learning_rate": 4.593608501136002e-05,
      "loss": 0.0913,
      "num_input_tokens_seen": 35159632,
      "step": 16285
    },
    {
      "epoch": 2.6574225122349104,
      "grad_norm": 0.06790251284837723,
      "learning_rate": 4.5932194014233356e-05,
      "loss": 0.1793,
      "num_input_tokens_seen": 35170000,
      "step": 16290
    },
    {
      "epoch": 2.6582381729200653,
      "grad_norm": 0.08304151892662048,
      "learning_rate": 4.592830132023684e-05,
      "loss": 0.0288,
      "num_input_tokens_seen": 35181360,
      "step": 16295
    },
    {
      "epoch": 2.65905383360522,
      "grad_norm": 0.13840016722679138,
      "learning_rate": 4.592440692968602e-05,
      "loss": 0.1375,
      "num_input_tokens_seen": 35191600,
      "step": 16300
    },
    {
      "epoch": 2.6598694942903753,
      "grad_norm": 0.1742582619190216,
      "learning_rate": 4.5920510842896624e-05,
      "loss": 0.015,
      "num_input_tokens_seen": 35201136,
      "step": 16305
    },
    {
      "epoch": 2.66068515497553,
      "grad_norm": 2.0974180698394775,
      "learning_rate": 4.591661306018446e-05,
      "loss": 0.0422,
      "num_input_tokens_seen": 35211600,
      "step": 16310
    },
    {
      "epoch": 2.661500815660685,
      "grad_norm": 0.11414952576160431,
      "learning_rate": 4.591271358186551e-05,
      "loss": 0.1467,
      "num_input_tokens_seen": 35222416,
      "step": 16315
    },
    {
      "epoch": 2.6623164763458402,
      "grad_norm": 0.6652671098709106,
      "learning_rate": 4.5908812408255884e-05,
      "loss": 0.1594,
      "num_input_tokens_seen": 35233904,
      "step": 16320
    },
    {
      "epoch": 2.663132137030995,
      "grad_norm": 0.1648857444524765,
      "learning_rate": 4.5904909539671836e-05,
      "loss": 0.1124,
      "num_input_tokens_seen": 35243984,
      "step": 16325
    },
    {
      "epoch": 2.6639477977161503,
      "grad_norm": 1.170156717300415,
      "learning_rate": 4.590100497642975e-05,
      "loss": 0.0634,
      "num_input_tokens_seen": 35255216,
      "step": 16330
    },
    {
      "epoch": 2.664763458401305,
      "grad_norm": 5.243762969970703,
      "learning_rate": 4.589709871884615e-05,
      "loss": 0.1903,
      "num_input_tokens_seen": 35265136,
      "step": 16335
    },
    {
      "epoch": 2.66557911908646,
      "grad_norm": 0.07611371576786041,
      "learning_rate": 4.58931907672377e-05,
      "loss": 0.1323,
      "num_input_tokens_seen": 35275600,
      "step": 16340
    },
    {
      "epoch": 2.6663947797716148,
      "grad_norm": 2.1288270950317383,
      "learning_rate": 4.588928112192119e-05,
      "loss": 0.1555,
      "num_input_tokens_seen": 35286928,
      "step": 16345
    },
    {
      "epoch": 2.66721044045677,
      "grad_norm": 1.1248494386672974,
      "learning_rate": 4.588536978321357e-05,
      "loss": 0.1184,
      "num_input_tokens_seen": 35299024,
      "step": 16350
    },
    {
      "epoch": 2.668026101141925,
      "grad_norm": 0.08379536122083664,
      "learning_rate": 4.588145675143189e-05,
      "loss": 0.0401,
      "num_input_tokens_seen": 35310320,
      "step": 16355
    },
    {
      "epoch": 2.66884176182708,
      "grad_norm": 0.08577514439821243,
      "learning_rate": 4.5877542026893395e-05,
      "loss": 0.0539,
      "num_input_tokens_seen": 35321104,
      "step": 16360
    },
    {
      "epoch": 2.669657422512235,
      "grad_norm": 5.488109588623047,
      "learning_rate": 4.5873625609915393e-05,
      "loss": 0.2237,
      "num_input_tokens_seen": 35332688,
      "step": 16365
    },
    {
      "epoch": 2.6704730831973897,
      "grad_norm": 0.062464434653520584,
      "learning_rate": 4.58697075008154e-05,
      "loss": 0.2095,
      "num_input_tokens_seen": 35342416,
      "step": 16370
    },
    {
      "epoch": 2.671288743882545,
      "grad_norm": 0.29279690980911255,
      "learning_rate": 4.586578769991102e-05,
      "loss": 0.1046,
      "num_input_tokens_seen": 35353616,
      "step": 16375
    },
    {
      "epoch": 2.6721044045677,
      "grad_norm": 0.9107227325439453,
      "learning_rate": 4.586186620752001e-05,
      "loss": 0.1074,
      "num_input_tokens_seen": 35364304,
      "step": 16380
    },
    {
      "epoch": 2.672920065252855,
      "grad_norm": 0.12089785933494568,
      "learning_rate": 4.585794302396028e-05,
      "loss": 0.1412,
      "num_input_tokens_seen": 35374992,
      "step": 16385
    },
    {
      "epoch": 2.67373572593801,
      "grad_norm": 3.918166160583496,
      "learning_rate": 4.585401814954986e-05,
      "loss": 0.2009,
      "num_input_tokens_seen": 35386544,
      "step": 16390
    },
    {
      "epoch": 2.6745513866231647,
      "grad_norm": 0.08335122466087341,
      "learning_rate": 4.5850091584606906e-05,
      "loss": 0.2132,
      "num_input_tokens_seen": 35397616,
      "step": 16395
    },
    {
      "epoch": 2.6753670473083195,
      "grad_norm": 0.2627863883972168,
      "learning_rate": 4.5846163329449745e-05,
      "loss": 0.2543,
      "num_input_tokens_seen": 35407408,
      "step": 16400
    },
    {
      "epoch": 2.676182707993475,
      "grad_norm": 1.354507565498352,
      "learning_rate": 4.584223338439681e-05,
      "loss": 0.0171,
      "num_input_tokens_seen": 35418512,
      "step": 16405
    },
    {
      "epoch": 2.6769983686786296,
      "grad_norm": 4.321389675140381,
      "learning_rate": 4.583830174976669e-05,
      "loss": 0.2351,
      "num_input_tokens_seen": 35429712,
      "step": 16410
    },
    {
      "epoch": 2.677814029363785,
      "grad_norm": 0.3772534728050232,
      "learning_rate": 4.5834368425878085e-05,
      "loss": 0.2127,
      "num_input_tokens_seen": 35440560,
      "step": 16415
    },
    {
      "epoch": 2.6786296900489397,
      "grad_norm": 0.24870356917381287,
      "learning_rate": 4.583043341304987e-05,
      "loss": 0.2236,
      "num_input_tokens_seen": 35451504,
      "step": 16420
    },
    {
      "epoch": 2.6794453507340945,
      "grad_norm": 1.4166717529296875,
      "learning_rate": 4.582649671160103e-05,
      "loss": 0.1816,
      "num_input_tokens_seen": 35462640,
      "step": 16425
    },
    {
      "epoch": 2.6802610114192493,
      "grad_norm": 0.1415262520313263,
      "learning_rate": 4.5822558321850696e-05,
      "loss": 0.1237,
      "num_input_tokens_seen": 35474160,
      "step": 16430
    },
    {
      "epoch": 2.6810766721044046,
      "grad_norm": 1.583168625831604,
      "learning_rate": 4.5818618244118126e-05,
      "loss": 0.1406,
      "num_input_tokens_seen": 35485200,
      "step": 16435
    },
    {
      "epoch": 2.6818923327895594,
      "grad_norm": 1.8406635522842407,
      "learning_rate": 4.581467647872273e-05,
      "loss": 0.4291,
      "num_input_tokens_seen": 35496176,
      "step": 16440
    },
    {
      "epoch": 2.6827079934747147,
      "grad_norm": 0.17736110091209412,
      "learning_rate": 4.5810733025984045e-05,
      "loss": 0.1222,
      "num_input_tokens_seen": 35506960,
      "step": 16445
    },
    {
      "epoch": 2.6835236541598695,
      "grad_norm": 2.8422389030456543,
      "learning_rate": 4.580678788622176e-05,
      "loss": 0.2197,
      "num_input_tokens_seen": 35517744,
      "step": 16450
    },
    {
      "epoch": 2.6843393148450243,
      "grad_norm": 3.310188055038452,
      "learning_rate": 4.580284105975566e-05,
      "loss": 0.2853,
      "num_input_tokens_seen": 35529680,
      "step": 16455
    },
    {
      "epoch": 2.6851549755301796,
      "grad_norm": 0.23108744621276855,
      "learning_rate": 4.5798892546905726e-05,
      "loss": 0.0904,
      "num_input_tokens_seen": 35540496,
      "step": 16460
    },
    {
      "epoch": 2.6859706362153344,
      "grad_norm": 4.71056604385376,
      "learning_rate": 4.579494234799202e-05,
      "loss": 0.1644,
      "num_input_tokens_seen": 35552176,
      "step": 16465
    },
    {
      "epoch": 2.6867862969004896,
      "grad_norm": 0.12348005175590515,
      "learning_rate": 4.579099046333477e-05,
      "loss": 0.0199,
      "num_input_tokens_seen": 35563440,
      "step": 16470
    },
    {
      "epoch": 2.6876019575856445,
      "grad_norm": 0.7517337799072266,
      "learning_rate": 4.5787036893254355e-05,
      "loss": 0.0169,
      "num_input_tokens_seen": 35574256,
      "step": 16475
    },
    {
      "epoch": 2.6884176182707993,
      "grad_norm": 1.4278310537338257,
      "learning_rate": 4.578308163807125e-05,
      "loss": 0.0544,
      "num_input_tokens_seen": 35584368,
      "step": 16480
    },
    {
      "epoch": 2.689233278955954,
      "grad_norm": 0.46861472725868225,
      "learning_rate": 4.577912469810609e-05,
      "loss": 0.0886,
      "num_input_tokens_seen": 35594896,
      "step": 16485
    },
    {
      "epoch": 2.6900489396411094,
      "grad_norm": 3.45877742767334,
      "learning_rate": 4.577516607367965e-05,
      "loss": 0.093,
      "num_input_tokens_seen": 35605488,
      "step": 16490
    },
    {
      "epoch": 2.690864600326264,
      "grad_norm": 3.337538242340088,
      "learning_rate": 4.577120576511285e-05,
      "loss": 0.087,
      "num_input_tokens_seen": 35617616,
      "step": 16495
    },
    {
      "epoch": 2.6916802610114194,
      "grad_norm": 0.3401094675064087,
      "learning_rate": 4.5767243772726706e-05,
      "loss": 0.2254,
      "num_input_tokens_seen": 35628304,
      "step": 16500
    },
    {
      "epoch": 2.6924959216965743,
      "grad_norm": 0.359669953584671,
      "learning_rate": 4.576328009684241e-05,
      "loss": 0.1823,
      "num_input_tokens_seen": 35640016,
      "step": 16505
    },
    {
      "epoch": 2.693311582381729,
      "grad_norm": 1.9160491228103638,
      "learning_rate": 4.5759314737781275e-05,
      "loss": 0.2197,
      "num_input_tokens_seen": 35651632,
      "step": 16510
    },
    {
      "epoch": 2.6941272430668843,
      "grad_norm": 0.5485318303108215,
      "learning_rate": 4.575534769586477e-05,
      "loss": 0.0122,
      "num_input_tokens_seen": 35661200,
      "step": 16515
    },
    {
      "epoch": 2.694942903752039,
      "grad_norm": 3.687964677810669,
      "learning_rate": 4.575137897141446e-05,
      "loss": 0.1268,
      "num_input_tokens_seen": 35671472,
      "step": 16520
    },
    {
      "epoch": 2.695758564437194,
      "grad_norm": 2.542691707611084,
      "learning_rate": 4.574740856475207e-05,
      "loss": 0.1762,
      "num_input_tokens_seen": 35682480,
      "step": 16525
    },
    {
      "epoch": 2.6965742251223492,
      "grad_norm": 5.13115930557251,
      "learning_rate": 4.574343647619949e-05,
      "loss": 0.3092,
      "num_input_tokens_seen": 35693168,
      "step": 16530
    },
    {
      "epoch": 2.697389885807504,
      "grad_norm": 0.35084402561187744,
      "learning_rate": 4.573946270607868e-05,
      "loss": 0.0771,
      "num_input_tokens_seen": 35704624,
      "step": 16535
    },
    {
      "epoch": 2.698205546492659,
      "grad_norm": 0.4560092091560364,
      "learning_rate": 4.573548725471181e-05,
      "loss": 0.0506,
      "num_input_tokens_seen": 35714768,
      "step": 16540
    },
    {
      "epoch": 2.699021207177814,
      "grad_norm": 0.17051959037780762,
      "learning_rate": 4.573151012242112e-05,
      "loss": 0.0698,
      "num_input_tokens_seen": 35724240,
      "step": 16545
    },
    {
      "epoch": 2.699836867862969,
      "grad_norm": 1.0198067426681519,
      "learning_rate": 4.5727531309529024e-05,
      "loss": 0.0676,
      "num_input_tokens_seen": 35735472,
      "step": 16550
    },
    {
      "epoch": 2.700652528548124,
      "grad_norm": 0.08668447285890579,
      "learning_rate": 4.5723550816358076e-05,
      "loss": 0.019,
      "num_input_tokens_seen": 35747376,
      "step": 16555
    },
    {
      "epoch": 2.701468189233279,
      "grad_norm": 0.316034197807312,
      "learning_rate": 4.571956864323095e-05,
      "loss": 0.0096,
      "num_input_tokens_seen": 35758960,
      "step": 16560
    },
    {
      "epoch": 2.702283849918434,
      "grad_norm": 3.647171974182129,
      "learning_rate": 4.571558479047046e-05,
      "loss": 0.1996,
      "num_input_tokens_seen": 35770512,
      "step": 16565
    },
    {
      "epoch": 2.7030995106035887,
      "grad_norm": 0.10150393843650818,
      "learning_rate": 4.571159925839956e-05,
      "loss": 0.1544,
      "num_input_tokens_seen": 35781424,
      "step": 16570
    },
    {
      "epoch": 2.703915171288744,
      "grad_norm": 4.908153533935547,
      "learning_rate": 4.570761204734133e-05,
      "loss": 0.1752,
      "num_input_tokens_seen": 35793200,
      "step": 16575
    },
    {
      "epoch": 2.7047308319738987,
      "grad_norm": 2.5196597576141357,
      "learning_rate": 4.5703623157619e-05,
      "loss": 0.0226,
      "num_input_tokens_seen": 35804880,
      "step": 16580
    },
    {
      "epoch": 2.705546492659054,
      "grad_norm": 0.11236696690320969,
      "learning_rate": 4.5699632589555924e-05,
      "loss": 0.2593,
      "num_input_tokens_seen": 35815376,
      "step": 16585
    },
    {
      "epoch": 2.706362153344209,
      "grad_norm": 2.4441187381744385,
      "learning_rate": 4.569564034347561e-05,
      "loss": 0.2319,
      "num_input_tokens_seen": 35825008,
      "step": 16590
    },
    {
      "epoch": 2.7071778140293636,
      "grad_norm": 2.784428596496582,
      "learning_rate": 4.5691646419701675e-05,
      "loss": 0.063,
      "num_input_tokens_seen": 35834384,
      "step": 16595
    },
    {
      "epoch": 2.707993474714519,
      "grad_norm": 4.784909725189209,
      "learning_rate": 4.5687650818557884e-05,
      "loss": 0.1562,
      "num_input_tokens_seen": 35844752,
      "step": 16600
    },
    {
      "epoch": 2.7088091353996737,
      "grad_norm": 0.10333660989999771,
      "learning_rate": 4.568365354036816e-05,
      "loss": 0.1813,
      "num_input_tokens_seen": 35855664,
      "step": 16605
    },
    {
      "epoch": 2.709624796084829,
      "grad_norm": 0.3954441249370575,
      "learning_rate": 4.567965458545653e-05,
      "loss": 0.0188,
      "num_input_tokens_seen": 35867472,
      "step": 16610
    },
    {
      "epoch": 2.710440456769984,
      "grad_norm": 0.10883083194494247,
      "learning_rate": 4.5675653954147174e-05,
      "loss": 0.1063,
      "num_input_tokens_seen": 35878544,
      "step": 16615
    },
    {
      "epoch": 2.7112561174551386,
      "grad_norm": 3.0300588607788086,
      "learning_rate": 4.56716516467644e-05,
      "loss": 0.1465,
      "num_input_tokens_seen": 35890704,
      "step": 16620
    },
    {
      "epoch": 2.7120717781402934,
      "grad_norm": 0.05359930172562599,
      "learning_rate": 4.5667647663632653e-05,
      "loss": 0.066,
      "num_input_tokens_seen": 35901424,
      "step": 16625
    },
    {
      "epoch": 2.7128874388254487,
      "grad_norm": 5.807322978973389,
      "learning_rate": 4.566364200507652e-05,
      "loss": 0.3546,
      "num_input_tokens_seen": 35912688,
      "step": 16630
    },
    {
      "epoch": 2.7137030995106035,
      "grad_norm": 2.9554502964019775,
      "learning_rate": 4.565963467142073e-05,
      "loss": 0.2036,
      "num_input_tokens_seen": 35924272,
      "step": 16635
    },
    {
      "epoch": 2.7145187601957588,
      "grad_norm": 0.1474665254354477,
      "learning_rate": 4.565562566299012e-05,
      "loss": 0.1732,
      "num_input_tokens_seen": 35935984,
      "step": 16640
    },
    {
      "epoch": 2.7153344208809136,
      "grad_norm": 1.2817227840423584,
      "learning_rate": 4.5651614980109684e-05,
      "loss": 0.2505,
      "num_input_tokens_seen": 35945840,
      "step": 16645
    },
    {
      "epoch": 2.7161500815660684,
      "grad_norm": 0.5527687072753906,
      "learning_rate": 4.564760262310456e-05,
      "loss": 0.116,
      "num_input_tokens_seen": 35956336,
      "step": 16650
    },
    {
      "epoch": 2.7169657422512232,
      "grad_norm": 0.08650974184274673,
      "learning_rate": 4.5643588592300004e-05,
      "loss": 0.0655,
      "num_input_tokens_seen": 35967664,
      "step": 16655
    },
    {
      "epoch": 2.7177814029363785,
      "grad_norm": 3.350778818130493,
      "learning_rate": 4.56395728880214e-05,
      "loss": 0.3088,
      "num_input_tokens_seen": 35978480,
      "step": 16660
    },
    {
      "epoch": 2.7185970636215333,
      "grad_norm": 0.14510881900787354,
      "learning_rate": 4.5635555510594304e-05,
      "loss": 0.1518,
      "num_input_tokens_seen": 35989040,
      "step": 16665
    },
    {
      "epoch": 2.7194127243066886,
      "grad_norm": 1.3831045627593994,
      "learning_rate": 4.563153646034437e-05,
      "loss": 0.1915,
      "num_input_tokens_seen": 35999792,
      "step": 16670
    },
    {
      "epoch": 2.7202283849918434,
      "grad_norm": 0.1926625370979309,
      "learning_rate": 4.5627515737597406e-05,
      "loss": 0.0127,
      "num_input_tokens_seen": 36009392,
      "step": 16675
    },
    {
      "epoch": 2.721044045676998,
      "grad_norm": 0.21117781102657318,
      "learning_rate": 4.562349334267936e-05,
      "loss": 0.1086,
      "num_input_tokens_seen": 36019952,
      "step": 16680
    },
    {
      "epoch": 2.7218597063621535,
      "grad_norm": 0.0967724546790123,
      "learning_rate": 4.5619469275916294e-05,
      "loss": 0.0757,
      "num_input_tokens_seen": 36031152,
      "step": 16685
    },
    {
      "epoch": 2.7226753670473083,
      "grad_norm": 1.1944202184677124,
      "learning_rate": 4.5615443537634425e-05,
      "loss": 0.1179,
      "num_input_tokens_seen": 36041808,
      "step": 16690
    },
    {
      "epoch": 2.7234910277324635,
      "grad_norm": 2.5175201892852783,
      "learning_rate": 4.561141612816011e-05,
      "loss": 0.1078,
      "num_input_tokens_seen": 36051248,
      "step": 16695
    },
    {
      "epoch": 2.7243066884176184,
      "grad_norm": 0.1099877655506134,
      "learning_rate": 4.560738704781982e-05,
      "loss": 0.0885,
      "num_input_tokens_seen": 36063024,
      "step": 16700
    },
    {
      "epoch": 2.725122349102773,
      "grad_norm": 0.34157779812812805,
      "learning_rate": 4.560335629694018e-05,
      "loss": 0.1709,
      "num_input_tokens_seen": 36074800,
      "step": 16705
    },
    {
      "epoch": 2.725938009787928,
      "grad_norm": 0.4502664804458618,
      "learning_rate": 4.559932387584792e-05,
      "loss": 0.0853,
      "num_input_tokens_seen": 36084784,
      "step": 16710
    },
    {
      "epoch": 2.7267536704730833,
      "grad_norm": 6.078094482421875,
      "learning_rate": 4.559528978486997e-05,
      "loss": 0.1225,
      "num_input_tokens_seen": 36095248,
      "step": 16715
    },
    {
      "epoch": 2.727569331158238,
      "grad_norm": 0.8107971549034119,
      "learning_rate": 4.5591254024333304e-05,
      "loss": 0.2715,
      "num_input_tokens_seen": 36106288,
      "step": 16720
    },
    {
      "epoch": 2.7283849918433933,
      "grad_norm": 0.6794746518135071,
      "learning_rate": 4.558721659456513e-05,
      "loss": 0.0457,
      "num_input_tokens_seen": 36116976,
      "step": 16725
    },
    {
      "epoch": 2.729200652528548,
      "grad_norm": 0.1730351448059082,
      "learning_rate": 4.558317749589271e-05,
      "loss": 0.1307,
      "num_input_tokens_seen": 36127504,
      "step": 16730
    },
    {
      "epoch": 2.730016313213703,
      "grad_norm": 0.06251662224531174,
      "learning_rate": 4.557913672864349e-05,
      "loss": 0.0141,
      "num_input_tokens_seen": 36137744,
      "step": 16735
    },
    {
      "epoch": 2.7308319738988582,
      "grad_norm": 1.2340346574783325,
      "learning_rate": 4.5575094293145025e-05,
      "loss": 0.1185,
      "num_input_tokens_seen": 36149328,
      "step": 16740
    },
    {
      "epoch": 2.731647634584013,
      "grad_norm": 0.1639896184206009,
      "learning_rate": 4.557105018972502e-05,
      "loss": 0.078,
      "num_input_tokens_seen": 36160688,
      "step": 16745
    },
    {
      "epoch": 2.732463295269168,
      "grad_norm": 0.26298972964286804,
      "learning_rate": 4.5567004418711314e-05,
      "loss": 0.0634,
      "num_input_tokens_seen": 36171056,
      "step": 16750
    },
    {
      "epoch": 2.733278955954323,
      "grad_norm": 0.09849293529987335,
      "learning_rate": 4.556295698043187e-05,
      "loss": 0.1476,
      "num_input_tokens_seen": 36181552,
      "step": 16755
    },
    {
      "epoch": 2.734094616639478,
      "grad_norm": 1.2345936298370361,
      "learning_rate": 4.55589078752148e-05,
      "loss": 0.0764,
      "num_input_tokens_seen": 36191472,
      "step": 16760
    },
    {
      "epoch": 2.7349102773246328,
      "grad_norm": 0.5437313914299011,
      "learning_rate": 4.5554857103388336e-05,
      "loss": 0.2333,
      "num_input_tokens_seen": 36204272,
      "step": 16765
    },
    {
      "epoch": 2.735725938009788,
      "grad_norm": 0.9578171372413635,
      "learning_rate": 4.555080466528087e-05,
      "loss": 0.19,
      "num_input_tokens_seen": 36216144,
      "step": 16770
    },
    {
      "epoch": 2.736541598694943,
      "grad_norm": 4.469488143920898,
      "learning_rate": 4.5546750561220896e-05,
      "loss": 0.0396,
      "num_input_tokens_seen": 36226672,
      "step": 16775
    },
    {
      "epoch": 2.737357259380098,
      "grad_norm": 6.144575119018555,
      "learning_rate": 4.554269479153708e-05,
      "loss": 0.2819,
      "num_input_tokens_seen": 36238448,
      "step": 16780
    },
    {
      "epoch": 2.738172920065253,
      "grad_norm": 0.42157989740371704,
      "learning_rate": 4.553863735655818e-05,
      "loss": 0.275,
      "num_input_tokens_seen": 36248400,
      "step": 16785
    },
    {
      "epoch": 2.7389885807504077,
      "grad_norm": 0.41651174426078796,
      "learning_rate": 4.553457825661313e-05,
      "loss": 0.0088,
      "num_input_tokens_seen": 36259824,
      "step": 16790
    },
    {
      "epoch": 2.7398042414355626,
      "grad_norm": 4.7950029373168945,
      "learning_rate": 4.553051749203097e-05,
      "loss": 0.0473,
      "num_input_tokens_seen": 36269168,
      "step": 16795
    },
    {
      "epoch": 2.740619902120718,
      "grad_norm": 1.291053056716919,
      "learning_rate": 4.5526455063140894e-05,
      "loss": 0.0389,
      "num_input_tokens_seen": 36278288,
      "step": 16800
    },
    {
      "epoch": 2.7414355628058726,
      "grad_norm": 0.9477439522743225,
      "learning_rate": 4.552239097027222e-05,
      "loss": 0.0774,
      "num_input_tokens_seen": 36289200,
      "step": 16805
    },
    {
      "epoch": 2.742251223491028,
      "grad_norm": 3.5565085411071777,
      "learning_rate": 4.551832521375441e-05,
      "loss": 0.1947,
      "num_input_tokens_seen": 36300112,
      "step": 16810
    },
    {
      "epoch": 2.7430668841761827,
      "grad_norm": 4.274855136871338,
      "learning_rate": 4.551425779391705e-05,
      "loss": 0.3973,
      "num_input_tokens_seen": 36309872,
      "step": 16815
    },
    {
      "epoch": 2.7438825448613375,
      "grad_norm": 0.06560499966144562,
      "learning_rate": 4.551018871108985e-05,
      "loss": 0.1101,
      "num_input_tokens_seen": 36320080,
      "step": 16820
    },
    {
      "epoch": 2.744698205546493,
      "grad_norm": 0.14905408024787903,
      "learning_rate": 4.55061179656027e-05,
      "loss": 0.2605,
      "num_input_tokens_seen": 36330448,
      "step": 16825
    },
    {
      "epoch": 2.7455138662316476,
      "grad_norm": 0.5720303654670715,
      "learning_rate": 4.550204555778558e-05,
      "loss": 0.2433,
      "num_input_tokens_seen": 36341008,
      "step": 16830
    },
    {
      "epoch": 2.746329526916803,
      "grad_norm": 0.10031141340732574,
      "learning_rate": 4.549797148796861e-05,
      "loss": 0.0955,
      "num_input_tokens_seen": 36351728,
      "step": 16835
    },
    {
      "epoch": 2.7471451876019577,
      "grad_norm": 0.16158132255077362,
      "learning_rate": 4.549389575648208e-05,
      "loss": 0.1842,
      "num_input_tokens_seen": 36362832,
      "step": 16840
    },
    {
      "epoch": 2.7479608482871125,
      "grad_norm": 6.217977523803711,
      "learning_rate": 4.548981836365636e-05,
      "loss": 0.1195,
      "num_input_tokens_seen": 36373520,
      "step": 16845
    },
    {
      "epoch": 2.7487765089722673,
      "grad_norm": 0.23625050485134125,
      "learning_rate": 4.5485739309822e-05,
      "loss": 0.0167,
      "num_input_tokens_seen": 36384912,
      "step": 16850
    },
    {
      "epoch": 2.7495921696574226,
      "grad_norm": 0.09042830765247345,
      "learning_rate": 4.548165859530968e-05,
      "loss": 0.0098,
      "num_input_tokens_seen": 36395760,
      "step": 16855
    },
    {
      "epoch": 2.7504078303425774,
      "grad_norm": 1.595046877861023,
      "learning_rate": 4.547757622045018e-05,
      "loss": 0.0277,
      "num_input_tokens_seen": 36406768,
      "step": 16860
    },
    {
      "epoch": 2.7512234910277327,
      "grad_norm": 0.9949770569801331,
      "learning_rate": 4.5473492185574465e-05,
      "loss": 0.0316,
      "num_input_tokens_seen": 36415728,
      "step": 16865
    },
    {
      "epoch": 2.7520391517128875,
      "grad_norm": 0.09952064603567123,
      "learning_rate": 4.546940649101358e-05,
      "loss": 0.3187,
      "num_input_tokens_seen": 36427728,
      "step": 16870
    },
    {
      "epoch": 2.7528548123980423,
      "grad_norm": 0.07224081456661224,
      "learning_rate": 4.546531913709874e-05,
      "loss": 0.2966,
      "num_input_tokens_seen": 36438640,
      "step": 16875
    },
    {
      "epoch": 2.753670473083197,
      "grad_norm": 10.022820472717285,
      "learning_rate": 4.5461230124161294e-05,
      "loss": 0.1469,
      "num_input_tokens_seen": 36448976,
      "step": 16880
    },
    {
      "epoch": 2.7544861337683524,
      "grad_norm": 0.37871983647346497,
      "learning_rate": 4.545713945253272e-05,
      "loss": 0.1956,
      "num_input_tokens_seen": 36458736,
      "step": 16885
    },
    {
      "epoch": 2.755301794453507,
      "grad_norm": 7.582609176635742,
      "learning_rate": 4.545304712254462e-05,
      "loss": 0.2638,
      "num_input_tokens_seen": 36468144,
      "step": 16890
    },
    {
      "epoch": 2.7561174551386625,
      "grad_norm": 0.13654311001300812,
      "learning_rate": 4.544895313452875e-05,
      "loss": 0.1795,
      "num_input_tokens_seen": 36478512,
      "step": 16895
    },
    {
      "epoch": 2.7569331158238173,
      "grad_norm": 3.9354288578033447,
      "learning_rate": 4.544485748881697e-05,
      "loss": 0.1404,
      "num_input_tokens_seen": 36490384,
      "step": 16900
    },
    {
      "epoch": 2.757748776508972,
      "grad_norm": 7.661076545715332,
      "learning_rate": 4.544076018574131e-05,
      "loss": 0.0703,
      "num_input_tokens_seen": 36500944,
      "step": 16905
    },
    {
      "epoch": 2.7585644371941274,
      "grad_norm": 1.766951322555542,
      "learning_rate": 4.5436661225633915e-05,
      "loss": 0.0863,
      "num_input_tokens_seen": 36511056,
      "step": 16910
    },
    {
      "epoch": 2.759380097879282,
      "grad_norm": 2.157850503921509,
      "learning_rate": 4.543256060882707e-05,
      "loss": 0.1478,
      "num_input_tokens_seen": 36522448,
      "step": 16915
    },
    {
      "epoch": 2.7601957585644374,
      "grad_norm": 0.15223313868045807,
      "learning_rate": 4.542845833565318e-05,
      "loss": 0.1927,
      "num_input_tokens_seen": 36533456,
      "step": 16920
    },
    {
      "epoch": 2.7610114192495923,
      "grad_norm": 0.4642239809036255,
      "learning_rate": 4.5424354406444815e-05,
      "loss": 0.1126,
      "num_input_tokens_seen": 36544656,
      "step": 16925
    },
    {
      "epoch": 2.761827079934747,
      "grad_norm": 3.68845272064209,
      "learning_rate": 4.542024882153464e-05,
      "loss": 0.1251,
      "num_input_tokens_seen": 36556144,
      "step": 16930
    },
    {
      "epoch": 2.762642740619902,
      "grad_norm": 0.17834614217281342,
      "learning_rate": 4.541614158125549e-05,
      "loss": 0.1239,
      "num_input_tokens_seen": 36566064,
      "step": 16935
    },
    {
      "epoch": 2.763458401305057,
      "grad_norm": 4.574592590332031,
      "learning_rate": 4.541203268594031e-05,
      "loss": 0.0395,
      "num_input_tokens_seen": 36576528,
      "step": 16940
    },
    {
      "epoch": 2.764274061990212,
      "grad_norm": 5.893138408660889,
      "learning_rate": 4.5407922135922194e-05,
      "loss": 0.1199,
      "num_input_tokens_seen": 36587856,
      "step": 16945
    },
    {
      "epoch": 2.7650897226753672,
      "grad_norm": 5.327587604522705,
      "learning_rate": 4.5403809931534355e-05,
      "loss": 0.2579,
      "num_input_tokens_seen": 36597936,
      "step": 16950
    },
    {
      "epoch": 2.765905383360522,
      "grad_norm": 0.3149024248123169,
      "learning_rate": 4.5399696073110166e-05,
      "loss": 0.0489,
      "num_input_tokens_seen": 36608688,
      "step": 16955
    },
    {
      "epoch": 2.766721044045677,
      "grad_norm": 1.3448281288146973,
      "learning_rate": 4.53955805609831e-05,
      "loss": 0.158,
      "num_input_tokens_seen": 36619824,
      "step": 16960
    },
    {
      "epoch": 2.767536704730832,
      "grad_norm": 3.0247461795806885,
      "learning_rate": 4.5391463395486784e-05,
      "loss": 0.1252,
      "num_input_tokens_seen": 36630960,
      "step": 16965
    },
    {
      "epoch": 2.768352365415987,
      "grad_norm": 0.11827442795038223,
      "learning_rate": 4.538734457695498e-05,
      "loss": 0.1118,
      "num_input_tokens_seen": 36640976,
      "step": 16970
    },
    {
      "epoch": 2.7691680261011418,
      "grad_norm": 0.21846012771129608,
      "learning_rate": 4.5383224105721586e-05,
      "loss": 0.2705,
      "num_input_tokens_seen": 36651792,
      "step": 16975
    },
    {
      "epoch": 2.769983686786297,
      "grad_norm": 3.764328718185425,
      "learning_rate": 4.537910198212061e-05,
      "loss": 0.0186,
      "num_input_tokens_seen": 36662896,
      "step": 16980
    },
    {
      "epoch": 2.770799347471452,
      "grad_norm": 4.463024616241455,
      "learning_rate": 4.537497820648624e-05,
      "loss": 0.1958,
      "num_input_tokens_seen": 36673040,
      "step": 16985
    },
    {
      "epoch": 2.7716150081566067,
      "grad_norm": 2.405142307281494,
      "learning_rate": 4.537085277915275e-05,
      "loss": 0.1598,
      "num_input_tokens_seen": 36684464,
      "step": 16990
    },
    {
      "epoch": 2.772430668841762,
      "grad_norm": 9.530285835266113,
      "learning_rate": 4.536672570045457e-05,
      "loss": 0.1109,
      "num_input_tokens_seen": 36696368,
      "step": 16995
    },
    {
      "epoch": 2.7732463295269167,
      "grad_norm": 0.11592987179756165,
      "learning_rate": 4.536259697072627e-05,
      "loss": 0.1465,
      "num_input_tokens_seen": 36707600,
      "step": 17000
    },
    {
      "epoch": 2.774061990212072,
      "grad_norm": 7.631938457489014,
      "learning_rate": 4.535846659030254e-05,
      "loss": 0.2427,
      "num_input_tokens_seen": 36718992,
      "step": 17005
    },
    {
      "epoch": 2.774877650897227,
      "grad_norm": 0.2692801356315613,
      "learning_rate": 4.5354334559518205e-05,
      "loss": 0.1457,
      "num_input_tokens_seen": 36729840,
      "step": 17010
    },
    {
      "epoch": 2.7756933115823816,
      "grad_norm": 0.17738287150859833,
      "learning_rate": 4.535020087870824e-05,
      "loss": 0.1265,
      "num_input_tokens_seen": 36738672,
      "step": 17015
    },
    {
      "epoch": 2.7765089722675365,
      "grad_norm": 2.427211284637451,
      "learning_rate": 4.5346065548207727e-05,
      "loss": 0.0315,
      "num_input_tokens_seen": 36749104,
      "step": 17020
    },
    {
      "epoch": 2.7773246329526917,
      "grad_norm": 0.10846050828695297,
      "learning_rate": 4.5341928568351915e-05,
      "loss": 0.112,
      "num_input_tokens_seen": 36760560,
      "step": 17025
    },
    {
      "epoch": 2.7781402936378465,
      "grad_norm": 0.5021817088127136,
      "learning_rate": 4.533778993947615e-05,
      "loss": 0.0255,
      "num_input_tokens_seen": 36771728,
      "step": 17030
    },
    {
      "epoch": 2.778955954323002,
      "grad_norm": 3.761016845703125,
      "learning_rate": 4.533364966191595e-05,
      "loss": 0.2266,
      "num_input_tokens_seen": 36783472,
      "step": 17035
    },
    {
      "epoch": 2.7797716150081566,
      "grad_norm": 0.07614640891551971,
      "learning_rate": 4.532950773600694e-05,
      "loss": 0.2203,
      "num_input_tokens_seen": 36794672,
      "step": 17040
    },
    {
      "epoch": 2.7805872756933114,
      "grad_norm": 0.1337439864873886,
      "learning_rate": 4.532536416208487e-05,
      "loss": 0.1636,
      "num_input_tokens_seen": 36805840,
      "step": 17045
    },
    {
      "epoch": 2.7814029363784667,
      "grad_norm": 0.40264835953712463,
      "learning_rate": 4.532121894048566e-05,
      "loss": 0.0561,
      "num_input_tokens_seen": 36816368,
      "step": 17050
    },
    {
      "epoch": 2.7822185970636215,
      "grad_norm": 3.4463798999786377,
      "learning_rate": 4.531707207154532e-05,
      "loss": 0.1587,
      "num_input_tokens_seen": 36827920,
      "step": 17055
    },
    {
      "epoch": 2.7830342577487768,
      "grad_norm": 4.410633087158203,
      "learning_rate": 4.531292355560004e-05,
      "loss": 0.3407,
      "num_input_tokens_seen": 36839312,
      "step": 17060
    },
    {
      "epoch": 2.7838499184339316,
      "grad_norm": 6.410935401916504,
      "learning_rate": 4.5308773392986115e-05,
      "loss": 0.1334,
      "num_input_tokens_seen": 36849264,
      "step": 17065
    },
    {
      "epoch": 2.7846655791190864,
      "grad_norm": 2.1469972133636475,
      "learning_rate": 4.530462158403996e-05,
      "loss": 0.2583,
      "num_input_tokens_seen": 36859664,
      "step": 17070
    },
    {
      "epoch": 2.7854812398042412,
      "grad_norm": 0.2848243713378906,
      "learning_rate": 4.5300468129098165e-05,
      "loss": 0.1154,
      "num_input_tokens_seen": 36869776,
      "step": 17075
    },
    {
      "epoch": 2.7862969004893965,
      "grad_norm": 1.9185163974761963,
      "learning_rate": 4.529631302849742e-05,
      "loss": 0.0949,
      "num_input_tokens_seen": 36880912,
      "step": 17080
    },
    {
      "epoch": 2.7871125611745513,
      "grad_norm": 3.354107141494751,
      "learning_rate": 4.529215628257455e-05,
      "loss": 0.2706,
      "num_input_tokens_seen": 36891056,
      "step": 17085
    },
    {
      "epoch": 2.7879282218597066,
      "grad_norm": 0.2763150632381439,
      "learning_rate": 4.528799789166654e-05,
      "loss": 0.1209,
      "num_input_tokens_seen": 36901456,
      "step": 17090
    },
    {
      "epoch": 2.7887438825448614,
      "grad_norm": 0.2002396583557129,
      "learning_rate": 4.5283837856110474e-05,
      "loss": 0.1359,
      "num_input_tokens_seen": 36912336,
      "step": 17095
    },
    {
      "epoch": 2.789559543230016,
      "grad_norm": 0.1000060886144638,
      "learning_rate": 4.5279676176243596e-05,
      "loss": 0.0315,
      "num_input_tokens_seen": 36921456,
      "step": 17100
    },
    {
      "epoch": 2.790375203915171,
      "grad_norm": 0.4621014893054962,
      "learning_rate": 4.527551285240327e-05,
      "loss": 0.0672,
      "num_input_tokens_seen": 36932272,
      "step": 17105
    },
    {
      "epoch": 2.7911908646003263,
      "grad_norm": 0.21004720032215118,
      "learning_rate": 4.527134788492698e-05,
      "loss": 0.0702,
      "num_input_tokens_seen": 36942640,
      "step": 17110
    },
    {
      "epoch": 2.792006525285481,
      "grad_norm": 2.599816083908081,
      "learning_rate": 4.526718127415239e-05,
      "loss": 0.3906,
      "num_input_tokens_seen": 36954192,
      "step": 17115
    },
    {
      "epoch": 2.7928221859706364,
      "grad_norm": 3.8914761543273926,
      "learning_rate": 4.5263013020417254e-05,
      "loss": 0.2187,
      "num_input_tokens_seen": 36964880,
      "step": 17120
    },
    {
      "epoch": 2.793637846655791,
      "grad_norm": 1.0694578886032104,
      "learning_rate": 4.5258843124059466e-05,
      "loss": 0.1012,
      "num_input_tokens_seen": 36976112,
      "step": 17125
    },
    {
      "epoch": 2.794453507340946,
      "grad_norm": 0.17898370325565338,
      "learning_rate": 4.5254671585417056e-05,
      "loss": 0.0571,
      "num_input_tokens_seen": 36985232,
      "step": 17130
    },
    {
      "epoch": 2.7952691680261013,
      "grad_norm": 0.751848578453064,
      "learning_rate": 4.52504984048282e-05,
      "loss": 0.0606,
      "num_input_tokens_seen": 36995376,
      "step": 17135
    },
    {
      "epoch": 2.796084828711256,
      "grad_norm": 1.4432446956634521,
      "learning_rate": 4.5246323582631196e-05,
      "loss": 0.0701,
      "num_input_tokens_seen": 37006576,
      "step": 17140
    },
    {
      "epoch": 2.7969004893964113,
      "grad_norm": 2.072270631790161,
      "learning_rate": 4.524214711916447e-05,
      "loss": 0.0836,
      "num_input_tokens_seen": 37017264,
      "step": 17145
    },
    {
      "epoch": 2.797716150081566,
      "grad_norm": 0.17519228160381317,
      "learning_rate": 4.523796901476659e-05,
      "loss": 0.0112,
      "num_input_tokens_seen": 37027472,
      "step": 17150
    },
    {
      "epoch": 2.798531810766721,
      "grad_norm": 0.1969202756881714,
      "learning_rate": 4.5233789269776264e-05,
      "loss": 0.0564,
      "num_input_tokens_seen": 37038768,
      "step": 17155
    },
    {
      "epoch": 2.799347471451876,
      "grad_norm": 0.6406100988388062,
      "learning_rate": 4.5229607884532306e-05,
      "loss": 0.0759,
      "num_input_tokens_seen": 37049424,
      "step": 17160
    },
    {
      "epoch": 2.800163132137031,
      "grad_norm": 7.187447547912598,
      "learning_rate": 4.522542485937369e-05,
      "loss": 0.3022,
      "num_input_tokens_seen": 37060912,
      "step": 17165
    },
    {
      "epoch": 2.800978792822186,
      "grad_norm": 2.4074254035949707,
      "learning_rate": 4.5221240194639514e-05,
      "loss": 0.1944,
      "num_input_tokens_seen": 37070032,
      "step": 17170
    },
    {
      "epoch": 2.801794453507341,
      "grad_norm": 9.948955535888672,
      "learning_rate": 4.5217053890669004e-05,
      "loss": 0.1661,
      "num_input_tokens_seen": 37079792,
      "step": 17175
    },
    {
      "epoch": 2.802610114192496,
      "grad_norm": 0.4232616722583771,
      "learning_rate": 4.521286594780152e-05,
      "loss": 0.0898,
      "num_input_tokens_seen": 37091344,
      "step": 17180
    },
    {
      "epoch": 2.8034257748776508,
      "grad_norm": 1.625488519668579,
      "learning_rate": 4.5208676366376574e-05,
      "loss": 0.0793,
      "num_input_tokens_seen": 37102800,
      "step": 17185
    },
    {
      "epoch": 2.804241435562806,
      "grad_norm": 0.9074686765670776,
      "learning_rate": 4.520448514673378e-05,
      "loss": 0.1862,
      "num_input_tokens_seen": 37114512,
      "step": 17190
    },
    {
      "epoch": 2.805057096247961,
      "grad_norm": 2.776407480239868,
      "learning_rate": 4.52002922892129e-05,
      "loss": 0.2865,
      "num_input_tokens_seen": 37125232,
      "step": 17195
    },
    {
      "epoch": 2.8058727569331157,
      "grad_norm": 2.4943649768829346,
      "learning_rate": 4.519609779415384e-05,
      "loss": 0.0979,
      "num_input_tokens_seen": 37136656,
      "step": 17200
    },
    {
      "epoch": 2.806688417618271,
      "grad_norm": 1.6606615781784058,
      "learning_rate": 4.519190166189661e-05,
      "loss": 0.1429,
      "num_input_tokens_seen": 37148112,
      "step": 17205
    },
    {
      "epoch": 2.8075040783034257,
      "grad_norm": 0.2653537392616272,
      "learning_rate": 4.518770389278138e-05,
      "loss": 0.0566,
      "num_input_tokens_seen": 37160208,
      "step": 17210
    },
    {
      "epoch": 2.8083197389885806,
      "grad_norm": 0.15048940479755402,
      "learning_rate": 4.5183504487148444e-05,
      "loss": 0.0767,
      "num_input_tokens_seen": 37170288,
      "step": 17215
    },
    {
      "epoch": 2.809135399673736,
      "grad_norm": 0.3598777651786804,
      "learning_rate": 4.517930344533822e-05,
      "loss": 0.0386,
      "num_input_tokens_seen": 37179408,
      "step": 17220
    },
    {
      "epoch": 2.8099510603588906,
      "grad_norm": 0.19925951957702637,
      "learning_rate": 4.517510076769128e-05,
      "loss": 0.1304,
      "num_input_tokens_seen": 37191472,
      "step": 17225
    },
    {
      "epoch": 2.810766721044046,
      "grad_norm": 0.2026115208864212,
      "learning_rate": 4.517089645454829e-05,
      "loss": 0.0549,
      "num_input_tokens_seen": 37200528,
      "step": 17230
    },
    {
      "epoch": 2.8115823817292007,
      "grad_norm": 0.1826828569173813,
      "learning_rate": 4.51666905062501e-05,
      "loss": 0.1201,
      "num_input_tokens_seen": 37210960,
      "step": 17235
    },
    {
      "epoch": 2.8123980424143555,
      "grad_norm": 0.0931023508310318,
      "learning_rate": 4.516248292313765e-05,
      "loss": 0.235,
      "num_input_tokens_seen": 37222192,
      "step": 17240
    },
    {
      "epoch": 2.8132137030995104,
      "grad_norm": 14.185015678405762,
      "learning_rate": 4.515827370555202e-05,
      "loss": 0.1758,
      "num_input_tokens_seen": 37234064,
      "step": 17245
    },
    {
      "epoch": 2.8140293637846656,
      "grad_norm": 0.6374616622924805,
      "learning_rate": 4.515406285383446e-05,
      "loss": 0.2708,
      "num_input_tokens_seen": 37245968,
      "step": 17250
    },
    {
      "epoch": 2.8148450244698204,
      "grad_norm": 3.457275867462158,
      "learning_rate": 4.514985036832628e-05,
      "loss": 0.2472,
      "num_input_tokens_seen": 37256432,
      "step": 17255
    },
    {
      "epoch": 2.8156606851549757,
      "grad_norm": 1.529738426208496,
      "learning_rate": 4.514563624936901e-05,
      "loss": 0.2225,
      "num_input_tokens_seen": 37267312,
      "step": 17260
    },
    {
      "epoch": 2.8164763458401305,
      "grad_norm": 0.34878918528556824,
      "learning_rate": 4.514142049730424e-05,
      "loss": 0.16,
      "num_input_tokens_seen": 37278256,
      "step": 17265
    },
    {
      "epoch": 2.8172920065252853,
      "grad_norm": 0.16346636414527893,
      "learning_rate": 4.513720311247374e-05,
      "loss": 0.095,
      "num_input_tokens_seen": 37289968,
      "step": 17270
    },
    {
      "epoch": 2.8181076672104406,
      "grad_norm": 0.13625754415988922,
      "learning_rate": 4.5132984095219364e-05,
      "loss": 0.071,
      "num_input_tokens_seen": 37301072,
      "step": 17275
    },
    {
      "epoch": 2.8189233278955954,
      "grad_norm": 4.504883289337158,
      "learning_rate": 4.512876344588315e-05,
      "loss": 0.2223,
      "num_input_tokens_seen": 37311664,
      "step": 17280
    },
    {
      "epoch": 2.8197389885807507,
      "grad_norm": 1.4539239406585693,
      "learning_rate": 4.512454116480724e-05,
      "loss": 0.0761,
      "num_input_tokens_seen": 37323120,
      "step": 17285
    },
    {
      "epoch": 2.8205546492659055,
      "grad_norm": 0.3599880635738373,
      "learning_rate": 4.512031725233391e-05,
      "loss": 0.0756,
      "num_input_tokens_seen": 37334672,
      "step": 17290
    },
    {
      "epoch": 2.8213703099510603,
      "grad_norm": 0.08150095492601395,
      "learning_rate": 4.5116091708805575e-05,
      "loss": 0.0938,
      "num_input_tokens_seen": 37346384,
      "step": 17295
    },
    {
      "epoch": 2.822185970636215,
      "grad_norm": 0.6873994469642639,
      "learning_rate": 4.5111864534564776e-05,
      "loss": 0.0132,
      "num_input_tokens_seen": 37358256,
      "step": 17300
    },
    {
      "epoch": 2.8230016313213704,
      "grad_norm": 5.348658084869385,
      "learning_rate": 4.510763572995419e-05,
      "loss": 0.1198,
      "num_input_tokens_seen": 37369968,
      "step": 17305
    },
    {
      "epoch": 2.823817292006525,
      "grad_norm": 0.5481806993484497,
      "learning_rate": 4.5103405295316634e-05,
      "loss": 0.2089,
      "num_input_tokens_seen": 37380432,
      "step": 17310
    },
    {
      "epoch": 2.8246329526916805,
      "grad_norm": 4.319050312042236,
      "learning_rate": 4.509917323099504e-05,
      "loss": 0.3818,
      "num_input_tokens_seen": 37390480,
      "step": 17315
    },
    {
      "epoch": 2.8254486133768353,
      "grad_norm": 0.35110506415367126,
      "learning_rate": 4.5094939537332475e-05,
      "loss": 0.1083,
      "num_input_tokens_seen": 37402608,
      "step": 17320
    },
    {
      "epoch": 2.82626427406199,
      "grad_norm": 0.6002646088600159,
      "learning_rate": 4.5090704214672155e-05,
      "loss": 0.0823,
      "num_input_tokens_seen": 37413936,
      "step": 17325
    },
    {
      "epoch": 2.827079934747145,
      "grad_norm": 0.06697563827037811,
      "learning_rate": 4.508646726335741e-05,
      "loss": 0.0499,
      "num_input_tokens_seen": 37426480,
      "step": 17330
    },
    {
      "epoch": 2.8278955954323,
      "grad_norm": 5.018412113189697,
      "learning_rate": 4.508222868373171e-05,
      "loss": 0.2247,
      "num_input_tokens_seen": 37438224,
      "step": 17335
    },
    {
      "epoch": 2.828711256117455,
      "grad_norm": 0.06362497806549072,
      "learning_rate": 4.507798847613866e-05,
      "loss": 0.0617,
      "num_input_tokens_seen": 37448176,
      "step": 17340
    },
    {
      "epoch": 2.8295269168026103,
      "grad_norm": 0.2531951367855072,
      "learning_rate": 4.507374664092199e-05,
      "loss": 0.1934,
      "num_input_tokens_seen": 37458672,
      "step": 17345
    },
    {
      "epoch": 2.830342577487765,
      "grad_norm": 0.12508904933929443,
      "learning_rate": 4.506950317842556e-05,
      "loss": 0.2338,
      "num_input_tokens_seen": 37469840,
      "step": 17350
    },
    {
      "epoch": 2.83115823817292,
      "grad_norm": 3.515825033187866,
      "learning_rate": 4.506525808899337e-05,
      "loss": 0.2209,
      "num_input_tokens_seen": 37479376,
      "step": 17355
    },
    {
      "epoch": 2.831973898858075,
      "grad_norm": 3.7540087699890137,
      "learning_rate": 4.506101137296955e-05,
      "loss": 0.1844,
      "num_input_tokens_seen": 37489392,
      "step": 17360
    },
    {
      "epoch": 2.83278955954323,
      "grad_norm": 1.872329831123352,
      "learning_rate": 4.505676303069837e-05,
      "loss": 0.1234,
      "num_input_tokens_seen": 37499504,
      "step": 17365
    },
    {
      "epoch": 2.8336052202283852,
      "grad_norm": 0.8481312990188599,
      "learning_rate": 4.5052513062524196e-05,
      "loss": 0.1183,
      "num_input_tokens_seen": 37510512,
      "step": 17370
    },
    {
      "epoch": 2.83442088091354,
      "grad_norm": 2.0243680477142334,
      "learning_rate": 4.504826146879158e-05,
      "loss": 0.1912,
      "num_input_tokens_seen": 37521424,
      "step": 17375
    },
    {
      "epoch": 2.835236541598695,
      "grad_norm": 2.5130465030670166,
      "learning_rate": 4.504400824984516e-05,
      "loss": 0.267,
      "num_input_tokens_seen": 37532976,
      "step": 17380
    },
    {
      "epoch": 2.8360522022838497,
      "grad_norm": 1.1350537538528442,
      "learning_rate": 4.503975340602973e-05,
      "loss": 0.2528,
      "num_input_tokens_seen": 37543504,
      "step": 17385
    },
    {
      "epoch": 2.836867862969005,
      "grad_norm": 0.19420993328094482,
      "learning_rate": 4.50354969376902e-05,
      "loss": 0.143,
      "num_input_tokens_seen": 37554000,
      "step": 17390
    },
    {
      "epoch": 2.8376835236541598,
      "grad_norm": 3.076580762863159,
      "learning_rate": 4.5031238845171644e-05,
      "loss": 0.1128,
      "num_input_tokens_seen": 37565360,
      "step": 17395
    },
    {
      "epoch": 2.838499184339315,
      "grad_norm": 0.17554321885108948,
      "learning_rate": 4.502697912881923e-05,
      "loss": 0.2298,
      "num_input_tokens_seen": 37576368,
      "step": 17400
    },
    {
      "epoch": 2.83931484502447,
      "grad_norm": 0.1670863777399063,
      "learning_rate": 4.502271778897825e-05,
      "loss": 0.1485,
      "num_input_tokens_seen": 37587664,
      "step": 17405
    },
    {
      "epoch": 2.8401305057096247,
      "grad_norm": 0.5998808741569519,
      "learning_rate": 4.50184548259942e-05,
      "loss": 0.2266,
      "num_input_tokens_seen": 37598256,
      "step": 17410
    },
    {
      "epoch": 2.84094616639478,
      "grad_norm": 0.2259002923965454,
      "learning_rate": 4.501419024021261e-05,
      "loss": 0.1715,
      "num_input_tokens_seen": 37608976,
      "step": 17415
    },
    {
      "epoch": 2.8417618270799347,
      "grad_norm": 1.2721748352050781,
      "learning_rate": 4.500992403197921e-05,
      "loss": 0.1692,
      "num_input_tokens_seen": 37618352,
      "step": 17420
    },
    {
      "epoch": 2.8425774877650896,
      "grad_norm": 0.22586800158023834,
      "learning_rate": 4.500565620163985e-05,
      "loss": 0.1697,
      "num_input_tokens_seen": 37628816,
      "step": 17425
    },
    {
      "epoch": 2.843393148450245,
      "grad_norm": 0.13375671207904816,
      "learning_rate": 4.500138674954047e-05,
      "loss": 0.0252,
      "num_input_tokens_seen": 37639824,
      "step": 17430
    },
    {
      "epoch": 2.8442088091353996,
      "grad_norm": 3.517141342163086,
      "learning_rate": 4.499711567602721e-05,
      "loss": 0.1266,
      "num_input_tokens_seen": 37651504,
      "step": 17435
    },
    {
      "epoch": 2.8450244698205545,
      "grad_norm": 0.2667512595653534,
      "learning_rate": 4.499284298144629e-05,
      "loss": 0.0342,
      "num_input_tokens_seen": 37662992,
      "step": 17440
    },
    {
      "epoch": 2.8458401305057097,
      "grad_norm": 0.11251013725996017,
      "learning_rate": 4.498856866614407e-05,
      "loss": 0.0498,
      "num_input_tokens_seen": 37673904,
      "step": 17445
    },
    {
      "epoch": 2.8466557911908645,
      "grad_norm": 3.6070191860198975,
      "learning_rate": 4.498429273046705e-05,
      "loss": 0.0733,
      "num_input_tokens_seen": 37684368,
      "step": 17450
    },
    {
      "epoch": 2.84747145187602,
      "grad_norm": 0.16577482223510742,
      "learning_rate": 4.498001517476187e-05,
      "loss": 0.0665,
      "num_input_tokens_seen": 37694480,
      "step": 17455
    },
    {
      "epoch": 2.8482871125611746,
      "grad_norm": 0.09861022979021072,
      "learning_rate": 4.497573599937528e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 37705040,
      "step": 17460
    },
    {
      "epoch": 2.8491027732463294,
      "grad_norm": 3.8634822368621826,
      "learning_rate": 4.497145520465417e-05,
      "loss": 0.2386,
      "num_input_tokens_seen": 37716304,
      "step": 17465
    },
    {
      "epoch": 2.8499184339314843,
      "grad_norm": 0.17338253557682037,
      "learning_rate": 4.4967172790945565e-05,
      "loss": 0.085,
      "num_input_tokens_seen": 37726352,
      "step": 17470
    },
    {
      "epoch": 2.8507340946166395,
      "grad_norm": 1.4482851028442383,
      "learning_rate": 4.496288875859663e-05,
      "loss": 0.0207,
      "num_input_tokens_seen": 37736048,
      "step": 17475
    },
    {
      "epoch": 2.8515497553017943,
      "grad_norm": 3.201935052871704,
      "learning_rate": 4.4958603107954635e-05,
      "loss": 0.1531,
      "num_input_tokens_seen": 37746704,
      "step": 17480
    },
    {
      "epoch": 2.8523654159869496,
      "grad_norm": 0.1465047299861908,
      "learning_rate": 4.4954315839367006e-05,
      "loss": 0.0595,
      "num_input_tokens_seen": 37758000,
      "step": 17485
    },
    {
      "epoch": 2.8531810766721044,
      "grad_norm": 0.07675353437662125,
      "learning_rate": 4.495002695318129e-05,
      "loss": 0.3736,
      "num_input_tokens_seen": 37768240,
      "step": 17490
    },
    {
      "epoch": 2.8539967373572592,
      "grad_norm": 0.2032974660396576,
      "learning_rate": 4.494573644974516e-05,
      "loss": 0.2442,
      "num_input_tokens_seen": 37778576,
      "step": 17495
    },
    {
      "epoch": 2.8548123980424145,
      "grad_norm": 0.1352575570344925,
      "learning_rate": 4.494144432940643e-05,
      "loss": 0.0453,
      "num_input_tokens_seen": 37788880,
      "step": 17500
    },
    {
      "epoch": 2.8556280587275693,
      "grad_norm": 0.09475905448198318,
      "learning_rate": 4.493715059251304e-05,
      "loss": 0.04,
      "num_input_tokens_seen": 37799856,
      "step": 17505
    },
    {
      "epoch": 2.8564437194127246,
      "grad_norm": 0.9383955597877502,
      "learning_rate": 4.4932855239413065e-05,
      "loss": 0.2561,
      "num_input_tokens_seen": 37809808,
      "step": 17510
    },
    {
      "epoch": 2.8572593800978794,
      "grad_norm": 4.063689708709717,
      "learning_rate": 4.49285582704547e-05,
      "loss": 0.2876,
      "num_input_tokens_seen": 37819856,
      "step": 17515
    },
    {
      "epoch": 2.858075040783034,
      "grad_norm": 2.331108331680298,
      "learning_rate": 4.492425968598629e-05,
      "loss": 0.1281,
      "num_input_tokens_seen": 37831792,
      "step": 17520
    },
    {
      "epoch": 2.858890701468189,
      "grad_norm": 2.068864583969116,
      "learning_rate": 4.49199594863563e-05,
      "loss": 0.0916,
      "num_input_tokens_seen": 37843280,
      "step": 17525
    },
    {
      "epoch": 2.8597063621533443,
      "grad_norm": 1.9548044204711914,
      "learning_rate": 4.4915657671913314e-05,
      "loss": 0.1012,
      "num_input_tokens_seen": 37854928,
      "step": 17530
    },
    {
      "epoch": 2.860522022838499,
      "grad_norm": 0.30348631739616394,
      "learning_rate": 4.491135424300607e-05,
      "loss": 0.1112,
      "num_input_tokens_seen": 37865424,
      "step": 17535
    },
    {
      "epoch": 2.8613376835236544,
      "grad_norm": 2.8615989685058594,
      "learning_rate": 4.490704919998342e-05,
      "loss": 0.0659,
      "num_input_tokens_seen": 37876624,
      "step": 17540
    },
    {
      "epoch": 2.862153344208809,
      "grad_norm": 2.7815873622894287,
      "learning_rate": 4.4902742543194356e-05,
      "loss": 0.1882,
      "num_input_tokens_seen": 37886896,
      "step": 17545
    },
    {
      "epoch": 2.862969004893964,
      "grad_norm": 2.147285223007202,
      "learning_rate": 4.4898434272988e-05,
      "loss": 0.3117,
      "num_input_tokens_seen": 37897424,
      "step": 17550
    },
    {
      "epoch": 2.863784665579119,
      "grad_norm": 0.15618032217025757,
      "learning_rate": 4.489412438971359e-05,
      "loss": 0.0881,
      "num_input_tokens_seen": 37908080,
      "step": 17555
    },
    {
      "epoch": 2.864600326264274,
      "grad_norm": 0.3206488788127899,
      "learning_rate": 4.488981289372052e-05,
      "loss": 0.0313,
      "num_input_tokens_seen": 37918768,
      "step": 17560
    },
    {
      "epoch": 2.865415986949429,
      "grad_norm": 2.207361936569214,
      "learning_rate": 4.488549978535829e-05,
      "loss": 0.2666,
      "num_input_tokens_seen": 37929488,
      "step": 17565
    },
    {
      "epoch": 2.866231647634584,
      "grad_norm": 0.21528835594654083,
      "learning_rate": 4.4881185064976553e-05,
      "loss": 0.1354,
      "num_input_tokens_seen": 37940560,
      "step": 17570
    },
    {
      "epoch": 2.867047308319739,
      "grad_norm": 0.8961257934570312,
      "learning_rate": 4.487686873292508e-05,
      "loss": 0.2215,
      "num_input_tokens_seen": 37951248,
      "step": 17575
    },
    {
      "epoch": 2.867862969004894,
      "grad_norm": 0.05624648556113243,
      "learning_rate": 4.487255078955378e-05,
      "loss": 0.0352,
      "num_input_tokens_seen": 37961936,
      "step": 17580
    },
    {
      "epoch": 2.868678629690049,
      "grad_norm": 4.384384632110596,
      "learning_rate": 4.486823123521267e-05,
      "loss": 0.2172,
      "num_input_tokens_seen": 37972560,
      "step": 17585
    },
    {
      "epoch": 2.869494290375204,
      "grad_norm": 0.31094902753829956,
      "learning_rate": 4.4863910070251927e-05,
      "loss": 0.0444,
      "num_input_tokens_seen": 37984016,
      "step": 17590
    },
    {
      "epoch": 2.870309951060359,
      "grad_norm": 0.17169949412345886,
      "learning_rate": 4.485958729502185e-05,
      "loss": 0.0879,
      "num_input_tokens_seen": 37994896,
      "step": 17595
    },
    {
      "epoch": 2.871125611745514,
      "grad_norm": 0.37429696321487427,
      "learning_rate": 4.485526290987286e-05,
      "loss": 0.0973,
      "num_input_tokens_seen": 38006832,
      "step": 17600
    },
    {
      "epoch": 2.8719412724306688,
      "grad_norm": 3.762223720550537,
      "learning_rate": 4.485093691515551e-05,
      "loss": 0.1567,
      "num_input_tokens_seen": 38018832,
      "step": 17605
    },
    {
      "epoch": 2.8727569331158236,
      "grad_norm": 0.08808736503124237,
      "learning_rate": 4.4846609311220494e-05,
      "loss": 0.2442,
      "num_input_tokens_seen": 38029520,
      "step": 17610
    },
    {
      "epoch": 2.873572593800979,
      "grad_norm": 0.18813839554786682,
      "learning_rate": 4.484228009841863e-05,
      "loss": 0.0607,
      "num_input_tokens_seen": 38039568,
      "step": 17615
    },
    {
      "epoch": 2.8743882544861337,
      "grad_norm": 0.12047629803419113,
      "learning_rate": 4.483794927710085e-05,
      "loss": 0.1261,
      "num_input_tokens_seen": 38051344,
      "step": 17620
    },
    {
      "epoch": 2.875203915171289,
      "grad_norm": 3.394768714904785,
      "learning_rate": 4.483361684761826e-05,
      "loss": 0.1065,
      "num_input_tokens_seen": 38062448,
      "step": 17625
    },
    {
      "epoch": 2.8760195758564437,
      "grad_norm": 1.3142900466918945,
      "learning_rate": 4.482928281032205e-05,
      "loss": 0.0506,
      "num_input_tokens_seen": 38073520,
      "step": 17630
    },
    {
      "epoch": 2.8768352365415986,
      "grad_norm": 4.224359035491943,
      "learning_rate": 4.482494716556356e-05,
      "loss": 0.1867,
      "num_input_tokens_seen": 38084400,
      "step": 17635
    },
    {
      "epoch": 2.877650897226754,
      "grad_norm": 0.025160379707813263,
      "learning_rate": 4.482060991369426e-05,
      "loss": 0.2161,
      "num_input_tokens_seen": 38094288,
      "step": 17640
    },
    {
      "epoch": 2.8784665579119086,
      "grad_norm": 0.8182905316352844,
      "learning_rate": 4.481627105506575e-05,
      "loss": 0.0301,
      "num_input_tokens_seen": 38104752,
      "step": 17645
    },
    {
      "epoch": 2.8792822185970635,
      "grad_norm": 0.20233500003814697,
      "learning_rate": 4.481193059002976e-05,
      "loss": 0.0753,
      "num_input_tokens_seen": 38113456,
      "step": 17650
    },
    {
      "epoch": 2.8800978792822187,
      "grad_norm": 0.43862026929855347,
      "learning_rate": 4.480758851893816e-05,
      "loss": 0.0842,
      "num_input_tokens_seen": 38123888,
      "step": 17655
    },
    {
      "epoch": 2.8809135399673735,
      "grad_norm": 2.028978109359741,
      "learning_rate": 4.480324484214293e-05,
      "loss": 0.0703,
      "num_input_tokens_seen": 38134448,
      "step": 17660
    },
    {
      "epoch": 2.8817292006525284,
      "grad_norm": 6.340385913848877,
      "learning_rate": 4.479889955999619e-05,
      "loss": 0.3515,
      "num_input_tokens_seen": 38145104,
      "step": 17665
    },
    {
      "epoch": 2.8825448613376836,
      "grad_norm": 2.516404390335083,
      "learning_rate": 4.4794552672850185e-05,
      "loss": 0.259,
      "num_input_tokens_seen": 38155984,
      "step": 17670
    },
    {
      "epoch": 2.8833605220228384,
      "grad_norm": 4.506988525390625,
      "learning_rate": 4.479020418105732e-05,
      "loss": 0.2115,
      "num_input_tokens_seen": 38167408,
      "step": 17675
    },
    {
      "epoch": 2.8841761827079937,
      "grad_norm": 6.194934844970703,
      "learning_rate": 4.478585408497008e-05,
      "loss": 0.123,
      "num_input_tokens_seen": 38178096,
      "step": 17680
    },
    {
      "epoch": 2.8849918433931485,
      "grad_norm": 0.3223642408847809,
      "learning_rate": 4.478150238494112e-05,
      "loss": 0.0133,
      "num_input_tokens_seen": 38188304,
      "step": 17685
    },
    {
      "epoch": 2.8858075040783033,
      "grad_norm": 4.470766544342041,
      "learning_rate": 4.47771490813232e-05,
      "loss": 0.1234,
      "num_input_tokens_seen": 38198512,
      "step": 17690
    },
    {
      "epoch": 2.886623164763458,
      "grad_norm": 0.050803132355213165,
      "learning_rate": 4.4772794174469234e-05,
      "loss": 0.1378,
      "num_input_tokens_seen": 38209840,
      "step": 17695
    },
    {
      "epoch": 2.8874388254486134,
      "grad_norm": 0.3100869357585907,
      "learning_rate": 4.4768437664732244e-05,
      "loss": 0.1365,
      "num_input_tokens_seen": 38219760,
      "step": 17700
    },
    {
      "epoch": 2.8882544861337682,
      "grad_norm": 0.9346982836723328,
      "learning_rate": 4.4764079552465385e-05,
      "loss": 0.0195,
      "num_input_tokens_seen": 38231376,
      "step": 17705
    },
    {
      "epoch": 2.8890701468189235,
      "grad_norm": 3.9480233192443848,
      "learning_rate": 4.475971983802196e-05,
      "loss": 0.1957,
      "num_input_tokens_seen": 38242096,
      "step": 17710
    },
    {
      "epoch": 2.8898858075040783,
      "grad_norm": 0.06706345081329346,
      "learning_rate": 4.475535852175539e-05,
      "loss": 0.0324,
      "num_input_tokens_seen": 38252624,
      "step": 17715
    },
    {
      "epoch": 2.890701468189233,
      "grad_norm": 2.782609701156616,
      "learning_rate": 4.475099560401922e-05,
      "loss": 0.0684,
      "num_input_tokens_seen": 38262864,
      "step": 17720
    },
    {
      "epoch": 2.8915171288743884,
      "grad_norm": 0.2172604203224182,
      "learning_rate": 4.474663108516713e-05,
      "loss": 0.0215,
      "num_input_tokens_seen": 38274640,
      "step": 17725
    },
    {
      "epoch": 2.892332789559543,
      "grad_norm": 0.06671618670225143,
      "learning_rate": 4.474226496555293e-05,
      "loss": 0.0152,
      "num_input_tokens_seen": 38286512,
      "step": 17730
    },
    {
      "epoch": 2.8931484502446985,
      "grad_norm": 0.036211103200912476,
      "learning_rate": 4.473789724553056e-05,
      "loss": 0.0822,
      "num_input_tokens_seen": 38297456,
      "step": 17735
    },
    {
      "epoch": 2.8939641109298533,
      "grad_norm": 0.7471785545349121,
      "learning_rate": 4.473352792545409e-05,
      "loss": 0.1271,
      "num_input_tokens_seen": 38308304,
      "step": 17740
    },
    {
      "epoch": 2.894779771615008,
      "grad_norm": 0.4336090683937073,
      "learning_rate": 4.4729157005677724e-05,
      "loss": 0.2564,
      "num_input_tokens_seen": 38318640,
      "step": 17745
    },
    {
      "epoch": 2.895595432300163,
      "grad_norm": 0.1255933791399002,
      "learning_rate": 4.472478448655578e-05,
      "loss": 0.0443,
      "num_input_tokens_seen": 38330320,
      "step": 17750
    },
    {
      "epoch": 2.896411092985318,
      "grad_norm": 0.5787214636802673,
      "learning_rate": 4.4720410368442724e-05,
      "loss": 0.0185,
      "num_input_tokens_seen": 38340432,
      "step": 17755
    },
    {
      "epoch": 2.897226753670473,
      "grad_norm": 5.940229892730713,
      "learning_rate": 4.471603465169314e-05,
      "loss": 0.1063,
      "num_input_tokens_seen": 38351760,
      "step": 17760
    },
    {
      "epoch": 2.8980424143556283,
      "grad_norm": 2.6881027221679688,
      "learning_rate": 4.471165733666176e-05,
      "loss": 0.0283,
      "num_input_tokens_seen": 38361136,
      "step": 17765
    },
    {
      "epoch": 2.898858075040783,
      "grad_norm": 0.05304047837853432,
      "learning_rate": 4.4707278423703415e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 38371408,
      "step": 17770
    },
    {
      "epoch": 2.899673735725938,
      "grad_norm": 4.1059064865112305,
      "learning_rate": 4.470289791317308e-05,
      "loss": 0.1215,
      "num_input_tokens_seen": 38383056,
      "step": 17775
    },
    {
      "epoch": 2.9004893964110927,
      "grad_norm": 0.48627591133117676,
      "learning_rate": 4.4698515805425876e-05,
      "loss": 0.0077,
      "num_input_tokens_seen": 38393840,
      "step": 17780
    },
    {
      "epoch": 2.901305057096248,
      "grad_norm": 0.21503886580467224,
      "learning_rate": 4.469413210081703e-05,
      "loss": 0.0651,
      "num_input_tokens_seen": 38406096,
      "step": 17785
    },
    {
      "epoch": 2.902120717781403,
      "grad_norm": 0.11739999055862427,
      "learning_rate": 4.468974679970191e-05,
      "loss": 0.1323,
      "num_input_tokens_seen": 38416912,
      "step": 17790
    },
    {
      "epoch": 2.902936378466558,
      "grad_norm": 6.5325541496276855,
      "learning_rate": 4.468535990243601e-05,
      "loss": 0.175,
      "num_input_tokens_seen": 38428016,
      "step": 17795
    },
    {
      "epoch": 2.903752039151713,
      "grad_norm": 5.3871307373046875,
      "learning_rate": 4.468097140937495e-05,
      "loss": 0.1288,
      "num_input_tokens_seen": 38438480,
      "step": 17800
    },
    {
      "epoch": 2.9045676998368677,
      "grad_norm": 3.131307601928711,
      "learning_rate": 4.467658132087449e-05,
      "loss": 0.0539,
      "num_input_tokens_seen": 38448112,
      "step": 17805
    },
    {
      "epoch": 2.905383360522023,
      "grad_norm": 4.033162593841553,
      "learning_rate": 4.4672189637290505e-05,
      "loss": 0.2172,
      "num_input_tokens_seen": 38457872,
      "step": 17810
    },
    {
      "epoch": 2.9061990212071778,
      "grad_norm": 0.10378050059080124,
      "learning_rate": 4.466779635897902e-05,
      "loss": 0.2298,
      "num_input_tokens_seen": 38469040,
      "step": 17815
    },
    {
      "epoch": 2.907014681892333,
      "grad_norm": 8.18344783782959,
      "learning_rate": 4.466340148629617e-05,
      "loss": 0.2725,
      "num_input_tokens_seen": 38479344,
      "step": 17820
    },
    {
      "epoch": 2.907830342577488,
      "grad_norm": 3.1844842433929443,
      "learning_rate": 4.465900501959822e-05,
      "loss": 0.1133,
      "num_input_tokens_seen": 38489552,
      "step": 17825
    },
    {
      "epoch": 2.9086460032626427,
      "grad_norm": 2.472390651702881,
      "learning_rate": 4.465460695924157e-05,
      "loss": 0.1878,
      "num_input_tokens_seen": 38500752,
      "step": 17830
    },
    {
      "epoch": 2.9094616639477975,
      "grad_norm": 3.5917232036590576,
      "learning_rate": 4.4650207305582756e-05,
      "loss": 0.1565,
      "num_input_tokens_seen": 38512432,
      "step": 17835
    },
    {
      "epoch": 2.9102773246329527,
      "grad_norm": 0.044487133622169495,
      "learning_rate": 4.464580605897844e-05,
      "loss": 0.115,
      "num_input_tokens_seen": 38522256,
      "step": 17840
    },
    {
      "epoch": 2.9110929853181076,
      "grad_norm": 0.19997821748256683,
      "learning_rate": 4.4641403219785396e-05,
      "loss": 0.2138,
      "num_input_tokens_seen": 38532784,
      "step": 17845
    },
    {
      "epoch": 2.911908646003263,
      "grad_norm": 4.2811408042907715,
      "learning_rate": 4.463699878836055e-05,
      "loss": 0.2576,
      "num_input_tokens_seen": 38542640,
      "step": 17850
    },
    {
      "epoch": 2.9127243066884176,
      "grad_norm": 3.429095506668091,
      "learning_rate": 4.463259276506095e-05,
      "loss": 0.1889,
      "num_input_tokens_seen": 38553008,
      "step": 17855
    },
    {
      "epoch": 2.9135399673735725,
      "grad_norm": 3.9942924976348877,
      "learning_rate": 4.462818515024376e-05,
      "loss": 0.2476,
      "num_input_tokens_seen": 38565200,
      "step": 17860
    },
    {
      "epoch": 2.9143556280587277,
      "grad_norm": 0.1372908353805542,
      "learning_rate": 4.462377594426629e-05,
      "loss": 0.1281,
      "num_input_tokens_seen": 38574736,
      "step": 17865
    },
    {
      "epoch": 2.9151712887438825,
      "grad_norm": 0.20757682621479034,
      "learning_rate": 4.461936514748597e-05,
      "loss": 0.0215,
      "num_input_tokens_seen": 38584496,
      "step": 17870
    },
    {
      "epoch": 2.9159869494290374,
      "grad_norm": 0.09978707879781723,
      "learning_rate": 4.4614952760260366e-05,
      "loss": 0.147,
      "num_input_tokens_seen": 38594448,
      "step": 17875
    },
    {
      "epoch": 2.9168026101141926,
      "grad_norm": 0.2199597954750061,
      "learning_rate": 4.4610538782947166e-05,
      "loss": 0.0174,
      "num_input_tokens_seen": 38604976,
      "step": 17880
    },
    {
      "epoch": 2.9176182707993474,
      "grad_norm": 0.1524445116519928,
      "learning_rate": 4.460612321590419e-05,
      "loss": 0.2321,
      "num_input_tokens_seen": 38615792,
      "step": 17885
    },
    {
      "epoch": 2.9184339314845023,
      "grad_norm": 2.3187215328216553,
      "learning_rate": 4.460170605948939e-05,
      "loss": 0.2173,
      "num_input_tokens_seen": 38626320,
      "step": 17890
    },
    {
      "epoch": 2.9192495921696575,
      "grad_norm": 3.7369632720947266,
      "learning_rate": 4.459728731406083e-05,
      "loss": 0.1052,
      "num_input_tokens_seen": 38637008,
      "step": 17895
    },
    {
      "epoch": 2.9200652528548123,
      "grad_norm": 1.549872636795044,
      "learning_rate": 4.4592866979976725e-05,
      "loss": 0.0231,
      "num_input_tokens_seen": 38647056,
      "step": 17900
    },
    {
      "epoch": 2.9208809135399676,
      "grad_norm": 9.33074951171875,
      "learning_rate": 4.458844505759542e-05,
      "loss": 0.4205,
      "num_input_tokens_seen": 38658512,
      "step": 17905
    },
    {
      "epoch": 2.9216965742251224,
      "grad_norm": 0.3138631582260132,
      "learning_rate": 4.4584021547275356e-05,
      "loss": 0.0993,
      "num_input_tokens_seen": 38669328,
      "step": 17910
    },
    {
      "epoch": 2.9225122349102772,
      "grad_norm": 0.30171555280685425,
      "learning_rate": 4.457959644937514e-05,
      "loss": 0.1021,
      "num_input_tokens_seen": 38680048,
      "step": 17915
    },
    {
      "epoch": 2.923327895595432,
      "grad_norm": 0.22459585964679718,
      "learning_rate": 4.457516976425349e-05,
      "loss": 0.2373,
      "num_input_tokens_seen": 38691952,
      "step": 17920
    },
    {
      "epoch": 2.9241435562805873,
      "grad_norm": 0.32993584871292114,
      "learning_rate": 4.457074149226926e-05,
      "loss": 0.0452,
      "num_input_tokens_seen": 38701872,
      "step": 17925
    },
    {
      "epoch": 2.924959216965742,
      "grad_norm": 0.06956195831298828,
      "learning_rate": 4.456631163378142e-05,
      "loss": 0.0127,
      "num_input_tokens_seen": 38713104,
      "step": 17930
    },
    {
      "epoch": 2.9257748776508974,
      "grad_norm": 0.13496209681034088,
      "learning_rate": 4.456188018914908e-05,
      "loss": 0.244,
      "num_input_tokens_seen": 38723952,
      "step": 17935
    },
    {
      "epoch": 2.926590538336052,
      "grad_norm": 0.15551023185253143,
      "learning_rate": 4.455744715873148e-05,
      "loss": 0.0215,
      "num_input_tokens_seen": 38735216,
      "step": 17940
    },
    {
      "epoch": 2.927406199021207,
      "grad_norm": 0.15168346464633942,
      "learning_rate": 4.455301254288797e-05,
      "loss": 0.1295,
      "num_input_tokens_seen": 38746032,
      "step": 17945
    },
    {
      "epoch": 2.9282218597063623,
      "grad_norm": 3.0744872093200684,
      "learning_rate": 4.454857634197806e-05,
      "loss": 0.1594,
      "num_input_tokens_seen": 38756464,
      "step": 17950
    },
    {
      "epoch": 2.929037520391517,
      "grad_norm": 1.041496992111206,
      "learning_rate": 4.4544138556361364e-05,
      "loss": 0.1836,
      "num_input_tokens_seen": 38767952,
      "step": 17955
    },
    {
      "epoch": 2.9298531810766724,
      "grad_norm": 0.09206711500883102,
      "learning_rate": 4.453969918639763e-05,
      "loss": 0.044,
      "num_input_tokens_seen": 38779312,
      "step": 17960
    },
    {
      "epoch": 2.930668841761827,
      "grad_norm": 0.10270342230796814,
      "learning_rate": 4.453525823244673e-05,
      "loss": 0.0388,
      "num_input_tokens_seen": 38790576,
      "step": 17965
    },
    {
      "epoch": 2.931484502446982,
      "grad_norm": 1.683138132095337,
      "learning_rate": 4.453081569486869e-05,
      "loss": 0.1008,
      "num_input_tokens_seen": 38802032,
      "step": 17970
    },
    {
      "epoch": 2.932300163132137,
      "grad_norm": 1.897834062576294,
      "learning_rate": 4.452637157402362e-05,
      "loss": 0.2771,
      "num_input_tokens_seen": 38812080,
      "step": 17975
    },
    {
      "epoch": 2.933115823817292,
      "grad_norm": 4.9028425216674805,
      "learning_rate": 4.45219258702718e-05,
      "loss": 0.2769,
      "num_input_tokens_seen": 38823664,
      "step": 17980
    },
    {
      "epoch": 2.933931484502447,
      "grad_norm": 3.6018197536468506,
      "learning_rate": 4.451747858397361e-05,
      "loss": 0.1651,
      "num_input_tokens_seen": 38833840,
      "step": 17985
    },
    {
      "epoch": 2.934747145187602,
      "grad_norm": 2.3101999759674072,
      "learning_rate": 4.451302971548958e-05,
      "loss": 0.1877,
      "num_input_tokens_seen": 38844400,
      "step": 17990
    },
    {
      "epoch": 2.935562805872757,
      "grad_norm": 0.25400373339653015,
      "learning_rate": 4.450857926518035e-05,
      "loss": 0.0593,
      "num_input_tokens_seen": 38855248,
      "step": 17995
    },
    {
      "epoch": 2.936378466557912,
      "grad_norm": 0.1637653261423111,
      "learning_rate": 4.45041272334067e-05,
      "loss": 0.099,
      "num_input_tokens_seen": 38867472,
      "step": 18000
    },
    {
      "epoch": 2.9371941272430666,
      "grad_norm": 0.24362194538116455,
      "learning_rate": 4.449967362052954e-05,
      "loss": 0.0279,
      "num_input_tokens_seen": 38878864,
      "step": 18005
    },
    {
      "epoch": 2.938009787928222,
      "grad_norm": 3.1802594661712646,
      "learning_rate": 4.449521842690989e-05,
      "loss": 0.1694,
      "num_input_tokens_seen": 38890032,
      "step": 18010
    },
    {
      "epoch": 2.9388254486133767,
      "grad_norm": 0.14970026910305023,
      "learning_rate": 4.449076165290892e-05,
      "loss": 0.0931,
      "num_input_tokens_seen": 38900624,
      "step": 18015
    },
    {
      "epoch": 2.939641109298532,
      "grad_norm": 0.10972364991903305,
      "learning_rate": 4.448630329888791e-05,
      "loss": 0.0187,
      "num_input_tokens_seen": 38911984,
      "step": 18020
    },
    {
      "epoch": 2.9404567699836868,
      "grad_norm": 1.4522041082382202,
      "learning_rate": 4.448184336520829e-05,
      "loss": 0.0246,
      "num_input_tokens_seen": 38922896,
      "step": 18025
    },
    {
      "epoch": 2.9412724306688416,
      "grad_norm": 4.203207492828369,
      "learning_rate": 4.447738185223158e-05,
      "loss": 0.2772,
      "num_input_tokens_seen": 38933872,
      "step": 18030
    },
    {
      "epoch": 2.942088091353997,
      "grad_norm": 8.245977401733398,
      "learning_rate": 4.447291876031949e-05,
      "loss": 0.0493,
      "num_input_tokens_seen": 38943088,
      "step": 18035
    },
    {
      "epoch": 2.9429037520391517,
      "grad_norm": 4.124063968658447,
      "learning_rate": 4.446845408983379e-05,
      "loss": 0.2852,
      "num_input_tokens_seen": 38954064,
      "step": 18040
    },
    {
      "epoch": 2.943719412724307,
      "grad_norm": 0.08290500938892365,
      "learning_rate": 4.446398784113642e-05,
      "loss": 0.0899,
      "num_input_tokens_seen": 38965904,
      "step": 18045
    },
    {
      "epoch": 2.9445350734094617,
      "grad_norm": 0.26444438099861145,
      "learning_rate": 4.445952001458944e-05,
      "loss": 0.1542,
      "num_input_tokens_seen": 38976080,
      "step": 18050
    },
    {
      "epoch": 2.9453507340946166,
      "grad_norm": 0.2069074660539627,
      "learning_rate": 4.445505061055503e-05,
      "loss": 0.0629,
      "num_input_tokens_seen": 38987344,
      "step": 18055
    },
    {
      "epoch": 2.9461663947797714,
      "grad_norm": 14.2250394821167,
      "learning_rate": 4.44505796293955e-05,
      "loss": 0.0465,
      "num_input_tokens_seen": 38999472,
      "step": 18060
    },
    {
      "epoch": 2.9469820554649266,
      "grad_norm": 0.08218816667795181,
      "learning_rate": 4.44461070714733e-05,
      "loss": 0.1194,
      "num_input_tokens_seen": 39010352,
      "step": 18065
    },
    {
      "epoch": 2.9477977161500815,
      "grad_norm": 0.08249372243881226,
      "learning_rate": 4.4441632937150984e-05,
      "loss": 0.1605,
      "num_input_tokens_seen": 39020816,
      "step": 18070
    },
    {
      "epoch": 2.9486133768352367,
      "grad_norm": 3.6566240787506104,
      "learning_rate": 4.443715722679126e-05,
      "loss": 0.157,
      "num_input_tokens_seen": 39032112,
      "step": 18075
    },
    {
      "epoch": 2.9494290375203915,
      "grad_norm": 0.048471830785274506,
      "learning_rate": 4.443267994075695e-05,
      "loss": 0.0101,
      "num_input_tokens_seen": 39043536,
      "step": 18080
    },
    {
      "epoch": 2.9502446982055464,
      "grad_norm": 3.1837503910064697,
      "learning_rate": 4.4428201079411004e-05,
      "loss": 0.122,
      "num_input_tokens_seen": 39053232,
      "step": 18085
    },
    {
      "epoch": 2.9510603588907016,
      "grad_norm": 2.7203781604766846,
      "learning_rate": 4.4423720643116495e-05,
      "loss": 0.0211,
      "num_input_tokens_seen": 39064368,
      "step": 18090
    },
    {
      "epoch": 2.9518760195758564,
      "grad_norm": 0.42375457286834717,
      "learning_rate": 4.441923863223664e-05,
      "loss": 0.3618,
      "num_input_tokens_seen": 39072560,
      "step": 18095
    },
    {
      "epoch": 2.9526916802610113,
      "grad_norm": 0.13614337146282196,
      "learning_rate": 4.441475504713477e-05,
      "loss": 0.0832,
      "num_input_tokens_seen": 39084080,
      "step": 18100
    },
    {
      "epoch": 2.9535073409461665,
      "grad_norm": 3.6559746265411377,
      "learning_rate": 4.4410269888174346e-05,
      "loss": 0.1442,
      "num_input_tokens_seen": 39095056,
      "step": 18105
    },
    {
      "epoch": 2.9543230016313213,
      "grad_norm": 0.13492968678474426,
      "learning_rate": 4.440578315571896e-05,
      "loss": 0.1121,
      "num_input_tokens_seen": 39106000,
      "step": 18110
    },
    {
      "epoch": 2.955138662316476,
      "grad_norm": 4.666341781616211,
      "learning_rate": 4.4401294850132324e-05,
      "loss": 0.235,
      "num_input_tokens_seen": 39115760,
      "step": 18115
    },
    {
      "epoch": 2.9559543230016314,
      "grad_norm": 6.095045566558838,
      "learning_rate": 4.439680497177829e-05,
      "loss": 0.2459,
      "num_input_tokens_seen": 39126160,
      "step": 18120
    },
    {
      "epoch": 2.9567699836867862,
      "grad_norm": 1.2581567764282227,
      "learning_rate": 4.439231352102082e-05,
      "loss": 0.0505,
      "num_input_tokens_seen": 39137264,
      "step": 18125
    },
    {
      "epoch": 2.9575856443719415,
      "grad_norm": 4.644974231719971,
      "learning_rate": 4.438782049822403e-05,
      "loss": 0.0847,
      "num_input_tokens_seen": 39147824,
      "step": 18130
    },
    {
      "epoch": 2.9584013050570963,
      "grad_norm": 0.1869252324104309,
      "learning_rate": 4.4383325903752124e-05,
      "loss": 0.1519,
      "num_input_tokens_seen": 39158768,
      "step": 18135
    },
    {
      "epoch": 2.959216965742251,
      "grad_norm": 2.9593968391418457,
      "learning_rate": 4.437882973796948e-05,
      "loss": 0.0903,
      "num_input_tokens_seen": 39168752,
      "step": 18140
    },
    {
      "epoch": 2.960032626427406,
      "grad_norm": 0.4411729872226715,
      "learning_rate": 4.437433200124057e-05,
      "loss": 0.0114,
      "num_input_tokens_seen": 39180528,
      "step": 18145
    },
    {
      "epoch": 2.960848287112561,
      "grad_norm": 4.563373565673828,
      "learning_rate": 4.4369832693930005e-05,
      "loss": 0.2417,
      "num_input_tokens_seen": 39192336,
      "step": 18150
    },
    {
      "epoch": 2.961663947797716,
      "grad_norm": 1.0392076969146729,
      "learning_rate": 4.436533181640252e-05,
      "loss": 0.1119,
      "num_input_tokens_seen": 39203056,
      "step": 18155
    },
    {
      "epoch": 2.9624796084828713,
      "grad_norm": 0.15708614885807037,
      "learning_rate": 4.436082936902297e-05,
      "loss": 0.1149,
      "num_input_tokens_seen": 39213104,
      "step": 18160
    },
    {
      "epoch": 2.963295269168026,
      "grad_norm": 2.713299036026001,
      "learning_rate": 4.435632535215637e-05,
      "loss": 0.2427,
      "num_input_tokens_seen": 39223792,
      "step": 18165
    },
    {
      "epoch": 2.964110929853181,
      "grad_norm": 4.85723352432251,
      "learning_rate": 4.435181976616781e-05,
      "loss": 0.1196,
      "num_input_tokens_seen": 39235248,
      "step": 18170
    },
    {
      "epoch": 2.964926590538336,
      "grad_norm": 0.08172395080327988,
      "learning_rate": 4.4347312611422555e-05,
      "loss": 0.1422,
      "num_input_tokens_seen": 39245424,
      "step": 18175
    },
    {
      "epoch": 2.965742251223491,
      "grad_norm": 0.10064270347356796,
      "learning_rate": 4.434280388828598e-05,
      "loss": 0.2157,
      "num_input_tokens_seen": 39255344,
      "step": 18180
    },
    {
      "epoch": 2.9665579119086463,
      "grad_norm": 5.2386555671691895,
      "learning_rate": 4.433829359712356e-05,
      "loss": 0.0972,
      "num_input_tokens_seen": 39266192,
      "step": 18185
    },
    {
      "epoch": 2.967373572593801,
      "grad_norm": 2.407827615737915,
      "learning_rate": 4.4333781738300954e-05,
      "loss": 0.2413,
      "num_input_tokens_seen": 39277040,
      "step": 18190
    },
    {
      "epoch": 2.968189233278956,
      "grad_norm": 3.0858025550842285,
      "learning_rate": 4.43292683121839e-05,
      "loss": 0.1982,
      "num_input_tokens_seen": 39286160,
      "step": 18195
    },
    {
      "epoch": 2.9690048939641107,
      "grad_norm": 0.13510720431804657,
      "learning_rate": 4.432475331913828e-05,
      "loss": 0.0432,
      "num_input_tokens_seen": 39296496,
      "step": 18200
    },
    {
      "epoch": 2.969820554649266,
      "grad_norm": 3.1260018348693848,
      "learning_rate": 4.4320236759530095e-05,
      "loss": 0.1484,
      "num_input_tokens_seen": 39306672,
      "step": 18205
    },
    {
      "epoch": 2.970636215334421,
      "grad_norm": 0.10233663767576218,
      "learning_rate": 4.43157186337255e-05,
      "loss": 0.0118,
      "num_input_tokens_seen": 39316816,
      "step": 18210
    },
    {
      "epoch": 2.971451876019576,
      "grad_norm": 0.25618240237236023,
      "learning_rate": 4.431119894209074e-05,
      "loss": 0.1324,
      "num_input_tokens_seen": 39327984,
      "step": 18215
    },
    {
      "epoch": 2.972267536704731,
      "grad_norm": 0.2764376699924469,
      "learning_rate": 4.430667768499221e-05,
      "loss": 0.1514,
      "num_input_tokens_seen": 39338832,
      "step": 18220
    },
    {
      "epoch": 2.9730831973898857,
      "grad_norm": 0.2718583047389984,
      "learning_rate": 4.4302154862796425e-05,
      "loss": 0.1349,
      "num_input_tokens_seen": 39349104,
      "step": 18225
    },
    {
      "epoch": 2.9738988580750405,
      "grad_norm": 3.6538009643554688,
      "learning_rate": 4.4297630475870025e-05,
      "loss": 0.3551,
      "num_input_tokens_seen": 39358640,
      "step": 18230
    },
    {
      "epoch": 2.9747145187601958,
      "grad_norm": 0.119206503033638,
      "learning_rate": 4.429310452457979e-05,
      "loss": 0.0875,
      "num_input_tokens_seen": 39368688,
      "step": 18235
    },
    {
      "epoch": 2.9755301794453506,
      "grad_norm": 0.44838809967041016,
      "learning_rate": 4.428857700929261e-05,
      "loss": 0.0963,
      "num_input_tokens_seen": 39378224,
      "step": 18240
    },
    {
      "epoch": 2.976345840130506,
      "grad_norm": 0.9642930030822754,
      "learning_rate": 4.428404793037551e-05,
      "loss": 0.0235,
      "num_input_tokens_seen": 39388368,
      "step": 18245
    },
    {
      "epoch": 2.9771615008156607,
      "grad_norm": 3.36867618560791,
      "learning_rate": 4.427951728819564e-05,
      "loss": 0.163,
      "num_input_tokens_seen": 39399120,
      "step": 18250
    },
    {
      "epoch": 2.9779771615008155,
      "grad_norm": 0.419706791639328,
      "learning_rate": 4.427498508312026e-05,
      "loss": 0.0774,
      "num_input_tokens_seen": 39408560,
      "step": 18255
    },
    {
      "epoch": 2.9787928221859707,
      "grad_norm": 1.9322048425674438,
      "learning_rate": 4.4270451315516807e-05,
      "loss": 0.0891,
      "num_input_tokens_seen": 39419664,
      "step": 18260
    },
    {
      "epoch": 2.9796084828711256,
      "grad_norm": 2.4042837619781494,
      "learning_rate": 4.426591598575278e-05,
      "loss": 0.0169,
      "num_input_tokens_seen": 39430480,
      "step": 18265
    },
    {
      "epoch": 2.980424143556281,
      "grad_norm": 0.16649606823921204,
      "learning_rate": 4.4261379094195856e-05,
      "loss": 0.0231,
      "num_input_tokens_seen": 39441008,
      "step": 18270
    },
    {
      "epoch": 2.9812398042414356,
      "grad_norm": 2.1626293659210205,
      "learning_rate": 4.425684064121381e-05,
      "loss": 0.0349,
      "num_input_tokens_seen": 39451984,
      "step": 18275
    },
    {
      "epoch": 2.9820554649265905,
      "grad_norm": 7.148900032043457,
      "learning_rate": 4.425230062717455e-05,
      "loss": 0.2486,
      "num_input_tokens_seen": 39461904,
      "step": 18280
    },
    {
      "epoch": 2.9828711256117453,
      "grad_norm": 2.3899693489074707,
      "learning_rate": 4.424775905244612e-05,
      "loss": 0.0257,
      "num_input_tokens_seen": 39472656,
      "step": 18285
    },
    {
      "epoch": 2.9836867862969005,
      "grad_norm": 0.0685814619064331,
      "learning_rate": 4.424321591739668e-05,
      "loss": 0.2119,
      "num_input_tokens_seen": 39483376,
      "step": 18290
    },
    {
      "epoch": 2.9845024469820554,
      "grad_norm": 0.15265725553035736,
      "learning_rate": 4.423867122239451e-05,
      "loss": 0.057,
      "num_input_tokens_seen": 39496144,
      "step": 18295
    },
    {
      "epoch": 2.9853181076672106,
      "grad_norm": 1.2704753875732422,
      "learning_rate": 4.4234124967808044e-05,
      "loss": 0.167,
      "num_input_tokens_seen": 39506128,
      "step": 18300
    },
    {
      "epoch": 2.9861337683523654,
      "grad_norm": 3.1465489864349365,
      "learning_rate": 4.4229577154005814e-05,
      "loss": 0.489,
      "num_input_tokens_seen": 39515440,
      "step": 18305
    },
    {
      "epoch": 2.9869494290375203,
      "grad_norm": 0.0706682801246643,
      "learning_rate": 4.4225027781356484e-05,
      "loss": 0.0114,
      "num_input_tokens_seen": 39526512,
      "step": 18310
    },
    {
      "epoch": 2.9877650897226755,
      "grad_norm": 0.04816197231411934,
      "learning_rate": 4.4220476850228866e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 39537136,
      "step": 18315
    },
    {
      "epoch": 2.9885807504078303,
      "grad_norm": 1.9526958465576172,
      "learning_rate": 4.421592436099186e-05,
      "loss": 0.0855,
      "num_input_tokens_seen": 39545808,
      "step": 18320
    },
    {
      "epoch": 2.9893964110929856,
      "grad_norm": 1.2589412927627563,
      "learning_rate": 4.4211370314014534e-05,
      "loss": 0.0655,
      "num_input_tokens_seen": 39558000,
      "step": 18325
    },
    {
      "epoch": 2.9902120717781404,
      "grad_norm": 0.10761820524930954,
      "learning_rate": 4.4206814709666046e-05,
      "loss": 0.2526,
      "num_input_tokens_seen": 39570000,
      "step": 18330
    },
    {
      "epoch": 2.9910277324632952,
      "grad_norm": 0.17023952305316925,
      "learning_rate": 4.4202257548315704e-05,
      "loss": 0.1917,
      "num_input_tokens_seen": 39580560,
      "step": 18335
    },
    {
      "epoch": 2.99184339314845,
      "grad_norm": 0.2935185730457306,
      "learning_rate": 4.4197698830332934e-05,
      "loss": 0.0068,
      "num_input_tokens_seen": 39591408,
      "step": 18340
    },
    {
      "epoch": 2.9926590538336053,
      "grad_norm": 2.8027877807617188,
      "learning_rate": 4.419313855608729e-05,
      "loss": 0.0838,
      "num_input_tokens_seen": 39601872,
      "step": 18345
    },
    {
      "epoch": 2.99347471451876,
      "grad_norm": 3.0456812381744385,
      "learning_rate": 4.418857672594845e-05,
      "loss": 0.0514,
      "num_input_tokens_seen": 39613008,
      "step": 18350
    },
    {
      "epoch": 2.9942903752039154,
      "grad_norm": 3.173734188079834,
      "learning_rate": 4.4184013340286215e-05,
      "loss": 0.092,
      "num_input_tokens_seen": 39624272,
      "step": 18355
    },
    {
      "epoch": 2.99510603588907,
      "grad_norm": 0.14348101615905762,
      "learning_rate": 4.417944839947053e-05,
      "loss": 0.0765,
      "num_input_tokens_seen": 39635216,
      "step": 18360
    },
    {
      "epoch": 2.995921696574225,
      "grad_norm": 0.1864471733570099,
      "learning_rate": 4.417488190387144e-05,
      "loss": 0.0615,
      "num_input_tokens_seen": 39645808,
      "step": 18365
    },
    {
      "epoch": 2.99673735725938,
      "grad_norm": 0.3066174387931824,
      "learning_rate": 4.417031385385911e-05,
      "loss": 0.1555,
      "num_input_tokens_seen": 39656560,
      "step": 18370
    },
    {
      "epoch": 2.997553017944535,
      "grad_norm": 2.119971752166748,
      "learning_rate": 4.416574424980389e-05,
      "loss": 0.0398,
      "num_input_tokens_seen": 39666704,
      "step": 18375
    },
    {
      "epoch": 2.99836867862969,
      "grad_norm": 1.2401740550994873,
      "learning_rate": 4.4161173092076194e-05,
      "loss": 0.0196,
      "num_input_tokens_seen": 39677008,
      "step": 18380
    },
    {
      "epoch": 2.999184339314845,
      "grad_norm": 0.6576679944992065,
      "learning_rate": 4.415660038104658e-05,
      "loss": 0.0093,
      "num_input_tokens_seen": 39686416,
      "step": 18385
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.03525367006659508,
      "learning_rate": 4.4152026117085735e-05,
      "loss": 0.1007,
      "num_input_tokens_seen": 39694112,
      "step": 18390
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.16290850937366486,
      "eval_runtime": 133.0304,
      "eval_samples_per_second": 20.484,
      "eval_steps_per_second": 5.127,
      "num_input_tokens_seen": 39694112,
      "step": 18390
    },
    {
      "epoch": 3.000815660685155,
      "grad_norm": 0.047092389315366745,
      "learning_rate": 4.4147450300564485e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 39703936,
      "step": 18395
    },
    {
      "epoch": 3.00163132137031,
      "grad_norm": 0.05611247941851616,
      "learning_rate": 4.414287293185376e-05,
      "loss": 0.0276,
      "num_input_tokens_seen": 39715968,
      "step": 18400
    },
    {
      "epoch": 3.002446982055465,
      "grad_norm": 0.40346306562423706,
      "learning_rate": 4.413829401132462e-05,
      "loss": 0.0844,
      "num_input_tokens_seen": 39726912,
      "step": 18405
    },
    {
      "epoch": 3.0032626427406197,
      "grad_norm": 3.6461217403411865,
      "learning_rate": 4.4133713539348266e-05,
      "loss": 0.2274,
      "num_input_tokens_seen": 39737504,
      "step": 18410
    },
    {
      "epoch": 3.004078303425775,
      "grad_norm": 3.9422435760498047,
      "learning_rate": 4.4129131516296006e-05,
      "loss": 0.4117,
      "num_input_tokens_seen": 39749792,
      "step": 18415
    },
    {
      "epoch": 3.00489396411093,
      "grad_norm": 6.620187759399414,
      "learning_rate": 4.412454794253928e-05,
      "loss": 0.0258,
      "num_input_tokens_seen": 39760928,
      "step": 18420
    },
    {
      "epoch": 3.0057096247960846,
      "grad_norm": 1.4317384958267212,
      "learning_rate": 4.4119962818449665e-05,
      "loss": 0.0109,
      "num_input_tokens_seen": 39772096,
      "step": 18425
    },
    {
      "epoch": 3.00652528548124,
      "grad_norm": 0.21592706441879272,
      "learning_rate": 4.411537614439886e-05,
      "loss": 0.0328,
      "num_input_tokens_seen": 39782528,
      "step": 18430
    },
    {
      "epoch": 3.0073409461663947,
      "grad_norm": 0.06887643784284592,
      "learning_rate": 4.4110787920758663e-05,
      "loss": 0.0025,
      "num_input_tokens_seen": 39794336,
      "step": 18435
    },
    {
      "epoch": 3.00815660685155,
      "grad_norm": 2.4659671783447266,
      "learning_rate": 4.4106198147901035e-05,
      "loss": 0.2849,
      "num_input_tokens_seen": 39805376,
      "step": 18440
    },
    {
      "epoch": 3.0089722675367048,
      "grad_norm": 3.8042080402374268,
      "learning_rate": 4.410160682619803e-05,
      "loss": 0.1076,
      "num_input_tokens_seen": 39816768,
      "step": 18445
    },
    {
      "epoch": 3.0097879282218596,
      "grad_norm": 0.10827802866697311,
      "learning_rate": 4.409701395602187e-05,
      "loss": 0.1606,
      "num_input_tokens_seen": 39827520,
      "step": 18450
    },
    {
      "epoch": 3.010603588907015,
      "grad_norm": 0.26161307096481323,
      "learning_rate": 4.4092419537744854e-05,
      "loss": 0.1583,
      "num_input_tokens_seen": 39839072,
      "step": 18455
    },
    {
      "epoch": 3.0114192495921697,
      "grad_norm": 0.17363031208515167,
      "learning_rate": 4.408782357173944e-05,
      "loss": 0.0936,
      "num_input_tokens_seen": 39850432,
      "step": 18460
    },
    {
      "epoch": 3.0122349102773245,
      "grad_norm": 0.05275749787688255,
      "learning_rate": 4.408322605837819e-05,
      "loss": 0.0348,
      "num_input_tokens_seen": 39861600,
      "step": 18465
    },
    {
      "epoch": 3.0130505709624797,
      "grad_norm": 5.259739875793457,
      "learning_rate": 4.407862699803381e-05,
      "loss": 0.0829,
      "num_input_tokens_seen": 39872288,
      "step": 18470
    },
    {
      "epoch": 3.0138662316476346,
      "grad_norm": 0.046283259987831116,
      "learning_rate": 4.4074026391079126e-05,
      "loss": 0.3053,
      "num_input_tokens_seen": 39881984,
      "step": 18475
    },
    {
      "epoch": 3.0146818923327894,
      "grad_norm": 0.01923038437962532,
      "learning_rate": 4.406942423788708e-05,
      "loss": 0.0363,
      "num_input_tokens_seen": 39893184,
      "step": 18480
    },
    {
      "epoch": 3.0154975530179446,
      "grad_norm": 0.07903698086738586,
      "learning_rate": 4.406482053883075e-05,
      "loss": 0.0186,
      "num_input_tokens_seen": 39903200,
      "step": 18485
    },
    {
      "epoch": 3.0163132137030995,
      "grad_norm": 0.0979006364941597,
      "learning_rate": 4.406021529428334e-05,
      "loss": 0.1102,
      "num_input_tokens_seen": 39913728,
      "step": 18490
    },
    {
      "epoch": 3.0171288743882543,
      "grad_norm": 0.5228192806243896,
      "learning_rate": 4.405560850461815e-05,
      "loss": 0.0121,
      "num_input_tokens_seen": 39924576,
      "step": 18495
    },
    {
      "epoch": 3.0179445350734095,
      "grad_norm": 3.690387725830078,
      "learning_rate": 4.405100017020866e-05,
      "loss": 0.1169,
      "num_input_tokens_seen": 39935808,
      "step": 18500
    },
    {
      "epoch": 3.0187601957585644,
      "grad_norm": 0.2534869909286499,
      "learning_rate": 4.4046390291428426e-05,
      "loss": 0.1099,
      "num_input_tokens_seen": 39946592,
      "step": 18505
    },
    {
      "epoch": 3.0195758564437196,
      "grad_norm": 0.17684774100780487,
      "learning_rate": 4.4041778868651155e-05,
      "loss": 0.0376,
      "num_input_tokens_seen": 39958592,
      "step": 18510
    },
    {
      "epoch": 3.0203915171288744,
      "grad_norm": 3.616046190261841,
      "learning_rate": 4.4037165902250676e-05,
      "loss": 0.1999,
      "num_input_tokens_seen": 39969760,
      "step": 18515
    },
    {
      "epoch": 3.0212071778140293,
      "grad_norm": 0.5237236618995667,
      "learning_rate": 4.403255139260093e-05,
      "loss": 0.113,
      "num_input_tokens_seen": 39980064,
      "step": 18520
    },
    {
      "epoch": 3.0220228384991845,
      "grad_norm": 3.0811939239501953,
      "learning_rate": 4.4027935340076005e-05,
      "loss": 0.0931,
      "num_input_tokens_seen": 39991840,
      "step": 18525
    },
    {
      "epoch": 3.0228384991843393,
      "grad_norm": 0.05356327071785927,
      "learning_rate": 4.402331774505009e-05,
      "loss": 0.0746,
      "num_input_tokens_seen": 40002912,
      "step": 18530
    },
    {
      "epoch": 3.023654159869494,
      "grad_norm": 0.1326315999031067,
      "learning_rate": 4.4018698607897515e-05,
      "loss": 0.2377,
      "num_input_tokens_seen": 40014624,
      "step": 18535
    },
    {
      "epoch": 3.0244698205546494,
      "grad_norm": 0.7073240280151367,
      "learning_rate": 4.4014077928992736e-05,
      "loss": 0.3098,
      "num_input_tokens_seen": 40025408,
      "step": 18540
    },
    {
      "epoch": 3.0252854812398042,
      "grad_norm": 0.3344561755657196,
      "learning_rate": 4.4009455708710315e-05,
      "loss": 0.1189,
      "num_input_tokens_seen": 40036352,
      "step": 18545
    },
    {
      "epoch": 3.026101141924959,
      "grad_norm": 0.2792600989341736,
      "learning_rate": 4.4004831947424967e-05,
      "loss": 0.0541,
      "num_input_tokens_seen": 40045888,
      "step": 18550
    },
    {
      "epoch": 3.0269168026101143,
      "grad_norm": 0.08360278606414795,
      "learning_rate": 4.400020664551151e-05,
      "loss": 0.1498,
      "num_input_tokens_seen": 40056928,
      "step": 18555
    },
    {
      "epoch": 3.027732463295269,
      "grad_norm": 0.09548623114824295,
      "learning_rate": 4.39955798033449e-05,
      "loss": 0.069,
      "num_input_tokens_seen": 40067456,
      "step": 18560
    },
    {
      "epoch": 3.028548123980424,
      "grad_norm": 0.08647414296865463,
      "learning_rate": 4.399095142130021e-05,
      "loss": 0.1047,
      "num_input_tokens_seen": 40077760,
      "step": 18565
    },
    {
      "epoch": 3.029363784665579,
      "grad_norm": 0.8288561701774597,
      "learning_rate": 4.398632149975263e-05,
      "loss": 0.0825,
      "num_input_tokens_seen": 40089376,
      "step": 18570
    },
    {
      "epoch": 3.030179445350734,
      "grad_norm": 2.954484224319458,
      "learning_rate": 4.39816900390775e-05,
      "loss": 0.1254,
      "num_input_tokens_seen": 40100544,
      "step": 18575
    },
    {
      "epoch": 3.0309951060358893,
      "grad_norm": 0.09243351966142654,
      "learning_rate": 4.397705703965026e-05,
      "loss": 0.019,
      "num_input_tokens_seen": 40111616,
      "step": 18580
    },
    {
      "epoch": 3.031810766721044,
      "grad_norm": 0.9688690900802612,
      "learning_rate": 4.397242250184649e-05,
      "loss": 0.168,
      "num_input_tokens_seen": 40123712,
      "step": 18585
    },
    {
      "epoch": 3.032626427406199,
      "grad_norm": 0.630936861038208,
      "learning_rate": 4.396778642604188e-05,
      "loss": 0.0185,
      "num_input_tokens_seen": 40134976,
      "step": 18590
    },
    {
      "epoch": 3.033442088091354,
      "grad_norm": 5.368227958679199,
      "learning_rate": 4.396314881261227e-05,
      "loss": 0.1821,
      "num_input_tokens_seen": 40144992,
      "step": 18595
    },
    {
      "epoch": 3.034257748776509,
      "grad_norm": 0.03500809147953987,
      "learning_rate": 4.39585096619336e-05,
      "loss": 0.0139,
      "num_input_tokens_seen": 40156928,
      "step": 18600
    },
    {
      "epoch": 3.035073409461664,
      "grad_norm": 2.2299718856811523,
      "learning_rate": 4.395386897438194e-05,
      "loss": 0.07,
      "num_input_tokens_seen": 40167264,
      "step": 18605
    },
    {
      "epoch": 3.035889070146819,
      "grad_norm": 3.590679883956909,
      "learning_rate": 4.3949226750333484e-05,
      "loss": 0.2646,
      "num_input_tokens_seen": 40178944,
      "step": 18610
    },
    {
      "epoch": 3.036704730831974,
      "grad_norm": 2.9252543449401855,
      "learning_rate": 4.3944582990164565e-05,
      "loss": 0.1848,
      "num_input_tokens_seen": 40190240,
      "step": 18615
    },
    {
      "epoch": 3.0375203915171287,
      "grad_norm": 4.0557146072387695,
      "learning_rate": 4.393993769425162e-05,
      "loss": 0.102,
      "num_input_tokens_seen": 40201696,
      "step": 18620
    },
    {
      "epoch": 3.038336052202284,
      "grad_norm": 0.2512882649898529,
      "learning_rate": 4.393529086297123e-05,
      "loss": 0.1558,
      "num_input_tokens_seen": 40211424,
      "step": 18625
    },
    {
      "epoch": 3.039151712887439,
      "grad_norm": 4.620949745178223,
      "learning_rate": 4.3930642496700084e-05,
      "loss": 0.0893,
      "num_input_tokens_seen": 40221888,
      "step": 18630
    },
    {
      "epoch": 3.0399673735725936,
      "grad_norm": 0.0515848733484745,
      "learning_rate": 4.392599259581501e-05,
      "loss": 0.0301,
      "num_input_tokens_seen": 40233120,
      "step": 18635
    },
    {
      "epoch": 3.040783034257749,
      "grad_norm": 4.30033540725708,
      "learning_rate": 4.392134116069294e-05,
      "loss": 0.1255,
      "num_input_tokens_seen": 40243104,
      "step": 18640
    },
    {
      "epoch": 3.0415986949429037,
      "grad_norm": 0.07945439964532852,
      "learning_rate": 4.391668819171095e-05,
      "loss": 0.1488,
      "num_input_tokens_seen": 40253344,
      "step": 18645
    },
    {
      "epoch": 3.0424143556280585,
      "grad_norm": 0.4420635402202606,
      "learning_rate": 4.391203368924623e-05,
      "loss": 0.2902,
      "num_input_tokens_seen": 40265280,
      "step": 18650
    },
    {
      "epoch": 3.0432300163132138,
      "grad_norm": 0.8691559433937073,
      "learning_rate": 4.3907377653676104e-05,
      "loss": 0.1047,
      "num_input_tokens_seen": 40276224,
      "step": 18655
    },
    {
      "epoch": 3.0440456769983686,
      "grad_norm": 0.17548291385173798,
      "learning_rate": 4.390272008537802e-05,
      "loss": 0.2017,
      "num_input_tokens_seen": 40287168,
      "step": 18660
    },
    {
      "epoch": 3.044861337683524,
      "grad_norm": 1.90533447265625,
      "learning_rate": 4.3898060984729526e-05,
      "loss": 0.0163,
      "num_input_tokens_seen": 40297184,
      "step": 18665
    },
    {
      "epoch": 3.0456769983686787,
      "grad_norm": 6.821253299713135,
      "learning_rate": 4.389340035210832e-05,
      "loss": 0.053,
      "num_input_tokens_seen": 40308128,
      "step": 18670
    },
    {
      "epoch": 3.0464926590538335,
      "grad_norm": 3.088578939437866,
      "learning_rate": 4.388873818789222e-05,
      "loss": 0.0741,
      "num_input_tokens_seen": 40320352,
      "step": 18675
    },
    {
      "epoch": 3.0473083197389887,
      "grad_norm": 0.9741634130477905,
      "learning_rate": 4.388407449245916e-05,
      "loss": 0.1692,
      "num_input_tokens_seen": 40331424,
      "step": 18680
    },
    {
      "epoch": 3.0481239804241436,
      "grad_norm": 1.4948443174362183,
      "learning_rate": 4.38794092661872e-05,
      "loss": 0.1048,
      "num_input_tokens_seen": 40343264,
      "step": 18685
    },
    {
      "epoch": 3.0489396411092984,
      "grad_norm": 5.092470169067383,
      "learning_rate": 4.3874742509454536e-05,
      "loss": 0.0909,
      "num_input_tokens_seen": 40356608,
      "step": 18690
    },
    {
      "epoch": 3.0497553017944536,
      "grad_norm": 1.0479038953781128,
      "learning_rate": 4.387007422263948e-05,
      "loss": 0.1159,
      "num_input_tokens_seen": 40367456,
      "step": 18695
    },
    {
      "epoch": 3.0505709624796085,
      "grad_norm": 5.6116719245910645,
      "learning_rate": 4.3865404406120456e-05,
      "loss": 0.0724,
      "num_input_tokens_seen": 40377728,
      "step": 18700
    },
    {
      "epoch": 3.0513866231647633,
      "grad_norm": 0.2902747690677643,
      "learning_rate": 4.3860733060276025e-05,
      "loss": 0.0591,
      "num_input_tokens_seen": 40389184,
      "step": 18705
    },
    {
      "epoch": 3.0522022838499185,
      "grad_norm": 3.059255838394165,
      "learning_rate": 4.385606018548488e-05,
      "loss": 0.0407,
      "num_input_tokens_seen": 40400512,
      "step": 18710
    },
    {
      "epoch": 3.0530179445350734,
      "grad_norm": 0.31856194138526917,
      "learning_rate": 4.385138578212582e-05,
      "loss": 0.0224,
      "num_input_tokens_seen": 40411008,
      "step": 18715
    },
    {
      "epoch": 3.053833605220228,
      "grad_norm": 5.904905319213867,
      "learning_rate": 4.384670985057778e-05,
      "loss": 0.1246,
      "num_input_tokens_seen": 40422528,
      "step": 18720
    },
    {
      "epoch": 3.0546492659053834,
      "grad_norm": 0.024655459448695183,
      "learning_rate": 4.3842032391219804e-05,
      "loss": 0.1097,
      "num_input_tokens_seen": 40432384,
      "step": 18725
    },
    {
      "epoch": 3.0554649265905383,
      "grad_norm": 0.06969005614519119,
      "learning_rate": 4.3837353404431086e-05,
      "loss": 0.2286,
      "num_input_tokens_seen": 40443424,
      "step": 18730
    },
    {
      "epoch": 3.0562805872756935,
      "grad_norm": 0.04249146208167076,
      "learning_rate": 4.383267289059092e-05,
      "loss": 0.0787,
      "num_input_tokens_seen": 40453760,
      "step": 18735
    },
    {
      "epoch": 3.0570962479608483,
      "grad_norm": 4.027807712554932,
      "learning_rate": 4.382799085007873e-05,
      "loss": 0.1105,
      "num_input_tokens_seen": 40464832,
      "step": 18740
    },
    {
      "epoch": 3.057911908646003,
      "grad_norm": 14.182214736938477,
      "learning_rate": 4.382330728327407e-05,
      "loss": 0.3135,
      "num_input_tokens_seen": 40475136,
      "step": 18745
    },
    {
      "epoch": 3.0587275693311584,
      "grad_norm": 2.4968860149383545,
      "learning_rate": 4.3818622190556624e-05,
      "loss": 0.2739,
      "num_input_tokens_seen": 40485856,
      "step": 18750
    },
    {
      "epoch": 3.0595432300163132,
      "grad_norm": 0.09472081065177917,
      "learning_rate": 4.381393557230617e-05,
      "loss": 0.153,
      "num_input_tokens_seen": 40497120,
      "step": 18755
    },
    {
      "epoch": 3.060358890701468,
      "grad_norm": 0.643191933631897,
      "learning_rate": 4.380924742890264e-05,
      "loss": 0.0556,
      "num_input_tokens_seen": 40508928,
      "step": 18760
    },
    {
      "epoch": 3.0611745513866233,
      "grad_norm": 0.0940224900841713,
      "learning_rate": 4.380455776072607e-05,
      "loss": 0.1985,
      "num_input_tokens_seen": 40519904,
      "step": 18765
    },
    {
      "epoch": 3.061990212071778,
      "grad_norm": 3.93025803565979,
      "learning_rate": 4.3799866568156634e-05,
      "loss": 0.248,
      "num_input_tokens_seen": 40529568,
      "step": 18770
    },
    {
      "epoch": 3.062805872756933,
      "grad_norm": 4.1050639152526855,
      "learning_rate": 4.379517385157463e-05,
      "loss": 0.119,
      "num_input_tokens_seen": 40540512,
      "step": 18775
    },
    {
      "epoch": 3.063621533442088,
      "grad_norm": 0.17521882057189941,
      "learning_rate": 4.3790479611360466e-05,
      "loss": 0.1883,
      "num_input_tokens_seen": 40550592,
      "step": 18780
    },
    {
      "epoch": 3.064437194127243,
      "grad_norm": 1.3301870822906494,
      "learning_rate": 4.378578384789469e-05,
      "loss": 0.0607,
      "num_input_tokens_seen": 40561824,
      "step": 18785
    },
    {
      "epoch": 3.065252854812398,
      "grad_norm": 0.2315429151058197,
      "learning_rate": 4.378108656155795e-05,
      "loss": 0.0135,
      "num_input_tokens_seen": 40572416,
      "step": 18790
    },
    {
      "epoch": 3.066068515497553,
      "grad_norm": 2.1123085021972656,
      "learning_rate": 4.377638775273104e-05,
      "loss": 0.2255,
      "num_input_tokens_seen": 40583712,
      "step": 18795
    },
    {
      "epoch": 3.066884176182708,
      "grad_norm": 5.286981105804443,
      "learning_rate": 4.3771687421794866e-05,
      "loss": 0.1539,
      "num_input_tokens_seen": 40594560,
      "step": 18800
    },
    {
      "epoch": 3.067699836867863,
      "grad_norm": 0.7339895963668823,
      "learning_rate": 4.3766985569130465e-05,
      "loss": 0.0547,
      "num_input_tokens_seen": 40605536,
      "step": 18805
    },
    {
      "epoch": 3.068515497553018,
      "grad_norm": 0.6658563613891602,
      "learning_rate": 4.376228219511899e-05,
      "loss": 0.0417,
      "num_input_tokens_seen": 40615328,
      "step": 18810
    },
    {
      "epoch": 3.069331158238173,
      "grad_norm": 4.639886379241943,
      "learning_rate": 4.375757730014172e-05,
      "loss": 0.0324,
      "num_input_tokens_seen": 40625504,
      "step": 18815
    },
    {
      "epoch": 3.070146818923328,
      "grad_norm": 2.805645704269409,
      "learning_rate": 4.3752870884580065e-05,
      "loss": 0.1696,
      "num_input_tokens_seen": 40636896,
      "step": 18820
    },
    {
      "epoch": 3.070962479608483,
      "grad_norm": 5.8893256187438965,
      "learning_rate": 4.374816294881554e-05,
      "loss": 0.1972,
      "num_input_tokens_seen": 40647456,
      "step": 18825
    },
    {
      "epoch": 3.0717781402936377,
      "grad_norm": 3.609346628189087,
      "learning_rate": 4.37434534932298e-05,
      "loss": 0.1405,
      "num_input_tokens_seen": 40656832,
      "step": 18830
    },
    {
      "epoch": 3.072593800978793,
      "grad_norm": 6.025660037994385,
      "learning_rate": 4.373874251820462e-05,
      "loss": 0.0531,
      "num_input_tokens_seen": 40666752,
      "step": 18835
    },
    {
      "epoch": 3.073409461663948,
      "grad_norm": 4.759237289428711,
      "learning_rate": 4.3734030024121886e-05,
      "loss": 0.0388,
      "num_input_tokens_seen": 40678368,
      "step": 18840
    },
    {
      "epoch": 3.0742251223491026,
      "grad_norm": 4.019664764404297,
      "learning_rate": 4.372931601136363e-05,
      "loss": 0.268,
      "num_input_tokens_seen": 40689632,
      "step": 18845
    },
    {
      "epoch": 3.075040783034258,
      "grad_norm": 0.5742751359939575,
      "learning_rate": 4.372460048031198e-05,
      "loss": 0.023,
      "num_input_tokens_seen": 40700576,
      "step": 18850
    },
    {
      "epoch": 3.0758564437194127,
      "grad_norm": 2.622826099395752,
      "learning_rate": 4.3719883431349206e-05,
      "loss": 0.1285,
      "num_input_tokens_seen": 40711488,
      "step": 18855
    },
    {
      "epoch": 3.0766721044045675,
      "grad_norm": 0.3004291355609894,
      "learning_rate": 4.3715164864857705e-05,
      "loss": 0.1862,
      "num_input_tokens_seen": 40720832,
      "step": 18860
    },
    {
      "epoch": 3.0774877650897228,
      "grad_norm": 5.004072666168213,
      "learning_rate": 4.371044478121998e-05,
      "loss": 0.1355,
      "num_input_tokens_seen": 40731200,
      "step": 18865
    },
    {
      "epoch": 3.0783034257748776,
      "grad_norm": 2.9772660732269287,
      "learning_rate": 4.370572318081866e-05,
      "loss": 0.0598,
      "num_input_tokens_seen": 40742432,
      "step": 18870
    },
    {
      "epoch": 3.0791190864600324,
      "grad_norm": 0.14002737402915955,
      "learning_rate": 4.3701000064036504e-05,
      "loss": 0.2778,
      "num_input_tokens_seen": 40753856,
      "step": 18875
    },
    {
      "epoch": 3.0799347471451877,
      "grad_norm": 1.761139988899231,
      "learning_rate": 4.3696275431256405e-05,
      "loss": 0.0989,
      "num_input_tokens_seen": 40762976,
      "step": 18880
    },
    {
      "epoch": 3.0807504078303425,
      "grad_norm": 3.3624911308288574,
      "learning_rate": 4.369154928286134e-05,
      "loss": 0.1211,
      "num_input_tokens_seen": 40773504,
      "step": 18885
    },
    {
      "epoch": 3.0815660685154977,
      "grad_norm": 0.22525611519813538,
      "learning_rate": 4.368682161923447e-05,
      "loss": 0.0058,
      "num_input_tokens_seen": 40784576,
      "step": 18890
    },
    {
      "epoch": 3.0823817292006526,
      "grad_norm": 0.24241799116134644,
      "learning_rate": 4.368209244075901e-05,
      "loss": 0.0236,
      "num_input_tokens_seen": 40794240,
      "step": 18895
    },
    {
      "epoch": 3.0831973898858074,
      "grad_norm": 0.05112004280090332,
      "learning_rate": 4.3677361747818355e-05,
      "loss": 0.0441,
      "num_input_tokens_seen": 40805600,
      "step": 18900
    },
    {
      "epoch": 3.0840130505709626,
      "grad_norm": 0.057318106293678284,
      "learning_rate": 4.3672629540795976e-05,
      "loss": 0.0868,
      "num_input_tokens_seen": 40816352,
      "step": 18905
    },
    {
      "epoch": 3.0848287112561175,
      "grad_norm": 0.14517413079738617,
      "learning_rate": 4.366789582007551e-05,
      "loss": 0.0669,
      "num_input_tokens_seen": 40826720,
      "step": 18910
    },
    {
      "epoch": 3.0856443719412723,
      "grad_norm": 3.7802202701568604,
      "learning_rate": 4.366316058604069e-05,
      "loss": 0.2246,
      "num_input_tokens_seen": 40837216,
      "step": 18915
    },
    {
      "epoch": 3.0864600326264275,
      "grad_norm": 0.45749226212501526,
      "learning_rate": 4.3658423839075376e-05,
      "loss": 0.0127,
      "num_input_tokens_seen": 40847904,
      "step": 18920
    },
    {
      "epoch": 3.0872756933115824,
      "grad_norm": 0.07833077013492584,
      "learning_rate": 4.3653685579563555e-05,
      "loss": 0.1618,
      "num_input_tokens_seen": 40858304,
      "step": 18925
    },
    {
      "epoch": 3.088091353996737,
      "grad_norm": 0.14516383409500122,
      "learning_rate": 4.364894580788932e-05,
      "loss": 0.0151,
      "num_input_tokens_seen": 40869088,
      "step": 18930
    },
    {
      "epoch": 3.0889070146818924,
      "grad_norm": 0.24459004402160645,
      "learning_rate": 4.364420452443693e-05,
      "loss": 0.0282,
      "num_input_tokens_seen": 40878976,
      "step": 18935
    },
    {
      "epoch": 3.0897226753670473,
      "grad_norm": 6.220860958099365,
      "learning_rate": 4.363946172959071e-05,
      "loss": 0.1223,
      "num_input_tokens_seen": 40890144,
      "step": 18940
    },
    {
      "epoch": 3.090538336052202,
      "grad_norm": 9.879328727722168,
      "learning_rate": 4.363471742373516e-05,
      "loss": 0.132,
      "num_input_tokens_seen": 40900448,
      "step": 18945
    },
    {
      "epoch": 3.0913539967373573,
      "grad_norm": 0.04250815510749817,
      "learning_rate": 4.3629971607254855e-05,
      "loss": 0.0092,
      "num_input_tokens_seen": 40910880,
      "step": 18950
    },
    {
      "epoch": 3.092169657422512,
      "grad_norm": 0.06580474227666855,
      "learning_rate": 4.362522428053453e-05,
      "loss": 0.0124,
      "num_input_tokens_seen": 40922752,
      "step": 18955
    },
    {
      "epoch": 3.0929853181076674,
      "grad_norm": 5.778861045837402,
      "learning_rate": 4.3620475443959016e-05,
      "loss": 0.1461,
      "num_input_tokens_seen": 40932640,
      "step": 18960
    },
    {
      "epoch": 3.0938009787928222,
      "grad_norm": 1.215570330619812,
      "learning_rate": 4.361572509791328e-05,
      "loss": 0.1187,
      "num_input_tokens_seen": 40943264,
      "step": 18965
    },
    {
      "epoch": 3.094616639477977,
      "grad_norm": 0.7755405306816101,
      "learning_rate": 4.361097324278242e-05,
      "loss": 0.0735,
      "num_input_tokens_seen": 40953664,
      "step": 18970
    },
    {
      "epoch": 3.0954323001631323,
      "grad_norm": 6.451382637023926,
      "learning_rate": 4.3606219878951623e-05,
      "loss": 0.1979,
      "num_input_tokens_seen": 40964544,
      "step": 18975
    },
    {
      "epoch": 3.096247960848287,
      "grad_norm": 0.0326329804956913,
      "learning_rate": 4.360146500680625e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 40975520,
      "step": 18980
    },
    {
      "epoch": 3.097063621533442,
      "grad_norm": 4.4509735107421875,
      "learning_rate": 4.3596708626731744e-05,
      "loss": 0.0525,
      "num_input_tokens_seen": 40985728,
      "step": 18985
    },
    {
      "epoch": 3.097879282218597,
      "grad_norm": 0.12521952390670776,
      "learning_rate": 4.359195073911367e-05,
      "loss": 0.1533,
      "num_input_tokens_seen": 40996064,
      "step": 18990
    },
    {
      "epoch": 3.098694942903752,
      "grad_norm": 0.04597676545381546,
      "learning_rate": 4.3587191344337735e-05,
      "loss": 0.0982,
      "num_input_tokens_seen": 41007456,
      "step": 18995
    },
    {
      "epoch": 3.099510603588907,
      "grad_norm": 4.002562522888184,
      "learning_rate": 4.358243044278976e-05,
      "loss": 0.2087,
      "num_input_tokens_seen": 41017792,
      "step": 19000
    },
    {
      "epoch": 3.100326264274062,
      "grad_norm": 0.048084646463394165,
      "learning_rate": 4.3577668034855684e-05,
      "loss": 0.0743,
      "num_input_tokens_seen": 41028512,
      "step": 19005
    },
    {
      "epoch": 3.101141924959217,
      "grad_norm": 0.241954043507576,
      "learning_rate": 4.357290412092158e-05,
      "loss": 0.0317,
      "num_input_tokens_seen": 41039520,
      "step": 19010
    },
    {
      "epoch": 3.1019575856443717,
      "grad_norm": 0.12843763828277588,
      "learning_rate": 4.356813870137363e-05,
      "loss": 0.2005,
      "num_input_tokens_seen": 41050336,
      "step": 19015
    },
    {
      "epoch": 3.102773246329527,
      "grad_norm": 0.0742432102560997,
      "learning_rate": 4.356337177659814e-05,
      "loss": 0.0028,
      "num_input_tokens_seen": 41061088,
      "step": 19020
    },
    {
      "epoch": 3.103588907014682,
      "grad_norm": 0.04520115256309509,
      "learning_rate": 4.355860334698154e-05,
      "loss": 0.2471,
      "num_input_tokens_seen": 41071680,
      "step": 19025
    },
    {
      "epoch": 3.104404567699837,
      "grad_norm": 0.03389989584684372,
      "learning_rate": 4.3553833412910395e-05,
      "loss": 0.1464,
      "num_input_tokens_seen": 41083744,
      "step": 19030
    },
    {
      "epoch": 3.105220228384992,
      "grad_norm": 3.7635552883148193,
      "learning_rate": 4.354906197477137e-05,
      "loss": 0.1775,
      "num_input_tokens_seen": 41094688,
      "step": 19035
    },
    {
      "epoch": 3.1060358890701467,
      "grad_norm": 0.04354064539074898,
      "learning_rate": 4.354428903295126e-05,
      "loss": 0.2955,
      "num_input_tokens_seen": 41106656,
      "step": 19040
    },
    {
      "epoch": 3.106851549755302,
      "grad_norm": 0.1443745642900467,
      "learning_rate": 4.353951458783699e-05,
      "loss": 0.1355,
      "num_input_tokens_seen": 41117280,
      "step": 19045
    },
    {
      "epoch": 3.107667210440457,
      "grad_norm": 0.5298668146133423,
      "learning_rate": 4.3534738639815606e-05,
      "loss": 0.1538,
      "num_input_tokens_seen": 41128064,
      "step": 19050
    },
    {
      "epoch": 3.1084828711256116,
      "grad_norm": 3.103580951690674,
      "learning_rate": 4.352996118927426e-05,
      "loss": 0.1054,
      "num_input_tokens_seen": 41137440,
      "step": 19055
    },
    {
      "epoch": 3.109298531810767,
      "grad_norm": 1.0356005430221558,
      "learning_rate": 4.3525182236600235e-05,
      "loss": 0.0878,
      "num_input_tokens_seen": 41147648,
      "step": 19060
    },
    {
      "epoch": 3.1101141924959217,
      "grad_norm": 2.1156842708587646,
      "learning_rate": 4.3520401782180954e-05,
      "loss": 0.0645,
      "num_input_tokens_seen": 41158688,
      "step": 19065
    },
    {
      "epoch": 3.1109298531810765,
      "grad_norm": 0.21462346613407135,
      "learning_rate": 4.351561982640392e-05,
      "loss": 0.0059,
      "num_input_tokens_seen": 41168160,
      "step": 19070
    },
    {
      "epoch": 3.1117455138662318,
      "grad_norm": 6.509387493133545,
      "learning_rate": 4.35108363696568e-05,
      "loss": 0.0507,
      "num_input_tokens_seen": 41179936,
      "step": 19075
    },
    {
      "epoch": 3.1125611745513866,
      "grad_norm": 0.0804077610373497,
      "learning_rate": 4.3506051412327364e-05,
      "loss": 0.1488,
      "num_input_tokens_seen": 41190528,
      "step": 19080
    },
    {
      "epoch": 3.1133768352365414,
      "grad_norm": 0.2413562536239624,
      "learning_rate": 4.3501264954803495e-05,
      "loss": 0.0673,
      "num_input_tokens_seen": 41200128,
      "step": 19085
    },
    {
      "epoch": 3.1141924959216967,
      "grad_norm": 3.4854648113250732,
      "learning_rate": 4.3496476997473216e-05,
      "loss": 0.1053,
      "num_input_tokens_seen": 41211040,
      "step": 19090
    },
    {
      "epoch": 3.1150081566068515,
      "grad_norm": 1.6438874006271362,
      "learning_rate": 4.349168754072467e-05,
      "loss": 0.3506,
      "num_input_tokens_seen": 41223392,
      "step": 19095
    },
    {
      "epoch": 3.1158238172920063,
      "grad_norm": 3.393418312072754,
      "learning_rate": 4.3486896584946094e-05,
      "loss": 0.0614,
      "num_input_tokens_seen": 41233344,
      "step": 19100
    },
    {
      "epoch": 3.1166394779771616,
      "grad_norm": 3.376964807510376,
      "learning_rate": 4.348210413052589e-05,
      "loss": 0.3814,
      "num_input_tokens_seen": 41244512,
      "step": 19105
    },
    {
      "epoch": 3.1174551386623164,
      "grad_norm": 2.494851589202881,
      "learning_rate": 4.3477310177852537e-05,
      "loss": 0.168,
      "num_input_tokens_seen": 41255360,
      "step": 19110
    },
    {
      "epoch": 3.1182707993474716,
      "grad_norm": 4.940349102020264,
      "learning_rate": 4.347251472731467e-05,
      "loss": 0.0301,
      "num_input_tokens_seen": 41265856,
      "step": 19115
    },
    {
      "epoch": 3.1190864600326265,
      "grad_norm": 0.06120862811803818,
      "learning_rate": 4.3467717779301046e-05,
      "loss": 0.1862,
      "num_input_tokens_seen": 41277216,
      "step": 19120
    },
    {
      "epoch": 3.1199021207177813,
      "grad_norm": 9.058176040649414,
      "learning_rate": 4.3462919334200494e-05,
      "loss": 0.4465,
      "num_input_tokens_seen": 41288256,
      "step": 19125
    },
    {
      "epoch": 3.1207177814029365,
      "grad_norm": 0.17716369032859802,
      "learning_rate": 4.345811939240203e-05,
      "loss": 0.0919,
      "num_input_tokens_seen": 41298784,
      "step": 19130
    },
    {
      "epoch": 3.1215334420880914,
      "grad_norm": 0.2572365403175354,
      "learning_rate": 4.3453317954294755e-05,
      "loss": 0.1556,
      "num_input_tokens_seen": 41309792,
      "step": 19135
    },
    {
      "epoch": 3.122349102773246,
      "grad_norm": 2.621544122695923,
      "learning_rate": 4.3448515020267896e-05,
      "loss": 0.1619,
      "num_input_tokens_seen": 41320320,
      "step": 19140
    },
    {
      "epoch": 3.1231647634584014,
      "grad_norm": 2.794127941131592,
      "learning_rate": 4.3443710590710795e-05,
      "loss": 0.0703,
      "num_input_tokens_seen": 41331008,
      "step": 19145
    },
    {
      "epoch": 3.1239804241435563,
      "grad_norm": 0.18797270953655243,
      "learning_rate": 4.343890466601294e-05,
      "loss": 0.2203,
      "num_input_tokens_seen": 41341888,
      "step": 19150
    },
    {
      "epoch": 3.124796084828711,
      "grad_norm": 0.22668831050395966,
      "learning_rate": 4.343409724656391e-05,
      "loss": 0.1158,
      "num_input_tokens_seen": 41351840,
      "step": 19155
    },
    {
      "epoch": 3.1256117455138663,
      "grad_norm": 0.4405692219734192,
      "learning_rate": 4.342928833275341e-05,
      "loss": 0.0139,
      "num_input_tokens_seen": 41362592,
      "step": 19160
    },
    {
      "epoch": 3.126427406199021,
      "grad_norm": 0.5118138194084167,
      "learning_rate": 4.342447792497131e-05,
      "loss": 0.1122,
      "num_input_tokens_seen": 41372704,
      "step": 19165
    },
    {
      "epoch": 3.1272430668841764,
      "grad_norm": 0.46270260214805603,
      "learning_rate": 4.341966602360754e-05,
      "loss": 0.134,
      "num_input_tokens_seen": 41383584,
      "step": 19170
    },
    {
      "epoch": 3.1280587275693312,
      "grad_norm": 3.503925323486328,
      "learning_rate": 4.3414852629052175e-05,
      "loss": 0.351,
      "num_input_tokens_seen": 41394976,
      "step": 19175
    },
    {
      "epoch": 3.128874388254486,
      "grad_norm": 6.204897880554199,
      "learning_rate": 4.341003774169542e-05,
      "loss": 0.1614,
      "num_input_tokens_seen": 41406016,
      "step": 19180
    },
    {
      "epoch": 3.1296900489396413,
      "grad_norm": 0.08034936338663101,
      "learning_rate": 4.34052213619276e-05,
      "loss": 0.0253,
      "num_input_tokens_seen": 41417376,
      "step": 19185
    },
    {
      "epoch": 3.130505709624796,
      "grad_norm": 1.7257933616638184,
      "learning_rate": 4.340040349013915e-05,
      "loss": 0.2314,
      "num_input_tokens_seen": 41428416,
      "step": 19190
    },
    {
      "epoch": 3.131321370309951,
      "grad_norm": 6.101888656616211,
      "learning_rate": 4.3395584126720626e-05,
      "loss": 0.1961,
      "num_input_tokens_seen": 41439552,
      "step": 19195
    },
    {
      "epoch": 3.132137030995106,
      "grad_norm": 0.16559572517871857,
      "learning_rate": 4.339076327206272e-05,
      "loss": 0.0191,
      "num_input_tokens_seen": 41448928,
      "step": 19200
    },
    {
      "epoch": 3.132952691680261,
      "grad_norm": 0.6203173398971558,
      "learning_rate": 4.338594092655622e-05,
      "loss": 0.0534,
      "num_input_tokens_seen": 41460096,
      "step": 19205
    },
    {
      "epoch": 3.133768352365416,
      "grad_norm": 5.801191329956055,
      "learning_rate": 4.338111709059206e-05,
      "loss": 0.1536,
      "num_input_tokens_seen": 41469696,
      "step": 19210
    },
    {
      "epoch": 3.134584013050571,
      "grad_norm": 0.12250760942697525,
      "learning_rate": 4.337629176456129e-05,
      "loss": 0.0289,
      "num_input_tokens_seen": 41480224,
      "step": 19215
    },
    {
      "epoch": 3.135399673735726,
      "grad_norm": 0.2107073813676834,
      "learning_rate": 4.337146494885507e-05,
      "loss": 0.1377,
      "num_input_tokens_seen": 41490176,
      "step": 19220
    },
    {
      "epoch": 3.1362153344208807,
      "grad_norm": 3.8480005264282227,
      "learning_rate": 4.336663664386468e-05,
      "loss": 0.1146,
      "num_input_tokens_seen": 41500928,
      "step": 19225
    },
    {
      "epoch": 3.137030995106036,
      "grad_norm": 0.11215127259492874,
      "learning_rate": 4.3361806849981535e-05,
      "loss": 0.0617,
      "num_input_tokens_seen": 41512608,
      "step": 19230
    },
    {
      "epoch": 3.137846655791191,
      "grad_norm": 0.7500024437904358,
      "learning_rate": 4.335697556759716e-05,
      "loss": 0.0204,
      "num_input_tokens_seen": 41522880,
      "step": 19235
    },
    {
      "epoch": 3.1386623164763456,
      "grad_norm": 0.3657032549381256,
      "learning_rate": 4.3352142797103204e-05,
      "loss": 0.1502,
      "num_input_tokens_seen": 41533632,
      "step": 19240
    },
    {
      "epoch": 3.139477977161501,
      "grad_norm": 3.9524943828582764,
      "learning_rate": 4.334730853889143e-05,
      "loss": 0.0479,
      "num_input_tokens_seen": 41546112,
      "step": 19245
    },
    {
      "epoch": 3.1402936378466557,
      "grad_norm": 4.626284599304199,
      "learning_rate": 4.3342472793353736e-05,
      "loss": 0.0369,
      "num_input_tokens_seen": 41557376,
      "step": 19250
    },
    {
      "epoch": 3.141109298531811,
      "grad_norm": 1.2742774486541748,
      "learning_rate": 4.333763556088213e-05,
      "loss": 0.2307,
      "num_input_tokens_seen": 41568160,
      "step": 19255
    },
    {
      "epoch": 3.141924959216966,
      "grad_norm": 0.5053208470344543,
      "learning_rate": 4.333279684186874e-05,
      "loss": 0.162,
      "num_input_tokens_seen": 41577696,
      "step": 19260
    },
    {
      "epoch": 3.1427406199021206,
      "grad_norm": 0.1496593952178955,
      "learning_rate": 4.332795663670581e-05,
      "loss": 0.0919,
      "num_input_tokens_seen": 41589088,
      "step": 19265
    },
    {
      "epoch": 3.143556280587276,
      "grad_norm": 2.1042985916137695,
      "learning_rate": 4.332311494578573e-05,
      "loss": 0.096,
      "num_input_tokens_seen": 41599904,
      "step": 19270
    },
    {
      "epoch": 3.1443719412724307,
      "grad_norm": 0.07737638801336288,
      "learning_rate": 4.331827176950098e-05,
      "loss": 0.0157,
      "num_input_tokens_seen": 41610784,
      "step": 19275
    },
    {
      "epoch": 3.1451876019575855,
      "grad_norm": 0.11145178228616714,
      "learning_rate": 4.3313427108244175e-05,
      "loss": 0.1237,
      "num_input_tokens_seen": 41620768,
      "step": 19280
    },
    {
      "epoch": 3.1460032626427408,
      "grad_norm": 0.1373949497938156,
      "learning_rate": 4.330858096240804e-05,
      "loss": 0.0953,
      "num_input_tokens_seen": 41632384,
      "step": 19285
    },
    {
      "epoch": 3.1468189233278956,
      "grad_norm": 3.2667558193206787,
      "learning_rate": 4.3303733332385446e-05,
      "loss": 0.136,
      "num_input_tokens_seen": 41643680,
      "step": 19290
    },
    {
      "epoch": 3.1476345840130504,
      "grad_norm": 2.046661615371704,
      "learning_rate": 4.329888421856936e-05,
      "loss": 0.1643,
      "num_input_tokens_seen": 41655424,
      "step": 19295
    },
    {
      "epoch": 3.1484502446982057,
      "grad_norm": 0.5549253225326538,
      "learning_rate": 4.3294033621352856e-05,
      "loss": 0.0374,
      "num_input_tokens_seen": 41666144,
      "step": 19300
    },
    {
      "epoch": 3.1492659053833605,
      "grad_norm": 5.20822286605835,
      "learning_rate": 4.3289181541129174e-05,
      "loss": 0.0326,
      "num_input_tokens_seen": 41677312,
      "step": 19305
    },
    {
      "epoch": 3.1500815660685153,
      "grad_norm": 4.700889587402344,
      "learning_rate": 4.328432797829164e-05,
      "loss": 0.1031,
      "num_input_tokens_seen": 41689344,
      "step": 19310
    },
    {
      "epoch": 3.1508972267536706,
      "grad_norm": 0.08882571756839752,
      "learning_rate": 4.3279472933233696e-05,
      "loss": 0.0107,
      "num_input_tokens_seen": 41700032,
      "step": 19315
    },
    {
      "epoch": 3.1517128874388254,
      "grad_norm": 0.04364703595638275,
      "learning_rate": 4.327461640634893e-05,
      "loss": 0.0488,
      "num_input_tokens_seen": 41711040,
      "step": 19320
    },
    {
      "epoch": 3.15252854812398,
      "grad_norm": 0.1867649257183075,
      "learning_rate": 4.3269758398031037e-05,
      "loss": 0.1591,
      "num_input_tokens_seen": 41721184,
      "step": 19325
    },
    {
      "epoch": 3.1533442088091355,
      "grad_norm": 2.5056307315826416,
      "learning_rate": 4.3264898908673826e-05,
      "loss": 0.2196,
      "num_input_tokens_seen": 41731616,
      "step": 19330
    },
    {
      "epoch": 3.1541598694942903,
      "grad_norm": 0.2886751890182495,
      "learning_rate": 4.3260037938671237e-05,
      "loss": 0.1046,
      "num_input_tokens_seen": 41741568,
      "step": 19335
    },
    {
      "epoch": 3.1549755301794455,
      "grad_norm": 4.4228081703186035,
      "learning_rate": 4.325517548841732e-05,
      "loss": 0.0868,
      "num_input_tokens_seen": 41752512,
      "step": 19340
    },
    {
      "epoch": 3.1557911908646004,
      "grad_norm": 4.792253017425537,
      "learning_rate": 4.3250311558306255e-05,
      "loss": 0.1609,
      "num_input_tokens_seen": 41763040,
      "step": 19345
    },
    {
      "epoch": 3.156606851549755,
      "grad_norm": 0.13030284643173218,
      "learning_rate": 4.324544614873233e-05,
      "loss": 0.3224,
      "num_input_tokens_seen": 41773472,
      "step": 19350
    },
    {
      "epoch": 3.1574225122349104,
      "grad_norm": 4.019415855407715,
      "learning_rate": 4.324057926008997e-05,
      "loss": 0.0577,
      "num_input_tokens_seen": 41784096,
      "step": 19355
    },
    {
      "epoch": 3.1582381729200653,
      "grad_norm": 0.10057320445775986,
      "learning_rate": 4.323571089277369e-05,
      "loss": 0.0088,
      "num_input_tokens_seen": 41795616,
      "step": 19360
    },
    {
      "epoch": 3.15905383360522,
      "grad_norm": 0.12926925718784332,
      "learning_rate": 4.3230841047178175e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 41806784,
      "step": 19365
    },
    {
      "epoch": 3.1598694942903753,
      "grad_norm": 6.900053024291992,
      "learning_rate": 4.3225969723698165e-05,
      "loss": 0.1477,
      "num_input_tokens_seen": 41817920,
      "step": 19370
    },
    {
      "epoch": 3.16068515497553,
      "grad_norm": 2.3402750492095947,
      "learning_rate": 4.322109692272858e-05,
      "loss": 0.227,
      "num_input_tokens_seen": 41827520,
      "step": 19375
    },
    {
      "epoch": 3.161500815660685,
      "grad_norm": 0.36514484882354736,
      "learning_rate": 4.321622264466443e-05,
      "loss": 0.1202,
      "num_input_tokens_seen": 41838496,
      "step": 19380
    },
    {
      "epoch": 3.1623164763458402,
      "grad_norm": 2.110487699508667,
      "learning_rate": 4.321134688990084e-05,
      "loss": 0.1345,
      "num_input_tokens_seen": 41848800,
      "step": 19385
    },
    {
      "epoch": 3.163132137030995,
      "grad_norm": 0.2392733246088028,
      "learning_rate": 4.320646965883307e-05,
      "loss": 0.2938,
      "num_input_tokens_seen": 41859616,
      "step": 19390
    },
    {
      "epoch": 3.1639477977161503,
      "grad_norm": 0.6725291013717651,
      "learning_rate": 4.320159095185648e-05,
      "loss": 0.1144,
      "num_input_tokens_seen": 41870432,
      "step": 19395
    },
    {
      "epoch": 3.164763458401305,
      "grad_norm": 1.0118898153305054,
      "learning_rate": 4.3196710769366585e-05,
      "loss": 0.1588,
      "num_input_tokens_seen": 41880960,
      "step": 19400
    },
    {
      "epoch": 3.16557911908646,
      "grad_norm": 3.306779384613037,
      "learning_rate": 4.3191829111758985e-05,
      "loss": 0.3511,
      "num_input_tokens_seen": 41891456,
      "step": 19405
    },
    {
      "epoch": 3.166394779771615,
      "grad_norm": 2.330289602279663,
      "learning_rate": 4.318694597942941e-05,
      "loss": 0.0267,
      "num_input_tokens_seen": 41903040,
      "step": 19410
    },
    {
      "epoch": 3.16721044045677,
      "grad_norm": 3.969630718231201,
      "learning_rate": 4.318206137277372e-05,
      "loss": 0.148,
      "num_input_tokens_seen": 41914272,
      "step": 19415
    },
    {
      "epoch": 3.168026101141925,
      "grad_norm": 2.5971837043762207,
      "learning_rate": 4.317717529218788e-05,
      "loss": 0.2292,
      "num_input_tokens_seen": 41925184,
      "step": 19420
    },
    {
      "epoch": 3.16884176182708,
      "grad_norm": 3.305995225906372,
      "learning_rate": 4.317228773806797e-05,
      "loss": 0.2062,
      "num_input_tokens_seen": 41936288,
      "step": 19425
    },
    {
      "epoch": 3.169657422512235,
      "grad_norm": 2.1929593086242676,
      "learning_rate": 4.316739871081021e-05,
      "loss": 0.0966,
      "num_input_tokens_seen": 41945984,
      "step": 19430
    },
    {
      "epoch": 3.1704730831973897,
      "grad_norm": 1.9077894687652588,
      "learning_rate": 4.3162508210810936e-05,
      "loss": 0.3034,
      "num_input_tokens_seen": 41956256,
      "step": 19435
    },
    {
      "epoch": 3.171288743882545,
      "grad_norm": 1.9926167726516724,
      "learning_rate": 4.31576162384666e-05,
      "loss": 0.0752,
      "num_input_tokens_seen": 41967040,
      "step": 19440
    },
    {
      "epoch": 3.1721044045677,
      "grad_norm": 0.08837661892175674,
      "learning_rate": 4.315272279417375e-05,
      "loss": 0.0368,
      "num_input_tokens_seen": 41977664,
      "step": 19445
    },
    {
      "epoch": 3.1729200652528546,
      "grad_norm": 12.569263458251953,
      "learning_rate": 4.314782787832908e-05,
      "loss": 0.1588,
      "num_input_tokens_seen": 41988384,
      "step": 19450
    },
    {
      "epoch": 3.17373572593801,
      "grad_norm": 6.0814690589904785,
      "learning_rate": 4.314293149132941e-05,
      "loss": 0.3066,
      "num_input_tokens_seen": 42000576,
      "step": 19455
    },
    {
      "epoch": 3.1745513866231647,
      "grad_norm": 0.45871320366859436,
      "learning_rate": 4.313803363357166e-05,
      "loss": 0.0765,
      "num_input_tokens_seen": 42010176,
      "step": 19460
    },
    {
      "epoch": 3.1753670473083195,
      "grad_norm": 0.300508052110672,
      "learning_rate": 4.313313430545286e-05,
      "loss": 0.081,
      "num_input_tokens_seen": 42021248,
      "step": 19465
    },
    {
      "epoch": 3.176182707993475,
      "grad_norm": 0.1853676736354828,
      "learning_rate": 4.3128233507370196e-05,
      "loss": 0.0504,
      "num_input_tokens_seen": 42031712,
      "step": 19470
    },
    {
      "epoch": 3.1769983686786296,
      "grad_norm": 0.3095267117023468,
      "learning_rate": 4.312333123972094e-05,
      "loss": 0.1787,
      "num_input_tokens_seen": 42042848,
      "step": 19475
    },
    {
      "epoch": 3.177814029363785,
      "grad_norm": 2.2687737941741943,
      "learning_rate": 4.31184275029025e-05,
      "loss": 0.1077,
      "num_input_tokens_seen": 42053856,
      "step": 19480
    },
    {
      "epoch": 3.1786296900489397,
      "grad_norm": 2.702927589416504,
      "learning_rate": 4.311352229731239e-05,
      "loss": 0.1128,
      "num_input_tokens_seen": 42064864,
      "step": 19485
    },
    {
      "epoch": 3.1794453507340945,
      "grad_norm": 2.2455344200134277,
      "learning_rate": 4.310861562334826e-05,
      "loss": 0.1222,
      "num_input_tokens_seen": 42076864,
      "step": 19490
    },
    {
      "epoch": 3.1802610114192498,
      "grad_norm": 0.5983144044876099,
      "learning_rate": 4.310370748140786e-05,
      "loss": 0.0858,
      "num_input_tokens_seen": 42086592,
      "step": 19495
    },
    {
      "epoch": 3.1810766721044046,
      "grad_norm": 0.12773123383522034,
      "learning_rate": 4.3098797871889075e-05,
      "loss": 0.1902,
      "num_input_tokens_seen": 42096736,
      "step": 19500
    },
    {
      "epoch": 3.1818923327895594,
      "grad_norm": 0.14167964458465576,
      "learning_rate": 4.30938867951899e-05,
      "loss": 0.1249,
      "num_input_tokens_seen": 42106592,
      "step": 19505
    },
    {
      "epoch": 3.1827079934747147,
      "grad_norm": 0.3585329055786133,
      "learning_rate": 4.308897425170846e-05,
      "loss": 0.0828,
      "num_input_tokens_seen": 42117024,
      "step": 19510
    },
    {
      "epoch": 3.1835236541598695,
      "grad_norm": 3.088606834411621,
      "learning_rate": 4.3084060241842984e-05,
      "loss": 0.0701,
      "num_input_tokens_seen": 42127776,
      "step": 19515
    },
    {
      "epoch": 3.1843393148450243,
      "grad_norm": 0.22292488813400269,
      "learning_rate": 4.307914476599182e-05,
      "loss": 0.0281,
      "num_input_tokens_seen": 42138976,
      "step": 19520
    },
    {
      "epoch": 3.1851549755301796,
      "grad_norm": 2.7585763931274414,
      "learning_rate": 4.307422782455346e-05,
      "loss": 0.0254,
      "num_input_tokens_seen": 42150752,
      "step": 19525
    },
    {
      "epoch": 3.1859706362153344,
      "grad_norm": 0.4058837592601776,
      "learning_rate": 4.306930941792648e-05,
      "loss": 0.0201,
      "num_input_tokens_seen": 42161696,
      "step": 19530
    },
    {
      "epoch": 3.186786296900489,
      "grad_norm": 3.5705759525299072,
      "learning_rate": 4.3064389546509585e-05,
      "loss": 0.1497,
      "num_input_tokens_seen": 42173312,
      "step": 19535
    },
    {
      "epoch": 3.1876019575856445,
      "grad_norm": 0.3486447334289551,
      "learning_rate": 4.305946821070163e-05,
      "loss": 0.2562,
      "num_input_tokens_seen": 42185344,
      "step": 19540
    },
    {
      "epoch": 3.1884176182707993,
      "grad_norm": 0.2385052740573883,
      "learning_rate": 4.3054545410901547e-05,
      "loss": 0.0955,
      "num_input_tokens_seen": 42196480,
      "step": 19545
    },
    {
      "epoch": 3.189233278955954,
      "grad_norm": 0.35330891609191895,
      "learning_rate": 4.30496211475084e-05,
      "loss": 0.0127,
      "num_input_tokens_seen": 42206624,
      "step": 19550
    },
    {
      "epoch": 3.1900489396411094,
      "grad_norm": 3.7042102813720703,
      "learning_rate": 4.3044695420921386e-05,
      "loss": 0.1845,
      "num_input_tokens_seen": 42218688,
      "step": 19555
    },
    {
      "epoch": 3.190864600326264,
      "grad_norm": 0.06464269757270813,
      "learning_rate": 4.30397682315398e-05,
      "loss": 0.0134,
      "num_input_tokens_seen": 42228896,
      "step": 19560
    },
    {
      "epoch": 3.1916802610114194,
      "grad_norm": 0.25477996468544006,
      "learning_rate": 4.303483957976306e-05,
      "loss": 0.2881,
      "num_input_tokens_seen": 42239424,
      "step": 19565
    },
    {
      "epoch": 3.1924959216965743,
      "grad_norm": 0.243189737200737,
      "learning_rate": 4.302990946599073e-05,
      "loss": 0.2144,
      "num_input_tokens_seen": 42250272,
      "step": 19570
    },
    {
      "epoch": 3.193311582381729,
      "grad_norm": 3.361558437347412,
      "learning_rate": 4.302497789062245e-05,
      "loss": 0.0704,
      "num_input_tokens_seen": 42261024,
      "step": 19575
    },
    {
      "epoch": 3.1941272430668843,
      "grad_norm": 0.07526399940252304,
      "learning_rate": 4.3020044854058e-05,
      "loss": 0.0393,
      "num_input_tokens_seen": 42271680,
      "step": 19580
    },
    {
      "epoch": 3.194942903752039,
      "grad_norm": 0.07639128714799881,
      "learning_rate": 4.3015110356697285e-05,
      "loss": 0.0525,
      "num_input_tokens_seen": 42281792,
      "step": 19585
    },
    {
      "epoch": 3.195758564437194,
      "grad_norm": 0.09000908583402634,
      "learning_rate": 4.301017439894032e-05,
      "loss": 0.0096,
      "num_input_tokens_seen": 42292128,
      "step": 19590
    },
    {
      "epoch": 3.1965742251223492,
      "grad_norm": 0.6561002135276794,
      "learning_rate": 4.300523698118722e-05,
      "loss": 0.0995,
      "num_input_tokens_seen": 42303232,
      "step": 19595
    },
    {
      "epoch": 3.197389885807504,
      "grad_norm": 0.0983612909913063,
      "learning_rate": 4.3000298103838274e-05,
      "loss": 0.0148,
      "num_input_tokens_seen": 42313056,
      "step": 19600
    },
    {
      "epoch": 3.198205546492659,
      "grad_norm": 0.11739430576562881,
      "learning_rate": 4.299535776729382e-05,
      "loss": 0.0558,
      "num_input_tokens_seen": 42324320,
      "step": 19605
    },
    {
      "epoch": 3.199021207177814,
      "grad_norm": 0.13181328773498535,
      "learning_rate": 4.2990415971954364e-05,
      "loss": 0.0366,
      "num_input_tokens_seen": 42332800,
      "step": 19610
    },
    {
      "epoch": 3.199836867862969,
      "grad_norm": 3.944260358810425,
      "learning_rate": 4.29854727182205e-05,
      "loss": 0.379,
      "num_input_tokens_seen": 42344448,
      "step": 19615
    },
    {
      "epoch": 3.200652528548124,
      "grad_norm": 0.08940883725881577,
      "learning_rate": 4.298052800649296e-05,
      "loss": 0.0591,
      "num_input_tokens_seen": 42355520,
      "step": 19620
    },
    {
      "epoch": 3.201468189233279,
      "grad_norm": 1.6371586322784424,
      "learning_rate": 4.297558183717259e-05,
      "loss": 0.0875,
      "num_input_tokens_seen": 42366656,
      "step": 19625
    },
    {
      "epoch": 3.202283849918434,
      "grad_norm": 0.17973874509334564,
      "learning_rate": 4.297063421066035e-05,
      "loss": 0.0371,
      "num_input_tokens_seen": 42377824,
      "step": 19630
    },
    {
      "epoch": 3.203099510603589,
      "grad_norm": 7.309290409088135,
      "learning_rate": 4.296568512735732e-05,
      "loss": 0.0664,
      "num_input_tokens_seen": 42388064,
      "step": 19635
    },
    {
      "epoch": 3.203915171288744,
      "grad_norm": 0.04178430140018463,
      "learning_rate": 4.29607345876647e-05,
      "loss": 0.1628,
      "num_input_tokens_seen": 42399648,
      "step": 19640
    },
    {
      "epoch": 3.2047308319738987,
      "grad_norm": 0.022029142826795578,
      "learning_rate": 4.2955782591983795e-05,
      "loss": 0.0573,
      "num_input_tokens_seen": 42409024,
      "step": 19645
    },
    {
      "epoch": 3.205546492659054,
      "grad_norm": 0.0897393524646759,
      "learning_rate": 4.295082914071604e-05,
      "loss": 0.0048,
      "num_input_tokens_seen": 42419552,
      "step": 19650
    },
    {
      "epoch": 3.206362153344209,
      "grad_norm": 4.748601913452148,
      "learning_rate": 4.294587423426301e-05,
      "loss": 0.2774,
      "num_input_tokens_seen": 42428992,
      "step": 19655
    },
    {
      "epoch": 3.2071778140293636,
      "grad_norm": 0.10691147297620773,
      "learning_rate": 4.294091787302634e-05,
      "loss": 0.0772,
      "num_input_tokens_seen": 42440736,
      "step": 19660
    },
    {
      "epoch": 3.207993474714519,
      "grad_norm": 0.3394257724285126,
      "learning_rate": 4.2935960057407855e-05,
      "loss": 0.0965,
      "num_input_tokens_seen": 42451168,
      "step": 19665
    },
    {
      "epoch": 3.2088091353996737,
      "grad_norm": 5.757011890411377,
      "learning_rate": 4.2931000787809426e-05,
      "loss": 0.0841,
      "num_input_tokens_seen": 42462528,
      "step": 19670
    },
    {
      "epoch": 3.2096247960848285,
      "grad_norm": 0.08598591387271881,
      "learning_rate": 4.29260400646331e-05,
      "loss": 0.0393,
      "num_input_tokens_seen": 42471648,
      "step": 19675
    },
    {
      "epoch": 3.210440456769984,
      "grad_norm": 1.1945375204086304,
      "learning_rate": 4.2921077888281014e-05,
      "loss": 0.0381,
      "num_input_tokens_seen": 42482944,
      "step": 19680
    },
    {
      "epoch": 3.2112561174551386,
      "grad_norm": 2.91871976852417,
      "learning_rate": 4.2916114259155414e-05,
      "loss": 0.1889,
      "num_input_tokens_seen": 42493152,
      "step": 19685
    },
    {
      "epoch": 3.2120717781402934,
      "grad_norm": 0.3258408010005951,
      "learning_rate": 4.291114917765869e-05,
      "loss": 0.0159,
      "num_input_tokens_seen": 42504544,
      "step": 19690
    },
    {
      "epoch": 3.2128874388254487,
      "grad_norm": 0.4206432104110718,
      "learning_rate": 4.290618264419334e-05,
      "loss": 0.1514,
      "num_input_tokens_seen": 42514496,
      "step": 19695
    },
    {
      "epoch": 3.2137030995106035,
      "grad_norm": 0.558193027973175,
      "learning_rate": 4.290121465916196e-05,
      "loss": 0.0871,
      "num_input_tokens_seen": 42524704,
      "step": 19700
    },
    {
      "epoch": 3.2145187601957588,
      "grad_norm": 6.601524829864502,
      "learning_rate": 4.2896245222967296e-05,
      "loss": 0.2024,
      "num_input_tokens_seen": 42534688,
      "step": 19705
    },
    {
      "epoch": 3.2153344208809136,
      "grad_norm": 2.0279533863067627,
      "learning_rate": 4.2891274336012186e-05,
      "loss": 0.2494,
      "num_input_tokens_seen": 42546016,
      "step": 19710
    },
    {
      "epoch": 3.2161500815660684,
      "grad_norm": 6.865815162658691,
      "learning_rate": 4.288630199869961e-05,
      "loss": 0.1239,
      "num_input_tokens_seen": 42556672,
      "step": 19715
    },
    {
      "epoch": 3.2169657422512237,
      "grad_norm": 0.076471708714962,
      "learning_rate": 4.2881328211432626e-05,
      "loss": 0.0609,
      "num_input_tokens_seen": 42567168,
      "step": 19720
    },
    {
      "epoch": 3.2177814029363785,
      "grad_norm": 0.09073363244533539,
      "learning_rate": 4.2876352974614456e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 42577760,
      "step": 19725
    },
    {
      "epoch": 3.2185970636215333,
      "grad_norm": 5.533669948577881,
      "learning_rate": 4.28713762886484e-05,
      "loss": 0.1846,
      "num_input_tokens_seen": 42588672,
      "step": 19730
    },
    {
      "epoch": 3.2194127243066886,
      "grad_norm": 1.4149954319000244,
      "learning_rate": 4.286639815393791e-05,
      "loss": 0.0151,
      "num_input_tokens_seen": 42600480,
      "step": 19735
    },
    {
      "epoch": 3.2202283849918434,
      "grad_norm": 0.03279214724898338,
      "learning_rate": 4.286141857088654e-05,
      "loss": 0.1365,
      "num_input_tokens_seen": 42612992,
      "step": 19740
    },
    {
      "epoch": 3.221044045676998,
      "grad_norm": 0.4979197680950165,
      "learning_rate": 4.285643753989794e-05,
      "loss": 0.0752,
      "num_input_tokens_seen": 42624576,
      "step": 19745
    },
    {
      "epoch": 3.2218597063621535,
      "grad_norm": 0.1591596007347107,
      "learning_rate": 4.2851455061375924e-05,
      "loss": 0.1735,
      "num_input_tokens_seen": 42636128,
      "step": 19750
    },
    {
      "epoch": 3.2226753670473083,
      "grad_norm": 0.08087214827537537,
      "learning_rate": 4.2846471135724376e-05,
      "loss": 0.1514,
      "num_input_tokens_seen": 42646304,
      "step": 19755
    },
    {
      "epoch": 3.223491027732463,
      "grad_norm": 1.0540143251419067,
      "learning_rate": 4.2841485763347324e-05,
      "loss": 0.2253,
      "num_input_tokens_seen": 42656576,
      "step": 19760
    },
    {
      "epoch": 3.2243066884176184,
      "grad_norm": 0.29019859433174133,
      "learning_rate": 4.2836498944648904e-05,
      "loss": 0.0052,
      "num_input_tokens_seen": 42666624,
      "step": 19765
    },
    {
      "epoch": 3.225122349102773,
      "grad_norm": 3.46596360206604,
      "learning_rate": 4.2831510680033394e-05,
      "loss": 0.1699,
      "num_input_tokens_seen": 42676096,
      "step": 19770
    },
    {
      "epoch": 3.225938009787928,
      "grad_norm": 0.8415746092796326,
      "learning_rate": 4.2826520969905134e-05,
      "loss": 0.1026,
      "num_input_tokens_seen": 42686752,
      "step": 19775
    },
    {
      "epoch": 3.2267536704730833,
      "grad_norm": 2.7905514240264893,
      "learning_rate": 4.282152981466865e-05,
      "loss": 0.1408,
      "num_input_tokens_seen": 42697792,
      "step": 19780
    },
    {
      "epoch": 3.227569331158238,
      "grad_norm": 0.28766247630119324,
      "learning_rate": 4.2816537214728524e-05,
      "loss": 0.141,
      "num_input_tokens_seen": 42708384,
      "step": 19785
    },
    {
      "epoch": 3.2283849918433933,
      "grad_norm": 0.3859138488769531,
      "learning_rate": 4.281154317048949e-05,
      "loss": 0.116,
      "num_input_tokens_seen": 42719136,
      "step": 19790
    },
    {
      "epoch": 3.229200652528548,
      "grad_norm": 0.06281393766403198,
      "learning_rate": 4.28065476823564e-05,
      "loss": 0.075,
      "num_input_tokens_seen": 42728896,
      "step": 19795
    },
    {
      "epoch": 3.230016313213703,
      "grad_norm": 2.613255739212036,
      "learning_rate": 4.2801550750734195e-05,
      "loss": 0.189,
      "num_input_tokens_seen": 42740512,
      "step": 19800
    },
    {
      "epoch": 3.2308319738988582,
      "grad_norm": 0.07633854448795319,
      "learning_rate": 4.279655237602796e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 42750240,
      "step": 19805
    },
    {
      "epoch": 3.231647634584013,
      "grad_norm": 0.10590975731611252,
      "learning_rate": 4.279155255864291e-05,
      "loss": 0.0057,
      "num_input_tokens_seen": 42760672,
      "step": 19810
    },
    {
      "epoch": 3.232463295269168,
      "grad_norm": 3.3129844665527344,
      "learning_rate": 4.2786551298984315e-05,
      "loss": 0.1996,
      "num_input_tokens_seen": 42771680,
      "step": 19815
    },
    {
      "epoch": 3.233278955954323,
      "grad_norm": 0.37032970786094666,
      "learning_rate": 4.278154859745763e-05,
      "loss": 0.1322,
      "num_input_tokens_seen": 42782048,
      "step": 19820
    },
    {
      "epoch": 3.234094616639478,
      "grad_norm": 3.0718038082122803,
      "learning_rate": 4.27765444544684e-05,
      "loss": 0.0622,
      "num_input_tokens_seen": 42792864,
      "step": 19825
    },
    {
      "epoch": 3.2349102773246328,
      "grad_norm": 0.06151893734931946,
      "learning_rate": 4.277153887042227e-05,
      "loss": 0.167,
      "num_input_tokens_seen": 42803904,
      "step": 19830
    },
    {
      "epoch": 3.235725938009788,
      "grad_norm": 1.7453343868255615,
      "learning_rate": 4.2766531845725036e-05,
      "loss": 0.0721,
      "num_input_tokens_seen": 42814720,
      "step": 19835
    },
    {
      "epoch": 3.236541598694943,
      "grad_norm": 0.5372622013092041,
      "learning_rate": 4.276152338078258e-05,
      "loss": 0.0171,
      "num_input_tokens_seen": 42825856,
      "step": 19840
    },
    {
      "epoch": 3.237357259380098,
      "grad_norm": 2.056267261505127,
      "learning_rate": 4.275651347600092e-05,
      "loss": 0.2859,
      "num_input_tokens_seen": 42837408,
      "step": 19845
    },
    {
      "epoch": 3.238172920065253,
      "grad_norm": 3.7286434173583984,
      "learning_rate": 4.275150213178618e-05,
      "loss": 0.0628,
      "num_input_tokens_seen": 42848960,
      "step": 19850
    },
    {
      "epoch": 3.2389885807504077,
      "grad_norm": 0.07230158150196075,
      "learning_rate": 4.27464893485446e-05,
      "loss": 0.1969,
      "num_input_tokens_seen": 42859168,
      "step": 19855
    },
    {
      "epoch": 3.239804241435563,
      "grad_norm": 0.23849955201148987,
      "learning_rate": 4.274147512668256e-05,
      "loss": 0.052,
      "num_input_tokens_seen": 42867872,
      "step": 19860
    },
    {
      "epoch": 3.240619902120718,
      "grad_norm": 3.9088425636291504,
      "learning_rate": 4.273645946660652e-05,
      "loss": 0.2549,
      "num_input_tokens_seen": 42879264,
      "step": 19865
    },
    {
      "epoch": 3.2414355628058726,
      "grad_norm": 0.10310837626457214,
      "learning_rate": 4.273144236872308e-05,
      "loss": 0.0131,
      "num_input_tokens_seen": 42890816,
      "step": 19870
    },
    {
      "epoch": 3.242251223491028,
      "grad_norm": 4.3532609939575195,
      "learning_rate": 4.2726423833438964e-05,
      "loss": 0.2167,
      "num_input_tokens_seen": 42902400,
      "step": 19875
    },
    {
      "epoch": 3.2430668841761827,
      "grad_norm": 4.053173065185547,
      "learning_rate": 4.272140386116098e-05,
      "loss": 0.4124,
      "num_input_tokens_seen": 42913152,
      "step": 19880
    },
    {
      "epoch": 3.2438825448613375,
      "grad_norm": 9.261698722839355,
      "learning_rate": 4.2716382452296086e-05,
      "loss": 0.1838,
      "num_input_tokens_seen": 42924672,
      "step": 19885
    },
    {
      "epoch": 3.244698205546493,
      "grad_norm": 0.16540081799030304,
      "learning_rate": 4.271135960725133e-05,
      "loss": 0.2583,
      "num_input_tokens_seen": 42934880,
      "step": 19890
    },
    {
      "epoch": 3.2455138662316476,
      "grad_norm": 0.07863563299179077,
      "learning_rate": 4.270633532643391e-05,
      "loss": 0.0355,
      "num_input_tokens_seen": 42946176,
      "step": 19895
    },
    {
      "epoch": 3.2463295269168024,
      "grad_norm": 0.2188352346420288,
      "learning_rate": 4.27013096102511e-05,
      "loss": 0.1408,
      "num_input_tokens_seen": 42958016,
      "step": 19900
    },
    {
      "epoch": 3.2471451876019577,
      "grad_norm": 0.30662140250205994,
      "learning_rate": 4.269628245911031e-05,
      "loss": 0.0544,
      "num_input_tokens_seen": 42968320,
      "step": 19905
    },
    {
      "epoch": 3.2479608482871125,
      "grad_norm": 3.4740748405456543,
      "learning_rate": 4.269125387341909e-05,
      "loss": 0.1067,
      "num_input_tokens_seen": 42980096,
      "step": 19910
    },
    {
      "epoch": 3.2487765089722673,
      "grad_norm": 2.7133607864379883,
      "learning_rate": 4.268622385358506e-05,
      "loss": 0.2278,
      "num_input_tokens_seen": 42990304,
      "step": 19915
    },
    {
      "epoch": 3.2495921696574226,
      "grad_norm": 0.09134632349014282,
      "learning_rate": 4.268119240001598e-05,
      "loss": 0.282,
      "num_input_tokens_seen": 43001280,
      "step": 19920
    },
    {
      "epoch": 3.2504078303425774,
      "grad_norm": 0.4034450352191925,
      "learning_rate": 4.267615951311974e-05,
      "loss": 0.0167,
      "num_input_tokens_seen": 43011584,
      "step": 19925
    },
    {
      "epoch": 3.2512234910277327,
      "grad_norm": 0.06232699751853943,
      "learning_rate": 4.267112519330432e-05,
      "loss": 0.1074,
      "num_input_tokens_seen": 43022432,
      "step": 19930
    },
    {
      "epoch": 3.2520391517128875,
      "grad_norm": 0.5810019373893738,
      "learning_rate": 4.266608944097782e-05,
      "loss": 0.1979,
      "num_input_tokens_seen": 43033792,
      "step": 19935
    },
    {
      "epoch": 3.2528548123980423,
      "grad_norm": 0.8372753858566284,
      "learning_rate": 4.266105225654848e-05,
      "loss": 0.2448,
      "num_input_tokens_seen": 43045504,
      "step": 19940
    },
    {
      "epoch": 3.2536704730831976,
      "grad_norm": 0.0908043310046196,
      "learning_rate": 4.265601364042463e-05,
      "loss": 0.1678,
      "num_input_tokens_seen": 43056544,
      "step": 19945
    },
    {
      "epoch": 3.2544861337683524,
      "grad_norm": 7.50385856628418,
      "learning_rate": 4.2650973593014734e-05,
      "loss": 0.1924,
      "num_input_tokens_seen": 43065184,
      "step": 19950
    },
    {
      "epoch": 3.255301794453507,
      "grad_norm": 0.24904431402683258,
      "learning_rate": 4.264593211472735e-05,
      "loss": 0.0456,
      "num_input_tokens_seen": 43074784,
      "step": 19955
    },
    {
      "epoch": 3.2561174551386625,
      "grad_norm": 3.333866596221924,
      "learning_rate": 4.264088920597118e-05,
      "loss": 0.0339,
      "num_input_tokens_seen": 43084864,
      "step": 19960
    },
    {
      "epoch": 3.2569331158238173,
      "grad_norm": 0.13684949278831482,
      "learning_rate": 4.263584486715503e-05,
      "loss": 0.2277,
      "num_input_tokens_seen": 43096576,
      "step": 19965
    },
    {
      "epoch": 3.257748776508972,
      "grad_norm": 4.079904556274414,
      "learning_rate": 4.2630799098687804e-05,
      "loss": 0.1449,
      "num_input_tokens_seen": 43106816,
      "step": 19970
    },
    {
      "epoch": 3.2585644371941274,
      "grad_norm": 5.152735233306885,
      "learning_rate": 4.262575190097854e-05,
      "loss": 0.0791,
      "num_input_tokens_seen": 43118304,
      "step": 19975
    },
    {
      "epoch": 3.259380097879282,
      "grad_norm": 0.04147205501794815,
      "learning_rate": 4.262070327443639e-05,
      "loss": 0.1603,
      "num_input_tokens_seen": 43128384,
      "step": 19980
    },
    {
      "epoch": 3.2601957585644374,
      "grad_norm": 4.397024631500244,
      "learning_rate": 4.261565321947064e-05,
      "loss": 0.234,
      "num_input_tokens_seen": 43139136,
      "step": 19985
    },
    {
      "epoch": 3.2610114192495923,
      "grad_norm": 0.466985821723938,
      "learning_rate": 4.261060173649065e-05,
      "loss": 0.0627,
      "num_input_tokens_seen": 43150048,
      "step": 19990
    },
    {
      "epoch": 3.261827079934747,
      "grad_norm": 4.005570888519287,
      "learning_rate": 4.260554882590594e-05,
      "loss": 0.1105,
      "num_input_tokens_seen": 43159456,
      "step": 19995
    },
    {
      "epoch": 3.262642740619902,
      "grad_norm": 6.014162063598633,
      "learning_rate": 4.2600494488126104e-05,
      "loss": 0.2537,
      "num_input_tokens_seen": 43170816,
      "step": 20000
    },
    {
      "epoch": 3.263458401305057,
      "grad_norm": 3.622558832168579,
      "learning_rate": 4.259543872356088e-05,
      "loss": 0.2441,
      "num_input_tokens_seen": 43182336,
      "step": 20005
    },
    {
      "epoch": 3.264274061990212,
      "grad_norm": 2.2375385761260986,
      "learning_rate": 4.259038153262012e-05,
      "loss": 0.2732,
      "num_input_tokens_seen": 43192384,
      "step": 20010
    },
    {
      "epoch": 3.2650897226753672,
      "grad_norm": 0.07627036422491074,
      "learning_rate": 4.2585322915713774e-05,
      "loss": 0.1285,
      "num_input_tokens_seen": 43203744,
      "step": 20015
    },
    {
      "epoch": 3.265905383360522,
      "grad_norm": 2.762985944747925,
      "learning_rate": 4.258026287325192e-05,
      "loss": 0.0786,
      "num_input_tokens_seen": 43213376,
      "step": 20020
    },
    {
      "epoch": 3.266721044045677,
      "grad_norm": 2.847794771194458,
      "learning_rate": 4.2575201405644764e-05,
      "loss": 0.2222,
      "num_input_tokens_seen": 43223744,
      "step": 20025
    },
    {
      "epoch": 3.267536704730832,
      "grad_norm": 1.276902675628662,
      "learning_rate": 4.257013851330261e-05,
      "loss": 0.0735,
      "num_input_tokens_seen": 43234752,
      "step": 20030
    },
    {
      "epoch": 3.268352365415987,
      "grad_norm": 2.2544126510620117,
      "learning_rate": 4.256507419663587e-05,
      "loss": 0.22,
      "num_input_tokens_seen": 43245280,
      "step": 20035
    },
    {
      "epoch": 3.2691680261011418,
      "grad_norm": 4.820090293884277,
      "learning_rate": 4.25600084560551e-05,
      "loss": 0.0622,
      "num_input_tokens_seen": 43256672,
      "step": 20040
    },
    {
      "epoch": 3.269983686786297,
      "grad_norm": 2.6052327156066895,
      "learning_rate": 4.255494129197094e-05,
      "loss": 0.2593,
      "num_input_tokens_seen": 43268288,
      "step": 20045
    },
    {
      "epoch": 3.270799347471452,
      "grad_norm": 0.15062707662582397,
      "learning_rate": 4.254987270479417e-05,
      "loss": 0.0118,
      "num_input_tokens_seen": 43278080,
      "step": 20050
    },
    {
      "epoch": 3.2716150081566067,
      "grad_norm": 0.06929773837327957,
      "learning_rate": 4.254480269493567e-05,
      "loss": 0.016,
      "num_input_tokens_seen": 43287392,
      "step": 20055
    },
    {
      "epoch": 3.272430668841762,
      "grad_norm": 0.9511746168136597,
      "learning_rate": 4.253973126280644e-05,
      "loss": 0.0457,
      "num_input_tokens_seen": 43297824,
      "step": 20060
    },
    {
      "epoch": 3.2732463295269167,
      "grad_norm": 3.381207227706909,
      "learning_rate": 4.2534658408817595e-05,
      "loss": 0.0696,
      "num_input_tokens_seen": 43308864,
      "step": 20065
    },
    {
      "epoch": 3.274061990212072,
      "grad_norm": 3.5614192485809326,
      "learning_rate": 4.252958413338038e-05,
      "loss": 0.1761,
      "num_input_tokens_seen": 43318912,
      "step": 20070
    },
    {
      "epoch": 3.274877650897227,
      "grad_norm": 0.08948476612567902,
      "learning_rate": 4.2524508436906124e-05,
      "loss": 0.0985,
      "num_input_tokens_seen": 43328896,
      "step": 20075
    },
    {
      "epoch": 3.2756933115823816,
      "grad_norm": 11.680952072143555,
      "learning_rate": 4.251943131980629e-05,
      "loss": 0.0872,
      "num_input_tokens_seen": 43340512,
      "step": 20080
    },
    {
      "epoch": 3.2765089722675365,
      "grad_norm": 2.4284684658050537,
      "learning_rate": 4.2514352782492475e-05,
      "loss": 0.122,
      "num_input_tokens_seen": 43351840,
      "step": 20085
    },
    {
      "epoch": 3.2773246329526917,
      "grad_norm": 4.431033611297607,
      "learning_rate": 4.250927282537635e-05,
      "loss": 0.3525,
      "num_input_tokens_seen": 43363328,
      "step": 20090
    },
    {
      "epoch": 3.2781402936378465,
      "grad_norm": 0.826481819152832,
      "learning_rate": 4.2504191448869716e-05,
      "loss": 0.1639,
      "num_input_tokens_seen": 43374688,
      "step": 20095
    },
    {
      "epoch": 3.278955954323002,
      "grad_norm": 0.4012637734413147,
      "learning_rate": 4.249910865338452e-05,
      "loss": 0.1135,
      "num_input_tokens_seen": 43384832,
      "step": 20100
    },
    {
      "epoch": 3.2797716150081566,
      "grad_norm": 0.3164308965206146,
      "learning_rate": 4.249402443933279e-05,
      "loss": 0.252,
      "num_input_tokens_seen": 43396352,
      "step": 20105
    },
    {
      "epoch": 3.2805872756933114,
      "grad_norm": 6.287999629974365,
      "learning_rate": 4.248893880712667e-05,
      "loss": 0.0945,
      "num_input_tokens_seen": 43407104,
      "step": 20110
    },
    {
      "epoch": 3.2814029363784667,
      "grad_norm": 0.17996253073215485,
      "learning_rate": 4.248385175717843e-05,
      "loss": 0.0275,
      "num_input_tokens_seen": 43418816,
      "step": 20115
    },
    {
      "epoch": 3.2822185970636215,
      "grad_norm": 0.10354874283075333,
      "learning_rate": 4.247876328990046e-05,
      "loss": 0.1144,
      "num_input_tokens_seen": 43429088,
      "step": 20120
    },
    {
      "epoch": 3.2830342577487763,
      "grad_norm": 3.017449140548706,
      "learning_rate": 4.247367340570525e-05,
      "loss": 0.0328,
      "num_input_tokens_seen": 43439744,
      "step": 20125
    },
    {
      "epoch": 3.2838499184339316,
      "grad_norm": 1.2335954904556274,
      "learning_rate": 4.2468582105005413e-05,
      "loss": 0.0128,
      "num_input_tokens_seen": 43451264,
      "step": 20130
    },
    {
      "epoch": 3.2846655791190864,
      "grad_norm": 0.0889868438243866,
      "learning_rate": 4.246348938821367e-05,
      "loss": 0.1561,
      "num_input_tokens_seen": 43461696,
      "step": 20135
    },
    {
      "epoch": 3.2854812398042412,
      "grad_norm": 0.8793284296989441,
      "learning_rate": 4.2458395255742875e-05,
      "loss": 0.0136,
      "num_input_tokens_seen": 43471904,
      "step": 20140
    },
    {
      "epoch": 3.2862969004893965,
      "grad_norm": 0.21098622679710388,
      "learning_rate": 4.245329970800597e-05,
      "loss": 0.2366,
      "num_input_tokens_seen": 43482400,
      "step": 20145
    },
    {
      "epoch": 3.2871125611745513,
      "grad_norm": 0.20438610017299652,
      "learning_rate": 4.244820274541604e-05,
      "loss": 0.184,
      "num_input_tokens_seen": 43492160,
      "step": 20150
    },
    {
      "epoch": 3.2879282218597066,
      "grad_norm": 0.6875913143157959,
      "learning_rate": 4.244310436838627e-05,
      "loss": 0.1147,
      "num_input_tokens_seen": 43502688,
      "step": 20155
    },
    {
      "epoch": 3.2887438825448614,
      "grad_norm": 2.2987444400787354,
      "learning_rate": 4.2438004577329946e-05,
      "loss": 0.2091,
      "num_input_tokens_seen": 43513184,
      "step": 20160
    },
    {
      "epoch": 3.289559543230016,
      "grad_norm": 3.6858136653900146,
      "learning_rate": 4.243290337266049e-05,
      "loss": 0.1854,
      "num_input_tokens_seen": 43522592,
      "step": 20165
    },
    {
      "epoch": 3.2903752039151715,
      "grad_norm": 3.0720269680023193,
      "learning_rate": 4.242780075479143e-05,
      "loss": 0.1067,
      "num_input_tokens_seen": 43533312,
      "step": 20170
    },
    {
      "epoch": 3.2911908646003263,
      "grad_norm": 3.2114734649658203,
      "learning_rate": 4.242269672413643e-05,
      "loss": 0.2091,
      "num_input_tokens_seen": 43543808,
      "step": 20175
    },
    {
      "epoch": 3.292006525285481,
      "grad_norm": 2.9163646697998047,
      "learning_rate": 4.241759128110922e-05,
      "loss": 0.3067,
      "num_input_tokens_seen": 43554528,
      "step": 20180
    },
    {
      "epoch": 3.2928221859706364,
      "grad_norm": 0.2551548480987549,
      "learning_rate": 4.241248442612368e-05,
      "loss": 0.0563,
      "num_input_tokens_seen": 43564960,
      "step": 20185
    },
    {
      "epoch": 3.293637846655791,
      "grad_norm": 0.07951763272285461,
      "learning_rate": 4.240737615959381e-05,
      "loss": 0.0559,
      "num_input_tokens_seen": 43576576,
      "step": 20190
    },
    {
      "epoch": 3.294453507340946,
      "grad_norm": 5.029916763305664,
      "learning_rate": 4.2402266481933706e-05,
      "loss": 0.1453,
      "num_input_tokens_seen": 43588576,
      "step": 20195
    },
    {
      "epoch": 3.2952691680261013,
      "grad_norm": 0.6786830425262451,
      "learning_rate": 4.2397155393557574e-05,
      "loss": 0.3138,
      "num_input_tokens_seen": 43598912,
      "step": 20200
    },
    {
      "epoch": 3.296084828711256,
      "grad_norm": 1.1113815307617188,
      "learning_rate": 4.239204289487976e-05,
      "loss": 0.0831,
      "num_input_tokens_seen": 43609568,
      "step": 20205
    },
    {
      "epoch": 3.2969004893964113,
      "grad_norm": 5.254055023193359,
      "learning_rate": 4.23869289863147e-05,
      "loss": 0.0951,
      "num_input_tokens_seen": 43619904,
      "step": 20210
    },
    {
      "epoch": 3.297716150081566,
      "grad_norm": 0.3169059455394745,
      "learning_rate": 4.238181366827696e-05,
      "loss": 0.13,
      "num_input_tokens_seen": 43630976,
      "step": 20215
    },
    {
      "epoch": 3.298531810766721,
      "grad_norm": 0.26758649945259094,
      "learning_rate": 4.237669694118121e-05,
      "loss": 0.0168,
      "num_input_tokens_seen": 43640384,
      "step": 20220
    },
    {
      "epoch": 3.299347471451876,
      "grad_norm": 1.9709473848342896,
      "learning_rate": 4.237157880544223e-05,
      "loss": 0.1413,
      "num_input_tokens_seen": 43650912,
      "step": 20225
    },
    {
      "epoch": 3.300163132137031,
      "grad_norm": 0.19510877132415771,
      "learning_rate": 4.2366459261474933e-05,
      "loss": 0.1766,
      "num_input_tokens_seen": 43662368,
      "step": 20230
    },
    {
      "epoch": 3.300978792822186,
      "grad_norm": 5.569932460784912,
      "learning_rate": 4.2361338309694335e-05,
      "loss": 0.1336,
      "num_input_tokens_seen": 43674048,
      "step": 20235
    },
    {
      "epoch": 3.301794453507341,
      "grad_norm": 0.298435240983963,
      "learning_rate": 4.235621595051556e-05,
      "loss": 0.1006,
      "num_input_tokens_seen": 43685120,
      "step": 20240
    },
    {
      "epoch": 3.302610114192496,
      "grad_norm": 0.10472102463245392,
      "learning_rate": 4.2351092184353855e-05,
      "loss": 0.0847,
      "num_input_tokens_seen": 43696288,
      "step": 20245
    },
    {
      "epoch": 3.3034257748776508,
      "grad_norm": 0.14596039056777954,
      "learning_rate": 4.234596701162458e-05,
      "loss": 0.129,
      "num_input_tokens_seen": 43708000,
      "step": 20250
    },
    {
      "epoch": 3.304241435562806,
      "grad_norm": 0.0793973058462143,
      "learning_rate": 4.2340840432743206e-05,
      "loss": 0.0357,
      "num_input_tokens_seen": 43720416,
      "step": 20255
    },
    {
      "epoch": 3.305057096247961,
      "grad_norm": 0.12220805883407593,
      "learning_rate": 4.2335712448125316e-05,
      "loss": 0.1799,
      "num_input_tokens_seen": 43730080,
      "step": 20260
    },
    {
      "epoch": 3.3058727569331157,
      "grad_norm": 3.793534994125366,
      "learning_rate": 4.233058305818662e-05,
      "loss": 0.0631,
      "num_input_tokens_seen": 43740896,
      "step": 20265
    },
    {
      "epoch": 3.306688417618271,
      "grad_norm": 7.794310569763184,
      "learning_rate": 4.232545226334293e-05,
      "loss": 0.1448,
      "num_input_tokens_seen": 43752128,
      "step": 20270
    },
    {
      "epoch": 3.3075040783034257,
      "grad_norm": 0.04614422470331192,
      "learning_rate": 4.232032006401017e-05,
      "loss": 0.0284,
      "num_input_tokens_seen": 43763008,
      "step": 20275
    },
    {
      "epoch": 3.3083197389885806,
      "grad_norm": 3.9614317417144775,
      "learning_rate": 4.231518646060438e-05,
      "loss": 0.1683,
      "num_input_tokens_seen": 43774144,
      "step": 20280
    },
    {
      "epoch": 3.309135399673736,
      "grad_norm": 3.1272754669189453,
      "learning_rate": 4.231005145354172e-05,
      "loss": 0.0643,
      "num_input_tokens_seen": 43785280,
      "step": 20285
    },
    {
      "epoch": 3.3099510603588906,
      "grad_norm": 0.057867251336574554,
      "learning_rate": 4.230491504323846e-05,
      "loss": 0.0143,
      "num_input_tokens_seen": 43796288,
      "step": 20290
    },
    {
      "epoch": 3.310766721044046,
      "grad_norm": 0.23088066279888153,
      "learning_rate": 4.229977723011097e-05,
      "loss": 0.0154,
      "num_input_tokens_seen": 43807392,
      "step": 20295
    },
    {
      "epoch": 3.3115823817292007,
      "grad_norm": 4.954625606536865,
      "learning_rate": 4.2294638014575774e-05,
      "loss": 0.2845,
      "num_input_tokens_seen": 43819328,
      "step": 20300
    },
    {
      "epoch": 3.3123980424143555,
      "grad_norm": 0.16805298626422882,
      "learning_rate": 4.228949739704946e-05,
      "loss": 0.0137,
      "num_input_tokens_seen": 43829632,
      "step": 20305
    },
    {
      "epoch": 3.3132137030995104,
      "grad_norm": 1.6958798170089722,
      "learning_rate": 4.228435537794877e-05,
      "loss": 0.0947,
      "num_input_tokens_seen": 43839520,
      "step": 20310
    },
    {
      "epoch": 3.3140293637846656,
      "grad_norm": 6.2627387046813965,
      "learning_rate": 4.227921195769053e-05,
      "loss": 0.041,
      "num_input_tokens_seen": 43850048,
      "step": 20315
    },
    {
      "epoch": 3.3148450244698204,
      "grad_norm": 0.14634250104427338,
      "learning_rate": 4.227406713669169e-05,
      "loss": 0.0083,
      "num_input_tokens_seen": 43860480,
      "step": 20320
    },
    {
      "epoch": 3.3156606851549757,
      "grad_norm": 0.07750158756971359,
      "learning_rate": 4.226892091536933e-05,
      "loss": 0.044,
      "num_input_tokens_seen": 43871456,
      "step": 20325
    },
    {
      "epoch": 3.3164763458401305,
      "grad_norm": 1.7951619625091553,
      "learning_rate": 4.226377329414061e-05,
      "loss": 0.1785,
      "num_input_tokens_seen": 43882208,
      "step": 20330
    },
    {
      "epoch": 3.3172920065252853,
      "grad_norm": 0.02641744166612625,
      "learning_rate": 4.225862427342283e-05,
      "loss": 0.0144,
      "num_input_tokens_seen": 43893568,
      "step": 20335
    },
    {
      "epoch": 3.3181076672104406,
      "grad_norm": 3.071866273880005,
      "learning_rate": 4.2253473853633405e-05,
      "loss": 0.2477,
      "num_input_tokens_seen": 43904736,
      "step": 20340
    },
    {
      "epoch": 3.3189233278955954,
      "grad_norm": 1.6703068017959595,
      "learning_rate": 4.2248322035189835e-05,
      "loss": 0.0634,
      "num_input_tokens_seen": 43915232,
      "step": 20345
    },
    {
      "epoch": 3.3197389885807502,
      "grad_norm": 0.08380986750125885,
      "learning_rate": 4.224316881850977e-05,
      "loss": 0.3356,
      "num_input_tokens_seen": 43925664,
      "step": 20350
    },
    {
      "epoch": 3.3205546492659055,
      "grad_norm": 2.1386399269104004,
      "learning_rate": 4.223801420401095e-05,
      "loss": 0.0725,
      "num_input_tokens_seen": 43936736,
      "step": 20355
    },
    {
      "epoch": 3.3213703099510603,
      "grad_norm": 3.1797494888305664,
      "learning_rate": 4.223285819211124e-05,
      "loss": 0.2569,
      "num_input_tokens_seen": 43948320,
      "step": 20360
    },
    {
      "epoch": 3.322185970636215,
      "grad_norm": 0.11439325660467148,
      "learning_rate": 4.2227700783228594e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 43958304,
      "step": 20365
    },
    {
      "epoch": 3.3230016313213704,
      "grad_norm": 0.0685453861951828,
      "learning_rate": 4.222254197778112e-05,
      "loss": 0.1345,
      "num_input_tokens_seen": 43969120,
      "step": 20370
    },
    {
      "epoch": 3.323817292006525,
      "grad_norm": 0.10162924230098724,
      "learning_rate": 4.2217381776187005e-05,
      "loss": 0.127,
      "num_input_tokens_seen": 43980768,
      "step": 20375
    },
    {
      "epoch": 3.3246329526916805,
      "grad_norm": 4.4270124435424805,
      "learning_rate": 4.2212220178864556e-05,
      "loss": 0.0686,
      "num_input_tokens_seen": 43991360,
      "step": 20380
    },
    {
      "epoch": 3.3254486133768353,
      "grad_norm": 0.6833563446998596,
      "learning_rate": 4.2207057186232215e-05,
      "loss": 0.0277,
      "num_input_tokens_seen": 44002464,
      "step": 20385
    },
    {
      "epoch": 3.32626427406199,
      "grad_norm": 0.2188410758972168,
      "learning_rate": 4.220189279870851e-05,
      "loss": 0.0136,
      "num_input_tokens_seen": 44011264,
      "step": 20390
    },
    {
      "epoch": 3.3270799347471454,
      "grad_norm": 0.09716194868087769,
      "learning_rate": 4.219672701671209e-05,
      "loss": 0.0373,
      "num_input_tokens_seen": 44021632,
      "step": 20395
    },
    {
      "epoch": 3.3278955954323,
      "grad_norm": 0.04455023258924484,
      "learning_rate": 4.219155984066171e-05,
      "loss": 0.1428,
      "num_input_tokens_seen": 44031872,
      "step": 20400
    },
    {
      "epoch": 3.328711256117455,
      "grad_norm": 0.22936658561229706,
      "learning_rate": 4.218639127097628e-05,
      "loss": 0.185,
      "num_input_tokens_seen": 44044064,
      "step": 20405
    },
    {
      "epoch": 3.3295269168026103,
      "grad_norm": 12.915611267089844,
      "learning_rate": 4.218122130807476e-05,
      "loss": 0.3073,
      "num_input_tokens_seen": 44055488,
      "step": 20410
    },
    {
      "epoch": 3.330342577487765,
      "grad_norm": 0.139049232006073,
      "learning_rate": 4.2176049952376265e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 44066496,
      "step": 20415
    },
    {
      "epoch": 3.33115823817292,
      "grad_norm": 0.07639367878437042,
      "learning_rate": 4.217087720430002e-05,
      "loss": 0.141,
      "num_input_tokens_seen": 44077664,
      "step": 20420
    },
    {
      "epoch": 3.331973898858075,
      "grad_norm": 0.4085349440574646,
      "learning_rate": 4.2165703064265335e-05,
      "loss": 0.1546,
      "num_input_tokens_seen": 44089344,
      "step": 20425
    },
    {
      "epoch": 3.33278955954323,
      "grad_norm": 0.16980627179145813,
      "learning_rate": 4.216052753269166e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 44100288,
      "step": 20430
    },
    {
      "epoch": 3.3336052202283852,
      "grad_norm": 0.954876184463501,
      "learning_rate": 4.215535060999856e-05,
      "loss": 0.0113,
      "num_input_tokens_seen": 44111040,
      "step": 20435
    },
    {
      "epoch": 3.33442088091354,
      "grad_norm": 0.2842998504638672,
      "learning_rate": 4.215017229660569e-05,
      "loss": 0.0206,
      "num_input_tokens_seen": 44122144,
      "step": 20440
    },
    {
      "epoch": 3.335236541598695,
      "grad_norm": 1.7308735847473145,
      "learning_rate": 4.214499259293283e-05,
      "loss": 0.1076,
      "num_input_tokens_seen": 44133344,
      "step": 20445
    },
    {
      "epoch": 3.3360522022838497,
      "grad_norm": 5.43765926361084,
      "learning_rate": 4.213981149939988e-05,
      "loss": 0.043,
      "num_input_tokens_seen": 44143520,
      "step": 20450
    },
    {
      "epoch": 3.336867862969005,
      "grad_norm": 0.058233749121427536,
      "learning_rate": 4.213462901642685e-05,
      "loss": 0.0145,
      "num_input_tokens_seen": 44154496,
      "step": 20455
    },
    {
      "epoch": 3.3376835236541598,
      "grad_norm": 0.04777275770902634,
      "learning_rate": 4.212944514443384e-05,
      "loss": 0.006,
      "num_input_tokens_seen": 44165504,
      "step": 20460
    },
    {
      "epoch": 3.338499184339315,
      "grad_norm": 1.7751857042312622,
      "learning_rate": 4.21242598838411e-05,
      "loss": 0.1912,
      "num_input_tokens_seen": 44177472,
      "step": 20465
    },
    {
      "epoch": 3.33931484502447,
      "grad_norm": 0.137112095952034,
      "learning_rate": 4.211907323506897e-05,
      "loss": 0.1756,
      "num_input_tokens_seen": 44187712,
      "step": 20470
    },
    {
      "epoch": 3.3401305057096247,
      "grad_norm": 0.9803479909896851,
      "learning_rate": 4.21138851985379e-05,
      "loss": 0.2745,
      "num_input_tokens_seen": 44198816,
      "step": 20475
    },
    {
      "epoch": 3.34094616639478,
      "grad_norm": 5.399055480957031,
      "learning_rate": 4.210869577466846e-05,
      "loss": 0.2211,
      "num_input_tokens_seen": 44209600,
      "step": 20480
    },
    {
      "epoch": 3.3417618270799347,
      "grad_norm": 5.411726474761963,
      "learning_rate": 4.210350496388133e-05,
      "loss": 0.0525,
      "num_input_tokens_seen": 44220768,
      "step": 20485
    },
    {
      "epoch": 3.3425774877650896,
      "grad_norm": 0.09991085529327393,
      "learning_rate": 4.2098312766597305e-05,
      "loss": 0.1057,
      "num_input_tokens_seen": 44231936,
      "step": 20490
    },
    {
      "epoch": 3.343393148450245,
      "grad_norm": 0.08061181008815765,
      "learning_rate": 4.209311918323729e-05,
      "loss": 0.1447,
      "num_input_tokens_seen": 44242912,
      "step": 20495
    },
    {
      "epoch": 3.3442088091353996,
      "grad_norm": 0.15167714655399323,
      "learning_rate": 4.208792421422231e-05,
      "loss": 0.0979,
      "num_input_tokens_seen": 44254144,
      "step": 20500
    },
    {
      "epoch": 3.3450244698205545,
      "grad_norm": 0.5698845982551575,
      "learning_rate": 4.208272785997348e-05,
      "loss": 0.0788,
      "num_input_tokens_seen": 44264832,
      "step": 20505
    },
    {
      "epoch": 3.3458401305057097,
      "grad_norm": 2.7397055625915527,
      "learning_rate": 4.207753012091207e-05,
      "loss": 0.1534,
      "num_input_tokens_seen": 44276800,
      "step": 20510
    },
    {
      "epoch": 3.3466557911908645,
      "grad_norm": 7.79945707321167,
      "learning_rate": 4.20723309974594e-05,
      "loss": 0.0505,
      "num_input_tokens_seen": 44287584,
      "step": 20515
    },
    {
      "epoch": 3.34747145187602,
      "grad_norm": 9.29866886138916,
      "learning_rate": 4.2067130490036964e-05,
      "loss": 0.0592,
      "num_input_tokens_seen": 44298688,
      "step": 20520
    },
    {
      "epoch": 3.3482871125611746,
      "grad_norm": 0.09406154602766037,
      "learning_rate": 4.206192859906633e-05,
      "loss": 0.0137,
      "num_input_tokens_seen": 44309184,
      "step": 20525
    },
    {
      "epoch": 3.3491027732463294,
      "grad_norm": 0.11522577702999115,
      "learning_rate": 4.205672532496919e-05,
      "loss": 0.2563,
      "num_input_tokens_seen": 44319872,
      "step": 20530
    },
    {
      "epoch": 3.3499184339314847,
      "grad_norm": 4.166998386383057,
      "learning_rate": 4.205152066816736e-05,
      "loss": 0.1554,
      "num_input_tokens_seen": 44330432,
      "step": 20535
    },
    {
      "epoch": 3.3507340946166395,
      "grad_norm": 0.469267874956131,
      "learning_rate": 4.204631462908274e-05,
      "loss": 0.227,
      "num_input_tokens_seen": 44341920,
      "step": 20540
    },
    {
      "epoch": 3.3515497553017943,
      "grad_norm": 0.7317011952400208,
      "learning_rate": 4.2041107208137366e-05,
      "loss": 0.0236,
      "num_input_tokens_seen": 44352096,
      "step": 20545
    },
    {
      "epoch": 3.3523654159869496,
      "grad_norm": 1.8006529808044434,
      "learning_rate": 4.203589840575337e-05,
      "loss": 0.0204,
      "num_input_tokens_seen": 44363456,
      "step": 20550
    },
    {
      "epoch": 3.3531810766721044,
      "grad_norm": 0.26886630058288574,
      "learning_rate": 4.203068822235302e-05,
      "loss": 0.0149,
      "num_input_tokens_seen": 44375168,
      "step": 20555
    },
    {
      "epoch": 3.3539967373572592,
      "grad_norm": 0.051335785537958145,
      "learning_rate": 4.2025476658358656e-05,
      "loss": 0.0156,
      "num_input_tokens_seen": 44385824,
      "step": 20560
    },
    {
      "epoch": 3.3548123980424145,
      "grad_norm": 2.6601674556732178,
      "learning_rate": 4.202026371419278e-05,
      "loss": 0.2979,
      "num_input_tokens_seen": 44396736,
      "step": 20565
    },
    {
      "epoch": 3.3556280587275693,
      "grad_norm": 0.06045924127101898,
      "learning_rate": 4.201504939027796e-05,
      "loss": 0.1814,
      "num_input_tokens_seen": 44407648,
      "step": 20570
    },
    {
      "epoch": 3.356443719412724,
      "grad_norm": 0.08100897073745728,
      "learning_rate": 4.20098336870369e-05,
      "loss": 0.1006,
      "num_input_tokens_seen": 44418336,
      "step": 20575
    },
    {
      "epoch": 3.3572593800978794,
      "grad_norm": 7.067513942718506,
      "learning_rate": 4.200461660489242e-05,
      "loss": 0.1193,
      "num_input_tokens_seen": 44428960,
      "step": 20580
    },
    {
      "epoch": 3.358075040783034,
      "grad_norm": 0.15315377712249756,
      "learning_rate": 4.199939814426744e-05,
      "loss": 0.0467,
      "num_input_tokens_seen": 44440608,
      "step": 20585
    },
    {
      "epoch": 3.358890701468189,
      "grad_norm": 0.11060906201601028,
      "learning_rate": 4.1994178305584996e-05,
      "loss": 0.1923,
      "num_input_tokens_seen": 44451040,
      "step": 20590
    },
    {
      "epoch": 3.3597063621533443,
      "grad_norm": 0.46185383200645447,
      "learning_rate": 4.198895708926822e-05,
      "loss": 0.0389,
      "num_input_tokens_seen": 44461792,
      "step": 20595
    },
    {
      "epoch": 3.360522022838499,
      "grad_norm": 0.028823737055063248,
      "learning_rate": 4.198373449574039e-05,
      "loss": 0.0052,
      "num_input_tokens_seen": 44472000,
      "step": 20600
    },
    {
      "epoch": 3.3613376835236544,
      "grad_norm": 6.627003192901611,
      "learning_rate": 4.197851052542486e-05,
      "loss": 0.0401,
      "num_input_tokens_seen": 44482752,
      "step": 20605
    },
    {
      "epoch": 3.362153344208809,
      "grad_norm": 0.37836965918540955,
      "learning_rate": 4.197328517874513e-05,
      "loss": 0.1861,
      "num_input_tokens_seen": 44494624,
      "step": 20610
    },
    {
      "epoch": 3.362969004893964,
      "grad_norm": 3.33638858795166,
      "learning_rate": 4.1968058456124756e-05,
      "loss": 0.2023,
      "num_input_tokens_seen": 44504256,
      "step": 20615
    },
    {
      "epoch": 3.3637846655791193,
      "grad_norm": 0.037092022597789764,
      "learning_rate": 4.196283035798749e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 44515680,
      "step": 20620
    },
    {
      "epoch": 3.364600326264274,
      "grad_norm": 0.5618161559104919,
      "learning_rate": 4.1957600884757124e-05,
      "loss": 0.0323,
      "num_input_tokens_seen": 44527360,
      "step": 20625
    },
    {
      "epoch": 3.365415986949429,
      "grad_norm": 0.06743231415748596,
      "learning_rate": 4.195237003685759e-05,
      "loss": 0.1649,
      "num_input_tokens_seen": 44537600,
      "step": 20630
    },
    {
      "epoch": 3.366231647634584,
      "grad_norm": 6.578611373901367,
      "learning_rate": 4.194713781471292e-05,
      "loss": 0.014,
      "num_input_tokens_seen": 44548704,
      "step": 20635
    },
    {
      "epoch": 3.367047308319739,
      "grad_norm": 0.08273762464523315,
      "learning_rate": 4.194190421874727e-05,
      "loss": 0.168,
      "num_input_tokens_seen": 44559584,
      "step": 20640
    },
    {
      "epoch": 3.367862969004894,
      "grad_norm": 0.4571586847305298,
      "learning_rate": 4.193666924938491e-05,
      "loss": 0.0049,
      "num_input_tokens_seen": 44568576,
      "step": 20645
    },
    {
      "epoch": 3.368678629690049,
      "grad_norm": 4.7706217765808105,
      "learning_rate": 4.1931432907050196e-05,
      "loss": 0.2462,
      "num_input_tokens_seen": 44579968,
      "step": 20650
    },
    {
      "epoch": 3.369494290375204,
      "grad_norm": 0.12259075790643692,
      "learning_rate": 4.192619519216763e-05,
      "loss": 0.2038,
      "num_input_tokens_seen": 44591552,
      "step": 20655
    },
    {
      "epoch": 3.370309951060359,
      "grad_norm": 1.7152997255325317,
      "learning_rate": 4.192095610516179e-05,
      "loss": 0.1175,
      "num_input_tokens_seen": 44602016,
      "step": 20660
    },
    {
      "epoch": 3.371125611745514,
      "grad_norm": 0.030470367521047592,
      "learning_rate": 4.1915715646457385e-05,
      "loss": 0.0843,
      "num_input_tokens_seen": 44611936,
      "step": 20665
    },
    {
      "epoch": 3.3719412724306688,
      "grad_norm": 0.10524343699216843,
      "learning_rate": 4.191047381647925e-05,
      "loss": 0.1735,
      "num_input_tokens_seen": 44623232,
      "step": 20670
    },
    {
      "epoch": 3.3727569331158236,
      "grad_norm": 0.05628953501582146,
      "learning_rate": 4.190523061565231e-05,
      "loss": 0.0332,
      "num_input_tokens_seen": 44634592,
      "step": 20675
    },
    {
      "epoch": 3.373572593800979,
      "grad_norm": 0.01846391148865223,
      "learning_rate": 4.189998604440159e-05,
      "loss": 0.0234,
      "num_input_tokens_seen": 44645568,
      "step": 20680
    },
    {
      "epoch": 3.3743882544861337,
      "grad_norm": 13.302069664001465,
      "learning_rate": 4.189474010315226e-05,
      "loss": 0.2192,
      "num_input_tokens_seen": 44656992,
      "step": 20685
    },
    {
      "epoch": 3.375203915171289,
      "grad_norm": 0.1590079665184021,
      "learning_rate": 4.188949279232958e-05,
      "loss": 0.069,
      "num_input_tokens_seen": 44669248,
      "step": 20690
    },
    {
      "epoch": 3.3760195758564437,
      "grad_norm": 0.13195693492889404,
      "learning_rate": 4.188424411235891e-05,
      "loss": 0.1071,
      "num_input_tokens_seen": 44680864,
      "step": 20695
    },
    {
      "epoch": 3.3768352365415986,
      "grad_norm": 0.0731479600071907,
      "learning_rate": 4.1878994063665734e-05,
      "loss": 0.1902,
      "num_input_tokens_seen": 44691456,
      "step": 20700
    },
    {
      "epoch": 3.377650897226754,
      "grad_norm": 0.04988702014088631,
      "learning_rate": 4.187374264667566e-05,
      "loss": 0.0075,
      "num_input_tokens_seen": 44702336,
      "step": 20705
    },
    {
      "epoch": 3.3784665579119086,
      "grad_norm": 3.4724490642547607,
      "learning_rate": 4.1868489861814394e-05,
      "loss": 0.4671,
      "num_input_tokens_seen": 44713728,
      "step": 20710
    },
    {
      "epoch": 3.3792822185970635,
      "grad_norm": 5.628291606903076,
      "learning_rate": 4.1863235709507755e-05,
      "loss": 0.1601,
      "num_input_tokens_seen": 44723968,
      "step": 20715
    },
    {
      "epoch": 3.3800978792822187,
      "grad_norm": 0.056657902896404266,
      "learning_rate": 4.1857980190181655e-05,
      "loss": 0.0375,
      "num_input_tokens_seen": 44734304,
      "step": 20720
    },
    {
      "epoch": 3.3809135399673735,
      "grad_norm": 4.896595001220703,
      "learning_rate": 4.1852723304262145e-05,
      "loss": 0.1615,
      "num_input_tokens_seen": 44745568,
      "step": 20725
    },
    {
      "epoch": 3.3817292006525284,
      "grad_norm": 4.006829738616943,
      "learning_rate": 4.1847465052175386e-05,
      "loss": 0.0521,
      "num_input_tokens_seen": 44757216,
      "step": 20730
    },
    {
      "epoch": 3.3825448613376836,
      "grad_norm": 0.04925549402832985,
      "learning_rate": 4.184220543434762e-05,
      "loss": 0.0489,
      "num_input_tokens_seen": 44767808,
      "step": 20735
    },
    {
      "epoch": 3.3833605220228384,
      "grad_norm": 5.684967994689941,
      "learning_rate": 4.1836944451205215e-05,
      "loss": 0.1919,
      "num_input_tokens_seen": 44778208,
      "step": 20740
    },
    {
      "epoch": 3.3841761827079937,
      "grad_norm": 0.06142764911055565,
      "learning_rate": 4.1831682103174676e-05,
      "loss": 0.0585,
      "num_input_tokens_seen": 44788032,
      "step": 20745
    },
    {
      "epoch": 3.3849918433931485,
      "grad_norm": 2.765772819519043,
      "learning_rate": 4.182641839068259e-05,
      "loss": 0.1909,
      "num_input_tokens_seen": 44797440,
      "step": 20750
    },
    {
      "epoch": 3.3858075040783033,
      "grad_norm": 7.4340009689331055,
      "learning_rate": 4.182115331415564e-05,
      "loss": 0.1822,
      "num_input_tokens_seen": 44809248,
      "step": 20755
    },
    {
      "epoch": 3.3866231647634586,
      "grad_norm": 7.273159980773926,
      "learning_rate": 4.1815886874020646e-05,
      "loss": 0.1483,
      "num_input_tokens_seen": 44820768,
      "step": 20760
    },
    {
      "epoch": 3.3874388254486134,
      "grad_norm": 0.15974025428295135,
      "learning_rate": 4.181061907070455e-05,
      "loss": 0.0072,
      "num_input_tokens_seen": 44832128,
      "step": 20765
    },
    {
      "epoch": 3.3882544861337682,
      "grad_norm": 2.6168227195739746,
      "learning_rate": 4.180534990463437e-05,
      "loss": 0.1866,
      "num_input_tokens_seen": 44840992,
      "step": 20770
    },
    {
      "epoch": 3.3890701468189235,
      "grad_norm": 0.29398149251937866,
      "learning_rate": 4.1800079376237265e-05,
      "loss": 0.0418,
      "num_input_tokens_seen": 44851232,
      "step": 20775
    },
    {
      "epoch": 3.3898858075040783,
      "grad_norm": 0.07715565711259842,
      "learning_rate": 4.179480748594048e-05,
      "loss": 0.0193,
      "num_input_tokens_seen": 44860672,
      "step": 20780
    },
    {
      "epoch": 3.390701468189233,
      "grad_norm": 0.3808293342590332,
      "learning_rate": 4.178953423417138e-05,
      "loss": 0.0299,
      "num_input_tokens_seen": 44870688,
      "step": 20785
    },
    {
      "epoch": 3.3915171288743884,
      "grad_norm": 3.0713024139404297,
      "learning_rate": 4.1784259621357444e-05,
      "loss": 0.0206,
      "num_input_tokens_seen": 44881760,
      "step": 20790
    },
    {
      "epoch": 3.392332789559543,
      "grad_norm": 0.355150431394577,
      "learning_rate": 4.1778983647926274e-05,
      "loss": 0.0104,
      "num_input_tokens_seen": 44892896,
      "step": 20795
    },
    {
      "epoch": 3.393148450244698,
      "grad_norm": 0.4394971430301666,
      "learning_rate": 4.177370631430554e-05,
      "loss": 0.0883,
      "num_input_tokens_seen": 44903296,
      "step": 20800
    },
    {
      "epoch": 3.3939641109298533,
      "grad_norm": 0.1047433614730835,
      "learning_rate": 4.176842762092307e-05,
      "loss": 0.1069,
      "num_input_tokens_seen": 44913664,
      "step": 20805
    },
    {
      "epoch": 3.394779771615008,
      "grad_norm": 0.9792728424072266,
      "learning_rate": 4.176314756820677e-05,
      "loss": 0.1567,
      "num_input_tokens_seen": 44924864,
      "step": 20810
    },
    {
      "epoch": 3.395595432300163,
      "grad_norm": 0.0860372856259346,
      "learning_rate": 4.175786615658468e-05,
      "loss": 0.0993,
      "num_input_tokens_seen": 44935008,
      "step": 20815
    },
    {
      "epoch": 3.396411092985318,
      "grad_norm": 0.03137561306357384,
      "learning_rate": 4.175258338648493e-05,
      "loss": 0.0732,
      "num_input_tokens_seen": 44945280,
      "step": 20820
    },
    {
      "epoch": 3.397226753670473,
      "grad_norm": 0.35096603631973267,
      "learning_rate": 4.174729925833576e-05,
      "loss": 0.0592,
      "num_input_tokens_seen": 44956064,
      "step": 20825
    },
    {
      "epoch": 3.3980424143556283,
      "grad_norm": 0.047019828110933304,
      "learning_rate": 4.174201377256555e-05,
      "loss": 0.0811,
      "num_input_tokens_seen": 44967424,
      "step": 20830
    },
    {
      "epoch": 3.398858075040783,
      "grad_norm": 0.12674105167388916,
      "learning_rate": 4.173672692960274e-05,
      "loss": 0.0743,
      "num_input_tokens_seen": 44977952,
      "step": 20835
    },
    {
      "epoch": 3.399673735725938,
      "grad_norm": 0.051974739879369736,
      "learning_rate": 4.173143872987594e-05,
      "loss": 0.1049,
      "num_input_tokens_seen": 44988384,
      "step": 20840
    },
    {
      "epoch": 3.400489396411093,
      "grad_norm": 0.02065962553024292,
      "learning_rate": 4.172614917381381e-05,
      "loss": 0.0057,
      "num_input_tokens_seen": 44998816,
      "step": 20845
    },
    {
      "epoch": 3.401305057096248,
      "grad_norm": 5.289228439331055,
      "learning_rate": 4.1720858261845166e-05,
      "loss": 0.1662,
      "num_input_tokens_seen": 45010112,
      "step": 20850
    },
    {
      "epoch": 3.402120717781403,
      "grad_norm": 9.325918197631836,
      "learning_rate": 4.171556599439891e-05,
      "loss": 0.2236,
      "num_input_tokens_seen": 45021184,
      "step": 20855
    },
    {
      "epoch": 3.402936378466558,
      "grad_norm": 0.10597573220729828,
      "learning_rate": 4.1710272371904055e-05,
      "loss": 0.2338,
      "num_input_tokens_seen": 45030976,
      "step": 20860
    },
    {
      "epoch": 3.403752039151713,
      "grad_norm": 1.7547309398651123,
      "learning_rate": 4.170497739478974e-05,
      "loss": 0.264,
      "num_input_tokens_seen": 45042400,
      "step": 20865
    },
    {
      "epoch": 3.4045676998368677,
      "grad_norm": 0.03804885968565941,
      "learning_rate": 4.16996810634852e-05,
      "loss": 0.067,
      "num_input_tokens_seen": 45052352,
      "step": 20870
    },
    {
      "epoch": 3.405383360522023,
      "grad_norm": 0.1667662113904953,
      "learning_rate": 4.1694383378419774e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 45063520,
      "step": 20875
    },
    {
      "epoch": 3.4061990212071778,
      "grad_norm": 10.003411293029785,
      "learning_rate": 4.168908434002292e-05,
      "loss": 0.0792,
      "num_input_tokens_seen": 45074144,
      "step": 20880
    },
    {
      "epoch": 3.407014681892333,
      "grad_norm": 0.23482690751552582,
      "learning_rate": 4.168378394872422e-05,
      "loss": 0.0184,
      "num_input_tokens_seen": 45084704,
      "step": 20885
    },
    {
      "epoch": 3.407830342577488,
      "grad_norm": 2.349637031555176,
      "learning_rate": 4.167848220495334e-05,
      "loss": 0.1031,
      "num_input_tokens_seen": 45095744,
      "step": 20890
    },
    {
      "epoch": 3.4086460032626427,
      "grad_norm": 0.12306543439626694,
      "learning_rate": 4.167317910914006e-05,
      "loss": 0.089,
      "num_input_tokens_seen": 45105952,
      "step": 20895
    },
    {
      "epoch": 3.4094616639477975,
      "grad_norm": 0.12144485861063004,
      "learning_rate": 4.166787466171429e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 45118080,
      "step": 20900
    },
    {
      "epoch": 3.4102773246329527,
      "grad_norm": 0.024729499593377113,
      "learning_rate": 4.166256886310602e-05,
      "loss": 0.0096,
      "num_input_tokens_seen": 45129024,
      "step": 20905
    },
    {
      "epoch": 3.4110929853181076,
      "grad_norm": 2.3089208602905273,
      "learning_rate": 4.165726171374538e-05,
      "loss": 0.0283,
      "num_input_tokens_seen": 45139200,
      "step": 20910
    },
    {
      "epoch": 3.411908646003263,
      "grad_norm": 0.04093559458851814,
      "learning_rate": 4.165195321406259e-05,
      "loss": 0.0837,
      "num_input_tokens_seen": 45150752,
      "step": 20915
    },
    {
      "epoch": 3.4127243066884176,
      "grad_norm": 0.11183979362249374,
      "learning_rate": 4.164664336448797e-05,
      "loss": 0.1029,
      "num_input_tokens_seen": 45162016,
      "step": 20920
    },
    {
      "epoch": 3.4135399673735725,
      "grad_norm": 3.6680314540863037,
      "learning_rate": 4.164133216545199e-05,
      "loss": 0.1584,
      "num_input_tokens_seen": 45171904,
      "step": 20925
    },
    {
      "epoch": 3.4143556280587277,
      "grad_norm": 7.581708908081055,
      "learning_rate": 4.163601961738517e-05,
      "loss": 0.2686,
      "num_input_tokens_seen": 45183712,
      "step": 20930
    },
    {
      "epoch": 3.4151712887438825,
      "grad_norm": 1.1461793184280396,
      "learning_rate": 4.16307057207182e-05,
      "loss": 0.0608,
      "num_input_tokens_seen": 45194656,
      "step": 20935
    },
    {
      "epoch": 3.4159869494290374,
      "grad_norm": 0.03561193495988846,
      "learning_rate": 4.162539047588183e-05,
      "loss": 0.2059,
      "num_input_tokens_seen": 45204800,
      "step": 20940
    },
    {
      "epoch": 3.4168026101141926,
      "grad_norm": 10.955655097961426,
      "learning_rate": 4.162007388330696e-05,
      "loss": 0.1946,
      "num_input_tokens_seen": 45215776,
      "step": 20945
    },
    {
      "epoch": 3.4176182707993474,
      "grad_norm": 0.05724068358540535,
      "learning_rate": 4.1614755943424575e-05,
      "loss": 0.0306,
      "num_input_tokens_seen": 45225952,
      "step": 20950
    },
    {
      "epoch": 3.4184339314845023,
      "grad_norm": 1.750388503074646,
      "learning_rate": 4.160943665666577e-05,
      "loss": 0.1012,
      "num_input_tokens_seen": 45237312,
      "step": 20955
    },
    {
      "epoch": 3.4192495921696575,
      "grad_norm": 0.2838281989097595,
      "learning_rate": 4.160411602346175e-05,
      "loss": 0.0107,
      "num_input_tokens_seen": 45246912,
      "step": 20960
    },
    {
      "epoch": 3.4200652528548123,
      "grad_norm": 0.1384785771369934,
      "learning_rate": 4.159879404424384e-05,
      "loss": 0.1127,
      "num_input_tokens_seen": 45257440,
      "step": 20965
    },
    {
      "epoch": 3.4208809135399676,
      "grad_norm": 0.13913112878799438,
      "learning_rate": 4.159347071944346e-05,
      "loss": 0.1909,
      "num_input_tokens_seen": 45268416,
      "step": 20970
    },
    {
      "epoch": 3.4216965742251224,
      "grad_norm": 0.32134515047073364,
      "learning_rate": 4.158814604949215e-05,
      "loss": 0.1516,
      "num_input_tokens_seen": 45278848,
      "step": 20975
    },
    {
      "epoch": 3.4225122349102772,
      "grad_norm": 0.4873961806297302,
      "learning_rate": 4.158282003482156e-05,
      "loss": 0.0175,
      "num_input_tokens_seen": 45291072,
      "step": 20980
    },
    {
      "epoch": 3.4233278955954325,
      "grad_norm": 0.3885081708431244,
      "learning_rate": 4.157749267586343e-05,
      "loss": 0.1169,
      "num_input_tokens_seen": 45302336,
      "step": 20985
    },
    {
      "epoch": 3.4241435562805873,
      "grad_norm": 0.17299550771713257,
      "learning_rate": 4.1572163973049624e-05,
      "loss": 0.0423,
      "num_input_tokens_seen": 45312864,
      "step": 20990
    },
    {
      "epoch": 3.424959216965742,
      "grad_norm": 0.3936578631401062,
      "learning_rate": 4.1566833926812135e-05,
      "loss": 0.1395,
      "num_input_tokens_seen": 45323744,
      "step": 20995
    },
    {
      "epoch": 3.4257748776508974,
      "grad_norm": 0.4577231705188751,
      "learning_rate": 4.1561502537583016e-05,
      "loss": 0.1863,
      "num_input_tokens_seen": 45333760,
      "step": 21000
    },
    {
      "epoch": 3.426590538336052,
      "grad_norm": 1.2642056941986084,
      "learning_rate": 4.155616980579447e-05,
      "loss": 0.2441,
      "num_input_tokens_seen": 45344896,
      "step": 21005
    },
    {
      "epoch": 3.427406199021207,
      "grad_norm": 9.072660446166992,
      "learning_rate": 4.155083573187881e-05,
      "loss": 0.1113,
      "num_input_tokens_seen": 45354336,
      "step": 21010
    },
    {
      "epoch": 3.4282218597063623,
      "grad_norm": 0.16346946358680725,
      "learning_rate": 4.154550031626842e-05,
      "loss": 0.3153,
      "num_input_tokens_seen": 45365376,
      "step": 21015
    },
    {
      "epoch": 3.429037520391517,
      "grad_norm": 3.710902452468872,
      "learning_rate": 4.1540163559395816e-05,
      "loss": 0.0215,
      "num_input_tokens_seen": 45376416,
      "step": 21020
    },
    {
      "epoch": 3.429853181076672,
      "grad_norm": 0.0782494992017746,
      "learning_rate": 4.153482546169364e-05,
      "loss": 0.0936,
      "num_input_tokens_seen": 45386272,
      "step": 21025
    },
    {
      "epoch": 3.430668841761827,
      "grad_norm": 0.04548351466655731,
      "learning_rate": 4.15294860235946e-05,
      "loss": 0.0538,
      "num_input_tokens_seen": 45395936,
      "step": 21030
    },
    {
      "epoch": 3.431484502446982,
      "grad_norm": 0.07497764378786087,
      "learning_rate": 4.152414524553156e-05,
      "loss": 0.0151,
      "num_input_tokens_seen": 45407680,
      "step": 21035
    },
    {
      "epoch": 3.432300163132137,
      "grad_norm": 2.2037370204925537,
      "learning_rate": 4.1518803127937464e-05,
      "loss": 0.0161,
      "num_input_tokens_seen": 45417088,
      "step": 21040
    },
    {
      "epoch": 3.433115823817292,
      "grad_norm": 0.6846911311149597,
      "learning_rate": 4.1513459671245384e-05,
      "loss": 0.0445,
      "num_input_tokens_seen": 45429536,
      "step": 21045
    },
    {
      "epoch": 3.433931484502447,
      "grad_norm": 0.07101361453533173,
      "learning_rate": 4.150811487588846e-05,
      "loss": 0.0205,
      "num_input_tokens_seen": 45440320,
      "step": 21050
    },
    {
      "epoch": 3.434747145187602,
      "grad_norm": 0.4925655722618103,
      "learning_rate": 4.150276874229999e-05,
      "loss": 0.2046,
      "num_input_tokens_seen": 45451008,
      "step": 21055
    },
    {
      "epoch": 3.435562805872757,
      "grad_norm": 1.419433355331421,
      "learning_rate": 4.149742127091335e-05,
      "loss": 0.0333,
      "num_input_tokens_seen": 45462432,
      "step": 21060
    },
    {
      "epoch": 3.436378466557912,
      "grad_norm": 8.319808006286621,
      "learning_rate": 4.149207246216203e-05,
      "loss": 0.1205,
      "num_input_tokens_seen": 45472544,
      "step": 21065
    },
    {
      "epoch": 3.437194127243067,
      "grad_norm": 0.07023496180772781,
      "learning_rate": 4.1486722316479635e-05,
      "loss": 0.0681,
      "num_input_tokens_seen": 45483168,
      "step": 21070
    },
    {
      "epoch": 3.438009787928222,
      "grad_norm": 0.05403876677155495,
      "learning_rate": 4.1481370834299884e-05,
      "loss": 0.0056,
      "num_input_tokens_seen": 45492832,
      "step": 21075
    },
    {
      "epoch": 3.4388254486133767,
      "grad_norm": 1.4059116840362549,
      "learning_rate": 4.1476018016056583e-05,
      "loss": 0.2701,
      "num_input_tokens_seen": 45504064,
      "step": 21080
    },
    {
      "epoch": 3.439641109298532,
      "grad_norm": 4.7639923095703125,
      "learning_rate": 4.1470663862183664e-05,
      "loss": 0.1012,
      "num_input_tokens_seen": 45514336,
      "step": 21085
    },
    {
      "epoch": 3.4404567699836868,
      "grad_norm": 0.06506458669900894,
      "learning_rate": 4.146530837311516e-05,
      "loss": 0.0895,
      "num_input_tokens_seen": 45524480,
      "step": 21090
    },
    {
      "epoch": 3.4412724306688416,
      "grad_norm": 11.314379692077637,
      "learning_rate": 4.145995154928521e-05,
      "loss": 0.3164,
      "num_input_tokens_seen": 45535264,
      "step": 21095
    },
    {
      "epoch": 3.442088091353997,
      "grad_norm": 7.392039775848389,
      "learning_rate": 4.1454593391128084e-05,
      "loss": 0.0397,
      "num_input_tokens_seen": 45546560,
      "step": 21100
    },
    {
      "epoch": 3.4429037520391517,
      "grad_norm": 0.08344903588294983,
      "learning_rate": 4.144923389907812e-05,
      "loss": 0.0747,
      "num_input_tokens_seen": 45558016,
      "step": 21105
    },
    {
      "epoch": 3.443719412724307,
      "grad_norm": 0.10498899966478348,
      "learning_rate": 4.1443873073569796e-05,
      "loss": 0.0587,
      "num_input_tokens_seen": 45568512,
      "step": 21110
    },
    {
      "epoch": 3.4445350734094617,
      "grad_norm": 0.8218194842338562,
      "learning_rate": 4.143851091503768e-05,
      "loss": 0.0851,
      "num_input_tokens_seen": 45579968,
      "step": 21115
    },
    {
      "epoch": 3.4453507340946166,
      "grad_norm": 0.4953779876232147,
      "learning_rate": 4.1433147423916466e-05,
      "loss": 0.0262,
      "num_input_tokens_seen": 45592064,
      "step": 21120
    },
    {
      "epoch": 3.4461663947797714,
      "grad_norm": 0.06428025662899017,
      "learning_rate": 4.1427782600640943e-05,
      "loss": 0.1159,
      "num_input_tokens_seen": 45603040,
      "step": 21125
    },
    {
      "epoch": 3.4469820554649266,
      "grad_norm": 0.13550199568271637,
      "learning_rate": 4.1422416445646e-05,
      "loss": 0.0592,
      "num_input_tokens_seen": 45613536,
      "step": 21130
    },
    {
      "epoch": 3.4477977161500815,
      "grad_norm": 0.29464489221572876,
      "learning_rate": 4.141704895936666e-05,
      "loss": 0.0622,
      "num_input_tokens_seen": 45625120,
      "step": 21135
    },
    {
      "epoch": 3.4486133768352367,
      "grad_norm": 6.6131815910339355,
      "learning_rate": 4.141168014223803e-05,
      "loss": 0.1254,
      "num_input_tokens_seen": 45635744,
      "step": 21140
    },
    {
      "epoch": 3.4494290375203915,
      "grad_norm": 0.18010494112968445,
      "learning_rate": 4.1406309994695335e-05,
      "loss": 0.1449,
      "num_input_tokens_seen": 45646432,
      "step": 21145
    },
    {
      "epoch": 3.4502446982055464,
      "grad_norm": 2.384899377822876,
      "learning_rate": 4.1400938517173905e-05,
      "loss": 0.0981,
      "num_input_tokens_seen": 45656576,
      "step": 21150
    },
    {
      "epoch": 3.4510603588907016,
      "grad_norm": 2.9363596439361572,
      "learning_rate": 4.139556571010919e-05,
      "loss": 0.2242,
      "num_input_tokens_seen": 45667904,
      "step": 21155
    },
    {
      "epoch": 3.4518760195758564,
      "grad_norm": 0.039616163820028305,
      "learning_rate": 4.139019157393672e-05,
      "loss": 0.011,
      "num_input_tokens_seen": 45678528,
      "step": 21160
    },
    {
      "epoch": 3.4526916802610113,
      "grad_norm": 3.5711658000946045,
      "learning_rate": 4.138481610909216e-05,
      "loss": 0.1952,
      "num_input_tokens_seen": 45689312,
      "step": 21165
    },
    {
      "epoch": 3.4535073409461665,
      "grad_norm": 1.4512710571289062,
      "learning_rate": 4.137943931601127e-05,
      "loss": 0.0097,
      "num_input_tokens_seen": 45700160,
      "step": 21170
    },
    {
      "epoch": 3.4543230016313213,
      "grad_norm": 8.78475570678711,
      "learning_rate": 4.1374061195129924e-05,
      "loss": 0.2915,
      "num_input_tokens_seen": 45710432,
      "step": 21175
    },
    {
      "epoch": 3.455138662316476,
      "grad_norm": 0.7189760804176331,
      "learning_rate": 4.13686817468841e-05,
      "loss": 0.0062,
      "num_input_tokens_seen": 45721248,
      "step": 21180
    },
    {
      "epoch": 3.4559543230016314,
      "grad_norm": 3.4032669067382812,
      "learning_rate": 4.1363300971709866e-05,
      "loss": 0.0931,
      "num_input_tokens_seen": 45732768,
      "step": 21185
    },
    {
      "epoch": 3.4567699836867862,
      "grad_norm": 0.06290104985237122,
      "learning_rate": 4.135791887004344e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 45743200,
      "step": 21190
    },
    {
      "epoch": 3.4575856443719415,
      "grad_norm": 0.16576777398586273,
      "learning_rate": 4.13525354423211e-05,
      "loss": 0.2485,
      "num_input_tokens_seen": 45753376,
      "step": 21195
    },
    {
      "epoch": 3.4584013050570963,
      "grad_norm": 3.3895223140716553,
      "learning_rate": 4.1347150688979275e-05,
      "loss": 0.0537,
      "num_input_tokens_seen": 45764064,
      "step": 21200
    },
    {
      "epoch": 3.459216965742251,
      "grad_norm": 6.020229339599609,
      "learning_rate": 4.134176461045447e-05,
      "loss": 0.2202,
      "num_input_tokens_seen": 45774944,
      "step": 21205
    },
    {
      "epoch": 3.4600326264274064,
      "grad_norm": 0.6488676071166992,
      "learning_rate": 4.133637720718331e-05,
      "loss": 0.0603,
      "num_input_tokens_seen": 45785856,
      "step": 21210
    },
    {
      "epoch": 3.460848287112561,
      "grad_norm": 5.132880687713623,
      "learning_rate": 4.133098847960252e-05,
      "loss": 0.3604,
      "num_input_tokens_seen": 45796352,
      "step": 21215
    },
    {
      "epoch": 3.461663947797716,
      "grad_norm": 5.736630916595459,
      "learning_rate": 4.1325598428148935e-05,
      "loss": 0.0395,
      "num_input_tokens_seen": 45808160,
      "step": 21220
    },
    {
      "epoch": 3.4624796084828713,
      "grad_norm": 0.06908325850963593,
      "learning_rate": 4.132020705325952e-05,
      "loss": 0.0071,
      "num_input_tokens_seen": 45819744,
      "step": 21225
    },
    {
      "epoch": 3.463295269168026,
      "grad_norm": 0.03077886253595352,
      "learning_rate": 4.13148143553713e-05,
      "loss": 0.2597,
      "num_input_tokens_seen": 45830048,
      "step": 21230
    },
    {
      "epoch": 3.464110929853181,
      "grad_norm": 2.3064441680908203,
      "learning_rate": 4.130942033492146e-05,
      "loss": 0.0756,
      "num_input_tokens_seen": 45841216,
      "step": 21235
    },
    {
      "epoch": 3.464926590538336,
      "grad_norm": 0.12238824367523193,
      "learning_rate": 4.1304024992347245e-05,
      "loss": 0.0749,
      "num_input_tokens_seen": 45850720,
      "step": 21240
    },
    {
      "epoch": 3.465742251223491,
      "grad_norm": 6.307570934295654,
      "learning_rate": 4.129862832808604e-05,
      "loss": 0.272,
      "num_input_tokens_seen": 45862272,
      "step": 21245
    },
    {
      "epoch": 3.466557911908646,
      "grad_norm": 0.08646532893180847,
      "learning_rate": 4.129323034257533e-05,
      "loss": 0.1786,
      "num_input_tokens_seen": 45872576,
      "step": 21250
    },
    {
      "epoch": 3.467373572593801,
      "grad_norm": 0.06172173470258713,
      "learning_rate": 4.128783103625269e-05,
      "loss": 0.2169,
      "num_input_tokens_seen": 45882816,
      "step": 21255
    },
    {
      "epoch": 3.468189233278956,
      "grad_norm": 0.13198105990886688,
      "learning_rate": 4.128243040955583e-05,
      "loss": 0.16,
      "num_input_tokens_seen": 45893856,
      "step": 21260
    },
    {
      "epoch": 3.4690048939641107,
      "grad_norm": 3.57318377494812,
      "learning_rate": 4.1277028462922535e-05,
      "loss": 0.1352,
      "num_input_tokens_seen": 45904128,
      "step": 21265
    },
    {
      "epoch": 3.469820554649266,
      "grad_norm": 6.988917827606201,
      "learning_rate": 4.127162519679073e-05,
      "loss": 0.1319,
      "num_input_tokens_seen": 45914624,
      "step": 21270
    },
    {
      "epoch": 3.470636215334421,
      "grad_norm": 1.898236870765686,
      "learning_rate": 4.126622061159843e-05,
      "loss": 0.1336,
      "num_input_tokens_seen": 45926432,
      "step": 21275
    },
    {
      "epoch": 3.471451876019576,
      "grad_norm": 0.3959386348724365,
      "learning_rate": 4.126081470778375e-05,
      "loss": 0.0191,
      "num_input_tokens_seen": 45936608,
      "step": 21280
    },
    {
      "epoch": 3.472267536704731,
      "grad_norm": 0.11599043756723404,
      "learning_rate": 4.125540748578491e-05,
      "loss": 0.1255,
      "num_input_tokens_seen": 45948768,
      "step": 21285
    },
    {
      "epoch": 3.4730831973898857,
      "grad_norm": 0.05390927568078041,
      "learning_rate": 4.124999894604028e-05,
      "loss": 0.0068,
      "num_input_tokens_seen": 45958080,
      "step": 21290
    },
    {
      "epoch": 3.473898858075041,
      "grad_norm": 4.834368705749512,
      "learning_rate": 4.124458908898827e-05,
      "loss": 0.1203,
      "num_input_tokens_seen": 45968768,
      "step": 21295
    },
    {
      "epoch": 3.4747145187601958,
      "grad_norm": 3.5230612754821777,
      "learning_rate": 4.1239177915067454e-05,
      "loss": 0.1929,
      "num_input_tokens_seen": 45979360,
      "step": 21300
    },
    {
      "epoch": 3.4755301794453506,
      "grad_norm": 1.043532133102417,
      "learning_rate": 4.123376542471648e-05,
      "loss": 0.0244,
      "num_input_tokens_seen": 45990880,
      "step": 21305
    },
    {
      "epoch": 3.476345840130506,
      "grad_norm": 0.046062592417001724,
      "learning_rate": 4.122835161837409e-05,
      "loss": 0.1462,
      "num_input_tokens_seen": 46001344,
      "step": 21310
    },
    {
      "epoch": 3.4771615008156607,
      "grad_norm": 1.381137728691101,
      "learning_rate": 4.12229364964792e-05,
      "loss": 0.0699,
      "num_input_tokens_seen": 46013152,
      "step": 21315
    },
    {
      "epoch": 3.4779771615008155,
      "grad_norm": 0.1660504937171936,
      "learning_rate": 4.121752005947076e-05,
      "loss": 0.1241,
      "num_input_tokens_seen": 46023424,
      "step": 21320
    },
    {
      "epoch": 3.4787928221859707,
      "grad_norm": 7.055490493774414,
      "learning_rate": 4.121210230778785e-05,
      "loss": 0.0877,
      "num_input_tokens_seen": 46033568,
      "step": 21325
    },
    {
      "epoch": 3.4796084828711256,
      "grad_norm": 3.550671100616455,
      "learning_rate": 4.120668324186967e-05,
      "loss": 0.0416,
      "num_input_tokens_seen": 46045088,
      "step": 21330
    },
    {
      "epoch": 3.480424143556281,
      "grad_norm": 0.07857126742601395,
      "learning_rate": 4.120126286215552e-05,
      "loss": 0.063,
      "num_input_tokens_seen": 46056640,
      "step": 21335
    },
    {
      "epoch": 3.4812398042414356,
      "grad_norm": 0.11759999394416809,
      "learning_rate": 4.119584116908478e-05,
      "loss": 0.0602,
      "num_input_tokens_seen": 46067008,
      "step": 21340
    },
    {
      "epoch": 3.4820554649265905,
      "grad_norm": 0.061887551099061966,
      "learning_rate": 4.1190418163097e-05,
      "loss": 0.0646,
      "num_input_tokens_seen": 46078048,
      "step": 21345
    },
    {
      "epoch": 3.4828711256117453,
      "grad_norm": 0.040443260222673416,
      "learning_rate": 4.118499384463176e-05,
      "loss": 0.0951,
      "num_input_tokens_seen": 46089568,
      "step": 21350
    },
    {
      "epoch": 3.4836867862969005,
      "grad_norm": 4.102940559387207,
      "learning_rate": 4.1179568214128805e-05,
      "loss": 0.3468,
      "num_input_tokens_seen": 46101024,
      "step": 21355
    },
    {
      "epoch": 3.4845024469820554,
      "grad_norm": 6.000858306884766,
      "learning_rate": 4.117414127202795e-05,
      "loss": 0.1659,
      "num_input_tokens_seen": 46112064,
      "step": 21360
    },
    {
      "epoch": 3.4853181076672106,
      "grad_norm": 12.418498039245605,
      "learning_rate": 4.116871301876914e-05,
      "loss": 0.1087,
      "num_input_tokens_seen": 46122080,
      "step": 21365
    },
    {
      "epoch": 3.4861337683523654,
      "grad_norm": 4.769489288330078,
      "learning_rate": 4.116328345479241e-05,
      "loss": 0.1946,
      "num_input_tokens_seen": 46132992,
      "step": 21370
    },
    {
      "epoch": 3.4869494290375203,
      "grad_norm": 4.555278778076172,
      "learning_rate": 4.115785258053792e-05,
      "loss": 0.2054,
      "num_input_tokens_seen": 46144576,
      "step": 21375
    },
    {
      "epoch": 3.4877650897226755,
      "grad_norm": 0.12587052583694458,
      "learning_rate": 4.1152420396445915e-05,
      "loss": 0.3145,
      "num_input_tokens_seen": 46155296,
      "step": 21380
    },
    {
      "epoch": 3.4885807504078303,
      "grad_norm": 4.871669769287109,
      "learning_rate": 4.1146986902956745e-05,
      "loss": 0.0611,
      "num_input_tokens_seen": 46166528,
      "step": 21385
    },
    {
      "epoch": 3.489396411092985,
      "grad_norm": 0.40100112557411194,
      "learning_rate": 4.1141552100510896e-05,
      "loss": 0.1811,
      "num_input_tokens_seen": 46176672,
      "step": 21390
    },
    {
      "epoch": 3.4902120717781404,
      "grad_norm": 0.15952937304973602,
      "learning_rate": 4.1136115989548926e-05,
      "loss": 0.0188,
      "num_input_tokens_seen": 46186880,
      "step": 21395
    },
    {
      "epoch": 3.4910277324632952,
      "grad_norm": 8.589343070983887,
      "learning_rate": 4.113067857051153e-05,
      "loss": 0.085,
      "num_input_tokens_seen": 46196320,
      "step": 21400
    },
    {
      "epoch": 3.49184339314845,
      "grad_norm": 0.05669253692030907,
      "learning_rate": 4.112523984383948e-05,
      "loss": 0.0679,
      "num_input_tokens_seen": 46207040,
      "step": 21405
    },
    {
      "epoch": 3.4926590538336053,
      "grad_norm": 1.8569155931472778,
      "learning_rate": 4.111979980997366e-05,
      "loss": 0.0304,
      "num_input_tokens_seen": 46217248,
      "step": 21410
    },
    {
      "epoch": 3.49347471451876,
      "grad_norm": 0.23351731896400452,
      "learning_rate": 4.1114358469355084e-05,
      "loss": 0.1226,
      "num_input_tokens_seen": 46226688,
      "step": 21415
    },
    {
      "epoch": 3.4942903752039154,
      "grad_norm": 0.1810547411441803,
      "learning_rate": 4.110891582242485e-05,
      "loss": 0.1612,
      "num_input_tokens_seen": 46236768,
      "step": 21420
    },
    {
      "epoch": 3.49510603588907,
      "grad_norm": 1.3555699586868286,
      "learning_rate": 4.1103471869624154e-05,
      "loss": 0.0848,
      "num_input_tokens_seen": 46247872,
      "step": 21425
    },
    {
      "epoch": 3.495921696574225,
      "grad_norm": 1.4279813766479492,
      "learning_rate": 4.109802661139433e-05,
      "loss": 0.1751,
      "num_input_tokens_seen": 46258688,
      "step": 21430
    },
    {
      "epoch": 3.4967373572593803,
      "grad_norm": 1.8895460367202759,
      "learning_rate": 4.109258004817679e-05,
      "loss": 0.0242,
      "num_input_tokens_seen": 46269504,
      "step": 21435
    },
    {
      "epoch": 3.497553017944535,
      "grad_norm": 4.239985942840576,
      "learning_rate": 4.1087132180413047e-05,
      "loss": 0.1051,
      "num_input_tokens_seen": 46280928,
      "step": 21440
    },
    {
      "epoch": 3.49836867862969,
      "grad_norm": 0.11252937465906143,
      "learning_rate": 4.108168300854475e-05,
      "loss": 0.1634,
      "num_input_tokens_seen": 46291648,
      "step": 21445
    },
    {
      "epoch": 3.499184339314845,
      "grad_norm": 0.35491886734962463,
      "learning_rate": 4.1076232533013635e-05,
      "loss": 0.0715,
      "num_input_tokens_seen": 46302592,
      "step": 21450
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.26530101895332336,
      "learning_rate": 4.1070780754261533e-05,
      "loss": 0.1521,
      "num_input_tokens_seen": 46313216,
      "step": 21455
    },
    {
      "epoch": 3.5,
      "eval_loss": 0.16028907895088196,
      "eval_runtime": 132.9607,
      "eval_samples_per_second": 20.495,
      "eval_steps_per_second": 5.129,
      "num_input_tokens_seen": 46313216,
      "step": 21455
    },
    {
      "epoch": 3.500815660685155,
      "grad_norm": 0.11428025364875793,
      "learning_rate": 4.10653276727304e-05,
      "loss": 0.1485,
      "num_input_tokens_seen": 46322656,
      "step": 21460
    },
    {
      "epoch": 3.50163132137031,
      "grad_norm": 0.42429232597351074,
      "learning_rate": 4.105987328886229e-05,
      "loss": 0.0923,
      "num_input_tokens_seen": 46332576,
      "step": 21465
    },
    {
      "epoch": 3.502446982055465,
      "grad_norm": 0.028669022023677826,
      "learning_rate": 4.1054417603099376e-05,
      "loss": 0.218,
      "num_input_tokens_seen": 46343584,
      "step": 21470
    },
    {
      "epoch": 3.50326264274062,
      "grad_norm": 0.09211162477731705,
      "learning_rate": 4.104896061588391e-05,
      "loss": 0.0127,
      "num_input_tokens_seen": 46354240,
      "step": 21475
    },
    {
      "epoch": 3.504078303425775,
      "grad_norm": 8.533767700195312,
      "learning_rate": 4.1043502327658256e-05,
      "loss": 0.2029,
      "num_input_tokens_seen": 46364736,
      "step": 21480
    },
    {
      "epoch": 3.50489396411093,
      "grad_norm": 0.18505850434303284,
      "learning_rate": 4.1038042738864906e-05,
      "loss": 0.183,
      "num_input_tokens_seen": 46374976,
      "step": 21485
    },
    {
      "epoch": 3.5057096247960846,
      "grad_norm": 2.84120774269104,
      "learning_rate": 4.103258184994644e-05,
      "loss": 0.1389,
      "num_input_tokens_seen": 46386080,
      "step": 21490
    },
    {
      "epoch": 3.50652528548124,
      "grad_norm": 0.7979764342308044,
      "learning_rate": 4.102711966134553e-05,
      "loss": 0.0688,
      "num_input_tokens_seen": 46397152,
      "step": 21495
    },
    {
      "epoch": 3.5073409461663947,
      "grad_norm": 0.1829701066017151,
      "learning_rate": 4.102165617350498e-05,
      "loss": 0.1268,
      "num_input_tokens_seen": 46408480,
      "step": 21500
    },
    {
      "epoch": 3.50815660685155,
      "grad_norm": 0.11440054327249527,
      "learning_rate": 4.101619138686769e-05,
      "loss": 0.1559,
      "num_input_tokens_seen": 46419712,
      "step": 21505
    },
    {
      "epoch": 3.5089722675367048,
      "grad_norm": 7.477168560028076,
      "learning_rate": 4.101072530187666e-05,
      "loss": 0.2169,
      "num_input_tokens_seen": 46431072,
      "step": 21510
    },
    {
      "epoch": 3.5097879282218596,
      "grad_norm": 0.15226133167743683,
      "learning_rate": 4.100525791897501e-05,
      "loss": 0.0295,
      "num_input_tokens_seen": 46441280,
      "step": 21515
    },
    {
      "epoch": 3.5106035889070144,
      "grad_norm": 0.23722317814826965,
      "learning_rate": 4.0999789238605925e-05,
      "loss": 0.2065,
      "num_input_tokens_seen": 46451808,
      "step": 21520
    },
    {
      "epoch": 3.5114192495921697,
      "grad_norm": 3.630800485610962,
      "learning_rate": 4.099431926121276e-05,
      "loss": 0.1932,
      "num_input_tokens_seen": 46463136,
      "step": 21525
    },
    {
      "epoch": 3.5122349102773245,
      "grad_norm": 4.703624725341797,
      "learning_rate": 4.098884798723891e-05,
      "loss": 0.1012,
      "num_input_tokens_seen": 46473280,
      "step": 21530
    },
    {
      "epoch": 3.5130505709624797,
      "grad_norm": 3.5786826610565186,
      "learning_rate": 4.098337541712791e-05,
      "loss": 0.0781,
      "num_input_tokens_seen": 46484032,
      "step": 21535
    },
    {
      "epoch": 3.5138662316476346,
      "grad_norm": 0.13141930103302002,
      "learning_rate": 4.0977901551323414e-05,
      "loss": 0.0406,
      "num_input_tokens_seen": 46495488,
      "step": 21540
    },
    {
      "epoch": 3.5146818923327894,
      "grad_norm": 3.897994041442871,
      "learning_rate": 4.097242639026914e-05,
      "loss": 0.2088,
      "num_input_tokens_seen": 46506944,
      "step": 21545
    },
    {
      "epoch": 3.5154975530179446,
      "grad_norm": 2.7126681804656982,
      "learning_rate": 4.0966949934408946e-05,
      "loss": 0.1499,
      "num_input_tokens_seen": 46517216,
      "step": 21550
    },
    {
      "epoch": 3.5163132137030995,
      "grad_norm": 4.191082954406738,
      "learning_rate": 4.0961472184186766e-05,
      "loss": 0.1604,
      "num_input_tokens_seen": 46528864,
      "step": 21555
    },
    {
      "epoch": 3.5171288743882547,
      "grad_norm": 0.14017659425735474,
      "learning_rate": 4.0955993140046665e-05,
      "loss": 0.1651,
      "num_input_tokens_seen": 46539072,
      "step": 21560
    },
    {
      "epoch": 3.5179445350734095,
      "grad_norm": 0.11831638962030411,
      "learning_rate": 4.095051280243281e-05,
      "loss": 0.0834,
      "num_input_tokens_seen": 46550016,
      "step": 21565
    },
    {
      "epoch": 3.5187601957585644,
      "grad_norm": 1.5843397378921509,
      "learning_rate": 4.0945031171789435e-05,
      "loss": 0.0641,
      "num_input_tokens_seen": 46559968,
      "step": 21570
    },
    {
      "epoch": 3.519575856443719,
      "grad_norm": 0.5666118860244751,
      "learning_rate": 4.0939548248560946e-05,
      "loss": 0.0396,
      "num_input_tokens_seen": 46571456,
      "step": 21575
    },
    {
      "epoch": 3.5203915171288744,
      "grad_norm": 0.08470553159713745,
      "learning_rate": 4.093406403319179e-05,
      "loss": 0.0682,
      "num_input_tokens_seen": 46583104,
      "step": 21580
    },
    {
      "epoch": 3.5212071778140293,
      "grad_norm": 0.33066871762275696,
      "learning_rate": 4.0928578526126566e-05,
      "loss": 0.1707,
      "num_input_tokens_seen": 46592384,
      "step": 21585
    },
    {
      "epoch": 3.5220228384991845,
      "grad_norm": 0.3490210175514221,
      "learning_rate": 4.092309172780994e-05,
      "loss": 0.0124,
      "num_input_tokens_seen": 46603488,
      "step": 21590
    },
    {
      "epoch": 3.5228384991843393,
      "grad_norm": 0.061410918831825256,
      "learning_rate": 4.09176036386867e-05,
      "loss": 0.2111,
      "num_input_tokens_seen": 46614528,
      "step": 21595
    },
    {
      "epoch": 3.523654159869494,
      "grad_norm": 3.608489513397217,
      "learning_rate": 4.091211425920175e-05,
      "loss": 0.0739,
      "num_input_tokens_seen": 46625920,
      "step": 21600
    },
    {
      "epoch": 3.5244698205546494,
      "grad_norm": 0.05101187154650688,
      "learning_rate": 4.090662358980009e-05,
      "loss": 0.0256,
      "num_input_tokens_seen": 46636288,
      "step": 21605
    },
    {
      "epoch": 3.5252854812398042,
      "grad_norm": 5.769841194152832,
      "learning_rate": 4.0901131630926794e-05,
      "loss": 0.2354,
      "num_input_tokens_seen": 46645536,
      "step": 21610
    },
    {
      "epoch": 3.5261011419249595,
      "grad_norm": 0.13154403865337372,
      "learning_rate": 4.089563838302709e-05,
      "loss": 0.0662,
      "num_input_tokens_seen": 46656224,
      "step": 21615
    },
    {
      "epoch": 3.5269168026101143,
      "grad_norm": 0.40239590406417847,
      "learning_rate": 4.089014384654629e-05,
      "loss": 0.0607,
      "num_input_tokens_seen": 46665920,
      "step": 21620
    },
    {
      "epoch": 3.527732463295269,
      "grad_norm": 1.9793695211410522,
      "learning_rate": 4.088464802192981e-05,
      "loss": 0.0313,
      "num_input_tokens_seen": 46676864,
      "step": 21625
    },
    {
      "epoch": 3.528548123980424,
      "grad_norm": 0.11798906326293945,
      "learning_rate": 4.0879150909623156e-05,
      "loss": 0.1958,
      "num_input_tokens_seen": 46688192,
      "step": 21630
    },
    {
      "epoch": 3.529363784665579,
      "grad_norm": 0.08413796126842499,
      "learning_rate": 4.0873652510071955e-05,
      "loss": 0.1456,
      "num_input_tokens_seen": 46698848,
      "step": 21635
    },
    {
      "epoch": 3.530179445350734,
      "grad_norm": 7.618236064910889,
      "learning_rate": 4.086815282372195e-05,
      "loss": 0.0819,
      "num_input_tokens_seen": 46709920,
      "step": 21640
    },
    {
      "epoch": 3.5309951060358893,
      "grad_norm": 4.7247114181518555,
      "learning_rate": 4.086265185101895e-05,
      "loss": 0.1243,
      "num_input_tokens_seen": 46721152,
      "step": 21645
    },
    {
      "epoch": 3.531810766721044,
      "grad_norm": 0.1500084549188614,
      "learning_rate": 4.0857149592408914e-05,
      "loss": 0.096,
      "num_input_tokens_seen": 46732320,
      "step": 21650
    },
    {
      "epoch": 3.532626427406199,
      "grad_norm": 0.060236118733882904,
      "learning_rate": 4.085164604833788e-05,
      "loss": 0.1269,
      "num_input_tokens_seen": 46744032,
      "step": 21655
    },
    {
      "epoch": 3.5334420880913537,
      "grad_norm": 0.4299047589302063,
      "learning_rate": 4.084614121925198e-05,
      "loss": 0.127,
      "num_input_tokens_seen": 46755584,
      "step": 21660
    },
    {
      "epoch": 3.534257748776509,
      "grad_norm": 3.1743924617767334,
      "learning_rate": 4.084063510559746e-05,
      "loss": 0.1116,
      "num_input_tokens_seen": 46766368,
      "step": 21665
    },
    {
      "epoch": 3.535073409461664,
      "grad_norm": 0.2604175806045532,
      "learning_rate": 4.0835127707820696e-05,
      "loss": 0.1737,
      "num_input_tokens_seen": 46777920,
      "step": 21670
    },
    {
      "epoch": 3.535889070146819,
      "grad_norm": 0.0569436214864254,
      "learning_rate": 4.0829619026368134e-05,
      "loss": 0.1881,
      "num_input_tokens_seen": 46789312,
      "step": 21675
    },
    {
      "epoch": 3.536704730831974,
      "grad_norm": 0.10592181980609894,
      "learning_rate": 4.0824109061686325e-05,
      "loss": 0.0114,
      "num_input_tokens_seen": 46800736,
      "step": 21680
    },
    {
      "epoch": 3.5375203915171287,
      "grad_norm": 7.927008152008057,
      "learning_rate": 4.081859781422195e-05,
      "loss": 0.2195,
      "num_input_tokens_seen": 46812096,
      "step": 21685
    },
    {
      "epoch": 3.538336052202284,
      "grad_norm": 0.1780281662940979,
      "learning_rate": 4.0813085284421774e-05,
      "loss": 0.1556,
      "num_input_tokens_seen": 46822624,
      "step": 21690
    },
    {
      "epoch": 3.539151712887439,
      "grad_norm": 10.250486373901367,
      "learning_rate": 4.080757147273267e-05,
      "loss": 0.1929,
      "num_input_tokens_seen": 46833760,
      "step": 21695
    },
    {
      "epoch": 3.539967373572594,
      "grad_norm": 3.7362842559814453,
      "learning_rate": 4.080205637960162e-05,
      "loss": 0.2167,
      "num_input_tokens_seen": 46845312,
      "step": 21700
    },
    {
      "epoch": 3.540783034257749,
      "grad_norm": 1.0604051351547241,
      "learning_rate": 4.07965400054757e-05,
      "loss": 0.0727,
      "num_input_tokens_seen": 46855904,
      "step": 21705
    },
    {
      "epoch": 3.5415986949429037,
      "grad_norm": 0.11310368776321411,
      "learning_rate": 4.0791022350802086e-05,
      "loss": 0.1851,
      "num_input_tokens_seen": 46865760,
      "step": 21710
    },
    {
      "epoch": 3.5424143556280585,
      "grad_norm": 0.5822985172271729,
      "learning_rate": 4.078550341602809e-05,
      "loss": 0.0106,
      "num_input_tokens_seen": 46877888,
      "step": 21715
    },
    {
      "epoch": 3.5432300163132138,
      "grad_norm": 0.23588088154792786,
      "learning_rate": 4.077998320160109e-05,
      "loss": 0.1103,
      "num_input_tokens_seen": 46889312,
      "step": 21720
    },
    {
      "epoch": 3.5440456769983686,
      "grad_norm": 2.986375093460083,
      "learning_rate": 4.077446170796858e-05,
      "loss": 0.1407,
      "num_input_tokens_seen": 46899392,
      "step": 21725
    },
    {
      "epoch": 3.544861337683524,
      "grad_norm": 2.9055681228637695,
      "learning_rate": 4.076893893557816e-05,
      "loss": 0.074,
      "num_input_tokens_seen": 46910784,
      "step": 21730
    },
    {
      "epoch": 3.5456769983686787,
      "grad_norm": 2.1018295288085938,
      "learning_rate": 4.076341488487755e-05,
      "loss": 0.2735,
      "num_input_tokens_seen": 46921920,
      "step": 21735
    },
    {
      "epoch": 3.5464926590538335,
      "grad_norm": 0.5600054860115051,
      "learning_rate": 4.0757889556314545e-05,
      "loss": 0.0757,
      "num_input_tokens_seen": 46931776,
      "step": 21740
    },
    {
      "epoch": 3.5473083197389887,
      "grad_norm": 1.423844575881958,
      "learning_rate": 4.0752362950337054e-05,
      "loss": 0.2423,
      "num_input_tokens_seen": 46942848,
      "step": 21745
    },
    {
      "epoch": 3.5481239804241436,
      "grad_norm": 0.08828958868980408,
      "learning_rate": 4.0746835067393096e-05,
      "loss": 0.0082,
      "num_input_tokens_seen": 46953408,
      "step": 21750
    },
    {
      "epoch": 3.5489396411092984,
      "grad_norm": 1.2527973651885986,
      "learning_rate": 4.074130590793079e-05,
      "loss": 0.0254,
      "num_input_tokens_seen": 46964416,
      "step": 21755
    },
    {
      "epoch": 3.5497553017944536,
      "grad_norm": 0.1187516376376152,
      "learning_rate": 4.073577547239836e-05,
      "loss": 0.022,
      "num_input_tokens_seen": 46974304,
      "step": 21760
    },
    {
      "epoch": 3.5505709624796085,
      "grad_norm": 0.17528195679187775,
      "learning_rate": 4.073024376124412e-05,
      "loss": 0.042,
      "num_input_tokens_seen": 46984256,
      "step": 21765
    },
    {
      "epoch": 3.5513866231647633,
      "grad_norm": 0.8355662822723389,
      "learning_rate": 4.072471077491651e-05,
      "loss": 0.2306,
      "num_input_tokens_seen": 46995424,
      "step": 21770
    },
    {
      "epoch": 3.5522022838499185,
      "grad_norm": 0.0951867401599884,
      "learning_rate": 4.071917651386406e-05,
      "loss": 0.3113,
      "num_input_tokens_seen": 47006880,
      "step": 21775
    },
    {
      "epoch": 3.5530179445350734,
      "grad_norm": 2.912970542907715,
      "learning_rate": 4.071364097853541e-05,
      "loss": 0.1603,
      "num_input_tokens_seen": 47016864,
      "step": 21780
    },
    {
      "epoch": 3.5538336052202286,
      "grad_norm": 3.3453893661499023,
      "learning_rate": 4.070810416937927e-05,
      "loss": 0.2121,
      "num_input_tokens_seen": 47029248,
      "step": 21785
    },
    {
      "epoch": 3.5546492659053834,
      "grad_norm": 0.22751952707767487,
      "learning_rate": 4.070256608684452e-05,
      "loss": 0.025,
      "num_input_tokens_seen": 47040480,
      "step": 21790
    },
    {
      "epoch": 3.5554649265905383,
      "grad_norm": 0.13204766809940338,
      "learning_rate": 4.069702673138009e-05,
      "loss": 0.0208,
      "num_input_tokens_seen": 47051872,
      "step": 21795
    },
    {
      "epoch": 3.556280587275693,
      "grad_norm": 0.09759233891963959,
      "learning_rate": 4.0691486103435025e-05,
      "loss": 0.0098,
      "num_input_tokens_seen": 47063008,
      "step": 21800
    },
    {
      "epoch": 3.5570962479608483,
      "grad_norm": 0.07521530240774155,
      "learning_rate": 4.0685944203458476e-05,
      "loss": 0.0319,
      "num_input_tokens_seen": 47073792,
      "step": 21805
    },
    {
      "epoch": 3.557911908646003,
      "grad_norm": 0.11585956811904907,
      "learning_rate": 4.06804010318997e-05,
      "loss": 0.0384,
      "num_input_tokens_seen": 47084832,
      "step": 21810
    },
    {
      "epoch": 3.5587275693311584,
      "grad_norm": 0.2817971408367157,
      "learning_rate": 4.0674856589208063e-05,
      "loss": 0.0096,
      "num_input_tokens_seen": 47096032,
      "step": 21815
    },
    {
      "epoch": 3.5595432300163132,
      "grad_norm": 0.0972599983215332,
      "learning_rate": 4.066931087583301e-05,
      "loss": 0.0799,
      "num_input_tokens_seen": 47106720,
      "step": 21820
    },
    {
      "epoch": 3.560358890701468,
      "grad_norm": 0.07077526301145554,
      "learning_rate": 4.0663763892224114e-05,
      "loss": 0.1032,
      "num_input_tokens_seen": 47118592,
      "step": 21825
    },
    {
      "epoch": 3.5611745513866233,
      "grad_norm": 4.707823753356934,
      "learning_rate": 4.065821563883104e-05,
      "loss": 0.1321,
      "num_input_tokens_seen": 47129792,
      "step": 21830
    },
    {
      "epoch": 3.561990212071778,
      "grad_norm": 0.07917211204767227,
      "learning_rate": 4.0652666116103556e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 47140224,
      "step": 21835
    },
    {
      "epoch": 3.5628058727569334,
      "grad_norm": 1.6662336587905884,
      "learning_rate": 4.064711532449153e-05,
      "loss": 0.0629,
      "num_input_tokens_seen": 47151904,
      "step": 21840
    },
    {
      "epoch": 3.563621533442088,
      "grad_norm": 2.8767783641815186,
      "learning_rate": 4.0641563264444946e-05,
      "loss": 0.0193,
      "num_input_tokens_seen": 47161920,
      "step": 21845
    },
    {
      "epoch": 3.564437194127243,
      "grad_norm": 0.08682331442832947,
      "learning_rate": 4.063600993641389e-05,
      "loss": 0.166,
      "num_input_tokens_seen": 47172128,
      "step": 21850
    },
    {
      "epoch": 3.565252854812398,
      "grad_norm": 9.139212608337402,
      "learning_rate": 4.0630455340848525e-05,
      "loss": 0.1559,
      "num_input_tokens_seen": 47182784,
      "step": 21855
    },
    {
      "epoch": 3.566068515497553,
      "grad_norm": 0.028107237070798874,
      "learning_rate": 4.062489947819914e-05,
      "loss": 0.0055,
      "num_input_tokens_seen": 47194336,
      "step": 21860
    },
    {
      "epoch": 3.566884176182708,
      "grad_norm": 3.5822339057922363,
      "learning_rate": 4.061934234891612e-05,
      "loss": 0.1516,
      "num_input_tokens_seen": 47204832,
      "step": 21865
    },
    {
      "epoch": 3.567699836867863,
      "grad_norm": 0.12957142293453217,
      "learning_rate": 4.0613783953449966e-05,
      "loss": 0.1401,
      "num_input_tokens_seen": 47215648,
      "step": 21870
    },
    {
      "epoch": 3.568515497553018,
      "grad_norm": 0.035445135086774826,
      "learning_rate": 4.0608224292251264e-05,
      "loss": 0.1497,
      "num_input_tokens_seen": 47226080,
      "step": 21875
    },
    {
      "epoch": 3.569331158238173,
      "grad_norm": 2.962271213531494,
      "learning_rate": 4.0602663365770696e-05,
      "loss": 0.1262,
      "num_input_tokens_seen": 47238112,
      "step": 21880
    },
    {
      "epoch": 3.5701468189233276,
      "grad_norm": 3.3195712566375732,
      "learning_rate": 4.0597101174459074e-05,
      "loss": 0.1438,
      "num_input_tokens_seen": 47249216,
      "step": 21885
    },
    {
      "epoch": 3.570962479608483,
      "grad_norm": 0.04456210136413574,
      "learning_rate": 4.0591537718767284e-05,
      "loss": 0.0894,
      "num_input_tokens_seen": 47259296,
      "step": 21890
    },
    {
      "epoch": 3.5717781402936377,
      "grad_norm": 0.8475660681724548,
      "learning_rate": 4.058597299914634e-05,
      "loss": 0.1626,
      "num_input_tokens_seen": 47270336,
      "step": 21895
    },
    {
      "epoch": 3.572593800978793,
      "grad_norm": 4.9658894538879395,
      "learning_rate": 4.0580407016047345e-05,
      "loss": 0.0797,
      "num_input_tokens_seen": 47280736,
      "step": 21900
    },
    {
      "epoch": 3.573409461663948,
      "grad_norm": 2.843942880630493,
      "learning_rate": 4.0574839769921504e-05,
      "loss": 0.1598,
      "num_input_tokens_seen": 47291328,
      "step": 21905
    },
    {
      "epoch": 3.5742251223491026,
      "grad_norm": 0.07666657119989395,
      "learning_rate": 4.056927126122012e-05,
      "loss": 0.115,
      "num_input_tokens_seen": 47302368,
      "step": 21910
    },
    {
      "epoch": 3.575040783034258,
      "grad_norm": 4.6631622314453125,
      "learning_rate": 4.056370149039461e-05,
      "loss": 0.032,
      "num_input_tokens_seen": 47312768,
      "step": 21915
    },
    {
      "epoch": 3.5758564437194127,
      "grad_norm": 0.055636707693338394,
      "learning_rate": 4.05581304578965e-05,
      "loss": 0.0083,
      "num_input_tokens_seen": 47323520,
      "step": 21920
    },
    {
      "epoch": 3.576672104404568,
      "grad_norm": 10.303662300109863,
      "learning_rate": 4.055255816417738e-05,
      "loss": 0.2338,
      "num_input_tokens_seen": 47334688,
      "step": 21925
    },
    {
      "epoch": 3.5774877650897228,
      "grad_norm": 0.2885773777961731,
      "learning_rate": 4.054698460968899e-05,
      "loss": 0.1213,
      "num_input_tokens_seen": 47345184,
      "step": 21930
    },
    {
      "epoch": 3.5783034257748776,
      "grad_norm": 2.2620084285736084,
      "learning_rate": 4.054140979488314e-05,
      "loss": 0.0632,
      "num_input_tokens_seen": 47356608,
      "step": 21935
    },
    {
      "epoch": 3.5791190864600324,
      "grad_norm": 4.410715579986572,
      "learning_rate": 4.0535833720211755e-05,
      "loss": 0.4028,
      "num_input_tokens_seen": 47367520,
      "step": 21940
    },
    {
      "epoch": 3.5799347471451877,
      "grad_norm": 5.771556377410889,
      "learning_rate": 4.053025638612686e-05,
      "loss": 0.1804,
      "num_input_tokens_seen": 47376288,
      "step": 21945
    },
    {
      "epoch": 3.5807504078303425,
      "grad_norm": 0.24861368536949158,
      "learning_rate": 4.052467779308058e-05,
      "loss": 0.0099,
      "num_input_tokens_seen": 47386592,
      "step": 21950
    },
    {
      "epoch": 3.5815660685154977,
      "grad_norm": 2.6123125553131104,
      "learning_rate": 4.051909794152515e-05,
      "loss": 0.2404,
      "num_input_tokens_seen": 47397792,
      "step": 21955
    },
    {
      "epoch": 3.5823817292006526,
      "grad_norm": 0.07052227109670639,
      "learning_rate": 4.05135168319129e-05,
      "loss": 0.0078,
      "num_input_tokens_seen": 47409280,
      "step": 21960
    },
    {
      "epoch": 3.5831973898858074,
      "grad_norm": 0.0400143526494503,
      "learning_rate": 4.050793446469626e-05,
      "loss": 0.0118,
      "num_input_tokens_seen": 47419680,
      "step": 21965
    },
    {
      "epoch": 3.5840130505709626,
      "grad_norm": 6.331526756286621,
      "learning_rate": 4.0502350840327764e-05,
      "loss": 0.0409,
      "num_input_tokens_seen": 47431296,
      "step": 21970
    },
    {
      "epoch": 3.5848287112561175,
      "grad_norm": 5.153724193572998,
      "learning_rate": 4.0496765959260055e-05,
      "loss": 0.0997,
      "num_input_tokens_seen": 47441824,
      "step": 21975
    },
    {
      "epoch": 3.5856443719412723,
      "grad_norm": 4.120921611785889,
      "learning_rate": 4.049117982194586e-05,
      "loss": 0.0082,
      "num_input_tokens_seen": 47452192,
      "step": 21980
    },
    {
      "epoch": 3.5864600326264275,
      "grad_norm": 4.24963903427124,
      "learning_rate": 4.048559242883804e-05,
      "loss": 0.1077,
      "num_input_tokens_seen": 47461696,
      "step": 21985
    },
    {
      "epoch": 3.5872756933115824,
      "grad_norm": 0.09849881380796432,
      "learning_rate": 4.0480003780389507e-05,
      "loss": 0.1782,
      "num_input_tokens_seen": 47472960,
      "step": 21990
    },
    {
      "epoch": 3.588091353996737,
      "grad_norm": 0.10503087937831879,
      "learning_rate": 4.0474413877053335e-05,
      "loss": 0.2302,
      "num_input_tokens_seen": 47483712,
      "step": 21995
    },
    {
      "epoch": 3.5889070146818924,
      "grad_norm": 0.1379232257604599,
      "learning_rate": 4.0468822719282654e-05,
      "loss": 0.1462,
      "num_input_tokens_seen": 47493728,
      "step": 22000
    },
    {
      "epoch": 3.5897226753670473,
      "grad_norm": 0.04909015819430351,
      "learning_rate": 4.046323030753071e-05,
      "loss": 0.0048,
      "num_input_tokens_seen": 47504928,
      "step": 22005
    },
    {
      "epoch": 3.5905383360522025,
      "grad_norm": 0.12813489139080048,
      "learning_rate": 4.045763664225087e-05,
      "loss": 0.1032,
      "num_input_tokens_seen": 47514880,
      "step": 22010
    },
    {
      "epoch": 3.5913539967373573,
      "grad_norm": 0.569116473197937,
      "learning_rate": 4.045204172389656e-05,
      "loss": 0.0439,
      "num_input_tokens_seen": 47526688,
      "step": 22015
    },
    {
      "epoch": 3.592169657422512,
      "grad_norm": 0.07062584906816483,
      "learning_rate": 4.044644555292135e-05,
      "loss": 0.0286,
      "num_input_tokens_seen": 47536768,
      "step": 22020
    },
    {
      "epoch": 3.592985318107667,
      "grad_norm": 0.05885003134608269,
      "learning_rate": 4.04408481297789e-05,
      "loss": 0.1038,
      "num_input_tokens_seen": 47548768,
      "step": 22025
    },
    {
      "epoch": 3.5938009787928222,
      "grad_norm": 0.07328484207391739,
      "learning_rate": 4.043524945492294e-05,
      "loss": 0.1124,
      "num_input_tokens_seen": 47559968,
      "step": 22030
    },
    {
      "epoch": 3.594616639477977,
      "grad_norm": 3.6764700412750244,
      "learning_rate": 4.042964952880734e-05,
      "loss": 0.1308,
      "num_input_tokens_seen": 47571040,
      "step": 22035
    },
    {
      "epoch": 3.5954323001631323,
      "grad_norm": 2.515673875808716,
      "learning_rate": 4.042404835188607e-05,
      "loss": 0.0795,
      "num_input_tokens_seen": 47582816,
      "step": 22040
    },
    {
      "epoch": 3.596247960848287,
      "grad_norm": 0.10273520648479462,
      "learning_rate": 4.041844592461318e-05,
      "loss": 0.0245,
      "num_input_tokens_seen": 47594112,
      "step": 22045
    },
    {
      "epoch": 3.597063621533442,
      "grad_norm": 0.09178955107927322,
      "learning_rate": 4.0412842247442815e-05,
      "loss": 0.1805,
      "num_input_tokens_seen": 47604320,
      "step": 22050
    },
    {
      "epoch": 3.597879282218597,
      "grad_norm": 0.048253919929265976,
      "learning_rate": 4.040723732082927e-05,
      "loss": 0.1487,
      "num_input_tokens_seen": 47613472,
      "step": 22055
    },
    {
      "epoch": 3.598694942903752,
      "grad_norm": 12.505542755126953,
      "learning_rate": 4.040163114522689e-05,
      "loss": 0.265,
      "num_input_tokens_seen": 47624096,
      "step": 22060
    },
    {
      "epoch": 3.5995106035889073,
      "grad_norm": 7.198615550994873,
      "learning_rate": 4.039602372109014e-05,
      "loss": 0.1578,
      "num_input_tokens_seen": 47635456,
      "step": 22065
    },
    {
      "epoch": 3.600326264274062,
      "grad_norm": 4.6033124923706055,
      "learning_rate": 4.0390415048873584e-05,
      "loss": 0.1346,
      "num_input_tokens_seen": 47644448,
      "step": 22070
    },
    {
      "epoch": 3.601141924959217,
      "grad_norm": 4.066956043243408,
      "learning_rate": 4.03848051290319e-05,
      "loss": 0.1952,
      "num_input_tokens_seen": 47654144,
      "step": 22075
    },
    {
      "epoch": 3.6019575856443717,
      "grad_norm": 0.04538601636886597,
      "learning_rate": 4.037919396201985e-05,
      "loss": 0.0186,
      "num_input_tokens_seen": 47664672,
      "step": 22080
    },
    {
      "epoch": 3.602773246329527,
      "grad_norm": 4.017793655395508,
      "learning_rate": 4.0373581548292305e-05,
      "loss": 0.1193,
      "num_input_tokens_seen": 47676512,
      "step": 22085
    },
    {
      "epoch": 3.603588907014682,
      "grad_norm": 4.2030463218688965,
      "learning_rate": 4.036796788830423e-05,
      "loss": 0.1719,
      "num_input_tokens_seen": 47687168,
      "step": 22090
    },
    {
      "epoch": 3.604404567699837,
      "grad_norm": 7.6258745193481445,
      "learning_rate": 4.036235298251071e-05,
      "loss": 0.2498,
      "num_input_tokens_seen": 47698208,
      "step": 22095
    },
    {
      "epoch": 3.605220228384992,
      "grad_norm": 1.149688720703125,
      "learning_rate": 4.03567368313669e-05,
      "loss": 0.0456,
      "num_input_tokens_seen": 47708448,
      "step": 22100
    },
    {
      "epoch": 3.6060358890701467,
      "grad_norm": 0.1677428036928177,
      "learning_rate": 4.035111943532808e-05,
      "loss": 0.0714,
      "num_input_tokens_seen": 47718624,
      "step": 22105
    },
    {
      "epoch": 3.6068515497553015,
      "grad_norm": 3.900057792663574,
      "learning_rate": 4.034550079484964e-05,
      "loss": 0.33,
      "num_input_tokens_seen": 47729184,
      "step": 22110
    },
    {
      "epoch": 3.607667210440457,
      "grad_norm": 0.27360770106315613,
      "learning_rate": 4.033988091038704e-05,
      "loss": 0.0163,
      "num_input_tokens_seen": 47740992,
      "step": 22115
    },
    {
      "epoch": 3.6084828711256116,
      "grad_norm": 0.5041369795799255,
      "learning_rate": 4.0334259782395855e-05,
      "loss": 0.0946,
      "num_input_tokens_seen": 47751520,
      "step": 22120
    },
    {
      "epoch": 3.609298531810767,
      "grad_norm": 0.6643962860107422,
      "learning_rate": 4.032863741133177e-05,
      "loss": 0.1931,
      "num_input_tokens_seen": 47761920,
      "step": 22125
    },
    {
      "epoch": 3.6101141924959217,
      "grad_norm": 0.24680005013942719,
      "learning_rate": 4.0323013797650556e-05,
      "loss": 0.2495,
      "num_input_tokens_seen": 47773376,
      "step": 22130
    },
    {
      "epoch": 3.6109298531810765,
      "grad_norm": 0.7548781633377075,
      "learning_rate": 4.0317388941808096e-05,
      "loss": 0.0122,
      "num_input_tokens_seen": 47784064,
      "step": 22135
    },
    {
      "epoch": 3.6117455138662318,
      "grad_norm": 3.6045496463775635,
      "learning_rate": 4.0311762844260377e-05,
      "loss": 0.0533,
      "num_input_tokens_seen": 47794592,
      "step": 22140
    },
    {
      "epoch": 3.6125611745513866,
      "grad_norm": 2.84478497505188,
      "learning_rate": 4.030613550546347e-05,
      "loss": 0.0654,
      "num_input_tokens_seen": 47805504,
      "step": 22145
    },
    {
      "epoch": 3.613376835236542,
      "grad_norm": 0.22830621898174286,
      "learning_rate": 4.030050692587355e-05,
      "loss": 0.0894,
      "num_input_tokens_seen": 47816128,
      "step": 22150
    },
    {
      "epoch": 3.6141924959216967,
      "grad_norm": 2.5198605060577393,
      "learning_rate": 4.02948771059469e-05,
      "loss": 0.1818,
      "num_input_tokens_seen": 47827264,
      "step": 22155
    },
    {
      "epoch": 3.6150081566068515,
      "grad_norm": 0.3472936451435089,
      "learning_rate": 4.028924604613991e-05,
      "loss": 0.0564,
      "num_input_tokens_seen": 47838208,
      "step": 22160
    },
    {
      "epoch": 3.6158238172920063,
      "grad_norm": 6.473494052886963,
      "learning_rate": 4.028361374690906e-05,
      "loss": 0.3111,
      "num_input_tokens_seen": 47848192,
      "step": 22165
    },
    {
      "epoch": 3.6166394779771616,
      "grad_norm": 0.1116599515080452,
      "learning_rate": 4.027798020871093e-05,
      "loss": 0.0812,
      "num_input_tokens_seen": 47858688,
      "step": 22170
    },
    {
      "epoch": 3.6174551386623164,
      "grad_norm": 0.28514230251312256,
      "learning_rate": 4.027234543200221e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 47870048,
      "step": 22175
    },
    {
      "epoch": 3.6182707993474716,
      "grad_norm": 3.8633968830108643,
      "learning_rate": 4.026670941723968e-05,
      "loss": 0.2806,
      "num_input_tokens_seen": 47881984,
      "step": 22180
    },
    {
      "epoch": 3.6190864600326265,
      "grad_norm": 1.2139700651168823,
      "learning_rate": 4.026107216488022e-05,
      "loss": 0.0192,
      "num_input_tokens_seen": 47893376,
      "step": 22185
    },
    {
      "epoch": 3.6199021207177813,
      "grad_norm": 0.20846295356750488,
      "learning_rate": 4.0255433675380803e-05,
      "loss": 0.1225,
      "num_input_tokens_seen": 47903680,
      "step": 22190
    },
    {
      "epoch": 3.6207177814029365,
      "grad_norm": 4.988443374633789,
      "learning_rate": 4.024979394919855e-05,
      "loss": 0.2028,
      "num_input_tokens_seen": 47914144,
      "step": 22195
    },
    {
      "epoch": 3.6215334420880914,
      "grad_norm": 0.16319090127944946,
      "learning_rate": 4.0244152986790604e-05,
      "loss": 0.0056,
      "num_input_tokens_seen": 47924224,
      "step": 22200
    },
    {
      "epoch": 3.622349102773246,
      "grad_norm": 0.17418339848518372,
      "learning_rate": 4.0238510788614276e-05,
      "loss": 0.0155,
      "num_input_tokens_seen": 47935264,
      "step": 22205
    },
    {
      "epoch": 3.6231647634584014,
      "grad_norm": 9.328039169311523,
      "learning_rate": 4.0232867355126934e-05,
      "loss": 0.1032,
      "num_input_tokens_seen": 47946432,
      "step": 22210
    },
    {
      "epoch": 3.6239804241435563,
      "grad_norm": 0.0981740653514862,
      "learning_rate": 4.0227222686786084e-05,
      "loss": 0.1034,
      "num_input_tokens_seen": 47956608,
      "step": 22215
    },
    {
      "epoch": 3.624796084828711,
      "grad_norm": 0.2725767195224762,
      "learning_rate": 4.02215767840493e-05,
      "loss": 0.2099,
      "num_input_tokens_seen": 47967552,
      "step": 22220
    },
    {
      "epoch": 3.6256117455138663,
      "grad_norm": 2.8347389698028564,
      "learning_rate": 4.021592964737427e-05,
      "loss": 0.1766,
      "num_input_tokens_seen": 47977888,
      "step": 22225
    },
    {
      "epoch": 3.626427406199021,
      "grad_norm": 0.5867570638656616,
      "learning_rate": 4.021028127721878e-05,
      "loss": 0.0112,
      "num_input_tokens_seen": 47988576,
      "step": 22230
    },
    {
      "epoch": 3.6272430668841764,
      "grad_norm": 6.528834819793701,
      "learning_rate": 4.020463167404071e-05,
      "loss": 0.3195,
      "num_input_tokens_seen": 47999776,
      "step": 22235
    },
    {
      "epoch": 3.6280587275693312,
      "grad_norm": 0.07322845607995987,
      "learning_rate": 4.019898083829804e-05,
      "loss": 0.16,
      "num_input_tokens_seen": 48010688,
      "step": 22240
    },
    {
      "epoch": 3.628874388254486,
      "grad_norm": 0.10080315917730331,
      "learning_rate": 4.019332877044888e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 48021152,
      "step": 22245
    },
    {
      "epoch": 3.629690048939641,
      "grad_norm": 3.8889048099517822,
      "learning_rate": 4.018767547095139e-05,
      "loss": 0.1934,
      "num_input_tokens_seen": 48030176,
      "step": 22250
    },
    {
      "epoch": 3.630505709624796,
      "grad_norm": 4.339661121368408,
      "learning_rate": 4.018202094026386e-05,
      "loss": 0.161,
      "num_input_tokens_seen": 48039232,
      "step": 22255
    },
    {
      "epoch": 3.631321370309951,
      "grad_norm": 0.1206539049744606,
      "learning_rate": 4.01763651788447e-05,
      "loss": 0.1115,
      "num_input_tokens_seen": 48049984,
      "step": 22260
    },
    {
      "epoch": 3.632137030995106,
      "grad_norm": 0.10961958020925522,
      "learning_rate": 4.017070818715235e-05,
      "loss": 0.0891,
      "num_input_tokens_seen": 48061088,
      "step": 22265
    },
    {
      "epoch": 3.632952691680261,
      "grad_norm": 2.189587354660034,
      "learning_rate": 4.016504996564544e-05,
      "loss": 0.116,
      "num_input_tokens_seen": 48072704,
      "step": 22270
    },
    {
      "epoch": 3.633768352365416,
      "grad_norm": 0.13103020191192627,
      "learning_rate": 4.015939051478262e-05,
      "loss": 0.0934,
      "num_input_tokens_seen": 48082848,
      "step": 22275
    },
    {
      "epoch": 3.634584013050571,
      "grad_norm": 1.8878916501998901,
      "learning_rate": 4.0153729835022685e-05,
      "loss": 0.1781,
      "num_input_tokens_seen": 48094464,
      "step": 22280
    },
    {
      "epoch": 3.635399673735726,
      "grad_norm": 2.901374101638794,
      "learning_rate": 4.014806792682453e-05,
      "loss": 0.1184,
      "num_input_tokens_seen": 48105088,
      "step": 22285
    },
    {
      "epoch": 3.636215334420881,
      "grad_norm": 3.0098648071289062,
      "learning_rate": 4.0142404790647124e-05,
      "loss": 0.1215,
      "num_input_tokens_seen": 48114432,
      "step": 22290
    },
    {
      "epoch": 3.637030995106036,
      "grad_norm": 6.524054527282715,
      "learning_rate": 4.0136740426949546e-05,
      "loss": 0.0509,
      "num_input_tokens_seen": 48124288,
      "step": 22295
    },
    {
      "epoch": 3.637846655791191,
      "grad_norm": 4.471999645233154,
      "learning_rate": 4.0131074836191e-05,
      "loss": 0.2926,
      "num_input_tokens_seen": 48134144,
      "step": 22300
    },
    {
      "epoch": 3.6386623164763456,
      "grad_norm": 0.30536070466041565,
      "learning_rate": 4.0125408018830744e-05,
      "loss": 0.0636,
      "num_input_tokens_seen": 48144416,
      "step": 22305
    },
    {
      "epoch": 3.639477977161501,
      "grad_norm": 0.401883602142334,
      "learning_rate": 4.011973997532818e-05,
      "loss": 0.0148,
      "num_input_tokens_seen": 48155584,
      "step": 22310
    },
    {
      "epoch": 3.6402936378466557,
      "grad_norm": 7.902167320251465,
      "learning_rate": 4.011407070614276e-05,
      "loss": 0.2724,
      "num_input_tokens_seen": 48167232,
      "step": 22315
    },
    {
      "epoch": 3.641109298531811,
      "grad_norm": 2.3360376358032227,
      "learning_rate": 4.010840021173409e-05,
      "loss": 0.148,
      "num_input_tokens_seen": 48178208,
      "step": 22320
    },
    {
      "epoch": 3.641924959216966,
      "grad_norm": 0.12659363448619843,
      "learning_rate": 4.010272849256184e-05,
      "loss": 0.0125,
      "num_input_tokens_seen": 48188896,
      "step": 22325
    },
    {
      "epoch": 3.6427406199021206,
      "grad_norm": 3.3409955501556396,
      "learning_rate": 4.0097055549085784e-05,
      "loss": 0.3578,
      "num_input_tokens_seen": 48200000,
      "step": 22330
    },
    {
      "epoch": 3.6435562805872754,
      "grad_norm": 2.3330581188201904,
      "learning_rate": 4.009138138176581e-05,
      "loss": 0.1743,
      "num_input_tokens_seen": 48211200,
      "step": 22335
    },
    {
      "epoch": 3.6443719412724307,
      "grad_norm": 4.646770477294922,
      "learning_rate": 4.008570599106188e-05,
      "loss": 0.2123,
      "num_input_tokens_seen": 48221504,
      "step": 22340
    },
    {
      "epoch": 3.6451876019575855,
      "grad_norm": 1.5298365354537964,
      "learning_rate": 4.008002937743409e-05,
      "loss": 0.0963,
      "num_input_tokens_seen": 48230976,
      "step": 22345
    },
    {
      "epoch": 3.6460032626427408,
      "grad_norm": 5.603265762329102,
      "learning_rate": 4.0074351541342595e-05,
      "loss": 0.0707,
      "num_input_tokens_seen": 48242368,
      "step": 22350
    },
    {
      "epoch": 3.6468189233278956,
      "grad_norm": 0.32391926646232605,
      "learning_rate": 4.006867248324767e-05,
      "loss": 0.0429,
      "num_input_tokens_seen": 48253696,
      "step": 22355
    },
    {
      "epoch": 3.6476345840130504,
      "grad_norm": 0.1386079043149948,
      "learning_rate": 4.006299220360971e-05,
      "loss": 0.1191,
      "num_input_tokens_seen": 48264768,
      "step": 22360
    },
    {
      "epoch": 3.6484502446982057,
      "grad_norm": 0.21881791949272156,
      "learning_rate": 4.0057310702889164e-05,
      "loss": 0.1884,
      "num_input_tokens_seen": 48275520,
      "step": 22365
    },
    {
      "epoch": 3.6492659053833605,
      "grad_norm": 0.826500654220581,
      "learning_rate": 4.005162798154661e-05,
      "loss": 0.0849,
      "num_input_tokens_seen": 48284960,
      "step": 22370
    },
    {
      "epoch": 3.6500815660685157,
      "grad_norm": 0.33012261986732483,
      "learning_rate": 4.004594404004273e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 48295840,
      "step": 22375
    },
    {
      "epoch": 3.6508972267536706,
      "grad_norm": 2.405694007873535,
      "learning_rate": 4.0040258878838284e-05,
      "loss": 0.0482,
      "num_input_tokens_seen": 48306880,
      "step": 22380
    },
    {
      "epoch": 3.6517128874388254,
      "grad_norm": 9.564769744873047,
      "learning_rate": 4.003457249839413e-05,
      "loss": 0.1081,
      "num_input_tokens_seen": 48318336,
      "step": 22385
    },
    {
      "epoch": 3.65252854812398,
      "grad_norm": 6.086526393890381,
      "learning_rate": 4.002888489917126e-05,
      "loss": 0.2148,
      "num_input_tokens_seen": 48329152,
      "step": 22390
    },
    {
      "epoch": 3.6533442088091355,
      "grad_norm": 1.8801467418670654,
      "learning_rate": 4.002319608163071e-05,
      "loss": 0.0184,
      "num_input_tokens_seen": 48339136,
      "step": 22395
    },
    {
      "epoch": 3.6541598694942903,
      "grad_norm": 0.10271084308624268,
      "learning_rate": 4.0017506046233664e-05,
      "loss": 0.0898,
      "num_input_tokens_seen": 48350208,
      "step": 22400
    },
    {
      "epoch": 3.6549755301794455,
      "grad_norm": 0.14470317959785461,
      "learning_rate": 4.001181479344138e-05,
      "loss": 0.3134,
      "num_input_tokens_seen": 48362016,
      "step": 22405
    },
    {
      "epoch": 3.6557911908646004,
      "grad_norm": 0.4115452766418457,
      "learning_rate": 4.000612232371522e-05,
      "loss": 0.1287,
      "num_input_tokens_seen": 48371424,
      "step": 22410
    },
    {
      "epoch": 3.656606851549755,
      "grad_norm": 0.11699778586626053,
      "learning_rate": 4.000042863751664e-05,
      "loss": 0.2665,
      "num_input_tokens_seen": 48383328,
      "step": 22415
    },
    {
      "epoch": 3.6574225122349104,
      "grad_norm": 0.1325097382068634,
      "learning_rate": 3.999473373530721e-05,
      "loss": 0.0129,
      "num_input_tokens_seen": 48394304,
      "step": 22420
    },
    {
      "epoch": 3.6582381729200653,
      "grad_norm": 0.4000729024410248,
      "learning_rate": 3.9989037617548575e-05,
      "loss": 0.0617,
      "num_input_tokens_seen": 48405376,
      "step": 22425
    },
    {
      "epoch": 3.65905383360522,
      "grad_norm": 2.915386915206909,
      "learning_rate": 3.9983340284702495e-05,
      "loss": 0.0663,
      "num_input_tokens_seen": 48416352,
      "step": 22430
    },
    {
      "epoch": 3.6598694942903753,
      "grad_norm": 1.3424534797668457,
      "learning_rate": 3.9977641737230833e-05,
      "loss": 0.2314,
      "num_input_tokens_seen": 48427488,
      "step": 22435
    },
    {
      "epoch": 3.66068515497553,
      "grad_norm": 0.15219749510288239,
      "learning_rate": 3.9971941975595535e-05,
      "loss": 0.0193,
      "num_input_tokens_seen": 48439520,
      "step": 22440
    },
    {
      "epoch": 3.661500815660685,
      "grad_norm": 0.43363040685653687,
      "learning_rate": 3.996624100025865e-05,
      "loss": 0.0262,
      "num_input_tokens_seen": 48449984,
      "step": 22445
    },
    {
      "epoch": 3.6623164763458402,
      "grad_norm": 4.819627285003662,
      "learning_rate": 3.9960538811682334e-05,
      "loss": 0.246,
      "num_input_tokens_seen": 48460672,
      "step": 22450
    },
    {
      "epoch": 3.663132137030995,
      "grad_norm": 5.342092514038086,
      "learning_rate": 3.9954835410328836e-05,
      "loss": 0.0607,
      "num_input_tokens_seen": 48471328,
      "step": 22455
    },
    {
      "epoch": 3.6639477977161503,
      "grad_norm": 0.058290671557188034,
      "learning_rate": 3.9949130796660496e-05,
      "loss": 0.0115,
      "num_input_tokens_seen": 48481664,
      "step": 22460
    },
    {
      "epoch": 3.664763458401305,
      "grad_norm": 0.13387128710746765,
      "learning_rate": 3.994342497113977e-05,
      "loss": 0.0682,
      "num_input_tokens_seen": 48492896,
      "step": 22465
    },
    {
      "epoch": 3.66557911908646,
      "grad_norm": 16.06231689453125,
      "learning_rate": 3.993771793422918e-05,
      "loss": 0.0848,
      "num_input_tokens_seen": 48503648,
      "step": 22470
    },
    {
      "epoch": 3.6663947797716148,
      "grad_norm": 1.8369101285934448,
      "learning_rate": 3.993200968639139e-05,
      "loss": 0.0649,
      "num_input_tokens_seen": 48514368,
      "step": 22475
    },
    {
      "epoch": 3.66721044045677,
      "grad_norm": 0.08171354234218597,
      "learning_rate": 3.9926300228089124e-05,
      "loss": 0.21,
      "num_input_tokens_seen": 48525504,
      "step": 22480
    },
    {
      "epoch": 3.668026101141925,
      "grad_norm": 0.33790531754493713,
      "learning_rate": 3.992058955978523e-05,
      "loss": 0.0831,
      "num_input_tokens_seen": 48534880,
      "step": 22485
    },
    {
      "epoch": 3.66884176182708,
      "grad_norm": 3.9071567058563232,
      "learning_rate": 3.9914877681942645e-05,
      "loss": 0.1641,
      "num_input_tokens_seen": 48546656,
      "step": 22490
    },
    {
      "epoch": 3.669657422512235,
      "grad_norm": 0.05480445548892021,
      "learning_rate": 3.99091645950244e-05,
      "loss": 0.0346,
      "num_input_tokens_seen": 48557952,
      "step": 22495
    },
    {
      "epoch": 3.6704730831973897,
      "grad_norm": 1.3236916065216064,
      "learning_rate": 3.990345029949361e-05,
      "loss": 0.0472,
      "num_input_tokens_seen": 48568832,
      "step": 22500
    },
    {
      "epoch": 3.671288743882545,
      "grad_norm": 8.661673545837402,
      "learning_rate": 3.9897734795813524e-05,
      "loss": 0.1859,
      "num_input_tokens_seen": 48578784,
      "step": 22505
    },
    {
      "epoch": 3.6721044045677,
      "grad_norm": 0.064410001039505,
      "learning_rate": 3.989201808444747e-05,
      "loss": 0.0965,
      "num_input_tokens_seen": 48590016,
      "step": 22510
    },
    {
      "epoch": 3.672920065252855,
      "grad_norm": 3.6234500408172607,
      "learning_rate": 3.988630016585887e-05,
      "loss": 0.0224,
      "num_input_tokens_seen": 48600608,
      "step": 22515
    },
    {
      "epoch": 3.67373572593801,
      "grad_norm": 0.05369186773896217,
      "learning_rate": 3.988058104051124e-05,
      "loss": 0.0091,
      "num_input_tokens_seen": 48610912,
      "step": 22520
    },
    {
      "epoch": 3.6745513866231647,
      "grad_norm": 0.2313581258058548,
      "learning_rate": 3.987486070886821e-05,
      "loss": 0.0782,
      "num_input_tokens_seen": 48622272,
      "step": 22525
    },
    {
      "epoch": 3.6753670473083195,
      "grad_norm": 0.08435726910829544,
      "learning_rate": 3.98691391713935e-05,
      "loss": 0.0125,
      "num_input_tokens_seen": 48633376,
      "step": 22530
    },
    {
      "epoch": 3.676182707993475,
      "grad_norm": 0.24193263053894043,
      "learning_rate": 3.986341642855092e-05,
      "loss": 0.1817,
      "num_input_tokens_seen": 48644608,
      "step": 22535
    },
    {
      "epoch": 3.6769983686786296,
      "grad_norm": 0.3538898825645447,
      "learning_rate": 3.985769248080439e-05,
      "loss": 0.0667,
      "num_input_tokens_seen": 48656448,
      "step": 22540
    },
    {
      "epoch": 3.677814029363785,
      "grad_norm": 2.382725954055786,
      "learning_rate": 3.9851967328617925e-05,
      "loss": 0.2129,
      "num_input_tokens_seen": 48667936,
      "step": 22545
    },
    {
      "epoch": 3.6786296900489397,
      "grad_norm": 0.2911075949668884,
      "learning_rate": 3.984624097245562e-05,
      "loss": 0.252,
      "num_input_tokens_seen": 48678240,
      "step": 22550
    },
    {
      "epoch": 3.6794453507340945,
      "grad_norm": 4.779908657073975,
      "learning_rate": 3.98405134127817e-05,
      "loss": 0.1643,
      "num_input_tokens_seen": 48687296,
      "step": 22555
    },
    {
      "epoch": 3.6802610114192493,
      "grad_norm": 0.07430237531661987,
      "learning_rate": 3.983478465006045e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 48698720,
      "step": 22560
    },
    {
      "epoch": 3.6810766721044046,
      "grad_norm": 3.6542553901672363,
      "learning_rate": 3.9829054684756304e-05,
      "loss": 0.1002,
      "num_input_tokens_seen": 48709696,
      "step": 22565
    },
    {
      "epoch": 3.6818923327895594,
      "grad_norm": 5.811183452606201,
      "learning_rate": 3.982332351733373e-05,
      "loss": 0.0659,
      "num_input_tokens_seen": 48720032,
      "step": 22570
    },
    {
      "epoch": 3.6827079934747147,
      "grad_norm": 3.3798558712005615,
      "learning_rate": 3.981759114825735e-05,
      "loss": 0.1565,
      "num_input_tokens_seen": 48729664,
      "step": 22575
    },
    {
      "epoch": 3.6835236541598695,
      "grad_norm": 0.1618494689464569,
      "learning_rate": 3.981185757799184e-05,
      "loss": 0.0121,
      "num_input_tokens_seen": 48742336,
      "step": 22580
    },
    {
      "epoch": 3.6843393148450243,
      "grad_norm": 0.2879089415073395,
      "learning_rate": 3.9806122807002e-05,
      "loss": 0.0097,
      "num_input_tokens_seen": 48754048,
      "step": 22585
    },
    {
      "epoch": 3.6851549755301796,
      "grad_norm": 0.32481124997138977,
      "learning_rate": 3.9800386835752726e-05,
      "loss": 0.0145,
      "num_input_tokens_seen": 48765344,
      "step": 22590
    },
    {
      "epoch": 3.6859706362153344,
      "grad_norm": 0.20661230385303497,
      "learning_rate": 3.979464966470899e-05,
      "loss": 0.131,
      "num_input_tokens_seen": 48777152,
      "step": 22595
    },
    {
      "epoch": 3.6867862969004896,
      "grad_norm": 0.6438668966293335,
      "learning_rate": 3.978891129433588e-05,
      "loss": 0.166,
      "num_input_tokens_seen": 48786880,
      "step": 22600
    },
    {
      "epoch": 3.6876019575856445,
      "grad_norm": 0.03673599287867546,
      "learning_rate": 3.97831717250986e-05,
      "loss": 0.0039,
      "num_input_tokens_seen": 48796256,
      "step": 22605
    },
    {
      "epoch": 3.6884176182707993,
      "grad_norm": 0.1118549332022667,
      "learning_rate": 3.97774309574624e-05,
      "loss": 0.1462,
      "num_input_tokens_seen": 48808096,
      "step": 22610
    },
    {
      "epoch": 3.689233278955954,
      "grad_norm": 2.930147647857666,
      "learning_rate": 3.977168899189267e-05,
      "loss": 0.1644,
      "num_input_tokens_seen": 48818848,
      "step": 22615
    },
    {
      "epoch": 3.6900489396411094,
      "grad_norm": 0.1776353120803833,
      "learning_rate": 3.9765945828854876e-05,
      "loss": 0.0123,
      "num_input_tokens_seen": 48831296,
      "step": 22620
    },
    {
      "epoch": 3.690864600326264,
      "grad_norm": 0.09796354919672012,
      "learning_rate": 3.97602014688146e-05,
      "loss": 0.1369,
      "num_input_tokens_seen": 48841536,
      "step": 22625
    },
    {
      "epoch": 3.6916802610114194,
      "grad_norm": 8.46008014678955,
      "learning_rate": 3.9754455912237486e-05,
      "loss": 0.329,
      "num_input_tokens_seen": 48853152,
      "step": 22630
    },
    {
      "epoch": 3.6924959216965743,
      "grad_norm": 0.1644945740699768,
      "learning_rate": 3.974870915958932e-05,
      "loss": 0.1864,
      "num_input_tokens_seen": 48862688,
      "step": 22635
    },
    {
      "epoch": 3.693311582381729,
      "grad_norm": 0.29013460874557495,
      "learning_rate": 3.974296121133596e-05,
      "loss": 0.0922,
      "num_input_tokens_seen": 48874688,
      "step": 22640
    },
    {
      "epoch": 3.6941272430668843,
      "grad_norm": 0.32632938027381897,
      "learning_rate": 3.9737212067943354e-05,
      "loss": 0.089,
      "num_input_tokens_seen": 48885856,
      "step": 22645
    },
    {
      "epoch": 3.694942903752039,
      "grad_norm": 0.1759522408246994,
      "learning_rate": 3.973146172987756e-05,
      "loss": 0.2164,
      "num_input_tokens_seen": 48896096,
      "step": 22650
    },
    {
      "epoch": 3.695758564437194,
      "grad_norm": 3.1296448707580566,
      "learning_rate": 3.9725710197604735e-05,
      "loss": 0.2253,
      "num_input_tokens_seen": 48907328,
      "step": 22655
    },
    {
      "epoch": 3.6965742251223492,
      "grad_norm": 6.081822872161865,
      "learning_rate": 3.971995747159113e-05,
      "loss": 0.052,
      "num_input_tokens_seen": 48918304,
      "step": 22660
    },
    {
      "epoch": 3.697389885807504,
      "grad_norm": 0.17746877670288086,
      "learning_rate": 3.971420355230308e-05,
      "loss": 0.0185,
      "num_input_tokens_seen": 48929440,
      "step": 22665
    },
    {
      "epoch": 3.698205546492659,
      "grad_norm": 0.16611170768737793,
      "learning_rate": 3.9708448440207026e-05,
      "loss": 0.0139,
      "num_input_tokens_seen": 48940256,
      "step": 22670
    },
    {
      "epoch": 3.699021207177814,
      "grad_norm": 0.11874650418758392,
      "learning_rate": 3.970269213576951e-05,
      "loss": 0.0557,
      "num_input_tokens_seen": 48949856,
      "step": 22675
    },
    {
      "epoch": 3.699836867862969,
      "grad_norm": 0.20596261322498322,
      "learning_rate": 3.969693463945717e-05,
      "loss": 0.0107,
      "num_input_tokens_seen": 48960768,
      "step": 22680
    },
    {
      "epoch": 3.700652528548124,
      "grad_norm": 8.530813217163086,
      "learning_rate": 3.9691175951736745e-05,
      "loss": 0.1962,
      "num_input_tokens_seen": 48971264,
      "step": 22685
    },
    {
      "epoch": 3.701468189233279,
      "grad_norm": 0.068712517619133,
      "learning_rate": 3.9685416073075045e-05,
      "loss": 0.1436,
      "num_input_tokens_seen": 48981632,
      "step": 22690
    },
    {
      "epoch": 3.702283849918434,
      "grad_norm": 1.9284168481826782,
      "learning_rate": 3.967965500393901e-05,
      "loss": 0.2985,
      "num_input_tokens_seen": 48993024,
      "step": 22695
    },
    {
      "epoch": 3.7030995106035887,
      "grad_norm": 0.6498374938964844,
      "learning_rate": 3.9673892744795655e-05,
      "loss": 0.2043,
      "num_input_tokens_seen": 49004064,
      "step": 22700
    },
    {
      "epoch": 3.703915171288744,
      "grad_norm": 2.0610876083374023,
      "learning_rate": 3.96681292961121e-05,
      "loss": 0.1693,
      "num_input_tokens_seen": 49016032,
      "step": 22705
    },
    {
      "epoch": 3.7047308319738987,
      "grad_norm": 2.4175643920898438,
      "learning_rate": 3.9662364658355555e-05,
      "loss": 0.0362,
      "num_input_tokens_seen": 49026336,
      "step": 22710
    },
    {
      "epoch": 3.705546492659054,
      "grad_norm": 5.219014644622803,
      "learning_rate": 3.965659883199334e-05,
      "loss": 0.105,
      "num_input_tokens_seen": 49037312,
      "step": 22715
    },
    {
      "epoch": 3.706362153344209,
      "grad_norm": 0.7204117774963379,
      "learning_rate": 3.9650831817492864e-05,
      "loss": 0.11,
      "num_input_tokens_seen": 49048480,
      "step": 22720
    },
    {
      "epoch": 3.7071778140293636,
      "grad_norm": 2.4042067527770996,
      "learning_rate": 3.964506361532161e-05,
      "loss": 0.4708,
      "num_input_tokens_seen": 49059840,
      "step": 22725
    },
    {
      "epoch": 3.707993474714519,
      "grad_norm": 0.5451491475105286,
      "learning_rate": 3.96392942259472e-05,
      "loss": 0.1008,
      "num_input_tokens_seen": 49070720,
      "step": 22730
    },
    {
      "epoch": 3.7088091353996737,
      "grad_norm": 0.12057767063379288,
      "learning_rate": 3.963352364983731e-05,
      "loss": 0.0816,
      "num_input_tokens_seen": 49081408,
      "step": 22735
    },
    {
      "epoch": 3.709624796084829,
      "grad_norm": 3.3763442039489746,
      "learning_rate": 3.962775188745975e-05,
      "loss": 0.0747,
      "num_input_tokens_seen": 49092416,
      "step": 22740
    },
    {
      "epoch": 3.710440456769984,
      "grad_norm": 1.9849827289581299,
      "learning_rate": 3.9621978939282405e-05,
      "loss": 0.1388,
      "num_input_tokens_seen": 49102464,
      "step": 22745
    },
    {
      "epoch": 3.7112561174551386,
      "grad_norm": 0.1819087713956833,
      "learning_rate": 3.961620480577325e-05,
      "loss": 0.166,
      "num_input_tokens_seen": 49113920,
      "step": 22750
    },
    {
      "epoch": 3.7120717781402934,
      "grad_norm": 0.08712363243103027,
      "learning_rate": 3.961042948740038e-05,
      "loss": 0.1381,
      "num_input_tokens_seen": 49124096,
      "step": 22755
    },
    {
      "epoch": 3.7128874388254487,
      "grad_norm": 0.08532737195491791,
      "learning_rate": 3.960465298463195e-05,
      "loss": 0.082,
      "num_input_tokens_seen": 49134048,
      "step": 22760
    },
    {
      "epoch": 3.7137030995106035,
      "grad_norm": 0.8071739673614502,
      "learning_rate": 3.959887529793625e-05,
      "loss": 0.0217,
      "num_input_tokens_seen": 49144800,
      "step": 22765
    },
    {
      "epoch": 3.7145187601957588,
      "grad_norm": 6.525815486907959,
      "learning_rate": 3.9593096427781665e-05,
      "loss": 0.1802,
      "num_input_tokens_seen": 49156128,
      "step": 22770
    },
    {
      "epoch": 3.7153344208809136,
      "grad_norm": 0.13750576972961426,
      "learning_rate": 3.958731637463662e-05,
      "loss": 0.1035,
      "num_input_tokens_seen": 49167168,
      "step": 22775
    },
    {
      "epoch": 3.7161500815660684,
      "grad_norm": 0.1299505978822708,
      "learning_rate": 3.958153513896969e-05,
      "loss": 0.0479,
      "num_input_tokens_seen": 49177248,
      "step": 22780
    },
    {
      "epoch": 3.7169657422512232,
      "grad_norm": 0.6548534631729126,
      "learning_rate": 3.957575272124954e-05,
      "loss": 0.0128,
      "num_input_tokens_seen": 49189568,
      "step": 22785
    },
    {
      "epoch": 3.7177814029363785,
      "grad_norm": 6.015018939971924,
      "learning_rate": 3.9569969121944925e-05,
      "loss": 0.1047,
      "num_input_tokens_seen": 49198912,
      "step": 22790
    },
    {
      "epoch": 3.7185970636215333,
      "grad_norm": 0.1631668359041214,
      "learning_rate": 3.956418434152467e-05,
      "loss": 0.2272,
      "num_input_tokens_seen": 49210112,
      "step": 22795
    },
    {
      "epoch": 3.7194127243066886,
      "grad_norm": 2.453158378601074,
      "learning_rate": 3.955839838045775e-05,
      "loss": 0.2212,
      "num_input_tokens_seen": 49221344,
      "step": 22800
    },
    {
      "epoch": 3.7202283849918434,
      "grad_norm": 0.03234280273318291,
      "learning_rate": 3.9552611239213185e-05,
      "loss": 0.0937,
      "num_input_tokens_seen": 49232064,
      "step": 22805
    },
    {
      "epoch": 3.721044045676998,
      "grad_norm": 3.2036361694335938,
      "learning_rate": 3.954682291826011e-05,
      "loss": 0.2731,
      "num_input_tokens_seen": 49242144,
      "step": 22810
    },
    {
      "epoch": 3.7218597063621535,
      "grad_norm": 0.5641581416130066,
      "learning_rate": 3.9541033418067765e-05,
      "loss": 0.0319,
      "num_input_tokens_seen": 49253248,
      "step": 22815
    },
    {
      "epoch": 3.7226753670473083,
      "grad_norm": 0.41947269439697266,
      "learning_rate": 3.953524273910546e-05,
      "loss": 0.1217,
      "num_input_tokens_seen": 49264384,
      "step": 22820
    },
    {
      "epoch": 3.7234910277324635,
      "grad_norm": 0.08137278258800507,
      "learning_rate": 3.952945088184264e-05,
      "loss": 0.0355,
      "num_input_tokens_seen": 49276416,
      "step": 22825
    },
    {
      "epoch": 3.7243066884176184,
      "grad_norm": 0.2503032386302948,
      "learning_rate": 3.952365784674881e-05,
      "loss": 0.0399,
      "num_input_tokens_seen": 49288800,
      "step": 22830
    },
    {
      "epoch": 3.725122349102773,
      "grad_norm": 5.778484344482422,
      "learning_rate": 3.951786363429357e-05,
      "loss": 0.1051,
      "num_input_tokens_seen": 49299328,
      "step": 22835
    },
    {
      "epoch": 3.725938009787928,
      "grad_norm": 2.861217737197876,
      "learning_rate": 3.951206824494665e-05,
      "loss": 0.1333,
      "num_input_tokens_seen": 49311360,
      "step": 22840
    },
    {
      "epoch": 3.7267536704730833,
      "grad_norm": 0.1345771700143814,
      "learning_rate": 3.950627167917784e-05,
      "loss": 0.2041,
      "num_input_tokens_seen": 49321632,
      "step": 22845
    },
    {
      "epoch": 3.727569331158238,
      "grad_norm": 5.471856117248535,
      "learning_rate": 3.950047393745705e-05,
      "loss": 0.0848,
      "num_input_tokens_seen": 49332256,
      "step": 22850
    },
    {
      "epoch": 3.7283849918433933,
      "grad_norm": 1.5649278163909912,
      "learning_rate": 3.949467502025426e-05,
      "loss": 0.1056,
      "num_input_tokens_seen": 49343360,
      "step": 22855
    },
    {
      "epoch": 3.729200652528548,
      "grad_norm": 0.0935976654291153,
      "learning_rate": 3.948887492803957e-05,
      "loss": 0.1103,
      "num_input_tokens_seen": 49354112,
      "step": 22860
    },
    {
      "epoch": 3.730016313213703,
      "grad_norm": 0.08797755092382431,
      "learning_rate": 3.948307366128316e-05,
      "loss": 0.1159,
      "num_input_tokens_seen": 49365120,
      "step": 22865
    },
    {
      "epoch": 3.7308319738988582,
      "grad_norm": 3.224078416824341,
      "learning_rate": 3.9477271220455323e-05,
      "loss": 0.3085,
      "num_input_tokens_seen": 49377600,
      "step": 22870
    },
    {
      "epoch": 3.731647634584013,
      "grad_norm": 0.08593336492776871,
      "learning_rate": 3.947146760602642e-05,
      "loss": 0.1162,
      "num_input_tokens_seen": 49387648,
      "step": 22875
    },
    {
      "epoch": 3.732463295269168,
      "grad_norm": 6.358371734619141,
      "learning_rate": 3.946566281846692e-05,
      "loss": 0.2713,
      "num_input_tokens_seen": 49398368,
      "step": 22880
    },
    {
      "epoch": 3.733278955954323,
      "grad_norm": 0.22296041250228882,
      "learning_rate": 3.9459856858247404e-05,
      "loss": 0.1504,
      "num_input_tokens_seen": 49408416,
      "step": 22885
    },
    {
      "epoch": 3.734094616639478,
      "grad_norm": 0.2840973138809204,
      "learning_rate": 3.945404972583851e-05,
      "loss": 0.239,
      "num_input_tokens_seen": 49417888,
      "step": 22890
    },
    {
      "epoch": 3.7349102773246328,
      "grad_norm": 0.3990428149700165,
      "learning_rate": 3.9448241421711004e-05,
      "loss": 0.2198,
      "num_input_tokens_seen": 49427296,
      "step": 22895
    },
    {
      "epoch": 3.735725938009788,
      "grad_norm": 0.13959169387817383,
      "learning_rate": 3.9442431946335755e-05,
      "loss": 0.209,
      "num_input_tokens_seen": 49437984,
      "step": 22900
    },
    {
      "epoch": 3.736541598694943,
      "grad_norm": 2.983022451400757,
      "learning_rate": 3.943662130018368e-05,
      "loss": 0.1907,
      "num_input_tokens_seen": 49448288,
      "step": 22905
    },
    {
      "epoch": 3.737357259380098,
      "grad_norm": 0.11356587707996368,
      "learning_rate": 3.943080948372583e-05,
      "loss": 0.1462,
      "num_input_tokens_seen": 49458624,
      "step": 22910
    },
    {
      "epoch": 3.738172920065253,
      "grad_norm": 3.975989818572998,
      "learning_rate": 3.942499649743335e-05,
      "loss": 0.0523,
      "num_input_tokens_seen": 49468032,
      "step": 22915
    },
    {
      "epoch": 3.7389885807504077,
      "grad_norm": 1.244232416152954,
      "learning_rate": 3.941918234177746e-05,
      "loss": 0.0235,
      "num_input_tokens_seen": 49479744,
      "step": 22920
    },
    {
      "epoch": 3.7398042414355626,
      "grad_norm": 7.66337251663208,
      "learning_rate": 3.941336701722949e-05,
      "loss": 0.1321,
      "num_input_tokens_seen": 49490688,
      "step": 22925
    },
    {
      "epoch": 3.740619902120718,
      "grad_norm": 1.0893155336380005,
      "learning_rate": 3.940755052426085e-05,
      "loss": 0.0255,
      "num_input_tokens_seen": 49499936,
      "step": 22930
    },
    {
      "epoch": 3.7414355628058726,
      "grad_norm": 0.09969876706600189,
      "learning_rate": 3.940173286334307e-05,
      "loss": 0.03,
      "num_input_tokens_seen": 49510080,
      "step": 22935
    },
    {
      "epoch": 3.742251223491028,
      "grad_norm": 0.971980094909668,
      "learning_rate": 3.9395914034947744e-05,
      "loss": 0.0408,
      "num_input_tokens_seen": 49521056,
      "step": 22940
    },
    {
      "epoch": 3.7430668841761827,
      "grad_norm": 0.25531283020973206,
      "learning_rate": 3.939009403954659e-05,
      "loss": 0.0144,
      "num_input_tokens_seen": 49531712,
      "step": 22945
    },
    {
      "epoch": 3.7438825448613375,
      "grad_norm": 1.407782793045044,
      "learning_rate": 3.9384272877611384e-05,
      "loss": 0.0522,
      "num_input_tokens_seen": 49542912,
      "step": 22950
    },
    {
      "epoch": 3.744698205546493,
      "grad_norm": 0.13444636762142181,
      "learning_rate": 3.9378450549614044e-05,
      "loss": 0.0092,
      "num_input_tokens_seen": 49553888,
      "step": 22955
    },
    {
      "epoch": 3.7455138662316476,
      "grad_norm": 1.0330255031585693,
      "learning_rate": 3.9372627056026544e-05,
      "loss": 0.0525,
      "num_input_tokens_seen": 49565376,
      "step": 22960
    },
    {
      "epoch": 3.746329526916803,
      "grad_norm": 0.7507807612419128,
      "learning_rate": 3.9366802397320966e-05,
      "loss": 0.0254,
      "num_input_tokens_seen": 49577184,
      "step": 22965
    },
    {
      "epoch": 3.7471451876019577,
      "grad_norm": 1.9532883167266846,
      "learning_rate": 3.9360976573969494e-05,
      "loss": 0.0124,
      "num_input_tokens_seen": 49588608,
      "step": 22970
    },
    {
      "epoch": 3.7479608482871125,
      "grad_norm": 0.11609118431806564,
      "learning_rate": 3.935514958644439e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 49599840,
      "step": 22975
    },
    {
      "epoch": 3.7487765089722673,
      "grad_norm": 0.15790703892707825,
      "learning_rate": 3.934932143521803e-05,
      "loss": 0.2568,
      "num_input_tokens_seen": 49611040,
      "step": 22980
    },
    {
      "epoch": 3.7495921696574226,
      "grad_norm": 0.09029453992843628,
      "learning_rate": 3.934349212076286e-05,
      "loss": 0.018,
      "num_input_tokens_seen": 49621472,
      "step": 22985
    },
    {
      "epoch": 3.7504078303425774,
      "grad_norm": 0.054871074855327606,
      "learning_rate": 3.933766164355145e-05,
      "loss": 0.1665,
      "num_input_tokens_seen": 49632832,
      "step": 22990
    },
    {
      "epoch": 3.7512234910277327,
      "grad_norm": 6.67854642868042,
      "learning_rate": 3.9331830004056424e-05,
      "loss": 0.3048,
      "num_input_tokens_seen": 49643136,
      "step": 22995
    },
    {
      "epoch": 3.7520391517128875,
      "grad_norm": 2.5512938499450684,
      "learning_rate": 3.932599720275055e-05,
      "loss": 0.1603,
      "num_input_tokens_seen": 49653920,
      "step": 23000
    },
    {
      "epoch": 3.7528548123980423,
      "grad_norm": 0.0809488296508789,
      "learning_rate": 3.9320163240106656e-05,
      "loss": 0.0827,
      "num_input_tokens_seen": 49664992,
      "step": 23005
    },
    {
      "epoch": 3.753670473083197,
      "grad_norm": 0.07908891886472702,
      "learning_rate": 3.931432811659766e-05,
      "loss": 0.0055,
      "num_input_tokens_seen": 49676000,
      "step": 23010
    },
    {
      "epoch": 3.7544861337683524,
      "grad_norm": 0.10362471640110016,
      "learning_rate": 3.9308491832696596e-05,
      "loss": 0.0187,
      "num_input_tokens_seen": 49687712,
      "step": 23015
    },
    {
      "epoch": 3.755301794453507,
      "grad_norm": 0.40513619780540466,
      "learning_rate": 3.930265438887659e-05,
      "loss": 0.0395,
      "num_input_tokens_seen": 49698944,
      "step": 23020
    },
    {
      "epoch": 3.7561174551386625,
      "grad_norm": 0.07383420318365097,
      "learning_rate": 3.929681578561084e-05,
      "loss": 0.0855,
      "num_input_tokens_seen": 49710048,
      "step": 23025
    },
    {
      "epoch": 3.7569331158238173,
      "grad_norm": 1.2087088823318481,
      "learning_rate": 3.929097602337267e-05,
      "loss": 0.2128,
      "num_input_tokens_seen": 49720672,
      "step": 23030
    },
    {
      "epoch": 3.757748776508972,
      "grad_norm": 0.9839165210723877,
      "learning_rate": 3.9285135102635474e-05,
      "loss": 0.0764,
      "num_input_tokens_seen": 49731776,
      "step": 23035
    },
    {
      "epoch": 3.7585644371941274,
      "grad_norm": 0.16747084259986877,
      "learning_rate": 3.9279293023872745e-05,
      "loss": 0.1409,
      "num_input_tokens_seen": 49742336,
      "step": 23040
    },
    {
      "epoch": 3.759380097879282,
      "grad_norm": 3.6237456798553467,
      "learning_rate": 3.927344978755806e-05,
      "loss": 0.3473,
      "num_input_tokens_seen": 49753408,
      "step": 23045
    },
    {
      "epoch": 3.7601957585644374,
      "grad_norm": 0.05490643531084061,
      "learning_rate": 3.926760539416512e-05,
      "loss": 0.05,
      "num_input_tokens_seen": 49764192,
      "step": 23050
    },
    {
      "epoch": 3.7610114192495923,
      "grad_norm": 1.0296424627304077,
      "learning_rate": 3.926175984416769e-05,
      "loss": 0.0124,
      "num_input_tokens_seen": 49775840,
      "step": 23055
    },
    {
      "epoch": 3.761827079934747,
      "grad_norm": 0.07540196925401688,
      "learning_rate": 3.9255913138039645e-05,
      "loss": 0.0601,
      "num_input_tokens_seen": 49785408,
      "step": 23060
    },
    {
      "epoch": 3.762642740619902,
      "grad_norm": 4.27272891998291,
      "learning_rate": 3.925006527625494e-05,
      "loss": 0.2503,
      "num_input_tokens_seen": 49796192,
      "step": 23065
    },
    {
      "epoch": 3.763458401305057,
      "grad_norm": 0.09080624580383301,
      "learning_rate": 3.924421625928765e-05,
      "loss": 0.0178,
      "num_input_tokens_seen": 49807296,
      "step": 23070
    },
    {
      "epoch": 3.764274061990212,
      "grad_norm": 10.499856948852539,
      "learning_rate": 3.923836608761192e-05,
      "loss": 0.1899,
      "num_input_tokens_seen": 49818368,
      "step": 23075
    },
    {
      "epoch": 3.7650897226753672,
      "grad_norm": 0.10141909867525101,
      "learning_rate": 3.923251476170198e-05,
      "loss": 0.1773,
      "num_input_tokens_seen": 49830368,
      "step": 23080
    },
    {
      "epoch": 3.765905383360522,
      "grad_norm": 6.361057281494141,
      "learning_rate": 3.922666228203218e-05,
      "loss": 0.1997,
      "num_input_tokens_seen": 49840960,
      "step": 23085
    },
    {
      "epoch": 3.766721044045677,
      "grad_norm": 0.16668111085891724,
      "learning_rate": 3.9220808649076954e-05,
      "loss": 0.3243,
      "num_input_tokens_seen": 49851776,
      "step": 23090
    },
    {
      "epoch": 3.767536704730832,
      "grad_norm": 0.5158731937408447,
      "learning_rate": 3.921495386331082e-05,
      "loss": 0.209,
      "num_input_tokens_seen": 49862336,
      "step": 23095
    },
    {
      "epoch": 3.768352365415987,
      "grad_norm": 8.285778045654297,
      "learning_rate": 3.9209097925208405e-05,
      "loss": 0.2586,
      "num_input_tokens_seen": 49874208,
      "step": 23100
    },
    {
      "epoch": 3.7691680261011418,
      "grad_norm": 0.31928977370262146,
      "learning_rate": 3.920324083524441e-05,
      "loss": 0.0141,
      "num_input_tokens_seen": 49885440,
      "step": 23105
    },
    {
      "epoch": 3.769983686786297,
      "grad_norm": 1.4121078252792358,
      "learning_rate": 3.919738259389365e-05,
      "loss": 0.0671,
      "num_input_tokens_seen": 49896832,
      "step": 23110
    },
    {
      "epoch": 3.770799347471452,
      "grad_norm": 0.08855217695236206,
      "learning_rate": 3.919152320163101e-05,
      "loss": 0.1848,
      "num_input_tokens_seen": 49907008,
      "step": 23115
    },
    {
      "epoch": 3.7716150081566067,
      "grad_norm": 0.10175371170043945,
      "learning_rate": 3.91856626589315e-05,
      "loss": 0.0132,
      "num_input_tokens_seen": 49917600,
      "step": 23120
    },
    {
      "epoch": 3.772430668841762,
      "grad_norm": 5.034108638763428,
      "learning_rate": 3.91798009662702e-05,
      "loss": 0.0776,
      "num_input_tokens_seen": 49929280,
      "step": 23125
    },
    {
      "epoch": 3.7732463295269167,
      "grad_norm": 1.5025721788406372,
      "learning_rate": 3.917393812412229e-05,
      "loss": 0.0291,
      "num_input_tokens_seen": 49940384,
      "step": 23130
    },
    {
      "epoch": 3.774061990212072,
      "grad_norm": 0.04287495091557503,
      "learning_rate": 3.916807413296303e-05,
      "loss": 0.2155,
      "num_input_tokens_seen": 49949696,
      "step": 23135
    },
    {
      "epoch": 3.774877650897227,
      "grad_norm": 8.513350486755371,
      "learning_rate": 3.916220899326779e-05,
      "loss": 0.1193,
      "num_input_tokens_seen": 49959680,
      "step": 23140
    },
    {
      "epoch": 3.7756933115823816,
      "grad_norm": 2.701164722442627,
      "learning_rate": 3.915634270551204e-05,
      "loss": 0.2091,
      "num_input_tokens_seen": 49970592,
      "step": 23145
    },
    {
      "epoch": 3.7765089722675365,
      "grad_norm": 3.0898141860961914,
      "learning_rate": 3.915047527017132e-05,
      "loss": 0.1448,
      "num_input_tokens_seen": 49981440,
      "step": 23150
    },
    {
      "epoch": 3.7773246329526917,
      "grad_norm": 0.4086742103099823,
      "learning_rate": 3.914460668772127e-05,
      "loss": 0.1051,
      "num_input_tokens_seen": 49993024,
      "step": 23155
    },
    {
      "epoch": 3.7781402936378465,
      "grad_norm": 7.631880283355713,
      "learning_rate": 3.913873695863763e-05,
      "loss": 0.1822,
      "num_input_tokens_seen": 50004704,
      "step": 23160
    },
    {
      "epoch": 3.778955954323002,
      "grad_norm": 0.4384767413139343,
      "learning_rate": 3.913286608339625e-05,
      "loss": 0.1034,
      "num_input_tokens_seen": 50015424,
      "step": 23165
    },
    {
      "epoch": 3.7797716150081566,
      "grad_norm": 4.022773742675781,
      "learning_rate": 3.9126994062473013e-05,
      "loss": 0.0819,
      "num_input_tokens_seen": 50026368,
      "step": 23170
    },
    {
      "epoch": 3.7805872756933114,
      "grad_norm": 1.4691871404647827,
      "learning_rate": 3.912112089634397e-05,
      "loss": 0.137,
      "num_input_tokens_seen": 50036512,
      "step": 23175
    },
    {
      "epoch": 3.7814029363784667,
      "grad_norm": 0.08211508393287659,
      "learning_rate": 3.911524658548522e-05,
      "loss": 0.0938,
      "num_input_tokens_seen": 50047776,
      "step": 23180
    },
    {
      "epoch": 3.7822185970636215,
      "grad_norm": 12.340832710266113,
      "learning_rate": 3.9109371130372956e-05,
      "loss": 0.2162,
      "num_input_tokens_seen": 50058048,
      "step": 23185
    },
    {
      "epoch": 3.7830342577487768,
      "grad_norm": 0.12297343462705612,
      "learning_rate": 3.910349453148348e-05,
      "loss": 0.0299,
      "num_input_tokens_seen": 50068992,
      "step": 23190
    },
    {
      "epoch": 3.7838499184339316,
      "grad_norm": 0.13493342697620392,
      "learning_rate": 3.909761678929318e-05,
      "loss": 0.0647,
      "num_input_tokens_seen": 50079776,
      "step": 23195
    },
    {
      "epoch": 3.7846655791190864,
      "grad_norm": 0.07279930263757706,
      "learning_rate": 3.909173790427852e-05,
      "loss": 0.0746,
      "num_input_tokens_seen": 50090592,
      "step": 23200
    },
    {
      "epoch": 3.7854812398042412,
      "grad_norm": 0.10993704199790955,
      "learning_rate": 3.90858578769161e-05,
      "loss": 0.1579,
      "num_input_tokens_seen": 50102240,
      "step": 23205
    },
    {
      "epoch": 3.7862969004893965,
      "grad_norm": 2.9637222290039062,
      "learning_rate": 3.907997670768256e-05,
      "loss": 0.1213,
      "num_input_tokens_seen": 50113472,
      "step": 23210
    },
    {
      "epoch": 3.7871125611745513,
      "grad_norm": 0.1488426774740219,
      "learning_rate": 3.907409439705467e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 50124224,
      "step": 23215
    },
    {
      "epoch": 3.7879282218597066,
      "grad_norm": 0.2778640389442444,
      "learning_rate": 3.9068210945509276e-05,
      "loss": 0.0135,
      "num_input_tokens_seen": 50135264,
      "step": 23220
    },
    {
      "epoch": 3.7887438825448614,
      "grad_norm": 3.5414040088653564,
      "learning_rate": 3.906232635352333e-05,
      "loss": 0.2533,
      "num_input_tokens_seen": 50145824,
      "step": 23225
    },
    {
      "epoch": 3.789559543230016,
      "grad_norm": 0.025893602520227432,
      "learning_rate": 3.9056440621573855e-05,
      "loss": 0.164,
      "num_input_tokens_seen": 50156864,
      "step": 23230
    },
    {
      "epoch": 3.790375203915171,
      "grad_norm": 0.4216342568397522,
      "learning_rate": 3.9050553750137975e-05,
      "loss": 0.1048,
      "num_input_tokens_seen": 50167872,
      "step": 23235
    },
    {
      "epoch": 3.7911908646003263,
      "grad_norm": 0.3165377974510193,
      "learning_rate": 3.904466573969292e-05,
      "loss": 0.0694,
      "num_input_tokens_seen": 50178880,
      "step": 23240
    },
    {
      "epoch": 3.792006525285481,
      "grad_norm": 2.1399779319763184,
      "learning_rate": 3.9038776590716e-05,
      "loss": 0.3976,
      "num_input_tokens_seen": 50189280,
      "step": 23245
    },
    {
      "epoch": 3.7928221859706364,
      "grad_norm": 0.06540483236312866,
      "learning_rate": 3.903288630368461e-05,
      "loss": 0.0216,
      "num_input_tokens_seen": 50200736,
      "step": 23250
    },
    {
      "epoch": 3.793637846655791,
      "grad_norm": 0.05924204736948013,
      "learning_rate": 3.902699487907626e-05,
      "loss": 0.0134,
      "num_input_tokens_seen": 50210208,
      "step": 23255
    },
    {
      "epoch": 3.794453507340946,
      "grad_norm": 0.39854496717453003,
      "learning_rate": 3.902110231736853e-05,
      "loss": 0.0584,
      "num_input_tokens_seen": 50220960,
      "step": 23260
    },
    {
      "epoch": 3.7952691680261013,
      "grad_norm": 0.11542457342147827,
      "learning_rate": 3.901520861903911e-05,
      "loss": 0.1137,
      "num_input_tokens_seen": 50232384,
      "step": 23265
    },
    {
      "epoch": 3.796084828711256,
      "grad_norm": 4.767996311187744,
      "learning_rate": 3.900931378456576e-05,
      "loss": 0.0676,
      "num_input_tokens_seen": 50243328,
      "step": 23270
    },
    {
      "epoch": 3.7969004893964113,
      "grad_norm": 0.15116938948631287,
      "learning_rate": 3.9003417814426346e-05,
      "loss": 0.1516,
      "num_input_tokens_seen": 50254080,
      "step": 23275
    },
    {
      "epoch": 3.797716150081566,
      "grad_norm": 0.06791775673627853,
      "learning_rate": 3.8997520709098845e-05,
      "loss": 0.0222,
      "num_input_tokens_seen": 50264384,
      "step": 23280
    },
    {
      "epoch": 3.798531810766721,
      "grad_norm": 7.044241428375244,
      "learning_rate": 3.899162246906129e-05,
      "loss": 0.1518,
      "num_input_tokens_seen": 50275232,
      "step": 23285
    },
    {
      "epoch": 3.799347471451876,
      "grad_norm": 3.8409976959228516,
      "learning_rate": 3.8985723094791814e-05,
      "loss": 0.1798,
      "num_input_tokens_seen": 50286048,
      "step": 23290
    },
    {
      "epoch": 3.800163132137031,
      "grad_norm": 0.14364583790302277,
      "learning_rate": 3.897982258676867e-05,
      "loss": 0.0591,
      "num_input_tokens_seen": 50296192,
      "step": 23295
    },
    {
      "epoch": 3.800978792822186,
      "grad_norm": 8.493257522583008,
      "learning_rate": 3.8973920945470174e-05,
      "loss": 0.204,
      "num_input_tokens_seen": 50305888,
      "step": 23300
    },
    {
      "epoch": 3.801794453507341,
      "grad_norm": 2.7930526733398438,
      "learning_rate": 3.896801817137474e-05,
      "loss": 0.0949,
      "num_input_tokens_seen": 50316640,
      "step": 23305
    },
    {
      "epoch": 3.802610114192496,
      "grad_norm": 0.30696114897727966,
      "learning_rate": 3.8962114264960894e-05,
      "loss": 0.1659,
      "num_input_tokens_seen": 50328832,
      "step": 23310
    },
    {
      "epoch": 3.8034257748776508,
      "grad_norm": 0.2347618192434311,
      "learning_rate": 3.8956209226707206e-05,
      "loss": 0.255,
      "num_input_tokens_seen": 50339136,
      "step": 23315
    },
    {
      "epoch": 3.804241435562806,
      "grad_norm": 0.03590773046016693,
      "learning_rate": 3.8950303057092386e-05,
      "loss": 0.1674,
      "num_input_tokens_seen": 50350336,
      "step": 23320
    },
    {
      "epoch": 3.805057096247961,
      "grad_norm": 3.0185699462890625,
      "learning_rate": 3.8944395756595225e-05,
      "loss": 0.1713,
      "num_input_tokens_seen": 50361376,
      "step": 23325
    },
    {
      "epoch": 3.8058727569331157,
      "grad_norm": 0.06552945077419281,
      "learning_rate": 3.893848732569458e-05,
      "loss": 0.063,
      "num_input_tokens_seen": 50372064,
      "step": 23330
    },
    {
      "epoch": 3.806688417618271,
      "grad_norm": 0.7695654630661011,
      "learning_rate": 3.893257776486944e-05,
      "loss": 0.0524,
      "num_input_tokens_seen": 50382304,
      "step": 23335
    },
    {
      "epoch": 3.8075040783034257,
      "grad_norm": 6.420783519744873,
      "learning_rate": 3.8926667074598846e-05,
      "loss": 0.1637,
      "num_input_tokens_seen": 50392416,
      "step": 23340
    },
    {
      "epoch": 3.8083197389885806,
      "grad_norm": 3.166912317276001,
      "learning_rate": 3.892075525536196e-05,
      "loss": 0.2588,
      "num_input_tokens_seen": 50402816,
      "step": 23345
    },
    {
      "epoch": 3.809135399673736,
      "grad_norm": 0.7989125847816467,
      "learning_rate": 3.891484230763802e-05,
      "loss": 0.1568,
      "num_input_tokens_seen": 50412224,
      "step": 23350
    },
    {
      "epoch": 3.8099510603588906,
      "grad_norm": 3.0957610607147217,
      "learning_rate": 3.890892823190636e-05,
      "loss": 0.3021,
      "num_input_tokens_seen": 50423680,
      "step": 23355
    },
    {
      "epoch": 3.810766721044046,
      "grad_norm": 0.22804246842861176,
      "learning_rate": 3.890301302864641e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 50435072,
      "step": 23360
    },
    {
      "epoch": 3.8115823817292007,
      "grad_norm": 6.268031120300293,
      "learning_rate": 3.889709669833767e-05,
      "loss": 0.2217,
      "num_input_tokens_seen": 50446336,
      "step": 23365
    },
    {
      "epoch": 3.8123980424143555,
      "grad_norm": 3.2093803882598877,
      "learning_rate": 3.8891179241459766e-05,
      "loss": 0.103,
      "num_input_tokens_seen": 50456928,
      "step": 23370
    },
    {
      "epoch": 3.8132137030995104,
      "grad_norm": 0.15560902655124664,
      "learning_rate": 3.888526065849238e-05,
      "loss": 0.142,
      "num_input_tokens_seen": 50467488,
      "step": 23375
    },
    {
      "epoch": 3.8140293637846656,
      "grad_norm": 0.392856627702713,
      "learning_rate": 3.887934094991531e-05,
      "loss": 0.0582,
      "num_input_tokens_seen": 50479168,
      "step": 23380
    },
    {
      "epoch": 3.8148450244698204,
      "grad_norm": 0.37020111083984375,
      "learning_rate": 3.887342011620845e-05,
      "loss": 0.0112,
      "num_input_tokens_seen": 50490048,
      "step": 23385
    },
    {
      "epoch": 3.8156606851549757,
      "grad_norm": 4.762162685394287,
      "learning_rate": 3.886749815785176e-05,
      "loss": 0.1138,
      "num_input_tokens_seen": 50500192,
      "step": 23390
    },
    {
      "epoch": 3.8164763458401305,
      "grad_norm": 0.07374925166368484,
      "learning_rate": 3.8861575075325304e-05,
      "loss": 0.0963,
      "num_input_tokens_seen": 50510944,
      "step": 23395
    },
    {
      "epoch": 3.8172920065252853,
      "grad_norm": 0.11507974565029144,
      "learning_rate": 3.8855650869109246e-05,
      "loss": 0.1208,
      "num_input_tokens_seen": 50521344,
      "step": 23400
    },
    {
      "epoch": 3.8181076672104406,
      "grad_norm": 0.07705114781856537,
      "learning_rate": 3.884972553968382e-05,
      "loss": 0.0464,
      "num_input_tokens_seen": 50532064,
      "step": 23405
    },
    {
      "epoch": 3.8189233278955954,
      "grad_norm": 0.29260510206222534,
      "learning_rate": 3.884379908752936e-05,
      "loss": 0.0628,
      "num_input_tokens_seen": 50542816,
      "step": 23410
    },
    {
      "epoch": 3.8197389885807507,
      "grad_norm": 0.06639407575130463,
      "learning_rate": 3.883787151312632e-05,
      "loss": 0.0111,
      "num_input_tokens_seen": 50553088,
      "step": 23415
    },
    {
      "epoch": 3.8205546492659055,
      "grad_norm": 0.12404583394527435,
      "learning_rate": 3.88319428169552e-05,
      "loss": 0.0105,
      "num_input_tokens_seen": 50561792,
      "step": 23420
    },
    {
      "epoch": 3.8213703099510603,
      "grad_norm": 1.8111509084701538,
      "learning_rate": 3.882601299949661e-05,
      "loss": 0.1381,
      "num_input_tokens_seen": 50573280,
      "step": 23425
    },
    {
      "epoch": 3.822185970636215,
      "grad_norm": 0.09226392954587936,
      "learning_rate": 3.882008206123125e-05,
      "loss": 0.0091,
      "num_input_tokens_seen": 50584384,
      "step": 23430
    },
    {
      "epoch": 3.8230016313213704,
      "grad_norm": 0.04823372885584831,
      "learning_rate": 3.881415000263991e-05,
      "loss": 0.0361,
      "num_input_tokens_seen": 50595904,
      "step": 23435
    },
    {
      "epoch": 3.823817292006525,
      "grad_norm": 0.5151596665382385,
      "learning_rate": 3.8808216824203494e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 50608224,
      "step": 23440
    },
    {
      "epoch": 3.8246329526916805,
      "grad_norm": 1.2669740915298462,
      "learning_rate": 3.880228252640295e-05,
      "loss": 0.3168,
      "num_input_tokens_seen": 50618912,
      "step": 23445
    },
    {
      "epoch": 3.8254486133768353,
      "grad_norm": 0.1803109049797058,
      "learning_rate": 3.879634710971935e-05,
      "loss": 0.0697,
      "num_input_tokens_seen": 50629728,
      "step": 23450
    },
    {
      "epoch": 3.82626427406199,
      "grad_norm": 0.08579455316066742,
      "learning_rate": 3.8790410574633854e-05,
      "loss": 0.1001,
      "num_input_tokens_seen": 50640096,
      "step": 23455
    },
    {
      "epoch": 3.827079934747145,
      "grad_norm": 0.0672377198934555,
      "learning_rate": 3.8784472921627715e-05,
      "loss": 0.1406,
      "num_input_tokens_seen": 50650176,
      "step": 23460
    },
    {
      "epoch": 3.8278955954323,
      "grad_norm": 0.051177676767110825,
      "learning_rate": 3.877853415118224e-05,
      "loss": 0.1531,
      "num_input_tokens_seen": 50661504,
      "step": 23465
    },
    {
      "epoch": 3.828711256117455,
      "grad_norm": 0.13691945374011993,
      "learning_rate": 3.877259426377889e-05,
      "loss": 0.0122,
      "num_input_tokens_seen": 50672384,
      "step": 23470
    },
    {
      "epoch": 3.8295269168026103,
      "grad_norm": 0.0938628613948822,
      "learning_rate": 3.8766653259899165e-05,
      "loss": 0.1482,
      "num_input_tokens_seen": 50683168,
      "step": 23475
    },
    {
      "epoch": 3.830342577487765,
      "grad_norm": 4.719187259674072,
      "learning_rate": 3.8760711140024677e-05,
      "loss": 0.2146,
      "num_input_tokens_seen": 50692512,
      "step": 23480
    },
    {
      "epoch": 3.83115823817292,
      "grad_norm": 0.07632511854171753,
      "learning_rate": 3.875476790463712e-05,
      "loss": 0.1284,
      "num_input_tokens_seen": 50703168,
      "step": 23485
    },
    {
      "epoch": 3.831973898858075,
      "grad_norm": 0.7380771636962891,
      "learning_rate": 3.8748823554218286e-05,
      "loss": 0.199,
      "num_input_tokens_seen": 50714112,
      "step": 23490
    },
    {
      "epoch": 3.83278955954323,
      "grad_norm": 3.486445426940918,
      "learning_rate": 3.8742878089250043e-05,
      "loss": 0.1281,
      "num_input_tokens_seen": 50724576,
      "step": 23495
    },
    {
      "epoch": 3.8336052202283852,
      "grad_norm": 0.6635816693305969,
      "learning_rate": 3.8736931510214385e-05,
      "loss": 0.1293,
      "num_input_tokens_seen": 50735392,
      "step": 23500
    },
    {
      "epoch": 3.83442088091354,
      "grad_norm": 0.1672944873571396,
      "learning_rate": 3.873098381759336e-05,
      "loss": 0.0404,
      "num_input_tokens_seen": 50747488,
      "step": 23505
    },
    {
      "epoch": 3.835236541598695,
      "grad_norm": 0.22422181069850922,
      "learning_rate": 3.872503501186911e-05,
      "loss": 0.0297,
      "num_input_tokens_seen": 50757120,
      "step": 23510
    },
    {
      "epoch": 3.8360522022838497,
      "grad_norm": 0.06843329221010208,
      "learning_rate": 3.871908509352388e-05,
      "loss": 0.0706,
      "num_input_tokens_seen": 50767136,
      "step": 23515
    },
    {
      "epoch": 3.836867862969005,
      "grad_norm": 0.8522431254386902,
      "learning_rate": 3.871313406304001e-05,
      "loss": 0.0101,
      "num_input_tokens_seen": 50777728,
      "step": 23520
    },
    {
      "epoch": 3.8376835236541598,
      "grad_norm": 2.191159725189209,
      "learning_rate": 3.87071819208999e-05,
      "loss": 0.2006,
      "num_input_tokens_seen": 50787488,
      "step": 23525
    },
    {
      "epoch": 3.838499184339315,
      "grad_norm": 3.2646467685699463,
      "learning_rate": 3.870122866758609e-05,
      "loss": 0.1007,
      "num_input_tokens_seen": 50797760,
      "step": 23530
    },
    {
      "epoch": 3.83931484502447,
      "grad_norm": 0.24704453349113464,
      "learning_rate": 3.869527430358116e-05,
      "loss": 0.0304,
      "num_input_tokens_seen": 50808832,
      "step": 23535
    },
    {
      "epoch": 3.8401305057096247,
      "grad_norm": 0.10138709843158722,
      "learning_rate": 3.8689318829367796e-05,
      "loss": 0.124,
      "num_input_tokens_seen": 50819296,
      "step": 23540
    },
    {
      "epoch": 3.84094616639478,
      "grad_norm": 0.11551807075738907,
      "learning_rate": 3.86833622454288e-05,
      "loss": 0.0586,
      "num_input_tokens_seen": 50831424,
      "step": 23545
    },
    {
      "epoch": 3.8417618270799347,
      "grad_norm": 0.2826504409313202,
      "learning_rate": 3.8677404552247024e-05,
      "loss": 0.0514,
      "num_input_tokens_seen": 50841152,
      "step": 23550
    },
    {
      "epoch": 3.8425774877650896,
      "grad_norm": 6.57863187789917,
      "learning_rate": 3.8671445750305444e-05,
      "loss": 0.3215,
      "num_input_tokens_seen": 50850752,
      "step": 23555
    },
    {
      "epoch": 3.843393148450245,
      "grad_norm": 1.516378402709961,
      "learning_rate": 3.8665485840087104e-05,
      "loss": 0.0714,
      "num_input_tokens_seen": 50860896,
      "step": 23560
    },
    {
      "epoch": 3.8442088091353996,
      "grad_norm": 3.2162604331970215,
      "learning_rate": 3.865952482207513e-05,
      "loss": 0.1335,
      "num_input_tokens_seen": 50871072,
      "step": 23565
    },
    {
      "epoch": 3.8450244698205545,
      "grad_norm": 2.1718997955322266,
      "learning_rate": 3.865356269675278e-05,
      "loss": 0.106,
      "num_input_tokens_seen": 50880960,
      "step": 23570
    },
    {
      "epoch": 3.8458401305057097,
      "grad_norm": 4.0175933837890625,
      "learning_rate": 3.8647599464603355e-05,
      "loss": 0.0306,
      "num_input_tokens_seen": 50890080,
      "step": 23575
    },
    {
      "epoch": 3.8466557911908645,
      "grad_norm": 0.11540604382753372,
      "learning_rate": 3.864163512611028e-05,
      "loss": 0.006,
      "num_input_tokens_seen": 50901216,
      "step": 23580
    },
    {
      "epoch": 3.84747145187602,
      "grad_norm": 10.32332992553711,
      "learning_rate": 3.863566968175703e-05,
      "loss": 0.0681,
      "num_input_tokens_seen": 50912416,
      "step": 23585
    },
    {
      "epoch": 3.8482871125611746,
      "grad_norm": 0.05607185512781143,
      "learning_rate": 3.862970313202722e-05,
      "loss": 0.0092,
      "num_input_tokens_seen": 50922752,
      "step": 23590
    },
    {
      "epoch": 3.8491027732463294,
      "grad_norm": 0.08821630477905273,
      "learning_rate": 3.86237354774045e-05,
      "loss": 0.004,
      "num_input_tokens_seen": 50935168,
      "step": 23595
    },
    {
      "epoch": 3.8499184339314843,
      "grad_norm": 4.020759105682373,
      "learning_rate": 3.861776671837267e-05,
      "loss": 0.282,
      "num_input_tokens_seen": 50946304,
      "step": 23600
    },
    {
      "epoch": 3.8507340946166395,
      "grad_norm": 0.6966497302055359,
      "learning_rate": 3.861179685541557e-05,
      "loss": 0.0111,
      "num_input_tokens_seen": 50955872,
      "step": 23605
    },
    {
      "epoch": 3.8515497553017943,
      "grad_norm": 3.10355806350708,
      "learning_rate": 3.8605825889017156e-05,
      "loss": 0.1657,
      "num_input_tokens_seen": 50967488,
      "step": 23610
    },
    {
      "epoch": 3.8523654159869496,
      "grad_norm": 0.09562846273183823,
      "learning_rate": 3.859985381966146e-05,
      "loss": 0.0234,
      "num_input_tokens_seen": 50977664,
      "step": 23615
    },
    {
      "epoch": 3.8531810766721044,
      "grad_norm": 0.3513879179954529,
      "learning_rate": 3.8593880647832606e-05,
      "loss": 0.0194,
      "num_input_tokens_seen": 50989920,
      "step": 23620
    },
    {
      "epoch": 3.8539967373572592,
      "grad_norm": 6.543063163757324,
      "learning_rate": 3.858790637401482e-05,
      "loss": 0.1168,
      "num_input_tokens_seen": 50999136,
      "step": 23625
    },
    {
      "epoch": 3.8548123980424145,
      "grad_norm": 0.021561764180660248,
      "learning_rate": 3.858193099869239e-05,
      "loss": 0.2842,
      "num_input_tokens_seen": 51010336,
      "step": 23630
    },
    {
      "epoch": 3.8556280587275693,
      "grad_norm": 2.674123764038086,
      "learning_rate": 3.857595452234971e-05,
      "loss": 0.218,
      "num_input_tokens_seen": 51021792,
      "step": 23635
    },
    {
      "epoch": 3.8564437194127246,
      "grad_norm": 0.04573666304349899,
      "learning_rate": 3.856997694547129e-05,
      "loss": 0.1464,
      "num_input_tokens_seen": 51032416,
      "step": 23640
    },
    {
      "epoch": 3.8572593800978794,
      "grad_norm": 7.868891716003418,
      "learning_rate": 3.856399826854168e-05,
      "loss": 0.1463,
      "num_input_tokens_seen": 51042304,
      "step": 23645
    },
    {
      "epoch": 3.858075040783034,
      "grad_norm": 1.4317147731781006,
      "learning_rate": 3.855801849204555e-05,
      "loss": 0.084,
      "num_input_tokens_seen": 51053664,
      "step": 23650
    },
    {
      "epoch": 3.858890701468189,
      "grad_norm": 0.1854161024093628,
      "learning_rate": 3.855203761646764e-05,
      "loss": 0.009,
      "num_input_tokens_seen": 51064608,
      "step": 23655
    },
    {
      "epoch": 3.8597063621533443,
      "grad_norm": 0.04124079644680023,
      "learning_rate": 3.85460556422928e-05,
      "loss": 0.1217,
      "num_input_tokens_seen": 51074208,
      "step": 23660
    },
    {
      "epoch": 3.860522022838499,
      "grad_norm": 2.4907431602478027,
      "learning_rate": 3.854007257000596e-05,
      "loss": 0.0648,
      "num_input_tokens_seen": 51085600,
      "step": 23665
    },
    {
      "epoch": 3.8613376835236544,
      "grad_norm": 0.08115028589963913,
      "learning_rate": 3.853408840009214e-05,
      "loss": 0.0973,
      "num_input_tokens_seen": 51097344,
      "step": 23670
    },
    {
      "epoch": 3.862153344208809,
      "grad_norm": 0.11729727685451508,
      "learning_rate": 3.8528103133036434e-05,
      "loss": 0.0077,
      "num_input_tokens_seen": 51109184,
      "step": 23675
    },
    {
      "epoch": 3.862969004893964,
      "grad_norm": 0.07816064357757568,
      "learning_rate": 3.8522116769324056e-05,
      "loss": 0.1096,
      "num_input_tokens_seen": 51120800,
      "step": 23680
    },
    {
      "epoch": 3.863784665579119,
      "grad_norm": 0.13324949145317078,
      "learning_rate": 3.851612930944027e-05,
      "loss": 0.0708,
      "num_input_tokens_seen": 51131008,
      "step": 23685
    },
    {
      "epoch": 3.864600326264274,
      "grad_norm": 0.4708535969257355,
      "learning_rate": 3.851014075387048e-05,
      "loss": 0.2543,
      "num_input_tokens_seen": 51141600,
      "step": 23690
    },
    {
      "epoch": 3.865415986949429,
      "grad_norm": 2.499551296234131,
      "learning_rate": 3.850415110310012e-05,
      "loss": 0.0298,
      "num_input_tokens_seen": 51152128,
      "step": 23695
    },
    {
      "epoch": 3.866231647634584,
      "grad_norm": 0.04783983528614044,
      "learning_rate": 3.8498160357614756e-05,
      "loss": 0.0438,
      "num_input_tokens_seen": 51163840,
      "step": 23700
    },
    {
      "epoch": 3.867047308319739,
      "grad_norm": 0.27930396795272827,
      "learning_rate": 3.8492168517900016e-05,
      "loss": 0.086,
      "num_input_tokens_seen": 51173984,
      "step": 23705
    },
    {
      "epoch": 3.867862969004894,
      "grad_norm": 4.76519250869751,
      "learning_rate": 3.8486175584441643e-05,
      "loss": 0.1064,
      "num_input_tokens_seen": 51185120,
      "step": 23710
    },
    {
      "epoch": 3.868678629690049,
      "grad_norm": 0.048098061233758926,
      "learning_rate": 3.8480181557725455e-05,
      "loss": 0.0231,
      "num_input_tokens_seen": 51195968,
      "step": 23715
    },
    {
      "epoch": 3.869494290375204,
      "grad_norm": 0.9078822135925293,
      "learning_rate": 3.847418643823735e-05,
      "loss": 0.0254,
      "num_input_tokens_seen": 51206368,
      "step": 23720
    },
    {
      "epoch": 3.870309951060359,
      "grad_norm": 0.14563687145709991,
      "learning_rate": 3.8468190226463316e-05,
      "loss": 0.1075,
      "num_input_tokens_seen": 51216928,
      "step": 23725
    },
    {
      "epoch": 3.871125611745514,
      "grad_norm": 3.4391939640045166,
      "learning_rate": 3.846219292288945e-05,
      "loss": 0.3805,
      "num_input_tokens_seen": 51227936,
      "step": 23730
    },
    {
      "epoch": 3.8719412724306688,
      "grad_norm": 0.7178727984428406,
      "learning_rate": 3.845619452800192e-05,
      "loss": 0.217,
      "num_input_tokens_seen": 51238880,
      "step": 23735
    },
    {
      "epoch": 3.8727569331158236,
      "grad_norm": 0.705920934677124,
      "learning_rate": 3.845019504228699e-05,
      "loss": 0.1067,
      "num_input_tokens_seen": 51250560,
      "step": 23740
    },
    {
      "epoch": 3.873572593800979,
      "grad_norm": 0.06280519813299179,
      "learning_rate": 3.8444194466230994e-05,
      "loss": 0.0058,
      "num_input_tokens_seen": 51261440,
      "step": 23745
    },
    {
      "epoch": 3.8743882544861337,
      "grad_norm": 0.23043088614940643,
      "learning_rate": 3.843819280032038e-05,
      "loss": 0.178,
      "num_input_tokens_seen": 51272640,
      "step": 23750
    },
    {
      "epoch": 3.875203915171289,
      "grad_norm": 2.443315267562866,
      "learning_rate": 3.843219004504168e-05,
      "loss": 0.3003,
      "num_input_tokens_seen": 51284064,
      "step": 23755
    },
    {
      "epoch": 3.8760195758564437,
      "grad_norm": 2.8800432682037354,
      "learning_rate": 3.84261862008815e-05,
      "loss": 0.1018,
      "num_input_tokens_seen": 51294368,
      "step": 23760
    },
    {
      "epoch": 3.8768352365415986,
      "grad_norm": 0.11355220526456833,
      "learning_rate": 3.8420181268326536e-05,
      "loss": 0.0793,
      "num_input_tokens_seen": 51304704,
      "step": 23765
    },
    {
      "epoch": 3.877650897226754,
      "grad_norm": 2.2404026985168457,
      "learning_rate": 3.841417524786359e-05,
      "loss": 0.0141,
      "num_input_tokens_seen": 51316128,
      "step": 23770
    },
    {
      "epoch": 3.8784665579119086,
      "grad_norm": 0.03165043145418167,
      "learning_rate": 3.840816813997954e-05,
      "loss": 0.0677,
      "num_input_tokens_seen": 51327712,
      "step": 23775
    },
    {
      "epoch": 3.8792822185970635,
      "grad_norm": 2.7214980125427246,
      "learning_rate": 3.8402159945161346e-05,
      "loss": 0.2191,
      "num_input_tokens_seen": 51338624,
      "step": 23780
    },
    {
      "epoch": 3.8800978792822187,
      "grad_norm": 0.05646583065390587,
      "learning_rate": 3.839615066389607e-05,
      "loss": 0.0186,
      "num_input_tokens_seen": 51348928,
      "step": 23785
    },
    {
      "epoch": 3.8809135399673735,
      "grad_norm": 6.865532875061035,
      "learning_rate": 3.839014029667084e-05,
      "loss": 0.1212,
      "num_input_tokens_seen": 51359744,
      "step": 23790
    },
    {
      "epoch": 3.8817292006525284,
      "grad_norm": 0.0790897086262703,
      "learning_rate": 3.83841288439729e-05,
      "loss": 0.252,
      "num_input_tokens_seen": 51371616,
      "step": 23795
    },
    {
      "epoch": 3.8825448613376836,
      "grad_norm": 0.24048492312431335,
      "learning_rate": 3.837811630628957e-05,
      "loss": 0.1602,
      "num_input_tokens_seen": 51382752,
      "step": 23800
    },
    {
      "epoch": 3.8833605220228384,
      "grad_norm": 6.1536431312561035,
      "learning_rate": 3.837210268410824e-05,
      "loss": 0.2929,
      "num_input_tokens_seen": 51393440,
      "step": 23805
    },
    {
      "epoch": 3.8841761827079937,
      "grad_norm": 1.6496332883834839,
      "learning_rate": 3.836608797791642e-05,
      "loss": 0.0169,
      "num_input_tokens_seen": 51401792,
      "step": 23810
    },
    {
      "epoch": 3.8849918433931485,
      "grad_norm": 1.3711801767349243,
      "learning_rate": 3.8360072188201704e-05,
      "loss": 0.2867,
      "num_input_tokens_seen": 51413216,
      "step": 23815
    },
    {
      "epoch": 3.8858075040783033,
      "grad_norm": 0.17884577810764313,
      "learning_rate": 3.835405531545173e-05,
      "loss": 0.1844,
      "num_input_tokens_seen": 51423840,
      "step": 23820
    },
    {
      "epoch": 3.886623164763458,
      "grad_norm": 0.10995449870824814,
      "learning_rate": 3.834803736015428e-05,
      "loss": 0.0403,
      "num_input_tokens_seen": 51434464,
      "step": 23825
    },
    {
      "epoch": 3.8874388254486134,
      "grad_norm": 0.39918452501296997,
      "learning_rate": 3.8342018322797205e-05,
      "loss": 0.2348,
      "num_input_tokens_seen": 51446112,
      "step": 23830
    },
    {
      "epoch": 3.8882544861337682,
      "grad_norm": 2.797009229660034,
      "learning_rate": 3.833599820386842e-05,
      "loss": 0.2407,
      "num_input_tokens_seen": 51456704,
      "step": 23835
    },
    {
      "epoch": 3.8890701468189235,
      "grad_norm": 3.1366398334503174,
      "learning_rate": 3.8329977003855956e-05,
      "loss": 0.1807,
      "num_input_tokens_seen": 51467424,
      "step": 23840
    },
    {
      "epoch": 3.8898858075040783,
      "grad_norm": 1.1464749574661255,
      "learning_rate": 3.832395472324791e-05,
      "loss": 0.0183,
      "num_input_tokens_seen": 51477792,
      "step": 23845
    },
    {
      "epoch": 3.890701468189233,
      "grad_norm": 2.3124589920043945,
      "learning_rate": 3.83179313625325e-05,
      "loss": 0.2955,
      "num_input_tokens_seen": 51486368,
      "step": 23850
    },
    {
      "epoch": 3.8915171288743884,
      "grad_norm": 5.189943790435791,
      "learning_rate": 3.8311906922198005e-05,
      "loss": 0.1469,
      "num_input_tokens_seen": 51496224,
      "step": 23855
    },
    {
      "epoch": 3.892332789559543,
      "grad_norm": 0.08746843785047531,
      "learning_rate": 3.830588140273278e-05,
      "loss": 0.1092,
      "num_input_tokens_seen": 51507872,
      "step": 23860
    },
    {
      "epoch": 3.8931484502446985,
      "grad_norm": 2.0948245525360107,
      "learning_rate": 3.829985480462529e-05,
      "loss": 0.0898,
      "num_input_tokens_seen": 51518112,
      "step": 23865
    },
    {
      "epoch": 3.8939641109298533,
      "grad_norm": 3.290347099304199,
      "learning_rate": 3.82938271283641e-05,
      "loss": 0.1162,
      "num_input_tokens_seen": 51529120,
      "step": 23870
    },
    {
      "epoch": 3.894779771615008,
      "grad_norm": 0.5662157535552979,
      "learning_rate": 3.828779837443783e-05,
      "loss": 0.12,
      "num_input_tokens_seen": 51540768,
      "step": 23875
    },
    {
      "epoch": 3.895595432300163,
      "grad_norm": 0.3609636127948761,
      "learning_rate": 3.8281768543335195e-05,
      "loss": 0.1168,
      "num_input_tokens_seen": 51550304,
      "step": 23880
    },
    {
      "epoch": 3.896411092985318,
      "grad_norm": 0.196279376745224,
      "learning_rate": 3.827573763554502e-05,
      "loss": 0.0881,
      "num_input_tokens_seen": 51561088,
      "step": 23885
    },
    {
      "epoch": 3.897226753670473,
      "grad_norm": 3.1661527156829834,
      "learning_rate": 3.826970565155618e-05,
      "loss": 0.1379,
      "num_input_tokens_seen": 51571264,
      "step": 23890
    },
    {
      "epoch": 3.8980424143556283,
      "grad_norm": 0.12482129037380219,
      "learning_rate": 3.8263672591857666e-05,
      "loss": 0.0682,
      "num_input_tokens_seen": 51582912,
      "step": 23895
    },
    {
      "epoch": 3.898858075040783,
      "grad_norm": 0.6632061004638672,
      "learning_rate": 3.825763845693857e-05,
      "loss": 0.0494,
      "num_input_tokens_seen": 51594144,
      "step": 23900
    },
    {
      "epoch": 3.899673735725938,
      "grad_norm": 0.1388709396123886,
      "learning_rate": 3.825160324728802e-05,
      "loss": 0.0762,
      "num_input_tokens_seen": 51604928,
      "step": 23905
    },
    {
      "epoch": 3.9004893964110927,
      "grad_norm": 4.40236234664917,
      "learning_rate": 3.824556696339528e-05,
      "loss": 0.264,
      "num_input_tokens_seen": 51614176,
      "step": 23910
    },
    {
      "epoch": 3.901305057096248,
      "grad_norm": 2.515265703201294,
      "learning_rate": 3.823952960574967e-05,
      "loss": 0.0735,
      "num_input_tokens_seen": 51624992,
      "step": 23915
    },
    {
      "epoch": 3.902120717781403,
      "grad_norm": 0.3718299865722656,
      "learning_rate": 3.823349117484062e-05,
      "loss": 0.0106,
      "num_input_tokens_seen": 51635744,
      "step": 23920
    },
    {
      "epoch": 3.902936378466558,
      "grad_norm": 0.35502177476882935,
      "learning_rate": 3.822745167115762e-05,
      "loss": 0.0169,
      "num_input_tokens_seen": 51645472,
      "step": 23925
    },
    {
      "epoch": 3.903752039151713,
      "grad_norm": 5.897215366363525,
      "learning_rate": 3.822141109519027e-05,
      "loss": 0.1144,
      "num_input_tokens_seen": 51656128,
      "step": 23930
    },
    {
      "epoch": 3.9045676998368677,
      "grad_norm": 3.5866503715515137,
      "learning_rate": 3.821536944742827e-05,
      "loss": 0.2927,
      "num_input_tokens_seen": 51666368,
      "step": 23935
    },
    {
      "epoch": 3.905383360522023,
      "grad_norm": 2.2468960285186768,
      "learning_rate": 3.820932672836135e-05,
      "loss": 0.0721,
      "num_input_tokens_seen": 51677792,
      "step": 23940
    },
    {
      "epoch": 3.9061990212071778,
      "grad_norm": 0.5014287233352661,
      "learning_rate": 3.820328293847939e-05,
      "loss": 0.2138,
      "num_input_tokens_seen": 51689280,
      "step": 23945
    },
    {
      "epoch": 3.907014681892333,
      "grad_norm": 0.36148637533187866,
      "learning_rate": 3.819723807827232e-05,
      "loss": 0.0615,
      "num_input_tokens_seen": 51699712,
      "step": 23950
    },
    {
      "epoch": 3.907830342577488,
      "grad_norm": 3.4324469566345215,
      "learning_rate": 3.8191192148230176e-05,
      "loss": 0.2222,
      "num_input_tokens_seen": 51710624,
      "step": 23955
    },
    {
      "epoch": 3.9086460032626427,
      "grad_norm": 4.112708568572998,
      "learning_rate": 3.818514514884306e-05,
      "loss": 0.0628,
      "num_input_tokens_seen": 51721952,
      "step": 23960
    },
    {
      "epoch": 3.9094616639477975,
      "grad_norm": 0.1018175259232521,
      "learning_rate": 3.8179097080601175e-05,
      "loss": 0.2149,
      "num_input_tokens_seen": 51731584,
      "step": 23965
    },
    {
      "epoch": 3.9102773246329527,
      "grad_norm": 0.1846858710050583,
      "learning_rate": 3.817304794399481e-05,
      "loss": 0.228,
      "num_input_tokens_seen": 51743232,
      "step": 23970
    },
    {
      "epoch": 3.9110929853181076,
      "grad_norm": 0.16165630519390106,
      "learning_rate": 3.816699773951434e-05,
      "loss": 0.0865,
      "num_input_tokens_seen": 51755712,
      "step": 23975
    },
    {
      "epoch": 3.911908646003263,
      "grad_norm": 0.1982184648513794,
      "learning_rate": 3.8160946467650226e-05,
      "loss": 0.0162,
      "num_input_tokens_seen": 51767776,
      "step": 23980
    },
    {
      "epoch": 3.9127243066884176,
      "grad_norm": 0.9768170118331909,
      "learning_rate": 3.815489412889302e-05,
      "loss": 0.109,
      "num_input_tokens_seen": 51778176,
      "step": 23985
    },
    {
      "epoch": 3.9135399673735725,
      "grad_norm": 0.1167348325252533,
      "learning_rate": 3.8148840723733335e-05,
      "loss": 0.1562,
      "num_input_tokens_seen": 51789024,
      "step": 23990
    },
    {
      "epoch": 3.9143556280587277,
      "grad_norm": 0.16741915047168732,
      "learning_rate": 3.814278625266191e-05,
      "loss": 0.0853,
      "num_input_tokens_seen": 51798752,
      "step": 23995
    },
    {
      "epoch": 3.9151712887438825,
      "grad_norm": 10.41820240020752,
      "learning_rate": 3.8136730716169554e-05,
      "loss": 0.161,
      "num_input_tokens_seen": 51809856,
      "step": 24000
    },
    {
      "epoch": 3.9159869494290374,
      "grad_norm": 0.1835808902978897,
      "learning_rate": 3.8130674114747146e-05,
      "loss": 0.0321,
      "num_input_tokens_seen": 51819584,
      "step": 24005
    },
    {
      "epoch": 3.9168026101141926,
      "grad_norm": 2.11114764213562,
      "learning_rate": 3.812461644888566e-05,
      "loss": 0.1169,
      "num_input_tokens_seen": 51831168,
      "step": 24010
    },
    {
      "epoch": 3.9176182707993474,
      "grad_norm": 0.175631582736969,
      "learning_rate": 3.8118557719076186e-05,
      "loss": 0.0438,
      "num_input_tokens_seen": 51839776,
      "step": 24015
    },
    {
      "epoch": 3.9184339314845023,
      "grad_norm": 0.08247873932123184,
      "learning_rate": 3.811249792580985e-05,
      "loss": 0.1423,
      "num_input_tokens_seen": 51850368,
      "step": 24020
    },
    {
      "epoch": 3.9192495921696575,
      "grad_norm": 3.1666617393493652,
      "learning_rate": 3.810643706957791e-05,
      "loss": 0.1349,
      "num_input_tokens_seen": 51861248,
      "step": 24025
    },
    {
      "epoch": 3.9200652528548123,
      "grad_norm": 9.254212379455566,
      "learning_rate": 3.810037515087167e-05,
      "loss": 0.3344,
      "num_input_tokens_seen": 51872928,
      "step": 24030
    },
    {
      "epoch": 3.9208809135399676,
      "grad_norm": 9.931520462036133,
      "learning_rate": 3.809431217018255e-05,
      "loss": 0.2647,
      "num_input_tokens_seen": 51883680,
      "step": 24035
    },
    {
      "epoch": 3.9216965742251224,
      "grad_norm": 2.439455270767212,
      "learning_rate": 3.8088248128002044e-05,
      "loss": 0.3082,
      "num_input_tokens_seen": 51895808,
      "step": 24040
    },
    {
      "epoch": 3.9225122349102772,
      "grad_norm": 0.23569348454475403,
      "learning_rate": 3.808218302482175e-05,
      "loss": 0.0596,
      "num_input_tokens_seen": 51905760,
      "step": 24045
    },
    {
      "epoch": 3.923327895595432,
      "grad_norm": 3.6026687622070312,
      "learning_rate": 3.8076116861133305e-05,
      "loss": 0.2412,
      "num_input_tokens_seen": 51916288,
      "step": 24050
    },
    {
      "epoch": 3.9241435562805873,
      "grad_norm": 0.42549461126327515,
      "learning_rate": 3.8070049637428485e-05,
      "loss": 0.0244,
      "num_input_tokens_seen": 51927040,
      "step": 24055
    },
    {
      "epoch": 3.924959216965742,
      "grad_norm": 0.1177360936999321,
      "learning_rate": 3.806398135419913e-05,
      "loss": 0.0116,
      "num_input_tokens_seen": 51937760,
      "step": 24060
    },
    {
      "epoch": 3.9257748776508974,
      "grad_norm": 0.8087987899780273,
      "learning_rate": 3.805791201193716e-05,
      "loss": 0.0162,
      "num_input_tokens_seen": 51948384,
      "step": 24065
    },
    {
      "epoch": 3.926590538336052,
      "grad_norm": 2.6735105514526367,
      "learning_rate": 3.8051841611134576e-05,
      "loss": 0.1596,
      "num_input_tokens_seen": 51958496,
      "step": 24070
    },
    {
      "epoch": 3.927406199021207,
      "grad_norm": 0.08874467760324478,
      "learning_rate": 3.804577015228349e-05,
      "loss": 0.0228,
      "num_input_tokens_seen": 51969216,
      "step": 24075
    },
    {
      "epoch": 3.9282218597063623,
      "grad_norm": 0.10837216675281525,
      "learning_rate": 3.803969763587609e-05,
      "loss": 0.0239,
      "num_input_tokens_seen": 51980768,
      "step": 24080
    },
    {
      "epoch": 3.929037520391517,
      "grad_norm": 0.42723798751831055,
      "learning_rate": 3.803362406240463e-05,
      "loss": 0.0496,
      "num_input_tokens_seen": 51990976,
      "step": 24085
    },
    {
      "epoch": 3.9298531810766724,
      "grad_norm": 8.741414070129395,
      "learning_rate": 3.802754943236148e-05,
      "loss": 0.1212,
      "num_input_tokens_seen": 52002240,
      "step": 24090
    },
    {
      "epoch": 3.930668841761827,
      "grad_norm": 4.983425140380859,
      "learning_rate": 3.8021473746239064e-05,
      "loss": 0.0816,
      "num_input_tokens_seen": 52013760,
      "step": 24095
    },
    {
      "epoch": 3.931484502446982,
      "grad_norm": 0.33898645639419556,
      "learning_rate": 3.801539700452992e-05,
      "loss": 0.016,
      "num_input_tokens_seen": 52024960,
      "step": 24100
    },
    {
      "epoch": 3.932300163132137,
      "grad_norm": 0.09568259865045547,
      "learning_rate": 3.800931920772666e-05,
      "loss": 0.0903,
      "num_input_tokens_seen": 52035872,
      "step": 24105
    },
    {
      "epoch": 3.933115823817292,
      "grad_norm": 0.11450260132551193,
      "learning_rate": 3.8003240356321965e-05,
      "loss": 0.2034,
      "num_input_tokens_seen": 52047136,
      "step": 24110
    },
    {
      "epoch": 3.933931484502447,
      "grad_norm": 0.05758999288082123,
      "learning_rate": 3.7997160450808634e-05,
      "loss": 0.2704,
      "num_input_tokens_seen": 52059040,
      "step": 24115
    },
    {
      "epoch": 3.934747145187602,
      "grad_norm": 0.12375502288341522,
      "learning_rate": 3.7991079491679524e-05,
      "loss": 0.1142,
      "num_input_tokens_seen": 52070176,
      "step": 24120
    },
    {
      "epoch": 3.935562805872757,
      "grad_norm": 0.07845445722341537,
      "learning_rate": 3.79849974794276e-05,
      "loss": 0.0391,
      "num_input_tokens_seen": 52080576,
      "step": 24125
    },
    {
      "epoch": 3.936378466557912,
      "grad_norm": 0.668676495552063,
      "learning_rate": 3.7978914414545895e-05,
      "loss": 0.073,
      "num_input_tokens_seen": 52091744,
      "step": 24130
    },
    {
      "epoch": 3.9371941272430666,
      "grad_norm": 0.23654919862747192,
      "learning_rate": 3.797283029752753e-05,
      "loss": 0.2787,
      "num_input_tokens_seen": 52102432,
      "step": 24135
    },
    {
      "epoch": 3.938009787928222,
      "grad_norm": 0.12085125595331192,
      "learning_rate": 3.796674512886573e-05,
      "loss": 0.0253,
      "num_input_tokens_seen": 52111968,
      "step": 24140
    },
    {
      "epoch": 3.9388254486133767,
      "grad_norm": 0.13203787803649902,
      "learning_rate": 3.7960658909053766e-05,
      "loss": 0.1109,
      "num_input_tokens_seen": 52123872,
      "step": 24145
    },
    {
      "epoch": 3.939641109298532,
      "grad_norm": 0.09453985840082169,
      "learning_rate": 3.7954571638585035e-05,
      "loss": 0.0118,
      "num_input_tokens_seen": 52134400,
      "step": 24150
    },
    {
      "epoch": 3.9404567699836868,
      "grad_norm": 0.08839213848114014,
      "learning_rate": 3.7948483317952985e-05,
      "loss": 0.0097,
      "num_input_tokens_seen": 52144448,
      "step": 24155
    },
    {
      "epoch": 3.9412724306688416,
      "grad_norm": 7.535945892333984,
      "learning_rate": 3.794239394765119e-05,
      "loss": 0.1754,
      "num_input_tokens_seen": 52155840,
      "step": 24160
    },
    {
      "epoch": 3.942088091353997,
      "grad_norm": 0.07784853130578995,
      "learning_rate": 3.793630352817327e-05,
      "loss": 0.0804,
      "num_input_tokens_seen": 52167968,
      "step": 24165
    },
    {
      "epoch": 3.9429037520391517,
      "grad_norm": 0.5558604001998901,
      "learning_rate": 3.7930212060012946e-05,
      "loss": 0.011,
      "num_input_tokens_seen": 52177792,
      "step": 24170
    },
    {
      "epoch": 3.943719412724307,
      "grad_norm": 4.437076091766357,
      "learning_rate": 3.792411954366402e-05,
      "loss": 0.0865,
      "num_input_tokens_seen": 52188992,
      "step": 24175
    },
    {
      "epoch": 3.9445350734094617,
      "grad_norm": 6.7131757736206055,
      "learning_rate": 3.791802597962039e-05,
      "loss": 0.1714,
      "num_input_tokens_seen": 52200192,
      "step": 24180
    },
    {
      "epoch": 3.9453507340946166,
      "grad_norm": 2.135201930999756,
      "learning_rate": 3.791193136837603e-05,
      "loss": 0.1607,
      "num_input_tokens_seen": 52210944,
      "step": 24185
    },
    {
      "epoch": 3.9461663947797714,
      "grad_norm": 0.322039932012558,
      "learning_rate": 3.7905835710425e-05,
      "loss": 0.1606,
      "num_input_tokens_seen": 52220960,
      "step": 24190
    },
    {
      "epoch": 3.9469820554649266,
      "grad_norm": 3.2532620429992676,
      "learning_rate": 3.789973900626145e-05,
      "loss": 0.236,
      "num_input_tokens_seen": 52231232,
      "step": 24195
    },
    {
      "epoch": 3.9477977161500815,
      "grad_norm": 0.06413562595844269,
      "learning_rate": 3.78936412563796e-05,
      "loss": 0.1216,
      "num_input_tokens_seen": 52240032,
      "step": 24200
    },
    {
      "epoch": 3.9486133768352367,
      "grad_norm": 0.05348264053463936,
      "learning_rate": 3.788754246127375e-05,
      "loss": 0.0141,
      "num_input_tokens_seen": 52251968,
      "step": 24205
    },
    {
      "epoch": 3.9494290375203915,
      "grad_norm": 0.026875484734773636,
      "learning_rate": 3.7881442621438333e-05,
      "loss": 0.0168,
      "num_input_tokens_seen": 52263072,
      "step": 24210
    },
    {
      "epoch": 3.9502446982055464,
      "grad_norm": 4.340539932250977,
      "learning_rate": 3.787534173736782e-05,
      "loss": 0.2256,
      "num_input_tokens_seen": 52274336,
      "step": 24215
    },
    {
      "epoch": 3.9510603588907016,
      "grad_norm": 8.347455024719238,
      "learning_rate": 3.786923980955678e-05,
      "loss": 0.0662,
      "num_input_tokens_seen": 52284992,
      "step": 24220
    },
    {
      "epoch": 3.9518760195758564,
      "grad_norm": 3.045938014984131,
      "learning_rate": 3.7863136838499855e-05,
      "loss": 0.0868,
      "num_input_tokens_seen": 52296032,
      "step": 24225
    },
    {
      "epoch": 3.9526916802610113,
      "grad_norm": 2.9014275074005127,
      "learning_rate": 3.785703282469179e-05,
      "loss": 0.2974,
      "num_input_tokens_seen": 52306432,
      "step": 24230
    },
    {
      "epoch": 3.9535073409461665,
      "grad_norm": 0.120819590985775,
      "learning_rate": 3.785092776862741e-05,
      "loss": 0.0123,
      "num_input_tokens_seen": 52316640,
      "step": 24235
    },
    {
      "epoch": 3.9543230016313213,
      "grad_norm": 5.4397149085998535,
      "learning_rate": 3.784482167080162e-05,
      "loss": 0.1867,
      "num_input_tokens_seen": 52327104,
      "step": 24240
    },
    {
      "epoch": 3.955138662316476,
      "grad_norm": 0.06622958928346634,
      "learning_rate": 3.783871453170941e-05,
      "loss": 0.1037,
      "num_input_tokens_seen": 52338080,
      "step": 24245
    },
    {
      "epoch": 3.9559543230016314,
      "grad_norm": 0.13515892624855042,
      "learning_rate": 3.783260635184586e-05,
      "loss": 0.4521,
      "num_input_tokens_seen": 52348576,
      "step": 24250
    },
    {
      "epoch": 3.9567699836867862,
      "grad_norm": 0.10123121738433838,
      "learning_rate": 3.782649713170613e-05,
      "loss": 0.1931,
      "num_input_tokens_seen": 52358848,
      "step": 24255
    },
    {
      "epoch": 3.9575856443719415,
      "grad_norm": 0.1138068288564682,
      "learning_rate": 3.7820386871785455e-05,
      "loss": 0.0617,
      "num_input_tokens_seen": 52369248,
      "step": 24260
    },
    {
      "epoch": 3.9584013050570963,
      "grad_norm": 4.640443801879883,
      "learning_rate": 3.7814275572579175e-05,
      "loss": 0.1401,
      "num_input_tokens_seen": 52380640,
      "step": 24265
    },
    {
      "epoch": 3.959216965742251,
      "grad_norm": 0.8769100904464722,
      "learning_rate": 3.780816323458269e-05,
      "loss": 0.1245,
      "num_input_tokens_seen": 52391136,
      "step": 24270
    },
    {
      "epoch": 3.960032626427406,
      "grad_norm": 0.24302352964878082,
      "learning_rate": 3.7802049858291515e-05,
      "loss": 0.07,
      "num_input_tokens_seen": 52401824,
      "step": 24275
    },
    {
      "epoch": 3.960848287112561,
      "grad_norm": 2.1642651557922363,
      "learning_rate": 3.779593544420122e-05,
      "loss": 0.4402,
      "num_input_tokens_seen": 52412576,
      "step": 24280
    },
    {
      "epoch": 3.961663947797716,
      "grad_norm": 7.140532493591309,
      "learning_rate": 3.7789819992807474e-05,
      "loss": 0.1253,
      "num_input_tokens_seen": 52422720,
      "step": 24285
    },
    {
      "epoch": 3.9624796084828713,
      "grad_norm": 5.465451240539551,
      "learning_rate": 3.778370350460601e-05,
      "loss": 0.1529,
      "num_input_tokens_seen": 52432800,
      "step": 24290
    },
    {
      "epoch": 3.963295269168026,
      "grad_norm": 0.20366117358207703,
      "learning_rate": 3.777758598009269e-05,
      "loss": 0.0684,
      "num_input_tokens_seen": 52443680,
      "step": 24295
    },
    {
      "epoch": 3.964110929853181,
      "grad_norm": 0.09867950528860092,
      "learning_rate": 3.777146741976342e-05,
      "loss": 0.1906,
      "num_input_tokens_seen": 52454976,
      "step": 24300
    },
    {
      "epoch": 3.964926590538336,
      "grad_norm": 4.806294918060303,
      "learning_rate": 3.776534782411419e-05,
      "loss": 0.2447,
      "num_input_tokens_seen": 52465376,
      "step": 24305
    },
    {
      "epoch": 3.965742251223491,
      "grad_norm": 0.7245295643806458,
      "learning_rate": 3.77592271936411e-05,
      "loss": 0.1734,
      "num_input_tokens_seen": 52477408,
      "step": 24310
    },
    {
      "epoch": 3.9665579119086463,
      "grad_norm": 0.575904130935669,
      "learning_rate": 3.775310552884031e-05,
      "loss": 0.0808,
      "num_input_tokens_seen": 52489472,
      "step": 24315
    },
    {
      "epoch": 3.967373572593801,
      "grad_norm": 0.27454447746276855,
      "learning_rate": 3.7746982830208075e-05,
      "loss": 0.0187,
      "num_input_tokens_seen": 52499648,
      "step": 24320
    },
    {
      "epoch": 3.968189233278956,
      "grad_norm": 0.15682311356067657,
      "learning_rate": 3.774085909824074e-05,
      "loss": 0.2283,
      "num_input_tokens_seen": 52508096,
      "step": 24325
    },
    {
      "epoch": 3.9690048939641107,
      "grad_norm": 0.3136940002441406,
      "learning_rate": 3.7734734333434726e-05,
      "loss": 0.0263,
      "num_input_tokens_seen": 52520224,
      "step": 24330
    },
    {
      "epoch": 3.969820554649266,
      "grad_norm": 0.11171292513608932,
      "learning_rate": 3.772860853628652e-05,
      "loss": 0.0889,
      "num_input_tokens_seen": 52531520,
      "step": 24335
    },
    {
      "epoch": 3.970636215334421,
      "grad_norm": 0.12412583827972412,
      "learning_rate": 3.772248170729272e-05,
      "loss": 0.2002,
      "num_input_tokens_seen": 52543040,
      "step": 24340
    },
    {
      "epoch": 3.971451876019576,
      "grad_norm": 2.202580451965332,
      "learning_rate": 3.771635384695001e-05,
      "loss": 0.4208,
      "num_input_tokens_seen": 52553760,
      "step": 24345
    },
    {
      "epoch": 3.972267536704731,
      "grad_norm": 0.40388861298561096,
      "learning_rate": 3.771022495575513e-05,
      "loss": 0.0678,
      "num_input_tokens_seen": 52564416,
      "step": 24350
    },
    {
      "epoch": 3.9730831973898857,
      "grad_norm": 3.802887201309204,
      "learning_rate": 3.770409503420492e-05,
      "loss": 0.1629,
      "num_input_tokens_seen": 52574880,
      "step": 24355
    },
    {
      "epoch": 3.9738988580750405,
      "grad_norm": 0.3135204017162323,
      "learning_rate": 3.769796408279631e-05,
      "loss": 0.2398,
      "num_input_tokens_seen": 52586464,
      "step": 24360
    },
    {
      "epoch": 3.9747145187601958,
      "grad_norm": 0.09923018515110016,
      "learning_rate": 3.76918321020263e-05,
      "loss": 0.0126,
      "num_input_tokens_seen": 52596608,
      "step": 24365
    },
    {
      "epoch": 3.9755301794453506,
      "grad_norm": 0.8163269758224487,
      "learning_rate": 3.768569909239199e-05,
      "loss": 0.093,
      "num_input_tokens_seen": 52607680,
      "step": 24370
    },
    {
      "epoch": 3.976345840130506,
      "grad_norm": 3.5307023525238037,
      "learning_rate": 3.767956505439054e-05,
      "loss": 0.2163,
      "num_input_tokens_seen": 52619648,
      "step": 24375
    },
    {
      "epoch": 3.9771615008156607,
      "grad_norm": 0.07754193991422653,
      "learning_rate": 3.767342998851921e-05,
      "loss": 0.0958,
      "num_input_tokens_seen": 52630528,
      "step": 24380
    },
    {
      "epoch": 3.9779771615008155,
      "grad_norm": 0.06371898204088211,
      "learning_rate": 3.766729389527535e-05,
      "loss": 0.0081,
      "num_input_tokens_seen": 52640608,
      "step": 24385
    },
    {
      "epoch": 3.9787928221859707,
      "grad_norm": 2.396165370941162,
      "learning_rate": 3.766115677515637e-05,
      "loss": 0.1193,
      "num_input_tokens_seen": 52651872,
      "step": 24390
    },
    {
      "epoch": 3.9796084828711256,
      "grad_norm": 0.4515656530857086,
      "learning_rate": 3.765501862865976e-05,
      "loss": 0.0623,
      "num_input_tokens_seen": 52663904,
      "step": 24395
    },
    {
      "epoch": 3.980424143556281,
      "grad_norm": 4.028966426849365,
      "learning_rate": 3.764887945628315e-05,
      "loss": 0.1423,
      "num_input_tokens_seen": 52674848,
      "step": 24400
    },
    {
      "epoch": 3.9812398042414356,
      "grad_norm": 3.8315951824188232,
      "learning_rate": 3.76427392585242e-05,
      "loss": 0.3372,
      "num_input_tokens_seen": 52686656,
      "step": 24405
    },
    {
      "epoch": 3.9820554649265905,
      "grad_norm": 0.09870853275060654,
      "learning_rate": 3.7636598035880633e-05,
      "loss": 0.0296,
      "num_input_tokens_seen": 52696544,
      "step": 24410
    },
    {
      "epoch": 3.9828711256117453,
      "grad_norm": 0.04006926342844963,
      "learning_rate": 3.763045578885033e-05,
      "loss": 0.0853,
      "num_input_tokens_seen": 52706912,
      "step": 24415
    },
    {
      "epoch": 3.9836867862969005,
      "grad_norm": 0.08128601312637329,
      "learning_rate": 3.762431251793118e-05,
      "loss": 0.1298,
      "num_input_tokens_seen": 52718016,
      "step": 24420
    },
    {
      "epoch": 3.9845024469820554,
      "grad_norm": 4.476246356964111,
      "learning_rate": 3.7618168223621215e-05,
      "loss": 0.4976,
      "num_input_tokens_seen": 52728288,
      "step": 24425
    },
    {
      "epoch": 3.9853181076672106,
      "grad_norm": 0.20408247411251068,
      "learning_rate": 3.761202290641851e-05,
      "loss": 0.0464,
      "num_input_tokens_seen": 52739104,
      "step": 24430
    },
    {
      "epoch": 3.9861337683523654,
      "grad_norm": 6.188340187072754,
      "learning_rate": 3.760587656682122e-05,
      "loss": 0.0654,
      "num_input_tokens_seen": 52749536,
      "step": 24435
    },
    {
      "epoch": 3.9869494290375203,
      "grad_norm": 3.187253952026367,
      "learning_rate": 3.759972920532762e-05,
      "loss": 0.2218,
      "num_input_tokens_seen": 52760416,
      "step": 24440
    },
    {
      "epoch": 3.9877650897226755,
      "grad_norm": 0.1321265697479248,
      "learning_rate": 3.759358082243604e-05,
      "loss": 0.1113,
      "num_input_tokens_seen": 52772160,
      "step": 24445
    },
    {
      "epoch": 3.9885807504078303,
      "grad_norm": 4.083325386047363,
      "learning_rate": 3.7587431418644906e-05,
      "loss": 0.0218,
      "num_input_tokens_seen": 52781984,
      "step": 24450
    },
    {
      "epoch": 3.9893964110929856,
      "grad_norm": 5.500246524810791,
      "learning_rate": 3.758128099445271e-05,
      "loss": 0.1482,
      "num_input_tokens_seen": 52791776,
      "step": 24455
    },
    {
      "epoch": 3.9902120717781404,
      "grad_norm": 0.1290796846151352,
      "learning_rate": 3.757512955035804e-05,
      "loss": 0.0392,
      "num_input_tokens_seen": 52802528,
      "step": 24460
    },
    {
      "epoch": 3.9910277324632952,
      "grad_norm": 1.1345018148422241,
      "learning_rate": 3.7568977086859566e-05,
      "loss": 0.0764,
      "num_input_tokens_seen": 52812640,
      "step": 24465
    },
    {
      "epoch": 3.99184339314845,
      "grad_norm": 0.26198068261146545,
      "learning_rate": 3.7562823604456035e-05,
      "loss": 0.1294,
      "num_input_tokens_seen": 52823744,
      "step": 24470
    },
    {
      "epoch": 3.9926590538336053,
      "grad_norm": 3.3135857582092285,
      "learning_rate": 3.7556669103646266e-05,
      "loss": 0.0971,
      "num_input_tokens_seen": 52834016,
      "step": 24475
    },
    {
      "epoch": 3.99347471451876,
      "grad_norm": 2.892432451248169,
      "learning_rate": 3.75505135849292e-05,
      "loss": 0.2205,
      "num_input_tokens_seen": 52845248,
      "step": 24480
    },
    {
      "epoch": 3.9942903752039154,
      "grad_norm": 0.3946531414985657,
      "learning_rate": 3.7544357048803824e-05,
      "loss": 0.0121,
      "num_input_tokens_seen": 52855104,
      "step": 24485
    },
    {
      "epoch": 3.99510603588907,
      "grad_norm": 0.11953412741422653,
      "learning_rate": 3.7538199495769214e-05,
      "loss": 0.0097,
      "num_input_tokens_seen": 52866048,
      "step": 24490
    },
    {
      "epoch": 3.995921696574225,
      "grad_norm": 0.09131952375173569,
      "learning_rate": 3.753204092632454e-05,
      "loss": 0.0429,
      "num_input_tokens_seen": 52876768,
      "step": 24495
    },
    {
      "epoch": 3.99673735725938,
      "grad_norm": 0.11290084570646286,
      "learning_rate": 3.752588134096903e-05,
      "loss": 0.1046,
      "num_input_tokens_seen": 52887872,
      "step": 24500
    },
    {
      "epoch": 3.997553017944535,
      "grad_norm": 7.183202266693115,
      "learning_rate": 3.751972074020202e-05,
      "loss": 0.1252,
      "num_input_tokens_seen": 52898624,
      "step": 24505
    },
    {
      "epoch": 3.99836867862969,
      "grad_norm": 0.767217755317688,
      "learning_rate": 3.751355912452294e-05,
      "loss": 0.1128,
      "num_input_tokens_seen": 52908768,
      "step": 24510
    },
    {
      "epoch": 3.999184339314845,
      "grad_norm": 0.08274319022893906,
      "learning_rate": 3.7507396494431246e-05,
      "loss": 0.038,
      "num_input_tokens_seen": 52920064,
      "step": 24515
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.06304477155208588,
      "learning_rate": 3.750123285042654e-05,
      "loss": 0.006,
      "num_input_tokens_seen": 52929744,
      "step": 24520
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.15006808936595917,
      "eval_runtime": 132.9283,
      "eval_samples_per_second": 20.5,
      "eval_steps_per_second": 5.131,
      "num_input_tokens_seen": 52929744,
      "step": 24520
    },
    {
      "epoch": 4.000815660685155,
      "grad_norm": 6.7583513259887695,
      "learning_rate": 3.749506819300846e-05,
      "loss": 0.1874,
      "num_input_tokens_seen": 52941840,
      "step": 24525
    },
    {
      "epoch": 4.00163132137031,
      "grad_norm": 0.07735206186771393,
      "learning_rate": 3.748890252267676e-05,
      "loss": 0.0937,
      "num_input_tokens_seen": 52952560,
      "step": 24530
    },
    {
      "epoch": 4.002446982055465,
      "grad_norm": 0.1584097295999527,
      "learning_rate": 3.748273583993126e-05,
      "loss": 0.1299,
      "num_input_tokens_seen": 52961968,
      "step": 24535
    },
    {
      "epoch": 4.00326264274062,
      "grad_norm": 8.218377113342285,
      "learning_rate": 3.747656814527185e-05,
      "loss": 0.1639,
      "num_input_tokens_seen": 52971408,
      "step": 24540
    },
    {
      "epoch": 4.004078303425775,
      "grad_norm": 0.07929620146751404,
      "learning_rate": 3.747039943919852e-05,
      "loss": 0.1938,
      "num_input_tokens_seen": 52982832,
      "step": 24545
    },
    {
      "epoch": 4.00489396411093,
      "grad_norm": 1.899898648262024,
      "learning_rate": 3.746422972221134e-05,
      "loss": 0.1135,
      "num_input_tokens_seen": 52993648,
      "step": 24550
    },
    {
      "epoch": 4.005709624796085,
      "grad_norm": 0.15819135308265686,
      "learning_rate": 3.745805899481045e-05,
      "loss": 0.1322,
      "num_input_tokens_seen": 53004272,
      "step": 24555
    },
    {
      "epoch": 4.006525285481239,
      "grad_norm": 4.0420684814453125,
      "learning_rate": 3.745188725749609e-05,
      "loss": 0.2079,
      "num_input_tokens_seen": 53013872,
      "step": 24560
    },
    {
      "epoch": 4.007340946166395,
      "grad_norm": 0.08327308297157288,
      "learning_rate": 3.744571451076856e-05,
      "loss": 0.0109,
      "num_input_tokens_seen": 53024912,
      "step": 24565
    },
    {
      "epoch": 4.00815660685155,
      "grad_norm": 6.343252658843994,
      "learning_rate": 3.7439540755128276e-05,
      "loss": 0.2233,
      "num_input_tokens_seen": 53037360,
      "step": 24570
    },
    {
      "epoch": 4.008972267536705,
      "grad_norm": 4.405014991760254,
      "learning_rate": 3.7433365991075695e-05,
      "loss": 0.1093,
      "num_input_tokens_seen": 53048688,
      "step": 24575
    },
    {
      "epoch": 4.00978792822186,
      "grad_norm": 0.39339569211006165,
      "learning_rate": 3.742719021911138e-05,
      "loss": 0.0156,
      "num_input_tokens_seen": 53059568,
      "step": 24580
    },
    {
      "epoch": 4.010603588907014,
      "grad_norm": 0.8301985859870911,
      "learning_rate": 3.742101343973598e-05,
      "loss": 0.0451,
      "num_input_tokens_seen": 53071312,
      "step": 24585
    },
    {
      "epoch": 4.011419249592169,
      "grad_norm": 3.7286832332611084,
      "learning_rate": 3.741483565345019e-05,
      "loss": 0.1269,
      "num_input_tokens_seen": 53082544,
      "step": 24590
    },
    {
      "epoch": 4.012234910277325,
      "grad_norm": 0.08277827501296997,
      "learning_rate": 3.740865686075484e-05,
      "loss": 0.0107,
      "num_input_tokens_seen": 53093168,
      "step": 24595
    },
    {
      "epoch": 4.01305057096248,
      "grad_norm": 0.09586529433727264,
      "learning_rate": 3.7402477062150795e-05,
      "loss": 0.1632,
      "num_input_tokens_seen": 53104688,
      "step": 24600
    },
    {
      "epoch": 4.013866231647635,
      "grad_norm": 0.16563250124454498,
      "learning_rate": 3.739629625813904e-05,
      "loss": 0.1802,
      "num_input_tokens_seen": 53116528,
      "step": 24605
    },
    {
      "epoch": 4.014681892332789,
      "grad_norm": 0.373230904340744,
      "learning_rate": 3.739011444922061e-05,
      "loss": 0.0248,
      "num_input_tokens_seen": 53128016,
      "step": 24610
    },
    {
      "epoch": 4.015497553017944,
      "grad_norm": 0.1066754013299942,
      "learning_rate": 3.7383931635896634e-05,
      "loss": 0.059,
      "num_input_tokens_seen": 53139120,
      "step": 24615
    },
    {
      "epoch": 4.0163132137031,
      "grad_norm": 0.12022466212511063,
      "learning_rate": 3.737774781866833e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 53149392,
      "step": 24620
    },
    {
      "epoch": 4.017128874388255,
      "grad_norm": 5.02327823638916,
      "learning_rate": 3.737156299803698e-05,
      "loss": 0.0226,
      "num_input_tokens_seen": 53160304,
      "step": 24625
    },
    {
      "epoch": 4.0179445350734095,
      "grad_norm": 2.427842140197754,
      "learning_rate": 3.7365377174503956e-05,
      "loss": 0.0662,
      "num_input_tokens_seen": 53171248,
      "step": 24630
    },
    {
      "epoch": 4.018760195758564,
      "grad_norm": 0.3013766407966614,
      "learning_rate": 3.7359190348570726e-05,
      "loss": 0.2066,
      "num_input_tokens_seen": 53183312,
      "step": 24635
    },
    {
      "epoch": 4.019575856443719,
      "grad_norm": 0.05398857966065407,
      "learning_rate": 3.735300252073881e-05,
      "loss": 0.1387,
      "num_input_tokens_seen": 53194192,
      "step": 24640
    },
    {
      "epoch": 4.020391517128874,
      "grad_norm": 0.1032828837633133,
      "learning_rate": 3.734681369150983e-05,
      "loss": 0.2984,
      "num_input_tokens_seen": 53205296,
      "step": 24645
    },
    {
      "epoch": 4.02120717781403,
      "grad_norm": 0.061879999935626984,
      "learning_rate": 3.7340623861385496e-05,
      "loss": 0.0551,
      "num_input_tokens_seen": 53215632,
      "step": 24650
    },
    {
      "epoch": 4.0220228384991845,
      "grad_norm": 0.23465439677238464,
      "learning_rate": 3.7334433030867564e-05,
      "loss": 0.1207,
      "num_input_tokens_seen": 53226704,
      "step": 24655
    },
    {
      "epoch": 4.022838499184339,
      "grad_norm": 0.050487224012613297,
      "learning_rate": 3.732824120045791e-05,
      "loss": 0.0071,
      "num_input_tokens_seen": 53237328,
      "step": 24660
    },
    {
      "epoch": 4.023654159869494,
      "grad_norm": 0.14931054413318634,
      "learning_rate": 3.732204837065847e-05,
      "loss": 0.0258,
      "num_input_tokens_seen": 53248912,
      "step": 24665
    },
    {
      "epoch": 4.024469820554649,
      "grad_norm": 0.04330739751458168,
      "learning_rate": 3.731585454197127e-05,
      "loss": 0.0847,
      "num_input_tokens_seen": 53260528,
      "step": 24670
    },
    {
      "epoch": 4.025285481239805,
      "grad_norm": 0.5237597823143005,
      "learning_rate": 3.7309659714898404e-05,
      "loss": 0.0995,
      "num_input_tokens_seen": 53272016,
      "step": 24675
    },
    {
      "epoch": 4.0261011419249595,
      "grad_norm": 0.34010252356529236,
      "learning_rate": 3.730346388994207e-05,
      "loss": 0.05,
      "num_input_tokens_seen": 53282640,
      "step": 24680
    },
    {
      "epoch": 4.026916802610114,
      "grad_norm": 0.019590429961681366,
      "learning_rate": 3.729726706760452e-05,
      "loss": 0.0682,
      "num_input_tokens_seen": 53293904,
      "step": 24685
    },
    {
      "epoch": 4.027732463295269,
      "grad_norm": 0.05539269000291824,
      "learning_rate": 3.729106924838812e-05,
      "loss": 0.0366,
      "num_input_tokens_seen": 53305904,
      "step": 24690
    },
    {
      "epoch": 4.028548123980424,
      "grad_norm": 0.0381401851773262,
      "learning_rate": 3.728487043279527e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 53316528,
      "step": 24695
    },
    {
      "epoch": 4.029363784665579,
      "grad_norm": 1.7421964406967163,
      "learning_rate": 3.727867062132849e-05,
      "loss": 0.1583,
      "num_input_tokens_seen": 53328464,
      "step": 24700
    },
    {
      "epoch": 4.0301794453507345,
      "grad_norm": 2.946322202682495,
      "learning_rate": 3.7272469814490376e-05,
      "loss": 0.1548,
      "num_input_tokens_seen": 53339440,
      "step": 24705
    },
    {
      "epoch": 4.030995106035889,
      "grad_norm": 0.048724982887506485,
      "learning_rate": 3.726626801278358e-05,
      "loss": 0.1217,
      "num_input_tokens_seen": 53350000,
      "step": 24710
    },
    {
      "epoch": 4.031810766721044,
      "grad_norm": 0.027808157727122307,
      "learning_rate": 3.726006521671086e-05,
      "loss": 0.0087,
      "num_input_tokens_seen": 53360528,
      "step": 24715
    },
    {
      "epoch": 4.032626427406199,
      "grad_norm": 0.04992597550153732,
      "learning_rate": 3.7253861426775056e-05,
      "loss": 0.071,
      "num_input_tokens_seen": 53371600,
      "step": 24720
    },
    {
      "epoch": 4.033442088091354,
      "grad_norm": 0.43019813299179077,
      "learning_rate": 3.7247656643479064e-05,
      "loss": 0.1201,
      "num_input_tokens_seen": 53382032,
      "step": 24725
    },
    {
      "epoch": 4.034257748776509,
      "grad_norm": 2.9925100803375244,
      "learning_rate": 3.724145086732588e-05,
      "loss": 0.1676,
      "num_input_tokens_seen": 53392816,
      "step": 24730
    },
    {
      "epoch": 4.035073409461664,
      "grad_norm": 0.16607451438903809,
      "learning_rate": 3.7235244098818576e-05,
      "loss": 0.0634,
      "num_input_tokens_seen": 53404624,
      "step": 24735
    },
    {
      "epoch": 4.035889070146819,
      "grad_norm": 4.58715295791626,
      "learning_rate": 3.722903633846031e-05,
      "loss": 0.1857,
      "num_input_tokens_seen": 53416048,
      "step": 24740
    },
    {
      "epoch": 4.036704730831974,
      "grad_norm": 11.995363235473633,
      "learning_rate": 3.72228275867543e-05,
      "loss": 0.0628,
      "num_input_tokens_seen": 53426704,
      "step": 24745
    },
    {
      "epoch": 4.037520391517129,
      "grad_norm": 4.142264366149902,
      "learning_rate": 3.721661784420387e-05,
      "loss": 0.1447,
      "num_input_tokens_seen": 53439248,
      "step": 24750
    },
    {
      "epoch": 4.0383360522022835,
      "grad_norm": 0.30402427911758423,
      "learning_rate": 3.721040711131242e-05,
      "loss": 0.2321,
      "num_input_tokens_seen": 53448400,
      "step": 24755
    },
    {
      "epoch": 4.039151712887439,
      "grad_norm": 8.091341018676758,
      "learning_rate": 3.72041953885834e-05,
      "loss": 0.2699,
      "num_input_tokens_seen": 53460336,
      "step": 24760
    },
    {
      "epoch": 4.039967373572594,
      "grad_norm": 3.046910047531128,
      "learning_rate": 3.719798267652038e-05,
      "loss": 0.1623,
      "num_input_tokens_seen": 53471568,
      "step": 24765
    },
    {
      "epoch": 4.040783034257749,
      "grad_norm": 0.10052645951509476,
      "learning_rate": 3.719176897562701e-05,
      "loss": 0.1048,
      "num_input_tokens_seen": 53483568,
      "step": 24770
    },
    {
      "epoch": 4.041598694942904,
      "grad_norm": 0.43055182695388794,
      "learning_rate": 3.718555428640697e-05,
      "loss": 0.0993,
      "num_input_tokens_seen": 53492848,
      "step": 24775
    },
    {
      "epoch": 4.0424143556280585,
      "grad_norm": 3.058279514312744,
      "learning_rate": 3.717933860936407e-05,
      "loss": 0.2066,
      "num_input_tokens_seen": 53502992,
      "step": 24780
    },
    {
      "epoch": 4.043230016313213,
      "grad_norm": 0.631894588470459,
      "learning_rate": 3.7173121945002197e-05,
      "loss": 0.0916,
      "num_input_tokens_seen": 53513264,
      "step": 24785
    },
    {
      "epoch": 4.044045676998369,
      "grad_norm": 0.27123937010765076,
      "learning_rate": 3.716690429382529e-05,
      "loss": 0.0888,
      "num_input_tokens_seen": 53523984,
      "step": 24790
    },
    {
      "epoch": 4.044861337683524,
      "grad_norm": 0.26726406812667847,
      "learning_rate": 3.716068565633738e-05,
      "loss": 0.0563,
      "num_input_tokens_seen": 53534672,
      "step": 24795
    },
    {
      "epoch": 4.045676998368679,
      "grad_norm": 0.06608258187770844,
      "learning_rate": 3.715446603304259e-05,
      "loss": 0.0395,
      "num_input_tokens_seen": 53546224,
      "step": 24800
    },
    {
      "epoch": 4.0464926590538335,
      "grad_norm": 0.05818143114447594,
      "learning_rate": 3.7148245424445114e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 53555952,
      "step": 24805
    },
    {
      "epoch": 4.047308319738988,
      "grad_norm": 0.06563541293144226,
      "learning_rate": 3.7142023831049226e-05,
      "loss": 0.1048,
      "num_input_tokens_seen": 53567024,
      "step": 24810
    },
    {
      "epoch": 4.048123980424143,
      "grad_norm": 0.14977294206619263,
      "learning_rate": 3.713580125335928e-05,
      "loss": 0.0826,
      "num_input_tokens_seen": 53578032,
      "step": 24815
    },
    {
      "epoch": 4.048939641109299,
      "grad_norm": 3.3613672256469727,
      "learning_rate": 3.7129577691879694e-05,
      "loss": 0.1246,
      "num_input_tokens_seen": 53589968,
      "step": 24820
    },
    {
      "epoch": 4.049755301794454,
      "grad_norm": 0.1577356606721878,
      "learning_rate": 3.712335314711501e-05,
      "loss": 0.0128,
      "num_input_tokens_seen": 53599472,
      "step": 24825
    },
    {
      "epoch": 4.0505709624796085,
      "grad_norm": 0.0451761893928051,
      "learning_rate": 3.7117127619569796e-05,
      "loss": 0.1854,
      "num_input_tokens_seen": 53611376,
      "step": 24830
    },
    {
      "epoch": 4.051386623164763,
      "grad_norm": 0.38689252734184265,
      "learning_rate": 3.7110901109748745e-05,
      "loss": 0.008,
      "num_input_tokens_seen": 53623760,
      "step": 24835
    },
    {
      "epoch": 4.052202283849918,
      "grad_norm": 6.373911380767822,
      "learning_rate": 3.710467361815659e-05,
      "loss": 0.3157,
      "num_input_tokens_seen": 53634416,
      "step": 24840
    },
    {
      "epoch": 4.053017944535074,
      "grad_norm": 0.07288794964551926,
      "learning_rate": 3.709844514529818e-05,
      "loss": 0.0646,
      "num_input_tokens_seen": 53644528,
      "step": 24845
    },
    {
      "epoch": 4.053833605220229,
      "grad_norm": 0.16024848818778992,
      "learning_rate": 3.709221569167842e-05,
      "loss": 0.0677,
      "num_input_tokens_seen": 53654480,
      "step": 24850
    },
    {
      "epoch": 4.054649265905383,
      "grad_norm": 0.06283432990312576,
      "learning_rate": 3.70859852578023e-05,
      "loss": 0.2092,
      "num_input_tokens_seen": 53665136,
      "step": 24855
    },
    {
      "epoch": 4.055464926590538,
      "grad_norm": 0.16095586121082306,
      "learning_rate": 3.70797538441749e-05,
      "loss": 0.2295,
      "num_input_tokens_seen": 53677104,
      "step": 24860
    },
    {
      "epoch": 4.056280587275693,
      "grad_norm": 0.05175670608878136,
      "learning_rate": 3.707352145130135e-05,
      "loss": 0.0215,
      "num_input_tokens_seen": 53688016,
      "step": 24865
    },
    {
      "epoch": 4.057096247960848,
      "grad_norm": 0.2593509256839752,
      "learning_rate": 3.706728807968689e-05,
      "loss": 0.1117,
      "num_input_tokens_seen": 53698544,
      "step": 24870
    },
    {
      "epoch": 4.057911908646004,
      "grad_norm": 6.267168045043945,
      "learning_rate": 3.706105372983683e-05,
      "loss": 0.0268,
      "num_input_tokens_seen": 53710000,
      "step": 24875
    },
    {
      "epoch": 4.058727569331158,
      "grad_norm": 0.06121160462498665,
      "learning_rate": 3.705481840225656e-05,
      "loss": 0.0126,
      "num_input_tokens_seen": 53720656,
      "step": 24880
    },
    {
      "epoch": 4.059543230016313,
      "grad_norm": 0.07203152030706406,
      "learning_rate": 3.704858209745155e-05,
      "loss": 0.1027,
      "num_input_tokens_seen": 53731280,
      "step": 24885
    },
    {
      "epoch": 4.060358890701468,
      "grad_norm": 2.044027090072632,
      "learning_rate": 3.704234481592733e-05,
      "loss": 0.0952,
      "num_input_tokens_seen": 53741712,
      "step": 24890
    },
    {
      "epoch": 4.061174551386623,
      "grad_norm": 0.1900075078010559,
      "learning_rate": 3.703610655818955e-05,
      "loss": 0.0872,
      "num_input_tokens_seen": 53753360,
      "step": 24895
    },
    {
      "epoch": 4.061990212071779,
      "grad_norm": 0.17512746155261993,
      "learning_rate": 3.702986732474389e-05,
      "loss": 0.1335,
      "num_input_tokens_seen": 53764112,
      "step": 24900
    },
    {
      "epoch": 4.062805872756933,
      "grad_norm": 0.19973358511924744,
      "learning_rate": 3.702362711609615e-05,
      "loss": 0.025,
      "num_input_tokens_seen": 53774320,
      "step": 24905
    },
    {
      "epoch": 4.063621533442088,
      "grad_norm": 0.05211591720581055,
      "learning_rate": 3.701738593275219e-05,
      "loss": 0.1012,
      "num_input_tokens_seen": 53784848,
      "step": 24910
    },
    {
      "epoch": 4.064437194127243,
      "grad_norm": 0.0706319734454155,
      "learning_rate": 3.701114377521795e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 53795632,
      "step": 24915
    },
    {
      "epoch": 4.065252854812398,
      "grad_norm": 0.8674832582473755,
      "learning_rate": 3.700490064399945e-05,
      "loss": 0.0962,
      "num_input_tokens_seen": 53806480,
      "step": 24920
    },
    {
      "epoch": 4.066068515497553,
      "grad_norm": 0.5811168551445007,
      "learning_rate": 3.6998656539602795e-05,
      "loss": 0.0128,
      "num_input_tokens_seen": 53816816,
      "step": 24925
    },
    {
      "epoch": 4.066884176182708,
      "grad_norm": 0.07982023805379868,
      "learning_rate": 3.699241146253416e-05,
      "loss": 0.0476,
      "num_input_tokens_seen": 53828720,
      "step": 24930
    },
    {
      "epoch": 4.067699836867863,
      "grad_norm": 0.31422337889671326,
      "learning_rate": 3.69861654132998e-05,
      "loss": 0.0843,
      "num_input_tokens_seen": 53840816,
      "step": 24935
    },
    {
      "epoch": 4.068515497553018,
      "grad_norm": 0.3347615897655487,
      "learning_rate": 3.6979918392406055e-05,
      "loss": 0.1054,
      "num_input_tokens_seen": 53851120,
      "step": 24940
    },
    {
      "epoch": 4.069331158238173,
      "grad_norm": 0.2270788550376892,
      "learning_rate": 3.697367040035934e-05,
      "loss": 0.1236,
      "num_input_tokens_seen": 53862288,
      "step": 24945
    },
    {
      "epoch": 4.070146818923328,
      "grad_norm": 5.550386905670166,
      "learning_rate": 3.696742143766615e-05,
      "loss": 0.3034,
      "num_input_tokens_seen": 53873584,
      "step": 24950
    },
    {
      "epoch": 4.0709624796084825,
      "grad_norm": 0.4365846812725067,
      "learning_rate": 3.696117150483306e-05,
      "loss": 0.3094,
      "num_input_tokens_seen": 53885328,
      "step": 24955
    },
    {
      "epoch": 4.071778140293638,
      "grad_norm": 0.3106023669242859,
      "learning_rate": 3.695492060236671e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 53895376,
      "step": 24960
    },
    {
      "epoch": 4.072593800978793,
      "grad_norm": 0.06049007922410965,
      "learning_rate": 3.694866873077384e-05,
      "loss": 0.0536,
      "num_input_tokens_seen": 53906640,
      "step": 24965
    },
    {
      "epoch": 4.073409461663948,
      "grad_norm": 3.8268356323242188,
      "learning_rate": 3.6942415890561254e-05,
      "loss": 0.2055,
      "num_input_tokens_seen": 53917264,
      "step": 24970
    },
    {
      "epoch": 4.074225122349103,
      "grad_norm": 0.09823447465896606,
      "learning_rate": 3.6936162082235844e-05,
      "loss": 0.1302,
      "num_input_tokens_seen": 53927984,
      "step": 24975
    },
    {
      "epoch": 4.075040783034257,
      "grad_norm": 0.029776480048894882,
      "learning_rate": 3.692990730630457e-05,
      "loss": 0.0094,
      "num_input_tokens_seen": 53938672,
      "step": 24980
    },
    {
      "epoch": 4.075856443719413,
      "grad_norm": 3.100522994995117,
      "learning_rate": 3.692365156327448e-05,
      "loss": 0.0714,
      "num_input_tokens_seen": 53951120,
      "step": 24985
    },
    {
      "epoch": 4.076672104404568,
      "grad_norm": 0.08029317110776901,
      "learning_rate": 3.691739485365269e-05,
      "loss": 0.0815,
      "num_input_tokens_seen": 53962192,
      "step": 24990
    },
    {
      "epoch": 4.077487765089723,
      "grad_norm": 1.666698694229126,
      "learning_rate": 3.691113717794641e-05,
      "loss": 0.1829,
      "num_input_tokens_seen": 53973488,
      "step": 24995
    },
    {
      "epoch": 4.078303425774878,
      "grad_norm": 0.1005997583270073,
      "learning_rate": 3.6904878536662904e-05,
      "loss": 0.1201,
      "num_input_tokens_seen": 53984272,
      "step": 25000
    },
    {
      "epoch": 4.079119086460032,
      "grad_norm": 3.5812594890594482,
      "learning_rate": 3.6898618930309556e-05,
      "loss": 0.1481,
      "num_input_tokens_seen": 53994704,
      "step": 25005
    },
    {
      "epoch": 4.079934747145187,
      "grad_norm": 0.07321606576442719,
      "learning_rate": 3.6892358359393767e-05,
      "loss": 0.098,
      "num_input_tokens_seen": 54005552,
      "step": 25010
    },
    {
      "epoch": 4.080750407830343,
      "grad_norm": 0.19606611132621765,
      "learning_rate": 3.688609682442308e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 54016560,
      "step": 25015
    },
    {
      "epoch": 4.081566068515498,
      "grad_norm": 2.725309133529663,
      "learning_rate": 3.687983432590507e-05,
      "loss": 0.0964,
      "num_input_tokens_seen": 54026736,
      "step": 25020
    },
    {
      "epoch": 4.082381729200653,
      "grad_norm": 5.751374244689941,
      "learning_rate": 3.6873570864347415e-05,
      "loss": 0.2511,
      "num_input_tokens_seen": 54036592,
      "step": 25025
    },
    {
      "epoch": 4.083197389885807,
      "grad_norm": 0.20788143575191498,
      "learning_rate": 3.686730644025786e-05,
      "loss": 0.3024,
      "num_input_tokens_seen": 54046864,
      "step": 25030
    },
    {
      "epoch": 4.084013050570962,
      "grad_norm": 5.681027889251709,
      "learning_rate": 3.686104105414423e-05,
      "loss": 0.0208,
      "num_input_tokens_seen": 54058448,
      "step": 25035
    },
    {
      "epoch": 4.084828711256117,
      "grad_norm": 8.688946723937988,
      "learning_rate": 3.6854774706514424e-05,
      "loss": 0.1365,
      "num_input_tokens_seen": 54068848,
      "step": 25040
    },
    {
      "epoch": 4.085644371941273,
      "grad_norm": 0.15534977614879608,
      "learning_rate": 3.684850739787644e-05,
      "loss": 0.1952,
      "num_input_tokens_seen": 54079504,
      "step": 25045
    },
    {
      "epoch": 4.0864600326264275,
      "grad_norm": 0.14457550644874573,
      "learning_rate": 3.684223912873832e-05,
      "loss": 0.0333,
      "num_input_tokens_seen": 54089424,
      "step": 25050
    },
    {
      "epoch": 4.087275693311582,
      "grad_norm": 0.07060840725898743,
      "learning_rate": 3.683596989960821e-05,
      "loss": 0.0221,
      "num_input_tokens_seen": 54100848,
      "step": 25055
    },
    {
      "epoch": 4.088091353996737,
      "grad_norm": 0.23421601951122284,
      "learning_rate": 3.682969971099433e-05,
      "loss": 0.0752,
      "num_input_tokens_seen": 54111376,
      "step": 25060
    },
    {
      "epoch": 4.088907014681892,
      "grad_norm": 0.18303559720516205,
      "learning_rate": 3.682342856340496e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 54123408,
      "step": 25065
    },
    {
      "epoch": 4.089722675367048,
      "grad_norm": 0.104855477809906,
      "learning_rate": 3.681715645734848e-05,
      "loss": 0.0814,
      "num_input_tokens_seen": 54133296,
      "step": 25070
    },
    {
      "epoch": 4.0905383360522025,
      "grad_norm": 0.08674453943967819,
      "learning_rate": 3.681088339333334e-05,
      "loss": 0.0476,
      "num_input_tokens_seen": 54144240,
      "step": 25075
    },
    {
      "epoch": 4.091353996737357,
      "grad_norm": 0.05705615505576134,
      "learning_rate": 3.680460937186807e-05,
      "loss": 0.089,
      "num_input_tokens_seen": 54157008,
      "step": 25080
    },
    {
      "epoch": 4.092169657422512,
      "grad_norm": 0.06719832122325897,
      "learning_rate": 3.679833439346126e-05,
      "loss": 0.0936,
      "num_input_tokens_seen": 54167792,
      "step": 25085
    },
    {
      "epoch": 4.092985318107667,
      "grad_norm": 0.2682736814022064,
      "learning_rate": 3.6792058458621607e-05,
      "loss": 0.1036,
      "num_input_tokens_seen": 54178672,
      "step": 25090
    },
    {
      "epoch": 4.093800978792822,
      "grad_norm": 0.1056482195854187,
      "learning_rate": 3.678578156785786e-05,
      "loss": 0.015,
      "num_input_tokens_seen": 54190288,
      "step": 25095
    },
    {
      "epoch": 4.0946166394779775,
      "grad_norm": 7.483541488647461,
      "learning_rate": 3.677950372167885e-05,
      "loss": 0.1145,
      "num_input_tokens_seen": 54199728,
      "step": 25100
    },
    {
      "epoch": 4.095432300163132,
      "grad_norm": 0.26063913106918335,
      "learning_rate": 3.677322492059352e-05,
      "loss": 0.0126,
      "num_input_tokens_seen": 54209840,
      "step": 25105
    },
    {
      "epoch": 4.096247960848287,
      "grad_norm": 0.05677470564842224,
      "learning_rate": 3.676694516511083e-05,
      "loss": 0.1141,
      "num_input_tokens_seen": 54219856,
      "step": 25110
    },
    {
      "epoch": 4.097063621533442,
      "grad_norm": 0.07580921798944473,
      "learning_rate": 3.676066445573986e-05,
      "loss": 0.0271,
      "num_input_tokens_seen": 54230096,
      "step": 25115
    },
    {
      "epoch": 4.097879282218597,
      "grad_norm": 0.11550819873809814,
      "learning_rate": 3.675438279298975e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 54241840,
      "step": 25120
    },
    {
      "epoch": 4.0986949429037525,
      "grad_norm": 4.404228687286377,
      "learning_rate": 3.674810017736974e-05,
      "loss": 0.037,
      "num_input_tokens_seen": 54251920,
      "step": 25125
    },
    {
      "epoch": 4.099510603588907,
      "grad_norm": 0.21475058794021606,
      "learning_rate": 3.674181660938911e-05,
      "loss": 0.1168,
      "num_input_tokens_seen": 54263344,
      "step": 25130
    },
    {
      "epoch": 4.100326264274062,
      "grad_norm": 4.548152923583984,
      "learning_rate": 3.6735532089557256e-05,
      "loss": 0.108,
      "num_input_tokens_seen": 54274384,
      "step": 25135
    },
    {
      "epoch": 4.101141924959217,
      "grad_norm": 0.06051735207438469,
      "learning_rate": 3.672924661838362e-05,
      "loss": 0.0645,
      "num_input_tokens_seen": 54285296,
      "step": 25140
    },
    {
      "epoch": 4.101957585644372,
      "grad_norm": 0.24951577186584473,
      "learning_rate": 3.672296019637774e-05,
      "loss": 0.1373,
      "num_input_tokens_seen": 54295504,
      "step": 25145
    },
    {
      "epoch": 4.102773246329527,
      "grad_norm": 4.169466972351074,
      "learning_rate": 3.6716672824049234e-05,
      "loss": 0.0927,
      "num_input_tokens_seen": 54306704,
      "step": 25150
    },
    {
      "epoch": 4.103588907014682,
      "grad_norm": 0.026811687275767326,
      "learning_rate": 3.671038450190777e-05,
      "loss": 0.1799,
      "num_input_tokens_seen": 54316720,
      "step": 25155
    },
    {
      "epoch": 4.104404567699837,
      "grad_norm": 4.2193284034729,
      "learning_rate": 3.670409523046312e-05,
      "loss": 0.196,
      "num_input_tokens_seen": 54327408,
      "step": 25160
    },
    {
      "epoch": 4.105220228384992,
      "grad_norm": 1.2365455627441406,
      "learning_rate": 3.669780501022513e-05,
      "loss": 0.0864,
      "num_input_tokens_seen": 54337104,
      "step": 25165
    },
    {
      "epoch": 4.106035889070147,
      "grad_norm": 0.19575998187065125,
      "learning_rate": 3.669151384170371e-05,
      "loss": 0.1946,
      "num_input_tokens_seen": 54347408,
      "step": 25170
    },
    {
      "epoch": 4.1068515497553015,
      "grad_norm": 2.7266340255737305,
      "learning_rate": 3.668522172540886e-05,
      "loss": 0.1258,
      "num_input_tokens_seen": 54357264,
      "step": 25175
    },
    {
      "epoch": 4.107667210440456,
      "grad_norm": 3.082364559173584,
      "learning_rate": 3.667892866185064e-05,
      "loss": 0.1118,
      "num_input_tokens_seen": 54368816,
      "step": 25180
    },
    {
      "epoch": 4.108482871125612,
      "grad_norm": 0.3306456208229065,
      "learning_rate": 3.6672634651539205e-05,
      "loss": 0.0559,
      "num_input_tokens_seen": 54380848,
      "step": 25185
    },
    {
      "epoch": 4.109298531810767,
      "grad_norm": 0.9651033878326416,
      "learning_rate": 3.6666339694984785e-05,
      "loss": 0.0198,
      "num_input_tokens_seen": 54391760,
      "step": 25190
    },
    {
      "epoch": 4.110114192495922,
      "grad_norm": 0.12155020982027054,
      "learning_rate": 3.666004379269766e-05,
      "loss": 0.2691,
      "num_input_tokens_seen": 54402384,
      "step": 25195
    },
    {
      "epoch": 4.1109298531810765,
      "grad_norm": 0.10422401875257492,
      "learning_rate": 3.665374694518824e-05,
      "loss": 0.0097,
      "num_input_tokens_seen": 54413680,
      "step": 25200
    },
    {
      "epoch": 4.111745513866231,
      "grad_norm": 2.983114242553711,
      "learning_rate": 3.664744915296695e-05,
      "loss": 0.0988,
      "num_input_tokens_seen": 54422640,
      "step": 25205
    },
    {
      "epoch": 4.112561174551387,
      "grad_norm": 3.5289039611816406,
      "learning_rate": 3.664115041654434e-05,
      "loss": 0.2287,
      "num_input_tokens_seen": 54434576,
      "step": 25210
    },
    {
      "epoch": 4.113376835236542,
      "grad_norm": 0.06660588830709457,
      "learning_rate": 3.663485073643102e-05,
      "loss": 0.0055,
      "num_input_tokens_seen": 54444272,
      "step": 25215
    },
    {
      "epoch": 4.114192495921697,
      "grad_norm": 1.9301166534423828,
      "learning_rate": 3.6628550113137635e-05,
      "loss": 0.0347,
      "num_input_tokens_seen": 54456016,
      "step": 25220
    },
    {
      "epoch": 4.1150081566068515,
      "grad_norm": 0.05434871092438698,
      "learning_rate": 3.6622248547175e-05,
      "loss": 0.0046,
      "num_input_tokens_seen": 54466448,
      "step": 25225
    },
    {
      "epoch": 4.115823817292006,
      "grad_norm": 1.4604432582855225,
      "learning_rate": 3.661594603905392e-05,
      "loss": 0.1336,
      "num_input_tokens_seen": 54476592,
      "step": 25230
    },
    {
      "epoch": 4.116639477977161,
      "grad_norm": 0.024946851655840874,
      "learning_rate": 3.660964258928532e-05,
      "loss": 0.1005,
      "num_input_tokens_seen": 54488048,
      "step": 25235
    },
    {
      "epoch": 4.117455138662317,
      "grad_norm": 0.168961301445961,
      "learning_rate": 3.660333819838018e-05,
      "loss": 0.0691,
      "num_input_tokens_seen": 54499120,
      "step": 25240
    },
    {
      "epoch": 4.118270799347472,
      "grad_norm": 0.37382256984710693,
      "learning_rate": 3.659703286684957e-05,
      "loss": 0.1189,
      "num_input_tokens_seen": 54510576,
      "step": 25245
    },
    {
      "epoch": 4.1190864600326265,
      "grad_norm": 0.09253478050231934,
      "learning_rate": 3.659072659520463e-05,
      "loss": 0.1668,
      "num_input_tokens_seen": 54522160,
      "step": 25250
    },
    {
      "epoch": 4.119902120717781,
      "grad_norm": 0.05152549222111702,
      "learning_rate": 3.658441938395659e-05,
      "loss": 0.0054,
      "num_input_tokens_seen": 54533872,
      "step": 25255
    },
    {
      "epoch": 4.120717781402936,
      "grad_norm": 2.148808002471924,
      "learning_rate": 3.6578111233616726e-05,
      "loss": 0.1147,
      "num_input_tokens_seen": 54545200,
      "step": 25260
    },
    {
      "epoch": 4.121533442088092,
      "grad_norm": 0.0854240134358406,
      "learning_rate": 3.657180214469643e-05,
      "loss": 0.1078,
      "num_input_tokens_seen": 54554960,
      "step": 25265
    },
    {
      "epoch": 4.122349102773247,
      "grad_norm": 0.11657576262950897,
      "learning_rate": 3.656549211770713e-05,
      "loss": 0.1865,
      "num_input_tokens_seen": 54565456,
      "step": 25270
    },
    {
      "epoch": 4.123164763458401,
      "grad_norm": 0.14645171165466309,
      "learning_rate": 3.655918115316036e-05,
      "loss": 0.0936,
      "num_input_tokens_seen": 54575536,
      "step": 25275
    },
    {
      "epoch": 4.123980424143556,
      "grad_norm": 0.08738540858030319,
      "learning_rate": 3.655286925156772e-05,
      "loss": 0.0146,
      "num_input_tokens_seen": 54585296,
      "step": 25280
    },
    {
      "epoch": 4.124796084828711,
      "grad_norm": 0.14348307251930237,
      "learning_rate": 3.654655641344087e-05,
      "loss": 0.181,
      "num_input_tokens_seen": 54595600,
      "step": 25285
    },
    {
      "epoch": 4.125611745513866,
      "grad_norm": 0.10797704011201859,
      "learning_rate": 3.654024263929157e-05,
      "loss": 0.0894,
      "num_input_tokens_seen": 54606160,
      "step": 25290
    },
    {
      "epoch": 4.126427406199022,
      "grad_norm": 0.2273145169019699,
      "learning_rate": 3.653392792963165e-05,
      "loss": 0.0373,
      "num_input_tokens_seen": 54617232,
      "step": 25295
    },
    {
      "epoch": 4.127243066884176,
      "grad_norm": 4.387463569641113,
      "learning_rate": 3.652761228497301e-05,
      "loss": 0.1154,
      "num_input_tokens_seen": 54628688,
      "step": 25300
    },
    {
      "epoch": 4.128058727569331,
      "grad_norm": 0.1348361074924469,
      "learning_rate": 3.652129570582763e-05,
      "loss": 0.1604,
      "num_input_tokens_seen": 54640016,
      "step": 25305
    },
    {
      "epoch": 4.128874388254486,
      "grad_norm": 0.07359375804662704,
      "learning_rate": 3.651497819270756e-05,
      "loss": 0.0725,
      "num_input_tokens_seen": 54648848,
      "step": 25310
    },
    {
      "epoch": 4.129690048939641,
      "grad_norm": 6.487445831298828,
      "learning_rate": 3.650865974612493e-05,
      "loss": 0.1315,
      "num_input_tokens_seen": 54659376,
      "step": 25315
    },
    {
      "epoch": 4.130505709624796,
      "grad_norm": 0.057771340012550354,
      "learning_rate": 3.650234036659195e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 54670928,
      "step": 25320
    },
    {
      "epoch": 4.131321370309951,
      "grad_norm": 0.08444222062826157,
      "learning_rate": 3.649602005462089e-05,
      "loss": 0.1,
      "num_input_tokens_seen": 54681584,
      "step": 25325
    },
    {
      "epoch": 4.132137030995106,
      "grad_norm": 0.08515111356973648,
      "learning_rate": 3.648969881072412e-05,
      "loss": 0.1271,
      "num_input_tokens_seen": 54692432,
      "step": 25330
    },
    {
      "epoch": 4.132952691680261,
      "grad_norm": 0.14698028564453125,
      "learning_rate": 3.648337663541407e-05,
      "loss": 0.1074,
      "num_input_tokens_seen": 54703152,
      "step": 25335
    },
    {
      "epoch": 4.133768352365416,
      "grad_norm": 0.1383892446756363,
      "learning_rate": 3.647705352920324e-05,
      "loss": 0.0131,
      "num_input_tokens_seen": 54714032,
      "step": 25340
    },
    {
      "epoch": 4.134584013050571,
      "grad_norm": 0.27039095759391785,
      "learning_rate": 3.647072949260422e-05,
      "loss": 0.0114,
      "num_input_tokens_seen": 54724272,
      "step": 25345
    },
    {
      "epoch": 4.135399673735726,
      "grad_norm": 0.4326794743537903,
      "learning_rate": 3.646440452612965e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 54734864,
      "step": 25350
    },
    {
      "epoch": 4.136215334420881,
      "grad_norm": 0.07697264105081558,
      "learning_rate": 3.645807863029229e-05,
      "loss": 0.2007,
      "num_input_tokens_seen": 54744464,
      "step": 25355
    },
    {
      "epoch": 4.137030995106036,
      "grad_norm": 7.296753406524658,
      "learning_rate": 3.645175180560495e-05,
      "loss": 0.2002,
      "num_input_tokens_seen": 54754672,
      "step": 25360
    },
    {
      "epoch": 4.137846655791191,
      "grad_norm": 0.060346946120262146,
      "learning_rate": 3.644542405258049e-05,
      "loss": 0.1618,
      "num_input_tokens_seen": 54766256,
      "step": 25365
    },
    {
      "epoch": 4.138662316476346,
      "grad_norm": 0.06213405355811119,
      "learning_rate": 3.643909537173188e-05,
      "loss": 0.0759,
      "num_input_tokens_seen": 54777296,
      "step": 25370
    },
    {
      "epoch": 4.1394779771615005,
      "grad_norm": 2.9957454204559326,
      "learning_rate": 3.643276576357216e-05,
      "loss": 0.2049,
      "num_input_tokens_seen": 54787728,
      "step": 25375
    },
    {
      "epoch": 4.140293637846656,
      "grad_norm": 0.13104800879955292,
      "learning_rate": 3.642643522861444e-05,
      "loss": 0.1201,
      "num_input_tokens_seen": 54798864,
      "step": 25380
    },
    {
      "epoch": 4.141109298531811,
      "grad_norm": 0.057072713971138,
      "learning_rate": 3.642010376737191e-05,
      "loss": 0.0628,
      "num_input_tokens_seen": 54809872,
      "step": 25385
    },
    {
      "epoch": 4.141924959216966,
      "grad_norm": 0.07199932634830475,
      "learning_rate": 3.641377138035782e-05,
      "loss": 0.1763,
      "num_input_tokens_seen": 54819248,
      "step": 25390
    },
    {
      "epoch": 4.142740619902121,
      "grad_norm": 0.149081289768219,
      "learning_rate": 3.640743806808551e-05,
      "loss": 0.0838,
      "num_input_tokens_seen": 54830448,
      "step": 25395
    },
    {
      "epoch": 4.143556280587275,
      "grad_norm": 0.13100646436214447,
      "learning_rate": 3.640110383106838e-05,
      "loss": 0.0525,
      "num_input_tokens_seen": 54841776,
      "step": 25400
    },
    {
      "epoch": 4.14437194127243,
      "grad_norm": 3.682749032974243,
      "learning_rate": 3.639476866981993e-05,
      "loss": 0.1583,
      "num_input_tokens_seen": 54851888,
      "step": 25405
    },
    {
      "epoch": 4.145187601957586,
      "grad_norm": 4.476661682128906,
      "learning_rate": 3.638843258485372e-05,
      "loss": 0.2497,
      "num_input_tokens_seen": 54861968,
      "step": 25410
    },
    {
      "epoch": 4.146003262642741,
      "grad_norm": 0.06056198850274086,
      "learning_rate": 3.638209557668337e-05,
      "loss": 0.0421,
      "num_input_tokens_seen": 54873104,
      "step": 25415
    },
    {
      "epoch": 4.146818923327896,
      "grad_norm": 0.10154382139444351,
      "learning_rate": 3.637575764582261e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 54883600,
      "step": 25420
    },
    {
      "epoch": 4.14763458401305,
      "grad_norm": 0.06345032900571823,
      "learning_rate": 3.636941879278522e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 54895184,
      "step": 25425
    },
    {
      "epoch": 4.148450244698205,
      "grad_norm": 0.2177605926990509,
      "learning_rate": 3.636307901808504e-05,
      "loss": 0.0904,
      "num_input_tokens_seen": 54906640,
      "step": 25430
    },
    {
      "epoch": 4.149265905383361,
      "grad_norm": 0.16488832235336304,
      "learning_rate": 3.635673832223603e-05,
      "loss": 0.1137,
      "num_input_tokens_seen": 54918192,
      "step": 25435
    },
    {
      "epoch": 4.150081566068516,
      "grad_norm": 0.16648943722248077,
      "learning_rate": 3.635039670575218e-05,
      "loss": 0.1023,
      "num_input_tokens_seen": 54927536,
      "step": 25440
    },
    {
      "epoch": 4.150897226753671,
      "grad_norm": 0.2220931351184845,
      "learning_rate": 3.6344054169147584e-05,
      "loss": 0.1194,
      "num_input_tokens_seen": 54938224,
      "step": 25445
    },
    {
      "epoch": 4.151712887438825,
      "grad_norm": 7.944691181182861,
      "learning_rate": 3.63377107129364e-05,
      "loss": 0.0566,
      "num_input_tokens_seen": 54949104,
      "step": 25450
    },
    {
      "epoch": 4.15252854812398,
      "grad_norm": 0.07716577500104904,
      "learning_rate": 3.633136633763286e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 54960848,
      "step": 25455
    },
    {
      "epoch": 4.153344208809135,
      "grad_norm": 5.695812702178955,
      "learning_rate": 3.632502104375127e-05,
      "loss": 0.1043,
      "num_input_tokens_seen": 54971600,
      "step": 25460
    },
    {
      "epoch": 4.154159869494291,
      "grad_norm": 0.19582857191562653,
      "learning_rate": 3.6318674831806e-05,
      "loss": 0.0072,
      "num_input_tokens_seen": 54981936,
      "step": 25465
    },
    {
      "epoch": 4.1549755301794455,
      "grad_norm": 3.1364433765411377,
      "learning_rate": 3.6312327702311536e-05,
      "loss": 0.0413,
      "num_input_tokens_seen": 54993552,
      "step": 25470
    },
    {
      "epoch": 4.1557911908646,
      "grad_norm": 0.0790497288107872,
      "learning_rate": 3.630597965578238e-05,
      "loss": 0.024,
      "num_input_tokens_seen": 55005200,
      "step": 25475
    },
    {
      "epoch": 4.156606851549755,
      "grad_norm": 0.11226534843444824,
      "learning_rate": 3.629963069273315e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 55016464,
      "step": 25480
    },
    {
      "epoch": 4.15742251223491,
      "grad_norm": 0.11973495781421661,
      "learning_rate": 3.6293280813678523e-05,
      "loss": 0.1225,
      "num_input_tokens_seen": 55026576,
      "step": 25485
    },
    {
      "epoch": 4.158238172920065,
      "grad_norm": 0.024105289950966835,
      "learning_rate": 3.628693001913325e-05,
      "loss": 0.0865,
      "num_input_tokens_seen": 55037904,
      "step": 25490
    },
    {
      "epoch": 4.1590538336052205,
      "grad_norm": 0.09786190837621689,
      "learning_rate": 3.6280578309612165e-05,
      "loss": 0.0863,
      "num_input_tokens_seen": 55047408,
      "step": 25495
    },
    {
      "epoch": 4.159869494290375,
      "grad_norm": 6.50067138671875,
      "learning_rate": 3.6274225685630156e-05,
      "loss": 0.2055,
      "num_input_tokens_seen": 55058064,
      "step": 25500
    },
    {
      "epoch": 4.16068515497553,
      "grad_norm": 0.45748159289360046,
      "learning_rate": 3.626787214770221e-05,
      "loss": 0.0278,
      "num_input_tokens_seen": 55069008,
      "step": 25505
    },
    {
      "epoch": 4.161500815660685,
      "grad_norm": 3.031940460205078,
      "learning_rate": 3.626151769634338e-05,
      "loss": 0.2087,
      "num_input_tokens_seen": 55079792,
      "step": 25510
    },
    {
      "epoch": 4.16231647634584,
      "grad_norm": 2.2870099544525146,
      "learning_rate": 3.6255162332068785e-05,
      "loss": 0.1916,
      "num_input_tokens_seen": 55090864,
      "step": 25515
    },
    {
      "epoch": 4.1631321370309955,
      "grad_norm": 0.695946455001831,
      "learning_rate": 3.624880605539362e-05,
      "loss": 0.1228,
      "num_input_tokens_seen": 55101904,
      "step": 25520
    },
    {
      "epoch": 4.16394779771615,
      "grad_norm": 5.246313571929932,
      "learning_rate": 3.6242448866833164e-05,
      "loss": 0.0776,
      "num_input_tokens_seen": 55112656,
      "step": 25525
    },
    {
      "epoch": 4.164763458401305,
      "grad_norm": 0.04840467497706413,
      "learning_rate": 3.623609076690275e-05,
      "loss": 0.2569,
      "num_input_tokens_seen": 55123024,
      "step": 25530
    },
    {
      "epoch": 4.16557911908646,
      "grad_norm": 0.11822468042373657,
      "learning_rate": 3.622973175611781e-05,
      "loss": 0.2145,
      "num_input_tokens_seen": 55133136,
      "step": 25535
    },
    {
      "epoch": 4.166394779771615,
      "grad_norm": 0.36142075061798096,
      "learning_rate": 3.622337183499384e-05,
      "loss": 0.062,
      "num_input_tokens_seen": 55142864,
      "step": 25540
    },
    {
      "epoch": 4.16721044045677,
      "grad_norm": 0.546535849571228,
      "learning_rate": 3.6217011004046404e-05,
      "loss": 0.06,
      "num_input_tokens_seen": 55152848,
      "step": 25545
    },
    {
      "epoch": 4.168026101141925,
      "grad_norm": 0.5461816191673279,
      "learning_rate": 3.621064926379114e-05,
      "loss": 0.1557,
      "num_input_tokens_seen": 55162032,
      "step": 25550
    },
    {
      "epoch": 4.16884176182708,
      "grad_norm": 0.1468852460384369,
      "learning_rate": 3.620428661474377e-05,
      "loss": 0.1167,
      "num_input_tokens_seen": 55173616,
      "step": 25555
    },
    {
      "epoch": 4.169657422512235,
      "grad_norm": 0.1524345725774765,
      "learning_rate": 3.619792305742006e-05,
      "loss": 0.098,
      "num_input_tokens_seen": 55185328,
      "step": 25560
    },
    {
      "epoch": 4.17047308319739,
      "grad_norm": 2.2733712196350098,
      "learning_rate": 3.619155859233589e-05,
      "loss": 0.1251,
      "num_input_tokens_seen": 55197072,
      "step": 25565
    },
    {
      "epoch": 4.171288743882545,
      "grad_norm": 0.15534906089305878,
      "learning_rate": 3.6185193220007214e-05,
      "loss": 0.0061,
      "num_input_tokens_seen": 55205936,
      "step": 25570
    },
    {
      "epoch": 4.1721044045677,
      "grad_norm": 1.2479537725448608,
      "learning_rate": 3.617882694095001e-05,
      "loss": 0.0121,
      "num_input_tokens_seen": 55216560,
      "step": 25575
    },
    {
      "epoch": 4.172920065252855,
      "grad_norm": 0.22745876014232635,
      "learning_rate": 3.617245975568038e-05,
      "loss": 0.0366,
      "num_input_tokens_seen": 55227504,
      "step": 25580
    },
    {
      "epoch": 4.17373572593801,
      "grad_norm": 0.9728760719299316,
      "learning_rate": 3.616609166471447e-05,
      "loss": 0.2422,
      "num_input_tokens_seen": 55238032,
      "step": 25585
    },
    {
      "epoch": 4.174551386623165,
      "grad_norm": 0.10870423913002014,
      "learning_rate": 3.615972266856851e-05,
      "loss": 0.0759,
      "num_input_tokens_seen": 55248080,
      "step": 25590
    },
    {
      "epoch": 4.1753670473083195,
      "grad_norm": 3.1872990131378174,
      "learning_rate": 3.6153352767758816e-05,
      "loss": 0.1896,
      "num_input_tokens_seen": 55259760,
      "step": 25595
    },
    {
      "epoch": 4.176182707993474,
      "grad_norm": 0.43499961495399475,
      "learning_rate": 3.6146981962801744e-05,
      "loss": 0.0128,
      "num_input_tokens_seen": 55270128,
      "step": 25600
    },
    {
      "epoch": 4.17699836867863,
      "grad_norm": 0.24309612810611725,
      "learning_rate": 3.6140610254213756e-05,
      "loss": 0.1526,
      "num_input_tokens_seen": 55280592,
      "step": 25605
    },
    {
      "epoch": 4.177814029363785,
      "grad_norm": 0.16260983049869537,
      "learning_rate": 3.613423764251138e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 55291344,
      "step": 25610
    },
    {
      "epoch": 4.17862969004894,
      "grad_norm": 2.815412998199463,
      "learning_rate": 3.61278641282112e-05,
      "loss": 0.112,
      "num_input_tokens_seen": 55301360,
      "step": 25615
    },
    {
      "epoch": 4.1794453507340945,
      "grad_norm": 11.62366008758545,
      "learning_rate": 3.612148971182989e-05,
      "loss": 0.0714,
      "num_input_tokens_seen": 55311696,
      "step": 25620
    },
    {
      "epoch": 4.180261011419249,
      "grad_norm": 0.08460749685764313,
      "learning_rate": 3.6115114393884206e-05,
      "loss": 0.1437,
      "num_input_tokens_seen": 55322704,
      "step": 25625
    },
    {
      "epoch": 4.181076672104404,
      "grad_norm": 3.9669132232666016,
      "learning_rate": 3.6108738174890944e-05,
      "loss": 0.0878,
      "num_input_tokens_seen": 55334704,
      "step": 25630
    },
    {
      "epoch": 4.18189233278956,
      "grad_norm": 0.20038148760795593,
      "learning_rate": 3.6102361055367e-05,
      "loss": 0.1143,
      "num_input_tokens_seen": 55344912,
      "step": 25635
    },
    {
      "epoch": 4.182707993474715,
      "grad_norm": 0.2454468011856079,
      "learning_rate": 3.609598303582934e-05,
      "loss": 0.1271,
      "num_input_tokens_seen": 55354832,
      "step": 25640
    },
    {
      "epoch": 4.1835236541598695,
      "grad_norm": 0.14531424641609192,
      "learning_rate": 3.608960411679499e-05,
      "loss": 0.0318,
      "num_input_tokens_seen": 55365776,
      "step": 25645
    },
    {
      "epoch": 4.184339314845024,
      "grad_norm": 0.06752365082502365,
      "learning_rate": 3.608322429878107e-05,
      "loss": 0.2215,
      "num_input_tokens_seen": 55376208,
      "step": 25650
    },
    {
      "epoch": 4.185154975530179,
      "grad_norm": 2.8704254627227783,
      "learning_rate": 3.6076843582304744e-05,
      "loss": 0.2243,
      "num_input_tokens_seen": 55386512,
      "step": 25655
    },
    {
      "epoch": 4.185970636215335,
      "grad_norm": 0.10375895351171494,
      "learning_rate": 3.607046196788328e-05,
      "loss": 0.0199,
      "num_input_tokens_seen": 55397168,
      "step": 25660
    },
    {
      "epoch": 4.18678629690049,
      "grad_norm": 0.07490267604589462,
      "learning_rate": 3.6064079456033996e-05,
      "loss": 0.2718,
      "num_input_tokens_seen": 55408912,
      "step": 25665
    },
    {
      "epoch": 4.1876019575856445,
      "grad_norm": 0.18197505176067352,
      "learning_rate": 3.6057696047274285e-05,
      "loss": 0.1544,
      "num_input_tokens_seen": 55418672,
      "step": 25670
    },
    {
      "epoch": 4.188417618270799,
      "grad_norm": 3.714529275894165,
      "learning_rate": 3.605131174212164e-05,
      "loss": 0.0537,
      "num_input_tokens_seen": 55430768,
      "step": 25675
    },
    {
      "epoch": 4.189233278955954,
      "grad_norm": 3.1004531383514404,
      "learning_rate": 3.604492654109357e-05,
      "loss": 0.09,
      "num_input_tokens_seen": 55441232,
      "step": 25680
    },
    {
      "epoch": 4.190048939641109,
      "grad_norm": 4.810756206512451,
      "learning_rate": 3.6038540444707734e-05,
      "loss": 0.1299,
      "num_input_tokens_seen": 55453040,
      "step": 25685
    },
    {
      "epoch": 4.190864600326265,
      "grad_norm": 0.07222369313240051,
      "learning_rate": 3.603215345348179e-05,
      "loss": 0.0728,
      "num_input_tokens_seen": 55462768,
      "step": 25690
    },
    {
      "epoch": 4.191680261011419,
      "grad_norm": 0.09615682065486908,
      "learning_rate": 3.602576556793352e-05,
      "loss": 0.03,
      "num_input_tokens_seen": 55474416,
      "step": 25695
    },
    {
      "epoch": 4.192495921696574,
      "grad_norm": 0.11471203714609146,
      "learning_rate": 3.601937678858074e-05,
      "loss": 0.0447,
      "num_input_tokens_seen": 55484272,
      "step": 25700
    },
    {
      "epoch": 4.193311582381729,
      "grad_norm": 0.2656436562538147,
      "learning_rate": 3.601298711594137e-05,
      "loss": 0.0209,
      "num_input_tokens_seen": 55496336,
      "step": 25705
    },
    {
      "epoch": 4.194127243066884,
      "grad_norm": 0.33770278096199036,
      "learning_rate": 3.6006596550533385e-05,
      "loss": 0.0088,
      "num_input_tokens_seen": 55508656,
      "step": 25710
    },
    {
      "epoch": 4.19494290375204,
      "grad_norm": 9.408333778381348,
      "learning_rate": 3.600020509287483e-05,
      "loss": 0.3765,
      "num_input_tokens_seen": 55519664,
      "step": 25715
    },
    {
      "epoch": 4.195758564437194,
      "grad_norm": 0.11777154356241226,
      "learning_rate": 3.599381274348385e-05,
      "loss": 0.1029,
      "num_input_tokens_seen": 55529744,
      "step": 25720
    },
    {
      "epoch": 4.196574225122349,
      "grad_norm": 0.18434153497219086,
      "learning_rate": 3.598741950287861e-05,
      "loss": 0.0072,
      "num_input_tokens_seen": 55540400,
      "step": 25725
    },
    {
      "epoch": 4.197389885807504,
      "grad_norm": 3.3441872596740723,
      "learning_rate": 3.5981025371577404e-05,
      "loss": 0.2807,
      "num_input_tokens_seen": 55551312,
      "step": 25730
    },
    {
      "epoch": 4.198205546492659,
      "grad_norm": 0.09962259978055954,
      "learning_rate": 3.5974630350098566e-05,
      "loss": 0.0252,
      "num_input_tokens_seen": 55562448,
      "step": 25735
    },
    {
      "epoch": 4.199021207177814,
      "grad_norm": 0.08699111640453339,
      "learning_rate": 3.5968234438960505e-05,
      "loss": 0.0126,
      "num_input_tokens_seen": 55572848,
      "step": 25740
    },
    {
      "epoch": 4.199836867862969,
      "grad_norm": 0.12356137484312057,
      "learning_rate": 3.5961837638681714e-05,
      "loss": 0.1189,
      "num_input_tokens_seen": 55582800,
      "step": 25745
    },
    {
      "epoch": 4.200652528548124,
      "grad_norm": 0.30556872487068176,
      "learning_rate": 3.595543994978073e-05,
      "loss": 0.1044,
      "num_input_tokens_seen": 55594096,
      "step": 25750
    },
    {
      "epoch": 4.201468189233279,
      "grad_norm": 0.13969768583774567,
      "learning_rate": 3.594904137277621e-05,
      "loss": 0.0817,
      "num_input_tokens_seen": 55605456,
      "step": 25755
    },
    {
      "epoch": 4.202283849918434,
      "grad_norm": 0.06743574142456055,
      "learning_rate": 3.594264190818683e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 55617072,
      "step": 25760
    },
    {
      "epoch": 4.203099510603589,
      "grad_norm": 0.10467995703220367,
      "learning_rate": 3.593624155653138e-05,
      "loss": 0.256,
      "num_input_tokens_seen": 55628176,
      "step": 25765
    },
    {
      "epoch": 4.2039151712887435,
      "grad_norm": 0.0721723809838295,
      "learning_rate": 3.592984031832871e-05,
      "loss": 0.0999,
      "num_input_tokens_seen": 55637744,
      "step": 25770
    },
    {
      "epoch": 4.204730831973899,
      "grad_norm": 0.09203427284955978,
      "learning_rate": 3.5923438194097715e-05,
      "loss": 0.0045,
      "num_input_tokens_seen": 55649104,
      "step": 25775
    },
    {
      "epoch": 4.205546492659054,
      "grad_norm": 11.121380805969238,
      "learning_rate": 3.591703518435739e-05,
      "loss": 0.1183,
      "num_input_tokens_seen": 55660272,
      "step": 25780
    },
    {
      "epoch": 4.206362153344209,
      "grad_norm": 0.15184751152992249,
      "learning_rate": 3.591063128962681e-05,
      "loss": 0.1472,
      "num_input_tokens_seen": 55670864,
      "step": 25785
    },
    {
      "epoch": 4.207177814029364,
      "grad_norm": 2.250476360321045,
      "learning_rate": 3.5904226510425095e-05,
      "loss": 0.0418,
      "num_input_tokens_seen": 55681456,
      "step": 25790
    },
    {
      "epoch": 4.2079934747145185,
      "grad_norm": 0.5987846851348877,
      "learning_rate": 3.5897820847271446e-05,
      "loss": 0.0123,
      "num_input_tokens_seen": 55692464,
      "step": 25795
    },
    {
      "epoch": 4.208809135399674,
      "grad_norm": 0.22947758436203003,
      "learning_rate": 3.5891414300685155e-05,
      "loss": 0.1404,
      "num_input_tokens_seen": 55702704,
      "step": 25800
    },
    {
      "epoch": 4.209624796084829,
      "grad_norm": 0.05595271289348602,
      "learning_rate": 3.588500687118555e-05,
      "loss": 0.0833,
      "num_input_tokens_seen": 55713040,
      "step": 25805
    },
    {
      "epoch": 4.210440456769984,
      "grad_norm": 0.13147498667240143,
      "learning_rate": 3.587859855929207e-05,
      "loss": 0.0299,
      "num_input_tokens_seen": 55723920,
      "step": 25810
    },
    {
      "epoch": 4.211256117455139,
      "grad_norm": 19.015289306640625,
      "learning_rate": 3.5872189365524175e-05,
      "loss": 0.0729,
      "num_input_tokens_seen": 55735440,
      "step": 25815
    },
    {
      "epoch": 4.212071778140293,
      "grad_norm": 0.1084061786532402,
      "learning_rate": 3.586577929040146e-05,
      "loss": 0.1183,
      "num_input_tokens_seen": 55745680,
      "step": 25820
    },
    {
      "epoch": 4.212887438825448,
      "grad_norm": 2.7874019145965576,
      "learning_rate": 3.5859368334443536e-05,
      "loss": 0.1077,
      "num_input_tokens_seen": 55757680,
      "step": 25825
    },
    {
      "epoch": 4.213703099510604,
      "grad_norm": 3.409447193145752,
      "learning_rate": 3.585295649817011e-05,
      "loss": 0.13,
      "num_input_tokens_seen": 55767536,
      "step": 25830
    },
    {
      "epoch": 4.214518760195759,
      "grad_norm": 0.1951976865530014,
      "learning_rate": 3.5846543782100974e-05,
      "loss": 0.1067,
      "num_input_tokens_seen": 55778032,
      "step": 25835
    },
    {
      "epoch": 4.215334420880914,
      "grad_norm": 0.08985646069049835,
      "learning_rate": 3.584013018675596e-05,
      "loss": 0.0909,
      "num_input_tokens_seen": 55790000,
      "step": 25840
    },
    {
      "epoch": 4.216150081566068,
      "grad_norm": 0.09971218556165695,
      "learning_rate": 3.583371571265498e-05,
      "loss": 0.1387,
      "num_input_tokens_seen": 55800688,
      "step": 25845
    },
    {
      "epoch": 4.216965742251223,
      "grad_norm": 0.10093860328197479,
      "learning_rate": 3.582730036031805e-05,
      "loss": 0.0071,
      "num_input_tokens_seen": 55812208,
      "step": 25850
    },
    {
      "epoch": 4.217781402936378,
      "grad_norm": 3.3075568675994873,
      "learning_rate": 3.582088413026521e-05,
      "loss": 0.1078,
      "num_input_tokens_seen": 55823120,
      "step": 25855
    },
    {
      "epoch": 4.218597063621534,
      "grad_norm": 0.04948318377137184,
      "learning_rate": 3.581446702301659e-05,
      "loss": 0.0178,
      "num_input_tokens_seen": 55834416,
      "step": 25860
    },
    {
      "epoch": 4.219412724306689,
      "grad_norm": 0.09975404292345047,
      "learning_rate": 3.5808049039092414e-05,
      "loss": 0.1671,
      "num_input_tokens_seen": 55846480,
      "step": 25865
    },
    {
      "epoch": 4.220228384991843,
      "grad_norm": 2.3931007385253906,
      "learning_rate": 3.580163017901295e-05,
      "loss": 0.1215,
      "num_input_tokens_seen": 55857200,
      "step": 25870
    },
    {
      "epoch": 4.221044045676998,
      "grad_norm": 0.10142625123262405,
      "learning_rate": 3.579521044329852e-05,
      "loss": 0.1154,
      "num_input_tokens_seen": 55868976,
      "step": 25875
    },
    {
      "epoch": 4.221859706362153,
      "grad_norm": 13.955004692077637,
      "learning_rate": 3.578878983246956e-05,
      "loss": 0.1048,
      "num_input_tokens_seen": 55881360,
      "step": 25880
    },
    {
      "epoch": 4.222675367047309,
      "grad_norm": 0.5234656929969788,
      "learning_rate": 3.578236834704656e-05,
      "loss": 0.1197,
      "num_input_tokens_seen": 55892208,
      "step": 25885
    },
    {
      "epoch": 4.2234910277324635,
      "grad_norm": 0.026215214282274246,
      "learning_rate": 3.577594598755006e-05,
      "loss": 0.0956,
      "num_input_tokens_seen": 55901936,
      "step": 25890
    },
    {
      "epoch": 4.224306688417618,
      "grad_norm": 4.179839611053467,
      "learning_rate": 3.5769522754500714e-05,
      "loss": 0.0344,
      "num_input_tokens_seen": 55912784,
      "step": 25895
    },
    {
      "epoch": 4.225122349102773,
      "grad_norm": 0.2951279282569885,
      "learning_rate": 3.5763098648419216e-05,
      "loss": 0.2639,
      "num_input_tokens_seen": 55924432,
      "step": 25900
    },
    {
      "epoch": 4.225938009787928,
      "grad_norm": 0.055586107075214386,
      "learning_rate": 3.575667366982631e-05,
      "loss": 0.1581,
      "num_input_tokens_seen": 55935216,
      "step": 25905
    },
    {
      "epoch": 4.226753670473083,
      "grad_norm": 0.04721270874142647,
      "learning_rate": 3.575024781924288e-05,
      "loss": 0.0291,
      "num_input_tokens_seen": 55945680,
      "step": 25910
    },
    {
      "epoch": 4.2275693311582385,
      "grad_norm": 1.1868259906768799,
      "learning_rate": 3.574382109718979e-05,
      "loss": 0.1081,
      "num_input_tokens_seen": 55956496,
      "step": 25915
    },
    {
      "epoch": 4.228384991843393,
      "grad_norm": 0.051986824721097946,
      "learning_rate": 3.573739350418806e-05,
      "loss": 0.0312,
      "num_input_tokens_seen": 55967280,
      "step": 25920
    },
    {
      "epoch": 4.229200652528548,
      "grad_norm": 0.11333861202001572,
      "learning_rate": 3.573096504075874e-05,
      "loss": 0.0825,
      "num_input_tokens_seen": 55978160,
      "step": 25925
    },
    {
      "epoch": 4.230016313213703,
      "grad_norm": 0.6075248718261719,
      "learning_rate": 3.572453570742294e-05,
      "loss": 0.0778,
      "num_input_tokens_seen": 55989808,
      "step": 25930
    },
    {
      "epoch": 4.230831973898858,
      "grad_norm": 0.05753014609217644,
      "learning_rate": 3.571810550470186e-05,
      "loss": 0.1058,
      "num_input_tokens_seen": 56000944,
      "step": 25935
    },
    {
      "epoch": 4.231647634584013,
      "grad_norm": 0.0715099647641182,
      "learning_rate": 3.571167443311676e-05,
      "loss": 0.0819,
      "num_input_tokens_seen": 56011408,
      "step": 25940
    },
    {
      "epoch": 4.232463295269168,
      "grad_norm": 3.7070655822753906,
      "learning_rate": 3.5705242493188986e-05,
      "loss": 0.0146,
      "num_input_tokens_seen": 56020944,
      "step": 25945
    },
    {
      "epoch": 4.233278955954323,
      "grad_norm": 0.06861277669668198,
      "learning_rate": 3.569880968543994e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 56031792,
      "step": 25950
    },
    {
      "epoch": 4.234094616639478,
      "grad_norm": 0.4699649512767792,
      "learning_rate": 3.569237601039109e-05,
      "loss": 0.0058,
      "num_input_tokens_seen": 56041936,
      "step": 25955
    },
    {
      "epoch": 4.234910277324633,
      "grad_norm": 0.14664636552333832,
      "learning_rate": 3.5685941468563985e-05,
      "loss": 0.0046,
      "num_input_tokens_seen": 56052272,
      "step": 25960
    },
    {
      "epoch": 4.235725938009788,
      "grad_norm": 0.028544507920742035,
      "learning_rate": 3.567950606048025e-05,
      "loss": 0.0023,
      "num_input_tokens_seen": 56061808,
      "step": 25965
    },
    {
      "epoch": 4.236541598694943,
      "grad_norm": 3.6018576622009277,
      "learning_rate": 3.5673069786661566e-05,
      "loss": 0.2213,
      "num_input_tokens_seen": 56071536,
      "step": 25970
    },
    {
      "epoch": 4.237357259380098,
      "grad_norm": 0.07787901163101196,
      "learning_rate": 3.566663264762969e-05,
      "loss": 0.1086,
      "num_input_tokens_seen": 56083568,
      "step": 25975
    },
    {
      "epoch": 4.238172920065253,
      "grad_norm": 0.11781366169452667,
      "learning_rate": 3.5660194643906455e-05,
      "loss": 0.0746,
      "num_input_tokens_seen": 56094000,
      "step": 25980
    },
    {
      "epoch": 4.238988580750408,
      "grad_norm": 2.923859119415283,
      "learning_rate": 3.5653755776013745e-05,
      "loss": 0.3323,
      "num_input_tokens_seen": 56103952,
      "step": 25985
    },
    {
      "epoch": 4.239804241435563,
      "grad_norm": 0.15193922817707062,
      "learning_rate": 3.5647316044473537e-05,
      "loss": 0.0133,
      "num_input_tokens_seen": 56114224,
      "step": 25990
    },
    {
      "epoch": 4.240619902120717,
      "grad_norm": 0.07131417095661163,
      "learning_rate": 3.564087544980786e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 56124464,
      "step": 25995
    },
    {
      "epoch": 4.241435562805873,
      "grad_norm": 0.09576962888240814,
      "learning_rate": 3.563443399253883e-05,
      "loss": 0.0119,
      "num_input_tokens_seen": 56135280,
      "step": 26000
    },
    {
      "epoch": 4.242251223491028,
      "grad_norm": 0.1686936914920807,
      "learning_rate": 3.5627991673188624e-05,
      "loss": 0.1132,
      "num_input_tokens_seen": 56144976,
      "step": 26005
    },
    {
      "epoch": 4.243066884176183,
      "grad_norm": 0.8898986577987671,
      "learning_rate": 3.562154849227949e-05,
      "loss": 0.287,
      "num_input_tokens_seen": 56156688,
      "step": 26010
    },
    {
      "epoch": 4.2438825448613375,
      "grad_norm": 0.09941082447767258,
      "learning_rate": 3.561510445033375e-05,
      "loss": 0.0211,
      "num_input_tokens_seen": 56167856,
      "step": 26015
    },
    {
      "epoch": 4.244698205546492,
      "grad_norm": 3.588923692703247,
      "learning_rate": 3.560865954787377e-05,
      "loss": 0.0275,
      "num_input_tokens_seen": 56178928,
      "step": 26020
    },
    {
      "epoch": 4.245513866231648,
      "grad_norm": 0.07287617772817612,
      "learning_rate": 3.5602213785422025e-05,
      "loss": 0.1913,
      "num_input_tokens_seen": 56189584,
      "step": 26025
    },
    {
      "epoch": 4.246329526916803,
      "grad_norm": 0.20097781717777252,
      "learning_rate": 3.5595767163501034e-05,
      "loss": 0.1521,
      "num_input_tokens_seen": 56200080,
      "step": 26030
    },
    {
      "epoch": 4.247145187601958,
      "grad_norm": 0.06811130046844482,
      "learning_rate": 3.5589319682633393e-05,
      "loss": 0.2815,
      "num_input_tokens_seen": 56210288,
      "step": 26035
    },
    {
      "epoch": 4.2479608482871125,
      "grad_norm": 0.15266764163970947,
      "learning_rate": 3.558287134334177e-05,
      "loss": 0.0271,
      "num_input_tokens_seen": 56220976,
      "step": 26040
    },
    {
      "epoch": 4.248776508972267,
      "grad_norm": 6.053004264831543,
      "learning_rate": 3.55764221461489e-05,
      "loss": 0.1249,
      "num_input_tokens_seen": 56232176,
      "step": 26045
    },
    {
      "epoch": 4.249592169657422,
      "grad_norm": 1.8790841102600098,
      "learning_rate": 3.556997209157759e-05,
      "loss": 0.2825,
      "num_input_tokens_seen": 56244400,
      "step": 26050
    },
    {
      "epoch": 4.250407830342578,
      "grad_norm": 0.5246086716651917,
      "learning_rate": 3.5563521180150704e-05,
      "loss": 0.0094,
      "num_input_tokens_seen": 56255952,
      "step": 26055
    },
    {
      "epoch": 4.251223491027733,
      "grad_norm": 0.1442943662405014,
      "learning_rate": 3.55570694123912e-05,
      "loss": 0.0878,
      "num_input_tokens_seen": 56267760,
      "step": 26060
    },
    {
      "epoch": 4.2520391517128875,
      "grad_norm": 0.23565569519996643,
      "learning_rate": 3.5550616788822074e-05,
      "loss": 0.0115,
      "num_input_tokens_seen": 56278800,
      "step": 26065
    },
    {
      "epoch": 4.252854812398042,
      "grad_norm": 0.09448880702257156,
      "learning_rate": 3.5544163309966425e-05,
      "loss": 0.0604,
      "num_input_tokens_seen": 56289552,
      "step": 26070
    },
    {
      "epoch": 4.253670473083197,
      "grad_norm": 0.12274815142154694,
      "learning_rate": 3.5537708976347386e-05,
      "loss": 0.1721,
      "num_input_tokens_seen": 56301808,
      "step": 26075
    },
    {
      "epoch": 4.254486133768353,
      "grad_norm": 2.4918394088745117,
      "learning_rate": 3.55312537884882e-05,
      "loss": 0.1046,
      "num_input_tokens_seen": 56311536,
      "step": 26080
    },
    {
      "epoch": 4.255301794453508,
      "grad_norm": 0.09813707321882248,
      "learning_rate": 3.552479774691215e-05,
      "loss": 0.0328,
      "num_input_tokens_seen": 56323344,
      "step": 26085
    },
    {
      "epoch": 4.2561174551386625,
      "grad_norm": 1.4146684408187866,
      "learning_rate": 3.5518340852142587e-05,
      "loss": 0.0293,
      "num_input_tokens_seen": 56333904,
      "step": 26090
    },
    {
      "epoch": 4.256933115823817,
      "grad_norm": 1.7968381643295288,
      "learning_rate": 3.5511883104702943e-05,
      "loss": 0.0094,
      "num_input_tokens_seen": 56343312,
      "step": 26095
    },
    {
      "epoch": 4.257748776508972,
      "grad_norm": 0.16410556435585022,
      "learning_rate": 3.5505424505116714e-05,
      "loss": 0.0462,
      "num_input_tokens_seen": 56354960,
      "step": 26100
    },
    {
      "epoch": 4.258564437194127,
      "grad_norm": 0.13683778047561646,
      "learning_rate": 3.549896505390748e-05,
      "loss": 0.0776,
      "num_input_tokens_seen": 56366032,
      "step": 26105
    },
    {
      "epoch": 4.259380097879283,
      "grad_norm": 1.570995569229126,
      "learning_rate": 3.549250475159887e-05,
      "loss": 0.0721,
      "num_input_tokens_seen": 56376496,
      "step": 26110
    },
    {
      "epoch": 4.260195758564437,
      "grad_norm": 0.06962514668703079,
      "learning_rate": 3.5486043598714576e-05,
      "loss": 0.1436,
      "num_input_tokens_seen": 56387408,
      "step": 26115
    },
    {
      "epoch": 4.261011419249592,
      "grad_norm": 0.1006123498082161,
      "learning_rate": 3.547958159577839e-05,
      "loss": 0.1313,
      "num_input_tokens_seen": 56397328,
      "step": 26120
    },
    {
      "epoch": 4.261827079934747,
      "grad_norm": 0.11071991920471191,
      "learning_rate": 3.547311874331414e-05,
      "loss": 0.2202,
      "num_input_tokens_seen": 56406640,
      "step": 26125
    },
    {
      "epoch": 4.262642740619902,
      "grad_norm": 7.6907057762146,
      "learning_rate": 3.546665504184575e-05,
      "loss": 0.1149,
      "num_input_tokens_seen": 56416400,
      "step": 26130
    },
    {
      "epoch": 4.263458401305057,
      "grad_norm": 3.0809524059295654,
      "learning_rate": 3.5460190491897195e-05,
      "loss": 0.1143,
      "num_input_tokens_seen": 56426320,
      "step": 26135
    },
    {
      "epoch": 4.264274061990212,
      "grad_norm": 1.5005754232406616,
      "learning_rate": 3.5453725093992526e-05,
      "loss": 0.3749,
      "num_input_tokens_seen": 56436400,
      "step": 26140
    },
    {
      "epoch": 4.265089722675367,
      "grad_norm": 3.1108760833740234,
      "learning_rate": 3.544725884865585e-05,
      "loss": 0.25,
      "num_input_tokens_seen": 56445712,
      "step": 26145
    },
    {
      "epoch": 4.265905383360522,
      "grad_norm": 0.10910908132791519,
      "learning_rate": 3.544079175641137e-05,
      "loss": 0.1843,
      "num_input_tokens_seen": 56456720,
      "step": 26150
    },
    {
      "epoch": 4.266721044045677,
      "grad_norm": 0.19026920199394226,
      "learning_rate": 3.543432381778333e-05,
      "loss": 0.0665,
      "num_input_tokens_seen": 56468464,
      "step": 26155
    },
    {
      "epoch": 4.267536704730832,
      "grad_norm": 4.13557767868042,
      "learning_rate": 3.5427855033296056e-05,
      "loss": 0.2386,
      "num_input_tokens_seen": 56478096,
      "step": 26160
    },
    {
      "epoch": 4.268352365415987,
      "grad_norm": 10.79997730255127,
      "learning_rate": 3.542138540347395e-05,
      "loss": 0.4069,
      "num_input_tokens_seen": 56489776,
      "step": 26165
    },
    {
      "epoch": 4.269168026101142,
      "grad_norm": 0.19795987010002136,
      "learning_rate": 3.5414914928841467e-05,
      "loss": 0.0928,
      "num_input_tokens_seen": 56501232,
      "step": 26170
    },
    {
      "epoch": 4.269983686786297,
      "grad_norm": 0.12328500300645828,
      "learning_rate": 3.540844360992313e-05,
      "loss": 0.0106,
      "num_input_tokens_seen": 56511664,
      "step": 26175
    },
    {
      "epoch": 4.270799347471452,
      "grad_norm": 1.53931725025177,
      "learning_rate": 3.5401971447243545e-05,
      "loss": 0.0804,
      "num_input_tokens_seen": 56522032,
      "step": 26180
    },
    {
      "epoch": 4.271615008156607,
      "grad_norm": 0.07922205328941345,
      "learning_rate": 3.539549844132737e-05,
      "loss": 0.2395,
      "num_input_tokens_seen": 56533008,
      "step": 26185
    },
    {
      "epoch": 4.2724306688417615,
      "grad_norm": 0.5737140774726868,
      "learning_rate": 3.538902459269935e-05,
      "loss": 0.024,
      "num_input_tokens_seen": 56544176,
      "step": 26190
    },
    {
      "epoch": 4.273246329526917,
      "grad_norm": 0.715808093547821,
      "learning_rate": 3.538254990188429e-05,
      "loss": 0.1182,
      "num_input_tokens_seen": 56555376,
      "step": 26195
    },
    {
      "epoch": 4.274061990212072,
      "grad_norm": 0.2927914559841156,
      "learning_rate": 3.5376074369407044e-05,
      "loss": 0.0973,
      "num_input_tokens_seen": 56566256,
      "step": 26200
    },
    {
      "epoch": 4.274877650897227,
      "grad_norm": 2.412468194961548,
      "learning_rate": 3.536959799579256e-05,
      "loss": 0.1044,
      "num_input_tokens_seen": 56576976,
      "step": 26205
    },
    {
      "epoch": 4.275693311582382,
      "grad_norm": 0.12590470910072327,
      "learning_rate": 3.5363120781565854e-05,
      "loss": 0.0996,
      "num_input_tokens_seen": 56587504,
      "step": 26210
    },
    {
      "epoch": 4.2765089722675365,
      "grad_norm": 2.6490683555603027,
      "learning_rate": 3.535664272725199e-05,
      "loss": 0.1726,
      "num_input_tokens_seen": 56598800,
      "step": 26215
    },
    {
      "epoch": 4.277324632952691,
      "grad_norm": 1.593904733657837,
      "learning_rate": 3.5350163833376124e-05,
      "loss": 0.0986,
      "num_input_tokens_seen": 56609584,
      "step": 26220
    },
    {
      "epoch": 4.278140293637847,
      "grad_norm": 0.04505256563425064,
      "learning_rate": 3.534368410046346e-05,
      "loss": 0.0606,
      "num_input_tokens_seen": 56619984,
      "step": 26225
    },
    {
      "epoch": 4.278955954323002,
      "grad_norm": 0.25533008575439453,
      "learning_rate": 3.5337203529039275e-05,
      "loss": 0.1114,
      "num_input_tokens_seen": 56630128,
      "step": 26230
    },
    {
      "epoch": 4.279771615008157,
      "grad_norm": 7.802379131317139,
      "learning_rate": 3.533072211962892e-05,
      "loss": 0.1193,
      "num_input_tokens_seen": 56641264,
      "step": 26235
    },
    {
      "epoch": 4.280587275693311,
      "grad_norm": 0.22463050484657288,
      "learning_rate": 3.532423987275782e-05,
      "loss": 0.015,
      "num_input_tokens_seen": 56652688,
      "step": 26240
    },
    {
      "epoch": 4.281402936378466,
      "grad_norm": 2.565871477127075,
      "learning_rate": 3.531775678895145e-05,
      "loss": 0.1115,
      "num_input_tokens_seen": 56663600,
      "step": 26245
    },
    {
      "epoch": 4.282218597063622,
      "grad_norm": 0.06665221601724625,
      "learning_rate": 3.531127286873536e-05,
      "loss": 0.0191,
      "num_input_tokens_seen": 56674544,
      "step": 26250
    },
    {
      "epoch": 4.283034257748777,
      "grad_norm": 0.09177578240633011,
      "learning_rate": 3.530478811263518e-05,
      "loss": 0.0072,
      "num_input_tokens_seen": 56685680,
      "step": 26255
    },
    {
      "epoch": 4.283849918433932,
      "grad_norm": 3.6111679077148438,
      "learning_rate": 3.529830252117657e-05,
      "loss": 0.1132,
      "num_input_tokens_seen": 56696752,
      "step": 26260
    },
    {
      "epoch": 4.284665579119086,
      "grad_norm": 0.15600493550300598,
      "learning_rate": 3.529181609488532e-05,
      "loss": 0.1766,
      "num_input_tokens_seen": 56707600,
      "step": 26265
    },
    {
      "epoch": 4.285481239804241,
      "grad_norm": 0.22444719076156616,
      "learning_rate": 3.528532883428724e-05,
      "loss": 0.2635,
      "num_input_tokens_seen": 56718608,
      "step": 26270
    },
    {
      "epoch": 4.286296900489396,
      "grad_norm": 0.18730668723583221,
      "learning_rate": 3.527884073990822e-05,
      "loss": 0.2498,
      "num_input_tokens_seen": 56729840,
      "step": 26275
    },
    {
      "epoch": 4.287112561174552,
      "grad_norm": 0.2111804485321045,
      "learning_rate": 3.52723518122742e-05,
      "loss": 0.037,
      "num_input_tokens_seen": 56739984,
      "step": 26280
    },
    {
      "epoch": 4.287928221859707,
      "grad_norm": 0.25164511799812317,
      "learning_rate": 3.526586205191123e-05,
      "loss": 0.2955,
      "num_input_tokens_seen": 56751248,
      "step": 26285
    },
    {
      "epoch": 4.288743882544861,
      "grad_norm": 0.22389717400074005,
      "learning_rate": 3.525937145934539e-05,
      "loss": 0.1524,
      "num_input_tokens_seen": 56762480,
      "step": 26290
    },
    {
      "epoch": 4.289559543230016,
      "grad_norm": 2.940941572189331,
      "learning_rate": 3.525288003510285e-05,
      "loss": 0.1536,
      "num_input_tokens_seen": 56772016,
      "step": 26295
    },
    {
      "epoch": 4.290375203915171,
      "grad_norm": 2.946580171585083,
      "learning_rate": 3.524638777970982e-05,
      "loss": 0.058,
      "num_input_tokens_seen": 56781168,
      "step": 26300
    },
    {
      "epoch": 4.291190864600326,
      "grad_norm": 6.9011006355285645,
      "learning_rate": 3.523989469369262e-05,
      "loss": 0.0667,
      "num_input_tokens_seen": 56792720,
      "step": 26305
    },
    {
      "epoch": 4.2920065252854815,
      "grad_norm": 0.10244137793779373,
      "learning_rate": 3.523340077757759e-05,
      "loss": 0.0907,
      "num_input_tokens_seen": 56803856,
      "step": 26310
    },
    {
      "epoch": 4.292822185970636,
      "grad_norm": 0.16750401258468628,
      "learning_rate": 3.522690603189117e-05,
      "loss": 0.0724,
      "num_input_tokens_seen": 56814480,
      "step": 26315
    },
    {
      "epoch": 4.293637846655791,
      "grad_norm": 0.27848297357559204,
      "learning_rate": 3.522041045715986e-05,
      "loss": 0.1096,
      "num_input_tokens_seen": 56824816,
      "step": 26320
    },
    {
      "epoch": 4.294453507340946,
      "grad_norm": 0.3545133173465729,
      "learning_rate": 3.521391405391022e-05,
      "loss": 0.1749,
      "num_input_tokens_seen": 56834928,
      "step": 26325
    },
    {
      "epoch": 4.295269168026101,
      "grad_norm": 0.13783428072929382,
      "learning_rate": 3.520741682266888e-05,
      "loss": 0.0112,
      "num_input_tokens_seen": 56845680,
      "step": 26330
    },
    {
      "epoch": 4.2960848287112565,
      "grad_norm": 0.05638476088643074,
      "learning_rate": 3.520091876396255e-05,
      "loss": 0.0618,
      "num_input_tokens_seen": 56856464,
      "step": 26335
    },
    {
      "epoch": 4.296900489396411,
      "grad_norm": 0.12468436360359192,
      "learning_rate": 3.5194419878317975e-05,
      "loss": 0.1081,
      "num_input_tokens_seen": 56867920,
      "step": 26340
    },
    {
      "epoch": 4.297716150081566,
      "grad_norm": 0.14122150838375092,
      "learning_rate": 3.518792016626201e-05,
      "loss": 0.0986,
      "num_input_tokens_seen": 56878672,
      "step": 26345
    },
    {
      "epoch": 4.298531810766721,
      "grad_norm": 0.297615647315979,
      "learning_rate": 3.518141962832153e-05,
      "loss": 0.0148,
      "num_input_tokens_seen": 56889616,
      "step": 26350
    },
    {
      "epoch": 4.299347471451876,
      "grad_norm": 0.07948726415634155,
      "learning_rate": 3.517491826502352e-05,
      "loss": 0.0796,
      "num_input_tokens_seen": 56899184,
      "step": 26355
    },
    {
      "epoch": 4.300163132137031,
      "grad_norm": 0.1359148621559143,
      "learning_rate": 3.516841607689501e-05,
      "loss": 0.0834,
      "num_input_tokens_seen": 56910480,
      "step": 26360
    },
    {
      "epoch": 4.300978792822186,
      "grad_norm": 0.29599788784980774,
      "learning_rate": 3.516191306446309e-05,
      "loss": 0.2221,
      "num_input_tokens_seen": 56921040,
      "step": 26365
    },
    {
      "epoch": 4.301794453507341,
      "grad_norm": 4.5030741691589355,
      "learning_rate": 3.5155409228254946e-05,
      "loss": 0.1198,
      "num_input_tokens_seen": 56931920,
      "step": 26370
    },
    {
      "epoch": 4.302610114192496,
      "grad_norm": 4.534426212310791,
      "learning_rate": 3.5148904568797805e-05,
      "loss": 0.094,
      "num_input_tokens_seen": 56942960,
      "step": 26375
    },
    {
      "epoch": 4.303425774877651,
      "grad_norm": 0.2612292766571045,
      "learning_rate": 3.514239908661896e-05,
      "loss": 0.1264,
      "num_input_tokens_seen": 56955728,
      "step": 26380
    },
    {
      "epoch": 4.304241435562806,
      "grad_norm": 18.73579216003418,
      "learning_rate": 3.513589278224577e-05,
      "loss": 0.0536,
      "num_input_tokens_seen": 56966224,
      "step": 26385
    },
    {
      "epoch": 4.30505709624796,
      "grad_norm": 3.6883773803710938,
      "learning_rate": 3.5129385656205696e-05,
      "loss": 0.2356,
      "num_input_tokens_seen": 56975440,
      "step": 26390
    },
    {
      "epoch": 4.305872756933116,
      "grad_norm": 7.598383903503418,
      "learning_rate": 3.512287770902623e-05,
      "loss": 0.0494,
      "num_input_tokens_seen": 56986416,
      "step": 26395
    },
    {
      "epoch": 4.306688417618271,
      "grad_norm": 0.12900906801223755,
      "learning_rate": 3.5116368941234924e-05,
      "loss": 0.0968,
      "num_input_tokens_seen": 56997392,
      "step": 26400
    },
    {
      "epoch": 4.307504078303426,
      "grad_norm": 0.1268405169248581,
      "learning_rate": 3.510985935335943e-05,
      "loss": 0.0726,
      "num_input_tokens_seen": 57008144,
      "step": 26405
    },
    {
      "epoch": 4.308319738988581,
      "grad_norm": 0.26509571075439453,
      "learning_rate": 3.510334894592743e-05,
      "loss": 0.0582,
      "num_input_tokens_seen": 57019472,
      "step": 26410
    },
    {
      "epoch": 4.309135399673735,
      "grad_norm": 0.08845847100019455,
      "learning_rate": 3.509683771946671e-05,
      "loss": 0.1454,
      "num_input_tokens_seen": 57029136,
      "step": 26415
    },
    {
      "epoch": 4.309951060358891,
      "grad_norm": 3.4568231105804443,
      "learning_rate": 3.509032567450508e-05,
      "loss": 0.0526,
      "num_input_tokens_seen": 57039504,
      "step": 26420
    },
    {
      "epoch": 4.310766721044046,
      "grad_norm": 0.6754986643791199,
      "learning_rate": 3.508381281157046e-05,
      "loss": 0.1198,
      "num_input_tokens_seen": 57050064,
      "step": 26425
    },
    {
      "epoch": 4.311582381729201,
      "grad_norm": 0.21514974534511566,
      "learning_rate": 3.507729913119081e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 57061232,
      "step": 26430
    },
    {
      "epoch": 4.3123980424143555,
      "grad_norm": 0.07706812769174576,
      "learning_rate": 3.507078463389417e-05,
      "loss": 0.1651,
      "num_input_tokens_seen": 57072016,
      "step": 26435
    },
    {
      "epoch": 4.31321370309951,
      "grad_norm": 2.491929292678833,
      "learning_rate": 3.506426932020861e-05,
      "loss": 0.0838,
      "num_input_tokens_seen": 57082576,
      "step": 26440
    },
    {
      "epoch": 4.314029363784665,
      "grad_norm": 0.3132483959197998,
      "learning_rate": 3.505775319066233e-05,
      "loss": 0.1013,
      "num_input_tokens_seen": 57093712,
      "step": 26445
    },
    {
      "epoch": 4.314845024469821,
      "grad_norm": 0.6698467135429382,
      "learning_rate": 3.5051236245783536e-05,
      "loss": 0.0183,
      "num_input_tokens_seen": 57104432,
      "step": 26450
    },
    {
      "epoch": 4.315660685154976,
      "grad_norm": 0.07937267422676086,
      "learning_rate": 3.5044718486100536e-05,
      "loss": 0.0059,
      "num_input_tokens_seen": 57114576,
      "step": 26455
    },
    {
      "epoch": 4.3164763458401305,
      "grad_norm": 4.7036943435668945,
      "learning_rate": 3.503819991214168e-05,
      "loss": 0.3437,
      "num_input_tokens_seen": 57125744,
      "step": 26460
    },
    {
      "epoch": 4.317292006525285,
      "grad_norm": 0.039076440036296844,
      "learning_rate": 3.503168052443542e-05,
      "loss": 0.2231,
      "num_input_tokens_seen": 57136784,
      "step": 26465
    },
    {
      "epoch": 4.31810766721044,
      "grad_norm": 0.3440026640892029,
      "learning_rate": 3.502516032351022e-05,
      "loss": 0.0521,
      "num_input_tokens_seen": 57147888,
      "step": 26470
    },
    {
      "epoch": 4.318923327895595,
      "grad_norm": 0.0627477690577507,
      "learning_rate": 3.501863930989467e-05,
      "loss": 0.065,
      "num_input_tokens_seen": 57158704,
      "step": 26475
    },
    {
      "epoch": 4.319738988580751,
      "grad_norm": 0.0356375128030777,
      "learning_rate": 3.501211748411738e-05,
      "loss": 0.0159,
      "num_input_tokens_seen": 57170096,
      "step": 26480
    },
    {
      "epoch": 4.3205546492659055,
      "grad_norm": 0.2010306566953659,
      "learning_rate": 3.500559484670705e-05,
      "loss": 0.0743,
      "num_input_tokens_seen": 57181008,
      "step": 26485
    },
    {
      "epoch": 4.32137030995106,
      "grad_norm": 3.861107587814331,
      "learning_rate": 3.499907139819242e-05,
      "loss": 0.0702,
      "num_input_tokens_seen": 57192752,
      "step": 26490
    },
    {
      "epoch": 4.322185970636215,
      "grad_norm": 3.188190221786499,
      "learning_rate": 3.499254713910234e-05,
      "loss": 0.1386,
      "num_input_tokens_seen": 57204976,
      "step": 26495
    },
    {
      "epoch": 4.32300163132137,
      "grad_norm": 7.758331298828125,
      "learning_rate": 3.498602206996569e-05,
      "loss": 0.184,
      "num_input_tokens_seen": 57216592,
      "step": 26500
    },
    {
      "epoch": 4.323817292006526,
      "grad_norm": 0.126559779047966,
      "learning_rate": 3.497949619131141e-05,
      "loss": 0.0089,
      "num_input_tokens_seen": 57227568,
      "step": 26505
    },
    {
      "epoch": 4.3246329526916805,
      "grad_norm": 3.4178578853607178,
      "learning_rate": 3.497296950366854e-05,
      "loss": 0.0823,
      "num_input_tokens_seen": 57238096,
      "step": 26510
    },
    {
      "epoch": 4.325448613376835,
      "grad_norm": 1.1395678520202637,
      "learning_rate": 3.4966442007566165e-05,
      "loss": 0.0142,
      "num_input_tokens_seen": 57250416,
      "step": 26515
    },
    {
      "epoch": 4.32626427406199,
      "grad_norm": 0.06961100548505783,
      "learning_rate": 3.495991370353342e-05,
      "loss": 0.1423,
      "num_input_tokens_seen": 57261008,
      "step": 26520
    },
    {
      "epoch": 4.327079934747145,
      "grad_norm": 6.854227542877197,
      "learning_rate": 3.4953384592099536e-05,
      "loss": 0.188,
      "num_input_tokens_seen": 57271344,
      "step": 26525
    },
    {
      "epoch": 4.327895595432301,
      "grad_norm": 8.178387641906738,
      "learning_rate": 3.494685467379381e-05,
      "loss": 0.0609,
      "num_input_tokens_seen": 57280752,
      "step": 26530
    },
    {
      "epoch": 4.328711256117455,
      "grad_norm": 0.1267123967409134,
      "learning_rate": 3.494032394914555e-05,
      "loss": 0.1506,
      "num_input_tokens_seen": 57290736,
      "step": 26535
    },
    {
      "epoch": 4.32952691680261,
      "grad_norm": 3.2645373344421387,
      "learning_rate": 3.493379241868421e-05,
      "loss": 0.147,
      "num_input_tokens_seen": 57301680,
      "step": 26540
    },
    {
      "epoch": 4.330342577487765,
      "grad_norm": 0.09174946695566177,
      "learning_rate": 3.492726008293925e-05,
      "loss": 0.1232,
      "num_input_tokens_seen": 57313200,
      "step": 26545
    },
    {
      "epoch": 4.33115823817292,
      "grad_norm": 4.232726097106934,
      "learning_rate": 3.4920726942440215e-05,
      "loss": 0.1457,
      "num_input_tokens_seen": 57323856,
      "step": 26550
    },
    {
      "epoch": 4.331973898858075,
      "grad_norm": 0.20001770555973053,
      "learning_rate": 3.4914192997716724e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 57334768,
      "step": 26555
    },
    {
      "epoch": 4.33278955954323,
      "grad_norm": 1.2704259157180786,
      "learning_rate": 3.4907658249298435e-05,
      "loss": 0.0722,
      "num_input_tokens_seen": 57344720,
      "step": 26560
    },
    {
      "epoch": 4.333605220228385,
      "grad_norm": 0.37678268551826477,
      "learning_rate": 3.4901122697715096e-05,
      "loss": 0.1788,
      "num_input_tokens_seen": 57356784,
      "step": 26565
    },
    {
      "epoch": 4.33442088091354,
      "grad_norm": 0.11859045177698135,
      "learning_rate": 3.4894586343496524e-05,
      "loss": 0.0226,
      "num_input_tokens_seen": 57367184,
      "step": 26570
    },
    {
      "epoch": 4.335236541598695,
      "grad_norm": 0.12833264470100403,
      "learning_rate": 3.4888049187172566e-05,
      "loss": 0.1555,
      "num_input_tokens_seen": 57377840,
      "step": 26575
    },
    {
      "epoch": 4.33605220228385,
      "grad_norm": 0.06211987882852554,
      "learning_rate": 3.4881511229273175e-05,
      "loss": 0.1053,
      "num_input_tokens_seen": 57389232,
      "step": 26580
    },
    {
      "epoch": 4.3368678629690045,
      "grad_norm": 0.6239936947822571,
      "learning_rate": 3.487497247032835e-05,
      "loss": 0.0939,
      "num_input_tokens_seen": 57400176,
      "step": 26585
    },
    {
      "epoch": 4.33768352365416,
      "grad_norm": 0.04757961258292198,
      "learning_rate": 3.4868432910868156e-05,
      "loss": 0.1114,
      "num_input_tokens_seen": 57410768,
      "step": 26590
    },
    {
      "epoch": 4.338499184339315,
      "grad_norm": 4.031280517578125,
      "learning_rate": 3.48618925514227e-05,
      "loss": 0.1682,
      "num_input_tokens_seen": 57421680,
      "step": 26595
    },
    {
      "epoch": 4.33931484502447,
      "grad_norm": 6.466372966766357,
      "learning_rate": 3.4855351392522214e-05,
      "loss": 0.216,
      "num_input_tokens_seen": 57431600,
      "step": 26600
    },
    {
      "epoch": 4.340130505709625,
      "grad_norm": 0.07889610528945923,
      "learning_rate": 3.4848809434696924e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 57442640,
      "step": 26605
    },
    {
      "epoch": 4.3409461663947795,
      "grad_norm": 2.756345510482788,
      "learning_rate": 3.484226667847718e-05,
      "loss": 0.0943,
      "num_input_tokens_seen": 57453072,
      "step": 26610
    },
    {
      "epoch": 4.341761827079935,
      "grad_norm": 1.4201195240020752,
      "learning_rate": 3.4835723124393347e-05,
      "loss": 0.0945,
      "num_input_tokens_seen": 57463792,
      "step": 26615
    },
    {
      "epoch": 4.34257748776509,
      "grad_norm": 0.17208625376224518,
      "learning_rate": 3.48291787729759e-05,
      "loss": 0.0832,
      "num_input_tokens_seen": 57474448,
      "step": 26620
    },
    {
      "epoch": 4.343393148450245,
      "grad_norm": 1.5106254816055298,
      "learning_rate": 3.482263362475535e-05,
      "loss": 0.0628,
      "num_input_tokens_seen": 57485904,
      "step": 26625
    },
    {
      "epoch": 4.3442088091354,
      "grad_norm": 0.17081543803215027,
      "learning_rate": 3.4816087680262275e-05,
      "loss": 0.1529,
      "num_input_tokens_seen": 57497008,
      "step": 26630
    },
    {
      "epoch": 4.3450244698205545,
      "grad_norm": 0.11218702793121338,
      "learning_rate": 3.480954094002733e-05,
      "loss": 0.0435,
      "num_input_tokens_seen": 57507920,
      "step": 26635
    },
    {
      "epoch": 4.345840130505709,
      "grad_norm": 0.3067167103290558,
      "learning_rate": 3.480299340458123e-05,
      "loss": 0.1803,
      "num_input_tokens_seen": 57518832,
      "step": 26640
    },
    {
      "epoch": 4.346655791190865,
      "grad_norm": 0.0654403418302536,
      "learning_rate": 3.479644507445473e-05,
      "loss": 0.0949,
      "num_input_tokens_seen": 57529808,
      "step": 26645
    },
    {
      "epoch": 4.34747145187602,
      "grad_norm": 3.237981081008911,
      "learning_rate": 3.4789895950178694e-05,
      "loss": 0.1251,
      "num_input_tokens_seen": 57540944,
      "step": 26650
    },
    {
      "epoch": 4.348287112561175,
      "grad_norm": 0.06239500269293785,
      "learning_rate": 3.478334603228401e-05,
      "loss": 0.1113,
      "num_input_tokens_seen": 57552176,
      "step": 26655
    },
    {
      "epoch": 4.349102773246329,
      "grad_norm": 0.1120411828160286,
      "learning_rate": 3.477679532130167e-05,
      "loss": 0.0927,
      "num_input_tokens_seen": 57563184,
      "step": 26660
    },
    {
      "epoch": 4.349918433931484,
      "grad_norm": 0.5621286034584045,
      "learning_rate": 3.4770243817762686e-05,
      "loss": 0.2926,
      "num_input_tokens_seen": 57574320,
      "step": 26665
    },
    {
      "epoch": 4.350734094616639,
      "grad_norm": 0.08487051725387573,
      "learning_rate": 3.476369152219817e-05,
      "loss": 0.1444,
      "num_input_tokens_seen": 57584944,
      "step": 26670
    },
    {
      "epoch": 4.351549755301795,
      "grad_norm": 5.538961410522461,
      "learning_rate": 3.4757138435139274e-05,
      "loss": 0.0384,
      "num_input_tokens_seen": 57595600,
      "step": 26675
    },
    {
      "epoch": 4.35236541598695,
      "grad_norm": 11.25765609741211,
      "learning_rate": 3.4750584557117234e-05,
      "loss": 0.3891,
      "num_input_tokens_seen": 57607472,
      "step": 26680
    },
    {
      "epoch": 4.353181076672104,
      "grad_norm": 1.2607146501541138,
      "learning_rate": 3.4744029888663326e-05,
      "loss": 0.3185,
      "num_input_tokens_seen": 57618800,
      "step": 26685
    },
    {
      "epoch": 4.353996737357259,
      "grad_norm": 0.4579668939113617,
      "learning_rate": 3.473747443030892e-05,
      "loss": 0.0086,
      "num_input_tokens_seen": 57629040,
      "step": 26690
    },
    {
      "epoch": 4.354812398042414,
      "grad_norm": 3.0267632007598877,
      "learning_rate": 3.473091818258543e-05,
      "loss": 0.1986,
      "num_input_tokens_seen": 57639728,
      "step": 26695
    },
    {
      "epoch": 4.35562805872757,
      "grad_norm": 1.9840439558029175,
      "learning_rate": 3.472436114602433e-05,
      "loss": 0.0924,
      "num_input_tokens_seen": 57650512,
      "step": 26700
    },
    {
      "epoch": 4.356443719412725,
      "grad_norm": 0.18485522270202637,
      "learning_rate": 3.471780332115719e-05,
      "loss": 0.1409,
      "num_input_tokens_seen": 57662096,
      "step": 26705
    },
    {
      "epoch": 4.357259380097879,
      "grad_norm": 0.09250368922948837,
      "learning_rate": 3.47112447085156e-05,
      "loss": 0.1419,
      "num_input_tokens_seen": 57672080,
      "step": 26710
    },
    {
      "epoch": 4.358075040783034,
      "grad_norm": 0.32140758633613586,
      "learning_rate": 3.470468530863123e-05,
      "loss": 0.2394,
      "num_input_tokens_seen": 57682704,
      "step": 26715
    },
    {
      "epoch": 4.358890701468189,
      "grad_norm": 3.253242254257202,
      "learning_rate": 3.469812512203584e-05,
      "loss": 0.2174,
      "num_input_tokens_seen": 57693520,
      "step": 26720
    },
    {
      "epoch": 4.359706362153344,
      "grad_norm": 0.4338360130786896,
      "learning_rate": 3.469156414926121e-05,
      "loss": 0.1294,
      "num_input_tokens_seen": 57703312,
      "step": 26725
    },
    {
      "epoch": 4.3605220228384995,
      "grad_norm": 3.105952024459839,
      "learning_rate": 3.4685002390839226e-05,
      "loss": 0.1444,
      "num_input_tokens_seen": 57714672,
      "step": 26730
    },
    {
      "epoch": 4.361337683523654,
      "grad_norm": 0.07433799654245377,
      "learning_rate": 3.467843984730179e-05,
      "loss": 0.009,
      "num_input_tokens_seen": 57724464,
      "step": 26735
    },
    {
      "epoch": 4.362153344208809,
      "grad_norm": 1.7410246133804321,
      "learning_rate": 3.467187651918093e-05,
      "loss": 0.017,
      "num_input_tokens_seen": 57735856,
      "step": 26740
    },
    {
      "epoch": 4.362969004893964,
      "grad_norm": 0.07141131907701492,
      "learning_rate": 3.466531240700868e-05,
      "loss": 0.0683,
      "num_input_tokens_seen": 57747440,
      "step": 26745
    },
    {
      "epoch": 4.363784665579119,
      "grad_norm": 0.12390885502099991,
      "learning_rate": 3.465874751131716e-05,
      "loss": 0.0592,
      "num_input_tokens_seen": 57757168,
      "step": 26750
    },
    {
      "epoch": 4.364600326264274,
      "grad_norm": 0.07554621249437332,
      "learning_rate": 3.4652181832638566e-05,
      "loss": 0.0097,
      "num_input_tokens_seen": 57768112,
      "step": 26755
    },
    {
      "epoch": 4.365415986949429,
      "grad_norm": 2.8957602977752686,
      "learning_rate": 3.464561537150513e-05,
      "loss": 0.1345,
      "num_input_tokens_seen": 57778480,
      "step": 26760
    },
    {
      "epoch": 4.366231647634584,
      "grad_norm": 0.10320143401622772,
      "learning_rate": 3.4639048128449175e-05,
      "loss": 0.0125,
      "num_input_tokens_seen": 57789328,
      "step": 26765
    },
    {
      "epoch": 4.367047308319739,
      "grad_norm": 0.08945643901824951,
      "learning_rate": 3.463248010400307e-05,
      "loss": 0.1218,
      "num_input_tokens_seen": 57800112,
      "step": 26770
    },
    {
      "epoch": 4.367862969004894,
      "grad_norm": 0.16159725189208984,
      "learning_rate": 3.462591129869925e-05,
      "loss": 0.1699,
      "num_input_tokens_seen": 57810064,
      "step": 26775
    },
    {
      "epoch": 4.368678629690049,
      "grad_norm": 0.7375674843788147,
      "learning_rate": 3.461934171307022e-05,
      "loss": 0.012,
      "num_input_tokens_seen": 57821680,
      "step": 26780
    },
    {
      "epoch": 4.369494290375204,
      "grad_norm": 0.18759872019290924,
      "learning_rate": 3.461277134764855e-05,
      "loss": 0.1464,
      "num_input_tokens_seen": 57832496,
      "step": 26785
    },
    {
      "epoch": 4.370309951060359,
      "grad_norm": 0.11277393251657486,
      "learning_rate": 3.460620020296684e-05,
      "loss": 0.2279,
      "num_input_tokens_seen": 57842768,
      "step": 26790
    },
    {
      "epoch": 4.371125611745514,
      "grad_norm": 3.195002794265747,
      "learning_rate": 3.459962827955782e-05,
      "loss": 0.1832,
      "num_input_tokens_seen": 57853008,
      "step": 26795
    },
    {
      "epoch": 4.371941272430669,
      "grad_norm": 6.2746357917785645,
      "learning_rate": 3.459305557795422e-05,
      "loss": 0.0506,
      "num_input_tokens_seen": 57863312,
      "step": 26800
    },
    {
      "epoch": 4.372756933115824,
      "grad_norm": 0.5368702411651611,
      "learning_rate": 3.458648209868886e-05,
      "loss": 0.0721,
      "num_input_tokens_seen": 57873552,
      "step": 26805
    },
    {
      "epoch": 4.373572593800978,
      "grad_norm": 13.515825271606445,
      "learning_rate": 3.4579907842294614e-05,
      "loss": 0.1683,
      "num_input_tokens_seen": 57884336,
      "step": 26810
    },
    {
      "epoch": 4.374388254486134,
      "grad_norm": 2.330129384994507,
      "learning_rate": 3.457333280930444e-05,
      "loss": 0.1345,
      "num_input_tokens_seen": 57894512,
      "step": 26815
    },
    {
      "epoch": 4.375203915171289,
      "grad_norm": 0.03426094353199005,
      "learning_rate": 3.456675700025132e-05,
      "loss": 0.0974,
      "num_input_tokens_seen": 57906576,
      "step": 26820
    },
    {
      "epoch": 4.376019575856444,
      "grad_norm": 0.16858018934726715,
      "learning_rate": 3.4560180415668354e-05,
      "loss": 0.071,
      "num_input_tokens_seen": 57916816,
      "step": 26825
    },
    {
      "epoch": 4.376835236541599,
      "grad_norm": 0.03578922525048256,
      "learning_rate": 3.455360305608865e-05,
      "loss": 0.0166,
      "num_input_tokens_seen": 57927088,
      "step": 26830
    },
    {
      "epoch": 4.377650897226753,
      "grad_norm": 0.17116525769233704,
      "learning_rate": 3.4547024922045405e-05,
      "loss": 0.0135,
      "num_input_tokens_seen": 57939120,
      "step": 26835
    },
    {
      "epoch": 4.378466557911908,
      "grad_norm": 0.2882421016693115,
      "learning_rate": 3.454044601407187e-05,
      "loss": 0.1348,
      "num_input_tokens_seen": 57949264,
      "step": 26840
    },
    {
      "epoch": 4.379282218597064,
      "grad_norm": 1.7336941957473755,
      "learning_rate": 3.453386633270138e-05,
      "loss": 0.224,
      "num_input_tokens_seen": 57961328,
      "step": 26845
    },
    {
      "epoch": 4.380097879282219,
      "grad_norm": 2.617790460586548,
      "learning_rate": 3.4527285878467305e-05,
      "loss": 0.2194,
      "num_input_tokens_seen": 57972816,
      "step": 26850
    },
    {
      "epoch": 4.3809135399673735,
      "grad_norm": 11.77711296081543,
      "learning_rate": 3.45207046519031e-05,
      "loss": 0.0778,
      "num_input_tokens_seen": 57983440,
      "step": 26855
    },
    {
      "epoch": 4.381729200652528,
      "grad_norm": 4.458096981048584,
      "learning_rate": 3.451412265354227e-05,
      "loss": 0.1339,
      "num_input_tokens_seen": 57994384,
      "step": 26860
    },
    {
      "epoch": 4.382544861337683,
      "grad_norm": 5.544656276702881,
      "learning_rate": 3.450753988391839e-05,
      "loss": 0.043,
      "num_input_tokens_seen": 58006192,
      "step": 26865
    },
    {
      "epoch": 4.383360522022839,
      "grad_norm": 0.11429605633020401,
      "learning_rate": 3.450095634356508e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 58016784,
      "step": 26870
    },
    {
      "epoch": 4.384176182707994,
      "grad_norm": 4.855502128601074,
      "learning_rate": 3.449437203301604e-05,
      "loss": 0.1091,
      "num_input_tokens_seen": 58026736,
      "step": 26875
    },
    {
      "epoch": 4.3849918433931485,
      "grad_norm": 0.0600491501390934,
      "learning_rate": 3.4487786952805035e-05,
      "loss": 0.0417,
      "num_input_tokens_seen": 58037200,
      "step": 26880
    },
    {
      "epoch": 4.385807504078303,
      "grad_norm": 0.2659836411476135,
      "learning_rate": 3.4481201103465875e-05,
      "loss": 0.0303,
      "num_input_tokens_seen": 58049200,
      "step": 26885
    },
    {
      "epoch": 4.386623164763458,
      "grad_norm": 0.12740908563137054,
      "learning_rate": 3.447461448553245e-05,
      "loss": 0.038,
      "num_input_tokens_seen": 58060144,
      "step": 26890
    },
    {
      "epoch": 4.387438825448613,
      "grad_norm": 0.18616020679473877,
      "learning_rate": 3.4468027099538694e-05,
      "loss": 0.2452,
      "num_input_tokens_seen": 58071600,
      "step": 26895
    },
    {
      "epoch": 4.388254486133769,
      "grad_norm": 1.6752793788909912,
      "learning_rate": 3.446143894601862e-05,
      "loss": 0.2502,
      "num_input_tokens_seen": 58081616,
      "step": 26900
    },
    {
      "epoch": 4.3890701468189235,
      "grad_norm": 0.16435900330543518,
      "learning_rate": 3.44548500255063e-05,
      "loss": 0.2265,
      "num_input_tokens_seen": 58092976,
      "step": 26905
    },
    {
      "epoch": 4.389885807504078,
      "grad_norm": 0.35967424511909485,
      "learning_rate": 3.444826033853587e-05,
      "loss": 0.0129,
      "num_input_tokens_seen": 58103536,
      "step": 26910
    },
    {
      "epoch": 4.390701468189233,
      "grad_norm": 0.11986248940229416,
      "learning_rate": 3.4441669885641517e-05,
      "loss": 0.0131,
      "num_input_tokens_seen": 58115504,
      "step": 26915
    },
    {
      "epoch": 4.391517128874388,
      "grad_norm": 0.09129441529512405,
      "learning_rate": 3.443507866735749e-05,
      "loss": 0.0819,
      "num_input_tokens_seen": 58125648,
      "step": 26920
    },
    {
      "epoch": 4.392332789559543,
      "grad_norm": 2.3310320377349854,
      "learning_rate": 3.4428486684218116e-05,
      "loss": 0.1177,
      "num_input_tokens_seen": 58136816,
      "step": 26925
    },
    {
      "epoch": 4.3931484502446985,
      "grad_norm": 3.101334810256958,
      "learning_rate": 3.442189393675777e-05,
      "loss": 0.155,
      "num_input_tokens_seen": 58146832,
      "step": 26930
    },
    {
      "epoch": 4.393964110929853,
      "grad_norm": 3.337653398513794,
      "learning_rate": 3.44153004255109e-05,
      "loss": 0.2613,
      "num_input_tokens_seen": 58156912,
      "step": 26935
    },
    {
      "epoch": 4.394779771615008,
      "grad_norm": 3.4108667373657227,
      "learning_rate": 3.4408706151012e-05,
      "loss": 0.0964,
      "num_input_tokens_seen": 58167856,
      "step": 26940
    },
    {
      "epoch": 4.395595432300163,
      "grad_norm": 1.8708680868148804,
      "learning_rate": 3.440211111379564e-05,
      "loss": 0.1649,
      "num_input_tokens_seen": 58178928,
      "step": 26945
    },
    {
      "epoch": 4.396411092985318,
      "grad_norm": 0.086298368871212,
      "learning_rate": 3.4395515314396445e-05,
      "loss": 0.0975,
      "num_input_tokens_seen": 58187888,
      "step": 26950
    },
    {
      "epoch": 4.397226753670473,
      "grad_norm": 0.3319890797138214,
      "learning_rate": 3.4388918753349106e-05,
      "loss": 0.0647,
      "num_input_tokens_seen": 58198160,
      "step": 26955
    },
    {
      "epoch": 4.398042414355628,
      "grad_norm": 0.09963146597146988,
      "learning_rate": 3.438232143118838e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 58209808,
      "step": 26960
    },
    {
      "epoch": 4.398858075040783,
      "grad_norm": 23.190380096435547,
      "learning_rate": 3.437572334844907e-05,
      "loss": 0.0889,
      "num_input_tokens_seen": 58221104,
      "step": 26965
    },
    {
      "epoch": 4.399673735725938,
      "grad_norm": 2.905056953430176,
      "learning_rate": 3.436912450566606e-05,
      "loss": 0.1671,
      "num_input_tokens_seen": 58230832,
      "step": 26970
    },
    {
      "epoch": 4.400489396411093,
      "grad_norm": 1.5367621183395386,
      "learning_rate": 3.436252490337428e-05,
      "loss": 0.1571,
      "num_input_tokens_seen": 58241328,
      "step": 26975
    },
    {
      "epoch": 4.401305057096248,
      "grad_norm": 0.13706767559051514,
      "learning_rate": 3.4355924542108716e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 58252720,
      "step": 26980
    },
    {
      "epoch": 4.402120717781403,
      "grad_norm": 1.714287281036377,
      "learning_rate": 3.4349323422404444e-05,
      "loss": 0.0284,
      "num_input_tokens_seen": 58263216,
      "step": 26985
    },
    {
      "epoch": 4.402936378466558,
      "grad_norm": 0.16424110531806946,
      "learning_rate": 3.434272154479657e-05,
      "loss": 0.0086,
      "num_input_tokens_seen": 58274320,
      "step": 26990
    },
    {
      "epoch": 4.403752039151713,
      "grad_norm": 0.08186185359954834,
      "learning_rate": 3.4336118909820295e-05,
      "loss": 0.2202,
      "num_input_tokens_seen": 58285392,
      "step": 26995
    },
    {
      "epoch": 4.404567699836868,
      "grad_norm": 3.257965326309204,
      "learning_rate": 3.432951551801084e-05,
      "loss": 0.2592,
      "num_input_tokens_seen": 58295728,
      "step": 27000
    },
    {
      "epoch": 4.4053833605220225,
      "grad_norm": 0.09292906522750854,
      "learning_rate": 3.432291136990352e-05,
      "loss": 0.1242,
      "num_input_tokens_seen": 58306704,
      "step": 27005
    },
    {
      "epoch": 4.406199021207178,
      "grad_norm": 0.21827851235866547,
      "learning_rate": 3.4316306466033704e-05,
      "loss": 0.1869,
      "num_input_tokens_seen": 58318288,
      "step": 27010
    },
    {
      "epoch": 4.407014681892333,
      "grad_norm": 1.5792090892791748,
      "learning_rate": 3.430970080693681e-05,
      "loss": 0.1153,
      "num_input_tokens_seen": 58330416,
      "step": 27015
    },
    {
      "epoch": 4.407830342577488,
      "grad_norm": 5.034025192260742,
      "learning_rate": 3.430309439314834e-05,
      "loss": 0.1188,
      "num_input_tokens_seen": 58341456,
      "step": 27020
    },
    {
      "epoch": 4.408646003262643,
      "grad_norm": 0.15225641429424286,
      "learning_rate": 3.4296487225203825e-05,
      "loss": 0.1919,
      "num_input_tokens_seen": 58351568,
      "step": 27025
    },
    {
      "epoch": 4.4094616639477975,
      "grad_norm": 2.4444918632507324,
      "learning_rate": 3.42898793036389e-05,
      "loss": 0.2354,
      "num_input_tokens_seen": 58361968,
      "step": 27030
    },
    {
      "epoch": 4.410277324632952,
      "grad_norm": 2.6613261699676514,
      "learning_rate": 3.428327062898921e-05,
      "loss": 0.1081,
      "num_input_tokens_seen": 58373200,
      "step": 27035
    },
    {
      "epoch": 4.411092985318108,
      "grad_norm": 0.1318122297525406,
      "learning_rate": 3.4276661201790506e-05,
      "loss": 0.0221,
      "num_input_tokens_seen": 58383632,
      "step": 27040
    },
    {
      "epoch": 4.411908646003263,
      "grad_norm": 2.835576295852661,
      "learning_rate": 3.427005102257857e-05,
      "loss": 0.1857,
      "num_input_tokens_seen": 58394768,
      "step": 27045
    },
    {
      "epoch": 4.412724306688418,
      "grad_norm": 7.886944770812988,
      "learning_rate": 3.426344009188927e-05,
      "loss": 0.0585,
      "num_input_tokens_seen": 58403856,
      "step": 27050
    },
    {
      "epoch": 4.4135399673735725,
      "grad_norm": 3.3899881839752197,
      "learning_rate": 3.425682841025851e-05,
      "loss": 0.2073,
      "num_input_tokens_seen": 58415152,
      "step": 27055
    },
    {
      "epoch": 4.414355628058727,
      "grad_norm": 4.721774578094482,
      "learning_rate": 3.4250215978222264e-05,
      "loss": 0.0383,
      "num_input_tokens_seen": 58425584,
      "step": 27060
    },
    {
      "epoch": 4.415171288743883,
      "grad_norm": 0.14894041419029236,
      "learning_rate": 3.424360279631659e-05,
      "loss": 0.1,
      "num_input_tokens_seen": 58435088,
      "step": 27065
    },
    {
      "epoch": 4.415986949429038,
      "grad_norm": 0.1305241733789444,
      "learning_rate": 3.423698886507756e-05,
      "loss": 0.0868,
      "num_input_tokens_seen": 58447088,
      "step": 27070
    },
    {
      "epoch": 4.416802610114193,
      "grad_norm": 7.27046537399292,
      "learning_rate": 3.4230374185041346e-05,
      "loss": 0.1465,
      "num_input_tokens_seen": 58458928,
      "step": 27075
    },
    {
      "epoch": 4.417618270799347,
      "grad_norm": 0.12002504616975784,
      "learning_rate": 3.4223758756744176e-05,
      "loss": 0.1507,
      "num_input_tokens_seen": 58470224,
      "step": 27080
    },
    {
      "epoch": 4.418433931484502,
      "grad_norm": 0.3602011203765869,
      "learning_rate": 3.421714258072231e-05,
      "loss": 0.0083,
      "num_input_tokens_seen": 58480816,
      "step": 27085
    },
    {
      "epoch": 4.419249592169657,
      "grad_norm": 0.3553735315799713,
      "learning_rate": 3.421052565751209e-05,
      "loss": 0.1393,
      "num_input_tokens_seen": 58492528,
      "step": 27090
    },
    {
      "epoch": 4.420065252854813,
      "grad_norm": 0.07079923152923584,
      "learning_rate": 3.420390798764995e-05,
      "loss": 0.0349,
      "num_input_tokens_seen": 58502640,
      "step": 27095
    },
    {
      "epoch": 4.420880913539968,
      "grad_norm": 14.102914810180664,
      "learning_rate": 3.4197289571672316e-05,
      "loss": 0.0637,
      "num_input_tokens_seen": 58514064,
      "step": 27100
    },
    {
      "epoch": 4.421696574225122,
      "grad_norm": 0.3079579174518585,
      "learning_rate": 3.4190670410115724e-05,
      "loss": 0.0062,
      "num_input_tokens_seen": 58526032,
      "step": 27105
    },
    {
      "epoch": 4.422512234910277,
      "grad_norm": 0.17233538627624512,
      "learning_rate": 3.418405050351674e-05,
      "loss": 0.0052,
      "num_input_tokens_seen": 58537904,
      "step": 27110
    },
    {
      "epoch": 4.423327895595432,
      "grad_norm": 3.1176302433013916,
      "learning_rate": 3.417742985241205e-05,
      "loss": 0.3075,
      "num_input_tokens_seen": 58548208,
      "step": 27115
    },
    {
      "epoch": 4.424143556280587,
      "grad_norm": 0.07439375668764114,
      "learning_rate": 3.417080845733831e-05,
      "loss": 0.2225,
      "num_input_tokens_seen": 58559568,
      "step": 27120
    },
    {
      "epoch": 4.424959216965743,
      "grad_norm": 1.2107059955596924,
      "learning_rate": 3.416418631883231e-05,
      "loss": 0.1219,
      "num_input_tokens_seen": 58570576,
      "step": 27125
    },
    {
      "epoch": 4.425774877650897,
      "grad_norm": 0.09766573458909988,
      "learning_rate": 3.415756343743088e-05,
      "loss": 0.0197,
      "num_input_tokens_seen": 58581680,
      "step": 27130
    },
    {
      "epoch": 4.426590538336052,
      "grad_norm": 0.06342066079378128,
      "learning_rate": 3.4150939813670886e-05,
      "loss": 0.0404,
      "num_input_tokens_seen": 58593040,
      "step": 27135
    },
    {
      "epoch": 4.427406199021207,
      "grad_norm": 0.046015411615371704,
      "learning_rate": 3.414431544808928e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 58603728,
      "step": 27140
    },
    {
      "epoch": 4.428221859706362,
      "grad_norm": 0.09383854269981384,
      "learning_rate": 3.413769034122306e-05,
      "loss": 0.0322,
      "num_input_tokens_seen": 58614640,
      "step": 27145
    },
    {
      "epoch": 4.4290375203915175,
      "grad_norm": 0.07216141372919083,
      "learning_rate": 3.41310644936093e-05,
      "loss": 0.0134,
      "num_input_tokens_seen": 58623536,
      "step": 27150
    },
    {
      "epoch": 4.429853181076672,
      "grad_norm": 5.684617519378662,
      "learning_rate": 3.412443790578512e-05,
      "loss": 0.0929,
      "num_input_tokens_seen": 58633648,
      "step": 27155
    },
    {
      "epoch": 4.430668841761827,
      "grad_norm": 0.039336834102869034,
      "learning_rate": 3.4117810578287704e-05,
      "loss": 0.1294,
      "num_input_tokens_seen": 58644624,
      "step": 27160
    },
    {
      "epoch": 4.431484502446982,
      "grad_norm": 0.17408223450183868,
      "learning_rate": 3.411118251165431e-05,
      "loss": 0.0368,
      "num_input_tokens_seen": 58655344,
      "step": 27165
    },
    {
      "epoch": 4.432300163132137,
      "grad_norm": 4.096160411834717,
      "learning_rate": 3.410455370642221e-05,
      "loss": 0.1473,
      "num_input_tokens_seen": 58666448,
      "step": 27170
    },
    {
      "epoch": 4.433115823817292,
      "grad_norm": 0.0826522707939148,
      "learning_rate": 3.409792416312881e-05,
      "loss": 0.0381,
      "num_input_tokens_seen": 58678672,
      "step": 27175
    },
    {
      "epoch": 4.433931484502447,
      "grad_norm": 1.2679916620254517,
      "learning_rate": 3.409129388231151e-05,
      "loss": 0.0126,
      "num_input_tokens_seen": 58690672,
      "step": 27180
    },
    {
      "epoch": 4.434747145187602,
      "grad_norm": 6.634708881378174,
      "learning_rate": 3.40846628645078e-05,
      "loss": 0.1661,
      "num_input_tokens_seen": 58702320,
      "step": 27185
    },
    {
      "epoch": 4.435562805872757,
      "grad_norm": 0.08634289354085922,
      "learning_rate": 3.407803111025522e-05,
      "loss": 0.0956,
      "num_input_tokens_seen": 58714192,
      "step": 27190
    },
    {
      "epoch": 4.436378466557912,
      "grad_norm": 0.0404016375541687,
      "learning_rate": 3.407139862009138e-05,
      "loss": 0.0887,
      "num_input_tokens_seen": 58725488,
      "step": 27195
    },
    {
      "epoch": 4.437194127243067,
      "grad_norm": 1.8363744020462036,
      "learning_rate": 3.406476539455394e-05,
      "loss": 0.1618,
      "num_input_tokens_seen": 58736208,
      "step": 27200
    },
    {
      "epoch": 4.438009787928221,
      "grad_norm": 0.05074001103639603,
      "learning_rate": 3.405813143418062e-05,
      "loss": 0.0049,
      "num_input_tokens_seen": 58746608,
      "step": 27205
    },
    {
      "epoch": 4.438825448613377,
      "grad_norm": 0.18867208063602448,
      "learning_rate": 3.4051496739509216e-05,
      "loss": 0.1844,
      "num_input_tokens_seen": 58756880,
      "step": 27210
    },
    {
      "epoch": 4.439641109298532,
      "grad_norm": 0.046085234731435776,
      "learning_rate": 3.404486131107754e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 58768080,
      "step": 27215
    },
    {
      "epoch": 4.440456769983687,
      "grad_norm": 2.8690378665924072,
      "learning_rate": 3.403822514942353e-05,
      "loss": 0.2246,
      "num_input_tokens_seen": 58778896,
      "step": 27220
    },
    {
      "epoch": 4.441272430668842,
      "grad_norm": 37.09941482543945,
      "learning_rate": 3.4031588255085126e-05,
      "loss": 0.3199,
      "num_input_tokens_seen": 58790032,
      "step": 27225
    },
    {
      "epoch": 4.442088091353996,
      "grad_norm": 0.10624881833791733,
      "learning_rate": 3.4024950628600345e-05,
      "loss": 0.0536,
      "num_input_tokens_seen": 58800880,
      "step": 27230
    },
    {
      "epoch": 4.442903752039152,
      "grad_norm": 0.03980366140604019,
      "learning_rate": 3.401831227050728e-05,
      "loss": 0.0121,
      "num_input_tokens_seen": 58811760,
      "step": 27235
    },
    {
      "epoch": 4.443719412724307,
      "grad_norm": 0.16718868911266327,
      "learning_rate": 3.401167318134406e-05,
      "loss": 0.1678,
      "num_input_tokens_seen": 58821808,
      "step": 27240
    },
    {
      "epoch": 4.444535073409462,
      "grad_norm": 0.4627486765384674,
      "learning_rate": 3.400503336164888e-05,
      "loss": 0.1981,
      "num_input_tokens_seen": 58832080,
      "step": 27245
    },
    {
      "epoch": 4.445350734094617,
      "grad_norm": 0.2281552255153656,
      "learning_rate": 3.3998392811960024e-05,
      "loss": 0.0237,
      "num_input_tokens_seen": 58842480,
      "step": 27250
    },
    {
      "epoch": 4.446166394779771,
      "grad_norm": 20.436752319335938,
      "learning_rate": 3.399175153281578e-05,
      "loss": 0.3219,
      "num_input_tokens_seen": 58852752,
      "step": 27255
    },
    {
      "epoch": 4.446982055464926,
      "grad_norm": 0.08264545351266861,
      "learning_rate": 3.3985109524754535e-05,
      "loss": 0.0609,
      "num_input_tokens_seen": 58863312,
      "step": 27260
    },
    {
      "epoch": 4.447797716150082,
      "grad_norm": 5.9884033203125,
      "learning_rate": 3.397846678831472e-05,
      "loss": 0.11,
      "num_input_tokens_seen": 58873104,
      "step": 27265
    },
    {
      "epoch": 4.448613376835237,
      "grad_norm": 2.786705255508423,
      "learning_rate": 3.397182332403482e-05,
      "loss": 0.0129,
      "num_input_tokens_seen": 58884624,
      "step": 27270
    },
    {
      "epoch": 4.4494290375203915,
      "grad_norm": 0.21680717170238495,
      "learning_rate": 3.3965179132453416e-05,
      "loss": 0.0276,
      "num_input_tokens_seen": 58894864,
      "step": 27275
    },
    {
      "epoch": 4.450244698205546,
      "grad_norm": 3.943552017211914,
      "learning_rate": 3.3958534214109095e-05,
      "loss": 0.1979,
      "num_input_tokens_seen": 58906512,
      "step": 27280
    },
    {
      "epoch": 4.451060358890701,
      "grad_norm": 0.06703973561525345,
      "learning_rate": 3.395188856954054e-05,
      "loss": 0.1125,
      "num_input_tokens_seen": 58917904,
      "step": 27285
    },
    {
      "epoch": 4.451876019575856,
      "grad_norm": 1.4853523969650269,
      "learning_rate": 3.394524219928647e-05,
      "loss": 0.081,
      "num_input_tokens_seen": 58928304,
      "step": 27290
    },
    {
      "epoch": 4.452691680261012,
      "grad_norm": 0.07906728982925415,
      "learning_rate": 3.3938595103885684e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 58939056,
      "step": 27295
    },
    {
      "epoch": 4.4535073409461665,
      "grad_norm": 0.12298695743083954,
      "learning_rate": 3.393194728387702e-05,
      "loss": 0.1072,
      "num_input_tokens_seen": 58950384,
      "step": 27300
    },
    {
      "epoch": 4.454323001631321,
      "grad_norm": 0.0752132460474968,
      "learning_rate": 3.39252987397994e-05,
      "loss": 0.0229,
      "num_input_tokens_seen": 58959952,
      "step": 27305
    },
    {
      "epoch": 4.455138662316476,
      "grad_norm": 7.1743950843811035,
      "learning_rate": 3.391864947219177e-05,
      "loss": 0.0405,
      "num_input_tokens_seen": 58970288,
      "step": 27310
    },
    {
      "epoch": 4.455954323001631,
      "grad_norm": 0.05831082910299301,
      "learning_rate": 3.391199948159315e-05,
      "loss": 0.1414,
      "num_input_tokens_seen": 58981936,
      "step": 27315
    },
    {
      "epoch": 4.456769983686787,
      "grad_norm": 0.03013616055250168,
      "learning_rate": 3.390534876854265e-05,
      "loss": 0.0076,
      "num_input_tokens_seen": 58992432,
      "step": 27320
    },
    {
      "epoch": 4.4575856443719415,
      "grad_norm": 0.12155994772911072,
      "learning_rate": 3.389869733357939e-05,
      "loss": 0.004,
      "num_input_tokens_seen": 59002832,
      "step": 27325
    },
    {
      "epoch": 4.458401305057096,
      "grad_norm": 6.801733016967773,
      "learning_rate": 3.389204517724256e-05,
      "loss": 0.0427,
      "num_input_tokens_seen": 59013936,
      "step": 27330
    },
    {
      "epoch": 4.459216965742251,
      "grad_norm": 0.1223423108458519,
      "learning_rate": 3.388539230007145e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 59023440,
      "step": 27335
    },
    {
      "epoch": 4.460032626427406,
      "grad_norm": 7.5102314949035645,
      "learning_rate": 3.387873870260534e-05,
      "loss": 0.0938,
      "num_input_tokens_seen": 59034224,
      "step": 27340
    },
    {
      "epoch": 4.460848287112561,
      "grad_norm": 0.18983300030231476,
      "learning_rate": 3.3872084385383626e-05,
      "loss": 0.0232,
      "num_input_tokens_seen": 59044016,
      "step": 27345
    },
    {
      "epoch": 4.4616639477977165,
      "grad_norm": 0.047964051365852356,
      "learning_rate": 3.3865429348945735e-05,
      "loss": 0.048,
      "num_input_tokens_seen": 59053648,
      "step": 27350
    },
    {
      "epoch": 4.462479608482871,
      "grad_norm": 0.09409400075674057,
      "learning_rate": 3.385877359383116e-05,
      "loss": 0.1408,
      "num_input_tokens_seen": 59064080,
      "step": 27355
    },
    {
      "epoch": 4.463295269168026,
      "grad_norm": 0.9640812873840332,
      "learning_rate": 3.385211712057945e-05,
      "loss": 0.2134,
      "num_input_tokens_seen": 59074672,
      "step": 27360
    },
    {
      "epoch": 4.464110929853181,
      "grad_norm": 2.773480176925659,
      "learning_rate": 3.384545992973021e-05,
      "loss": 0.2477,
      "num_input_tokens_seen": 59084784,
      "step": 27365
    },
    {
      "epoch": 4.464926590538336,
      "grad_norm": 0.2358299195766449,
      "learning_rate": 3.383880202182311e-05,
      "loss": 0.017,
      "num_input_tokens_seen": 59095504,
      "step": 27370
    },
    {
      "epoch": 4.465742251223491,
      "grad_norm": 0.06360097974538803,
      "learning_rate": 3.3832143397397855e-05,
      "loss": 0.006,
      "num_input_tokens_seen": 59105616,
      "step": 27375
    },
    {
      "epoch": 4.466557911908646,
      "grad_norm": 10.953752517700195,
      "learning_rate": 3.382548405699426e-05,
      "loss": 0.3359,
      "num_input_tokens_seen": 59114576,
      "step": 27380
    },
    {
      "epoch": 4.467373572593801,
      "grad_norm": 0.07115211337804794,
      "learning_rate": 3.3818824001152135e-05,
      "loss": 0.1698,
      "num_input_tokens_seen": 59124976,
      "step": 27385
    },
    {
      "epoch": 4.468189233278956,
      "grad_norm": 0.07822829484939575,
      "learning_rate": 3.38121632304114e-05,
      "loss": 0.0975,
      "num_input_tokens_seen": 59136336,
      "step": 27390
    },
    {
      "epoch": 4.469004893964111,
      "grad_norm": 0.29284754395484924,
      "learning_rate": 3.3805501745312e-05,
      "loss": 0.1261,
      "num_input_tokens_seen": 59146960,
      "step": 27395
    },
    {
      "epoch": 4.4698205546492655,
      "grad_norm": 1.5599113702774048,
      "learning_rate": 3.379883954639394e-05,
      "loss": 0.0087,
      "num_input_tokens_seen": 59158096,
      "step": 27400
    },
    {
      "epoch": 4.470636215334421,
      "grad_norm": 0.06289583444595337,
      "learning_rate": 3.379217663419731e-05,
      "loss": 0.0693,
      "num_input_tokens_seen": 59168912,
      "step": 27405
    },
    {
      "epoch": 4.471451876019576,
      "grad_norm": 0.05054587125778198,
      "learning_rate": 3.378551300926222e-05,
      "loss": 0.008,
      "num_input_tokens_seen": 59179728,
      "step": 27410
    },
    {
      "epoch": 4.472267536704731,
      "grad_norm": 4.775998592376709,
      "learning_rate": 3.3778848672128884e-05,
      "loss": 0.2334,
      "num_input_tokens_seen": 59191312,
      "step": 27415
    },
    {
      "epoch": 4.473083197389886,
      "grad_norm": 3.131594657897949,
      "learning_rate": 3.3772183623337524e-05,
      "loss": 0.013,
      "num_input_tokens_seen": 59202352,
      "step": 27420
    },
    {
      "epoch": 4.4738988580750405,
      "grad_norm": 0.06370081752538681,
      "learning_rate": 3.3765517863428456e-05,
      "loss": 0.0405,
      "num_input_tokens_seen": 59212560,
      "step": 27425
    },
    {
      "epoch": 4.474714518760196,
      "grad_norm": 0.916146993637085,
      "learning_rate": 3.375885139294202e-05,
      "loss": 0.1004,
      "num_input_tokens_seen": 59221968,
      "step": 27430
    },
    {
      "epoch": 4.475530179445351,
      "grad_norm": 5.828142166137695,
      "learning_rate": 3.375218421241866e-05,
      "loss": 0.1806,
      "num_input_tokens_seen": 59232848,
      "step": 27435
    },
    {
      "epoch": 4.476345840130506,
      "grad_norm": 0.044280849397182465,
      "learning_rate": 3.3745516322398834e-05,
      "loss": 0.1405,
      "num_input_tokens_seen": 59243984,
      "step": 27440
    },
    {
      "epoch": 4.477161500815661,
      "grad_norm": 0.1461818963289261,
      "learning_rate": 3.373884772342308e-05,
      "loss": 0.1658,
      "num_input_tokens_seen": 59253840,
      "step": 27445
    },
    {
      "epoch": 4.4779771615008155,
      "grad_norm": 0.08795010298490524,
      "learning_rate": 3.3732178416032e-05,
      "loss": 0.0094,
      "num_input_tokens_seen": 59264400,
      "step": 27450
    },
    {
      "epoch": 4.47879282218597,
      "grad_norm": 3.8043558597564697,
      "learning_rate": 3.372550840076622e-05,
      "loss": 0.271,
      "num_input_tokens_seen": 59275728,
      "step": 27455
    },
    {
      "epoch": 4.479608482871126,
      "grad_norm": 0.06280068308115005,
      "learning_rate": 3.371883767816646e-05,
      "loss": 0.1255,
      "num_input_tokens_seen": 59286896,
      "step": 27460
    },
    {
      "epoch": 4.480424143556281,
      "grad_norm": 0.3041984736919403,
      "learning_rate": 3.371216624877348e-05,
      "loss": 0.0791,
      "num_input_tokens_seen": 59297808,
      "step": 27465
    },
    {
      "epoch": 4.481239804241436,
      "grad_norm": 0.3431588113307953,
      "learning_rate": 3.370549411312809e-05,
      "loss": 0.085,
      "num_input_tokens_seen": 59308752,
      "step": 27470
    },
    {
      "epoch": 4.4820554649265905,
      "grad_norm": 8.118783950805664,
      "learning_rate": 3.3698821271771186e-05,
      "loss": 0.13,
      "num_input_tokens_seen": 59318320,
      "step": 27475
    },
    {
      "epoch": 4.482871125611745,
      "grad_norm": 4.661539554595947,
      "learning_rate": 3.369214772524369e-05,
      "loss": 0.1307,
      "num_input_tokens_seen": 59328464,
      "step": 27480
    },
    {
      "epoch": 4.4836867862969,
      "grad_norm": 0.4202403128147125,
      "learning_rate": 3.3685473474086584e-05,
      "loss": 0.0906,
      "num_input_tokens_seen": 59338000,
      "step": 27485
    },
    {
      "epoch": 4.484502446982056,
      "grad_norm": 0.6933928728103638,
      "learning_rate": 3.3678798518840946e-05,
      "loss": 0.1531,
      "num_input_tokens_seen": 59347792,
      "step": 27490
    },
    {
      "epoch": 4.485318107667211,
      "grad_norm": 0.10543301701545715,
      "learning_rate": 3.367212286004786e-05,
      "loss": 0.0093,
      "num_input_tokens_seen": 59357648,
      "step": 27495
    },
    {
      "epoch": 4.486133768352365,
      "grad_norm": 0.08430114388465881,
      "learning_rate": 3.366544649824849e-05,
      "loss": 0.0321,
      "num_input_tokens_seen": 59367760,
      "step": 27500
    },
    {
      "epoch": 4.48694942903752,
      "grad_norm": 0.11894170939922333,
      "learning_rate": 3.365876943398406e-05,
      "loss": 0.0485,
      "num_input_tokens_seen": 59378416,
      "step": 27505
    },
    {
      "epoch": 4.487765089722675,
      "grad_norm": 1.0292885303497314,
      "learning_rate": 3.365209166779585e-05,
      "loss": 0.039,
      "num_input_tokens_seen": 59388944,
      "step": 27510
    },
    {
      "epoch": 4.488580750407831,
      "grad_norm": 0.06280895322561264,
      "learning_rate": 3.3645413200225175e-05,
      "loss": 0.1303,
      "num_input_tokens_seen": 59397776,
      "step": 27515
    },
    {
      "epoch": 4.489396411092986,
      "grad_norm": 1.400780439376831,
      "learning_rate": 3.363873403181346e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 59408464,
      "step": 27520
    },
    {
      "epoch": 4.49021207177814,
      "grad_norm": 0.0447203665971756,
      "learning_rate": 3.363205416310212e-05,
      "loss": 0.2334,
      "num_input_tokens_seen": 59419888,
      "step": 27525
    },
    {
      "epoch": 4.491027732463295,
      "grad_norm": 0.022025197744369507,
      "learning_rate": 3.362537359463267e-05,
      "loss": 0.1503,
      "num_input_tokens_seen": 59429712,
      "step": 27530
    },
    {
      "epoch": 4.49184339314845,
      "grad_norm": 0.044437769800424576,
      "learning_rate": 3.361869232694666e-05,
      "loss": 0.0692,
      "num_input_tokens_seen": 59440688,
      "step": 27535
    },
    {
      "epoch": 4.492659053833605,
      "grad_norm": 0.05670817941427231,
      "learning_rate": 3.3612010360585744e-05,
      "loss": 0.0124,
      "num_input_tokens_seen": 59451120,
      "step": 27540
    },
    {
      "epoch": 4.493474714518761,
      "grad_norm": 0.0395425446331501,
      "learning_rate": 3.360532769609156e-05,
      "loss": 0.0974,
      "num_input_tokens_seen": 59461776,
      "step": 27545
    },
    {
      "epoch": 4.494290375203915,
      "grad_norm": 0.0732116624712944,
      "learning_rate": 3.359864433400585e-05,
      "loss": 0.2992,
      "num_input_tokens_seen": 59472464,
      "step": 27550
    },
    {
      "epoch": 4.49510603588907,
      "grad_norm": 0.42865869402885437,
      "learning_rate": 3.3591960274870394e-05,
      "loss": 0.3121,
      "num_input_tokens_seen": 59482704,
      "step": 27555
    },
    {
      "epoch": 4.495921696574225,
      "grad_norm": 5.648609161376953,
      "learning_rate": 3.3585275519227046e-05,
      "loss": 0.1836,
      "num_input_tokens_seen": 59492976,
      "step": 27560
    },
    {
      "epoch": 4.49673735725938,
      "grad_norm": 3.8055789470672607,
      "learning_rate": 3.357859006761771e-05,
      "loss": 0.0356,
      "num_input_tokens_seen": 59503344,
      "step": 27565
    },
    {
      "epoch": 4.497553017944535,
      "grad_norm": 0.21388722956180573,
      "learning_rate": 3.357190392058433e-05,
      "loss": 0.006,
      "num_input_tokens_seen": 59514896,
      "step": 27570
    },
    {
      "epoch": 4.49836867862969,
      "grad_norm": 0.39768272638320923,
      "learning_rate": 3.356521707866893e-05,
      "loss": 0.2554,
      "num_input_tokens_seen": 59527408,
      "step": 27575
    },
    {
      "epoch": 4.499184339314845,
      "grad_norm": 3.3803622722625732,
      "learning_rate": 3.3558529542413574e-05,
      "loss": 0.1233,
      "num_input_tokens_seen": 59537520,
      "step": 27580
    },
    {
      "epoch": 4.5,
      "grad_norm": 1.0554314851760864,
      "learning_rate": 3.3551841312360386e-05,
      "loss": 0.3003,
      "num_input_tokens_seen": 59549072,
      "step": 27585
    },
    {
      "epoch": 4.5,
      "eval_loss": 0.15887264907360077,
      "eval_runtime": 132.9408,
      "eval_samples_per_second": 20.498,
      "eval_steps_per_second": 5.13,
      "num_input_tokens_seen": 59549072,
      "step": 27585
    },
    {
      "epoch": 4.500815660685155,
      "grad_norm": 0.08097808808088303,
      "learning_rate": 3.354515238905155e-05,
      "loss": 0.0688,
      "num_input_tokens_seen": 59559824,
      "step": 27590
    },
    {
      "epoch": 4.50163132137031,
      "grad_norm": 0.2921116352081299,
      "learning_rate": 3.35384627730293e-05,
      "loss": 0.1252,
      "num_input_tokens_seen": 59569776,
      "step": 27595
    },
    {
      "epoch": 4.502446982055465,
      "grad_norm": 4.903281211853027,
      "learning_rate": 3.353177246483594e-05,
      "loss": 0.1634,
      "num_input_tokens_seen": 59579184,
      "step": 27600
    },
    {
      "epoch": 4.50326264274062,
      "grad_norm": 0.1140977144241333,
      "learning_rate": 3.352508146501381e-05,
      "loss": 0.0486,
      "num_input_tokens_seen": 59590416,
      "step": 27605
    },
    {
      "epoch": 4.504078303425775,
      "grad_norm": 0.10739248991012573,
      "learning_rate": 3.3518389774105326e-05,
      "loss": 0.0719,
      "num_input_tokens_seen": 59601136,
      "step": 27610
    },
    {
      "epoch": 4.50489396411093,
      "grad_norm": 2.9648189544677734,
      "learning_rate": 3.351169739265294e-05,
      "loss": 0.2624,
      "num_input_tokens_seen": 59612816,
      "step": 27615
    },
    {
      "epoch": 4.505709624796085,
      "grad_norm": 0.13738565146923065,
      "learning_rate": 3.350500432119917e-05,
      "loss": 0.0152,
      "num_input_tokens_seen": 59624080,
      "step": 27620
    },
    {
      "epoch": 4.506525285481239,
      "grad_norm": 3.4406023025512695,
      "learning_rate": 3.3498310560286604e-05,
      "loss": 0.0309,
      "num_input_tokens_seen": 59634608,
      "step": 27625
    },
    {
      "epoch": 4.507340946166395,
      "grad_norm": 0.14925315976142883,
      "learning_rate": 3.349161611045786e-05,
      "loss": 0.014,
      "num_input_tokens_seen": 59645200,
      "step": 27630
    },
    {
      "epoch": 4.50815660685155,
      "grad_norm": 2.3995683193206787,
      "learning_rate": 3.348492097225563e-05,
      "loss": 0.1583,
      "num_input_tokens_seen": 59654736,
      "step": 27635
    },
    {
      "epoch": 4.508972267536705,
      "grad_norm": 0.0835544764995575,
      "learning_rate": 3.347822514622265e-05,
      "loss": 0.1303,
      "num_input_tokens_seen": 59664624,
      "step": 27640
    },
    {
      "epoch": 4.50978792822186,
      "grad_norm": 0.06443583965301514,
      "learning_rate": 3.347152863290173e-05,
      "loss": 0.2365,
      "num_input_tokens_seen": 59674480,
      "step": 27645
    },
    {
      "epoch": 4.510603588907014,
      "grad_norm": 0.34996533393859863,
      "learning_rate": 3.346483143283571e-05,
      "loss": 0.0096,
      "num_input_tokens_seen": 59685392,
      "step": 27650
    },
    {
      "epoch": 4.511419249592169,
      "grad_norm": 1.139493703842163,
      "learning_rate": 3.3458133546567506e-05,
      "loss": 0.014,
      "num_input_tokens_seen": 59695280,
      "step": 27655
    },
    {
      "epoch": 4.512234910277325,
      "grad_norm": 0.2241903841495514,
      "learning_rate": 3.345143497464007e-05,
      "loss": 0.1069,
      "num_input_tokens_seen": 59705872,
      "step": 27660
    },
    {
      "epoch": 4.51305057096248,
      "grad_norm": 0.07191357016563416,
      "learning_rate": 3.344473571759645e-05,
      "loss": 0.0936,
      "num_input_tokens_seen": 59717936,
      "step": 27665
    },
    {
      "epoch": 4.513866231647635,
      "grad_norm": 6.5147199630737305,
      "learning_rate": 3.343803577597969e-05,
      "loss": 0.0942,
      "num_input_tokens_seen": 59728144,
      "step": 27670
    },
    {
      "epoch": 4.514681892332789,
      "grad_norm": 4.543582439422607,
      "learning_rate": 3.343133515033295e-05,
      "loss": 0.1052,
      "num_input_tokens_seen": 59739888,
      "step": 27675
    },
    {
      "epoch": 4.515497553017944,
      "grad_norm": 0.026386456564068794,
      "learning_rate": 3.342463384119939e-05,
      "loss": 0.0073,
      "num_input_tokens_seen": 59750320,
      "step": 27680
    },
    {
      "epoch": 4.5163132137031,
      "grad_norm": 0.4252329170703888,
      "learning_rate": 3.3417931849122275e-05,
      "loss": 0.0351,
      "num_input_tokens_seen": 59761520,
      "step": 27685
    },
    {
      "epoch": 4.517128874388255,
      "grad_norm": 0.2085205316543579,
      "learning_rate": 3.341122917464489e-05,
      "loss": 0.1756,
      "num_input_tokens_seen": 59771632,
      "step": 27690
    },
    {
      "epoch": 4.5179445350734095,
      "grad_norm": 2.8549644947052,
      "learning_rate": 3.340452581831057e-05,
      "loss": 0.0695,
      "num_input_tokens_seen": 59781744,
      "step": 27695
    },
    {
      "epoch": 4.518760195758564,
      "grad_norm": 4.925718307495117,
      "learning_rate": 3.3397821780662764e-05,
      "loss": 0.115,
      "num_input_tokens_seen": 59791760,
      "step": 27700
    },
    {
      "epoch": 4.519575856443719,
      "grad_norm": 0.08845800161361694,
      "learning_rate": 3.3391117062244913e-05,
      "loss": 0.1119,
      "num_input_tokens_seen": 59801872,
      "step": 27705
    },
    {
      "epoch": 4.520391517128875,
      "grad_norm": 0.1270817369222641,
      "learning_rate": 3.338441166360054e-05,
      "loss": 0.0802,
      "num_input_tokens_seen": 59811152,
      "step": 27710
    },
    {
      "epoch": 4.52120717781403,
      "grad_norm": 0.10203168541193008,
      "learning_rate": 3.33777055852732e-05,
      "loss": 0.0859,
      "num_input_tokens_seen": 59821264,
      "step": 27715
    },
    {
      "epoch": 4.5220228384991845,
      "grad_norm": 8.421981811523438,
      "learning_rate": 3.3370998827806543e-05,
      "loss": 0.1104,
      "num_input_tokens_seen": 59832336,
      "step": 27720
    },
    {
      "epoch": 4.522838499184339,
      "grad_norm": 0.15523754060268402,
      "learning_rate": 3.336429139174425e-05,
      "loss": 0.0492,
      "num_input_tokens_seen": 59842288,
      "step": 27725
    },
    {
      "epoch": 4.523654159869494,
      "grad_norm": 3.196800947189331,
      "learning_rate": 3.335758327763006e-05,
      "loss": 0.124,
      "num_input_tokens_seen": 59852912,
      "step": 27730
    },
    {
      "epoch": 4.524469820554649,
      "grad_norm": 0.06876061856746674,
      "learning_rate": 3.335087448600776e-05,
      "loss": 0.0119,
      "num_input_tokens_seen": 59863280,
      "step": 27735
    },
    {
      "epoch": 4.525285481239804,
      "grad_norm": 0.10988235473632812,
      "learning_rate": 3.33441650174212e-05,
      "loss": 0.0783,
      "num_input_tokens_seen": 59874288,
      "step": 27740
    },
    {
      "epoch": 4.5261011419249595,
      "grad_norm": 7.433391571044922,
      "learning_rate": 3.3337454872414294e-05,
      "loss": 0.0836,
      "num_input_tokens_seen": 59886000,
      "step": 27745
    },
    {
      "epoch": 4.526916802610114,
      "grad_norm": 4.332917213439941,
      "learning_rate": 3.333074405153098e-05,
      "loss": 0.0186,
      "num_input_tokens_seen": 59896432,
      "step": 27750
    },
    {
      "epoch": 4.527732463295269,
      "grad_norm": 7.137960433959961,
      "learning_rate": 3.332403255531529e-05,
      "loss": 0.2452,
      "num_input_tokens_seen": 59906704,
      "step": 27755
    },
    {
      "epoch": 4.528548123980424,
      "grad_norm": 3.40251088142395,
      "learning_rate": 3.331732038431129e-05,
      "loss": 0.0759,
      "num_input_tokens_seen": 59917392,
      "step": 27760
    },
    {
      "epoch": 4.529363784665579,
      "grad_norm": 0.0433967299759388,
      "learning_rate": 3.3310607539063096e-05,
      "loss": 0.2353,
      "num_input_tokens_seen": 59927984,
      "step": 27765
    },
    {
      "epoch": 4.5301794453507345,
      "grad_norm": 0.17781051993370056,
      "learning_rate": 3.3303894020114886e-05,
      "loss": 0.0743,
      "num_input_tokens_seen": 59938160,
      "step": 27770
    },
    {
      "epoch": 4.530995106035889,
      "grad_norm": 0.1469729095697403,
      "learning_rate": 3.329717982801089e-05,
      "loss": 0.0073,
      "num_input_tokens_seen": 59950000,
      "step": 27775
    },
    {
      "epoch": 4.531810766721044,
      "grad_norm": 0.36192265152931213,
      "learning_rate": 3.32904649632954e-05,
      "loss": 0.0058,
      "num_input_tokens_seen": 59960720,
      "step": 27780
    },
    {
      "epoch": 4.532626427406199,
      "grad_norm": 0.03430166840553284,
      "learning_rate": 3.328374942651275e-05,
      "loss": 0.004,
      "num_input_tokens_seen": 59970864,
      "step": 27785
    },
    {
      "epoch": 4.533442088091354,
      "grad_norm": 4.052310466766357,
      "learning_rate": 3.3277033218207346e-05,
      "loss": 0.1539,
      "num_input_tokens_seen": 59982256,
      "step": 27790
    },
    {
      "epoch": 4.5342577487765094,
      "grad_norm": 0.07250082492828369,
      "learning_rate": 3.327031633892363e-05,
      "loss": 0.1253,
      "num_input_tokens_seen": 59993584,
      "step": 27795
    },
    {
      "epoch": 4.535073409461664,
      "grad_norm": 0.08707235008478165,
      "learning_rate": 3.32635987892061e-05,
      "loss": 0.0591,
      "num_input_tokens_seen": 60004720,
      "step": 27800
    },
    {
      "epoch": 4.535889070146819,
      "grad_norm": 0.17127718031406403,
      "learning_rate": 3.3256880569599335e-05,
      "loss": 0.1543,
      "num_input_tokens_seen": 60015120,
      "step": 27805
    },
    {
      "epoch": 4.536704730831974,
      "grad_norm": 0.04245676100254059,
      "learning_rate": 3.325016168064794e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 60024752,
      "step": 27810
    },
    {
      "epoch": 4.537520391517129,
      "grad_norm": 0.213352769613266,
      "learning_rate": 3.324344212289657e-05,
      "loss": 0.0901,
      "num_input_tokens_seen": 60037072,
      "step": 27815
    },
    {
      "epoch": 4.5383360522022835,
      "grad_norm": 0.06449801474809647,
      "learning_rate": 3.3236721896889954e-05,
      "loss": 0.0429,
      "num_input_tokens_seen": 60048144,
      "step": 27820
    },
    {
      "epoch": 4.539151712887438,
      "grad_norm": 0.10689418762922287,
      "learning_rate": 3.323000100317287e-05,
      "loss": 0.1278,
      "num_input_tokens_seen": 60058544,
      "step": 27825
    },
    {
      "epoch": 4.539967373572594,
      "grad_norm": 5.112598419189453,
      "learning_rate": 3.3223279442290146e-05,
      "loss": 0.019,
      "num_input_tokens_seen": 60068688,
      "step": 27830
    },
    {
      "epoch": 4.540783034257749,
      "grad_norm": 0.2425864189863205,
      "learning_rate": 3.321655721478667e-05,
      "loss": 0.0157,
      "num_input_tokens_seen": 60079664,
      "step": 27835
    },
    {
      "epoch": 4.541598694942904,
      "grad_norm": 6.997628688812256,
      "learning_rate": 3.320983432120737e-05,
      "loss": 0.1535,
      "num_input_tokens_seen": 60089296,
      "step": 27840
    },
    {
      "epoch": 4.5424143556280585,
      "grad_norm": 0.0939311757683754,
      "learning_rate": 3.320311076209724e-05,
      "loss": 0.1832,
      "num_input_tokens_seen": 60099568,
      "step": 27845
    },
    {
      "epoch": 4.543230016313213,
      "grad_norm": 0.20239602029323578,
      "learning_rate": 3.3196386538001346e-05,
      "loss": 0.0991,
      "num_input_tokens_seen": 60110160,
      "step": 27850
    },
    {
      "epoch": 4.544045676998369,
      "grad_norm": 12.30156421661377,
      "learning_rate": 3.3189661649464754e-05,
      "loss": 0.0411,
      "num_input_tokens_seen": 60120880,
      "step": 27855
    },
    {
      "epoch": 4.544861337683524,
      "grad_norm": 0.08484846353530884,
      "learning_rate": 3.318293609703264e-05,
      "loss": 0.0251,
      "num_input_tokens_seen": 60132336,
      "step": 27860
    },
    {
      "epoch": 4.545676998368679,
      "grad_norm": 1.118166446685791,
      "learning_rate": 3.3176209881250206e-05,
      "loss": 0.0092,
      "num_input_tokens_seen": 60142608,
      "step": 27865
    },
    {
      "epoch": 4.5464926590538335,
      "grad_norm": 2.9867448806762695,
      "learning_rate": 3.3169483002662714e-05,
      "loss": 0.0851,
      "num_input_tokens_seen": 60153744,
      "step": 27870
    },
    {
      "epoch": 4.547308319738988,
      "grad_norm": 0.055605363100767136,
      "learning_rate": 3.316275546181548e-05,
      "loss": 0.0391,
      "num_input_tokens_seen": 60166288,
      "step": 27875
    },
    {
      "epoch": 4.548123980424144,
      "grad_norm": 3.6087186336517334,
      "learning_rate": 3.315602725925387e-05,
      "loss": 0.2699,
      "num_input_tokens_seen": 60176208,
      "step": 27880
    },
    {
      "epoch": 4.548939641109299,
      "grad_norm": 4.053160190582275,
      "learning_rate": 3.314929839552331e-05,
      "loss": 0.1586,
      "num_input_tokens_seen": 60187088,
      "step": 27885
    },
    {
      "epoch": 4.549755301794454,
      "grad_norm": 0.20458002388477325,
      "learning_rate": 3.314256887116927e-05,
      "loss": 0.2275,
      "num_input_tokens_seen": 60197488,
      "step": 27890
    },
    {
      "epoch": 4.5505709624796085,
      "grad_norm": 0.27609655261039734,
      "learning_rate": 3.313583868673728e-05,
      "loss": 0.0312,
      "num_input_tokens_seen": 60208816,
      "step": 27895
    },
    {
      "epoch": 4.551386623164763,
      "grad_norm": 0.4201211631298065,
      "learning_rate": 3.312910784277293e-05,
      "loss": 0.0431,
      "num_input_tokens_seen": 60219728,
      "step": 27900
    },
    {
      "epoch": 4.552202283849918,
      "grad_norm": 4.90434455871582,
      "learning_rate": 3.312237633982185e-05,
      "loss": 0.0687,
      "num_input_tokens_seen": 60230288,
      "step": 27905
    },
    {
      "epoch": 4.553017944535073,
      "grad_norm": 0.03794803097844124,
      "learning_rate": 3.3115644178429725e-05,
      "loss": 0.0302,
      "num_input_tokens_seen": 60242384,
      "step": 27910
    },
    {
      "epoch": 4.553833605220229,
      "grad_norm": 0.02601797692477703,
      "learning_rate": 3.310891135914231e-05,
      "loss": 0.0415,
      "num_input_tokens_seen": 60253360,
      "step": 27915
    },
    {
      "epoch": 4.554649265905383,
      "grad_norm": 0.05846745893359184,
      "learning_rate": 3.31021778825054e-05,
      "loss": 0.1983,
      "num_input_tokens_seen": 60263440,
      "step": 27920
    },
    {
      "epoch": 4.555464926590538,
      "grad_norm": 7.333343505859375,
      "learning_rate": 3.309544374906484e-05,
      "loss": 0.0208,
      "num_input_tokens_seen": 60273392,
      "step": 27925
    },
    {
      "epoch": 4.556280587275693,
      "grad_norm": 0.13697630167007446,
      "learning_rate": 3.308870895936652e-05,
      "loss": 0.19,
      "num_input_tokens_seen": 60285424,
      "step": 27930
    },
    {
      "epoch": 4.557096247960848,
      "grad_norm": 5.5662102699279785,
      "learning_rate": 3.308197351395643e-05,
      "loss": 0.078,
      "num_input_tokens_seen": 60295248,
      "step": 27935
    },
    {
      "epoch": 4.557911908646004,
      "grad_norm": 0.052226513624191284,
      "learning_rate": 3.3075237413380545e-05,
      "loss": 0.1936,
      "num_input_tokens_seen": 60306320,
      "step": 27940
    },
    {
      "epoch": 4.558727569331158,
      "grad_norm": 0.09828644245862961,
      "learning_rate": 3.306850065818494e-05,
      "loss": 0.0806,
      "num_input_tokens_seen": 60315600,
      "step": 27945
    },
    {
      "epoch": 4.559543230016313,
      "grad_norm": 0.05902625992894173,
      "learning_rate": 3.3061763248915744e-05,
      "loss": 0.2177,
      "num_input_tokens_seen": 60326608,
      "step": 27950
    },
    {
      "epoch": 4.560358890701468,
      "grad_norm": 0.0987459197640419,
      "learning_rate": 3.305502518611911e-05,
      "loss": 0.3177,
      "num_input_tokens_seen": 60336944,
      "step": 27955
    },
    {
      "epoch": 4.561174551386623,
      "grad_norm": 0.16019931435585022,
      "learning_rate": 3.304828647034126e-05,
      "loss": 0.1571,
      "num_input_tokens_seen": 60348656,
      "step": 27960
    },
    {
      "epoch": 4.561990212071779,
      "grad_norm": 5.781607627868652,
      "learning_rate": 3.304154710212847e-05,
      "loss": 0.0661,
      "num_input_tokens_seen": 60359568,
      "step": 27965
    },
    {
      "epoch": 4.562805872756933,
      "grad_norm": 10.70506763458252,
      "learning_rate": 3.303480708202708e-05,
      "loss": 0.1256,
      "num_input_tokens_seen": 60370192,
      "step": 27970
    },
    {
      "epoch": 4.563621533442088,
      "grad_norm": 0.08649314939975739,
      "learning_rate": 3.3028066410583456e-05,
      "loss": 0.0207,
      "num_input_tokens_seen": 60381136,
      "step": 27975
    },
    {
      "epoch": 4.564437194127243,
      "grad_norm": 0.43787726759910583,
      "learning_rate": 3.3021325088344036e-05,
      "loss": 0.1025,
      "num_input_tokens_seen": 60392304,
      "step": 27980
    },
    {
      "epoch": 4.565252854812398,
      "grad_norm": 0.07810968160629272,
      "learning_rate": 3.3014583115855304e-05,
      "loss": 0.2413,
      "num_input_tokens_seen": 60402544,
      "step": 27985
    },
    {
      "epoch": 4.566068515497553,
      "grad_norm": 0.13535502552986145,
      "learning_rate": 3.3007840493663794e-05,
      "loss": 0.021,
      "num_input_tokens_seen": 60413488,
      "step": 27990
    },
    {
      "epoch": 4.566884176182708,
      "grad_norm": 0.15852278470993042,
      "learning_rate": 3.30010972223161e-05,
      "loss": 0.0677,
      "num_input_tokens_seen": 60423696,
      "step": 27995
    },
    {
      "epoch": 4.567699836867863,
      "grad_norm": 0.3404339551925659,
      "learning_rate": 3.2994353302358875e-05,
      "loss": 0.0126,
      "num_input_tokens_seen": 60434576,
      "step": 28000
    },
    {
      "epoch": 4.568515497553018,
      "grad_norm": 17.157421112060547,
      "learning_rate": 3.298760873433881e-05,
      "loss": 0.1839,
      "num_input_tokens_seen": 60443856,
      "step": 28005
    },
    {
      "epoch": 4.569331158238173,
      "grad_norm": 3.211148262023926,
      "learning_rate": 3.298086351880265e-05,
      "loss": 0.1236,
      "num_input_tokens_seen": 60454864,
      "step": 28010
    },
    {
      "epoch": 4.570146818923328,
      "grad_norm": 5.151402473449707,
      "learning_rate": 3.2974117656297194e-05,
      "loss": 0.131,
      "num_input_tokens_seen": 60465456,
      "step": 28015
    },
    {
      "epoch": 4.5709624796084825,
      "grad_norm": 0.16944673657417297,
      "learning_rate": 3.2967371147369306e-05,
      "loss": 0.1117,
      "num_input_tokens_seen": 60476816,
      "step": 28020
    },
    {
      "epoch": 4.571778140293638,
      "grad_norm": 0.49499160051345825,
      "learning_rate": 3.296062399256587e-05,
      "loss": 0.103,
      "num_input_tokens_seen": 60488784,
      "step": 28025
    },
    {
      "epoch": 4.572593800978793,
      "grad_norm": 0.4088912308216095,
      "learning_rate": 3.295387619243389e-05,
      "loss": 0.0229,
      "num_input_tokens_seen": 60500656,
      "step": 28030
    },
    {
      "epoch": 4.573409461663948,
      "grad_norm": 0.9882709383964539,
      "learning_rate": 3.294712774752033e-05,
      "loss": 0.0098,
      "num_input_tokens_seen": 60511664,
      "step": 28035
    },
    {
      "epoch": 4.574225122349103,
      "grad_norm": 7.481235027313232,
      "learning_rate": 3.2940378658372276e-05,
      "loss": 0.1471,
      "num_input_tokens_seen": 60522000,
      "step": 28040
    },
    {
      "epoch": 4.575040783034257,
      "grad_norm": 0.19363392889499664,
      "learning_rate": 3.293362892553684e-05,
      "loss": 0.0344,
      "num_input_tokens_seen": 60532688,
      "step": 28045
    },
    {
      "epoch": 4.575856443719413,
      "grad_norm": 0.08438599854707718,
      "learning_rate": 3.292687854956119e-05,
      "loss": 0.0973,
      "num_input_tokens_seen": 60543152,
      "step": 28050
    },
    {
      "epoch": 4.576672104404568,
      "grad_norm": 0.08524743467569351,
      "learning_rate": 3.292012753099254e-05,
      "loss": 0.0623,
      "num_input_tokens_seen": 60552272,
      "step": 28055
    },
    {
      "epoch": 4.577487765089723,
      "grad_norm": 0.06825464218854904,
      "learning_rate": 3.2913375870378165e-05,
      "loss": 0.1963,
      "num_input_tokens_seen": 60563888,
      "step": 28060
    },
    {
      "epoch": 4.578303425774878,
      "grad_norm": 0.15410315990447998,
      "learning_rate": 3.2906623568265396e-05,
      "loss": 0.3716,
      "num_input_tokens_seen": 60574576,
      "step": 28065
    },
    {
      "epoch": 4.579119086460032,
      "grad_norm": 12.65908145904541,
      "learning_rate": 3.289987062520159e-05,
      "loss": 0.0591,
      "num_input_tokens_seen": 60584848,
      "step": 28070
    },
    {
      "epoch": 4.579934747145187,
      "grad_norm": 2.266758918762207,
      "learning_rate": 3.28931170417342e-05,
      "loss": 0.197,
      "num_input_tokens_seen": 60595312,
      "step": 28075
    },
    {
      "epoch": 4.580750407830343,
      "grad_norm": 0.5498905181884766,
      "learning_rate": 3.288636281841069e-05,
      "loss": 0.0149,
      "num_input_tokens_seen": 60604848,
      "step": 28080
    },
    {
      "epoch": 4.581566068515498,
      "grad_norm": 0.09909345954656601,
      "learning_rate": 3.287960795577859e-05,
      "loss": 0.0449,
      "num_input_tokens_seen": 60614640,
      "step": 28085
    },
    {
      "epoch": 4.582381729200653,
      "grad_norm": 0.5386938452720642,
      "learning_rate": 3.2872852454385495e-05,
      "loss": 0.0078,
      "num_input_tokens_seen": 60624912,
      "step": 28090
    },
    {
      "epoch": 4.583197389885807,
      "grad_norm": 0.22791962325572968,
      "learning_rate": 3.2866096314779035e-05,
      "loss": 0.0075,
      "num_input_tokens_seen": 60637968,
      "step": 28095
    },
    {
      "epoch": 4.584013050570962,
      "grad_norm": 0.03843766823410988,
      "learning_rate": 3.285933953750689e-05,
      "loss": 0.1,
      "num_input_tokens_seen": 60648816,
      "step": 28100
    },
    {
      "epoch": 4.584828711256117,
      "grad_norm": 2.0750088691711426,
      "learning_rate": 3.28525821231168e-05,
      "loss": 0.0982,
      "num_input_tokens_seen": 60659888,
      "step": 28105
    },
    {
      "epoch": 4.585644371941273,
      "grad_norm": 4.184640407562256,
      "learning_rate": 3.284582407215657e-05,
      "loss": 0.1045,
      "num_input_tokens_seen": 60671472,
      "step": 28110
    },
    {
      "epoch": 4.5864600326264275,
      "grad_norm": 7.943884372711182,
      "learning_rate": 3.283906538517403e-05,
      "loss": 0.119,
      "num_input_tokens_seen": 60683696,
      "step": 28115
    },
    {
      "epoch": 4.587275693311582,
      "grad_norm": 5.281688690185547,
      "learning_rate": 3.283230606271707e-05,
      "loss": 0.1073,
      "num_input_tokens_seen": 60694832,
      "step": 28120
    },
    {
      "epoch": 4.588091353996737,
      "grad_norm": 0.15140409767627716,
      "learning_rate": 3.2825546105333634e-05,
      "loss": 0.1108,
      "num_input_tokens_seen": 60706192,
      "step": 28125
    },
    {
      "epoch": 4.588907014681892,
      "grad_norm": 0.09975605458021164,
      "learning_rate": 3.281878551357174e-05,
      "loss": 0.0073,
      "num_input_tokens_seen": 60716880,
      "step": 28130
    },
    {
      "epoch": 4.589722675367048,
      "grad_norm": 4.524142265319824,
      "learning_rate": 3.281202428797941e-05,
      "loss": 0.2411,
      "num_input_tokens_seen": 60727792,
      "step": 28135
    },
    {
      "epoch": 4.5905383360522025,
      "grad_norm": 5.362515449523926,
      "learning_rate": 3.2805262429104755e-05,
      "loss": 0.2446,
      "num_input_tokens_seen": 60738800,
      "step": 28140
    },
    {
      "epoch": 4.591353996737357,
      "grad_norm": 17.26320457458496,
      "learning_rate": 3.279849993749593e-05,
      "loss": 0.1968,
      "num_input_tokens_seen": 60748336,
      "step": 28145
    },
    {
      "epoch": 4.592169657422512,
      "grad_norm": 0.10141994804143906,
      "learning_rate": 3.279173681370112e-05,
      "loss": 0.1395,
      "num_input_tokens_seen": 60759728,
      "step": 28150
    },
    {
      "epoch": 4.592985318107667,
      "grad_norm": 0.8254680037498474,
      "learning_rate": 3.27849730582686e-05,
      "loss": 0.0778,
      "num_input_tokens_seen": 60771184,
      "step": 28155
    },
    {
      "epoch": 4.593800978792823,
      "grad_norm": 0.06635220348834991,
      "learning_rate": 3.2778208671746654e-05,
      "loss": 0.014,
      "num_input_tokens_seen": 60782864,
      "step": 28160
    },
    {
      "epoch": 4.5946166394779775,
      "grad_norm": 13.53496265411377,
      "learning_rate": 3.277144365468365e-05,
      "loss": 0.2678,
      "num_input_tokens_seen": 60794800,
      "step": 28165
    },
    {
      "epoch": 4.595432300163132,
      "grad_norm": 0.04601912200450897,
      "learning_rate": 3.2764678007627994e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 60806224,
      "step": 28170
    },
    {
      "epoch": 4.596247960848287,
      "grad_norm": 0.028869764879345894,
      "learning_rate": 3.275791173112814e-05,
      "loss": 0.1233,
      "num_input_tokens_seen": 60817232,
      "step": 28175
    },
    {
      "epoch": 4.597063621533442,
      "grad_norm": 0.49962377548217773,
      "learning_rate": 3.2751144825732595e-05,
      "loss": 0.1578,
      "num_input_tokens_seen": 60828400,
      "step": 28180
    },
    {
      "epoch": 4.597879282218597,
      "grad_norm": 0.15380997955799103,
      "learning_rate": 3.274437729198992e-05,
      "loss": 0.2603,
      "num_input_tokens_seen": 60840080,
      "step": 28185
    },
    {
      "epoch": 4.598694942903752,
      "grad_norm": 0.30230018496513367,
      "learning_rate": 3.273760913044873e-05,
      "loss": 0.0198,
      "num_input_tokens_seen": 60851088,
      "step": 28190
    },
    {
      "epoch": 4.599510603588907,
      "grad_norm": 0.15211479365825653,
      "learning_rate": 3.273084034165769e-05,
      "loss": 0.12,
      "num_input_tokens_seen": 60862096,
      "step": 28195
    },
    {
      "epoch": 4.600326264274062,
      "grad_norm": 0.040675919502973557,
      "learning_rate": 3.2724070926165495e-05,
      "loss": 0.2047,
      "num_input_tokens_seen": 60872848,
      "step": 28200
    },
    {
      "epoch": 4.601141924959217,
      "grad_norm": 0.2294074147939682,
      "learning_rate": 3.271730088452093e-05,
      "loss": 0.0039,
      "num_input_tokens_seen": 60883760,
      "step": 28205
    },
    {
      "epoch": 4.601957585644372,
      "grad_norm": 3.3079845905303955,
      "learning_rate": 3.2710530217272794e-05,
      "loss": 0.1291,
      "num_input_tokens_seen": 60893456,
      "step": 28210
    },
    {
      "epoch": 4.602773246329527,
      "grad_norm": 0.5316904187202454,
      "learning_rate": 3.270375892496995e-05,
      "loss": 0.0295,
      "num_input_tokens_seen": 60905136,
      "step": 28215
    },
    {
      "epoch": 4.603588907014682,
      "grad_norm": 9.550395011901855,
      "learning_rate": 3.2696987008161325e-05,
      "loss": 0.1919,
      "num_input_tokens_seen": 60915088,
      "step": 28220
    },
    {
      "epoch": 4.604404567699837,
      "grad_norm": 0.0909288227558136,
      "learning_rate": 3.269021446739588e-05,
      "loss": 0.0407,
      "num_input_tokens_seen": 60926704,
      "step": 28225
    },
    {
      "epoch": 4.605220228384992,
      "grad_norm": 3.365037202835083,
      "learning_rate": 3.268344130322262e-05,
      "loss": 0.0203,
      "num_input_tokens_seen": 60937808,
      "step": 28230
    },
    {
      "epoch": 4.606035889070147,
      "grad_norm": 2.849531412124634,
      "learning_rate": 3.2676667516190634e-05,
      "loss": 0.0208,
      "num_input_tokens_seen": 60948112,
      "step": 28235
    },
    {
      "epoch": 4.6068515497553015,
      "grad_norm": 6.55972146987915,
      "learning_rate": 3.266989310684902e-05,
      "loss": 0.3289,
      "num_input_tokens_seen": 60959312,
      "step": 28240
    },
    {
      "epoch": 4.607667210440457,
      "grad_norm": 1.8439161777496338,
      "learning_rate": 3.266311807574697e-05,
      "loss": 0.0294,
      "num_input_tokens_seen": 60970256,
      "step": 28245
    },
    {
      "epoch": 4.608482871125612,
      "grad_norm": 0.03487294167280197,
      "learning_rate": 3.265634242343367e-05,
      "loss": 0.0056,
      "num_input_tokens_seen": 60980752,
      "step": 28250
    },
    {
      "epoch": 4.609298531810767,
      "grad_norm": 0.035358358174562454,
      "learning_rate": 3.264956615045841e-05,
      "loss": 0.0259,
      "num_input_tokens_seen": 60991344,
      "step": 28255
    },
    {
      "epoch": 4.610114192495922,
      "grad_norm": 0.0793282613158226,
      "learning_rate": 3.26427892573705e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 61001264,
      "step": 28260
    },
    {
      "epoch": 4.6109298531810765,
      "grad_norm": 0.04088451713323593,
      "learning_rate": 3.263601174471932e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 61011408,
      "step": 28265
    },
    {
      "epoch": 4.611745513866231,
      "grad_norm": 2.698221206665039,
      "learning_rate": 3.262923361305429e-05,
      "loss": 0.0359,
      "num_input_tokens_seen": 61020944,
      "step": 28270
    },
    {
      "epoch": 4.612561174551386,
      "grad_norm": 2.7350680828094482,
      "learning_rate": 3.262245486292486e-05,
      "loss": 0.075,
      "num_input_tokens_seen": 61031248,
      "step": 28275
    },
    {
      "epoch": 4.613376835236542,
      "grad_norm": 3.3778626918792725,
      "learning_rate": 3.261567549488056e-05,
      "loss": 0.166,
      "num_input_tokens_seen": 61042320,
      "step": 28280
    },
    {
      "epoch": 4.614192495921697,
      "grad_norm": 7.231352806091309,
      "learning_rate": 3.260889550947098e-05,
      "loss": 0.2126,
      "num_input_tokens_seen": 61053520,
      "step": 28285
    },
    {
      "epoch": 4.6150081566068515,
      "grad_norm": 0.32570576667785645,
      "learning_rate": 3.260211490724571e-05,
      "loss": 0.0126,
      "num_input_tokens_seen": 61062960,
      "step": 28290
    },
    {
      "epoch": 4.615823817292006,
      "grad_norm": 0.09390170872211456,
      "learning_rate": 3.259533368875444e-05,
      "loss": 0.0041,
      "num_input_tokens_seen": 61073392,
      "step": 28295
    },
    {
      "epoch": 4.616639477977161,
      "grad_norm": 0.01974472776055336,
      "learning_rate": 3.2588551854546876e-05,
      "loss": 0.0355,
      "num_input_tokens_seen": 61083984,
      "step": 28300
    },
    {
      "epoch": 4.617455138662317,
      "grad_norm": 8.095100402832031,
      "learning_rate": 3.2581769405172805e-05,
      "loss": 0.0251,
      "num_input_tokens_seen": 61095280,
      "step": 28305
    },
    {
      "epoch": 4.618270799347472,
      "grad_norm": 0.06068865582346916,
      "learning_rate": 3.2574986341182026e-05,
      "loss": 0.0151,
      "num_input_tokens_seen": 61107024,
      "step": 28310
    },
    {
      "epoch": 4.6190864600326265,
      "grad_norm": 0.045194510370492935,
      "learning_rate": 3.256820266312442e-05,
      "loss": 0.0635,
      "num_input_tokens_seen": 61117552,
      "step": 28315
    },
    {
      "epoch": 4.619902120717781,
      "grad_norm": 3.254822254180908,
      "learning_rate": 3.256141837154991e-05,
      "loss": 0.0813,
      "num_input_tokens_seen": 61129040,
      "step": 28320
    },
    {
      "epoch": 4.620717781402936,
      "grad_norm": 0.08180870115756989,
      "learning_rate": 3.255463346700846e-05,
      "loss": 0.0587,
      "num_input_tokens_seen": 61139664,
      "step": 28325
    },
    {
      "epoch": 4.621533442088092,
      "grad_norm": 8.497345924377441,
      "learning_rate": 3.254784795005008e-05,
      "loss": 0.3768,
      "num_input_tokens_seen": 61151728,
      "step": 28330
    },
    {
      "epoch": 4.622349102773247,
      "grad_norm": 17.47870445251465,
      "learning_rate": 3.254106182122486e-05,
      "loss": 0.1989,
      "num_input_tokens_seen": 61162448,
      "step": 28335
    },
    {
      "epoch": 4.623164763458401,
      "grad_norm": 3.7979838848114014,
      "learning_rate": 3.2534275081082896e-05,
      "loss": 0.1276,
      "num_input_tokens_seen": 61173136,
      "step": 28340
    },
    {
      "epoch": 4.623980424143556,
      "grad_norm": 8.182798385620117,
      "learning_rate": 3.252748773017437e-05,
      "loss": 0.2062,
      "num_input_tokens_seen": 61183120,
      "step": 28345
    },
    {
      "epoch": 4.624796084828711,
      "grad_norm": 0.25087881088256836,
      "learning_rate": 3.2520699769049496e-05,
      "loss": 0.0131,
      "num_input_tokens_seen": 61194000,
      "step": 28350
    },
    {
      "epoch": 4.625611745513866,
      "grad_norm": 8.378771781921387,
      "learning_rate": 3.251391119825854e-05,
      "loss": 0.0302,
      "num_input_tokens_seen": 61204688,
      "step": 28355
    },
    {
      "epoch": 4.626427406199021,
      "grad_norm": 3.5833957195281982,
      "learning_rate": 3.2507122018351815e-05,
      "loss": 0.191,
      "num_input_tokens_seen": 61215376,
      "step": 28360
    },
    {
      "epoch": 4.627243066884176,
      "grad_norm": 1.5312604904174805,
      "learning_rate": 3.250033222987969e-05,
      "loss": 0.0091,
      "num_input_tokens_seen": 61226256,
      "step": 28365
    },
    {
      "epoch": 4.628058727569331,
      "grad_norm": 0.10052099823951721,
      "learning_rate": 3.2493541833392575e-05,
      "loss": 0.0093,
      "num_input_tokens_seen": 61238192,
      "step": 28370
    },
    {
      "epoch": 4.628874388254486,
      "grad_norm": 0.09227970242500305,
      "learning_rate": 3.2486750829440946e-05,
      "loss": 0.1166,
      "num_input_tokens_seen": 61248944,
      "step": 28375
    },
    {
      "epoch": 4.629690048939641,
      "grad_norm": 0.04053603112697601,
      "learning_rate": 3.2479959218575295e-05,
      "loss": 0.0964,
      "num_input_tokens_seen": 61260752,
      "step": 28380
    },
    {
      "epoch": 4.630505709624796,
      "grad_norm": 2.4717376232147217,
      "learning_rate": 3.24731670013462e-05,
      "loss": 0.3137,
      "num_input_tokens_seen": 61271984,
      "step": 28385
    },
    {
      "epoch": 4.631321370309951,
      "grad_norm": 0.34996065497398376,
      "learning_rate": 3.246637417830427e-05,
      "loss": 0.0561,
      "num_input_tokens_seen": 61283248,
      "step": 28390
    },
    {
      "epoch": 4.632137030995106,
      "grad_norm": 0.12414489686489105,
      "learning_rate": 3.245958075000017e-05,
      "loss": 0.0059,
      "num_input_tokens_seen": 61294512,
      "step": 28395
    },
    {
      "epoch": 4.632952691680261,
      "grad_norm": 5.165646553039551,
      "learning_rate": 3.24527867169846e-05,
      "loss": 0.0387,
      "num_input_tokens_seen": 61304944,
      "step": 28400
    },
    {
      "epoch": 4.633768352365416,
      "grad_norm": 0.13805988430976868,
      "learning_rate": 3.244599207980833e-05,
      "loss": 0.0068,
      "num_input_tokens_seen": 61315408,
      "step": 28405
    },
    {
      "epoch": 4.634584013050571,
      "grad_norm": 0.4154154658317566,
      "learning_rate": 3.243919683902216e-05,
      "loss": 0.2016,
      "num_input_tokens_seen": 61326000,
      "step": 28410
    },
    {
      "epoch": 4.635399673735726,
      "grad_norm": 0.060973986983299255,
      "learning_rate": 3.2432400995176934e-05,
      "loss": 0.1803,
      "num_input_tokens_seen": 61335984,
      "step": 28415
    },
    {
      "epoch": 4.636215334420881,
      "grad_norm": 0.16784125566482544,
      "learning_rate": 3.242560454882359e-05,
      "loss": 0.0277,
      "num_input_tokens_seen": 61347344,
      "step": 28420
    },
    {
      "epoch": 4.637030995106036,
      "grad_norm": 3.4738075733184814,
      "learning_rate": 3.241880750051306e-05,
      "loss": 0.2198,
      "num_input_tokens_seen": 61358352,
      "step": 28425
    },
    {
      "epoch": 4.637846655791191,
      "grad_norm": 0.04191743582487106,
      "learning_rate": 3.241200985079634e-05,
      "loss": 0.0093,
      "num_input_tokens_seen": 61369168,
      "step": 28430
    },
    {
      "epoch": 4.638662316476346,
      "grad_norm": 3.6386828422546387,
      "learning_rate": 3.2405211600224503e-05,
      "loss": 0.2868,
      "num_input_tokens_seen": 61379984,
      "step": 28435
    },
    {
      "epoch": 4.6394779771615005,
      "grad_norm": 0.11262372881174088,
      "learning_rate": 3.239841274934863e-05,
      "loss": 0.0778,
      "num_input_tokens_seen": 61391760,
      "step": 28440
    },
    {
      "epoch": 4.640293637846656,
      "grad_norm": 0.34785395860671997,
      "learning_rate": 3.239161329871989e-05,
      "loss": 0.1569,
      "num_input_tokens_seen": 61401680,
      "step": 28445
    },
    {
      "epoch": 4.641109298531811,
      "grad_norm": 0.08060356229543686,
      "learning_rate": 3.2384813248889475e-05,
      "loss": 0.0088,
      "num_input_tokens_seen": 61411696,
      "step": 28450
    },
    {
      "epoch": 4.641924959216966,
      "grad_norm": 0.478834867477417,
      "learning_rate": 3.2378012600408625e-05,
      "loss": 0.0909,
      "num_input_tokens_seen": 61421968,
      "step": 28455
    },
    {
      "epoch": 4.642740619902121,
      "grad_norm": 5.574655055999756,
      "learning_rate": 3.2371211353828636e-05,
      "loss": 0.0306,
      "num_input_tokens_seen": 61430960,
      "step": 28460
    },
    {
      "epoch": 4.643556280587275,
      "grad_norm": 0.6605509519577026,
      "learning_rate": 3.236440950970085e-05,
      "loss": 0.3075,
      "num_input_tokens_seen": 61442032,
      "step": 28465
    },
    {
      "epoch": 4.64437194127243,
      "grad_norm": 3.7971913814544678,
      "learning_rate": 3.2357607068576664e-05,
      "loss": 0.2212,
      "num_input_tokens_seen": 61453136,
      "step": 28470
    },
    {
      "epoch": 4.645187601957586,
      "grad_norm": 6.3657941818237305,
      "learning_rate": 3.2350804031007524e-05,
      "loss": 0.1452,
      "num_input_tokens_seen": 61464048,
      "step": 28475
    },
    {
      "epoch": 4.646003262642741,
      "grad_norm": 0.04483633488416672,
      "learning_rate": 3.234400039754491e-05,
      "loss": 0.1186,
      "num_input_tokens_seen": 61475344,
      "step": 28480
    },
    {
      "epoch": 4.646818923327896,
      "grad_norm": 0.25604698061943054,
      "learning_rate": 3.2337196168740356e-05,
      "loss": 0.0275,
      "num_input_tokens_seen": 61485168,
      "step": 28485
    },
    {
      "epoch": 4.64763458401305,
      "grad_norm": 4.546419620513916,
      "learning_rate": 3.233039134514545e-05,
      "loss": 0.2129,
      "num_input_tokens_seen": 61496304,
      "step": 28490
    },
    {
      "epoch": 4.648450244698205,
      "grad_norm": 6.424332618713379,
      "learning_rate": 3.2323585927311825e-05,
      "loss": 0.0382,
      "num_input_tokens_seen": 61507088,
      "step": 28495
    },
    {
      "epoch": 4.649265905383361,
      "grad_norm": 2.768878936767578,
      "learning_rate": 3.231677991579118e-05,
      "loss": 0.1253,
      "num_input_tokens_seen": 61518032,
      "step": 28500
    },
    {
      "epoch": 4.650081566068516,
      "grad_norm": 0.47725772857666016,
      "learning_rate": 3.230997331113521e-05,
      "loss": 0.0204,
      "num_input_tokens_seen": 61529424,
      "step": 28505
    },
    {
      "epoch": 4.650897226753671,
      "grad_norm": 0.20057564973831177,
      "learning_rate": 3.230316611389573e-05,
      "loss": 0.0122,
      "num_input_tokens_seen": 61540976,
      "step": 28510
    },
    {
      "epoch": 4.651712887438825,
      "grad_norm": 0.1292143017053604,
      "learning_rate": 3.229635832462454e-05,
      "loss": 0.0137,
      "num_input_tokens_seen": 61552368,
      "step": 28515
    },
    {
      "epoch": 4.65252854812398,
      "grad_norm": 0.1415448933839798,
      "learning_rate": 3.228954994387352e-05,
      "loss": 0.0073,
      "num_input_tokens_seen": 61563184,
      "step": 28520
    },
    {
      "epoch": 4.653344208809135,
      "grad_norm": 4.261139392852783,
      "learning_rate": 3.2282740972194606e-05,
      "loss": 0.1982,
      "num_input_tokens_seen": 61572720,
      "step": 28525
    },
    {
      "epoch": 4.654159869494291,
      "grad_norm": 0.08284708112478256,
      "learning_rate": 3.2275931410139755e-05,
      "loss": 0.0975,
      "num_input_tokens_seen": 61584176,
      "step": 28530
    },
    {
      "epoch": 4.6549755301794455,
      "grad_norm": 0.12331119179725647,
      "learning_rate": 3.226912125826098e-05,
      "loss": 0.0922,
      "num_input_tokens_seen": 61595152,
      "step": 28535
    },
    {
      "epoch": 4.6557911908646,
      "grad_norm": 0.21844713389873505,
      "learning_rate": 3.226231051711035e-05,
      "loss": 0.083,
      "num_input_tokens_seen": 61606032,
      "step": 28540
    },
    {
      "epoch": 4.656606851549755,
      "grad_norm": 0.05081196874380112,
      "learning_rate": 3.225549918723999e-05,
      "loss": 0.0321,
      "num_input_tokens_seen": 61617680,
      "step": 28545
    },
    {
      "epoch": 4.65742251223491,
      "grad_norm": 0.1778022050857544,
      "learning_rate": 3.224868726920205e-05,
      "loss": 0.1595,
      "num_input_tokens_seen": 61628080,
      "step": 28550
    },
    {
      "epoch": 4.658238172920065,
      "grad_norm": 0.19769565761089325,
      "learning_rate": 3.224187476354873e-05,
      "loss": 0.2647,
      "num_input_tokens_seen": 61639440,
      "step": 28555
    },
    {
      "epoch": 4.6590538336052205,
      "grad_norm": 8.99872875213623,
      "learning_rate": 3.223506167083231e-05,
      "loss": 0.1861,
      "num_input_tokens_seen": 61650544,
      "step": 28560
    },
    {
      "epoch": 4.659869494290375,
      "grad_norm": 0.1628834456205368,
      "learning_rate": 3.222824799160508e-05,
      "loss": 0.2301,
      "num_input_tokens_seen": 61661232,
      "step": 28565
    },
    {
      "epoch": 4.66068515497553,
      "grad_norm": 1.367659330368042,
      "learning_rate": 3.222143372641938e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 61671984,
      "step": 28570
    },
    {
      "epoch": 4.661500815660685,
      "grad_norm": 0.21327966451644897,
      "learning_rate": 3.2214618875827626e-05,
      "loss": 0.04,
      "num_input_tokens_seen": 61682160,
      "step": 28575
    },
    {
      "epoch": 4.66231647634584,
      "grad_norm": 0.03936144337058067,
      "learning_rate": 3.220780344038227e-05,
      "loss": 0.1217,
      "num_input_tokens_seen": 61692112,
      "step": 28580
    },
    {
      "epoch": 4.6631321370309955,
      "grad_norm": 0.05543329194188118,
      "learning_rate": 3.220098742063578e-05,
      "loss": 0.1951,
      "num_input_tokens_seen": 61703312,
      "step": 28585
    },
    {
      "epoch": 4.66394779771615,
      "grad_norm": 2.2213213443756104,
      "learning_rate": 3.219417081714072e-05,
      "loss": 0.0089,
      "num_input_tokens_seen": 61713616,
      "step": 28590
    },
    {
      "epoch": 4.664763458401305,
      "grad_norm": 0.05598160997033119,
      "learning_rate": 3.218735363044967e-05,
      "loss": 0.2321,
      "num_input_tokens_seen": 61724080,
      "step": 28595
    },
    {
      "epoch": 4.66557911908646,
      "grad_norm": 4.151381015777588,
      "learning_rate": 3.218053586111526e-05,
      "loss": 0.1904,
      "num_input_tokens_seen": 61735216,
      "step": 28600
    },
    {
      "epoch": 4.666394779771615,
      "grad_norm": 0.36127719283103943,
      "learning_rate": 3.217371750969019e-05,
      "loss": 0.0418,
      "num_input_tokens_seen": 61746736,
      "step": 28605
    },
    {
      "epoch": 4.6672104404567705,
      "grad_norm": 4.928699493408203,
      "learning_rate": 3.2166898576727176e-05,
      "loss": 0.1783,
      "num_input_tokens_seen": 61758032,
      "step": 28610
    },
    {
      "epoch": 4.668026101141925,
      "grad_norm": 2.978797674179077,
      "learning_rate": 3.2160079062779005e-05,
      "loss": 0.2182,
      "num_input_tokens_seen": 61768624,
      "step": 28615
    },
    {
      "epoch": 4.66884176182708,
      "grad_norm": 1.138555645942688,
      "learning_rate": 3.215325896839848e-05,
      "loss": 0.0115,
      "num_input_tokens_seen": 61779408,
      "step": 28620
    },
    {
      "epoch": 4.669657422512235,
      "grad_norm": 1.297851800918579,
      "learning_rate": 3.2146438294138505e-05,
      "loss": 0.0284,
      "num_input_tokens_seen": 61790320,
      "step": 28625
    },
    {
      "epoch": 4.67047308319739,
      "grad_norm": 0.3052097260951996,
      "learning_rate": 3.2139617040551966e-05,
      "loss": 0.0303,
      "num_input_tokens_seen": 61801968,
      "step": 28630
    },
    {
      "epoch": 4.671288743882545,
      "grad_norm": 0.12733018398284912,
      "learning_rate": 3.2132795208191853e-05,
      "loss": 0.18,
      "num_input_tokens_seen": 61812944,
      "step": 28635
    },
    {
      "epoch": 4.672104404567699,
      "grad_norm": 6.405884265899658,
      "learning_rate": 3.212597279761116e-05,
      "loss": 0.0533,
      "num_input_tokens_seen": 61824912,
      "step": 28640
    },
    {
      "epoch": 4.672920065252855,
      "grad_norm": 0.052328865975141525,
      "learning_rate": 3.211914980936296e-05,
      "loss": 0.1802,
      "num_input_tokens_seen": 61835120,
      "step": 28645
    },
    {
      "epoch": 4.67373572593801,
      "grad_norm": 0.046687010675668716,
      "learning_rate": 3.2112326244000355e-05,
      "loss": 0.0724,
      "num_input_tokens_seen": 61845680,
      "step": 28650
    },
    {
      "epoch": 4.674551386623165,
      "grad_norm": 3.757732629776001,
      "learning_rate": 3.2105502102076494e-05,
      "loss": 0.1822,
      "num_input_tokens_seen": 61856368,
      "step": 28655
    },
    {
      "epoch": 4.6753670473083195,
      "grad_norm": 0.09236893057823181,
      "learning_rate": 3.209867738414459e-05,
      "loss": 0.0629,
      "num_input_tokens_seen": 61866448,
      "step": 28660
    },
    {
      "epoch": 4.676182707993474,
      "grad_norm": 0.2082487940788269,
      "learning_rate": 3.2091852090757865e-05,
      "loss": 0.1419,
      "num_input_tokens_seen": 61877040,
      "step": 28665
    },
    {
      "epoch": 4.67699836867863,
      "grad_norm": 3.1891257762908936,
      "learning_rate": 3.208502622246964e-05,
      "loss": 0.1606,
      "num_input_tokens_seen": 61887472,
      "step": 28670
    },
    {
      "epoch": 4.677814029363785,
      "grad_norm": 0.5192645788192749,
      "learning_rate": 3.207819977983323e-05,
      "loss": 0.0086,
      "num_input_tokens_seen": 61898864,
      "step": 28675
    },
    {
      "epoch": 4.67862969004894,
      "grad_norm": 0.9870857000350952,
      "learning_rate": 3.207137276340203e-05,
      "loss": 0.0099,
      "num_input_tokens_seen": 61910160,
      "step": 28680
    },
    {
      "epoch": 4.6794453507340945,
      "grad_norm": 3.560532569885254,
      "learning_rate": 3.206454517372949e-05,
      "loss": 0.3913,
      "num_input_tokens_seen": 61920176,
      "step": 28685
    },
    {
      "epoch": 4.680261011419249,
      "grad_norm": 3.4725193977355957,
      "learning_rate": 3.205771701136906e-05,
      "loss": 0.2192,
      "num_input_tokens_seen": 61930992,
      "step": 28690
    },
    {
      "epoch": 4.681076672104405,
      "grad_norm": 0.3666292428970337,
      "learning_rate": 3.205088827687428e-05,
      "loss": 0.1232,
      "num_input_tokens_seen": 61941744,
      "step": 28695
    },
    {
      "epoch": 4.68189233278956,
      "grad_norm": 0.22167155146598816,
      "learning_rate": 3.204405897079872e-05,
      "loss": 0.1114,
      "num_input_tokens_seen": 61951760,
      "step": 28700
    },
    {
      "epoch": 4.682707993474715,
      "grad_norm": 1.9301999807357788,
      "learning_rate": 3.2037229093696e-05,
      "loss": 0.3247,
      "num_input_tokens_seen": 61961040,
      "step": 28705
    },
    {
      "epoch": 4.6835236541598695,
      "grad_norm": 0.08697057515382767,
      "learning_rate": 3.203039864611978e-05,
      "loss": 0.0152,
      "num_input_tokens_seen": 61971856,
      "step": 28710
    },
    {
      "epoch": 4.684339314845024,
      "grad_norm": 0.22499242424964905,
      "learning_rate": 3.202356762862377e-05,
      "loss": 0.0114,
      "num_input_tokens_seen": 61983760,
      "step": 28715
    },
    {
      "epoch": 4.685154975530179,
      "grad_norm": 0.2075299471616745,
      "learning_rate": 3.201673604176174e-05,
      "loss": 0.0102,
      "num_input_tokens_seen": 61994480,
      "step": 28720
    },
    {
      "epoch": 4.685970636215334,
      "grad_norm": 3.208085060119629,
      "learning_rate": 3.2009903886087476e-05,
      "loss": 0.1886,
      "num_input_tokens_seen": 62005360,
      "step": 28725
    },
    {
      "epoch": 4.68678629690049,
      "grad_norm": 0.27816641330718994,
      "learning_rate": 3.200307116215485e-05,
      "loss": 0.1239,
      "num_input_tokens_seen": 62016976,
      "step": 28730
    },
    {
      "epoch": 4.6876019575856445,
      "grad_norm": 0.21742834150791168,
      "learning_rate": 3.1996237870517734e-05,
      "loss": 0.0602,
      "num_input_tokens_seen": 62028176,
      "step": 28735
    },
    {
      "epoch": 4.688417618270799,
      "grad_norm": 0.1057346984744072,
      "learning_rate": 3.198940401173007e-05,
      "loss": 0.1379,
      "num_input_tokens_seen": 62038672,
      "step": 28740
    },
    {
      "epoch": 4.689233278955954,
      "grad_norm": 0.11908625811338425,
      "learning_rate": 3.198256958634586e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 62049424,
      "step": 28745
    },
    {
      "epoch": 4.690048939641109,
      "grad_norm": 0.7854670286178589,
      "learning_rate": 3.197573459491913e-05,
      "loss": 0.0882,
      "num_input_tokens_seen": 62060112,
      "step": 28750
    },
    {
      "epoch": 4.690864600326265,
      "grad_norm": 0.11598500609397888,
      "learning_rate": 3.1968899038003965e-05,
      "loss": 0.1178,
      "num_input_tokens_seen": 62071728,
      "step": 28755
    },
    {
      "epoch": 4.691680261011419,
      "grad_norm": 0.1409139782190323,
      "learning_rate": 3.196206291615447e-05,
      "loss": 0.2613,
      "num_input_tokens_seen": 62081680,
      "step": 28760
    },
    {
      "epoch": 4.692495921696574,
      "grad_norm": 3.294092893600464,
      "learning_rate": 3.195522622992484e-05,
      "loss": 0.0389,
      "num_input_tokens_seen": 62091984,
      "step": 28765
    },
    {
      "epoch": 4.693311582381729,
      "grad_norm": 0.08398202806711197,
      "learning_rate": 3.1948388979869277e-05,
      "loss": 0.0969,
      "num_input_tokens_seen": 62102864,
      "step": 28770
    },
    {
      "epoch": 4.694127243066884,
      "grad_norm": 2.166158437728882,
      "learning_rate": 3.194155116654205e-05,
      "loss": 0.1787,
      "num_input_tokens_seen": 62114256,
      "step": 28775
    },
    {
      "epoch": 4.69494290375204,
      "grad_norm": 0.26545998454093933,
      "learning_rate": 3.193471279049746e-05,
      "loss": 0.011,
      "num_input_tokens_seen": 62124816,
      "step": 28780
    },
    {
      "epoch": 4.695758564437194,
      "grad_norm": 0.15841230750083923,
      "learning_rate": 3.192787385228987e-05,
      "loss": 0.0413,
      "num_input_tokens_seen": 62136368,
      "step": 28785
    },
    {
      "epoch": 4.696574225122349,
      "grad_norm": 0.11272495239973068,
      "learning_rate": 3.192103435247368e-05,
      "loss": 0.2126,
      "num_input_tokens_seen": 62147024,
      "step": 28790
    },
    {
      "epoch": 4.697389885807504,
      "grad_norm": 0.5528010129928589,
      "learning_rate": 3.1914194291603313e-05,
      "loss": 0.0297,
      "num_input_tokens_seen": 62156848,
      "step": 28795
    },
    {
      "epoch": 4.698205546492659,
      "grad_norm": 0.16787166893482208,
      "learning_rate": 3.190735367023328e-05,
      "loss": 0.0539,
      "num_input_tokens_seen": 62168496,
      "step": 28800
    },
    {
      "epoch": 4.699021207177814,
      "grad_norm": 1.6341058015823364,
      "learning_rate": 3.1900512488918114e-05,
      "loss": 0.1337,
      "num_input_tokens_seen": 62178800,
      "step": 28805
    },
    {
      "epoch": 4.699836867862969,
      "grad_norm": 0.21060863137245178,
      "learning_rate": 3.189367074821239e-05,
      "loss": 0.0353,
      "num_input_tokens_seen": 62190768,
      "step": 28810
    },
    {
      "epoch": 4.700652528548124,
      "grad_norm": 0.07222887128591537,
      "learning_rate": 3.1886828448670734e-05,
      "loss": 0.0087,
      "num_input_tokens_seen": 62201648,
      "step": 28815
    },
    {
      "epoch": 4.701468189233279,
      "grad_norm": 0.5213015675544739,
      "learning_rate": 3.1879985590847824e-05,
      "loss": 0.1385,
      "num_input_tokens_seen": 62212816,
      "step": 28820
    },
    {
      "epoch": 4.702283849918434,
      "grad_norm": 0.09626034647226334,
      "learning_rate": 3.187314217529838e-05,
      "loss": 0.0857,
      "num_input_tokens_seen": 62222992,
      "step": 28825
    },
    {
      "epoch": 4.703099510603589,
      "grad_norm": 8.10428237915039,
      "learning_rate": 3.1866298202577157e-05,
      "loss": 0.1228,
      "num_input_tokens_seen": 62234192,
      "step": 28830
    },
    {
      "epoch": 4.7039151712887435,
      "grad_norm": 0.03548339381814003,
      "learning_rate": 3.185945367323895e-05,
      "loss": 0.2025,
      "num_input_tokens_seen": 62244848,
      "step": 28835
    },
    {
      "epoch": 4.704730831973899,
      "grad_norm": 0.6064218878746033,
      "learning_rate": 3.185260858783864e-05,
      "loss": 0.1151,
      "num_input_tokens_seen": 62255376,
      "step": 28840
    },
    {
      "epoch": 4.705546492659054,
      "grad_norm": 4.100457668304443,
      "learning_rate": 3.1845762946931093e-05,
      "loss": 0.0526,
      "num_input_tokens_seen": 62266192,
      "step": 28845
    },
    {
      "epoch": 4.706362153344209,
      "grad_norm": 0.10805456340312958,
      "learning_rate": 3.183891675107128e-05,
      "loss": 0.0158,
      "num_input_tokens_seen": 62277232,
      "step": 28850
    },
    {
      "epoch": 4.707177814029364,
      "grad_norm": 0.08018821477890015,
      "learning_rate": 3.183207000081416e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 62288368,
      "step": 28855
    },
    {
      "epoch": 4.7079934747145185,
      "grad_norm": 1.5426335334777832,
      "learning_rate": 3.1825222696714796e-05,
      "loss": 0.1832,
      "num_input_tokens_seen": 62299024,
      "step": 28860
    },
    {
      "epoch": 4.708809135399674,
      "grad_norm": 4.21427583694458,
      "learning_rate": 3.1818374839328236e-05,
      "loss": 0.2846,
      "num_input_tokens_seen": 62309136,
      "step": 28865
    },
    {
      "epoch": 4.709624796084829,
      "grad_norm": 3.1382803916931152,
      "learning_rate": 3.181152642920962e-05,
      "loss": 0.1796,
      "num_input_tokens_seen": 62320656,
      "step": 28870
    },
    {
      "epoch": 4.710440456769984,
      "grad_norm": 0.0743112787604332,
      "learning_rate": 3.180467746691411e-05,
      "loss": 0.0119,
      "num_input_tokens_seen": 62330896,
      "step": 28875
    },
    {
      "epoch": 4.711256117455139,
      "grad_norm": 0.10520720481872559,
      "learning_rate": 3.1797827952996914e-05,
      "loss": 0.0609,
      "num_input_tokens_seen": 62342032,
      "step": 28880
    },
    {
      "epoch": 4.712071778140293,
      "grad_norm": 0.07115879654884338,
      "learning_rate": 3.1790977888013294e-05,
      "loss": 0.1152,
      "num_input_tokens_seen": 62351984,
      "step": 28885
    },
    {
      "epoch": 4.712887438825448,
      "grad_norm": 5.284401893615723,
      "learning_rate": 3.178412727251856e-05,
      "loss": 0.0579,
      "num_input_tokens_seen": 62362256,
      "step": 28890
    },
    {
      "epoch": 4.713703099510604,
      "grad_norm": 2.9111032485961914,
      "learning_rate": 3.177727610706804e-05,
      "loss": 0.1628,
      "num_input_tokens_seen": 62373040,
      "step": 28895
    },
    {
      "epoch": 4.714518760195759,
      "grad_norm": 0.05595320835709572,
      "learning_rate": 3.177042439221713e-05,
      "loss": 0.0899,
      "num_input_tokens_seen": 62384304,
      "step": 28900
    },
    {
      "epoch": 4.715334420880914,
      "grad_norm": 3.172632932662964,
      "learning_rate": 3.176357212852127e-05,
      "loss": 0.0814,
      "num_input_tokens_seen": 62394864,
      "step": 28905
    },
    {
      "epoch": 4.716150081566068,
      "grad_norm": 0.17778953909873962,
      "learning_rate": 3.175671931653593e-05,
      "loss": 0.0177,
      "num_input_tokens_seen": 62405040,
      "step": 28910
    },
    {
      "epoch": 4.716965742251223,
      "grad_norm": 0.05175602808594704,
      "learning_rate": 3.174986595681664e-05,
      "loss": 0.0738,
      "num_input_tokens_seen": 62417040,
      "step": 28915
    },
    {
      "epoch": 4.717781402936378,
      "grad_norm": 4.293681621551514,
      "learning_rate": 3.174301204991896e-05,
      "loss": 0.2695,
      "num_input_tokens_seen": 62427088,
      "step": 28920
    },
    {
      "epoch": 4.718597063621534,
      "grad_norm": 0.14565585553646088,
      "learning_rate": 3.173615759639852e-05,
      "loss": 0.1103,
      "num_input_tokens_seen": 62436816,
      "step": 28925
    },
    {
      "epoch": 4.719412724306689,
      "grad_norm": 0.08410273492336273,
      "learning_rate": 3.1729302596810965e-05,
      "loss": 0.0122,
      "num_input_tokens_seen": 62447248,
      "step": 28930
    },
    {
      "epoch": 4.720228384991843,
      "grad_norm": 0.11749638617038727,
      "learning_rate": 3.172244705171199e-05,
      "loss": 0.1997,
      "num_input_tokens_seen": 62456368,
      "step": 28935
    },
    {
      "epoch": 4.721044045676998,
      "grad_norm": 0.820249080657959,
      "learning_rate": 3.171559096165736e-05,
      "loss": 0.0587,
      "num_input_tokens_seen": 62467088,
      "step": 28940
    },
    {
      "epoch": 4.721859706362153,
      "grad_norm": 3.017606258392334,
      "learning_rate": 3.170873432720285e-05,
      "loss": 0.3186,
      "num_input_tokens_seen": 62476752,
      "step": 28945
    },
    {
      "epoch": 4.722675367047309,
      "grad_norm": 0.0420236699283123,
      "learning_rate": 3.170187714890429e-05,
      "loss": 0.0757,
      "num_input_tokens_seen": 62487152,
      "step": 28950
    },
    {
      "epoch": 4.7234910277324635,
      "grad_norm": 0.128378227353096,
      "learning_rate": 3.1695019427317564e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 62498544,
      "step": 28955
    },
    {
      "epoch": 4.724306688417618,
      "grad_norm": 0.039971981197595596,
      "learning_rate": 3.1688161162998595e-05,
      "loss": 0.0351,
      "num_input_tokens_seen": 62509488,
      "step": 28960
    },
    {
      "epoch": 4.725122349102773,
      "grad_norm": 0.21320219337940216,
      "learning_rate": 3.1681302356503337e-05,
      "loss": 0.074,
      "num_input_tokens_seen": 62520464,
      "step": 28965
    },
    {
      "epoch": 4.725938009787928,
      "grad_norm": 0.39001935720443726,
      "learning_rate": 3.167444300838782e-05,
      "loss": 0.3243,
      "num_input_tokens_seen": 62531536,
      "step": 28970
    },
    {
      "epoch": 4.726753670473083,
      "grad_norm": 2.2165637016296387,
      "learning_rate": 3.1667583119208085e-05,
      "loss": 0.0718,
      "num_input_tokens_seen": 62542320,
      "step": 28975
    },
    {
      "epoch": 4.7275693311582385,
      "grad_norm": 0.07460815459489822,
      "learning_rate": 3.1660722689520225e-05,
      "loss": 0.0782,
      "num_input_tokens_seen": 62551760,
      "step": 28980
    },
    {
      "epoch": 4.728384991843393,
      "grad_norm": 0.41532909870147705,
      "learning_rate": 3.16538617198804e-05,
      "loss": 0.014,
      "num_input_tokens_seen": 62562512,
      "step": 28985
    },
    {
      "epoch": 4.729200652528548,
      "grad_norm": 0.062238991260528564,
      "learning_rate": 3.164700021084478e-05,
      "loss": 0.15,
      "num_input_tokens_seen": 62573648,
      "step": 28990
    },
    {
      "epoch": 4.730016313213703,
      "grad_norm": 0.23913052678108215,
      "learning_rate": 3.164013816296959e-05,
      "loss": 0.179,
      "num_input_tokens_seen": 62584592,
      "step": 28995
    },
    {
      "epoch": 4.730831973898858,
      "grad_norm": 0.0670330822467804,
      "learning_rate": 3.163327557681111e-05,
      "loss": 0.0145,
      "num_input_tokens_seen": 62596496,
      "step": 29000
    },
    {
      "epoch": 4.731647634584013,
      "grad_norm": 0.11926860362291336,
      "learning_rate": 3.162641245292566e-05,
      "loss": 0.0289,
      "num_input_tokens_seen": 62607312,
      "step": 29005
    },
    {
      "epoch": 4.732463295269168,
      "grad_norm": 2.4738662242889404,
      "learning_rate": 3.161954879186959e-05,
      "loss": 0.0152,
      "num_input_tokens_seen": 62617680,
      "step": 29010
    },
    {
      "epoch": 4.733278955954323,
      "grad_norm": 5.717556953430176,
      "learning_rate": 3.161268459419931e-05,
      "loss": 0.0376,
      "num_input_tokens_seen": 62627664,
      "step": 29015
    },
    {
      "epoch": 4.734094616639478,
      "grad_norm": 5.469749927520752,
      "learning_rate": 3.160581986047127e-05,
      "loss": 0.2754,
      "num_input_tokens_seen": 62638352,
      "step": 29020
    },
    {
      "epoch": 4.734910277324633,
      "grad_norm": 0.16168759763240814,
      "learning_rate": 3.1598954591241934e-05,
      "loss": 0.1976,
      "num_input_tokens_seen": 62649520,
      "step": 29025
    },
    {
      "epoch": 4.735725938009788,
      "grad_norm": 0.15394924581050873,
      "learning_rate": 3.159208878706787e-05,
      "loss": 0.0558,
      "num_input_tokens_seen": 62660080,
      "step": 29030
    },
    {
      "epoch": 4.736541598694943,
      "grad_norm": 0.07229402661323547,
      "learning_rate": 3.1585222448505644e-05,
      "loss": 0.3034,
      "num_input_tokens_seen": 62669808,
      "step": 29035
    },
    {
      "epoch": 4.737357259380098,
      "grad_norm": 0.34821203351020813,
      "learning_rate": 3.1578355576111864e-05,
      "loss": 0.0129,
      "num_input_tokens_seen": 62681136,
      "step": 29040
    },
    {
      "epoch": 4.738172920065253,
      "grad_norm": 0.16153861582279205,
      "learning_rate": 3.157148817044321e-05,
      "loss": 0.2218,
      "num_input_tokens_seen": 62692144,
      "step": 29045
    },
    {
      "epoch": 4.738988580750408,
      "grad_norm": 0.10921628028154373,
      "learning_rate": 3.156462023205638e-05,
      "loss": 0.1657,
      "num_input_tokens_seen": 62701360,
      "step": 29050
    },
    {
      "epoch": 4.739804241435563,
      "grad_norm": 4.022568225860596,
      "learning_rate": 3.155775176150812e-05,
      "loss": 0.141,
      "num_input_tokens_seen": 62713488,
      "step": 29055
    },
    {
      "epoch": 4.740619902120718,
      "grad_norm": 0.04539196193218231,
      "learning_rate": 3.1550882759355246e-05,
      "loss": 0.1429,
      "num_input_tokens_seen": 62725264,
      "step": 29060
    },
    {
      "epoch": 4.741435562805873,
      "grad_norm": 0.3306664824485779,
      "learning_rate": 3.154401322615456e-05,
      "loss": 0.0185,
      "num_input_tokens_seen": 62735312,
      "step": 29065
    },
    {
      "epoch": 4.742251223491028,
      "grad_norm": 0.12463626265525818,
      "learning_rate": 3.153714316246297e-05,
      "loss": 0.1587,
      "num_input_tokens_seen": 62745744,
      "step": 29070
    },
    {
      "epoch": 4.743066884176183,
      "grad_norm": 5.152036190032959,
      "learning_rate": 3.153027256883737e-05,
      "loss": 0.1146,
      "num_input_tokens_seen": 62754960,
      "step": 29075
    },
    {
      "epoch": 4.7438825448613375,
      "grad_norm": 0.3472740948200226,
      "learning_rate": 3.152340144583475e-05,
      "loss": 0.1383,
      "num_input_tokens_seen": 62765712,
      "step": 29080
    },
    {
      "epoch": 4.744698205546492,
      "grad_norm": 0.1714463233947754,
      "learning_rate": 3.151652979401211e-05,
      "loss": 0.0352,
      "num_input_tokens_seen": 62776784,
      "step": 29085
    },
    {
      "epoch": 4.745513866231647,
      "grad_norm": 0.12423329800367355,
      "learning_rate": 3.15096576139265e-05,
      "loss": 0.2172,
      "num_input_tokens_seen": 62787984,
      "step": 29090
    },
    {
      "epoch": 4.746329526916803,
      "grad_norm": 0.10747270286083221,
      "learning_rate": 3.150278490613501e-05,
      "loss": 0.0812,
      "num_input_tokens_seen": 62797680,
      "step": 29095
    },
    {
      "epoch": 4.747145187601958,
      "grad_norm": 0.11853362619876862,
      "learning_rate": 3.149591167119479e-05,
      "loss": 0.0124,
      "num_input_tokens_seen": 62808112,
      "step": 29100
    },
    {
      "epoch": 4.7479608482871125,
      "grad_norm": 0.5263053178787231,
      "learning_rate": 3.148903790966301e-05,
      "loss": 0.2872,
      "num_input_tokens_seen": 62818928,
      "step": 29105
    },
    {
      "epoch": 4.748776508972267,
      "grad_norm": 0.15484373271465302,
      "learning_rate": 3.148216362209688e-05,
      "loss": 0.1805,
      "num_input_tokens_seen": 62829872,
      "step": 29110
    },
    {
      "epoch": 4.749592169657422,
      "grad_norm": 0.1782502830028534,
      "learning_rate": 3.1475288809053684e-05,
      "loss": 0.0874,
      "num_input_tokens_seen": 62840144,
      "step": 29115
    },
    {
      "epoch": 4.750407830342578,
      "grad_norm": 0.16154274344444275,
      "learning_rate": 3.146841347109072e-05,
      "loss": 0.1157,
      "num_input_tokens_seen": 62851856,
      "step": 29120
    },
    {
      "epoch": 4.751223491027733,
      "grad_norm": 0.45126184821128845,
      "learning_rate": 3.146153760876534e-05,
      "loss": 0.1545,
      "num_input_tokens_seen": 62863056,
      "step": 29125
    },
    {
      "epoch": 4.7520391517128875,
      "grad_norm": 7.471928119659424,
      "learning_rate": 3.145466122263494e-05,
      "loss": 0.149,
      "num_input_tokens_seen": 62872592,
      "step": 29130
    },
    {
      "epoch": 4.752854812398042,
      "grad_norm": 3.426064968109131,
      "learning_rate": 3.144778431325694e-05,
      "loss": 0.199,
      "num_input_tokens_seen": 62883600,
      "step": 29135
    },
    {
      "epoch": 4.753670473083197,
      "grad_norm": 11.00782585144043,
      "learning_rate": 3.1440906881188835e-05,
      "loss": 0.1378,
      "num_input_tokens_seen": 62894160,
      "step": 29140
    },
    {
      "epoch": 4.754486133768353,
      "grad_norm": 1.6691228151321411,
      "learning_rate": 3.143402892698814e-05,
      "loss": 0.0193,
      "num_input_tokens_seen": 62905648,
      "step": 29145
    },
    {
      "epoch": 4.755301794453508,
      "grad_norm": 0.09684643894433975,
      "learning_rate": 3.142715045121241e-05,
      "loss": 0.0216,
      "num_input_tokens_seen": 62918256,
      "step": 29150
    },
    {
      "epoch": 4.7561174551386625,
      "grad_norm": 6.829464912414551,
      "learning_rate": 3.142027145441926e-05,
      "loss": 0.2386,
      "num_input_tokens_seen": 62929200,
      "step": 29155
    },
    {
      "epoch": 4.756933115823817,
      "grad_norm": 0.9654311537742615,
      "learning_rate": 3.141339193716633e-05,
      "loss": 0.0162,
      "num_input_tokens_seen": 62940016,
      "step": 29160
    },
    {
      "epoch": 4.757748776508972,
      "grad_norm": 0.12815995514392853,
      "learning_rate": 3.1406511900011295e-05,
      "loss": 0.0096,
      "num_input_tokens_seen": 62949744,
      "step": 29165
    },
    {
      "epoch": 4.758564437194127,
      "grad_norm": 0.07554585486650467,
      "learning_rate": 3.139963134351191e-05,
      "loss": 0.0767,
      "num_input_tokens_seen": 62960848,
      "step": 29170
    },
    {
      "epoch": 4.759380097879282,
      "grad_norm": 0.09591441601514816,
      "learning_rate": 3.139275026822594e-05,
      "loss": 0.3065,
      "num_input_tokens_seen": 62971984,
      "step": 29175
    },
    {
      "epoch": 4.760195758564437,
      "grad_norm": 0.20106206834316254,
      "learning_rate": 3.138586867471118e-05,
      "loss": 0.0198,
      "num_input_tokens_seen": 62982224,
      "step": 29180
    },
    {
      "epoch": 4.761011419249592,
      "grad_norm": 3.4516098499298096,
      "learning_rate": 3.137898656352551e-05,
      "loss": 0.1415,
      "num_input_tokens_seen": 62992464,
      "step": 29185
    },
    {
      "epoch": 4.761827079934747,
      "grad_norm": 0.18744824826717377,
      "learning_rate": 3.137210393522683e-05,
      "loss": 0.0219,
      "num_input_tokens_seen": 63002960,
      "step": 29190
    },
    {
      "epoch": 4.762642740619902,
      "grad_norm": 0.335336297750473,
      "learning_rate": 3.136522079037307e-05,
      "loss": 0.1888,
      "num_input_tokens_seen": 63013520,
      "step": 29195
    },
    {
      "epoch": 4.763458401305057,
      "grad_norm": 2.757715940475464,
      "learning_rate": 3.135833712952222e-05,
      "loss": 0.1149,
      "num_input_tokens_seen": 63023760,
      "step": 29200
    },
    {
      "epoch": 4.764274061990212,
      "grad_norm": 0.05723390355706215,
      "learning_rate": 3.135145295323229e-05,
      "loss": 0.0981,
      "num_input_tokens_seen": 63035248,
      "step": 29205
    },
    {
      "epoch": 4.765089722675367,
      "grad_norm": 0.28717201948165894,
      "learning_rate": 3.1344568262061366e-05,
      "loss": 0.0862,
      "num_input_tokens_seen": 63045904,
      "step": 29210
    },
    {
      "epoch": 4.765905383360522,
      "grad_norm": 4.344303607940674,
      "learning_rate": 3.133768305656755e-05,
      "loss": 0.1569,
      "num_input_tokens_seen": 63056336,
      "step": 29215
    },
    {
      "epoch": 4.766721044045677,
      "grad_norm": 9.279243469238281,
      "learning_rate": 3.1330797337308984e-05,
      "loss": 0.2002,
      "num_input_tokens_seen": 63067376,
      "step": 29220
    },
    {
      "epoch": 4.767536704730832,
      "grad_norm": 0.12709756195545197,
      "learning_rate": 3.1323911104843865e-05,
      "loss": 0.1951,
      "num_input_tokens_seen": 63077456,
      "step": 29225
    },
    {
      "epoch": 4.768352365415987,
      "grad_norm": 0.16337549686431885,
      "learning_rate": 3.131702435973042e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 63087024,
      "step": 29230
    },
    {
      "epoch": 4.769168026101142,
      "grad_norm": 0.1980210691690445,
      "learning_rate": 3.1310137102526926e-05,
      "loss": 0.0942,
      "num_input_tokens_seen": 63098608,
      "step": 29235
    },
    {
      "epoch": 4.769983686786297,
      "grad_norm": 0.052752479910850525,
      "learning_rate": 3.13032493337917e-05,
      "loss": 0.0842,
      "num_input_tokens_seen": 63108144,
      "step": 29240
    },
    {
      "epoch": 4.770799347471452,
      "grad_norm": 2.5380256175994873,
      "learning_rate": 3.129636105408311e-05,
      "loss": 0.2796,
      "num_input_tokens_seen": 63118736,
      "step": 29245
    },
    {
      "epoch": 4.771615008156607,
      "grad_norm": 0.22159433364868164,
      "learning_rate": 3.128947226395954e-05,
      "loss": 0.0082,
      "num_input_tokens_seen": 63128656,
      "step": 29250
    },
    {
      "epoch": 4.7724306688417615,
      "grad_norm": 0.07380012422800064,
      "learning_rate": 3.1282582963979434e-05,
      "loss": 0.1333,
      "num_input_tokens_seen": 63139632,
      "step": 29255
    },
    {
      "epoch": 4.773246329526917,
      "grad_norm": 0.10260308533906937,
      "learning_rate": 3.127569315470128e-05,
      "loss": 0.0118,
      "num_input_tokens_seen": 63150128,
      "step": 29260
    },
    {
      "epoch": 4.774061990212072,
      "grad_norm": 0.5949441194534302,
      "learning_rate": 3.12688028366836e-05,
      "loss": 0.0929,
      "num_input_tokens_seen": 63160112,
      "step": 29265
    },
    {
      "epoch": 4.774877650897227,
      "grad_norm": 0.2540006637573242,
      "learning_rate": 3.126191201048494e-05,
      "loss": 0.119,
      "num_input_tokens_seen": 63172592,
      "step": 29270
    },
    {
      "epoch": 4.775693311582382,
      "grad_norm": 0.8490146994590759,
      "learning_rate": 3.125502067666393e-05,
      "loss": 0.1109,
      "num_input_tokens_seen": 63184304,
      "step": 29275
    },
    {
      "epoch": 4.7765089722675365,
      "grad_norm": 0.06298528611660004,
      "learning_rate": 3.1248128835779206e-05,
      "loss": 0.0083,
      "num_input_tokens_seen": 63195184,
      "step": 29280
    },
    {
      "epoch": 4.777324632952691,
      "grad_norm": 0.07878922671079636,
      "learning_rate": 3.124123648838946e-05,
      "loss": 0.0465,
      "num_input_tokens_seen": 63206064,
      "step": 29285
    },
    {
      "epoch": 4.778140293637847,
      "grad_norm": 0.09068889915943146,
      "learning_rate": 3.123434363505341e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 63215536,
      "step": 29290
    },
    {
      "epoch": 4.778955954323002,
      "grad_norm": 11.19113540649414,
      "learning_rate": 3.122745027632983e-05,
      "loss": 0.1293,
      "num_input_tokens_seen": 63226480,
      "step": 29295
    },
    {
      "epoch": 4.779771615008157,
      "grad_norm": 1.2498172521591187,
      "learning_rate": 3.1220556412777536e-05,
      "loss": 0.083,
      "num_input_tokens_seen": 63237200,
      "step": 29300
    },
    {
      "epoch": 4.780587275693311,
      "grad_norm": 0.10224765539169312,
      "learning_rate": 3.121366204495538e-05,
      "loss": 0.2689,
      "num_input_tokens_seen": 63247632,
      "step": 29305
    },
    {
      "epoch": 4.781402936378466,
      "grad_norm": 0.10505528002977371,
      "learning_rate": 3.120676717342225e-05,
      "loss": 0.1855,
      "num_input_tokens_seen": 63258864,
      "step": 29310
    },
    {
      "epoch": 4.782218597063622,
      "grad_norm": 1.2177871465682983,
      "learning_rate": 3.119987179873707e-05,
      "loss": 0.0732,
      "num_input_tokens_seen": 63270928,
      "step": 29315
    },
    {
      "epoch": 4.783034257748777,
      "grad_norm": 3.561885356903076,
      "learning_rate": 3.119297592145884e-05,
      "loss": 0.2115,
      "num_input_tokens_seen": 63281200,
      "step": 29320
    },
    {
      "epoch": 4.783849918433932,
      "grad_norm": 0.6261371970176697,
      "learning_rate": 3.1186079542146554e-05,
      "loss": 0.0488,
      "num_input_tokens_seen": 63291184,
      "step": 29325
    },
    {
      "epoch": 4.784665579119086,
      "grad_norm": 23.49911117553711,
      "learning_rate": 3.117918266135927e-05,
      "loss": 0.1411,
      "num_input_tokens_seen": 63302544,
      "step": 29330
    },
    {
      "epoch": 4.785481239804241,
      "grad_norm": 0.10632102936506271,
      "learning_rate": 3.1172285279656085e-05,
      "loss": 0.0995,
      "num_input_tokens_seen": 63312464,
      "step": 29335
    },
    {
      "epoch": 4.786296900489396,
      "grad_norm": 0.11808846890926361,
      "learning_rate": 3.1165387397596136e-05,
      "loss": 0.3068,
      "num_input_tokens_seen": 63323856,
      "step": 29340
    },
    {
      "epoch": 4.787112561174552,
      "grad_norm": 0.0858970656991005,
      "learning_rate": 3.1158489015738604e-05,
      "loss": 0.0119,
      "num_input_tokens_seen": 63335664,
      "step": 29345
    },
    {
      "epoch": 4.787928221859707,
      "grad_norm": 0.13045793771743774,
      "learning_rate": 3.1151590134642705e-05,
      "loss": 0.0056,
      "num_input_tokens_seen": 63347632,
      "step": 29350
    },
    {
      "epoch": 4.788743882544861,
      "grad_norm": 6.656271457672119,
      "learning_rate": 3.114469075486769e-05,
      "loss": 0.1221,
      "num_input_tokens_seen": 63359152,
      "step": 29355
    },
    {
      "epoch": 4.789559543230016,
      "grad_norm": 0.08970692753791809,
      "learning_rate": 3.113779087697287e-05,
      "loss": 0.013,
      "num_input_tokens_seen": 63370224,
      "step": 29360
    },
    {
      "epoch": 4.790375203915171,
      "grad_norm": 5.6270751953125,
      "learning_rate": 3.1130890501517586e-05,
      "loss": 0.0565,
      "num_input_tokens_seen": 63380784,
      "step": 29365
    },
    {
      "epoch": 4.791190864600326,
      "grad_norm": 0.20602378249168396,
      "learning_rate": 3.11239896290612e-05,
      "loss": 0.2154,
      "num_input_tokens_seen": 63392656,
      "step": 29370
    },
    {
      "epoch": 4.7920065252854815,
      "grad_norm": 0.1836508810520172,
      "learning_rate": 3.111708826016315e-05,
      "loss": 0.0466,
      "num_input_tokens_seen": 63403440,
      "step": 29375
    },
    {
      "epoch": 4.792822185970636,
      "grad_norm": 3.018244504928589,
      "learning_rate": 3.1110186395382885e-05,
      "loss": 0.1263,
      "num_input_tokens_seen": 63413968,
      "step": 29380
    },
    {
      "epoch": 4.793637846655791,
      "grad_norm": 0.5818954706192017,
      "learning_rate": 3.1103284035279905e-05,
      "loss": 0.1695,
      "num_input_tokens_seen": 63424752,
      "step": 29385
    },
    {
      "epoch": 4.794453507340946,
      "grad_norm": 0.061303652822971344,
      "learning_rate": 3.109638118041376e-05,
      "loss": 0.0091,
      "num_input_tokens_seen": 63435792,
      "step": 29390
    },
    {
      "epoch": 4.795269168026101,
      "grad_norm": 0.49282169342041016,
      "learning_rate": 3.108947783134402e-05,
      "loss": 0.0105,
      "num_input_tokens_seen": 63448016,
      "step": 29395
    },
    {
      "epoch": 4.7960848287112565,
      "grad_norm": 0.08261608332395554,
      "learning_rate": 3.10825739886303e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 63456880,
      "step": 29400
    },
    {
      "epoch": 4.796900489396411,
      "grad_norm": 3.3087964057922363,
      "learning_rate": 3.107566965283228e-05,
      "loss": 0.1434,
      "num_input_tokens_seen": 63467504,
      "step": 29405
    },
    {
      "epoch": 4.797716150081566,
      "grad_norm": 2.5278067588806152,
      "learning_rate": 3.106876482450964e-05,
      "loss": 0.1676,
      "num_input_tokens_seen": 63478288,
      "step": 29410
    },
    {
      "epoch": 4.798531810766721,
      "grad_norm": 0.04726960510015488,
      "learning_rate": 3.106185950422215e-05,
      "loss": 0.0193,
      "num_input_tokens_seen": 63487888,
      "step": 29415
    },
    {
      "epoch": 4.799347471451876,
      "grad_norm": 0.040033210068941116,
      "learning_rate": 3.105495369252956e-05,
      "loss": 0.0194,
      "num_input_tokens_seen": 63498576,
      "step": 29420
    },
    {
      "epoch": 4.800163132137031,
      "grad_norm": 0.04523498937487602,
      "learning_rate": 3.104804738999169e-05,
      "loss": 0.1039,
      "num_input_tokens_seen": 63509872,
      "step": 29425
    },
    {
      "epoch": 4.800978792822186,
      "grad_norm": 0.1345680207014084,
      "learning_rate": 3.1041140597168425e-05,
      "loss": 0.0041,
      "num_input_tokens_seen": 63521328,
      "step": 29430
    },
    {
      "epoch": 4.801794453507341,
      "grad_norm": 1.8691577911376953,
      "learning_rate": 3.1034233314619647e-05,
      "loss": 0.1054,
      "num_input_tokens_seen": 63531248,
      "step": 29435
    },
    {
      "epoch": 4.802610114192496,
      "grad_norm": 5.198378562927246,
      "learning_rate": 3.102732554290531e-05,
      "loss": 0.2767,
      "num_input_tokens_seen": 63541648,
      "step": 29440
    },
    {
      "epoch": 4.803425774877651,
      "grad_norm": 0.24566799402236938,
      "learning_rate": 3.102041728258537e-05,
      "loss": 0.0073,
      "num_input_tokens_seen": 63551408,
      "step": 29445
    },
    {
      "epoch": 4.804241435562806,
      "grad_norm": 0.1768108606338501,
      "learning_rate": 3.101350853421986e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 63562160,
      "step": 29450
    },
    {
      "epoch": 4.80505709624796,
      "grad_norm": 0.045698415488004684,
      "learning_rate": 3.1006599298368826e-05,
      "loss": 0.0792,
      "num_input_tokens_seen": 63573648,
      "step": 29455
    },
    {
      "epoch": 4.805872756933116,
      "grad_norm": 0.0276840440928936,
      "learning_rate": 3.099968957559239e-05,
      "loss": 0.1144,
      "num_input_tokens_seen": 63584048,
      "step": 29460
    },
    {
      "epoch": 4.806688417618271,
      "grad_norm": 0.159894198179245,
      "learning_rate": 3.0992779366450666e-05,
      "loss": 0.0532,
      "num_input_tokens_seen": 63595344,
      "step": 29465
    },
    {
      "epoch": 4.807504078303426,
      "grad_norm": 3.1524837017059326,
      "learning_rate": 3.098586867150385e-05,
      "loss": 0.1979,
      "num_input_tokens_seen": 63606800,
      "step": 29470
    },
    {
      "epoch": 4.808319738988581,
      "grad_norm": 4.080893039703369,
      "learning_rate": 3.097895749131214e-05,
      "loss": 0.1287,
      "num_input_tokens_seen": 63617456,
      "step": 29475
    },
    {
      "epoch": 4.809135399673735,
      "grad_norm": 0.25066328048706055,
      "learning_rate": 3.09720458264358e-05,
      "loss": 0.397,
      "num_input_tokens_seen": 63628336,
      "step": 29480
    },
    {
      "epoch": 4.809951060358891,
      "grad_norm": 0.0336231030523777,
      "learning_rate": 3.096513367743513e-05,
      "loss": 0.0741,
      "num_input_tokens_seen": 63640720,
      "step": 29485
    },
    {
      "epoch": 4.810766721044046,
      "grad_norm": 0.30254265666007996,
      "learning_rate": 3.095822104487045e-05,
      "loss": 0.0118,
      "num_input_tokens_seen": 63650960,
      "step": 29490
    },
    {
      "epoch": 4.811582381729201,
      "grad_norm": 3.7460927963256836,
      "learning_rate": 3.0951307929302136e-05,
      "loss": 0.1093,
      "num_input_tokens_seen": 63662192,
      "step": 29495
    },
    {
      "epoch": 4.8123980424143555,
      "grad_norm": 5.232872486114502,
      "learning_rate": 3.094439433129061e-05,
      "loss": 0.0899,
      "num_input_tokens_seen": 63673232,
      "step": 29500
    },
    {
      "epoch": 4.81321370309951,
      "grad_norm": 0.10796443372964859,
      "learning_rate": 3.093748025139632e-05,
      "loss": 0.0054,
      "num_input_tokens_seen": 63683728,
      "step": 29505
    },
    {
      "epoch": 4.814029363784666,
      "grad_norm": 4.561418533325195,
      "learning_rate": 3.093056569017975e-05,
      "loss": 0.0574,
      "num_input_tokens_seen": 63694768,
      "step": 29510
    },
    {
      "epoch": 4.814845024469821,
      "grad_norm": 0.11669047176837921,
      "learning_rate": 3.0923650648201436e-05,
      "loss": 0.2051,
      "num_input_tokens_seen": 63705488,
      "step": 29515
    },
    {
      "epoch": 4.815660685154976,
      "grad_norm": 0.6478515267372131,
      "learning_rate": 3.0916735126021945e-05,
      "loss": 0.2099,
      "num_input_tokens_seen": 63715600,
      "step": 29520
    },
    {
      "epoch": 4.8164763458401305,
      "grad_norm": 4.888887405395508,
      "learning_rate": 3.090981912420188e-05,
      "loss": 0.2501,
      "num_input_tokens_seen": 63726000,
      "step": 29525
    },
    {
      "epoch": 4.817292006525285,
      "grad_norm": 4.986412048339844,
      "learning_rate": 3.09029026433019e-05,
      "loss": 0.1143,
      "num_input_tokens_seen": 63736400,
      "step": 29530
    },
    {
      "epoch": 4.81810766721044,
      "grad_norm": 4.063265800476074,
      "learning_rate": 3.0895985683882675e-05,
      "loss": 0.4035,
      "num_input_tokens_seen": 63747600,
      "step": 29535
    },
    {
      "epoch": 4.818923327895595,
      "grad_norm": 0.17317841947078705,
      "learning_rate": 3.088906824650493e-05,
      "loss": 0.011,
      "num_input_tokens_seen": 63758416,
      "step": 29540
    },
    {
      "epoch": 4.819738988580751,
      "grad_norm": 0.1033623069524765,
      "learning_rate": 3.088215033172944e-05,
      "loss": 0.1442,
      "num_input_tokens_seen": 63768976,
      "step": 29545
    },
    {
      "epoch": 4.8205546492659055,
      "grad_norm": 0.15044410526752472,
      "learning_rate": 3.087523194011699e-05,
      "loss": 0.0491,
      "num_input_tokens_seen": 63778896,
      "step": 29550
    },
    {
      "epoch": 4.82137030995106,
      "grad_norm": 4.59242057800293,
      "learning_rate": 3.086831307222844e-05,
      "loss": 0.0512,
      "num_input_tokens_seen": 63790384,
      "step": 29555
    },
    {
      "epoch": 4.822185970636215,
      "grad_norm": 0.14458999037742615,
      "learning_rate": 3.086139372862464e-05,
      "loss": 0.0138,
      "num_input_tokens_seen": 63800688,
      "step": 29560
    },
    {
      "epoch": 4.82300163132137,
      "grad_norm": 0.08712341636419296,
      "learning_rate": 3.085447390986653e-05,
      "loss": 0.0484,
      "num_input_tokens_seen": 63811856,
      "step": 29565
    },
    {
      "epoch": 4.823817292006526,
      "grad_norm": 3.085439682006836,
      "learning_rate": 3.084755361651507e-05,
      "loss": 0.1413,
      "num_input_tokens_seen": 63822576,
      "step": 29570
    },
    {
      "epoch": 4.8246329526916805,
      "grad_norm": 0.8713060617446899,
      "learning_rate": 3.0840632849131236e-05,
      "loss": 0.0139,
      "num_input_tokens_seen": 63833168,
      "step": 29575
    },
    {
      "epoch": 4.825448613376835,
      "grad_norm": 0.1782616525888443,
      "learning_rate": 3.083371160827606e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 63843312,
      "step": 29580
    },
    {
      "epoch": 4.82626427406199,
      "grad_norm": 1.5676994323730469,
      "learning_rate": 3.082678989451063e-05,
      "loss": 0.3691,
      "num_input_tokens_seen": 63853904,
      "step": 29585
    },
    {
      "epoch": 4.827079934747145,
      "grad_norm": 4.571804046630859,
      "learning_rate": 3.081986770839605e-05,
      "loss": 0.4312,
      "num_input_tokens_seen": 63865072,
      "step": 29590
    },
    {
      "epoch": 4.827895595432301,
      "grad_norm": 0.29922136664390564,
      "learning_rate": 3.0812945050493464e-05,
      "loss": 0.0204,
      "num_input_tokens_seen": 63875760,
      "step": 29595
    },
    {
      "epoch": 4.828711256117455,
      "grad_norm": 0.1453588306903839,
      "learning_rate": 3.080602192136405e-05,
      "loss": 0.0312,
      "num_input_tokens_seen": 63887216,
      "step": 29600
    },
    {
      "epoch": 4.82952691680261,
      "grad_norm": 0.07486657053232193,
      "learning_rate": 3.079909832156905e-05,
      "loss": 0.0815,
      "num_input_tokens_seen": 63897104,
      "step": 29605
    },
    {
      "epoch": 4.830342577487765,
      "grad_norm": 0.0959646925330162,
      "learning_rate": 3.0792174251669706e-05,
      "loss": 0.1833,
      "num_input_tokens_seen": 63907984,
      "step": 29610
    },
    {
      "epoch": 4.83115823817292,
      "grad_norm": 0.22850702702999115,
      "learning_rate": 3.078524971222733e-05,
      "loss": 0.0527,
      "num_input_tokens_seen": 63918640,
      "step": 29615
    },
    {
      "epoch": 4.831973898858075,
      "grad_norm": 13.872844696044922,
      "learning_rate": 3.0778324703803256e-05,
      "loss": 0.203,
      "num_input_tokens_seen": 63929136,
      "step": 29620
    },
    {
      "epoch": 4.8327895595432295,
      "grad_norm": 0.11571940034627914,
      "learning_rate": 3.0771399226958865e-05,
      "loss": 0.2644,
      "num_input_tokens_seen": 63940464,
      "step": 29625
    },
    {
      "epoch": 4.833605220228385,
      "grad_norm": 0.6639408469200134,
      "learning_rate": 3.076447328225557e-05,
      "loss": 0.1808,
      "num_input_tokens_seen": 63950512,
      "step": 29630
    },
    {
      "epoch": 4.83442088091354,
      "grad_norm": 3.395463466644287,
      "learning_rate": 3.075754687025482e-05,
      "loss": 0.32,
      "num_input_tokens_seen": 63961840,
      "step": 29635
    },
    {
      "epoch": 4.835236541598695,
      "grad_norm": 0.3087729215621948,
      "learning_rate": 3.0750619991518115e-05,
      "loss": 0.032,
      "num_input_tokens_seen": 63971440,
      "step": 29640
    },
    {
      "epoch": 4.83605220228385,
      "grad_norm": 0.18898051977157593,
      "learning_rate": 3.074369264660697e-05,
      "loss": 0.1064,
      "num_input_tokens_seen": 63982608,
      "step": 29645
    },
    {
      "epoch": 4.8368678629690045,
      "grad_norm": 0.1280173659324646,
      "learning_rate": 3.0736764836082954e-05,
      "loss": 0.2161,
      "num_input_tokens_seen": 63994000,
      "step": 29650
    },
    {
      "epoch": 4.83768352365416,
      "grad_norm": 3.3714702129364014,
      "learning_rate": 3.072983656050767e-05,
      "loss": 0.1225,
      "num_input_tokens_seen": 64005008,
      "step": 29655
    },
    {
      "epoch": 4.838499184339315,
      "grad_norm": 0.44245752692222595,
      "learning_rate": 3.072290782044276e-05,
      "loss": 0.0651,
      "num_input_tokens_seen": 64014448,
      "step": 29660
    },
    {
      "epoch": 4.83931484502447,
      "grad_norm": 0.14941665530204773,
      "learning_rate": 3.0715978616449906e-05,
      "loss": 0.0101,
      "num_input_tokens_seen": 64026672,
      "step": 29665
    },
    {
      "epoch": 4.840130505709625,
      "grad_norm": 0.1486716866493225,
      "learning_rate": 3.070904894909083e-05,
      "loss": 0.1468,
      "num_input_tokens_seen": 64036752,
      "step": 29670
    },
    {
      "epoch": 4.8409461663947795,
      "grad_norm": 0.22281208634376526,
      "learning_rate": 3.070211881892727e-05,
      "loss": 0.2368,
      "num_input_tokens_seen": 64047760,
      "step": 29675
    },
    {
      "epoch": 4.841761827079935,
      "grad_norm": 0.10058537125587463,
      "learning_rate": 3.069518822652103e-05,
      "loss": 0.1777,
      "num_input_tokens_seen": 64059664,
      "step": 29680
    },
    {
      "epoch": 4.84257748776509,
      "grad_norm": 1.5558687448501587,
      "learning_rate": 3.0688257172433944e-05,
      "loss": 0.0738,
      "num_input_tokens_seen": 64071408,
      "step": 29685
    },
    {
      "epoch": 4.843393148450245,
      "grad_norm": 0.1454678624868393,
      "learning_rate": 3.068132565722786e-05,
      "loss": 0.098,
      "num_input_tokens_seen": 64082448,
      "step": 29690
    },
    {
      "epoch": 4.8442088091354,
      "grad_norm": 0.22452178597450256,
      "learning_rate": 3.06743936814647e-05,
      "loss": 0.0198,
      "num_input_tokens_seen": 64094320,
      "step": 29695
    },
    {
      "epoch": 4.8450244698205545,
      "grad_norm": 0.21330443024635315,
      "learning_rate": 3.0667461245706386e-05,
      "loss": 0.1539,
      "num_input_tokens_seen": 64105392,
      "step": 29700
    },
    {
      "epoch": 4.845840130505709,
      "grad_norm": 0.13264863193035126,
      "learning_rate": 3.066052835051491e-05,
      "loss": 0.1373,
      "num_input_tokens_seen": 64117008,
      "step": 29705
    },
    {
      "epoch": 4.846655791190865,
      "grad_norm": 0.4915766716003418,
      "learning_rate": 3.065359499645228e-05,
      "loss": 0.4959,
      "num_input_tokens_seen": 64128016,
      "step": 29710
    },
    {
      "epoch": 4.84747145187602,
      "grad_norm": 1.5233148336410522,
      "learning_rate": 3.064666118408057e-05,
      "loss": 0.0606,
      "num_input_tokens_seen": 64138448,
      "step": 29715
    },
    {
      "epoch": 4.848287112561175,
      "grad_norm": 0.1547406017780304,
      "learning_rate": 3.0639726913961833e-05,
      "loss": 0.0099,
      "num_input_tokens_seen": 64148848,
      "step": 29720
    },
    {
      "epoch": 4.849102773246329,
      "grad_norm": 1.3181960582733154,
      "learning_rate": 3.0632792186658225e-05,
      "loss": 0.1275,
      "num_input_tokens_seen": 64159280,
      "step": 29725
    },
    {
      "epoch": 4.849918433931484,
      "grad_norm": 0.19170302152633667,
      "learning_rate": 3.062585700273191e-05,
      "loss": 0.0658,
      "num_input_tokens_seen": 64170384,
      "step": 29730
    },
    {
      "epoch": 4.850734094616639,
      "grad_norm": 0.1350255161523819,
      "learning_rate": 3.0618921362745075e-05,
      "loss": 0.163,
      "num_input_tokens_seen": 64180112,
      "step": 29735
    },
    {
      "epoch": 4.851549755301795,
      "grad_norm": 0.6598547101020813,
      "learning_rate": 3.061198526725996e-05,
      "loss": 0.1019,
      "num_input_tokens_seen": 64190032,
      "step": 29740
    },
    {
      "epoch": 4.85236541598695,
      "grad_norm": 0.937228262424469,
      "learning_rate": 3.060504871683885e-05,
      "loss": 0.1113,
      "num_input_tokens_seen": 64201936,
      "step": 29745
    },
    {
      "epoch": 4.853181076672104,
      "grad_norm": 1.5486509799957275,
      "learning_rate": 3.059811171204404e-05,
      "loss": 0.06,
      "num_input_tokens_seen": 64212784,
      "step": 29750
    },
    {
      "epoch": 4.853996737357259,
      "grad_norm": 0.13088580965995789,
      "learning_rate": 3.0591174253437904e-05,
      "loss": 0.0781,
      "num_input_tokens_seen": 64224208,
      "step": 29755
    },
    {
      "epoch": 4.854812398042414,
      "grad_norm": 0.9528616070747375,
      "learning_rate": 3.05842363415828e-05,
      "loss": 0.0992,
      "num_input_tokens_seen": 64235792,
      "step": 29760
    },
    {
      "epoch": 4.85562805872757,
      "grad_norm": 0.058940738439559937,
      "learning_rate": 3.057729797704118e-05,
      "loss": 0.1113,
      "num_input_tokens_seen": 64246512,
      "step": 29765
    },
    {
      "epoch": 4.856443719412725,
      "grad_norm": 0.12711629271507263,
      "learning_rate": 3.057035916037548e-05,
      "loss": 0.0145,
      "num_input_tokens_seen": 64256624,
      "step": 29770
    },
    {
      "epoch": 4.857259380097879,
      "grad_norm": 0.2545081675052643,
      "learning_rate": 3.05634198921482e-05,
      "loss": 0.149,
      "num_input_tokens_seen": 64266544,
      "step": 29775
    },
    {
      "epoch": 4.858075040783034,
      "grad_norm": 3.3272788524627686,
      "learning_rate": 3.055648017292188e-05,
      "loss": 0.2232,
      "num_input_tokens_seen": 64277904,
      "step": 29780
    },
    {
      "epoch": 4.858890701468189,
      "grad_norm": 0.10898911207914352,
      "learning_rate": 3.0549540003259084e-05,
      "loss": 0.0329,
      "num_input_tokens_seen": 64288464,
      "step": 29785
    },
    {
      "epoch": 4.859706362153344,
      "grad_norm": 0.07357125729322433,
      "learning_rate": 3.054259938372242e-05,
      "loss": 0.229,
      "num_input_tokens_seen": 64299408,
      "step": 29790
    },
    {
      "epoch": 4.8605220228384995,
      "grad_norm": 0.28173163533210754,
      "learning_rate": 3.0535658314874515e-05,
      "loss": 0.1425,
      "num_input_tokens_seen": 64310224,
      "step": 29795
    },
    {
      "epoch": 4.861337683523654,
      "grad_norm": 4.659884452819824,
      "learning_rate": 3.0528716797278064e-05,
      "loss": 0.0321,
      "num_input_tokens_seen": 64321520,
      "step": 29800
    },
    {
      "epoch": 4.862153344208809,
      "grad_norm": 3.7471280097961426,
      "learning_rate": 3.052177483149578e-05,
      "loss": 0.2272,
      "num_input_tokens_seen": 64332880,
      "step": 29805
    },
    {
      "epoch": 4.862969004893964,
      "grad_norm": 0.1539532095193863,
      "learning_rate": 3.0514832418090406e-05,
      "loss": 0.0778,
      "num_input_tokens_seen": 64342576,
      "step": 29810
    },
    {
      "epoch": 4.863784665579119,
      "grad_norm": 0.2837701141834259,
      "learning_rate": 3.050788955762474e-05,
      "loss": 0.0697,
      "num_input_tokens_seen": 64353936,
      "step": 29815
    },
    {
      "epoch": 4.864600326264274,
      "grad_norm": 0.044804736971855164,
      "learning_rate": 3.05009462506616e-05,
      "loss": 0.069,
      "num_input_tokens_seen": 64364592,
      "step": 29820
    },
    {
      "epoch": 4.865415986949429,
      "grad_norm": 0.2907470762729645,
      "learning_rate": 3.049400249776384e-05,
      "loss": 0.0766,
      "num_input_tokens_seen": 64375696,
      "step": 29825
    },
    {
      "epoch": 4.866231647634584,
      "grad_norm": 0.12099996954202652,
      "learning_rate": 3.0487058299494363e-05,
      "loss": 0.0481,
      "num_input_tokens_seen": 64387440,
      "step": 29830
    },
    {
      "epoch": 4.867047308319739,
      "grad_norm": 0.10837902128696442,
      "learning_rate": 3.0480113656416103e-05,
      "loss": 0.029,
      "num_input_tokens_seen": 64399088,
      "step": 29835
    },
    {
      "epoch": 4.867862969004894,
      "grad_norm": 0.2713727355003357,
      "learning_rate": 3.047316856909202e-05,
      "loss": 0.3061,
      "num_input_tokens_seen": 64409712,
      "step": 29840
    },
    {
      "epoch": 4.868678629690049,
      "grad_norm": 0.18403325974941254,
      "learning_rate": 3.0466223038085128e-05,
      "loss": 0.227,
      "num_input_tokens_seen": 64421200,
      "step": 29845
    },
    {
      "epoch": 4.869494290375204,
      "grad_norm": 0.1828850656747818,
      "learning_rate": 3.0459277063958457e-05,
      "loss": 0.2055,
      "num_input_tokens_seen": 64431344,
      "step": 29850
    },
    {
      "epoch": 4.870309951060359,
      "grad_norm": 0.1920708417892456,
      "learning_rate": 3.0452330647275086e-05,
      "loss": 0.0306,
      "num_input_tokens_seen": 64442928,
      "step": 29855
    },
    {
      "epoch": 4.871125611745514,
      "grad_norm": 0.04284502938389778,
      "learning_rate": 3.044538378859813e-05,
      "loss": 0.0869,
      "num_input_tokens_seen": 64453968,
      "step": 29860
    },
    {
      "epoch": 4.871941272430669,
      "grad_norm": 0.06283725053071976,
      "learning_rate": 3.0438436488490736e-05,
      "loss": 0.0837,
      "num_input_tokens_seen": 64465008,
      "step": 29865
    },
    {
      "epoch": 4.872756933115824,
      "grad_norm": 0.14269883930683136,
      "learning_rate": 3.0431488747516085e-05,
      "loss": 0.0905,
      "num_input_tokens_seen": 64474832,
      "step": 29870
    },
    {
      "epoch": 4.873572593800979,
      "grad_norm": 3.301142930984497,
      "learning_rate": 3.0424540566237398e-05,
      "loss": 0.0312,
      "num_input_tokens_seen": 64485776,
      "step": 29875
    },
    {
      "epoch": 4.874388254486134,
      "grad_norm": 0.09250722080469131,
      "learning_rate": 3.041759194521792e-05,
      "loss": 0.1696,
      "num_input_tokens_seen": 64497392,
      "step": 29880
    },
    {
      "epoch": 4.875203915171289,
      "grad_norm": 0.20140580832958221,
      "learning_rate": 3.0410642885020957e-05,
      "loss": 0.0221,
      "num_input_tokens_seen": 64508176,
      "step": 29885
    },
    {
      "epoch": 4.876019575856444,
      "grad_norm": 2.741020679473877,
      "learning_rate": 3.040369338620983e-05,
      "loss": 0.2064,
      "num_input_tokens_seen": 64517808,
      "step": 29890
    },
    {
      "epoch": 4.876835236541599,
      "grad_norm": 9.838257789611816,
      "learning_rate": 3.0396743449347893e-05,
      "loss": 0.2057,
      "num_input_tokens_seen": 64527696,
      "step": 29895
    },
    {
      "epoch": 4.877650897226753,
      "grad_norm": 0.29459795355796814,
      "learning_rate": 3.0389793074998553e-05,
      "loss": 0.0092,
      "num_input_tokens_seen": 64538448,
      "step": 29900
    },
    {
      "epoch": 4.878466557911908,
      "grad_norm": 8.487483978271484,
      "learning_rate": 3.038284226372524e-05,
      "loss": 0.044,
      "num_input_tokens_seen": 64548784,
      "step": 29905
    },
    {
      "epoch": 4.879282218597064,
      "grad_norm": 0.05116702616214752,
      "learning_rate": 3.0375891016091424e-05,
      "loss": 0.0292,
      "num_input_tokens_seen": 64559152,
      "step": 29910
    },
    {
      "epoch": 4.880097879282219,
      "grad_norm": 0.11584285646677017,
      "learning_rate": 3.0368939332660603e-05,
      "loss": 0.0437,
      "num_input_tokens_seen": 64570096,
      "step": 29915
    },
    {
      "epoch": 4.8809135399673735,
      "grad_norm": 0.029633769765496254,
      "learning_rate": 3.036198721399631e-05,
      "loss": 0.0711,
      "num_input_tokens_seen": 64581744,
      "step": 29920
    },
    {
      "epoch": 4.881729200652528,
      "grad_norm": 4.062506198883057,
      "learning_rate": 3.035503466066214e-05,
      "loss": 0.0718,
      "num_input_tokens_seen": 64591856,
      "step": 29925
    },
    {
      "epoch": 4.882544861337683,
      "grad_norm": 0.05810140445828438,
      "learning_rate": 3.0348081673221678e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 64602192,
      "step": 29930
    },
    {
      "epoch": 4.883360522022839,
      "grad_norm": 0.07992871105670929,
      "learning_rate": 3.034112825223858e-05,
      "loss": 0.0151,
      "num_input_tokens_seen": 64612976,
      "step": 29935
    },
    {
      "epoch": 4.884176182707994,
      "grad_norm": 0.1341329663991928,
      "learning_rate": 3.0334174398276532e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 64623984,
      "step": 29940
    },
    {
      "epoch": 4.8849918433931485,
      "grad_norm": 4.25712776184082,
      "learning_rate": 3.032722011189924e-05,
      "loss": 0.0702,
      "num_input_tokens_seen": 64634992,
      "step": 29945
    },
    {
      "epoch": 4.885807504078303,
      "grad_norm": 0.12009090185165405,
      "learning_rate": 3.032026539367046e-05,
      "loss": 0.1729,
      "num_input_tokens_seen": 64645840,
      "step": 29950
    },
    {
      "epoch": 4.886623164763458,
      "grad_norm": 0.11088258028030396,
      "learning_rate": 3.0313310244153968e-05,
      "loss": 0.0295,
      "num_input_tokens_seen": 64656016,
      "step": 29955
    },
    {
      "epoch": 4.887438825448614,
      "grad_norm": 0.11191460490226746,
      "learning_rate": 3.0306354663913588e-05,
      "loss": 0.0801,
      "num_input_tokens_seen": 64667760,
      "step": 29960
    },
    {
      "epoch": 4.888254486133769,
      "grad_norm": 4.154870510101318,
      "learning_rate": 3.029939865351317e-05,
      "loss": 0.12,
      "num_input_tokens_seen": 64677872,
      "step": 29965
    },
    {
      "epoch": 4.8890701468189235,
      "grad_norm": 0.05185890570282936,
      "learning_rate": 3.0292442213516613e-05,
      "loss": 0.0034,
      "num_input_tokens_seen": 64688784,
      "step": 29970
    },
    {
      "epoch": 4.889885807504078,
      "grad_norm": 0.09086186438798904,
      "learning_rate": 3.0285485344487834e-05,
      "loss": 0.2251,
      "num_input_tokens_seen": 64698576,
      "step": 29975
    },
    {
      "epoch": 4.890701468189233,
      "grad_norm": 0.06559891998767853,
      "learning_rate": 3.02785280469908e-05,
      "loss": 0.0041,
      "num_input_tokens_seen": 64709520,
      "step": 29980
    },
    {
      "epoch": 4.891517128874388,
      "grad_norm": 0.1890016347169876,
      "learning_rate": 3.0271570321589494e-05,
      "loss": 0.0125,
      "num_input_tokens_seen": 64720592,
      "step": 29985
    },
    {
      "epoch": 4.892332789559543,
      "grad_norm": 2.3863577842712402,
      "learning_rate": 3.026461216884795e-05,
      "loss": 0.0922,
      "num_input_tokens_seen": 64731568,
      "step": 29990
    },
    {
      "epoch": 4.8931484502446985,
      "grad_norm": 5.497276782989502,
      "learning_rate": 3.025765358933024e-05,
      "loss": 0.2474,
      "num_input_tokens_seen": 64742256,
      "step": 29995
    },
    {
      "epoch": 4.893964110929853,
      "grad_norm": 0.04517051950097084,
      "learning_rate": 3.0250694583600448e-05,
      "loss": 0.0048,
      "num_input_tokens_seen": 64752208,
      "step": 30000
    },
    {
      "epoch": 4.894779771615008,
      "grad_norm": 0.0528087392449379,
      "learning_rate": 3.024373515222271e-05,
      "loss": 0.2048,
      "num_input_tokens_seen": 64761168,
      "step": 30005
    },
    {
      "epoch": 4.895595432300163,
      "grad_norm": 0.08989369869232178,
      "learning_rate": 3.0236775295761194e-05,
      "loss": 0.0963,
      "num_input_tokens_seen": 64773040,
      "step": 30010
    },
    {
      "epoch": 4.896411092985318,
      "grad_norm": 0.1310116946697235,
      "learning_rate": 3.02298150147801e-05,
      "loss": 0.2538,
      "num_input_tokens_seen": 64785584,
      "step": 30015
    },
    {
      "epoch": 4.897226753670473,
      "grad_norm": 0.09187537431716919,
      "learning_rate": 3.022285430984367e-05,
      "loss": 0.076,
      "num_input_tokens_seen": 64797648,
      "step": 30020
    },
    {
      "epoch": 4.898042414355628,
      "grad_norm": 2.4482696056365967,
      "learning_rate": 3.0215893181516163e-05,
      "loss": 0.0103,
      "num_input_tokens_seen": 64808912,
      "step": 30025
    },
    {
      "epoch": 4.898858075040783,
      "grad_norm": 0.2752697467803955,
      "learning_rate": 3.02089316303619e-05,
      "loss": 0.2575,
      "num_input_tokens_seen": 64820336,
      "step": 30030
    },
    {
      "epoch": 4.899673735725938,
      "grad_norm": 0.044006578624248505,
      "learning_rate": 3.0201969656945196e-05,
      "loss": 0.0275,
      "num_input_tokens_seen": 64831600,
      "step": 30035
    },
    {
      "epoch": 4.900489396411093,
      "grad_norm": 0.11853750795125961,
      "learning_rate": 3.0195007261830438e-05,
      "loss": 0.0452,
      "num_input_tokens_seen": 64842000,
      "step": 30040
    },
    {
      "epoch": 4.901305057096248,
      "grad_norm": 1.8368370532989502,
      "learning_rate": 3.018804444558204e-05,
      "loss": 0.0742,
      "num_input_tokens_seen": 64853456,
      "step": 30045
    },
    {
      "epoch": 4.902120717781403,
      "grad_norm": 0.20973731577396393,
      "learning_rate": 3.018108120876443e-05,
      "loss": 0.0836,
      "num_input_tokens_seen": 64863952,
      "step": 30050
    },
    {
      "epoch": 4.902936378466558,
      "grad_norm": 8.172491073608398,
      "learning_rate": 3.0174117551942087e-05,
      "loss": 0.1811,
      "num_input_tokens_seen": 64874128,
      "step": 30055
    },
    {
      "epoch": 4.903752039151713,
      "grad_norm": 0.05342257767915726,
      "learning_rate": 3.0167153475679527e-05,
      "loss": 0.0182,
      "num_input_tokens_seen": 64883856,
      "step": 30060
    },
    {
      "epoch": 4.904567699836868,
      "grad_norm": 2.8744332790374756,
      "learning_rate": 3.0160188980541288e-05,
      "loss": 0.1477,
      "num_input_tokens_seen": 64894736,
      "step": 30065
    },
    {
      "epoch": 4.9053833605220225,
      "grad_norm": 0.08954238891601562,
      "learning_rate": 3.0153224067091952e-05,
      "loss": 0.0055,
      "num_input_tokens_seen": 64905808,
      "step": 30070
    },
    {
      "epoch": 4.906199021207177,
      "grad_norm": 0.4676968455314636,
      "learning_rate": 3.0146258735896117e-05,
      "loss": 0.2841,
      "num_input_tokens_seen": 64917104,
      "step": 30075
    },
    {
      "epoch": 4.907014681892333,
      "grad_norm": 0.09429818391799927,
      "learning_rate": 3.0139292987518443e-05,
      "loss": 0.0045,
      "num_input_tokens_seen": 64926800,
      "step": 30080
    },
    {
      "epoch": 4.907830342577488,
      "grad_norm": 0.05108118802309036,
      "learning_rate": 3.0132326822523606e-05,
      "loss": 0.1334,
      "num_input_tokens_seen": 64937904,
      "step": 30085
    },
    {
      "epoch": 4.908646003262643,
      "grad_norm": 0.02007273957133293,
      "learning_rate": 3.01253602414763e-05,
      "loss": 0.2642,
      "num_input_tokens_seen": 64948592,
      "step": 30090
    },
    {
      "epoch": 4.9094616639477975,
      "grad_norm": 0.3647564649581909,
      "learning_rate": 3.0118393244941302e-05,
      "loss": 0.0283,
      "num_input_tokens_seen": 64959152,
      "step": 30095
    },
    {
      "epoch": 4.910277324632952,
      "grad_norm": 0.09454967081546783,
      "learning_rate": 3.011142583348337e-05,
      "loss": 0.4647,
      "num_input_tokens_seen": 64970736,
      "step": 30100
    },
    {
      "epoch": 4.911092985318108,
      "grad_norm": 6.534158706665039,
      "learning_rate": 3.010445800766733e-05,
      "loss": 0.1598,
      "num_input_tokens_seen": 64982032,
      "step": 30105
    },
    {
      "epoch": 4.911908646003263,
      "grad_norm": 0.4977462589740753,
      "learning_rate": 3.0097489768058022e-05,
      "loss": 0.0744,
      "num_input_tokens_seen": 64994448,
      "step": 30110
    },
    {
      "epoch": 4.912724306688418,
      "grad_norm": 0.10170254111289978,
      "learning_rate": 3.0090521115220327e-05,
      "loss": 0.1168,
      "num_input_tokens_seen": 65005744,
      "step": 30115
    },
    {
      "epoch": 4.9135399673735725,
      "grad_norm": 0.05971534177660942,
      "learning_rate": 3.0083552049719167e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 65016848,
      "step": 30120
    },
    {
      "epoch": 4.914355628058727,
      "grad_norm": 0.925605833530426,
      "learning_rate": 3.0076582572119473e-05,
      "loss": 0.1789,
      "num_input_tokens_seen": 65027600,
      "step": 30125
    },
    {
      "epoch": 4.915171288743883,
      "grad_norm": 0.20594875514507294,
      "learning_rate": 3.0069612682986247e-05,
      "loss": 0.1097,
      "num_input_tokens_seen": 65039120,
      "step": 30130
    },
    {
      "epoch": 4.915986949429038,
      "grad_norm": 11.441937446594238,
      "learning_rate": 3.0062642382884494e-05,
      "loss": 0.0927,
      "num_input_tokens_seen": 65050512,
      "step": 30135
    },
    {
      "epoch": 4.916802610114193,
      "grad_norm": 3.2262916564941406,
      "learning_rate": 3.005567167237926e-05,
      "loss": 0.1628,
      "num_input_tokens_seen": 65059888,
      "step": 30140
    },
    {
      "epoch": 4.917618270799347,
      "grad_norm": 0.2380727231502533,
      "learning_rate": 3.004870055203562e-05,
      "loss": 0.0341,
      "num_input_tokens_seen": 65071120,
      "step": 30145
    },
    {
      "epoch": 4.918433931484502,
      "grad_norm": 0.08102700114250183,
      "learning_rate": 3.0041729022418702e-05,
      "loss": 0.1206,
      "num_input_tokens_seen": 65081744,
      "step": 30150
    },
    {
      "epoch": 4.919249592169657,
      "grad_norm": 0.11428258568048477,
      "learning_rate": 3.003475708409365e-05,
      "loss": 0.0859,
      "num_input_tokens_seen": 65093040,
      "step": 30155
    },
    {
      "epoch": 4.920065252854813,
      "grad_norm": 0.03618955612182617,
      "learning_rate": 3.0027784737625646e-05,
      "loss": 0.1927,
      "num_input_tokens_seen": 65103056,
      "step": 30160
    },
    {
      "epoch": 4.920880913539968,
      "grad_norm": 0.0515863411128521,
      "learning_rate": 3.00208119835799e-05,
      "loss": 0.0133,
      "num_input_tokens_seen": 65113904,
      "step": 30165
    },
    {
      "epoch": 4.921696574225122,
      "grad_norm": 3.6284677982330322,
      "learning_rate": 3.0013838822521655e-05,
      "loss": 0.2296,
      "num_input_tokens_seen": 65124016,
      "step": 30170
    },
    {
      "epoch": 4.922512234910277,
      "grad_norm": 0.09662287682294846,
      "learning_rate": 3.0006865255016192e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 65135120,
      "step": 30175
    },
    {
      "epoch": 4.923327895595432,
      "grad_norm": 0.07900609076023102,
      "learning_rate": 2.9999891281628832e-05,
      "loss": 0.2014,
      "num_input_tokens_seen": 65144784,
      "step": 30180
    },
    {
      "epoch": 4.924143556280587,
      "grad_norm": 6.857663154602051,
      "learning_rate": 2.9992916902924917e-05,
      "loss": 0.1551,
      "num_input_tokens_seen": 65154928,
      "step": 30185
    },
    {
      "epoch": 4.924959216965743,
      "grad_norm": 0.1024380773305893,
      "learning_rate": 2.998594211946982e-05,
      "loss": 0.0079,
      "num_input_tokens_seen": 65165808,
      "step": 30190
    },
    {
      "epoch": 4.925774877650897,
      "grad_norm": 0.13860879838466644,
      "learning_rate": 2.9978966931828957e-05,
      "loss": 0.0472,
      "num_input_tokens_seen": 65175952,
      "step": 30195
    },
    {
      "epoch": 4.926590538336052,
      "grad_norm": 0.13814367353916168,
      "learning_rate": 2.9971991340567773e-05,
      "loss": 0.1278,
      "num_input_tokens_seen": 65185904,
      "step": 30200
    },
    {
      "epoch": 4.927406199021207,
      "grad_norm": 0.0338091216981411,
      "learning_rate": 2.9965015346251747e-05,
      "loss": 0.04,
      "num_input_tokens_seen": 65196848,
      "step": 30205
    },
    {
      "epoch": 4.928221859706362,
      "grad_norm": 0.2112356722354889,
      "learning_rate": 2.995803894944637e-05,
      "loss": 0.2068,
      "num_input_tokens_seen": 65207440,
      "step": 30210
    },
    {
      "epoch": 4.9290375203915175,
      "grad_norm": 0.06825273483991623,
      "learning_rate": 2.9951062150717212e-05,
      "loss": 0.0128,
      "num_input_tokens_seen": 65217904,
      "step": 30215
    },
    {
      "epoch": 4.929853181076672,
      "grad_norm": 2.894876480102539,
      "learning_rate": 2.994408495062983e-05,
      "loss": 0.0186,
      "num_input_tokens_seen": 65228464,
      "step": 30220
    },
    {
      "epoch": 4.930668841761827,
      "grad_norm": 3.265835762023926,
      "learning_rate": 2.9937107349749842e-05,
      "loss": 0.1604,
      "num_input_tokens_seen": 65239024,
      "step": 30225
    },
    {
      "epoch": 4.931484502446982,
      "grad_norm": 10.262737274169922,
      "learning_rate": 2.9930129348642877e-05,
      "loss": 0.0816,
      "num_input_tokens_seen": 65249872,
      "step": 30230
    },
    {
      "epoch": 4.932300163132137,
      "grad_norm": 0.12191098928451538,
      "learning_rate": 2.9923150947874613e-05,
      "loss": 0.2379,
      "num_input_tokens_seen": 65260560,
      "step": 30235
    },
    {
      "epoch": 4.933115823817292,
      "grad_norm": 1.5043644905090332,
      "learning_rate": 2.991617214801075e-05,
      "loss": 0.0125,
      "num_input_tokens_seen": 65272016,
      "step": 30240
    },
    {
      "epoch": 4.933931484502447,
      "grad_norm": 0.11814262717962265,
      "learning_rate": 2.9909192949617036e-05,
      "loss": 0.0103,
      "num_input_tokens_seen": 65281136,
      "step": 30245
    },
    {
      "epoch": 4.934747145187602,
      "grad_norm": 3.2631006240844727,
      "learning_rate": 2.9902213353259223e-05,
      "loss": 0.1033,
      "num_input_tokens_seen": 65290640,
      "step": 30250
    },
    {
      "epoch": 4.935562805872757,
      "grad_norm": 11.115103721618652,
      "learning_rate": 2.989523335950313e-05,
      "loss": 0.2474,
      "num_input_tokens_seen": 65302032,
      "step": 30255
    },
    {
      "epoch": 4.936378466557912,
      "grad_norm": 0.1490999013185501,
      "learning_rate": 2.9888252968914576e-05,
      "loss": 0.2117,
      "num_input_tokens_seen": 65313168,
      "step": 30260
    },
    {
      "epoch": 4.937194127243067,
      "grad_norm": 0.15903590619564056,
      "learning_rate": 2.988127218205944e-05,
      "loss": 0.0053,
      "num_input_tokens_seen": 65324528,
      "step": 30265
    },
    {
      "epoch": 4.938009787928221,
      "grad_norm": 4.044406890869141,
      "learning_rate": 2.9874290999503606e-05,
      "loss": 0.0204,
      "num_input_tokens_seen": 65335568,
      "step": 30270
    },
    {
      "epoch": 4.938825448613377,
      "grad_norm": 3.8695523738861084,
      "learning_rate": 2.9867309421813018e-05,
      "loss": 0.2925,
      "num_input_tokens_seen": 65346032,
      "step": 30275
    },
    {
      "epoch": 4.939641109298532,
      "grad_norm": 0.18034009635448456,
      "learning_rate": 2.9860327449553626e-05,
      "loss": 0.1214,
      "num_input_tokens_seen": 65357712,
      "step": 30280
    },
    {
      "epoch": 4.940456769983687,
      "grad_norm": 0.035709358751773834,
      "learning_rate": 2.9853345083291434e-05,
      "loss": 0.0385,
      "num_input_tokens_seen": 65369552,
      "step": 30285
    },
    {
      "epoch": 4.941272430668842,
      "grad_norm": 10.548879623413086,
      "learning_rate": 2.9846362323592463e-05,
      "loss": 0.1767,
      "num_input_tokens_seen": 65380912,
      "step": 30290
    },
    {
      "epoch": 4.942088091353996,
      "grad_norm": 0.2996707558631897,
      "learning_rate": 2.9839379171022776e-05,
      "loss": 0.3709,
      "num_input_tokens_seen": 65392080,
      "step": 30295
    },
    {
      "epoch": 4.942903752039152,
      "grad_norm": 0.05010681226849556,
      "learning_rate": 2.983239562614845e-05,
      "loss": 0.0053,
      "num_input_tokens_seen": 65402256,
      "step": 30300
    },
    {
      "epoch": 4.943719412724307,
      "grad_norm": 1.4723745584487915,
      "learning_rate": 2.982541168953562e-05,
      "loss": 0.1478,
      "num_input_tokens_seen": 65413456,
      "step": 30305
    },
    {
      "epoch": 4.944535073409462,
      "grad_norm": 0.14409276843070984,
      "learning_rate": 2.9818427361750434e-05,
      "loss": 0.0575,
      "num_input_tokens_seen": 65424912,
      "step": 30310
    },
    {
      "epoch": 4.945350734094617,
      "grad_norm": 0.11816307157278061,
      "learning_rate": 2.9811442643359076e-05,
      "loss": 0.0052,
      "num_input_tokens_seen": 65434896,
      "step": 30315
    },
    {
      "epoch": 4.946166394779771,
      "grad_norm": 7.738298416137695,
      "learning_rate": 2.9804457534927772e-05,
      "loss": 0.0473,
      "num_input_tokens_seen": 65445424,
      "step": 30320
    },
    {
      "epoch": 4.946982055464927,
      "grad_norm": 0.046474017202854156,
      "learning_rate": 2.9797472037022757e-05,
      "loss": 0.089,
      "num_input_tokens_seen": 65454672,
      "step": 30325
    },
    {
      "epoch": 4.947797716150082,
      "grad_norm": 0.07026860117912292,
      "learning_rate": 2.9790486150210316e-05,
      "loss": 0.0908,
      "num_input_tokens_seen": 65465776,
      "step": 30330
    },
    {
      "epoch": 4.948613376835237,
      "grad_norm": 5.883081912994385,
      "learning_rate": 2.9783499875056766e-05,
      "loss": 0.3138,
      "num_input_tokens_seen": 65477168,
      "step": 30335
    },
    {
      "epoch": 4.9494290375203915,
      "grad_norm": 1.2314774990081787,
      "learning_rate": 2.9776513212128442e-05,
      "loss": 0.0062,
      "num_input_tokens_seen": 65487248,
      "step": 30340
    },
    {
      "epoch": 4.950244698205546,
      "grad_norm": 0.18590541183948517,
      "learning_rate": 2.976952616199172e-05,
      "loss": 0.0106,
      "num_input_tokens_seen": 65497360,
      "step": 30345
    },
    {
      "epoch": 4.951060358890701,
      "grad_norm": 0.14897702634334564,
      "learning_rate": 2.9762538725213007e-05,
      "loss": 0.1456,
      "num_input_tokens_seen": 65507728,
      "step": 30350
    },
    {
      "epoch": 4.951876019575856,
      "grad_norm": 0.5946313142776489,
      "learning_rate": 2.9755550902358737e-05,
      "loss": 0.0991,
      "num_input_tokens_seen": 65517104,
      "step": 30355
    },
    {
      "epoch": 4.952691680261012,
      "grad_norm": 3.074784755706787,
      "learning_rate": 2.9748562693995386e-05,
      "loss": 0.204,
      "num_input_tokens_seen": 65529168,
      "step": 30360
    },
    {
      "epoch": 4.9535073409461665,
      "grad_norm": 3.872284173965454,
      "learning_rate": 2.974157410068944e-05,
      "loss": 0.0447,
      "num_input_tokens_seen": 65538960,
      "step": 30365
    },
    {
      "epoch": 4.954323001631321,
      "grad_norm": 0.016039861366152763,
      "learning_rate": 2.9734585123007446e-05,
      "loss": 0.0053,
      "num_input_tokens_seen": 65549392,
      "step": 30370
    },
    {
      "epoch": 4.955138662316476,
      "grad_norm": 3.6199285984039307,
      "learning_rate": 2.9727595761515958e-05,
      "loss": 0.1062,
      "num_input_tokens_seen": 65559824,
      "step": 30375
    },
    {
      "epoch": 4.955954323001631,
      "grad_norm": 12.265833854675293,
      "learning_rate": 2.9720606016781577e-05,
      "loss": 0.2147,
      "num_input_tokens_seen": 65570896,
      "step": 30380
    },
    {
      "epoch": 4.956769983686787,
      "grad_norm": 0.09133773297071457,
      "learning_rate": 2.9713615889370917e-05,
      "loss": 0.1078,
      "num_input_tokens_seen": 65582384,
      "step": 30385
    },
    {
      "epoch": 4.9575856443719415,
      "grad_norm": 6.277388572692871,
      "learning_rate": 2.9706625379850627e-05,
      "loss": 0.3821,
      "num_input_tokens_seen": 65593264,
      "step": 30390
    },
    {
      "epoch": 4.958401305057096,
      "grad_norm": 0.1470021903514862,
      "learning_rate": 2.9699634488787415e-05,
      "loss": 0.2508,
      "num_input_tokens_seen": 65604016,
      "step": 30395
    },
    {
      "epoch": 4.959216965742251,
      "grad_norm": 1.6399931907653809,
      "learning_rate": 2.9692643216747978e-05,
      "loss": 0.0532,
      "num_input_tokens_seen": 65613616,
      "step": 30400
    },
    {
      "epoch": 4.960032626427406,
      "grad_norm": 3.120595932006836,
      "learning_rate": 2.9685651564299077e-05,
      "loss": 0.2575,
      "num_input_tokens_seen": 65624816,
      "step": 30405
    },
    {
      "epoch": 4.960848287112562,
      "grad_norm": 0.289460688829422,
      "learning_rate": 2.9678659532007475e-05,
      "loss": 0.1482,
      "num_input_tokens_seen": 65635792,
      "step": 30410
    },
    {
      "epoch": 4.9616639477977165,
      "grad_norm": 1.108140230178833,
      "learning_rate": 2.967166712044e-05,
      "loss": 0.1776,
      "num_input_tokens_seen": 65646512,
      "step": 30415
    },
    {
      "epoch": 4.962479608482871,
      "grad_norm": 1.021653413772583,
      "learning_rate": 2.9664674330163485e-05,
      "loss": 0.0673,
      "num_input_tokens_seen": 65656976,
      "step": 30420
    },
    {
      "epoch": 4.963295269168026,
      "grad_norm": 0.5491104125976562,
      "learning_rate": 2.96576811617448e-05,
      "loss": 0.0539,
      "num_input_tokens_seen": 65667792,
      "step": 30425
    },
    {
      "epoch": 4.964110929853181,
      "grad_norm": 0.3017660081386566,
      "learning_rate": 2.9650687615750843e-05,
      "loss": 0.1247,
      "num_input_tokens_seen": 65679088,
      "step": 30430
    },
    {
      "epoch": 4.964926590538336,
      "grad_norm": 1.055377721786499,
      "learning_rate": 2.964369369274856e-05,
      "loss": 0.0698,
      "num_input_tokens_seen": 65689840,
      "step": 30435
    },
    {
      "epoch": 4.9657422512234906,
      "grad_norm": 0.3545195758342743,
      "learning_rate": 2.963669939330489e-05,
      "loss": 0.1265,
      "num_input_tokens_seen": 65700528,
      "step": 30440
    },
    {
      "epoch": 4.966557911908646,
      "grad_norm": 0.11180157959461212,
      "learning_rate": 2.962970471798685e-05,
      "loss": 0.009,
      "num_input_tokens_seen": 65710128,
      "step": 30445
    },
    {
      "epoch": 4.967373572593801,
      "grad_norm": 0.06180586665868759,
      "learning_rate": 2.9622709667361455e-05,
      "loss": 0.2663,
      "num_input_tokens_seen": 65722064,
      "step": 30450
    },
    {
      "epoch": 4.968189233278956,
      "grad_norm": 0.6542524695396423,
      "learning_rate": 2.9615714241995758e-05,
      "loss": 0.0117,
      "num_input_tokens_seen": 65732336,
      "step": 30455
    },
    {
      "epoch": 4.969004893964111,
      "grad_norm": 0.08314738422632217,
      "learning_rate": 2.9608718442456844e-05,
      "loss": 0.0121,
      "num_input_tokens_seen": 65743504,
      "step": 30460
    },
    {
      "epoch": 4.9698205546492655,
      "grad_norm": 0.39737963676452637,
      "learning_rate": 2.9601722269311827e-05,
      "loss": 0.0133,
      "num_input_tokens_seen": 65754608,
      "step": 30465
    },
    {
      "epoch": 4.970636215334421,
      "grad_norm": 0.9284301996231079,
      "learning_rate": 2.9594725723127855e-05,
      "loss": 0.1873,
      "num_input_tokens_seen": 65765584,
      "step": 30470
    },
    {
      "epoch": 4.971451876019576,
      "grad_norm": 0.0789085254073143,
      "learning_rate": 2.9587728804472104e-05,
      "loss": 0.2233,
      "num_input_tokens_seen": 65776816,
      "step": 30475
    },
    {
      "epoch": 4.972267536704731,
      "grad_norm": 3.007207155227661,
      "learning_rate": 2.9580731513911773e-05,
      "loss": 0.2211,
      "num_input_tokens_seen": 65788336,
      "step": 30480
    },
    {
      "epoch": 4.973083197389886,
      "grad_norm": 0.045617252588272095,
      "learning_rate": 2.9573733852014112e-05,
      "loss": 0.22,
      "num_input_tokens_seen": 65798704,
      "step": 30485
    },
    {
      "epoch": 4.9738988580750405,
      "grad_norm": 0.6935256123542786,
      "learning_rate": 2.9566735819346376e-05,
      "loss": 0.0459,
      "num_input_tokens_seen": 65808816,
      "step": 30490
    },
    {
      "epoch": 4.974714518760196,
      "grad_norm": 0.8358042240142822,
      "learning_rate": 2.9559737416475863e-05,
      "loss": 0.0905,
      "num_input_tokens_seen": 65819088,
      "step": 30495
    },
    {
      "epoch": 4.975530179445351,
      "grad_norm": 2.981520414352417,
      "learning_rate": 2.9552738643969896e-05,
      "loss": 0.1319,
      "num_input_tokens_seen": 65830256,
      "step": 30500
    },
    {
      "epoch": 4.976345840130506,
      "grad_norm": 0.02853558212518692,
      "learning_rate": 2.9545739502395835e-05,
      "loss": 0.0606,
      "num_input_tokens_seen": 65839792,
      "step": 30505
    },
    {
      "epoch": 4.977161500815661,
      "grad_norm": 0.12231956422328949,
      "learning_rate": 2.9538739992321062e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 65850512,
      "step": 30510
    },
    {
      "epoch": 4.9779771615008155,
      "grad_norm": 0.06114667281508446,
      "learning_rate": 2.9531740114313e-05,
      "loss": 0.1149,
      "num_input_tokens_seen": 65861584,
      "step": 30515
    },
    {
      "epoch": 4.97879282218597,
      "grad_norm": 1.5656062364578247,
      "learning_rate": 2.9524739868939088e-05,
      "loss": 0.3395,
      "num_input_tokens_seen": 65872880,
      "step": 30520
    },
    {
      "epoch": 4.979608482871125,
      "grad_norm": 0.08660892397165298,
      "learning_rate": 2.9517739256766803e-05,
      "loss": 0.0996,
      "num_input_tokens_seen": 65882544,
      "step": 30525
    },
    {
      "epoch": 4.980424143556281,
      "grad_norm": 0.043251391500234604,
      "learning_rate": 2.9510738278363652e-05,
      "loss": 0.0092,
      "num_input_tokens_seen": 65893456,
      "step": 30530
    },
    {
      "epoch": 4.981239804241436,
      "grad_norm": 0.7454972267150879,
      "learning_rate": 2.950373693429717e-05,
      "loss": 0.0158,
      "num_input_tokens_seen": 65905264,
      "step": 30535
    },
    {
      "epoch": 4.9820554649265905,
      "grad_norm": 4.430110454559326,
      "learning_rate": 2.949673522513492e-05,
      "loss": 0.2665,
      "num_input_tokens_seen": 65916400,
      "step": 30540
    },
    {
      "epoch": 4.982871125611745,
      "grad_norm": 0.11500086635351181,
      "learning_rate": 2.9489733151444497e-05,
      "loss": 0.1166,
      "num_input_tokens_seen": 65928272,
      "step": 30545
    },
    {
      "epoch": 4.9836867862969,
      "grad_norm": 0.3506146967411041,
      "learning_rate": 2.9482730713793526e-05,
      "loss": 0.0104,
      "num_input_tokens_seen": 65939792,
      "step": 30550
    },
    {
      "epoch": 4.984502446982056,
      "grad_norm": 0.10922025144100189,
      "learning_rate": 2.9475727912749656e-05,
      "loss": 0.1016,
      "num_input_tokens_seen": 65950576,
      "step": 30555
    },
    {
      "epoch": 4.985318107667211,
      "grad_norm": 9.96968936920166,
      "learning_rate": 2.946872474888058e-05,
      "loss": 0.3188,
      "num_input_tokens_seen": 65960240,
      "step": 30560
    },
    {
      "epoch": 4.986133768352365,
      "grad_norm": 0.38060101866722107,
      "learning_rate": 2.9461721222753992e-05,
      "loss": 0.2204,
      "num_input_tokens_seen": 65970544,
      "step": 30565
    },
    {
      "epoch": 4.98694942903752,
      "grad_norm": 0.05818890407681465,
      "learning_rate": 2.9454717334937638e-05,
      "loss": 0.0043,
      "num_input_tokens_seen": 65982768,
      "step": 30570
    },
    {
      "epoch": 4.987765089722675,
      "grad_norm": 0.23830565810203552,
      "learning_rate": 2.94477130859993e-05,
      "loss": 0.0388,
      "num_input_tokens_seen": 65993680,
      "step": 30575
    },
    {
      "epoch": 4.988580750407831,
      "grad_norm": 2.3207828998565674,
      "learning_rate": 2.9440708476506773e-05,
      "loss": 0.2399,
      "num_input_tokens_seen": 66005136,
      "step": 30580
    },
    {
      "epoch": 4.989396411092986,
      "grad_norm": 1.1624802350997925,
      "learning_rate": 2.943370350702789e-05,
      "loss": 0.151,
      "num_input_tokens_seen": 66015376,
      "step": 30585
    },
    {
      "epoch": 4.99021207177814,
      "grad_norm": 0.3364430069923401,
      "learning_rate": 2.9426698178130495e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 66025904,
      "step": 30590
    },
    {
      "epoch": 4.991027732463295,
      "grad_norm": 0.17804332077503204,
      "learning_rate": 2.9419692490382488e-05,
      "loss": 0.0859,
      "num_input_tokens_seen": 66036560,
      "step": 30595
    },
    {
      "epoch": 4.99184339314845,
      "grad_norm": 1.2029184103012085,
      "learning_rate": 2.9412686444351782e-05,
      "loss": 0.1147,
      "num_input_tokens_seen": 66047856,
      "step": 30600
    },
    {
      "epoch": 4.992659053833605,
      "grad_norm": 0.0865970030426979,
      "learning_rate": 2.9405680040606326e-05,
      "loss": 0.2751,
      "num_input_tokens_seen": 66058736,
      "step": 30605
    },
    {
      "epoch": 4.993474714518761,
      "grad_norm": 4.403013706207275,
      "learning_rate": 2.939867327971409e-05,
      "loss": 0.0799,
      "num_input_tokens_seen": 66068848,
      "step": 30610
    },
    {
      "epoch": 4.994290375203915,
      "grad_norm": 12.007497787475586,
      "learning_rate": 2.939166616224308e-05,
      "loss": 0.0594,
      "num_input_tokens_seen": 66079760,
      "step": 30615
    },
    {
      "epoch": 4.99510603588907,
      "grad_norm": 0.08808305114507675,
      "learning_rate": 2.938465868876133e-05,
      "loss": 0.1133,
      "num_input_tokens_seen": 66090128,
      "step": 30620
    },
    {
      "epoch": 4.995921696574225,
      "grad_norm": 15.229327201843262,
      "learning_rate": 2.9377650859836892e-05,
      "loss": 0.0856,
      "num_input_tokens_seen": 66100304,
      "step": 30625
    },
    {
      "epoch": 4.99673735725938,
      "grad_norm": 1.8007338047027588,
      "learning_rate": 2.9370642676037867e-05,
      "loss": 0.2761,
      "num_input_tokens_seen": 66111696,
      "step": 30630
    },
    {
      "epoch": 4.997553017944535,
      "grad_norm": 1.2313814163208008,
      "learning_rate": 2.936363413793237e-05,
      "loss": 0.2358,
      "num_input_tokens_seen": 66122064,
      "step": 30635
    },
    {
      "epoch": 4.99836867862969,
      "grad_norm": 0.06142730638384819,
      "learning_rate": 2.9356625246088554e-05,
      "loss": 0.0265,
      "num_input_tokens_seen": 66131376,
      "step": 30640
    },
    {
      "epoch": 4.999184339314845,
      "grad_norm": 0.09649068117141724,
      "learning_rate": 2.9349616001074588e-05,
      "loss": 0.2084,
      "num_input_tokens_seen": 66142800,
      "step": 30645
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.20904871821403503,
      "learning_rate": 2.934260640345867e-05,
      "loss": 0.1177,
      "num_input_tokens_seen": 66152480,
      "step": 30650
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.15915174782276154,
      "eval_runtime": 132.9324,
      "eval_samples_per_second": 20.499,
      "eval_steps_per_second": 5.13,
      "num_input_tokens_seen": 66152480,
      "step": 30650
    },
    {
      "epoch": 5.000815660685155,
      "grad_norm": 1.0603883266448975,
      "learning_rate": 2.9335596453809055e-05,
      "loss": 0.01,
      "num_input_tokens_seen": 66162912,
      "step": 30655
    },
    {
      "epoch": 5.00163132137031,
      "grad_norm": 6.042734146118164,
      "learning_rate": 2.9328586152693986e-05,
      "loss": 0.0805,
      "num_input_tokens_seen": 66173312,
      "step": 30660
    },
    {
      "epoch": 5.002446982055465,
      "grad_norm": 0.6006112098693848,
      "learning_rate": 2.9321575500681757e-05,
      "loss": 0.1409,
      "num_input_tokens_seen": 66184832,
      "step": 30665
    },
    {
      "epoch": 5.00326264274062,
      "grad_norm": 0.22609871625900269,
      "learning_rate": 2.93145644983407e-05,
      "loss": 0.0269,
      "num_input_tokens_seen": 66196544,
      "step": 30670
    },
    {
      "epoch": 5.004078303425775,
      "grad_norm": 0.08074972778558731,
      "learning_rate": 2.9307553146239146e-05,
      "loss": 0.0122,
      "num_input_tokens_seen": 66207904,
      "step": 30675
    },
    {
      "epoch": 5.00489396411093,
      "grad_norm": 0.10380563139915466,
      "learning_rate": 2.930054144494548e-05,
      "loss": 0.1043,
      "num_input_tokens_seen": 66218752,
      "step": 30680
    },
    {
      "epoch": 5.005709624796085,
      "grad_norm": 0.12653124332427979,
      "learning_rate": 2.9293529395028102e-05,
      "loss": 0.0058,
      "num_input_tokens_seen": 66230528,
      "step": 30685
    },
    {
      "epoch": 5.006525285481239,
      "grad_norm": 5.836536884307861,
      "learning_rate": 2.928651699705545e-05,
      "loss": 0.0815,
      "num_input_tokens_seen": 66240768,
      "step": 30690
    },
    {
      "epoch": 5.007340946166395,
      "grad_norm": 3.5016517639160156,
      "learning_rate": 2.927950425159598e-05,
      "loss": 0.1495,
      "num_input_tokens_seen": 66252672,
      "step": 30695
    },
    {
      "epoch": 5.00815660685155,
      "grad_norm": 0.23449905216693878,
      "learning_rate": 2.927249115921818e-05,
      "loss": 0.0862,
      "num_input_tokens_seen": 66263520,
      "step": 30700
    },
    {
      "epoch": 5.008972267536705,
      "grad_norm": 0.032858956605196,
      "learning_rate": 2.926547772049057e-05,
      "loss": 0.0033,
      "num_input_tokens_seen": 66274752,
      "step": 30705
    },
    {
      "epoch": 5.00978792822186,
      "grad_norm": 0.19907324016094208,
      "learning_rate": 2.9258463935981696e-05,
      "loss": 0.3923,
      "num_input_tokens_seen": 66285440,
      "step": 30710
    },
    {
      "epoch": 5.010603588907014,
      "grad_norm": 0.20320872962474823,
      "learning_rate": 2.9251449806260122e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 66295040,
      "step": 30715
    },
    {
      "epoch": 5.011419249592169,
      "grad_norm": 0.09244179725646973,
      "learning_rate": 2.9244435331894454e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 66304384,
      "step": 30720
    },
    {
      "epoch": 5.012234910277325,
      "grad_norm": 0.0688881129026413,
      "learning_rate": 2.9237420513453328e-05,
      "loss": 0.1721,
      "num_input_tokens_seen": 66315648,
      "step": 30725
    },
    {
      "epoch": 5.01305057096248,
      "grad_norm": 0.08274415135383606,
      "learning_rate": 2.9230405351505386e-05,
      "loss": 0.0121,
      "num_input_tokens_seen": 66325664,
      "step": 30730
    },
    {
      "epoch": 5.013866231647635,
      "grad_norm": 0.14355804026126862,
      "learning_rate": 2.9223389846619326e-05,
      "loss": 0.0457,
      "num_input_tokens_seen": 66336544,
      "step": 30735
    },
    {
      "epoch": 5.014681892332789,
      "grad_norm": 0.4062250852584839,
      "learning_rate": 2.921637399936386e-05,
      "loss": 0.0097,
      "num_input_tokens_seen": 66347360,
      "step": 30740
    },
    {
      "epoch": 5.015497553017944,
      "grad_norm": 0.1017821654677391,
      "learning_rate": 2.920935781030772e-05,
      "loss": 0.0094,
      "num_input_tokens_seen": 66356992,
      "step": 30745
    },
    {
      "epoch": 5.0163132137031,
      "grad_norm": 0.06947299093008041,
      "learning_rate": 2.9202341280019675e-05,
      "loss": 0.0073,
      "num_input_tokens_seen": 66368032,
      "step": 30750
    },
    {
      "epoch": 5.017128874388255,
      "grad_norm": 1.3397068977355957,
      "learning_rate": 2.9195324409068525e-05,
      "loss": 0.1538,
      "num_input_tokens_seen": 66379424,
      "step": 30755
    },
    {
      "epoch": 5.0179445350734095,
      "grad_norm": 0.055242735892534256,
      "learning_rate": 2.9188307198023095e-05,
      "loss": 0.0166,
      "num_input_tokens_seen": 66390752,
      "step": 30760
    },
    {
      "epoch": 5.018760195758564,
      "grad_norm": 0.14783857762813568,
      "learning_rate": 2.918128964745223e-05,
      "loss": 0.0706,
      "num_input_tokens_seen": 66403200,
      "step": 30765
    },
    {
      "epoch": 5.019575856443719,
      "grad_norm": 0.038159240037202835,
      "learning_rate": 2.9174271757924814e-05,
      "loss": 0.0716,
      "num_input_tokens_seen": 66412768,
      "step": 30770
    },
    {
      "epoch": 5.020391517128874,
      "grad_norm": 3.5265026092529297,
      "learning_rate": 2.9167253530009748e-05,
      "loss": 0.1267,
      "num_input_tokens_seen": 66424448,
      "step": 30775
    },
    {
      "epoch": 5.02120717781403,
      "grad_norm": 0.13227763772010803,
      "learning_rate": 2.9160234964275963e-05,
      "loss": 0.218,
      "num_input_tokens_seen": 66434400,
      "step": 30780
    },
    {
      "epoch": 5.0220228384991845,
      "grad_norm": 0.04481755942106247,
      "learning_rate": 2.915321606129242e-05,
      "loss": 0.298,
      "num_input_tokens_seen": 66445440,
      "step": 30785
    },
    {
      "epoch": 5.022838499184339,
      "grad_norm": 2.932248115539551,
      "learning_rate": 2.9146196821628113e-05,
      "loss": 0.238,
      "num_input_tokens_seen": 66455776,
      "step": 30790
    },
    {
      "epoch": 5.023654159869494,
      "grad_norm": 0.03845330327749252,
      "learning_rate": 2.9139177245852056e-05,
      "loss": 0.0043,
      "num_input_tokens_seen": 66467616,
      "step": 30795
    },
    {
      "epoch": 5.024469820554649,
      "grad_norm": 0.18302397429943085,
      "learning_rate": 2.9132157334533295e-05,
      "loss": 0.0971,
      "num_input_tokens_seen": 66477056,
      "step": 30800
    },
    {
      "epoch": 5.025285481239805,
      "grad_norm": 0.14941471815109253,
      "learning_rate": 2.9125137088240885e-05,
      "loss": 0.1101,
      "num_input_tokens_seen": 66488352,
      "step": 30805
    },
    {
      "epoch": 5.0261011419249595,
      "grad_norm": 0.18477576971054077,
      "learning_rate": 2.9118116507543936e-05,
      "loss": 0.1074,
      "num_input_tokens_seen": 66497824,
      "step": 30810
    },
    {
      "epoch": 5.026916802610114,
      "grad_norm": 0.09586291015148163,
      "learning_rate": 2.9111095593011567e-05,
      "loss": 0.0776,
      "num_input_tokens_seen": 66509920,
      "step": 30815
    },
    {
      "epoch": 5.027732463295269,
      "grad_norm": 0.32964301109313965,
      "learning_rate": 2.9104074345212933e-05,
      "loss": 0.0103,
      "num_input_tokens_seen": 66519520,
      "step": 30820
    },
    {
      "epoch": 5.028548123980424,
      "grad_norm": 0.11387702077627182,
      "learning_rate": 2.9097052764717196e-05,
      "loss": 0.0161,
      "num_input_tokens_seen": 66529792,
      "step": 30825
    },
    {
      "epoch": 5.029363784665579,
      "grad_norm": 2.490086078643799,
      "learning_rate": 2.9090030852093586e-05,
      "loss": 0.1166,
      "num_input_tokens_seen": 66540416,
      "step": 30830
    },
    {
      "epoch": 5.0301794453507345,
      "grad_norm": 10.41690444946289,
      "learning_rate": 2.9083008607911322e-05,
      "loss": 0.0685,
      "num_input_tokens_seen": 66549280,
      "step": 30835
    },
    {
      "epoch": 5.030995106035889,
      "grad_norm": 0.26020798087120056,
      "learning_rate": 2.9075986032739656e-05,
      "loss": 0.0096,
      "num_input_tokens_seen": 66560640,
      "step": 30840
    },
    {
      "epoch": 5.031810766721044,
      "grad_norm": 0.4090158939361572,
      "learning_rate": 2.9068963127147886e-05,
      "loss": 0.1129,
      "num_input_tokens_seen": 66571808,
      "step": 30845
    },
    {
      "epoch": 5.032626427406199,
      "grad_norm": 11.681133270263672,
      "learning_rate": 2.906193989170532e-05,
      "loss": 0.2865,
      "num_input_tokens_seen": 66583424,
      "step": 30850
    },
    {
      "epoch": 5.033442088091354,
      "grad_norm": 23.934995651245117,
      "learning_rate": 2.9054916326981297e-05,
      "loss": 0.2309,
      "num_input_tokens_seen": 66593984,
      "step": 30855
    },
    {
      "epoch": 5.034257748776509,
      "grad_norm": 0.1975080817937851,
      "learning_rate": 2.9047892433545176e-05,
      "loss": 0.0191,
      "num_input_tokens_seen": 66604704,
      "step": 30860
    },
    {
      "epoch": 5.035073409461664,
      "grad_norm": 3.619813919067383,
      "learning_rate": 2.9040868211966364e-05,
      "loss": 0.2261,
      "num_input_tokens_seen": 66615808,
      "step": 30865
    },
    {
      "epoch": 5.035889070146819,
      "grad_norm": 0.3861522078514099,
      "learning_rate": 2.903384366281427e-05,
      "loss": 0.008,
      "num_input_tokens_seen": 66627008,
      "step": 30870
    },
    {
      "epoch": 5.036704730831974,
      "grad_norm": 0.08608371764421463,
      "learning_rate": 2.902681878665834e-05,
      "loss": 0.088,
      "num_input_tokens_seen": 66638464,
      "step": 30875
    },
    {
      "epoch": 5.037520391517129,
      "grad_norm": 0.2712276577949524,
      "learning_rate": 2.9019793584068046e-05,
      "loss": 0.1403,
      "num_input_tokens_seen": 66648544,
      "step": 30880
    },
    {
      "epoch": 5.0383360522022835,
      "grad_norm": 0.07102425396442413,
      "learning_rate": 2.9012768055612887e-05,
      "loss": 0.1543,
      "num_input_tokens_seen": 66660384,
      "step": 30885
    },
    {
      "epoch": 5.039151712887439,
      "grad_norm": 0.1890052855014801,
      "learning_rate": 2.9005742201862385e-05,
      "loss": 0.1326,
      "num_input_tokens_seen": 66671136,
      "step": 30890
    },
    {
      "epoch": 5.039967373572594,
      "grad_norm": 2.965296506881714,
      "learning_rate": 2.8998716023386096e-05,
      "loss": 0.201,
      "num_input_tokens_seen": 66681664,
      "step": 30895
    },
    {
      "epoch": 5.040783034257749,
      "grad_norm": 0.07390269637107849,
      "learning_rate": 2.8991689520753605e-05,
      "loss": 0.0141,
      "num_input_tokens_seen": 66691968,
      "step": 30900
    },
    {
      "epoch": 5.041598694942904,
      "grad_norm": 2.248039722442627,
      "learning_rate": 2.8984662694534504e-05,
      "loss": 0.0174,
      "num_input_tokens_seen": 66703488,
      "step": 30905
    },
    {
      "epoch": 5.0424143556280585,
      "grad_norm": 0.05766557902097702,
      "learning_rate": 2.897763554529842e-05,
      "loss": 0.0839,
      "num_input_tokens_seen": 66714304,
      "step": 30910
    },
    {
      "epoch": 5.043230016313213,
      "grad_norm": 2.0388343334198,
      "learning_rate": 2.8970608073615026e-05,
      "loss": 0.1338,
      "num_input_tokens_seen": 66724864,
      "step": 30915
    },
    {
      "epoch": 5.044045676998369,
      "grad_norm": 0.09663654118776321,
      "learning_rate": 2.8963580280053992e-05,
      "loss": 0.1101,
      "num_input_tokens_seen": 66736736,
      "step": 30920
    },
    {
      "epoch": 5.044861337683524,
      "grad_norm": 0.6017317771911621,
      "learning_rate": 2.8956552165185023e-05,
      "loss": 0.1653,
      "num_input_tokens_seen": 66748128,
      "step": 30925
    },
    {
      "epoch": 5.045676998368679,
      "grad_norm": 0.11222974956035614,
      "learning_rate": 2.894952372957787e-05,
      "loss": 0.0072,
      "num_input_tokens_seen": 66759200,
      "step": 30930
    },
    {
      "epoch": 5.0464926590538335,
      "grad_norm": 0.08082642406225204,
      "learning_rate": 2.894249497380228e-05,
      "loss": 0.1505,
      "num_input_tokens_seen": 66770720,
      "step": 30935
    },
    {
      "epoch": 5.047308319738988,
      "grad_norm": 16.161096572875977,
      "learning_rate": 2.893546589842805e-05,
      "loss": 0.1574,
      "num_input_tokens_seen": 66781312,
      "step": 30940
    },
    {
      "epoch": 5.048123980424143,
      "grad_norm": 0.13353952765464783,
      "learning_rate": 2.892843650402497e-05,
      "loss": 0.0102,
      "num_input_tokens_seen": 66792864,
      "step": 30945
    },
    {
      "epoch": 5.048939641109299,
      "grad_norm": 8.055357933044434,
      "learning_rate": 2.8921406791162902e-05,
      "loss": 0.2173,
      "num_input_tokens_seen": 66802976,
      "step": 30950
    },
    {
      "epoch": 5.049755301794454,
      "grad_norm": 0.23622195422649384,
      "learning_rate": 2.891437676041171e-05,
      "loss": 0.0082,
      "num_input_tokens_seen": 66812768,
      "step": 30955
    },
    {
      "epoch": 5.0505709624796085,
      "grad_norm": 0.10315897315740585,
      "learning_rate": 2.890734641234127e-05,
      "loss": 0.0993,
      "num_input_tokens_seen": 66823072,
      "step": 30960
    },
    {
      "epoch": 5.051386623164763,
      "grad_norm": 0.9988123178482056,
      "learning_rate": 2.8900315747521507e-05,
      "loss": 0.0113,
      "num_input_tokens_seen": 66835200,
      "step": 30965
    },
    {
      "epoch": 5.052202283849918,
      "grad_norm": 0.21393397450447083,
      "learning_rate": 2.8893284766522353e-05,
      "loss": 0.009,
      "num_input_tokens_seen": 66846528,
      "step": 30970
    },
    {
      "epoch": 5.053017944535074,
      "grad_norm": 1.640389084815979,
      "learning_rate": 2.8886253469913787e-05,
      "loss": 0.1719,
      "num_input_tokens_seen": 66858080,
      "step": 30975
    },
    {
      "epoch": 5.053833605220229,
      "grad_norm": 0.2589954435825348,
      "learning_rate": 2.8879221858265794e-05,
      "loss": 0.0182,
      "num_input_tokens_seen": 66869888,
      "step": 30980
    },
    {
      "epoch": 5.054649265905383,
      "grad_norm": 18.620851516723633,
      "learning_rate": 2.8872189932148392e-05,
      "loss": 0.2544,
      "num_input_tokens_seen": 66880256,
      "step": 30985
    },
    {
      "epoch": 5.055464926590538,
      "grad_norm": 0.7262380719184875,
      "learning_rate": 2.8865157692131633e-05,
      "loss": 0.1503,
      "num_input_tokens_seen": 66890784,
      "step": 30990
    },
    {
      "epoch": 5.056280587275693,
      "grad_norm": 0.15699176490306854,
      "learning_rate": 2.8858125138785568e-05,
      "loss": 0.0112,
      "num_input_tokens_seen": 66901920,
      "step": 30995
    },
    {
      "epoch": 5.057096247960848,
      "grad_norm": 0.6760200262069702,
      "learning_rate": 2.8851092272680313e-05,
      "loss": 0.1433,
      "num_input_tokens_seen": 66912608,
      "step": 31000
    },
    {
      "epoch": 5.057911908646004,
      "grad_norm": 0.12168726325035095,
      "learning_rate": 2.8844059094385977e-05,
      "loss": 0.1217,
      "num_input_tokens_seen": 66922336,
      "step": 31005
    },
    {
      "epoch": 5.058727569331158,
      "grad_norm": 0.2774670422077179,
      "learning_rate": 2.883702560447271e-05,
      "loss": 0.1797,
      "num_input_tokens_seen": 66932960,
      "step": 31010
    },
    {
      "epoch": 5.059543230016313,
      "grad_norm": 0.12740923464298248,
      "learning_rate": 2.8829991803510675e-05,
      "loss": 0.1272,
      "num_input_tokens_seen": 66944448,
      "step": 31015
    },
    {
      "epoch": 5.060358890701468,
      "grad_norm": 0.5450809001922607,
      "learning_rate": 2.8822957692070073e-05,
      "loss": 0.0596,
      "num_input_tokens_seen": 66955264,
      "step": 31020
    },
    {
      "epoch": 5.061174551386623,
      "grad_norm": 0.08875981718301773,
      "learning_rate": 2.8815923270721124e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 66965024,
      "step": 31025
    },
    {
      "epoch": 5.061990212071779,
      "grad_norm": 0.08391879498958588,
      "learning_rate": 2.8808888540034067e-05,
      "loss": 0.0838,
      "num_input_tokens_seen": 66975072,
      "step": 31030
    },
    {
      "epoch": 5.062805872756933,
      "grad_norm": 0.03846724331378937,
      "learning_rate": 2.8801853500579183e-05,
      "loss": 0.1336,
      "num_input_tokens_seen": 66987232,
      "step": 31035
    },
    {
      "epoch": 5.063621533442088,
      "grad_norm": 2.5398213863372803,
      "learning_rate": 2.879481815292676e-05,
      "loss": 0.2049,
      "num_input_tokens_seen": 66995968,
      "step": 31040
    },
    {
      "epoch": 5.064437194127243,
      "grad_norm": 4.764420032501221,
      "learning_rate": 2.878778249764713e-05,
      "loss": 0.2104,
      "num_input_tokens_seen": 67006912,
      "step": 31045
    },
    {
      "epoch": 5.065252854812398,
      "grad_norm": 0.1985485851764679,
      "learning_rate": 2.8780746535310623e-05,
      "loss": 0.0223,
      "num_input_tokens_seen": 67017824,
      "step": 31050
    },
    {
      "epoch": 5.066068515497553,
      "grad_norm": 0.08837421983480453,
      "learning_rate": 2.8773710266487623e-05,
      "loss": 0.008,
      "num_input_tokens_seen": 67028736,
      "step": 31055
    },
    {
      "epoch": 5.066884176182708,
      "grad_norm": 0.10470012575387955,
      "learning_rate": 2.8766673691748524e-05,
      "loss": 0.048,
      "num_input_tokens_seen": 67039936,
      "step": 31060
    },
    {
      "epoch": 5.067699836867863,
      "grad_norm": 0.11755790561437607,
      "learning_rate": 2.875963681166373e-05,
      "loss": 0.0116,
      "num_input_tokens_seen": 67051616,
      "step": 31065
    },
    {
      "epoch": 5.068515497553018,
      "grad_norm": 0.062179673463106155,
      "learning_rate": 2.8752599626803717e-05,
      "loss": 0.0514,
      "num_input_tokens_seen": 67062240,
      "step": 31070
    },
    {
      "epoch": 5.069331158238173,
      "grad_norm": 0.05381360277533531,
      "learning_rate": 2.874556213773893e-05,
      "loss": 0.0173,
      "num_input_tokens_seen": 67073408,
      "step": 31075
    },
    {
      "epoch": 5.070146818923328,
      "grad_norm": 0.10158298909664154,
      "learning_rate": 2.8738524345039876e-05,
      "loss": 0.1015,
      "num_input_tokens_seen": 67084896,
      "step": 31080
    },
    {
      "epoch": 5.0709624796084825,
      "grad_norm": 0.07138735055923462,
      "learning_rate": 2.8731486249277062e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 67095040,
      "step": 31085
    },
    {
      "epoch": 5.071778140293638,
      "grad_norm": 0.0807000994682312,
      "learning_rate": 2.8724447851021047e-05,
      "loss": 0.0992,
      "num_input_tokens_seen": 67106336,
      "step": 31090
    },
    {
      "epoch": 5.072593800978793,
      "grad_norm": 0.09665209800004959,
      "learning_rate": 2.871740915084239e-05,
      "loss": 0.2611,
      "num_input_tokens_seen": 67116832,
      "step": 31095
    },
    {
      "epoch": 5.073409461663948,
      "grad_norm": 11.138023376464844,
      "learning_rate": 2.871037014931168e-05,
      "loss": 0.1062,
      "num_input_tokens_seen": 67128736,
      "step": 31100
    },
    {
      "epoch": 5.074225122349103,
      "grad_norm": 3.8535048961639404,
      "learning_rate": 2.870333084699954e-05,
      "loss": 0.0221,
      "num_input_tokens_seen": 67140128,
      "step": 31105
    },
    {
      "epoch": 5.075040783034257,
      "grad_norm": 0.25655168294906616,
      "learning_rate": 2.8696291244476613e-05,
      "loss": 0.0776,
      "num_input_tokens_seen": 67151232,
      "step": 31110
    },
    {
      "epoch": 5.075856443719413,
      "grad_norm": 0.08060576021671295,
      "learning_rate": 2.8689251342313562e-05,
      "loss": 0.1069,
      "num_input_tokens_seen": 67161984,
      "step": 31115
    },
    {
      "epoch": 5.076672104404568,
      "grad_norm": 1.157850980758667,
      "learning_rate": 2.8682211141081084e-05,
      "loss": 0.1153,
      "num_input_tokens_seen": 67172224,
      "step": 31120
    },
    {
      "epoch": 5.077487765089723,
      "grad_norm": 0.04678581282496452,
      "learning_rate": 2.867517064134988e-05,
      "loss": 0.0092,
      "num_input_tokens_seen": 67182560,
      "step": 31125
    },
    {
      "epoch": 5.078303425774878,
      "grad_norm": 0.0874355211853981,
      "learning_rate": 2.8668129843690704e-05,
      "loss": 0.0052,
      "num_input_tokens_seen": 67193024,
      "step": 31130
    },
    {
      "epoch": 5.079119086460032,
      "grad_norm": 0.08548480272293091,
      "learning_rate": 2.86610887486743e-05,
      "loss": 0.2387,
      "num_input_tokens_seen": 67204096,
      "step": 31135
    },
    {
      "epoch": 5.079934747145187,
      "grad_norm": 0.10910390317440033,
      "learning_rate": 2.8654047356871473e-05,
      "loss": 0.0158,
      "num_input_tokens_seen": 67214432,
      "step": 31140
    },
    {
      "epoch": 5.080750407830343,
      "grad_norm": 0.32175755500793457,
      "learning_rate": 2.8647005668853027e-05,
      "loss": 0.0158,
      "num_input_tokens_seen": 67225792,
      "step": 31145
    },
    {
      "epoch": 5.081566068515498,
      "grad_norm": 1.7120707035064697,
      "learning_rate": 2.8639963685189795e-05,
      "loss": 0.2072,
      "num_input_tokens_seen": 67236128,
      "step": 31150
    },
    {
      "epoch": 5.082381729200653,
      "grad_norm": 0.13650250434875488,
      "learning_rate": 2.8632921406452635e-05,
      "loss": 0.0841,
      "num_input_tokens_seen": 67247072,
      "step": 31155
    },
    {
      "epoch": 5.083197389885807,
      "grad_norm": 0.14863349497318268,
      "learning_rate": 2.862587883321244e-05,
      "loss": 0.0944,
      "num_input_tokens_seen": 67257344,
      "step": 31160
    },
    {
      "epoch": 5.084013050570962,
      "grad_norm": 6.298161506652832,
      "learning_rate": 2.8618835966040104e-05,
      "loss": 0.0831,
      "num_input_tokens_seen": 67266848,
      "step": 31165
    },
    {
      "epoch": 5.084828711256117,
      "grad_norm": 0.17926152050495148,
      "learning_rate": 2.8611792805506565e-05,
      "loss": 0.2324,
      "num_input_tokens_seen": 67277056,
      "step": 31170
    },
    {
      "epoch": 5.085644371941273,
      "grad_norm": 0.08532644063234329,
      "learning_rate": 2.8604749352182776e-05,
      "loss": 0.065,
      "num_input_tokens_seen": 67288672,
      "step": 31175
    },
    {
      "epoch": 5.0864600326264275,
      "grad_norm": 0.06849771738052368,
      "learning_rate": 2.8597705606639707e-05,
      "loss": 0.0989,
      "num_input_tokens_seen": 67299424,
      "step": 31180
    },
    {
      "epoch": 5.087275693311582,
      "grad_norm": 0.16645334661006927,
      "learning_rate": 2.8590661569448372e-05,
      "loss": 0.013,
      "num_input_tokens_seen": 67309056,
      "step": 31185
    },
    {
      "epoch": 5.088091353996737,
      "grad_norm": 0.2268446981906891,
      "learning_rate": 2.8583617241179796e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 67320320,
      "step": 31190
    },
    {
      "epoch": 5.088907014681892,
      "grad_norm": 0.165228009223938,
      "learning_rate": 2.8576572622405017e-05,
      "loss": 0.2108,
      "num_input_tokens_seen": 67330944,
      "step": 31195
    },
    {
      "epoch": 5.089722675367048,
      "grad_norm": 0.0722666010260582,
      "learning_rate": 2.856952771369512e-05,
      "loss": 0.0045,
      "num_input_tokens_seen": 67342720,
      "step": 31200
    },
    {
      "epoch": 5.0905383360522025,
      "grad_norm": 4.162609100341797,
      "learning_rate": 2.856248251562119e-05,
      "loss": 0.1851,
      "num_input_tokens_seen": 67353856,
      "step": 31205
    },
    {
      "epoch": 5.091353996737357,
      "grad_norm": 0.02475403994321823,
      "learning_rate": 2.8555437028754352e-05,
      "loss": 0.2968,
      "num_input_tokens_seen": 67364384,
      "step": 31210
    },
    {
      "epoch": 5.092169657422512,
      "grad_norm": 0.10372800379991531,
      "learning_rate": 2.8548391253665746e-05,
      "loss": 0.2329,
      "num_input_tokens_seen": 67375328,
      "step": 31215
    },
    {
      "epoch": 5.092985318107667,
      "grad_norm": 0.17950288951396942,
      "learning_rate": 2.854134519092654e-05,
      "loss": 0.0115,
      "num_input_tokens_seen": 67386240,
      "step": 31220
    },
    {
      "epoch": 5.093800978792822,
      "grad_norm": 0.15762710571289062,
      "learning_rate": 2.8534298841107925e-05,
      "loss": 0.0208,
      "num_input_tokens_seen": 67397760,
      "step": 31225
    },
    {
      "epoch": 5.0946166394779775,
      "grad_norm": 0.045956000685691833,
      "learning_rate": 2.8527252204781117e-05,
      "loss": 0.0207,
      "num_input_tokens_seen": 67408768,
      "step": 31230
    },
    {
      "epoch": 5.095432300163132,
      "grad_norm": 7.062222480773926,
      "learning_rate": 2.852020528251735e-05,
      "loss": 0.141,
      "num_input_tokens_seen": 67419456,
      "step": 31235
    },
    {
      "epoch": 5.096247960848287,
      "grad_norm": 0.2240462750196457,
      "learning_rate": 2.8513158074887875e-05,
      "loss": 0.1079,
      "num_input_tokens_seen": 67430528,
      "step": 31240
    },
    {
      "epoch": 5.097063621533442,
      "grad_norm": 0.3925155997276306,
      "learning_rate": 2.8506110582463984e-05,
      "loss": 0.0875,
      "num_input_tokens_seen": 67441472,
      "step": 31245
    },
    {
      "epoch": 5.097879282218597,
      "grad_norm": 3.969595432281494,
      "learning_rate": 2.8499062805816984e-05,
      "loss": 0.249,
      "num_input_tokens_seen": 67451808,
      "step": 31250
    },
    {
      "epoch": 5.0986949429037525,
      "grad_norm": 0.21328844130039215,
      "learning_rate": 2.8492014745518192e-05,
      "loss": 0.0764,
      "num_input_tokens_seen": 67463104,
      "step": 31255
    },
    {
      "epoch": 5.099510603588907,
      "grad_norm": 0.07602519541978836,
      "learning_rate": 2.8484966402138968e-05,
      "loss": 0.0097,
      "num_input_tokens_seen": 67474144,
      "step": 31260
    },
    {
      "epoch": 5.100326264274062,
      "grad_norm": 0.20172946155071259,
      "learning_rate": 2.8477917776250683e-05,
      "loss": 0.0227,
      "num_input_tokens_seen": 67484096,
      "step": 31265
    },
    {
      "epoch": 5.101141924959217,
      "grad_norm": 0.03285258263349533,
      "learning_rate": 2.847086886842474e-05,
      "loss": 0.1259,
      "num_input_tokens_seen": 67493664,
      "step": 31270
    },
    {
      "epoch": 5.101957585644372,
      "grad_norm": 1.429862141609192,
      "learning_rate": 2.8463819679232555e-05,
      "loss": 0.3046,
      "num_input_tokens_seen": 67504768,
      "step": 31275
    },
    {
      "epoch": 5.102773246329527,
      "grad_norm": 0.3224422335624695,
      "learning_rate": 2.845677020924557e-05,
      "loss": 0.0953,
      "num_input_tokens_seen": 67514848,
      "step": 31280
    },
    {
      "epoch": 5.103588907014682,
      "grad_norm": 0.0700070783495903,
      "learning_rate": 2.8449720459035256e-05,
      "loss": 0.0796,
      "num_input_tokens_seen": 67525248,
      "step": 31285
    },
    {
      "epoch": 5.104404567699837,
      "grad_norm": 2.38926100730896,
      "learning_rate": 2.8442670429173098e-05,
      "loss": 0.0528,
      "num_input_tokens_seen": 67534656,
      "step": 31290
    },
    {
      "epoch": 5.105220228384992,
      "grad_norm": 5.326434135437012,
      "learning_rate": 2.8435620120230595e-05,
      "loss": 0.2816,
      "num_input_tokens_seen": 67545024,
      "step": 31295
    },
    {
      "epoch": 5.106035889070147,
      "grad_norm": 6.8628644943237305,
      "learning_rate": 2.84285695327793e-05,
      "loss": 0.2522,
      "num_input_tokens_seen": 67556640,
      "step": 31300
    },
    {
      "epoch": 5.1068515497553015,
      "grad_norm": 0.13472425937652588,
      "learning_rate": 2.842151866739077e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 67568032,
      "step": 31305
    },
    {
      "epoch": 5.107667210440456,
      "grad_norm": 0.11969398707151413,
      "learning_rate": 2.8414467524636568e-05,
      "loss": 0.0087,
      "num_input_tokens_seen": 67578688,
      "step": 31310
    },
    {
      "epoch": 5.108482871125612,
      "grad_norm": 0.055297184735536575,
      "learning_rate": 2.8407416105088304e-05,
      "loss": 0.0137,
      "num_input_tokens_seen": 67589664,
      "step": 31315
    },
    {
      "epoch": 5.109298531810767,
      "grad_norm": 0.9463332295417786,
      "learning_rate": 2.8400364409317604e-05,
      "loss": 0.1658,
      "num_input_tokens_seen": 67600736,
      "step": 31320
    },
    {
      "epoch": 5.110114192495922,
      "grad_norm": 0.21723707020282745,
      "learning_rate": 2.8393312437896112e-05,
      "loss": 0.2344,
      "num_input_tokens_seen": 67612288,
      "step": 31325
    },
    {
      "epoch": 5.1109298531810765,
      "grad_norm": 3.053849220275879,
      "learning_rate": 2.8386260191395497e-05,
      "loss": 0.2375,
      "num_input_tokens_seen": 67622592,
      "step": 31330
    },
    {
      "epoch": 5.111745513866231,
      "grad_norm": 0.13577203452587128,
      "learning_rate": 2.837920767038745e-05,
      "loss": 0.1754,
      "num_input_tokens_seen": 67633568,
      "step": 31335
    },
    {
      "epoch": 5.112561174551387,
      "grad_norm": 0.09780284017324448,
      "learning_rate": 2.837215487544368e-05,
      "loss": 0.1429,
      "num_input_tokens_seen": 67644000,
      "step": 31340
    },
    {
      "epoch": 5.113376835236542,
      "grad_norm": 7.872146129608154,
      "learning_rate": 2.836510180713593e-05,
      "loss": 0.2021,
      "num_input_tokens_seen": 67654816,
      "step": 31345
    },
    {
      "epoch": 5.114192495921697,
      "grad_norm": 2.2904276847839355,
      "learning_rate": 2.835804846603595e-05,
      "loss": 0.3047,
      "num_input_tokens_seen": 67666304,
      "step": 31350
    },
    {
      "epoch": 5.1150081566068515,
      "grad_norm": 4.582032203674316,
      "learning_rate": 2.8350994852715522e-05,
      "loss": 0.1335,
      "num_input_tokens_seen": 67677536,
      "step": 31355
    },
    {
      "epoch": 5.115823817292006,
      "grad_norm": 0.11408593505620956,
      "learning_rate": 2.8343940967746453e-05,
      "loss": 0.1285,
      "num_input_tokens_seen": 67688608,
      "step": 31360
    },
    {
      "epoch": 5.116639477977161,
      "grad_norm": 19.40458106994629,
      "learning_rate": 2.8336886811700548e-05,
      "loss": 0.1566,
      "num_input_tokens_seen": 67699200,
      "step": 31365
    },
    {
      "epoch": 5.117455138662317,
      "grad_norm": 0.1368633657693863,
      "learning_rate": 2.8329832385149678e-05,
      "loss": 0.022,
      "num_input_tokens_seen": 67709568,
      "step": 31370
    },
    {
      "epoch": 5.118270799347472,
      "grad_norm": 2.977189540863037,
      "learning_rate": 2.8322777688665704e-05,
      "loss": 0.1955,
      "num_input_tokens_seen": 67719360,
      "step": 31375
    },
    {
      "epoch": 5.1190864600326265,
      "grad_norm": 0.13537751138210297,
      "learning_rate": 2.83157227228205e-05,
      "loss": 0.065,
      "num_input_tokens_seen": 67729472,
      "step": 31380
    },
    {
      "epoch": 5.119902120717781,
      "grad_norm": 5.512944221496582,
      "learning_rate": 2.830866748818599e-05,
      "loss": 0.0586,
      "num_input_tokens_seen": 67739424,
      "step": 31385
    },
    {
      "epoch": 5.120717781402936,
      "grad_norm": 13.24183177947998,
      "learning_rate": 2.8301611985334115e-05,
      "loss": 0.0685,
      "num_input_tokens_seen": 67750624,
      "step": 31390
    },
    {
      "epoch": 5.121533442088092,
      "grad_norm": 0.27740323543548584,
      "learning_rate": 2.8294556214836813e-05,
      "loss": 0.0966,
      "num_input_tokens_seen": 67760960,
      "step": 31395
    },
    {
      "epoch": 5.122349102773247,
      "grad_norm": 0.24310491979122162,
      "learning_rate": 2.828750017726607e-05,
      "loss": 0.0623,
      "num_input_tokens_seen": 67771232,
      "step": 31400
    },
    {
      "epoch": 5.123164763458401,
      "grad_norm": 3.1149415969848633,
      "learning_rate": 2.8280443873193884e-05,
      "loss": 0.0122,
      "num_input_tokens_seen": 67781408,
      "step": 31405
    },
    {
      "epoch": 5.123980424143556,
      "grad_norm": 0.09213629364967346,
      "learning_rate": 2.8273387303192266e-05,
      "loss": 0.1982,
      "num_input_tokens_seen": 67793472,
      "step": 31410
    },
    {
      "epoch": 5.124796084828711,
      "grad_norm": 0.2630845904350281,
      "learning_rate": 2.8266330467833274e-05,
      "loss": 0.0068,
      "num_input_tokens_seen": 67804704,
      "step": 31415
    },
    {
      "epoch": 5.125611745513866,
      "grad_norm": 0.10485062748193741,
      "learning_rate": 2.8259273367688954e-05,
      "loss": 0.1833,
      "num_input_tokens_seen": 67815840,
      "step": 31420
    },
    {
      "epoch": 5.126427406199022,
      "grad_norm": 0.11413134634494781,
      "learning_rate": 2.8252216003331395e-05,
      "loss": 0.0076,
      "num_input_tokens_seen": 67827008,
      "step": 31425
    },
    {
      "epoch": 5.127243066884176,
      "grad_norm": 2.9589438438415527,
      "learning_rate": 2.824515837533271e-05,
      "loss": 0.3522,
      "num_input_tokens_seen": 67837024,
      "step": 31430
    },
    {
      "epoch": 5.128058727569331,
      "grad_norm": 0.2477777749300003,
      "learning_rate": 2.823810048426502e-05,
      "loss": 0.0164,
      "num_input_tokens_seen": 67847360,
      "step": 31435
    },
    {
      "epoch": 5.128874388254486,
      "grad_norm": 0.05343763157725334,
      "learning_rate": 2.8231042330700473e-05,
      "loss": 0.341,
      "num_input_tokens_seen": 67857152,
      "step": 31440
    },
    {
      "epoch": 5.129690048939641,
      "grad_norm": 3.411064386367798,
      "learning_rate": 2.822398391521125e-05,
      "loss": 0.1293,
      "num_input_tokens_seen": 67867680,
      "step": 31445
    },
    {
      "epoch": 5.130505709624796,
      "grad_norm": 0.09728110581636429,
      "learning_rate": 2.8216925238369518e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 67877632,
      "step": 31450
    },
    {
      "epoch": 5.131321370309951,
      "grad_norm": 0.0923701673746109,
      "learning_rate": 2.820986630074751e-05,
      "loss": 0.1656,
      "num_input_tokens_seen": 67886912,
      "step": 31455
    },
    {
      "epoch": 5.132137030995106,
      "grad_norm": 0.14128021895885468,
      "learning_rate": 2.8202807102917457e-05,
      "loss": 0.0103,
      "num_input_tokens_seen": 67895040,
      "step": 31460
    },
    {
      "epoch": 5.132952691680261,
      "grad_norm": 0.20301920175552368,
      "learning_rate": 2.8195747645451605e-05,
      "loss": 0.0616,
      "num_input_tokens_seen": 67906336,
      "step": 31465
    },
    {
      "epoch": 5.133768352365416,
      "grad_norm": 0.07375723868608475,
      "learning_rate": 2.8188687928922237e-05,
      "loss": 0.1307,
      "num_input_tokens_seen": 67916992,
      "step": 31470
    },
    {
      "epoch": 5.134584013050571,
      "grad_norm": 0.12706264853477478,
      "learning_rate": 2.818162795390164e-05,
      "loss": 0.0992,
      "num_input_tokens_seen": 67928352,
      "step": 31475
    },
    {
      "epoch": 5.135399673735726,
      "grad_norm": 0.8841932415962219,
      "learning_rate": 2.817456772096214e-05,
      "loss": 0.2177,
      "num_input_tokens_seen": 67940160,
      "step": 31480
    },
    {
      "epoch": 5.136215334420881,
      "grad_norm": 1.2765356302261353,
      "learning_rate": 2.8167507230676077e-05,
      "loss": 0.0696,
      "num_input_tokens_seen": 67950304,
      "step": 31485
    },
    {
      "epoch": 5.137030995106036,
      "grad_norm": 0.15975184738636017,
      "learning_rate": 2.8160446483615804e-05,
      "loss": 0.1641,
      "num_input_tokens_seen": 67960224,
      "step": 31490
    },
    {
      "epoch": 5.137846655791191,
      "grad_norm": 2.555025577545166,
      "learning_rate": 2.8153385480353705e-05,
      "loss": 0.1185,
      "num_input_tokens_seen": 67971616,
      "step": 31495
    },
    {
      "epoch": 5.138662316476346,
      "grad_norm": 0.24455052614212036,
      "learning_rate": 2.814632422146218e-05,
      "loss": 0.1452,
      "num_input_tokens_seen": 67982048,
      "step": 31500
    },
    {
      "epoch": 5.1394779771615005,
      "grad_norm": 0.1774958372116089,
      "learning_rate": 2.8139262707513647e-05,
      "loss": 0.1026,
      "num_input_tokens_seen": 67991584,
      "step": 31505
    },
    {
      "epoch": 5.140293637846656,
      "grad_norm": 6.054966449737549,
      "learning_rate": 2.813220093908055e-05,
      "loss": 0.1188,
      "num_input_tokens_seen": 68002208,
      "step": 31510
    },
    {
      "epoch": 5.141109298531811,
      "grad_norm": 0.17508752644062042,
      "learning_rate": 2.812513891673535e-05,
      "loss": 0.0089,
      "num_input_tokens_seen": 68013344,
      "step": 31515
    },
    {
      "epoch": 5.141924959216966,
      "grad_norm": 1.0667266845703125,
      "learning_rate": 2.8118076641050535e-05,
      "loss": 0.1915,
      "num_input_tokens_seen": 68025056,
      "step": 31520
    },
    {
      "epoch": 5.142740619902121,
      "grad_norm": 0.3312191069126129,
      "learning_rate": 2.81110141125986e-05,
      "loss": 0.0102,
      "num_input_tokens_seen": 68035296,
      "step": 31525
    },
    {
      "epoch": 5.143556280587275,
      "grad_norm": 2.1736700534820557,
      "learning_rate": 2.8103951331952083e-05,
      "loss": 0.0901,
      "num_input_tokens_seen": 68046080,
      "step": 31530
    },
    {
      "epoch": 5.14437194127243,
      "grad_norm": 0.08169755339622498,
      "learning_rate": 2.8096888299683515e-05,
      "loss": 0.0885,
      "num_input_tokens_seen": 68057344,
      "step": 31535
    },
    {
      "epoch": 5.145187601957586,
      "grad_norm": 0.13582096993923187,
      "learning_rate": 2.8089825016365478e-05,
      "loss": 0.127,
      "num_input_tokens_seen": 68067488,
      "step": 31540
    },
    {
      "epoch": 5.146003262642741,
      "grad_norm": 14.365918159484863,
      "learning_rate": 2.808276148257054e-05,
      "loss": 0.151,
      "num_input_tokens_seen": 68079584,
      "step": 31545
    },
    {
      "epoch": 5.146818923327896,
      "grad_norm": 0.16848717629909515,
      "learning_rate": 2.807569769887132e-05,
      "loss": 0.2657,
      "num_input_tokens_seen": 68090816,
      "step": 31550
    },
    {
      "epoch": 5.14763458401305,
      "grad_norm": 0.10022691637277603,
      "learning_rate": 2.8068633665840438e-05,
      "loss": 0.0765,
      "num_input_tokens_seen": 68102592,
      "step": 31555
    },
    {
      "epoch": 5.148450244698205,
      "grad_norm": 0.2540022134780884,
      "learning_rate": 2.806156938405054e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 68113664,
      "step": 31560
    },
    {
      "epoch": 5.149265905383361,
      "grad_norm": 0.15770824253559113,
      "learning_rate": 2.8054504854074293e-05,
      "loss": 0.0684,
      "num_input_tokens_seen": 68124480,
      "step": 31565
    },
    {
      "epoch": 5.150081566068516,
      "grad_norm": 0.11198746412992477,
      "learning_rate": 2.8047440076484383e-05,
      "loss": 0.093,
      "num_input_tokens_seen": 68135424,
      "step": 31570
    },
    {
      "epoch": 5.150897226753671,
      "grad_norm": 0.7075061202049255,
      "learning_rate": 2.8040375051853522e-05,
      "loss": 0.1386,
      "num_input_tokens_seen": 68146784,
      "step": 31575
    },
    {
      "epoch": 5.151712887438825,
      "grad_norm": 0.45923352241516113,
      "learning_rate": 2.803330978075443e-05,
      "loss": 0.1452,
      "num_input_tokens_seen": 68155744,
      "step": 31580
    },
    {
      "epoch": 5.15252854812398,
      "grad_norm": 0.06245943903923035,
      "learning_rate": 2.802624426375985e-05,
      "loss": 0.0054,
      "num_input_tokens_seen": 68167168,
      "step": 31585
    },
    {
      "epoch": 5.153344208809135,
      "grad_norm": 3.847533702850342,
      "learning_rate": 2.801917850144256e-05,
      "loss": 0.1779,
      "num_input_tokens_seen": 68178496,
      "step": 31590
    },
    {
      "epoch": 5.154159869494291,
      "grad_norm": 0.1625690907239914,
      "learning_rate": 2.8012112494375342e-05,
      "loss": 0.0659,
      "num_input_tokens_seen": 68189568,
      "step": 31595
    },
    {
      "epoch": 5.1549755301794455,
      "grad_norm": 0.17873123288154602,
      "learning_rate": 2.8005046243131005e-05,
      "loss": 0.0522,
      "num_input_tokens_seen": 68198624,
      "step": 31600
    },
    {
      "epoch": 5.1557911908646,
      "grad_norm": 0.1790742725133896,
      "learning_rate": 2.7997979748282364e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 68210528,
      "step": 31605
    },
    {
      "epoch": 5.156606851549755,
      "grad_norm": 0.09914734214544296,
      "learning_rate": 2.7990913010402282e-05,
      "loss": 0.0718,
      "num_input_tokens_seen": 68220384,
      "step": 31610
    },
    {
      "epoch": 5.15742251223491,
      "grad_norm": 0.21887736022472382,
      "learning_rate": 2.798384603006361e-05,
      "loss": 0.1765,
      "num_input_tokens_seen": 68230464,
      "step": 31615
    },
    {
      "epoch": 5.158238172920065,
      "grad_norm": 0.1654205173254013,
      "learning_rate": 2.7976778807839245e-05,
      "loss": 0.1839,
      "num_input_tokens_seen": 68240384,
      "step": 31620
    },
    {
      "epoch": 5.1590538336052205,
      "grad_norm": 0.12954449653625488,
      "learning_rate": 2.796971134430208e-05,
      "loss": 0.0274,
      "num_input_tokens_seen": 68251840,
      "step": 31625
    },
    {
      "epoch": 5.159869494290375,
      "grad_norm": 1.3999621868133545,
      "learning_rate": 2.7962643640025044e-05,
      "loss": 0.1217,
      "num_input_tokens_seen": 68263008,
      "step": 31630
    },
    {
      "epoch": 5.16068515497553,
      "grad_norm": 0.18048496544361115,
      "learning_rate": 2.7955575695581083e-05,
      "loss": 0.0967,
      "num_input_tokens_seen": 68273920,
      "step": 31635
    },
    {
      "epoch": 5.161500815660685,
      "grad_norm": 1.6067979335784912,
      "learning_rate": 2.794850751154316e-05,
      "loss": 0.1178,
      "num_input_tokens_seen": 68284256,
      "step": 31640
    },
    {
      "epoch": 5.16231647634584,
      "grad_norm": 0.06456427276134491,
      "learning_rate": 2.794143908848426e-05,
      "loss": 0.1148,
      "num_input_tokens_seen": 68296000,
      "step": 31645
    },
    {
      "epoch": 5.1631321370309955,
      "grad_norm": 0.11345984786748886,
      "learning_rate": 2.7934370426977385e-05,
      "loss": 0.013,
      "num_input_tokens_seen": 68308224,
      "step": 31650
    },
    {
      "epoch": 5.16394779771615,
      "grad_norm": 0.10962020605802536,
      "learning_rate": 2.792730152759555e-05,
      "loss": 0.0314,
      "num_input_tokens_seen": 68318560,
      "step": 31655
    },
    {
      "epoch": 5.164763458401305,
      "grad_norm": 0.11335140466690063,
      "learning_rate": 2.7920232390911805e-05,
      "loss": 0.1019,
      "num_input_tokens_seen": 68328256,
      "step": 31660
    },
    {
      "epoch": 5.16557911908646,
      "grad_norm": 0.08123160898685455,
      "learning_rate": 2.79131630174992e-05,
      "loss": 0.0973,
      "num_input_tokens_seen": 68339648,
      "step": 31665
    },
    {
      "epoch": 5.166394779771615,
      "grad_norm": 12.271551132202148,
      "learning_rate": 2.790609340793082e-05,
      "loss": 0.1062,
      "num_input_tokens_seen": 68351072,
      "step": 31670
    },
    {
      "epoch": 5.16721044045677,
      "grad_norm": 10.418120384216309,
      "learning_rate": 2.789902356277977e-05,
      "loss": 0.054,
      "num_input_tokens_seen": 68360832,
      "step": 31675
    },
    {
      "epoch": 5.168026101141925,
      "grad_norm": 2.957167387008667,
      "learning_rate": 2.7891953482619148e-05,
      "loss": 0.137,
      "num_input_tokens_seen": 68370016,
      "step": 31680
    },
    {
      "epoch": 5.16884176182708,
      "grad_norm": 0.08972841501235962,
      "learning_rate": 2.788488316802211e-05,
      "loss": 0.064,
      "num_input_tokens_seen": 68380768,
      "step": 31685
    },
    {
      "epoch": 5.169657422512235,
      "grad_norm": 0.12127567827701569,
      "learning_rate": 2.78778126195618e-05,
      "loss": 0.0072,
      "num_input_tokens_seen": 68392448,
      "step": 31690
    },
    {
      "epoch": 5.17047308319739,
      "grad_norm": 0.649427592754364,
      "learning_rate": 2.7870741837811404e-05,
      "loss": 0.1453,
      "num_input_tokens_seen": 68402016,
      "step": 31695
    },
    {
      "epoch": 5.171288743882545,
      "grad_norm": 0.31435349583625793,
      "learning_rate": 2.7863670823344106e-05,
      "loss": 0.0956,
      "num_input_tokens_seen": 68412704,
      "step": 31700
    },
    {
      "epoch": 5.1721044045677,
      "grad_norm": 0.11511395126581192,
      "learning_rate": 2.7856599576733124e-05,
      "loss": 0.1296,
      "num_input_tokens_seen": 68424448,
      "step": 31705
    },
    {
      "epoch": 5.172920065252855,
      "grad_norm": 0.131520077586174,
      "learning_rate": 2.7849528098551682e-05,
      "loss": 0.212,
      "num_input_tokens_seen": 68435168,
      "step": 31710
    },
    {
      "epoch": 5.17373572593801,
      "grad_norm": 0.3217465579509735,
      "learning_rate": 2.7842456389373032e-05,
      "loss": 0.1175,
      "num_input_tokens_seen": 68445696,
      "step": 31715
    },
    {
      "epoch": 5.174551386623165,
      "grad_norm": 3.1008524894714355,
      "learning_rate": 2.783538444977045e-05,
      "loss": 0.0683,
      "num_input_tokens_seen": 68455936,
      "step": 31720
    },
    {
      "epoch": 5.1753670473083195,
      "grad_norm": 0.08708109706640244,
      "learning_rate": 2.7828312280317214e-05,
      "loss": 0.0753,
      "num_input_tokens_seen": 68465344,
      "step": 31725
    },
    {
      "epoch": 5.176182707993474,
      "grad_norm": 0.11733467131853104,
      "learning_rate": 2.782123988158664e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 68475936,
      "step": 31730
    },
    {
      "epoch": 5.17699836867863,
      "grad_norm": 0.33614835143089294,
      "learning_rate": 2.781416725415204e-05,
      "loss": 0.0123,
      "num_input_tokens_seen": 68486496,
      "step": 31735
    },
    {
      "epoch": 5.177814029363785,
      "grad_norm": 4.399515151977539,
      "learning_rate": 2.780709439858677e-05,
      "loss": 0.019,
      "num_input_tokens_seen": 68497408,
      "step": 31740
    },
    {
      "epoch": 5.17862969004894,
      "grad_norm": 3.38046932220459,
      "learning_rate": 2.7800021315464176e-05,
      "loss": 0.0222,
      "num_input_tokens_seen": 68506720,
      "step": 31745
    },
    {
      "epoch": 5.1794453507340945,
      "grad_norm": 3.1816532611846924,
      "learning_rate": 2.779294800535765e-05,
      "loss": 0.1885,
      "num_input_tokens_seen": 68516800,
      "step": 31750
    },
    {
      "epoch": 5.180261011419249,
      "grad_norm": 3.5422580242156982,
      "learning_rate": 2.778587446884059e-05,
      "loss": 0.0761,
      "num_input_tokens_seen": 68525120,
      "step": 31755
    },
    {
      "epoch": 5.181076672104404,
      "grad_norm": 0.10872621089220047,
      "learning_rate": 2.777880070648641e-05,
      "loss": 0.063,
      "num_input_tokens_seen": 68535872,
      "step": 31760
    },
    {
      "epoch": 5.18189233278956,
      "grad_norm": 0.17329590022563934,
      "learning_rate": 2.777172671886854e-05,
      "loss": 0.0395,
      "num_input_tokens_seen": 68545760,
      "step": 31765
    },
    {
      "epoch": 5.182707993474715,
      "grad_norm": 0.13874921202659607,
      "learning_rate": 2.776465250656044e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 68555552,
      "step": 31770
    },
    {
      "epoch": 5.1835236541598695,
      "grad_norm": 9.618099212646484,
      "learning_rate": 2.7757578070135588e-05,
      "loss": 0.0973,
      "num_input_tokens_seen": 68565760,
      "step": 31775
    },
    {
      "epoch": 5.184339314845024,
      "grad_norm": 0.17414389550685883,
      "learning_rate": 2.775050341016746e-05,
      "loss": 0.0045,
      "num_input_tokens_seen": 68576160,
      "step": 31780
    },
    {
      "epoch": 5.185154975530179,
      "grad_norm": 3.377487897872925,
      "learning_rate": 2.774342852722957e-05,
      "loss": 0.3322,
      "num_input_tokens_seen": 68587360,
      "step": 31785
    },
    {
      "epoch": 5.185970636215335,
      "grad_norm": 0.9131451845169067,
      "learning_rate": 2.7736353421895445e-05,
      "loss": 0.0215,
      "num_input_tokens_seen": 68598336,
      "step": 31790
    },
    {
      "epoch": 5.18678629690049,
      "grad_norm": 0.06972074508666992,
      "learning_rate": 2.772927809473862e-05,
      "loss": 0.1982,
      "num_input_tokens_seen": 68610560,
      "step": 31795
    },
    {
      "epoch": 5.1876019575856445,
      "grad_norm": 0.153375044465065,
      "learning_rate": 2.7722202546332676e-05,
      "loss": 0.0079,
      "num_input_tokens_seen": 68621472,
      "step": 31800
    },
    {
      "epoch": 5.188417618270799,
      "grad_norm": 3.2225635051727295,
      "learning_rate": 2.7715126777251177e-05,
      "loss": 0.2064,
      "num_input_tokens_seen": 68632064,
      "step": 31805
    },
    {
      "epoch": 5.189233278955954,
      "grad_norm": 0.3280313313007355,
      "learning_rate": 2.7708050788067724e-05,
      "loss": 0.112,
      "num_input_tokens_seen": 68642784,
      "step": 31810
    },
    {
      "epoch": 5.190048939641109,
      "grad_norm": 0.27965638041496277,
      "learning_rate": 2.7700974579355933e-05,
      "loss": 0.098,
      "num_input_tokens_seen": 68652448,
      "step": 31815
    },
    {
      "epoch": 5.190864600326265,
      "grad_norm": 3.0123000144958496,
      "learning_rate": 2.769389815168944e-05,
      "loss": 0.1138,
      "num_input_tokens_seen": 68662656,
      "step": 31820
    },
    {
      "epoch": 5.191680261011419,
      "grad_norm": 0.11257915198802948,
      "learning_rate": 2.7686821505641893e-05,
      "loss": 0.0073,
      "num_input_tokens_seen": 68673248,
      "step": 31825
    },
    {
      "epoch": 5.192495921696574,
      "grad_norm": 2.494020462036133,
      "learning_rate": 2.7679744641786963e-05,
      "loss": 0.0867,
      "num_input_tokens_seen": 68683776,
      "step": 31830
    },
    {
      "epoch": 5.193311582381729,
      "grad_norm": 0.20498842000961304,
      "learning_rate": 2.7672667560698328e-05,
      "loss": 0.0584,
      "num_input_tokens_seen": 68694112,
      "step": 31835
    },
    {
      "epoch": 5.194127243066884,
      "grad_norm": 4.805168628692627,
      "learning_rate": 2.7665590262949707e-05,
      "loss": 0.2257,
      "num_input_tokens_seen": 68704512,
      "step": 31840
    },
    {
      "epoch": 5.19494290375204,
      "grad_norm": 0.2649928033351898,
      "learning_rate": 2.7658512749114816e-05,
      "loss": 0.1219,
      "num_input_tokens_seen": 68715776,
      "step": 31845
    },
    {
      "epoch": 5.195758564437194,
      "grad_norm": 0.6095231771469116,
      "learning_rate": 2.7651435019767384e-05,
      "loss": 0.1598,
      "num_input_tokens_seen": 68725888,
      "step": 31850
    },
    {
      "epoch": 5.196574225122349,
      "grad_norm": 0.1108047366142273,
      "learning_rate": 2.764435707548118e-05,
      "loss": 0.1038,
      "num_input_tokens_seen": 68736864,
      "step": 31855
    },
    {
      "epoch": 5.197389885807504,
      "grad_norm": 0.11246475577354431,
      "learning_rate": 2.7637278916829977e-05,
      "loss": 0.0156,
      "num_input_tokens_seen": 68747264,
      "step": 31860
    },
    {
      "epoch": 5.198205546492659,
      "grad_norm": 0.062301430851221085,
      "learning_rate": 2.7630200544387562e-05,
      "loss": 0.2893,
      "num_input_tokens_seen": 68757792,
      "step": 31865
    },
    {
      "epoch": 5.199021207177814,
      "grad_norm": 0.531463623046875,
      "learning_rate": 2.762312195872775e-05,
      "loss": 0.1987,
      "num_input_tokens_seen": 68769440,
      "step": 31870
    },
    {
      "epoch": 5.199836867862969,
      "grad_norm": 2.7355659008026123,
      "learning_rate": 2.761604316042436e-05,
      "loss": 0.1167,
      "num_input_tokens_seen": 68780224,
      "step": 31875
    },
    {
      "epoch": 5.200652528548124,
      "grad_norm": 6.167346954345703,
      "learning_rate": 2.760896415005123e-05,
      "loss": 0.0425,
      "num_input_tokens_seen": 68791008,
      "step": 31880
    },
    {
      "epoch": 5.201468189233279,
      "grad_norm": 0.10047260671854019,
      "learning_rate": 2.7601884928182238e-05,
      "loss": 0.0165,
      "num_input_tokens_seen": 68801088,
      "step": 31885
    },
    {
      "epoch": 5.202283849918434,
      "grad_norm": 0.17573612928390503,
      "learning_rate": 2.759480549539125e-05,
      "loss": 0.0088,
      "num_input_tokens_seen": 68812128,
      "step": 31890
    },
    {
      "epoch": 5.203099510603589,
      "grad_norm": 2.3614935874938965,
      "learning_rate": 2.758772585225216e-05,
      "loss": 0.026,
      "num_input_tokens_seen": 68822688,
      "step": 31895
    },
    {
      "epoch": 5.2039151712887435,
      "grad_norm": 30.16387939453125,
      "learning_rate": 2.7580645999338885e-05,
      "loss": 0.2327,
      "num_input_tokens_seen": 68832896,
      "step": 31900
    },
    {
      "epoch": 5.204730831973899,
      "grad_norm": 0.14647795259952545,
      "learning_rate": 2.757356593722534e-05,
      "loss": 0.0248,
      "num_input_tokens_seen": 68844224,
      "step": 31905
    },
    {
      "epoch": 5.205546492659054,
      "grad_norm": 0.11921296268701553,
      "learning_rate": 2.7566485666485496e-05,
      "loss": 0.1069,
      "num_input_tokens_seen": 68856512,
      "step": 31910
    },
    {
      "epoch": 5.206362153344209,
      "grad_norm": 0.13035118579864502,
      "learning_rate": 2.75594051876933e-05,
      "loss": 0.0794,
      "num_input_tokens_seen": 68867552,
      "step": 31915
    },
    {
      "epoch": 5.207177814029364,
      "grad_norm": 0.6677807569503784,
      "learning_rate": 2.755232450142272e-05,
      "loss": 0.0437,
      "num_input_tokens_seen": 68877536,
      "step": 31920
    },
    {
      "epoch": 5.2079934747145185,
      "grad_norm": 5.915046215057373,
      "learning_rate": 2.754524360824778e-05,
      "loss": 0.3672,
      "num_input_tokens_seen": 68887360,
      "step": 31925
    },
    {
      "epoch": 5.208809135399674,
      "grad_norm": 0.9929408431053162,
      "learning_rate": 2.7538162508742472e-05,
      "loss": 0.1934,
      "num_input_tokens_seen": 68899328,
      "step": 31930
    },
    {
      "epoch": 5.209624796084829,
      "grad_norm": 0.0498029850423336,
      "learning_rate": 2.7531081203480834e-05,
      "loss": 0.023,
      "num_input_tokens_seen": 68909888,
      "step": 31935
    },
    {
      "epoch": 5.210440456769984,
      "grad_norm": 12.45026683807373,
      "learning_rate": 2.7523999693036916e-05,
      "loss": 0.189,
      "num_input_tokens_seen": 68920416,
      "step": 31940
    },
    {
      "epoch": 5.211256117455139,
      "grad_norm": 0.17802025377750397,
      "learning_rate": 2.7516917977984773e-05,
      "loss": 0.0187,
      "num_input_tokens_seen": 68931936,
      "step": 31945
    },
    {
      "epoch": 5.212071778140293,
      "grad_norm": 12.071422576904297,
      "learning_rate": 2.7509836058898487e-05,
      "loss": 0.2195,
      "num_input_tokens_seen": 68942912,
      "step": 31950
    },
    {
      "epoch": 5.212887438825448,
      "grad_norm": 0.10195551812648773,
      "learning_rate": 2.750275393635215e-05,
      "loss": 0.314,
      "num_input_tokens_seen": 68954400,
      "step": 31955
    },
    {
      "epoch": 5.213703099510604,
      "grad_norm": 4.505454063415527,
      "learning_rate": 2.7495671610919886e-05,
      "loss": 0.1015,
      "num_input_tokens_seen": 68964032,
      "step": 31960
    },
    {
      "epoch": 5.214518760195759,
      "grad_norm": 0.11572358757257462,
      "learning_rate": 2.748858908317582e-05,
      "loss": 0.0083,
      "num_input_tokens_seen": 68974944,
      "step": 31965
    },
    {
      "epoch": 5.215334420880914,
      "grad_norm": 0.1790105402469635,
      "learning_rate": 2.7481506353694092e-05,
      "loss": 0.021,
      "num_input_tokens_seen": 68985856,
      "step": 31970
    },
    {
      "epoch": 5.216150081566068,
      "grad_norm": 0.15279224514961243,
      "learning_rate": 2.7474423423048873e-05,
      "loss": 0.0088,
      "num_input_tokens_seen": 68996768,
      "step": 31975
    },
    {
      "epoch": 5.216965742251223,
      "grad_norm": 0.16828513145446777,
      "learning_rate": 2.746734029181433e-05,
      "loss": 0.0227,
      "num_input_tokens_seen": 69008064,
      "step": 31980
    },
    {
      "epoch": 5.217781402936378,
      "grad_norm": 0.07797053456306458,
      "learning_rate": 2.7460256960564668e-05,
      "loss": 0.1128,
      "num_input_tokens_seen": 69018240,
      "step": 31985
    },
    {
      "epoch": 5.218597063621534,
      "grad_norm": 0.19876162707805634,
      "learning_rate": 2.7453173429874096e-05,
      "loss": 0.1801,
      "num_input_tokens_seen": 69028128,
      "step": 31990
    },
    {
      "epoch": 5.219412724306689,
      "grad_norm": 0.5721792578697205,
      "learning_rate": 2.744608970031683e-05,
      "loss": 0.0687,
      "num_input_tokens_seen": 69038720,
      "step": 31995
    },
    {
      "epoch": 5.220228384991843,
      "grad_norm": 0.0984506830573082,
      "learning_rate": 2.7439005772467126e-05,
      "loss": 0.004,
      "num_input_tokens_seen": 69050368,
      "step": 32000
    },
    {
      "epoch": 5.221044045676998,
      "grad_norm": 0.04918161779642105,
      "learning_rate": 2.743192164689924e-05,
      "loss": 0.0043,
      "num_input_tokens_seen": 69060672,
      "step": 32005
    },
    {
      "epoch": 5.221859706362153,
      "grad_norm": 0.16197675466537476,
      "learning_rate": 2.742483732418744e-05,
      "loss": 0.2101,
      "num_input_tokens_seen": 69072480,
      "step": 32010
    },
    {
      "epoch": 5.222675367047309,
      "grad_norm": 0.11991266161203384,
      "learning_rate": 2.7417752804906027e-05,
      "loss": 0.0746,
      "num_input_tokens_seen": 69083232,
      "step": 32015
    },
    {
      "epoch": 5.2234910277324635,
      "grad_norm": 0.42736032605171204,
      "learning_rate": 2.7410668089629304e-05,
      "loss": 0.008,
      "num_input_tokens_seen": 69094912,
      "step": 32020
    },
    {
      "epoch": 5.224306688417618,
      "grad_norm": 0.07446371018886566,
      "learning_rate": 2.7403583178931597e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 69106432,
      "step": 32025
    },
    {
      "epoch": 5.225122349102773,
      "grad_norm": 7.186981201171875,
      "learning_rate": 2.7396498073387245e-05,
      "loss": 0.0339,
      "num_input_tokens_seen": 69117184,
      "step": 32030
    },
    {
      "epoch": 5.225938009787928,
      "grad_norm": 0.08522886037826538,
      "learning_rate": 2.7389412773570595e-05,
      "loss": 0.0258,
      "num_input_tokens_seen": 69127520,
      "step": 32035
    },
    {
      "epoch": 5.226753670473083,
      "grad_norm": 0.02839471399784088,
      "learning_rate": 2.738232728005602e-05,
      "loss": 0.0092,
      "num_input_tokens_seen": 69139104,
      "step": 32040
    },
    {
      "epoch": 5.2275693311582385,
      "grad_norm": 3.8303990364074707,
      "learning_rate": 2.737524159341791e-05,
      "loss": 0.114,
      "num_input_tokens_seen": 69149312,
      "step": 32045
    },
    {
      "epoch": 5.228384991843393,
      "grad_norm": 0.08804535120725632,
      "learning_rate": 2.7368155714230663e-05,
      "loss": 0.1289,
      "num_input_tokens_seen": 69160576,
      "step": 32050
    },
    {
      "epoch": 5.229200652528548,
      "grad_norm": 0.2264700084924698,
      "learning_rate": 2.7361069643068698e-05,
      "loss": 0.0056,
      "num_input_tokens_seen": 69171360,
      "step": 32055
    },
    {
      "epoch": 5.230016313213703,
      "grad_norm": 0.2846857011318207,
      "learning_rate": 2.7353983380506444e-05,
      "loss": 0.3607,
      "num_input_tokens_seen": 69183360,
      "step": 32060
    },
    {
      "epoch": 5.230831973898858,
      "grad_norm": 0.09883857518434525,
      "learning_rate": 2.734689692711836e-05,
      "loss": 0.1259,
      "num_input_tokens_seen": 69195392,
      "step": 32065
    },
    {
      "epoch": 5.231647634584013,
      "grad_norm": 0.09415154159069061,
      "learning_rate": 2.73398102834789e-05,
      "loss": 0.0037,
      "num_input_tokens_seen": 69206816,
      "step": 32070
    },
    {
      "epoch": 5.232463295269168,
      "grad_norm": 0.13084068894386292,
      "learning_rate": 2.7332723450162544e-05,
      "loss": 0.0041,
      "num_input_tokens_seen": 69217248,
      "step": 32075
    },
    {
      "epoch": 5.233278955954323,
      "grad_norm": 4.655344009399414,
      "learning_rate": 2.7325636427743788e-05,
      "loss": 0.2244,
      "num_input_tokens_seen": 69228480,
      "step": 32080
    },
    {
      "epoch": 5.234094616639478,
      "grad_norm": 10.137812614440918,
      "learning_rate": 2.731854921679715e-05,
      "loss": 0.2596,
      "num_input_tokens_seen": 69240832,
      "step": 32085
    },
    {
      "epoch": 5.234910277324633,
      "grad_norm": 0.17076519131660461,
      "learning_rate": 2.7311461817897143e-05,
      "loss": 0.0161,
      "num_input_tokens_seen": 69250976,
      "step": 32090
    },
    {
      "epoch": 5.235725938009788,
      "grad_norm": 0.07735753059387207,
      "learning_rate": 2.7304374231618318e-05,
      "loss": 0.0434,
      "num_input_tokens_seen": 69261184,
      "step": 32095
    },
    {
      "epoch": 5.236541598694943,
      "grad_norm": 0.6737622618675232,
      "learning_rate": 2.729728645853522e-05,
      "loss": 0.2213,
      "num_input_tokens_seen": 69273024,
      "step": 32100
    },
    {
      "epoch": 5.237357259380098,
      "grad_norm": 0.08265198767185211,
      "learning_rate": 2.729019849922243e-05,
      "loss": 0.0046,
      "num_input_tokens_seen": 69285280,
      "step": 32105
    },
    {
      "epoch": 5.238172920065253,
      "grad_norm": 0.08026156574487686,
      "learning_rate": 2.7283110354254526e-05,
      "loss": 0.0283,
      "num_input_tokens_seen": 69295680,
      "step": 32110
    },
    {
      "epoch": 5.238988580750408,
      "grad_norm": 2.73895001411438,
      "learning_rate": 2.727602202420611e-05,
      "loss": 0.1615,
      "num_input_tokens_seen": 69305888,
      "step": 32115
    },
    {
      "epoch": 5.239804241435563,
      "grad_norm": 3.35025954246521,
      "learning_rate": 2.7268933509651806e-05,
      "loss": 0.0164,
      "num_input_tokens_seen": 69317856,
      "step": 32120
    },
    {
      "epoch": 5.240619902120717,
      "grad_norm": 13.326385498046875,
      "learning_rate": 2.7261844811166236e-05,
      "loss": 0.2848,
      "num_input_tokens_seen": 69329088,
      "step": 32125
    },
    {
      "epoch": 5.241435562805873,
      "grad_norm": 2.5656604766845703,
      "learning_rate": 2.725475592932405e-05,
      "loss": 0.0907,
      "num_input_tokens_seen": 69340544,
      "step": 32130
    },
    {
      "epoch": 5.242251223491028,
      "grad_norm": 2.8082115650177,
      "learning_rate": 2.724766686469991e-05,
      "loss": 0.2859,
      "num_input_tokens_seen": 69351104,
      "step": 32135
    },
    {
      "epoch": 5.243066884176183,
      "grad_norm": 0.03363499417901039,
      "learning_rate": 2.724057761786849e-05,
      "loss": 0.1132,
      "num_input_tokens_seen": 69362848,
      "step": 32140
    },
    {
      "epoch": 5.2438825448613375,
      "grad_norm": 2.582084894180298,
      "learning_rate": 2.7233488189404478e-05,
      "loss": 0.1285,
      "num_input_tokens_seen": 69373504,
      "step": 32145
    },
    {
      "epoch": 5.244698205546492,
      "grad_norm": 0.06234988570213318,
      "learning_rate": 2.7226398579882573e-05,
      "loss": 0.008,
      "num_input_tokens_seen": 69384256,
      "step": 32150
    },
    {
      "epoch": 5.245513866231648,
      "grad_norm": 0.13269774615764618,
      "learning_rate": 2.7219308789877513e-05,
      "loss": 0.1043,
      "num_input_tokens_seen": 69394880,
      "step": 32155
    },
    {
      "epoch": 5.246329526916803,
      "grad_norm": 0.15573489665985107,
      "learning_rate": 2.7212218819964013e-05,
      "loss": 0.1152,
      "num_input_tokens_seen": 69405120,
      "step": 32160
    },
    {
      "epoch": 5.247145187601958,
      "grad_norm": 0.040256571024656296,
      "learning_rate": 2.720512867071684e-05,
      "loss": 0.0387,
      "num_input_tokens_seen": 69416640,
      "step": 32165
    },
    {
      "epoch": 5.2479608482871125,
      "grad_norm": 1.0483318567276,
      "learning_rate": 2.719803834271074e-05,
      "loss": 0.2663,
      "num_input_tokens_seen": 69426912,
      "step": 32170
    },
    {
      "epoch": 5.248776508972267,
      "grad_norm": 0.050562601536512375,
      "learning_rate": 2.7190947836520502e-05,
      "loss": 0.1559,
      "num_input_tokens_seen": 69436512,
      "step": 32175
    },
    {
      "epoch": 5.249592169657422,
      "grad_norm": 0.1760358065366745,
      "learning_rate": 2.718385715272092e-05,
      "loss": 0.4186,
      "num_input_tokens_seen": 69447424,
      "step": 32180
    },
    {
      "epoch": 5.250407830342578,
      "grad_norm": 6.044267654418945,
      "learning_rate": 2.7176766291886792e-05,
      "loss": 0.1837,
      "num_input_tokens_seen": 69459136,
      "step": 32185
    },
    {
      "epoch": 5.251223491027733,
      "grad_norm": 12.450900077819824,
      "learning_rate": 2.7169675254592947e-05,
      "loss": 0.1447,
      "num_input_tokens_seen": 69470048,
      "step": 32190
    },
    {
      "epoch": 5.2520391517128875,
      "grad_norm": 5.102044582366943,
      "learning_rate": 2.716258404141421e-05,
      "loss": 0.0608,
      "num_input_tokens_seen": 69480992,
      "step": 32195
    },
    {
      "epoch": 5.252854812398042,
      "grad_norm": 21.830202102661133,
      "learning_rate": 2.7155492652925446e-05,
      "loss": 0.0556,
      "num_input_tokens_seen": 69491936,
      "step": 32200
    },
    {
      "epoch": 5.253670473083197,
      "grad_norm": 2.6970579624176025,
      "learning_rate": 2.714840108970151e-05,
      "loss": 0.2538,
      "num_input_tokens_seen": 69500896,
      "step": 32205
    },
    {
      "epoch": 5.254486133768353,
      "grad_norm": 8.061768531799316,
      "learning_rate": 2.7141309352317278e-05,
      "loss": 0.0383,
      "num_input_tokens_seen": 69509888,
      "step": 32210
    },
    {
      "epoch": 5.255301794453508,
      "grad_norm": 0.07931341230869293,
      "learning_rate": 2.7134217441347647e-05,
      "loss": 0.0095,
      "num_input_tokens_seen": 69519840,
      "step": 32215
    },
    {
      "epoch": 5.2561174551386625,
      "grad_norm": 11.895001411437988,
      "learning_rate": 2.7127125357367515e-05,
      "loss": 0.0714,
      "num_input_tokens_seen": 69530400,
      "step": 32220
    },
    {
      "epoch": 5.256933115823817,
      "grad_norm": 0.06481499969959259,
      "learning_rate": 2.7120033100951814e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 69540992,
      "step": 32225
    },
    {
      "epoch": 5.257748776508972,
      "grad_norm": 0.3081865608692169,
      "learning_rate": 2.7112940672675473e-05,
      "loss": 0.0856,
      "num_input_tokens_seen": 69552704,
      "step": 32230
    },
    {
      "epoch": 5.258564437194127,
      "grad_norm": 1.779091477394104,
      "learning_rate": 2.7105848073113433e-05,
      "loss": 0.0184,
      "num_input_tokens_seen": 69563648,
      "step": 32235
    },
    {
      "epoch": 5.259380097879283,
      "grad_norm": 3.2973453998565674,
      "learning_rate": 2.709875530284067e-05,
      "loss": 0.0693,
      "num_input_tokens_seen": 69574880,
      "step": 32240
    },
    {
      "epoch": 5.260195758564437,
      "grad_norm": 6.793567180633545,
      "learning_rate": 2.7091662362432153e-05,
      "loss": 0.2804,
      "num_input_tokens_seen": 69586496,
      "step": 32245
    },
    {
      "epoch": 5.261011419249592,
      "grad_norm": 1.9933561086654663,
      "learning_rate": 2.7084569252462873e-05,
      "loss": 0.0954,
      "num_input_tokens_seen": 69596704,
      "step": 32250
    },
    {
      "epoch": 5.261827079934747,
      "grad_norm": 4.946976661682129,
      "learning_rate": 2.7077475973507832e-05,
      "loss": 0.235,
      "num_input_tokens_seen": 69606912,
      "step": 32255
    },
    {
      "epoch": 5.262642740619902,
      "grad_norm": 6.156367778778076,
      "learning_rate": 2.7070382526142045e-05,
      "loss": 0.3343,
      "num_input_tokens_seen": 69618880,
      "step": 32260
    },
    {
      "epoch": 5.263458401305057,
      "grad_norm": 2.610217809677124,
      "learning_rate": 2.706328891094055e-05,
      "loss": 0.075,
      "num_input_tokens_seen": 69629440,
      "step": 32265
    },
    {
      "epoch": 5.264274061990212,
      "grad_norm": 0.1646733283996582,
      "learning_rate": 2.7056195128478384e-05,
      "loss": 0.0419,
      "num_input_tokens_seen": 69640128,
      "step": 32270
    },
    {
      "epoch": 5.265089722675367,
      "grad_norm": 1.4181983470916748,
      "learning_rate": 2.7049101179330605e-05,
      "loss": 0.0451,
      "num_input_tokens_seen": 69650816,
      "step": 32275
    },
    {
      "epoch": 5.265905383360522,
      "grad_norm": 0.19968171417713165,
      "learning_rate": 2.7042007064072288e-05,
      "loss": 0.1225,
      "num_input_tokens_seen": 69661376,
      "step": 32280
    },
    {
      "epoch": 5.266721044045677,
      "grad_norm": 3.8542120456695557,
      "learning_rate": 2.703491278327852e-05,
      "loss": 0.0796,
      "num_input_tokens_seen": 69671200,
      "step": 32285
    },
    {
      "epoch": 5.267536704730832,
      "grad_norm": 2.813113212585449,
      "learning_rate": 2.7027818337524396e-05,
      "loss": 0.182,
      "num_input_tokens_seen": 69682048,
      "step": 32290
    },
    {
      "epoch": 5.268352365415987,
      "grad_norm": 0.17513194680213928,
      "learning_rate": 2.7020723727385027e-05,
      "loss": 0.0331,
      "num_input_tokens_seen": 69694208,
      "step": 32295
    },
    {
      "epoch": 5.269168026101142,
      "grad_norm": 0.10825838148593903,
      "learning_rate": 2.7013628953435544e-05,
      "loss": 0.0772,
      "num_input_tokens_seen": 69702848,
      "step": 32300
    },
    {
      "epoch": 5.269983686786297,
      "grad_norm": 0.9168631434440613,
      "learning_rate": 2.7006534016251072e-05,
      "loss": 0.0105,
      "num_input_tokens_seen": 69714080,
      "step": 32305
    },
    {
      "epoch": 5.270799347471452,
      "grad_norm": 0.14680083096027374,
      "learning_rate": 2.6999438916406777e-05,
      "loss": 0.0493,
      "num_input_tokens_seen": 69725408,
      "step": 32310
    },
    {
      "epoch": 5.271615008156607,
      "grad_norm": 0.29254454374313354,
      "learning_rate": 2.6992343654477825e-05,
      "loss": 0.008,
      "num_input_tokens_seen": 69736704,
      "step": 32315
    },
    {
      "epoch": 5.2724306688417615,
      "grad_norm": 0.31529340147972107,
      "learning_rate": 2.6985248231039378e-05,
      "loss": 0.0111,
      "num_input_tokens_seen": 69747264,
      "step": 32320
    },
    {
      "epoch": 5.273246329526917,
      "grad_norm": 0.7765026688575745,
      "learning_rate": 2.6978152646666644e-05,
      "loss": 0.0907,
      "num_input_tokens_seen": 69758592,
      "step": 32325
    },
    {
      "epoch": 5.274061990212072,
      "grad_norm": 0.2740999460220337,
      "learning_rate": 2.697105690193481e-05,
      "loss": 0.0996,
      "num_input_tokens_seen": 69769088,
      "step": 32330
    },
    {
      "epoch": 5.274877650897227,
      "grad_norm": 0.20411568880081177,
      "learning_rate": 2.696396099741911e-05,
      "loss": 0.0103,
      "num_input_tokens_seen": 69780064,
      "step": 32335
    },
    {
      "epoch": 5.275693311582382,
      "grad_norm": 0.0828116163611412,
      "learning_rate": 2.695686493369476e-05,
      "loss": 0.1759,
      "num_input_tokens_seen": 69792064,
      "step": 32340
    },
    {
      "epoch": 5.2765089722675365,
      "grad_norm": 0.8827752470970154,
      "learning_rate": 2.6949768711337015e-05,
      "loss": 0.1659,
      "num_input_tokens_seen": 69802304,
      "step": 32345
    },
    {
      "epoch": 5.277324632952691,
      "grad_norm": 0.10121121257543564,
      "learning_rate": 2.6942672330921124e-05,
      "loss": 0.0183,
      "num_input_tokens_seen": 69811104,
      "step": 32350
    },
    {
      "epoch": 5.278140293637847,
      "grad_norm": 0.07794969528913498,
      "learning_rate": 2.693557579302236e-05,
      "loss": 0.0108,
      "num_input_tokens_seen": 69823072,
      "step": 32355
    },
    {
      "epoch": 5.278955954323002,
      "grad_norm": 1.5654137134552002,
      "learning_rate": 2.6928479098216e-05,
      "loss": 0.076,
      "num_input_tokens_seen": 69835104,
      "step": 32360
    },
    {
      "epoch": 5.279771615008157,
      "grad_norm": 0.2712656557559967,
      "learning_rate": 2.6921382247077336e-05,
      "loss": 0.0235,
      "num_input_tokens_seen": 69846784,
      "step": 32365
    },
    {
      "epoch": 5.280587275693311,
      "grad_norm": 0.18609561026096344,
      "learning_rate": 2.691428524018168e-05,
      "loss": 0.0113,
      "num_input_tokens_seen": 69857024,
      "step": 32370
    },
    {
      "epoch": 5.281402936378466,
      "grad_norm": 0.08994683623313904,
      "learning_rate": 2.6907188078104352e-05,
      "loss": 0.0045,
      "num_input_tokens_seen": 69866624,
      "step": 32375
    },
    {
      "epoch": 5.282218597063622,
      "grad_norm": 0.39006930589675903,
      "learning_rate": 2.690009076142067e-05,
      "loss": 0.0389,
      "num_input_tokens_seen": 69878816,
      "step": 32380
    },
    {
      "epoch": 5.283034257748777,
      "grad_norm": 0.05270082503557205,
      "learning_rate": 2.6892993290706e-05,
      "loss": 0.0967,
      "num_input_tokens_seen": 69889184,
      "step": 32385
    },
    {
      "epoch": 5.283849918433932,
      "grad_norm": 0.22735679149627686,
      "learning_rate": 2.6885895666535684e-05,
      "loss": 0.1762,
      "num_input_tokens_seen": 69900160,
      "step": 32390
    },
    {
      "epoch": 5.284665579119086,
      "grad_norm": 0.024947430938482285,
      "learning_rate": 2.6878797889485096e-05,
      "loss": 0.1381,
      "num_input_tokens_seen": 69910848,
      "step": 32395
    },
    {
      "epoch": 5.285481239804241,
      "grad_norm": 0.04396042972803116,
      "learning_rate": 2.687169996012962e-05,
      "loss": 0.0053,
      "num_input_tokens_seen": 69921856,
      "step": 32400
    },
    {
      "epoch": 5.286296900489396,
      "grad_norm": 0.05487280711531639,
      "learning_rate": 2.6864601879044653e-05,
      "loss": 0.0675,
      "num_input_tokens_seen": 69932768,
      "step": 32405
    },
    {
      "epoch": 5.287112561174552,
      "grad_norm": 2.912031888961792,
      "learning_rate": 2.6857503646805593e-05,
      "loss": 0.1006,
      "num_input_tokens_seen": 69943488,
      "step": 32410
    },
    {
      "epoch": 5.287928221859707,
      "grad_norm": 9.795567512512207,
      "learning_rate": 2.6850405263987867e-05,
      "loss": 0.101,
      "num_input_tokens_seen": 69953312,
      "step": 32415
    },
    {
      "epoch": 5.288743882544861,
      "grad_norm": 4.187163352966309,
      "learning_rate": 2.6843306731166894e-05,
      "loss": 0.0863,
      "num_input_tokens_seen": 69963072,
      "step": 32420
    },
    {
      "epoch": 5.289559543230016,
      "grad_norm": 0.03776213154196739,
      "learning_rate": 2.6836208048918132e-05,
      "loss": 0.1004,
      "num_input_tokens_seen": 69973600,
      "step": 32425
    },
    {
      "epoch": 5.290375203915171,
      "grad_norm": 0.17811918258666992,
      "learning_rate": 2.682910921781702e-05,
      "loss": 0.0504,
      "num_input_tokens_seen": 69984384,
      "step": 32430
    },
    {
      "epoch": 5.291190864600326,
      "grad_norm": 0.16918852925300598,
      "learning_rate": 2.682201023843904e-05,
      "loss": 0.0149,
      "num_input_tokens_seen": 69995616,
      "step": 32435
    },
    {
      "epoch": 5.2920065252854815,
      "grad_norm": 2.7108592987060547,
      "learning_rate": 2.6814911111359665e-05,
      "loss": 0.0701,
      "num_input_tokens_seen": 70006432,
      "step": 32440
    },
    {
      "epoch": 5.292822185970636,
      "grad_norm": 3.337555408477783,
      "learning_rate": 2.6807811837154383e-05,
      "loss": 0.2409,
      "num_input_tokens_seen": 70017120,
      "step": 32445
    },
    {
      "epoch": 5.293637846655791,
      "grad_norm": 0.2774198651313782,
      "learning_rate": 2.6800712416398705e-05,
      "loss": 0.1345,
      "num_input_tokens_seen": 70027744,
      "step": 32450
    },
    {
      "epoch": 5.294453507340946,
      "grad_norm": 4.527461051940918,
      "learning_rate": 2.6793612849668138e-05,
      "loss": 0.0884,
      "num_input_tokens_seen": 70039360,
      "step": 32455
    },
    {
      "epoch": 5.295269168026101,
      "grad_norm": 0.035836637020111084,
      "learning_rate": 2.6786513137538216e-05,
      "loss": 0.249,
      "num_input_tokens_seen": 70049952,
      "step": 32460
    },
    {
      "epoch": 5.2960848287112565,
      "grad_norm": 0.10114829242229462,
      "learning_rate": 2.677941328058447e-05,
      "loss": 0.1256,
      "num_input_tokens_seen": 70059904,
      "step": 32465
    },
    {
      "epoch": 5.296900489396411,
      "grad_norm": 0.07702568918466568,
      "learning_rate": 2.677231327938246e-05,
      "loss": 0.1912,
      "num_input_tokens_seen": 70070464,
      "step": 32470
    },
    {
      "epoch": 5.297716150081566,
      "grad_norm": 1.3145924806594849,
      "learning_rate": 2.676521313450774e-05,
      "loss": 0.0132,
      "num_input_tokens_seen": 70081472,
      "step": 32475
    },
    {
      "epoch": 5.298531810766721,
      "grad_norm": 0.15544812381267548,
      "learning_rate": 2.6758112846535888e-05,
      "loss": 0.0579,
      "num_input_tokens_seen": 70091360,
      "step": 32480
    },
    {
      "epoch": 5.299347471451876,
      "grad_norm": 0.7376506328582764,
      "learning_rate": 2.6751012416042487e-05,
      "loss": 0.1002,
      "num_input_tokens_seen": 70101088,
      "step": 32485
    },
    {
      "epoch": 5.300163132137031,
      "grad_norm": 0.09198694676160812,
      "learning_rate": 2.674391184360313e-05,
      "loss": 0.039,
      "num_input_tokens_seen": 70112608,
      "step": 32490
    },
    {
      "epoch": 5.300978792822186,
      "grad_norm": 0.05273007974028587,
      "learning_rate": 2.6736811129793438e-05,
      "loss": 0.0071,
      "num_input_tokens_seen": 70124128,
      "step": 32495
    },
    {
      "epoch": 5.301794453507341,
      "grad_norm": 3.622174024581909,
      "learning_rate": 2.6729710275189024e-05,
      "loss": 0.2176,
      "num_input_tokens_seen": 70136736,
      "step": 32500
    },
    {
      "epoch": 5.302610114192496,
      "grad_norm": 0.08061046898365021,
      "learning_rate": 2.672260928036552e-05,
      "loss": 0.064,
      "num_input_tokens_seen": 70148768,
      "step": 32505
    },
    {
      "epoch": 5.303425774877651,
      "grad_norm": 3.023188352584839,
      "learning_rate": 2.671550814589856e-05,
      "loss": 0.1831,
      "num_input_tokens_seen": 70158592,
      "step": 32510
    },
    {
      "epoch": 5.304241435562806,
      "grad_norm": 0.07232240587472916,
      "learning_rate": 2.6708406872363813e-05,
      "loss": 0.0991,
      "num_input_tokens_seen": 70169856,
      "step": 32515
    },
    {
      "epoch": 5.30505709624796,
      "grad_norm": 0.1896428018808365,
      "learning_rate": 2.670130546033693e-05,
      "loss": 0.0068,
      "num_input_tokens_seen": 70179680,
      "step": 32520
    },
    {
      "epoch": 5.305872756933116,
      "grad_norm": 0.2053537368774414,
      "learning_rate": 2.6694203910393594e-05,
      "loss": 0.1245,
      "num_input_tokens_seen": 70191104,
      "step": 32525
    },
    {
      "epoch": 5.306688417618271,
      "grad_norm": 0.10377363860607147,
      "learning_rate": 2.66871022231095e-05,
      "loss": 0.1242,
      "num_input_tokens_seen": 70201728,
      "step": 32530
    },
    {
      "epoch": 5.307504078303426,
      "grad_norm": 0.12116867303848267,
      "learning_rate": 2.6680000399060327e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 70213088,
      "step": 32535
    },
    {
      "epoch": 5.308319738988581,
      "grad_norm": 0.14606209099292755,
      "learning_rate": 2.6672898438821808e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 70223648,
      "step": 32540
    },
    {
      "epoch": 5.309135399673735,
      "grad_norm": 0.9420191049575806,
      "learning_rate": 2.666579634296965e-05,
      "loss": 0.0239,
      "num_input_tokens_seen": 70234272,
      "step": 32545
    },
    {
      "epoch": 5.309951060358891,
      "grad_norm": 0.267930805683136,
      "learning_rate": 2.6658694112079586e-05,
      "loss": 0.1197,
      "num_input_tokens_seen": 70244800,
      "step": 32550
    },
    {
      "epoch": 5.310766721044046,
      "grad_norm": 0.0489763505756855,
      "learning_rate": 2.6651591746727363e-05,
      "loss": 0.1675,
      "num_input_tokens_seen": 70255360,
      "step": 32555
    },
    {
      "epoch": 5.311582381729201,
      "grad_norm": 0.06418640166521072,
      "learning_rate": 2.6644489247488735e-05,
      "loss": 0.022,
      "num_input_tokens_seen": 70265888,
      "step": 32560
    },
    {
      "epoch": 5.3123980424143555,
      "grad_norm": 0.09622830152511597,
      "learning_rate": 2.6637386614939464e-05,
      "loss": 0.1332,
      "num_input_tokens_seen": 70276544,
      "step": 32565
    },
    {
      "epoch": 5.31321370309951,
      "grad_norm": 0.08779608458280563,
      "learning_rate": 2.6630283849655326e-05,
      "loss": 0.0889,
      "num_input_tokens_seen": 70287904,
      "step": 32570
    },
    {
      "epoch": 5.314029363784665,
      "grad_norm": 0.14219705760478973,
      "learning_rate": 2.6623180952212106e-05,
      "loss": 0.0728,
      "num_input_tokens_seen": 70299872,
      "step": 32575
    },
    {
      "epoch": 5.314845024469821,
      "grad_norm": 0.05864930897951126,
      "learning_rate": 2.66160779231856e-05,
      "loss": 0.1205,
      "num_input_tokens_seen": 70310624,
      "step": 32580
    },
    {
      "epoch": 5.315660685154976,
      "grad_norm": 0.11006398499011993,
      "learning_rate": 2.660897476315162e-05,
      "loss": 0.0054,
      "num_input_tokens_seen": 70321312,
      "step": 32585
    },
    {
      "epoch": 5.3164763458401305,
      "grad_norm": 0.12669697403907776,
      "learning_rate": 2.6601871472685985e-05,
      "loss": 0.0715,
      "num_input_tokens_seen": 70332384,
      "step": 32590
    },
    {
      "epoch": 5.317292006525285,
      "grad_norm": 1.7560433149337769,
      "learning_rate": 2.659476805236451e-05,
      "loss": 0.0444,
      "num_input_tokens_seen": 70343232,
      "step": 32595
    },
    {
      "epoch": 5.31810766721044,
      "grad_norm": 3.5992536544799805,
      "learning_rate": 2.6587664502763054e-05,
      "loss": 0.4086,
      "num_input_tokens_seen": 70355104,
      "step": 32600
    },
    {
      "epoch": 5.318923327895595,
      "grad_norm": 3.6785888671875,
      "learning_rate": 2.6580560824457457e-05,
      "loss": 0.2076,
      "num_input_tokens_seen": 70366400,
      "step": 32605
    },
    {
      "epoch": 5.319738988580751,
      "grad_norm": 0.10741819441318512,
      "learning_rate": 2.657345701802358e-05,
      "loss": 0.1342,
      "num_input_tokens_seen": 70378400,
      "step": 32610
    },
    {
      "epoch": 5.3205546492659055,
      "grad_norm": 0.1747598648071289,
      "learning_rate": 2.6566353084037295e-05,
      "loss": 0.0091,
      "num_input_tokens_seen": 70389952,
      "step": 32615
    },
    {
      "epoch": 5.32137030995106,
      "grad_norm": 0.151988685131073,
      "learning_rate": 2.6559249023074474e-05,
      "loss": 0.0898,
      "num_input_tokens_seen": 70401408,
      "step": 32620
    },
    {
      "epoch": 5.322185970636215,
      "grad_norm": 0.061704110354185104,
      "learning_rate": 2.6552144835711017e-05,
      "loss": 0.1951,
      "num_input_tokens_seen": 70412640,
      "step": 32625
    },
    {
      "epoch": 5.32300163132137,
      "grad_norm": 0.3974476456642151,
      "learning_rate": 2.6545040522522828e-05,
      "loss": 0.0756,
      "num_input_tokens_seen": 70422592,
      "step": 32630
    },
    {
      "epoch": 5.323817292006526,
      "grad_norm": 0.20766876637935638,
      "learning_rate": 2.653793608408582e-05,
      "loss": 0.1215,
      "num_input_tokens_seen": 70433856,
      "step": 32635
    },
    {
      "epoch": 5.3246329526916805,
      "grad_norm": 9.215160369873047,
      "learning_rate": 2.6530831520975903e-05,
      "loss": 0.163,
      "num_input_tokens_seen": 70445376,
      "step": 32640
    },
    {
      "epoch": 5.325448613376835,
      "grad_norm": 4.210348129272461,
      "learning_rate": 2.652372683376902e-05,
      "loss": 0.2986,
      "num_input_tokens_seen": 70456384,
      "step": 32645
    },
    {
      "epoch": 5.32626427406199,
      "grad_norm": 0.09094883501529694,
      "learning_rate": 2.65166220230411e-05,
      "loss": 0.1897,
      "num_input_tokens_seen": 70467040,
      "step": 32650
    },
    {
      "epoch": 5.327079934747145,
      "grad_norm": 0.2546117901802063,
      "learning_rate": 2.650951708936811e-05,
      "loss": 0.1962,
      "num_input_tokens_seen": 70477248,
      "step": 32655
    },
    {
      "epoch": 5.327895595432301,
      "grad_norm": 0.08630216121673584,
      "learning_rate": 2.6502412033326e-05,
      "loss": 0.1757,
      "num_input_tokens_seen": 70487648,
      "step": 32660
    },
    {
      "epoch": 5.328711256117455,
      "grad_norm": 6.063177108764648,
      "learning_rate": 2.6495306855490754e-05,
      "loss": 0.1292,
      "num_input_tokens_seen": 70499296,
      "step": 32665
    },
    {
      "epoch": 5.32952691680261,
      "grad_norm": 2.3382294178009033,
      "learning_rate": 2.6488201556438346e-05,
      "loss": 0.088,
      "num_input_tokens_seen": 70509664,
      "step": 32670
    },
    {
      "epoch": 5.330342577487765,
      "grad_norm": 0.6118956804275513,
      "learning_rate": 2.648109613674477e-05,
      "loss": 0.017,
      "num_input_tokens_seen": 70520064,
      "step": 32675
    },
    {
      "epoch": 5.33115823817292,
      "grad_norm": 2.301363945007324,
      "learning_rate": 2.647399059698602e-05,
      "loss": 0.1289,
      "num_input_tokens_seen": 70531104,
      "step": 32680
    },
    {
      "epoch": 5.331973898858075,
      "grad_norm": 0.10155824571847916,
      "learning_rate": 2.646688493773812e-05,
      "loss": 0.0855,
      "num_input_tokens_seen": 70542528,
      "step": 32685
    },
    {
      "epoch": 5.33278955954323,
      "grad_norm": 2.3470518589019775,
      "learning_rate": 2.6459779159577077e-05,
      "loss": 0.0781,
      "num_input_tokens_seen": 70553216,
      "step": 32690
    },
    {
      "epoch": 5.333605220228385,
      "grad_norm": 3.9175291061401367,
      "learning_rate": 2.645267326307893e-05,
      "loss": 0.2545,
      "num_input_tokens_seen": 70565504,
      "step": 32695
    },
    {
      "epoch": 5.33442088091354,
      "grad_norm": 0.26578038930892944,
      "learning_rate": 2.6445567248819726e-05,
      "loss": 0.112,
      "num_input_tokens_seen": 70577376,
      "step": 32700
    },
    {
      "epoch": 5.335236541598695,
      "grad_norm": 0.4504992961883545,
      "learning_rate": 2.643846111737549e-05,
      "loss": 0.1135,
      "num_input_tokens_seen": 70588960,
      "step": 32705
    },
    {
      "epoch": 5.33605220228385,
      "grad_norm": 0.229661762714386,
      "learning_rate": 2.643135486932231e-05,
      "loss": 0.0112,
      "num_input_tokens_seen": 70600192,
      "step": 32710
    },
    {
      "epoch": 5.3368678629690045,
      "grad_norm": 0.12986430525779724,
      "learning_rate": 2.642424850523624e-05,
      "loss": 0.1258,
      "num_input_tokens_seen": 70610816,
      "step": 32715
    },
    {
      "epoch": 5.33768352365416,
      "grad_norm": 3.3342740535736084,
      "learning_rate": 2.641714202569336e-05,
      "loss": 0.3,
      "num_input_tokens_seen": 70621280,
      "step": 32720
    },
    {
      "epoch": 5.338499184339315,
      "grad_norm": 0.16055022180080414,
      "learning_rate": 2.6410035431269754e-05,
      "loss": 0.14,
      "num_input_tokens_seen": 70631968,
      "step": 32725
    },
    {
      "epoch": 5.33931484502447,
      "grad_norm": 0.05201416462659836,
      "learning_rate": 2.6402928722541524e-05,
      "loss": 0.0999,
      "num_input_tokens_seen": 70643296,
      "step": 32730
    },
    {
      "epoch": 5.340130505709625,
      "grad_norm": 0.1715608537197113,
      "learning_rate": 2.6395821900084772e-05,
      "loss": 0.0128,
      "num_input_tokens_seen": 70652576,
      "step": 32735
    },
    {
      "epoch": 5.3409461663947795,
      "grad_norm": 0.15612873435020447,
      "learning_rate": 2.638871496447562e-05,
      "loss": 0.0635,
      "num_input_tokens_seen": 70663424,
      "step": 32740
    },
    {
      "epoch": 5.341761827079935,
      "grad_norm": 3.142871856689453,
      "learning_rate": 2.638160791629018e-05,
      "loss": 0.0558,
      "num_input_tokens_seen": 70673536,
      "step": 32745
    },
    {
      "epoch": 5.34257748776509,
      "grad_norm": 1.802825689315796,
      "learning_rate": 2.6374500756104594e-05,
      "loss": 0.0178,
      "num_input_tokens_seen": 70683712,
      "step": 32750
    },
    {
      "epoch": 5.343393148450245,
      "grad_norm": 2.878145933151245,
      "learning_rate": 2.6367393484494994e-05,
      "loss": 0.1787,
      "num_input_tokens_seen": 70693856,
      "step": 32755
    },
    {
      "epoch": 5.3442088091354,
      "grad_norm": 0.1917629837989807,
      "learning_rate": 2.636028610203755e-05,
      "loss": 0.0087,
      "num_input_tokens_seen": 70705024,
      "step": 32760
    },
    {
      "epoch": 5.3450244698205545,
      "grad_norm": 0.16410884261131287,
      "learning_rate": 2.635317860930841e-05,
      "loss": 0.0105,
      "num_input_tokens_seen": 70715040,
      "step": 32765
    },
    {
      "epoch": 5.345840130505709,
      "grad_norm": 12.318873405456543,
      "learning_rate": 2.6346071006883748e-05,
      "loss": 0.0714,
      "num_input_tokens_seen": 70726080,
      "step": 32770
    },
    {
      "epoch": 5.346655791190865,
      "grad_norm": 0.12560924887657166,
      "learning_rate": 2.6338963295339737e-05,
      "loss": 0.0146,
      "num_input_tokens_seen": 70736800,
      "step": 32775
    },
    {
      "epoch": 5.34747145187602,
      "grad_norm": 19.746686935424805,
      "learning_rate": 2.633185547525257e-05,
      "loss": 0.0734,
      "num_input_tokens_seen": 70748928,
      "step": 32780
    },
    {
      "epoch": 5.348287112561175,
      "grad_norm": 0.23885926604270935,
      "learning_rate": 2.6324747547198443e-05,
      "loss": 0.0997,
      "num_input_tokens_seen": 70760736,
      "step": 32785
    },
    {
      "epoch": 5.349102773246329,
      "grad_norm": 2.257115364074707,
      "learning_rate": 2.631763951175355e-05,
      "loss": 0.1691,
      "num_input_tokens_seen": 70771552,
      "step": 32790
    },
    {
      "epoch": 5.349918433931484,
      "grad_norm": 1.4539097547531128,
      "learning_rate": 2.6310531369494118e-05,
      "loss": 0.0141,
      "num_input_tokens_seen": 70782656,
      "step": 32795
    },
    {
      "epoch": 5.350734094616639,
      "grad_norm": 13.407401084899902,
      "learning_rate": 2.630342312099637e-05,
      "loss": 0.1118,
      "num_input_tokens_seen": 70792736,
      "step": 32800
    },
    {
      "epoch": 5.351549755301795,
      "grad_norm": 0.21261943876743317,
      "learning_rate": 2.629631476683652e-05,
      "loss": 0.1466,
      "num_input_tokens_seen": 70804032,
      "step": 32805
    },
    {
      "epoch": 5.35236541598695,
      "grad_norm": 1.5478466749191284,
      "learning_rate": 2.6289206307590815e-05,
      "loss": 0.0089,
      "num_input_tokens_seen": 70814208,
      "step": 32810
    },
    {
      "epoch": 5.353181076672104,
      "grad_norm": 2.2880988121032715,
      "learning_rate": 2.6282097743835517e-05,
      "loss": 0.1866,
      "num_input_tokens_seen": 70825312,
      "step": 32815
    },
    {
      "epoch": 5.353996737357259,
      "grad_norm": 0.7886966466903687,
      "learning_rate": 2.627498907614686e-05,
      "loss": 0.1134,
      "num_input_tokens_seen": 70835680,
      "step": 32820
    },
    {
      "epoch": 5.354812398042414,
      "grad_norm": 0.09569641202688217,
      "learning_rate": 2.6267880305101127e-05,
      "loss": 0.0423,
      "num_input_tokens_seen": 70846336,
      "step": 32825
    },
    {
      "epoch": 5.35562805872757,
      "grad_norm": 2.4989962577819824,
      "learning_rate": 2.626077143127458e-05,
      "loss": 0.1074,
      "num_input_tokens_seen": 70858016,
      "step": 32830
    },
    {
      "epoch": 5.356443719412725,
      "grad_norm": 3.8092448711395264,
      "learning_rate": 2.6253662455243504e-05,
      "loss": 0.2957,
      "num_input_tokens_seen": 70868704,
      "step": 32835
    },
    {
      "epoch": 5.357259380097879,
      "grad_norm": 0.07583662122488022,
      "learning_rate": 2.6246553377584186e-05,
      "loss": 0.0413,
      "num_input_tokens_seen": 70879392,
      "step": 32840
    },
    {
      "epoch": 5.358075040783034,
      "grad_norm": 2.4972646236419678,
      "learning_rate": 2.623944419887293e-05,
      "loss": 0.1723,
      "num_input_tokens_seen": 70890848,
      "step": 32845
    },
    {
      "epoch": 5.358890701468189,
      "grad_norm": 0.3224732577800751,
      "learning_rate": 2.6232334919686035e-05,
      "loss": 0.0123,
      "num_input_tokens_seen": 70901472,
      "step": 32850
    },
    {
      "epoch": 5.359706362153344,
      "grad_norm": 10.250856399536133,
      "learning_rate": 2.6225225540599825e-05,
      "loss": 0.2318,
      "num_input_tokens_seen": 70911904,
      "step": 32855
    },
    {
      "epoch": 5.3605220228384995,
      "grad_norm": 0.19227977097034454,
      "learning_rate": 2.6218116062190605e-05,
      "loss": 0.0105,
      "num_input_tokens_seen": 70922880,
      "step": 32860
    },
    {
      "epoch": 5.361337683523654,
      "grad_norm": 0.1123371422290802,
      "learning_rate": 2.621100648503472e-05,
      "loss": 0.0705,
      "num_input_tokens_seen": 70933472,
      "step": 32865
    },
    {
      "epoch": 5.362153344208809,
      "grad_norm": 0.0657767653465271,
      "learning_rate": 2.6203896809708512e-05,
      "loss": 0.1193,
      "num_input_tokens_seen": 70944864,
      "step": 32870
    },
    {
      "epoch": 5.362969004893964,
      "grad_norm": 8.928384780883789,
      "learning_rate": 2.619678703678832e-05,
      "loss": 0.0632,
      "num_input_tokens_seen": 70956512,
      "step": 32875
    },
    {
      "epoch": 5.363784665579119,
      "grad_norm": 0.13426414132118225,
      "learning_rate": 2.618967716685049e-05,
      "loss": 0.0059,
      "num_input_tokens_seen": 70966400,
      "step": 32880
    },
    {
      "epoch": 5.364600326264274,
      "grad_norm": 13.635554313659668,
      "learning_rate": 2.6182567200471396e-05,
      "loss": 0.0216,
      "num_input_tokens_seen": 70977120,
      "step": 32885
    },
    {
      "epoch": 5.365415986949429,
      "grad_norm": 0.16418886184692383,
      "learning_rate": 2.6175457138227404e-05,
      "loss": 0.1141,
      "num_input_tokens_seen": 70987712,
      "step": 32890
    },
    {
      "epoch": 5.366231647634584,
      "grad_norm": 0.2703512907028198,
      "learning_rate": 2.6168346980694896e-05,
      "loss": 0.0088,
      "num_input_tokens_seen": 70998080,
      "step": 32895
    },
    {
      "epoch": 5.367047308319739,
      "grad_norm": 0.12325150519609451,
      "learning_rate": 2.6161236728450257e-05,
      "loss": 0.2325,
      "num_input_tokens_seen": 71009568,
      "step": 32900
    },
    {
      "epoch": 5.367862969004894,
      "grad_norm": 3.2493736743927,
      "learning_rate": 2.6154126382069866e-05,
      "loss": 0.2135,
      "num_input_tokens_seen": 71021728,
      "step": 32905
    },
    {
      "epoch": 5.368678629690049,
      "grad_norm": 0.23775611817836761,
      "learning_rate": 2.6147015942130143e-05,
      "loss": 0.0112,
      "num_input_tokens_seen": 71032704,
      "step": 32910
    },
    {
      "epoch": 5.369494290375204,
      "grad_norm": 0.13886013627052307,
      "learning_rate": 2.6139905409207475e-05,
      "loss": 0.0079,
      "num_input_tokens_seen": 71043808,
      "step": 32915
    },
    {
      "epoch": 5.370309951060359,
      "grad_norm": 0.10095523297786713,
      "learning_rate": 2.61327947838783e-05,
      "loss": 0.2003,
      "num_input_tokens_seen": 71054752,
      "step": 32920
    },
    {
      "epoch": 5.371125611745514,
      "grad_norm": 0.23826496303081512,
      "learning_rate": 2.6125684066719036e-05,
      "loss": 0.1355,
      "num_input_tokens_seen": 71065216,
      "step": 32925
    },
    {
      "epoch": 5.371941272430669,
      "grad_norm": 0.20730678737163544,
      "learning_rate": 2.6118573258306106e-05,
      "loss": 0.1139,
      "num_input_tokens_seen": 71074752,
      "step": 32930
    },
    {
      "epoch": 5.372756933115824,
      "grad_norm": 0.11309021711349487,
      "learning_rate": 2.6111462359215944e-05,
      "loss": 0.0195,
      "num_input_tokens_seen": 71084864,
      "step": 32935
    },
    {
      "epoch": 5.373572593800978,
      "grad_norm": 0.2552671730518341,
      "learning_rate": 2.6104351370025014e-05,
      "loss": 0.1077,
      "num_input_tokens_seen": 71095392,
      "step": 32940
    },
    {
      "epoch": 5.374388254486134,
      "grad_norm": 0.07608038187026978,
      "learning_rate": 2.6097240291309756e-05,
      "loss": 0.1251,
      "num_input_tokens_seen": 71105472,
      "step": 32945
    },
    {
      "epoch": 5.375203915171289,
      "grad_norm": 0.12142956256866455,
      "learning_rate": 2.6090129123646633e-05,
      "loss": 0.1885,
      "num_input_tokens_seen": 71116448,
      "step": 32950
    },
    {
      "epoch": 5.376019575856444,
      "grad_norm": 0.08452937006950378,
      "learning_rate": 2.6083017867612115e-05,
      "loss": 0.0086,
      "num_input_tokens_seen": 71128032,
      "step": 32955
    },
    {
      "epoch": 5.376835236541599,
      "grad_norm": 0.1397397518157959,
      "learning_rate": 2.6075906523782666e-05,
      "loss": 0.0198,
      "num_input_tokens_seen": 71140448,
      "step": 32960
    },
    {
      "epoch": 5.377650897226753,
      "grad_norm": 0.1688927561044693,
      "learning_rate": 2.6068795092734783e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 71150720,
      "step": 32965
    },
    {
      "epoch": 5.378466557911908,
      "grad_norm": 0.32195374369621277,
      "learning_rate": 2.6061683575044937e-05,
      "loss": 0.0985,
      "num_input_tokens_seen": 71161920,
      "step": 32970
    },
    {
      "epoch": 5.379282218597064,
      "grad_norm": 0.14487984776496887,
      "learning_rate": 2.605457197128964e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 71173888,
      "step": 32975
    },
    {
      "epoch": 5.380097879282219,
      "grad_norm": 0.05538102984428406,
      "learning_rate": 2.6047460282045388e-05,
      "loss": 0.0078,
      "num_input_tokens_seen": 71185984,
      "step": 32980
    },
    {
      "epoch": 5.3809135399673735,
      "grad_norm": 0.16724319756031036,
      "learning_rate": 2.604034850788869e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 71196640,
      "step": 32985
    },
    {
      "epoch": 5.381729200652528,
      "grad_norm": 0.04707736149430275,
      "learning_rate": 2.6033236649396063e-05,
      "loss": 0.0034,
      "num_input_tokens_seen": 71208000,
      "step": 32990
    },
    {
      "epoch": 5.382544861337683,
      "grad_norm": 0.18590906262397766,
      "learning_rate": 2.6026124707144033e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 71219616,
      "step": 32995
    },
    {
      "epoch": 5.383360522022839,
      "grad_norm": 0.07881730049848557,
      "learning_rate": 2.6019012681709127e-05,
      "loss": 0.1255,
      "num_input_tokens_seen": 71230880,
      "step": 33000
    },
    {
      "epoch": 5.384176182707994,
      "grad_norm": 3.265533685684204,
      "learning_rate": 2.601190057366788e-05,
      "loss": 0.248,
      "num_input_tokens_seen": 71240800,
      "step": 33005
    },
    {
      "epoch": 5.3849918433931485,
      "grad_norm": 0.11597271263599396,
      "learning_rate": 2.600478838359684e-05,
      "loss": 0.1397,
      "num_input_tokens_seen": 71250784,
      "step": 33010
    },
    {
      "epoch": 5.385807504078303,
      "grad_norm": 0.06675510108470917,
      "learning_rate": 2.5997676112072557e-05,
      "loss": 0.0938,
      "num_input_tokens_seen": 71260576,
      "step": 33015
    },
    {
      "epoch": 5.386623164763458,
      "grad_norm": 0.37705445289611816,
      "learning_rate": 2.5990563759671575e-05,
      "loss": 0.0073,
      "num_input_tokens_seen": 71270976,
      "step": 33020
    },
    {
      "epoch": 5.387438825448613,
      "grad_norm": 0.04514387622475624,
      "learning_rate": 2.598345132697048e-05,
      "loss": 0.082,
      "num_input_tokens_seen": 71281216,
      "step": 33025
    },
    {
      "epoch": 5.388254486133769,
      "grad_norm": 0.4911499321460724,
      "learning_rate": 2.597633881454583e-05,
      "loss": 0.2121,
      "num_input_tokens_seen": 71292064,
      "step": 33030
    },
    {
      "epoch": 5.3890701468189235,
      "grad_norm": 0.19001545011997223,
      "learning_rate": 2.5969226222974196e-05,
      "loss": 0.1146,
      "num_input_tokens_seen": 71303392,
      "step": 33035
    },
    {
      "epoch": 5.389885807504078,
      "grad_norm": 0.0863967165350914,
      "learning_rate": 2.5962113552832173e-05,
      "loss": 0.0243,
      "num_input_tokens_seen": 71314016,
      "step": 33040
    },
    {
      "epoch": 5.390701468189233,
      "grad_norm": 0.10328985005617142,
      "learning_rate": 2.5955000804696345e-05,
      "loss": 0.0575,
      "num_input_tokens_seen": 71326016,
      "step": 33045
    },
    {
      "epoch": 5.391517128874388,
      "grad_norm": 1.9994709491729736,
      "learning_rate": 2.5947887979143304e-05,
      "loss": 0.1321,
      "num_input_tokens_seen": 71335840,
      "step": 33050
    },
    {
      "epoch": 5.392332789559543,
      "grad_norm": 0.18004560470581055,
      "learning_rate": 2.594077507674965e-05,
      "loss": 0.0847,
      "num_input_tokens_seen": 71347072,
      "step": 33055
    },
    {
      "epoch": 5.3931484502446985,
      "grad_norm": 11.753279685974121,
      "learning_rate": 2.5933662098091997e-05,
      "loss": 0.069,
      "num_input_tokens_seen": 71358464,
      "step": 33060
    },
    {
      "epoch": 5.393964110929853,
      "grad_norm": 9.915629386901855,
      "learning_rate": 2.5926549043746962e-05,
      "loss": 0.0235,
      "num_input_tokens_seen": 71367776,
      "step": 33065
    },
    {
      "epoch": 5.394779771615008,
      "grad_norm": 0.18175454437732697,
      "learning_rate": 2.591943591429115e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 71378368,
      "step": 33070
    },
    {
      "epoch": 5.395595432300163,
      "grad_norm": 3.6783509254455566,
      "learning_rate": 2.5912322710301202e-05,
      "loss": 0.1407,
      "num_input_tokens_seen": 71389632,
      "step": 33075
    },
    {
      "epoch": 5.396411092985318,
      "grad_norm": 0.12951412796974182,
      "learning_rate": 2.590520943235375e-05,
      "loss": 0.0037,
      "num_input_tokens_seen": 71400064,
      "step": 33080
    },
    {
      "epoch": 5.397226753670473,
      "grad_norm": 0.09587570279836655,
      "learning_rate": 2.5898096081025424e-05,
      "loss": 0.0096,
      "num_input_tokens_seen": 71410784,
      "step": 33085
    },
    {
      "epoch": 5.398042414355628,
      "grad_norm": 5.226691246032715,
      "learning_rate": 2.589098265689287e-05,
      "loss": 0.227,
      "num_input_tokens_seen": 71421632,
      "step": 33090
    },
    {
      "epoch": 5.398858075040783,
      "grad_norm": 0.03244301676750183,
      "learning_rate": 2.5883869160532743e-05,
      "loss": 0.2446,
      "num_input_tokens_seen": 71432640,
      "step": 33095
    },
    {
      "epoch": 5.399673735725938,
      "grad_norm": 0.13412439823150635,
      "learning_rate": 2.58767555925217e-05,
      "loss": 0.0039,
      "num_input_tokens_seen": 71442816,
      "step": 33100
    },
    {
      "epoch": 5.400489396411093,
      "grad_norm": 0.16147126257419586,
      "learning_rate": 2.5869641953436402e-05,
      "loss": 0.0198,
      "num_input_tokens_seen": 71453568,
      "step": 33105
    },
    {
      "epoch": 5.401305057096248,
      "grad_norm": 0.09940747916698456,
      "learning_rate": 2.5862528243853513e-05,
      "loss": 0.0154,
      "num_input_tokens_seen": 71464448,
      "step": 33110
    },
    {
      "epoch": 5.402120717781403,
      "grad_norm": 0.026617012917995453,
      "learning_rate": 2.5855414464349707e-05,
      "loss": 0.1371,
      "num_input_tokens_seen": 71475264,
      "step": 33115
    },
    {
      "epoch": 5.402936378466558,
      "grad_norm": 0.08824633061885834,
      "learning_rate": 2.5848300615501663e-05,
      "loss": 0.0712,
      "num_input_tokens_seen": 71484704,
      "step": 33120
    },
    {
      "epoch": 5.403752039151713,
      "grad_norm": 10.036717414855957,
      "learning_rate": 2.5841186697886065e-05,
      "loss": 0.1291,
      "num_input_tokens_seen": 71494592,
      "step": 33125
    },
    {
      "epoch": 5.404567699836868,
      "grad_norm": 3.598909378051758,
      "learning_rate": 2.583407271207961e-05,
      "loss": 0.2294,
      "num_input_tokens_seen": 71505568,
      "step": 33130
    },
    {
      "epoch": 5.4053833605220225,
      "grad_norm": 0.037693921476602554,
      "learning_rate": 2.582695865865899e-05,
      "loss": 0.1229,
      "num_input_tokens_seen": 71515872,
      "step": 33135
    },
    {
      "epoch": 5.406199021207178,
      "grad_norm": 0.03884103149175644,
      "learning_rate": 2.5819844538200906e-05,
      "loss": 0.0429,
      "num_input_tokens_seen": 71526720,
      "step": 33140
    },
    {
      "epoch": 5.407014681892333,
      "grad_norm": 0.0533999465405941,
      "learning_rate": 2.5812730351282056e-05,
      "loss": 0.096,
      "num_input_tokens_seen": 71537856,
      "step": 33145
    },
    {
      "epoch": 5.407830342577488,
      "grad_norm": 0.11679085344076157,
      "learning_rate": 2.5805616098479167e-05,
      "loss": 0.0177,
      "num_input_tokens_seen": 71546624,
      "step": 33150
    },
    {
      "epoch": 5.408646003262643,
      "grad_norm": 0.03162815421819687,
      "learning_rate": 2.5798501780368944e-05,
      "loss": 0.201,
      "num_input_tokens_seen": 71556864,
      "step": 33155
    },
    {
      "epoch": 5.4094616639477975,
      "grad_norm": 1.8352751731872559,
      "learning_rate": 2.5791387397528123e-05,
      "loss": 0.0315,
      "num_input_tokens_seen": 71566848,
      "step": 33160
    },
    {
      "epoch": 5.410277324632952,
      "grad_norm": 2.4174234867095947,
      "learning_rate": 2.578427295053341e-05,
      "loss": 0.1758,
      "num_input_tokens_seen": 71576544,
      "step": 33165
    },
    {
      "epoch": 5.411092985318108,
      "grad_norm": 0.07857690751552582,
      "learning_rate": 2.5777158439961564e-05,
      "loss": 0.0328,
      "num_input_tokens_seen": 71586816,
      "step": 33170
    },
    {
      "epoch": 5.411908646003263,
      "grad_norm": 6.798092842102051,
      "learning_rate": 2.577004386638931e-05,
      "loss": 0.0316,
      "num_input_tokens_seen": 71596480,
      "step": 33175
    },
    {
      "epoch": 5.412724306688418,
      "grad_norm": 0.40470385551452637,
      "learning_rate": 2.576292923039339e-05,
      "loss": 0.0332,
      "num_input_tokens_seen": 71607968,
      "step": 33180
    },
    {
      "epoch": 5.4135399673735725,
      "grad_norm": 3.3327794075012207,
      "learning_rate": 2.5755814532550553e-05,
      "loss": 0.2206,
      "num_input_tokens_seen": 71618560,
      "step": 33185
    },
    {
      "epoch": 5.414355628058727,
      "grad_norm": 0.14963901042938232,
      "learning_rate": 2.574869977343756e-05,
      "loss": 0.0054,
      "num_input_tokens_seen": 71629952,
      "step": 33190
    },
    {
      "epoch": 5.415171288743883,
      "grad_norm": 0.42876407504081726,
      "learning_rate": 2.574158495363117e-05,
      "loss": 0.0304,
      "num_input_tokens_seen": 71639296,
      "step": 33195
    },
    {
      "epoch": 5.415986949429038,
      "grad_norm": 0.11662980914115906,
      "learning_rate": 2.5734470073708133e-05,
      "loss": 0.0899,
      "num_input_tokens_seen": 71649760,
      "step": 33200
    },
    {
      "epoch": 5.416802610114193,
      "grad_norm": 0.3094348907470703,
      "learning_rate": 2.572735513424523e-05,
      "loss": 0.0565,
      "num_input_tokens_seen": 71660800,
      "step": 33205
    },
    {
      "epoch": 5.417618270799347,
      "grad_norm": 0.10273489356040955,
      "learning_rate": 2.5720240135819223e-05,
      "loss": 0.0554,
      "num_input_tokens_seen": 71671104,
      "step": 33210
    },
    {
      "epoch": 5.418433931484502,
      "grad_norm": 3.2516162395477295,
      "learning_rate": 2.57131250790069e-05,
      "loss": 0.1653,
      "num_input_tokens_seen": 71681568,
      "step": 33215
    },
    {
      "epoch": 5.419249592169657,
      "grad_norm": 0.19021402299404144,
      "learning_rate": 2.570600996438504e-05,
      "loss": 0.1874,
      "num_input_tokens_seen": 71691008,
      "step": 33220
    },
    {
      "epoch": 5.420065252854813,
      "grad_norm": 0.20147915184497833,
      "learning_rate": 2.5698894792530432e-05,
      "loss": 0.0104,
      "num_input_tokens_seen": 71702112,
      "step": 33225
    },
    {
      "epoch": 5.420880913539968,
      "grad_norm": 0.3195922374725342,
      "learning_rate": 2.5691779564019862e-05,
      "loss": 0.0191,
      "num_input_tokens_seen": 71713696,
      "step": 33230
    },
    {
      "epoch": 5.421696574225122,
      "grad_norm": 0.41486889123916626,
      "learning_rate": 2.5684664279430125e-05,
      "loss": 0.2334,
      "num_input_tokens_seen": 71724800,
      "step": 33235
    },
    {
      "epoch": 5.422512234910277,
      "grad_norm": 10.956829071044922,
      "learning_rate": 2.5677548939338035e-05,
      "loss": 0.0639,
      "num_input_tokens_seen": 71736128,
      "step": 33240
    },
    {
      "epoch": 5.423327895595432,
      "grad_norm": 0.10004019737243652,
      "learning_rate": 2.5670433544320388e-05,
      "loss": 0.0899,
      "num_input_tokens_seen": 71746912,
      "step": 33245
    },
    {
      "epoch": 5.424143556280587,
      "grad_norm": 3.6930129528045654,
      "learning_rate": 2.5663318094953997e-05,
      "loss": 0.2368,
      "num_input_tokens_seen": 71757888,
      "step": 33250
    },
    {
      "epoch": 5.424959216965743,
      "grad_norm": 0.07590536773204803,
      "learning_rate": 2.5656202591815675e-05,
      "loss": 0.1169,
      "num_input_tokens_seen": 71769504,
      "step": 33255
    },
    {
      "epoch": 5.425774877650897,
      "grad_norm": 11.46955394744873,
      "learning_rate": 2.5649087035482243e-05,
      "loss": 0.0598,
      "num_input_tokens_seen": 71780352,
      "step": 33260
    },
    {
      "epoch": 5.426590538336052,
      "grad_norm": 2.9528470039367676,
      "learning_rate": 2.5641971426530525e-05,
      "loss": 0.1247,
      "num_input_tokens_seen": 71791488,
      "step": 33265
    },
    {
      "epoch": 5.427406199021207,
      "grad_norm": 0.3688742220401764,
      "learning_rate": 2.5634855765537347e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 71801728,
      "step": 33270
    },
    {
      "epoch": 5.428221859706362,
      "grad_norm": 11.852249145507812,
      "learning_rate": 2.5627740053079534e-05,
      "loss": 0.1007,
      "num_input_tokens_seen": 71812096,
      "step": 33275
    },
    {
      "epoch": 5.4290375203915175,
      "grad_norm": 0.11403562873601913,
      "learning_rate": 2.562062428973393e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 71822752,
      "step": 33280
    },
    {
      "epoch": 5.429853181076672,
      "grad_norm": 0.026809722185134888,
      "learning_rate": 2.5613508476077365e-05,
      "loss": 0.0073,
      "num_input_tokens_seen": 71833056,
      "step": 33285
    },
    {
      "epoch": 5.430668841761827,
      "grad_norm": 0.06105045601725578,
      "learning_rate": 2.5606392612686697e-05,
      "loss": 0.1306,
      "num_input_tokens_seen": 71842528,
      "step": 33290
    },
    {
      "epoch": 5.431484502446982,
      "grad_norm": 0.19938939809799194,
      "learning_rate": 2.5599276700138764e-05,
      "loss": 0.0688,
      "num_input_tokens_seen": 71853440,
      "step": 33295
    },
    {
      "epoch": 5.432300163132137,
      "grad_norm": 0.07202395796775818,
      "learning_rate": 2.5592160739010425e-05,
      "loss": 0.0405,
      "num_input_tokens_seen": 71863040,
      "step": 33300
    },
    {
      "epoch": 5.433115823817292,
      "grad_norm": 0.044230278581380844,
      "learning_rate": 2.5585044729878526e-05,
      "loss": 0.0284,
      "num_input_tokens_seen": 71873792,
      "step": 33305
    },
    {
      "epoch": 5.433931484502447,
      "grad_norm": 4.130042552947998,
      "learning_rate": 2.557792867331994e-05,
      "loss": 0.0519,
      "num_input_tokens_seen": 71885856,
      "step": 33310
    },
    {
      "epoch": 5.434747145187602,
      "grad_norm": 0.1673230081796646,
      "learning_rate": 2.5570812569911518e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 71896416,
      "step": 33315
    },
    {
      "epoch": 5.435562805872757,
      "grad_norm": 0.05843973904848099,
      "learning_rate": 2.556369642023013e-05,
      "loss": 0.0845,
      "num_input_tokens_seen": 71906976,
      "step": 33320
    },
    {
      "epoch": 5.436378466557912,
      "grad_norm": 0.03759612888097763,
      "learning_rate": 2.5556580224852655e-05,
      "loss": 0.037,
      "num_input_tokens_seen": 71916864,
      "step": 33325
    },
    {
      "epoch": 5.437194127243067,
      "grad_norm": 0.16183309257030487,
      "learning_rate": 2.5549463984355964e-05,
      "loss": 0.0048,
      "num_input_tokens_seen": 71928288,
      "step": 33330
    },
    {
      "epoch": 5.438009787928221,
      "grad_norm": 0.09352131932973862,
      "learning_rate": 2.5542347699316933e-05,
      "loss": 0.0933,
      "num_input_tokens_seen": 71939104,
      "step": 33335
    },
    {
      "epoch": 5.438825448613377,
      "grad_norm": 0.7050575017929077,
      "learning_rate": 2.553523137031244e-05,
      "loss": 0.0098,
      "num_input_tokens_seen": 71950208,
      "step": 33340
    },
    {
      "epoch": 5.439641109298532,
      "grad_norm": 3.3034229278564453,
      "learning_rate": 2.5528114997919384e-05,
      "loss": 0.3338,
      "num_input_tokens_seen": 71961312,
      "step": 33345
    },
    {
      "epoch": 5.440456769983687,
      "grad_norm": 3.80537748336792,
      "learning_rate": 2.5520998582714645e-05,
      "loss": 0.28,
      "num_input_tokens_seen": 71973248,
      "step": 33350
    },
    {
      "epoch": 5.441272430668842,
      "grad_norm": 0.05213324353098869,
      "learning_rate": 2.5513882125275113e-05,
      "loss": 0.0399,
      "num_input_tokens_seen": 71983648,
      "step": 33355
    },
    {
      "epoch": 5.442088091353996,
      "grad_norm": 0.021955057978630066,
      "learning_rate": 2.5506765626177697e-05,
      "loss": 0.0565,
      "num_input_tokens_seen": 71995072,
      "step": 33360
    },
    {
      "epoch": 5.442903752039152,
      "grad_norm": 18.607219696044922,
      "learning_rate": 2.5499649085999282e-05,
      "loss": 0.1689,
      "num_input_tokens_seen": 72006112,
      "step": 33365
    },
    {
      "epoch": 5.443719412724307,
      "grad_norm": 0.10573983937501907,
      "learning_rate": 2.549253250531678e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 72017184,
      "step": 33370
    },
    {
      "epoch": 5.444535073409462,
      "grad_norm": 3.3179361820220947,
      "learning_rate": 2.548541588470709e-05,
      "loss": 0.0748,
      "num_input_tokens_seen": 72028352,
      "step": 33375
    },
    {
      "epoch": 5.445350734094617,
      "grad_norm": 0.037115976214408875,
      "learning_rate": 2.547829922474713e-05,
      "loss": 0.0365,
      "num_input_tokens_seen": 72037664,
      "step": 33380
    },
    {
      "epoch": 5.446166394779771,
      "grad_norm": 0.151457279920578,
      "learning_rate": 2.5471182526013805e-05,
      "loss": 0.0537,
      "num_input_tokens_seen": 72049120,
      "step": 33385
    },
    {
      "epoch": 5.446982055464926,
      "grad_norm": 0.39002159237861633,
      "learning_rate": 2.546406578908403e-05,
      "loss": 0.1165,
      "num_input_tokens_seen": 72059104,
      "step": 33390
    },
    {
      "epoch": 5.447797716150082,
      "grad_norm": 3.467186450958252,
      "learning_rate": 2.545694901453473e-05,
      "loss": 0.1031,
      "num_input_tokens_seen": 72070048,
      "step": 33395
    },
    {
      "epoch": 5.448613376835237,
      "grad_norm": 0.12871865928173065,
      "learning_rate": 2.5449832202942832e-05,
      "loss": 0.265,
      "num_input_tokens_seen": 72080928,
      "step": 33400
    },
    {
      "epoch": 5.4494290375203915,
      "grad_norm": 0.06301422417163849,
      "learning_rate": 2.5442715354885237e-05,
      "loss": 0.0519,
      "num_input_tokens_seen": 72091488,
      "step": 33405
    },
    {
      "epoch": 5.450244698205546,
      "grad_norm": 5.510315895080566,
      "learning_rate": 2.5435598470938903e-05,
      "loss": 0.0918,
      "num_input_tokens_seen": 72102528,
      "step": 33410
    },
    {
      "epoch": 5.451060358890701,
      "grad_norm": 1.7421666383743286,
      "learning_rate": 2.5428481551680745e-05,
      "loss": 0.0615,
      "num_input_tokens_seen": 72112992,
      "step": 33415
    },
    {
      "epoch": 5.451876019575856,
      "grad_norm": 0.5023473501205444,
      "learning_rate": 2.5421364597687696e-05,
      "loss": 0.0449,
      "num_input_tokens_seen": 72124096,
      "step": 33420
    },
    {
      "epoch": 5.452691680261012,
      "grad_norm": 0.3875141739845276,
      "learning_rate": 2.5414247609536696e-05,
      "loss": 0.044,
      "num_input_tokens_seen": 72134624,
      "step": 33425
    },
    {
      "epoch": 5.4535073409461665,
      "grad_norm": 2.727522611618042,
      "learning_rate": 2.5407130587804685e-05,
      "loss": 0.2095,
      "num_input_tokens_seen": 72145344,
      "step": 33430
    },
    {
      "epoch": 5.454323001631321,
      "grad_norm": 0.17522837221622467,
      "learning_rate": 2.5400013533068594e-05,
      "loss": 0.0596,
      "num_input_tokens_seen": 72157312,
      "step": 33435
    },
    {
      "epoch": 5.455138662316476,
      "grad_norm": 16.254186630249023,
      "learning_rate": 2.5392896445905385e-05,
      "loss": 0.2993,
      "num_input_tokens_seen": 72168032,
      "step": 33440
    },
    {
      "epoch": 5.455954323001631,
      "grad_norm": 0.13682357966899872,
      "learning_rate": 2.538577932689199e-05,
      "loss": 0.0372,
      "num_input_tokens_seen": 72178944,
      "step": 33445
    },
    {
      "epoch": 5.456769983686787,
      "grad_norm": 5.391839027404785,
      "learning_rate": 2.537866217660537e-05,
      "loss": 0.0823,
      "num_input_tokens_seen": 72188640,
      "step": 33450
    },
    {
      "epoch": 5.4575856443719415,
      "grad_norm": 0.13826815783977509,
      "learning_rate": 2.5371544995622472e-05,
      "loss": 0.1648,
      "num_input_tokens_seen": 72198720,
      "step": 33455
    },
    {
      "epoch": 5.458401305057096,
      "grad_norm": 0.025393128395080566,
      "learning_rate": 2.536442778452025e-05,
      "loss": 0.0092,
      "num_input_tokens_seen": 72208704,
      "step": 33460
    },
    {
      "epoch": 5.459216965742251,
      "grad_norm": 0.035799961537122726,
      "learning_rate": 2.5357310543875667e-05,
      "loss": 0.1064,
      "num_input_tokens_seen": 72218688,
      "step": 33465
    },
    {
      "epoch": 5.460032626427406,
      "grad_norm": 0.5198113918304443,
      "learning_rate": 2.5350193274265678e-05,
      "loss": 0.0613,
      "num_input_tokens_seen": 72230336,
      "step": 33470
    },
    {
      "epoch": 5.460848287112561,
      "grad_norm": 0.08530508726835251,
      "learning_rate": 2.5343075976267234e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 72241024,
      "step": 33475
    },
    {
      "epoch": 5.4616639477977165,
      "grad_norm": 1.217314600944519,
      "learning_rate": 2.533595865045732e-05,
      "loss": 0.0833,
      "num_input_tokens_seen": 72252064,
      "step": 33480
    },
    {
      "epoch": 5.462479608482871,
      "grad_norm": 2.1794791221618652,
      "learning_rate": 2.532884129741289e-05,
      "loss": 0.2085,
      "num_input_tokens_seen": 72263616,
      "step": 33485
    },
    {
      "epoch": 5.463295269168026,
      "grad_norm": 4.713282108306885,
      "learning_rate": 2.5321723917710923e-05,
      "loss": 0.2499,
      "num_input_tokens_seen": 72274784,
      "step": 33490
    },
    {
      "epoch": 5.464110929853181,
      "grad_norm": 0.04681149125099182,
      "learning_rate": 2.531460651192838e-05,
      "loss": 0.1314,
      "num_input_tokens_seen": 72284064,
      "step": 33495
    },
    {
      "epoch": 5.464926590538336,
      "grad_norm": 2.854098320007324,
      "learning_rate": 2.5307489080642227e-05,
      "loss": 0.1856,
      "num_input_tokens_seen": 72295616,
      "step": 33500
    },
    {
      "epoch": 5.465742251223491,
      "grad_norm": 0.08297780901193619,
      "learning_rate": 2.530037162442946e-05,
      "loss": 0.1858,
      "num_input_tokens_seen": 72305952,
      "step": 33505
    },
    {
      "epoch": 5.466557911908646,
      "grad_norm": 0.11627074331045151,
      "learning_rate": 2.529325414386704e-05,
      "loss": 0.0695,
      "num_input_tokens_seen": 72316512,
      "step": 33510
    },
    {
      "epoch": 5.467373572593801,
      "grad_norm": 0.10977844893932343,
      "learning_rate": 2.5286136639531956e-05,
      "loss": 0.1574,
      "num_input_tokens_seen": 72327328,
      "step": 33515
    },
    {
      "epoch": 5.468189233278956,
      "grad_norm": 4.253857135772705,
      "learning_rate": 2.527901911200118e-05,
      "loss": 0.0769,
      "num_input_tokens_seen": 72338560,
      "step": 33520
    },
    {
      "epoch": 5.469004893964111,
      "grad_norm": 4.271973133087158,
      "learning_rate": 2.5271901561851703e-05,
      "loss": 0.0998,
      "num_input_tokens_seen": 72349312,
      "step": 33525
    },
    {
      "epoch": 5.4698205546492655,
      "grad_norm": 8.721693992614746,
      "learning_rate": 2.52647839896605e-05,
      "loss": 0.0377,
      "num_input_tokens_seen": 72361024,
      "step": 33530
    },
    {
      "epoch": 5.470636215334421,
      "grad_norm": 0.10647906363010406,
      "learning_rate": 2.525766639600457e-05,
      "loss": 0.2605,
      "num_input_tokens_seen": 72372352,
      "step": 33535
    },
    {
      "epoch": 5.471451876019576,
      "grad_norm": 0.11971849203109741,
      "learning_rate": 2.525054878146089e-05,
      "loss": 0.0552,
      "num_input_tokens_seen": 72383136,
      "step": 33540
    },
    {
      "epoch": 5.472267536704731,
      "grad_norm": 0.08364076167345047,
      "learning_rate": 2.5243431146606456e-05,
      "loss": 0.0049,
      "num_input_tokens_seen": 72392224,
      "step": 33545
    },
    {
      "epoch": 5.473083197389886,
      "grad_norm": 0.12079733610153198,
      "learning_rate": 2.5236313492018254e-05,
      "loss": 0.0166,
      "num_input_tokens_seen": 72402176,
      "step": 33550
    },
    {
      "epoch": 5.4738988580750405,
      "grad_norm": 10.084371566772461,
      "learning_rate": 2.5229195818273284e-05,
      "loss": 0.135,
      "num_input_tokens_seen": 72413792,
      "step": 33555
    },
    {
      "epoch": 5.474714518760196,
      "grad_norm": 9.428359031677246,
      "learning_rate": 2.5222078125948534e-05,
      "loss": 0.1116,
      "num_input_tokens_seen": 72425792,
      "step": 33560
    },
    {
      "epoch": 5.475530179445351,
      "grad_norm": 0.036408375948667526,
      "learning_rate": 2.5214960415621007e-05,
      "loss": 0.0036,
      "num_input_tokens_seen": 72436864,
      "step": 33565
    },
    {
      "epoch": 5.476345840130506,
      "grad_norm": 0.14001904428005219,
      "learning_rate": 2.5207842687867705e-05,
      "loss": 0.0056,
      "num_input_tokens_seen": 72448096,
      "step": 33570
    },
    {
      "epoch": 5.477161500815661,
      "grad_norm": 0.4629044234752655,
      "learning_rate": 2.5200724943265614e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 72459776,
      "step": 33575
    },
    {
      "epoch": 5.4779771615008155,
      "grad_norm": 3.62270450592041,
      "learning_rate": 2.519360718239174e-05,
      "loss": 0.2214,
      "num_input_tokens_seen": 72470528,
      "step": 33580
    },
    {
      "epoch": 5.47879282218597,
      "grad_norm": 11.740579605102539,
      "learning_rate": 2.5186489405823087e-05,
      "loss": 0.435,
      "num_input_tokens_seen": 72482144,
      "step": 33585
    },
    {
      "epoch": 5.479608482871126,
      "grad_norm": 4.328428745269775,
      "learning_rate": 2.517937161413666e-05,
      "loss": 0.1729,
      "num_input_tokens_seen": 72493504,
      "step": 33590
    },
    {
      "epoch": 5.480424143556281,
      "grad_norm": 0.04599461331963539,
      "learning_rate": 2.517225380790946e-05,
      "loss": 0.1162,
      "num_input_tokens_seen": 72503904,
      "step": 33595
    },
    {
      "epoch": 5.481239804241436,
      "grad_norm": 4.032310485839844,
      "learning_rate": 2.5165135987718486e-05,
      "loss": 0.2707,
      "num_input_tokens_seen": 72514560,
      "step": 33600
    },
    {
      "epoch": 5.4820554649265905,
      "grad_norm": 0.14988665282726288,
      "learning_rate": 2.515801815414075e-05,
      "loss": 0.1035,
      "num_input_tokens_seen": 72526080,
      "step": 33605
    },
    {
      "epoch": 5.482871125611745,
      "grad_norm": 0.06683401018381119,
      "learning_rate": 2.5150900307753267e-05,
      "loss": 0.0934,
      "num_input_tokens_seen": 72536992,
      "step": 33610
    },
    {
      "epoch": 5.4836867862969,
      "grad_norm": 0.2525116801261902,
      "learning_rate": 2.5143782449133036e-05,
      "loss": 0.1114,
      "num_input_tokens_seen": 72548224,
      "step": 33615
    },
    {
      "epoch": 5.484502446982056,
      "grad_norm": 0.0889025330543518,
      "learning_rate": 2.5136664578857072e-05,
      "loss": 0.0169,
      "num_input_tokens_seen": 72559744,
      "step": 33620
    },
    {
      "epoch": 5.485318107667211,
      "grad_norm": 2.2922513484954834,
      "learning_rate": 2.5129546697502382e-05,
      "loss": 0.0187,
      "num_input_tokens_seen": 72570528,
      "step": 33625
    },
    {
      "epoch": 5.486133768352365,
      "grad_norm": 0.5152546167373657,
      "learning_rate": 2.512242880564598e-05,
      "loss": 0.0991,
      "num_input_tokens_seen": 72582112,
      "step": 33630
    },
    {
      "epoch": 5.48694942903752,
      "grad_norm": 0.09374929219484329,
      "learning_rate": 2.5115310903864874e-05,
      "loss": 0.1365,
      "num_input_tokens_seen": 72592736,
      "step": 33635
    },
    {
      "epoch": 5.487765089722675,
      "grad_norm": 0.0640566498041153,
      "learning_rate": 2.510819299273609e-05,
      "loss": 0.0885,
      "num_input_tokens_seen": 72604096,
      "step": 33640
    },
    {
      "epoch": 5.488580750407831,
      "grad_norm": 2.2651100158691406,
      "learning_rate": 2.510107507283663e-05,
      "loss": 0.0139,
      "num_input_tokens_seen": 72614560,
      "step": 33645
    },
    {
      "epoch": 5.489396411092986,
      "grad_norm": 4.540492534637451,
      "learning_rate": 2.5093957144743507e-05,
      "loss": 0.2888,
      "num_input_tokens_seen": 72625536,
      "step": 33650
    },
    {
      "epoch": 5.49021207177814,
      "grad_norm": 0.2596532702445984,
      "learning_rate": 2.5086839209033747e-05,
      "loss": 0.1229,
      "num_input_tokens_seen": 72636384,
      "step": 33655
    },
    {
      "epoch": 5.491027732463295,
      "grad_norm": 9.449118614196777,
      "learning_rate": 2.507972126628435e-05,
      "loss": 0.2381,
      "num_input_tokens_seen": 72646272,
      "step": 33660
    },
    {
      "epoch": 5.49184339314845,
      "grad_norm": 0.06537148356437683,
      "learning_rate": 2.5072603317072353e-05,
      "loss": 0.0975,
      "num_input_tokens_seen": 72657184,
      "step": 33665
    },
    {
      "epoch": 5.492659053833605,
      "grad_norm": 0.25678354501724243,
      "learning_rate": 2.5065485361974754e-05,
      "loss": 0.1541,
      "num_input_tokens_seen": 72668544,
      "step": 33670
    },
    {
      "epoch": 5.493474714518761,
      "grad_norm": 0.0459088459610939,
      "learning_rate": 2.505836740156859e-05,
      "loss": 0.0104,
      "num_input_tokens_seen": 72679104,
      "step": 33675
    },
    {
      "epoch": 5.494290375203915,
      "grad_norm": 1.9900023937225342,
      "learning_rate": 2.5051249436430862e-05,
      "loss": 0.2362,
      "num_input_tokens_seen": 72689856,
      "step": 33680
    },
    {
      "epoch": 5.49510603588907,
      "grad_norm": 0.3316431939601898,
      "learning_rate": 2.5044131467138597e-05,
      "loss": 0.0106,
      "num_input_tokens_seen": 72701312,
      "step": 33685
    },
    {
      "epoch": 5.495921696574225,
      "grad_norm": 0.058297768235206604,
      "learning_rate": 2.5037013494268814e-05,
      "loss": 0.0049,
      "num_input_tokens_seen": 72711808,
      "step": 33690
    },
    {
      "epoch": 5.49673735725938,
      "grad_norm": 0.16702599823474884,
      "learning_rate": 2.502989551839852e-05,
      "loss": 0.0562,
      "num_input_tokens_seen": 72722528,
      "step": 33695
    },
    {
      "epoch": 5.497553017944535,
      "grad_norm": 10.069947242736816,
      "learning_rate": 2.5022777540104752e-05,
      "loss": 0.1806,
      "num_input_tokens_seen": 72733408,
      "step": 33700
    },
    {
      "epoch": 5.49836867862969,
      "grad_norm": 11.7423677444458,
      "learning_rate": 2.5015659559964516e-05,
      "loss": 0.0934,
      "num_input_tokens_seen": 72744544,
      "step": 33705
    },
    {
      "epoch": 5.499184339314845,
      "grad_norm": 0.18311762809753418,
      "learning_rate": 2.5008541578554838e-05,
      "loss": 0.0133,
      "num_input_tokens_seen": 72755712,
      "step": 33710
    },
    {
      "epoch": 5.5,
      "grad_norm": 0.5082947015762329,
      "learning_rate": 2.5001423596452738e-05,
      "loss": 0.0486,
      "num_input_tokens_seen": 72765696,
      "step": 33715
    },
    {
      "epoch": 5.5,
      "eval_loss": 0.16719458997249603,
      "eval_runtime": 132.9383,
      "eval_samples_per_second": 20.498,
      "eval_steps_per_second": 5.13,
      "num_input_tokens_seen": 72765696,
      "step": 33715
    },
    {
      "epoch": 5.500815660685155,
      "grad_norm": 0.16994374990463257,
      "learning_rate": 2.4994305614235228e-05,
      "loss": 0.0077,
      "num_input_tokens_seen": 72776576,
      "step": 33720
    },
    {
      "epoch": 5.50163132137031,
      "grad_norm": 0.14518482983112335,
      "learning_rate": 2.498718763247934e-05,
      "loss": 0.0109,
      "num_input_tokens_seen": 72786112,
      "step": 33725
    },
    {
      "epoch": 5.502446982055465,
      "grad_norm": 0.035523671656847,
      "learning_rate": 2.4980069651762085e-05,
      "loss": 0.0184,
      "num_input_tokens_seen": 72796384,
      "step": 33730
    },
    {
      "epoch": 5.50326264274062,
      "grad_norm": 0.06263341754674911,
      "learning_rate": 2.4972951672660487e-05,
      "loss": 0.0131,
      "num_input_tokens_seen": 72806720,
      "step": 33735
    },
    {
      "epoch": 5.504078303425775,
      "grad_norm": 6.641420841217041,
      "learning_rate": 2.4965833695751563e-05,
      "loss": 0.0889,
      "num_input_tokens_seen": 72817408,
      "step": 33740
    },
    {
      "epoch": 5.50489396411093,
      "grad_norm": 4.616932392120361,
      "learning_rate": 2.4958715721612335e-05,
      "loss": 0.1256,
      "num_input_tokens_seen": 72828096,
      "step": 33745
    },
    {
      "epoch": 5.505709624796085,
      "grad_norm": 0.09794651716947556,
      "learning_rate": 2.495159775081982e-05,
      "loss": 0.0048,
      "num_input_tokens_seen": 72839040,
      "step": 33750
    },
    {
      "epoch": 5.506525285481239,
      "grad_norm": 1.5033702850341797,
      "learning_rate": 2.4944479783951037e-05,
      "loss": 0.0201,
      "num_input_tokens_seen": 72849728,
      "step": 33755
    },
    {
      "epoch": 5.507340946166395,
      "grad_norm": 0.0542055144906044,
      "learning_rate": 2.4937361821583e-05,
      "loss": 0.1689,
      "num_input_tokens_seen": 72859872,
      "step": 33760
    },
    {
      "epoch": 5.50815660685155,
      "grad_norm": 12.139034271240234,
      "learning_rate": 2.4930243864292736e-05,
      "loss": 0.3785,
      "num_input_tokens_seen": 72871648,
      "step": 33765
    },
    {
      "epoch": 5.508972267536705,
      "grad_norm": 0.08845608681440353,
      "learning_rate": 2.492312591265726e-05,
      "loss": 0.0111,
      "num_input_tokens_seen": 72883136,
      "step": 33770
    },
    {
      "epoch": 5.50978792822186,
      "grad_norm": 0.36479949951171875,
      "learning_rate": 2.4916007967253576e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 72892896,
      "step": 33775
    },
    {
      "epoch": 5.510603588907014,
      "grad_norm": 0.020641397684812546,
      "learning_rate": 2.490889002865872e-05,
      "loss": 0.1182,
      "num_input_tokens_seen": 72904576,
      "step": 33780
    },
    {
      "epoch": 5.511419249592169,
      "grad_norm": 0.07545993477106094,
      "learning_rate": 2.4901772097449703e-05,
      "loss": 0.0762,
      "num_input_tokens_seen": 72914016,
      "step": 33785
    },
    {
      "epoch": 5.512234910277325,
      "grad_norm": 0.06554381549358368,
      "learning_rate": 2.4894654174203535e-05,
      "loss": 0.0166,
      "num_input_tokens_seen": 72925152,
      "step": 33790
    },
    {
      "epoch": 5.51305057096248,
      "grad_norm": 0.7819629311561584,
      "learning_rate": 2.488753625949723e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 72936320,
      "step": 33795
    },
    {
      "epoch": 5.513866231647635,
      "grad_norm": 0.027263063937425613,
      "learning_rate": 2.488041835390781e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 72946848,
      "step": 33800
    },
    {
      "epoch": 5.514681892332789,
      "grad_norm": 0.08070861548185349,
      "learning_rate": 2.4873300458012285e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 72958208,
      "step": 33805
    },
    {
      "epoch": 5.515497553017944,
      "grad_norm": 4.997661113739014,
      "learning_rate": 2.486618257238767e-05,
      "loss": 0.0146,
      "num_input_tokens_seen": 72969152,
      "step": 33810
    },
    {
      "epoch": 5.5163132137031,
      "grad_norm": 0.07219992578029633,
      "learning_rate": 2.4859064697610977e-05,
      "loss": 0.237,
      "num_input_tokens_seen": 72981664,
      "step": 33815
    },
    {
      "epoch": 5.517128874388255,
      "grad_norm": 0.29128319025039673,
      "learning_rate": 2.485194683425921e-05,
      "loss": 0.0061,
      "num_input_tokens_seen": 72992224,
      "step": 33820
    },
    {
      "epoch": 5.5179445350734095,
      "grad_norm": 0.06747906655073166,
      "learning_rate": 2.4844828982909388e-05,
      "loss": 0.0388,
      "num_input_tokens_seen": 73003712,
      "step": 33825
    },
    {
      "epoch": 5.518760195758564,
      "grad_norm": 0.09900006651878357,
      "learning_rate": 2.4837711144138514e-05,
      "loss": 0.259,
      "num_input_tokens_seen": 73015328,
      "step": 33830
    },
    {
      "epoch": 5.519575856443719,
      "grad_norm": 3.314185857772827,
      "learning_rate": 2.48305933185236e-05,
      "loss": 0.1504,
      "num_input_tokens_seen": 73026304,
      "step": 33835
    },
    {
      "epoch": 5.520391517128875,
      "grad_norm": 0.2923809587955475,
      "learning_rate": 2.4823475506641646e-05,
      "loss": 0.2657,
      "num_input_tokens_seen": 73038688,
      "step": 33840
    },
    {
      "epoch": 5.52120717781403,
      "grad_norm": 0.23544256389141083,
      "learning_rate": 2.481635770906967e-05,
      "loss": 0.0075,
      "num_input_tokens_seen": 73049824,
      "step": 33845
    },
    {
      "epoch": 5.5220228384991845,
      "grad_norm": 0.06933756172657013,
      "learning_rate": 2.4809239926384664e-05,
      "loss": 0.1956,
      "num_input_tokens_seen": 73061760,
      "step": 33850
    },
    {
      "epoch": 5.522838499184339,
      "grad_norm": 0.17645519971847534,
      "learning_rate": 2.480212215916364e-05,
      "loss": 0.1016,
      "num_input_tokens_seen": 73072704,
      "step": 33855
    },
    {
      "epoch": 5.523654159869494,
      "grad_norm": 3.188039541244507,
      "learning_rate": 2.4795004407983593e-05,
      "loss": 0.1201,
      "num_input_tokens_seen": 73084032,
      "step": 33860
    },
    {
      "epoch": 5.524469820554649,
      "grad_norm": 0.21747814118862152,
      "learning_rate": 2.4787886673421536e-05,
      "loss": 0.0762,
      "num_input_tokens_seen": 73094592,
      "step": 33865
    },
    {
      "epoch": 5.525285481239804,
      "grad_norm": 0.2958066165447235,
      "learning_rate": 2.4780768956054457e-05,
      "loss": 0.1048,
      "num_input_tokens_seen": 73106304,
      "step": 33870
    },
    {
      "epoch": 5.5261011419249595,
      "grad_norm": 0.08281079679727554,
      "learning_rate": 2.477365125645936e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 73117088,
      "step": 33875
    },
    {
      "epoch": 5.526916802610114,
      "grad_norm": 0.08794347196817398,
      "learning_rate": 2.4766533575213242e-05,
      "loss": 0.0521,
      "num_input_tokens_seen": 73127776,
      "step": 33880
    },
    {
      "epoch": 5.527732463295269,
      "grad_norm": 1.4676828384399414,
      "learning_rate": 2.4759415912893096e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 73139904,
      "step": 33885
    },
    {
      "epoch": 5.528548123980424,
      "grad_norm": 0.09515149891376495,
      "learning_rate": 2.4752298270075918e-05,
      "loss": 0.0054,
      "num_input_tokens_seen": 73150688,
      "step": 33890
    },
    {
      "epoch": 5.529363784665579,
      "grad_norm": 0.16909362375736237,
      "learning_rate": 2.47451806473387e-05,
      "loss": 0.0183,
      "num_input_tokens_seen": 73160576,
      "step": 33895
    },
    {
      "epoch": 5.5301794453507345,
      "grad_norm": 0.14664624631404877,
      "learning_rate": 2.4738063045258415e-05,
      "loss": 0.1083,
      "num_input_tokens_seen": 73170272,
      "step": 33900
    },
    {
      "epoch": 5.530995106035889,
      "grad_norm": 0.5503304600715637,
      "learning_rate": 2.4730945464412085e-05,
      "loss": 0.23,
      "num_input_tokens_seen": 73181760,
      "step": 33905
    },
    {
      "epoch": 5.531810766721044,
      "grad_norm": 0.024620776996016502,
      "learning_rate": 2.472382790537668e-05,
      "loss": 0.0087,
      "num_input_tokens_seen": 73192096,
      "step": 33910
    },
    {
      "epoch": 5.532626427406199,
      "grad_norm": 3.65716290473938,
      "learning_rate": 2.4716710368729187e-05,
      "loss": 0.1757,
      "num_input_tokens_seen": 73201408,
      "step": 33915
    },
    {
      "epoch": 5.533442088091354,
      "grad_norm": 0.3498358130455017,
      "learning_rate": 2.4709592855046587e-05,
      "loss": 0.0301,
      "num_input_tokens_seen": 73211936,
      "step": 33920
    },
    {
      "epoch": 5.5342577487765094,
      "grad_norm": 16.28522491455078,
      "learning_rate": 2.4702475364905864e-05,
      "loss": 0.1952,
      "num_input_tokens_seen": 73223392,
      "step": 33925
    },
    {
      "epoch": 5.535073409461664,
      "grad_norm": 4.879078388214111,
      "learning_rate": 2.4695357898883998e-05,
      "loss": 0.1818,
      "num_input_tokens_seen": 73233920,
      "step": 33930
    },
    {
      "epoch": 5.535889070146819,
      "grad_norm": 0.27701303362846375,
      "learning_rate": 2.4688240457557967e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 73244288,
      "step": 33935
    },
    {
      "epoch": 5.536704730831974,
      "grad_norm": 2.499258279800415,
      "learning_rate": 2.4681123041504746e-05,
      "loss": 0.3055,
      "num_input_tokens_seen": 73256256,
      "step": 33940
    },
    {
      "epoch": 5.537520391517129,
      "grad_norm": 2.6599695682525635,
      "learning_rate": 2.46740056513013e-05,
      "loss": 0.2024,
      "num_input_tokens_seen": 73267264,
      "step": 33945
    },
    {
      "epoch": 5.5383360522022835,
      "grad_norm": 0.18573834002017975,
      "learning_rate": 2.466688828752462e-05,
      "loss": 0.1245,
      "num_input_tokens_seen": 73279168,
      "step": 33950
    },
    {
      "epoch": 5.539151712887438,
      "grad_norm": 2.9579179286956787,
      "learning_rate": 2.4659770950751666e-05,
      "loss": 0.1006,
      "num_input_tokens_seen": 73288128,
      "step": 33955
    },
    {
      "epoch": 5.539967373572594,
      "grad_norm": 5.895270824432373,
      "learning_rate": 2.4652653641559404e-05,
      "loss": 0.263,
      "num_input_tokens_seen": 73297472,
      "step": 33960
    },
    {
      "epoch": 5.540783034257749,
      "grad_norm": 0.10312424600124359,
      "learning_rate": 2.46455363605248e-05,
      "loss": 0.0883,
      "num_input_tokens_seen": 73309504,
      "step": 33965
    },
    {
      "epoch": 5.541598694942904,
      "grad_norm": 0.10254838317632675,
      "learning_rate": 2.4638419108224817e-05,
      "loss": 0.0273,
      "num_input_tokens_seen": 73320192,
      "step": 33970
    },
    {
      "epoch": 5.5424143556280585,
      "grad_norm": 0.04222483187913895,
      "learning_rate": 2.4631301885236415e-05,
      "loss": 0.1335,
      "num_input_tokens_seen": 73331168,
      "step": 33975
    },
    {
      "epoch": 5.543230016313213,
      "grad_norm": 10.71635627746582,
      "learning_rate": 2.4624184692136554e-05,
      "loss": 0.1051,
      "num_input_tokens_seen": 73342944,
      "step": 33980
    },
    {
      "epoch": 5.544045676998369,
      "grad_norm": 0.07811756432056427,
      "learning_rate": 2.4617067529502188e-05,
      "loss": 0.0046,
      "num_input_tokens_seen": 73353920,
      "step": 33985
    },
    {
      "epoch": 5.544861337683524,
      "grad_norm": 3.223130226135254,
      "learning_rate": 2.460995039791027e-05,
      "loss": 0.3277,
      "num_input_tokens_seen": 73366432,
      "step": 33990
    },
    {
      "epoch": 5.545676998368679,
      "grad_norm": 0.07836192101240158,
      "learning_rate": 2.4602833297937755e-05,
      "loss": 0.0763,
      "num_input_tokens_seen": 73377824,
      "step": 33995
    },
    {
      "epoch": 5.5464926590538335,
      "grad_norm": 0.31072986125946045,
      "learning_rate": 2.4595716230161586e-05,
      "loss": 0.0137,
      "num_input_tokens_seen": 73387840,
      "step": 34000
    },
    {
      "epoch": 5.547308319738988,
      "grad_norm": 0.13992208242416382,
      "learning_rate": 2.45885991951587e-05,
      "loss": 0.0282,
      "num_input_tokens_seen": 73398304,
      "step": 34005
    },
    {
      "epoch": 5.548123980424144,
      "grad_norm": 8.604073524475098,
      "learning_rate": 2.458148219350606e-05,
      "loss": 0.2722,
      "num_input_tokens_seen": 73410016,
      "step": 34010
    },
    {
      "epoch": 5.548939641109299,
      "grad_norm": 33.65258026123047,
      "learning_rate": 2.45743652257806e-05,
      "loss": 0.0335,
      "num_input_tokens_seen": 73420960,
      "step": 34015
    },
    {
      "epoch": 5.549755301794454,
      "grad_norm": 0.2566872537136078,
      "learning_rate": 2.4567248292559253e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 73431712,
      "step": 34020
    },
    {
      "epoch": 5.5505709624796085,
      "grad_norm": 4.51474142074585,
      "learning_rate": 2.4560131394418958e-05,
      "loss": 0.2098,
      "num_input_tokens_seen": 73441376,
      "step": 34025
    },
    {
      "epoch": 5.551386623164763,
      "grad_norm": 0.10575133562088013,
      "learning_rate": 2.4553014531936632e-05,
      "loss": 0.0309,
      "num_input_tokens_seen": 73450752,
      "step": 34030
    },
    {
      "epoch": 5.552202283849918,
      "grad_norm": 3.058199167251587,
      "learning_rate": 2.4545897705689223e-05,
      "loss": 0.1061,
      "num_input_tokens_seen": 73461952,
      "step": 34035
    },
    {
      "epoch": 5.553017944535073,
      "grad_norm": 6.396117210388184,
      "learning_rate": 2.4538780916253657e-05,
      "loss": 0.0286,
      "num_input_tokens_seen": 73472064,
      "step": 34040
    },
    {
      "epoch": 5.553833605220229,
      "grad_norm": 0.06466473639011383,
      "learning_rate": 2.4531664164206843e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 73483200,
      "step": 34045
    },
    {
      "epoch": 5.554649265905383,
      "grad_norm": 0.3745845556259155,
      "learning_rate": 2.4524547450125713e-05,
      "loss": 0.0039,
      "num_input_tokens_seen": 73494080,
      "step": 34050
    },
    {
      "epoch": 5.555464926590538,
      "grad_norm": 0.0476202592253685,
      "learning_rate": 2.4517430774587174e-05,
      "loss": 0.2585,
      "num_input_tokens_seen": 73503616,
      "step": 34055
    },
    {
      "epoch": 5.556280587275693,
      "grad_norm": 0.22952498495578766,
      "learning_rate": 2.4510314138168146e-05,
      "loss": 0.0652,
      "num_input_tokens_seen": 73515328,
      "step": 34060
    },
    {
      "epoch": 5.557096247960848,
      "grad_norm": 0.07610568404197693,
      "learning_rate": 2.4503197541445545e-05,
      "loss": 0.1514,
      "num_input_tokens_seen": 73527520,
      "step": 34065
    },
    {
      "epoch": 5.557911908646004,
      "grad_norm": 0.9586765170097351,
      "learning_rate": 2.4496080984996264e-05,
      "loss": 0.179,
      "num_input_tokens_seen": 73538688,
      "step": 34070
    },
    {
      "epoch": 5.558727569331158,
      "grad_norm": 4.535788059234619,
      "learning_rate": 2.448896446939722e-05,
      "loss": 0.0607,
      "num_input_tokens_seen": 73548544,
      "step": 34075
    },
    {
      "epoch": 5.559543230016313,
      "grad_norm": 0.20011794567108154,
      "learning_rate": 2.4481847995225307e-05,
      "loss": 0.0705,
      "num_input_tokens_seen": 73559712,
      "step": 34080
    },
    {
      "epoch": 5.560358890701468,
      "grad_norm": 0.09488652646541595,
      "learning_rate": 2.4474731563057426e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 73570304,
      "step": 34085
    },
    {
      "epoch": 5.561174551386623,
      "grad_norm": 2.0041866302490234,
      "learning_rate": 2.446761517347046e-05,
      "loss": 0.1299,
      "num_input_tokens_seen": 73581056,
      "step": 34090
    },
    {
      "epoch": 5.561990212071779,
      "grad_norm": 0.039340272545814514,
      "learning_rate": 2.446049882704132e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 73592064,
      "step": 34095
    },
    {
      "epoch": 5.562805872756933,
      "grad_norm": 0.13631676137447357,
      "learning_rate": 2.4453382524346882e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 73603296,
      "step": 34100
    },
    {
      "epoch": 5.563621533442088,
      "grad_norm": 0.2204047441482544,
      "learning_rate": 2.444626626596403e-05,
      "loss": 0.0689,
      "num_input_tokens_seen": 73612704,
      "step": 34105
    },
    {
      "epoch": 5.564437194127243,
      "grad_norm": 3.499032735824585,
      "learning_rate": 2.4439150052469644e-05,
      "loss": 0.1091,
      "num_input_tokens_seen": 73623168,
      "step": 34110
    },
    {
      "epoch": 5.565252854812398,
      "grad_norm": 0.06525424122810364,
      "learning_rate": 2.4432033884440585e-05,
      "loss": 0.3741,
      "num_input_tokens_seen": 73633088,
      "step": 34115
    },
    {
      "epoch": 5.566068515497553,
      "grad_norm": 0.24328632652759552,
      "learning_rate": 2.4424917762453757e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 73644512,
      "step": 34120
    },
    {
      "epoch": 5.566884176182708,
      "grad_norm": 0.05231905356049538,
      "learning_rate": 2.4417801687086013e-05,
      "loss": 0.0376,
      "num_input_tokens_seen": 73655008,
      "step": 34125
    },
    {
      "epoch": 5.567699836867863,
      "grad_norm": 0.5238412618637085,
      "learning_rate": 2.4410685658914213e-05,
      "loss": 0.0184,
      "num_input_tokens_seen": 73665216,
      "step": 34130
    },
    {
      "epoch": 5.568515497553018,
      "grad_norm": 0.07202116400003433,
      "learning_rate": 2.4403569678515227e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 73676384,
      "step": 34135
    },
    {
      "epoch": 5.569331158238173,
      "grad_norm": 0.1429869532585144,
      "learning_rate": 2.4396453746465912e-05,
      "loss": 0.2107,
      "num_input_tokens_seen": 73687456,
      "step": 34140
    },
    {
      "epoch": 5.570146818923328,
      "grad_norm": 0.3126033544540405,
      "learning_rate": 2.4389337863343117e-05,
      "loss": 0.0808,
      "num_input_tokens_seen": 73699424,
      "step": 34145
    },
    {
      "epoch": 5.5709624796084825,
      "grad_norm": 0.1548863798379898,
      "learning_rate": 2.4382222029723693e-05,
      "loss": 0.0059,
      "num_input_tokens_seen": 73710400,
      "step": 34150
    },
    {
      "epoch": 5.571778140293638,
      "grad_norm": 0.1327655166387558,
      "learning_rate": 2.4375106246184484e-05,
      "loss": 0.3141,
      "num_input_tokens_seen": 73721184,
      "step": 34155
    },
    {
      "epoch": 5.572593800978793,
      "grad_norm": 0.3220544457435608,
      "learning_rate": 2.4367990513302336e-05,
      "loss": 0.047,
      "num_input_tokens_seen": 73732032,
      "step": 34160
    },
    {
      "epoch": 5.573409461663948,
      "grad_norm": 4.705662727355957,
      "learning_rate": 2.4360874831654083e-05,
      "loss": 0.2368,
      "num_input_tokens_seen": 73742944,
      "step": 34165
    },
    {
      "epoch": 5.574225122349103,
      "grad_norm": 0.08677390217781067,
      "learning_rate": 2.4353759201816555e-05,
      "loss": 0.0184,
      "num_input_tokens_seen": 73753312,
      "step": 34170
    },
    {
      "epoch": 5.575040783034257,
      "grad_norm": 0.255723237991333,
      "learning_rate": 2.4346643624366586e-05,
      "loss": 0.0963,
      "num_input_tokens_seen": 73764480,
      "step": 34175
    },
    {
      "epoch": 5.575856443719413,
      "grad_norm": 4.120612621307373,
      "learning_rate": 2.4339528099881e-05,
      "loss": 0.1556,
      "num_input_tokens_seen": 73774496,
      "step": 34180
    },
    {
      "epoch": 5.576672104404568,
      "grad_norm": 0.13706041872501373,
      "learning_rate": 2.433241262893662e-05,
      "loss": 0.0919,
      "num_input_tokens_seen": 73786688,
      "step": 34185
    },
    {
      "epoch": 5.577487765089723,
      "grad_norm": 0.13830256462097168,
      "learning_rate": 2.432529721211026e-05,
      "loss": 0.134,
      "num_input_tokens_seen": 73798080,
      "step": 34190
    },
    {
      "epoch": 5.578303425774878,
      "grad_norm": 0.3326249420642853,
      "learning_rate": 2.4318181849978733e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 73808736,
      "step": 34195
    },
    {
      "epoch": 5.579119086460032,
      "grad_norm": 0.2934521436691284,
      "learning_rate": 2.4311066543118842e-05,
      "loss": 0.0188,
      "num_input_tokens_seen": 73819232,
      "step": 34200
    },
    {
      "epoch": 5.579934747145187,
      "grad_norm": 0.2670780420303345,
      "learning_rate": 2.4303951292107395e-05,
      "loss": 0.0091,
      "num_input_tokens_seen": 73829312,
      "step": 34205
    },
    {
      "epoch": 5.580750407830343,
      "grad_norm": 0.0827723890542984,
      "learning_rate": 2.4296836097521186e-05,
      "loss": 0.01,
      "num_input_tokens_seen": 73840320,
      "step": 34210
    },
    {
      "epoch": 5.581566068515498,
      "grad_norm": 0.13001371920108795,
      "learning_rate": 2.4289720959937008e-05,
      "loss": 0.0435,
      "num_input_tokens_seen": 73850464,
      "step": 34215
    },
    {
      "epoch": 5.582381729200653,
      "grad_norm": 0.05852844938635826,
      "learning_rate": 2.4282605879931647e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 73861152,
      "step": 34220
    },
    {
      "epoch": 5.583197389885807,
      "grad_norm": 0.036576732993125916,
      "learning_rate": 2.4275490858081903e-05,
      "loss": 0.0025,
      "num_input_tokens_seen": 73872288,
      "step": 34225
    },
    {
      "epoch": 5.584013050570962,
      "grad_norm": 0.10926913470029831,
      "learning_rate": 2.4268375894964544e-05,
      "loss": 0.0077,
      "num_input_tokens_seen": 73882720,
      "step": 34230
    },
    {
      "epoch": 5.584828711256117,
      "grad_norm": 0.08241516351699829,
      "learning_rate": 2.426126099115635e-05,
      "loss": 0.1167,
      "num_input_tokens_seen": 73892448,
      "step": 34235
    },
    {
      "epoch": 5.585644371941273,
      "grad_norm": 4.633983135223389,
      "learning_rate": 2.4254146147234087e-05,
      "loss": 0.2502,
      "num_input_tokens_seen": 73904224,
      "step": 34240
    },
    {
      "epoch": 5.5864600326264275,
      "grad_norm": 3.368659496307373,
      "learning_rate": 2.4247031363774523e-05,
      "loss": 0.1073,
      "num_input_tokens_seen": 73915424,
      "step": 34245
    },
    {
      "epoch": 5.587275693311582,
      "grad_norm": 0.18018129467964172,
      "learning_rate": 2.4239916641354417e-05,
      "loss": 0.1531,
      "num_input_tokens_seen": 73925344,
      "step": 34250
    },
    {
      "epoch": 5.588091353996737,
      "grad_norm": 0.1320713758468628,
      "learning_rate": 2.4232801980550523e-05,
      "loss": 0.0566,
      "num_input_tokens_seen": 73935904,
      "step": 34255
    },
    {
      "epoch": 5.588907014681892,
      "grad_norm": 0.05851376801729202,
      "learning_rate": 2.422568738193959e-05,
      "loss": 0.0033,
      "num_input_tokens_seen": 73947296,
      "step": 34260
    },
    {
      "epoch": 5.589722675367048,
      "grad_norm": 0.072971411049366,
      "learning_rate": 2.421857284609837e-05,
      "loss": 0.0053,
      "num_input_tokens_seen": 73955680,
      "step": 34265
    },
    {
      "epoch": 5.5905383360522025,
      "grad_norm": 3.2490286827087402,
      "learning_rate": 2.42114583736036e-05,
      "loss": 0.02,
      "num_input_tokens_seen": 73966304,
      "step": 34270
    },
    {
      "epoch": 5.591353996737357,
      "grad_norm": 0.03564507141709328,
      "learning_rate": 2.4204343965032015e-05,
      "loss": 0.0049,
      "num_input_tokens_seen": 73977120,
      "step": 34275
    },
    {
      "epoch": 5.592169657422512,
      "grad_norm": 0.06907259672880173,
      "learning_rate": 2.4197229620960347e-05,
      "loss": 0.0692,
      "num_input_tokens_seen": 73987648,
      "step": 34280
    },
    {
      "epoch": 5.592985318107667,
      "grad_norm": 0.8568238615989685,
      "learning_rate": 2.4190115341965316e-05,
      "loss": 0.0646,
      "num_input_tokens_seen": 73997312,
      "step": 34285
    },
    {
      "epoch": 5.593800978792823,
      "grad_norm": 0.08649122714996338,
      "learning_rate": 2.418300112862365e-05,
      "loss": 0.1882,
      "num_input_tokens_seen": 74009056,
      "step": 34290
    },
    {
      "epoch": 5.5946166394779775,
      "grad_norm": 6.20982027053833,
      "learning_rate": 2.4175886981512054e-05,
      "loss": 0.0242,
      "num_input_tokens_seen": 74019488,
      "step": 34295
    },
    {
      "epoch": 5.595432300163132,
      "grad_norm": 0.0463801734149456,
      "learning_rate": 2.416877290120724e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 74030880,
      "step": 34300
    },
    {
      "epoch": 5.596247960848287,
      "grad_norm": 0.05491442233324051,
      "learning_rate": 2.4161658888285916e-05,
      "loss": 0.1812,
      "num_input_tokens_seen": 74042080,
      "step": 34305
    },
    {
      "epoch": 5.597063621533442,
      "grad_norm": 0.015205527655780315,
      "learning_rate": 2.4154544943324772e-05,
      "loss": 0.0934,
      "num_input_tokens_seen": 74052448,
      "step": 34310
    },
    {
      "epoch": 5.597879282218597,
      "grad_norm": 11.813061714172363,
      "learning_rate": 2.414743106690051e-05,
      "loss": 0.0904,
      "num_input_tokens_seen": 74063392,
      "step": 34315
    },
    {
      "epoch": 5.598694942903752,
      "grad_norm": 14.201874732971191,
      "learning_rate": 2.41403172595898e-05,
      "loss": 0.1326,
      "num_input_tokens_seen": 74074720,
      "step": 34320
    },
    {
      "epoch": 5.599510603588907,
      "grad_norm": 4.2750959396362305,
      "learning_rate": 2.413320352196934e-05,
      "loss": 0.2027,
      "num_input_tokens_seen": 74084512,
      "step": 34325
    },
    {
      "epoch": 5.600326264274062,
      "grad_norm": 0.09404604882001877,
      "learning_rate": 2.4126089854615802e-05,
      "loss": 0.0753,
      "num_input_tokens_seen": 74095520,
      "step": 34330
    },
    {
      "epoch": 5.601141924959217,
      "grad_norm": 0.12843450903892517,
      "learning_rate": 2.411897625810586e-05,
      "loss": 0.1805,
      "num_input_tokens_seen": 74105792,
      "step": 34335
    },
    {
      "epoch": 5.601957585644372,
      "grad_norm": 0.033789101988077164,
      "learning_rate": 2.4111862733016164e-05,
      "loss": 0.0054,
      "num_input_tokens_seen": 74116224,
      "step": 34340
    },
    {
      "epoch": 5.602773246329527,
      "grad_norm": 3.307924509048462,
      "learning_rate": 2.4104749279923383e-05,
      "loss": 0.0375,
      "num_input_tokens_seen": 74127232,
      "step": 34345
    },
    {
      "epoch": 5.603588907014682,
      "grad_norm": 3.4159066677093506,
      "learning_rate": 2.409763589940417e-05,
      "loss": 0.1128,
      "num_input_tokens_seen": 74137088,
      "step": 34350
    },
    {
      "epoch": 5.604404567699837,
      "grad_norm": 0.04170740395784378,
      "learning_rate": 2.4090522592035172e-05,
      "loss": 0.0971,
      "num_input_tokens_seen": 74148128,
      "step": 34355
    },
    {
      "epoch": 5.605220228384992,
      "grad_norm": 0.06444224715232849,
      "learning_rate": 2.408340935839303e-05,
      "loss": 0.0072,
      "num_input_tokens_seen": 74158432,
      "step": 34360
    },
    {
      "epoch": 5.606035889070147,
      "grad_norm": 9.898969650268555,
      "learning_rate": 2.407629619905437e-05,
      "loss": 0.1866,
      "num_input_tokens_seen": 74169376,
      "step": 34365
    },
    {
      "epoch": 5.6068515497553015,
      "grad_norm": 4.78087043762207,
      "learning_rate": 2.406918311459583e-05,
      "loss": 0.0765,
      "num_input_tokens_seen": 74180352,
      "step": 34370
    },
    {
      "epoch": 5.607667210440457,
      "grad_norm": 0.11153988540172577,
      "learning_rate": 2.406207010559403e-05,
      "loss": 0.0068,
      "num_input_tokens_seen": 74190304,
      "step": 34375
    },
    {
      "epoch": 5.608482871125612,
      "grad_norm": 4.656994819641113,
      "learning_rate": 2.4054957172625584e-05,
      "loss": 0.2249,
      "num_input_tokens_seen": 74201504,
      "step": 34380
    },
    {
      "epoch": 5.609298531810767,
      "grad_norm": 3.565504550933838,
      "learning_rate": 2.4047844316267104e-05,
      "loss": 0.388,
      "num_input_tokens_seen": 74211968,
      "step": 34385
    },
    {
      "epoch": 5.610114192495922,
      "grad_norm": 0.014209095388650894,
      "learning_rate": 2.40407315370952e-05,
      "loss": 0.1318,
      "num_input_tokens_seen": 74223232,
      "step": 34390
    },
    {
      "epoch": 5.6109298531810765,
      "grad_norm": 3.1884076595306396,
      "learning_rate": 2.4033618835686462e-05,
      "loss": 0.1181,
      "num_input_tokens_seen": 74234272,
      "step": 34395
    },
    {
      "epoch": 5.611745513866231,
      "grad_norm": 0.039724018424749374,
      "learning_rate": 2.4026506212617485e-05,
      "loss": 0.0118,
      "num_input_tokens_seen": 74244608,
      "step": 34400
    },
    {
      "epoch": 5.612561174551386,
      "grad_norm": 4.550290107727051,
      "learning_rate": 2.4019393668464846e-05,
      "loss": 0.165,
      "num_input_tokens_seen": 74255104,
      "step": 34405
    },
    {
      "epoch": 5.613376835236542,
      "grad_norm": 0.07191260904073715,
      "learning_rate": 2.4012281203805138e-05,
      "loss": 0.008,
      "num_input_tokens_seen": 74265792,
      "step": 34410
    },
    {
      "epoch": 5.614192495921697,
      "grad_norm": 0.23330551385879517,
      "learning_rate": 2.4005168819214926e-05,
      "loss": 0.191,
      "num_input_tokens_seen": 74277120,
      "step": 34415
    },
    {
      "epoch": 5.6150081566068515,
      "grad_norm": 0.023702984675765038,
      "learning_rate": 2.3998056515270782e-05,
      "loss": 0.0083,
      "num_input_tokens_seen": 74286848,
      "step": 34420
    },
    {
      "epoch": 5.615823817292006,
      "grad_norm": 0.6659714579582214,
      "learning_rate": 2.3990944292549257e-05,
      "loss": 0.0075,
      "num_input_tokens_seen": 74296992,
      "step": 34425
    },
    {
      "epoch": 5.616639477977161,
      "grad_norm": 0.07955005764961243,
      "learning_rate": 2.3983832151626897e-05,
      "loss": 0.0786,
      "num_input_tokens_seen": 74307680,
      "step": 34430
    },
    {
      "epoch": 5.617455138662317,
      "grad_norm": 3.4622609615325928,
      "learning_rate": 2.397672009308027e-05,
      "loss": 0.3743,
      "num_input_tokens_seen": 74318368,
      "step": 34435
    },
    {
      "epoch": 5.618270799347472,
      "grad_norm": 0.051373761147260666,
      "learning_rate": 2.3969608117485906e-05,
      "loss": 0.0758,
      "num_input_tokens_seen": 74328544,
      "step": 34440
    },
    {
      "epoch": 5.6190864600326265,
      "grad_norm": 0.12117135524749756,
      "learning_rate": 2.3962496225420335e-05,
      "loss": 0.0345,
      "num_input_tokens_seen": 74339904,
      "step": 34445
    },
    {
      "epoch": 5.619902120717781,
      "grad_norm": 0.08636009693145752,
      "learning_rate": 2.3955384417460084e-05,
      "loss": 0.0128,
      "num_input_tokens_seen": 74350784,
      "step": 34450
    },
    {
      "epoch": 5.620717781402936,
      "grad_norm": 0.04565194249153137,
      "learning_rate": 2.3948272694181673e-05,
      "loss": 0.1465,
      "num_input_tokens_seen": 74361408,
      "step": 34455
    },
    {
      "epoch": 5.621533442088092,
      "grad_norm": 0.06937927007675171,
      "learning_rate": 2.3941161056161612e-05,
      "loss": 0.2219,
      "num_input_tokens_seen": 74371904,
      "step": 34460
    },
    {
      "epoch": 5.622349102773247,
      "grad_norm": 0.06344190984964371,
      "learning_rate": 2.393404950397641e-05,
      "loss": 0.2713,
      "num_input_tokens_seen": 74383232,
      "step": 34465
    },
    {
      "epoch": 5.623164763458401,
      "grad_norm": 0.1508975625038147,
      "learning_rate": 2.3926938038202565e-05,
      "loss": 0.0963,
      "num_input_tokens_seen": 74394240,
      "step": 34470
    },
    {
      "epoch": 5.623980424143556,
      "grad_norm": 0.060986749827861786,
      "learning_rate": 2.3919826659416564e-05,
      "loss": 0.0041,
      "num_input_tokens_seen": 74404160,
      "step": 34475
    },
    {
      "epoch": 5.624796084828711,
      "grad_norm": 9.748682975769043,
      "learning_rate": 2.3912715368194895e-05,
      "loss": 0.3195,
      "num_input_tokens_seen": 74415904,
      "step": 34480
    },
    {
      "epoch": 5.625611745513866,
      "grad_norm": 0.14794349670410156,
      "learning_rate": 2.3905604165114038e-05,
      "loss": 0.044,
      "num_input_tokens_seen": 74426336,
      "step": 34485
    },
    {
      "epoch": 5.626427406199021,
      "grad_norm": 0.2329591065645218,
      "learning_rate": 2.3898493050750453e-05,
      "loss": 0.0738,
      "num_input_tokens_seen": 74434592,
      "step": 34490
    },
    {
      "epoch": 5.627243066884176,
      "grad_norm": 0.11005127429962158,
      "learning_rate": 2.3891382025680616e-05,
      "loss": 0.0661,
      "num_input_tokens_seen": 74446464,
      "step": 34495
    },
    {
      "epoch": 5.628058727569331,
      "grad_norm": 0.14500364661216736,
      "learning_rate": 2.388427109048098e-05,
      "loss": 0.1189,
      "num_input_tokens_seen": 74457056,
      "step": 34500
    },
    {
      "epoch": 5.628874388254486,
      "grad_norm": 1.0034968852996826,
      "learning_rate": 2.3877160245727988e-05,
      "loss": 0.0059,
      "num_input_tokens_seen": 74467584,
      "step": 34505
    },
    {
      "epoch": 5.629690048939641,
      "grad_norm": 0.17740269005298615,
      "learning_rate": 2.3870049491998082e-05,
      "loss": 0.0201,
      "num_input_tokens_seen": 74477376,
      "step": 34510
    },
    {
      "epoch": 5.630505709624796,
      "grad_norm": 0.6840254664421082,
      "learning_rate": 2.3862938829867698e-05,
      "loss": 0.229,
      "num_input_tokens_seen": 74487648,
      "step": 34515
    },
    {
      "epoch": 5.631321370309951,
      "grad_norm": 4.9663166999816895,
      "learning_rate": 2.3855828259913262e-05,
      "loss": 0.0929,
      "num_input_tokens_seen": 74498496,
      "step": 34520
    },
    {
      "epoch": 5.632137030995106,
      "grad_norm": 13.157565116882324,
      "learning_rate": 2.3848717782711194e-05,
      "loss": 0.1202,
      "num_input_tokens_seen": 74509664,
      "step": 34525
    },
    {
      "epoch": 5.632952691680261,
      "grad_norm": 0.19897565245628357,
      "learning_rate": 2.3841607398837902e-05,
      "loss": 0.0507,
      "num_input_tokens_seen": 74519808,
      "step": 34530
    },
    {
      "epoch": 5.633768352365416,
      "grad_norm": 2.7781496047973633,
      "learning_rate": 2.3834497108869797e-05,
      "loss": 0.0264,
      "num_input_tokens_seen": 74531840,
      "step": 34535
    },
    {
      "epoch": 5.634584013050571,
      "grad_norm": 0.18439719080924988,
      "learning_rate": 2.3827386913383254e-05,
      "loss": 0.1879,
      "num_input_tokens_seen": 74543168,
      "step": 34540
    },
    {
      "epoch": 5.635399673735726,
      "grad_norm": 0.28179651498794556,
      "learning_rate": 2.3820276812954688e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 74553792,
      "step": 34545
    },
    {
      "epoch": 5.636215334420881,
      "grad_norm": 0.07636716961860657,
      "learning_rate": 2.3813166808160472e-05,
      "loss": 0.085,
      "num_input_tokens_seen": 74565312,
      "step": 34550
    },
    {
      "epoch": 5.637030995106036,
      "grad_norm": 0.07693570852279663,
      "learning_rate": 2.3806056899576978e-05,
      "loss": 0.3763,
      "num_input_tokens_seen": 74576256,
      "step": 34555
    },
    {
      "epoch": 5.637846655791191,
      "grad_norm": 0.07625547051429749,
      "learning_rate": 2.3798947087780567e-05,
      "loss": 0.1703,
      "num_input_tokens_seen": 74586560,
      "step": 34560
    },
    {
      "epoch": 5.638662316476346,
      "grad_norm": 7.158646106719971,
      "learning_rate": 2.37918373733476e-05,
      "loss": 0.0801,
      "num_input_tokens_seen": 74597792,
      "step": 34565
    },
    {
      "epoch": 5.6394779771615005,
      "grad_norm": 33.091461181640625,
      "learning_rate": 2.3784727756854425e-05,
      "loss": 0.3652,
      "num_input_tokens_seen": 74609504,
      "step": 34570
    },
    {
      "epoch": 5.640293637846656,
      "grad_norm": 0.06495015323162079,
      "learning_rate": 2.377761823887738e-05,
      "loss": 0.1862,
      "num_input_tokens_seen": 74620160,
      "step": 34575
    },
    {
      "epoch": 5.641109298531811,
      "grad_norm": 23.01750946044922,
      "learning_rate": 2.3770508819992807e-05,
      "loss": 0.0941,
      "num_input_tokens_seen": 74631072,
      "step": 34580
    },
    {
      "epoch": 5.641924959216966,
      "grad_norm": 0.04679006710648537,
      "learning_rate": 2.376339950077703e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 74641088,
      "step": 34585
    },
    {
      "epoch": 5.642740619902121,
      "grad_norm": 0.220841184258461,
      "learning_rate": 2.3756290281806358e-05,
      "loss": 0.0244,
      "num_input_tokens_seen": 74650560,
      "step": 34590
    },
    {
      "epoch": 5.643556280587275,
      "grad_norm": 0.10817943513393402,
      "learning_rate": 2.3749181163657114e-05,
      "loss": 0.1005,
      "num_input_tokens_seen": 74660608,
      "step": 34595
    },
    {
      "epoch": 5.64437194127243,
      "grad_norm": 0.15172620117664337,
      "learning_rate": 2.3742072146905587e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 74672064,
      "step": 34600
    },
    {
      "epoch": 5.645187601957586,
      "grad_norm": 4.3850908279418945,
      "learning_rate": 2.3734963232128072e-05,
      "loss": 0.0824,
      "num_input_tokens_seen": 74683200,
      "step": 34605
    },
    {
      "epoch": 5.646003262642741,
      "grad_norm": 0.14718104898929596,
      "learning_rate": 2.372785441990086e-05,
      "loss": 0.1096,
      "num_input_tokens_seen": 74694016,
      "step": 34610
    },
    {
      "epoch": 5.646818923327896,
      "grad_norm": 1.600773811340332,
      "learning_rate": 2.3720745710800225e-05,
      "loss": 0.4462,
      "num_input_tokens_seen": 74704096,
      "step": 34615
    },
    {
      "epoch": 5.64763458401305,
      "grad_norm": 0.0689341202378273,
      "learning_rate": 2.371363710540243e-05,
      "loss": 0.1558,
      "num_input_tokens_seen": 74714528,
      "step": 34620
    },
    {
      "epoch": 5.648450244698205,
      "grad_norm": 0.4660826027393341,
      "learning_rate": 2.370652860428374e-05,
      "loss": 0.281,
      "num_input_tokens_seen": 74725888,
      "step": 34625
    },
    {
      "epoch": 5.649265905383361,
      "grad_norm": 0.4790792167186737,
      "learning_rate": 2.3699420208020403e-05,
      "loss": 0.0261,
      "num_input_tokens_seen": 74736384,
      "step": 34630
    },
    {
      "epoch": 5.650081566068516,
      "grad_norm": 3.3586387634277344,
      "learning_rate": 2.3692311917188658e-05,
      "loss": 0.2205,
      "num_input_tokens_seen": 74748064,
      "step": 34635
    },
    {
      "epoch": 5.650897226753671,
      "grad_norm": 0.03131727874279022,
      "learning_rate": 2.3685203732364754e-05,
      "loss": 0.0594,
      "num_input_tokens_seen": 74758720,
      "step": 34640
    },
    {
      "epoch": 5.651712887438825,
      "grad_norm": 0.06394805014133453,
      "learning_rate": 2.3678095654124893e-05,
      "loss": 0.0058,
      "num_input_tokens_seen": 74770208,
      "step": 34645
    },
    {
      "epoch": 5.65252854812398,
      "grad_norm": 0.09772054105997086,
      "learning_rate": 2.3670987683045317e-05,
      "loss": 0.0046,
      "num_input_tokens_seen": 74781568,
      "step": 34650
    },
    {
      "epoch": 5.653344208809135,
      "grad_norm": 0.07734900712966919,
      "learning_rate": 2.366387981970222e-05,
      "loss": 0.1028,
      "num_input_tokens_seen": 74792608,
      "step": 34655
    },
    {
      "epoch": 5.654159869494291,
      "grad_norm": 0.30056944489479065,
      "learning_rate": 2.36567720646718e-05,
      "loss": 0.0053,
      "num_input_tokens_seen": 74801760,
      "step": 34660
    },
    {
      "epoch": 5.6549755301794455,
      "grad_norm": 0.09153929352760315,
      "learning_rate": 2.3649664418530258e-05,
      "loss": 0.0277,
      "num_input_tokens_seen": 74812256,
      "step": 34665
    },
    {
      "epoch": 5.6557911908646,
      "grad_norm": 0.07793822884559631,
      "learning_rate": 2.364255688185377e-05,
      "loss": 0.0704,
      "num_input_tokens_seen": 74822848,
      "step": 34670
    },
    {
      "epoch": 5.656606851549755,
      "grad_norm": 0.06883666664361954,
      "learning_rate": 2.3635449455218506e-05,
      "loss": 0.0998,
      "num_input_tokens_seen": 74832992,
      "step": 34675
    },
    {
      "epoch": 5.65742251223491,
      "grad_norm": 0.08330200612545013,
      "learning_rate": 2.3628342139200636e-05,
      "loss": 0.0607,
      "num_input_tokens_seen": 74843136,
      "step": 34680
    },
    {
      "epoch": 5.658238172920065,
      "grad_norm": 0.017148641869425774,
      "learning_rate": 2.362123493437631e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 74852800,
      "step": 34685
    },
    {
      "epoch": 5.6590538336052205,
      "grad_norm": 0.19309087097644806,
      "learning_rate": 2.3614127841321677e-05,
      "loss": 0.073,
      "num_input_tokens_seen": 74864512,
      "step": 34690
    },
    {
      "epoch": 5.659869494290375,
      "grad_norm": 6.053650856018066,
      "learning_rate": 2.3607020860612872e-05,
      "loss": 0.0083,
      "num_input_tokens_seen": 74875936,
      "step": 34695
    },
    {
      "epoch": 5.66068515497553,
      "grad_norm": 0.07050441205501556,
      "learning_rate": 2.3599913992826023e-05,
      "loss": 0.1225,
      "num_input_tokens_seen": 74885504,
      "step": 34700
    },
    {
      "epoch": 5.661500815660685,
      "grad_norm": 10.587239265441895,
      "learning_rate": 2.3592807238537253e-05,
      "loss": 0.1015,
      "num_input_tokens_seen": 74897632,
      "step": 34705
    },
    {
      "epoch": 5.66231647634584,
      "grad_norm": 5.930193901062012,
      "learning_rate": 2.3585700598322665e-05,
      "loss": 0.0922,
      "num_input_tokens_seen": 74908672,
      "step": 34710
    },
    {
      "epoch": 5.6631321370309955,
      "grad_norm": 3.4469075202941895,
      "learning_rate": 2.3578594072758363e-05,
      "loss": 0.1463,
      "num_input_tokens_seen": 74919104,
      "step": 34715
    },
    {
      "epoch": 5.66394779771615,
      "grad_norm": 10.311992645263672,
      "learning_rate": 2.3571487662420433e-05,
      "loss": 0.0571,
      "num_input_tokens_seen": 74929952,
      "step": 34720
    },
    {
      "epoch": 5.664763458401305,
      "grad_norm": 11.221634864807129,
      "learning_rate": 2.3564381367884965e-05,
      "loss": 0.1078,
      "num_input_tokens_seen": 74941760,
      "step": 34725
    },
    {
      "epoch": 5.66557911908646,
      "grad_norm": 0.6332096457481384,
      "learning_rate": 2.3557275189728032e-05,
      "loss": 0.0771,
      "num_input_tokens_seen": 74952000,
      "step": 34730
    },
    {
      "epoch": 5.666394779771615,
      "grad_norm": 0.05423618480563164,
      "learning_rate": 2.3550169128525688e-05,
      "loss": 0.0086,
      "num_input_tokens_seen": 74963008,
      "step": 34735
    },
    {
      "epoch": 5.6672104404567705,
      "grad_norm": 1.9471640586853027,
      "learning_rate": 2.3543063184853994e-05,
      "loss": 0.0095,
      "num_input_tokens_seen": 74973408,
      "step": 34740
    },
    {
      "epoch": 5.668026101141925,
      "grad_norm": 0.691474974155426,
      "learning_rate": 2.353595735928899e-05,
      "loss": 0.4548,
      "num_input_tokens_seen": 74983744,
      "step": 34745
    },
    {
      "epoch": 5.66884176182708,
      "grad_norm": 0.10274942964315414,
      "learning_rate": 2.3528851652406697e-05,
      "loss": 0.1179,
      "num_input_tokens_seen": 74993952,
      "step": 34750
    },
    {
      "epoch": 5.669657422512235,
      "grad_norm": 0.07681412994861603,
      "learning_rate": 2.3521746064783168e-05,
      "loss": 0.0111,
      "num_input_tokens_seen": 75004128,
      "step": 34755
    },
    {
      "epoch": 5.67047308319739,
      "grad_norm": 2.1961638927459717,
      "learning_rate": 2.3514640596994404e-05,
      "loss": 0.0706,
      "num_input_tokens_seen": 75014496,
      "step": 34760
    },
    {
      "epoch": 5.671288743882545,
      "grad_norm": 0.0980810821056366,
      "learning_rate": 2.350753524961641e-05,
      "loss": 0.3276,
      "num_input_tokens_seen": 75024704,
      "step": 34765
    },
    {
      "epoch": 5.672104404567699,
      "grad_norm": 0.07521381974220276,
      "learning_rate": 2.3500430023225174e-05,
      "loss": 0.0969,
      "num_input_tokens_seen": 75035488,
      "step": 34770
    },
    {
      "epoch": 5.672920065252855,
      "grad_norm": 0.062481507658958435,
      "learning_rate": 2.3493324918396696e-05,
      "loss": 0.0655,
      "num_input_tokens_seen": 75048192,
      "step": 34775
    },
    {
      "epoch": 5.67373572593801,
      "grad_norm": 0.060710132122039795,
      "learning_rate": 2.3486219935706944e-05,
      "loss": 0.1326,
      "num_input_tokens_seen": 75059680,
      "step": 34780
    },
    {
      "epoch": 5.674551386623165,
      "grad_norm": 6.397144794464111,
      "learning_rate": 2.3479115075731886e-05,
      "loss": 0.0355,
      "num_input_tokens_seen": 75069696,
      "step": 34785
    },
    {
      "epoch": 5.6753670473083195,
      "grad_norm": 0.37986424565315247,
      "learning_rate": 2.3472010339047474e-05,
      "loss": 0.292,
      "num_input_tokens_seen": 75080512,
      "step": 34790
    },
    {
      "epoch": 5.676182707993474,
      "grad_norm": 1.0515350103378296,
      "learning_rate": 2.3464905726229657e-05,
      "loss": 0.1161,
      "num_input_tokens_seen": 75092000,
      "step": 34795
    },
    {
      "epoch": 5.67699836867863,
      "grad_norm": 0.05977501720190048,
      "learning_rate": 2.3457801237854367e-05,
      "loss": 0.114,
      "num_input_tokens_seen": 75102208,
      "step": 34800
    },
    {
      "epoch": 5.677814029363785,
      "grad_norm": 0.07490808516740799,
      "learning_rate": 2.345069687449754e-05,
      "loss": 0.1285,
      "num_input_tokens_seen": 75112960,
      "step": 34805
    },
    {
      "epoch": 5.67862969004894,
      "grad_norm": 0.1772504597902298,
      "learning_rate": 2.3443592636735085e-05,
      "loss": 0.0438,
      "num_input_tokens_seen": 75125824,
      "step": 34810
    },
    {
      "epoch": 5.6794453507340945,
      "grad_norm": 0.06352439522743225,
      "learning_rate": 2.3436488525142906e-05,
      "loss": 0.151,
      "num_input_tokens_seen": 75136896,
      "step": 34815
    },
    {
      "epoch": 5.680261011419249,
      "grad_norm": 8.625276565551758,
      "learning_rate": 2.3429384540296902e-05,
      "loss": 0.0414,
      "num_input_tokens_seen": 75146528,
      "step": 34820
    },
    {
      "epoch": 5.681076672104405,
      "grad_norm": 0.10543076694011688,
      "learning_rate": 2.3422280682772953e-05,
      "loss": 0.0634,
      "num_input_tokens_seen": 75157184,
      "step": 34825
    },
    {
      "epoch": 5.68189233278956,
      "grad_norm": 9.880266189575195,
      "learning_rate": 2.341517695314694e-05,
      "loss": 0.0378,
      "num_input_tokens_seen": 75166016,
      "step": 34830
    },
    {
      "epoch": 5.682707993474715,
      "grad_norm": 0.06467192620038986,
      "learning_rate": 2.3408073351994726e-05,
      "loss": 0.2627,
      "num_input_tokens_seen": 75177824,
      "step": 34835
    },
    {
      "epoch": 5.6835236541598695,
      "grad_norm": 0.27344152331352234,
      "learning_rate": 2.340096987989216e-05,
      "loss": 0.1237,
      "num_input_tokens_seen": 75188128,
      "step": 34840
    },
    {
      "epoch": 5.684339314845024,
      "grad_norm": 0.054349880665540695,
      "learning_rate": 2.3393866537415093e-05,
      "loss": 0.064,
      "num_input_tokens_seen": 75198016,
      "step": 34845
    },
    {
      "epoch": 5.685154975530179,
      "grad_norm": 0.05116664618253708,
      "learning_rate": 2.3386763325139353e-05,
      "loss": 0.0093,
      "num_input_tokens_seen": 75208768,
      "step": 34850
    },
    {
      "epoch": 5.685970636215334,
      "grad_norm": 0.057738080620765686,
      "learning_rate": 2.337966024364076e-05,
      "loss": 0.0105,
      "num_input_tokens_seen": 75220064,
      "step": 34855
    },
    {
      "epoch": 5.68678629690049,
      "grad_norm": 3.193617582321167,
      "learning_rate": 2.337255729349512e-05,
      "loss": 0.0188,
      "num_input_tokens_seen": 75231552,
      "step": 34860
    },
    {
      "epoch": 5.6876019575856445,
      "grad_norm": 1.0187528133392334,
      "learning_rate": 2.3365454475278257e-05,
      "loss": 0.0601,
      "num_input_tokens_seen": 75242080,
      "step": 34865
    },
    {
      "epoch": 5.688417618270799,
      "grad_norm": 0.03958727419376373,
      "learning_rate": 2.3358351789565945e-05,
      "loss": 0.0723,
      "num_input_tokens_seen": 75253312,
      "step": 34870
    },
    {
      "epoch": 5.689233278955954,
      "grad_norm": 0.12234574556350708,
      "learning_rate": 2.335124923693397e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 75265376,
      "step": 34875
    },
    {
      "epoch": 5.690048939641109,
      "grad_norm": 0.1843361258506775,
      "learning_rate": 2.334414681795809e-05,
      "loss": 0.2009,
      "num_input_tokens_seen": 75275872,
      "step": 34880
    },
    {
      "epoch": 5.690864600326265,
      "grad_norm": 0.19655779004096985,
      "learning_rate": 2.3337044533214068e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 75288096,
      "step": 34885
    },
    {
      "epoch": 5.691680261011419,
      "grad_norm": 0.36933252215385437,
      "learning_rate": 2.3329942383277665e-05,
      "loss": 0.0082,
      "num_input_tokens_seen": 75298112,
      "step": 34890
    },
    {
      "epoch": 5.692495921696574,
      "grad_norm": 0.1846073865890503,
      "learning_rate": 2.3322840368724598e-05,
      "loss": 0.1659,
      "num_input_tokens_seen": 75308992,
      "step": 34895
    },
    {
      "epoch": 5.693311582381729,
      "grad_norm": 0.1440029889345169,
      "learning_rate": 2.3315738490130606e-05,
      "loss": 0.1509,
      "num_input_tokens_seen": 75320352,
      "step": 34900
    },
    {
      "epoch": 5.694127243066884,
      "grad_norm": 0.1343105584383011,
      "learning_rate": 2.3308636748071395e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 75333056,
      "step": 34905
    },
    {
      "epoch": 5.69494290375204,
      "grad_norm": 0.07324358075857162,
      "learning_rate": 2.3301535143122675e-05,
      "loss": 0.1334,
      "num_input_tokens_seen": 75345248,
      "step": 34910
    },
    {
      "epoch": 5.695758564437194,
      "grad_norm": 10.516416549682617,
      "learning_rate": 2.3294433675860134e-05,
      "loss": 0.2258,
      "num_input_tokens_seen": 75356576,
      "step": 34915
    },
    {
      "epoch": 5.696574225122349,
      "grad_norm": 0.055170800536870956,
      "learning_rate": 2.328733234685945e-05,
      "loss": 0.0173,
      "num_input_tokens_seen": 75368000,
      "step": 34920
    },
    {
      "epoch": 5.697389885807504,
      "grad_norm": 1.2303236722946167,
      "learning_rate": 2.3280231156696297e-05,
      "loss": 0.019,
      "num_input_tokens_seen": 75378208,
      "step": 34925
    },
    {
      "epoch": 5.698205546492659,
      "grad_norm": 6.7660231590271,
      "learning_rate": 2.3273130105946333e-05,
      "loss": 0.087,
      "num_input_tokens_seen": 75390112,
      "step": 34930
    },
    {
      "epoch": 5.699021207177814,
      "grad_norm": 0.06041080877184868,
      "learning_rate": 2.3266029195185204e-05,
      "loss": 0.0052,
      "num_input_tokens_seen": 75401824,
      "step": 34935
    },
    {
      "epoch": 5.699836867862969,
      "grad_norm": 2.1426451206207275,
      "learning_rate": 2.3258928424988548e-05,
      "loss": 0.1177,
      "num_input_tokens_seen": 75412384,
      "step": 34940
    },
    {
      "epoch": 5.700652528548124,
      "grad_norm": 0.7164055109024048,
      "learning_rate": 2.325182779593198e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 75423648,
      "step": 34945
    },
    {
      "epoch": 5.701468189233279,
      "grad_norm": 5.581310272216797,
      "learning_rate": 2.3244727308591126e-05,
      "loss": 0.2702,
      "num_input_tokens_seen": 75435232,
      "step": 34950
    },
    {
      "epoch": 5.702283849918434,
      "grad_norm": 1.0950038433074951,
      "learning_rate": 2.3237626963541588e-05,
      "loss": 0.0081,
      "num_input_tokens_seen": 75445952,
      "step": 34955
    },
    {
      "epoch": 5.703099510603589,
      "grad_norm": 0.08078707754611969,
      "learning_rate": 2.3230526761358944e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 75457696,
      "step": 34960
    },
    {
      "epoch": 5.7039151712887435,
      "grad_norm": 0.10439500957727432,
      "learning_rate": 2.3223426702618776e-05,
      "loss": 0.0796,
      "num_input_tokens_seen": 75467424,
      "step": 34965
    },
    {
      "epoch": 5.704730831973899,
      "grad_norm": 0.06901131570339203,
      "learning_rate": 2.3216326787896652e-05,
      "loss": 0.1305,
      "num_input_tokens_seen": 75476928,
      "step": 34970
    },
    {
      "epoch": 5.705546492659054,
      "grad_norm": 11.365278244018555,
      "learning_rate": 2.3209227017768137e-05,
      "loss": 0.0197,
      "num_input_tokens_seen": 75486624,
      "step": 34975
    },
    {
      "epoch": 5.706362153344209,
      "grad_norm": 0.10638049244880676,
      "learning_rate": 2.3202127392808768e-05,
      "loss": 0.1906,
      "num_input_tokens_seen": 75497888,
      "step": 34980
    },
    {
      "epoch": 5.707177814029364,
      "grad_norm": 0.10540467500686646,
      "learning_rate": 2.319502791359407e-05,
      "loss": 0.0617,
      "num_input_tokens_seen": 75508352,
      "step": 34985
    },
    {
      "epoch": 5.7079934747145185,
      "grad_norm": 0.16034981608390808,
      "learning_rate": 2.3187928580699573e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 75520416,
      "step": 34990
    },
    {
      "epoch": 5.708809135399674,
      "grad_norm": 0.09266732633113861,
      "learning_rate": 2.3180829394700775e-05,
      "loss": 0.1377,
      "num_input_tokens_seen": 75531168,
      "step": 34995
    },
    {
      "epoch": 5.709624796084829,
      "grad_norm": 0.029344908893108368,
      "learning_rate": 2.317373035617318e-05,
      "loss": 0.0652,
      "num_input_tokens_seen": 75541632,
      "step": 35000
    },
    {
      "epoch": 5.710440456769984,
      "grad_norm": 0.09086069464683533,
      "learning_rate": 2.3166631465692264e-05,
      "loss": 0.07,
      "num_input_tokens_seen": 75552224,
      "step": 35005
    },
    {
      "epoch": 5.711256117455139,
      "grad_norm": 0.066799096763134,
      "learning_rate": 2.3159532723833508e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 75562080,
      "step": 35010
    },
    {
      "epoch": 5.712071778140293,
      "grad_norm": 0.09885165840387344,
      "learning_rate": 2.3152434131172368e-05,
      "loss": 0.0062,
      "num_input_tokens_seen": 75571616,
      "step": 35015
    },
    {
      "epoch": 5.712887438825448,
      "grad_norm": 0.08804667741060257,
      "learning_rate": 2.3145335688284288e-05,
      "loss": 0.0512,
      "num_input_tokens_seen": 75581952,
      "step": 35020
    },
    {
      "epoch": 5.713703099510604,
      "grad_norm": 0.07176681607961655,
      "learning_rate": 2.3138237395744712e-05,
      "loss": 0.0781,
      "num_input_tokens_seen": 75592448,
      "step": 35025
    },
    {
      "epoch": 5.714518760195759,
      "grad_norm": 0.2840748727321625,
      "learning_rate": 2.313113925412905e-05,
      "loss": 0.0096,
      "num_input_tokens_seen": 75603872,
      "step": 35030
    },
    {
      "epoch": 5.715334420880914,
      "grad_norm": 0.04150667041540146,
      "learning_rate": 2.312404126401273e-05,
      "loss": 0.1128,
      "num_input_tokens_seen": 75614752,
      "step": 35035
    },
    {
      "epoch": 5.716150081566068,
      "grad_norm": 0.07693462073802948,
      "learning_rate": 2.3116943425971144e-05,
      "loss": 0.004,
      "num_input_tokens_seen": 75625600,
      "step": 35040
    },
    {
      "epoch": 5.716965742251223,
      "grad_norm": 5.008595943450928,
      "learning_rate": 2.3109845740579676e-05,
      "loss": 0.2178,
      "num_input_tokens_seen": 75635648,
      "step": 35045
    },
    {
      "epoch": 5.717781402936378,
      "grad_norm": 0.06779158115386963,
      "learning_rate": 2.3102748208413706e-05,
      "loss": 0.089,
      "num_input_tokens_seen": 75647680,
      "step": 35050
    },
    {
      "epoch": 5.718597063621534,
      "grad_norm": 0.12235770374536514,
      "learning_rate": 2.3095650830048595e-05,
      "loss": 0.0835,
      "num_input_tokens_seen": 75658080,
      "step": 35055
    },
    {
      "epoch": 5.719412724306689,
      "grad_norm": 0.05873510614037514,
      "learning_rate": 2.3088553606059686e-05,
      "loss": 0.1574,
      "num_input_tokens_seen": 75669920,
      "step": 35060
    },
    {
      "epoch": 5.720228384991843,
      "grad_norm": 0.04165768623352051,
      "learning_rate": 2.308145653702232e-05,
      "loss": 0.0149,
      "num_input_tokens_seen": 75681792,
      "step": 35065
    },
    {
      "epoch": 5.721044045676998,
      "grad_norm": 0.04350676387548447,
      "learning_rate": 2.307435962351181e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 75692992,
      "step": 35070
    },
    {
      "epoch": 5.721859706362153,
      "grad_norm": 0.036871396005153656,
      "learning_rate": 2.3067262866103492e-05,
      "loss": 0.0045,
      "num_input_tokens_seen": 75704608,
      "step": 35075
    },
    {
      "epoch": 5.722675367047309,
      "grad_norm": 0.3714873790740967,
      "learning_rate": 2.3060166265372654e-05,
      "loss": 0.0238,
      "num_input_tokens_seen": 75715328,
      "step": 35080
    },
    {
      "epoch": 5.7234910277324635,
      "grad_norm": 3.1402053833007812,
      "learning_rate": 2.3053069821894578e-05,
      "loss": 0.1874,
      "num_input_tokens_seen": 75725376,
      "step": 35085
    },
    {
      "epoch": 5.724306688417618,
      "grad_norm": 0.044017672538757324,
      "learning_rate": 2.3045973536244543e-05,
      "loss": 0.0289,
      "num_input_tokens_seen": 75736512,
      "step": 35090
    },
    {
      "epoch": 5.725122349102773,
      "grad_norm": 0.032331086695194244,
      "learning_rate": 2.303887740899781e-05,
      "loss": 0.0105,
      "num_input_tokens_seen": 75747552,
      "step": 35095
    },
    {
      "epoch": 5.725938009787928,
      "grad_norm": 2.916252613067627,
      "learning_rate": 2.3031781440729623e-05,
      "loss": 0.1272,
      "num_input_tokens_seen": 75759136,
      "step": 35100
    },
    {
      "epoch": 5.726753670473083,
      "grad_norm": 0.052721064537763596,
      "learning_rate": 2.3024685632015218e-05,
      "loss": 0.1054,
      "num_input_tokens_seen": 75770208,
      "step": 35105
    },
    {
      "epoch": 5.7275693311582385,
      "grad_norm": 0.08456026017665863,
      "learning_rate": 2.3017589983429817e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 75780800,
      "step": 35110
    },
    {
      "epoch": 5.728384991843393,
      "grad_norm": 0.09905166923999786,
      "learning_rate": 2.301049449554863e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 75791328,
      "step": 35115
    },
    {
      "epoch": 5.729200652528548,
      "grad_norm": 0.20791390538215637,
      "learning_rate": 2.3003399168946855e-05,
      "loss": 0.3981,
      "num_input_tokens_seen": 75802688,
      "step": 35120
    },
    {
      "epoch": 5.730016313213703,
      "grad_norm": 0.06286616623401642,
      "learning_rate": 2.2996304004199677e-05,
      "loss": 0.1911,
      "num_input_tokens_seen": 75812640,
      "step": 35125
    },
    {
      "epoch": 5.730831973898858,
      "grad_norm": 0.29400545358657837,
      "learning_rate": 2.298920900188226e-05,
      "loss": 0.0224,
      "num_input_tokens_seen": 75823680,
      "step": 35130
    },
    {
      "epoch": 5.731647634584013,
      "grad_norm": 0.09266781806945801,
      "learning_rate": 2.2982114162569766e-05,
      "loss": 0.0076,
      "num_input_tokens_seen": 75834208,
      "step": 35135
    },
    {
      "epoch": 5.732463295269168,
      "grad_norm": 0.027921630069613457,
      "learning_rate": 2.2975019486837334e-05,
      "loss": 0.0931,
      "num_input_tokens_seen": 75846432,
      "step": 35140
    },
    {
      "epoch": 5.733278955954323,
      "grad_norm": 0.08629529178142548,
      "learning_rate": 2.29679249752601e-05,
      "loss": 0.202,
      "num_input_tokens_seen": 75856960,
      "step": 35145
    },
    {
      "epoch": 5.734094616639478,
      "grad_norm": 0.16682440042495728,
      "learning_rate": 2.2960830628413175e-05,
      "loss": 0.0054,
      "num_input_tokens_seen": 75868256,
      "step": 35150
    },
    {
      "epoch": 5.734910277324633,
      "grad_norm": 0.09931493550539017,
      "learning_rate": 2.295373644687167e-05,
      "loss": 0.0618,
      "num_input_tokens_seen": 75879808,
      "step": 35155
    },
    {
      "epoch": 5.735725938009788,
      "grad_norm": 5.3806843757629395,
      "learning_rate": 2.294664243121067e-05,
      "loss": 0.3896,
      "num_input_tokens_seen": 75891168,
      "step": 35160
    },
    {
      "epoch": 5.736541598694943,
      "grad_norm": 0.0984605923295021,
      "learning_rate": 2.2939548582005253e-05,
      "loss": 0.2172,
      "num_input_tokens_seen": 75902080,
      "step": 35165
    },
    {
      "epoch": 5.737357259380098,
      "grad_norm": 1.2241793870925903,
      "learning_rate": 2.293245489983048e-05,
      "loss": 0.014,
      "num_input_tokens_seen": 75913568,
      "step": 35170
    },
    {
      "epoch": 5.738172920065253,
      "grad_norm": 0.05884980410337448,
      "learning_rate": 2.2925361385261402e-05,
      "loss": 0.1283,
      "num_input_tokens_seen": 75923552,
      "step": 35175
    },
    {
      "epoch": 5.738988580750408,
      "grad_norm": 14.16159439086914,
      "learning_rate": 2.2918268038873055e-05,
      "loss": 0.1801,
      "num_input_tokens_seen": 75933856,
      "step": 35180
    },
    {
      "epoch": 5.739804241435563,
      "grad_norm": 0.08795402199029922,
      "learning_rate": 2.291117486124047e-05,
      "loss": 0.1888,
      "num_input_tokens_seen": 75944896,
      "step": 35185
    },
    {
      "epoch": 5.740619902120718,
      "grad_norm": 2.2796897888183594,
      "learning_rate": 2.290408185293865e-05,
      "loss": 0.1177,
      "num_input_tokens_seen": 75954880,
      "step": 35190
    },
    {
      "epoch": 5.741435562805873,
      "grad_norm": 0.11136547476053238,
      "learning_rate": 2.2896989014542584e-05,
      "loss": 0.0575,
      "num_input_tokens_seen": 75964800,
      "step": 35195
    },
    {
      "epoch": 5.742251223491028,
      "grad_norm": 5.454838752746582,
      "learning_rate": 2.2889896346627256e-05,
      "loss": 0.1566,
      "num_input_tokens_seen": 75975264,
      "step": 35200
    },
    {
      "epoch": 5.743066884176183,
      "grad_norm": 0.059465523809194565,
      "learning_rate": 2.2882803849767646e-05,
      "loss": 0.008,
      "num_input_tokens_seen": 75986976,
      "step": 35205
    },
    {
      "epoch": 5.7438825448613375,
      "grad_norm": 0.1806742250919342,
      "learning_rate": 2.2875711524538697e-05,
      "loss": 0.0201,
      "num_input_tokens_seen": 75998752,
      "step": 35210
    },
    {
      "epoch": 5.744698205546492,
      "grad_norm": 0.9701160788536072,
      "learning_rate": 2.2868619371515348e-05,
      "loss": 0.0964,
      "num_input_tokens_seen": 76010688,
      "step": 35215
    },
    {
      "epoch": 5.745513866231647,
      "grad_norm": 2.5485148429870605,
      "learning_rate": 2.2861527391272526e-05,
      "loss": 0.2846,
      "num_input_tokens_seen": 76021408,
      "step": 35220
    },
    {
      "epoch": 5.746329526916803,
      "grad_norm": 0.0871826633810997,
      "learning_rate": 2.285443558438515e-05,
      "loss": 0.1911,
      "num_input_tokens_seen": 76032480,
      "step": 35225
    },
    {
      "epoch": 5.747145187601958,
      "grad_norm": 2.6616318225860596,
      "learning_rate": 2.2847343951428106e-05,
      "loss": 0.1337,
      "num_input_tokens_seen": 76043072,
      "step": 35230
    },
    {
      "epoch": 5.7479608482871125,
      "grad_norm": 0.1521041989326477,
      "learning_rate": 2.284025249297629e-05,
      "loss": 0.1055,
      "num_input_tokens_seen": 76054208,
      "step": 35235
    },
    {
      "epoch": 5.748776508972267,
      "grad_norm": 0.1400117129087448,
      "learning_rate": 2.2833161209604557e-05,
      "loss": 0.0237,
      "num_input_tokens_seen": 76066080,
      "step": 35240
    },
    {
      "epoch": 5.749592169657422,
      "grad_norm": 8.059832572937012,
      "learning_rate": 2.2826070101887777e-05,
      "loss": 0.11,
      "num_input_tokens_seen": 76077088,
      "step": 35245
    },
    {
      "epoch": 5.750407830342578,
      "grad_norm": 0.08581706136465073,
      "learning_rate": 2.2818979170400785e-05,
      "loss": 0.1219,
      "num_input_tokens_seen": 76087104,
      "step": 35250
    },
    {
      "epoch": 5.751223491027733,
      "grad_norm": 4.261509418487549,
      "learning_rate": 2.2811888415718405e-05,
      "loss": 0.0208,
      "num_input_tokens_seen": 76098560,
      "step": 35255
    },
    {
      "epoch": 5.7520391517128875,
      "grad_norm": 0.7076281905174255,
      "learning_rate": 2.2804797838415448e-05,
      "loss": 0.0129,
      "num_input_tokens_seen": 76109472,
      "step": 35260
    },
    {
      "epoch": 5.752854812398042,
      "grad_norm": 6.184848308563232,
      "learning_rate": 2.2797707439066724e-05,
      "loss": 0.0707,
      "num_input_tokens_seen": 76119744,
      "step": 35265
    },
    {
      "epoch": 5.753670473083197,
      "grad_norm": 0.08045632392168045,
      "learning_rate": 2.2790617218247005e-05,
      "loss": 0.0201,
      "num_input_tokens_seen": 76129664,
      "step": 35270
    },
    {
      "epoch": 5.754486133768353,
      "grad_norm": 0.09801807254552841,
      "learning_rate": 2.278352717653107e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 76140384,
      "step": 35275
    },
    {
      "epoch": 5.755301794453508,
      "grad_norm": 2.68274188041687,
      "learning_rate": 2.2776437314493666e-05,
      "loss": 0.1434,
      "num_input_tokens_seen": 76150624,
      "step": 35280
    },
    {
      "epoch": 5.7561174551386625,
      "grad_norm": 0.15872056782245636,
      "learning_rate": 2.2769347632709523e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 76161952,
      "step": 35285
    },
    {
      "epoch": 5.756933115823817,
      "grad_norm": 0.19446246325969696,
      "learning_rate": 2.276225813175339e-05,
      "loss": 0.0688,
      "num_input_tokens_seen": 76173600,
      "step": 35290
    },
    {
      "epoch": 5.757748776508972,
      "grad_norm": 0.1902131289243698,
      "learning_rate": 2.275516881219997e-05,
      "loss": 0.0055,
      "num_input_tokens_seen": 76183808,
      "step": 35295
    },
    {
      "epoch": 5.758564437194127,
      "grad_norm": 3.389545202255249,
      "learning_rate": 2.2748079674623954e-05,
      "loss": 0.1351,
      "num_input_tokens_seen": 76194752,
      "step": 35300
    },
    {
      "epoch": 5.759380097879282,
      "grad_norm": 8.538814544677734,
      "learning_rate": 2.2740990719600026e-05,
      "loss": 0.1226,
      "num_input_tokens_seen": 76205696,
      "step": 35305
    },
    {
      "epoch": 5.760195758564437,
      "grad_norm": 0.15769486129283905,
      "learning_rate": 2.2733901947702852e-05,
      "loss": 0.1312,
      "num_input_tokens_seen": 76215072,
      "step": 35310
    },
    {
      "epoch": 5.761011419249592,
      "grad_norm": 0.24559558928012848,
      "learning_rate": 2.2726813359507084e-05,
      "loss": 0.1406,
      "num_input_tokens_seen": 76225248,
      "step": 35315
    },
    {
      "epoch": 5.761827079934747,
      "grad_norm": 7.306979179382324,
      "learning_rate": 2.271972495558736e-05,
      "loss": 0.1356,
      "num_input_tokens_seen": 76235552,
      "step": 35320
    },
    {
      "epoch": 5.762642740619902,
      "grad_norm": 3.5641539096832275,
      "learning_rate": 2.27126367365183e-05,
      "loss": 0.1759,
      "num_input_tokens_seen": 76246624,
      "step": 35325
    },
    {
      "epoch": 5.763458401305057,
      "grad_norm": 33.02423858642578,
      "learning_rate": 2.2705548702874512e-05,
      "loss": 0.1133,
      "num_input_tokens_seen": 76256352,
      "step": 35330
    },
    {
      "epoch": 5.764274061990212,
      "grad_norm": 0.10651031136512756,
      "learning_rate": 2.269846085523059e-05,
      "loss": 0.1168,
      "num_input_tokens_seen": 76266912,
      "step": 35335
    },
    {
      "epoch": 5.765089722675367,
      "grad_norm": 0.11089621484279633,
      "learning_rate": 2.2691373194161107e-05,
      "loss": 0.2604,
      "num_input_tokens_seen": 76277344,
      "step": 35340
    },
    {
      "epoch": 5.765905383360522,
      "grad_norm": 0.3724483847618103,
      "learning_rate": 2.2684285720240624e-05,
      "loss": 0.2363,
      "num_input_tokens_seen": 76288544,
      "step": 35345
    },
    {
      "epoch": 5.766721044045677,
      "grad_norm": 5.472805976867676,
      "learning_rate": 2.2677198434043695e-05,
      "loss": 0.1657,
      "num_input_tokens_seen": 76299744,
      "step": 35350
    },
    {
      "epoch": 5.767536704730832,
      "grad_norm": 0.9301806688308716,
      "learning_rate": 2.2670111336144844e-05,
      "loss": 0.1473,
      "num_input_tokens_seen": 76309280,
      "step": 35355
    },
    {
      "epoch": 5.768352365415987,
      "grad_norm": 0.2854974865913391,
      "learning_rate": 2.2663024427118592e-05,
      "loss": 0.071,
      "num_input_tokens_seen": 76319072,
      "step": 35360
    },
    {
      "epoch": 5.769168026101142,
      "grad_norm": 7.550900459289551,
      "learning_rate": 2.2655937707539437e-05,
      "loss": 0.0871,
      "num_input_tokens_seen": 76329696,
      "step": 35365
    },
    {
      "epoch": 5.769983686786297,
      "grad_norm": 0.12121465802192688,
      "learning_rate": 2.2648851177981868e-05,
      "loss": 0.1767,
      "num_input_tokens_seen": 76339904,
      "step": 35370
    },
    {
      "epoch": 5.770799347471452,
      "grad_norm": 0.18952710926532745,
      "learning_rate": 2.264176483902035e-05,
      "loss": 0.0636,
      "num_input_tokens_seen": 76351776,
      "step": 35375
    },
    {
      "epoch": 5.771615008156607,
      "grad_norm": 0.11591441184282303,
      "learning_rate": 2.263467869122934e-05,
      "loss": 0.1627,
      "num_input_tokens_seen": 76362208,
      "step": 35380
    },
    {
      "epoch": 5.7724306688417615,
      "grad_norm": 0.19922132790088654,
      "learning_rate": 2.262759273518327e-05,
      "loss": 0.0083,
      "num_input_tokens_seen": 76374624,
      "step": 35385
    },
    {
      "epoch": 5.773246329526917,
      "grad_norm": 0.07203475385904312,
      "learning_rate": 2.262050697145657e-05,
      "loss": 0.025,
      "num_input_tokens_seen": 76384064,
      "step": 35390
    },
    {
      "epoch": 5.774061990212072,
      "grad_norm": 8.112296104431152,
      "learning_rate": 2.2613421400623653e-05,
      "loss": 0.0815,
      "num_input_tokens_seen": 76394976,
      "step": 35395
    },
    {
      "epoch": 5.774877650897227,
      "grad_norm": 1.2074600458145142,
      "learning_rate": 2.2606336023258907e-05,
      "loss": 0.2279,
      "num_input_tokens_seen": 76407136,
      "step": 35400
    },
    {
      "epoch": 5.775693311582382,
      "grad_norm": 1.015274167060852,
      "learning_rate": 2.259925083993671e-05,
      "loss": 0.0182,
      "num_input_tokens_seen": 76417536,
      "step": 35405
    },
    {
      "epoch": 5.7765089722675365,
      "grad_norm": 0.033531554043293,
      "learning_rate": 2.2592165851231423e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 76428576,
      "step": 35410
    },
    {
      "epoch": 5.777324632952691,
      "grad_norm": 2.3847498893737793,
      "learning_rate": 2.2585081057717387e-05,
      "loss": 0.0937,
      "num_input_tokens_seen": 76439040,
      "step": 35415
    },
    {
      "epoch": 5.778140293637847,
      "grad_norm": 0.7646045088768005,
      "learning_rate": 2.2577996459968935e-05,
      "loss": 0.2439,
      "num_input_tokens_seen": 76450624,
      "step": 35420
    },
    {
      "epoch": 5.778955954323002,
      "grad_norm": 0.11400401592254639,
      "learning_rate": 2.2570912058560375e-05,
      "loss": 0.1651,
      "num_input_tokens_seen": 76459840,
      "step": 35425
    },
    {
      "epoch": 5.779771615008157,
      "grad_norm": 5.900925159454346,
      "learning_rate": 2.2563827854066007e-05,
      "loss": 0.2483,
      "num_input_tokens_seen": 76470688,
      "step": 35430
    },
    {
      "epoch": 5.780587275693311,
      "grad_norm": 0.12469127029180527,
      "learning_rate": 2.2556743847060118e-05,
      "loss": 0.2205,
      "num_input_tokens_seen": 76481728,
      "step": 35435
    },
    {
      "epoch": 5.781402936378466,
      "grad_norm": 4.723505020141602,
      "learning_rate": 2.254966003811697e-05,
      "loss": 0.1852,
      "num_input_tokens_seen": 76492032,
      "step": 35440
    },
    {
      "epoch": 5.782218597063622,
      "grad_norm": 0.492357075214386,
      "learning_rate": 2.2542576427810813e-05,
      "loss": 0.0147,
      "num_input_tokens_seen": 76502784,
      "step": 35445
    },
    {
      "epoch": 5.783034257748777,
      "grad_norm": 0.08330883085727692,
      "learning_rate": 2.2535493016715882e-05,
      "loss": 0.0254,
      "num_input_tokens_seen": 76513856,
      "step": 35450
    },
    {
      "epoch": 5.783849918433932,
      "grad_norm": 0.22224728763103485,
      "learning_rate": 2.2528409805406388e-05,
      "loss": 0.0112,
      "num_input_tokens_seen": 76525056,
      "step": 35455
    },
    {
      "epoch": 5.784665579119086,
      "grad_norm": 4.219114780426025,
      "learning_rate": 2.2521326794456537e-05,
      "loss": 0.1197,
      "num_input_tokens_seen": 76535584,
      "step": 35460
    },
    {
      "epoch": 5.785481239804241,
      "grad_norm": 12.117701530456543,
      "learning_rate": 2.2514243984440512e-05,
      "loss": 0.1731,
      "num_input_tokens_seen": 76547168,
      "step": 35465
    },
    {
      "epoch": 5.786296900489396,
      "grad_norm": 0.08506445586681366,
      "learning_rate": 2.2507161375932484e-05,
      "loss": 0.0174,
      "num_input_tokens_seen": 76557888,
      "step": 35470
    },
    {
      "epoch": 5.787112561174552,
      "grad_norm": 0.16435596346855164,
      "learning_rate": 2.25000789695066e-05,
      "loss": 0.0159,
      "num_input_tokens_seen": 76568864,
      "step": 35475
    },
    {
      "epoch": 5.787928221859707,
      "grad_norm": 0.21710620820522308,
      "learning_rate": 2.2492996765737004e-05,
      "loss": 0.012,
      "num_input_tokens_seen": 76579296,
      "step": 35480
    },
    {
      "epoch": 5.788743882544861,
      "grad_norm": 0.04540694132447243,
      "learning_rate": 2.2485914765197807e-05,
      "loss": 0.0091,
      "num_input_tokens_seen": 76589824,
      "step": 35485
    },
    {
      "epoch": 5.789559543230016,
      "grad_norm": 0.17728549242019653,
      "learning_rate": 2.247883296846311e-05,
      "loss": 0.1958,
      "num_input_tokens_seen": 76599744,
      "step": 35490
    },
    {
      "epoch": 5.790375203915171,
      "grad_norm": 0.3285267949104309,
      "learning_rate": 2.2471751376107006e-05,
      "loss": 0.0901,
      "num_input_tokens_seen": 76612000,
      "step": 35495
    },
    {
      "epoch": 5.791190864600326,
      "grad_norm": 0.1349720060825348,
      "learning_rate": 2.246466998870357e-05,
      "loss": 0.0504,
      "num_input_tokens_seen": 76623936,
      "step": 35500
    },
    {
      "epoch": 5.7920065252854815,
      "grad_norm": 0.16432398557662964,
      "learning_rate": 2.245758880682685e-05,
      "loss": 0.1017,
      "num_input_tokens_seen": 76634240,
      "step": 35505
    },
    {
      "epoch": 5.792822185970636,
      "grad_norm": 0.15062400698661804,
      "learning_rate": 2.2450507831050876e-05,
      "loss": 0.0936,
      "num_input_tokens_seen": 76645344,
      "step": 35510
    },
    {
      "epoch": 5.793637846655791,
      "grad_norm": 0.05078652873635292,
      "learning_rate": 2.2443427061949672e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 76656000,
      "step": 35515
    },
    {
      "epoch": 5.794453507340946,
      "grad_norm": 4.0731916427612305,
      "learning_rate": 2.2436346500097247e-05,
      "loss": 0.1775,
      "num_input_tokens_seen": 76666976,
      "step": 35520
    },
    {
      "epoch": 5.795269168026101,
      "grad_norm": 19.542570114135742,
      "learning_rate": 2.2429266146067582e-05,
      "loss": 0.1237,
      "num_input_tokens_seen": 76676192,
      "step": 35525
    },
    {
      "epoch": 5.7960848287112565,
      "grad_norm": 1.1487821340560913,
      "learning_rate": 2.242218600043465e-05,
      "loss": 0.0618,
      "num_input_tokens_seen": 76687104,
      "step": 35530
    },
    {
      "epoch": 5.796900489396411,
      "grad_norm": 12.550699234008789,
      "learning_rate": 2.2415106063772394e-05,
      "loss": 0.0512,
      "num_input_tokens_seen": 76699104,
      "step": 35535
    },
    {
      "epoch": 5.797716150081566,
      "grad_norm": 0.05929354205727577,
      "learning_rate": 2.240802633665476e-05,
      "loss": 0.2031,
      "num_input_tokens_seen": 76710464,
      "step": 35540
    },
    {
      "epoch": 5.798531810766721,
      "grad_norm": 0.07123995572328568,
      "learning_rate": 2.2400946819655663e-05,
      "loss": 0.009,
      "num_input_tokens_seen": 76720864,
      "step": 35545
    },
    {
      "epoch": 5.799347471451876,
      "grad_norm": 0.061252374202013016,
      "learning_rate": 2.2393867513349002e-05,
      "loss": 0.1249,
      "num_input_tokens_seen": 76731520,
      "step": 35550
    },
    {
      "epoch": 5.800163132137031,
      "grad_norm": 3.8870437145233154,
      "learning_rate": 2.238678841830867e-05,
      "loss": 0.1198,
      "num_input_tokens_seen": 76741536,
      "step": 35555
    },
    {
      "epoch": 5.800978792822186,
      "grad_norm": 0.067466601729393,
      "learning_rate": 2.2379709535108524e-05,
      "loss": 0.0693,
      "num_input_tokens_seen": 76751904,
      "step": 35560
    },
    {
      "epoch": 5.801794453507341,
      "grad_norm": 2.9204559326171875,
      "learning_rate": 2.2372630864322416e-05,
      "loss": 0.2492,
      "num_input_tokens_seen": 76761504,
      "step": 35565
    },
    {
      "epoch": 5.802610114192496,
      "grad_norm": 0.11660844832658768,
      "learning_rate": 2.2365552406524183e-05,
      "loss": 0.0035,
      "num_input_tokens_seen": 76771968,
      "step": 35570
    },
    {
      "epoch": 5.803425774877651,
      "grad_norm": 4.045225143432617,
      "learning_rate": 2.2358474162287635e-05,
      "loss": 0.1328,
      "num_input_tokens_seen": 76782432,
      "step": 35575
    },
    {
      "epoch": 5.804241435562806,
      "grad_norm": 10.254171371459961,
      "learning_rate": 2.235139613218658e-05,
      "loss": 0.1031,
      "num_input_tokens_seen": 76793152,
      "step": 35580
    },
    {
      "epoch": 5.80505709624796,
      "grad_norm": 9.46228313446045,
      "learning_rate": 2.234431831679479e-05,
      "loss": 0.1023,
      "num_input_tokens_seen": 76804000,
      "step": 35585
    },
    {
      "epoch": 5.805872756933116,
      "grad_norm": 0.13323760032653809,
      "learning_rate": 2.2337240716686035e-05,
      "loss": 0.0071,
      "num_input_tokens_seen": 76814432,
      "step": 35590
    },
    {
      "epoch": 5.806688417618271,
      "grad_norm": 3.207540512084961,
      "learning_rate": 2.2330163332434056e-05,
      "loss": 0.0341,
      "num_input_tokens_seen": 76825472,
      "step": 35595
    },
    {
      "epoch": 5.807504078303426,
      "grad_norm": 0.17430220544338226,
      "learning_rate": 2.2323086164612584e-05,
      "loss": 0.0048,
      "num_input_tokens_seen": 76837568,
      "step": 35600
    },
    {
      "epoch": 5.808319738988581,
      "grad_norm": 0.1959105283021927,
      "learning_rate": 2.2316009213795323e-05,
      "loss": 0.006,
      "num_input_tokens_seen": 76848768,
      "step": 35605
    },
    {
      "epoch": 5.809135399673735,
      "grad_norm": 3.450186014175415,
      "learning_rate": 2.230893248055598e-05,
      "loss": 0.0121,
      "num_input_tokens_seen": 76860000,
      "step": 35610
    },
    {
      "epoch": 5.809951060358891,
      "grad_norm": 0.04870615154504776,
      "learning_rate": 2.2301855965468226e-05,
      "loss": 0.0984,
      "num_input_tokens_seen": 76871904,
      "step": 35615
    },
    {
      "epoch": 5.810766721044046,
      "grad_norm": 0.1246744841337204,
      "learning_rate": 2.2294779669105716e-05,
      "loss": 0.0748,
      "num_input_tokens_seen": 76881504,
      "step": 35620
    },
    {
      "epoch": 5.811582381729201,
      "grad_norm": 6.546163082122803,
      "learning_rate": 2.2287703592042096e-05,
      "loss": 0.2706,
      "num_input_tokens_seen": 76894080,
      "step": 35625
    },
    {
      "epoch": 5.8123980424143555,
      "grad_norm": 0.32414501905441284,
      "learning_rate": 2.2280627734850984e-05,
      "loss": 0.0746,
      "num_input_tokens_seen": 76904672,
      "step": 35630
    },
    {
      "epoch": 5.81321370309951,
      "grad_norm": 0.26913467049598694,
      "learning_rate": 2.2273552098105983e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 76914976,
      "step": 35635
    },
    {
      "epoch": 5.814029363784666,
      "grad_norm": 15.564455032348633,
      "learning_rate": 2.2266476682380685e-05,
      "loss": 0.2012,
      "num_input_tokens_seen": 76925824,
      "step": 35640
    },
    {
      "epoch": 5.814845024469821,
      "grad_norm": 0.3071824312210083,
      "learning_rate": 2.2259401488248658e-05,
      "loss": 0.0614,
      "num_input_tokens_seen": 76936160,
      "step": 35645
    },
    {
      "epoch": 5.815660685154976,
      "grad_norm": 7.369630336761475,
      "learning_rate": 2.225232651628345e-05,
      "loss": 0.1465,
      "num_input_tokens_seen": 76946016,
      "step": 35650
    },
    {
      "epoch": 5.8164763458401305,
      "grad_norm": 0.2431706339120865,
      "learning_rate": 2.2245251767058595e-05,
      "loss": 0.0753,
      "num_input_tokens_seen": 76956704,
      "step": 35655
    },
    {
      "epoch": 5.817292006525285,
      "grad_norm": 0.23439978063106537,
      "learning_rate": 2.2238177241147607e-05,
      "loss": 0.0147,
      "num_input_tokens_seen": 76967360,
      "step": 35660
    },
    {
      "epoch": 5.81810766721044,
      "grad_norm": 0.7164652347564697,
      "learning_rate": 2.223110293912399e-05,
      "loss": 0.0953,
      "num_input_tokens_seen": 76979008,
      "step": 35665
    },
    {
      "epoch": 5.818923327895595,
      "grad_norm": 0.04254661127924919,
      "learning_rate": 2.2224028861561215e-05,
      "loss": 0.0941,
      "num_input_tokens_seen": 76988800,
      "step": 35670
    },
    {
      "epoch": 5.819738988580751,
      "grad_norm": 19.6538028717041,
      "learning_rate": 2.2216955009032747e-05,
      "loss": 0.2615,
      "num_input_tokens_seen": 76999616,
      "step": 35675
    },
    {
      "epoch": 5.8205546492659055,
      "grad_norm": 0.13810043036937714,
      "learning_rate": 2.2209881382112026e-05,
      "loss": 0.0046,
      "num_input_tokens_seen": 77010656,
      "step": 35680
    },
    {
      "epoch": 5.82137030995106,
      "grad_norm": 0.025951286777853966,
      "learning_rate": 2.220280798137248e-05,
      "loss": 0.0232,
      "num_input_tokens_seen": 77021920,
      "step": 35685
    },
    {
      "epoch": 5.822185970636215,
      "grad_norm": 6.147768497467041,
      "learning_rate": 2.219573480738751e-05,
      "loss": 0.012,
      "num_input_tokens_seen": 77032416,
      "step": 35690
    },
    {
      "epoch": 5.82300163132137,
      "grad_norm": 5.249512195587158,
      "learning_rate": 2.2188661860730507e-05,
      "loss": 0.0562,
      "num_input_tokens_seen": 77042944,
      "step": 35695
    },
    {
      "epoch": 5.823817292006526,
      "grad_norm": 0.30161696672439575,
      "learning_rate": 2.2181589141974836e-05,
      "loss": 0.0366,
      "num_input_tokens_seen": 77051584,
      "step": 35700
    },
    {
      "epoch": 5.8246329526916805,
      "grad_norm": 3.3952252864837646,
      "learning_rate": 2.217451665169385e-05,
      "loss": 0.1948,
      "num_input_tokens_seen": 77062048,
      "step": 35705
    },
    {
      "epoch": 5.825448613376835,
      "grad_norm": 0.2144870162010193,
      "learning_rate": 2.216744439046087e-05,
      "loss": 0.1756,
      "num_input_tokens_seen": 77074080,
      "step": 35710
    },
    {
      "epoch": 5.82626427406199,
      "grad_norm": 0.20196601748466492,
      "learning_rate": 2.2160372358849234e-05,
      "loss": 0.3953,
      "num_input_tokens_seen": 77085632,
      "step": 35715
    },
    {
      "epoch": 5.827079934747145,
      "grad_norm": 0.05781104043126106,
      "learning_rate": 2.215330055743222e-05,
      "loss": 0.0032,
      "num_input_tokens_seen": 77097664,
      "step": 35720
    },
    {
      "epoch": 5.827895595432301,
      "grad_norm": 0.18307341635227203,
      "learning_rate": 2.2146228986783105e-05,
      "loss": 0.0092,
      "num_input_tokens_seen": 77109568,
      "step": 35725
    },
    {
      "epoch": 5.828711256117455,
      "grad_norm": 4.371525764465332,
      "learning_rate": 2.213915764747515e-05,
      "loss": 0.1864,
      "num_input_tokens_seen": 77119648,
      "step": 35730
    },
    {
      "epoch": 5.82952691680261,
      "grad_norm": 1.8286266326904297,
      "learning_rate": 2.2132086540081593e-05,
      "loss": 0.0056,
      "num_input_tokens_seen": 77132064,
      "step": 35735
    },
    {
      "epoch": 5.830342577487765,
      "grad_norm": 0.1642422080039978,
      "learning_rate": 2.212501566517565e-05,
      "loss": 0.0221,
      "num_input_tokens_seen": 77142880,
      "step": 35740
    },
    {
      "epoch": 5.83115823817292,
      "grad_norm": 3.268336534500122,
      "learning_rate": 2.211794502333052e-05,
      "loss": 0.1043,
      "num_input_tokens_seen": 77154272,
      "step": 35745
    },
    {
      "epoch": 5.831973898858075,
      "grad_norm": 3.634614944458008,
      "learning_rate": 2.2110874615119396e-05,
      "loss": 0.3382,
      "num_input_tokens_seen": 77165280,
      "step": 35750
    },
    {
      "epoch": 5.8327895595432295,
      "grad_norm": 0.4623981714248657,
      "learning_rate": 2.2103804441115434e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 77176032,
      "step": 35755
    },
    {
      "epoch": 5.833605220228385,
      "grad_norm": 0.05410349369049072,
      "learning_rate": 2.209673450189178e-05,
      "loss": 0.1039,
      "num_input_tokens_seen": 77186880,
      "step": 35760
    },
    {
      "epoch": 5.83442088091354,
      "grad_norm": 8.331046104431152,
      "learning_rate": 2.208966479802156e-05,
      "loss": 0.1679,
      "num_input_tokens_seen": 77196704,
      "step": 35765
    },
    {
      "epoch": 5.835236541598695,
      "grad_norm": 1.2005671262741089,
      "learning_rate": 2.2082595330077878e-05,
      "loss": 0.011,
      "num_input_tokens_seen": 77208064,
      "step": 35770
    },
    {
      "epoch": 5.83605220228385,
      "grad_norm": 0.37973007559776306,
      "learning_rate": 2.2075526098633816e-05,
      "loss": 0.0972,
      "num_input_tokens_seen": 77219328,
      "step": 35775
    },
    {
      "epoch": 5.8368678629690045,
      "grad_norm": 0.15165221691131592,
      "learning_rate": 2.206845710426245e-05,
      "loss": 0.0747,
      "num_input_tokens_seen": 77230592,
      "step": 35780
    },
    {
      "epoch": 5.83768352365416,
      "grad_norm": 0.22922126948833466,
      "learning_rate": 2.2061388347536828e-05,
      "loss": 0.1342,
      "num_input_tokens_seen": 77240992,
      "step": 35785
    },
    {
      "epoch": 5.838499184339315,
      "grad_norm": 0.3719537854194641,
      "learning_rate": 2.2054319829029975e-05,
      "loss": 0.0888,
      "num_input_tokens_seen": 77250144,
      "step": 35790
    },
    {
      "epoch": 5.83931484502447,
      "grad_norm": 0.06579183042049408,
      "learning_rate": 2.2047251549314907e-05,
      "loss": 0.1123,
      "num_input_tokens_seen": 77262464,
      "step": 35795
    },
    {
      "epoch": 5.840130505709625,
      "grad_norm": 0.586622416973114,
      "learning_rate": 2.20401835089646e-05,
      "loss": 0.1145,
      "num_input_tokens_seen": 77273248,
      "step": 35800
    },
    {
      "epoch": 5.8409461663947795,
      "grad_norm": 6.7092413902282715,
      "learning_rate": 2.2033115708552044e-05,
      "loss": 0.18,
      "num_input_tokens_seen": 77284032,
      "step": 35805
    },
    {
      "epoch": 5.841761827079935,
      "grad_norm": 2.1209661960601807,
      "learning_rate": 2.202604814865018e-05,
      "loss": 0.0114,
      "num_input_tokens_seen": 77295072,
      "step": 35810
    },
    {
      "epoch": 5.84257748776509,
      "grad_norm": 5.148712158203125,
      "learning_rate": 2.201898082983194e-05,
      "loss": 0.2185,
      "num_input_tokens_seen": 77306208,
      "step": 35815
    },
    {
      "epoch": 5.843393148450245,
      "grad_norm": 0.052786875516176224,
      "learning_rate": 2.2011913752670242e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 77317024,
      "step": 35820
    },
    {
      "epoch": 5.8442088091354,
      "grad_norm": 2.9350016117095947,
      "learning_rate": 2.2004846917737978e-05,
      "loss": 0.1611,
      "num_input_tokens_seen": 77327648,
      "step": 35825
    },
    {
      "epoch": 5.8450244698205545,
      "grad_norm": 0.11455246806144714,
      "learning_rate": 2.1997780325608013e-05,
      "loss": 0.0899,
      "num_input_tokens_seen": 77338304,
      "step": 35830
    },
    {
      "epoch": 5.845840130505709,
      "grad_norm": 0.07974403351545334,
      "learning_rate": 2.1990713976853216e-05,
      "loss": 0.0844,
      "num_input_tokens_seen": 77348768,
      "step": 35835
    },
    {
      "epoch": 5.846655791190865,
      "grad_norm": 0.0672159492969513,
      "learning_rate": 2.1983647872046412e-05,
      "loss": 0.1166,
      "num_input_tokens_seen": 77358208,
      "step": 35840
    },
    {
      "epoch": 5.84747145187602,
      "grad_norm": 0.10648999363183975,
      "learning_rate": 2.1976582011760415e-05,
      "loss": 0.008,
      "num_input_tokens_seen": 77368896,
      "step": 35845
    },
    {
      "epoch": 5.848287112561175,
      "grad_norm": 0.3450216054916382,
      "learning_rate": 2.1969516396568023e-05,
      "loss": 0.008,
      "num_input_tokens_seen": 77379616,
      "step": 35850
    },
    {
      "epoch": 5.849102773246329,
      "grad_norm": 6.794552326202393,
      "learning_rate": 2.196245102704201e-05,
      "loss": 0.0232,
      "num_input_tokens_seen": 77390400,
      "step": 35855
    },
    {
      "epoch": 5.849918433931484,
      "grad_norm": 4.051516056060791,
      "learning_rate": 2.1955385903755127e-05,
      "loss": 0.0822,
      "num_input_tokens_seen": 77400544,
      "step": 35860
    },
    {
      "epoch": 5.850734094616639,
      "grad_norm": 0.050954125821590424,
      "learning_rate": 2.1948321027280108e-05,
      "loss": 0.0322,
      "num_input_tokens_seen": 77411136,
      "step": 35865
    },
    {
      "epoch": 5.851549755301795,
      "grad_norm": 0.21890828013420105,
      "learning_rate": 2.1941256398189676e-05,
      "loss": 0.1633,
      "num_input_tokens_seen": 77422240,
      "step": 35870
    },
    {
      "epoch": 5.85236541598695,
      "grad_norm": 0.1725703775882721,
      "learning_rate": 2.1934192017056515e-05,
      "loss": 0.0885,
      "num_input_tokens_seen": 77431648,
      "step": 35875
    },
    {
      "epoch": 5.853181076672104,
      "grad_norm": 0.2586349546909332,
      "learning_rate": 2.1927127884453307e-05,
      "loss": 0.0061,
      "num_input_tokens_seen": 77443680,
      "step": 35880
    },
    {
      "epoch": 5.853996737357259,
      "grad_norm": 0.13698704540729523,
      "learning_rate": 2.19200640009527e-05,
      "loss": 0.1708,
      "num_input_tokens_seen": 77453824,
      "step": 35885
    },
    {
      "epoch": 5.854812398042414,
      "grad_norm": 0.05802655220031738,
      "learning_rate": 2.1913000367127337e-05,
      "loss": 0.1927,
      "num_input_tokens_seen": 77464512,
      "step": 35890
    },
    {
      "epoch": 5.85562805872757,
      "grad_norm": 10.350048065185547,
      "learning_rate": 2.190593698354983e-05,
      "loss": 0.2048,
      "num_input_tokens_seen": 77475584,
      "step": 35895
    },
    {
      "epoch": 5.856443719412725,
      "grad_norm": 12.542141914367676,
      "learning_rate": 2.1898873850792768e-05,
      "loss": 0.0516,
      "num_input_tokens_seen": 77486464,
      "step": 35900
    },
    {
      "epoch": 5.857259380097879,
      "grad_norm": 3.6151926517486572,
      "learning_rate": 2.1891810969428724e-05,
      "loss": 0.1128,
      "num_input_tokens_seen": 77495168,
      "step": 35905
    },
    {
      "epoch": 5.858075040783034,
      "grad_norm": 9.008210182189941,
      "learning_rate": 2.1884748340030255e-05,
      "loss": 0.1673,
      "num_input_tokens_seen": 77505632,
      "step": 35910
    },
    {
      "epoch": 5.858890701468189,
      "grad_norm": 0.17407682538032532,
      "learning_rate": 2.1877685963169893e-05,
      "loss": 0.0957,
      "num_input_tokens_seen": 77516512,
      "step": 35915
    },
    {
      "epoch": 5.859706362153344,
      "grad_norm": 20.76967430114746,
      "learning_rate": 2.187062383942015e-05,
      "loss": 0.0564,
      "num_input_tokens_seen": 77528352,
      "step": 35920
    },
    {
      "epoch": 5.8605220228384995,
      "grad_norm": 0.37421369552612305,
      "learning_rate": 2.186356196935351e-05,
      "loss": 0.069,
      "num_input_tokens_seen": 77537728,
      "step": 35925
    },
    {
      "epoch": 5.861337683523654,
      "grad_norm": 0.04438697174191475,
      "learning_rate": 2.185650035354245e-05,
      "loss": 0.1387,
      "num_input_tokens_seen": 77549216,
      "step": 35930
    },
    {
      "epoch": 5.862153344208809,
      "grad_norm": 3.2634708881378174,
      "learning_rate": 2.184943899255943e-05,
      "loss": 0.2825,
      "num_input_tokens_seen": 77560288,
      "step": 35935
    },
    {
      "epoch": 5.862969004893964,
      "grad_norm": 16.22699737548828,
      "learning_rate": 2.1842377886976873e-05,
      "loss": 0.0552,
      "num_input_tokens_seen": 77571968,
      "step": 35940
    },
    {
      "epoch": 5.863784665579119,
      "grad_norm": 0.07784358412027359,
      "learning_rate": 2.183531703736718e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 77583872,
      "step": 35945
    },
    {
      "epoch": 5.864600326264274,
      "grad_norm": 0.07082092016935349,
      "learning_rate": 2.182825644430275e-05,
      "loss": 0.1043,
      "num_input_tokens_seen": 77595232,
      "step": 35950
    },
    {
      "epoch": 5.865415986949429,
      "grad_norm": 0.057193465530872345,
      "learning_rate": 2.1821196108355944e-05,
      "loss": 0.004,
      "num_input_tokens_seen": 77605696,
      "step": 35955
    },
    {
      "epoch": 5.866231647634584,
      "grad_norm": 0.2835041582584381,
      "learning_rate": 2.181413603009911e-05,
      "loss": 0.0368,
      "num_input_tokens_seen": 77616192,
      "step": 35960
    },
    {
      "epoch": 5.867047308319739,
      "grad_norm": 1.8883839845657349,
      "learning_rate": 2.1807076210104575e-05,
      "loss": 0.0098,
      "num_input_tokens_seen": 77626688,
      "step": 35965
    },
    {
      "epoch": 5.867862969004894,
      "grad_norm": 0.06178255379199982,
      "learning_rate": 2.1800016648944638e-05,
      "loss": 0.0253,
      "num_input_tokens_seen": 77637152,
      "step": 35970
    },
    {
      "epoch": 5.868678629690049,
      "grad_norm": 3.3733088970184326,
      "learning_rate": 2.1792957347191594e-05,
      "loss": 0.0982,
      "num_input_tokens_seen": 77647328,
      "step": 35975
    },
    {
      "epoch": 5.869494290375204,
      "grad_norm": 0.06617254763841629,
      "learning_rate": 2.1785898305417698e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 77657824,
      "step": 35980
    },
    {
      "epoch": 5.870309951060359,
      "grad_norm": 0.06605422496795654,
      "learning_rate": 2.1778839524195195e-05,
      "loss": 0.0493,
      "num_input_tokens_seen": 77668768,
      "step": 35985
    },
    {
      "epoch": 5.871125611745514,
      "grad_norm": 0.24393099546432495,
      "learning_rate": 2.1771781004096304e-05,
      "loss": 0.1428,
      "num_input_tokens_seen": 77679040,
      "step": 35990
    },
    {
      "epoch": 5.871941272430669,
      "grad_norm": 0.06658919900655746,
      "learning_rate": 2.1764722745693223e-05,
      "loss": 0.0349,
      "num_input_tokens_seen": 77690304,
      "step": 35995
    },
    {
      "epoch": 5.872756933115824,
      "grad_norm": 3.923738479614258,
      "learning_rate": 2.1757664749558132e-05,
      "loss": 0.1157,
      "num_input_tokens_seen": 77700480,
      "step": 36000
    },
    {
      "epoch": 5.873572593800979,
      "grad_norm": 0.4210411012172699,
      "learning_rate": 2.1750607016263192e-05,
      "loss": 0.0086,
      "num_input_tokens_seen": 77711552,
      "step": 36005
    },
    {
      "epoch": 5.874388254486134,
      "grad_norm": 0.21074756979942322,
      "learning_rate": 2.1743549546380527e-05,
      "loss": 0.1148,
      "num_input_tokens_seen": 77722944,
      "step": 36010
    },
    {
      "epoch": 5.875203915171289,
      "grad_norm": 0.02160373516380787,
      "learning_rate": 2.1736492340482267e-05,
      "loss": 0.0082,
      "num_input_tokens_seen": 77735072,
      "step": 36015
    },
    {
      "epoch": 5.876019575856444,
      "grad_norm": 0.042782049626111984,
      "learning_rate": 2.172943539914049e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 77746848,
      "step": 36020
    },
    {
      "epoch": 5.876835236541599,
      "grad_norm": 0.07420703768730164,
      "learning_rate": 2.172237872292728e-05,
      "loss": 0.1185,
      "num_input_tokens_seen": 77757664,
      "step": 36025
    },
    {
      "epoch": 5.877650897226753,
      "grad_norm": 0.049979452043771744,
      "learning_rate": 2.1715322312414664e-05,
      "loss": 0.0241,
      "num_input_tokens_seen": 77768288,
      "step": 36030
    },
    {
      "epoch": 5.878466557911908,
      "grad_norm": 0.027508003637194633,
      "learning_rate": 2.1708266168174703e-05,
      "loss": 0.1088,
      "num_input_tokens_seen": 77778080,
      "step": 36035
    },
    {
      "epoch": 5.879282218597064,
      "grad_norm": 3.2788803577423096,
      "learning_rate": 2.170121029077939e-05,
      "loss": 0.2079,
      "num_input_tokens_seen": 77790784,
      "step": 36040
    },
    {
      "epoch": 5.880097879282219,
      "grad_norm": 0.21825259923934937,
      "learning_rate": 2.1694154680800706e-05,
      "loss": 0.0829,
      "num_input_tokens_seen": 77801344,
      "step": 36045
    },
    {
      "epoch": 5.8809135399673735,
      "grad_norm": 0.8757532238960266,
      "learning_rate": 2.168709933881062e-05,
      "loss": 0.0593,
      "num_input_tokens_seen": 77811392,
      "step": 36050
    },
    {
      "epoch": 5.881729200652528,
      "grad_norm": 5.703310489654541,
      "learning_rate": 2.168004426538106e-05,
      "loss": 0.1182,
      "num_input_tokens_seen": 77821312,
      "step": 36055
    },
    {
      "epoch": 5.882544861337683,
      "grad_norm": 0.0797046422958374,
      "learning_rate": 2.1672989461083964e-05,
      "loss": 0.2148,
      "num_input_tokens_seen": 77831872,
      "step": 36060
    },
    {
      "epoch": 5.883360522022839,
      "grad_norm": 0.18410950899124146,
      "learning_rate": 2.1665934926491226e-05,
      "loss": 0.1024,
      "num_input_tokens_seen": 77842752,
      "step": 36065
    },
    {
      "epoch": 5.884176182707994,
      "grad_norm": 0.05765313655138016,
      "learning_rate": 2.1658880662174717e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 77853696,
      "step": 36070
    },
    {
      "epoch": 5.8849918433931485,
      "grad_norm": 31.744882583618164,
      "learning_rate": 2.1651826668706297e-05,
      "loss": 0.2001,
      "num_input_tokens_seen": 77864288,
      "step": 36075
    },
    {
      "epoch": 5.885807504078303,
      "grad_norm": 0.1185903251171112,
      "learning_rate": 2.1644772946657795e-05,
      "loss": 0.0989,
      "num_input_tokens_seen": 77874880,
      "step": 36080
    },
    {
      "epoch": 5.886623164763458,
      "grad_norm": 0.07939574122428894,
      "learning_rate": 2.163771949660102e-05,
      "loss": 0.0043,
      "num_input_tokens_seen": 77886688,
      "step": 36085
    },
    {
      "epoch": 5.887438825448614,
      "grad_norm": 0.15546125173568726,
      "learning_rate": 2.1630666319107767e-05,
      "loss": 0.0839,
      "num_input_tokens_seen": 77897376,
      "step": 36090
    },
    {
      "epoch": 5.888254486133769,
      "grad_norm": 0.21585629880428314,
      "learning_rate": 2.1623613414749797e-05,
      "loss": 0.0233,
      "num_input_tokens_seen": 77908736,
      "step": 36095
    },
    {
      "epoch": 5.8890701468189235,
      "grad_norm": 0.05212084576487541,
      "learning_rate": 2.1616560784098856e-05,
      "loss": 0.2501,
      "num_input_tokens_seen": 77920352,
      "step": 36100
    },
    {
      "epoch": 5.889885807504078,
      "grad_norm": 0.07802629470825195,
      "learning_rate": 2.160950842772666e-05,
      "loss": 0.2069,
      "num_input_tokens_seen": 77931360,
      "step": 36105
    },
    {
      "epoch": 5.890701468189233,
      "grad_norm": 0.07027492672204971,
      "learning_rate": 2.160245634620492e-05,
      "loss": 0.0748,
      "num_input_tokens_seen": 77942208,
      "step": 36110
    },
    {
      "epoch": 5.891517128874388,
      "grad_norm": 0.032340358942747116,
      "learning_rate": 2.1595404540105295e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 77953248,
      "step": 36115
    },
    {
      "epoch": 5.892332789559543,
      "grad_norm": 0.056712131947278976,
      "learning_rate": 2.1588353009999464e-05,
      "loss": 0.0043,
      "num_input_tokens_seen": 77963584,
      "step": 36120
    },
    {
      "epoch": 5.8931484502446985,
      "grad_norm": 21.15962028503418,
      "learning_rate": 2.158130175645905e-05,
      "loss": 0.0409,
      "num_input_tokens_seen": 77973920,
      "step": 36125
    },
    {
      "epoch": 5.893964110929853,
      "grad_norm": 0.0940650999546051,
      "learning_rate": 2.1574250780055654e-05,
      "loss": 0.0596,
      "num_input_tokens_seen": 77984736,
      "step": 36130
    },
    {
      "epoch": 5.894779771615008,
      "grad_norm": 3.097771406173706,
      "learning_rate": 2.156720008136087e-05,
      "loss": 0.1741,
      "num_input_tokens_seen": 77996160,
      "step": 36135
    },
    {
      "epoch": 5.895595432300163,
      "grad_norm": 1.6625806093215942,
      "learning_rate": 2.156014966094627e-05,
      "loss": 0.1153,
      "num_input_tokens_seen": 78006592,
      "step": 36140
    },
    {
      "epoch": 5.896411092985318,
      "grad_norm": 0.02249511331319809,
      "learning_rate": 2.1553099519383394e-05,
      "loss": 0.0952,
      "num_input_tokens_seen": 78015936,
      "step": 36145
    },
    {
      "epoch": 5.897226753670473,
      "grad_norm": 3.918808698654175,
      "learning_rate": 2.154604965724376e-05,
      "loss": 0.3321,
      "num_input_tokens_seen": 78028160,
      "step": 36150
    },
    {
      "epoch": 5.898042414355628,
      "grad_norm": 0.044293925166130066,
      "learning_rate": 2.1539000075098868e-05,
      "loss": 0.0118,
      "num_input_tokens_seen": 78039712,
      "step": 36155
    },
    {
      "epoch": 5.898858075040783,
      "grad_norm": 0.05846496298909187,
      "learning_rate": 2.1531950773520187e-05,
      "loss": 0.01,
      "num_input_tokens_seen": 78051936,
      "step": 36160
    },
    {
      "epoch": 5.899673735725938,
      "grad_norm": 0.14292560517787933,
      "learning_rate": 2.1524901753079176e-05,
      "loss": 0.2394,
      "num_input_tokens_seen": 78062464,
      "step": 36165
    },
    {
      "epoch": 5.900489396411093,
      "grad_norm": 0.3116625249385834,
      "learning_rate": 2.1517853014347262e-05,
      "loss": 0.0095,
      "num_input_tokens_seen": 78072352,
      "step": 36170
    },
    {
      "epoch": 5.901305057096248,
      "grad_norm": 0.06515955924987793,
      "learning_rate": 2.1510804557895847e-05,
      "loss": 0.0567,
      "num_input_tokens_seen": 78082816,
      "step": 36175
    },
    {
      "epoch": 5.902120717781403,
      "grad_norm": 2.8720502853393555,
      "learning_rate": 2.1503756384296323e-05,
      "loss": 0.2682,
      "num_input_tokens_seen": 78092288,
      "step": 36180
    },
    {
      "epoch": 5.902936378466558,
      "grad_norm": 0.4986286759376526,
      "learning_rate": 2.1496708494120043e-05,
      "loss": 0.0928,
      "num_input_tokens_seen": 78104608,
      "step": 36185
    },
    {
      "epoch": 5.903752039151713,
      "grad_norm": 0.01598019152879715,
      "learning_rate": 2.148966088793835e-05,
      "loss": 0.0145,
      "num_input_tokens_seen": 78116160,
      "step": 36190
    },
    {
      "epoch": 5.904567699836868,
      "grad_norm": 1.459521770477295,
      "learning_rate": 2.1482613566322558e-05,
      "loss": 0.2332,
      "num_input_tokens_seen": 78127200,
      "step": 36195
    },
    {
      "epoch": 5.9053833605220225,
      "grad_norm": 4.28659200668335,
      "learning_rate": 2.147556652984395e-05,
      "loss": 0.2466,
      "num_input_tokens_seen": 78136928,
      "step": 36200
    },
    {
      "epoch": 5.906199021207177,
      "grad_norm": 0.06842447817325592,
      "learning_rate": 2.1468519779073805e-05,
      "loss": 0.0482,
      "num_input_tokens_seen": 78148704,
      "step": 36205
    },
    {
      "epoch": 5.907014681892333,
      "grad_norm": 0.8169599175453186,
      "learning_rate": 2.146147331458337e-05,
      "loss": 0.0213,
      "num_input_tokens_seen": 78158752,
      "step": 36210
    },
    {
      "epoch": 5.907830342577488,
      "grad_norm": 0.37449151277542114,
      "learning_rate": 2.1454427136943858e-05,
      "loss": 0.3498,
      "num_input_tokens_seen": 78167968,
      "step": 36215
    },
    {
      "epoch": 5.908646003262643,
      "grad_norm": 3.2668418884277344,
      "learning_rate": 2.1447381246726473e-05,
      "loss": 0.1727,
      "num_input_tokens_seen": 78178688,
      "step": 36220
    },
    {
      "epoch": 5.9094616639477975,
      "grad_norm": 3.2942874431610107,
      "learning_rate": 2.144033564450239e-05,
      "loss": 0.1158,
      "num_input_tokens_seen": 78189376,
      "step": 36225
    },
    {
      "epoch": 5.910277324632952,
      "grad_norm": 0.04981905594468117,
      "learning_rate": 2.1433290330842764e-05,
      "loss": 0.0791,
      "num_input_tokens_seen": 78200000,
      "step": 36230
    },
    {
      "epoch": 5.911092985318108,
      "grad_norm": 6.152901649475098,
      "learning_rate": 2.142624530631872e-05,
      "loss": 0.1143,
      "num_input_tokens_seen": 78210592,
      "step": 36235
    },
    {
      "epoch": 5.911908646003263,
      "grad_norm": 3.7300734519958496,
      "learning_rate": 2.1419200571501363e-05,
      "loss": 0.0654,
      "num_input_tokens_seen": 78220896,
      "step": 36240
    },
    {
      "epoch": 5.912724306688418,
      "grad_norm": 11.314860343933105,
      "learning_rate": 2.141215612696177e-05,
      "loss": 0.0129,
      "num_input_tokens_seen": 78233088,
      "step": 36245
    },
    {
      "epoch": 5.9135399673735725,
      "grad_norm": 6.528377056121826,
      "learning_rate": 2.140511197327101e-05,
      "loss": 0.2043,
      "num_input_tokens_seen": 78243680,
      "step": 36250
    },
    {
      "epoch": 5.914355628058727,
      "grad_norm": 0.046376537531614304,
      "learning_rate": 2.139806811100012e-05,
      "loss": 0.0192,
      "num_input_tokens_seen": 78253344,
      "step": 36255
    },
    {
      "epoch": 5.915171288743883,
      "grad_norm": 0.03144540637731552,
      "learning_rate": 2.1391024540720102e-05,
      "loss": 0.1141,
      "num_input_tokens_seen": 78264864,
      "step": 36260
    },
    {
      "epoch": 5.915986949429038,
      "grad_norm": 0.15380115807056427,
      "learning_rate": 2.1383981263001947e-05,
      "loss": 0.0134,
      "num_input_tokens_seen": 78276000,
      "step": 36265
    },
    {
      "epoch": 5.916802610114193,
      "grad_norm": 3.3430352210998535,
      "learning_rate": 2.1376938278416615e-05,
      "loss": 0.1266,
      "num_input_tokens_seen": 78287360,
      "step": 36270
    },
    {
      "epoch": 5.917618270799347,
      "grad_norm": 0.07964406907558441,
      "learning_rate": 2.1369895587535052e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 78296608,
      "step": 36275
    },
    {
      "epoch": 5.918433931484502,
      "grad_norm": 2.8070991039276123,
      "learning_rate": 2.1362853190928172e-05,
      "loss": 0.0666,
      "num_input_tokens_seen": 78306848,
      "step": 36280
    },
    {
      "epoch": 5.919249592169657,
      "grad_norm": 0.9714038968086243,
      "learning_rate": 2.135581108916686e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 78317504,
      "step": 36285
    },
    {
      "epoch": 5.920065252854813,
      "grad_norm": 2.707551956176758,
      "learning_rate": 2.1348769282822e-05,
      "loss": 0.136,
      "num_input_tokens_seen": 78327360,
      "step": 36290
    },
    {
      "epoch": 5.920880913539968,
      "grad_norm": 0.10459740459918976,
      "learning_rate": 2.1341727772464425e-05,
      "loss": 0.1302,
      "num_input_tokens_seen": 78338656,
      "step": 36295
    },
    {
      "epoch": 5.921696574225122,
      "grad_norm": 1.3902908563613892,
      "learning_rate": 2.1334686558664964e-05,
      "loss": 0.0922,
      "num_input_tokens_seen": 78350464,
      "step": 36300
    },
    {
      "epoch": 5.922512234910277,
      "grad_norm": 3.597879648208618,
      "learning_rate": 2.1327645641994404e-05,
      "loss": 0.1789,
      "num_input_tokens_seen": 78360480,
      "step": 36305
    },
    {
      "epoch": 5.923327895595432,
      "grad_norm": 4.231083869934082,
      "learning_rate": 2.1320605023023522e-05,
      "loss": 0.0886,
      "num_input_tokens_seen": 78371264,
      "step": 36310
    },
    {
      "epoch": 5.924143556280587,
      "grad_norm": 3.2040674686431885,
      "learning_rate": 2.1313564702323064e-05,
      "loss": 0.1933,
      "num_input_tokens_seen": 78382176,
      "step": 36315
    },
    {
      "epoch": 5.924959216965743,
      "grad_norm": 0.34265559911727905,
      "learning_rate": 2.1306524680463758e-05,
      "loss": 0.2899,
      "num_input_tokens_seen": 78392704,
      "step": 36320
    },
    {
      "epoch": 5.925774877650897,
      "grad_norm": 0.8359548449516296,
      "learning_rate": 2.12994849580163e-05,
      "loss": 0.2695,
      "num_input_tokens_seen": 78404544,
      "step": 36325
    },
    {
      "epoch": 5.926590538336052,
      "grad_norm": 0.05384374409914017,
      "learning_rate": 2.129244553555137e-05,
      "loss": 0.0999,
      "num_input_tokens_seen": 78414400,
      "step": 36330
    },
    {
      "epoch": 5.927406199021207,
      "grad_norm": 0.1213223859667778,
      "learning_rate": 2.1285406413639616e-05,
      "loss": 0.0138,
      "num_input_tokens_seen": 78425664,
      "step": 36335
    },
    {
      "epoch": 5.928221859706362,
      "grad_norm": 0.28292325139045715,
      "learning_rate": 2.1278367592851668e-05,
      "loss": 0.01,
      "num_input_tokens_seen": 78437120,
      "step": 36340
    },
    {
      "epoch": 5.9290375203915175,
      "grad_norm": 3.51906156539917,
      "learning_rate": 2.1271329073758118e-05,
      "loss": 0.0215,
      "num_input_tokens_seen": 78447456,
      "step": 36345
    },
    {
      "epoch": 5.929853181076672,
      "grad_norm": 0.1483282446861267,
      "learning_rate": 2.1264290856929553e-05,
      "loss": 0.1043,
      "num_input_tokens_seen": 78457760,
      "step": 36350
    },
    {
      "epoch": 5.930668841761827,
      "grad_norm": 0.12273486703634262,
      "learning_rate": 2.125725294293653e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 78469248,
      "step": 36355
    },
    {
      "epoch": 5.931484502446982,
      "grad_norm": 0.05951046571135521,
      "learning_rate": 2.1250215332349575e-05,
      "loss": 0.1769,
      "num_input_tokens_seen": 78480864,
      "step": 36360
    },
    {
      "epoch": 5.932300163132137,
      "grad_norm": 3.0192644596099854,
      "learning_rate": 2.1243178025739193e-05,
      "loss": 0.1787,
      "num_input_tokens_seen": 78492224,
      "step": 36365
    },
    {
      "epoch": 5.933115823817292,
      "grad_norm": 0.5985738039016724,
      "learning_rate": 2.1236141023675855e-05,
      "loss": 0.0061,
      "num_input_tokens_seen": 78503072,
      "step": 36370
    },
    {
      "epoch": 5.933931484502447,
      "grad_norm": 0.15368081629276276,
      "learning_rate": 2.122910432673003e-05,
      "loss": 0.1992,
      "num_input_tokens_seen": 78511744,
      "step": 36375
    },
    {
      "epoch": 5.934747145187602,
      "grad_norm": 0.7402132153511047,
      "learning_rate": 2.122206793547214e-05,
      "loss": 0.1226,
      "num_input_tokens_seen": 78522176,
      "step": 36380
    },
    {
      "epoch": 5.935562805872757,
      "grad_norm": 0.0874972864985466,
      "learning_rate": 2.1215031850472593e-05,
      "loss": 0.0079,
      "num_input_tokens_seen": 78532928,
      "step": 36385
    },
    {
      "epoch": 5.936378466557912,
      "grad_norm": 0.14864759147167206,
      "learning_rate": 2.120799607230177e-05,
      "loss": 0.1188,
      "num_input_tokens_seen": 78543552,
      "step": 36390
    },
    {
      "epoch": 5.937194127243067,
      "grad_norm": 0.14092212915420532,
      "learning_rate": 2.1200960601530022e-05,
      "loss": 0.0116,
      "num_input_tokens_seen": 78553824,
      "step": 36395
    },
    {
      "epoch": 5.938009787928221,
      "grad_norm": 4.489637851715088,
      "learning_rate": 2.119392543872769e-05,
      "loss": 0.1303,
      "num_input_tokens_seen": 78565248,
      "step": 36400
    },
    {
      "epoch": 5.938825448613377,
      "grad_norm": 0.9154524803161621,
      "learning_rate": 2.1186890584465068e-05,
      "loss": 0.1014,
      "num_input_tokens_seen": 78577440,
      "step": 36405
    },
    {
      "epoch": 5.939641109298532,
      "grad_norm": 0.2515009641647339,
      "learning_rate": 2.1179856039312446e-05,
      "loss": 0.1446,
      "num_input_tokens_seen": 78588160,
      "step": 36410
    },
    {
      "epoch": 5.940456769983687,
      "grad_norm": 4.479397773742676,
      "learning_rate": 2.1172821803840077e-05,
      "loss": 0.1346,
      "num_input_tokens_seen": 78598976,
      "step": 36415
    },
    {
      "epoch": 5.941272430668842,
      "grad_norm": 1.1386377811431885,
      "learning_rate": 2.116578787861819e-05,
      "loss": 0.1998,
      "num_input_tokens_seen": 78608768,
      "step": 36420
    },
    {
      "epoch": 5.942088091353996,
      "grad_norm": 0.6867659687995911,
      "learning_rate": 2.1158754264216992e-05,
      "loss": 0.0703,
      "num_input_tokens_seen": 78620736,
      "step": 36425
    },
    {
      "epoch": 5.942903752039152,
      "grad_norm": 0.16088555753231049,
      "learning_rate": 2.1151720961206657e-05,
      "loss": 0.1494,
      "num_input_tokens_seen": 78629216,
      "step": 36430
    },
    {
      "epoch": 5.943719412724307,
      "grad_norm": 0.11808604747056961,
      "learning_rate": 2.1144687970157357e-05,
      "loss": 0.0134,
      "num_input_tokens_seen": 78639040,
      "step": 36435
    },
    {
      "epoch": 5.944535073409462,
      "grad_norm": 0.10915657877922058,
      "learning_rate": 2.1137655291639206e-05,
      "loss": 0.039,
      "num_input_tokens_seen": 78650624,
      "step": 36440
    },
    {
      "epoch": 5.945350734094617,
      "grad_norm": 0.13303032517433167,
      "learning_rate": 2.113062292622232e-05,
      "loss": 0.0145,
      "num_input_tokens_seen": 78662368,
      "step": 36445
    },
    {
      "epoch": 5.946166394779771,
      "grad_norm": 3.116680860519409,
      "learning_rate": 2.112359087447677e-05,
      "loss": 0.2113,
      "num_input_tokens_seen": 78673984,
      "step": 36450
    },
    {
      "epoch": 5.946982055464927,
      "grad_norm": 0.17844155430793762,
      "learning_rate": 2.11165591369726e-05,
      "loss": 0.0083,
      "num_input_tokens_seen": 78683328,
      "step": 36455
    },
    {
      "epoch": 5.947797716150082,
      "grad_norm": 17.343334197998047,
      "learning_rate": 2.110952771427986e-05,
      "loss": 0.039,
      "num_input_tokens_seen": 78694496,
      "step": 36460
    },
    {
      "epoch": 5.948613376835237,
      "grad_norm": 0.1877734214067459,
      "learning_rate": 2.110249660696855e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 78704832,
      "step": 36465
    },
    {
      "epoch": 5.9494290375203915,
      "grad_norm": 13.953911781311035,
      "learning_rate": 2.1095465815608637e-05,
      "loss": 0.132,
      "num_input_tokens_seen": 78715840,
      "step": 36470
    },
    {
      "epoch": 5.950244698205546,
      "grad_norm": 0.11150404065847397,
      "learning_rate": 2.1088435340770074e-05,
      "loss": 0.0852,
      "num_input_tokens_seen": 78726592,
      "step": 36475
    },
    {
      "epoch": 5.951060358890701,
      "grad_norm": 0.060464903712272644,
      "learning_rate": 2.108140518302279e-05,
      "loss": 0.1296,
      "num_input_tokens_seen": 78738048,
      "step": 36480
    },
    {
      "epoch": 5.951876019575856,
      "grad_norm": 6.174735069274902,
      "learning_rate": 2.107437534293669e-05,
      "loss": 0.112,
      "num_input_tokens_seen": 78749312,
      "step": 36485
    },
    {
      "epoch": 5.952691680261012,
      "grad_norm": 0.06503239274024963,
      "learning_rate": 2.106734582108164e-05,
      "loss": 0.2002,
      "num_input_tokens_seen": 78760512,
      "step": 36490
    },
    {
      "epoch": 5.9535073409461665,
      "grad_norm": 0.266341894865036,
      "learning_rate": 2.1060316618027493e-05,
      "loss": 0.0518,
      "num_input_tokens_seen": 78771072,
      "step": 36495
    },
    {
      "epoch": 5.954323001631321,
      "grad_norm": 0.05134322866797447,
      "learning_rate": 2.105328773434407e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 78781376,
      "step": 36500
    },
    {
      "epoch": 5.955138662316476,
      "grad_norm": 0.18325527012348175,
      "learning_rate": 2.1046259170601167e-05,
      "loss": 0.1604,
      "num_input_tokens_seen": 78792640,
      "step": 36505
    },
    {
      "epoch": 5.955954323001631,
      "grad_norm": 2.8084986209869385,
      "learning_rate": 2.1039230927368556e-05,
      "loss": 0.1192,
      "num_input_tokens_seen": 78803776,
      "step": 36510
    },
    {
      "epoch": 5.956769983686787,
      "grad_norm": 3.9901211261749268,
      "learning_rate": 2.1032203005215978e-05,
      "loss": 0.0995,
      "num_input_tokens_seen": 78813440,
      "step": 36515
    },
    {
      "epoch": 5.9575856443719415,
      "grad_norm": 0.11165003478527069,
      "learning_rate": 2.1025175404713167e-05,
      "loss": 0.2142,
      "num_input_tokens_seen": 78824736,
      "step": 36520
    },
    {
      "epoch": 5.958401305057096,
      "grad_norm": 0.04973118007183075,
      "learning_rate": 2.1018148126429797e-05,
      "loss": 0.1009,
      "num_input_tokens_seen": 78833920,
      "step": 36525
    },
    {
      "epoch": 5.959216965742251,
      "grad_norm": 0.05530374497175217,
      "learning_rate": 2.101112117093555e-05,
      "loss": 0.0538,
      "num_input_tokens_seen": 78845376,
      "step": 36530
    },
    {
      "epoch": 5.960032626427406,
      "grad_norm": 5.85557746887207,
      "learning_rate": 2.1004094538800058e-05,
      "loss": 0.0815,
      "num_input_tokens_seen": 78858016,
      "step": 36535
    },
    {
      "epoch": 5.960848287112562,
      "grad_norm": 0.14480511844158173,
      "learning_rate": 2.099706823059294e-05,
      "loss": 0.0884,
      "num_input_tokens_seen": 78869216,
      "step": 36540
    },
    {
      "epoch": 5.9616639477977165,
      "grad_norm": 0.5307287573814392,
      "learning_rate": 2.0990042246883777e-05,
      "loss": 0.0812,
      "num_input_tokens_seen": 78880096,
      "step": 36545
    },
    {
      "epoch": 5.962479608482871,
      "grad_norm": 0.09958609193563461,
      "learning_rate": 2.0983016588242145e-05,
      "loss": 0.1191,
      "num_input_tokens_seen": 78889664,
      "step": 36550
    },
    {
      "epoch": 5.963295269168026,
      "grad_norm": 0.035562992095947266,
      "learning_rate": 2.0975991255237562e-05,
      "loss": 0.0854,
      "num_input_tokens_seen": 78902240,
      "step": 36555
    },
    {
      "epoch": 5.964110929853181,
      "grad_norm": 5.524087905883789,
      "learning_rate": 2.0968966248439536e-05,
      "loss": 0.0837,
      "num_input_tokens_seen": 78911872,
      "step": 36560
    },
    {
      "epoch": 5.964926590538336,
      "grad_norm": 0.17915554344654083,
      "learning_rate": 2.096194156841757e-05,
      "loss": 0.0053,
      "num_input_tokens_seen": 78922304,
      "step": 36565
    },
    {
      "epoch": 5.9657422512234906,
      "grad_norm": 0.11609411239624023,
      "learning_rate": 2.0954917215741113e-05,
      "loss": 0.0124,
      "num_input_tokens_seen": 78933408,
      "step": 36570
    },
    {
      "epoch": 5.966557911908646,
      "grad_norm": 0.09525105357170105,
      "learning_rate": 2.0947893190979588e-05,
      "loss": 0.1277,
      "num_input_tokens_seen": 78944704,
      "step": 36575
    },
    {
      "epoch": 5.967373572593801,
      "grad_norm": 0.12271397560834885,
      "learning_rate": 2.09408694947024e-05,
      "loss": 0.0382,
      "num_input_tokens_seen": 78955584,
      "step": 36580
    },
    {
      "epoch": 5.968189233278956,
      "grad_norm": 2.0783212184906006,
      "learning_rate": 2.0933846127478928e-05,
      "loss": 0.0751,
      "num_input_tokens_seen": 78966272,
      "step": 36585
    },
    {
      "epoch": 5.969004893964111,
      "grad_norm": 0.09040044993162155,
      "learning_rate": 2.092682308987852e-05,
      "loss": 0.1292,
      "num_input_tokens_seen": 78976608,
      "step": 36590
    },
    {
      "epoch": 5.9698205546492655,
      "grad_norm": 0.09519308060407639,
      "learning_rate": 2.0919800382470503e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 78987360,
      "step": 36595
    },
    {
      "epoch": 5.970636215334421,
      "grad_norm": 0.10772380232810974,
      "learning_rate": 2.091277800582416e-05,
      "loss": 0.0503,
      "num_input_tokens_seen": 78999040,
      "step": 36600
    },
    {
      "epoch": 5.971451876019576,
      "grad_norm": 0.14227809011936188,
      "learning_rate": 2.0905755960508778e-05,
      "loss": 0.0045,
      "num_input_tokens_seen": 79010112,
      "step": 36605
    },
    {
      "epoch": 5.972267536704731,
      "grad_norm": 0.15200011432170868,
      "learning_rate": 2.0898734247093592e-05,
      "loss": 0.0059,
      "num_input_tokens_seen": 79020416,
      "step": 36610
    },
    {
      "epoch": 5.973083197389886,
      "grad_norm": 7.089992046356201,
      "learning_rate": 2.0891712866147812e-05,
      "loss": 0.2296,
      "num_input_tokens_seen": 79031648,
      "step": 36615
    },
    {
      "epoch": 5.9738988580750405,
      "grad_norm": 0.09132542461156845,
      "learning_rate": 2.088469181824064e-05,
      "loss": 0.0704,
      "num_input_tokens_seen": 79042560,
      "step": 36620
    },
    {
      "epoch": 5.974714518760196,
      "grad_norm": 0.3920039236545563,
      "learning_rate": 2.0877671103941228e-05,
      "loss": 0.1477,
      "num_input_tokens_seen": 79052672,
      "step": 36625
    },
    {
      "epoch": 5.975530179445351,
      "grad_norm": 2.6713600158691406,
      "learning_rate": 2.0870650723818706e-05,
      "loss": 0.3068,
      "num_input_tokens_seen": 79063744,
      "step": 36630
    },
    {
      "epoch": 5.976345840130506,
      "grad_norm": 3.448038101196289,
      "learning_rate": 2.0863630678442196e-05,
      "loss": 0.0742,
      "num_input_tokens_seen": 79075424,
      "step": 36635
    },
    {
      "epoch": 5.977161500815661,
      "grad_norm": 0.18815290927886963,
      "learning_rate": 2.0856610968380768e-05,
      "loss": 0.0229,
      "num_input_tokens_seen": 79085792,
      "step": 36640
    },
    {
      "epoch": 5.9779771615008155,
      "grad_norm": 0.16216164827346802,
      "learning_rate": 2.0849591594203482e-05,
      "loss": 0.0118,
      "num_input_tokens_seen": 79096704,
      "step": 36645
    },
    {
      "epoch": 5.97879282218597,
      "grad_norm": 2.9226410388946533,
      "learning_rate": 2.0842572556479355e-05,
      "loss": 0.2137,
      "num_input_tokens_seen": 79107328,
      "step": 36650
    },
    {
      "epoch": 5.979608482871125,
      "grad_norm": 0.35585007071495056,
      "learning_rate": 2.0835553855777396e-05,
      "loss": 0.0105,
      "num_input_tokens_seen": 79118144,
      "step": 36655
    },
    {
      "epoch": 5.980424143556281,
      "grad_norm": 0.021596642211079597,
      "learning_rate": 2.082853549266656e-05,
      "loss": 0.1753,
      "num_input_tokens_seen": 79129568,
      "step": 36660
    },
    {
      "epoch": 5.981239804241436,
      "grad_norm": 0.1883329153060913,
      "learning_rate": 2.0821517467715815e-05,
      "loss": 0.0965,
      "num_input_tokens_seen": 79139936,
      "step": 36665
    },
    {
      "epoch": 5.9820554649265905,
      "grad_norm": 0.22896647453308105,
      "learning_rate": 2.0814499781494057e-05,
      "loss": 0.0597,
      "num_input_tokens_seen": 79149664,
      "step": 36670
    },
    {
      "epoch": 5.982871125611745,
      "grad_norm": 0.12242002040147781,
      "learning_rate": 2.0807482434570187e-05,
      "loss": 0.0531,
      "num_input_tokens_seen": 79161280,
      "step": 36675
    },
    {
      "epoch": 5.9836867862969,
      "grad_norm": 0.0653303861618042,
      "learning_rate": 2.080046542751307e-05,
      "loss": 0.1036,
      "num_input_tokens_seen": 79171488,
      "step": 36680
    },
    {
      "epoch": 5.984502446982056,
      "grad_norm": 0.14149434864521027,
      "learning_rate": 2.079344876089152e-05,
      "loss": 0.0097,
      "num_input_tokens_seen": 79182656,
      "step": 36685
    },
    {
      "epoch": 5.985318107667211,
      "grad_norm": 0.21198756992816925,
      "learning_rate": 2.078643243527437e-05,
      "loss": 0.2048,
      "num_input_tokens_seen": 79194016,
      "step": 36690
    },
    {
      "epoch": 5.986133768352365,
      "grad_norm": 1.6793668270111084,
      "learning_rate": 2.0779416451230382e-05,
      "loss": 0.1456,
      "num_input_tokens_seen": 79205472,
      "step": 36695
    },
    {
      "epoch": 5.98694942903752,
      "grad_norm": 0.07833819836378098,
      "learning_rate": 2.0772400809328314e-05,
      "loss": 0.008,
      "num_input_tokens_seen": 79215904,
      "step": 36700
    },
    {
      "epoch": 5.987765089722675,
      "grad_norm": 0.06340375542640686,
      "learning_rate": 2.0765385510136884e-05,
      "loss": 0.0598,
      "num_input_tokens_seen": 79227776,
      "step": 36705
    },
    {
      "epoch": 5.988580750407831,
      "grad_norm": 0.02490225061774254,
      "learning_rate": 2.0758370554224793e-05,
      "loss": 0.1367,
      "num_input_tokens_seen": 79239488,
      "step": 36710
    },
    {
      "epoch": 5.989396411092986,
      "grad_norm": 11.206292152404785,
      "learning_rate": 2.0751355942160706e-05,
      "loss": 0.1416,
      "num_input_tokens_seen": 79249600,
      "step": 36715
    },
    {
      "epoch": 5.99021207177814,
      "grad_norm": 0.42250674962997437,
      "learning_rate": 2.0744341674513264e-05,
      "loss": 0.1338,
      "num_input_tokens_seen": 79259840,
      "step": 36720
    },
    {
      "epoch": 5.991027732463295,
      "grad_norm": 0.16194994747638702,
      "learning_rate": 2.0737327751851075e-05,
      "loss": 0.0856,
      "num_input_tokens_seen": 79271488,
      "step": 36725
    },
    {
      "epoch": 5.99184339314845,
      "grad_norm": 0.08617598563432693,
      "learning_rate": 2.0730314174742733e-05,
      "loss": 0.0053,
      "num_input_tokens_seen": 79282752,
      "step": 36730
    },
    {
      "epoch": 5.992659053833605,
      "grad_norm": 2.626254081726074,
      "learning_rate": 2.0723300943756783e-05,
      "loss": 0.0878,
      "num_input_tokens_seen": 79294336,
      "step": 36735
    },
    {
      "epoch": 5.993474714518761,
      "grad_norm": 2.470318555831909,
      "learning_rate": 2.0716288059461764e-05,
      "loss": 0.3177,
      "num_input_tokens_seen": 79306272,
      "step": 36740
    },
    {
      "epoch": 5.994290375203915,
      "grad_norm": 0.18605534732341766,
      "learning_rate": 2.0709275522426158e-05,
      "loss": 0.0187,
      "num_input_tokens_seen": 79317120,
      "step": 36745
    },
    {
      "epoch": 5.99510603588907,
      "grad_norm": 0.06405258178710938,
      "learning_rate": 2.070226333321846e-05,
      "loss": 0.0039,
      "num_input_tokens_seen": 79327200,
      "step": 36750
    },
    {
      "epoch": 5.995921696574225,
      "grad_norm": 8.52464771270752,
      "learning_rate": 2.06952514924071e-05,
      "loss": 0.1118,
      "num_input_tokens_seen": 79337888,
      "step": 36755
    },
    {
      "epoch": 5.99673735725938,
      "grad_norm": 0.3218190371990204,
      "learning_rate": 2.0688240000560498e-05,
      "loss": 0.1347,
      "num_input_tokens_seen": 79348480,
      "step": 36760
    },
    {
      "epoch": 5.997553017944535,
      "grad_norm": 0.3105626106262207,
      "learning_rate": 2.0681228858247038e-05,
      "loss": 0.1995,
      "num_input_tokens_seen": 79360480,
      "step": 36765
    },
    {
      "epoch": 5.99836867862969,
      "grad_norm": 0.2892025411128998,
      "learning_rate": 2.067421806603508e-05,
      "loss": 0.0095,
      "num_input_tokens_seen": 79370912,
      "step": 36770
    },
    {
      "epoch": 5.999184339314845,
      "grad_norm": 0.10774438828229904,
      "learning_rate": 2.0667207624492943e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 79380288,
      "step": 36775
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.07041917741298676,
      "learning_rate": 2.0660197534188952e-05,
      "loss": 0.0755,
      "num_input_tokens_seen": 79389648,
      "step": 36780
    },
    {
      "epoch": 6.0,
      "eval_loss": 0.17715702950954437,
      "eval_runtime": 132.8411,
      "eval_samples_per_second": 20.513,
      "eval_steps_per_second": 5.134,
      "num_input_tokens_seen": 79389648,
      "step": 36780
    },
    {
      "epoch": 6.000815660685155,
      "grad_norm": 0.17613597214221954,
      "learning_rate": 2.065318779569137e-05,
      "loss": 0.0061,
      "num_input_tokens_seen": 79400656,
      "step": 36785
    },
    {
      "epoch": 6.00163132137031,
      "grad_norm": 14.462206840515137,
      "learning_rate": 2.064617840956844e-05,
      "loss": 0.0414,
      "num_input_tokens_seen": 79412656,
      "step": 36790
    },
    {
      "epoch": 6.002446982055465,
      "grad_norm": 0.06690766662359238,
      "learning_rate": 2.063916937638838e-05,
      "loss": 0.0082,
      "num_input_tokens_seen": 79424048,
      "step": 36795
    },
    {
      "epoch": 6.00326264274062,
      "grad_norm": 0.13082489371299744,
      "learning_rate": 2.063216069671937e-05,
      "loss": 0.0717,
      "num_input_tokens_seen": 79434896,
      "step": 36800
    },
    {
      "epoch": 6.004078303425775,
      "grad_norm": 0.05276201665401459,
      "learning_rate": 2.0625152371129585e-05,
      "loss": 0.1992,
      "num_input_tokens_seen": 79445680,
      "step": 36805
    },
    {
      "epoch": 6.00489396411093,
      "grad_norm": 0.7351275682449341,
      "learning_rate": 2.0618144400187142e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 79456592,
      "step": 36810
    },
    {
      "epoch": 6.005709624796085,
      "grad_norm": 5.1559271812438965,
      "learning_rate": 2.061113678446015e-05,
      "loss": 0.0637,
      "num_input_tokens_seen": 79466992,
      "step": 36815
    },
    {
      "epoch": 6.006525285481239,
      "grad_norm": 0.0848996490240097,
      "learning_rate": 2.0604129524516676e-05,
      "loss": 0.0058,
      "num_input_tokens_seen": 79477168,
      "step": 36820
    },
    {
      "epoch": 6.007340946166395,
      "grad_norm": 6.075421333312988,
      "learning_rate": 2.059712262092477e-05,
      "loss": 0.1392,
      "num_input_tokens_seen": 79488368,
      "step": 36825
    },
    {
      "epoch": 6.00815660685155,
      "grad_norm": 0.07987889647483826,
      "learning_rate": 2.0590116074252438e-05,
      "loss": 0.0938,
      "num_input_tokens_seen": 79498736,
      "step": 36830
    },
    {
      "epoch": 6.008972267536705,
      "grad_norm": 0.07463080435991287,
      "learning_rate": 2.058310988506768e-05,
      "loss": 0.2583,
      "num_input_tokens_seen": 79508656,
      "step": 36835
    },
    {
      "epoch": 6.00978792822186,
      "grad_norm": 2.687781572341919,
      "learning_rate": 2.057610405393844e-05,
      "loss": 0.21,
      "num_input_tokens_seen": 79519696,
      "step": 36840
    },
    {
      "epoch": 6.010603588907014,
      "grad_norm": 0.08558306097984314,
      "learning_rate": 2.0569098581432655e-05,
      "loss": 0.2366,
      "num_input_tokens_seen": 79529776,
      "step": 36845
    },
    {
      "epoch": 6.011419249592169,
      "grad_norm": 2.86208438873291,
      "learning_rate": 2.0562093468118225e-05,
      "loss": 0.1569,
      "num_input_tokens_seen": 79541392,
      "step": 36850
    },
    {
      "epoch": 6.012234910277325,
      "grad_norm": 1.8412063121795654,
      "learning_rate": 2.055508871456301e-05,
      "loss": 0.0642,
      "num_input_tokens_seen": 79552720,
      "step": 36855
    },
    {
      "epoch": 6.01305057096248,
      "grad_norm": 0.2715950012207031,
      "learning_rate": 2.054808432133486e-05,
      "loss": 0.2212,
      "num_input_tokens_seen": 79563888,
      "step": 36860
    },
    {
      "epoch": 6.013866231647635,
      "grad_norm": 0.06896059960126877,
      "learning_rate": 2.0541080289001584e-05,
      "loss": 0.1458,
      "num_input_tokens_seen": 79575760,
      "step": 36865
    },
    {
      "epoch": 6.014681892332789,
      "grad_norm": 0.25679296255111694,
      "learning_rate": 2.0534076618130965e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 79587408,
      "step": 36870
    },
    {
      "epoch": 6.015497553017944,
      "grad_norm": 2.22693133354187,
      "learning_rate": 2.0527073309290755e-05,
      "loss": 0.0755,
      "num_input_tokens_seen": 79598352,
      "step": 36875
    },
    {
      "epoch": 6.0163132137031,
      "grad_norm": 3.286787509918213,
      "learning_rate": 2.0520070363048667e-05,
      "loss": 0.1579,
      "num_input_tokens_seen": 79609520,
      "step": 36880
    },
    {
      "epoch": 6.017128874388255,
      "grad_norm": 0.23035003244876862,
      "learning_rate": 2.0513067779972415e-05,
      "loss": 0.0243,
      "num_input_tokens_seen": 79621712,
      "step": 36885
    },
    {
      "epoch": 6.0179445350734095,
      "grad_norm": 0.1889280080795288,
      "learning_rate": 2.0506065560629655e-05,
      "loss": 0.0503,
      "num_input_tokens_seen": 79630576,
      "step": 36890
    },
    {
      "epoch": 6.018760195758564,
      "grad_norm": 0.10859799385070801,
      "learning_rate": 2.0499063705588024e-05,
      "loss": 0.1151,
      "num_input_tokens_seen": 79641360,
      "step": 36895
    },
    {
      "epoch": 6.019575856443719,
      "grad_norm": 0.1558310091495514,
      "learning_rate": 2.0492062215415125e-05,
      "loss": 0.0075,
      "num_input_tokens_seen": 79651792,
      "step": 36900
    },
    {
      "epoch": 6.020391517128874,
      "grad_norm": 0.08306483179330826,
      "learning_rate": 2.048506109067854e-05,
      "loss": 0.0338,
      "num_input_tokens_seen": 79661488,
      "step": 36905
    },
    {
      "epoch": 6.02120717781403,
      "grad_norm": 0.15285880863666534,
      "learning_rate": 2.047806033194581e-05,
      "loss": 0.1687,
      "num_input_tokens_seen": 79671856,
      "step": 36910
    },
    {
      "epoch": 6.0220228384991845,
      "grad_norm": 0.028413239866495132,
      "learning_rate": 2.0471059939784447e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 79681232,
      "step": 36915
    },
    {
      "epoch": 6.022838499184339,
      "grad_norm": 9.373331069946289,
      "learning_rate": 2.046405991476195e-05,
      "loss": 0.3035,
      "num_input_tokens_seen": 79691728,
      "step": 36920
    },
    {
      "epoch": 6.023654159869494,
      "grad_norm": 0.07682496309280396,
      "learning_rate": 2.045706025744577e-05,
      "loss": 0.0141,
      "num_input_tokens_seen": 79702672,
      "step": 36925
    },
    {
      "epoch": 6.024469820554649,
      "grad_norm": 0.07791969925165176,
      "learning_rate": 2.045006096840334e-05,
      "loss": 0.0059,
      "num_input_tokens_seen": 79713552,
      "step": 36930
    },
    {
      "epoch": 6.025285481239805,
      "grad_norm": 0.03607427701354027,
      "learning_rate": 2.044306204820205e-05,
      "loss": 0.0057,
      "num_input_tokens_seen": 79725296,
      "step": 36935
    },
    {
      "epoch": 6.0261011419249595,
      "grad_norm": 3.750582218170166,
      "learning_rate": 2.0436063497409274e-05,
      "loss": 0.1002,
      "num_input_tokens_seen": 79735792,
      "step": 36940
    },
    {
      "epoch": 6.026916802610114,
      "grad_norm": 0.12163159996271133,
      "learning_rate": 2.042906531659235e-05,
      "loss": 0.0039,
      "num_input_tokens_seen": 79746576,
      "step": 36945
    },
    {
      "epoch": 6.027732463295269,
      "grad_norm": 3.5392658710479736,
      "learning_rate": 2.042206750631858e-05,
      "loss": 0.1053,
      "num_input_tokens_seen": 79756496,
      "step": 36950
    },
    {
      "epoch": 6.028548123980424,
      "grad_norm": 0.13679811358451843,
      "learning_rate": 2.041507006715525e-05,
      "loss": 0.0346,
      "num_input_tokens_seen": 79767600,
      "step": 36955
    },
    {
      "epoch": 6.029363784665579,
      "grad_norm": 0.03501264750957489,
      "learning_rate": 2.0408072999669604e-05,
      "loss": 0.1284,
      "num_input_tokens_seen": 79779056,
      "step": 36960
    },
    {
      "epoch": 6.0301794453507345,
      "grad_norm": 4.524255752563477,
      "learning_rate": 2.0401076304428857e-05,
      "loss": 0.2575,
      "num_input_tokens_seen": 79790672,
      "step": 36965
    },
    {
      "epoch": 6.030995106035889,
      "grad_norm": 23.41586685180664,
      "learning_rate": 2.0394079982000198e-05,
      "loss": 0.1424,
      "num_input_tokens_seen": 79800144,
      "step": 36970
    },
    {
      "epoch": 6.031810766721044,
      "grad_norm": 4.102931022644043,
      "learning_rate": 2.0387084032950787e-05,
      "loss": 0.2014,
      "num_input_tokens_seen": 79811056,
      "step": 36975
    },
    {
      "epoch": 6.032626427406199,
      "grad_norm": 0.0753416195511818,
      "learning_rate": 2.038008845784775e-05,
      "loss": 0.12,
      "num_input_tokens_seen": 79821616,
      "step": 36980
    },
    {
      "epoch": 6.033442088091354,
      "grad_norm": 0.20337048172950745,
      "learning_rate": 2.0373093257258184e-05,
      "loss": 0.1088,
      "num_input_tokens_seen": 79833104,
      "step": 36985
    },
    {
      "epoch": 6.034257748776509,
      "grad_norm": 0.09057068079710007,
      "learning_rate": 2.0366098431749152e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 79842832,
      "step": 36990
    },
    {
      "epoch": 6.035073409461664,
      "grad_norm": 0.07087525725364685,
      "learning_rate": 2.0359103981887695e-05,
      "loss": 0.0056,
      "num_input_tokens_seen": 79853840,
      "step": 36995
    },
    {
      "epoch": 6.035889070146819,
      "grad_norm": 0.17854323983192444,
      "learning_rate": 2.035210990824082e-05,
      "loss": 0.095,
      "num_input_tokens_seen": 79865488,
      "step": 37000
    },
    {
      "epoch": 6.036704730831974,
      "grad_norm": 0.583503246307373,
      "learning_rate": 2.0345116211375496e-05,
      "loss": 0.2521,
      "num_input_tokens_seen": 79874640,
      "step": 37005
    },
    {
      "epoch": 6.037520391517129,
      "grad_norm": 0.37706518173217773,
      "learning_rate": 2.0338122891858677e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 79885552,
      "step": 37010
    },
    {
      "epoch": 6.0383360522022835,
      "grad_norm": 0.09744422882795334,
      "learning_rate": 2.0331129950257266e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 79896336,
      "step": 37015
    },
    {
      "epoch": 6.039151712887439,
      "grad_norm": 10.079150199890137,
      "learning_rate": 2.0324137387138152e-05,
      "loss": 0.1087,
      "num_input_tokens_seen": 79907184,
      "step": 37020
    },
    {
      "epoch": 6.039967373572594,
      "grad_norm": 5.770218372344971,
      "learning_rate": 2.031714520306819e-05,
      "loss": 0.1118,
      "num_input_tokens_seen": 79917936,
      "step": 37025
    },
    {
      "epoch": 6.040783034257749,
      "grad_norm": 17.64241600036621,
      "learning_rate": 2.0310153398614192e-05,
      "loss": 0.2081,
      "num_input_tokens_seen": 79930000,
      "step": 37030
    },
    {
      "epoch": 6.041598694942904,
      "grad_norm": 0.13271000981330872,
      "learning_rate": 2.030316197434296e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 79941456,
      "step": 37035
    },
    {
      "epoch": 6.0424143556280585,
      "grad_norm": 0.11974164098501205,
      "learning_rate": 2.0296170930821245e-05,
      "loss": 0.018,
      "num_input_tokens_seen": 79951440,
      "step": 37040
    },
    {
      "epoch": 6.043230016313213,
      "grad_norm": 0.050197869539260864,
      "learning_rate": 2.028918026861579e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 79962064,
      "step": 37045
    },
    {
      "epoch": 6.044045676998369,
      "grad_norm": 0.09011278301477432,
      "learning_rate": 2.0282189988293276e-05,
      "loss": 0.0035,
      "num_input_tokens_seen": 79974160,
      "step": 37050
    },
    {
      "epoch": 6.044861337683524,
      "grad_norm": 0.041562121361494064,
      "learning_rate": 2.0275200090420376e-05,
      "loss": 0.0028,
      "num_input_tokens_seen": 79985072,
      "step": 37055
    },
    {
      "epoch": 6.045676998368679,
      "grad_norm": 4.7424421310424805,
      "learning_rate": 2.026821057556374e-05,
      "loss": 0.0817,
      "num_input_tokens_seen": 79996784,
      "step": 37060
    },
    {
      "epoch": 6.0464926590538335,
      "grad_norm": 0.10275337845087051,
      "learning_rate": 2.026122144428996e-05,
      "loss": 0.0043,
      "num_input_tokens_seen": 80008496,
      "step": 37065
    },
    {
      "epoch": 6.047308319738988,
      "grad_norm": 3.4851765632629395,
      "learning_rate": 2.0254232697165616e-05,
      "loss": 0.1343,
      "num_input_tokens_seen": 80019600,
      "step": 37070
    },
    {
      "epoch": 6.048123980424143,
      "grad_norm": 0.07727508246898651,
      "learning_rate": 2.0247244334757248e-05,
      "loss": 0.0035,
      "num_input_tokens_seen": 80030864,
      "step": 37075
    },
    {
      "epoch": 6.048939641109299,
      "grad_norm": 2.237130880355835,
      "learning_rate": 2.0240256357631367e-05,
      "loss": 0.0088,
      "num_input_tokens_seen": 80041936,
      "step": 37080
    },
    {
      "epoch": 6.049755301794454,
      "grad_norm": 0.9212360382080078,
      "learning_rate": 2.023326876635446e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 80053552,
      "step": 37085
    },
    {
      "epoch": 6.0505709624796085,
      "grad_norm": 2.9676239490509033,
      "learning_rate": 2.022628156149297e-05,
      "loss": 0.2996,
      "num_input_tokens_seen": 80064688,
      "step": 37090
    },
    {
      "epoch": 6.051386623164763,
      "grad_norm": 0.2867843210697174,
      "learning_rate": 2.021929474361331e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 80076176,
      "step": 37095
    },
    {
      "epoch": 6.052202283849918,
      "grad_norm": 0.06813370436429977,
      "learning_rate": 2.0212308313281886e-05,
      "loss": 0.0049,
      "num_input_tokens_seen": 80085968,
      "step": 37100
    },
    {
      "epoch": 6.053017944535074,
      "grad_norm": 2.913398027420044,
      "learning_rate": 2.0205322271065042e-05,
      "loss": 0.3051,
      "num_input_tokens_seen": 80096048,
      "step": 37105
    },
    {
      "epoch": 6.053833605220229,
      "grad_norm": 37.044898986816406,
      "learning_rate": 2.01983366175291e-05,
      "loss": 0.0585,
      "num_input_tokens_seen": 80106608,
      "step": 37110
    },
    {
      "epoch": 6.054649265905383,
      "grad_norm": 0.02120068296790123,
      "learning_rate": 2.0191351353240363e-05,
      "loss": 0.1304,
      "num_input_tokens_seen": 80116400,
      "step": 37115
    },
    {
      "epoch": 6.055464926590538,
      "grad_norm": 11.320032119750977,
      "learning_rate": 2.0184366478765078e-05,
      "loss": 0.1768,
      "num_input_tokens_seen": 80127024,
      "step": 37120
    },
    {
      "epoch": 6.056280587275693,
      "grad_norm": 0.06965632736682892,
      "learning_rate": 2.017738199466948e-05,
      "loss": 0.1013,
      "num_input_tokens_seen": 80137936,
      "step": 37125
    },
    {
      "epoch": 6.057096247960848,
      "grad_norm": 6.816587448120117,
      "learning_rate": 2.0170397901519766e-05,
      "loss": 0.0887,
      "num_input_tokens_seen": 80148752,
      "step": 37130
    },
    {
      "epoch": 6.057911908646004,
      "grad_norm": 0.11578565090894699,
      "learning_rate": 2.016341419988211e-05,
      "loss": 0.1381,
      "num_input_tokens_seen": 80159856,
      "step": 37135
    },
    {
      "epoch": 6.058727569331158,
      "grad_norm": 2.500230312347412,
      "learning_rate": 2.0156430890322627e-05,
      "loss": 0.0974,
      "num_input_tokens_seen": 80172656,
      "step": 37140
    },
    {
      "epoch": 6.059543230016313,
      "grad_norm": 0.04798300191760063,
      "learning_rate": 2.0149447973407443e-05,
      "loss": 0.0823,
      "num_input_tokens_seen": 80184240,
      "step": 37145
    },
    {
      "epoch": 6.060358890701468,
      "grad_norm": 0.13084805011749268,
      "learning_rate": 2.0142465449702612e-05,
      "loss": 0.024,
      "num_input_tokens_seen": 80195184,
      "step": 37150
    },
    {
      "epoch": 6.061174551386623,
      "grad_norm": 0.17297035455703735,
      "learning_rate": 2.0135483319774183e-05,
      "loss": 0.0075,
      "num_input_tokens_seen": 80205296,
      "step": 37155
    },
    {
      "epoch": 6.061990212071779,
      "grad_norm": 0.02693340927362442,
      "learning_rate": 2.012850158418816e-05,
      "loss": 0.0184,
      "num_input_tokens_seen": 80216496,
      "step": 37160
    },
    {
      "epoch": 6.062805872756933,
      "grad_norm": 13.927870750427246,
      "learning_rate": 2.0121520243510512e-05,
      "loss": 0.0787,
      "num_input_tokens_seen": 80227280,
      "step": 37165
    },
    {
      "epoch": 6.063621533442088,
      "grad_norm": 0.19517157971858978,
      "learning_rate": 2.0114539298307188e-05,
      "loss": 0.1433,
      "num_input_tokens_seen": 80238704,
      "step": 37170
    },
    {
      "epoch": 6.064437194127243,
      "grad_norm": 3.330983877182007,
      "learning_rate": 2.0107558749144096e-05,
      "loss": 0.1432,
      "num_input_tokens_seen": 80248560,
      "step": 37175
    },
    {
      "epoch": 6.065252854812398,
      "grad_norm": 0.045291390269994736,
      "learning_rate": 2.0100578596587116e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 80258448,
      "step": 37180
    },
    {
      "epoch": 6.066068515497553,
      "grad_norm": 0.05958879366517067,
      "learning_rate": 2.0093598841202092e-05,
      "loss": 0.0055,
      "num_input_tokens_seen": 80269168,
      "step": 37185
    },
    {
      "epoch": 6.066884176182708,
      "grad_norm": 0.08888333290815353,
      "learning_rate": 2.0086619483554847e-05,
      "loss": 0.0052,
      "num_input_tokens_seen": 80278672,
      "step": 37190
    },
    {
      "epoch": 6.067699836867863,
      "grad_norm": 0.04317786172032356,
      "learning_rate": 2.0079640524211153e-05,
      "loss": 0.0201,
      "num_input_tokens_seen": 80290128,
      "step": 37195
    },
    {
      "epoch": 6.068515497553018,
      "grad_norm": 0.04700427129864693,
      "learning_rate": 2.0072661963736752e-05,
      "loss": 0.2334,
      "num_input_tokens_seen": 80300816,
      "step": 37200
    },
    {
      "epoch": 6.069331158238173,
      "grad_norm": 0.13251464068889618,
      "learning_rate": 2.006568380269739e-05,
      "loss": 0.0027,
      "num_input_tokens_seen": 80311120,
      "step": 37205
    },
    {
      "epoch": 6.070146818923328,
      "grad_norm": 0.3197738528251648,
      "learning_rate": 2.005870604165873e-05,
      "loss": 0.0059,
      "num_input_tokens_seen": 80321968,
      "step": 37210
    },
    {
      "epoch": 6.0709624796084825,
      "grad_norm": 3.1156668663024902,
      "learning_rate": 2.005172868118643e-05,
      "loss": 0.1079,
      "num_input_tokens_seen": 80332176,
      "step": 37215
    },
    {
      "epoch": 6.071778140293638,
      "grad_norm": 0.3527423143386841,
      "learning_rate": 2.004475172184611e-05,
      "loss": 0.0043,
      "num_input_tokens_seen": 80343376,
      "step": 37220
    },
    {
      "epoch": 6.072593800978793,
      "grad_norm": 0.1022476777434349,
      "learning_rate": 2.0037775164203356e-05,
      "loss": 0.0034,
      "num_input_tokens_seen": 80354032,
      "step": 37225
    },
    {
      "epoch": 6.073409461663948,
      "grad_norm": 0.08814200013875961,
      "learning_rate": 2.0030799008823727e-05,
      "loss": 0.0036,
      "num_input_tokens_seen": 80365072,
      "step": 37230
    },
    {
      "epoch": 6.074225122349103,
      "grad_norm": 0.051949575543403625,
      "learning_rate": 2.0023823256272748e-05,
      "loss": 0.0057,
      "num_input_tokens_seen": 80374512,
      "step": 37235
    },
    {
      "epoch": 6.075040783034257,
      "grad_norm": 0.028245143592357635,
      "learning_rate": 2.00168479071159e-05,
      "loss": 0.0777,
      "num_input_tokens_seen": 80384368,
      "step": 37240
    },
    {
      "epoch": 6.075856443719413,
      "grad_norm": 5.9619951248168945,
      "learning_rate": 2.0009872961918648e-05,
      "loss": 0.0166,
      "num_input_tokens_seen": 80395152,
      "step": 37245
    },
    {
      "epoch": 6.076672104404568,
      "grad_norm": 2.1715729236602783,
      "learning_rate": 2.0002898421246414e-05,
      "loss": 0.016,
      "num_input_tokens_seen": 80406288,
      "step": 37250
    },
    {
      "epoch": 6.077487765089723,
      "grad_norm": 0.043432507663965225,
      "learning_rate": 1.9995924285664587e-05,
      "loss": 0.0255,
      "num_input_tokens_seen": 80417040,
      "step": 37255
    },
    {
      "epoch": 6.078303425774878,
      "grad_norm": 0.0427730567753315,
      "learning_rate": 1.9988950555738528e-05,
      "loss": 0.1516,
      "num_input_tokens_seen": 80428112,
      "step": 37260
    },
    {
      "epoch": 6.079119086460032,
      "grad_norm": 0.06148674339056015,
      "learning_rate": 1.9981977232033563e-05,
      "loss": 0.0062,
      "num_input_tokens_seen": 80438288,
      "step": 37265
    },
    {
      "epoch": 6.079934747145187,
      "grad_norm": 0.1648724526166916,
      "learning_rate": 1.9975004315114988e-05,
      "loss": 0.5308,
      "num_input_tokens_seen": 80449360,
      "step": 37270
    },
    {
      "epoch": 6.080750407830343,
      "grad_norm": 1.8730205297470093,
      "learning_rate": 1.9968031805548056e-05,
      "loss": 0.006,
      "num_input_tokens_seen": 80460752,
      "step": 37275
    },
    {
      "epoch": 6.081566068515498,
      "grad_norm": 5.308781147003174,
      "learning_rate": 1.9961059703898e-05,
      "loss": 0.0523,
      "num_input_tokens_seen": 80472048,
      "step": 37280
    },
    {
      "epoch": 6.082381729200653,
      "grad_norm": 0.09825458377599716,
      "learning_rate": 1.9954088010730003e-05,
      "loss": 0.0878,
      "num_input_tokens_seen": 80481840,
      "step": 37285
    },
    {
      "epoch": 6.083197389885807,
      "grad_norm": 0.14295031130313873,
      "learning_rate": 1.994711672660924e-05,
      "loss": 0.0027,
      "num_input_tokens_seen": 80493648,
      "step": 37290
    },
    {
      "epoch": 6.084013050570962,
      "grad_norm": 0.1177147701382637,
      "learning_rate": 1.9940145852100836e-05,
      "loss": 0.1494,
      "num_input_tokens_seen": 80503024,
      "step": 37295
    },
    {
      "epoch": 6.084828711256117,
      "grad_norm": 0.327179491519928,
      "learning_rate": 1.993317538776988e-05,
      "loss": 0.1426,
      "num_input_tokens_seen": 80513808,
      "step": 37300
    },
    {
      "epoch": 6.085644371941273,
      "grad_norm": 0.3292040526866913,
      "learning_rate": 1.9926205334181443e-05,
      "loss": 0.2491,
      "num_input_tokens_seen": 80524176,
      "step": 37305
    },
    {
      "epoch": 6.0864600326264275,
      "grad_norm": 0.0920260027050972,
      "learning_rate": 1.9919235691900526e-05,
      "loss": 0.0907,
      "num_input_tokens_seen": 80536176,
      "step": 37310
    },
    {
      "epoch": 6.087275693311582,
      "grad_norm": 0.026817047968506813,
      "learning_rate": 1.991226646149216e-05,
      "loss": 0.0177,
      "num_input_tokens_seen": 80546960,
      "step": 37315
    },
    {
      "epoch": 6.088091353996737,
      "grad_norm": 4.843094825744629,
      "learning_rate": 1.9905297643521287e-05,
      "loss": 0.1909,
      "num_input_tokens_seen": 80557936,
      "step": 37320
    },
    {
      "epoch": 6.088907014681892,
      "grad_norm": 8.50800609588623,
      "learning_rate": 1.9898329238552838e-05,
      "loss": 0.0974,
      "num_input_tokens_seen": 80567728,
      "step": 37325
    },
    {
      "epoch": 6.089722675367048,
      "grad_norm": 0.040082208812236786,
      "learning_rate": 1.9891361247151706e-05,
      "loss": 0.1036,
      "num_input_tokens_seen": 80579184,
      "step": 37330
    },
    {
      "epoch": 6.0905383360522025,
      "grad_norm": 6.693157196044922,
      "learning_rate": 1.9884393669882752e-05,
      "loss": 0.0201,
      "num_input_tokens_seen": 80589872,
      "step": 37335
    },
    {
      "epoch": 6.091353996737357,
      "grad_norm": 0.11779052764177322,
      "learning_rate": 1.9877426507310802e-05,
      "loss": 0.0045,
      "num_input_tokens_seen": 80599888,
      "step": 37340
    },
    {
      "epoch": 6.092169657422512,
      "grad_norm": 4.852125644683838,
      "learning_rate": 1.9870459760000654e-05,
      "loss": 0.1395,
      "num_input_tokens_seen": 80610384,
      "step": 37345
    },
    {
      "epoch": 6.092985318107667,
      "grad_norm": 0.09516799449920654,
      "learning_rate": 1.9863493428517066e-05,
      "loss": 0.1118,
      "num_input_tokens_seen": 80621456,
      "step": 37350
    },
    {
      "epoch": 6.093800978792822,
      "grad_norm": 0.06654660403728485,
      "learning_rate": 1.985652751342476e-05,
      "loss": 0.0973,
      "num_input_tokens_seen": 80632208,
      "step": 37355
    },
    {
      "epoch": 6.0946166394779775,
      "grad_norm": 0.09963220357894897,
      "learning_rate": 1.984956201528843e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 80642736,
      "step": 37360
    },
    {
      "epoch": 6.095432300163132,
      "grad_norm": 3.8635575771331787,
      "learning_rate": 1.984259693467274e-05,
      "loss": 0.2489,
      "num_input_tokens_seen": 80654064,
      "step": 37365
    },
    {
      "epoch": 6.096247960848287,
      "grad_norm": 7.422365188598633,
      "learning_rate": 1.9835632272142305e-05,
      "loss": 0.0577,
      "num_input_tokens_seen": 80664880,
      "step": 37370
    },
    {
      "epoch": 6.097063621533442,
      "grad_norm": 0.2181326150894165,
      "learning_rate": 1.9828668028261726e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 80676720,
      "step": 37375
    },
    {
      "epoch": 6.097879282218597,
      "grad_norm": 0.5527575612068176,
      "learning_rate": 1.9821704203595554e-05,
      "loss": 0.124,
      "num_input_tokens_seen": 80688240,
      "step": 37380
    },
    {
      "epoch": 6.0986949429037525,
      "grad_norm": 0.1534019410610199,
      "learning_rate": 1.9814740798708316e-05,
      "loss": 0.0278,
      "num_input_tokens_seen": 80699216,
      "step": 37385
    },
    {
      "epoch": 6.099510603588907,
      "grad_norm": 0.16948339343070984,
      "learning_rate": 1.98077778141645e-05,
      "loss": 0.174,
      "num_input_tokens_seen": 80710192,
      "step": 37390
    },
    {
      "epoch": 6.100326264274062,
      "grad_norm": 3.207892894744873,
      "learning_rate": 1.9800815250528557e-05,
      "loss": 0.2233,
      "num_input_tokens_seen": 80720880,
      "step": 37395
    },
    {
      "epoch": 6.101141924959217,
      "grad_norm": 0.2363218069076538,
      "learning_rate": 1.979385310836491e-05,
      "loss": 0.0034,
      "num_input_tokens_seen": 80731760,
      "step": 37400
    },
    {
      "epoch": 6.101957585644372,
      "grad_norm": 0.11209364235401154,
      "learning_rate": 1.9786891388237945e-05,
      "loss": 0.006,
      "num_input_tokens_seen": 80741712,
      "step": 37405
    },
    {
      "epoch": 6.102773246329527,
      "grad_norm": 5.0404815673828125,
      "learning_rate": 1.9779930090712017e-05,
      "loss": 0.2451,
      "num_input_tokens_seen": 80753264,
      "step": 37410
    },
    {
      "epoch": 6.103588907014682,
      "grad_norm": 4.946499824523926,
      "learning_rate": 1.9772969216351433e-05,
      "loss": 0.0371,
      "num_input_tokens_seen": 80764816,
      "step": 37415
    },
    {
      "epoch": 6.104404567699837,
      "grad_norm": 12.903267860412598,
      "learning_rate": 1.9766008765720493e-05,
      "loss": 0.3257,
      "num_input_tokens_seen": 80774320,
      "step": 37420
    },
    {
      "epoch": 6.105220228384992,
      "grad_norm": 0.01140553504228592,
      "learning_rate": 1.975904873938344e-05,
      "loss": 0.0061,
      "num_input_tokens_seen": 80784976,
      "step": 37425
    },
    {
      "epoch": 6.106035889070147,
      "grad_norm": 0.08069048821926117,
      "learning_rate": 1.9752089137904492e-05,
      "loss": 0.0974,
      "num_input_tokens_seen": 80794672,
      "step": 37430
    },
    {
      "epoch": 6.1068515497553015,
      "grad_norm": 0.1736743301153183,
      "learning_rate": 1.9745129961847824e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 80805424,
      "step": 37435
    },
    {
      "epoch": 6.107667210440456,
      "grad_norm": 5.19975471496582,
      "learning_rate": 1.9738171211777584e-05,
      "loss": 0.0125,
      "num_input_tokens_seen": 80816240,
      "step": 37440
    },
    {
      "epoch": 6.108482871125612,
      "grad_norm": 0.08598069101572037,
      "learning_rate": 1.9731212888257883e-05,
      "loss": 0.1493,
      "num_input_tokens_seen": 80825744,
      "step": 37445
    },
    {
      "epoch": 6.109298531810767,
      "grad_norm": 2.454923629760742,
      "learning_rate": 1.97242549918528e-05,
      "loss": 0.0927,
      "num_input_tokens_seen": 80835984,
      "step": 37450
    },
    {
      "epoch": 6.110114192495922,
      "grad_norm": 16.96182632446289,
      "learning_rate": 1.9717297523126373e-05,
      "loss": 0.0505,
      "num_input_tokens_seen": 80847184,
      "step": 37455
    },
    {
      "epoch": 6.1109298531810765,
      "grad_norm": 0.0514359325170517,
      "learning_rate": 1.9710340482642615e-05,
      "loss": 0.0323,
      "num_input_tokens_seen": 80858608,
      "step": 37460
    },
    {
      "epoch": 6.111745513866231,
      "grad_norm": 0.05719045549631119,
      "learning_rate": 1.9703383870965496e-05,
      "loss": 0.1911,
      "num_input_tokens_seen": 80869584,
      "step": 37465
    },
    {
      "epoch": 6.112561174551387,
      "grad_norm": 3.358921766281128,
      "learning_rate": 1.969642768865896e-05,
      "loss": 0.1088,
      "num_input_tokens_seen": 80879408,
      "step": 37470
    },
    {
      "epoch": 6.113376835236542,
      "grad_norm": 0.10545467585325241,
      "learning_rate": 1.9689471936286902e-05,
      "loss": 0.1177,
      "num_input_tokens_seen": 80890032,
      "step": 37475
    },
    {
      "epoch": 6.114192495921697,
      "grad_norm": 0.15293854475021362,
      "learning_rate": 1.9682516614413194e-05,
      "loss": 0.1575,
      "num_input_tokens_seen": 80902160,
      "step": 37480
    },
    {
      "epoch": 6.1150081566068515,
      "grad_norm": 0.06207267940044403,
      "learning_rate": 1.967556172360167e-05,
      "loss": 0.2369,
      "num_input_tokens_seen": 80914032,
      "step": 37485
    },
    {
      "epoch": 6.115823817292006,
      "grad_norm": 0.340839684009552,
      "learning_rate": 1.966860726441613e-05,
      "loss": 0.0175,
      "num_input_tokens_seen": 80924912,
      "step": 37490
    },
    {
      "epoch": 6.116639477977161,
      "grad_norm": 0.25051626563072205,
      "learning_rate": 1.9661653237420337e-05,
      "loss": 0.1899,
      "num_input_tokens_seen": 80934544,
      "step": 37495
    },
    {
      "epoch": 6.117455138662317,
      "grad_norm": 4.973474979400635,
      "learning_rate": 1.9654699643178016e-05,
      "loss": 0.1761,
      "num_input_tokens_seen": 80945456,
      "step": 37500
    },
    {
      "epoch": 6.118270799347472,
      "grad_norm": 0.14203764498233795,
      "learning_rate": 1.9647746482252866e-05,
      "loss": 0.0928,
      "num_input_tokens_seen": 80956368,
      "step": 37505
    },
    {
      "epoch": 6.1190864600326265,
      "grad_norm": 0.04542820528149605,
      "learning_rate": 1.9640793755208542e-05,
      "loss": 0.0037,
      "num_input_tokens_seen": 80967280,
      "step": 37510
    },
    {
      "epoch": 6.119902120717781,
      "grad_norm": 0.0741165429353714,
      "learning_rate": 1.9633841462608664e-05,
      "loss": 0.0102,
      "num_input_tokens_seen": 80978320,
      "step": 37515
    },
    {
      "epoch": 6.120717781402936,
      "grad_norm": 2.9917116165161133,
      "learning_rate": 1.9626889605016827e-05,
      "loss": 0.1145,
      "num_input_tokens_seen": 80988624,
      "step": 37520
    },
    {
      "epoch": 6.121533442088092,
      "grad_norm": 0.1031784862279892,
      "learning_rate": 1.9619938182996585e-05,
      "loss": 0.1684,
      "num_input_tokens_seen": 81000560,
      "step": 37525
    },
    {
      "epoch": 6.122349102773247,
      "grad_norm": 0.08577301353216171,
      "learning_rate": 1.961298719711145e-05,
      "loss": 0.003,
      "num_input_tokens_seen": 81011408,
      "step": 37530
    },
    {
      "epoch": 6.123164763458401,
      "grad_norm": 0.047973308712244034,
      "learning_rate": 1.9606036647924907e-05,
      "loss": 0.1376,
      "num_input_tokens_seen": 81023472,
      "step": 37535
    },
    {
      "epoch": 6.123980424143556,
      "grad_norm": 5.070557594299316,
      "learning_rate": 1.95990865360004e-05,
      "loss": 0.2248,
      "num_input_tokens_seen": 81033968,
      "step": 37540
    },
    {
      "epoch": 6.124796084828711,
      "grad_norm": 0.14994460344314575,
      "learning_rate": 1.9592136861901344e-05,
      "loss": 0.0054,
      "num_input_tokens_seen": 81044304,
      "step": 37545
    },
    {
      "epoch": 6.125611745513866,
      "grad_norm": 0.12661321461200714,
      "learning_rate": 1.9585187626191113e-05,
      "loss": 0.0299,
      "num_input_tokens_seen": 81055056,
      "step": 37550
    },
    {
      "epoch": 6.126427406199022,
      "grad_norm": 4.4868364334106445,
      "learning_rate": 1.9578238829433048e-05,
      "loss": 0.0798,
      "num_input_tokens_seen": 81066544,
      "step": 37555
    },
    {
      "epoch": 6.127243066884176,
      "grad_norm": 0.13730965554714203,
      "learning_rate": 1.9571290472190456e-05,
      "loss": 0.0144,
      "num_input_tokens_seen": 81077360,
      "step": 37560
    },
    {
      "epoch": 6.128058727569331,
      "grad_norm": 4.3978071212768555,
      "learning_rate": 1.95643425550266e-05,
      "loss": 0.0086,
      "num_input_tokens_seen": 81088816,
      "step": 37565
    },
    {
      "epoch": 6.128874388254486,
      "grad_norm": 0.03902905806899071,
      "learning_rate": 1.955739507850472e-05,
      "loss": 0.1386,
      "num_input_tokens_seen": 81101488,
      "step": 37570
    },
    {
      "epoch": 6.129690048939641,
      "grad_norm": 4.845416069030762,
      "learning_rate": 1.9550448043188007e-05,
      "loss": 0.0309,
      "num_input_tokens_seen": 81111568,
      "step": 37575
    },
    {
      "epoch": 6.130505709624796,
      "grad_norm": 0.12704837322235107,
      "learning_rate": 1.954350144963963e-05,
      "loss": 0.0099,
      "num_input_tokens_seen": 81121840,
      "step": 37580
    },
    {
      "epoch": 6.131321370309951,
      "grad_norm": 0.33101361989974976,
      "learning_rate": 1.9536555298422706e-05,
      "loss": 0.0079,
      "num_input_tokens_seen": 81132720,
      "step": 37585
    },
    {
      "epoch": 6.132137030995106,
      "grad_norm": 3.2499988079071045,
      "learning_rate": 1.9529609590100337e-05,
      "loss": 0.0928,
      "num_input_tokens_seen": 81143536,
      "step": 37590
    },
    {
      "epoch": 6.132952691680261,
      "grad_norm": 8.56700611114502,
      "learning_rate": 1.952266432523557e-05,
      "loss": 0.2774,
      "num_input_tokens_seen": 81154096,
      "step": 37595
    },
    {
      "epoch": 6.133768352365416,
      "grad_norm": 0.16566801071166992,
      "learning_rate": 1.9515719504391415e-05,
      "loss": 0.0043,
      "num_input_tokens_seen": 81165104,
      "step": 37600
    },
    {
      "epoch": 6.134584013050571,
      "grad_norm": 19.968868255615234,
      "learning_rate": 1.950877512813087e-05,
      "loss": 0.2139,
      "num_input_tokens_seen": 81175952,
      "step": 37605
    },
    {
      "epoch": 6.135399673735726,
      "grad_norm": 3.177694082260132,
      "learning_rate": 1.950183119701688e-05,
      "loss": 0.2195,
      "num_input_tokens_seen": 81186608,
      "step": 37610
    },
    {
      "epoch": 6.136215334420881,
      "grad_norm": 0.065876305103302,
      "learning_rate": 1.949488771161235e-05,
      "loss": 0.0079,
      "num_input_tokens_seen": 81198128,
      "step": 37615
    },
    {
      "epoch": 6.137030995106036,
      "grad_norm": 2.483712911605835,
      "learning_rate": 1.948794467248015e-05,
      "loss": 0.12,
      "num_input_tokens_seen": 81207792,
      "step": 37620
    },
    {
      "epoch": 6.137846655791191,
      "grad_norm": 0.3169977366924286,
      "learning_rate": 1.9481002080183114e-05,
      "loss": 0.0072,
      "num_input_tokens_seen": 81219152,
      "step": 37625
    },
    {
      "epoch": 6.138662316476346,
      "grad_norm": 0.12151922285556793,
      "learning_rate": 1.947405993528406e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 81230800,
      "step": 37630
    },
    {
      "epoch": 6.1394779771615005,
      "grad_norm": 0.17825913429260254,
      "learning_rate": 1.9467118238345752e-05,
      "loss": 0.1631,
      "num_input_tokens_seen": 81243536,
      "step": 37635
    },
    {
      "epoch": 6.140293637846656,
      "grad_norm": 0.031340572983026505,
      "learning_rate": 1.946017698993091e-05,
      "loss": 0.0049,
      "num_input_tokens_seen": 81255344,
      "step": 37640
    },
    {
      "epoch": 6.141109298531811,
      "grad_norm": 0.10641069710254669,
      "learning_rate": 1.945323619060223e-05,
      "loss": 0.0057,
      "num_input_tokens_seen": 81266544,
      "step": 37645
    },
    {
      "epoch": 6.141924959216966,
      "grad_norm": 10.570899963378906,
      "learning_rate": 1.944629584092237e-05,
      "loss": 0.1512,
      "num_input_tokens_seen": 81276976,
      "step": 37650
    },
    {
      "epoch": 6.142740619902121,
      "grad_norm": 0.10499405115842819,
      "learning_rate": 1.943935594145395e-05,
      "loss": 0.0875,
      "num_input_tokens_seen": 81288656,
      "step": 37655
    },
    {
      "epoch": 6.143556280587275,
      "grad_norm": 3.573042154312134,
      "learning_rate": 1.9432416492759548e-05,
      "loss": 0.2075,
      "num_input_tokens_seen": 81298928,
      "step": 37660
    },
    {
      "epoch": 6.14437194127243,
      "grad_norm": 0.08196359872817993,
      "learning_rate": 1.9425477495401716e-05,
      "loss": 0.1534,
      "num_input_tokens_seen": 81310224,
      "step": 37665
    },
    {
      "epoch": 6.145187601957586,
      "grad_norm": 0.10421831905841827,
      "learning_rate": 1.9418538949942962e-05,
      "loss": 0.0048,
      "num_input_tokens_seen": 81321616,
      "step": 37670
    },
    {
      "epoch": 6.146003262642741,
      "grad_norm": 0.13622203469276428,
      "learning_rate": 1.9411600856945763e-05,
      "loss": 0.1136,
      "num_input_tokens_seen": 81332656,
      "step": 37675
    },
    {
      "epoch": 6.146818923327896,
      "grad_norm": 5.768846035003662,
      "learning_rate": 1.940466321697255e-05,
      "loss": 0.1429,
      "num_input_tokens_seen": 81343440,
      "step": 37680
    },
    {
      "epoch": 6.14763458401305,
      "grad_norm": 1.4352920055389404,
      "learning_rate": 1.9397726030585726e-05,
      "loss": 0.2253,
      "num_input_tokens_seen": 81354672,
      "step": 37685
    },
    {
      "epoch": 6.148450244698205,
      "grad_norm": 0.26600053906440735,
      "learning_rate": 1.939078929834766e-05,
      "loss": 0.0978,
      "num_input_tokens_seen": 81365808,
      "step": 37690
    },
    {
      "epoch": 6.149265905383361,
      "grad_norm": 0.7572985887527466,
      "learning_rate": 1.9383853020820674e-05,
      "loss": 0.0255,
      "num_input_tokens_seen": 81375856,
      "step": 37695
    },
    {
      "epoch": 6.150081566068516,
      "grad_norm": 0.920245885848999,
      "learning_rate": 1.9376917198567058e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 81387184,
      "step": 37700
    },
    {
      "epoch": 6.150897226753671,
      "grad_norm": 0.09726397693157196,
      "learning_rate": 1.9369981832149064e-05,
      "loss": 0.0946,
      "num_input_tokens_seen": 81397712,
      "step": 37705
    },
    {
      "epoch": 6.151712887438825,
      "grad_norm": 0.1214255541563034,
      "learning_rate": 1.936304692212891e-05,
      "loss": 0.1656,
      "num_input_tokens_seen": 81408592,
      "step": 37710
    },
    {
      "epoch": 6.15252854812398,
      "grad_norm": 7.828940391540527,
      "learning_rate": 1.9356112469068776e-05,
      "loss": 0.1424,
      "num_input_tokens_seen": 81420176,
      "step": 37715
    },
    {
      "epoch": 6.153344208809135,
      "grad_norm": 0.5535727143287659,
      "learning_rate": 1.93491784735308e-05,
      "loss": 0.116,
      "num_input_tokens_seen": 81431920,
      "step": 37720
    },
    {
      "epoch": 6.154159869494291,
      "grad_norm": 0.5543473958969116,
      "learning_rate": 1.934224493607709e-05,
      "loss": 0.2108,
      "num_input_tokens_seen": 81443312,
      "step": 37725
    },
    {
      "epoch": 6.1549755301794455,
      "grad_norm": 0.05757756158709526,
      "learning_rate": 1.933531185726971e-05,
      "loss": 0.0058,
      "num_input_tokens_seen": 81454736,
      "step": 37730
    },
    {
      "epoch": 6.1557911908646,
      "grad_norm": 4.306589603424072,
      "learning_rate": 1.9328379237670684e-05,
      "loss": 0.4875,
      "num_input_tokens_seen": 81465968,
      "step": 37735
    },
    {
      "epoch": 6.156606851549755,
      "grad_norm": 0.12160925567150116,
      "learning_rate": 1.9321447077842026e-05,
      "loss": 0.1139,
      "num_input_tokens_seen": 81477264,
      "step": 37740
    },
    {
      "epoch": 6.15742251223491,
      "grad_norm": 0.12456405162811279,
      "learning_rate": 1.931451537834568e-05,
      "loss": 0.0267,
      "num_input_tokens_seen": 81488080,
      "step": 37745
    },
    {
      "epoch": 6.158238172920065,
      "grad_norm": 0.10891862213611603,
      "learning_rate": 1.9307584139743564e-05,
      "loss": 0.185,
      "num_input_tokens_seen": 81499536,
      "step": 37750
    },
    {
      "epoch": 6.1590538336052205,
      "grad_norm": 0.15445049107074738,
      "learning_rate": 1.9300653362597564e-05,
      "loss": 0.013,
      "num_input_tokens_seen": 81511184,
      "step": 37755
    },
    {
      "epoch": 6.159869494290375,
      "grad_norm": 0.16697649657726288,
      "learning_rate": 1.929372304746952e-05,
      "loss": 0.4588,
      "num_input_tokens_seen": 81521648,
      "step": 37760
    },
    {
      "epoch": 6.16068515497553,
      "grad_norm": 0.06527233868837357,
      "learning_rate": 1.928679319492124e-05,
      "loss": 0.1165,
      "num_input_tokens_seen": 81532016,
      "step": 37765
    },
    {
      "epoch": 6.161500815660685,
      "grad_norm": 6.2396416664123535,
      "learning_rate": 1.9279863805514482e-05,
      "loss": 0.0152,
      "num_input_tokens_seen": 81543440,
      "step": 37770
    },
    {
      "epoch": 6.16231647634584,
      "grad_norm": 4.716691970825195,
      "learning_rate": 1.9272934879810994e-05,
      "loss": 0.1288,
      "num_input_tokens_seen": 81553680,
      "step": 37775
    },
    {
      "epoch": 6.1631321370309955,
      "grad_norm": 0.41060569882392883,
      "learning_rate": 1.9266006418372464e-05,
      "loss": 0.0061,
      "num_input_tokens_seen": 81564944,
      "step": 37780
    },
    {
      "epoch": 6.16394779771615,
      "grad_norm": 0.46858978271484375,
      "learning_rate": 1.925907842176055e-05,
      "loss": 0.1307,
      "num_input_tokens_seen": 81575376,
      "step": 37785
    },
    {
      "epoch": 6.164763458401305,
      "grad_norm": 4.114612102508545,
      "learning_rate": 1.925215089053687e-05,
      "loss": 0.1889,
      "num_input_tokens_seen": 81585840,
      "step": 37790
    },
    {
      "epoch": 6.16557911908646,
      "grad_norm": 6.160334587097168,
      "learning_rate": 1.9245223825262997e-05,
      "loss": 0.046,
      "num_input_tokens_seen": 81597072,
      "step": 37795
    },
    {
      "epoch": 6.166394779771615,
      "grad_norm": 0.13958598673343658,
      "learning_rate": 1.9238297226500483e-05,
      "loss": 0.1977,
      "num_input_tokens_seen": 81608336,
      "step": 37800
    },
    {
      "epoch": 6.16721044045677,
      "grad_norm": 0.1675865799188614,
      "learning_rate": 1.923137109481083e-05,
      "loss": 0.1069,
      "num_input_tokens_seen": 81619440,
      "step": 37805
    },
    {
      "epoch": 6.168026101141925,
      "grad_norm": 0.4309624433517456,
      "learning_rate": 1.9224445430755507e-05,
      "loss": 0.1008,
      "num_input_tokens_seen": 81629648,
      "step": 37810
    },
    {
      "epoch": 6.16884176182708,
      "grad_norm": 0.09904403984546661,
      "learning_rate": 1.9217520234895943e-05,
      "loss": 0.271,
      "num_input_tokens_seen": 81639408,
      "step": 37815
    },
    {
      "epoch": 6.169657422512235,
      "grad_norm": 0.6715161800384521,
      "learning_rate": 1.9210595507793526e-05,
      "loss": 0.0853,
      "num_input_tokens_seen": 81650704,
      "step": 37820
    },
    {
      "epoch": 6.17047308319739,
      "grad_norm": 0.08807911723852158,
      "learning_rate": 1.9203671250009612e-05,
      "loss": 0.0105,
      "num_input_tokens_seen": 81661328,
      "step": 37825
    },
    {
      "epoch": 6.171288743882545,
      "grad_norm": 0.11950883269309998,
      "learning_rate": 1.9196747462105517e-05,
      "loss": 0.0106,
      "num_input_tokens_seen": 81671728,
      "step": 37830
    },
    {
      "epoch": 6.1721044045677,
      "grad_norm": 4.313073635101318,
      "learning_rate": 1.918982414464252e-05,
      "loss": 0.0249,
      "num_input_tokens_seen": 81681936,
      "step": 37835
    },
    {
      "epoch": 6.172920065252855,
      "grad_norm": 12.713224411010742,
      "learning_rate": 1.918290129818185e-05,
      "loss": 0.1223,
      "num_input_tokens_seen": 81692592,
      "step": 37840
    },
    {
      "epoch": 6.17373572593801,
      "grad_norm": 10.112577438354492,
      "learning_rate": 1.9175978923284727e-05,
      "loss": 0.0976,
      "num_input_tokens_seen": 81703696,
      "step": 37845
    },
    {
      "epoch": 6.174551386623165,
      "grad_norm": 0.14462950825691223,
      "learning_rate": 1.91690570205123e-05,
      "loss": 0.0598,
      "num_input_tokens_seen": 81714448,
      "step": 37850
    },
    {
      "epoch": 6.1753670473083195,
      "grad_norm": 0.09021087735891342,
      "learning_rate": 1.916213559042569e-05,
      "loss": 0.1472,
      "num_input_tokens_seen": 81724560,
      "step": 37855
    },
    {
      "epoch": 6.176182707993474,
      "grad_norm": 0.21085111796855927,
      "learning_rate": 1.9155214633586e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 81734384,
      "step": 37860
    },
    {
      "epoch": 6.17699836867863,
      "grad_norm": 0.23198255896568298,
      "learning_rate": 1.9148294150554266e-05,
      "loss": 0.0678,
      "num_input_tokens_seen": 81745552,
      "step": 37865
    },
    {
      "epoch": 6.177814029363785,
      "grad_norm": 0.14588935673236847,
      "learning_rate": 1.9141374141891498e-05,
      "loss": 0.0212,
      "num_input_tokens_seen": 81756144,
      "step": 37870
    },
    {
      "epoch": 6.17862969004894,
      "grad_norm": 4.942633152008057,
      "learning_rate": 1.913445460815867e-05,
      "loss": 0.1059,
      "num_input_tokens_seen": 81767024,
      "step": 37875
    },
    {
      "epoch": 6.1794453507340945,
      "grad_norm": 6.1705217361450195,
      "learning_rate": 1.9127535549916715e-05,
      "loss": 0.1619,
      "num_input_tokens_seen": 81777488,
      "step": 37880
    },
    {
      "epoch": 6.180261011419249,
      "grad_norm": 0.34688469767570496,
      "learning_rate": 1.912061696772652e-05,
      "loss": 0.2104,
      "num_input_tokens_seen": 81788624,
      "step": 37885
    },
    {
      "epoch": 6.181076672104404,
      "grad_norm": 0.09854024648666382,
      "learning_rate": 1.911369886214895e-05,
      "loss": 0.0537,
      "num_input_tokens_seen": 81798224,
      "step": 37890
    },
    {
      "epoch": 6.18189233278956,
      "grad_norm": 0.08823223412036896,
      "learning_rate": 1.9106781233744813e-05,
      "loss": 0.0121,
      "num_input_tokens_seen": 81808976,
      "step": 37895
    },
    {
      "epoch": 6.182707993474715,
      "grad_norm": 4.729548931121826,
      "learning_rate": 1.9099864083074892e-05,
      "loss": 0.1242,
      "num_input_tokens_seen": 81819760,
      "step": 37900
    },
    {
      "epoch": 6.1835236541598695,
      "grad_norm": 0.09086601436138153,
      "learning_rate": 1.9092947410699927e-05,
      "loss": 0.1173,
      "num_input_tokens_seen": 81830416,
      "step": 37905
    },
    {
      "epoch": 6.184339314845024,
      "grad_norm": 2.592741012573242,
      "learning_rate": 1.9086031217180618e-05,
      "loss": 0.1232,
      "num_input_tokens_seen": 81842416,
      "step": 37910
    },
    {
      "epoch": 6.185154975530179,
      "grad_norm": 0.08322153985500336,
      "learning_rate": 1.9079115503077617e-05,
      "loss": 0.0227,
      "num_input_tokens_seen": 81852080,
      "step": 37915
    },
    {
      "epoch": 6.185970636215335,
      "grad_norm": 0.15700191259384155,
      "learning_rate": 1.9072200268951562e-05,
      "loss": 0.1221,
      "num_input_tokens_seen": 81862096,
      "step": 37920
    },
    {
      "epoch": 6.18678629690049,
      "grad_norm": 6.111955642700195,
      "learning_rate": 1.906528551536303e-05,
      "loss": 0.0185,
      "num_input_tokens_seen": 81872816,
      "step": 37925
    },
    {
      "epoch": 6.1876019575856445,
      "grad_norm": 3.117820978164673,
      "learning_rate": 1.905837124287257e-05,
      "loss": 0.1134,
      "num_input_tokens_seen": 81883120,
      "step": 37930
    },
    {
      "epoch": 6.188417618270799,
      "grad_norm": 0.17419977486133575,
      "learning_rate": 1.905145745204068e-05,
      "loss": 0.0966,
      "num_input_tokens_seen": 81893904,
      "step": 37935
    },
    {
      "epoch": 6.189233278955954,
      "grad_norm": 0.053510863333940506,
      "learning_rate": 1.9044544143427832e-05,
      "loss": 0.1066,
      "num_input_tokens_seen": 81904368,
      "step": 37940
    },
    {
      "epoch": 6.190048939641109,
      "grad_norm": 0.12100216001272202,
      "learning_rate": 1.9037631317594445e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 81915664,
      "step": 37945
    },
    {
      "epoch": 6.190864600326265,
      "grad_norm": 0.14096873998641968,
      "learning_rate": 1.9030718975100927e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 81926736,
      "step": 37950
    },
    {
      "epoch": 6.191680261011419,
      "grad_norm": 0.059876978397369385,
      "learning_rate": 1.9023807116507615e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 81937552,
      "step": 37955
    },
    {
      "epoch": 6.192495921696574,
      "grad_norm": 0.12152976542711258,
      "learning_rate": 1.9016895742374824e-05,
      "loss": 0.2442,
      "num_input_tokens_seen": 81947120,
      "step": 37960
    },
    {
      "epoch": 6.193311582381729,
      "grad_norm": 2.926598072052002,
      "learning_rate": 1.900998485326282e-05,
      "loss": 0.1074,
      "num_input_tokens_seen": 81955216,
      "step": 37965
    },
    {
      "epoch": 6.194127243066884,
      "grad_norm": 0.1282525509595871,
      "learning_rate": 1.9003074449731835e-05,
      "loss": 0.224,
      "num_input_tokens_seen": 81965360,
      "step": 37970
    },
    {
      "epoch": 6.19494290375204,
      "grad_norm": 15.548892974853516,
      "learning_rate": 1.8996164532342065e-05,
      "loss": 0.1526,
      "num_input_tokens_seen": 81976112,
      "step": 37975
    },
    {
      "epoch": 6.195758564437194,
      "grad_norm": 4.837267875671387,
      "learning_rate": 1.8989255101653662e-05,
      "loss": 0.0853,
      "num_input_tokens_seen": 81987568,
      "step": 37980
    },
    {
      "epoch": 6.196574225122349,
      "grad_norm": 4.110472679138184,
      "learning_rate": 1.898234615822674e-05,
      "loss": 0.191,
      "num_input_tokens_seen": 81998096,
      "step": 37985
    },
    {
      "epoch": 6.197389885807504,
      "grad_norm": 0.08067582547664642,
      "learning_rate": 1.8975437702621368e-05,
      "loss": 0.0045,
      "num_input_tokens_seen": 82009872,
      "step": 37990
    },
    {
      "epoch": 6.198205546492659,
      "grad_norm": 0.13552775979042053,
      "learning_rate": 1.8968529735397582e-05,
      "loss": 0.093,
      "num_input_tokens_seen": 82020080,
      "step": 37995
    },
    {
      "epoch": 6.199021207177814,
      "grad_norm": 0.049594298005104065,
      "learning_rate": 1.896162225711538e-05,
      "loss": 0.0054,
      "num_input_tokens_seen": 82030416,
      "step": 38000
    },
    {
      "epoch": 6.199836867862969,
      "grad_norm": 0.14990796148777008,
      "learning_rate": 1.895471526833472e-05,
      "loss": 0.0102,
      "num_input_tokens_seen": 82041296,
      "step": 38005
    },
    {
      "epoch": 6.200652528548124,
      "grad_norm": 0.11612722277641296,
      "learning_rate": 1.8947808769615512e-05,
      "loss": 0.051,
      "num_input_tokens_seen": 82051024,
      "step": 38010
    },
    {
      "epoch": 6.201468189233279,
      "grad_norm": 0.11736045777797699,
      "learning_rate": 1.8940902761517638e-05,
      "loss": 0.3306,
      "num_input_tokens_seen": 82062288,
      "step": 38015
    },
    {
      "epoch": 6.202283849918434,
      "grad_norm": 0.05441810190677643,
      "learning_rate": 1.8933997244600923e-05,
      "loss": 0.0059,
      "num_input_tokens_seen": 82073808,
      "step": 38020
    },
    {
      "epoch": 6.203099510603589,
      "grad_norm": 0.08446001261472702,
      "learning_rate": 1.8927092219425174e-05,
      "loss": 0.0974,
      "num_input_tokens_seen": 82083728,
      "step": 38025
    },
    {
      "epoch": 6.2039151712887435,
      "grad_norm": 7.407721519470215,
      "learning_rate": 1.892018768655014e-05,
      "loss": 0.2053,
      "num_input_tokens_seen": 82095600,
      "step": 38030
    },
    {
      "epoch": 6.204730831973899,
      "grad_norm": 0.10007230192422867,
      "learning_rate": 1.8913283646535547e-05,
      "loss": 0.059,
      "num_input_tokens_seen": 82105136,
      "step": 38035
    },
    {
      "epoch": 6.205546492659054,
      "grad_norm": 0.07161112129688263,
      "learning_rate": 1.890638009994106e-05,
      "loss": 0.0057,
      "num_input_tokens_seen": 82115504,
      "step": 38040
    },
    {
      "epoch": 6.206362153344209,
      "grad_norm": 2.146233558654785,
      "learning_rate": 1.889947704732632e-05,
      "loss": 0.2134,
      "num_input_tokens_seen": 82126864,
      "step": 38045
    },
    {
      "epoch": 6.207177814029364,
      "grad_norm": 0.19240115582942963,
      "learning_rate": 1.889257448925093e-05,
      "loss": 0.14,
      "num_input_tokens_seen": 82137520,
      "step": 38050
    },
    {
      "epoch": 6.2079934747145185,
      "grad_norm": 0.16441723704338074,
      "learning_rate": 1.8885672426274424e-05,
      "loss": 0.0147,
      "num_input_tokens_seen": 82148976,
      "step": 38055
    },
    {
      "epoch": 6.208809135399674,
      "grad_norm": 0.1422182321548462,
      "learning_rate": 1.8878770858956353e-05,
      "loss": 0.0508,
      "num_input_tokens_seen": 82160496,
      "step": 38060
    },
    {
      "epoch": 6.209624796084829,
      "grad_norm": 3.2957746982574463,
      "learning_rate": 1.8871869787856166e-05,
      "loss": 0.1198,
      "num_input_tokens_seen": 82171440,
      "step": 38065
    },
    {
      "epoch": 6.210440456769984,
      "grad_norm": 0.1262081265449524,
      "learning_rate": 1.886496921353331e-05,
      "loss": 0.2352,
      "num_input_tokens_seen": 82182704,
      "step": 38070
    },
    {
      "epoch": 6.211256117455139,
      "grad_norm": 0.09778767079114914,
      "learning_rate": 1.8858069136547186e-05,
      "loss": 0.0512,
      "num_input_tokens_seen": 82193392,
      "step": 38075
    },
    {
      "epoch": 6.212071778140293,
      "grad_norm": 4.510008811950684,
      "learning_rate": 1.8851169557457128e-05,
      "loss": 0.224,
      "num_input_tokens_seen": 82204080,
      "step": 38080
    },
    {
      "epoch": 6.212887438825448,
      "grad_norm": 0.08541733026504517,
      "learning_rate": 1.8844270476822473e-05,
      "loss": 0.0058,
      "num_input_tokens_seen": 82215440,
      "step": 38085
    },
    {
      "epoch": 6.213703099510604,
      "grad_norm": 2.2472951412200928,
      "learning_rate": 1.883737189520249e-05,
      "loss": 0.1546,
      "num_input_tokens_seen": 82226832,
      "step": 38090
    },
    {
      "epoch": 6.214518760195759,
      "grad_norm": 0.031898822635412216,
      "learning_rate": 1.88304738131564e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 82237296,
      "step": 38095
    },
    {
      "epoch": 6.215334420880914,
      "grad_norm": 0.3057255446910858,
      "learning_rate": 1.8823576231243418e-05,
      "loss": 0.1126,
      "num_input_tokens_seen": 82248400,
      "step": 38100
    },
    {
      "epoch": 6.216150081566068,
      "grad_norm": 0.13367220759391785,
      "learning_rate": 1.881667915002268e-05,
      "loss": 0.1254,
      "num_input_tokens_seen": 82259088,
      "step": 38105
    },
    {
      "epoch": 6.216965742251223,
      "grad_norm": 0.06978476792573929,
      "learning_rate": 1.8809782570053304e-05,
      "loss": 0.1414,
      "num_input_tokens_seen": 82269424,
      "step": 38110
    },
    {
      "epoch": 6.217781402936378,
      "grad_norm": 24.588245391845703,
      "learning_rate": 1.880288649189436e-05,
      "loss": 0.246,
      "num_input_tokens_seen": 82280144,
      "step": 38115
    },
    {
      "epoch": 6.218597063621534,
      "grad_norm": 0.10955842584371567,
      "learning_rate": 1.8795990916104886e-05,
      "loss": 0.0894,
      "num_input_tokens_seen": 82290672,
      "step": 38120
    },
    {
      "epoch": 6.219412724306689,
      "grad_norm": 0.8793649077415466,
      "learning_rate": 1.8789095843243863e-05,
      "loss": 0.133,
      "num_input_tokens_seen": 82301456,
      "step": 38125
    },
    {
      "epoch": 6.220228384991843,
      "grad_norm": 0.9195159673690796,
      "learning_rate": 1.878220127387025e-05,
      "loss": 0.0103,
      "num_input_tokens_seen": 82311696,
      "step": 38130
    },
    {
      "epoch": 6.221044045676998,
      "grad_norm": 4.79920768737793,
      "learning_rate": 1.8775307208542946e-05,
      "loss": 0.0985,
      "num_input_tokens_seen": 82322320,
      "step": 38135
    },
    {
      "epoch": 6.221859706362153,
      "grad_norm": 1.8701971769332886,
      "learning_rate": 1.8768413647820817e-05,
      "loss": 0.0674,
      "num_input_tokens_seen": 82332976,
      "step": 38140
    },
    {
      "epoch": 6.222675367047309,
      "grad_norm": 6.5250244140625,
      "learning_rate": 1.8761520592262704e-05,
      "loss": 0.3446,
      "num_input_tokens_seen": 82343920,
      "step": 38145
    },
    {
      "epoch": 6.2234910277324635,
      "grad_norm": 0.5002713799476624,
      "learning_rate": 1.8754628042427387e-05,
      "loss": 0.0098,
      "num_input_tokens_seen": 82354928,
      "step": 38150
    },
    {
      "epoch": 6.224306688417618,
      "grad_norm": 0.15075264871120453,
      "learning_rate": 1.8747735998873604e-05,
      "loss": 0.0087,
      "num_input_tokens_seen": 82366544,
      "step": 38155
    },
    {
      "epoch": 6.225122349102773,
      "grad_norm": 0.19868357479572296,
      "learning_rate": 1.8740844462160064e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 82377520,
      "step": 38160
    },
    {
      "epoch": 6.225938009787928,
      "grad_norm": 0.09077688306570053,
      "learning_rate": 1.873395343284543e-05,
      "loss": 0.008,
      "num_input_tokens_seen": 82388240,
      "step": 38165
    },
    {
      "epoch": 6.226753670473083,
      "grad_norm": 0.11144839972257614,
      "learning_rate": 1.872706291148833e-05,
      "loss": 0.2211,
      "num_input_tokens_seen": 82399984,
      "step": 38170
    },
    {
      "epoch": 6.2275693311582385,
      "grad_norm": 0.09272262454032898,
      "learning_rate": 1.8720172898647338e-05,
      "loss": 0.1108,
      "num_input_tokens_seen": 82410992,
      "step": 38175
    },
    {
      "epoch": 6.228384991843393,
      "grad_norm": 0.11452114582061768,
      "learning_rate": 1.8713283394880993e-05,
      "loss": 0.0093,
      "num_input_tokens_seen": 82422224,
      "step": 38180
    },
    {
      "epoch": 6.229200652528548,
      "grad_norm": 15.37417984008789,
      "learning_rate": 1.8706394400747796e-05,
      "loss": 0.0592,
      "num_input_tokens_seen": 82432624,
      "step": 38185
    },
    {
      "epoch": 6.230016313213703,
      "grad_norm": 0.6620145440101624,
      "learning_rate": 1.8699505916806205e-05,
      "loss": 0.1835,
      "num_input_tokens_seen": 82443696,
      "step": 38190
    },
    {
      "epoch": 6.230831973898858,
      "grad_norm": 0.16932103037834167,
      "learning_rate": 1.869261794361463e-05,
      "loss": 0.049,
      "num_input_tokens_seen": 82454736,
      "step": 38195
    },
    {
      "epoch": 6.231647634584013,
      "grad_norm": 2.5973780155181885,
      "learning_rate": 1.8685730481731444e-05,
      "loss": 0.2247,
      "num_input_tokens_seen": 82465072,
      "step": 38200
    },
    {
      "epoch": 6.232463295269168,
      "grad_norm": 0.0736318901181221,
      "learning_rate": 1.867884353171499e-05,
      "loss": 0.0141,
      "num_input_tokens_seen": 82475760,
      "step": 38205
    },
    {
      "epoch": 6.233278955954323,
      "grad_norm": 0.14232690632343292,
      "learning_rate": 1.867195709412355e-05,
      "loss": 0.1354,
      "num_input_tokens_seen": 82487056,
      "step": 38210
    },
    {
      "epoch": 6.234094616639478,
      "grad_norm": 0.03688379377126694,
      "learning_rate": 1.8665071169515375e-05,
      "loss": 0.1439,
      "num_input_tokens_seen": 82498544,
      "step": 38215
    },
    {
      "epoch": 6.234910277324633,
      "grad_norm": 0.058887895196676254,
      "learning_rate": 1.8658185758448676e-05,
      "loss": 0.2208,
      "num_input_tokens_seen": 82510096,
      "step": 38220
    },
    {
      "epoch": 6.235725938009788,
      "grad_norm": 0.29461947083473206,
      "learning_rate": 1.8651300861481614e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 82520752,
      "step": 38225
    },
    {
      "epoch": 6.236541598694943,
      "grad_norm": 0.17150340974330902,
      "learning_rate": 1.8644416479172316e-05,
      "loss": 0.0939,
      "num_input_tokens_seen": 82531760,
      "step": 38230
    },
    {
      "epoch": 6.237357259380098,
      "grad_norm": 0.05188937112689018,
      "learning_rate": 1.8637532612078872e-05,
      "loss": 0.0077,
      "num_input_tokens_seen": 82542576,
      "step": 38235
    },
    {
      "epoch": 6.238172920065253,
      "grad_norm": 0.08637218177318573,
      "learning_rate": 1.8630649260759315e-05,
      "loss": 0.1278,
      "num_input_tokens_seen": 82553168,
      "step": 38240
    },
    {
      "epoch": 6.238988580750408,
      "grad_norm": 0.11446234583854675,
      "learning_rate": 1.8623766425771648e-05,
      "loss": 0.0941,
      "num_input_tokens_seen": 82563280,
      "step": 38245
    },
    {
      "epoch": 6.239804241435563,
      "grad_norm": 0.056999627500772476,
      "learning_rate": 1.8616884107673823e-05,
      "loss": 0.1353,
      "num_input_tokens_seen": 82575120,
      "step": 38250
    },
    {
      "epoch": 6.240619902120717,
      "grad_norm": 22.947662353515625,
      "learning_rate": 1.8610002307023767e-05,
      "loss": 0.1476,
      "num_input_tokens_seen": 82586032,
      "step": 38255
    },
    {
      "epoch": 6.241435562805873,
      "grad_norm": 16.191913604736328,
      "learning_rate": 1.860312102437934e-05,
      "loss": 0.155,
      "num_input_tokens_seen": 82597040,
      "step": 38260
    },
    {
      "epoch": 6.242251223491028,
      "grad_norm": 0.0527874194085598,
      "learning_rate": 1.859624026029837e-05,
      "loss": 0.0062,
      "num_input_tokens_seen": 82608400,
      "step": 38265
    },
    {
      "epoch": 6.243066884176183,
      "grad_norm": 5.220399379730225,
      "learning_rate": 1.8589360015338668e-05,
      "loss": 0.1912,
      "num_input_tokens_seen": 82618000,
      "step": 38270
    },
    {
      "epoch": 6.2438825448613375,
      "grad_norm": 0.774243950843811,
      "learning_rate": 1.8582480290057975e-05,
      "loss": 0.121,
      "num_input_tokens_seen": 82628880,
      "step": 38275
    },
    {
      "epoch": 6.244698205546492,
      "grad_norm": 0.09298164397478104,
      "learning_rate": 1.8575601085013988e-05,
      "loss": 0.0856,
      "num_input_tokens_seen": 82639216,
      "step": 38280
    },
    {
      "epoch": 6.245513866231648,
      "grad_norm": 0.14385484158992767,
      "learning_rate": 1.8568722400764377e-05,
      "loss": 0.014,
      "num_input_tokens_seen": 82649264,
      "step": 38285
    },
    {
      "epoch": 6.246329526916803,
      "grad_norm": 0.10688867419958115,
      "learning_rate": 1.8561844237866756e-05,
      "loss": 0.1382,
      "num_input_tokens_seen": 82660112,
      "step": 38290
    },
    {
      "epoch": 6.247145187601958,
      "grad_norm": 0.13206766545772552,
      "learning_rate": 1.855496659687871e-05,
      "loss": 0.0149,
      "num_input_tokens_seen": 82671408,
      "step": 38295
    },
    {
      "epoch": 6.2479608482871125,
      "grad_norm": 0.20640970766544342,
      "learning_rate": 1.8548089478357774e-05,
      "loss": 0.0078,
      "num_input_tokens_seen": 82682928,
      "step": 38300
    },
    {
      "epoch": 6.248776508972267,
      "grad_norm": 0.174403578042984,
      "learning_rate": 1.8541212882861442e-05,
      "loss": 0.0919,
      "num_input_tokens_seen": 82694864,
      "step": 38305
    },
    {
      "epoch": 6.249592169657422,
      "grad_norm": 0.18106922507286072,
      "learning_rate": 1.853433681094716e-05,
      "loss": 0.0087,
      "num_input_tokens_seen": 82705296,
      "step": 38310
    },
    {
      "epoch": 6.250407830342578,
      "grad_norm": 0.28391170501708984,
      "learning_rate": 1.8527461263172346e-05,
      "loss": 0.0968,
      "num_input_tokens_seen": 82716304,
      "step": 38315
    },
    {
      "epoch": 6.251223491027733,
      "grad_norm": 0.04587554931640625,
      "learning_rate": 1.852058624009436e-05,
      "loss": 0.0088,
      "num_input_tokens_seen": 82728272,
      "step": 38320
    },
    {
      "epoch": 6.2520391517128875,
      "grad_norm": 2.93485164642334,
      "learning_rate": 1.8513711742270535e-05,
      "loss": 0.1192,
      "num_input_tokens_seen": 82738896,
      "step": 38325
    },
    {
      "epoch": 6.252854812398042,
      "grad_norm": 0.26074308156967163,
      "learning_rate": 1.8506837770258147e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 82749712,
      "step": 38330
    },
    {
      "epoch": 6.253670473083197,
      "grad_norm": 0.06070653349161148,
      "learning_rate": 1.8499964324614434e-05,
      "loss": 0.0089,
      "num_input_tokens_seen": 82758608,
      "step": 38335
    },
    {
      "epoch": 6.254486133768353,
      "grad_norm": 0.1550820916891098,
      "learning_rate": 1.8493091405896595e-05,
      "loss": 0.0104,
      "num_input_tokens_seen": 82770288,
      "step": 38340
    },
    {
      "epoch": 6.255301794453508,
      "grad_norm": 3.547386646270752,
      "learning_rate": 1.8486219014661782e-05,
      "loss": 0.1853,
      "num_input_tokens_seen": 82782448,
      "step": 38345
    },
    {
      "epoch": 6.2561174551386625,
      "grad_norm": 0.12808428704738617,
      "learning_rate": 1.8479347151467106e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 82793264,
      "step": 38350
    },
    {
      "epoch": 6.256933115823817,
      "grad_norm": 0.10375116765499115,
      "learning_rate": 1.8472475816869634e-05,
      "loss": 0.0068,
      "num_input_tokens_seen": 82804688,
      "step": 38355
    },
    {
      "epoch": 6.257748776508972,
      "grad_norm": 0.1232416108250618,
      "learning_rate": 1.8465605011426395e-05,
      "loss": 0.003,
      "num_input_tokens_seen": 82816336,
      "step": 38360
    },
    {
      "epoch": 6.258564437194127,
      "grad_norm": 0.08774631470441818,
      "learning_rate": 1.8458734735694366e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 82827408,
      "step": 38365
    },
    {
      "epoch": 6.259380097879283,
      "grad_norm": 3.3483340740203857,
      "learning_rate": 1.8451864990230488e-05,
      "loss": 0.1968,
      "num_input_tokens_seen": 82838128,
      "step": 38370
    },
    {
      "epoch": 6.260195758564437,
      "grad_norm": 0.05783867463469505,
      "learning_rate": 1.8444995775591654e-05,
      "loss": 0.0254,
      "num_input_tokens_seen": 82849072,
      "step": 38375
    },
    {
      "epoch": 6.261011419249592,
      "grad_norm": 0.1645868420600891,
      "learning_rate": 1.8438127092334732e-05,
      "loss": 0.0048,
      "num_input_tokens_seen": 82859088,
      "step": 38380
    },
    {
      "epoch": 6.261827079934747,
      "grad_norm": 0.08044086396694183,
      "learning_rate": 1.843125894101652e-05,
      "loss": 0.0045,
      "num_input_tokens_seen": 82870480,
      "step": 38385
    },
    {
      "epoch": 6.262642740619902,
      "grad_norm": 3.3642702102661133,
      "learning_rate": 1.8424391322193787e-05,
      "loss": 0.1054,
      "num_input_tokens_seen": 82880720,
      "step": 38390
    },
    {
      "epoch": 6.263458401305057,
      "grad_norm": 0.07266976684331894,
      "learning_rate": 1.8417524236423257e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 82891536,
      "step": 38395
    },
    {
      "epoch": 6.264274061990212,
      "grad_norm": 13.846406936645508,
      "learning_rate": 1.8410657684261613e-05,
      "loss": 0.2432,
      "num_input_tokens_seen": 82903056,
      "step": 38400
    },
    {
      "epoch": 6.265089722675367,
      "grad_norm": 6.636152744293213,
      "learning_rate": 1.840379166626549e-05,
      "loss": 0.1464,
      "num_input_tokens_seen": 82912912,
      "step": 38405
    },
    {
      "epoch": 6.265905383360522,
      "grad_norm": 7.255515098571777,
      "learning_rate": 1.8396926182991485e-05,
      "loss": 0.1737,
      "num_input_tokens_seen": 82923664,
      "step": 38410
    },
    {
      "epoch": 6.266721044045677,
      "grad_norm": 0.1796352118253708,
      "learning_rate": 1.8390061234996147e-05,
      "loss": 0.0075,
      "num_input_tokens_seen": 82934256,
      "step": 38415
    },
    {
      "epoch": 6.267536704730832,
      "grad_norm": 4.006227016448975,
      "learning_rate": 1.8383196822835984e-05,
      "loss": 0.2556,
      "num_input_tokens_seen": 82945712,
      "step": 38420
    },
    {
      "epoch": 6.268352365415987,
      "grad_norm": 0.11999749392271042,
      "learning_rate": 1.837633294706746e-05,
      "loss": 0.01,
      "num_input_tokens_seen": 82956720,
      "step": 38425
    },
    {
      "epoch": 6.269168026101142,
      "grad_norm": 0.09215438365936279,
      "learning_rate": 1.8369469608246993e-05,
      "loss": 0.0551,
      "num_input_tokens_seen": 82967856,
      "step": 38430
    },
    {
      "epoch": 6.269983686786297,
      "grad_norm": 0.3104093372821808,
      "learning_rate": 1.8362606806930964e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 82979472,
      "step": 38435
    },
    {
      "epoch": 6.270799347471452,
      "grad_norm": 0.18833382427692413,
      "learning_rate": 1.835574454367571e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 82989264,
      "step": 38440
    },
    {
      "epoch": 6.271615008156607,
      "grad_norm": 0.8444291949272156,
      "learning_rate": 1.834888281903751e-05,
      "loss": 0.1443,
      "num_input_tokens_seen": 83000368,
      "step": 38445
    },
    {
      "epoch": 6.2724306688417615,
      "grad_norm": 0.038760919123888016,
      "learning_rate": 1.8342021633572617e-05,
      "loss": 0.0711,
      "num_input_tokens_seen": 83011376,
      "step": 38450
    },
    {
      "epoch": 6.273246329526917,
      "grad_norm": 0.12532900273799896,
      "learning_rate": 1.833516098783723e-05,
      "loss": 0.1228,
      "num_input_tokens_seen": 83021008,
      "step": 38455
    },
    {
      "epoch": 6.274061990212072,
      "grad_norm": 0.0743989646434784,
      "learning_rate": 1.832830088238751e-05,
      "loss": 0.1924,
      "num_input_tokens_seen": 83031888,
      "step": 38460
    },
    {
      "epoch": 6.274877650897227,
      "grad_norm": 0.13158778846263885,
      "learning_rate": 1.832144131777958e-05,
      "loss": 0.1937,
      "num_input_tokens_seen": 83042032,
      "step": 38465
    },
    {
      "epoch": 6.275693311582382,
      "grad_norm": 0.07835894078016281,
      "learning_rate": 1.83145822945695e-05,
      "loss": 0.2449,
      "num_input_tokens_seen": 83051920,
      "step": 38470
    },
    {
      "epoch": 6.2765089722675365,
      "grad_norm": 0.09954798221588135,
      "learning_rate": 1.8307723813313298e-05,
      "loss": 0.2882,
      "num_input_tokens_seen": 83063280,
      "step": 38475
    },
    {
      "epoch": 6.277324632952691,
      "grad_norm": 0.15163902938365936,
      "learning_rate": 1.8300865874566953e-05,
      "loss": 0.0783,
      "num_input_tokens_seen": 83072112,
      "step": 38480
    },
    {
      "epoch": 6.278140293637847,
      "grad_norm": 0.15219746530056,
      "learning_rate": 1.829400847888642e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 83083728,
      "step": 38485
    },
    {
      "epoch": 6.278955954323002,
      "grad_norm": 27.522193908691406,
      "learning_rate": 1.8287151626827586e-05,
      "loss": 0.3392,
      "num_input_tokens_seen": 83094256,
      "step": 38490
    },
    {
      "epoch": 6.279771615008157,
      "grad_norm": 0.11280357092618942,
      "learning_rate": 1.8280295318946304e-05,
      "loss": 0.1118,
      "num_input_tokens_seen": 83105552,
      "step": 38495
    },
    {
      "epoch": 6.280587275693311,
      "grad_norm": 0.0732632502913475,
      "learning_rate": 1.827343955579838e-05,
      "loss": 0.0165,
      "num_input_tokens_seen": 83117296,
      "step": 38500
    },
    {
      "epoch": 6.281402936378466,
      "grad_norm": 0.12993024289608002,
      "learning_rate": 1.8266584337939568e-05,
      "loss": 0.1247,
      "num_input_tokens_seen": 83127824,
      "step": 38505
    },
    {
      "epoch": 6.282218597063622,
      "grad_norm": 0.354735404253006,
      "learning_rate": 1.82597296659256e-05,
      "loss": 0.173,
      "num_input_tokens_seen": 83138608,
      "step": 38510
    },
    {
      "epoch": 6.283034257748777,
      "grad_norm": 0.7837084531784058,
      "learning_rate": 1.8252875540312143e-05,
      "loss": 0.1228,
      "num_input_tokens_seen": 83149936,
      "step": 38515
    },
    {
      "epoch": 6.283849918433932,
      "grad_norm": 3.745943069458008,
      "learning_rate": 1.824602196165483e-05,
      "loss": 0.1244,
      "num_input_tokens_seen": 83160528,
      "step": 38520
    },
    {
      "epoch": 6.284665579119086,
      "grad_norm": 0.27920007705688477,
      "learning_rate": 1.823916893050925e-05,
      "loss": 0.1329,
      "num_input_tokens_seen": 83170640,
      "step": 38525
    },
    {
      "epoch": 6.285481239804241,
      "grad_norm": 0.08819136768579483,
      "learning_rate": 1.8232316447430936e-05,
      "loss": 0.0157,
      "num_input_tokens_seen": 83182128,
      "step": 38530
    },
    {
      "epoch": 6.286296900489396,
      "grad_norm": 0.5468014478683472,
      "learning_rate": 1.822546451297539e-05,
      "loss": 0.1199,
      "num_input_tokens_seen": 83192368,
      "step": 38535
    },
    {
      "epoch": 6.287112561174552,
      "grad_norm": 0.15090271830558777,
      "learning_rate": 1.8218613127698058e-05,
      "loss": 0.1052,
      "num_input_tokens_seen": 83203024,
      "step": 38540
    },
    {
      "epoch": 6.287928221859707,
      "grad_norm": 2.385889768600464,
      "learning_rate": 1.8211762292154362e-05,
      "loss": 0.2688,
      "num_input_tokens_seen": 83214160,
      "step": 38545
    },
    {
      "epoch": 6.288743882544861,
      "grad_norm": 0.05856755003333092,
      "learning_rate": 1.820491200689966e-05,
      "loss": 0.1684,
      "num_input_tokens_seen": 83224784,
      "step": 38550
    },
    {
      "epoch": 6.289559543230016,
      "grad_norm": 0.28382059931755066,
      "learning_rate": 1.8198062272489263e-05,
      "loss": 0.0408,
      "num_input_tokens_seen": 83235344,
      "step": 38555
    },
    {
      "epoch": 6.290375203915171,
      "grad_norm": 0.16582660377025604,
      "learning_rate": 1.8191213089478455e-05,
      "loss": 0.2156,
      "num_input_tokens_seen": 83246384,
      "step": 38560
    },
    {
      "epoch": 6.291190864600326,
      "grad_norm": 0.6182851791381836,
      "learning_rate": 1.818436445842246e-05,
      "loss": 0.2381,
      "num_input_tokens_seen": 83257520,
      "step": 38565
    },
    {
      "epoch": 6.2920065252854815,
      "grad_norm": 0.17569006979465485,
      "learning_rate": 1.8177516379876463e-05,
      "loss": 0.1326,
      "num_input_tokens_seen": 83267824,
      "step": 38570
    },
    {
      "epoch": 6.292822185970636,
      "grad_norm": 0.11772775650024414,
      "learning_rate": 1.817066885439561e-05,
      "loss": 0.0093,
      "num_input_tokens_seen": 83277904,
      "step": 38575
    },
    {
      "epoch": 6.293637846655791,
      "grad_norm": 0.06668476015329361,
      "learning_rate": 1.8163821882534986e-05,
      "loss": 0.0043,
      "num_input_tokens_seen": 83287280,
      "step": 38580
    },
    {
      "epoch": 6.294453507340946,
      "grad_norm": 2.960515022277832,
      "learning_rate": 1.815697546484964e-05,
      "loss": 0.2435,
      "num_input_tokens_seen": 83297520,
      "step": 38585
    },
    {
      "epoch": 6.295269168026101,
      "grad_norm": 0.06321658194065094,
      "learning_rate": 1.8150129601894592e-05,
      "loss": 0.012,
      "num_input_tokens_seen": 83308400,
      "step": 38590
    },
    {
      "epoch": 6.2960848287112565,
      "grad_norm": 0.2736243009567261,
      "learning_rate": 1.8143284294224794e-05,
      "loss": 0.1795,
      "num_input_tokens_seen": 83321008,
      "step": 38595
    },
    {
      "epoch": 6.296900489396411,
      "grad_norm": 0.1401585191488266,
      "learning_rate": 1.813643954239516e-05,
      "loss": 0.1012,
      "num_input_tokens_seen": 83333264,
      "step": 38600
    },
    {
      "epoch": 6.297716150081566,
      "grad_norm": 3.16827392578125,
      "learning_rate": 1.8129595346960568e-05,
      "loss": 0.0562,
      "num_input_tokens_seen": 83343856,
      "step": 38605
    },
    {
      "epoch": 6.298531810766721,
      "grad_norm": 0.158759206533432,
      "learning_rate": 1.812275170847583e-05,
      "loss": 0.0087,
      "num_input_tokens_seen": 83354832,
      "step": 38610
    },
    {
      "epoch": 6.299347471451876,
      "grad_norm": 0.03607413172721863,
      "learning_rate": 1.8115908627495742e-05,
      "loss": 0.0905,
      "num_input_tokens_seen": 83365488,
      "step": 38615
    },
    {
      "epoch": 6.300163132137031,
      "grad_norm": 1.4858367443084717,
      "learning_rate": 1.8109066104575023e-05,
      "loss": 0.119,
      "num_input_tokens_seen": 83376688,
      "step": 38620
    },
    {
      "epoch": 6.300978792822186,
      "grad_norm": 12.213780403137207,
      "learning_rate": 1.810222414026837e-05,
      "loss": 0.0787,
      "num_input_tokens_seen": 83387056,
      "step": 38625
    },
    {
      "epoch": 6.301794453507341,
      "grad_norm": 2.129650592803955,
      "learning_rate": 1.809538273513043e-05,
      "loss": 0.0181,
      "num_input_tokens_seen": 83398224,
      "step": 38630
    },
    {
      "epoch": 6.302610114192496,
      "grad_norm": 0.9154722690582275,
      "learning_rate": 1.8088541889715795e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 83408880,
      "step": 38635
    },
    {
      "epoch": 6.303425774877651,
      "grad_norm": 0.13156336545944214,
      "learning_rate": 1.8081701604579025e-05,
      "loss": 0.2049,
      "num_input_tokens_seen": 83419504,
      "step": 38640
    },
    {
      "epoch": 6.304241435562806,
      "grad_norm": 0.0758514329791069,
      "learning_rate": 1.807486188027463e-05,
      "loss": 0.0055,
      "num_input_tokens_seen": 83429296,
      "step": 38645
    },
    {
      "epoch": 6.30505709624796,
      "grad_norm": 0.018398450687527657,
      "learning_rate": 1.8068022717357066e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 83440912,
      "step": 38650
    },
    {
      "epoch": 6.305872756933116,
      "grad_norm": 0.160930797457695,
      "learning_rate": 1.8061184116380754e-05,
      "loss": 0.0898,
      "num_input_tokens_seen": 83451568,
      "step": 38655
    },
    {
      "epoch": 6.306688417618271,
      "grad_norm": 0.0657789558172226,
      "learning_rate": 1.8054346077900065e-05,
      "loss": 0.0196,
      "num_input_tokens_seen": 83461200,
      "step": 38660
    },
    {
      "epoch": 6.307504078303426,
      "grad_norm": 0.06973383575677872,
      "learning_rate": 1.8047508602469322e-05,
      "loss": 0.0092,
      "num_input_tokens_seen": 83472016,
      "step": 38665
    },
    {
      "epoch": 6.308319738988581,
      "grad_norm": 0.2847124934196472,
      "learning_rate": 1.804067169064281e-05,
      "loss": 0.1104,
      "num_input_tokens_seen": 83482896,
      "step": 38670
    },
    {
      "epoch": 6.309135399673735,
      "grad_norm": 3.6392717361450195,
      "learning_rate": 1.8033835342974763e-05,
      "loss": 0.3532,
      "num_input_tokens_seen": 83494384,
      "step": 38675
    },
    {
      "epoch": 6.309951060358891,
      "grad_norm": 0.10749179869890213,
      "learning_rate": 1.8026999560019366e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 83504848,
      "step": 38680
    },
    {
      "epoch": 6.310766721044046,
      "grad_norm": 0.03465574234724045,
      "learning_rate": 1.8020164342330763e-05,
      "loss": 0.0922,
      "num_input_tokens_seen": 83516208,
      "step": 38685
    },
    {
      "epoch": 6.311582381729201,
      "grad_norm": 9.664535522460938,
      "learning_rate": 1.8013329690463056e-05,
      "loss": 0.1194,
      "num_input_tokens_seen": 83527056,
      "step": 38690
    },
    {
      "epoch": 6.3123980424143555,
      "grad_norm": 0.35942038893699646,
      "learning_rate": 1.8006495604970295e-05,
      "loss": 0.0083,
      "num_input_tokens_seen": 83537392,
      "step": 38695
    },
    {
      "epoch": 6.31321370309951,
      "grad_norm": 0.09744260460138321,
      "learning_rate": 1.7999662086406484e-05,
      "loss": 0.0749,
      "num_input_tokens_seen": 83548944,
      "step": 38700
    },
    {
      "epoch": 6.314029363784665,
      "grad_norm": 0.19223381578922272,
      "learning_rate": 1.799282913532559e-05,
      "loss": 0.0115,
      "num_input_tokens_seen": 83558832,
      "step": 38705
    },
    {
      "epoch": 6.314845024469821,
      "grad_norm": 0.1924734115600586,
      "learning_rate": 1.798599675228151e-05,
      "loss": 0.1044,
      "num_input_tokens_seen": 83567696,
      "step": 38710
    },
    {
      "epoch": 6.315660685154976,
      "grad_norm": 0.03546510264277458,
      "learning_rate": 1.7979164937828127e-05,
      "loss": 0.0055,
      "num_input_tokens_seen": 83578480,
      "step": 38715
    },
    {
      "epoch": 6.3164763458401305,
      "grad_norm": 0.02767333947122097,
      "learning_rate": 1.797233369251926e-05,
      "loss": 0.1412,
      "num_input_tokens_seen": 83588496,
      "step": 38720
    },
    {
      "epoch": 6.317292006525285,
      "grad_norm": 0.07581031322479248,
      "learning_rate": 1.796550301690868e-05,
      "loss": 0.063,
      "num_input_tokens_seen": 83599376,
      "step": 38725
    },
    {
      "epoch": 6.31810766721044,
      "grad_norm": 0.11516465246677399,
      "learning_rate": 1.7958672911550117e-05,
      "loss": 0.0805,
      "num_input_tokens_seen": 83610032,
      "step": 38730
    },
    {
      "epoch": 6.318923327895595,
      "grad_norm": 0.08749879896640778,
      "learning_rate": 1.7951843376997256e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 83621392,
      "step": 38735
    },
    {
      "epoch": 6.319738988580751,
      "grad_norm": 4.46632194519043,
      "learning_rate": 1.7945014413803737e-05,
      "loss": 0.0197,
      "num_input_tokens_seen": 83632368,
      "step": 38740
    },
    {
      "epoch": 6.3205546492659055,
      "grad_norm": 0.09072194993495941,
      "learning_rate": 1.7938186022523144e-05,
      "loss": 0.003,
      "num_input_tokens_seen": 83642800,
      "step": 38745
    },
    {
      "epoch": 6.32137030995106,
      "grad_norm": 1.0660722255706787,
      "learning_rate": 1.793135820370902e-05,
      "loss": 0.0938,
      "num_input_tokens_seen": 83653552,
      "step": 38750
    },
    {
      "epoch": 6.322185970636215,
      "grad_norm": 0.13588771224021912,
      "learning_rate": 1.792453095791487e-05,
      "loss": 0.0076,
      "num_input_tokens_seen": 83664624,
      "step": 38755
    },
    {
      "epoch": 6.32300163132137,
      "grad_norm": 0.11264606565237045,
      "learning_rate": 1.791770428569414e-05,
      "loss": 0.0045,
      "num_input_tokens_seen": 83675536,
      "step": 38760
    },
    {
      "epoch": 6.323817292006526,
      "grad_norm": 0.019025340676307678,
      "learning_rate": 1.7910878187600232e-05,
      "loss": 0.0034,
      "num_input_tokens_seen": 83685712,
      "step": 38765
    },
    {
      "epoch": 6.3246329526916805,
      "grad_norm": 0.05009276419878006,
      "learning_rate": 1.790405266418651e-05,
      "loss": 0.0983,
      "num_input_tokens_seen": 83696112,
      "step": 38770
    },
    {
      "epoch": 6.325448613376835,
      "grad_norm": 0.08625942468643188,
      "learning_rate": 1.789722771600628e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 83706608,
      "step": 38775
    },
    {
      "epoch": 6.32626427406199,
      "grad_norm": 0.0683288499712944,
      "learning_rate": 1.789040334361282e-05,
      "loss": 0.1019,
      "num_input_tokens_seen": 83718288,
      "step": 38780
    },
    {
      "epoch": 6.327079934747145,
      "grad_norm": 6.77628755569458,
      "learning_rate": 1.788357954755933e-05,
      "loss": 0.0862,
      "num_input_tokens_seen": 83729328,
      "step": 38785
    },
    {
      "epoch": 6.327895595432301,
      "grad_norm": 3.364013671875,
      "learning_rate": 1.7876756328398998e-05,
      "loss": 0.2365,
      "num_input_tokens_seen": 83740016,
      "step": 38790
    },
    {
      "epoch": 6.328711256117455,
      "grad_norm": 4.250802993774414,
      "learning_rate": 1.7869933686684938e-05,
      "loss": 0.0146,
      "num_input_tokens_seen": 83751120,
      "step": 38795
    },
    {
      "epoch": 6.32952691680261,
      "grad_norm": 0.09552743285894394,
      "learning_rate": 1.786311162297022e-05,
      "loss": 0.1188,
      "num_input_tokens_seen": 83760880,
      "step": 38800
    },
    {
      "epoch": 6.330342577487765,
      "grad_norm": 0.10108266025781631,
      "learning_rate": 1.7856290137807893e-05,
      "loss": 0.1115,
      "num_input_tokens_seen": 83771440,
      "step": 38805
    },
    {
      "epoch": 6.33115823817292,
      "grad_norm": 0.03804726153612137,
      "learning_rate": 1.7849469231750936e-05,
      "loss": 0.0989,
      "num_input_tokens_seen": 83781424,
      "step": 38810
    },
    {
      "epoch": 6.331973898858075,
      "grad_norm": 0.0366843044757843,
      "learning_rate": 1.784264890535229e-05,
      "loss": 0.1035,
      "num_input_tokens_seen": 83792848,
      "step": 38815
    },
    {
      "epoch": 6.33278955954323,
      "grad_norm": 0.2906888723373413,
      "learning_rate": 1.7835829159164835e-05,
      "loss": 0.1995,
      "num_input_tokens_seen": 83803152,
      "step": 38820
    },
    {
      "epoch": 6.333605220228385,
      "grad_norm": 0.18760044872760773,
      "learning_rate": 1.7829009993741418e-05,
      "loss": 0.0818,
      "num_input_tokens_seen": 83814672,
      "step": 38825
    },
    {
      "epoch": 6.33442088091354,
      "grad_norm": 0.08920400589704514,
      "learning_rate": 1.782219140963484e-05,
      "loss": 0.1291,
      "num_input_tokens_seen": 83825904,
      "step": 38830
    },
    {
      "epoch": 6.335236541598695,
      "grad_norm": 0.0731261745095253,
      "learning_rate": 1.781537340739784e-05,
      "loss": 0.0027,
      "num_input_tokens_seen": 83837936,
      "step": 38835
    },
    {
      "epoch": 6.33605220228385,
      "grad_norm": 5.2528228759765625,
      "learning_rate": 1.780855598758313e-05,
      "loss": 0.1678,
      "num_input_tokens_seen": 83848944,
      "step": 38840
    },
    {
      "epoch": 6.3368678629690045,
      "grad_norm": 4.945896148681641,
      "learning_rate": 1.7801739150743363e-05,
      "loss": 0.113,
      "num_input_tokens_seen": 83861136,
      "step": 38845
    },
    {
      "epoch": 6.33768352365416,
      "grad_norm": 5.178552150726318,
      "learning_rate": 1.7794922897431145e-05,
      "loss": 0.1642,
      "num_input_tokens_seen": 83873136,
      "step": 38850
    },
    {
      "epoch": 6.338499184339315,
      "grad_norm": 0.13779862225055695,
      "learning_rate": 1.7788107228199023e-05,
      "loss": 0.0867,
      "num_input_tokens_seen": 83884496,
      "step": 38855
    },
    {
      "epoch": 6.33931484502447,
      "grad_norm": 0.08722608536481857,
      "learning_rate": 1.7781292143599532e-05,
      "loss": 0.0036,
      "num_input_tokens_seen": 83895664,
      "step": 38860
    },
    {
      "epoch": 6.340130505709625,
      "grad_norm": 0.6137920618057251,
      "learning_rate": 1.7774477644185125e-05,
      "loss": 0.1541,
      "num_input_tokens_seen": 83907664,
      "step": 38865
    },
    {
      "epoch": 6.3409461663947795,
      "grad_norm": 0.1464776247739792,
      "learning_rate": 1.7767663730508222e-05,
      "loss": 0.1053,
      "num_input_tokens_seen": 83918896,
      "step": 38870
    },
    {
      "epoch": 6.341761827079935,
      "grad_norm": 0.21143722534179688,
      "learning_rate": 1.7760850403121195e-05,
      "loss": 0.081,
      "num_input_tokens_seen": 83929008,
      "step": 38875
    },
    {
      "epoch": 6.34257748776509,
      "grad_norm": 0.07302909344434738,
      "learning_rate": 1.7754037662576365e-05,
      "loss": 0.0509,
      "num_input_tokens_seen": 83937776,
      "step": 38880
    },
    {
      "epoch": 6.343393148450245,
      "grad_norm": 0.4305923879146576,
      "learning_rate": 1.7747225509426008e-05,
      "loss": 0.0052,
      "num_input_tokens_seen": 83948816,
      "step": 38885
    },
    {
      "epoch": 6.3442088091354,
      "grad_norm": 0.07678848505020142,
      "learning_rate": 1.774041394422235e-05,
      "loss": 0.1067,
      "num_input_tokens_seen": 83959440,
      "step": 38890
    },
    {
      "epoch": 6.3450244698205545,
      "grad_norm": 0.060962971299886703,
      "learning_rate": 1.7733602967517578e-05,
      "loss": 0.004,
      "num_input_tokens_seen": 83971248,
      "step": 38895
    },
    {
      "epoch": 6.345840130505709,
      "grad_norm": 0.05522932857275009,
      "learning_rate": 1.772679257986381e-05,
      "loss": 0.0049,
      "num_input_tokens_seen": 83981584,
      "step": 38900
    },
    {
      "epoch": 6.346655791190865,
      "grad_norm": 0.15104833245277405,
      "learning_rate": 1.7719982781813135e-05,
      "loss": 0.1058,
      "num_input_tokens_seen": 83992112,
      "step": 38905
    },
    {
      "epoch": 6.34747145187602,
      "grad_norm": 0.43337759375572205,
      "learning_rate": 1.7713173573917603e-05,
      "loss": 0.0049,
      "num_input_tokens_seen": 84003472,
      "step": 38910
    },
    {
      "epoch": 6.348287112561175,
      "grad_norm": 0.06316080689430237,
      "learning_rate": 1.7706364956729195e-05,
      "loss": 0.0724,
      "num_input_tokens_seen": 84013840,
      "step": 38915
    },
    {
      "epoch": 6.349102773246329,
      "grad_norm": 0.0270449910312891,
      "learning_rate": 1.769955693079985e-05,
      "loss": 0.0408,
      "num_input_tokens_seen": 84024912,
      "step": 38920
    },
    {
      "epoch": 6.349918433931484,
      "grad_norm": 0.011509484611451626,
      "learning_rate": 1.769274949668146e-05,
      "loss": 0.0884,
      "num_input_tokens_seen": 84035632,
      "step": 38925
    },
    {
      "epoch": 6.350734094616639,
      "grad_norm": 3.131605863571167,
      "learning_rate": 1.7685942654925876e-05,
      "loss": 0.0764,
      "num_input_tokens_seen": 84046448,
      "step": 38930
    },
    {
      "epoch": 6.351549755301795,
      "grad_norm": 0.15195944905281067,
      "learning_rate": 1.767913640608489e-05,
      "loss": 0.1173,
      "num_input_tokens_seen": 84057648,
      "step": 38935
    },
    {
      "epoch": 6.35236541598695,
      "grad_norm": 4.266190528869629,
      "learning_rate": 1.7672330750710247e-05,
      "loss": 0.1128,
      "num_input_tokens_seen": 84068880,
      "step": 38940
    },
    {
      "epoch": 6.353181076672104,
      "grad_norm": 0.09167669713497162,
      "learning_rate": 1.766552568935366e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 84079696,
      "step": 38945
    },
    {
      "epoch": 6.353996737357259,
      "grad_norm": 3.26637864112854,
      "learning_rate": 1.7658721222566775e-05,
      "loss": 0.1575,
      "num_input_tokens_seen": 84089456,
      "step": 38950
    },
    {
      "epoch": 6.354812398042414,
      "grad_norm": 0.09386517852544785,
      "learning_rate": 1.76519173509012e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 84099440,
      "step": 38955
    },
    {
      "epoch": 6.35562805872757,
      "grad_norm": 0.8102290034294128,
      "learning_rate": 1.764511407490848e-05,
      "loss": 0.242,
      "num_input_tokens_seen": 84109232,
      "step": 38960
    },
    {
      "epoch": 6.356443719412725,
      "grad_norm": 0.09411946684122086,
      "learning_rate": 1.763831139514014e-05,
      "loss": 0.1275,
      "num_input_tokens_seen": 84121712,
      "step": 38965
    },
    {
      "epoch": 6.357259380097879,
      "grad_norm": 0.06968369334936142,
      "learning_rate": 1.7631509312147626e-05,
      "loss": 0.052,
      "num_input_tokens_seen": 84132688,
      "step": 38970
    },
    {
      "epoch": 6.358075040783034,
      "grad_norm": 9.786787986755371,
      "learning_rate": 1.7624707826482356e-05,
      "loss": 0.0797,
      "num_input_tokens_seen": 84142544,
      "step": 38975
    },
    {
      "epoch": 6.358890701468189,
      "grad_norm": 6.882516384124756,
      "learning_rate": 1.7617906938695694e-05,
      "loss": 0.2608,
      "num_input_tokens_seen": 84151312,
      "step": 38980
    },
    {
      "epoch": 6.359706362153344,
      "grad_norm": 2.8158648014068604,
      "learning_rate": 1.761110664933895e-05,
      "loss": 0.1112,
      "num_input_tokens_seen": 84162512,
      "step": 38985
    },
    {
      "epoch": 6.3605220228384995,
      "grad_norm": 0.48013320565223694,
      "learning_rate": 1.760430695896339e-05,
      "loss": 0.0105,
      "num_input_tokens_seen": 84174480,
      "step": 38990
    },
    {
      "epoch": 6.361337683523654,
      "grad_norm": 0.06856291741132736,
      "learning_rate": 1.7597507868120227e-05,
      "loss": 0.0101,
      "num_input_tokens_seen": 84185712,
      "step": 38995
    },
    {
      "epoch": 6.362153344208809,
      "grad_norm": 0.05834678187966347,
      "learning_rate": 1.7590709377360648e-05,
      "loss": 0.1768,
      "num_input_tokens_seen": 84195824,
      "step": 39000
    },
    {
      "epoch": 6.362969004893964,
      "grad_norm": 2.5577142238616943,
      "learning_rate": 1.7583911487235753e-05,
      "loss": 0.1976,
      "num_input_tokens_seen": 84207344,
      "step": 39005
    },
    {
      "epoch": 6.363784665579119,
      "grad_norm": 0.03688184916973114,
      "learning_rate": 1.7577114198296623e-05,
      "loss": 0.0036,
      "num_input_tokens_seen": 84218512,
      "step": 39010
    },
    {
      "epoch": 6.364600326264274,
      "grad_norm": 0.2032366544008255,
      "learning_rate": 1.757031751109428e-05,
      "loss": 0.006,
      "num_input_tokens_seen": 84229520,
      "step": 39015
    },
    {
      "epoch": 6.365415986949429,
      "grad_norm": 0.01645730994641781,
      "learning_rate": 1.75635214261797e-05,
      "loss": 0.1333,
      "num_input_tokens_seen": 84240752,
      "step": 39020
    },
    {
      "epoch": 6.366231647634584,
      "grad_norm": 0.12931299209594727,
      "learning_rate": 1.7556725944103803e-05,
      "loss": 0.1052,
      "num_input_tokens_seen": 84250224,
      "step": 39025
    },
    {
      "epoch": 6.367047308319739,
      "grad_norm": 0.09920920431613922,
      "learning_rate": 1.754993106541747e-05,
      "loss": 0.0026,
      "num_input_tokens_seen": 84260784,
      "step": 39030
    },
    {
      "epoch": 6.367862969004894,
      "grad_norm": 0.21248573064804077,
      "learning_rate": 1.7543136790671524e-05,
      "loss": 0.0563,
      "num_input_tokens_seen": 84272560,
      "step": 39035
    },
    {
      "epoch": 6.368678629690049,
      "grad_norm": 0.04769877716898918,
      "learning_rate": 1.753634312041675e-05,
      "loss": 0.006,
      "num_input_tokens_seen": 84282640,
      "step": 39040
    },
    {
      "epoch": 6.369494290375204,
      "grad_norm": 0.1672208607196808,
      "learning_rate": 1.752955005520387e-05,
      "loss": 0.2046,
      "num_input_tokens_seen": 84293936,
      "step": 39045
    },
    {
      "epoch": 6.370309951060359,
      "grad_norm": 0.2145567387342453,
      "learning_rate": 1.7522757595583567e-05,
      "loss": 0.229,
      "num_input_tokens_seen": 84303376,
      "step": 39050
    },
    {
      "epoch": 6.371125611745514,
      "grad_norm": 0.07079474627971649,
      "learning_rate": 1.751596574210647e-05,
      "loss": 0.11,
      "num_input_tokens_seen": 84313968,
      "step": 39055
    },
    {
      "epoch": 6.371941272430669,
      "grad_norm": 0.6985549926757812,
      "learning_rate": 1.750917449532317e-05,
      "loss": 0.1168,
      "num_input_tokens_seen": 84325520,
      "step": 39060
    },
    {
      "epoch": 6.372756933115824,
      "grad_norm": 0.1702076643705368,
      "learning_rate": 1.7502383855784187e-05,
      "loss": 0.0043,
      "num_input_tokens_seen": 84336496,
      "step": 39065
    },
    {
      "epoch": 6.373572593800978,
      "grad_norm": 0.08414614200592041,
      "learning_rate": 1.7495593824040014e-05,
      "loss": 0.1257,
      "num_input_tokens_seen": 84346928,
      "step": 39070
    },
    {
      "epoch": 6.374388254486134,
      "grad_norm": 13.162398338317871,
      "learning_rate": 1.7488804400641084e-05,
      "loss": 0.0405,
      "num_input_tokens_seen": 84358032,
      "step": 39075
    },
    {
      "epoch": 6.375203915171289,
      "grad_norm": 0.10966675728559494,
      "learning_rate": 1.7482015586137774e-05,
      "loss": 0.0055,
      "num_input_tokens_seen": 84368816,
      "step": 39080
    },
    {
      "epoch": 6.376019575856444,
      "grad_norm": 0.1741647720336914,
      "learning_rate": 1.7475227381080434e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 84378800,
      "step": 39085
    },
    {
      "epoch": 6.376835236541599,
      "grad_norm": 0.05604543536901474,
      "learning_rate": 1.746843978601934e-05,
      "loss": 0.0057,
      "num_input_tokens_seen": 84389584,
      "step": 39090
    },
    {
      "epoch": 6.377650897226753,
      "grad_norm": 0.08365931361913681,
      "learning_rate": 1.746165280150473e-05,
      "loss": 0.2504,
      "num_input_tokens_seen": 84398768,
      "step": 39095
    },
    {
      "epoch": 6.378466557911908,
      "grad_norm": 0.09252443164587021,
      "learning_rate": 1.7454866428086797e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 84408464,
      "step": 39100
    },
    {
      "epoch": 6.379282218597064,
      "grad_norm": 0.052541542798280716,
      "learning_rate": 1.7448080666315675e-05,
      "loss": 0.0031,
      "num_input_tokens_seen": 84419600,
      "step": 39105
    },
    {
      "epoch": 6.380097879282219,
      "grad_norm": 3.5105550289154053,
      "learning_rate": 1.744129551674145e-05,
      "loss": 0.0097,
      "num_input_tokens_seen": 84428784,
      "step": 39110
    },
    {
      "epoch": 6.3809135399673735,
      "grad_norm": 0.08369387686252594,
      "learning_rate": 1.7434510979914166e-05,
      "loss": 0.008,
      "num_input_tokens_seen": 84439696,
      "step": 39115
    },
    {
      "epoch": 6.381729200652528,
      "grad_norm": 2.1654927730560303,
      "learning_rate": 1.7427727056383795e-05,
      "loss": 0.1293,
      "num_input_tokens_seen": 84451280,
      "step": 39120
    },
    {
      "epoch": 6.382544861337683,
      "grad_norm": 3.6602649688720703,
      "learning_rate": 1.74209437467003e-05,
      "loss": 0.1376,
      "num_input_tokens_seen": 84462512,
      "step": 39125
    },
    {
      "epoch": 6.383360522022839,
      "grad_norm": 0.12523126602172852,
      "learning_rate": 1.7414161051413565e-05,
      "loss": 0.0077,
      "num_input_tokens_seen": 84473616,
      "step": 39130
    },
    {
      "epoch": 6.384176182707994,
      "grad_norm": 0.3088226020336151,
      "learning_rate": 1.740737897107342e-05,
      "loss": 0.0184,
      "num_input_tokens_seen": 84485776,
      "step": 39135
    },
    {
      "epoch": 6.3849918433931485,
      "grad_norm": 0.13221058249473572,
      "learning_rate": 1.7400597506229667e-05,
      "loss": 0.0338,
      "num_input_tokens_seen": 84498064,
      "step": 39140
    },
    {
      "epoch": 6.385807504078303,
      "grad_norm": 0.04836839437484741,
      "learning_rate": 1.739381665743203e-05,
      "loss": 0.2277,
      "num_input_tokens_seen": 84509296,
      "step": 39145
    },
    {
      "epoch": 6.386623164763458,
      "grad_norm": 7.938859939575195,
      "learning_rate": 1.7387036425230214e-05,
      "loss": 0.0136,
      "num_input_tokens_seen": 84519888,
      "step": 39150
    },
    {
      "epoch": 6.387438825448613,
      "grad_norm": 0.42295539379119873,
      "learning_rate": 1.7380256810173854e-05,
      "loss": 0.104,
      "num_input_tokens_seen": 84530352,
      "step": 39155
    },
    {
      "epoch": 6.388254486133769,
      "grad_norm": 0.05365495756268501,
      "learning_rate": 1.7373477812812538e-05,
      "loss": 0.1059,
      "num_input_tokens_seen": 84540912,
      "step": 39160
    },
    {
      "epoch": 6.3890701468189235,
      "grad_norm": 0.09460272639989853,
      "learning_rate": 1.73666994336958e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 84551760,
      "step": 39165
    },
    {
      "epoch": 6.389885807504078,
      "grad_norm": 0.10183282941579819,
      "learning_rate": 1.735992167337314e-05,
      "loss": 0.0425,
      "num_input_tokens_seen": 84562960,
      "step": 39170
    },
    {
      "epoch": 6.390701468189233,
      "grad_norm": 0.26614123582839966,
      "learning_rate": 1.7353144532394e-05,
      "loss": 0.1172,
      "num_input_tokens_seen": 84573264,
      "step": 39175
    },
    {
      "epoch": 6.391517128874388,
      "grad_norm": 0.1066962406039238,
      "learning_rate": 1.734636801130776e-05,
      "loss": 0.0045,
      "num_input_tokens_seen": 84583632,
      "step": 39180
    },
    {
      "epoch": 6.392332789559543,
      "grad_norm": 4.822566509246826,
      "learning_rate": 1.7339592110663768e-05,
      "loss": 0.4313,
      "num_input_tokens_seen": 84594960,
      "step": 39185
    },
    {
      "epoch": 6.3931484502446985,
      "grad_norm": 0.14477436244487762,
      "learning_rate": 1.7332816831011307e-05,
      "loss": 0.081,
      "num_input_tokens_seen": 84604912,
      "step": 39190
    },
    {
      "epoch": 6.393964110929853,
      "grad_norm": 0.11814351379871368,
      "learning_rate": 1.7326042172899616e-05,
      "loss": 0.1229,
      "num_input_tokens_seen": 84615344,
      "step": 39195
    },
    {
      "epoch": 6.394779771615008,
      "grad_norm": 1.5876661539077759,
      "learning_rate": 1.731926813687788e-05,
      "loss": 0.0108,
      "num_input_tokens_seen": 84625744,
      "step": 39200
    },
    {
      "epoch": 6.395595432300163,
      "grad_norm": 2.3105623722076416,
      "learning_rate": 1.7312494723495243e-05,
      "loss": 0.0859,
      "num_input_tokens_seen": 84636784,
      "step": 39205
    },
    {
      "epoch": 6.396411092985318,
      "grad_norm": 3.1567940711975098,
      "learning_rate": 1.730572193330079e-05,
      "loss": 0.0882,
      "num_input_tokens_seen": 84648656,
      "step": 39210
    },
    {
      "epoch": 6.397226753670473,
      "grad_norm": 0.052460797131061554,
      "learning_rate": 1.7298949766843558e-05,
      "loss": 0.0041,
      "num_input_tokens_seen": 84659728,
      "step": 39215
    },
    {
      "epoch": 6.398042414355628,
      "grad_norm": 0.10526681691408157,
      "learning_rate": 1.7292178224672528e-05,
      "loss": 0.0048,
      "num_input_tokens_seen": 84670672,
      "step": 39220
    },
    {
      "epoch": 6.398858075040783,
      "grad_norm": 7.077473163604736,
      "learning_rate": 1.7285407307336636e-05,
      "loss": 0.1098,
      "num_input_tokens_seen": 84679952,
      "step": 39225
    },
    {
      "epoch": 6.399673735725938,
      "grad_norm": 0.07950261980295181,
      "learning_rate": 1.727863701538478e-05,
      "loss": 0.0831,
      "num_input_tokens_seen": 84691024,
      "step": 39230
    },
    {
      "epoch": 6.400489396411093,
      "grad_norm": 0.08905600011348724,
      "learning_rate": 1.7271867349365782e-05,
      "loss": 0.1561,
      "num_input_tokens_seen": 84700976,
      "step": 39235
    },
    {
      "epoch": 6.401305057096248,
      "grad_norm": 0.06938436627388,
      "learning_rate": 1.7265098309828433e-05,
      "loss": 0.0041,
      "num_input_tokens_seen": 84710896,
      "step": 39240
    },
    {
      "epoch": 6.402120717781403,
      "grad_norm": 0.11138947308063507,
      "learning_rate": 1.725832989732146e-05,
      "loss": 0.1157,
      "num_input_tokens_seen": 84722448,
      "step": 39245
    },
    {
      "epoch": 6.402936378466558,
      "grad_norm": 5.175948619842529,
      "learning_rate": 1.725156211239354e-05,
      "loss": 0.2433,
      "num_input_tokens_seen": 84733168,
      "step": 39250
    },
    {
      "epoch": 6.403752039151713,
      "grad_norm": 3.1428029537200928,
      "learning_rate": 1.7244794955593316e-05,
      "loss": 0.0252,
      "num_input_tokens_seen": 84744176,
      "step": 39255
    },
    {
      "epoch": 6.404567699836868,
      "grad_norm": 0.12407192587852478,
      "learning_rate": 1.7238028427469363e-05,
      "loss": 0.0062,
      "num_input_tokens_seen": 84754704,
      "step": 39260
    },
    {
      "epoch": 6.4053833605220225,
      "grad_norm": 2.8569209575653076,
      "learning_rate": 1.7231262528570207e-05,
      "loss": 0.1144,
      "num_input_tokens_seen": 84765584,
      "step": 39265
    },
    {
      "epoch": 6.406199021207178,
      "grad_norm": 4.624263763427734,
      "learning_rate": 1.7224497259444334e-05,
      "loss": 0.0716,
      "num_input_tokens_seen": 84776144,
      "step": 39270
    },
    {
      "epoch": 6.407014681892333,
      "grad_norm": 0.07321374118328094,
      "learning_rate": 1.7217732620640163e-05,
      "loss": 0.1548,
      "num_input_tokens_seen": 84785488,
      "step": 39275
    },
    {
      "epoch": 6.407830342577488,
      "grad_norm": 5.30738639831543,
      "learning_rate": 1.721096861270607e-05,
      "loss": 0.1574,
      "num_input_tokens_seen": 84796080,
      "step": 39280
    },
    {
      "epoch": 6.408646003262643,
      "grad_norm": 0.20457549393177032,
      "learning_rate": 1.7204205236190385e-05,
      "loss": 0.1102,
      "num_input_tokens_seen": 84805520,
      "step": 39285
    },
    {
      "epoch": 6.4094616639477975,
      "grad_norm": 0.08466614037752151,
      "learning_rate": 1.719744249164138e-05,
      "loss": 0.0071,
      "num_input_tokens_seen": 84816464,
      "step": 39290
    },
    {
      "epoch": 6.410277324632952,
      "grad_norm": 0.17640598118305206,
      "learning_rate": 1.7190680379607278e-05,
      "loss": 0.12,
      "num_input_tokens_seen": 84826640,
      "step": 39295
    },
    {
      "epoch": 6.411092985318108,
      "grad_norm": 1.222657561302185,
      "learning_rate": 1.718391890063624e-05,
      "loss": 0.225,
      "num_input_tokens_seen": 84836304,
      "step": 39300
    },
    {
      "epoch": 6.411908646003263,
      "grad_norm": 0.1656293421983719,
      "learning_rate": 1.7177158055276405e-05,
      "loss": 0.0208,
      "num_input_tokens_seen": 84847280,
      "step": 39305
    },
    {
      "epoch": 6.412724306688418,
      "grad_norm": 0.08772674947977066,
      "learning_rate": 1.717039784407582e-05,
      "loss": 0.2311,
      "num_input_tokens_seen": 84859696,
      "step": 39310
    },
    {
      "epoch": 6.4135399673735725,
      "grad_norm": 0.2088811844587326,
      "learning_rate": 1.7163638267582516e-05,
      "loss": 0.0081,
      "num_input_tokens_seen": 84870576,
      "step": 39315
    },
    {
      "epoch": 6.414355628058727,
      "grad_norm": 0.04336438328027725,
      "learning_rate": 1.715687932634446e-05,
      "loss": 0.0867,
      "num_input_tokens_seen": 84882224,
      "step": 39320
    },
    {
      "epoch": 6.415171288743883,
      "grad_norm": 5.473361492156982,
      "learning_rate": 1.715012102090956e-05,
      "loss": 0.1718,
      "num_input_tokens_seen": 84892496,
      "step": 39325
    },
    {
      "epoch": 6.415986949429038,
      "grad_norm": 0.15484346449375153,
      "learning_rate": 1.714336335182567e-05,
      "loss": 0.1921,
      "num_input_tokens_seen": 84903440,
      "step": 39330
    },
    {
      "epoch": 6.416802610114193,
      "grad_norm": 0.5261359810829163,
      "learning_rate": 1.7136606319640616e-05,
      "loss": 0.0083,
      "num_input_tokens_seen": 84914608,
      "step": 39335
    },
    {
      "epoch": 6.417618270799347,
      "grad_norm": 0.11928585916757584,
      "learning_rate": 1.7129849924902157e-05,
      "loss": 0.0054,
      "num_input_tokens_seen": 84926768,
      "step": 39340
    },
    {
      "epoch": 6.418433931484502,
      "grad_norm": 3.9390835762023926,
      "learning_rate": 1.7123094168157994e-05,
      "loss": 0.1103,
      "num_input_tokens_seen": 84937488,
      "step": 39345
    },
    {
      "epoch": 6.419249592169657,
      "grad_norm": 0.059945400804281235,
      "learning_rate": 1.7116339049955788e-05,
      "loss": 0.1013,
      "num_input_tokens_seen": 84949840,
      "step": 39350
    },
    {
      "epoch": 6.420065252854813,
      "grad_norm": 0.22643250226974487,
      "learning_rate": 1.7109584570843136e-05,
      "loss": 0.182,
      "num_input_tokens_seen": 84959088,
      "step": 39355
    },
    {
      "epoch": 6.420880913539968,
      "grad_norm": 0.08659963309764862,
      "learning_rate": 1.7102830731367593e-05,
      "loss": 0.0172,
      "num_input_tokens_seen": 84969232,
      "step": 39360
    },
    {
      "epoch": 6.421696574225122,
      "grad_norm": 0.10305386781692505,
      "learning_rate": 1.7096077532076666e-05,
      "loss": 0.0115,
      "num_input_tokens_seen": 84981488,
      "step": 39365
    },
    {
      "epoch": 6.422512234910277,
      "grad_norm": 2.7358717918395996,
      "learning_rate": 1.7089324973517794e-05,
      "loss": 0.1204,
      "num_input_tokens_seen": 84992688,
      "step": 39370
    },
    {
      "epoch": 6.423327895595432,
      "grad_norm": 0.08726153522729874,
      "learning_rate": 1.708257305623838e-05,
      "loss": 0.0053,
      "num_input_tokens_seen": 85004144,
      "step": 39375
    },
    {
      "epoch": 6.424143556280587,
      "grad_norm": 11.125823974609375,
      "learning_rate": 1.7075821780785766e-05,
      "loss": 0.1448,
      "num_input_tokens_seen": 85015408,
      "step": 39380
    },
    {
      "epoch": 6.424959216965743,
      "grad_norm": 0.11659998446702957,
      "learning_rate": 1.7069071147707248e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 85026064,
      "step": 39385
    },
    {
      "epoch": 6.425774877650897,
      "grad_norm": 0.13623273372650146,
      "learning_rate": 1.706232115755006e-05,
      "loss": 0.0875,
      "num_input_tokens_seen": 85037232,
      "step": 39390
    },
    {
      "epoch": 6.426590538336052,
      "grad_norm": 3.484802007675171,
      "learning_rate": 1.705557181086139e-05,
      "loss": 0.2018,
      "num_input_tokens_seen": 85047888,
      "step": 39395
    },
    {
      "epoch": 6.427406199021207,
      "grad_norm": 0.16145901381969452,
      "learning_rate": 1.704882310818839e-05,
      "loss": 0.1123,
      "num_input_tokens_seen": 85058768,
      "step": 39400
    },
    {
      "epoch": 6.428221859706362,
      "grad_norm": 2.734320878982544,
      "learning_rate": 1.704207505007813e-05,
      "loss": 0.0654,
      "num_input_tokens_seen": 85069296,
      "step": 39405
    },
    {
      "epoch": 6.4290375203915175,
      "grad_norm": 0.09999395906925201,
      "learning_rate": 1.703532763707764e-05,
      "loss": 0.1187,
      "num_input_tokens_seen": 85080528,
      "step": 39410
    },
    {
      "epoch": 6.429853181076672,
      "grad_norm": 0.06624927371740341,
      "learning_rate": 1.7028580869733905e-05,
      "loss": 0.075,
      "num_input_tokens_seen": 85091024,
      "step": 39415
    },
    {
      "epoch": 6.430668841761827,
      "grad_norm": 0.10834236443042755,
      "learning_rate": 1.702183474859385e-05,
      "loss": 0.0677,
      "num_input_tokens_seen": 85100656,
      "step": 39420
    },
    {
      "epoch": 6.431484502446982,
      "grad_norm": 0.18645232915878296,
      "learning_rate": 1.7015089274204354e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 85111856,
      "step": 39425
    },
    {
      "epoch": 6.432300163132137,
      "grad_norm": 2.205200433731079,
      "learning_rate": 1.7008344447112238e-05,
      "loss": 0.1318,
      "num_input_tokens_seen": 85120944,
      "step": 39430
    },
    {
      "epoch": 6.433115823817292,
      "grad_norm": 13.182847023010254,
      "learning_rate": 1.7001600267864266e-05,
      "loss": 0.0437,
      "num_input_tokens_seen": 85131440,
      "step": 39435
    },
    {
      "epoch": 6.433931484502447,
      "grad_norm": 0.16795465350151062,
      "learning_rate": 1.6994856737007154e-05,
      "loss": 0.0846,
      "num_input_tokens_seen": 85140720,
      "step": 39440
    },
    {
      "epoch": 6.434747145187602,
      "grad_norm": 1.0521131753921509,
      "learning_rate": 1.698811385508758e-05,
      "loss": 0.1491,
      "num_input_tokens_seen": 85151152,
      "step": 39445
    },
    {
      "epoch": 6.435562805872757,
      "grad_norm": 4.9170050621032715,
      "learning_rate": 1.698137162265215e-05,
      "loss": 0.084,
      "num_input_tokens_seen": 85161392,
      "step": 39450
    },
    {
      "epoch": 6.436378466557912,
      "grad_norm": 0.37777280807495117,
      "learning_rate": 1.6974630040247425e-05,
      "loss": 0.017,
      "num_input_tokens_seen": 85172656,
      "step": 39455
    },
    {
      "epoch": 6.437194127243067,
      "grad_norm": 0.09394480288028717,
      "learning_rate": 1.6967889108419903e-05,
      "loss": 0.0075,
      "num_input_tokens_seen": 85184784,
      "step": 39460
    },
    {
      "epoch": 6.438009787928221,
      "grad_norm": 0.31116095185279846,
      "learning_rate": 1.696114882771605e-05,
      "loss": 0.0886,
      "num_input_tokens_seen": 85195248,
      "step": 39465
    },
    {
      "epoch": 6.438825448613377,
      "grad_norm": 0.1480165719985962,
      "learning_rate": 1.695440919868226e-05,
      "loss": 0.0989,
      "num_input_tokens_seen": 85205040,
      "step": 39470
    },
    {
      "epoch": 6.439641109298532,
      "grad_norm": 0.08026082068681717,
      "learning_rate": 1.694767022186488e-05,
      "loss": 0.006,
      "num_input_tokens_seen": 85215344,
      "step": 39475
    },
    {
      "epoch": 6.440456769983687,
      "grad_norm": 0.1445658802986145,
      "learning_rate": 1.6940931897810208e-05,
      "loss": 0.0797,
      "num_input_tokens_seen": 85226352,
      "step": 39480
    },
    {
      "epoch": 6.441272430668842,
      "grad_norm": 0.275978684425354,
      "learning_rate": 1.693419422706449e-05,
      "loss": 0.0173,
      "num_input_tokens_seen": 85235408,
      "step": 39485
    },
    {
      "epoch": 6.442088091353996,
      "grad_norm": 0.07860295474529266,
      "learning_rate": 1.6927457210173915e-05,
      "loss": 0.2019,
      "num_input_tokens_seen": 85246352,
      "step": 39490
    },
    {
      "epoch": 6.442903752039152,
      "grad_norm": 0.24194453656673431,
      "learning_rate": 1.6920720847684617e-05,
      "loss": 0.1945,
      "num_input_tokens_seen": 85256368,
      "step": 39495
    },
    {
      "epoch": 6.443719412724307,
      "grad_norm": 0.029847651720046997,
      "learning_rate": 1.6913985140142682e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 85267856,
      "step": 39500
    },
    {
      "epoch": 6.444535073409462,
      "grad_norm": 0.25842732191085815,
      "learning_rate": 1.690725008809414e-05,
      "loss": 0.1997,
      "num_input_tokens_seen": 85280112,
      "step": 39505
    },
    {
      "epoch": 6.445350734094617,
      "grad_norm": 0.16432420909404755,
      "learning_rate": 1.6900515692084966e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 85288688,
      "step": 39510
    },
    {
      "epoch": 6.446166394779771,
      "grad_norm": 0.12695054709911346,
      "learning_rate": 1.689378195266109e-05,
      "loss": 0.1807,
      "num_input_tokens_seen": 85299888,
      "step": 39515
    },
    {
      "epoch": 6.446982055464926,
      "grad_norm": 0.10619836300611496,
      "learning_rate": 1.6887048870368377e-05,
      "loss": 0.0095,
      "num_input_tokens_seen": 85310128,
      "step": 39520
    },
    {
      "epoch": 6.447797716150082,
      "grad_norm": 0.16509254276752472,
      "learning_rate": 1.688031644575265e-05,
      "loss": 0.0054,
      "num_input_tokens_seen": 85321072,
      "step": 39525
    },
    {
      "epoch": 6.448613376835237,
      "grad_norm": 2.5598042011260986,
      "learning_rate": 1.6873584679359665e-05,
      "loss": 0.0671,
      "num_input_tokens_seen": 85333008,
      "step": 39530
    },
    {
      "epoch": 6.4494290375203915,
      "grad_norm": 0.24327965080738068,
      "learning_rate": 1.686685357173514e-05,
      "loss": 0.0348,
      "num_input_tokens_seen": 85344048,
      "step": 39535
    },
    {
      "epoch": 6.450244698205546,
      "grad_norm": 0.04776002839207649,
      "learning_rate": 1.6860123123424733e-05,
      "loss": 0.1062,
      "num_input_tokens_seen": 85353136,
      "step": 39540
    },
    {
      "epoch": 6.451060358890701,
      "grad_norm": 2.810372829437256,
      "learning_rate": 1.6853393334974044e-05,
      "loss": 0.088,
      "num_input_tokens_seen": 85364016,
      "step": 39545
    },
    {
      "epoch": 6.451876019575856,
      "grad_norm": 0.2053820639848709,
      "learning_rate": 1.684666420692863e-05,
      "loss": 0.0082,
      "num_input_tokens_seen": 85374928,
      "step": 39550
    },
    {
      "epoch": 6.452691680261012,
      "grad_norm": 0.31872835755348206,
      "learning_rate": 1.6839935739833986e-05,
      "loss": 0.0413,
      "num_input_tokens_seen": 85383888,
      "step": 39555
    },
    {
      "epoch": 6.4535073409461665,
      "grad_norm": 12.280909538269043,
      "learning_rate": 1.683320793423555e-05,
      "loss": 0.2488,
      "num_input_tokens_seen": 85394736,
      "step": 39560
    },
    {
      "epoch": 6.454323001631321,
      "grad_norm": 0.08581593632698059,
      "learning_rate": 1.6826480790678718e-05,
      "loss": 0.0832,
      "num_input_tokens_seen": 85405232,
      "step": 39565
    },
    {
      "epoch": 6.455138662316476,
      "grad_norm": 0.34644970297813416,
      "learning_rate": 1.681975430970883e-05,
      "loss": 0.0947,
      "num_input_tokens_seen": 85416176,
      "step": 39570
    },
    {
      "epoch": 6.455954323001631,
      "grad_norm": 1.6972507238388062,
      "learning_rate": 1.681302849187116e-05,
      "loss": 0.0154,
      "num_input_tokens_seen": 85426800,
      "step": 39575
    },
    {
      "epoch": 6.456769983686787,
      "grad_norm": 1.9096978902816772,
      "learning_rate": 1.6806303337710942e-05,
      "loss": 0.0099,
      "num_input_tokens_seen": 85437520,
      "step": 39580
    },
    {
      "epoch": 6.4575856443719415,
      "grad_norm": 0.1439582258462906,
      "learning_rate": 1.679957884777335e-05,
      "loss": 0.0729,
      "num_input_tokens_seen": 85447888,
      "step": 39585
    },
    {
      "epoch": 6.458401305057096,
      "grad_norm": 0.03080224059522152,
      "learning_rate": 1.6792855022603508e-05,
      "loss": 0.0125,
      "num_input_tokens_seen": 85459344,
      "step": 39590
    },
    {
      "epoch": 6.459216965742251,
      "grad_norm": 0.29059574007987976,
      "learning_rate": 1.678613186274648e-05,
      "loss": 0.1,
      "num_input_tokens_seen": 85469744,
      "step": 39595
    },
    {
      "epoch": 6.460032626427406,
      "grad_norm": 0.1453476995229721,
      "learning_rate": 1.6779409368747274e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 85481168,
      "step": 39600
    },
    {
      "epoch": 6.460848287112561,
      "grad_norm": 0.05045260116457939,
      "learning_rate": 1.677268754115086e-05,
      "loss": 0.0118,
      "num_input_tokens_seen": 85492976,
      "step": 39605
    },
    {
      "epoch": 6.4616639477977165,
      "grad_norm": 0.7699715495109558,
      "learning_rate": 1.676596638050214e-05,
      "loss": 0.0819,
      "num_input_tokens_seen": 85502896,
      "step": 39610
    },
    {
      "epoch": 6.462479608482871,
      "grad_norm": 3.924278736114502,
      "learning_rate": 1.6759245887345966e-05,
      "loss": 0.3947,
      "num_input_tokens_seen": 85513680,
      "step": 39615
    },
    {
      "epoch": 6.463295269168026,
      "grad_norm": 0.11376676708459854,
      "learning_rate": 1.6752526062227127e-05,
      "loss": 0.1543,
      "num_input_tokens_seen": 85524784,
      "step": 39620
    },
    {
      "epoch": 6.464110929853181,
      "grad_norm": 0.5455036759376526,
      "learning_rate": 1.674580690569037e-05,
      "loss": 0.1193,
      "num_input_tokens_seen": 85536208,
      "step": 39625
    },
    {
      "epoch": 6.464926590538336,
      "grad_norm": 0.059227459132671356,
      "learning_rate": 1.6739088418280395e-05,
      "loss": 0.0086,
      "num_input_tokens_seen": 85545936,
      "step": 39630
    },
    {
      "epoch": 6.465742251223491,
      "grad_norm": 0.07577138394117355,
      "learning_rate": 1.6732370600541823e-05,
      "loss": 0.0985,
      "num_input_tokens_seen": 85556080,
      "step": 39635
    },
    {
      "epoch": 6.466557911908646,
      "grad_norm": 5.8748779296875,
      "learning_rate": 1.6725653453019244e-05,
      "loss": 0.2306,
      "num_input_tokens_seen": 85565840,
      "step": 39640
    },
    {
      "epoch": 6.467373572593801,
      "grad_norm": 0.06989464908838272,
      "learning_rate": 1.6718936976257177e-05,
      "loss": 0.1748,
      "num_input_tokens_seen": 85576912,
      "step": 39645
    },
    {
      "epoch": 6.468189233278956,
      "grad_norm": 0.04279068857431412,
      "learning_rate": 1.6712221170800087e-05,
      "loss": 0.0391,
      "num_input_tokens_seen": 85587760,
      "step": 39650
    },
    {
      "epoch": 6.469004893964111,
      "grad_norm": 0.025165436789393425,
      "learning_rate": 1.670550603719241e-05,
      "loss": 0.018,
      "num_input_tokens_seen": 85597552,
      "step": 39655
    },
    {
      "epoch": 6.4698205546492655,
      "grad_norm": 3.684547185897827,
      "learning_rate": 1.66987915759785e-05,
      "loss": 0.0706,
      "num_input_tokens_seen": 85608368,
      "step": 39660
    },
    {
      "epoch": 6.470636215334421,
      "grad_norm": 0.04838266968727112,
      "learning_rate": 1.6692077787702666e-05,
      "loss": 0.0086,
      "num_input_tokens_seen": 85618096,
      "step": 39665
    },
    {
      "epoch": 6.471451876019576,
      "grad_norm": 0.11319730430841446,
      "learning_rate": 1.6685364672909163e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 85628656,
      "step": 39670
    },
    {
      "epoch": 6.472267536704731,
      "grad_norm": 0.1918221116065979,
      "learning_rate": 1.6678652232142185e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 85639632,
      "step": 39675
    },
    {
      "epoch": 6.473083197389886,
      "grad_norm": 0.0761067196726799,
      "learning_rate": 1.667194046594588e-05,
      "loss": 0.1456,
      "num_input_tokens_seen": 85650832,
      "step": 39680
    },
    {
      "epoch": 6.4738988580750405,
      "grad_norm": 0.16649939119815826,
      "learning_rate": 1.666522937486433e-05,
      "loss": 0.1,
      "num_input_tokens_seen": 85661296,
      "step": 39685
    },
    {
      "epoch": 6.474714518760196,
      "grad_norm": 0.25331294536590576,
      "learning_rate": 1.6658518959441584e-05,
      "loss": 0.0984,
      "num_input_tokens_seen": 85672368,
      "step": 39690
    },
    {
      "epoch": 6.475530179445351,
      "grad_norm": 0.01694580353796482,
      "learning_rate": 1.6651809220221614e-05,
      "loss": 0.0746,
      "num_input_tokens_seen": 85682160,
      "step": 39695
    },
    {
      "epoch": 6.476345840130506,
      "grad_norm": 13.475996017456055,
      "learning_rate": 1.664510015774835e-05,
      "loss": 0.0439,
      "num_input_tokens_seen": 85692656,
      "step": 39700
    },
    {
      "epoch": 6.477161500815661,
      "grad_norm": 29.608610153198242,
      "learning_rate": 1.6638391772565658e-05,
      "loss": 0.018,
      "num_input_tokens_seen": 85703600,
      "step": 39705
    },
    {
      "epoch": 6.4779771615008155,
      "grad_norm": 7.677433013916016,
      "learning_rate": 1.6631684065217344e-05,
      "loss": 0.1436,
      "num_input_tokens_seen": 85714512,
      "step": 39710
    },
    {
      "epoch": 6.47879282218597,
      "grad_norm": 0.29250794649124146,
      "learning_rate": 1.662497703624719e-05,
      "loss": 0.155,
      "num_input_tokens_seen": 85723888,
      "step": 39715
    },
    {
      "epoch": 6.479608482871126,
      "grad_norm": 0.07598171383142471,
      "learning_rate": 1.6618270686198895e-05,
      "loss": 0.0032,
      "num_input_tokens_seen": 85734992,
      "step": 39720
    },
    {
      "epoch": 6.480424143556281,
      "grad_norm": 0.2042560577392578,
      "learning_rate": 1.6611565015616106e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 85746768,
      "step": 39725
    },
    {
      "epoch": 6.481239804241436,
      "grad_norm": 0.14365331828594208,
      "learning_rate": 1.6604860025042412e-05,
      "loss": 0.0052,
      "num_input_tokens_seen": 85758352,
      "step": 39730
    },
    {
      "epoch": 6.4820554649265905,
      "grad_norm": 0.04808463901281357,
      "learning_rate": 1.6598155715021368e-05,
      "loss": 0.0027,
      "num_input_tokens_seen": 85769424,
      "step": 39735
    },
    {
      "epoch": 6.482871125611745,
      "grad_norm": 0.14188307523727417,
      "learning_rate": 1.6591452086096448e-05,
      "loss": 0.077,
      "num_input_tokens_seen": 85780816,
      "step": 39740
    },
    {
      "epoch": 6.4836867862969,
      "grad_norm": 0.06382682919502258,
      "learning_rate": 1.658474913881109e-05,
      "loss": 0.1917,
      "num_input_tokens_seen": 85792528,
      "step": 39745
    },
    {
      "epoch": 6.484502446982056,
      "grad_norm": 0.11008022725582123,
      "learning_rate": 1.6578046873708663e-05,
      "loss": 0.0052,
      "num_input_tokens_seen": 85802448,
      "step": 39750
    },
    {
      "epoch": 6.485318107667211,
      "grad_norm": 0.25377973914146423,
      "learning_rate": 1.657134529133248e-05,
      "loss": 0.0046,
      "num_input_tokens_seen": 85813104,
      "step": 39755
    },
    {
      "epoch": 6.486133768352365,
      "grad_norm": 0.17505641281604767,
      "learning_rate": 1.6564644392225824e-05,
      "loss": 0.1367,
      "num_input_tokens_seen": 85823248,
      "step": 39760
    },
    {
      "epoch": 6.48694942903752,
      "grad_norm": 0.04072978347539902,
      "learning_rate": 1.6557944176931894e-05,
      "loss": 0.2781,
      "num_input_tokens_seen": 85834384,
      "step": 39765
    },
    {
      "epoch": 6.487765089722675,
      "grad_norm": 0.038568198680877686,
      "learning_rate": 1.6551244645993847e-05,
      "loss": 0.0132,
      "num_input_tokens_seen": 85844848,
      "step": 39770
    },
    {
      "epoch": 6.488580750407831,
      "grad_norm": 2.7014260292053223,
      "learning_rate": 1.654454579995477e-05,
      "loss": 0.2609,
      "num_input_tokens_seen": 85856528,
      "step": 39775
    },
    {
      "epoch": 6.489396411092986,
      "grad_norm": 0.07561331242322922,
      "learning_rate": 1.653784763935772e-05,
      "loss": 0.0027,
      "num_input_tokens_seen": 85868112,
      "step": 39780
    },
    {
      "epoch": 6.49021207177814,
      "grad_norm": 0.1559215486049652,
      "learning_rate": 1.6531150164745674e-05,
      "loss": 0.1453,
      "num_input_tokens_seen": 85878256,
      "step": 39785
    },
    {
      "epoch": 6.491027732463295,
      "grad_norm": 0.0716010108590126,
      "learning_rate": 1.6524453376661568e-05,
      "loss": 0.0033,
      "num_input_tokens_seen": 85887440,
      "step": 39790
    },
    {
      "epoch": 6.49184339314845,
      "grad_norm": 28.87321662902832,
      "learning_rate": 1.6517757275648267e-05,
      "loss": 0.0391,
      "num_input_tokens_seen": 85897872,
      "step": 39795
    },
    {
      "epoch": 6.492659053833605,
      "grad_norm": 0.023398280143737793,
      "learning_rate": 1.6511061862248605e-05,
      "loss": 0.1002,
      "num_input_tokens_seen": 85909776,
      "step": 39800
    },
    {
      "epoch": 6.493474714518761,
      "grad_norm": 3.4813265800476074,
      "learning_rate": 1.6504367137005344e-05,
      "loss": 0.2511,
      "num_input_tokens_seen": 85919984,
      "step": 39805
    },
    {
      "epoch": 6.494290375203915,
      "grad_norm": 0.14940987527370453,
      "learning_rate": 1.649767310046119e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 85931632,
      "step": 39810
    },
    {
      "epoch": 6.49510603588907,
      "grad_norm": 0.2095457911491394,
      "learning_rate": 1.649097975315879e-05,
      "loss": 0.0763,
      "num_input_tokens_seen": 85944176,
      "step": 39815
    },
    {
      "epoch": 6.495921696574225,
      "grad_norm": 0.09960245341062546,
      "learning_rate": 1.648428709564075e-05,
      "loss": 0.0039,
      "num_input_tokens_seen": 85956016,
      "step": 39820
    },
    {
      "epoch": 6.49673735725938,
      "grad_norm": 0.10123313218355179,
      "learning_rate": 1.6477595128449605e-05,
      "loss": 0.0062,
      "num_input_tokens_seen": 85966704,
      "step": 39825
    },
    {
      "epoch": 6.497553017944535,
      "grad_norm": 0.029882868751883507,
      "learning_rate": 1.647090385212784e-05,
      "loss": 0.2219,
      "num_input_tokens_seen": 85976048,
      "step": 39830
    },
    {
      "epoch": 6.49836867862969,
      "grad_norm": 0.04088251665234566,
      "learning_rate": 1.6464213267217888e-05,
      "loss": 0.1234,
      "num_input_tokens_seen": 85985424,
      "step": 39835
    },
    {
      "epoch": 6.499184339314845,
      "grad_norm": 0.08394760638475418,
      "learning_rate": 1.6457523374262117e-05,
      "loss": 0.1349,
      "num_input_tokens_seen": 85996976,
      "step": 39840
    },
    {
      "epoch": 6.5,
      "grad_norm": 0.040804579854011536,
      "learning_rate": 1.645083417380284e-05,
      "loss": 0.0772,
      "num_input_tokens_seen": 86008784,
      "step": 39845
    },
    {
      "epoch": 6.5,
      "eval_loss": 0.19124996662139893,
      "eval_runtime": 132.9357,
      "eval_samples_per_second": 20.499,
      "eval_steps_per_second": 5.13,
      "num_input_tokens_seen": 86008784,
      "step": 39845
    },
    {
      "epoch": 6.500815660685155,
      "grad_norm": 6.5842061042785645,
      "learning_rate": 1.644414566638233e-05,
      "loss": 0.1526,
      "num_input_tokens_seen": 86019568,
      "step": 39850
    },
    {
      "epoch": 6.50163132137031,
      "grad_norm": 0.17074818909168243,
      "learning_rate": 1.643745785254278e-05,
      "loss": 0.019,
      "num_input_tokens_seen": 86030448,
      "step": 39855
    },
    {
      "epoch": 6.502446982055465,
      "grad_norm": 20.010343551635742,
      "learning_rate": 1.6430770732826346e-05,
      "loss": 0.1846,
      "num_input_tokens_seen": 86041040,
      "step": 39860
    },
    {
      "epoch": 6.50326264274062,
      "grad_norm": 0.07051025331020355,
      "learning_rate": 1.6424084307775107e-05,
      "loss": 0.0026,
      "num_input_tokens_seen": 86052624,
      "step": 39865
    },
    {
      "epoch": 6.504078303425775,
      "grad_norm": 0.05950159206986427,
      "learning_rate": 1.6417398577931116e-05,
      "loss": 0.1023,
      "num_input_tokens_seen": 86063472,
      "step": 39870
    },
    {
      "epoch": 6.50489396411093,
      "grad_norm": 0.0516861230134964,
      "learning_rate": 1.6410713543836342e-05,
      "loss": 0.0062,
      "num_input_tokens_seen": 86074320,
      "step": 39875
    },
    {
      "epoch": 6.505709624796085,
      "grad_norm": 0.06555325537919998,
      "learning_rate": 1.6404029206032708e-05,
      "loss": 0.0344,
      "num_input_tokens_seen": 86086320,
      "step": 39880
    },
    {
      "epoch": 6.506525285481239,
      "grad_norm": 0.04884251207113266,
      "learning_rate": 1.6397345565062082e-05,
      "loss": 0.1264,
      "num_input_tokens_seen": 86098000,
      "step": 39885
    },
    {
      "epoch": 6.507340946166395,
      "grad_norm": 0.09672345966100693,
      "learning_rate": 1.639066262146628e-05,
      "loss": 0.0865,
      "num_input_tokens_seen": 86107088,
      "step": 39890
    },
    {
      "epoch": 6.50815660685155,
      "grad_norm": 0.01781996712088585,
      "learning_rate": 1.6383980375787044e-05,
      "loss": 0.1243,
      "num_input_tokens_seen": 86118096,
      "step": 39895
    },
    {
      "epoch": 6.508972267536705,
      "grad_norm": 20.226181030273438,
      "learning_rate": 1.637729882856608e-05,
      "loss": 0.1434,
      "num_input_tokens_seen": 86129136,
      "step": 39900
    },
    {
      "epoch": 6.50978792822186,
      "grad_norm": 3.394411325454712,
      "learning_rate": 1.6370617980345022e-05,
      "loss": 0.1982,
      "num_input_tokens_seen": 86140816,
      "step": 39905
    },
    {
      "epoch": 6.510603588907014,
      "grad_norm": 0.07280711084604263,
      "learning_rate": 1.6363937831665458e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 86151664,
      "step": 39910
    },
    {
      "epoch": 6.511419249592169,
      "grad_norm": 0.14743319153785706,
      "learning_rate": 1.635725838306891e-05,
      "loss": 0.0221,
      "num_input_tokens_seen": 86161744,
      "step": 39915
    },
    {
      "epoch": 6.512234910277325,
      "grad_norm": 3.7788338661193848,
      "learning_rate": 1.6350579635096852e-05,
      "loss": 0.0837,
      "num_input_tokens_seen": 86171312,
      "step": 39920
    },
    {
      "epoch": 6.51305057096248,
      "grad_norm": 0.04754680395126343,
      "learning_rate": 1.6343901588290695e-05,
      "loss": 0.0461,
      "num_input_tokens_seen": 86181392,
      "step": 39925
    },
    {
      "epoch": 6.513866231647635,
      "grad_norm": 0.19998221099376678,
      "learning_rate": 1.633722424319179e-05,
      "loss": 0.0321,
      "num_input_tokens_seen": 86191728,
      "step": 39930
    },
    {
      "epoch": 6.514681892332789,
      "grad_norm": 0.0944763571023941,
      "learning_rate": 1.633054760034145e-05,
      "loss": 0.0039,
      "num_input_tokens_seen": 86201712,
      "step": 39935
    },
    {
      "epoch": 6.515497553017944,
      "grad_norm": 0.1905948519706726,
      "learning_rate": 1.6323871660280904e-05,
      "loss": 0.3161,
      "num_input_tokens_seen": 86212464,
      "step": 39940
    },
    {
      "epoch": 6.5163132137031,
      "grad_norm": 18.064556121826172,
      "learning_rate": 1.6317196423551347e-05,
      "loss": 0.1353,
      "num_input_tokens_seen": 86224208,
      "step": 39945
    },
    {
      "epoch": 6.517128874388255,
      "grad_norm": 12.106965065002441,
      "learning_rate": 1.6310521890693904e-05,
      "loss": 0.0226,
      "num_input_tokens_seen": 86234544,
      "step": 39950
    },
    {
      "epoch": 6.5179445350734095,
      "grad_norm": 2.743065595626831,
      "learning_rate": 1.6303848062249643e-05,
      "loss": 0.1557,
      "num_input_tokens_seen": 86245776,
      "step": 39955
    },
    {
      "epoch": 6.518760195758564,
      "grad_norm": 0.08378918468952179,
      "learning_rate": 1.6297174938759584e-05,
      "loss": 0.1994,
      "num_input_tokens_seen": 86256976,
      "step": 39960
    },
    {
      "epoch": 6.519575856443719,
      "grad_norm": 0.12126666307449341,
      "learning_rate": 1.6290502520764687e-05,
      "loss": 0.2295,
      "num_input_tokens_seen": 86267568,
      "step": 39965
    },
    {
      "epoch": 6.520391517128875,
      "grad_norm": 0.15529920160770416,
      "learning_rate": 1.6283830808805832e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 86278768,
      "step": 39970
    },
    {
      "epoch": 6.52120717781403,
      "grad_norm": 0.20121616125106812,
      "learning_rate": 1.6277159803423888e-05,
      "loss": 0.0369,
      "num_input_tokens_seen": 86288080,
      "step": 39975
    },
    {
      "epoch": 6.5220228384991845,
      "grad_norm": 0.12498260289430618,
      "learning_rate": 1.627048950515963e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 86299120,
      "step": 39980
    },
    {
      "epoch": 6.522838499184339,
      "grad_norm": 0.11112265288829803,
      "learning_rate": 1.6263819914553786e-05,
      "loss": 0.0152,
      "num_input_tokens_seen": 86311056,
      "step": 39985
    },
    {
      "epoch": 6.523654159869494,
      "grad_norm": 10.087481498718262,
      "learning_rate": 1.6257151032147028e-05,
      "loss": 0.1839,
      "num_input_tokens_seen": 86321904,
      "step": 39990
    },
    {
      "epoch": 6.524469820554649,
      "grad_norm": 0.02778514288365841,
      "learning_rate": 1.6250482858479964e-05,
      "loss": 0.1932,
      "num_input_tokens_seen": 86332880,
      "step": 39995
    },
    {
      "epoch": 6.525285481239804,
      "grad_norm": 0.33237969875335693,
      "learning_rate": 1.624381539409316e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 86344080,
      "step": 40000
    },
    {
      "epoch": 6.5261011419249595,
      "grad_norm": 0.25928637385368347,
      "learning_rate": 1.6237148639527106e-05,
      "loss": 0.0295,
      "num_input_tokens_seen": 86356240,
      "step": 40005
    },
    {
      "epoch": 6.526916802610114,
      "grad_norm": 6.4005537033081055,
      "learning_rate": 1.6230482595322244e-05,
      "loss": 0.0609,
      "num_input_tokens_seen": 86367824,
      "step": 40010
    },
    {
      "epoch": 6.527732463295269,
      "grad_norm": 0.1521807760000229,
      "learning_rate": 1.6223817262018958e-05,
      "loss": 0.2179,
      "num_input_tokens_seen": 86379632,
      "step": 40015
    },
    {
      "epoch": 6.528548123980424,
      "grad_norm": 0.05539650097489357,
      "learning_rate": 1.6217152640157577e-05,
      "loss": 0.0226,
      "num_input_tokens_seen": 86391248,
      "step": 40020
    },
    {
      "epoch": 6.529363784665579,
      "grad_norm": 0.11192484945058823,
      "learning_rate": 1.621048873027836e-05,
      "loss": 0.1208,
      "num_input_tokens_seen": 86401104,
      "step": 40025
    },
    {
      "epoch": 6.5301794453507345,
      "grad_norm": 3.5296645164489746,
      "learning_rate": 1.6203825532921533e-05,
      "loss": 0.2194,
      "num_input_tokens_seen": 86412528,
      "step": 40030
    },
    {
      "epoch": 6.530995106035889,
      "grad_norm": 0.1914474219083786,
      "learning_rate": 1.6197163048627237e-05,
      "loss": 0.0661,
      "num_input_tokens_seen": 86422800,
      "step": 40035
    },
    {
      "epoch": 6.531810766721044,
      "grad_norm": 0.0820464938879013,
      "learning_rate": 1.619050127793557e-05,
      "loss": 0.0039,
      "num_input_tokens_seen": 86434448,
      "step": 40040
    },
    {
      "epoch": 6.532626427406199,
      "grad_norm": 0.08316051214933395,
      "learning_rate": 1.6183840221386567e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 86447312,
      "step": 40045
    },
    {
      "epoch": 6.533442088091354,
      "grad_norm": 7.715195655822754,
      "learning_rate": 1.617717987952021e-05,
      "loss": 0.0774,
      "num_input_tokens_seen": 86458800,
      "step": 40050
    },
    {
      "epoch": 6.5342577487765094,
      "grad_norm": 2.059783697128296,
      "learning_rate": 1.6170520252876416e-05,
      "loss": 0.1462,
      "num_input_tokens_seen": 86470928,
      "step": 40055
    },
    {
      "epoch": 6.535073409461664,
      "grad_norm": 0.06226469203829765,
      "learning_rate": 1.616386134199505e-05,
      "loss": 0.095,
      "num_input_tokens_seen": 86481520,
      "step": 40060
    },
    {
      "epoch": 6.535889070146819,
      "grad_norm": 0.19757278263568878,
      "learning_rate": 1.6157203147415923e-05,
      "loss": 0.0083,
      "num_input_tokens_seen": 86492080,
      "step": 40065
    },
    {
      "epoch": 6.536704730831974,
      "grad_norm": 0.0957663506269455,
      "learning_rate": 1.6150545669678773e-05,
      "loss": 0.0601,
      "num_input_tokens_seen": 86503280,
      "step": 40070
    },
    {
      "epoch": 6.537520391517129,
      "grad_norm": 14.515009880065918,
      "learning_rate": 1.6143888909323286e-05,
      "loss": 0.1451,
      "num_input_tokens_seen": 86513808,
      "step": 40075
    },
    {
      "epoch": 6.5383360522022835,
      "grad_norm": 0.060574427247047424,
      "learning_rate": 1.6137232866889107e-05,
      "loss": 0.0187,
      "num_input_tokens_seen": 86524944,
      "step": 40080
    },
    {
      "epoch": 6.539151712887438,
      "grad_norm": 0.021970512345433235,
      "learning_rate": 1.6130577542915798e-05,
      "loss": 0.2364,
      "num_input_tokens_seen": 86535824,
      "step": 40085
    },
    {
      "epoch": 6.539967373572594,
      "grad_norm": 8.479681015014648,
      "learning_rate": 1.6123922937942883e-05,
      "loss": 0.1004,
      "num_input_tokens_seen": 86546992,
      "step": 40090
    },
    {
      "epoch": 6.540783034257749,
      "grad_norm": 0.028135670349001884,
      "learning_rate": 1.6117269052509803e-05,
      "loss": 0.0025,
      "num_input_tokens_seen": 86558352,
      "step": 40095
    },
    {
      "epoch": 6.541598694942904,
      "grad_norm": 0.06854220479726791,
      "learning_rate": 1.6110615887155972e-05,
      "loss": 0.0821,
      "num_input_tokens_seen": 86569424,
      "step": 40100
    },
    {
      "epoch": 6.5424143556280585,
      "grad_norm": 0.35423922538757324,
      "learning_rate": 1.6103963442420717e-05,
      "loss": 0.1701,
      "num_input_tokens_seen": 86579024,
      "step": 40105
    },
    {
      "epoch": 6.543230016313213,
      "grad_norm": 0.07758591324090958,
      "learning_rate": 1.6097311718843322e-05,
      "loss": 0.1315,
      "num_input_tokens_seen": 86590832,
      "step": 40110
    },
    {
      "epoch": 6.544045676998369,
      "grad_norm": 7.269680976867676,
      "learning_rate": 1.6090660716963014e-05,
      "loss": 0.0177,
      "num_input_tokens_seen": 86602736,
      "step": 40115
    },
    {
      "epoch": 6.544861337683524,
      "grad_norm": 4.9752373695373535,
      "learning_rate": 1.608401043731895e-05,
      "loss": 0.0565,
      "num_input_tokens_seen": 86613712,
      "step": 40120
    },
    {
      "epoch": 6.545676998368679,
      "grad_norm": 0.049367327243089676,
      "learning_rate": 1.6077360880450244e-05,
      "loss": 0.2344,
      "num_input_tokens_seen": 86624112,
      "step": 40125
    },
    {
      "epoch": 6.5464926590538335,
      "grad_norm": 0.042861804366111755,
      "learning_rate": 1.6070712046895936e-05,
      "loss": 0.0039,
      "num_input_tokens_seen": 86634000,
      "step": 40130
    },
    {
      "epoch": 6.547308319738988,
      "grad_norm": 28.21422004699707,
      "learning_rate": 1.6064063937195017e-05,
      "loss": 0.2787,
      "num_input_tokens_seen": 86644816,
      "step": 40135
    },
    {
      "epoch": 6.548123980424144,
      "grad_norm": 17.396093368530273,
      "learning_rate": 1.6057416551886418e-05,
      "loss": 0.2133,
      "num_input_tokens_seen": 86656496,
      "step": 40140
    },
    {
      "epoch": 6.548939641109299,
      "grad_norm": 0.055646199733018875,
      "learning_rate": 1.6050769891509005e-05,
      "loss": 0.1225,
      "num_input_tokens_seen": 86668400,
      "step": 40145
    },
    {
      "epoch": 6.549755301794454,
      "grad_norm": 0.18408013880252838,
      "learning_rate": 1.6044123956601593e-05,
      "loss": 0.0061,
      "num_input_tokens_seen": 86678832,
      "step": 40150
    },
    {
      "epoch": 6.5505709624796085,
      "grad_norm": 5.477224349975586,
      "learning_rate": 1.6037478747702932e-05,
      "loss": 0.0607,
      "num_input_tokens_seen": 86689040,
      "step": 40155
    },
    {
      "epoch": 6.551386623164763,
      "grad_norm": 0.05760050565004349,
      "learning_rate": 1.6030834265351724e-05,
      "loss": 0.1876,
      "num_input_tokens_seen": 86699952,
      "step": 40160
    },
    {
      "epoch": 6.552202283849918,
      "grad_norm": 7.261889934539795,
      "learning_rate": 1.602419051008659e-05,
      "loss": 0.0195,
      "num_input_tokens_seen": 86710320,
      "step": 40165
    },
    {
      "epoch": 6.553017944535073,
      "grad_norm": 0.09678348898887634,
      "learning_rate": 1.6017547482446127e-05,
      "loss": 0.0087,
      "num_input_tokens_seen": 86721648,
      "step": 40170
    },
    {
      "epoch": 6.553833605220229,
      "grad_norm": 0.29837679862976074,
      "learning_rate": 1.6010905182968837e-05,
      "loss": 0.0772,
      "num_input_tokens_seen": 86732560,
      "step": 40175
    },
    {
      "epoch": 6.554649265905383,
      "grad_norm": 0.06096908822655678,
      "learning_rate": 1.6004263612193182e-05,
      "loss": 0.0037,
      "num_input_tokens_seen": 86743408,
      "step": 40180
    },
    {
      "epoch": 6.555464926590538,
      "grad_norm": 0.37986379861831665,
      "learning_rate": 1.599762277065756e-05,
      "loss": 0.1418,
      "num_input_tokens_seen": 86754384,
      "step": 40185
    },
    {
      "epoch": 6.556280587275693,
      "grad_norm": 0.5057947635650635,
      "learning_rate": 1.599098265890031e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 86764592,
      "step": 40190
    },
    {
      "epoch": 6.557096247960848,
      "grad_norm": 0.09633487462997437,
      "learning_rate": 1.598434327745973e-05,
      "loss": 0.0865,
      "num_input_tokens_seen": 86776176,
      "step": 40195
    },
    {
      "epoch": 6.557911908646004,
      "grad_norm": 5.3133320808410645,
      "learning_rate": 1.5977704626874023e-05,
      "loss": 0.1039,
      "num_input_tokens_seen": 86786320,
      "step": 40200
    },
    {
      "epoch": 6.558727569331158,
      "grad_norm": 0.04846911132335663,
      "learning_rate": 1.597106670768136e-05,
      "loss": 0.0037,
      "num_input_tokens_seen": 86798288,
      "step": 40205
    },
    {
      "epoch": 6.559543230016313,
      "grad_norm": 0.03388461098074913,
      "learning_rate": 1.5964429520419836e-05,
      "loss": 0.1113,
      "num_input_tokens_seen": 86809456,
      "step": 40210
    },
    {
      "epoch": 6.560358890701468,
      "grad_norm": 0.18287904560565948,
      "learning_rate": 1.595779306562751e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 86819824,
      "step": 40215
    },
    {
      "epoch": 6.561174551386623,
      "grad_norm": 6.340177536010742,
      "learning_rate": 1.5951157343842352e-05,
      "loss": 0.1767,
      "num_input_tokens_seen": 86829872,
      "step": 40220
    },
    {
      "epoch": 6.561990212071779,
      "grad_norm": 0.07319547235965729,
      "learning_rate": 1.5944522355602297e-05,
      "loss": 0.1392,
      "num_input_tokens_seen": 86839472,
      "step": 40225
    },
    {
      "epoch": 6.562805872756933,
      "grad_norm": 0.2601003646850586,
      "learning_rate": 1.59378881014452e-05,
      "loss": 0.0076,
      "num_input_tokens_seen": 86851184,
      "step": 40230
    },
    {
      "epoch": 6.563621533442088,
      "grad_norm": 0.035068362951278687,
      "learning_rate": 1.5931254581908882e-05,
      "loss": 0.0967,
      "num_input_tokens_seen": 86861136,
      "step": 40235
    },
    {
      "epoch": 6.564437194127243,
      "grad_norm": 0.04618893191218376,
      "learning_rate": 1.592462179753108e-05,
      "loss": 0.1418,
      "num_input_tokens_seen": 86871344,
      "step": 40240
    },
    {
      "epoch": 6.565252854812398,
      "grad_norm": 3.044525146484375,
      "learning_rate": 1.591798974884948e-05,
      "loss": 0.0782,
      "num_input_tokens_seen": 86881488,
      "step": 40245
    },
    {
      "epoch": 6.566068515497553,
      "grad_norm": 0.13312385976314545,
      "learning_rate": 1.5911358436401708e-05,
      "loss": 0.1363,
      "num_input_tokens_seen": 86893264,
      "step": 40250
    },
    {
      "epoch": 6.566884176182708,
      "grad_norm": 0.05873178318142891,
      "learning_rate": 1.5904727860725344e-05,
      "loss": 0.0663,
      "num_input_tokens_seen": 86904208,
      "step": 40255
    },
    {
      "epoch": 6.567699836867863,
      "grad_norm": 0.6661403179168701,
      "learning_rate": 1.589809802235789e-05,
      "loss": 0.011,
      "num_input_tokens_seen": 86916080,
      "step": 40260
    },
    {
      "epoch": 6.568515497553018,
      "grad_norm": 0.07328290492296219,
      "learning_rate": 1.589146892183679e-05,
      "loss": 0.0072,
      "num_input_tokens_seen": 86926640,
      "step": 40265
    },
    {
      "epoch": 6.569331158238173,
      "grad_norm": 0.05128014460206032,
      "learning_rate": 1.5884840559699436e-05,
      "loss": 0.205,
      "num_input_tokens_seen": 86936688,
      "step": 40270
    },
    {
      "epoch": 6.570146818923328,
      "grad_norm": 0.050329506397247314,
      "learning_rate": 1.5878212936483156e-05,
      "loss": 0.0086,
      "num_input_tokens_seen": 86948272,
      "step": 40275
    },
    {
      "epoch": 6.5709624796084825,
      "grad_norm": 8.136697769165039,
      "learning_rate": 1.5871586052725216e-05,
      "loss": 0.1234,
      "num_input_tokens_seen": 86959184,
      "step": 40280
    },
    {
      "epoch": 6.571778140293638,
      "grad_norm": 0.24187760055065155,
      "learning_rate": 1.5864959908962832e-05,
      "loss": 0.0698,
      "num_input_tokens_seen": 86970032,
      "step": 40285
    },
    {
      "epoch": 6.572593800978793,
      "grad_norm": 0.05244762822985649,
      "learning_rate": 1.5858334505733137e-05,
      "loss": 0.0918,
      "num_input_tokens_seen": 86980048,
      "step": 40290
    },
    {
      "epoch": 6.573409461663948,
      "grad_norm": 0.05218328908085823,
      "learning_rate": 1.585170984357324e-05,
      "loss": 0.0058,
      "num_input_tokens_seen": 86991888,
      "step": 40295
    },
    {
      "epoch": 6.574225122349103,
      "grad_norm": 0.09422808885574341,
      "learning_rate": 1.5845085923020165e-05,
      "loss": 0.0041,
      "num_input_tokens_seen": 87003056,
      "step": 40300
    },
    {
      "epoch": 6.575040783034257,
      "grad_norm": 0.9793643951416016,
      "learning_rate": 1.5838462744610872e-05,
      "loss": 0.0053,
      "num_input_tokens_seen": 87014192,
      "step": 40305
    },
    {
      "epoch": 6.575856443719413,
      "grad_norm": 0.214778870344162,
      "learning_rate": 1.5831840308882276e-05,
      "loss": 0.1694,
      "num_input_tokens_seen": 87024112,
      "step": 40310
    },
    {
      "epoch": 6.576672104404568,
      "grad_norm": 0.04145102947950363,
      "learning_rate": 1.5825218616371224e-05,
      "loss": 0.088,
      "num_input_tokens_seen": 87034896,
      "step": 40315
    },
    {
      "epoch": 6.577487765089723,
      "grad_norm": 0.6659348607063293,
      "learning_rate": 1.5818597667614503e-05,
      "loss": 0.1042,
      "num_input_tokens_seen": 87045392,
      "step": 40320
    },
    {
      "epoch": 6.578303425774878,
      "grad_norm": 0.1729201376438141,
      "learning_rate": 1.581197746314884e-05,
      "loss": 0.0058,
      "num_input_tokens_seen": 87056432,
      "step": 40325
    },
    {
      "epoch": 6.579119086460032,
      "grad_norm": 0.08217202872037888,
      "learning_rate": 1.5805358003510902e-05,
      "loss": 0.0949,
      "num_input_tokens_seen": 87067440,
      "step": 40330
    },
    {
      "epoch": 6.579934747145187,
      "grad_norm": 0.03858206793665886,
      "learning_rate": 1.5798739289237298e-05,
      "loss": 0.1033,
      "num_input_tokens_seen": 87078064,
      "step": 40335
    },
    {
      "epoch": 6.580750407830343,
      "grad_norm": 0.11875259131193161,
      "learning_rate": 1.5792121320864573e-05,
      "loss": 0.0035,
      "num_input_tokens_seen": 87087920,
      "step": 40340
    },
    {
      "epoch": 6.581566068515498,
      "grad_norm": 0.06483682245016098,
      "learning_rate": 1.5785504098929217e-05,
      "loss": 0.004,
      "num_input_tokens_seen": 87098032,
      "step": 40345
    },
    {
      "epoch": 6.582381729200653,
      "grad_norm": 0.20756934583187103,
      "learning_rate": 1.5778887623967654e-05,
      "loss": 0.1881,
      "num_input_tokens_seen": 87108304,
      "step": 40350
    },
    {
      "epoch": 6.583197389885807,
      "grad_norm": 0.1414896547794342,
      "learning_rate": 1.5772271896516245e-05,
      "loss": 0.0131,
      "num_input_tokens_seen": 87119088,
      "step": 40355
    },
    {
      "epoch": 6.584013050570962,
      "grad_norm": 0.5052565336227417,
      "learning_rate": 1.57656569171113e-05,
      "loss": 0.215,
      "num_input_tokens_seen": 87130032,
      "step": 40360
    },
    {
      "epoch": 6.584828711256117,
      "grad_norm": 0.08235272020101547,
      "learning_rate": 1.5759042686289056e-05,
      "loss": 0.2778,
      "num_input_tokens_seen": 87141104,
      "step": 40365
    },
    {
      "epoch": 6.585644371941273,
      "grad_norm": 0.22911354899406433,
      "learning_rate": 1.5752429204585702e-05,
      "loss": 0.1331,
      "num_input_tokens_seen": 87151472,
      "step": 40370
    },
    {
      "epoch": 6.5864600326264275,
      "grad_norm": 5.025905609130859,
      "learning_rate": 1.5745816472537355e-05,
      "loss": 0.1019,
      "num_input_tokens_seen": 87161680,
      "step": 40375
    },
    {
      "epoch": 6.587275693311582,
      "grad_norm": 0.27062445878982544,
      "learning_rate": 1.5739204490680085e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 87172080,
      "step": 40380
    },
    {
      "epoch": 6.588091353996737,
      "grad_norm": 0.38178327679634094,
      "learning_rate": 1.5732593259549885e-05,
      "loss": 0.0073,
      "num_input_tokens_seen": 87181168,
      "step": 40385
    },
    {
      "epoch": 6.588907014681892,
      "grad_norm": 0.05766642466187477,
      "learning_rate": 1.57259827796827e-05,
      "loss": 0.0917,
      "num_input_tokens_seen": 87192144,
      "step": 40390
    },
    {
      "epoch": 6.589722675367048,
      "grad_norm": 0.1844758838415146,
      "learning_rate": 1.5719373051614393e-05,
      "loss": 0.1106,
      "num_input_tokens_seen": 87203344,
      "step": 40395
    },
    {
      "epoch": 6.5905383360522025,
      "grad_norm": 2.9884049892425537,
      "learning_rate": 1.571276407588081e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 87214416,
      "step": 40400
    },
    {
      "epoch": 6.591353996737357,
      "grad_norm": 0.058588188141584396,
      "learning_rate": 1.570615585301769e-05,
      "loss": 0.0818,
      "num_input_tokens_seen": 87225168,
      "step": 40405
    },
    {
      "epoch": 6.592169657422512,
      "grad_norm": 0.17275620996952057,
      "learning_rate": 1.5699548383560736e-05,
      "loss": 0.1647,
      "num_input_tokens_seen": 87237232,
      "step": 40410
    },
    {
      "epoch": 6.592985318107667,
      "grad_norm": 16.883686065673828,
      "learning_rate": 1.569294166804558e-05,
      "loss": 0.4044,
      "num_input_tokens_seen": 87247696,
      "step": 40415
    },
    {
      "epoch": 6.593800978792823,
      "grad_norm": 3.278045177459717,
      "learning_rate": 1.5686335707007794e-05,
      "loss": 0.2006,
      "num_input_tokens_seen": 87258448,
      "step": 40420
    },
    {
      "epoch": 6.5946166394779775,
      "grad_norm": 0.420439213514328,
      "learning_rate": 1.5679730500982892e-05,
      "loss": 0.0061,
      "num_input_tokens_seen": 87268880,
      "step": 40425
    },
    {
      "epoch": 6.595432300163132,
      "grad_norm": 2.7619595527648926,
      "learning_rate": 1.5673126050506327e-05,
      "loss": 0.0093,
      "num_input_tokens_seen": 87279568,
      "step": 40430
    },
    {
      "epoch": 6.596247960848287,
      "grad_norm": 0.09254773706197739,
      "learning_rate": 1.5666522356113488e-05,
      "loss": 0.1424,
      "num_input_tokens_seen": 87289776,
      "step": 40435
    },
    {
      "epoch": 6.597063621533442,
      "grad_norm": 0.029101338237524033,
      "learning_rate": 1.5659919418339707e-05,
      "loss": 0.1755,
      "num_input_tokens_seen": 87301488,
      "step": 40440
    },
    {
      "epoch": 6.597879282218597,
      "grad_norm": 0.014889135956764221,
      "learning_rate": 1.565331723772025e-05,
      "loss": 0.0076,
      "num_input_tokens_seen": 87312112,
      "step": 40445
    },
    {
      "epoch": 6.598694942903752,
      "grad_norm": 0.3790788948535919,
      "learning_rate": 1.5646715814790318e-05,
      "loss": 0.0089,
      "num_input_tokens_seen": 87322416,
      "step": 40450
    },
    {
      "epoch": 6.599510603588907,
      "grad_norm": 0.11865105479955673,
      "learning_rate": 1.5640115150085067e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 87333040,
      "step": 40455
    },
    {
      "epoch": 6.600326264274062,
      "grad_norm": 0.03850835561752319,
      "learning_rate": 1.5633515244139567e-05,
      "loss": 0.1136,
      "num_input_tokens_seen": 87344112,
      "step": 40460
    },
    {
      "epoch": 6.601141924959217,
      "grad_norm": 0.07840988039970398,
      "learning_rate": 1.562691609748885e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 87353680,
      "step": 40465
    },
    {
      "epoch": 6.601957585644372,
      "grad_norm": 0.13444779813289642,
      "learning_rate": 1.562031771066787e-05,
      "loss": 0.1358,
      "num_input_tokens_seen": 87363536,
      "step": 40470
    },
    {
      "epoch": 6.602773246329527,
      "grad_norm": 0.09518253803253174,
      "learning_rate": 1.561372008421153e-05,
      "loss": 0.0678,
      "num_input_tokens_seen": 87374992,
      "step": 40475
    },
    {
      "epoch": 6.603588907014682,
      "grad_norm": 0.01630476862192154,
      "learning_rate": 1.560712321865466e-05,
      "loss": 0.052,
      "num_input_tokens_seen": 87386448,
      "step": 40480
    },
    {
      "epoch": 6.604404567699837,
      "grad_norm": 1.866984248161316,
      "learning_rate": 1.5600527114532042e-05,
      "loss": 0.3241,
      "num_input_tokens_seen": 87395760,
      "step": 40485
    },
    {
      "epoch": 6.605220228384992,
      "grad_norm": 0.2067229002714157,
      "learning_rate": 1.5593931772378395e-05,
      "loss": 0.0071,
      "num_input_tokens_seen": 87407088,
      "step": 40490
    },
    {
      "epoch": 6.606035889070147,
      "grad_norm": 1.2671343088150024,
      "learning_rate": 1.5587337192728365e-05,
      "loss": 0.0068,
      "num_input_tokens_seen": 87418256,
      "step": 40495
    },
    {
      "epoch": 6.6068515497553015,
      "grad_norm": 0.12746548652648926,
      "learning_rate": 1.5580743376116536e-05,
      "loss": 0.0112,
      "num_input_tokens_seen": 87427984,
      "step": 40500
    },
    {
      "epoch": 6.607667210440457,
      "grad_norm": 0.029078975319862366,
      "learning_rate": 1.5574150323077432e-05,
      "loss": 0.1386,
      "num_input_tokens_seen": 87437424,
      "step": 40505
    },
    {
      "epoch": 6.608482871125612,
      "grad_norm": 0.23436488211154938,
      "learning_rate": 1.556755803414554e-05,
      "loss": 0.0287,
      "num_input_tokens_seen": 87447504,
      "step": 40510
    },
    {
      "epoch": 6.609298531810767,
      "grad_norm": 0.10258165746927261,
      "learning_rate": 1.5560966509855256e-05,
      "loss": 0.0888,
      "num_input_tokens_seen": 87457456,
      "step": 40515
    },
    {
      "epoch": 6.610114192495922,
      "grad_norm": 5.345180034637451,
      "learning_rate": 1.5554375750740917e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 87468464,
      "step": 40520
    },
    {
      "epoch": 6.6109298531810765,
      "grad_norm": 0.16994620859622955,
      "learning_rate": 1.554778575733681e-05,
      "loss": 0.0253,
      "num_input_tokens_seen": 87479504,
      "step": 40525
    },
    {
      "epoch": 6.611745513866231,
      "grad_norm": 7.721708297729492,
      "learning_rate": 1.5541196530177148e-05,
      "loss": 0.0619,
      "num_input_tokens_seen": 87489840,
      "step": 40530
    },
    {
      "epoch": 6.612561174551386,
      "grad_norm": 0.07135919481515884,
      "learning_rate": 1.5534608069796085e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 87501424,
      "step": 40535
    },
    {
      "epoch": 6.613376835236542,
      "grad_norm": 0.06724556535482407,
      "learning_rate": 1.5528020376727725e-05,
      "loss": 0.0055,
      "num_input_tokens_seen": 87512688,
      "step": 40540
    },
    {
      "epoch": 6.614192495921697,
      "grad_norm": 0.07323489338159561,
      "learning_rate": 1.5521433451506088e-05,
      "loss": 0.1014,
      "num_input_tokens_seen": 87523056,
      "step": 40545
    },
    {
      "epoch": 6.6150081566068515,
      "grad_norm": 0.1892024278640747,
      "learning_rate": 1.5514847294665152e-05,
      "loss": 0.0923,
      "num_input_tokens_seen": 87532784,
      "step": 40550
    },
    {
      "epoch": 6.615823817292006,
      "grad_norm": 0.2373567819595337,
      "learning_rate": 1.5508261906738824e-05,
      "loss": 0.0598,
      "num_input_tokens_seen": 87542224,
      "step": 40555
    },
    {
      "epoch": 6.616639477977161,
      "grad_norm": 19.97500991821289,
      "learning_rate": 1.5501677288260943e-05,
      "loss": 0.0474,
      "num_input_tokens_seen": 87554224,
      "step": 40560
    },
    {
      "epoch": 6.617455138662317,
      "grad_norm": 18.255889892578125,
      "learning_rate": 1.549509343976529e-05,
      "loss": 0.0801,
      "num_input_tokens_seen": 87564944,
      "step": 40565
    },
    {
      "epoch": 6.618270799347472,
      "grad_norm": 0.02366139553487301,
      "learning_rate": 1.5488510361785597e-05,
      "loss": 0.0125,
      "num_input_tokens_seen": 87575696,
      "step": 40570
    },
    {
      "epoch": 6.6190864600326265,
      "grad_norm": 0.08095688372850418,
      "learning_rate": 1.5481928054855512e-05,
      "loss": 0.0815,
      "num_input_tokens_seen": 87587216,
      "step": 40575
    },
    {
      "epoch": 6.619902120717781,
      "grad_norm": 0.04319370537996292,
      "learning_rate": 1.5475346519508637e-05,
      "loss": 0.093,
      "num_input_tokens_seen": 87597872,
      "step": 40580
    },
    {
      "epoch": 6.620717781402936,
      "grad_norm": 0.10578560084104538,
      "learning_rate": 1.5468765756278498e-05,
      "loss": 0.0729,
      "num_input_tokens_seen": 87608816,
      "step": 40585
    },
    {
      "epoch": 6.621533442088092,
      "grad_norm": 0.1031089797616005,
      "learning_rate": 1.5462185765698568e-05,
      "loss": 0.1104,
      "num_input_tokens_seen": 87619600,
      "step": 40590
    },
    {
      "epoch": 6.622349102773247,
      "grad_norm": 3.261246919631958,
      "learning_rate": 1.5455606548302253e-05,
      "loss": 0.283,
      "num_input_tokens_seen": 87629168,
      "step": 40595
    },
    {
      "epoch": 6.623164763458401,
      "grad_norm": 0.06577812135219574,
      "learning_rate": 1.5449028104622905e-05,
      "loss": 0.0057,
      "num_input_tokens_seen": 87640144,
      "step": 40600
    },
    {
      "epoch": 6.623980424143556,
      "grad_norm": 4.057028293609619,
      "learning_rate": 1.5442450435193795e-05,
      "loss": 0.0098,
      "num_input_tokens_seen": 87650128,
      "step": 40605
    },
    {
      "epoch": 6.624796084828711,
      "grad_norm": 0.23669201135635376,
      "learning_rate": 1.5435873540548135e-05,
      "loss": 0.1881,
      "num_input_tokens_seen": 87661264,
      "step": 40610
    },
    {
      "epoch": 6.625611745513866,
      "grad_norm": 0.04815550148487091,
      "learning_rate": 1.5429297421219107e-05,
      "loss": 0.0274,
      "num_input_tokens_seen": 87672464,
      "step": 40615
    },
    {
      "epoch": 6.626427406199021,
      "grad_norm": 0.16993646323680878,
      "learning_rate": 1.5422722077739794e-05,
      "loss": 0.1498,
      "num_input_tokens_seen": 87684400,
      "step": 40620
    },
    {
      "epoch": 6.627243066884176,
      "grad_norm": 0.2493388056755066,
      "learning_rate": 1.541614751064322e-05,
      "loss": 0.2131,
      "num_input_tokens_seen": 87695728,
      "step": 40625
    },
    {
      "epoch": 6.628058727569331,
      "grad_norm": 0.01710120402276516,
      "learning_rate": 1.5409573720462357e-05,
      "loss": 0.0041,
      "num_input_tokens_seen": 87705456,
      "step": 40630
    },
    {
      "epoch": 6.628874388254486,
      "grad_norm": 0.039348945021629333,
      "learning_rate": 1.540300070773011e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 87716816,
      "step": 40635
    },
    {
      "epoch": 6.629690048939641,
      "grad_norm": 0.020207500085234642,
      "learning_rate": 1.539642847297932e-05,
      "loss": 0.0528,
      "num_input_tokens_seen": 87727216,
      "step": 40640
    },
    {
      "epoch": 6.630505709624796,
      "grad_norm": 0.2506010830402374,
      "learning_rate": 1.5389857016742764e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 87737040,
      "step": 40645
    },
    {
      "epoch": 6.631321370309951,
      "grad_norm": 1.0845345258712769,
      "learning_rate": 1.538328633955316e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 87746928,
      "step": 40650
    },
    {
      "epoch": 6.632137030995106,
      "grad_norm": 0.03723951801657677,
      "learning_rate": 1.5376716441943162e-05,
      "loss": 0.0958,
      "num_input_tokens_seen": 87757424,
      "step": 40655
    },
    {
      "epoch": 6.632952691680261,
      "grad_norm": 12.476139068603516,
      "learning_rate": 1.5370147324445354e-05,
      "loss": 0.0884,
      "num_input_tokens_seen": 87768816,
      "step": 40660
    },
    {
      "epoch": 6.633768352365416,
      "grad_norm": 0.04074963927268982,
      "learning_rate": 1.536357898759227e-05,
      "loss": 0.1074,
      "num_input_tokens_seen": 87780912,
      "step": 40665
    },
    {
      "epoch": 6.634584013050571,
      "grad_norm": 4.669559001922607,
      "learning_rate": 1.535701143191637e-05,
      "loss": 0.3011,
      "num_input_tokens_seen": 87791920,
      "step": 40670
    },
    {
      "epoch": 6.635399673735726,
      "grad_norm": 0.04353715851902962,
      "learning_rate": 1.535044465795005e-05,
      "loss": 0.0058,
      "num_input_tokens_seen": 87802352,
      "step": 40675
    },
    {
      "epoch": 6.636215334420881,
      "grad_norm": 0.022636588662862778,
      "learning_rate": 1.534387866622564e-05,
      "loss": 0.0021,
      "num_input_tokens_seen": 87813488,
      "step": 40680
    },
    {
      "epoch": 6.637030995106036,
      "grad_norm": 0.1221766248345375,
      "learning_rate": 1.5337313457275428e-05,
      "loss": 0.3061,
      "num_input_tokens_seen": 87824016,
      "step": 40685
    },
    {
      "epoch": 6.637846655791191,
      "grad_norm": 0.10250678658485413,
      "learning_rate": 1.533074903163161e-05,
      "loss": 0.4608,
      "num_input_tokens_seen": 87835120,
      "step": 40690
    },
    {
      "epoch": 6.638662316476346,
      "grad_norm": 0.03980075195431709,
      "learning_rate": 1.5324185389826338e-05,
      "loss": 0.1061,
      "num_input_tokens_seen": 87846480,
      "step": 40695
    },
    {
      "epoch": 6.6394779771615005,
      "grad_norm": 0.10017578303813934,
      "learning_rate": 1.5317622532391694e-05,
      "loss": 0.0365,
      "num_input_tokens_seen": 87856304,
      "step": 40700
    },
    {
      "epoch": 6.640293637846656,
      "grad_norm": 4.295700550079346,
      "learning_rate": 1.53110604598597e-05,
      "loss": 0.1861,
      "num_input_tokens_seen": 87868080,
      "step": 40705
    },
    {
      "epoch": 6.641109298531811,
      "grad_norm": 0.11103116720914841,
      "learning_rate": 1.5304499172762293e-05,
      "loss": 0.0411,
      "num_input_tokens_seen": 87878416,
      "step": 40710
    },
    {
      "epoch": 6.641924959216966,
      "grad_norm": 0.033165328204631805,
      "learning_rate": 1.5297938671631386e-05,
      "loss": 0.1826,
      "num_input_tokens_seen": 87888720,
      "step": 40715
    },
    {
      "epoch": 6.642740619902121,
      "grad_norm": 0.18278607726097107,
      "learning_rate": 1.5291378956998793e-05,
      "loss": 0.0148,
      "num_input_tokens_seen": 87900528,
      "step": 40720
    },
    {
      "epoch": 6.643556280587275,
      "grad_norm": 2.620253562927246,
      "learning_rate": 1.528482002939629e-05,
      "loss": 0.2184,
      "num_input_tokens_seen": 87911984,
      "step": 40725
    },
    {
      "epoch": 6.64437194127243,
      "grad_norm": 2.582531690597534,
      "learning_rate": 1.5278261889355568e-05,
      "loss": 0.1257,
      "num_input_tokens_seen": 87922608,
      "step": 40730
    },
    {
      "epoch": 6.645187601957586,
      "grad_norm": 0.1558326929807663,
      "learning_rate": 1.527170453740826e-05,
      "loss": 0.1289,
      "num_input_tokens_seen": 87933936,
      "step": 40735
    },
    {
      "epoch": 6.646003262642741,
      "grad_norm": 0.0765334889292717,
      "learning_rate": 1.5265147974085947e-05,
      "loss": 0.0199,
      "num_input_tokens_seen": 87945488,
      "step": 40740
    },
    {
      "epoch": 6.646818923327896,
      "grad_norm": 0.06828638911247253,
      "learning_rate": 1.5258592199920135e-05,
      "loss": 0.1816,
      "num_input_tokens_seen": 87956112,
      "step": 40745
    },
    {
      "epoch": 6.64763458401305,
      "grad_norm": 3.1311378479003906,
      "learning_rate": 1.5252037215442266e-05,
      "loss": 0.1812,
      "num_input_tokens_seen": 87965584,
      "step": 40750
    },
    {
      "epoch": 6.648450244698205,
      "grad_norm": 0.07298579066991806,
      "learning_rate": 1.5245483021183722e-05,
      "loss": 0.0043,
      "num_input_tokens_seen": 87975056,
      "step": 40755
    },
    {
      "epoch": 6.649265905383361,
      "grad_norm": 0.019851306453347206,
      "learning_rate": 1.5238929617675817e-05,
      "loss": 0.0954,
      "num_input_tokens_seen": 87985712,
      "step": 40760
    },
    {
      "epoch": 6.650081566068516,
      "grad_norm": 4.504660129547119,
      "learning_rate": 1.5232377005449805e-05,
      "loss": 0.0127,
      "num_input_tokens_seen": 87996336,
      "step": 40765
    },
    {
      "epoch": 6.650897226753671,
      "grad_norm": 0.20256765186786652,
      "learning_rate": 1.5225825185036874e-05,
      "loss": 0.0573,
      "num_input_tokens_seen": 88007184,
      "step": 40770
    },
    {
      "epoch": 6.651712887438825,
      "grad_norm": 0.33704063296318054,
      "learning_rate": 1.5219274156968143e-05,
      "loss": 0.2346,
      "num_input_tokens_seen": 88017008,
      "step": 40775
    },
    {
      "epoch": 6.65252854812398,
      "grad_norm": 0.22632668912410736,
      "learning_rate": 1.521272392177468e-05,
      "loss": 0.0071,
      "num_input_tokens_seen": 88027760,
      "step": 40780
    },
    {
      "epoch": 6.653344208809135,
      "grad_norm": 16.721834182739258,
      "learning_rate": 1.5206174479987475e-05,
      "loss": 0.1451,
      "num_input_tokens_seen": 88039696,
      "step": 40785
    },
    {
      "epoch": 6.654159869494291,
      "grad_norm": 0.07452557235956192,
      "learning_rate": 1.5199625832137459e-05,
      "loss": 0.0636,
      "num_input_tokens_seen": 88050544,
      "step": 40790
    },
    {
      "epoch": 6.6549755301794455,
      "grad_norm": 0.050509676337242126,
      "learning_rate": 1.5193077978755499e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 88061840,
      "step": 40795
    },
    {
      "epoch": 6.6557911908646,
      "grad_norm": 0.6662347912788391,
      "learning_rate": 1.5186530920372399e-05,
      "loss": 0.0087,
      "num_input_tokens_seen": 88072816,
      "step": 40800
    },
    {
      "epoch": 6.656606851549755,
      "grad_norm": 0.13393627107143402,
      "learning_rate": 1.5179984657518895e-05,
      "loss": 0.1109,
      "num_input_tokens_seen": 88083536,
      "step": 40805
    },
    {
      "epoch": 6.65742251223491,
      "grad_norm": 13.53052806854248,
      "learning_rate": 1.5173439190725663e-05,
      "loss": 0.2261,
      "num_input_tokens_seen": 88094704,
      "step": 40810
    },
    {
      "epoch": 6.658238172920065,
      "grad_norm": 4.617764472961426,
      "learning_rate": 1.5166894520523305e-05,
      "loss": 0.1982,
      "num_input_tokens_seen": 88106288,
      "step": 40815
    },
    {
      "epoch": 6.6590538336052205,
      "grad_norm": 0.02458474598824978,
      "learning_rate": 1.5160350647442367e-05,
      "loss": 0.0167,
      "num_input_tokens_seen": 88114960,
      "step": 40820
    },
    {
      "epoch": 6.659869494290375,
      "grad_norm": 0.12827251851558685,
      "learning_rate": 1.5153807572013338e-05,
      "loss": 0.0916,
      "num_input_tokens_seen": 88124944,
      "step": 40825
    },
    {
      "epoch": 6.66068515497553,
      "grad_norm": 0.15690867602825165,
      "learning_rate": 1.5147265294766624e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 88135056,
      "step": 40830
    },
    {
      "epoch": 6.661500815660685,
      "grad_norm": 0.16413655877113342,
      "learning_rate": 1.5140723816232583e-05,
      "loss": 0.0103,
      "num_input_tokens_seen": 88146768,
      "step": 40835
    },
    {
      "epoch": 6.66231647634584,
      "grad_norm": 0.045289475470781326,
      "learning_rate": 1.5134183136941487e-05,
      "loss": 0.3974,
      "num_input_tokens_seen": 88157968,
      "step": 40840
    },
    {
      "epoch": 6.6631321370309955,
      "grad_norm": 0.02990245632827282,
      "learning_rate": 1.5127643257423572e-05,
      "loss": 0.0989,
      "num_input_tokens_seen": 88168912,
      "step": 40845
    },
    {
      "epoch": 6.66394779771615,
      "grad_norm": 0.5999934077262878,
      "learning_rate": 1.5121104178208984e-05,
      "loss": 0.0082,
      "num_input_tokens_seen": 88179632,
      "step": 40850
    },
    {
      "epoch": 6.664763458401305,
      "grad_norm": 0.05874243006110191,
      "learning_rate": 1.5114565899827815e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 88191472,
      "step": 40855
    },
    {
      "epoch": 6.66557911908646,
      "grad_norm": 0.04252789542078972,
      "learning_rate": 1.5108028422810094e-05,
      "loss": 0.021,
      "num_input_tokens_seen": 88202800,
      "step": 40860
    },
    {
      "epoch": 6.666394779771615,
      "grad_norm": 2.9396400451660156,
      "learning_rate": 1.510149174768578e-05,
      "loss": 0.1816,
      "num_input_tokens_seen": 88212848,
      "step": 40865
    },
    {
      "epoch": 6.6672104404567705,
      "grad_norm": 0.07159342616796494,
      "learning_rate": 1.5094955874984767e-05,
      "loss": 0.1988,
      "num_input_tokens_seen": 88224112,
      "step": 40870
    },
    {
      "epoch": 6.668026101141925,
      "grad_norm": 0.09100861102342606,
      "learning_rate": 1.5088420805236892e-05,
      "loss": 0.1974,
      "num_input_tokens_seen": 88234704,
      "step": 40875
    },
    {
      "epoch": 6.66884176182708,
      "grad_norm": 0.057788897305727005,
      "learning_rate": 1.5081886538971911e-05,
      "loss": 0.0077,
      "num_input_tokens_seen": 88247216,
      "step": 40880
    },
    {
      "epoch": 6.669657422512235,
      "grad_norm": 0.4550677537918091,
      "learning_rate": 1.5075353076719536e-05,
      "loss": 0.1011,
      "num_input_tokens_seen": 88259664,
      "step": 40885
    },
    {
      "epoch": 6.67047308319739,
      "grad_norm": 0.662420392036438,
      "learning_rate": 1.50688204190094e-05,
      "loss": 0.1004,
      "num_input_tokens_seen": 88270128,
      "step": 40890
    },
    {
      "epoch": 6.671288743882545,
      "grad_norm": 0.08329256623983383,
      "learning_rate": 1.5062288566371069e-05,
      "loss": 0.0041,
      "num_input_tokens_seen": 88280176,
      "step": 40895
    },
    {
      "epoch": 6.672104404567699,
      "grad_norm": 7.898841857910156,
      "learning_rate": 1.5055757519334048e-05,
      "loss": 0.1837,
      "num_input_tokens_seen": 88290288,
      "step": 40900
    },
    {
      "epoch": 6.672920065252855,
      "grad_norm": 0.05887956917285919,
      "learning_rate": 1.5049227278427782e-05,
      "loss": 0.0879,
      "num_input_tokens_seen": 88301296,
      "step": 40905
    },
    {
      "epoch": 6.67373572593801,
      "grad_norm": 2.963451862335205,
      "learning_rate": 1.504269784418164e-05,
      "loss": 0.0772,
      "num_input_tokens_seen": 88312752,
      "step": 40910
    },
    {
      "epoch": 6.674551386623165,
      "grad_norm": 0.16043300926685333,
      "learning_rate": 1.5036169217124938e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 88324144,
      "step": 40915
    },
    {
      "epoch": 6.6753670473083195,
      "grad_norm": 0.09560660272836685,
      "learning_rate": 1.5029641397786912e-05,
      "loss": 0.0822,
      "num_input_tokens_seen": 88335088,
      "step": 40920
    },
    {
      "epoch": 6.676182707993474,
      "grad_norm": 0.162864550948143,
      "learning_rate": 1.5023114386696746e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 88345360,
      "step": 40925
    },
    {
      "epoch": 6.67699836867863,
      "grad_norm": 3.8639090061187744,
      "learning_rate": 1.5016588184383536e-05,
      "loss": 0.2153,
      "num_input_tokens_seen": 88355888,
      "step": 40930
    },
    {
      "epoch": 6.677814029363785,
      "grad_norm": 0.04227403923869133,
      "learning_rate": 1.5010062791376355e-05,
      "loss": 0.1095,
      "num_input_tokens_seen": 88366832,
      "step": 40935
    },
    {
      "epoch": 6.67862969004894,
      "grad_norm": 5.405738353729248,
      "learning_rate": 1.5003538208204173e-05,
      "loss": 0.2397,
      "num_input_tokens_seen": 88378320,
      "step": 40940
    },
    {
      "epoch": 6.6794453507340945,
      "grad_norm": 3.2369942665100098,
      "learning_rate": 1.4997014435395906e-05,
      "loss": 0.0957,
      "num_input_tokens_seen": 88388656,
      "step": 40945
    },
    {
      "epoch": 6.680261011419249,
      "grad_norm": 3.6323814392089844,
      "learning_rate": 1.4990491473480403e-05,
      "loss": 0.3306,
      "num_input_tokens_seen": 88399632,
      "step": 40950
    },
    {
      "epoch": 6.681076672104405,
      "grad_norm": 0.3965166509151459,
      "learning_rate": 1.4983969322986446e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 88410032,
      "step": 40955
    },
    {
      "epoch": 6.68189233278956,
      "grad_norm": 0.416993111371994,
      "learning_rate": 1.497744798444276e-05,
      "loss": 0.0125,
      "num_input_tokens_seen": 88421008,
      "step": 40960
    },
    {
      "epoch": 6.682707993474715,
      "grad_norm": 0.1507231742143631,
      "learning_rate": 1.497092745837799e-05,
      "loss": 0.0068,
      "num_input_tokens_seen": 88432144,
      "step": 40965
    },
    {
      "epoch": 6.6835236541598695,
      "grad_norm": 0.09753464162349701,
      "learning_rate": 1.496440774532073e-05,
      "loss": 0.0503,
      "num_input_tokens_seen": 88442576,
      "step": 40970
    },
    {
      "epoch": 6.684339314845024,
      "grad_norm": 0.6746512055397034,
      "learning_rate": 1.49578888457995e-05,
      "loss": 0.1647,
      "num_input_tokens_seen": 88452048,
      "step": 40975
    },
    {
      "epoch": 6.685154975530179,
      "grad_norm": 0.06260435283184052,
      "learning_rate": 1.4951370760342754e-05,
      "loss": 0.288,
      "num_input_tokens_seen": 88463280,
      "step": 40980
    },
    {
      "epoch": 6.685970636215334,
      "grad_norm": 0.0902155265212059,
      "learning_rate": 1.4944853489478878e-05,
      "loss": 0.04,
      "num_input_tokens_seen": 88473488,
      "step": 40985
    },
    {
      "epoch": 6.68678629690049,
      "grad_norm": 0.03820433467626572,
      "learning_rate": 1.4938337033736196e-05,
      "loss": 0.0031,
      "num_input_tokens_seen": 88484304,
      "step": 40990
    },
    {
      "epoch": 6.6876019575856445,
      "grad_norm": 0.04262197017669678,
      "learning_rate": 1.4931821393642969e-05,
      "loss": 0.1728,
      "num_input_tokens_seen": 88495088,
      "step": 40995
    },
    {
      "epoch": 6.688417618270799,
      "grad_norm": 0.04590749740600586,
      "learning_rate": 1.4925306569727385e-05,
      "loss": 0.2215,
      "num_input_tokens_seen": 88506512,
      "step": 41000
    },
    {
      "epoch": 6.689233278955954,
      "grad_norm": 0.8048055768013,
      "learning_rate": 1.491879256251757e-05,
      "loss": 0.0059,
      "num_input_tokens_seen": 88518192,
      "step": 41005
    },
    {
      "epoch": 6.690048939641109,
      "grad_norm": 0.17440329492092133,
      "learning_rate": 1.4912279372541577e-05,
      "loss": 0.0108,
      "num_input_tokens_seen": 88526928,
      "step": 41010
    },
    {
      "epoch": 6.690864600326265,
      "grad_norm": 0.10211269557476044,
      "learning_rate": 1.4905767000327409e-05,
      "loss": 0.1417,
      "num_input_tokens_seen": 88537616,
      "step": 41015
    },
    {
      "epoch": 6.691680261011419,
      "grad_norm": 0.033806685358285904,
      "learning_rate": 1.4899255446402982e-05,
      "loss": 0.1643,
      "num_input_tokens_seen": 88547408,
      "step": 41020
    },
    {
      "epoch": 6.692495921696574,
      "grad_norm": 0.044099051505327225,
      "learning_rate": 1.4892744711296152e-05,
      "loss": 0.0039,
      "num_input_tokens_seen": 88557232,
      "step": 41025
    },
    {
      "epoch": 6.693311582381729,
      "grad_norm": 0.11672758311033249,
      "learning_rate": 1.488623479553473e-05,
      "loss": 0.0699,
      "num_input_tokens_seen": 88567632,
      "step": 41030
    },
    {
      "epoch": 6.694127243066884,
      "grad_norm": 0.11783801764249802,
      "learning_rate": 1.4879725699646424e-05,
      "loss": 0.055,
      "num_input_tokens_seen": 88579312,
      "step": 41035
    },
    {
      "epoch": 6.69494290375204,
      "grad_norm": 2.3646669387817383,
      "learning_rate": 1.4873217424158906e-05,
      "loss": 0.0096,
      "num_input_tokens_seen": 88590608,
      "step": 41040
    },
    {
      "epoch": 6.695758564437194,
      "grad_norm": 0.20109257102012634,
      "learning_rate": 1.4866709969599767e-05,
      "loss": 0.0073,
      "num_input_tokens_seen": 88601904,
      "step": 41045
    },
    {
      "epoch": 6.696574225122349,
      "grad_norm": 0.02105763740837574,
      "learning_rate": 1.486020333649653e-05,
      "loss": 0.0045,
      "num_input_tokens_seen": 88611088,
      "step": 41050
    },
    {
      "epoch": 6.697389885807504,
      "grad_norm": 4.496692180633545,
      "learning_rate": 1.4853697525376665e-05,
      "loss": 0.2224,
      "num_input_tokens_seen": 88623120,
      "step": 41055
    },
    {
      "epoch": 6.698205546492659,
      "grad_norm": 1.6089199781417847,
      "learning_rate": 1.484719253676756e-05,
      "loss": 0.2066,
      "num_input_tokens_seen": 88634064,
      "step": 41060
    },
    {
      "epoch": 6.699021207177814,
      "grad_norm": 0.14514349400997162,
      "learning_rate": 1.4840688371196543e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 88645392,
      "step": 41065
    },
    {
      "epoch": 6.699836867862969,
      "grad_norm": 0.1685430258512497,
      "learning_rate": 1.4834185029190873e-05,
      "loss": 0.0049,
      "num_input_tokens_seen": 88657808,
      "step": 41070
    },
    {
      "epoch": 6.700652528548124,
      "grad_norm": 0.14244039356708527,
      "learning_rate": 1.4827682511277746e-05,
      "loss": 0.0054,
      "num_input_tokens_seen": 88668048,
      "step": 41075
    },
    {
      "epoch": 6.701468189233279,
      "grad_norm": 0.05467163026332855,
      "learning_rate": 1.4821180817984288e-05,
      "loss": 0.0053,
      "num_input_tokens_seen": 88677456,
      "step": 41080
    },
    {
      "epoch": 6.702283849918434,
      "grad_norm": 5.239294528961182,
      "learning_rate": 1.4814679949837563e-05,
      "loss": 0.3309,
      "num_input_tokens_seen": 88688592,
      "step": 41085
    },
    {
      "epoch": 6.703099510603589,
      "grad_norm": 0.11179067194461823,
      "learning_rate": 1.4808179907364555e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 88699152,
      "step": 41090
    },
    {
      "epoch": 6.7039151712887435,
      "grad_norm": 0.09347817301750183,
      "learning_rate": 1.48016806910922e-05,
      "loss": 0.1037,
      "num_input_tokens_seen": 88709776,
      "step": 41095
    },
    {
      "epoch": 6.704730831973899,
      "grad_norm": 10.745227813720703,
      "learning_rate": 1.4795182301547356e-05,
      "loss": 0.2152,
      "num_input_tokens_seen": 88721328,
      "step": 41100
    },
    {
      "epoch": 6.705546492659054,
      "grad_norm": 3.5864243507385254,
      "learning_rate": 1.4788684739256808e-05,
      "loss": 0.2329,
      "num_input_tokens_seen": 88732432,
      "step": 41105
    },
    {
      "epoch": 6.706362153344209,
      "grad_norm": 0.04763638973236084,
      "learning_rate": 1.4782188004747289e-05,
      "loss": 0.0087,
      "num_input_tokens_seen": 88743312,
      "step": 41110
    },
    {
      "epoch": 6.707177814029364,
      "grad_norm": 4.77978515625,
      "learning_rate": 1.4775692098545451e-05,
      "loss": 0.016,
      "num_input_tokens_seen": 88754096,
      "step": 41115
    },
    {
      "epoch": 6.7079934747145185,
      "grad_norm": 3.673560380935669,
      "learning_rate": 1.4769197021177896e-05,
      "loss": 0.0635,
      "num_input_tokens_seen": 88765200,
      "step": 41120
    },
    {
      "epoch": 6.708809135399674,
      "grad_norm": 0.10142073780298233,
      "learning_rate": 1.476270277317114e-05,
      "loss": 0.1696,
      "num_input_tokens_seen": 88777136,
      "step": 41125
    },
    {
      "epoch": 6.709624796084829,
      "grad_norm": 16.93016242980957,
      "learning_rate": 1.475620935505164e-05,
      "loss": 0.1775,
      "num_input_tokens_seen": 88789040,
      "step": 41130
    },
    {
      "epoch": 6.710440456769984,
      "grad_norm": 0.30513206124305725,
      "learning_rate": 1.4749716767345784e-05,
      "loss": 0.2889,
      "num_input_tokens_seen": 88799792,
      "step": 41135
    },
    {
      "epoch": 6.711256117455139,
      "grad_norm": 1.2527740001678467,
      "learning_rate": 1.4743225010579889e-05,
      "loss": 0.0834,
      "num_input_tokens_seen": 88810224,
      "step": 41140
    },
    {
      "epoch": 6.712071778140293,
      "grad_norm": 0.1730339080095291,
      "learning_rate": 1.4736734085280226e-05,
      "loss": 0.2215,
      "num_input_tokens_seen": 88820112,
      "step": 41145
    },
    {
      "epoch": 6.712887438825448,
      "grad_norm": 14.265073776245117,
      "learning_rate": 1.4730243991972976e-05,
      "loss": 0.075,
      "num_input_tokens_seen": 88830864,
      "step": 41150
    },
    {
      "epoch": 6.713703099510604,
      "grad_norm": 0.20754019916057587,
      "learning_rate": 1.4723754731184253e-05,
      "loss": 0.006,
      "num_input_tokens_seen": 88842096,
      "step": 41155
    },
    {
      "epoch": 6.714518760195759,
      "grad_norm": 6.739881992340088,
      "learning_rate": 1.4717266303440113e-05,
      "loss": 0.1074,
      "num_input_tokens_seen": 88853232,
      "step": 41160
    },
    {
      "epoch": 6.715334420880914,
      "grad_norm": 0.1268126666545868,
      "learning_rate": 1.471077870926654e-05,
      "loss": 0.102,
      "num_input_tokens_seen": 88863408,
      "step": 41165
    },
    {
      "epoch": 6.716150081566068,
      "grad_norm": 2.2937026023864746,
      "learning_rate": 1.4704291949189452e-05,
      "loss": 0.1632,
      "num_input_tokens_seen": 88873648,
      "step": 41170
    },
    {
      "epoch": 6.716965742251223,
      "grad_norm": 0.015560554340481758,
      "learning_rate": 1.46978060237347e-05,
      "loss": 0.1059,
      "num_input_tokens_seen": 88884720,
      "step": 41175
    },
    {
      "epoch": 6.717781402936378,
      "grad_norm": 0.11674555391073227,
      "learning_rate": 1.4691320933428066e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 88895632,
      "step": 41180
    },
    {
      "epoch": 6.718597063621534,
      "grad_norm": 0.16365036368370056,
      "learning_rate": 1.4684836678795259e-05,
      "loss": 0.1994,
      "num_input_tokens_seen": 88905456,
      "step": 41185
    },
    {
      "epoch": 6.719412724306689,
      "grad_norm": 0.12613674998283386,
      "learning_rate": 1.4678353260361927e-05,
      "loss": 0.1081,
      "num_input_tokens_seen": 88916848,
      "step": 41190
    },
    {
      "epoch": 6.720228384991843,
      "grad_norm": 0.10388201475143433,
      "learning_rate": 1.4671870678653653e-05,
      "loss": 0.0087,
      "num_input_tokens_seen": 88928240,
      "step": 41195
    },
    {
      "epoch": 6.721044045676998,
      "grad_norm": 0.19808514416217804,
      "learning_rate": 1.466538893419595e-05,
      "loss": 0.0811,
      "num_input_tokens_seen": 88940016,
      "step": 41200
    },
    {
      "epoch": 6.721859706362153,
      "grad_norm": 0.07586171478033066,
      "learning_rate": 1.4658908027514256e-05,
      "loss": 0.1817,
      "num_input_tokens_seen": 88950544,
      "step": 41205
    },
    {
      "epoch": 6.722675367047309,
      "grad_norm": 0.21056191623210907,
      "learning_rate": 1.4652427959133947e-05,
      "loss": 0.1585,
      "num_input_tokens_seen": 88961520,
      "step": 41210
    },
    {
      "epoch": 6.7234910277324635,
      "grad_norm": 0.1211901530623436,
      "learning_rate": 1.4645948729580331e-05,
      "loss": 0.1525,
      "num_input_tokens_seen": 88971952,
      "step": 41215
    },
    {
      "epoch": 6.724306688417618,
      "grad_norm": 0.38611674308776855,
      "learning_rate": 1.4639470339378647e-05,
      "loss": 0.087,
      "num_input_tokens_seen": 88981840,
      "step": 41220
    },
    {
      "epoch": 6.725122349102773,
      "grad_norm": 12.179522514343262,
      "learning_rate": 1.4632992789054064e-05,
      "loss": 0.1006,
      "num_input_tokens_seen": 88992656,
      "step": 41225
    },
    {
      "epoch": 6.725938009787928,
      "grad_norm": 3.596252679824829,
      "learning_rate": 1.4626516079131692e-05,
      "loss": 0.3164,
      "num_input_tokens_seen": 89004432,
      "step": 41230
    },
    {
      "epoch": 6.726753670473083,
      "grad_norm": 0.08227868378162384,
      "learning_rate": 1.4620040210136557e-05,
      "loss": 0.0592,
      "num_input_tokens_seen": 89014736,
      "step": 41235
    },
    {
      "epoch": 6.7275693311582385,
      "grad_norm": 0.05020721256732941,
      "learning_rate": 1.461356518259363e-05,
      "loss": 0.1544,
      "num_input_tokens_seen": 89024976,
      "step": 41240
    },
    {
      "epoch": 6.728384991843393,
      "grad_norm": 0.09834147989749908,
      "learning_rate": 1.4607090997027812e-05,
      "loss": 0.0034,
      "num_input_tokens_seen": 89035440,
      "step": 41245
    },
    {
      "epoch": 6.729200652528548,
      "grad_norm": 7.497015476226807,
      "learning_rate": 1.4600617653963918e-05,
      "loss": 0.1151,
      "num_input_tokens_seen": 89046608,
      "step": 41250
    },
    {
      "epoch": 6.730016313213703,
      "grad_norm": 3.570956230163574,
      "learning_rate": 1.4594145153926737e-05,
      "loss": 0.1994,
      "num_input_tokens_seen": 89056912,
      "step": 41255
    },
    {
      "epoch": 6.730831973898858,
      "grad_norm": 0.0936165452003479,
      "learning_rate": 1.4587673497440946e-05,
      "loss": 0.1032,
      "num_input_tokens_seen": 89068048,
      "step": 41260
    },
    {
      "epoch": 6.731647634584013,
      "grad_norm": 0.18565014004707336,
      "learning_rate": 1.458120268503117e-05,
      "loss": 0.1172,
      "num_input_tokens_seen": 89078544,
      "step": 41265
    },
    {
      "epoch": 6.732463295269168,
      "grad_norm": 5.346120357513428,
      "learning_rate": 1.4574732717221972e-05,
      "loss": 0.105,
      "num_input_tokens_seen": 89090768,
      "step": 41270
    },
    {
      "epoch": 6.733278955954323,
      "grad_norm": 0.6010496020317078,
      "learning_rate": 1.456826359453784e-05,
      "loss": 0.1191,
      "num_input_tokens_seen": 89101040,
      "step": 41275
    },
    {
      "epoch": 6.734094616639478,
      "grad_norm": 5.97984504699707,
      "learning_rate": 1.4561795317503185e-05,
      "loss": 0.029,
      "num_input_tokens_seen": 89112112,
      "step": 41280
    },
    {
      "epoch": 6.734910277324633,
      "grad_norm": 0.1399141401052475,
      "learning_rate": 1.455532788664237e-05,
      "loss": 0.1909,
      "num_input_tokens_seen": 89123056,
      "step": 41285
    },
    {
      "epoch": 6.735725938009788,
      "grad_norm": 4.638060569763184,
      "learning_rate": 1.4548861302479672e-05,
      "loss": 0.1535,
      "num_input_tokens_seen": 89135120,
      "step": 41290
    },
    {
      "epoch": 6.736541598694943,
      "grad_norm": 0.05866732448339462,
      "learning_rate": 1.4542395565539302e-05,
      "loss": 0.1598,
      "num_input_tokens_seen": 89147184,
      "step": 41295
    },
    {
      "epoch": 6.737357259380098,
      "grad_norm": 6.314901351928711,
      "learning_rate": 1.453593067634541e-05,
      "loss": 0.065,
      "num_input_tokens_seen": 89156912,
      "step": 41300
    },
    {
      "epoch": 6.738172920065253,
      "grad_norm": 0.07997670024633408,
      "learning_rate": 1.4529466635422063e-05,
      "loss": 0.0063,
      "num_input_tokens_seen": 89167184,
      "step": 41305
    },
    {
      "epoch": 6.738988580750408,
      "grad_norm": 0.11030901223421097,
      "learning_rate": 1.4523003443293285e-05,
      "loss": 0.0767,
      "num_input_tokens_seen": 89178448,
      "step": 41310
    },
    {
      "epoch": 6.739804241435563,
      "grad_norm": 0.14322669804096222,
      "learning_rate": 1.4516541100483008e-05,
      "loss": 0.0584,
      "num_input_tokens_seen": 89189072,
      "step": 41315
    },
    {
      "epoch": 6.740619902120718,
      "grad_norm": 0.036629918962717056,
      "learning_rate": 1.4510079607515104e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 89200496,
      "step": 41320
    },
    {
      "epoch": 6.741435562805873,
      "grad_norm": 0.09099335968494415,
      "learning_rate": 1.4503618964913368e-05,
      "loss": 0.0633,
      "num_input_tokens_seen": 89209840,
      "step": 41325
    },
    {
      "epoch": 6.742251223491028,
      "grad_norm": 9.764974594116211,
      "learning_rate": 1.4497159173201541e-05,
      "loss": 0.2554,
      "num_input_tokens_seen": 89220240,
      "step": 41330
    },
    {
      "epoch": 6.743066884176183,
      "grad_norm": 0.08768236637115479,
      "learning_rate": 1.4490700232903281e-05,
      "loss": 0.0054,
      "num_input_tokens_seen": 89232400,
      "step": 41335
    },
    {
      "epoch": 6.7438825448613375,
      "grad_norm": 0.12594589591026306,
      "learning_rate": 1.4484242144542184e-05,
      "loss": 0.2616,
      "num_input_tokens_seen": 89242608,
      "step": 41340
    },
    {
      "epoch": 6.744698205546492,
      "grad_norm": 0.053064584732055664,
      "learning_rate": 1.4477784908641775e-05,
      "loss": 0.0098,
      "num_input_tokens_seen": 89253328,
      "step": 41345
    },
    {
      "epoch": 6.745513866231647,
      "grad_norm": 0.07828322052955627,
      "learning_rate": 1.4471328525725512e-05,
      "loss": 0.1012,
      "num_input_tokens_seen": 89263984,
      "step": 41350
    },
    {
      "epoch": 6.746329526916803,
      "grad_norm": 7.68565034866333,
      "learning_rate": 1.446487299631677e-05,
      "loss": 0.0523,
      "num_input_tokens_seen": 89275856,
      "step": 41355
    },
    {
      "epoch": 6.747145187601958,
      "grad_norm": 5.093059062957764,
      "learning_rate": 1.4458418320938886e-05,
      "loss": 0.0586,
      "num_input_tokens_seen": 89287024,
      "step": 41360
    },
    {
      "epoch": 6.7479608482871125,
      "grad_norm": 0.12296286225318909,
      "learning_rate": 1.4451964500115101e-05,
      "loss": 0.1454,
      "num_input_tokens_seen": 89298032,
      "step": 41365
    },
    {
      "epoch": 6.748776508972267,
      "grad_norm": 0.1551298350095749,
      "learning_rate": 1.4445511534368595e-05,
      "loss": 0.006,
      "num_input_tokens_seen": 89309424,
      "step": 41370
    },
    {
      "epoch": 6.749592169657422,
      "grad_norm": 0.21927422285079956,
      "learning_rate": 1.4439059424222474e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 89320880,
      "step": 41375
    },
    {
      "epoch": 6.750407830342578,
      "grad_norm": 1.674319863319397,
      "learning_rate": 1.4432608170199785e-05,
      "loss": 0.184,
      "num_input_tokens_seen": 89332624,
      "step": 41380
    },
    {
      "epoch": 6.751223491027733,
      "grad_norm": 3.81581449508667,
      "learning_rate": 1.4426157772823495e-05,
      "loss": 0.0857,
      "num_input_tokens_seen": 89343344,
      "step": 41385
    },
    {
      "epoch": 6.7520391517128875,
      "grad_norm": 3.1091864109039307,
      "learning_rate": 1.4419708232616508e-05,
      "loss": 0.1051,
      "num_input_tokens_seen": 89355344,
      "step": 41390
    },
    {
      "epoch": 6.752854812398042,
      "grad_norm": 0.2115515172481537,
      "learning_rate": 1.4413259550101654e-05,
      "loss": 0.1125,
      "num_input_tokens_seen": 89366160,
      "step": 41395
    },
    {
      "epoch": 6.753670473083197,
      "grad_norm": 0.17943738400936127,
      "learning_rate": 1.4406811725801696e-05,
      "loss": 0.0703,
      "num_input_tokens_seen": 89376976,
      "step": 41400
    },
    {
      "epoch": 6.754486133768353,
      "grad_norm": 7.111088275909424,
      "learning_rate": 1.4400364760239333e-05,
      "loss": 0.1543,
      "num_input_tokens_seen": 89388176,
      "step": 41405
    },
    {
      "epoch": 6.755301794453508,
      "grad_norm": 0.24837414920330048,
      "learning_rate": 1.4393918653937183e-05,
      "loss": 0.0209,
      "num_input_tokens_seen": 89399152,
      "step": 41410
    },
    {
      "epoch": 6.7561174551386625,
      "grad_norm": 20.415496826171875,
      "learning_rate": 1.4387473407417801e-05,
      "loss": 0.2762,
      "num_input_tokens_seen": 89409968,
      "step": 41415
    },
    {
      "epoch": 6.756933115823817,
      "grad_norm": 0.060787633061409,
      "learning_rate": 1.438102902120367e-05,
      "loss": 0.0688,
      "num_input_tokens_seen": 89420560,
      "step": 41420
    },
    {
      "epoch": 6.757748776508972,
      "grad_norm": 7.669675350189209,
      "learning_rate": 1.437458549581721e-05,
      "loss": 0.1184,
      "num_input_tokens_seen": 89431952,
      "step": 41425
    },
    {
      "epoch": 6.758564437194127,
      "grad_norm": 0.10015132278203964,
      "learning_rate": 1.4368142831780763e-05,
      "loss": 0.0103,
      "num_input_tokens_seen": 89443152,
      "step": 41430
    },
    {
      "epoch": 6.759380097879282,
      "grad_norm": 0.23086363077163696,
      "learning_rate": 1.4361701029616598e-05,
      "loss": 0.0093,
      "num_input_tokens_seen": 89454576,
      "step": 41435
    },
    {
      "epoch": 6.760195758564437,
      "grad_norm": 0.07244721055030823,
      "learning_rate": 1.4355260089846931e-05,
      "loss": 0.0045,
      "num_input_tokens_seen": 89465840,
      "step": 41440
    },
    {
      "epoch": 6.761011419249592,
      "grad_norm": 0.056052159518003464,
      "learning_rate": 1.434882001299389e-05,
      "loss": 0.0043,
      "num_input_tokens_seen": 89476400,
      "step": 41445
    },
    {
      "epoch": 6.761827079934747,
      "grad_norm": 0.4685874879360199,
      "learning_rate": 1.4342380799579533e-05,
      "loss": 0.2262,
      "num_input_tokens_seen": 89487568,
      "step": 41450
    },
    {
      "epoch": 6.762642740619902,
      "grad_norm": 0.2143806368112564,
      "learning_rate": 1.4335942450125872e-05,
      "loss": 0.0116,
      "num_input_tokens_seen": 89496400,
      "step": 41455
    },
    {
      "epoch": 6.763458401305057,
      "grad_norm": 0.03215618431568146,
      "learning_rate": 1.4329504965154827e-05,
      "loss": 0.0177,
      "num_input_tokens_seen": 89507440,
      "step": 41460
    },
    {
      "epoch": 6.764274061990212,
      "grad_norm": 0.19878214597702026,
      "learning_rate": 1.4323068345188253e-05,
      "loss": 0.1351,
      "num_input_tokens_seen": 89516624,
      "step": 41465
    },
    {
      "epoch": 6.765089722675367,
      "grad_norm": 0.01781969517469406,
      "learning_rate": 1.431663259074793e-05,
      "loss": 0.0676,
      "num_input_tokens_seen": 89526352,
      "step": 41470
    },
    {
      "epoch": 6.765905383360522,
      "grad_norm": 0.06626878678798676,
      "learning_rate": 1.4310197702355572e-05,
      "loss": 0.1046,
      "num_input_tokens_seen": 89538128,
      "step": 41475
    },
    {
      "epoch": 6.766721044045677,
      "grad_norm": 0.23306971788406372,
      "learning_rate": 1.430376368053283e-05,
      "loss": 0.0868,
      "num_input_tokens_seen": 89549136,
      "step": 41480
    },
    {
      "epoch": 6.767536704730832,
      "grad_norm": 0.0779542401432991,
      "learning_rate": 1.429733052580128e-05,
      "loss": 0.009,
      "num_input_tokens_seen": 89559728,
      "step": 41485
    },
    {
      "epoch": 6.768352365415987,
      "grad_norm": 0.12217021733522415,
      "learning_rate": 1.4290898238682421e-05,
      "loss": 0.0192,
      "num_input_tokens_seen": 89569904,
      "step": 41490
    },
    {
      "epoch": 6.769168026101142,
      "grad_norm": 0.02566716820001602,
      "learning_rate": 1.428446681969769e-05,
      "loss": 0.0951,
      "num_input_tokens_seen": 89581008,
      "step": 41495
    },
    {
      "epoch": 6.769983686786297,
      "grad_norm": 0.14071223139762878,
      "learning_rate": 1.427803626936845e-05,
      "loss": 0.315,
      "num_input_tokens_seen": 89591728,
      "step": 41500
    },
    {
      "epoch": 6.770799347471452,
      "grad_norm": 0.05853862687945366,
      "learning_rate": 1.4271606588215988e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 89601456,
      "step": 41505
    },
    {
      "epoch": 6.771615008156607,
      "grad_norm": 0.203269362449646,
      "learning_rate": 1.4265177776761534e-05,
      "loss": 0.0807,
      "num_input_tokens_seen": 89612368,
      "step": 41510
    },
    {
      "epoch": 6.7724306688417615,
      "grad_norm": 0.41243788599967957,
      "learning_rate": 1.4258749835526235e-05,
      "loss": 0.0117,
      "num_input_tokens_seen": 89623312,
      "step": 41515
    },
    {
      "epoch": 6.773246329526917,
      "grad_norm": 3.9026591777801514,
      "learning_rate": 1.4252322765031179e-05,
      "loss": 0.1886,
      "num_input_tokens_seen": 89633936,
      "step": 41520
    },
    {
      "epoch": 6.774061990212072,
      "grad_norm": 0.14258667826652527,
      "learning_rate": 1.4245896565797373e-05,
      "loss": 0.0959,
      "num_input_tokens_seen": 89643696,
      "step": 41525
    },
    {
      "epoch": 6.774877650897227,
      "grad_norm": 0.23517794907093048,
      "learning_rate": 1.4239471238345753e-05,
      "loss": 0.0128,
      "num_input_tokens_seen": 89653936,
      "step": 41530
    },
    {
      "epoch": 6.775693311582382,
      "grad_norm": 0.13160397112369537,
      "learning_rate": 1.4233046783197195e-05,
      "loss": 0.013,
      "num_input_tokens_seen": 89664752,
      "step": 41535
    },
    {
      "epoch": 6.7765089722675365,
      "grad_norm": 0.06285054236650467,
      "learning_rate": 1.4226623200872496e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 89675984,
      "step": 41540
    },
    {
      "epoch": 6.777324632952691,
      "grad_norm": 0.12321215122938156,
      "learning_rate": 1.4220200491892383e-05,
      "loss": 0.0031,
      "num_input_tokens_seen": 89686704,
      "step": 41545
    },
    {
      "epoch": 6.778140293637847,
      "grad_norm": 1.5065103769302368,
      "learning_rate": 1.4213778656777515e-05,
      "loss": 0.094,
      "num_input_tokens_seen": 89697040,
      "step": 41550
    },
    {
      "epoch": 6.778955954323002,
      "grad_norm": 0.07444112747907639,
      "learning_rate": 1.4207357696048479e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 89708208,
      "step": 41555
    },
    {
      "epoch": 6.779771615008157,
      "grad_norm": 0.07855966687202454,
      "learning_rate": 1.4200937610225787e-05,
      "loss": 0.0073,
      "num_input_tokens_seen": 89719120,
      "step": 41560
    },
    {
      "epoch": 6.780587275693311,
      "grad_norm": 0.09173068404197693,
      "learning_rate": 1.4194518399829887e-05,
      "loss": 0.0724,
      "num_input_tokens_seen": 89730384,
      "step": 41565
    },
    {
      "epoch": 6.781402936378466,
      "grad_norm": 0.08013544976711273,
      "learning_rate": 1.4188100065381144e-05,
      "loss": 0.0035,
      "num_input_tokens_seen": 89742128,
      "step": 41570
    },
    {
      "epoch": 6.782218597063622,
      "grad_norm": 0.03320332244038582,
      "learning_rate": 1.4181682607399877e-05,
      "loss": 0.1222,
      "num_input_tokens_seen": 89752944,
      "step": 41575
    },
    {
      "epoch": 6.783034257748777,
      "grad_norm": 0.058275748044252396,
      "learning_rate": 1.4175266026406308e-05,
      "loss": 0.2549,
      "num_input_tokens_seen": 89762448,
      "step": 41580
    },
    {
      "epoch": 6.783849918433932,
      "grad_norm": 0.07470196485519409,
      "learning_rate": 1.4168850322920602e-05,
      "loss": 0.1627,
      "num_input_tokens_seen": 89773200,
      "step": 41585
    },
    {
      "epoch": 6.784665579119086,
      "grad_norm": 0.08631259202957153,
      "learning_rate": 1.4162435497462842e-05,
      "loss": 0.0104,
      "num_input_tokens_seen": 89783952,
      "step": 41590
    },
    {
      "epoch": 6.785481239804241,
      "grad_norm": 0.16717347502708435,
      "learning_rate": 1.415602155055305e-05,
      "loss": 0.0285,
      "num_input_tokens_seen": 89796528,
      "step": 41595
    },
    {
      "epoch": 6.786296900489396,
      "grad_norm": 0.13983605802059174,
      "learning_rate": 1.4149608482711177e-05,
      "loss": 0.0049,
      "num_input_tokens_seen": 89807664,
      "step": 41600
    },
    {
      "epoch": 6.787112561174552,
      "grad_norm": 5.405811309814453,
      "learning_rate": 1.4143196294457092e-05,
      "loss": 0.0504,
      "num_input_tokens_seen": 89817552,
      "step": 41605
    },
    {
      "epoch": 6.787928221859707,
      "grad_norm": 0.14414404332637787,
      "learning_rate": 1.4136784986310603e-05,
      "loss": 0.0279,
      "num_input_tokens_seen": 89829008,
      "step": 41610
    },
    {
      "epoch": 6.788743882544861,
      "grad_norm": 0.06346258521080017,
      "learning_rate": 1.4130374558791442e-05,
      "loss": 0.1968,
      "num_input_tokens_seen": 89839856,
      "step": 41615
    },
    {
      "epoch": 6.789559543230016,
      "grad_norm": 0.22917187213897705,
      "learning_rate": 1.412396501241926e-05,
      "loss": 0.0934,
      "num_input_tokens_seen": 89851312,
      "step": 41620
    },
    {
      "epoch": 6.790375203915171,
      "grad_norm": 0.09620549529790878,
      "learning_rate": 1.411755634771367e-05,
      "loss": 0.1002,
      "num_input_tokens_seen": 89863248,
      "step": 41625
    },
    {
      "epoch": 6.791190864600326,
      "grad_norm": 0.09660594910383224,
      "learning_rate": 1.411114856519418e-05,
      "loss": 0.0886,
      "num_input_tokens_seen": 89875344,
      "step": 41630
    },
    {
      "epoch": 6.7920065252854815,
      "grad_norm": 0.053771648555994034,
      "learning_rate": 1.4104741665380236e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 89887056,
      "step": 41635
    },
    {
      "epoch": 6.792822185970636,
      "grad_norm": 0.07990163564682007,
      "learning_rate": 1.4098335648791216e-05,
      "loss": 0.1245,
      "num_input_tokens_seen": 89898064,
      "step": 41640
    },
    {
      "epoch": 6.793637846655791,
      "grad_norm": 0.12838514149188995,
      "learning_rate": 1.4091930515946422e-05,
      "loss": 0.1057,
      "num_input_tokens_seen": 89908816,
      "step": 41645
    },
    {
      "epoch": 6.794453507340946,
      "grad_norm": 0.1484857052564621,
      "learning_rate": 1.4085526267365084e-05,
      "loss": 0.0046,
      "num_input_tokens_seen": 89918896,
      "step": 41650
    },
    {
      "epoch": 6.795269168026101,
      "grad_norm": 2.432342290878296,
      "learning_rate": 1.4079122903566371e-05,
      "loss": 0.0083,
      "num_input_tokens_seen": 89928880,
      "step": 41655
    },
    {
      "epoch": 6.7960848287112565,
      "grad_norm": 0.1658189594745636,
      "learning_rate": 1.4072720425069364e-05,
      "loss": 0.0115,
      "num_input_tokens_seen": 89939952,
      "step": 41660
    },
    {
      "epoch": 6.796900489396411,
      "grad_norm": 4.673414707183838,
      "learning_rate": 1.4066318832393086e-05,
      "loss": 0.2901,
      "num_input_tokens_seen": 89949360,
      "step": 41665
    },
    {
      "epoch": 6.797716150081566,
      "grad_norm": 0.26324915885925293,
      "learning_rate": 1.4059918126056478e-05,
      "loss": 0.0923,
      "num_input_tokens_seen": 89960144,
      "step": 41670
    },
    {
      "epoch": 6.798531810766721,
      "grad_norm": 3.5412778854370117,
      "learning_rate": 1.405351830657841e-05,
      "loss": 0.2425,
      "num_input_tokens_seen": 89971184,
      "step": 41675
    },
    {
      "epoch": 6.799347471451876,
      "grad_norm": 0.0986003503203392,
      "learning_rate": 1.4047119374477696e-05,
      "loss": 0.0439,
      "num_input_tokens_seen": 89982288,
      "step": 41680
    },
    {
      "epoch": 6.800163132137031,
      "grad_norm": 0.0386163592338562,
      "learning_rate": 1.4040721330273062e-05,
      "loss": 0.0072,
      "num_input_tokens_seen": 89993744,
      "step": 41685
    },
    {
      "epoch": 6.800978792822186,
      "grad_norm": 6.793071746826172,
      "learning_rate": 1.4034324174483166e-05,
      "loss": 0.0158,
      "num_input_tokens_seen": 90005232,
      "step": 41690
    },
    {
      "epoch": 6.801794453507341,
      "grad_norm": 0.28747808933258057,
      "learning_rate": 1.4027927907626586e-05,
      "loss": 0.0381,
      "num_input_tokens_seen": 90016240,
      "step": 41695
    },
    {
      "epoch": 6.802610114192496,
      "grad_norm": 0.035620465874671936,
      "learning_rate": 1.4021532530221846e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 90026736,
      "step": 41700
    },
    {
      "epoch": 6.803425774877651,
      "grad_norm": 0.2508222460746765,
      "learning_rate": 1.4015138042787381e-05,
      "loss": 0.0764,
      "num_input_tokens_seen": 90037296,
      "step": 41705
    },
    {
      "epoch": 6.804241435562806,
      "grad_norm": 0.18858025968074799,
      "learning_rate": 1.4008744445841566e-05,
      "loss": 0.0094,
      "num_input_tokens_seen": 90048176,
      "step": 41710
    },
    {
      "epoch": 6.80505709624796,
      "grad_norm": 0.03687680512666702,
      "learning_rate": 1.4002351739902691e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 90059056,
      "step": 41715
    },
    {
      "epoch": 6.805872756933116,
      "grad_norm": 0.13411162793636322,
      "learning_rate": 1.3995959925488988e-05,
      "loss": 0.0035,
      "num_input_tokens_seen": 90069456,
      "step": 41720
    },
    {
      "epoch": 6.806688417618271,
      "grad_norm": 0.06725441664457321,
      "learning_rate": 1.3989569003118609e-05,
      "loss": 0.0025,
      "num_input_tokens_seen": 90081840,
      "step": 41725
    },
    {
      "epoch": 6.807504078303426,
      "grad_norm": 0.20298072695732117,
      "learning_rate": 1.398317897330963e-05,
      "loss": 0.0803,
      "num_input_tokens_seen": 90091600,
      "step": 41730
    },
    {
      "epoch": 6.808319738988581,
      "grad_norm": 0.11427846550941467,
      "learning_rate": 1.3976789836580062e-05,
      "loss": 0.07,
      "num_input_tokens_seen": 90103344,
      "step": 41735
    },
    {
      "epoch": 6.809135399673735,
      "grad_norm": 0.9090158343315125,
      "learning_rate": 1.3970401593447843e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 90114352,
      "step": 41740
    },
    {
      "epoch": 6.809951060358891,
      "grad_norm": 11.22180461883545,
      "learning_rate": 1.3964014244430834e-05,
      "loss": 0.0845,
      "num_input_tokens_seen": 90125584,
      "step": 41745
    },
    {
      "epoch": 6.810766721044046,
      "grad_norm": 0.09490711987018585,
      "learning_rate": 1.3957627790046826e-05,
      "loss": 0.1107,
      "num_input_tokens_seen": 90136304,
      "step": 41750
    },
    {
      "epoch": 6.811582381729201,
      "grad_norm": 0.18473337590694427,
      "learning_rate": 1.3951242230813538e-05,
      "loss": 0.0164,
      "num_input_tokens_seen": 90147088,
      "step": 41755
    },
    {
      "epoch": 6.8123980424143555,
      "grad_norm": 0.04754093289375305,
      "learning_rate": 1.3944857567248615e-05,
      "loss": 0.2755,
      "num_input_tokens_seen": 90158512,
      "step": 41760
    },
    {
      "epoch": 6.81321370309951,
      "grad_norm": 1.2219489812850952,
      "learning_rate": 1.3938473799869622e-05,
      "loss": 0.0719,
      "num_input_tokens_seen": 90168880,
      "step": 41765
    },
    {
      "epoch": 6.814029363784666,
      "grad_norm": 0.24588292837142944,
      "learning_rate": 1.3932090929194075e-05,
      "loss": 0.0057,
      "num_input_tokens_seen": 90179696,
      "step": 41770
    },
    {
      "epoch": 6.814845024469821,
      "grad_norm": 2.426314353942871,
      "learning_rate": 1.39257089557394e-05,
      "loss": 0.0891,
      "num_input_tokens_seen": 90190960,
      "step": 41775
    },
    {
      "epoch": 6.815660685154976,
      "grad_norm": 3.5945773124694824,
      "learning_rate": 1.3919327880022945e-05,
      "loss": 0.1121,
      "num_input_tokens_seen": 90202288,
      "step": 41780
    },
    {
      "epoch": 6.8164763458401305,
      "grad_norm": 0.07115895301103592,
      "learning_rate": 1.3912947702561995e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 90214288,
      "step": 41785
    },
    {
      "epoch": 6.817292006525285,
      "grad_norm": 4.984953880310059,
      "learning_rate": 1.390656842387375e-05,
      "loss": 0.2038,
      "num_input_tokens_seen": 90225168,
      "step": 41790
    },
    {
      "epoch": 6.81810766721044,
      "grad_norm": 0.02814999222755432,
      "learning_rate": 1.3900190044475364e-05,
      "loss": 0.0937,
      "num_input_tokens_seen": 90235760,
      "step": 41795
    },
    {
      "epoch": 6.818923327895595,
      "grad_norm": 0.02262449637055397,
      "learning_rate": 1.3893812564883896e-05,
      "loss": 0.009,
      "num_input_tokens_seen": 90246864,
      "step": 41800
    },
    {
      "epoch": 6.819738988580751,
      "grad_norm": 0.050474390387535095,
      "learning_rate": 1.3887435985616332e-05,
      "loss": 0.2542,
      "num_input_tokens_seen": 90257680,
      "step": 41805
    },
    {
      "epoch": 6.8205546492659055,
      "grad_norm": 0.050444345921278,
      "learning_rate": 1.3881060307189592e-05,
      "loss": 0.2553,
      "num_input_tokens_seen": 90268976,
      "step": 41810
    },
    {
      "epoch": 6.82137030995106,
      "grad_norm": 2.1560733318328857,
      "learning_rate": 1.387468553012052e-05,
      "loss": 0.1103,
      "num_input_tokens_seen": 90279696,
      "step": 41815
    },
    {
      "epoch": 6.822185970636215,
      "grad_norm": 0.0733235776424408,
      "learning_rate": 1.386831165492589e-05,
      "loss": 0.0283,
      "num_input_tokens_seen": 90288624,
      "step": 41820
    },
    {
      "epoch": 6.82300163132137,
      "grad_norm": 0.024894678965210915,
      "learning_rate": 1.3861938682122396e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 90299984,
      "step": 41825
    },
    {
      "epoch": 6.823817292006526,
      "grad_norm": 0.06028543412685394,
      "learning_rate": 1.3855566612226666e-05,
      "loss": 0.0026,
      "num_input_tokens_seen": 90309264,
      "step": 41830
    },
    {
      "epoch": 6.8246329526916805,
      "grad_norm": 1.5850801467895508,
      "learning_rate": 1.3849195445755258e-05,
      "loss": 0.131,
      "num_input_tokens_seen": 90320528,
      "step": 41835
    },
    {
      "epoch": 6.825448613376835,
      "grad_norm": 0.035399507731199265,
      "learning_rate": 1.3842825183224642e-05,
      "loss": 0.0648,
      "num_input_tokens_seen": 90331184,
      "step": 41840
    },
    {
      "epoch": 6.82626427406199,
      "grad_norm": 0.030352847650647163,
      "learning_rate": 1.3836455825151229e-05,
      "loss": 0.0048,
      "num_input_tokens_seen": 90341200,
      "step": 41845
    },
    {
      "epoch": 6.827079934747145,
      "grad_norm": 0.06919962167739868,
      "learning_rate": 1.3830087372051347e-05,
      "loss": 0.1411,
      "num_input_tokens_seen": 90350992,
      "step": 41850
    },
    {
      "epoch": 6.827895595432301,
      "grad_norm": 3.3427679538726807,
      "learning_rate": 1.3823719824441262e-05,
      "loss": 0.1648,
      "num_input_tokens_seen": 90361520,
      "step": 41855
    },
    {
      "epoch": 6.828711256117455,
      "grad_norm": 0.37280991673469543,
      "learning_rate": 1.3817353182837151e-05,
      "loss": 0.1894,
      "num_input_tokens_seen": 90372112,
      "step": 41860
    },
    {
      "epoch": 6.82952691680261,
      "grad_norm": 0.3510379195213318,
      "learning_rate": 1.3810987447755136e-05,
      "loss": 0.2406,
      "num_input_tokens_seen": 90383856,
      "step": 41865
    },
    {
      "epoch": 6.830342577487765,
      "grad_norm": 0.110009104013443,
      "learning_rate": 1.380462261971125e-05,
      "loss": 0.0737,
      "num_input_tokens_seen": 90393552,
      "step": 41870
    },
    {
      "epoch": 6.83115823817292,
      "grad_norm": 4.968212604522705,
      "learning_rate": 1.3798258699221456e-05,
      "loss": 0.2073,
      "num_input_tokens_seen": 90403056,
      "step": 41875
    },
    {
      "epoch": 6.831973898858075,
      "grad_norm": 0.08078257739543915,
      "learning_rate": 1.3791895686801653e-05,
      "loss": 0.1244,
      "num_input_tokens_seen": 90414160,
      "step": 41880
    },
    {
      "epoch": 6.8327895595432295,
      "grad_norm": 3.059316635131836,
      "learning_rate": 1.3785533582967642e-05,
      "loss": 0.1318,
      "num_input_tokens_seen": 90426160,
      "step": 41885
    },
    {
      "epoch": 6.833605220228385,
      "grad_norm": 4.10181999206543,
      "learning_rate": 1.3779172388235192e-05,
      "loss": 0.4578,
      "num_input_tokens_seen": 90436816,
      "step": 41890
    },
    {
      "epoch": 6.83442088091354,
      "grad_norm": 0.15112021565437317,
      "learning_rate": 1.3772812103119964e-05,
      "loss": 0.0875,
      "num_input_tokens_seen": 90448144,
      "step": 41895
    },
    {
      "epoch": 6.835236541598695,
      "grad_norm": 4.008183002471924,
      "learning_rate": 1.376645272813755e-05,
      "loss": 0.0098,
      "num_input_tokens_seen": 90458352,
      "step": 41900
    },
    {
      "epoch": 6.83605220228385,
      "grad_norm": 3.7091166973114014,
      "learning_rate": 1.3760094263803474e-05,
      "loss": 0.1987,
      "num_input_tokens_seen": 90469264,
      "step": 41905
    },
    {
      "epoch": 6.8368678629690045,
      "grad_norm": 0.07551616430282593,
      "learning_rate": 1.3753736710633192e-05,
      "loss": 0.1271,
      "num_input_tokens_seen": 90478832,
      "step": 41910
    },
    {
      "epoch": 6.83768352365416,
      "grad_norm": 0.03895062580704689,
      "learning_rate": 1.3747380069142073e-05,
      "loss": 0.0048,
      "num_input_tokens_seen": 90489840,
      "step": 41915
    },
    {
      "epoch": 6.838499184339315,
      "grad_norm": 3.905212640762329,
      "learning_rate": 1.3741024339845426e-05,
      "loss": 0.1333,
      "num_input_tokens_seen": 90501232,
      "step": 41920
    },
    {
      "epoch": 6.83931484502447,
      "grad_norm": 0.134656623005867,
      "learning_rate": 1.373466952325847e-05,
      "loss": 0.0041,
      "num_input_tokens_seen": 90512560,
      "step": 41925
    },
    {
      "epoch": 6.840130505709625,
      "grad_norm": 0.07766005396842957,
      "learning_rate": 1.372831561989636e-05,
      "loss": 0.0088,
      "num_input_tokens_seen": 90523856,
      "step": 41930
    },
    {
      "epoch": 6.8409461663947795,
      "grad_norm": 0.042651377618312836,
      "learning_rate": 1.3721962630274171e-05,
      "loss": 0.1277,
      "num_input_tokens_seen": 90532720,
      "step": 41935
    },
    {
      "epoch": 6.841761827079935,
      "grad_norm": 0.0998101532459259,
      "learning_rate": 1.3715610554906922e-05,
      "loss": 0.147,
      "num_input_tokens_seen": 90543888,
      "step": 41940
    },
    {
      "epoch": 6.84257748776509,
      "grad_norm": 6.188546657562256,
      "learning_rate": 1.370925939430954e-05,
      "loss": 0.1926,
      "num_input_tokens_seen": 90553712,
      "step": 41945
    },
    {
      "epoch": 6.843393148450245,
      "grad_norm": 8.392866134643555,
      "learning_rate": 1.3702909148996878e-05,
      "loss": 0.0173,
      "num_input_tokens_seen": 90563920,
      "step": 41950
    },
    {
      "epoch": 6.8442088091354,
      "grad_norm": 0.11783468723297119,
      "learning_rate": 1.3696559819483722e-05,
      "loss": 0.0079,
      "num_input_tokens_seen": 90574544,
      "step": 41955
    },
    {
      "epoch": 6.8450244698205545,
      "grad_norm": 0.43951496481895447,
      "learning_rate": 1.3690211406284784e-05,
      "loss": 0.0281,
      "num_input_tokens_seen": 90586384,
      "step": 41960
    },
    {
      "epoch": 6.845840130505709,
      "grad_norm": 0.09904633462429047,
      "learning_rate": 1.368386390991469e-05,
      "loss": 0.0976,
      "num_input_tokens_seen": 90597776,
      "step": 41965
    },
    {
      "epoch": 6.846655791190865,
      "grad_norm": 0.14969505369663239,
      "learning_rate": 1.3677517330888007e-05,
      "loss": 0.0062,
      "num_input_tokens_seen": 90610128,
      "step": 41970
    },
    {
      "epoch": 6.84747145187602,
      "grad_norm": 0.09711721539497375,
      "learning_rate": 1.3671171669719218e-05,
      "loss": 0.0949,
      "num_input_tokens_seen": 90621168,
      "step": 41975
    },
    {
      "epoch": 6.848287112561175,
      "grad_norm": 0.2408400923013687,
      "learning_rate": 1.3664826926922736e-05,
      "loss": 0.0131,
      "num_input_tokens_seen": 90631536,
      "step": 41980
    },
    {
      "epoch": 6.849102773246329,
      "grad_norm": 10.967161178588867,
      "learning_rate": 1.3658483103012898e-05,
      "loss": 0.1604,
      "num_input_tokens_seen": 90641168,
      "step": 41985
    },
    {
      "epoch": 6.849918433931484,
      "grad_norm": 0.03806580603122711,
      "learning_rate": 1.3652140198503966e-05,
      "loss": 0.0056,
      "num_input_tokens_seen": 90651632,
      "step": 41990
    },
    {
      "epoch": 6.850734094616639,
      "grad_norm": 0.08589682728052139,
      "learning_rate": 1.364579821391012e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 90662128,
      "step": 41995
    },
    {
      "epoch": 6.851549755301795,
      "grad_norm": 0.08156748116016388,
      "learning_rate": 1.3639457149745489e-05,
      "loss": 0.0117,
      "num_input_tokens_seen": 90673808,
      "step": 42000
    },
    {
      "epoch": 6.85236541598695,
      "grad_norm": 2.5086851119995117,
      "learning_rate": 1.3633117006524102e-05,
      "loss": 0.0272,
      "num_input_tokens_seen": 90684752,
      "step": 42005
    },
    {
      "epoch": 6.853181076672104,
      "grad_norm": 0.07278906553983688,
      "learning_rate": 1.3626777784759925e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 90695312,
      "step": 42010
    },
    {
      "epoch": 6.853996737357259,
      "grad_norm": 4.181410789489746,
      "learning_rate": 1.3620439484966851e-05,
      "loss": 0.0125,
      "num_input_tokens_seen": 90705520,
      "step": 42015
    },
    {
      "epoch": 6.854812398042414,
      "grad_norm": 0.05359490588307381,
      "learning_rate": 1.3614102107658693e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 90717584,
      "step": 42020
    },
    {
      "epoch": 6.85562805872757,
      "grad_norm": 0.11791082471609116,
      "learning_rate": 1.3607765653349185e-05,
      "loss": 0.0124,
      "num_input_tokens_seen": 90728496,
      "step": 42025
    },
    {
      "epoch": 6.856443719412725,
      "grad_norm": 3.67244029045105,
      "learning_rate": 1.3601430122552e-05,
      "loss": 0.0962,
      "num_input_tokens_seen": 90738960,
      "step": 42030
    },
    {
      "epoch": 6.857259380097879,
      "grad_norm": 0.059158798307180405,
      "learning_rate": 1.3595095515780726e-05,
      "loss": 0.0693,
      "num_input_tokens_seen": 90750352,
      "step": 42035
    },
    {
      "epoch": 6.858075040783034,
      "grad_norm": 0.04778822138905525,
      "learning_rate": 1.3588761833548875e-05,
      "loss": 0.1266,
      "num_input_tokens_seen": 90761040,
      "step": 42040
    },
    {
      "epoch": 6.858890701468189,
      "grad_norm": 6.498258113861084,
      "learning_rate": 1.358242907636989e-05,
      "loss": 0.0789,
      "num_input_tokens_seen": 90770992,
      "step": 42045
    },
    {
      "epoch": 6.859706362153344,
      "grad_norm": 0.565824568271637,
      "learning_rate": 1.3576097244757138e-05,
      "loss": 0.0371,
      "num_input_tokens_seen": 90779088,
      "step": 42050
    },
    {
      "epoch": 6.8605220228384995,
      "grad_norm": 0.04997954145073891,
      "learning_rate": 1.3569766339223907e-05,
      "loss": 0.232,
      "num_input_tokens_seen": 90789616,
      "step": 42055
    },
    {
      "epoch": 6.861337683523654,
      "grad_norm": 3.0516979694366455,
      "learning_rate": 1.3563436360283412e-05,
      "loss": 0.0071,
      "num_input_tokens_seen": 90800496,
      "step": 42060
    },
    {
      "epoch": 6.862153344208809,
      "grad_norm": 31.080303192138672,
      "learning_rate": 1.3557107308448796e-05,
      "loss": 0.2309,
      "num_input_tokens_seen": 90809808,
      "step": 42065
    },
    {
      "epoch": 6.862969004893964,
      "grad_norm": 0.1256628930568695,
      "learning_rate": 1.355077918423312e-05,
      "loss": 0.1463,
      "num_input_tokens_seen": 90820880,
      "step": 42070
    },
    {
      "epoch": 6.863784665579119,
      "grad_norm": 0.07377687096595764,
      "learning_rate": 1.3544451988149376e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 90831696,
      "step": 42075
    },
    {
      "epoch": 6.864600326264274,
      "grad_norm": 0.10171285271644592,
      "learning_rate": 1.353812572071047e-05,
      "loss": 0.0035,
      "num_input_tokens_seen": 90842512,
      "step": 42080
    },
    {
      "epoch": 6.865415986949429,
      "grad_norm": 3.5421407222747803,
      "learning_rate": 1.353180038242926e-05,
      "loss": 0.0799,
      "num_input_tokens_seen": 90852784,
      "step": 42085
    },
    {
      "epoch": 6.866231647634584,
      "grad_norm": 0.4722101390361786,
      "learning_rate": 1.3525475973818502e-05,
      "loss": 0.1369,
      "num_input_tokens_seen": 90863280,
      "step": 42090
    },
    {
      "epoch": 6.867047308319739,
      "grad_norm": 0.054875221103429794,
      "learning_rate": 1.351915249539088e-05,
      "loss": 0.2059,
      "num_input_tokens_seen": 90874160,
      "step": 42095
    },
    {
      "epoch": 6.867862969004894,
      "grad_norm": 0.10923886299133301,
      "learning_rate": 1.3512829947659011e-05,
      "loss": 0.2623,
      "num_input_tokens_seen": 90884752,
      "step": 42100
    },
    {
      "epoch": 6.868678629690049,
      "grad_norm": 0.11440841108560562,
      "learning_rate": 1.350650833113542e-05,
      "loss": 0.1509,
      "num_input_tokens_seen": 90894704,
      "step": 42105
    },
    {
      "epoch": 6.869494290375204,
      "grad_norm": 0.06628463417291641,
      "learning_rate": 1.3500187646332593e-05,
      "loss": 0.1294,
      "num_input_tokens_seen": 90904528,
      "step": 42110
    },
    {
      "epoch": 6.870309951060359,
      "grad_norm": 0.19406814873218536,
      "learning_rate": 1.3493867893762904e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 90915696,
      "step": 42115
    },
    {
      "epoch": 6.871125611745514,
      "grad_norm": 0.2199442833662033,
      "learning_rate": 1.3487549073938666e-05,
      "loss": 0.0075,
      "num_input_tokens_seen": 90925776,
      "step": 42120
    },
    {
      "epoch": 6.871941272430669,
      "grad_norm": 0.09250465780496597,
      "learning_rate": 1.3481231187372111e-05,
      "loss": 0.0033,
      "num_input_tokens_seen": 90936144,
      "step": 42125
    },
    {
      "epoch": 6.872756933115824,
      "grad_norm": 0.23881153762340546,
      "learning_rate": 1.3474914234575406e-05,
      "loss": 0.1522,
      "num_input_tokens_seen": 90946480,
      "step": 42130
    },
    {
      "epoch": 6.873572593800979,
      "grad_norm": 0.8363812565803528,
      "learning_rate": 1.346859821606063e-05,
      "loss": 0.0061,
      "num_input_tokens_seen": 90956720,
      "step": 42135
    },
    {
      "epoch": 6.874388254486134,
      "grad_norm": 0.14776800572872162,
      "learning_rate": 1.3462283132339787e-05,
      "loss": 0.0839,
      "num_input_tokens_seen": 90967312,
      "step": 42140
    },
    {
      "epoch": 6.875203915171289,
      "grad_norm": 10.871723175048828,
      "learning_rate": 1.3455968983924822e-05,
      "loss": 0.2272,
      "num_input_tokens_seen": 90978000,
      "step": 42145
    },
    {
      "epoch": 6.876019575856444,
      "grad_norm": 0.21014146506786346,
      "learning_rate": 1.344965577132758e-05,
      "loss": 0.0062,
      "num_input_tokens_seen": 90988304,
      "step": 42150
    },
    {
      "epoch": 6.876835236541599,
      "grad_norm": 12.95048999786377,
      "learning_rate": 1.344334349505985e-05,
      "loss": 0.0869,
      "num_input_tokens_seen": 90998288,
      "step": 42155
    },
    {
      "epoch": 6.877650897226753,
      "grad_norm": 0.05227712169289589,
      "learning_rate": 1.3437032155633333e-05,
      "loss": 0.0193,
      "num_input_tokens_seen": 91009936,
      "step": 42160
    },
    {
      "epoch": 6.878466557911908,
      "grad_norm": 0.6685779094696045,
      "learning_rate": 1.343072175355966e-05,
      "loss": 0.0092,
      "num_input_tokens_seen": 91021456,
      "step": 42165
    },
    {
      "epoch": 6.879282218597064,
      "grad_norm": 0.1707424372434616,
      "learning_rate": 1.342441228935038e-05,
      "loss": 0.1517,
      "num_input_tokens_seen": 91032784,
      "step": 42170
    },
    {
      "epoch": 6.880097879282219,
      "grad_norm": 11.286739349365234,
      "learning_rate": 1.3418103763516979e-05,
      "loss": 0.0617,
      "num_input_tokens_seen": 91043952,
      "step": 42175
    },
    {
      "epoch": 6.8809135399673735,
      "grad_norm": 2.9973649978637695,
      "learning_rate": 1.3411796176570852e-05,
      "loss": 0.1612,
      "num_input_tokens_seen": 91054704,
      "step": 42180
    },
    {
      "epoch": 6.881729200652528,
      "grad_norm": 0.3737064599990845,
      "learning_rate": 1.3405489529023322e-05,
      "loss": 0.1331,
      "num_input_tokens_seen": 91064176,
      "step": 42185
    },
    {
      "epoch": 6.882544861337683,
      "grad_norm": 5.6691741943359375,
      "learning_rate": 1.339918382138564e-05,
      "loss": 0.1819,
      "num_input_tokens_seen": 91074768,
      "step": 42190
    },
    {
      "epoch": 6.883360522022839,
      "grad_norm": 2.48360276222229,
      "learning_rate": 1.3392879054168983e-05,
      "loss": 0.155,
      "num_input_tokens_seen": 91085520,
      "step": 42195
    },
    {
      "epoch": 6.884176182707994,
      "grad_norm": 0.31449663639068604,
      "learning_rate": 1.3386575227884443e-05,
      "loss": 0.1577,
      "num_input_tokens_seen": 91094800,
      "step": 42200
    },
    {
      "epoch": 6.8849918433931485,
      "grad_norm": 0.4666809141635895,
      "learning_rate": 1.3380272343043032e-05,
      "loss": 0.0081,
      "num_input_tokens_seen": 91105840,
      "step": 42205
    },
    {
      "epoch": 6.885807504078303,
      "grad_norm": 0.06634185463190079,
      "learning_rate": 1.337397040015571e-05,
      "loss": 0.2304,
      "num_input_tokens_seen": 91116624,
      "step": 42210
    },
    {
      "epoch": 6.886623164763458,
      "grad_norm": 9.664610862731934,
      "learning_rate": 1.336766939973334e-05,
      "loss": 0.1508,
      "num_input_tokens_seen": 91128176,
      "step": 42215
    },
    {
      "epoch": 6.887438825448614,
      "grad_norm": 0.19557489454746246,
      "learning_rate": 1.3361369342286706e-05,
      "loss": 0.219,
      "num_input_tokens_seen": 91138544,
      "step": 42220
    },
    {
      "epoch": 6.888254486133769,
      "grad_norm": 0.1942441165447235,
      "learning_rate": 1.3355070228326533e-05,
      "loss": 0.1022,
      "num_input_tokens_seen": 91148624,
      "step": 42225
    },
    {
      "epoch": 6.8890701468189235,
      "grad_norm": 0.08193667978048325,
      "learning_rate": 1.3348772058363448e-05,
      "loss": 0.075,
      "num_input_tokens_seen": 91158672,
      "step": 42230
    },
    {
      "epoch": 6.889885807504078,
      "grad_norm": 1.206705093383789,
      "learning_rate": 1.334247483290802e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 91169648,
      "step": 42235
    },
    {
      "epoch": 6.890701468189233,
      "grad_norm": 0.5195077061653137,
      "learning_rate": 1.3336178552470729e-05,
      "loss": 0.2202,
      "num_input_tokens_seen": 91180496,
      "step": 42240
    },
    {
      "epoch": 6.891517128874388,
      "grad_norm": 10.173349380493164,
      "learning_rate": 1.332988321756198e-05,
      "loss": 0.0865,
      "num_input_tokens_seen": 91191440,
      "step": 42245
    },
    {
      "epoch": 6.892332789559543,
      "grad_norm": 0.12923379242420197,
      "learning_rate": 1.3323588828692119e-05,
      "loss": 0.107,
      "num_input_tokens_seen": 91201584,
      "step": 42250
    },
    {
      "epoch": 6.8931484502446985,
      "grad_norm": 0.28006529808044434,
      "learning_rate": 1.3317295386371396e-05,
      "loss": 0.0053,
      "num_input_tokens_seen": 91213104,
      "step": 42255
    },
    {
      "epoch": 6.893964110929853,
      "grad_norm": 0.3573981523513794,
      "learning_rate": 1.3311002891109981e-05,
      "loss": 0.0077,
      "num_input_tokens_seen": 91224560,
      "step": 42260
    },
    {
      "epoch": 6.894779771615008,
      "grad_norm": 0.05163750424981117,
      "learning_rate": 1.3304711343417985e-05,
      "loss": 0.1145,
      "num_input_tokens_seen": 91235824,
      "step": 42265
    },
    {
      "epoch": 6.895595432300163,
      "grad_norm": 12.171839714050293,
      "learning_rate": 1.329842074380543e-05,
      "loss": 0.0396,
      "num_input_tokens_seen": 91247600,
      "step": 42270
    },
    {
      "epoch": 6.896411092985318,
      "grad_norm": 0.12578676640987396,
      "learning_rate": 1.3292131092782259e-05,
      "loss": 0.0811,
      "num_input_tokens_seen": 91259120,
      "step": 42275
    },
    {
      "epoch": 6.897226753670473,
      "grad_norm": 0.23984800279140472,
      "learning_rate": 1.328584239085835e-05,
      "loss": 0.0057,
      "num_input_tokens_seen": 91269040,
      "step": 42280
    },
    {
      "epoch": 6.898042414355628,
      "grad_norm": 0.13835729658603668,
      "learning_rate": 1.327955463854349e-05,
      "loss": 0.0879,
      "num_input_tokens_seen": 91280208,
      "step": 42285
    },
    {
      "epoch": 6.898858075040783,
      "grad_norm": 0.08634456992149353,
      "learning_rate": 1.3273267836347403e-05,
      "loss": 0.009,
      "num_input_tokens_seen": 91291088,
      "step": 42290
    },
    {
      "epoch": 6.899673735725938,
      "grad_norm": 0.10134554654359818,
      "learning_rate": 1.3266981984779725e-05,
      "loss": 0.0348,
      "num_input_tokens_seen": 91302032,
      "step": 42295
    },
    {
      "epoch": 6.900489396411093,
      "grad_norm": 0.05540751665830612,
      "learning_rate": 1.3260697084350018e-05,
      "loss": 0.0072,
      "num_input_tokens_seen": 91313552,
      "step": 42300
    },
    {
      "epoch": 6.901305057096248,
      "grad_norm": 0.057743240147829056,
      "learning_rate": 1.3254413135567773e-05,
      "loss": 0.0076,
      "num_input_tokens_seen": 91323728,
      "step": 42305
    },
    {
      "epoch": 6.902120717781403,
      "grad_norm": 0.11667633801698685,
      "learning_rate": 1.3248130138942394e-05,
      "loss": 0.1382,
      "num_input_tokens_seen": 91334960,
      "step": 42310
    },
    {
      "epoch": 6.902936378466558,
      "grad_norm": 0.08370362222194672,
      "learning_rate": 1.3241848094983206e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 91346896,
      "step": 42315
    },
    {
      "epoch": 6.903752039151713,
      "grad_norm": 0.03880749270319939,
      "learning_rate": 1.3235567004199481e-05,
      "loss": 0.0251,
      "num_input_tokens_seen": 91357168,
      "step": 42320
    },
    {
      "epoch": 6.904567699836868,
      "grad_norm": 5.930319786071777,
      "learning_rate": 1.3229286867100388e-05,
      "loss": 0.1583,
      "num_input_tokens_seen": 91368688,
      "step": 42325
    },
    {
      "epoch": 6.9053833605220225,
      "grad_norm": 0.18616336584091187,
      "learning_rate": 1.3223007684195021e-05,
      "loss": 0.0103,
      "num_input_tokens_seen": 91379600,
      "step": 42330
    },
    {
      "epoch": 6.906199021207177,
      "grad_norm": 0.08522260934114456,
      "learning_rate": 1.321672945599241e-05,
      "loss": 0.0102,
      "num_input_tokens_seen": 91389872,
      "step": 42335
    },
    {
      "epoch": 6.907014681892333,
      "grad_norm": 4.693546295166016,
      "learning_rate": 1.3210452183001497e-05,
      "loss": 0.2053,
      "num_input_tokens_seen": 91400944,
      "step": 42340
    },
    {
      "epoch": 6.907830342577488,
      "grad_norm": 2.370194673538208,
      "learning_rate": 1.320417586573115e-05,
      "loss": 0.2174,
      "num_input_tokens_seen": 91412080,
      "step": 42345
    },
    {
      "epoch": 6.908646003262643,
      "grad_norm": 0.07777683436870575,
      "learning_rate": 1.3197900504690161e-05,
      "loss": 0.0052,
      "num_input_tokens_seen": 91423952,
      "step": 42350
    },
    {
      "epoch": 6.9094616639477975,
      "grad_norm": 0.10456414520740509,
      "learning_rate": 1.3191626100387238e-05,
      "loss": 0.182,
      "num_input_tokens_seen": 91434480,
      "step": 42355
    },
    {
      "epoch": 6.910277324632952,
      "grad_norm": 1.8496055603027344,
      "learning_rate": 1.3185352653331018e-05,
      "loss": 0.1313,
      "num_input_tokens_seen": 91444464,
      "step": 42360
    },
    {
      "epoch": 6.911092985318108,
      "grad_norm": 0.10061901807785034,
      "learning_rate": 1.3179080164030059e-05,
      "loss": 0.1324,
      "num_input_tokens_seen": 91453104,
      "step": 42365
    },
    {
      "epoch": 6.911908646003263,
      "grad_norm": 0.04748468101024628,
      "learning_rate": 1.317280863299284e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 91463120,
      "step": 42370
    },
    {
      "epoch": 6.912724306688418,
      "grad_norm": 0.10149021446704865,
      "learning_rate": 1.3166538060727765e-05,
      "loss": 0.0154,
      "num_input_tokens_seen": 91473104,
      "step": 42375
    },
    {
      "epoch": 6.9135399673735725,
      "grad_norm": 4.338562488555908,
      "learning_rate": 1.3160268447743157e-05,
      "loss": 0.3162,
      "num_input_tokens_seen": 91484304,
      "step": 42380
    },
    {
      "epoch": 6.914355628058727,
      "grad_norm": 0.06958422064781189,
      "learning_rate": 1.3153999794547261e-05,
      "loss": 0.0054,
      "num_input_tokens_seen": 91495536,
      "step": 42385
    },
    {
      "epoch": 6.915171288743883,
      "grad_norm": 0.06562705338001251,
      "learning_rate": 1.3147732101648242e-05,
      "loss": 0.0159,
      "num_input_tokens_seen": 91505392,
      "step": 42390
    },
    {
      "epoch": 6.915986949429038,
      "grad_norm": 0.04994618520140648,
      "learning_rate": 1.31414653695542e-05,
      "loss": 0.0056,
      "num_input_tokens_seen": 91517200,
      "step": 42395
    },
    {
      "epoch": 6.916802610114193,
      "grad_norm": 0.043880391865968704,
      "learning_rate": 1.3135199598773152e-05,
      "loss": 0.0593,
      "num_input_tokens_seen": 91528592,
      "step": 42400
    },
    {
      "epoch": 6.917618270799347,
      "grad_norm": 0.11109756678342819,
      "learning_rate": 1.3128934789813021e-05,
      "loss": 0.0039,
      "num_input_tokens_seen": 91539216,
      "step": 42405
    },
    {
      "epoch": 6.918433931484502,
      "grad_norm": 0.024895496666431427,
      "learning_rate": 1.312267094318167e-05,
      "loss": 0.268,
      "num_input_tokens_seen": 91549616,
      "step": 42410
    },
    {
      "epoch": 6.919249592169657,
      "grad_norm": 0.08413500338792801,
      "learning_rate": 1.3116408059386881e-05,
      "loss": 0.1347,
      "num_input_tokens_seen": 91560496,
      "step": 42415
    },
    {
      "epoch": 6.920065252854813,
      "grad_norm": 0.23153434693813324,
      "learning_rate": 1.3110146138936335e-05,
      "loss": 0.0992,
      "num_input_tokens_seen": 91571408,
      "step": 42420
    },
    {
      "epoch": 6.920880913539968,
      "grad_norm": 0.1570703238248825,
      "learning_rate": 1.3103885182337688e-05,
      "loss": 0.0072,
      "num_input_tokens_seen": 91580496,
      "step": 42425
    },
    {
      "epoch": 6.921696574225122,
      "grad_norm": 0.08957034349441528,
      "learning_rate": 1.3097625190098464e-05,
      "loss": 0.1445,
      "num_input_tokens_seen": 91591984,
      "step": 42430
    },
    {
      "epoch": 6.922512234910277,
      "grad_norm": 11.272125244140625,
      "learning_rate": 1.3091366162726136e-05,
      "loss": 0.0367,
      "num_input_tokens_seen": 91602672,
      "step": 42435
    },
    {
      "epoch": 6.923327895595432,
      "grad_norm": 27.047489166259766,
      "learning_rate": 1.3085108100728089e-05,
      "loss": 0.0175,
      "num_input_tokens_seen": 91612752,
      "step": 42440
    },
    {
      "epoch": 6.924143556280587,
      "grad_norm": 0.09969615936279297,
      "learning_rate": 1.3078851004611636e-05,
      "loss": 0.0091,
      "num_input_tokens_seen": 91623184,
      "step": 42445
    },
    {
      "epoch": 6.924959216965743,
      "grad_norm": 0.0777190700173378,
      "learning_rate": 1.3072594874884008e-05,
      "loss": 0.0797,
      "num_input_tokens_seen": 91633488,
      "step": 42450
    },
    {
      "epoch": 6.925774877650897,
      "grad_norm": 0.27353769540786743,
      "learning_rate": 1.3066339712052359e-05,
      "loss": 0.2706,
      "num_input_tokens_seen": 91645456,
      "step": 42455
    },
    {
      "epoch": 6.926590538336052,
      "grad_norm": 0.11094507575035095,
      "learning_rate": 1.3060085516623763e-05,
      "loss": 0.1277,
      "num_input_tokens_seen": 91656016,
      "step": 42460
    },
    {
      "epoch": 6.927406199021207,
      "grad_norm": 4.023979663848877,
      "learning_rate": 1.3053832289105216e-05,
      "loss": 0.1015,
      "num_input_tokens_seen": 91667728,
      "step": 42465
    },
    {
      "epoch": 6.928221859706362,
      "grad_norm": 0.09130234271287918,
      "learning_rate": 1.3047580030003642e-05,
      "loss": 0.0736,
      "num_input_tokens_seen": 91679280,
      "step": 42470
    },
    {
      "epoch": 6.9290375203915175,
      "grad_norm": 0.11625941097736359,
      "learning_rate": 1.3041328739825873e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 91690032,
      "step": 42475
    },
    {
      "epoch": 6.929853181076672,
      "grad_norm": 4.393698692321777,
      "learning_rate": 1.3035078419078675e-05,
      "loss": 0.0745,
      "num_input_tokens_seen": 91700944,
      "step": 42480
    },
    {
      "epoch": 6.930668841761827,
      "grad_norm": 5.182872772216797,
      "learning_rate": 1.302882906826873e-05,
      "loss": 0.0418,
      "num_input_tokens_seen": 91710096,
      "step": 42485
    },
    {
      "epoch": 6.931484502446982,
      "grad_norm": 0.2051047831773758,
      "learning_rate": 1.3022580687902641e-05,
      "loss": 0.1138,
      "num_input_tokens_seen": 91719856,
      "step": 42490
    },
    {
      "epoch": 6.932300163132137,
      "grad_norm": 3.923438787460327,
      "learning_rate": 1.3016333278486936e-05,
      "loss": 0.2626,
      "num_input_tokens_seen": 91730448,
      "step": 42495
    },
    {
      "epoch": 6.933115823817292,
      "grad_norm": 4.099771499633789,
      "learning_rate": 1.301008684052806e-05,
      "loss": 0.2064,
      "num_input_tokens_seen": 91742256,
      "step": 42500
    },
    {
      "epoch": 6.933931484502447,
      "grad_norm": 19.56888771057129,
      "learning_rate": 1.300384137453238e-05,
      "loss": 0.0806,
      "num_input_tokens_seen": 91752848,
      "step": 42505
    },
    {
      "epoch": 6.934747145187602,
      "grad_norm": 0.1936420500278473,
      "learning_rate": 1.2997596881006185e-05,
      "loss": 0.0705,
      "num_input_tokens_seen": 91764240,
      "step": 42510
    },
    {
      "epoch": 6.935562805872757,
      "grad_norm": 0.12515167891979218,
      "learning_rate": 1.2991353360455688e-05,
      "loss": 0.0289,
      "num_input_tokens_seen": 91775696,
      "step": 42515
    },
    {
      "epoch": 6.936378466557912,
      "grad_norm": 0.05261050537228584,
      "learning_rate": 1.298511081338702e-05,
      "loss": 0.0052,
      "num_input_tokens_seen": 91786576,
      "step": 42520
    },
    {
      "epoch": 6.937194127243067,
      "grad_norm": 0.09552355855703354,
      "learning_rate": 1.2978869240306219e-05,
      "loss": 0.0104,
      "num_input_tokens_seen": 91796880,
      "step": 42525
    },
    {
      "epoch": 6.938009787928221,
      "grad_norm": 0.08294784277677536,
      "learning_rate": 1.2972628641719286e-05,
      "loss": 0.1157,
      "num_input_tokens_seen": 91807344,
      "step": 42530
    },
    {
      "epoch": 6.938825448613377,
      "grad_norm": 0.05821099877357483,
      "learning_rate": 1.2966389018132097e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 91818640,
      "step": 42535
    },
    {
      "epoch": 6.939641109298532,
      "grad_norm": 0.10651091486215591,
      "learning_rate": 1.2960150370050475e-05,
      "loss": 0.0031,
      "num_input_tokens_seen": 91829616,
      "step": 42540
    },
    {
      "epoch": 6.940456769983687,
      "grad_norm": 4.577032089233398,
      "learning_rate": 1.2953912697980152e-05,
      "loss": 0.104,
      "num_input_tokens_seen": 91839376,
      "step": 42545
    },
    {
      "epoch": 6.941272430668842,
      "grad_norm": 0.03873153030872345,
      "learning_rate": 1.2947676002426789e-05,
      "loss": 0.1172,
      "num_input_tokens_seen": 91849776,
      "step": 42550
    },
    {
      "epoch": 6.942088091353996,
      "grad_norm": 10.644501686096191,
      "learning_rate": 1.2941440283895961e-05,
      "loss": 0.301,
      "num_input_tokens_seen": 91861520,
      "step": 42555
    },
    {
      "epoch": 6.942903752039152,
      "grad_norm": 0.07911406457424164,
      "learning_rate": 1.2935205542893158e-05,
      "loss": 0.1147,
      "num_input_tokens_seen": 91872880,
      "step": 42560
    },
    {
      "epoch": 6.943719412724307,
      "grad_norm": 0.13447058200836182,
      "learning_rate": 1.2928971779923821e-05,
      "loss": 0.1414,
      "num_input_tokens_seen": 91884016,
      "step": 42565
    },
    {
      "epoch": 6.944535073409462,
      "grad_norm": 8.785443305969238,
      "learning_rate": 1.2922738995493277e-05,
      "loss": 0.3427,
      "num_input_tokens_seen": 91895504,
      "step": 42570
    },
    {
      "epoch": 6.945350734094617,
      "grad_norm": 0.05817743390798569,
      "learning_rate": 1.2916507190106792e-05,
      "loss": 0.0049,
      "num_input_tokens_seen": 91906288,
      "step": 42575
    },
    {
      "epoch": 6.946166394779771,
      "grad_norm": 3.8127334117889404,
      "learning_rate": 1.2910276364269546e-05,
      "loss": 0.2059,
      "num_input_tokens_seen": 91916752,
      "step": 42580
    },
    {
      "epoch": 6.946982055464927,
      "grad_norm": 4.929322242736816,
      "learning_rate": 1.2904046518486637e-05,
      "loss": 0.1455,
      "num_input_tokens_seen": 91926992,
      "step": 42585
    },
    {
      "epoch": 6.947797716150082,
      "grad_norm": 0.04898161441087723,
      "learning_rate": 1.2897817653263095e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 91935504,
      "step": 42590
    },
    {
      "epoch": 6.948613376835237,
      "grad_norm": 4.509163856506348,
      "learning_rate": 1.2891589769103856e-05,
      "loss": 0.375,
      "num_input_tokens_seen": 91946320,
      "step": 42595
    },
    {
      "epoch": 6.9494290375203915,
      "grad_norm": 7.143014907836914,
      "learning_rate": 1.288536286651379e-05,
      "loss": 0.0135,
      "num_input_tokens_seen": 91957008,
      "step": 42600
    },
    {
      "epoch": 6.950244698205546,
      "grad_norm": 3.1989221572875977,
      "learning_rate": 1.2879136945997677e-05,
      "loss": 0.1246,
      "num_input_tokens_seen": 91967824,
      "step": 42605
    },
    {
      "epoch": 6.951060358890701,
      "grad_norm": 4.313570022583008,
      "learning_rate": 1.2872912008060228e-05,
      "loss": 0.1304,
      "num_input_tokens_seen": 91979024,
      "step": 42610
    },
    {
      "epoch": 6.951876019575856,
      "grad_norm": 0.044972293078899384,
      "learning_rate": 1.286668805320606e-05,
      "loss": 0.1614,
      "num_input_tokens_seen": 91990288,
      "step": 42615
    },
    {
      "epoch": 6.952691680261012,
      "grad_norm": 0.11908325552940369,
      "learning_rate": 1.2860465081939727e-05,
      "loss": 0.1117,
      "num_input_tokens_seen": 92000592,
      "step": 42620
    },
    {
      "epoch": 6.9535073409461665,
      "grad_norm": 0.04871833324432373,
      "learning_rate": 1.2854243094765683e-05,
      "loss": 0.1082,
      "num_input_tokens_seen": 92009520,
      "step": 42625
    },
    {
      "epoch": 6.954323001631321,
      "grad_norm": 0.04810227081179619,
      "learning_rate": 1.2848022092188328e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 92020848,
      "step": 42630
    },
    {
      "epoch": 6.955138662316476,
      "grad_norm": 4.308464050292969,
      "learning_rate": 1.2841802074711945e-05,
      "loss": 0.1247,
      "num_input_tokens_seen": 92030384,
      "step": 42635
    },
    {
      "epoch": 6.955954323001631,
      "grad_norm": 3.530879020690918,
      "learning_rate": 1.2835583042840788e-05,
      "loss": 0.2087,
      "num_input_tokens_seen": 92040528,
      "step": 42640
    },
    {
      "epoch": 6.956769983686787,
      "grad_norm": 0.21271125972270966,
      "learning_rate": 1.282936499707899e-05,
      "loss": 0.0844,
      "num_input_tokens_seen": 92052368,
      "step": 42645
    },
    {
      "epoch": 6.9575856443719415,
      "grad_norm": 0.08056087046861649,
      "learning_rate": 1.282314793793062e-05,
      "loss": 0.1743,
      "num_input_tokens_seen": 92063152,
      "step": 42650
    },
    {
      "epoch": 6.958401305057096,
      "grad_norm": 0.16285933554172516,
      "learning_rate": 1.2816931865899662e-05,
      "loss": 0.004,
      "num_input_tokens_seen": 92073296,
      "step": 42655
    },
    {
      "epoch": 6.959216965742251,
      "grad_norm": 1.8300361633300781,
      "learning_rate": 1.2810716781490024e-05,
      "loss": 0.0709,
      "num_input_tokens_seen": 92084176,
      "step": 42660
    },
    {
      "epoch": 6.960032626427406,
      "grad_norm": 0.6765586733818054,
      "learning_rate": 1.2804502685205532e-05,
      "loss": 0.0549,
      "num_input_tokens_seen": 92095344,
      "step": 42665
    },
    {
      "epoch": 6.960848287112562,
      "grad_norm": 0.2839624285697937,
      "learning_rate": 1.2798289577549932e-05,
      "loss": 0.006,
      "num_input_tokens_seen": 92106384,
      "step": 42670
    },
    {
      "epoch": 6.9616639477977165,
      "grad_norm": 0.16660496592521667,
      "learning_rate": 1.2792077459026886e-05,
      "loss": 0.007,
      "num_input_tokens_seen": 92117104,
      "step": 42675
    },
    {
      "epoch": 6.962479608482871,
      "grad_norm": 0.31639882922172546,
      "learning_rate": 1.278586633013999e-05,
      "loss": 0.027,
      "num_input_tokens_seen": 92127600,
      "step": 42680
    },
    {
      "epoch": 6.963295269168026,
      "grad_norm": 0.11238505691289902,
      "learning_rate": 1.2779656191392736e-05,
      "loss": 0.0747,
      "num_input_tokens_seen": 92138448,
      "step": 42685
    },
    {
      "epoch": 6.964110929853181,
      "grad_norm": 0.08163265883922577,
      "learning_rate": 1.2773447043288561e-05,
      "loss": 0.004,
      "num_input_tokens_seen": 92148144,
      "step": 42690
    },
    {
      "epoch": 6.964926590538336,
      "grad_norm": 4.075665473937988,
      "learning_rate": 1.2767238886330805e-05,
      "loss": 0.2935,
      "num_input_tokens_seen": 92158640,
      "step": 42695
    },
    {
      "epoch": 6.9657422512234906,
      "grad_norm": 0.09558428823947906,
      "learning_rate": 1.2761031721022732e-05,
      "loss": 0.2357,
      "num_input_tokens_seen": 92169008,
      "step": 42700
    },
    {
      "epoch": 6.966557911908646,
      "grad_norm": 0.06287585943937302,
      "learning_rate": 1.2754825547867519e-05,
      "loss": 0.0745,
      "num_input_tokens_seen": 92180432,
      "step": 42705
    },
    {
      "epoch": 6.967373572593801,
      "grad_norm": 0.18176084756851196,
      "learning_rate": 1.2748620367368286e-05,
      "loss": 0.1636,
      "num_input_tokens_seen": 92191376,
      "step": 42710
    },
    {
      "epoch": 6.968189233278956,
      "grad_norm": 0.0947713851928711,
      "learning_rate": 1.2742416180028053e-05,
      "loss": 0.0105,
      "num_input_tokens_seen": 92202256,
      "step": 42715
    },
    {
      "epoch": 6.969004893964111,
      "grad_norm": 0.1489112675189972,
      "learning_rate": 1.2736212986349755e-05,
      "loss": 0.0027,
      "num_input_tokens_seen": 92212400,
      "step": 42720
    },
    {
      "epoch": 6.9698205546492655,
      "grad_norm": 3.682657241821289,
      "learning_rate": 1.2730010786836261e-05,
      "loss": 0.3949,
      "num_input_tokens_seen": 92223216,
      "step": 42725
    },
    {
      "epoch": 6.970636215334421,
      "grad_norm": 0.04183153063058853,
      "learning_rate": 1.2723809581990348e-05,
      "loss": 0.0655,
      "num_input_tokens_seen": 92234928,
      "step": 42730
    },
    {
      "epoch": 6.971451876019576,
      "grad_norm": 0.15970873832702637,
      "learning_rate": 1.271760937231472e-05,
      "loss": 0.1058,
      "num_input_tokens_seen": 92245392,
      "step": 42735
    },
    {
      "epoch": 6.972267536704731,
      "grad_norm": 4.388833045959473,
      "learning_rate": 1.2711410158311987e-05,
      "loss": 0.1514,
      "num_input_tokens_seen": 92255824,
      "step": 42740
    },
    {
      "epoch": 6.973083197389886,
      "grad_norm": 0.14887931942939758,
      "learning_rate": 1.2705211940484707e-05,
      "loss": 0.0071,
      "num_input_tokens_seen": 92265904,
      "step": 42745
    },
    {
      "epoch": 6.9738988580750405,
      "grad_norm": 0.0658448114991188,
      "learning_rate": 1.2699014719335329e-05,
      "loss": 0.0976,
      "num_input_tokens_seen": 92276176,
      "step": 42750
    },
    {
      "epoch": 6.974714518760196,
      "grad_norm": 0.06925688683986664,
      "learning_rate": 1.2692818495366236e-05,
      "loss": 0.0708,
      "num_input_tokens_seen": 92285744,
      "step": 42755
    },
    {
      "epoch": 6.975530179445351,
      "grad_norm": 0.17164693772792816,
      "learning_rate": 1.2686623269079717e-05,
      "loss": 0.0052,
      "num_input_tokens_seen": 92298128,
      "step": 42760
    },
    {
      "epoch": 6.976345840130506,
      "grad_norm": 0.07573288679122925,
      "learning_rate": 1.2680429040977998e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 92308752,
      "step": 42765
    },
    {
      "epoch": 6.977161500815661,
      "grad_norm": 3.413907527923584,
      "learning_rate": 1.2674235811563206e-05,
      "loss": 0.2298,
      "num_input_tokens_seen": 92319632,
      "step": 42770
    },
    {
      "epoch": 6.9779771615008155,
      "grad_norm": 0.15602654218673706,
      "learning_rate": 1.2668043581337401e-05,
      "loss": 0.0099,
      "num_input_tokens_seen": 92330576,
      "step": 42775
    },
    {
      "epoch": 6.97879282218597,
      "grad_norm": 2.75740909576416,
      "learning_rate": 1.2661852350802556e-05,
      "loss": 0.0679,
      "num_input_tokens_seen": 92340304,
      "step": 42780
    },
    {
      "epoch": 6.979608482871125,
      "grad_norm": 0.07793080806732178,
      "learning_rate": 1.2655662120460564e-05,
      "loss": 0.1154,
      "num_input_tokens_seen": 92350576,
      "step": 42785
    },
    {
      "epoch": 6.980424143556281,
      "grad_norm": 1.0050971508026123,
      "learning_rate": 1.2649472890813232e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 92359664,
      "step": 42790
    },
    {
      "epoch": 6.981239804241436,
      "grad_norm": 0.11205387860536575,
      "learning_rate": 1.2643284662362295e-05,
      "loss": 0.0571,
      "num_input_tokens_seen": 92370032,
      "step": 42795
    },
    {
      "epoch": 6.9820554649265905,
      "grad_norm": 0.12443529069423676,
      "learning_rate": 1.2637097435609402e-05,
      "loss": 0.203,
      "num_input_tokens_seen": 92380976,
      "step": 42800
    },
    {
      "epoch": 6.982871125611745,
      "grad_norm": 0.21364746987819672,
      "learning_rate": 1.2630911211056116e-05,
      "loss": 0.1235,
      "num_input_tokens_seen": 92393136,
      "step": 42805
    },
    {
      "epoch": 6.9836867862969,
      "grad_norm": 0.053184982389211655,
      "learning_rate": 1.2624725989203929e-05,
      "loss": 0.126,
      "num_input_tokens_seen": 92404208,
      "step": 42810
    },
    {
      "epoch": 6.984502446982056,
      "grad_norm": 0.05223655328154564,
      "learning_rate": 1.2618541770554243e-05,
      "loss": 0.1379,
      "num_input_tokens_seen": 92415152,
      "step": 42815
    },
    {
      "epoch": 6.985318107667211,
      "grad_norm": 0.09687217324972153,
      "learning_rate": 1.2612358555608388e-05,
      "loss": 0.0041,
      "num_input_tokens_seen": 92426192,
      "step": 42820
    },
    {
      "epoch": 6.986133768352365,
      "grad_norm": 0.08545750379562378,
      "learning_rate": 1.2606176344867598e-05,
      "loss": 0.0904,
      "num_input_tokens_seen": 92438064,
      "step": 42825
    },
    {
      "epoch": 6.98694942903752,
      "grad_norm": 0.6400485634803772,
      "learning_rate": 1.2599995138833043e-05,
      "loss": 0.0087,
      "num_input_tokens_seen": 92447280,
      "step": 42830
    },
    {
      "epoch": 6.987765089722675,
      "grad_norm": 0.17474797368049622,
      "learning_rate": 1.25938149380058e-05,
      "loss": 0.1147,
      "num_input_tokens_seen": 92458416,
      "step": 42835
    },
    {
      "epoch": 6.988580750407831,
      "grad_norm": 0.10868343710899353,
      "learning_rate": 1.258763574288686e-05,
      "loss": 0.0848,
      "num_input_tokens_seen": 92469680,
      "step": 42840
    },
    {
      "epoch": 6.989396411092986,
      "grad_norm": 0.5525995492935181,
      "learning_rate": 1.2581457553977144e-05,
      "loss": 0.0097,
      "num_input_tokens_seen": 92480592,
      "step": 42845
    },
    {
      "epoch": 6.99021207177814,
      "grad_norm": 0.22638683021068573,
      "learning_rate": 1.2575280371777496e-05,
      "loss": 0.0082,
      "num_input_tokens_seen": 92492336,
      "step": 42850
    },
    {
      "epoch": 6.991027732463295,
      "grad_norm": 0.4266146719455719,
      "learning_rate": 1.2569104196788665e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 92502608,
      "step": 42855
    },
    {
      "epoch": 6.99184339314845,
      "grad_norm": 0.11602754145860672,
      "learning_rate": 1.256292902951132e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 92513520,
      "step": 42860
    },
    {
      "epoch": 6.992659053833605,
      "grad_norm": 0.20767870545387268,
      "learning_rate": 1.2556754870446053e-05,
      "loss": 0.0556,
      "num_input_tokens_seen": 92524656,
      "step": 42865
    },
    {
      "epoch": 6.993474714518761,
      "grad_norm": 7.158193111419678,
      "learning_rate": 1.255058172009337e-05,
      "loss": 0.0401,
      "num_input_tokens_seen": 92535408,
      "step": 42870
    },
    {
      "epoch": 6.994290375203915,
      "grad_norm": 0.16819290816783905,
      "learning_rate": 1.2544409578953697e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 92545744,
      "step": 42875
    },
    {
      "epoch": 6.99510603588907,
      "grad_norm": 5.802420139312744,
      "learning_rate": 1.253823844752739e-05,
      "loss": 0.1006,
      "num_input_tokens_seen": 92556432,
      "step": 42880
    },
    {
      "epoch": 6.995921696574225,
      "grad_norm": 0.24903814494609833,
      "learning_rate": 1.25320683263147e-05,
      "loss": 0.016,
      "num_input_tokens_seen": 92568592,
      "step": 42885
    },
    {
      "epoch": 6.99673735725938,
      "grad_norm": 0.1295042634010315,
      "learning_rate": 1.2525899215815818e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 92579632,
      "step": 42890
    },
    {
      "epoch": 6.997553017944535,
      "grad_norm": 0.09052778035402298,
      "learning_rate": 1.2519731116530837e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 92591344,
      "step": 42895
    },
    {
      "epoch": 6.99836867862969,
      "grad_norm": 0.0461747832596302,
      "learning_rate": 1.2513564028959777e-05,
      "loss": 0.0991,
      "num_input_tokens_seen": 92601264,
      "step": 42900
    },
    {
      "epoch": 6.999184339314845,
      "grad_norm": 0.690471351146698,
      "learning_rate": 1.2507397953602574e-05,
      "loss": 0.1705,
      "num_input_tokens_seen": 92612464,
      "step": 42905
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.1497431844472885,
      "learning_rate": 1.2501232890959075e-05,
      "loss": 0.0286,
      "num_input_tokens_seen": 92621824,
      "step": 42910
    },
    {
      "epoch": 7.0,
      "eval_loss": 0.18837185204029083,
      "eval_runtime": 132.8718,
      "eval_samples_per_second": 20.508,
      "eval_steps_per_second": 5.133,
      "num_input_tokens_seen": 92621824,
      "step": 42910
    },
    {
      "epoch": 7.000815660685155,
      "grad_norm": 0.0485120564699173,
      "learning_rate": 1.2495068841529058e-05,
      "loss": 0.0976,
      "num_input_tokens_seen": 92633024,
      "step": 42915
    },
    {
      "epoch": 7.00163132137031,
      "grad_norm": 0.08123528957366943,
      "learning_rate": 1.248890580581221e-05,
      "loss": 0.0077,
      "num_input_tokens_seen": 92644768,
      "step": 42920
    },
    {
      "epoch": 7.002446982055465,
      "grad_norm": 0.024659737944602966,
      "learning_rate": 1.248274378430814e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 92654240,
      "step": 42925
    },
    {
      "epoch": 7.00326264274062,
      "grad_norm": 0.18686439096927643,
      "learning_rate": 1.2476582777516368e-05,
      "loss": 0.0041,
      "num_input_tokens_seen": 92664288,
      "step": 42930
    },
    {
      "epoch": 7.004078303425775,
      "grad_norm": 0.0459061935544014,
      "learning_rate": 1.2470422785936339e-05,
      "loss": 0.174,
      "num_input_tokens_seen": 92675104,
      "step": 42935
    },
    {
      "epoch": 7.00489396411093,
      "grad_norm": 0.1195080503821373,
      "learning_rate": 1.2464263810067417e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 92685056,
      "step": 42940
    },
    {
      "epoch": 7.005709624796085,
      "grad_norm": 17.653226852416992,
      "learning_rate": 1.2458105850408874e-05,
      "loss": 0.0493,
      "num_input_tokens_seen": 92696000,
      "step": 42945
    },
    {
      "epoch": 7.006525285481239,
      "grad_norm": 0.033694829791784286,
      "learning_rate": 1.2451948907459907e-05,
      "loss": 0.1407,
      "num_input_tokens_seen": 92706272,
      "step": 42950
    },
    {
      "epoch": 7.007340946166395,
      "grad_norm": 0.027108220383524895,
      "learning_rate": 1.2445792981719622e-05,
      "loss": 0.1758,
      "num_input_tokens_seen": 92717120,
      "step": 42955
    },
    {
      "epoch": 7.00815660685155,
      "grad_norm": 0.09192747622728348,
      "learning_rate": 1.2439638073687065e-05,
      "loss": 0.0031,
      "num_input_tokens_seen": 92726432,
      "step": 42960
    },
    {
      "epoch": 7.008972267536705,
      "grad_norm": 0.1266232281923294,
      "learning_rate": 1.2433484183861178e-05,
      "loss": 0.1223,
      "num_input_tokens_seen": 92736896,
      "step": 42965
    },
    {
      "epoch": 7.00978792822186,
      "grad_norm": 0.0833207368850708,
      "learning_rate": 1.2427331312740822e-05,
      "loss": 0.0136,
      "num_input_tokens_seen": 92747904,
      "step": 42970
    },
    {
      "epoch": 7.010603588907014,
      "grad_norm": 0.15957383811473846,
      "learning_rate": 1.2421179460824787e-05,
      "loss": 0.1064,
      "num_input_tokens_seen": 92757920,
      "step": 42975
    },
    {
      "epoch": 7.011419249592169,
      "grad_norm": 0.02515152096748352,
      "learning_rate": 1.2415028628611767e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 92769664,
      "step": 42980
    },
    {
      "epoch": 7.012234910277325,
      "grad_norm": 0.10467785596847534,
      "learning_rate": 1.2408878816600384e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 92780256,
      "step": 42985
    },
    {
      "epoch": 7.01305057096248,
      "grad_norm": 12.657437324523926,
      "learning_rate": 1.2402730025289166e-05,
      "loss": 0.0618,
      "num_input_tokens_seen": 92791744,
      "step": 42990
    },
    {
      "epoch": 7.013866231647635,
      "grad_norm": 11.175701141357422,
      "learning_rate": 1.2396582255176575e-05,
      "loss": 0.0851,
      "num_input_tokens_seen": 92800928,
      "step": 42995
    },
    {
      "epoch": 7.014681892332789,
      "grad_norm": 0.08104083687067032,
      "learning_rate": 1.2390435506760973e-05,
      "loss": 0.0479,
      "num_input_tokens_seen": 92812352,
      "step": 43000
    },
    {
      "epoch": 7.015497553017944,
      "grad_norm": 0.2647564709186554,
      "learning_rate": 1.238428978054065e-05,
      "loss": 0.0471,
      "num_input_tokens_seen": 92822944,
      "step": 43005
    },
    {
      "epoch": 7.0163132137031,
      "grad_norm": 0.14446577429771423,
      "learning_rate": 1.2378145077013808e-05,
      "loss": 0.1382,
      "num_input_tokens_seen": 92832960,
      "step": 43010
    },
    {
      "epoch": 7.017128874388255,
      "grad_norm": 0.05882642790675163,
      "learning_rate": 1.237200139667857e-05,
      "loss": 0.2418,
      "num_input_tokens_seen": 92843584,
      "step": 43015
    },
    {
      "epoch": 7.0179445350734095,
      "grad_norm": 0.14573557674884796,
      "learning_rate": 1.2365858740032962e-05,
      "loss": 0.0055,
      "num_input_tokens_seen": 92853440,
      "step": 43020
    },
    {
      "epoch": 7.018760195758564,
      "grad_norm": 0.1150248795747757,
      "learning_rate": 1.2359717107574959e-05,
      "loss": 0.0524,
      "num_input_tokens_seen": 92862112,
      "step": 43025
    },
    {
      "epoch": 7.019575856443719,
      "grad_norm": 0.024813566356897354,
      "learning_rate": 1.2353576499802425e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 92872000,
      "step": 43030
    },
    {
      "epoch": 7.020391517128874,
      "grad_norm": 0.09132643789052963,
      "learning_rate": 1.2347436917213145e-05,
      "loss": 0.0039,
      "num_input_tokens_seen": 92881856,
      "step": 43035
    },
    {
      "epoch": 7.02120717781403,
      "grad_norm": 0.10275115072727203,
      "learning_rate": 1.2341298360304828e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 92892768,
      "step": 43040
    },
    {
      "epoch": 7.0220228384991845,
      "grad_norm": 0.059842515736818314,
      "learning_rate": 1.2335160829575096e-05,
      "loss": 0.0961,
      "num_input_tokens_seen": 92904384,
      "step": 43045
    },
    {
      "epoch": 7.022838499184339,
      "grad_norm": 0.16072654724121094,
      "learning_rate": 1.2329024325521488e-05,
      "loss": 0.0097,
      "num_input_tokens_seen": 92916160,
      "step": 43050
    },
    {
      "epoch": 7.023654159869494,
      "grad_norm": 0.14826837182044983,
      "learning_rate": 1.2322888848641458e-05,
      "loss": 0.12,
      "num_input_tokens_seen": 92924640,
      "step": 43055
    },
    {
      "epoch": 7.024469820554649,
      "grad_norm": 0.04622304067015648,
      "learning_rate": 1.2316754399432374e-05,
      "loss": 0.0033,
      "num_input_tokens_seen": 92935264,
      "step": 43060
    },
    {
      "epoch": 7.025285481239805,
      "grad_norm": 0.05373028293251991,
      "learning_rate": 1.231062097839154e-05,
      "loss": 0.0033,
      "num_input_tokens_seen": 92947232,
      "step": 43065
    },
    {
      "epoch": 7.0261011419249595,
      "grad_norm": 0.031497180461883545,
      "learning_rate": 1.2304488586016156e-05,
      "loss": 0.1275,
      "num_input_tokens_seen": 92958816,
      "step": 43070
    },
    {
      "epoch": 7.026916802610114,
      "grad_norm": 2.944779634475708,
      "learning_rate": 1.2298357222803341e-05,
      "loss": 0.0124,
      "num_input_tokens_seen": 92969824,
      "step": 43075
    },
    {
      "epoch": 7.027732463295269,
      "grad_norm": 0.08053677529096603,
      "learning_rate": 1.2292226889250142e-05,
      "loss": 0.0715,
      "num_input_tokens_seen": 92980928,
      "step": 43080
    },
    {
      "epoch": 7.028548123980424,
      "grad_norm": 0.4775858521461487,
      "learning_rate": 1.2286097585853507e-05,
      "loss": 0.0756,
      "num_input_tokens_seen": 92992480,
      "step": 43085
    },
    {
      "epoch": 7.029363784665579,
      "grad_norm": 0.06219562515616417,
      "learning_rate": 1.2279969313110313e-05,
      "loss": 0.1079,
      "num_input_tokens_seen": 93003232,
      "step": 43090
    },
    {
      "epoch": 7.0301794453507345,
      "grad_norm": 5.719261169433594,
      "learning_rate": 1.2273842071517344e-05,
      "loss": 0.1312,
      "num_input_tokens_seen": 93013952,
      "step": 43095
    },
    {
      "epoch": 7.030995106035889,
      "grad_norm": 37.29789733886719,
      "learning_rate": 1.2267715861571311e-05,
      "loss": 0.0797,
      "num_input_tokens_seen": 93024064,
      "step": 43100
    },
    {
      "epoch": 7.031810766721044,
      "grad_norm": 0.1041211411356926,
      "learning_rate": 1.2261590683768831e-05,
      "loss": 0.003,
      "num_input_tokens_seen": 93036128,
      "step": 43105
    },
    {
      "epoch": 7.032626427406199,
      "grad_norm": 9.541900634765625,
      "learning_rate": 1.2255466538606447e-05,
      "loss": 0.1325,
      "num_input_tokens_seen": 93047904,
      "step": 43110
    },
    {
      "epoch": 7.033442088091354,
      "grad_norm": 0.06114305928349495,
      "learning_rate": 1.224934342658061e-05,
      "loss": 0.2022,
      "num_input_tokens_seen": 93057568,
      "step": 43115
    },
    {
      "epoch": 7.034257748776509,
      "grad_norm": 0.09569920599460602,
      "learning_rate": 1.224322134818769e-05,
      "loss": 0.1111,
      "num_input_tokens_seen": 93068192,
      "step": 43120
    },
    {
      "epoch": 7.035073409461664,
      "grad_norm": 0.04119197651743889,
      "learning_rate": 1.2237100303923977e-05,
      "loss": 0.0036,
      "num_input_tokens_seen": 93080032,
      "step": 43125
    },
    {
      "epoch": 7.035889070146819,
      "grad_norm": 18.058048248291016,
      "learning_rate": 1.2230980294285669e-05,
      "loss": 0.1067,
      "num_input_tokens_seen": 93090784,
      "step": 43130
    },
    {
      "epoch": 7.036704730831974,
      "grad_norm": 5.157989025115967,
      "learning_rate": 1.2224861319768887e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 93102176,
      "step": 43135
    },
    {
      "epoch": 7.037520391517129,
      "grad_norm": 0.05058105289936066,
      "learning_rate": 1.2218743380869669e-05,
      "loss": 0.0046,
      "num_input_tokens_seen": 93112928,
      "step": 43140
    },
    {
      "epoch": 7.0383360522022835,
      "grad_norm": 0.07497899979352951,
      "learning_rate": 1.2212626478083964e-05,
      "loss": 0.0596,
      "num_input_tokens_seen": 93123872,
      "step": 43145
    },
    {
      "epoch": 7.039151712887439,
      "grad_norm": 0.08604057133197784,
      "learning_rate": 1.2206510611907638e-05,
      "loss": 0.011,
      "num_input_tokens_seen": 93134368,
      "step": 43150
    },
    {
      "epoch": 7.039967373572594,
      "grad_norm": 4.297252655029297,
      "learning_rate": 1.2200395782836477e-05,
      "loss": 0.0941,
      "num_input_tokens_seen": 93146720,
      "step": 43155
    },
    {
      "epoch": 7.040783034257749,
      "grad_norm": 0.10059039294719696,
      "learning_rate": 1.2194281991366176e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 93157632,
      "step": 43160
    },
    {
      "epoch": 7.041598694942904,
      "grad_norm": 0.06647174060344696,
      "learning_rate": 1.2188169237992345e-05,
      "loss": 0.0033,
      "num_input_tokens_seen": 93167264,
      "step": 43165
    },
    {
      "epoch": 7.0424143556280585,
      "grad_norm": 0.44864416122436523,
      "learning_rate": 1.218205752321053e-05,
      "loss": 0.0305,
      "num_input_tokens_seen": 93179136,
      "step": 43170
    },
    {
      "epoch": 7.043230016313213,
      "grad_norm": 0.0658923089504242,
      "learning_rate": 1.217594684751617e-05,
      "loss": 0.1252,
      "num_input_tokens_seen": 93189536,
      "step": 43175
    },
    {
      "epoch": 7.044045676998369,
      "grad_norm": 0.04984398931264877,
      "learning_rate": 1.2169837211404627e-05,
      "loss": 0.0014,
      "num_input_tokens_seen": 93201568,
      "step": 43180
    },
    {
      "epoch": 7.044861337683524,
      "grad_norm": 0.034007783979177475,
      "learning_rate": 1.2163728615371181e-05,
      "loss": 0.0191,
      "num_input_tokens_seen": 93212128,
      "step": 43185
    },
    {
      "epoch": 7.045676998368679,
      "grad_norm": 6.431211471557617,
      "learning_rate": 1.2157621059911014e-05,
      "loss": 0.2947,
      "num_input_tokens_seen": 93222784,
      "step": 43190
    },
    {
      "epoch": 7.0464926590538335,
      "grad_norm": 0.05510122701525688,
      "learning_rate": 1.2151514545519254e-05,
      "loss": 0.0028,
      "num_input_tokens_seen": 93234016,
      "step": 43195
    },
    {
      "epoch": 7.047308319738988,
      "grad_norm": 4.3655595779418945,
      "learning_rate": 1.214540907269092e-05,
      "loss": 0.1949,
      "num_input_tokens_seen": 93243200,
      "step": 43200
    },
    {
      "epoch": 7.048123980424143,
      "grad_norm": 0.061375848948955536,
      "learning_rate": 1.2139304641920946e-05,
      "loss": 0.0016,
      "num_input_tokens_seen": 93254432,
      "step": 43205
    },
    {
      "epoch": 7.048939641109299,
      "grad_norm": 0.0294011402875185,
      "learning_rate": 1.2133201253704196e-05,
      "loss": 0.0037,
      "num_input_tokens_seen": 93263360,
      "step": 43210
    },
    {
      "epoch": 7.049755301794454,
      "grad_norm": 0.03514668717980385,
      "learning_rate": 1.2127098908535434e-05,
      "loss": 0.1108,
      "num_input_tokens_seen": 93273376,
      "step": 43215
    },
    {
      "epoch": 7.0505709624796085,
      "grad_norm": 0.09905792772769928,
      "learning_rate": 1.2120997606909354e-05,
      "loss": 0.091,
      "num_input_tokens_seen": 93284416,
      "step": 43220
    },
    {
      "epoch": 7.051386623164763,
      "grad_norm": 0.1089557558298111,
      "learning_rate": 1.2114897349320553e-05,
      "loss": 0.0242,
      "num_input_tokens_seen": 93295776,
      "step": 43225
    },
    {
      "epoch": 7.052202283849918,
      "grad_norm": 8.377816200256348,
      "learning_rate": 1.2108798136263555e-05,
      "loss": 0.0256,
      "num_input_tokens_seen": 93305408,
      "step": 43230
    },
    {
      "epoch": 7.053017944535074,
      "grad_norm": 7.742931842803955,
      "learning_rate": 1.2102699968232787e-05,
      "loss": 0.4057,
      "num_input_tokens_seen": 93316192,
      "step": 43235
    },
    {
      "epoch": 7.053833605220229,
      "grad_norm": 0.05888286232948303,
      "learning_rate": 1.2096602845722598e-05,
      "loss": 0.132,
      "num_input_tokens_seen": 93327072,
      "step": 43240
    },
    {
      "epoch": 7.054649265905383,
      "grad_norm": 0.2003031075000763,
      "learning_rate": 1.2090506769227256e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 93338368,
      "step": 43245
    },
    {
      "epoch": 7.055464926590538,
      "grad_norm": 0.4757084250450134,
      "learning_rate": 1.2084411739240936e-05,
      "loss": 0.011,
      "num_input_tokens_seen": 93349952,
      "step": 43250
    },
    {
      "epoch": 7.056280587275693,
      "grad_norm": 3.10579514503479,
      "learning_rate": 1.2078317756257735e-05,
      "loss": 0.1048,
      "num_input_tokens_seen": 93360480,
      "step": 43255
    },
    {
      "epoch": 7.057096247960848,
      "grad_norm": 8.451691627502441,
      "learning_rate": 1.207222482077166e-05,
      "loss": 0.0438,
      "num_input_tokens_seen": 93371104,
      "step": 43260
    },
    {
      "epoch": 7.057911908646004,
      "grad_norm": 10.191644668579102,
      "learning_rate": 1.2066132933276636e-05,
      "loss": 0.2268,
      "num_input_tokens_seen": 93381792,
      "step": 43265
    },
    {
      "epoch": 7.058727569331158,
      "grad_norm": 0.05645562708377838,
      "learning_rate": 1.2060042094266495e-05,
      "loss": 0.0019,
      "num_input_tokens_seen": 93393024,
      "step": 43270
    },
    {
      "epoch": 7.059543230016313,
      "grad_norm": 0.03519802913069725,
      "learning_rate": 1.2053952304235002e-05,
      "loss": 0.0435,
      "num_input_tokens_seen": 93403968,
      "step": 43275
    },
    {
      "epoch": 7.060358890701468,
      "grad_norm": 0.07684657722711563,
      "learning_rate": 1.2047863563675826e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 93414336,
      "step": 43280
    },
    {
      "epoch": 7.061174551386623,
      "grad_norm": 0.0919998362660408,
      "learning_rate": 1.204177587308255e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 93425312,
      "step": 43285
    },
    {
      "epoch": 7.061990212071779,
      "grad_norm": 3.51995587348938,
      "learning_rate": 1.2035689232948669e-05,
      "loss": 0.1389,
      "num_input_tokens_seen": 93434368,
      "step": 43290
    },
    {
      "epoch": 7.062805872756933,
      "grad_norm": 4.85670804977417,
      "learning_rate": 1.20296036437676e-05,
      "loss": 0.1302,
      "num_input_tokens_seen": 93445696,
      "step": 43295
    },
    {
      "epoch": 7.063621533442088,
      "grad_norm": 0.11340179294347763,
      "learning_rate": 1.2023519106032672e-05,
      "loss": 0.0035,
      "num_input_tokens_seen": 93456768,
      "step": 43300
    },
    {
      "epoch": 7.064437194127243,
      "grad_norm": 0.027684850618243217,
      "learning_rate": 1.2017435620237125e-05,
      "loss": 0.1451,
      "num_input_tokens_seen": 93468320,
      "step": 43305
    },
    {
      "epoch": 7.065252854812398,
      "grad_norm": 0.04040546715259552,
      "learning_rate": 1.201135318687412e-05,
      "loss": 0.0833,
      "num_input_tokens_seen": 93480832,
      "step": 43310
    },
    {
      "epoch": 7.066068515497553,
      "grad_norm": 0.05264955013990402,
      "learning_rate": 1.2005271806436727e-05,
      "loss": 0.0303,
      "num_input_tokens_seen": 93490880,
      "step": 43315
    },
    {
      "epoch": 7.066884176182708,
      "grad_norm": 0.030342532321810722,
      "learning_rate": 1.199919147941794e-05,
      "loss": 0.0019,
      "num_input_tokens_seen": 93501600,
      "step": 43320
    },
    {
      "epoch": 7.067699836867863,
      "grad_norm": 0.29926538467407227,
      "learning_rate": 1.1993112206310656e-05,
      "loss": 0.0794,
      "num_input_tokens_seen": 93512832,
      "step": 43325
    },
    {
      "epoch": 7.068515497553018,
      "grad_norm": 0.18785369396209717,
      "learning_rate": 1.1987033987607681e-05,
      "loss": 0.0204,
      "num_input_tokens_seen": 93524416,
      "step": 43330
    },
    {
      "epoch": 7.069331158238173,
      "grad_norm": 0.02750498801469803,
      "learning_rate": 1.198095682380177e-05,
      "loss": 0.0923,
      "num_input_tokens_seen": 93535744,
      "step": 43335
    },
    {
      "epoch": 7.070146818923328,
      "grad_norm": 0.06883826106786728,
      "learning_rate": 1.1974880715385557e-05,
      "loss": 0.0036,
      "num_input_tokens_seen": 93546016,
      "step": 43340
    },
    {
      "epoch": 7.0709624796084825,
      "grad_norm": 3.369436502456665,
      "learning_rate": 1.1968805662851601e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 93555040,
      "step": 43345
    },
    {
      "epoch": 7.071778140293638,
      "grad_norm": 3.999937057495117,
      "learning_rate": 1.1962731666692378e-05,
      "loss": 0.1094,
      "num_input_tokens_seen": 93565184,
      "step": 43350
    },
    {
      "epoch": 7.072593800978793,
      "grad_norm": 0.06369879096746445,
      "learning_rate": 1.1956658727400277e-05,
      "loss": 0.0048,
      "num_input_tokens_seen": 93576352,
      "step": 43355
    },
    {
      "epoch": 7.073409461663948,
      "grad_norm": 0.036028988659381866,
      "learning_rate": 1.1950586845467602e-05,
      "loss": 0.2048,
      "num_input_tokens_seen": 93586880,
      "step": 43360
    },
    {
      "epoch": 7.074225122349103,
      "grad_norm": 0.20576518774032593,
      "learning_rate": 1.1944516021386565e-05,
      "loss": 0.074,
      "num_input_tokens_seen": 93598560,
      "step": 43365
    },
    {
      "epoch": 7.075040783034257,
      "grad_norm": 8.174752235412598,
      "learning_rate": 1.1938446255649305e-05,
      "loss": 0.2437,
      "num_input_tokens_seen": 93608928,
      "step": 43370
    },
    {
      "epoch": 7.075856443719413,
      "grad_norm": 0.08758328855037689,
      "learning_rate": 1.1932377548747867e-05,
      "loss": 0.266,
      "num_input_tokens_seen": 93619552,
      "step": 43375
    },
    {
      "epoch": 7.076672104404568,
      "grad_norm": 0.20168966054916382,
      "learning_rate": 1.1926309901174196e-05,
      "loss": 0.0036,
      "num_input_tokens_seen": 93631264,
      "step": 43380
    },
    {
      "epoch": 7.077487765089723,
      "grad_norm": 9.433473587036133,
      "learning_rate": 1.192024331342019e-05,
      "loss": 0.0733,
      "num_input_tokens_seen": 93641792,
      "step": 43385
    },
    {
      "epoch": 7.078303425774878,
      "grad_norm": 0.04689888656139374,
      "learning_rate": 1.1914177785977629e-05,
      "loss": 0.0873,
      "num_input_tokens_seen": 93652992,
      "step": 43390
    },
    {
      "epoch": 7.079119086460032,
      "grad_norm": 0.12399861961603165,
      "learning_rate": 1.1908113319338212e-05,
      "loss": 0.0029,
      "num_input_tokens_seen": 93663552,
      "step": 43395
    },
    {
      "epoch": 7.079934747145187,
      "grad_norm": 1.831578254699707,
      "learning_rate": 1.1902049913993558e-05,
      "loss": 0.1098,
      "num_input_tokens_seen": 93674272,
      "step": 43400
    },
    {
      "epoch": 7.080750407830343,
      "grad_norm": 0.038481879979372025,
      "learning_rate": 1.1895987570435196e-05,
      "loss": 0.0035,
      "num_input_tokens_seen": 93685536,
      "step": 43405
    },
    {
      "epoch": 7.081566068515498,
      "grad_norm": 0.17415465414524078,
      "learning_rate": 1.188992628915457e-05,
      "loss": 0.0071,
      "num_input_tokens_seen": 93696640,
      "step": 43410
    },
    {
      "epoch": 7.082381729200653,
      "grad_norm": 29.55493927001953,
      "learning_rate": 1.1883866070643041e-05,
      "loss": 0.053,
      "num_input_tokens_seen": 93707008,
      "step": 43415
    },
    {
      "epoch": 7.083197389885807,
      "grad_norm": 0.9635235667228699,
      "learning_rate": 1.1877806915391875e-05,
      "loss": 0.1307,
      "num_input_tokens_seen": 93717376,
      "step": 43420
    },
    {
      "epoch": 7.084013050570962,
      "grad_norm": 18.65500831604004,
      "learning_rate": 1.1871748823892264e-05,
      "loss": 0.028,
      "num_input_tokens_seen": 93728032,
      "step": 43425
    },
    {
      "epoch": 7.084828711256117,
      "grad_norm": 0.0626855194568634,
      "learning_rate": 1.1865691796635306e-05,
      "loss": 0.1448,
      "num_input_tokens_seen": 93737792,
      "step": 43430
    },
    {
      "epoch": 7.085644371941273,
      "grad_norm": 0.1425950974225998,
      "learning_rate": 1.1859635834112012e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 93749088,
      "step": 43435
    },
    {
      "epoch": 7.0864600326264275,
      "grad_norm": 4.997335910797119,
      "learning_rate": 1.1853580936813313e-05,
      "loss": 0.0561,
      "num_input_tokens_seen": 93760512,
      "step": 43440
    },
    {
      "epoch": 7.087275693311582,
      "grad_norm": 0.13220566511154175,
      "learning_rate": 1.1847527105230047e-05,
      "loss": 0.1945,
      "num_input_tokens_seen": 93771712,
      "step": 43445
    },
    {
      "epoch": 7.088091353996737,
      "grad_norm": 0.06322714686393738,
      "learning_rate": 1.1841474339852968e-05,
      "loss": 0.0813,
      "num_input_tokens_seen": 93783008,
      "step": 43450
    },
    {
      "epoch": 7.088907014681892,
      "grad_norm": 5.114476680755615,
      "learning_rate": 1.1835422641172744e-05,
      "loss": 0.0604,
      "num_input_tokens_seen": 93792928,
      "step": 43455
    },
    {
      "epoch": 7.089722675367048,
      "grad_norm": 0.04860479384660721,
      "learning_rate": 1.182937200967996e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 93803872,
      "step": 43460
    },
    {
      "epoch": 7.0905383360522025,
      "grad_norm": 1.2112655639648438,
      "learning_rate": 1.1823322445865103e-05,
      "loss": 0.0438,
      "num_input_tokens_seen": 93815584,
      "step": 43465
    },
    {
      "epoch": 7.091353996737357,
      "grad_norm": 0.4199487864971161,
      "learning_rate": 1.1817273950218591e-05,
      "loss": 0.0869,
      "num_input_tokens_seen": 93827040,
      "step": 43470
    },
    {
      "epoch": 7.092169657422512,
      "grad_norm": 0.050442811101675034,
      "learning_rate": 1.1811226523230731e-05,
      "loss": 0.086,
      "num_input_tokens_seen": 93836800,
      "step": 43475
    },
    {
      "epoch": 7.092985318107667,
      "grad_norm": 14.030454635620117,
      "learning_rate": 1.1805180165391774e-05,
      "loss": 0.093,
      "num_input_tokens_seen": 93848032,
      "step": 43480
    },
    {
      "epoch": 7.093800978792822,
      "grad_norm": 4.291860103607178,
      "learning_rate": 1.1799134877191867e-05,
      "loss": 0.1784,
      "num_input_tokens_seen": 93858720,
      "step": 43485
    },
    {
      "epoch": 7.0946166394779775,
      "grad_norm": 0.052896466106176376,
      "learning_rate": 1.1793090659121065e-05,
      "loss": 0.0646,
      "num_input_tokens_seen": 93869696,
      "step": 43490
    },
    {
      "epoch": 7.095432300163132,
      "grad_norm": 0.11294753104448318,
      "learning_rate": 1.1787047511669347e-05,
      "loss": 0.1258,
      "num_input_tokens_seen": 93880672,
      "step": 43495
    },
    {
      "epoch": 7.096247960848287,
      "grad_norm": 3.0636801719665527,
      "learning_rate": 1.178100543532659e-05,
      "loss": 0.0719,
      "num_input_tokens_seen": 93891840,
      "step": 43500
    },
    {
      "epoch": 7.097063621533442,
      "grad_norm": 0.06495791673660278,
      "learning_rate": 1.1774964430582614e-05,
      "loss": 0.1446,
      "num_input_tokens_seen": 93900928,
      "step": 43505
    },
    {
      "epoch": 7.097879282218597,
      "grad_norm": 4.291179656982422,
      "learning_rate": 1.1768924497927123e-05,
      "loss": 0.0927,
      "num_input_tokens_seen": 93912480,
      "step": 43510
    },
    {
      "epoch": 7.0986949429037525,
      "grad_norm": 0.0967203751206398,
      "learning_rate": 1.1762885637849746e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 93924192,
      "step": 43515
    },
    {
      "epoch": 7.099510603588907,
      "grad_norm": 0.10868646204471588,
      "learning_rate": 1.1756847850840024e-05,
      "loss": 0.0052,
      "num_input_tokens_seen": 93934304,
      "step": 43520
    },
    {
      "epoch": 7.100326264274062,
      "grad_norm": 0.04877633601427078,
      "learning_rate": 1.1750811137387414e-05,
      "loss": 0.0986,
      "num_input_tokens_seen": 93944032,
      "step": 43525
    },
    {
      "epoch": 7.101141924959217,
      "grad_norm": 0.857296884059906,
      "learning_rate": 1.1744775497981273e-05,
      "loss": 0.0053,
      "num_input_tokens_seen": 93954304,
      "step": 43530
    },
    {
      "epoch": 7.101957585644372,
      "grad_norm": 0.02552896924316883,
      "learning_rate": 1.173874093311089e-05,
      "loss": 0.0902,
      "num_input_tokens_seen": 93964864,
      "step": 43535
    },
    {
      "epoch": 7.102773246329527,
      "grad_norm": 1.6049836874008179,
      "learning_rate": 1.1732707443265453e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 93976448,
      "step": 43540
    },
    {
      "epoch": 7.103588907014682,
      "grad_norm": 0.12095363438129425,
      "learning_rate": 1.172667502893407e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 93986848,
      "step": 43545
    },
    {
      "epoch": 7.104404567699837,
      "grad_norm": 0.09445792436599731,
      "learning_rate": 1.1720643690605754e-05,
      "loss": 0.0081,
      "num_input_tokens_seen": 93997472,
      "step": 43550
    },
    {
      "epoch": 7.105220228384992,
      "grad_norm": 0.01909150369465351,
      "learning_rate": 1.1714613428769442e-05,
      "loss": 0.0024,
      "num_input_tokens_seen": 94007744,
      "step": 43555
    },
    {
      "epoch": 7.106035889070147,
      "grad_norm": 12.791037559509277,
      "learning_rate": 1.1708584243913972e-05,
      "loss": 0.3366,
      "num_input_tokens_seen": 94018880,
      "step": 43560
    },
    {
      "epoch": 7.1068515497553015,
      "grad_norm": 0.028335576876997948,
      "learning_rate": 1.1702556136528106e-05,
      "loss": 0.0035,
      "num_input_tokens_seen": 94030112,
      "step": 43565
    },
    {
      "epoch": 7.107667210440456,
      "grad_norm": 0.1255069226026535,
      "learning_rate": 1.169652910710051e-05,
      "loss": 0.1141,
      "num_input_tokens_seen": 94041120,
      "step": 43570
    },
    {
      "epoch": 7.108482871125612,
      "grad_norm": 0.20495577156543732,
      "learning_rate": 1.1690503156119764e-05,
      "loss": 0.0487,
      "num_input_tokens_seen": 94052928,
      "step": 43575
    },
    {
      "epoch": 7.109298531810767,
      "grad_norm": 0.05670922249555588,
      "learning_rate": 1.1684478284074365e-05,
      "loss": 0.0772,
      "num_input_tokens_seen": 94064480,
      "step": 43580
    },
    {
      "epoch": 7.110114192495922,
      "grad_norm": 0.06790587306022644,
      "learning_rate": 1.1678454491452717e-05,
      "loss": 0.0988,
      "num_input_tokens_seen": 94076064,
      "step": 43585
    },
    {
      "epoch": 7.1109298531810765,
      "grad_norm": 0.05412507802248001,
      "learning_rate": 1.1672431778743133e-05,
      "loss": 0.0257,
      "num_input_tokens_seen": 94085216,
      "step": 43590
    },
    {
      "epoch": 7.111745513866231,
      "grad_norm": 3.800899028778076,
      "learning_rate": 1.1666410146433861e-05,
      "loss": 0.2593,
      "num_input_tokens_seen": 94095904,
      "step": 43595
    },
    {
      "epoch": 7.112561174551387,
      "grad_norm": 2.408957004547119,
      "learning_rate": 1.1660389595013038e-05,
      "loss": 0.1236,
      "num_input_tokens_seen": 94108288,
      "step": 43600
    },
    {
      "epoch": 7.113376835236542,
      "grad_norm": 0.08000515401363373,
      "learning_rate": 1.165437012496872e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 94117440,
      "step": 43605
    },
    {
      "epoch": 7.114192495921697,
      "grad_norm": 0.02687612548470497,
      "learning_rate": 1.1648351736788871e-05,
      "loss": 0.0875,
      "num_input_tokens_seen": 94127488,
      "step": 43610
    },
    {
      "epoch": 7.1150081566068515,
      "grad_norm": 0.21786536276340485,
      "learning_rate": 1.1642334430961377e-05,
      "loss": 0.0863,
      "num_input_tokens_seen": 94137824,
      "step": 43615
    },
    {
      "epoch": 7.115823817292006,
      "grad_norm": 0.27159833908081055,
      "learning_rate": 1.163631820797403e-05,
      "loss": 0.2043,
      "num_input_tokens_seen": 94149632,
      "step": 43620
    },
    {
      "epoch": 7.116639477977161,
      "grad_norm": 0.08399226516485214,
      "learning_rate": 1.1630303068314538e-05,
      "loss": 0.0145,
      "num_input_tokens_seen": 94161184,
      "step": 43625
    },
    {
      "epoch": 7.117455138662317,
      "grad_norm": 0.07265343517065048,
      "learning_rate": 1.1624289012470513e-05,
      "loss": 0.0459,
      "num_input_tokens_seen": 94170656,
      "step": 43630
    },
    {
      "epoch": 7.118270799347472,
      "grad_norm": 0.04176503047347069,
      "learning_rate": 1.161827604092949e-05,
      "loss": 0.0045,
      "num_input_tokens_seen": 94180384,
      "step": 43635
    },
    {
      "epoch": 7.1190864600326265,
      "grad_norm": 0.036481671035289764,
      "learning_rate": 1.161226415417891e-05,
      "loss": 0.0033,
      "num_input_tokens_seen": 94191456,
      "step": 43640
    },
    {
      "epoch": 7.119902120717781,
      "grad_norm": 0.08592111617326736,
      "learning_rate": 1.1606253352706118e-05,
      "loss": 0.1274,
      "num_input_tokens_seen": 94202112,
      "step": 43645
    },
    {
      "epoch": 7.120717781402936,
      "grad_norm": 3.752108573913574,
      "learning_rate": 1.1600243636998396e-05,
      "loss": 0.2679,
      "num_input_tokens_seen": 94213312,
      "step": 43650
    },
    {
      "epoch": 7.121533442088092,
      "grad_norm": 0.04887622222304344,
      "learning_rate": 1.1594235007542914e-05,
      "loss": 0.0061,
      "num_input_tokens_seen": 94225248,
      "step": 43655
    },
    {
      "epoch": 7.122349102773247,
      "grad_norm": 3.3832032680511475,
      "learning_rate": 1.1588227464826763e-05,
      "loss": 0.1378,
      "num_input_tokens_seen": 94236416,
      "step": 43660
    },
    {
      "epoch": 7.123164763458401,
      "grad_norm": 0.09872160851955414,
      "learning_rate": 1.1582221009336944e-05,
      "loss": 0.0412,
      "num_input_tokens_seen": 94247040,
      "step": 43665
    },
    {
      "epoch": 7.123980424143556,
      "grad_norm": 1.166061520576477,
      "learning_rate": 1.157621564156037e-05,
      "loss": 0.0091,
      "num_input_tokens_seen": 94257472,
      "step": 43670
    },
    {
      "epoch": 7.124796084828711,
      "grad_norm": 3.643521547317505,
      "learning_rate": 1.157021136198387e-05,
      "loss": 0.1308,
      "num_input_tokens_seen": 94267776,
      "step": 43675
    },
    {
      "epoch": 7.125611745513866,
      "grad_norm": 3.188948392868042,
      "learning_rate": 1.156420817109418e-05,
      "loss": 0.0884,
      "num_input_tokens_seen": 94279552,
      "step": 43680
    },
    {
      "epoch": 7.126427406199022,
      "grad_norm": 0.033812519162893295,
      "learning_rate": 1.1558206069377945e-05,
      "loss": 0.0273,
      "num_input_tokens_seen": 94291424,
      "step": 43685
    },
    {
      "epoch": 7.127243066884176,
      "grad_norm": 0.656554639339447,
      "learning_rate": 1.155220505732173e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 94302080,
      "step": 43690
    },
    {
      "epoch": 7.128058727569331,
      "grad_norm": 0.08332250267267227,
      "learning_rate": 1.1546205135412008e-05,
      "loss": 0.0918,
      "num_input_tokens_seen": 94313440,
      "step": 43695
    },
    {
      "epoch": 7.128874388254486,
      "grad_norm": 0.11393269151449203,
      "learning_rate": 1.1540206304135152e-05,
      "loss": 0.0604,
      "num_input_tokens_seen": 94322784,
      "step": 43700
    },
    {
      "epoch": 7.129690048939641,
      "grad_norm": 7.067556381225586,
      "learning_rate": 1.1534208563977475e-05,
      "loss": 0.1465,
      "num_input_tokens_seen": 94333280,
      "step": 43705
    },
    {
      "epoch": 7.130505709624796,
      "grad_norm": 5.529153347015381,
      "learning_rate": 1.1528211915425177e-05,
      "loss": 0.0854,
      "num_input_tokens_seen": 94344032,
      "step": 43710
    },
    {
      "epoch": 7.131321370309951,
      "grad_norm": 0.07599502056837082,
      "learning_rate": 1.1522216358964377e-05,
      "loss": 0.145,
      "num_input_tokens_seen": 94354816,
      "step": 43715
    },
    {
      "epoch": 7.132137030995106,
      "grad_norm": 0.10091648250818253,
      "learning_rate": 1.1516221895081104e-05,
      "loss": 0.1838,
      "num_input_tokens_seen": 94364896,
      "step": 43720
    },
    {
      "epoch": 7.132952691680261,
      "grad_norm": 0.20659762620925903,
      "learning_rate": 1.1510228524261302e-05,
      "loss": 0.3342,
      "num_input_tokens_seen": 94375648,
      "step": 43725
    },
    {
      "epoch": 7.133768352365416,
      "grad_norm": 0.0431620255112648,
      "learning_rate": 1.1504236246990819e-05,
      "loss": 0.1825,
      "num_input_tokens_seen": 94386816,
      "step": 43730
    },
    {
      "epoch": 7.134584013050571,
      "grad_norm": 0.1603519767522812,
      "learning_rate": 1.1498245063755425e-05,
      "loss": 0.0096,
      "num_input_tokens_seen": 94397856,
      "step": 43735
    },
    {
      "epoch": 7.135399673735726,
      "grad_norm": 3.901911735534668,
      "learning_rate": 1.1492254975040792e-05,
      "loss": 0.1674,
      "num_input_tokens_seen": 94408128,
      "step": 43740
    },
    {
      "epoch": 7.136215334420881,
      "grad_norm": 0.023856377229094505,
      "learning_rate": 1.148626598133251e-05,
      "loss": 0.1777,
      "num_input_tokens_seen": 94418912,
      "step": 43745
    },
    {
      "epoch": 7.137030995106036,
      "grad_norm": 0.08474874496459961,
      "learning_rate": 1.1480278083116074e-05,
      "loss": 0.0062,
      "num_input_tokens_seen": 94428576,
      "step": 43750
    },
    {
      "epoch": 7.137846655791191,
      "grad_norm": 0.03850866109132767,
      "learning_rate": 1.1474291280876894e-05,
      "loss": 0.2975,
      "num_input_tokens_seen": 94439232,
      "step": 43755
    },
    {
      "epoch": 7.138662316476346,
      "grad_norm": 5.355396747589111,
      "learning_rate": 1.1468305575100294e-05,
      "loss": 0.0709,
      "num_input_tokens_seen": 94450144,
      "step": 43760
    },
    {
      "epoch": 7.1394779771615005,
      "grad_norm": 0.050152700394392014,
      "learning_rate": 1.1462320966271503e-05,
      "loss": 0.0068,
      "num_input_tokens_seen": 94461664,
      "step": 43765
    },
    {
      "epoch": 7.140293637846656,
      "grad_norm": 0.057023610919713974,
      "learning_rate": 1.1456337454875663e-05,
      "loss": 0.0032,
      "num_input_tokens_seen": 94471392,
      "step": 43770
    },
    {
      "epoch": 7.141109298531811,
      "grad_norm": 0.15296074748039246,
      "learning_rate": 1.1450355041397829e-05,
      "loss": 0.1481,
      "num_input_tokens_seen": 94482848,
      "step": 43775
    },
    {
      "epoch": 7.141924959216966,
      "grad_norm": 0.09270456433296204,
      "learning_rate": 1.1444373726322966e-05,
      "loss": 0.1107,
      "num_input_tokens_seen": 94493792,
      "step": 43780
    },
    {
      "epoch": 7.142740619902121,
      "grad_norm": 0.08430933207273483,
      "learning_rate": 1.143839351013595e-05,
      "loss": 0.0044,
      "num_input_tokens_seen": 94504448,
      "step": 43785
    },
    {
      "epoch": 7.143556280587275,
      "grad_norm": 0.2279704511165619,
      "learning_rate": 1.1432414393321556e-05,
      "loss": 0.1016,
      "num_input_tokens_seen": 94515296,
      "step": 43790
    },
    {
      "epoch": 7.14437194127243,
      "grad_norm": 16.787500381469727,
      "learning_rate": 1.1426436376364502e-05,
      "loss": 0.0347,
      "num_input_tokens_seen": 94525984,
      "step": 43795
    },
    {
      "epoch": 7.145187601957586,
      "grad_norm": 0.16307874023914337,
      "learning_rate": 1.1420459459749385e-05,
      "loss": 0.0947,
      "num_input_tokens_seen": 94536512,
      "step": 43800
    },
    {
      "epoch": 7.146003262642741,
      "grad_norm": 0.0962166041135788,
      "learning_rate": 1.141448364396073e-05,
      "loss": 0.0046,
      "num_input_tokens_seen": 94546208,
      "step": 43805
    },
    {
      "epoch": 7.146818923327896,
      "grad_norm": 0.22677689790725708,
      "learning_rate": 1.1408508929482961e-05,
      "loss": 0.0061,
      "num_input_tokens_seen": 94558656,
      "step": 43810
    },
    {
      "epoch": 7.14763458401305,
      "grad_norm": 0.35085445642471313,
      "learning_rate": 1.1402535316800414e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 94568832,
      "step": 43815
    },
    {
      "epoch": 7.148450244698205,
      "grad_norm": 3.750436305999756,
      "learning_rate": 1.1396562806397354e-05,
      "loss": 0.1152,
      "num_input_tokens_seen": 94580512,
      "step": 43820
    },
    {
      "epoch": 7.149265905383361,
      "grad_norm": 4.464899063110352,
      "learning_rate": 1.1390591398757935e-05,
      "loss": 0.1694,
      "num_input_tokens_seen": 94592608,
      "step": 43825
    },
    {
      "epoch": 7.150081566068516,
      "grad_norm": 20.690568923950195,
      "learning_rate": 1.1384621094366232e-05,
      "loss": 0.0443,
      "num_input_tokens_seen": 94603200,
      "step": 43830
    },
    {
      "epoch": 7.150897226753671,
      "grad_norm": 0.12341461330652237,
      "learning_rate": 1.1378651893706227e-05,
      "loss": 0.0258,
      "num_input_tokens_seen": 94614656,
      "step": 43835
    },
    {
      "epoch": 7.151712887438825,
      "grad_norm": 0.2521572709083557,
      "learning_rate": 1.1372683797261814e-05,
      "loss": 0.0068,
      "num_input_tokens_seen": 94624512,
      "step": 43840
    },
    {
      "epoch": 7.15252854812398,
      "grad_norm": 0.051836270838975906,
      "learning_rate": 1.1366716805516794e-05,
      "loss": 0.1523,
      "num_input_tokens_seen": 94635360,
      "step": 43845
    },
    {
      "epoch": 7.153344208809135,
      "grad_norm": 0.06448886543512344,
      "learning_rate": 1.1360750918954887e-05,
      "loss": 0.0077,
      "num_input_tokens_seen": 94645440,
      "step": 43850
    },
    {
      "epoch": 7.154159869494291,
      "grad_norm": 0.042032964527606964,
      "learning_rate": 1.1354786138059715e-05,
      "loss": 0.2204,
      "num_input_tokens_seen": 94656832,
      "step": 43855
    },
    {
      "epoch": 7.1549755301794455,
      "grad_norm": 0.07452675700187683,
      "learning_rate": 1.1348822463314815e-05,
      "loss": 0.004,
      "num_input_tokens_seen": 94666816,
      "step": 43860
    },
    {
      "epoch": 7.1557911908646,
      "grad_norm": 0.13034167885780334,
      "learning_rate": 1.1342859895203629e-05,
      "loss": 0.102,
      "num_input_tokens_seen": 94677888,
      "step": 43865
    },
    {
      "epoch": 7.156606851549755,
      "grad_norm": 1.121642827987671,
      "learning_rate": 1.1336898434209517e-05,
      "loss": 0.0067,
      "num_input_tokens_seen": 94689088,
      "step": 43870
    },
    {
      "epoch": 7.15742251223491,
      "grad_norm": 0.04426287114620209,
      "learning_rate": 1.1330938080815743e-05,
      "loss": 0.0964,
      "num_input_tokens_seen": 94700992,
      "step": 43875
    },
    {
      "epoch": 7.158238172920065,
      "grad_norm": 0.2092503011226654,
      "learning_rate": 1.1324978835505483e-05,
      "loss": 0.0131,
      "num_input_tokens_seen": 94710112,
      "step": 43880
    },
    {
      "epoch": 7.1590538336052205,
      "grad_norm": 0.04995310306549072,
      "learning_rate": 1.1319020698761828e-05,
      "loss": 0.1868,
      "num_input_tokens_seen": 94720544,
      "step": 43885
    },
    {
      "epoch": 7.159869494290375,
      "grad_norm": 0.11302343010902405,
      "learning_rate": 1.1313063671067769e-05,
      "loss": 0.012,
      "num_input_tokens_seen": 94730496,
      "step": 43890
    },
    {
      "epoch": 7.16068515497553,
      "grad_norm": 0.07707829773426056,
      "learning_rate": 1.1307107752906218e-05,
      "loss": 0.2294,
      "num_input_tokens_seen": 94741696,
      "step": 43895
    },
    {
      "epoch": 7.161500815660685,
      "grad_norm": 0.0993894636631012,
      "learning_rate": 1.1301152944759988e-05,
      "loss": 0.0075,
      "num_input_tokens_seen": 94752000,
      "step": 43900
    },
    {
      "epoch": 7.16231647634584,
      "grad_norm": 0.08817077428102493,
      "learning_rate": 1.129519924711181e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 94763264,
      "step": 43905
    },
    {
      "epoch": 7.1631321370309955,
      "grad_norm": 3.8644626140594482,
      "learning_rate": 1.1289246660444306e-05,
      "loss": 0.2182,
      "num_input_tokens_seen": 94773504,
      "step": 43910
    },
    {
      "epoch": 7.16394779771615,
      "grad_norm": 0.10693071782588959,
      "learning_rate": 1.1283295185240048e-05,
      "loss": 0.1286,
      "num_input_tokens_seen": 94783328,
      "step": 43915
    },
    {
      "epoch": 7.164763458401305,
      "grad_norm": 0.1141429990530014,
      "learning_rate": 1.1277344821981475e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 94795584,
      "step": 43920
    },
    {
      "epoch": 7.16557911908646,
      "grad_norm": 0.03173322603106499,
      "learning_rate": 1.1271395571150964e-05,
      "loss": 0.0419,
      "num_input_tokens_seen": 94805408,
      "step": 43925
    },
    {
      "epoch": 7.166394779771615,
      "grad_norm": 0.1410226970911026,
      "learning_rate": 1.1265447433230784e-05,
      "loss": 0.0809,
      "num_input_tokens_seen": 94817024,
      "step": 43930
    },
    {
      "epoch": 7.16721044045677,
      "grad_norm": 4.855565547943115,
      "learning_rate": 1.1259500408703124e-05,
      "loss": 0.3698,
      "num_input_tokens_seen": 94828480,
      "step": 43935
    },
    {
      "epoch": 7.168026101141925,
      "grad_norm": 0.32295486330986023,
      "learning_rate": 1.1253554498050078e-05,
      "loss": 0.1422,
      "num_input_tokens_seen": 94839552,
      "step": 43940
    },
    {
      "epoch": 7.16884176182708,
      "grad_norm": 0.09155721217393875,
      "learning_rate": 1.1247609701753656e-05,
      "loss": 0.127,
      "num_input_tokens_seen": 94851200,
      "step": 43945
    },
    {
      "epoch": 7.169657422512235,
      "grad_norm": 0.08383266627788544,
      "learning_rate": 1.1241666020295768e-05,
      "loss": 0.2916,
      "num_input_tokens_seen": 94863744,
      "step": 43950
    },
    {
      "epoch": 7.17047308319739,
      "grad_norm": 0.3444756865501404,
      "learning_rate": 1.1235723454158242e-05,
      "loss": 0.084,
      "num_input_tokens_seen": 94873504,
      "step": 43955
    },
    {
      "epoch": 7.171288743882545,
      "grad_norm": 0.026276571676135063,
      "learning_rate": 1.1229782003822803e-05,
      "loss": 0.0992,
      "num_input_tokens_seen": 94883232,
      "step": 43960
    },
    {
      "epoch": 7.1721044045677,
      "grad_norm": 0.2174517661333084,
      "learning_rate": 1.1223841669771113e-05,
      "loss": 0.0146,
      "num_input_tokens_seen": 94893984,
      "step": 43965
    },
    {
      "epoch": 7.172920065252855,
      "grad_norm": 0.09940383583307266,
      "learning_rate": 1.121790245248472e-05,
      "loss": 0.1552,
      "num_input_tokens_seen": 94904960,
      "step": 43970
    },
    {
      "epoch": 7.17373572593801,
      "grad_norm": 3.099872350692749,
      "learning_rate": 1.1211964352445078e-05,
      "loss": 0.1952,
      "num_input_tokens_seen": 94914656,
      "step": 43975
    },
    {
      "epoch": 7.174551386623165,
      "grad_norm": 0.16598708927631378,
      "learning_rate": 1.120602737013357e-05,
      "loss": 0.0845,
      "num_input_tokens_seen": 94925632,
      "step": 43980
    },
    {
      "epoch": 7.1753670473083195,
      "grad_norm": 0.2137947827577591,
      "learning_rate": 1.120009150603147e-05,
      "loss": 0.0076,
      "num_input_tokens_seen": 94937376,
      "step": 43985
    },
    {
      "epoch": 7.176182707993474,
      "grad_norm": 20.040090560913086,
      "learning_rate": 1.1194156760619976e-05,
      "loss": 0.0403,
      "num_input_tokens_seen": 94948608,
      "step": 43990
    },
    {
      "epoch": 7.17699836867863,
      "grad_norm": 0.06532534956932068,
      "learning_rate": 1.1188223134380183e-05,
      "loss": 0.0057,
      "num_input_tokens_seen": 94958368,
      "step": 43995
    },
    {
      "epoch": 7.177814029363785,
      "grad_norm": 0.06551416218280792,
      "learning_rate": 1.1182290627793105e-05,
      "loss": 0.0031,
      "num_input_tokens_seen": 94968576,
      "step": 44000
    },
    {
      "epoch": 7.17862969004894,
      "grad_norm": 30.835641860961914,
      "learning_rate": 1.1176359241339656e-05,
      "loss": 0.1642,
      "num_input_tokens_seen": 94979552,
      "step": 44005
    },
    {
      "epoch": 7.1794453507340945,
      "grad_norm": 0.33559077978134155,
      "learning_rate": 1.1170428975500668e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 94990656,
      "step": 44010
    },
    {
      "epoch": 7.180261011419249,
      "grad_norm": 0.1564725637435913,
      "learning_rate": 1.116449983075688e-05,
      "loss": 0.1024,
      "num_input_tokens_seen": 95000480,
      "step": 44015
    },
    {
      "epoch": 7.181076672104404,
      "grad_norm": 0.18704740703105927,
      "learning_rate": 1.1158571807588924e-05,
      "loss": 0.0098,
      "num_input_tokens_seen": 95011584,
      "step": 44020
    },
    {
      "epoch": 7.18189233278956,
      "grad_norm": 0.07209386676549911,
      "learning_rate": 1.115264490647738e-05,
      "loss": 0.1841,
      "num_input_tokens_seen": 95021824,
      "step": 44025
    },
    {
      "epoch": 7.182707993474715,
      "grad_norm": 0.079127736389637,
      "learning_rate": 1.11467191279027e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 95033696,
      "step": 44030
    },
    {
      "epoch": 7.1835236541598695,
      "grad_norm": 5.181544780731201,
      "learning_rate": 1.1140794472345259e-05,
      "loss": 0.0906,
      "num_input_tokens_seen": 95043936,
      "step": 44035
    },
    {
      "epoch": 7.184339314845024,
      "grad_norm": 0.13229092955589294,
      "learning_rate": 1.1134870940285339e-05,
      "loss": 0.0051,
      "num_input_tokens_seen": 95055040,
      "step": 44040
    },
    {
      "epoch": 7.185154975530179,
      "grad_norm": 2.937776803970337,
      "learning_rate": 1.112894853220313e-05,
      "loss": 0.1312,
      "num_input_tokens_seen": 95066368,
      "step": 44045
    },
    {
      "epoch": 7.185970636215335,
      "grad_norm": 2.413372755050659,
      "learning_rate": 1.1123027248578736e-05,
      "loss": 0.2587,
      "num_input_tokens_seen": 95077728,
      "step": 44050
    },
    {
      "epoch": 7.18678629690049,
      "grad_norm": 0.17712894082069397,
      "learning_rate": 1.1117107089892162e-05,
      "loss": 0.0071,
      "num_input_tokens_seen": 95088160,
      "step": 44055
    },
    {
      "epoch": 7.1876019575856445,
      "grad_norm": 3.4987986087799072,
      "learning_rate": 1.1111188056623328e-05,
      "loss": 0.1675,
      "num_input_tokens_seen": 95099424,
      "step": 44060
    },
    {
      "epoch": 7.188417618270799,
      "grad_norm": 0.4862080514431,
      "learning_rate": 1.1105270149252062e-05,
      "loss": 0.1127,
      "num_input_tokens_seen": 95109600,
      "step": 44065
    },
    {
      "epoch": 7.189233278955954,
      "grad_norm": 0.11356819421052933,
      "learning_rate": 1.10993533682581e-05,
      "loss": 0.1124,
      "num_input_tokens_seen": 95119424,
      "step": 44070
    },
    {
      "epoch": 7.190048939641109,
      "grad_norm": 18.320451736450195,
      "learning_rate": 1.1093437714121085e-05,
      "loss": 0.0375,
      "num_input_tokens_seen": 95130368,
      "step": 44075
    },
    {
      "epoch": 7.190864600326265,
      "grad_norm": 7.893442630767822,
      "learning_rate": 1.108752318732057e-05,
      "loss": 0.0127,
      "num_input_tokens_seen": 95142400,
      "step": 44080
    },
    {
      "epoch": 7.191680261011419,
      "grad_norm": 0.03855288028717041,
      "learning_rate": 1.1081609788336014e-05,
      "loss": 0.2769,
      "num_input_tokens_seen": 95153472,
      "step": 44085
    },
    {
      "epoch": 7.192495921696574,
      "grad_norm": 9.533473014831543,
      "learning_rate": 1.1075697517646794e-05,
      "loss": 0.1633,
      "num_input_tokens_seen": 95163072,
      "step": 44090
    },
    {
      "epoch": 7.193311582381729,
      "grad_norm": 0.13133949041366577,
      "learning_rate": 1.1069786375732181e-05,
      "loss": 0.0058,
      "num_input_tokens_seen": 95174656,
      "step": 44095
    },
    {
      "epoch": 7.194127243066884,
      "grad_norm": 1.044507622718811,
      "learning_rate": 1.1063876363071368e-05,
      "loss": 0.0558,
      "num_input_tokens_seen": 95184960,
      "step": 44100
    },
    {
      "epoch": 7.19494290375204,
      "grad_norm": 3.8548696041107178,
      "learning_rate": 1.1057967480143438e-05,
      "loss": 0.1389,
      "num_input_tokens_seen": 95194848,
      "step": 44105
    },
    {
      "epoch": 7.195758564437194,
      "grad_norm": 0.8807480335235596,
      "learning_rate": 1.1052059727427414e-05,
      "loss": 0.0106,
      "num_input_tokens_seen": 95207072,
      "step": 44110
    },
    {
      "epoch": 7.196574225122349,
      "grad_norm": 0.046531785279512405,
      "learning_rate": 1.1046153105402199e-05,
      "loss": 0.1024,
      "num_input_tokens_seen": 95217920,
      "step": 44115
    },
    {
      "epoch": 7.197389885807504,
      "grad_norm": 0.1135723739862442,
      "learning_rate": 1.1040247614546617e-05,
      "loss": 0.0287,
      "num_input_tokens_seen": 95228224,
      "step": 44120
    },
    {
      "epoch": 7.198205546492659,
      "grad_norm": 0.09810398519039154,
      "learning_rate": 1.1034343255339391e-05,
      "loss": 0.0989,
      "num_input_tokens_seen": 95239008,
      "step": 44125
    },
    {
      "epoch": 7.199021207177814,
      "grad_norm": 0.08268575370311737,
      "learning_rate": 1.1028440028259154e-05,
      "loss": 0.293,
      "num_input_tokens_seen": 95248800,
      "step": 44130
    },
    {
      "epoch": 7.199836867862969,
      "grad_norm": 0.21904577314853668,
      "learning_rate": 1.1022537933784472e-05,
      "loss": 0.121,
      "num_input_tokens_seen": 95259840,
      "step": 44135
    },
    {
      "epoch": 7.200652528548124,
      "grad_norm": 10.687139511108398,
      "learning_rate": 1.1016636972393782e-05,
      "loss": 0.1494,
      "num_input_tokens_seen": 95270848,
      "step": 44140
    },
    {
      "epoch": 7.201468189233279,
      "grad_norm": 0.04208752140402794,
      "learning_rate": 1.101073714456545e-05,
      "loss": 0.0104,
      "num_input_tokens_seen": 95281472,
      "step": 44145
    },
    {
      "epoch": 7.202283849918434,
      "grad_norm": 0.07681772857904434,
      "learning_rate": 1.1004838450777747e-05,
      "loss": 0.1983,
      "num_input_tokens_seen": 95292160,
      "step": 44150
    },
    {
      "epoch": 7.203099510603589,
      "grad_norm": 3.470168113708496,
      "learning_rate": 1.099894089150885e-05,
      "loss": 0.1027,
      "num_input_tokens_seen": 95302208,
      "step": 44155
    },
    {
      "epoch": 7.2039151712887435,
      "grad_norm": 0.12584130465984344,
      "learning_rate": 1.0993044467236843e-05,
      "loss": 0.2133,
      "num_input_tokens_seen": 95312032,
      "step": 44160
    },
    {
      "epoch": 7.204730831973899,
      "grad_norm": 0.26527076959609985,
      "learning_rate": 1.0987149178439726e-05,
      "loss": 0.1062,
      "num_input_tokens_seen": 95323744,
      "step": 44165
    },
    {
      "epoch": 7.205546492659054,
      "grad_norm": 0.12256963551044464,
      "learning_rate": 1.0981255025595394e-05,
      "loss": 0.1266,
      "num_input_tokens_seen": 95333856,
      "step": 44170
    },
    {
      "epoch": 7.206362153344209,
      "grad_norm": 0.020157380029559135,
      "learning_rate": 1.097536200918166e-05,
      "loss": 0.0073,
      "num_input_tokens_seen": 95344864,
      "step": 44175
    },
    {
      "epoch": 7.207177814029364,
      "grad_norm": 0.18879617750644684,
      "learning_rate": 1.0969470129676243e-05,
      "loss": 0.1129,
      "num_input_tokens_seen": 95355520,
      "step": 44180
    },
    {
      "epoch": 7.2079934747145185,
      "grad_norm": 0.1413317173719406,
      "learning_rate": 1.096357938755677e-05,
      "loss": 0.0101,
      "num_input_tokens_seen": 95366752,
      "step": 44185
    },
    {
      "epoch": 7.208809135399674,
      "grad_norm": 0.0799066349864006,
      "learning_rate": 1.0957689783300767e-05,
      "loss": 0.0089,
      "num_input_tokens_seen": 95378304,
      "step": 44190
    },
    {
      "epoch": 7.209624796084829,
      "grad_norm": 3.785212755203247,
      "learning_rate": 1.0951801317385682e-05,
      "loss": 0.1014,
      "num_input_tokens_seen": 95389024,
      "step": 44195
    },
    {
      "epoch": 7.210440456769984,
      "grad_norm": 0.11698129028081894,
      "learning_rate": 1.0945913990288862e-05,
      "loss": 0.0065,
      "num_input_tokens_seen": 95398656,
      "step": 44200
    },
    {
      "epoch": 7.211256117455139,
      "grad_norm": 0.16734565794467926,
      "learning_rate": 1.0940027802487565e-05,
      "loss": 0.0894,
      "num_input_tokens_seen": 95408160,
      "step": 44205
    },
    {
      "epoch": 7.212071778140293,
      "grad_norm": 0.2755676805973053,
      "learning_rate": 1.0934142754458954e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 95419936,
      "step": 44210
    },
    {
      "epoch": 7.212887438825448,
      "grad_norm": 0.10861805826425552,
      "learning_rate": 1.0928258846680097e-05,
      "loss": 0.155,
      "num_input_tokens_seen": 95429952,
      "step": 44215
    },
    {
      "epoch": 7.213703099510604,
      "grad_norm": 0.13752296566963196,
      "learning_rate": 1.092237607962798e-05,
      "loss": 0.0071,
      "num_input_tokens_seen": 95441024,
      "step": 44220
    },
    {
      "epoch": 7.214518760195759,
      "grad_norm": 0.36296749114990234,
      "learning_rate": 1.0916494453779489e-05,
      "loss": 0.0056,
      "num_input_tokens_seen": 95452448,
      "step": 44225
    },
    {
      "epoch": 7.215334420880914,
      "grad_norm": 0.06946375966072083,
      "learning_rate": 1.0910613969611406e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 95462432,
      "step": 44230
    },
    {
      "epoch": 7.216150081566068,
      "grad_norm": 1.6671926975250244,
      "learning_rate": 1.0904734627600448e-05,
      "loss": 0.0131,
      "num_input_tokens_seen": 95473312,
      "step": 44235
    },
    {
      "epoch": 7.216965742251223,
      "grad_norm": 11.202441215515137,
      "learning_rate": 1.0898856428223225e-05,
      "loss": 0.1112,
      "num_input_tokens_seen": 95482624,
      "step": 44240
    },
    {
      "epoch": 7.217781402936378,
      "grad_norm": 0.06486225128173828,
      "learning_rate": 1.0892979371956246e-05,
      "loss": 0.0032,
      "num_input_tokens_seen": 95494304,
      "step": 44245
    },
    {
      "epoch": 7.218597063621534,
      "grad_norm": 0.12349742650985718,
      "learning_rate": 1.088710345927594e-05,
      "loss": 0.006,
      "num_input_tokens_seen": 95505024,
      "step": 44250
    },
    {
      "epoch": 7.219412724306689,
      "grad_norm": 0.8538283109664917,
      "learning_rate": 1.0881228690658634e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 95516224,
      "step": 44255
    },
    {
      "epoch": 7.220228384991843,
      "grad_norm": 0.049786556512117386,
      "learning_rate": 1.087535506658057e-05,
      "loss": 0.0028,
      "num_input_tokens_seen": 95527168,
      "step": 44260
    },
    {
      "epoch": 7.221044045676998,
      "grad_norm": 0.08179260045289993,
      "learning_rate": 1.086948258751789e-05,
      "loss": 0.0795,
      "num_input_tokens_seen": 95537152,
      "step": 44265
    },
    {
      "epoch": 7.221859706362153,
      "grad_norm": 10.488143920898438,
      "learning_rate": 1.0863611253946651e-05,
      "loss": 0.0797,
      "num_input_tokens_seen": 95549248,
      "step": 44270
    },
    {
      "epoch": 7.222675367047309,
      "grad_norm": 4.596770763397217,
      "learning_rate": 1.08577410663428e-05,
      "loss": 0.0929,
      "num_input_tokens_seen": 95560704,
      "step": 44275
    },
    {
      "epoch": 7.2234910277324635,
      "grad_norm": 0.021962575614452362,
      "learning_rate": 1.0851872025182225e-05,
      "loss": 0.0155,
      "num_input_tokens_seen": 95572864,
      "step": 44280
    },
    {
      "epoch": 7.224306688417618,
      "grad_norm": 0.9708815217018127,
      "learning_rate": 1.084600413094069e-05,
      "loss": 0.0589,
      "num_input_tokens_seen": 95582944,
      "step": 44285
    },
    {
      "epoch": 7.225122349102773,
      "grad_norm": 0.07511153817176819,
      "learning_rate": 1.0840137384093876e-05,
      "loss": 0.0054,
      "num_input_tokens_seen": 95593472,
      "step": 44290
    },
    {
      "epoch": 7.225938009787928,
      "grad_norm": 1.9997107982635498,
      "learning_rate": 1.0834271785117376e-05,
      "loss": 0.115,
      "num_input_tokens_seen": 95604064,
      "step": 44295
    },
    {
      "epoch": 7.226753670473083,
      "grad_norm": 0.026333212852478027,
      "learning_rate": 1.0828407334486676e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 95616192,
      "step": 44300
    },
    {
      "epoch": 7.2275693311582385,
      "grad_norm": 0.3405097424983978,
      "learning_rate": 1.0822544032677187e-05,
      "loss": 0.0114,
      "num_input_tokens_seen": 95626976,
      "step": 44305
    },
    {
      "epoch": 7.228384991843393,
      "grad_norm": 0.08290238678455353,
      "learning_rate": 1.0816681880164215e-05,
      "loss": 0.0041,
      "num_input_tokens_seen": 95635328,
      "step": 44310
    },
    {
      "epoch": 7.229200652528548,
      "grad_norm": 0.12996086478233337,
      "learning_rate": 1.0810820877422973e-05,
      "loss": 0.0345,
      "num_input_tokens_seen": 95645248,
      "step": 44315
    },
    {
      "epoch": 7.230016313213703,
      "grad_norm": 0.2046303153038025,
      "learning_rate": 1.0804961024928587e-05,
      "loss": 0.1687,
      "num_input_tokens_seen": 95656928,
      "step": 44320
    },
    {
      "epoch": 7.230831973898858,
      "grad_norm": 0.09022431820631027,
      "learning_rate": 1.0799102323156082e-05,
      "loss": 0.0028,
      "num_input_tokens_seen": 95667200,
      "step": 44325
    },
    {
      "epoch": 7.231647634584013,
      "grad_norm": 0.1085621789097786,
      "learning_rate": 1.0793244772580402e-05,
      "loss": 0.2045,
      "num_input_tokens_seen": 95677248,
      "step": 44330
    },
    {
      "epoch": 7.232463295269168,
      "grad_norm": 0.1027090772986412,
      "learning_rate": 1.0787388373676374e-05,
      "loss": 0.0984,
      "num_input_tokens_seen": 95687104,
      "step": 44335
    },
    {
      "epoch": 7.233278955954323,
      "grad_norm": 0.12635067105293274,
      "learning_rate": 1.0781533126918767e-05,
      "loss": 0.107,
      "num_input_tokens_seen": 95697792,
      "step": 44340
    },
    {
      "epoch": 7.234094616639478,
      "grad_norm": 0.08168936520814896,
      "learning_rate": 1.077567903278223e-05,
      "loss": 0.1504,
      "num_input_tokens_seen": 95708800,
      "step": 44345
    },
    {
      "epoch": 7.234910277324633,
      "grad_norm": 0.057498153299093246,
      "learning_rate": 1.0769826091741323e-05,
      "loss": 0.0113,
      "num_input_tokens_seen": 95719296,
      "step": 44350
    },
    {
      "epoch": 7.235725938009788,
      "grad_norm": 0.045036740601062775,
      "learning_rate": 1.0763974304270516e-05,
      "loss": 0.0309,
      "num_input_tokens_seen": 95730336,
      "step": 44355
    },
    {
      "epoch": 7.236541598694943,
      "grad_norm": 0.09333490580320358,
      "learning_rate": 1.0758123670844186e-05,
      "loss": 0.356,
      "num_input_tokens_seen": 95741760,
      "step": 44360
    },
    {
      "epoch": 7.237357259380098,
      "grad_norm": 0.07128450274467468,
      "learning_rate": 1.0752274191936611e-05,
      "loss": 0.1928,
      "num_input_tokens_seen": 95752896,
      "step": 44365
    },
    {
      "epoch": 7.238172920065253,
      "grad_norm": 20.4687442779541,
      "learning_rate": 1.0746425868021986e-05,
      "loss": 0.3655,
      "num_input_tokens_seen": 95763968,
      "step": 44370
    },
    {
      "epoch": 7.238988580750408,
      "grad_norm": 0.08805769681930542,
      "learning_rate": 1.07405786995744e-05,
      "loss": 0.0604,
      "num_input_tokens_seen": 95774528,
      "step": 44375
    },
    {
      "epoch": 7.239804241435563,
      "grad_norm": 0.07074488699436188,
      "learning_rate": 1.0734732687067856e-05,
      "loss": 0.0038,
      "num_input_tokens_seen": 95785600,
      "step": 44380
    },
    {
      "epoch": 7.240619902120717,
      "grad_norm": 0.7448371052742004,
      "learning_rate": 1.0728887830976261e-05,
      "loss": 0.1705,
      "num_input_tokens_seen": 95795104,
      "step": 44385
    },
    {
      "epoch": 7.241435562805873,
      "grad_norm": 0.060588471591472626,
      "learning_rate": 1.0723044131773433e-05,
      "loss": 0.0794,
      "num_input_tokens_seen": 95807104,
      "step": 44390
    },
    {
      "epoch": 7.242251223491028,
      "grad_norm": 0.10568378120660782,
      "learning_rate": 1.0717201589933085e-05,
      "loss": 0.1016,
      "num_input_tokens_seen": 95816832,
      "step": 44395
    },
    {
      "epoch": 7.243066884176183,
      "grad_norm": 0.09632561355829239,
      "learning_rate": 1.0711360205928847e-05,
      "loss": 0.004,
      "num_input_tokens_seen": 95826688,
      "step": 44400
    },
    {
      "epoch": 7.2438825448613375,
      "grad_norm": 0.04835936427116394,
      "learning_rate": 1.070551998023425e-05,
      "loss": 0.1001,
      "num_input_tokens_seen": 95835904,
      "step": 44405
    },
    {
      "epoch": 7.244698205546492,
      "grad_norm": 3.753042459487915,
      "learning_rate": 1.0699680913322736e-05,
      "loss": 0.1757,
      "num_input_tokens_seen": 95847136,
      "step": 44410
    },
    {
      "epoch": 7.245513866231648,
      "grad_norm": 0.05000664293766022,
      "learning_rate": 1.0693843005667633e-05,
      "loss": 0.2018,
      "num_input_tokens_seen": 95858368,
      "step": 44415
    },
    {
      "epoch": 7.246329526916803,
      "grad_norm": 0.04885876178741455,
      "learning_rate": 1.0688006257742214e-05,
      "loss": 0.133,
      "num_input_tokens_seen": 95868544,
      "step": 44420
    },
    {
      "epoch": 7.247145187601958,
      "grad_norm": 0.11001405119895935,
      "learning_rate": 1.0682170670019628e-05,
      "loss": 0.0048,
      "num_input_tokens_seen": 95878944,
      "step": 44425
    },
    {
      "epoch": 7.2479608482871125,
      "grad_norm": 4.917969226837158,
      "learning_rate": 1.0676336242972934e-05,
      "loss": 0.0092,
      "num_input_tokens_seen": 95889792,
      "step": 44430
    },
    {
      "epoch": 7.248776508972267,
      "grad_norm": 0.060306940227746964,
      "learning_rate": 1.0670502977075103e-05,
      "loss": 0.1261,
      "num_input_tokens_seen": 95900896,
      "step": 44435
    },
    {
      "epoch": 7.249592169657422,
      "grad_norm": 0.27811291813850403,
      "learning_rate": 1.0664670872799006e-05,
      "loss": 0.0085,
      "num_input_tokens_seen": 95911808,
      "step": 44440
    },
    {
      "epoch": 7.250407830342578,
      "grad_norm": 0.09314201772212982,
      "learning_rate": 1.065883993061742e-05,
      "loss": 0.1007,
      "num_input_tokens_seen": 95921696,
      "step": 44445
    },
    {
      "epoch": 7.251223491027733,
      "grad_norm": 0.112932488322258,
      "learning_rate": 1.065301015100304e-05,
      "loss": 0.0883,
      "num_input_tokens_seen": 95931104,
      "step": 44450
    },
    {
      "epoch": 7.2520391517128875,
      "grad_norm": 0.17163510620594025,
      "learning_rate": 1.0647181534428455e-05,
      "loss": 0.205,
      "num_input_tokens_seen": 95941376,
      "step": 44455
    },
    {
      "epoch": 7.252854812398042,
      "grad_norm": 0.21094530820846558,
      "learning_rate": 1.0641354081366161e-05,
      "loss": 0.0217,
      "num_input_tokens_seen": 95950496,
      "step": 44460
    },
    {
      "epoch": 7.253670473083197,
      "grad_norm": 4.583967685699463,
      "learning_rate": 1.0635527792288558e-05,
      "loss": 0.0568,
      "num_input_tokens_seen": 95961824,
      "step": 44465
    },
    {
      "epoch": 7.254486133768353,
      "grad_norm": 0.3027603030204773,
      "learning_rate": 1.0629702667667959e-05,
      "loss": 0.0733,
      "num_input_tokens_seen": 95972768,
      "step": 44470
    },
    {
      "epoch": 7.255301794453508,
      "grad_norm": 0.38592761754989624,
      "learning_rate": 1.0623878707976575e-05,
      "loss": 0.0104,
      "num_input_tokens_seen": 95985024,
      "step": 44475
    },
    {
      "epoch": 7.2561174551386625,
      "grad_norm": 0.1534576267004013,
      "learning_rate": 1.0618055913686525e-05,
      "loss": 0.005,
      "num_input_tokens_seen": 95996768,
      "step": 44480
    },
    {
      "epoch": 7.256933115823817,
      "grad_norm": 12.545076370239258,
      "learning_rate": 1.061223428526984e-05,
      "loss": 0.0407,
      "num_input_tokens_seen": 96007328,
      "step": 44485
    },
    {
      "epoch": 7.257748776508972,
      "grad_norm": 0.14382946491241455,
      "learning_rate": 1.0606413823198444e-05,
      "loss": 0.0059,
      "num_input_tokens_seen": 96017472,
      "step": 44490
    },
    {
      "epoch": 7.258564437194127,
      "grad_norm": 3.0954854488372803,
      "learning_rate": 1.0600594527944174e-05,
      "loss": 0.006,
      "num_input_tokens_seen": 96026048,
      "step": 44495
    },
    {
      "epoch": 7.259380097879283,
      "grad_norm": 0.0807352140545845,
      "learning_rate": 1.0594776399978776e-05,
      "loss": 0.008,
      "num_input_tokens_seen": 96037024,
      "step": 44500
    },
    {
      "epoch": 7.260195758564437,
      "grad_norm": 0.10835316777229309,
      "learning_rate": 1.0588959439773893e-05,
      "loss": 0.0039,
      "num_input_tokens_seen": 96048640,
      "step": 44505
    },
    {
      "epoch": 7.261011419249592,
      "grad_norm": 0.8585956692695618,
      "learning_rate": 1.058314364780108e-05,
      "loss": 0.0794,
      "num_input_tokens_seen": 96059968,
      "step": 44510
    },
    {
      "epoch": 7.261827079934747,
      "grad_norm": 0.14995978772640228,
      "learning_rate": 1.0577329024531792e-05,
      "loss": 0.1314,
      "num_input_tokens_seen": 96070592,
      "step": 44515
    },
    {
      "epoch": 7.262642740619902,
      "grad_norm": 4.186001777648926,
      "learning_rate": 1.0571515570437396e-05,
      "loss": 0.0596,
      "num_input_tokens_seen": 96080768,
      "step": 44520
    },
    {
      "epoch": 7.263458401305057,
      "grad_norm": 0.19002677500247955,
      "learning_rate": 1.0565703285989154e-05,
      "loss": 0.099,
      "num_input_tokens_seen": 96091584,
      "step": 44525
    },
    {
      "epoch": 7.264274061990212,
      "grad_norm": 0.21625880897045135,
      "learning_rate": 1.0559892171658245e-05,
      "loss": 0.0037,
      "num_input_tokens_seen": 96103264,
      "step": 44530
    },
    {
      "epoch": 7.265089722675367,
      "grad_norm": 0.08097948133945465,
      "learning_rate": 1.0554082227915743e-05,
      "loss": 0.2587,
      "num_input_tokens_seen": 96112992,
      "step": 44535
    },
    {
      "epoch": 7.265905383360522,
      "grad_norm": 0.18207955360412598,
      "learning_rate": 1.0548273455232634e-05,
      "loss": 0.1109,
      "num_input_tokens_seen": 96124480,
      "step": 44540
    },
    {
      "epoch": 7.266721044045677,
      "grad_norm": 0.4924478530883789,
      "learning_rate": 1.0542465854079806e-05,
      "loss": 0.0302,
      "num_input_tokens_seen": 96134848,
      "step": 44545
    },
    {
      "epoch": 7.267536704730832,
      "grad_norm": 0.08942606300115585,
      "learning_rate": 1.0536659424928044e-05,
      "loss": 0.0074,
      "num_input_tokens_seen": 96146176,
      "step": 44550
    },
    {
      "epoch": 7.268352365415987,
      "grad_norm": 0.08673729747533798,
      "learning_rate": 1.0530854168248064e-05,
      "loss": 0.017,
      "num_input_tokens_seen": 96157824,
      "step": 44555
    },
    {
      "epoch": 7.269168026101142,
      "grad_norm": 0.011961083859205246,
      "learning_rate": 1.052505008451046e-05,
      "loss": 0.0036,
      "num_input_tokens_seen": 96168992,
      "step": 44560
    },
    {
      "epoch": 7.269983686786297,
      "grad_norm": 1.6557157039642334,
      "learning_rate": 1.0519247174185742e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 96179712,
      "step": 44565
    },
    {
      "epoch": 7.270799347471452,
      "grad_norm": 0.4414486885070801,
      "learning_rate": 1.0513445437744323e-05,
      "loss": 0.0958,
      "num_input_tokens_seen": 96190656,
      "step": 44570
    },
    {
      "epoch": 7.271615008156607,
      "grad_norm": 4.245144367218018,
      "learning_rate": 1.0507644875656523e-05,
      "loss": 0.0887,
      "num_input_tokens_seen": 96201120,
      "step": 44575
    },
    {
      "epoch": 7.2724306688417615,
      "grad_norm": 0.10975703597068787,
      "learning_rate": 1.0501845488392558e-05,
      "loss": 0.0034,
      "num_input_tokens_seen": 96210304,
      "step": 44580
    },
    {
      "epoch": 7.273246329526917,
      "grad_norm": 0.08201498538255692,
      "learning_rate": 1.0496047276422554e-05,
      "loss": 0.1695,
      "num_input_tokens_seen": 96221472,
      "step": 44585
    },
    {
      "epoch": 7.274061990212072,
      "grad_norm": 0.04187267646193504,
      "learning_rate": 1.0490250240216562e-05,
      "loss": 0.0037,
      "num_input_tokens_seen": 96232736,
      "step": 44590
    },
    {
      "epoch": 7.274877650897227,
      "grad_norm": 0.12512241303920746,
      "learning_rate": 1.0484454380244505e-05,
      "loss": 0.2199,
      "num_input_tokens_seen": 96244352,
      "step": 44595
    },
    {
      "epoch": 7.275693311582382,
      "grad_norm": 0.023977143689990044,
      "learning_rate": 1.0478659696976225e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 96254240,
      "step": 44600
    },
    {
      "epoch": 7.2765089722675365,
      "grad_norm": 0.38099923729896545,
      "learning_rate": 1.0472866190881473e-05,
      "loss": 0.0032,
      "num_input_tokens_seen": 96265696,
      "step": 44605
    },
    {
      "epoch": 7.277324632952691,
      "grad_norm": 0.1553783416748047,
      "learning_rate": 1.0467073862429897e-05,
      "loss": 0.1281,
      "num_input_tokens_seen": 96277664,
      "step": 44610
    },
    {
      "epoch": 7.278140293637847,
      "grad_norm": 0.9362660050392151,
      "learning_rate": 1.0461282712091053e-05,
      "loss": 0.0076,
      "num_input_tokens_seen": 96288448,
      "step": 44615
    },
    {
      "epoch": 7.278955954323002,
      "grad_norm": 0.1234993264079094,
      "learning_rate": 1.0455492740334399e-05,
      "loss": 0.0035,
      "num_input_tokens_seen": 96298432,
      "step": 44620
    },
    {
      "epoch": 7.279771615008157,
      "grad_norm": 0.007566182874143124,
      "learning_rate": 1.0449703947629305e-05,
      "loss": 0.0066,
      "num_input_tokens_seen": 96309568,
      "step": 44625
    },
    {
      "epoch": 7.280587275693311,
      "grad_norm": 0.09912440925836563,
      "learning_rate": 1.0443916334445034e-05,
      "loss": 0.0019,
      "num_input_tokens_seen": 96319968,
      "step": 44630
    },
    {
      "epoch": 7.281402936378466,
      "grad_norm": 0.07563468813896179,
      "learning_rate": 1.0438129901250762e-05,
      "loss": 0.0039,
      "num_input_tokens_seen": 96331232,
      "step": 44635
    },
    {
      "epoch": 7.282218597063622,
      "grad_norm": 0.3058144450187683,
      "learning_rate": 1.0432344648515569e-05,
      "loss": 0.0046,
      "num_input_tokens_seen": 96341696,
      "step": 44640
    },
    {
      "epoch": 7.283034257748777,
      "grad_norm": 0.07371566444635391,
      "learning_rate": 1.042656057670843e-05,
      "loss": 0.1109,
      "num_input_tokens_seen": 96352160,
      "step": 44645
    },
    {
      "epoch": 7.283849918433932,
      "grad_norm": 0.10091523081064224,
      "learning_rate": 1.042077768629824e-05,
      "loss": 0.0042,
      "num_input_tokens_seen": 96362592,
      "step": 44650
    },
    {
      "epoch": 7.284665579119086,
      "grad_norm": 4.34228515625,
      "learning_rate": 1.0414995977753772e-05,
      "loss": 0.0769,
      "num_input_tokens_seen": 96375424,
      "step": 44655
    },
    {
      "epoch": 7.285481239804241,
      "grad_norm": 0.24394764006137848,
      "learning_rate": 1.0409215451543746e-05,
      "loss": 0.1543,
      "num_input_tokens_seen": 96387296,
      "step": 44660
    },
    {
      "epoch": 7.286296900489396,
      "grad_norm": 4.6436004638671875,
      "learning_rate": 1.0403436108136747e-05,
      "loss": 0.1301,
      "num_input_tokens_seen": 96397632,
      "step": 44665
    },
    {
      "epoch": 7.287112561174552,
      "grad_norm": 0.01952068693935871,
      "learning_rate": 1.039765794800128e-05,
      "loss": 0.0031,
      "num_input_tokens_seen": 96408096,
      "step": 44670
    },
    {
      "epoch": 7.287928221859707,
      "grad_norm": 0.018300577998161316,
      "learning_rate": 1.0391880971605749e-05,
      "loss": 0.0037,
      "num_input_tokens_seen": 96417408,
      "step": 44675
    },
    {
      "epoch": 7.288743882544861,
      "grad_norm": 0.2479696422815323,
      "learning_rate": 1.0386105179418467e-05,
      "loss": 0.1203,
      "num_input_tokens_seen": 96427328,
      "step": 44680
    },
    {
      "epoch": 7.289559543230016,
      "grad_norm": 0.08731896430253983,
      "learning_rate": 1.0380330571907654e-05,
      "loss": 0.0028,
      "num_input_tokens_seen": 96437920,
      "step": 44685
    },
    {
      "epoch": 7.290375203915171,
      "grad_norm": 4.4533305168151855,
      "learning_rate": 1.037455714954142e-05,
      "loss": 0.0815,
      "num_input_tokens_seen": 96447872,
      "step": 44690
    },
    {
      "epoch": 7.291190864600326,
      "grad_norm": 0.08045049011707306,
      "learning_rate": 1.0368784912787794e-05,
      "loss": 0.0029,
      "num_input_tokens_seen": 96458592,
      "step": 44695
    },
    {
      "epoch": 7.2920065252854815,
      "grad_norm": 0.09697709232568741,
      "learning_rate": 1.03630138621147e-05,
      "loss": 0.0393,
      "num_input_tokens_seen": 96469312,
      "step": 44700
    },
    {
      "epoch": 7.292822185970636,
      "grad_norm": 0.0496763177216053,
      "learning_rate": 1.035724399798997e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 96478976,
      "step": 44705
    },
    {
      "epoch": 7.293637846655791,
      "grad_norm": 0.04743843898177147,
      "learning_rate": 1.035147532088134e-05,
      "loss": 0.0176,
      "num_input_tokens_seen": 96489952,
      "step": 44710
    },
    {
      "epoch": 7.294453507340946,
      "grad_norm": 0.024325326085090637,
      "learning_rate": 1.0345707831256443e-05,
      "loss": 0.1312,
      "num_input_tokens_seen": 96500672,
      "step": 44715
    },
    {
      "epoch": 7.295269168026101,
      "grad_norm": 5.337332248687744,
      "learning_rate": 1.0339941529582828e-05,
      "loss": 0.2575,
      "num_input_tokens_seen": 96511488,
      "step": 44720
    },
    {
      "epoch": 7.2960848287112565,
      "grad_norm": 0.0264720618724823,
      "learning_rate": 1.0334176416327935e-05,
      "loss": 0.1006,
      "num_input_tokens_seen": 96522336,
      "step": 44725
    },
    {
      "epoch": 7.296900489396411,
      "grad_norm": 0.0639074295759201,
      "learning_rate": 1.0328412491959104e-05,
      "loss": 0.2508,
      "num_input_tokens_seen": 96533088,
      "step": 44730
    },
    {
      "epoch": 7.297716150081566,
      "grad_norm": 0.10456375032663345,
      "learning_rate": 1.0322649756943611e-05,
      "loss": 0.0031,
      "num_input_tokens_seen": 96543552,
      "step": 44735
    },
    {
      "epoch": 7.298531810766721,
      "grad_norm": 0.02855144441127777,
      "learning_rate": 1.0316888211748601e-05,
      "loss": 0.1592,
      "num_input_tokens_seen": 96554432,
      "step": 44740
    },
    {
      "epoch": 7.299347471451876,
      "grad_norm": 28.35318374633789,
      "learning_rate": 1.0311127856841136e-05,
      "loss": 0.061,
      "num_input_tokens_seen": 96565536,
      "step": 44745
    },
    {
      "epoch": 7.300163132137031,
      "grad_norm": 0.12948469817638397,
      "learning_rate": 1.0305368692688174e-05,
      "loss": 0.0048,
      "num_input_tokens_seen": 96576576,
      "step": 44750
    },
    {
      "epoch": 7.300978792822186,
      "grad_norm": 17.168794631958008,
      "learning_rate": 1.0299610719756587e-05,
      "loss": 0.0408,
      "num_input_tokens_seen": 96588640,
      "step": 44755
    },
    {
      "epoch": 7.301794453507341,
      "grad_norm": 0.06320144981145859,
      "learning_rate": 1.0293853938513142e-05,
      "loss": 0.0941,
      "num_input_tokens_seen": 96600160,
      "step": 44760
    },
    {
      "epoch": 7.302610114192496,
      "grad_norm": 0.07234900444746017,
      "learning_rate": 1.028809834942451e-05,
      "loss": 0.3319,
      "num_input_tokens_seen": 96610944,
      "step": 44765
    },
    {
      "epoch": 7.303425774877651,
      "grad_norm": 2.9788599014282227,
      "learning_rate": 1.028234395295728e-05,
      "loss": 0.1095,
      "num_input_tokens_seen": 96620960,
      "step": 44770
    },
    {
      "epoch": 7.304241435562806,
      "grad_norm": 0.05842946842312813,
      "learning_rate": 1.0276590749577924e-05,
      "loss": 0.2737,
      "num_input_tokens_seen": 96630592,
      "step": 44775
    },
    {
      "epoch": 7.30505709624796,
      "grad_norm": 0.01867961883544922,
      "learning_rate": 1.027083873975283e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 96642368,
      "step": 44780
    },
    {
      "epoch": 7.305872756933116,
      "grad_norm": 0.20065642893314362,
      "learning_rate": 1.0265087923948283e-05,
      "loss": 0.0117,
      "num_input_tokens_seen": 96654464,
      "step": 44785
    },
    {
      "epoch": 7.306688417618271,
      "grad_norm": 16.948871612548828,
      "learning_rate": 1.0259338302630472e-05,
      "loss": 0.1915,
      "num_input_tokens_seen": 96664992,
      "step": 44790
    },
    {
      "epoch": 7.307504078303426,
      "grad_norm": 0.08691882342100143,
      "learning_rate": 1.025358987626549e-05,
      "loss": 0.1411,
      "num_input_tokens_seen": 96675168,
      "step": 44795
    },
    {
      "epoch": 7.308319738988581,
      "grad_norm": 20.442317962646484,
      "learning_rate": 1.0247842645319339e-05,
      "loss": 0.0929,
      "num_input_tokens_seen": 96684992,
      "step": 44800
    },
    {
      "epoch": 7.309135399673735,
      "grad_norm": 0.11712927371263504,
      "learning_rate": 1.0242096610257911e-05,
      "loss": 0.0043,
      "num_input_tokens_seen": 96695808,
      "step": 44805
    },
    {
      "epoch": 7.309951060358891,
      "grad_norm": 0.2057601511478424,
      "learning_rate": 1.0236351771547014e-05,
      "loss": 0.0928,
      "num_input_tokens_seen": 96707840,
      "step": 44810
    },
    {
      "epoch": 7.310766721044046,
      "grad_norm": 5.6143999099731445,
      "learning_rate": 1.0230608129652355e-05,
      "loss": 0.1065,
      "num_input_tokens_seen": 96718784,
      "step": 44815
    },
    {
      "epoch": 7.311582381729201,
      "grad_norm": 0.10936146974563599,
      "learning_rate": 1.022486568503954e-05,
      "loss": 0.003,
      "num_input_tokens_seen": 96729376,
      "step": 44820
    },
    {
      "epoch": 7.3123980424143555,
      "grad_norm": 0.06403083354234695,
      "learning_rate": 1.0219124438174076e-05,
      "loss": 0.0043,
      "num_input_tokens_seen": 96740640,
      "step": 44825
    },
    {
      "epoch": 7.31321370309951,
      "grad_norm": 0.021619712933897972,
      "learning_rate": 1.0213384389521385e-05,
      "loss": 0.0052,
      "num_input_tokens_seen": 96751104,
      "step": 44830
    },
    {
      "epoch": 7.314029363784665,
      "grad_norm": 0.05380615219473839,
      "learning_rate": 1.0207645539546784e-05,
      "loss": 0.2215,
      "num_input_tokens_seen": 96760672,
      "step": 44835
    },
    {
      "epoch": 7.314845024469821,
      "grad_norm": 0.08868793398141861,
      "learning_rate": 1.020190788871549e-05,
      "loss": 0.0137,
      "num_input_tokens_seen": 96771840,
      "step": 44840
    },
    {
      "epoch": 7.315660685154976,
      "grad_norm": 0.0648154690861702,
      "learning_rate": 1.0196171437492627e-05,
      "loss": 0.1151,
      "num_input_tokens_seen": 96782592,
      "step": 44845
    },
    {
      "epoch": 7.3164763458401305,
      "grad_norm": 0.03126976639032364,
      "learning_rate": 1.0190436186343218e-05,
      "loss": 0.1971,
      "num_input_tokens_seen": 96794016,
      "step": 44850
    },
    {
      "epoch": 7.317292006525285,
      "grad_norm": 0.10227061808109283,
      "learning_rate": 1.01847021357322e-05,
      "loss": 0.1948,
      "num_input_tokens_seen": 96803968,
      "step": 44855
    },
    {
      "epoch": 7.31810766721044,
      "grad_norm": 0.17148098349571228,
      "learning_rate": 1.0178969286124396e-05,
      "loss": 0.1504,
      "num_input_tokens_seen": 96814304,
      "step": 44860
    },
    {
      "epoch": 7.318923327895595,
      "grad_norm": 0.04482850804924965,
      "learning_rate": 1.0173237637984542e-05,
      "loss": 0.0046,
      "num_input_tokens_seen": 96824128,
      "step": 44865
    },
    {
      "epoch": 7.319738988580751,
      "grad_norm": 0.06380118429660797,
      "learning_rate": 1.016750719177727e-05,
      "loss": 0.113,
      "num_input_tokens_seen": 96835712,
      "step": 44870
    },
    {
      "epoch": 7.3205546492659055,
      "grad_norm": 0.08428000658750534,
      "learning_rate": 1.016177794796713e-05,
      "loss": 0.0032,
      "num_input_tokens_seen": 96846272,
      "step": 44875
    },
    {
      "epoch": 7.32137030995106,
      "grad_norm": 0.07979920506477356,
      "learning_rate": 1.0156049907018562e-05,
      "loss": 0.0058,
      "num_input_tokens_seen": 96857408,
      "step": 44880
    },
    {
      "epoch": 7.322185970636215,
      "grad_norm": 0.06582298129796982,
      "learning_rate": 1.0150323069395901e-05,
      "loss": 0.1055,
      "num_input_tokens_seen": 96868032,
      "step": 44885
    },
    {
      "epoch": 7.32300163132137,
      "grad_norm": 0.053582530468702316,
      "learning_rate": 1.01445974355634e-05,
      "loss": 0.1086,
      "num_input_tokens_seen": 96879136,
      "step": 44890
    },
    {
      "epoch": 7.323817292006526,
      "grad_norm": 0.08738100528717041,
      "learning_rate": 1.0138873005985208e-05,
      "loss": 0.0854,
      "num_input_tokens_seen": 96890016,
      "step": 44895
    },
    {
      "epoch": 7.3246329526916805,
      "grad_norm": 0.057709500193595886,
      "learning_rate": 1.0133149781125365e-05,
      "loss": 0.1186,
      "num_input_tokens_seen": 96900160,
      "step": 44900
    },
    {
      "epoch": 7.325448613376835,
      "grad_norm": 0.277441143989563,
      "learning_rate": 1.0127427761447842e-05,
      "loss": 0.0941,
      "num_input_tokens_seen": 96909856,
      "step": 44905
    },
    {
      "epoch": 7.32626427406199,
      "grad_norm": 0.19494812190532684,
      "learning_rate": 1.012170694741649e-05,
      "loss": 0.0052,
      "num_input_tokens_seen": 96921280,
      "step": 44910
    },
    {
      "epoch": 7.327079934747145,
      "grad_norm": 6.171853542327881,
      "learning_rate": 1.0115987339495061e-05,
      "loss": 0.0378,
      "num_input_tokens_seen": 96931136,
      "step": 44915
    },
    {
      "epoch": 7.327895595432301,
      "grad_norm": 0.04178924486041069,
      "learning_rate": 1.0110268938147222e-05,
      "loss": 0.0029,
      "num_input_tokens_seen": 96942112,
      "step": 44920
    },
    {
      "epoch": 7.328711256117455,
      "grad_norm": 0.10311181098222733,
      "learning_rate": 1.0104551743836532e-05,
      "loss": 0.0384,
      "num_input_tokens_seen": 96953664,
      "step": 44925
    },
    {
      "epoch": 7.32952691680261,
      "grad_norm": 0.05509466305375099,
      "learning_rate": 1.0098835757026457e-05,
      "loss": 0.0775,
      "num_input_tokens_seen": 96964800,
      "step": 44930
    },
    {
      "epoch": 7.330342577487765,
      "grad_norm": 0.2352639138698578,
      "learning_rate": 1.009312097818036e-05,
      "loss": 0.0084,
      "num_input_tokens_seen": 96975680,
      "step": 44935
    },
    {
      "epoch": 7.33115823817292,
      "grad_norm": 0.07198192179203033,
      "learning_rate": 1.0087407407761515e-05,
      "loss": 0.0589,
      "num_input_tokens_seen": 96986496,
      "step": 44940
    },
    {
      "epoch": 7.331973898858075,
      "grad_norm": 0.13894616067409515,
      "learning_rate": 1.0081695046233091e-05,
      "loss": 0.0047,
      "num_input_tokens_seen": 96997184,
      "step": 44945
    },
    {
      "epoch": 7.33278955954323,
      "grad_norm": 0.12491953372955322,
      "learning_rate": 1.0075983894058163e-05,
      "loss": 0.0037,
      "num_input_tokens_seen": 97008416,
      "step": 44950
    },
    {
      "epoch": 7.333605220228385,
      "grad_norm": 0.06016465276479721,
      "learning_rate": 1.0070273951699704e-05,
      "loss": 0.0048,
      "num_input_tokens_seen": 97020000,
      "step": 44955
    },
    {
      "epoch": 7.33442088091354,
      "grad_norm": 0.557112455368042,
      "learning_rate": 1.0064565219620593e-05,
      "loss": 0.0069,
      "num_input_tokens_seen": 97030752,
      "step": 44960
    },
    {
      "epoch": 7.335236541598695,
      "grad_norm": 0.12895114719867706,
      "learning_rate": 1.0058857698283603e-05,
      "loss": 0.3145,
      "num_input_tokens_seen": 97040416,
      "step": 44965
    },
    {
      "epoch": 7.33605220228385,
      "grad_norm": 0.05453105270862579,
      "learning_rate": 1.0053151388151418e-05,
      "loss": 0.214,
      "num_input_tokens_seen": 97051936,
      "step": 44970
    },
    {
      "epoch": 7.3368678629690045,
      "grad_norm": 0.07821512222290039,
      "learning_rate": 1.0047446289686615e-05,
      "loss": 0.0877,
      "num_input_tokens_seen": 97060864,
      "step": 44975
    },
    {
      "epoch": 7.33768352365416,
      "grad_norm": 0.22982530295848846,
      "learning_rate": 1.0041742403351693e-05,
      "loss": 0.01,
      "num_input_tokens_seen": 97073376,
      "step": 44980
    },
    {
      "epoch": 7.338499184339315,
      "grad_norm": 0.11923679709434509,
      "learning_rate": 1.0036039729609029e-05,
      "loss": 0.0064,
      "num_input_tokens_seen": 97085696,
      "step": 44985
    },
    {
      "epoch": 7.33931484502447,
      "grad_norm": 0.05471346154808998,
      "learning_rate": 1.0030338268920911e-05,
      "loss": 0.0111,
      "num_input_tokens_seen": 97097856,
      "step": 44990
    },
    {
      "epoch": 7.340130505709625,
      "grad_norm": 0.04758811369538307,
      "learning_rate": 1.0024638021749527e-05,
      "loss": 0.0056,
      "num_input_tokens_seen": 97109280,
      "step": 44995
    },
    {
      "epoch": 7.3409461663947795,
      "grad_norm": 0.05021590366959572,
      "learning_rate": 1.001893898855697e-05,
      "loss": 0.1206,
      "num_input_tokens_seen": 97121728,
      "step": 45000
    },
    {
      "epoch": 7.341761827079935,
      "grad_norm": 0.07956917583942413,
      "learning_rate": 1.0013241169805232e-05,
      "loss": 0.0037,
      "num_input_tokens_seen": 97132480,
      "step": 45005
    },
    {
      "epoch": 7.34257748776509,
      "grad_norm": 0.06605811417102814,
      "learning_rate": 1.0007544565956206e-05,
      "loss": 0.0028,
      "num_input_tokens_seen": 97143904,
      "step": 45010
    },
    {
      "epoch": 7.343393148450245,
      "grad_norm": 0.0318470299243927,
      "learning_rate": 1.0001849177471687e-05,
      "loss": 0.0037,
      "num_input_tokens_seen": 97154816,
      "step": 45015
    },
    {
      "epoch": 7.3442088091354,
      "grad_norm": 22.52471923828125,
      "learning_rate": 9.996155004813376e-06,
      "loss": 0.0245,
      "num_input_tokens_seen": 97165664,
      "step": 45020
    },
    {
      "epoch": 7.3450244698205545,
      "grad_norm": 0.047088466584682465,
      "learning_rate": 9.99046204844287e-06,
      "loss": 0.1246,
      "num_input_tokens_seen": 97174976,
      "step": 45025
    },
    {
      "epoch": 7.345840130505709,
      "grad_norm": 0.02915360778570175,
      "learning_rate": 9.984770308821664e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 97186720,
      "step": 45030
    },
    {
      "epoch": 7.346655791190865,
      "grad_norm": 0.21505773067474365,
      "learning_rate": 9.979079786411167e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 97196608,
      "step": 45035
    },
    {
      "epoch": 7.34747145187602,
      "grad_norm": 9.541070938110352,
      "learning_rate": 9.973390481672676e-06,
      "loss": 0.0138,
      "num_input_tokens_seen": 97208352,
      "step": 45040
    },
    {
      "epoch": 7.348287112561175,
      "grad_norm": 0.13850171864032745,
      "learning_rate": 9.967702395067388e-06,
      "loss": 0.0101,
      "num_input_tokens_seen": 97218048,
      "step": 45045
    },
    {
      "epoch": 7.349102773246329,
      "grad_norm": 0.03252837806940079,
      "learning_rate": 9.962015527056429e-06,
      "loss": 0.1767,
      "num_input_tokens_seen": 97229920,
      "step": 45050
    },
    {
      "epoch": 7.349918433931484,
      "grad_norm": 0.04658183827996254,
      "learning_rate": 9.95632987810079e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 97242816,
      "step": 45055
    },
    {
      "epoch": 7.350734094616639,
      "grad_norm": 0.05596194788813591,
      "learning_rate": 9.950645448661381e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 97254656,
      "step": 45060
    },
    {
      "epoch": 7.351549755301795,
      "grad_norm": 0.015002463944256306,
      "learning_rate": 9.944962239199013e-06,
      "loss": 0.1525,
      "num_input_tokens_seen": 97266432,
      "step": 45065
    },
    {
      "epoch": 7.35236541598695,
      "grad_norm": 0.7433382272720337,
      "learning_rate": 9.939280250174396e-06,
      "loss": 0.0126,
      "num_input_tokens_seen": 97278080,
      "step": 45070
    },
    {
      "epoch": 7.353181076672104,
      "grad_norm": 4.4932451248168945,
      "learning_rate": 9.933599482048136e-06,
      "loss": 0.0914,
      "num_input_tokens_seen": 97288672,
      "step": 45075
    },
    {
      "epoch": 7.353996737357259,
      "grad_norm": 0.0814744308590889,
      "learning_rate": 9.927919935280752e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 97300736,
      "step": 45080
    },
    {
      "epoch": 7.354812398042414,
      "grad_norm": 0.06717698276042938,
      "learning_rate": 9.922241610332641e-06,
      "loss": 0.0441,
      "num_input_tokens_seen": 97311808,
      "step": 45085
    },
    {
      "epoch": 7.35562805872757,
      "grad_norm": 0.036793094128370285,
      "learning_rate": 9.91656450766414e-06,
      "loss": 0.1405,
      "num_input_tokens_seen": 97321120,
      "step": 45090
    },
    {
      "epoch": 7.356443719412725,
      "grad_norm": 0.11527273803949356,
      "learning_rate": 9.91088862773545e-06,
      "loss": 0.1126,
      "num_input_tokens_seen": 97331712,
      "step": 45095
    },
    {
      "epoch": 7.357259380097879,
      "grad_norm": 0.12357182055711746,
      "learning_rate": 9.90521397100669e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 97342720,
      "step": 45100
    },
    {
      "epoch": 7.358075040783034,
      "grad_norm": 0.10513553768396378,
      "learning_rate": 9.899540537937879e-06,
      "loss": 0.0063,
      "num_input_tokens_seen": 97353920,
      "step": 45105
    },
    {
      "epoch": 7.358890701468189,
      "grad_norm": 0.35938525199890137,
      "learning_rate": 9.893868328988928e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 97364320,
      "step": 45110
    },
    {
      "epoch": 7.359706362153344,
      "grad_norm": 0.026927918195724487,
      "learning_rate": 9.888197344619657e-06,
      "loss": 0.007,
      "num_input_tokens_seen": 97374656,
      "step": 45115
    },
    {
      "epoch": 7.3605220228384995,
      "grad_norm": 0.25100481510162354,
      "learning_rate": 9.882527585289788e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 97386272,
      "step": 45120
    },
    {
      "epoch": 7.361337683523654,
      "grad_norm": 4.940239906311035,
      "learning_rate": 9.876859051458937e-06,
      "loss": 0.0699,
      "num_input_tokens_seen": 97396992,
      "step": 45125
    },
    {
      "epoch": 7.362153344208809,
      "grad_norm": 0.03793177381157875,
      "learning_rate": 9.871191743586624e-06,
      "loss": 0.1292,
      "num_input_tokens_seen": 97407296,
      "step": 45130
    },
    {
      "epoch": 7.362969004893964,
      "grad_norm": 0.03328360989689827,
      "learning_rate": 9.865525662132274e-06,
      "loss": 0.2463,
      "num_input_tokens_seen": 97417824,
      "step": 45135
    },
    {
      "epoch": 7.363784665579119,
      "grad_norm": 0.011244897730648518,
      "learning_rate": 9.859860807555204e-06,
      "loss": 0.0019,
      "num_input_tokens_seen": 97428096,
      "step": 45140
    },
    {
      "epoch": 7.364600326264274,
      "grad_norm": 3.699322462081909,
      "learning_rate": 9.854197180314639e-06,
      "loss": 0.0702,
      "num_input_tokens_seen": 97439680,
      "step": 45145
    },
    {
      "epoch": 7.365415986949429,
      "grad_norm": 0.06068158149719238,
      "learning_rate": 9.848534780869698e-06,
      "loss": 0.1016,
      "num_input_tokens_seen": 97450880,
      "step": 45150
    },
    {
      "epoch": 7.366231647634584,
      "grad_norm": 10.884716033935547,
      "learning_rate": 9.842873609679404e-06,
      "loss": 0.4667,
      "num_input_tokens_seen": 97461568,
      "step": 45155
    },
    {
      "epoch": 7.367047308319739,
      "grad_norm": 0.027186408638954163,
      "learning_rate": 9.837213667202682e-06,
      "loss": 0.2195,
      "num_input_tokens_seen": 97472416,
      "step": 45160
    },
    {
      "epoch": 7.367862969004894,
      "grad_norm": 0.09181984513998032,
      "learning_rate": 9.83155495389836e-06,
      "loss": 0.1054,
      "num_input_tokens_seen": 97483744,
      "step": 45165
    },
    {
      "epoch": 7.368678629690049,
      "grad_norm": 0.04382199048995972,
      "learning_rate": 9.82589747022515e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 97494752,
      "step": 45170
    },
    {
      "epoch": 7.369494290375204,
      "grad_norm": 0.01950172521173954,
      "learning_rate": 9.82024121664169e-06,
      "loss": 0.002,
      "num_input_tokens_seen": 97505056,
      "step": 45175
    },
    {
      "epoch": 7.370309951060359,
      "grad_norm": 0.4417515993118286,
      "learning_rate": 9.814586193606496e-06,
      "loss": 0.146,
      "num_input_tokens_seen": 97516352,
      "step": 45180
    },
    {
      "epoch": 7.371125611745514,
      "grad_norm": 0.030991017818450928,
      "learning_rate": 9.808932401577994e-06,
      "loss": 0.1999,
      "num_input_tokens_seen": 97526784,
      "step": 45185
    },
    {
      "epoch": 7.371941272430669,
      "grad_norm": 0.46684086322784424,
      "learning_rate": 9.803279841014501e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 97538368,
      "step": 45190
    },
    {
      "epoch": 7.372756933115824,
      "grad_norm": 3.5872461795806885,
      "learning_rate": 9.797628512374262e-06,
      "loss": 0.1035,
      "num_input_tokens_seen": 97549440,
      "step": 45195
    },
    {
      "epoch": 7.373572593800978,
      "grad_norm": 0.10208668559789658,
      "learning_rate": 9.791978416115393e-06,
      "loss": 0.0788,
      "num_input_tokens_seen": 97559264,
      "step": 45200
    },
    {
      "epoch": 7.374388254486134,
      "grad_norm": 0.15362916886806488,
      "learning_rate": 9.786329552695916e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 97569312,
      "step": 45205
    },
    {
      "epoch": 7.375203915171289,
      "grad_norm": 0.10528465360403061,
      "learning_rate": 9.780681922573759e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 97579712,
      "step": 45210
    },
    {
      "epoch": 7.376019575856444,
      "grad_norm": 0.09068232774734497,
      "learning_rate": 9.775035526206741e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 97590912,
      "step": 45215
    },
    {
      "epoch": 7.376835236541599,
      "grad_norm": 0.060372915118932724,
      "learning_rate": 9.7693903640526e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 97602176,
      "step": 45220
    },
    {
      "epoch": 7.377650897226753,
      "grad_norm": 0.1669987291097641,
      "learning_rate": 9.763746436568957e-06,
      "loss": 0.1188,
      "num_input_tokens_seen": 97612128,
      "step": 45225
    },
    {
      "epoch": 7.378466557911908,
      "grad_norm": 0.1839768886566162,
      "learning_rate": 9.758103744213334e-06,
      "loss": 0.1168,
      "num_input_tokens_seen": 97622528,
      "step": 45230
    },
    {
      "epoch": 7.379282218597064,
      "grad_norm": 0.06200070306658745,
      "learning_rate": 9.752462287443163e-06,
      "loss": 0.0407,
      "num_input_tokens_seen": 97634304,
      "step": 45235
    },
    {
      "epoch": 7.380097879282219,
      "grad_norm": 0.2495262175798416,
      "learning_rate": 9.746822066715757e-06,
      "loss": 0.1293,
      "num_input_tokens_seen": 97644864,
      "step": 45240
    },
    {
      "epoch": 7.3809135399673735,
      "grad_norm": 0.12641514837741852,
      "learning_rate": 9.741183082488354e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 97656768,
      "step": 45245
    },
    {
      "epoch": 7.381729200652528,
      "grad_norm": 8.78662109375,
      "learning_rate": 9.73554533521807e-06,
      "loss": 0.25,
      "num_input_tokens_seen": 97668416,
      "step": 45250
    },
    {
      "epoch": 7.382544861337683,
      "grad_norm": 1.1876436471939087,
      "learning_rate": 9.729908825361933e-06,
      "loss": 0.0059,
      "num_input_tokens_seen": 97679392,
      "step": 45255
    },
    {
      "epoch": 7.383360522022839,
      "grad_norm": 0.351886510848999,
      "learning_rate": 9.724273553376864e-06,
      "loss": 0.2233,
      "num_input_tokens_seen": 97689312,
      "step": 45260
    },
    {
      "epoch": 7.384176182707994,
      "grad_norm": 0.07911387830972672,
      "learning_rate": 9.718639519719695e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 97700448,
      "step": 45265
    },
    {
      "epoch": 7.3849918433931485,
      "grad_norm": 13.688436508178711,
      "learning_rate": 9.713006724847137e-06,
      "loss": 0.0675,
      "num_input_tokens_seen": 97710528,
      "step": 45270
    },
    {
      "epoch": 7.385807504078303,
      "grad_norm": 0.13461512327194214,
      "learning_rate": 9.70737516921582e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 97718720,
      "step": 45275
    },
    {
      "epoch": 7.386623164763458,
      "grad_norm": 0.03747570514678955,
      "learning_rate": 9.70174485328227e-06,
      "loss": 0.1638,
      "num_input_tokens_seen": 97727744,
      "step": 45280
    },
    {
      "epoch": 7.387438825448613,
      "grad_norm": 0.05718259885907173,
      "learning_rate": 9.6961157775029e-06,
      "loss": 0.0022,
      "num_input_tokens_seen": 97738112,
      "step": 45285
    },
    {
      "epoch": 7.388254486133769,
      "grad_norm": 0.07959671318531036,
      "learning_rate": 9.69048794233404e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 97748320,
      "step": 45290
    },
    {
      "epoch": 7.3890701468189235,
      "grad_norm": 0.05740109831094742,
      "learning_rate": 9.684861348231897e-06,
      "loss": 0.0962,
      "num_input_tokens_seen": 97760480,
      "step": 45295
    },
    {
      "epoch": 7.389885807504078,
      "grad_norm": 0.1578749567270279,
      "learning_rate": 9.67923599565261e-06,
      "loss": 0.1199,
      "num_input_tokens_seen": 97770912,
      "step": 45300
    },
    {
      "epoch": 7.390701468189233,
      "grad_norm": 6.2333855628967285,
      "learning_rate": 9.673611885052189e-06,
      "loss": 0.4056,
      "num_input_tokens_seen": 97781984,
      "step": 45305
    },
    {
      "epoch": 7.391517128874388,
      "grad_norm": 3.5526957511901855,
      "learning_rate": 9.667989016886555e-06,
      "loss": 0.1209,
      "num_input_tokens_seen": 97793152,
      "step": 45310
    },
    {
      "epoch": 7.392332789559543,
      "grad_norm": 0.04285155236721039,
      "learning_rate": 9.662367391611526e-06,
      "loss": 0.1422,
      "num_input_tokens_seen": 97804896,
      "step": 45315
    },
    {
      "epoch": 7.3931484502446985,
      "grad_norm": 0.09960167855024338,
      "learning_rate": 9.656747009682817e-06,
      "loss": 0.0832,
      "num_input_tokens_seen": 97815008,
      "step": 45320
    },
    {
      "epoch": 7.393964110929853,
      "grad_norm": 4.54679012298584,
      "learning_rate": 9.651127871556049e-06,
      "loss": 0.0096,
      "num_input_tokens_seen": 97826400,
      "step": 45325
    },
    {
      "epoch": 7.394779771615008,
      "grad_norm": 0.09120772033929825,
      "learning_rate": 9.645509977686731e-06,
      "loss": 0.1545,
      "num_input_tokens_seen": 97837024,
      "step": 45330
    },
    {
      "epoch": 7.395595432300163,
      "grad_norm": 0.05697239562869072,
      "learning_rate": 9.639893328530283e-06,
      "loss": 0.0107,
      "num_input_tokens_seen": 97847616,
      "step": 45335
    },
    {
      "epoch": 7.396411092985318,
      "grad_norm": 0.03465592488646507,
      "learning_rate": 9.63427792454202e-06,
      "loss": 0.0696,
      "num_input_tokens_seen": 97858464,
      "step": 45340
    },
    {
      "epoch": 7.397226753670473,
      "grad_norm": 13.479477882385254,
      "learning_rate": 9.628663766177154e-06,
      "loss": 0.0887,
      "num_input_tokens_seen": 97868224,
      "step": 45345
    },
    {
      "epoch": 7.398042414355628,
      "grad_norm": 0.41337093710899353,
      "learning_rate": 9.623050853890795e-06,
      "loss": 0.1268,
      "num_input_tokens_seen": 97879520,
      "step": 45350
    },
    {
      "epoch": 7.398858075040783,
      "grad_norm": 0.1932287961244583,
      "learning_rate": 9.617439188137956e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 97889568,
      "step": 45355
    },
    {
      "epoch": 7.399673735725938,
      "grad_norm": 0.1612764447927475,
      "learning_rate": 9.611828769373538e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 97899840,
      "step": 45360
    },
    {
      "epoch": 7.400489396411093,
      "grad_norm": 3.143439292907715,
      "learning_rate": 9.606219598052366e-06,
      "loss": 0.1091,
      "num_input_tokens_seen": 97910112,
      "step": 45365
    },
    {
      "epoch": 7.401305057096248,
      "grad_norm": 0.42711320519447327,
      "learning_rate": 9.600611674629143e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 97921664,
      "step": 45370
    },
    {
      "epoch": 7.402120717781403,
      "grad_norm": 1.0538039207458496,
      "learning_rate": 9.595004999558471e-06,
      "loss": 0.0961,
      "num_input_tokens_seen": 97934112,
      "step": 45375
    },
    {
      "epoch": 7.402936378466558,
      "grad_norm": 0.22625838220119476,
      "learning_rate": 9.58939957329486e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 97945760,
      "step": 45380
    },
    {
      "epoch": 7.403752039151713,
      "grad_norm": 0.06716348975896835,
      "learning_rate": 9.58379539629271e-06,
      "loss": 0.3946,
      "num_input_tokens_seen": 97956320,
      "step": 45385
    },
    {
      "epoch": 7.404567699836868,
      "grad_norm": 0.2427644580602646,
      "learning_rate": 9.578192469006328e-06,
      "loss": 0.1536,
      "num_input_tokens_seen": 97966368,
      "step": 45390
    },
    {
      "epoch": 7.4053833605220225,
      "grad_norm": 0.08670927584171295,
      "learning_rate": 9.572590791889916e-06,
      "loss": 0.1181,
      "num_input_tokens_seen": 97976736,
      "step": 45395
    },
    {
      "epoch": 7.406199021207178,
      "grad_norm": 0.09502453356981277,
      "learning_rate": 9.56699036539756e-06,
      "loss": 0.0772,
      "num_input_tokens_seen": 97987616,
      "step": 45400
    },
    {
      "epoch": 7.407014681892333,
      "grad_norm": 0.05197295546531677,
      "learning_rate": 9.561391189983281e-06,
      "loss": 0.0732,
      "num_input_tokens_seen": 97999424,
      "step": 45405
    },
    {
      "epoch": 7.407830342577488,
      "grad_norm": 0.1300973743200302,
      "learning_rate": 9.555793266100969e-06,
      "loss": 0.1041,
      "num_input_tokens_seen": 98010688,
      "step": 45410
    },
    {
      "epoch": 7.408646003262643,
      "grad_norm": 32.43983840942383,
      "learning_rate": 9.550196594204413e-06,
      "loss": 0.0672,
      "num_input_tokens_seen": 98022304,
      "step": 45415
    },
    {
      "epoch": 7.4094616639477975,
      "grad_norm": 0.5860154628753662,
      "learning_rate": 9.544601174747316e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 98032448,
      "step": 45420
    },
    {
      "epoch": 7.410277324632952,
      "grad_norm": 0.3548831045627594,
      "learning_rate": 9.539007008183267e-06,
      "loss": 0.1053,
      "num_input_tokens_seen": 98043936,
      "step": 45425
    },
    {
      "epoch": 7.411092985318108,
      "grad_norm": 0.09791775792837143,
      "learning_rate": 9.533414094965759e-06,
      "loss": 0.1139,
      "num_input_tokens_seen": 98054880,
      "step": 45430
    },
    {
      "epoch": 7.411908646003263,
      "grad_norm": 0.21384207904338837,
      "learning_rate": 9.527822435548181e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 98065440,
      "step": 45435
    },
    {
      "epoch": 7.412724306688418,
      "grad_norm": 0.06359481066465378,
      "learning_rate": 9.522232030383822e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 98076256,
      "step": 45440
    },
    {
      "epoch": 7.4135399673735725,
      "grad_norm": 17.562448501586914,
      "learning_rate": 9.516642879925865e-06,
      "loss": 0.0194,
      "num_input_tokens_seen": 98087648,
      "step": 45445
    },
    {
      "epoch": 7.414355628058727,
      "grad_norm": 0.07706096768379211,
      "learning_rate": 9.511054984627402e-06,
      "loss": 0.106,
      "num_input_tokens_seen": 98097952,
      "step": 45450
    },
    {
      "epoch": 7.415171288743883,
      "grad_norm": 0.21106182038784027,
      "learning_rate": 9.50546834494141e-06,
      "loss": 0.0957,
      "num_input_tokens_seen": 98109440,
      "step": 45455
    },
    {
      "epoch": 7.415986949429038,
      "grad_norm": 2.599562168121338,
      "learning_rate": 9.499882961320771e-06,
      "loss": 0.0087,
      "num_input_tokens_seen": 98120160,
      "step": 45460
    },
    {
      "epoch": 7.416802610114193,
      "grad_norm": 0.1349298506975174,
      "learning_rate": 9.494298834218268e-06,
      "loss": 0.0819,
      "num_input_tokens_seen": 98131488,
      "step": 45465
    },
    {
      "epoch": 7.417618270799347,
      "grad_norm": 0.030637389048933983,
      "learning_rate": 9.488715964086575e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 98142528,
      "step": 45470
    },
    {
      "epoch": 7.418433931484502,
      "grad_norm": 0.07534648478031158,
      "learning_rate": 9.48313435137827e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 98153472,
      "step": 45475
    },
    {
      "epoch": 7.419249592169657,
      "grad_norm": 0.07768328487873077,
      "learning_rate": 9.47755399654583e-06,
      "loss": 0.1942,
      "num_input_tokens_seen": 98162784,
      "step": 45480
    },
    {
      "epoch": 7.420065252854813,
      "grad_norm": 0.10556311905384064,
      "learning_rate": 9.47197490004162e-06,
      "loss": 0.1105,
      "num_input_tokens_seen": 98173376,
      "step": 45485
    },
    {
      "epoch": 7.420880913539968,
      "grad_norm": 0.1772138774394989,
      "learning_rate": 9.466397062317911e-06,
      "loss": 0.1223,
      "num_input_tokens_seen": 98183872,
      "step": 45490
    },
    {
      "epoch": 7.421696574225122,
      "grad_norm": 0.07352486997842789,
      "learning_rate": 9.460820483826874e-06,
      "loss": 0.1066,
      "num_input_tokens_seen": 98195200,
      "step": 45495
    },
    {
      "epoch": 7.422512234910277,
      "grad_norm": 0.30517813563346863,
      "learning_rate": 9.455245165020565e-06,
      "loss": 0.1547,
      "num_input_tokens_seen": 98205344,
      "step": 45500
    },
    {
      "epoch": 7.423327895595432,
      "grad_norm": 0.0715760663151741,
      "learning_rate": 9.449671106350966e-06,
      "loss": 0.07,
      "num_input_tokens_seen": 98216896,
      "step": 45505
    },
    {
      "epoch": 7.424143556280587,
      "grad_norm": 0.9469295144081116,
      "learning_rate": 9.444098308269931e-06,
      "loss": 0.1897,
      "num_input_tokens_seen": 98228896,
      "step": 45510
    },
    {
      "epoch": 7.424959216965743,
      "grad_norm": 3.0780670642852783,
      "learning_rate": 9.438526771229212e-06,
      "loss": 0.2251,
      "num_input_tokens_seen": 98239488,
      "step": 45515
    },
    {
      "epoch": 7.425774877650897,
      "grad_norm": 0.08582847565412521,
      "learning_rate": 9.432956495680473e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 98251232,
      "step": 45520
    },
    {
      "epoch": 7.426590538336052,
      "grad_norm": 1.2630985975265503,
      "learning_rate": 9.42738748207526e-06,
      "loss": 0.01,
      "num_input_tokens_seen": 98262432,
      "step": 45525
    },
    {
      "epoch": 7.427406199021207,
      "grad_norm": 0.013918722979724407,
      "learning_rate": 9.421819730865039e-06,
      "loss": 0.3242,
      "num_input_tokens_seen": 98272320,
      "step": 45530
    },
    {
      "epoch": 7.428221859706362,
      "grad_norm": 0.07709713280200958,
      "learning_rate": 9.416253242501155e-06,
      "loss": 0.0107,
      "num_input_tokens_seen": 98283904,
      "step": 45535
    },
    {
      "epoch": 7.4290375203915175,
      "grad_norm": 3.635129690170288,
      "learning_rate": 9.410688017434852e-06,
      "loss": 0.1949,
      "num_input_tokens_seen": 98295840,
      "step": 45540
    },
    {
      "epoch": 7.429853181076672,
      "grad_norm": 0.04166160151362419,
      "learning_rate": 9.40512405611728e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 98307072,
      "step": 45545
    },
    {
      "epoch": 7.430668841761827,
      "grad_norm": 0.42729049921035767,
      "learning_rate": 9.399561358999479e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 98318144,
      "step": 45550
    },
    {
      "epoch": 7.431484502446982,
      "grad_norm": 0.3752131164073944,
      "learning_rate": 9.393999926532387e-06,
      "loss": 0.1307,
      "num_input_tokens_seen": 98329728,
      "step": 45555
    },
    {
      "epoch": 7.432300163132137,
      "grad_norm": 0.18965347111225128,
      "learning_rate": 9.38843975916685e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 98340256,
      "step": 45560
    },
    {
      "epoch": 7.433115823817292,
      "grad_norm": 0.024965984746813774,
      "learning_rate": 9.382880857353596e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 98350976,
      "step": 45565
    },
    {
      "epoch": 7.433931484502447,
      "grad_norm": 0.24310645461082458,
      "learning_rate": 9.37732322154326e-06,
      "loss": 0.1285,
      "num_input_tokens_seen": 98361536,
      "step": 45570
    },
    {
      "epoch": 7.434747145187602,
      "grad_norm": 0.08490164577960968,
      "learning_rate": 9.371766852186373e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 98372064,
      "step": 45575
    },
    {
      "epoch": 7.435562805872757,
      "grad_norm": 17.166574478149414,
      "learning_rate": 9.366211749733361e-06,
      "loss": 0.1932,
      "num_input_tokens_seen": 98382688,
      "step": 45580
    },
    {
      "epoch": 7.436378466557912,
      "grad_norm": 0.17929571866989136,
      "learning_rate": 9.36065791463455e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 98393312,
      "step": 45585
    },
    {
      "epoch": 7.437194127243067,
      "grad_norm": 0.12863053381443024,
      "learning_rate": 9.355105347340163e-06,
      "loss": 0.1902,
      "num_input_tokens_seen": 98403136,
      "step": 45590
    },
    {
      "epoch": 7.438009787928221,
      "grad_norm": 0.02911662869155407,
      "learning_rate": 9.349554048300316e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 98414752,
      "step": 45595
    },
    {
      "epoch": 7.438825448613377,
      "grad_norm": 0.019138168543577194,
      "learning_rate": 9.344004017965027e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 98426496,
      "step": 45600
    },
    {
      "epoch": 7.439641109298532,
      "grad_norm": 0.1090579703450203,
      "learning_rate": 9.338455256784212e-06,
      "loss": 0.1387,
      "num_input_tokens_seen": 98437696,
      "step": 45605
    },
    {
      "epoch": 7.440456769983687,
      "grad_norm": 0.5317416191101074,
      "learning_rate": 9.33290776520768e-06,
      "loss": 0.1597,
      "num_input_tokens_seen": 98448640,
      "step": 45610
    },
    {
      "epoch": 7.441272430668842,
      "grad_norm": 0.07390889525413513,
      "learning_rate": 9.32736154368513e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 98459360,
      "step": 45615
    },
    {
      "epoch": 7.442088091353996,
      "grad_norm": 0.20405705273151398,
      "learning_rate": 9.32181659266618e-06,
      "loss": 0.1511,
      "num_input_tokens_seen": 98470560,
      "step": 45620
    },
    {
      "epoch": 7.442903752039152,
      "grad_norm": 0.07471613585948944,
      "learning_rate": 9.316272912600332e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 98482432,
      "step": 45625
    },
    {
      "epoch": 7.443719412724307,
      "grad_norm": 17.78740119934082,
      "learning_rate": 9.310730503936976e-06,
      "loss": 0.0247,
      "num_input_tokens_seen": 98493056,
      "step": 45630
    },
    {
      "epoch": 7.444535073409462,
      "grad_norm": 0.04233328253030777,
      "learning_rate": 9.305189367125416e-06,
      "loss": 0.1337,
      "num_input_tokens_seen": 98503904,
      "step": 45635
    },
    {
      "epoch": 7.445350734094617,
      "grad_norm": 0.03839850798249245,
      "learning_rate": 9.299649502614838e-06,
      "loss": 0.0717,
      "num_input_tokens_seen": 98514848,
      "step": 45640
    },
    {
      "epoch": 7.446166394779771,
      "grad_norm": 12.67846393585205,
      "learning_rate": 9.294110910854337e-06,
      "loss": 0.0374,
      "num_input_tokens_seen": 98525696,
      "step": 45645
    },
    {
      "epoch": 7.446982055464926,
      "grad_norm": 0.03788448125123978,
      "learning_rate": 9.288573592292893e-06,
      "loss": 0.0381,
      "num_input_tokens_seen": 98536672,
      "step": 45650
    },
    {
      "epoch": 7.447797716150082,
      "grad_norm": 0.08515933156013489,
      "learning_rate": 9.283037547379394e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 98546912,
      "step": 45655
    },
    {
      "epoch": 7.448613376835237,
      "grad_norm": 9.248990058898926,
      "learning_rate": 9.27750277656262e-06,
      "loss": 0.4888,
      "num_input_tokens_seen": 98558336,
      "step": 45660
    },
    {
      "epoch": 7.4494290375203915,
      "grad_norm": 0.059895455837249756,
      "learning_rate": 9.271969280291243e-06,
      "loss": 0.0425,
      "num_input_tokens_seen": 98568096,
      "step": 45665
    },
    {
      "epoch": 7.450244698205546,
      "grad_norm": 0.034142039716243744,
      "learning_rate": 9.266437059013834e-06,
      "loss": 0.0977,
      "num_input_tokens_seen": 98579072,
      "step": 45670
    },
    {
      "epoch": 7.451060358890701,
      "grad_norm": 0.27393639087677,
      "learning_rate": 9.260906113178875e-06,
      "loss": 0.1577,
      "num_input_tokens_seen": 98589600,
      "step": 45675
    },
    {
      "epoch": 7.451876019575856,
      "grad_norm": 3.8771159648895264,
      "learning_rate": 9.255376443234725e-06,
      "loss": 0.2061,
      "num_input_tokens_seen": 98600864,
      "step": 45680
    },
    {
      "epoch": 7.452691680261012,
      "grad_norm": 0.08974248170852661,
      "learning_rate": 9.24984804962965e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 98611968,
      "step": 45685
    },
    {
      "epoch": 7.4535073409461665,
      "grad_norm": 0.11940401792526245,
      "learning_rate": 9.244320932811806e-06,
      "loss": 0.1428,
      "num_input_tokens_seen": 98622912,
      "step": 45690
    },
    {
      "epoch": 7.454323001631321,
      "grad_norm": 3.759678363800049,
      "learning_rate": 9.238795093229252e-06,
      "loss": 0.4739,
      "num_input_tokens_seen": 98634592,
      "step": 45695
    },
    {
      "epoch": 7.455138662316476,
      "grad_norm": 0.05598155036568642,
      "learning_rate": 9.233270531329937e-06,
      "loss": 0.121,
      "num_input_tokens_seen": 98644096,
      "step": 45700
    },
    {
      "epoch": 7.455954323001631,
      "grad_norm": 0.1402728259563446,
      "learning_rate": 9.227747247561713e-06,
      "loss": 0.1641,
      "num_input_tokens_seen": 98654400,
      "step": 45705
    },
    {
      "epoch": 7.456769983686787,
      "grad_norm": 0.08239126205444336,
      "learning_rate": 9.222225242372326e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 98665568,
      "step": 45710
    },
    {
      "epoch": 7.4575856443719415,
      "grad_norm": 0.10888636112213135,
      "learning_rate": 9.216704516209417e-06,
      "loss": 0.1002,
      "num_input_tokens_seen": 98674624,
      "step": 45715
    },
    {
      "epoch": 7.458401305057096,
      "grad_norm": 13.814203262329102,
      "learning_rate": 9.211185069520514e-06,
      "loss": 0.1048,
      "num_input_tokens_seen": 98685536,
      "step": 45720
    },
    {
      "epoch": 7.459216965742251,
      "grad_norm": 0.08124645799398422,
      "learning_rate": 9.205666902753071e-06,
      "loss": 0.1477,
      "num_input_tokens_seen": 98696352,
      "step": 45725
    },
    {
      "epoch": 7.460032626427406,
      "grad_norm": 0.11474943906068802,
      "learning_rate": 9.200150016354406e-06,
      "loss": 0.0704,
      "num_input_tokens_seen": 98707808,
      "step": 45730
    },
    {
      "epoch": 7.460848287112561,
      "grad_norm": 13.604469299316406,
      "learning_rate": 9.19463441077175e-06,
      "loss": 0.0962,
      "num_input_tokens_seen": 98718112,
      "step": 45735
    },
    {
      "epoch": 7.4616639477977165,
      "grad_norm": 3.5292141437530518,
      "learning_rate": 9.189120086452224e-06,
      "loss": 0.1251,
      "num_input_tokens_seen": 98729312,
      "step": 45740
    },
    {
      "epoch": 7.462479608482871,
      "grad_norm": 0.07247146964073181,
      "learning_rate": 9.183607043842846e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 98739168,
      "step": 45745
    },
    {
      "epoch": 7.463295269168026,
      "grad_norm": 17.904502868652344,
      "learning_rate": 9.178095283390533e-06,
      "loss": 0.4384,
      "num_input_tokens_seen": 98749376,
      "step": 45750
    },
    {
      "epoch": 7.464110929853181,
      "grad_norm": 0.0919409990310669,
      "learning_rate": 9.172584805542098e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 98760928,
      "step": 45755
    },
    {
      "epoch": 7.464926590538336,
      "grad_norm": 0.1532522439956665,
      "learning_rate": 9.167075610744244e-06,
      "loss": 0.0898,
      "num_input_tokens_seen": 98770560,
      "step": 45760
    },
    {
      "epoch": 7.465742251223491,
      "grad_norm": 0.1458083689212799,
      "learning_rate": 9.161567699443577e-06,
      "loss": 0.1232,
      "num_input_tokens_seen": 98780512,
      "step": 45765
    },
    {
      "epoch": 7.466557911908646,
      "grad_norm": 0.6249791383743286,
      "learning_rate": 9.156061072086597e-06,
      "loss": 0.0134,
      "num_input_tokens_seen": 98792192,
      "step": 45770
    },
    {
      "epoch": 7.467373572593801,
      "grad_norm": 6.886234760284424,
      "learning_rate": 9.150555729119697e-06,
      "loss": 0.143,
      "num_input_tokens_seen": 98802752,
      "step": 45775
    },
    {
      "epoch": 7.468189233278956,
      "grad_norm": 0.17001597583293915,
      "learning_rate": 9.14505167098917e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 98813600,
      "step": 45780
    },
    {
      "epoch": 7.469004893964111,
      "grad_norm": 0.011153833009302616,
      "learning_rate": 9.139548898141198e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 98825280,
      "step": 45785
    },
    {
      "epoch": 7.4698205546492655,
      "grad_norm": 2.9685122966766357,
      "learning_rate": 9.13404741102187e-06,
      "loss": 0.1782,
      "num_input_tokens_seen": 98836480,
      "step": 45790
    },
    {
      "epoch": 7.470636215334421,
      "grad_norm": 15.426504135131836,
      "learning_rate": 9.128547210077162e-06,
      "loss": 0.0538,
      "num_input_tokens_seen": 98847424,
      "step": 45795
    },
    {
      "epoch": 7.471451876019576,
      "grad_norm": 26.57598876953125,
      "learning_rate": 9.123048295752948e-06,
      "loss": 0.0576,
      "num_input_tokens_seen": 98858016,
      "step": 45800
    },
    {
      "epoch": 7.472267536704731,
      "grad_norm": 0.04468873515725136,
      "learning_rate": 9.117550668494998e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 98868864,
      "step": 45805
    },
    {
      "epoch": 7.473083197389886,
      "grad_norm": 4.287403583526611,
      "learning_rate": 9.112054328748975e-06,
      "loss": 0.1979,
      "num_input_tokens_seen": 98879936,
      "step": 45810
    },
    {
      "epoch": 7.4738988580750405,
      "grad_norm": 0.07600153237581253,
      "learning_rate": 9.106559276960439e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 98891328,
      "step": 45815
    },
    {
      "epoch": 7.474714518760196,
      "grad_norm": 0.15871362388134003,
      "learning_rate": 9.101065513574856e-06,
      "loss": 0.0059,
      "num_input_tokens_seen": 98902592,
      "step": 45820
    },
    {
      "epoch": 7.475530179445351,
      "grad_norm": 0.10289601236581802,
      "learning_rate": 9.095573039037574e-06,
      "loss": 0.0084,
      "num_input_tokens_seen": 98913440,
      "step": 45825
    },
    {
      "epoch": 7.476345840130506,
      "grad_norm": 0.1871442198753357,
      "learning_rate": 9.090081853793838e-06,
      "loss": 0.0861,
      "num_input_tokens_seen": 98923808,
      "step": 45830
    },
    {
      "epoch": 7.477161500815661,
      "grad_norm": 0.09045414626598358,
      "learning_rate": 9.084591958288797e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 98933856,
      "step": 45835
    },
    {
      "epoch": 7.4779771615008155,
      "grad_norm": 35.73626708984375,
      "learning_rate": 9.079103352967471e-06,
      "loss": 0.0645,
      "num_input_tokens_seen": 98944960,
      "step": 45840
    },
    {
      "epoch": 7.47879282218597,
      "grad_norm": 6.172349452972412,
      "learning_rate": 9.073616038274823e-06,
      "loss": 0.0824,
      "num_input_tokens_seen": 98956512,
      "step": 45845
    },
    {
      "epoch": 7.479608482871126,
      "grad_norm": 0.2083701193332672,
      "learning_rate": 9.068130014655665e-06,
      "loss": 0.0919,
      "num_input_tokens_seen": 98967520,
      "step": 45850
    },
    {
      "epoch": 7.480424143556281,
      "grad_norm": 0.11684663593769073,
      "learning_rate": 9.06264528255473e-06,
      "loss": 0.1304,
      "num_input_tokens_seen": 98977920,
      "step": 45855
    },
    {
      "epoch": 7.481239804241436,
      "grad_norm": 0.06528680771589279,
      "learning_rate": 9.057161842416628e-06,
      "loss": 0.0855,
      "num_input_tokens_seen": 98987392,
      "step": 45860
    },
    {
      "epoch": 7.4820554649265905,
      "grad_norm": 0.1682402789592743,
      "learning_rate": 9.051679694685885e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 98998304,
      "step": 45865
    },
    {
      "epoch": 7.482871125611745,
      "grad_norm": 0.20442993938922882,
      "learning_rate": 9.046198839806905e-06,
      "loss": 0.1825,
      "num_input_tokens_seen": 99008480,
      "step": 45870
    },
    {
      "epoch": 7.4836867862969,
      "grad_norm": 0.10275133699178696,
      "learning_rate": 9.040719278223997e-06,
      "loss": 0.1771,
      "num_input_tokens_seen": 99019584,
      "step": 45875
    },
    {
      "epoch": 7.484502446982056,
      "grad_norm": 4.467346668243408,
      "learning_rate": 9.03524101038136e-06,
      "loss": 0.2166,
      "num_input_tokens_seen": 99029920,
      "step": 45880
    },
    {
      "epoch": 7.485318107667211,
      "grad_norm": 2.960280656814575,
      "learning_rate": 9.02976403672309e-06,
      "loss": 0.3622,
      "num_input_tokens_seen": 99040288,
      "step": 45885
    },
    {
      "epoch": 7.486133768352365,
      "grad_norm": 0.10947456955909729,
      "learning_rate": 9.02428835769318e-06,
      "loss": 0.0076,
      "num_input_tokens_seen": 99053024,
      "step": 45890
    },
    {
      "epoch": 7.48694942903752,
      "grad_norm": 0.026944268494844437,
      "learning_rate": 9.018813973735515e-06,
      "loss": 0.112,
      "num_input_tokens_seen": 99062720,
      "step": 45895
    },
    {
      "epoch": 7.487765089722675,
      "grad_norm": 0.05024728551506996,
      "learning_rate": 9.013340885293878e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 99073728,
      "step": 45900
    },
    {
      "epoch": 7.488580750407831,
      "grad_norm": 0.13830679655075073,
      "learning_rate": 9.007869092811944e-06,
      "loss": 0.4181,
      "num_input_tokens_seen": 99084704,
      "step": 45905
    },
    {
      "epoch": 7.489396411092986,
      "grad_norm": 4.213515758514404,
      "learning_rate": 9.002398596733287e-06,
      "loss": 0.3499,
      "num_input_tokens_seen": 99095744,
      "step": 45910
    },
    {
      "epoch": 7.49021207177814,
      "grad_norm": 0.06781812012195587,
      "learning_rate": 8.996929397501366e-06,
      "loss": 0.1147,
      "num_input_tokens_seen": 99107168,
      "step": 45915
    },
    {
      "epoch": 7.491027732463295,
      "grad_norm": 0.04535306990146637,
      "learning_rate": 8.99146149555955e-06,
      "loss": 0.0146,
      "num_input_tokens_seen": 99118208,
      "step": 45920
    },
    {
      "epoch": 7.49184339314845,
      "grad_norm": 0.11362104117870331,
      "learning_rate": 8.98599489135109e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 99129152,
      "step": 45925
    },
    {
      "epoch": 7.492659053833605,
      "grad_norm": 0.13913558423519135,
      "learning_rate": 8.980529585319142e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 99140192,
      "step": 45930
    },
    {
      "epoch": 7.493474714518761,
      "grad_norm": 3.604703187942505,
      "learning_rate": 8.975065577906735e-06,
      "loss": 0.2062,
      "num_input_tokens_seen": 99150368,
      "step": 45935
    },
    {
      "epoch": 7.494290375203915,
      "grad_norm": 0.04341362416744232,
      "learning_rate": 8.969602869556834e-06,
      "loss": 0.1583,
      "num_input_tokens_seen": 99161728,
      "step": 45940
    },
    {
      "epoch": 7.49510603588907,
      "grad_norm": 4.822989463806152,
      "learning_rate": 8.964141460712258e-06,
      "loss": 0.1043,
      "num_input_tokens_seen": 99173024,
      "step": 45945
    },
    {
      "epoch": 7.495921696574225,
      "grad_norm": 1.2872027158737183,
      "learning_rate": 8.958681351815742e-06,
      "loss": 0.1475,
      "num_input_tokens_seen": 99183008,
      "step": 45950
    },
    {
      "epoch": 7.49673735725938,
      "grad_norm": 0.03739134222269058,
      "learning_rate": 8.95322254330991e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 99194976,
      "step": 45955
    },
    {
      "epoch": 7.497553017944535,
      "grad_norm": 5.683506011962891,
      "learning_rate": 8.947765035637278e-06,
      "loss": 0.0957,
      "num_input_tokens_seen": 99205760,
      "step": 45960
    },
    {
      "epoch": 7.49836867862969,
      "grad_norm": 0.14213387668132782,
      "learning_rate": 8.942308829240262e-06,
      "loss": 0.0785,
      "num_input_tokens_seen": 99215808,
      "step": 45965
    },
    {
      "epoch": 7.499184339314845,
      "grad_norm": 0.0866631269454956,
      "learning_rate": 8.936853924561167e-06,
      "loss": 0.1909,
      "num_input_tokens_seen": 99226848,
      "step": 45970
    },
    {
      "epoch": 7.5,
      "grad_norm": 0.13099905848503113,
      "learning_rate": 8.931400322042193e-06,
      "loss": 0.1522,
      "num_input_tokens_seen": 99237152,
      "step": 45975
    },
    {
      "epoch": 7.5,
      "eval_loss": 0.18871694803237915,
      "eval_runtime": 132.9402,
      "eval_samples_per_second": 20.498,
      "eval_steps_per_second": 5.13,
      "num_input_tokens_seen": 99237152,
      "step": 45975
    },
    {
      "epoch": 7.500815660685155,
      "grad_norm": 4.312325954437256,
      "learning_rate": 8.925948022125446e-06,
      "loss": 0.2061,
      "num_input_tokens_seen": 99247936,
      "step": 45980
    },
    {
      "epoch": 7.50163132137031,
      "grad_norm": 0.1773970127105713,
      "learning_rate": 8.9204970252529e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 99259456,
      "step": 45985
    },
    {
      "epoch": 7.502446982055465,
      "grad_norm": 0.06699678301811218,
      "learning_rate": 8.91504733186646e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 99269760,
      "step": 45990
    },
    {
      "epoch": 7.50326264274062,
      "grad_norm": 0.09259167313575745,
      "learning_rate": 8.909598942407898e-06,
      "loss": 0.0242,
      "num_input_tokens_seen": 99280256,
      "step": 45995
    },
    {
      "epoch": 7.504078303425775,
      "grad_norm": 0.12489908188581467,
      "learning_rate": 8.904151857318888e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 99289600,
      "step": 46000
    },
    {
      "epoch": 7.50489396411093,
      "grad_norm": 0.06093369424343109,
      "learning_rate": 8.898706077040997e-06,
      "loss": 0.2861,
      "num_input_tokens_seen": 99300128,
      "step": 46005
    },
    {
      "epoch": 7.505709624796085,
      "grad_norm": 3.503791093826294,
      "learning_rate": 8.893261602015687e-06,
      "loss": 0.1008,
      "num_input_tokens_seen": 99311680,
      "step": 46010
    },
    {
      "epoch": 7.506525285481239,
      "grad_norm": 0.23494867980480194,
      "learning_rate": 8.88781843268432e-06,
      "loss": 0.2047,
      "num_input_tokens_seen": 99322976,
      "step": 46015
    },
    {
      "epoch": 7.507340946166395,
      "grad_norm": 0.05065242946147919,
      "learning_rate": 8.88237656948814e-06,
      "loss": 0.1271,
      "num_input_tokens_seen": 99335584,
      "step": 46020
    },
    {
      "epoch": 7.50815660685155,
      "grad_norm": 0.0380345843732357,
      "learning_rate": 8.876936012868297e-06,
      "loss": 0.0798,
      "num_input_tokens_seen": 99346528,
      "step": 46025
    },
    {
      "epoch": 7.508972267536705,
      "grad_norm": 0.08900675177574158,
      "learning_rate": 8.87149676326583e-06,
      "loss": 0.0399,
      "num_input_tokens_seen": 99357280,
      "step": 46030
    },
    {
      "epoch": 7.50978792822186,
      "grad_norm": 0.11268915981054306,
      "learning_rate": 8.866058821121667e-06,
      "loss": 0.1257,
      "num_input_tokens_seen": 99368192,
      "step": 46035
    },
    {
      "epoch": 7.510603588907014,
      "grad_norm": 0.2183750420808792,
      "learning_rate": 8.860622186876632e-06,
      "loss": 0.0063,
      "num_input_tokens_seen": 99379104,
      "step": 46040
    },
    {
      "epoch": 7.511419249592169,
      "grad_norm": 29.613920211791992,
      "learning_rate": 8.855186860971462e-06,
      "loss": 0.2434,
      "num_input_tokens_seen": 99391200,
      "step": 46045
    },
    {
      "epoch": 7.512234910277325,
      "grad_norm": 0.12104889750480652,
      "learning_rate": 8.849752843846762e-06,
      "loss": 0.0953,
      "num_input_tokens_seen": 99402272,
      "step": 46050
    },
    {
      "epoch": 7.51305057096248,
      "grad_norm": 0.29253631830215454,
      "learning_rate": 8.844320135943042e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 99414016,
      "step": 46055
    },
    {
      "epoch": 7.513866231647635,
      "grad_norm": 0.14891695976257324,
      "learning_rate": 8.838888737700707e-06,
      "loss": 0.1026,
      "num_input_tokens_seen": 99423424,
      "step": 46060
    },
    {
      "epoch": 7.514681892332789,
      "grad_norm": 0.1279086023569107,
      "learning_rate": 8.833458649560051e-06,
      "loss": 0.1157,
      "num_input_tokens_seen": 99434592,
      "step": 46065
    },
    {
      "epoch": 7.515497553017944,
      "grad_norm": 2.201295852661133,
      "learning_rate": 8.828029871961263e-06,
      "loss": 0.0895,
      "num_input_tokens_seen": 99444096,
      "step": 46070
    },
    {
      "epoch": 7.5163132137031,
      "grad_norm": 0.08931440860033035,
      "learning_rate": 8.82260240534443e-06,
      "loss": 0.1461,
      "num_input_tokens_seen": 99455328,
      "step": 46075
    },
    {
      "epoch": 7.517128874388255,
      "grad_norm": 19.845977783203125,
      "learning_rate": 8.817176250149528e-06,
      "loss": 0.162,
      "num_input_tokens_seen": 99465152,
      "step": 46080
    },
    {
      "epoch": 7.5179445350734095,
      "grad_norm": 0.06507433950901031,
      "learning_rate": 8.811751406816432e-06,
      "loss": 0.0029,
      "num_input_tokens_seen": 99475072,
      "step": 46085
    },
    {
      "epoch": 7.518760195758564,
      "grad_norm": 0.5574439167976379,
      "learning_rate": 8.806327875784906e-06,
      "loss": 0.0802,
      "num_input_tokens_seen": 99485344,
      "step": 46090
    },
    {
      "epoch": 7.519575856443719,
      "grad_norm": 0.22275541722774506,
      "learning_rate": 8.800905657494607e-06,
      "loss": 0.104,
      "num_input_tokens_seen": 99496864,
      "step": 46095
    },
    {
      "epoch": 7.520391517128875,
      "grad_norm": 0.13205687701702118,
      "learning_rate": 8.795484752385088e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 99507520,
      "step": 46100
    },
    {
      "epoch": 7.52120717781403,
      "grad_norm": 0.09482310712337494,
      "learning_rate": 8.790065160895797e-06,
      "loss": 0.1108,
      "num_input_tokens_seen": 99517664,
      "step": 46105
    },
    {
      "epoch": 7.5220228384991845,
      "grad_norm": 0.09053410589694977,
      "learning_rate": 8.784646883466072e-06,
      "loss": 0.069,
      "num_input_tokens_seen": 99527392,
      "step": 46110
    },
    {
      "epoch": 7.522838499184339,
      "grad_norm": 0.06341391056776047,
      "learning_rate": 8.779229920535148e-06,
      "loss": 0.134,
      "num_input_tokens_seen": 99538784,
      "step": 46115
    },
    {
      "epoch": 7.523654159869494,
      "grad_norm": 1.591096043586731,
      "learning_rate": 8.77381427254215e-06,
      "loss": 0.1277,
      "num_input_tokens_seen": 99549024,
      "step": 46120
    },
    {
      "epoch": 7.524469820554649,
      "grad_norm": 0.19831347465515137,
      "learning_rate": 8.768399939926095e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 99560480,
      "step": 46125
    },
    {
      "epoch": 7.525285481239804,
      "grad_norm": 2.8108818531036377,
      "learning_rate": 8.762986923125894e-06,
      "loss": 0.1244,
      "num_input_tokens_seen": 99571968,
      "step": 46130
    },
    {
      "epoch": 7.5261011419249595,
      "grad_norm": 0.07803460955619812,
      "learning_rate": 8.757575222580364e-06,
      "loss": 0.1148,
      "num_input_tokens_seen": 99582112,
      "step": 46135
    },
    {
      "epoch": 7.526916802610114,
      "grad_norm": 0.1227453425526619,
      "learning_rate": 8.752164838728203e-06,
      "loss": 0.1457,
      "num_input_tokens_seen": 99591360,
      "step": 46140
    },
    {
      "epoch": 7.527732463295269,
      "grad_norm": 0.37101203203201294,
      "learning_rate": 8.746755772007998e-06,
      "loss": 0.0168,
      "num_input_tokens_seen": 99601696,
      "step": 46145
    },
    {
      "epoch": 7.528548123980424,
      "grad_norm": 2.539811134338379,
      "learning_rate": 8.74134802285824e-06,
      "loss": 0.1978,
      "num_input_tokens_seen": 99612928,
      "step": 46150
    },
    {
      "epoch": 7.529363784665579,
      "grad_norm": 0.15568888187408447,
      "learning_rate": 8.735941591717297e-06,
      "loss": 0.0187,
      "num_input_tokens_seen": 99622720,
      "step": 46155
    },
    {
      "epoch": 7.5301794453507345,
      "grad_norm": 0.15910722315311432,
      "learning_rate": 8.730536479023463e-06,
      "loss": 0.3713,
      "num_input_tokens_seen": 99633888,
      "step": 46160
    },
    {
      "epoch": 7.530995106035889,
      "grad_norm": 3.501009702682495,
      "learning_rate": 8.72513268521489e-06,
      "loss": 0.1287,
      "num_input_tokens_seen": 99644960,
      "step": 46165
    },
    {
      "epoch": 7.531810766721044,
      "grad_norm": 0.09123460203409195,
      "learning_rate": 8.719730210729638e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 99656096,
      "step": 46170
    },
    {
      "epoch": 7.532626427406199,
      "grad_norm": 3.6207470893859863,
      "learning_rate": 8.714329056005663e-06,
      "loss": 0.0933,
      "num_input_tokens_seen": 99666752,
      "step": 46175
    },
    {
      "epoch": 7.533442088091354,
      "grad_norm": 14.625901222229004,
      "learning_rate": 8.708929221480808e-06,
      "loss": 0.1012,
      "num_input_tokens_seen": 99678016,
      "step": 46180
    },
    {
      "epoch": 7.5342577487765094,
      "grad_norm": 0.0866745337843895,
      "learning_rate": 8.703530707592807e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 99687456,
      "step": 46185
    },
    {
      "epoch": 7.535073409461664,
      "grad_norm": 0.11230406910181046,
      "learning_rate": 8.698133514779297e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 99699584,
      "step": 46190
    },
    {
      "epoch": 7.535889070146819,
      "grad_norm": 0.05673373490571976,
      "learning_rate": 8.692737643477796e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 99710624,
      "step": 46195
    },
    {
      "epoch": 7.536704730831974,
      "grad_norm": 12.691792488098145,
      "learning_rate": 8.687343094125726e-06,
      "loss": 0.1306,
      "num_input_tokens_seen": 99720704,
      "step": 46200
    },
    {
      "epoch": 7.537520391517129,
      "grad_norm": 0.04475383833050728,
      "learning_rate": 8.681949867160396e-06,
      "loss": 0.1146,
      "num_input_tokens_seen": 99729984,
      "step": 46205
    },
    {
      "epoch": 7.5383360522022835,
      "grad_norm": 0.08457732945680618,
      "learning_rate": 8.676557963019005e-06,
      "loss": 0.1244,
      "num_input_tokens_seen": 99741632,
      "step": 46210
    },
    {
      "epoch": 7.539151712887438,
      "grad_norm": 0.08247902244329453,
      "learning_rate": 8.67116738213865e-06,
      "loss": 0.1136,
      "num_input_tokens_seen": 99750688,
      "step": 46215
    },
    {
      "epoch": 7.539967373572594,
      "grad_norm": 0.1298837661743164,
      "learning_rate": 8.66577812495632e-06,
      "loss": 0.1652,
      "num_input_tokens_seen": 99761440,
      "step": 46220
    },
    {
      "epoch": 7.540783034257749,
      "grad_norm": 7.871242046356201,
      "learning_rate": 8.660390191908892e-06,
      "loss": 0.0594,
      "num_input_tokens_seen": 99770912,
      "step": 46225
    },
    {
      "epoch": 7.541598694942904,
      "grad_norm": 2.9351773262023926,
      "learning_rate": 8.655003583433144e-06,
      "loss": 0.1058,
      "num_input_tokens_seen": 99782368,
      "step": 46230
    },
    {
      "epoch": 7.5424143556280585,
      "grad_norm": 0.15129609405994415,
      "learning_rate": 8.649618299965736e-06,
      "loss": 0.0652,
      "num_input_tokens_seen": 99794560,
      "step": 46235
    },
    {
      "epoch": 7.543230016313213,
      "grad_norm": 0.03578079119324684,
      "learning_rate": 8.644234341943232e-06,
      "loss": 0.091,
      "num_input_tokens_seen": 99805184,
      "step": 46240
    },
    {
      "epoch": 7.544045676998369,
      "grad_norm": 0.09130024164915085,
      "learning_rate": 8.638851709802082e-06,
      "loss": 0.1788,
      "num_input_tokens_seen": 99816160,
      "step": 46245
    },
    {
      "epoch": 7.544861337683524,
      "grad_norm": 0.16685356199741364,
      "learning_rate": 8.633470403978625e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 99826784,
      "step": 46250
    },
    {
      "epoch": 7.545676998368679,
      "grad_norm": 0.07381764054298401,
      "learning_rate": 8.628090424909091e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 99837920,
      "step": 46255
    },
    {
      "epoch": 7.5464926590538335,
      "grad_norm": 41.165855407714844,
      "learning_rate": 8.62271177302963e-06,
      "loss": 0.1403,
      "num_input_tokens_seen": 99848672,
      "step": 46260
    },
    {
      "epoch": 7.547308319738988,
      "grad_norm": 0.10340555012226105,
      "learning_rate": 8.617334448776246e-06,
      "loss": 0.1952,
      "num_input_tokens_seen": 99861216,
      "step": 46265
    },
    {
      "epoch": 7.548123980424144,
      "grad_norm": 0.08121224492788315,
      "learning_rate": 8.611958452584859e-06,
      "loss": 0.1579,
      "num_input_tokens_seen": 99870784,
      "step": 46270
    },
    {
      "epoch": 7.548939641109299,
      "grad_norm": 5.6692094802856445,
      "learning_rate": 8.60658378489127e-06,
      "loss": 0.0235,
      "num_input_tokens_seen": 99882272,
      "step": 46275
    },
    {
      "epoch": 7.549755301794454,
      "grad_norm": 0.13333682715892792,
      "learning_rate": 8.60121044613118e-06,
      "loss": 0.0808,
      "num_input_tokens_seen": 99892992,
      "step": 46280
    },
    {
      "epoch": 7.5505709624796085,
      "grad_norm": 0.04906648024916649,
      "learning_rate": 8.595838436740178e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 99903744,
      "step": 46285
    },
    {
      "epoch": 7.551386623164763,
      "grad_norm": 0.05480366200208664,
      "learning_rate": 8.590467757153744e-06,
      "loss": 0.1292,
      "num_input_tokens_seen": 99914688,
      "step": 46290
    },
    {
      "epoch": 7.552202283849918,
      "grad_norm": 0.1344582438468933,
      "learning_rate": 8.585098407807258e-06,
      "loss": 0.0082,
      "num_input_tokens_seen": 99925632,
      "step": 46295
    },
    {
      "epoch": 7.553017944535073,
      "grad_norm": 0.9332029223442078,
      "learning_rate": 8.579730389135973e-06,
      "loss": 0.0619,
      "num_input_tokens_seen": 99935552,
      "step": 46300
    },
    {
      "epoch": 7.553833605220229,
      "grad_norm": 0.239906445145607,
      "learning_rate": 8.574363701575067e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 99946880,
      "step": 46305
    },
    {
      "epoch": 7.554649265905383,
      "grad_norm": 0.14199669659137726,
      "learning_rate": 8.568998345559581e-06,
      "loss": 0.1089,
      "num_input_tokens_seen": 99957792,
      "step": 46310
    },
    {
      "epoch": 7.555464926590538,
      "grad_norm": 0.14231956005096436,
      "learning_rate": 8.56363432152446e-06,
      "loss": 0.1839,
      "num_input_tokens_seen": 99967168,
      "step": 46315
    },
    {
      "epoch": 7.556280587275693,
      "grad_norm": 0.13341651856899261,
      "learning_rate": 8.55827162990454e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 99978784,
      "step": 46320
    },
    {
      "epoch": 7.557096247960848,
      "grad_norm": 0.09734013676643372,
      "learning_rate": 8.552910271134545e-06,
      "loss": 0.1143,
      "num_input_tokens_seen": 99990144,
      "step": 46325
    },
    {
      "epoch": 7.557911908646004,
      "grad_norm": 0.05781449005007744,
      "learning_rate": 8.547550245649095e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 100001568,
      "step": 46330
    },
    {
      "epoch": 7.558727569331158,
      "grad_norm": 3.64250111579895,
      "learning_rate": 8.542191553882701e-06,
      "loss": 0.1238,
      "num_input_tokens_seen": 100012128,
      "step": 46335
    },
    {
      "epoch": 7.559543230016313,
      "grad_norm": 0.06417738646268845,
      "learning_rate": 8.536834196269766e-06,
      "loss": 0.1774,
      "num_input_tokens_seen": 100023936,
      "step": 46340
    },
    {
      "epoch": 7.560358890701468,
      "grad_norm": 3.0087335109710693,
      "learning_rate": 8.531478173244583e-06,
      "loss": 0.0928,
      "num_input_tokens_seen": 100035552,
      "step": 46345
    },
    {
      "epoch": 7.561174551386623,
      "grad_norm": 0.16188861429691315,
      "learning_rate": 8.52612348524134e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 100045408,
      "step": 46350
    },
    {
      "epoch": 7.561990212071779,
      "grad_norm": 0.13789072632789612,
      "learning_rate": 8.520770132694118e-06,
      "loss": 0.1002,
      "num_input_tokens_seen": 100057184,
      "step": 46355
    },
    {
      "epoch": 7.562805872756933,
      "grad_norm": 0.15002018213272095,
      "learning_rate": 8.515418116036872e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 100068384,
      "step": 46360
    },
    {
      "epoch": 7.563621533442088,
      "grad_norm": 1.1017075777053833,
      "learning_rate": 8.510067435703484e-06,
      "loss": 0.1029,
      "num_input_tokens_seen": 100079488,
      "step": 46365
    },
    {
      "epoch": 7.564437194127243,
      "grad_norm": 0.10850369185209274,
      "learning_rate": 8.5047180921277e-06,
      "loss": 0.1924,
      "num_input_tokens_seen": 100090720,
      "step": 46370
    },
    {
      "epoch": 7.565252854812398,
      "grad_norm": 0.33265987038612366,
      "learning_rate": 8.499370085743163e-06,
      "loss": 0.0203,
      "num_input_tokens_seen": 100101248,
      "step": 46375
    },
    {
      "epoch": 7.566068515497553,
      "grad_norm": 0.06324081122875214,
      "learning_rate": 8.49402341698341e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 100112224,
      "step": 46380
    },
    {
      "epoch": 7.566884176182708,
      "grad_norm": 0.10510724782943726,
      "learning_rate": 8.48867808628187e-06,
      "loss": 0.0127,
      "num_input_tokens_seen": 100121632,
      "step": 46385
    },
    {
      "epoch": 7.567699836867863,
      "grad_norm": 0.13526450097560883,
      "learning_rate": 8.483334094071862e-06,
      "loss": 0.083,
      "num_input_tokens_seen": 100133024,
      "step": 46390
    },
    {
      "epoch": 7.568515497553018,
      "grad_norm": 0.2898355722427368,
      "learning_rate": 8.477991440786597e-06,
      "loss": 0.1588,
      "num_input_tokens_seen": 100143008,
      "step": 46395
    },
    {
      "epoch": 7.569331158238173,
      "grad_norm": 0.04432517662644386,
      "learning_rate": 8.472650126859177e-06,
      "loss": 0.2262,
      "num_input_tokens_seen": 100153408,
      "step": 46400
    },
    {
      "epoch": 7.570146818923328,
      "grad_norm": 14.88448715209961,
      "learning_rate": 8.467310152722599e-06,
      "loss": 0.2278,
      "num_input_tokens_seen": 100163744,
      "step": 46405
    },
    {
      "epoch": 7.5709624796084825,
      "grad_norm": 0.14386697113513947,
      "learning_rate": 8.461971518809744e-06,
      "loss": 0.0539,
      "num_input_tokens_seen": 100172608,
      "step": 46410
    },
    {
      "epoch": 7.571778140293638,
      "grad_norm": 0.10405105352401733,
      "learning_rate": 8.456634225553389e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 100182400,
      "step": 46415
    },
    {
      "epoch": 7.572593800978793,
      "grad_norm": 0.5253939628601074,
      "learning_rate": 8.451298273386207e-06,
      "loss": 0.0125,
      "num_input_tokens_seen": 100191936,
      "step": 46420
    },
    {
      "epoch": 7.573409461663948,
      "grad_norm": 0.07708753645420074,
      "learning_rate": 8.445963662740752e-06,
      "loss": 0.0891,
      "num_input_tokens_seen": 100202048,
      "step": 46425
    },
    {
      "epoch": 7.574225122349103,
      "grad_norm": 0.14142228662967682,
      "learning_rate": 8.440630394049479e-06,
      "loss": 0.0964,
      "num_input_tokens_seen": 100213632,
      "step": 46430
    },
    {
      "epoch": 7.575040783034257,
      "grad_norm": 0.08559701591730118,
      "learning_rate": 8.435298467744726e-06,
      "loss": 0.1407,
      "num_input_tokens_seen": 100224096,
      "step": 46435
    },
    {
      "epoch": 7.575856443719413,
      "grad_norm": 0.09634073078632355,
      "learning_rate": 8.429967884258721e-06,
      "loss": 0.1044,
      "num_input_tokens_seen": 100234368,
      "step": 46440
    },
    {
      "epoch": 7.576672104404568,
      "grad_norm": 0.05761324614286423,
      "learning_rate": 8.424638644023603e-06,
      "loss": 0.1281,
      "num_input_tokens_seen": 100244256,
      "step": 46445
    },
    {
      "epoch": 7.577487765089723,
      "grad_norm": 0.2969125211238861,
      "learning_rate": 8.419310747471377e-06,
      "loss": 0.0093,
      "num_input_tokens_seen": 100255360,
      "step": 46450
    },
    {
      "epoch": 7.578303425774878,
      "grad_norm": 0.06729293614625931,
      "learning_rate": 8.413984195033953e-06,
      "loss": 0.0615,
      "num_input_tokens_seen": 100265824,
      "step": 46455
    },
    {
      "epoch": 7.579119086460032,
      "grad_norm": 0.31247928738594055,
      "learning_rate": 8.408658987143125e-06,
      "loss": 0.1944,
      "num_input_tokens_seen": 100276512,
      "step": 46460
    },
    {
      "epoch": 7.579934747145187,
      "grad_norm": 0.028771821409463882,
      "learning_rate": 8.403335124230586e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 100286560,
      "step": 46465
    },
    {
      "epoch": 7.580750407830343,
      "grad_norm": 0.08469962328672409,
      "learning_rate": 8.3980126067279e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 100297856,
      "step": 46470
    },
    {
      "epoch": 7.581566068515498,
      "grad_norm": 0.07814224064350128,
      "learning_rate": 8.392691435066563e-06,
      "loss": 0.2223,
      "num_input_tokens_seen": 100308352,
      "step": 46475
    },
    {
      "epoch": 7.582381729200653,
      "grad_norm": 0.4001646935939789,
      "learning_rate": 8.387371609677921e-06,
      "loss": 0.0137,
      "num_input_tokens_seen": 100317536,
      "step": 46480
    },
    {
      "epoch": 7.583197389885807,
      "grad_norm": 0.11351838707923889,
      "learning_rate": 8.382053130993226e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 100327808,
      "step": 46485
    },
    {
      "epoch": 7.584013050570962,
      "grad_norm": 0.07939837872982025,
      "learning_rate": 8.376735999443624e-06,
      "loss": 0.0928,
      "num_input_tokens_seen": 100339712,
      "step": 46490
    },
    {
      "epoch": 7.584828711256117,
      "grad_norm": 0.592789351940155,
      "learning_rate": 8.371420215460149e-06,
      "loss": 0.1337,
      "num_input_tokens_seen": 100349696,
      "step": 46495
    },
    {
      "epoch": 7.585644371941273,
      "grad_norm": 0.10668357461690903,
      "learning_rate": 8.366105779473723e-06,
      "loss": 0.139,
      "num_input_tokens_seen": 100360160,
      "step": 46500
    },
    {
      "epoch": 7.5864600326264275,
      "grad_norm": 1.1885261535644531,
      "learning_rate": 8.360792691915163e-06,
      "loss": 0.0101,
      "num_input_tokens_seen": 100370848,
      "step": 46505
    },
    {
      "epoch": 7.587275693311582,
      "grad_norm": 0.5904202461242676,
      "learning_rate": 8.35548095321517e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 100381248,
      "step": 46510
    },
    {
      "epoch": 7.588091353996737,
      "grad_norm": 19.061939239501953,
      "learning_rate": 8.350170563804349e-06,
      "loss": 0.0812,
      "num_input_tokens_seen": 100389888,
      "step": 46515
    },
    {
      "epoch": 7.588907014681892,
      "grad_norm": 0.1632000356912613,
      "learning_rate": 8.344861524113178e-06,
      "loss": 0.0426,
      "num_input_tokens_seen": 100399840,
      "step": 46520
    },
    {
      "epoch": 7.589722675367048,
      "grad_norm": 0.202743798494339,
      "learning_rate": 8.339553834572043e-06,
      "loss": 0.099,
      "num_input_tokens_seen": 100410848,
      "step": 46525
    },
    {
      "epoch": 7.5905383360522025,
      "grad_norm": 4.82681941986084,
      "learning_rate": 8.334247495611208e-06,
      "loss": 0.2846,
      "num_input_tokens_seen": 100422368,
      "step": 46530
    },
    {
      "epoch": 7.591353996737357,
      "grad_norm": 0.11173932254314423,
      "learning_rate": 8.32894250766083e-06,
      "loss": 0.1146,
      "num_input_tokens_seen": 100433504,
      "step": 46535
    },
    {
      "epoch": 7.592169657422512,
      "grad_norm": 0.11460330337285995,
      "learning_rate": 8.323638871150962e-06,
      "loss": 0.0093,
      "num_input_tokens_seen": 100443584,
      "step": 46540
    },
    {
      "epoch": 7.592985318107667,
      "grad_norm": 3.662522315979004,
      "learning_rate": 8.31833658651154e-06,
      "loss": 0.1426,
      "num_input_tokens_seen": 100453216,
      "step": 46545
    },
    {
      "epoch": 7.593800978792823,
      "grad_norm": 0.18638724088668823,
      "learning_rate": 8.313035654172399e-06,
      "loss": 0.0771,
      "num_input_tokens_seen": 100463296,
      "step": 46550
    },
    {
      "epoch": 7.5946166394779775,
      "grad_norm": 6.057276248931885,
      "learning_rate": 8.307736074563257e-06,
      "loss": 0.2101,
      "num_input_tokens_seen": 100473728,
      "step": 46555
    },
    {
      "epoch": 7.595432300163132,
      "grad_norm": 0.34686973690986633,
      "learning_rate": 8.302437848113722e-06,
      "loss": 0.1298,
      "num_input_tokens_seen": 100485344,
      "step": 46560
    },
    {
      "epoch": 7.596247960848287,
      "grad_norm": 3.977979898452759,
      "learning_rate": 8.297140975253302e-06,
      "loss": 0.1283,
      "num_input_tokens_seen": 100495552,
      "step": 46565
    },
    {
      "epoch": 7.597063621533442,
      "grad_norm": 0.0953311026096344,
      "learning_rate": 8.291845456411378e-06,
      "loss": 0.1078,
      "num_input_tokens_seen": 100506944,
      "step": 46570
    },
    {
      "epoch": 7.597879282218597,
      "grad_norm": 0.14356254041194916,
      "learning_rate": 8.286551292017233e-06,
      "loss": 0.1116,
      "num_input_tokens_seen": 100517760,
      "step": 46575
    },
    {
      "epoch": 7.598694942903752,
      "grad_norm": 0.42716148495674133,
      "learning_rate": 8.281258482500052e-06,
      "loss": 0.0942,
      "num_input_tokens_seen": 100528768,
      "step": 46580
    },
    {
      "epoch": 7.599510603588907,
      "grad_norm": 5.533297061920166,
      "learning_rate": 8.275967028288886e-06,
      "loss": 0.0964,
      "num_input_tokens_seen": 100539392,
      "step": 46585
    },
    {
      "epoch": 7.600326264274062,
      "grad_norm": 0.12495343387126923,
      "learning_rate": 8.270676929812692e-06,
      "loss": 0.1495,
      "num_input_tokens_seen": 100550176,
      "step": 46590
    },
    {
      "epoch": 7.601141924959217,
      "grad_norm": 3.771775722503662,
      "learning_rate": 8.265388187500309e-06,
      "loss": 0.1383,
      "num_input_tokens_seen": 100561216,
      "step": 46595
    },
    {
      "epoch": 7.601957585644372,
      "grad_norm": 0.0826186090707779,
      "learning_rate": 8.26010080178047e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 100571648,
      "step": 46600
    },
    {
      "epoch": 7.602773246329527,
      "grad_norm": 2.264418125152588,
      "learning_rate": 8.254814773081798e-06,
      "loss": 0.1827,
      "num_input_tokens_seen": 100583776,
      "step": 46605
    },
    {
      "epoch": 7.603588907014682,
      "grad_norm": 0.17515042424201965,
      "learning_rate": 8.249530101832795e-06,
      "loss": 0.0135,
      "num_input_tokens_seen": 100592992,
      "step": 46610
    },
    {
      "epoch": 7.604404567699837,
      "grad_norm": 0.08349408209323883,
      "learning_rate": 8.244246788461882e-06,
      "loss": 0.1155,
      "num_input_tokens_seen": 100603680,
      "step": 46615
    },
    {
      "epoch": 7.605220228384992,
      "grad_norm": 3.912369728088379,
      "learning_rate": 8.238964833397341e-06,
      "loss": 0.1567,
      "num_input_tokens_seen": 100615808,
      "step": 46620
    },
    {
      "epoch": 7.606035889070147,
      "grad_norm": 7.789304733276367,
      "learning_rate": 8.233684237067358e-06,
      "loss": 0.1143,
      "num_input_tokens_seen": 100625696,
      "step": 46625
    },
    {
      "epoch": 7.6068515497553015,
      "grad_norm": 0.26968061923980713,
      "learning_rate": 8.2284049999e-06,
      "loss": 0.1099,
      "num_input_tokens_seen": 100635968,
      "step": 46630
    },
    {
      "epoch": 7.607667210440457,
      "grad_norm": 0.10891762375831604,
      "learning_rate": 8.223127122323231e-06,
      "loss": 0.0378,
      "num_input_tokens_seen": 100647104,
      "step": 46635
    },
    {
      "epoch": 7.608482871125612,
      "grad_norm": 0.026683762669563293,
      "learning_rate": 8.217850604764903e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 100656992,
      "step": 46640
    },
    {
      "epoch": 7.609298531810767,
      "grad_norm": 4.219992637634277,
      "learning_rate": 8.212575447652757e-06,
      "loss": 0.1318,
      "num_input_tokens_seen": 100668288,
      "step": 46645
    },
    {
      "epoch": 7.610114192495922,
      "grad_norm": 0.18180352449417114,
      "learning_rate": 8.207301651414423e-06,
      "loss": 0.216,
      "num_input_tokens_seen": 100680736,
      "step": 46650
    },
    {
      "epoch": 7.6109298531810765,
      "grad_norm": 0.15708036720752716,
      "learning_rate": 8.202029216477425e-06,
      "loss": 0.0098,
      "num_input_tokens_seen": 100692000,
      "step": 46655
    },
    {
      "epoch": 7.611745513866231,
      "grad_norm": 1.7249727249145508,
      "learning_rate": 8.196758143269168e-06,
      "loss": 0.1963,
      "num_input_tokens_seen": 100701440,
      "step": 46660
    },
    {
      "epoch": 7.612561174551386,
      "grad_norm": 0.10883460938930511,
      "learning_rate": 8.191488432216957e-06,
      "loss": 0.1372,
      "num_input_tokens_seen": 100712224,
      "step": 46665
    },
    {
      "epoch": 7.613376835236542,
      "grad_norm": 0.7182273268699646,
      "learning_rate": 8.186220083747975e-06,
      "loss": 0.0079,
      "num_input_tokens_seen": 100722912,
      "step": 46670
    },
    {
      "epoch": 7.614192495921697,
      "grad_norm": 0.07891812920570374,
      "learning_rate": 8.18095309828931e-06,
      "loss": 0.0434,
      "num_input_tokens_seen": 100732384,
      "step": 46675
    },
    {
      "epoch": 7.6150081566068515,
      "grad_norm": 4.916957855224609,
      "learning_rate": 8.175687476267915e-06,
      "loss": 0.2123,
      "num_input_tokens_seen": 100743328,
      "step": 46680
    },
    {
      "epoch": 7.615823817292006,
      "grad_norm": 0.1271146982908249,
      "learning_rate": 8.170423218110667e-06,
      "loss": 0.1043,
      "num_input_tokens_seen": 100754944,
      "step": 46685
    },
    {
      "epoch": 7.616639477977161,
      "grad_norm": 4.182375431060791,
      "learning_rate": 8.165160324244305e-06,
      "loss": 0.319,
      "num_input_tokens_seen": 100767200,
      "step": 46690
    },
    {
      "epoch": 7.617455138662317,
      "grad_norm": 4.3861494064331055,
      "learning_rate": 8.15989879509547e-06,
      "loss": 0.1397,
      "num_input_tokens_seen": 100777408,
      "step": 46695
    },
    {
      "epoch": 7.618270799347472,
      "grad_norm": 0.12118413299322128,
      "learning_rate": 8.15463863109068e-06,
      "loss": 0.2319,
      "num_input_tokens_seen": 100787584,
      "step": 46700
    },
    {
      "epoch": 7.6190864600326265,
      "grad_norm": 0.24924133718013763,
      "learning_rate": 8.149379832656356e-06,
      "loss": 0.0904,
      "num_input_tokens_seen": 100796800,
      "step": 46705
    },
    {
      "epoch": 7.619902120717781,
      "grad_norm": 0.3655216693878174,
      "learning_rate": 8.144122400218804e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 100806528,
      "step": 46710
    },
    {
      "epoch": 7.620717781402936,
      "grad_norm": 0.18927684426307678,
      "learning_rate": 8.138866334204215e-06,
      "loss": 0.0129,
      "num_input_tokens_seen": 100817120,
      "step": 46715
    },
    {
      "epoch": 7.621533442088092,
      "grad_norm": 5.13013219833374,
      "learning_rate": 8.133611635038674e-06,
      "loss": 0.226,
      "num_input_tokens_seen": 100828000,
      "step": 46720
    },
    {
      "epoch": 7.622349102773247,
      "grad_norm": 0.08743203431367874,
      "learning_rate": 8.12835830314815e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 100839424,
      "step": 46725
    },
    {
      "epoch": 7.623164763458401,
      "grad_norm": 3.035060405731201,
      "learning_rate": 8.123106338958511e-06,
      "loss": 0.0833,
      "num_input_tokens_seen": 100849472,
      "step": 46730
    },
    {
      "epoch": 7.623980424143556,
      "grad_norm": 8.867013931274414,
      "learning_rate": 8.117855742895506e-06,
      "loss": 0.063,
      "num_input_tokens_seen": 100859232,
      "step": 46735
    },
    {
      "epoch": 7.624796084828711,
      "grad_norm": 0.35741615295410156,
      "learning_rate": 8.112606515384772e-06,
      "loss": 0.0085,
      "num_input_tokens_seen": 100869888,
      "step": 46740
    },
    {
      "epoch": 7.625611745513866,
      "grad_norm": 0.07205919176340103,
      "learning_rate": 8.107358656851838e-06,
      "loss": 0.26,
      "num_input_tokens_seen": 100881440,
      "step": 46745
    },
    {
      "epoch": 7.626427406199021,
      "grad_norm": 0.17037171125411987,
      "learning_rate": 8.102112167722125e-06,
      "loss": 0.1244,
      "num_input_tokens_seen": 100892160,
      "step": 46750
    },
    {
      "epoch": 7.627243066884176,
      "grad_norm": 0.18843121826648712,
      "learning_rate": 8.096867048420932e-06,
      "loss": 0.1191,
      "num_input_tokens_seen": 100901280,
      "step": 46755
    },
    {
      "epoch": 7.628058727569331,
      "grad_norm": 10.705406188964844,
      "learning_rate": 8.091623299373467e-06,
      "loss": 0.1362,
      "num_input_tokens_seen": 100913440,
      "step": 46760
    },
    {
      "epoch": 7.628874388254486,
      "grad_norm": 0.07794339954853058,
      "learning_rate": 8.08638092100481e-06,
      "loss": 0.091,
      "num_input_tokens_seen": 100923840,
      "step": 46765
    },
    {
      "epoch": 7.629690048939641,
      "grad_norm": 0.14688681066036224,
      "learning_rate": 8.081139913739936e-06,
      "loss": 0.0079,
      "num_input_tokens_seen": 100933824,
      "step": 46770
    },
    {
      "epoch": 7.630505709624796,
      "grad_norm": 0.2018391191959381,
      "learning_rate": 8.075900278003703e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 100943424,
      "step": 46775
    },
    {
      "epoch": 7.631321370309951,
      "grad_norm": 0.33399730920791626,
      "learning_rate": 8.07066201422087e-06,
      "loss": 0.0667,
      "num_input_tokens_seen": 100955040,
      "step": 46780
    },
    {
      "epoch": 7.632137030995106,
      "grad_norm": 13.152643203735352,
      "learning_rate": 8.065425122816061e-06,
      "loss": 0.0238,
      "num_input_tokens_seen": 100966848,
      "step": 46785
    },
    {
      "epoch": 7.632952691680261,
      "grad_norm": 0.0843852087855339,
      "learning_rate": 8.060189604213827e-06,
      "loss": 0.0583,
      "num_input_tokens_seen": 100977888,
      "step": 46790
    },
    {
      "epoch": 7.633768352365416,
      "grad_norm": 0.12785997986793518,
      "learning_rate": 8.054955458838576e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 100988736,
      "step": 46795
    },
    {
      "epoch": 7.634584013050571,
      "grad_norm": 0.10067712515592575,
      "learning_rate": 8.049722687114611e-06,
      "loss": 0.0902,
      "num_input_tokens_seen": 100999360,
      "step": 46800
    },
    {
      "epoch": 7.635399673735726,
      "grad_norm": 0.03972485288977623,
      "learning_rate": 8.044491289466133e-06,
      "loss": 0.006,
      "num_input_tokens_seen": 101011104,
      "step": 46805
    },
    {
      "epoch": 7.636215334420881,
      "grad_norm": 0.1417984664440155,
      "learning_rate": 8.039261266317219e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 101021664,
      "step": 46810
    },
    {
      "epoch": 7.637030995106036,
      "grad_norm": 0.13314402103424072,
      "learning_rate": 8.034032618091846e-06,
      "loss": 0.1178,
      "num_input_tokens_seen": 101032576,
      "step": 46815
    },
    {
      "epoch": 7.637846655791191,
      "grad_norm": 0.18647818267345428,
      "learning_rate": 8.028805345213875e-06,
      "loss": 0.0855,
      "num_input_tokens_seen": 101043040,
      "step": 46820
    },
    {
      "epoch": 7.638662316476346,
      "grad_norm": 0.14198535680770874,
      "learning_rate": 8.023579448107053e-06,
      "loss": 0.2243,
      "num_input_tokens_seen": 101054400,
      "step": 46825
    },
    {
      "epoch": 7.6394779771615005,
      "grad_norm": 0.14789186418056488,
      "learning_rate": 8.018354927195017e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 101064896,
      "step": 46830
    },
    {
      "epoch": 7.640293637846656,
      "grad_norm": 0.11056870967149734,
      "learning_rate": 8.013131782901295e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 101074848,
      "step": 46835
    },
    {
      "epoch": 7.641109298531811,
      "grad_norm": 0.03951028361916542,
      "learning_rate": 8.007910015649304e-06,
      "loss": 0.1201,
      "num_input_tokens_seen": 101085312,
      "step": 46840
    },
    {
      "epoch": 7.641924959216966,
      "grad_norm": 0.09886524826288223,
      "learning_rate": 8.002689625862342e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 101095584,
      "step": 46845
    },
    {
      "epoch": 7.642740619902121,
      "grad_norm": 0.10876501351594925,
      "learning_rate": 7.997470613963601e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 101105760,
      "step": 46850
    },
    {
      "epoch": 7.643556280587275,
      "grad_norm": 0.07424748688936234,
      "learning_rate": 7.992252980376164e-06,
      "loss": 0.0852,
      "num_input_tokens_seen": 101116576,
      "step": 46855
    },
    {
      "epoch": 7.64437194127243,
      "grad_norm": 0.09375285357236862,
      "learning_rate": 7.987036725522995e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 101127840,
      "step": 46860
    },
    {
      "epoch": 7.645187601957586,
      "grad_norm": 0.15094710886478424,
      "learning_rate": 7.981821849826954e-06,
      "loss": 0.135,
      "num_input_tokens_seen": 101140192,
      "step": 46865
    },
    {
      "epoch": 7.646003262642741,
      "grad_norm": 2.2446706295013428,
      "learning_rate": 7.976608353710782e-06,
      "loss": 0.1672,
      "num_input_tokens_seen": 101150336,
      "step": 46870
    },
    {
      "epoch": 7.646818923327896,
      "grad_norm": 0.10287173092365265,
      "learning_rate": 7.971396237597114e-06,
      "loss": 0.0075,
      "num_input_tokens_seen": 101161088,
      "step": 46875
    },
    {
      "epoch": 7.64763458401305,
      "grad_norm": 13.358624458312988,
      "learning_rate": 7.966185501908469e-06,
      "loss": 0.0326,
      "num_input_tokens_seen": 101171776,
      "step": 46880
    },
    {
      "epoch": 7.648450244698205,
      "grad_norm": 0.06350710988044739,
      "learning_rate": 7.960976147067254e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 101180544,
      "step": 46885
    },
    {
      "epoch": 7.649265905383361,
      "grad_norm": 0.39150890707969666,
      "learning_rate": 7.95576817349577e-06,
      "loss": 0.008,
      "num_input_tokens_seen": 101192448,
      "step": 46890
    },
    {
      "epoch": 7.650081566068516,
      "grad_norm": 0.07359413802623749,
      "learning_rate": 7.95056158161619e-06,
      "loss": 0.0165,
      "num_input_tokens_seen": 101203168,
      "step": 46895
    },
    {
      "epoch": 7.650897226753671,
      "grad_norm": 0.5913971662521362,
      "learning_rate": 7.945356371850604e-06,
      "loss": 0.0067,
      "num_input_tokens_seen": 101213216,
      "step": 46900
    },
    {
      "epoch": 7.651712887438825,
      "grad_norm": 0.238386869430542,
      "learning_rate": 7.940152544620966e-06,
      "loss": 0.0444,
      "num_input_tokens_seen": 101224960,
      "step": 46905
    },
    {
      "epoch": 7.65252854812398,
      "grad_norm": 3.611781597137451,
      "learning_rate": 7.934950100349123e-06,
      "loss": 0.1098,
      "num_input_tokens_seen": 101235296,
      "step": 46910
    },
    {
      "epoch": 7.653344208809135,
      "grad_norm": 14.35745620727539,
      "learning_rate": 7.929749039456813e-06,
      "loss": 0.1562,
      "num_input_tokens_seen": 101246784,
      "step": 46915
    },
    {
      "epoch": 7.654159869494291,
      "grad_norm": 0.1352587789297104,
      "learning_rate": 7.924549362365658e-06,
      "loss": 0.1563,
      "num_input_tokens_seen": 101257440,
      "step": 46920
    },
    {
      "epoch": 7.6549755301794455,
      "grad_norm": 0.047789525240659714,
      "learning_rate": 7.919351069497163e-06,
      "loss": 0.0067,
      "num_input_tokens_seen": 101268672,
      "step": 46925
    },
    {
      "epoch": 7.6557911908646,
      "grad_norm": 0.049748245626688004,
      "learning_rate": 7.914154161272746e-06,
      "loss": 0.095,
      "num_input_tokens_seen": 101279168,
      "step": 46930
    },
    {
      "epoch": 7.656606851549755,
      "grad_norm": 5.892858505249023,
      "learning_rate": 7.908958638113687e-06,
      "loss": 0.1287,
      "num_input_tokens_seen": 101290336,
      "step": 46935
    },
    {
      "epoch": 7.65742251223491,
      "grad_norm": 0.07121598720550537,
      "learning_rate": 7.903764500441157e-06,
      "loss": 0.2274,
      "num_input_tokens_seen": 101301344,
      "step": 46940
    },
    {
      "epoch": 7.658238172920065,
      "grad_norm": 0.1507032960653305,
      "learning_rate": 7.898571748676223e-06,
      "loss": 0.1458,
      "num_input_tokens_seen": 101312160,
      "step": 46945
    },
    {
      "epoch": 7.6590538336052205,
      "grad_norm": 3.4130213260650635,
      "learning_rate": 7.893380383239835e-06,
      "loss": 0.1016,
      "num_input_tokens_seen": 101322720,
      "step": 46950
    },
    {
      "epoch": 7.659869494290375,
      "grad_norm": 0.13280156254768372,
      "learning_rate": 7.888190404552832e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 101334048,
      "step": 46955
    },
    {
      "epoch": 7.66068515497553,
      "grad_norm": 3.296661615371704,
      "learning_rate": 7.883001813035937e-06,
      "loss": 0.2718,
      "num_input_tokens_seen": 101345760,
      "step": 46960
    },
    {
      "epoch": 7.661500815660685,
      "grad_norm": 3.592658281326294,
      "learning_rate": 7.877814609109769e-06,
      "loss": 0.1009,
      "num_input_tokens_seen": 101356416,
      "step": 46965
    },
    {
      "epoch": 7.66231647634584,
      "grad_norm": 3.6723384857177734,
      "learning_rate": 7.872628793194823e-06,
      "loss": 0.3296,
      "num_input_tokens_seen": 101367360,
      "step": 46970
    },
    {
      "epoch": 7.6631321370309955,
      "grad_norm": 0.21846187114715576,
      "learning_rate": 7.86744436571149e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 101378848,
      "step": 46975
    },
    {
      "epoch": 7.66394779771615,
      "grad_norm": 0.10069020837545395,
      "learning_rate": 7.86226132708005e-06,
      "loss": 0.0951,
      "num_input_tokens_seen": 101389184,
      "step": 46980
    },
    {
      "epoch": 7.664763458401305,
      "grad_norm": 0.11880723387002945,
      "learning_rate": 7.85707967772066e-06,
      "loss": 0.1587,
      "num_input_tokens_seen": 101399840,
      "step": 46985
    },
    {
      "epoch": 7.66557911908646,
      "grad_norm": 0.3594798147678375,
      "learning_rate": 7.851899418053374e-06,
      "loss": 0.1191,
      "num_input_tokens_seen": 101410112,
      "step": 46990
    },
    {
      "epoch": 7.666394779771615,
      "grad_norm": 0.17258435487747192,
      "learning_rate": 7.846720548498132e-06,
      "loss": 0.0814,
      "num_input_tokens_seen": 101420224,
      "step": 46995
    },
    {
      "epoch": 7.6672104404567705,
      "grad_norm": 0.15103264153003693,
      "learning_rate": 7.841543069474747e-06,
      "loss": 0.0225,
      "num_input_tokens_seen": 101430080,
      "step": 47000
    },
    {
      "epoch": 7.668026101141925,
      "grad_norm": 0.15950919687747955,
      "learning_rate": 7.836366981402951e-06,
      "loss": 0.0619,
      "num_input_tokens_seen": 101441856,
      "step": 47005
    },
    {
      "epoch": 7.66884176182708,
      "grad_norm": 4.0336995124816895,
      "learning_rate": 7.831192284702334e-06,
      "loss": 0.2092,
      "num_input_tokens_seen": 101451232,
      "step": 47010
    },
    {
      "epoch": 7.669657422512235,
      "grad_norm": 0.1788661777973175,
      "learning_rate": 7.826018979792385e-06,
      "loss": 0.0923,
      "num_input_tokens_seen": 101461792,
      "step": 47015
    },
    {
      "epoch": 7.67047308319739,
      "grad_norm": 0.2176932394504547,
      "learning_rate": 7.820847067092477e-06,
      "loss": 0.2219,
      "num_input_tokens_seen": 101472096,
      "step": 47020
    },
    {
      "epoch": 7.671288743882545,
      "grad_norm": 0.24234457314014435,
      "learning_rate": 7.815676547021871e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 101482784,
      "step": 47025
    },
    {
      "epoch": 7.672104404567699,
      "grad_norm": 0.15751397609710693,
      "learning_rate": 7.810507419999716e-06,
      "loss": 0.0795,
      "num_input_tokens_seen": 101493248,
      "step": 47030
    },
    {
      "epoch": 7.672920065252855,
      "grad_norm": 0.13083268702030182,
      "learning_rate": 7.805339686445051e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 101504128,
      "step": 47035
    },
    {
      "epoch": 7.67373572593801,
      "grad_norm": 0.28730228543281555,
      "learning_rate": 7.800173346776793e-06,
      "loss": 0.0085,
      "num_input_tokens_seen": 101515808,
      "step": 47040
    },
    {
      "epoch": 7.674551386623165,
      "grad_norm": 0.059784840792417526,
      "learning_rate": 7.795008401413756e-06,
      "loss": 0.142,
      "num_input_tokens_seen": 101526976,
      "step": 47045
    },
    {
      "epoch": 7.6753670473083195,
      "grad_norm": 0.08952692151069641,
      "learning_rate": 7.789844850774636e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 101537440,
      "step": 47050
    },
    {
      "epoch": 7.676182707993474,
      "grad_norm": 0.19022005796432495,
      "learning_rate": 7.784682695278014e-06,
      "loss": 0.094,
      "num_input_tokens_seen": 101549280,
      "step": 47055
    },
    {
      "epoch": 7.67699836867863,
      "grad_norm": 0.11355360597372055,
      "learning_rate": 7.779521935342363e-06,
      "loss": 0.0075,
      "num_input_tokens_seen": 101558720,
      "step": 47060
    },
    {
      "epoch": 7.677814029363785,
      "grad_norm": 0.03983641788363457,
      "learning_rate": 7.77436257138604e-06,
      "loss": 0.2278,
      "num_input_tokens_seen": 101570400,
      "step": 47065
    },
    {
      "epoch": 7.67862969004894,
      "grad_norm": 3.2674293518066406,
      "learning_rate": 7.769204603827282e-06,
      "loss": 0.1675,
      "num_input_tokens_seen": 101581472,
      "step": 47070
    },
    {
      "epoch": 7.6794453507340945,
      "grad_norm": 0.07456944137811661,
      "learning_rate": 7.764048033084235e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 101592480,
      "step": 47075
    },
    {
      "epoch": 7.680261011419249,
      "grad_norm": 6.097672462463379,
      "learning_rate": 7.758892859574906e-06,
      "loss": 0.0866,
      "num_input_tokens_seen": 101601952,
      "step": 47080
    },
    {
      "epoch": 7.681076672104405,
      "grad_norm": 0.1380632221698761,
      "learning_rate": 7.753739083717204e-06,
      "loss": 0.0369,
      "num_input_tokens_seen": 101612384,
      "step": 47085
    },
    {
      "epoch": 7.68189233278956,
      "grad_norm": 0.21851445734500885,
      "learning_rate": 7.748586705928917e-06,
      "loss": 0.0109,
      "num_input_tokens_seen": 101623872,
      "step": 47090
    },
    {
      "epoch": 7.682707993474715,
      "grad_norm": 0.11787327378988266,
      "learning_rate": 7.743435726627726e-06,
      "loss": 0.0424,
      "num_input_tokens_seen": 101635072,
      "step": 47095
    },
    {
      "epoch": 7.6835236541598695,
      "grad_norm": 0.09346489608287811,
      "learning_rate": 7.738286146231194e-06,
      "loss": 0.1052,
      "num_input_tokens_seen": 101644896,
      "step": 47100
    },
    {
      "epoch": 7.684339314845024,
      "grad_norm": 0.11581163108348846,
      "learning_rate": 7.733137965156764e-06,
      "loss": 0.0059,
      "num_input_tokens_seen": 101655584,
      "step": 47105
    },
    {
      "epoch": 7.685154975530179,
      "grad_norm": 0.1141398698091507,
      "learning_rate": 7.727991183821792e-06,
      "loss": 0.0172,
      "num_input_tokens_seen": 101666912,
      "step": 47110
    },
    {
      "epoch": 7.685970636215334,
      "grad_norm": 0.057288192212581635,
      "learning_rate": 7.722845802643489e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 101676768,
      "step": 47115
    },
    {
      "epoch": 7.68678629690049,
      "grad_norm": 0.08691194653511047,
      "learning_rate": 7.71770182203897e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 101687936,
      "step": 47120
    },
    {
      "epoch": 7.6876019575856445,
      "grad_norm": 3.4730427265167236,
      "learning_rate": 7.71255924242523e-06,
      "loss": 0.394,
      "num_input_tokens_seen": 101698592,
      "step": 47125
    },
    {
      "epoch": 7.688417618270799,
      "grad_norm": 0.6738741397857666,
      "learning_rate": 7.707418064219152e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 101709312,
      "step": 47130
    },
    {
      "epoch": 7.689233278955954,
      "grad_norm": 0.06230625882744789,
      "learning_rate": 7.702278287837509e-06,
      "loss": 0.0915,
      "num_input_tokens_seen": 101720512,
      "step": 47135
    },
    {
      "epoch": 7.690048939641109,
      "grad_norm": 0.05752212926745415,
      "learning_rate": 7.697139913696955e-06,
      "loss": 0.0314,
      "num_input_tokens_seen": 101732640,
      "step": 47140
    },
    {
      "epoch": 7.690864600326265,
      "grad_norm": 0.0707167237997055,
      "learning_rate": 7.692002942214035e-06,
      "loss": 0.0088,
      "num_input_tokens_seen": 101742400,
      "step": 47145
    },
    {
      "epoch": 7.691680261011419,
      "grad_norm": 0.07063805311918259,
      "learning_rate": 7.686867373805176e-06,
      "loss": 0.3276,
      "num_input_tokens_seen": 101751936,
      "step": 47150
    },
    {
      "epoch": 7.692495921696574,
      "grad_norm": 8.095948219299316,
      "learning_rate": 7.681733208886693e-06,
      "loss": 0.1036,
      "num_input_tokens_seen": 101763136,
      "step": 47155
    },
    {
      "epoch": 7.693311582381729,
      "grad_norm": 1.144657850265503,
      "learning_rate": 7.676600447874788e-06,
      "loss": 0.0264,
      "num_input_tokens_seen": 101773440,
      "step": 47160
    },
    {
      "epoch": 7.694127243066884,
      "grad_norm": 0.17921125888824463,
      "learning_rate": 7.67146909118555e-06,
      "loss": 0.2127,
      "num_input_tokens_seen": 101783264,
      "step": 47165
    },
    {
      "epoch": 7.69494290375204,
      "grad_norm": 0.08309762924909592,
      "learning_rate": 7.666339139234949e-06,
      "loss": 0.0865,
      "num_input_tokens_seen": 101793856,
      "step": 47170
    },
    {
      "epoch": 7.695758564437194,
      "grad_norm": 0.13982199132442474,
      "learning_rate": 7.66121059243885e-06,
      "loss": 0.1073,
      "num_input_tokens_seen": 101804832,
      "step": 47175
    },
    {
      "epoch": 7.696574225122349,
      "grad_norm": 0.1988714635372162,
      "learning_rate": 7.656083451212995e-06,
      "loss": 0.1206,
      "num_input_tokens_seen": 101815808,
      "step": 47180
    },
    {
      "epoch": 7.697389885807504,
      "grad_norm": 0.04910849779844284,
      "learning_rate": 7.650957715973017e-06,
      "loss": 0.0895,
      "num_input_tokens_seen": 101825120,
      "step": 47185
    },
    {
      "epoch": 7.698205546492659,
      "grad_norm": 0.6744894981384277,
      "learning_rate": 7.645833387134437e-06,
      "loss": 0.0997,
      "num_input_tokens_seen": 101836416,
      "step": 47190
    },
    {
      "epoch": 7.699021207177814,
      "grad_norm": 0.16725163161754608,
      "learning_rate": 7.640710465112654e-06,
      "loss": 0.0979,
      "num_input_tokens_seen": 101846240,
      "step": 47195
    },
    {
      "epoch": 7.699836867862969,
      "grad_norm": 0.12873879075050354,
      "learning_rate": 7.635588950322964e-06,
      "loss": 0.0226,
      "num_input_tokens_seen": 101857152,
      "step": 47200
    },
    {
      "epoch": 7.700652528548124,
      "grad_norm": 0.1449815183877945,
      "learning_rate": 7.630468843180538e-06,
      "loss": 0.0943,
      "num_input_tokens_seen": 101867296,
      "step": 47205
    },
    {
      "epoch": 7.701468189233279,
      "grad_norm": 0.20121796429157257,
      "learning_rate": 7.625350144100441e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 101876000,
      "step": 47210
    },
    {
      "epoch": 7.702283849918434,
      "grad_norm": 1.4556446075439453,
      "learning_rate": 7.620232853497611e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 101886080,
      "step": 47215
    },
    {
      "epoch": 7.703099510603589,
      "grad_norm": 0.19958502054214478,
      "learning_rate": 7.615116971786895e-06,
      "loss": 0.0085,
      "num_input_tokens_seen": 101896480,
      "step": 47220
    },
    {
      "epoch": 7.7039151712887435,
      "grad_norm": 0.017091643065214157,
      "learning_rate": 7.610002499383012e-06,
      "loss": 0.0477,
      "num_input_tokens_seen": 101907136,
      "step": 47225
    },
    {
      "epoch": 7.704730831973899,
      "grad_norm": 0.07241757214069366,
      "learning_rate": 7.60488943670056e-06,
      "loss": 0.1033,
      "num_input_tokens_seen": 101918624,
      "step": 47230
    },
    {
      "epoch": 7.705546492659054,
      "grad_norm": 0.07324308902025223,
      "learning_rate": 7.59977778415403e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 101929408,
      "step": 47235
    },
    {
      "epoch": 7.706362153344209,
      "grad_norm": 0.05124499648809433,
      "learning_rate": 7.594667542157796e-06,
      "loss": 0.0059,
      "num_input_tokens_seen": 101940736,
      "step": 47240
    },
    {
      "epoch": 7.707177814029364,
      "grad_norm": 4.194717884063721,
      "learning_rate": 7.5895587111261325e-06,
      "loss": 0.0208,
      "num_input_tokens_seen": 101950848,
      "step": 47245
    },
    {
      "epoch": 7.7079934747145185,
      "grad_norm": 15.6417236328125,
      "learning_rate": 7.584451291473177e-06,
      "loss": 0.1458,
      "num_input_tokens_seen": 101961376,
      "step": 47250
    },
    {
      "epoch": 7.708809135399674,
      "grad_norm": 0.12029201537370682,
      "learning_rate": 7.579345283612968e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 101972448,
      "step": 47255
    },
    {
      "epoch": 7.709624796084829,
      "grad_norm": 0.1028146743774414,
      "learning_rate": 7.574240687959422e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 101983712,
      "step": 47260
    },
    {
      "epoch": 7.710440456769984,
      "grad_norm": 0.1645936667919159,
      "learning_rate": 7.56913750492634e-06,
      "loss": 0.1216,
      "num_input_tokens_seen": 101993792,
      "step": 47265
    },
    {
      "epoch": 7.711256117455139,
      "grad_norm": 0.09468498826026917,
      "learning_rate": 7.564035734927419e-06,
      "loss": 0.0236,
      "num_input_tokens_seen": 102004800,
      "step": 47270
    },
    {
      "epoch": 7.712071778140293,
      "grad_norm": 10.034408569335938,
      "learning_rate": 7.558935378376228e-06,
      "loss": 0.1065,
      "num_input_tokens_seen": 102015936,
      "step": 47275
    },
    {
      "epoch": 7.712887438825448,
      "grad_norm": 0.3068498373031616,
      "learning_rate": 7.553836435686232e-06,
      "loss": 0.1882,
      "num_input_tokens_seen": 102026592,
      "step": 47280
    },
    {
      "epoch": 7.713703099510604,
      "grad_norm": 6.699936389923096,
      "learning_rate": 7.5487389072707744e-06,
      "loss": 0.0686,
      "num_input_tokens_seen": 102036448,
      "step": 47285
    },
    {
      "epoch": 7.714518760195759,
      "grad_norm": 3.9150755405426025,
      "learning_rate": 7.543642793543088e-06,
      "loss": 0.1662,
      "num_input_tokens_seen": 102048800,
      "step": 47290
    },
    {
      "epoch": 7.715334420880914,
      "grad_norm": 0.05595221742987633,
      "learning_rate": 7.53854809491629e-06,
      "loss": 0.0428,
      "num_input_tokens_seen": 102060000,
      "step": 47295
    },
    {
      "epoch": 7.716150081566068,
      "grad_norm": 0.1571471095085144,
      "learning_rate": 7.533454811803381e-06,
      "loss": 0.0086,
      "num_input_tokens_seen": 102071104,
      "step": 47300
    },
    {
      "epoch": 7.716965742251223,
      "grad_norm": 4.362936973571777,
      "learning_rate": 7.528362944617251e-06,
      "loss": 0.1903,
      "num_input_tokens_seen": 102082880,
      "step": 47305
    },
    {
      "epoch": 7.717781402936378,
      "grad_norm": 0.07679007202386856,
      "learning_rate": 7.523272493770669e-06,
      "loss": 0.0853,
      "num_input_tokens_seen": 102094816,
      "step": 47310
    },
    {
      "epoch": 7.718597063621534,
      "grad_norm": 0.12573713064193726,
      "learning_rate": 7.518183459676295e-06,
      "loss": 0.0322,
      "num_input_tokens_seen": 102105632,
      "step": 47315
    },
    {
      "epoch": 7.719412724306689,
      "grad_norm": 3.7796518802642822,
      "learning_rate": 7.513095842746665e-06,
      "loss": 0.1788,
      "num_input_tokens_seen": 102115808,
      "step": 47320
    },
    {
      "epoch": 7.720228384991843,
      "grad_norm": 8.606292724609375,
      "learning_rate": 7.5080096433942204e-06,
      "loss": 0.0794,
      "num_input_tokens_seen": 102127392,
      "step": 47325
    },
    {
      "epoch": 7.721044045676998,
      "grad_norm": 0.5781602263450623,
      "learning_rate": 7.502924862031269e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 102139136,
      "step": 47330
    },
    {
      "epoch": 7.721859706362153,
      "grad_norm": 0.03822608292102814,
      "learning_rate": 7.497841499070005e-06,
      "loss": 0.2096,
      "num_input_tokens_seen": 102150688,
      "step": 47335
    },
    {
      "epoch": 7.722675367047309,
      "grad_norm": 3.7553579807281494,
      "learning_rate": 7.4927595549225155e-06,
      "loss": 0.411,
      "num_input_tokens_seen": 102161568,
      "step": 47340
    },
    {
      "epoch": 7.7234910277324635,
      "grad_norm": 0.05443377047777176,
      "learning_rate": 7.487679030000769e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 102173440,
      "step": 47345
    },
    {
      "epoch": 7.724306688417618,
      "grad_norm": 0.08578251302242279,
      "learning_rate": 7.482599924716613e-06,
      "loss": 0.0029,
      "num_input_tokens_seen": 102185376,
      "step": 47350
    },
    {
      "epoch": 7.725122349102773,
      "grad_norm": 0.20941054821014404,
      "learning_rate": 7.477522239481793e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 102195456,
      "step": 47355
    },
    {
      "epoch": 7.725938009787928,
      "grad_norm": 3.4213364124298096,
      "learning_rate": 7.472445974707928e-06,
      "loss": 0.0877,
      "num_input_tokens_seen": 102206080,
      "step": 47360
    },
    {
      "epoch": 7.726753670473083,
      "grad_norm": 5.20107889175415,
      "learning_rate": 7.467371130806524e-06,
      "loss": 0.1009,
      "num_input_tokens_seen": 102216736,
      "step": 47365
    },
    {
      "epoch": 7.7275693311582385,
      "grad_norm": 0.06946857273578644,
      "learning_rate": 7.462297708188978e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 102227904,
      "step": 47370
    },
    {
      "epoch": 7.728384991843393,
      "grad_norm": 1.2200431823730469,
      "learning_rate": 7.457225707266566e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 102239200,
      "step": 47375
    },
    {
      "epoch": 7.729200652528548,
      "grad_norm": 0.06667802482843399,
      "learning_rate": 7.452155128450447e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 102251264,
      "step": 47380
    },
    {
      "epoch": 7.730016313213703,
      "grad_norm": 0.0479462705552578,
      "learning_rate": 7.447085972151663e-06,
      "loss": 0.0991,
      "num_input_tokens_seen": 102262432,
      "step": 47385
    },
    {
      "epoch": 7.730831973898858,
      "grad_norm": 0.32065635919570923,
      "learning_rate": 7.4420182387811596e-06,
      "loss": 0.0699,
      "num_input_tokens_seen": 102274912,
      "step": 47390
    },
    {
      "epoch": 7.731647634584013,
      "grad_norm": 0.13018104434013367,
      "learning_rate": 7.436951928749747e-06,
      "loss": 0.0122,
      "num_input_tokens_seen": 102285696,
      "step": 47395
    },
    {
      "epoch": 7.732463295269168,
      "grad_norm": 4.171314239501953,
      "learning_rate": 7.431887042468125e-06,
      "loss": 0.0592,
      "num_input_tokens_seen": 102296160,
      "step": 47400
    },
    {
      "epoch": 7.733278955954323,
      "grad_norm": 0.17030827701091766,
      "learning_rate": 7.426823580346881e-06,
      "loss": 0.0143,
      "num_input_tokens_seen": 102307712,
      "step": 47405
    },
    {
      "epoch": 7.734094616639478,
      "grad_norm": 0.03789321705698967,
      "learning_rate": 7.421761542796479e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 102320032,
      "step": 47410
    },
    {
      "epoch": 7.734910277324633,
      "grad_norm": 0.07053886353969574,
      "learning_rate": 7.41670093022728e-06,
      "loss": 0.147,
      "num_input_tokens_seen": 102330656,
      "step": 47415
    },
    {
      "epoch": 7.735725938009788,
      "grad_norm": 0.12778106331825256,
      "learning_rate": 7.411641743049522e-06,
      "loss": 0.1072,
      "num_input_tokens_seen": 102340704,
      "step": 47420
    },
    {
      "epoch": 7.736541598694943,
      "grad_norm": 0.06977693736553192,
      "learning_rate": 7.406583981673315e-06,
      "loss": 0.0512,
      "num_input_tokens_seen": 102350752,
      "step": 47425
    },
    {
      "epoch": 7.737357259380098,
      "grad_norm": 0.02339315600693226,
      "learning_rate": 7.401527646508691e-06,
      "loss": 0.0768,
      "num_input_tokens_seen": 102360800,
      "step": 47430
    },
    {
      "epoch": 7.738172920065253,
      "grad_norm": 0.150446355342865,
      "learning_rate": 7.396472737965526e-06,
      "loss": 0.0613,
      "num_input_tokens_seen": 102372096,
      "step": 47435
    },
    {
      "epoch": 7.738988580750408,
      "grad_norm": 0.06412381678819656,
      "learning_rate": 7.391419256453602e-06,
      "loss": 0.3177,
      "num_input_tokens_seen": 102382112,
      "step": 47440
    },
    {
      "epoch": 7.739804241435563,
      "grad_norm": 0.7351124882698059,
      "learning_rate": 7.386367202382577e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 102391392,
      "step": 47445
    },
    {
      "epoch": 7.740619902120718,
      "grad_norm": 2.7334470748901367,
      "learning_rate": 7.3813165761619975e-06,
      "loss": 0.0851,
      "num_input_tokens_seen": 102402720,
      "step": 47450
    },
    {
      "epoch": 7.741435562805873,
      "grad_norm": 0.056419868022203445,
      "learning_rate": 7.376267378201293e-06,
      "loss": 0.2647,
      "num_input_tokens_seen": 102412992,
      "step": 47455
    },
    {
      "epoch": 7.742251223491028,
      "grad_norm": 2.9997851848602295,
      "learning_rate": 7.371219608909777e-06,
      "loss": 0.1289,
      "num_input_tokens_seen": 102424416,
      "step": 47460
    },
    {
      "epoch": 7.743066884176183,
      "grad_norm": 0.060516271740198135,
      "learning_rate": 7.366173268696646e-06,
      "loss": 0.1282,
      "num_input_tokens_seen": 102435296,
      "step": 47465
    },
    {
      "epoch": 7.7438825448613375,
      "grad_norm": 4.344020843505859,
      "learning_rate": 7.3611283579709835e-06,
      "loss": 0.0793,
      "num_input_tokens_seen": 102445856,
      "step": 47470
    },
    {
      "epoch": 7.744698205546492,
      "grad_norm": 0.1005115956068039,
      "learning_rate": 7.356084877141756e-06,
      "loss": 0.0025,
      "num_input_tokens_seen": 102457024,
      "step": 47475
    },
    {
      "epoch": 7.745513866231647,
      "grad_norm": 9.595219612121582,
      "learning_rate": 7.35104282661781e-06,
      "loss": 0.06,
      "num_input_tokens_seen": 102467744,
      "step": 47480
    },
    {
      "epoch": 7.746329526916803,
      "grad_norm": 9.318653106689453,
      "learning_rate": 7.346002206807887e-06,
      "loss": 0.0219,
      "num_input_tokens_seen": 102477408,
      "step": 47485
    },
    {
      "epoch": 7.747145187601958,
      "grad_norm": 5.1646881103515625,
      "learning_rate": 7.340963018120597e-06,
      "loss": 0.0741,
      "num_input_tokens_seen": 102487840,
      "step": 47490
    },
    {
      "epoch": 7.7479608482871125,
      "grad_norm": 0.06552334129810333,
      "learning_rate": 7.335925260964446e-06,
      "loss": 0.0026,
      "num_input_tokens_seen": 102498624,
      "step": 47495
    },
    {
      "epoch": 7.748776508972267,
      "grad_norm": 0.14217686653137207,
      "learning_rate": 7.330888935747821e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 102509952,
      "step": 47500
    },
    {
      "epoch": 7.749592169657422,
      "grad_norm": 0.03769100457429886,
      "learning_rate": 7.325854042878991e-06,
      "loss": 0.0022,
      "num_input_tokens_seen": 102521120,
      "step": 47505
    },
    {
      "epoch": 7.750407830342578,
      "grad_norm": 0.14183053374290466,
      "learning_rate": 7.320820582766108e-06,
      "loss": 0.2305,
      "num_input_tokens_seen": 102532832,
      "step": 47510
    },
    {
      "epoch": 7.751223491027733,
      "grad_norm": 0.407478392124176,
      "learning_rate": 7.315788555817215e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 102542752,
      "step": 47515
    },
    {
      "epoch": 7.7520391517128875,
      "grad_norm": 0.07920314371585846,
      "learning_rate": 7.3107579624402286e-06,
      "loss": 0.0072,
      "num_input_tokens_seen": 102553920,
      "step": 47520
    },
    {
      "epoch": 7.752854812398042,
      "grad_norm": 0.12087984383106232,
      "learning_rate": 7.305728803042949e-06,
      "loss": 0.1148,
      "num_input_tokens_seen": 102565088,
      "step": 47525
    },
    {
      "epoch": 7.753670473083197,
      "grad_norm": 0.10578302294015884,
      "learning_rate": 7.3007010780330785e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 102576704,
      "step": 47530
    },
    {
      "epoch": 7.754486133768353,
      "grad_norm": 0.05380365997552872,
      "learning_rate": 7.295674787818188e-06,
      "loss": 0.2452,
      "num_input_tokens_seen": 102588000,
      "step": 47535
    },
    {
      "epoch": 7.755301794453508,
      "grad_norm": 1.0776162147521973,
      "learning_rate": 7.290649932805726e-06,
      "loss": 0.0142,
      "num_input_tokens_seen": 102598144,
      "step": 47540
    },
    {
      "epoch": 7.7561174551386625,
      "grad_norm": 0.0860818549990654,
      "learning_rate": 7.285626513403038e-06,
      "loss": 0.0297,
      "num_input_tokens_seen": 102608256,
      "step": 47545
    },
    {
      "epoch": 7.756933115823817,
      "grad_norm": 0.14374639093875885,
      "learning_rate": 7.2806045300173484e-06,
      "loss": 0.0559,
      "num_input_tokens_seen": 102618528,
      "step": 47550
    },
    {
      "epoch": 7.757748776508972,
      "grad_norm": 0.7806813716888428,
      "learning_rate": 7.275583983055753e-06,
      "loss": 0.167,
      "num_input_tokens_seen": 102629632,
      "step": 47555
    },
    {
      "epoch": 7.758564437194127,
      "grad_norm": 0.09024453163146973,
      "learning_rate": 7.2705648729252615e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 102639712,
      "step": 47560
    },
    {
      "epoch": 7.759380097879282,
      "grad_norm": 0.06424576789140701,
      "learning_rate": 7.265547200032738e-06,
      "loss": 0.009,
      "num_input_tokens_seen": 102650272,
      "step": 47565
    },
    {
      "epoch": 7.760195758564437,
      "grad_norm": 0.13625803589820862,
      "learning_rate": 7.260530964784945e-06,
      "loss": 0.1459,
      "num_input_tokens_seen": 102661216,
      "step": 47570
    },
    {
      "epoch": 7.761011419249592,
      "grad_norm": 0.09976625442504883,
      "learning_rate": 7.2555161675885195e-06,
      "loss": 0.0838,
      "num_input_tokens_seen": 102671200,
      "step": 47575
    },
    {
      "epoch": 7.761827079934747,
      "grad_norm": 0.12595100700855255,
      "learning_rate": 7.250502808849988e-06,
      "loss": 0.1217,
      "num_input_tokens_seen": 102682144,
      "step": 47580
    },
    {
      "epoch": 7.762642740619902,
      "grad_norm": 3.1571035385131836,
      "learning_rate": 7.2454908889757586e-06,
      "loss": 0.1977,
      "num_input_tokens_seen": 102693376,
      "step": 47585
    },
    {
      "epoch": 7.763458401305057,
      "grad_norm": 0.11334747821092606,
      "learning_rate": 7.240480408372125e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 102704800,
      "step": 47590
    },
    {
      "epoch": 7.764274061990212,
      "grad_norm": 2.676896572113037,
      "learning_rate": 7.235471367445257e-06,
      "loss": 0.2064,
      "num_input_tokens_seen": 102715776,
      "step": 47595
    },
    {
      "epoch": 7.765089722675367,
      "grad_norm": 10.041728973388672,
      "learning_rate": 7.2304637666012195e-06,
      "loss": 0.2402,
      "num_input_tokens_seen": 102727232,
      "step": 47600
    },
    {
      "epoch": 7.765905383360522,
      "grad_norm": 0.1420239955186844,
      "learning_rate": 7.22545760624595e-06,
      "loss": 0.015,
      "num_input_tokens_seen": 102737568,
      "step": 47605
    },
    {
      "epoch": 7.766721044045677,
      "grad_norm": 0.06168798729777336,
      "learning_rate": 7.2204528867852725e-06,
      "loss": 0.2718,
      "num_input_tokens_seen": 102749728,
      "step": 47610
    },
    {
      "epoch": 7.767536704730832,
      "grad_norm": 1.830137848854065,
      "learning_rate": 7.215449608624899e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 102760256,
      "step": 47615
    },
    {
      "epoch": 7.768352365415987,
      "grad_norm": 0.15203669667243958,
      "learning_rate": 7.210447772170418e-06,
      "loss": 0.0383,
      "num_input_tokens_seen": 102771648,
      "step": 47620
    },
    {
      "epoch": 7.769168026101142,
      "grad_norm": 0.1616964191198349,
      "learning_rate": 7.205447377827301e-06,
      "loss": 0.1632,
      "num_input_tokens_seen": 102782912,
      "step": 47625
    },
    {
      "epoch": 7.769983686786297,
      "grad_norm": 0.06401637196540833,
      "learning_rate": 7.200448426000911e-06,
      "loss": 0.1192,
      "num_input_tokens_seen": 102795200,
      "step": 47630
    },
    {
      "epoch": 7.770799347471452,
      "grad_norm": 0.3197227418422699,
      "learning_rate": 7.195450917096483e-06,
      "loss": 0.1945,
      "num_input_tokens_seen": 102805600,
      "step": 47635
    },
    {
      "epoch": 7.771615008156607,
      "grad_norm": 3.529085397720337,
      "learning_rate": 7.190454851519138e-06,
      "loss": 0.1202,
      "num_input_tokens_seen": 102816800,
      "step": 47640
    },
    {
      "epoch": 7.7724306688417615,
      "grad_norm": 0.09904611110687256,
      "learning_rate": 7.185460229673893e-06,
      "loss": 0.105,
      "num_input_tokens_seen": 102827968,
      "step": 47645
    },
    {
      "epoch": 7.773246329526917,
      "grad_norm": 0.019798442721366882,
      "learning_rate": 7.180467051965634e-06,
      "loss": 0.0173,
      "num_input_tokens_seen": 102838400,
      "step": 47650
    },
    {
      "epoch": 7.774061990212072,
      "grad_norm": 0.40912890434265137,
      "learning_rate": 7.17547531879913e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 102848704,
      "step": 47655
    },
    {
      "epoch": 7.774877650897227,
      "grad_norm": 0.08395379036664963,
      "learning_rate": 7.170485030579038e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 102859296,
      "step": 47660
    },
    {
      "epoch": 7.775693311582382,
      "grad_norm": 0.03093654103577137,
      "learning_rate": 7.165496187709894e-06,
      "loss": 0.0088,
      "num_input_tokens_seen": 102870592,
      "step": 47665
    },
    {
      "epoch": 7.7765089722675365,
      "grad_norm": 7.923911094665527,
      "learning_rate": 7.160508790596121e-06,
      "loss": 0.1545,
      "num_input_tokens_seen": 102881568,
      "step": 47670
    },
    {
      "epoch": 7.777324632952691,
      "grad_norm": 12.883262634277344,
      "learning_rate": 7.155522839642023e-06,
      "loss": 0.14,
      "num_input_tokens_seen": 102892320,
      "step": 47675
    },
    {
      "epoch": 7.778140293637847,
      "grad_norm": 1.4507733583450317,
      "learning_rate": 7.150538335251786e-06,
      "loss": 0.1976,
      "num_input_tokens_seen": 102903456,
      "step": 47680
    },
    {
      "epoch": 7.778955954323002,
      "grad_norm": 0.08981003612279892,
      "learning_rate": 7.1455552778294775e-06,
      "loss": 0.08,
      "num_input_tokens_seen": 102914208,
      "step": 47685
    },
    {
      "epoch": 7.779771615008157,
      "grad_norm": 5.994715690612793,
      "learning_rate": 7.140573667779052e-06,
      "loss": 0.1083,
      "num_input_tokens_seen": 102925152,
      "step": 47690
    },
    {
      "epoch": 7.780587275693311,
      "grad_norm": 0.1755041629076004,
      "learning_rate": 7.1355935055043314e-06,
      "loss": 0.1214,
      "num_input_tokens_seen": 102936064,
      "step": 47695
    },
    {
      "epoch": 7.781402936378466,
      "grad_norm": 0.062174130231142044,
      "learning_rate": 7.130614791409057e-06,
      "loss": 0.0663,
      "num_input_tokens_seen": 102947200,
      "step": 47700
    },
    {
      "epoch": 7.782218597063622,
      "grad_norm": 3.0506484508514404,
      "learning_rate": 7.125637525896814e-06,
      "loss": 0.1461,
      "num_input_tokens_seen": 102957216,
      "step": 47705
    },
    {
      "epoch": 7.783034257748777,
      "grad_norm": 0.035387951880693436,
      "learning_rate": 7.1206617093710845e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 102968160,
      "step": 47710
    },
    {
      "epoch": 7.783849918433932,
      "grad_norm": 0.10718975961208344,
      "learning_rate": 7.115687342235239e-06,
      "loss": 0.0384,
      "num_input_tokens_seen": 102978912,
      "step": 47715
    },
    {
      "epoch": 7.784665579119086,
      "grad_norm": 7.703681468963623,
      "learning_rate": 7.11071442489252e-06,
      "loss": 0.374,
      "num_input_tokens_seen": 102989568,
      "step": 47720
    },
    {
      "epoch": 7.785481239804241,
      "grad_norm": 0.05116169899702072,
      "learning_rate": 7.1057429577460584e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 103000512,
      "step": 47725
    },
    {
      "epoch": 7.786296900489396,
      "grad_norm": 0.051105331629514694,
      "learning_rate": 7.100772941198869e-06,
      "loss": 0.125,
      "num_input_tokens_seen": 103012320,
      "step": 47730
    },
    {
      "epoch": 7.787112561174552,
      "grad_norm": 0.026301920413970947,
      "learning_rate": 7.095804375653844e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 103023936,
      "step": 47735
    },
    {
      "epoch": 7.787928221859707,
      "grad_norm": 11.353914260864258,
      "learning_rate": 7.090837261513764e-06,
      "loss": 0.0834,
      "num_input_tokens_seen": 103033696,
      "step": 47740
    },
    {
      "epoch": 7.788743882544861,
      "grad_norm": 5.692740440368652,
      "learning_rate": 7.085871599181274e-06,
      "loss": 0.1808,
      "num_input_tokens_seen": 103045280,
      "step": 47745
    },
    {
      "epoch": 7.789559543230016,
      "grad_norm": 0.1049162745475769,
      "learning_rate": 7.0809073890589356e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 103055584,
      "step": 47750
    },
    {
      "epoch": 7.790375203915171,
      "grad_norm": 0.1505405306816101,
      "learning_rate": 7.075944631549167e-06,
      "loss": 0.0808,
      "num_input_tokens_seen": 103066880,
      "step": 47755
    },
    {
      "epoch": 7.791190864600326,
      "grad_norm": 0.05198020488023758,
      "learning_rate": 7.07098332705427e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 103077568,
      "step": 47760
    },
    {
      "epoch": 7.7920065252854815,
      "grad_norm": 0.0652122050523758,
      "learning_rate": 7.066023475976438e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 103088672,
      "step": 47765
    },
    {
      "epoch": 7.792822185970636,
      "grad_norm": 5.139396667480469,
      "learning_rate": 7.061065078717738e-06,
      "loss": 0.119,
      "num_input_tokens_seen": 103098496,
      "step": 47770
    },
    {
      "epoch": 7.793637846655791,
      "grad_norm": 3.967467784881592,
      "learning_rate": 7.056108135680123e-06,
      "loss": 0.2102,
      "num_input_tokens_seen": 103109856,
      "step": 47775
    },
    {
      "epoch": 7.794453507340946,
      "grad_norm": 0.09969165921211243,
      "learning_rate": 7.05115264726543e-06,
      "loss": 0.1134,
      "num_input_tokens_seen": 103120032,
      "step": 47780
    },
    {
      "epoch": 7.795269168026101,
      "grad_norm": 4.543612003326416,
      "learning_rate": 7.046198613875374e-06,
      "loss": 0.1984,
      "num_input_tokens_seen": 103131072,
      "step": 47785
    },
    {
      "epoch": 7.7960848287112565,
      "grad_norm": 3.157122850418091,
      "learning_rate": 7.0412460359115555e-06,
      "loss": 0.241,
      "num_input_tokens_seen": 103142240,
      "step": 47790
    },
    {
      "epoch": 7.796900489396411,
      "grad_norm": 5.2432541847229,
      "learning_rate": 7.0362949137754565e-06,
      "loss": 0.1828,
      "num_input_tokens_seen": 103152960,
      "step": 47795
    },
    {
      "epoch": 7.797716150081566,
      "grad_norm": 0.23773658275604248,
      "learning_rate": 7.031345247868437e-06,
      "loss": 0.1168,
      "num_input_tokens_seen": 103164352,
      "step": 47800
    },
    {
      "epoch": 7.798531810766721,
      "grad_norm": 27.523258209228516,
      "learning_rate": 7.026397038591745e-06,
      "loss": 0.1761,
      "num_input_tokens_seen": 103175680,
      "step": 47805
    },
    {
      "epoch": 7.799347471451876,
      "grad_norm": 0.08605588227510452,
      "learning_rate": 7.021450286346503e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 103185952,
      "step": 47810
    },
    {
      "epoch": 7.800163132137031,
      "grad_norm": 3.771977663040161,
      "learning_rate": 7.016504991533726e-06,
      "loss": 0.32,
      "num_input_tokens_seen": 103195840,
      "step": 47815
    },
    {
      "epoch": 7.800978792822186,
      "grad_norm": 0.09736517071723938,
      "learning_rate": 7.011561154554303e-06,
      "loss": 0.0859,
      "num_input_tokens_seen": 103207104,
      "step": 47820
    },
    {
      "epoch": 7.801794453507341,
      "grad_norm": 0.1261928230524063,
      "learning_rate": 7.006618775809001e-06,
      "loss": 0.0963,
      "num_input_tokens_seen": 103218624,
      "step": 47825
    },
    {
      "epoch": 7.802610114192496,
      "grad_norm": 0.15593859553337097,
      "learning_rate": 7.001677855698482e-06,
      "loss": 0.0132,
      "num_input_tokens_seen": 103230592,
      "step": 47830
    },
    {
      "epoch": 7.803425774877651,
      "grad_norm": 0.16541704535484314,
      "learning_rate": 6.996738394623279e-06,
      "loss": 0.2095,
      "num_input_tokens_seen": 103241984,
      "step": 47835
    },
    {
      "epoch": 7.804241435562806,
      "grad_norm": 6.937445163726807,
      "learning_rate": 6.991800392983799e-06,
      "loss": 0.2821,
      "num_input_tokens_seen": 103252800,
      "step": 47840
    },
    {
      "epoch": 7.80505709624796,
      "grad_norm": 0.06043032184243202,
      "learning_rate": 6.9868638511803615e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 103265440,
      "step": 47845
    },
    {
      "epoch": 7.805872756933116,
      "grad_norm": 0.2144368439912796,
      "learning_rate": 6.9819287696131355e-06,
      "loss": 0.0094,
      "num_input_tokens_seen": 103274976,
      "step": 47850
    },
    {
      "epoch": 7.806688417618271,
      "grad_norm": 0.1217925027012825,
      "learning_rate": 6.9769951486821885e-06,
      "loss": 0.0168,
      "num_input_tokens_seen": 103285344,
      "step": 47855
    },
    {
      "epoch": 7.807504078303426,
      "grad_norm": 0.174509659409523,
      "learning_rate": 6.972062988787462e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 103296576,
      "step": 47860
    },
    {
      "epoch": 7.808319738988581,
      "grad_norm": 0.14375446736812592,
      "learning_rate": 6.9671322903287765e-06,
      "loss": 0.1212,
      "num_input_tokens_seen": 103306880,
      "step": 47865
    },
    {
      "epoch": 7.809135399673735,
      "grad_norm": 0.13414366543293,
      "learning_rate": 6.962203053705851e-06,
      "loss": 0.061,
      "num_input_tokens_seen": 103317760,
      "step": 47870
    },
    {
      "epoch": 7.809951060358891,
      "grad_norm": 5.264692306518555,
      "learning_rate": 6.957275279318268e-06,
      "loss": 0.1892,
      "num_input_tokens_seen": 103328096,
      "step": 47875
    },
    {
      "epoch": 7.810766721044046,
      "grad_norm": 4.860972881317139,
      "learning_rate": 6.9523489675655e-06,
      "loss": 0.1944,
      "num_input_tokens_seen": 103338560,
      "step": 47880
    },
    {
      "epoch": 7.811582381729201,
      "grad_norm": 5.393228054046631,
      "learning_rate": 6.9474241188468985e-06,
      "loss": 0.1765,
      "num_input_tokens_seen": 103349824,
      "step": 47885
    },
    {
      "epoch": 7.8123980424143555,
      "grad_norm": 0.14536328613758087,
      "learning_rate": 6.942500733561694e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 103361056,
      "step": 47890
    },
    {
      "epoch": 7.81321370309951,
      "grad_norm": 0.07104796916246414,
      "learning_rate": 6.937578812109005e-06,
      "loss": 0.0186,
      "num_input_tokens_seen": 103371520,
      "step": 47895
    },
    {
      "epoch": 7.814029363784666,
      "grad_norm": 1.0618921518325806,
      "learning_rate": 6.932658354887825e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 103382784,
      "step": 47900
    },
    {
      "epoch": 7.814845024469821,
      "grad_norm": 0.1532667577266693,
      "learning_rate": 6.927739362297028e-06,
      "loss": 0.2538,
      "num_input_tokens_seen": 103393376,
      "step": 47905
    },
    {
      "epoch": 7.815660685154976,
      "grad_norm": 0.04023406654596329,
      "learning_rate": 6.92282183473538e-06,
      "loss": 0.1004,
      "num_input_tokens_seen": 103404000,
      "step": 47910
    },
    {
      "epoch": 7.8164763458401305,
      "grad_norm": 0.021517453715205193,
      "learning_rate": 6.917905772601516e-06,
      "loss": 0.0859,
      "num_input_tokens_seen": 103414848,
      "step": 47915
    },
    {
      "epoch": 7.817292006525285,
      "grad_norm": 0.06931926310062408,
      "learning_rate": 6.912991176293957e-06,
      "loss": 0.0091,
      "num_input_tokens_seen": 103426144,
      "step": 47920
    },
    {
      "epoch": 7.81810766721044,
      "grad_norm": 0.09444573521614075,
      "learning_rate": 6.908078046211105e-06,
      "loss": 0.082,
      "num_input_tokens_seen": 103435584,
      "step": 47925
    },
    {
      "epoch": 7.818923327895595,
      "grad_norm": 0.04047401621937752,
      "learning_rate": 6.903166382751244e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 103446464,
      "step": 47930
    },
    {
      "epoch": 7.819738988580751,
      "grad_norm": 0.858544647693634,
      "learning_rate": 6.8982561863125405e-06,
      "loss": 0.2054,
      "num_input_tokens_seen": 103457248,
      "step": 47935
    },
    {
      "epoch": 7.8205546492659055,
      "grad_norm": 1.1933023929595947,
      "learning_rate": 6.893347457293036e-06,
      "loss": 0.1077,
      "num_input_tokens_seen": 103468128,
      "step": 47940
    },
    {
      "epoch": 7.82137030995106,
      "grad_norm": 0.04212098568677902,
      "learning_rate": 6.888440196090659e-06,
      "loss": 0.1546,
      "num_input_tokens_seen": 103478656,
      "step": 47945
    },
    {
      "epoch": 7.822185970636215,
      "grad_norm": 0.17217549681663513,
      "learning_rate": 6.8835344031032175e-06,
      "loss": 0.0422,
      "num_input_tokens_seen": 103489472,
      "step": 47950
    },
    {
      "epoch": 7.82300163132137,
      "grad_norm": 2.7660112380981445,
      "learning_rate": 6.878630078728399e-06,
      "loss": 0.233,
      "num_input_tokens_seen": 103500384,
      "step": 47955
    },
    {
      "epoch": 7.823817292006526,
      "grad_norm": 3.457015037536621,
      "learning_rate": 6.873727223363766e-06,
      "loss": 0.2738,
      "num_input_tokens_seen": 103510432,
      "step": 47960
    },
    {
      "epoch": 7.8246329526916805,
      "grad_norm": 0.10388149321079254,
      "learning_rate": 6.868825837406784e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 103520544,
      "step": 47965
    },
    {
      "epoch": 7.825448613376835,
      "grad_norm": 2.991614818572998,
      "learning_rate": 6.8639259212547764e-06,
      "loss": 0.1083,
      "num_input_tokens_seen": 103531264,
      "step": 47970
    },
    {
      "epoch": 7.82626427406199,
      "grad_norm": 0.08335016667842865,
      "learning_rate": 6.859027475304955e-06,
      "loss": 0.0059,
      "num_input_tokens_seen": 103542176,
      "step": 47975
    },
    {
      "epoch": 7.827079934747145,
      "grad_norm": 0.11649288237094879,
      "learning_rate": 6.854130499954411e-06,
      "loss": 0.0086,
      "num_input_tokens_seen": 103552608,
      "step": 47980
    },
    {
      "epoch": 7.827895595432301,
      "grad_norm": 0.1272554099559784,
      "learning_rate": 6.849234995600121e-06,
      "loss": 0.1022,
      "num_input_tokens_seen": 103563552,
      "step": 47985
    },
    {
      "epoch": 7.828711256117455,
      "grad_norm": 0.14843489229679108,
      "learning_rate": 6.84434096263894e-06,
      "loss": 0.1328,
      "num_input_tokens_seen": 103575296,
      "step": 47990
    },
    {
      "epoch": 7.82952691680261,
      "grad_norm": 0.1605493575334549,
      "learning_rate": 6.839448401467599e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 103586304,
      "step": 47995
    },
    {
      "epoch": 7.830342577487765,
      "grad_norm": 3.597090721130371,
      "learning_rate": 6.834557312482717e-06,
      "loss": 0.1724,
      "num_input_tokens_seen": 103596768,
      "step": 48000
    },
    {
      "epoch": 7.83115823817292,
      "grad_norm": 11.16240406036377,
      "learning_rate": 6.8296676960807906e-06,
      "loss": 0.1301,
      "num_input_tokens_seen": 103607808,
      "step": 48005
    },
    {
      "epoch": 7.831973898858075,
      "grad_norm": 0.18940110504627228,
      "learning_rate": 6.824779552658189e-06,
      "loss": 0.09,
      "num_input_tokens_seen": 103618848,
      "step": 48010
    },
    {
      "epoch": 7.8327895595432295,
      "grad_norm": 0.22899290919303894,
      "learning_rate": 6.819892882611184e-06,
      "loss": 0.0242,
      "num_input_tokens_seen": 103629600,
      "step": 48015
    },
    {
      "epoch": 7.833605220228385,
      "grad_norm": 0.11677180975675583,
      "learning_rate": 6.8150076863359054e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 103639744,
      "step": 48020
    },
    {
      "epoch": 7.83442088091354,
      "grad_norm": 0.09852508455514908,
      "learning_rate": 6.810123964228374e-06,
      "loss": 0.1979,
      "num_input_tokens_seen": 103649472,
      "step": 48025
    },
    {
      "epoch": 7.835236541598695,
      "grad_norm": 3.817969560623169,
      "learning_rate": 6.8052417166844905e-06,
      "loss": 0.0794,
      "num_input_tokens_seen": 103659968,
      "step": 48030
    },
    {
      "epoch": 7.83605220228385,
      "grad_norm": 0.9190351366996765,
      "learning_rate": 6.800360944100031e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 103671584,
      "step": 48035
    },
    {
      "epoch": 7.8368678629690045,
      "grad_norm": 3.154757499694824,
      "learning_rate": 6.795481646870658e-06,
      "loss": 0.1681,
      "num_input_tokens_seen": 103681984,
      "step": 48040
    },
    {
      "epoch": 7.83768352365416,
      "grad_norm": 0.13393941521644592,
      "learning_rate": 6.790603825391912e-06,
      "loss": 0.2417,
      "num_input_tokens_seen": 103692544,
      "step": 48045
    },
    {
      "epoch": 7.838499184339315,
      "grad_norm": 0.05127906799316406,
      "learning_rate": 6.785727480059212e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 103703488,
      "step": 48050
    },
    {
      "epoch": 7.83931484502447,
      "grad_norm": 0.05056251958012581,
      "learning_rate": 6.78085261126786e-06,
      "loss": 0.008,
      "num_input_tokens_seen": 103714464,
      "step": 48055
    },
    {
      "epoch": 7.840130505709625,
      "grad_norm": 0.2773749530315399,
      "learning_rate": 6.775979219413042e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 103725184,
      "step": 48060
    },
    {
      "epoch": 7.8409461663947795,
      "grad_norm": 0.4287366569042206,
      "learning_rate": 6.771107304889807e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 103735520,
      "step": 48065
    },
    {
      "epoch": 7.841761827079935,
      "grad_norm": 7.799998760223389,
      "learning_rate": 6.766236868093112e-06,
      "loss": 0.0993,
      "num_input_tokens_seen": 103747072,
      "step": 48070
    },
    {
      "epoch": 7.84257748776509,
      "grad_norm": 0.17426146566867828,
      "learning_rate": 6.761367909417776e-06,
      "loss": 0.101,
      "num_input_tokens_seen": 103757632,
      "step": 48075
    },
    {
      "epoch": 7.843393148450245,
      "grad_norm": 6.004482269287109,
      "learning_rate": 6.756500429258497e-06,
      "loss": 0.1365,
      "num_input_tokens_seen": 103768576,
      "step": 48080
    },
    {
      "epoch": 7.8442088091354,
      "grad_norm": 0.04564683511853218,
      "learning_rate": 6.751634428009862e-06,
      "loss": 0.1335,
      "num_input_tokens_seen": 103779840,
      "step": 48085
    },
    {
      "epoch": 7.8450244698205545,
      "grad_norm": 0.14614471793174744,
      "learning_rate": 6.7467699060663305e-06,
      "loss": 0.1128,
      "num_input_tokens_seen": 103789664,
      "step": 48090
    },
    {
      "epoch": 7.845840130505709,
      "grad_norm": 0.05062619596719742,
      "learning_rate": 6.741906863822248e-06,
      "loss": 0.1937,
      "num_input_tokens_seen": 103800832,
      "step": 48095
    },
    {
      "epoch": 7.846655791190865,
      "grad_norm": 1.9862357378005981,
      "learning_rate": 6.737045301671832e-06,
      "loss": 0.0629,
      "num_input_tokens_seen": 103811616,
      "step": 48100
    },
    {
      "epoch": 7.84747145187602,
      "grad_norm": 0.442433625459671,
      "learning_rate": 6.7321852200091935e-06,
      "loss": 0.3016,
      "num_input_tokens_seen": 103822464,
      "step": 48105
    },
    {
      "epoch": 7.848287112561175,
      "grad_norm": 0.09706331789493561,
      "learning_rate": 6.727326619228308e-06,
      "loss": 0.006,
      "num_input_tokens_seen": 103833984,
      "step": 48110
    },
    {
      "epoch": 7.849102773246329,
      "grad_norm": 0.04700395464897156,
      "learning_rate": 6.722469499723042e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 103845664,
      "step": 48115
    },
    {
      "epoch": 7.849918433931484,
      "grad_norm": 0.26358553767204285,
      "learning_rate": 6.717613861887137e-06,
      "loss": 0.392,
      "num_input_tokens_seen": 103855936,
      "step": 48120
    },
    {
      "epoch": 7.850734094616639,
      "grad_norm": 13.11994743347168,
      "learning_rate": 6.712759706114219e-06,
      "loss": 0.0195,
      "num_input_tokens_seen": 103866304,
      "step": 48125
    },
    {
      "epoch": 7.851549755301795,
      "grad_norm": 0.14919701218605042,
      "learning_rate": 6.707907032797786e-06,
      "loss": 0.0119,
      "num_input_tokens_seen": 103877280,
      "step": 48130
    },
    {
      "epoch": 7.85236541598695,
      "grad_norm": 4.979850769042969,
      "learning_rate": 6.703055842331221e-06,
      "loss": 0.1614,
      "num_input_tokens_seen": 103888192,
      "step": 48135
    },
    {
      "epoch": 7.853181076672104,
      "grad_norm": 0.3740044832229614,
      "learning_rate": 6.698206135107787e-06,
      "loss": 0.0678,
      "num_input_tokens_seen": 103897568,
      "step": 48140
    },
    {
      "epoch": 7.853996737357259,
      "grad_norm": 0.21097080409526825,
      "learning_rate": 6.6933579115206284e-06,
      "loss": 0.0069,
      "num_input_tokens_seen": 103908448,
      "step": 48145
    },
    {
      "epoch": 7.854812398042414,
      "grad_norm": 22.14005470275879,
      "learning_rate": 6.6885111719627635e-06,
      "loss": 0.069,
      "num_input_tokens_seen": 103917856,
      "step": 48150
    },
    {
      "epoch": 7.85562805872757,
      "grad_norm": 0.09356860816478729,
      "learning_rate": 6.683665916827087e-06,
      "loss": 0.119,
      "num_input_tokens_seen": 103929792,
      "step": 48155
    },
    {
      "epoch": 7.856443719412725,
      "grad_norm": 26.533403396606445,
      "learning_rate": 6.678822146506394e-06,
      "loss": 0.1676,
      "num_input_tokens_seen": 103939488,
      "step": 48160
    },
    {
      "epoch": 7.857259380097879,
      "grad_norm": 0.1259954422712326,
      "learning_rate": 6.67397986139334e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 103949600,
      "step": 48165
    },
    {
      "epoch": 7.858075040783034,
      "grad_norm": 0.32674962282180786,
      "learning_rate": 6.669139061880464e-06,
      "loss": 0.2732,
      "num_input_tokens_seen": 103959936,
      "step": 48170
    },
    {
      "epoch": 7.858890701468189,
      "grad_norm": 0.1663704812526703,
      "learning_rate": 6.664299748360184e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 103971136,
      "step": 48175
    },
    {
      "epoch": 7.859706362153344,
      "grad_norm": 0.1198531985282898,
      "learning_rate": 6.659461921224794e-06,
      "loss": 0.1066,
      "num_input_tokens_seen": 103981760,
      "step": 48180
    },
    {
      "epoch": 7.8605220228384995,
      "grad_norm": 0.5928356051445007,
      "learning_rate": 6.654625580866486e-06,
      "loss": 0.0093,
      "num_input_tokens_seen": 103992480,
      "step": 48185
    },
    {
      "epoch": 7.861337683523654,
      "grad_norm": 1.333174228668213,
      "learning_rate": 6.649790727677313e-06,
      "loss": 0.0108,
      "num_input_tokens_seen": 104002720,
      "step": 48190
    },
    {
      "epoch": 7.862153344208809,
      "grad_norm": 3.8798046112060547,
      "learning_rate": 6.644957362049212e-06,
      "loss": 0.1092,
      "num_input_tokens_seen": 104011968,
      "step": 48195
    },
    {
      "epoch": 7.862969004893964,
      "grad_norm": 7.625709533691406,
      "learning_rate": 6.640125484373999e-06,
      "loss": 0.1359,
      "num_input_tokens_seen": 104023328,
      "step": 48200
    },
    {
      "epoch": 7.863784665579119,
      "grad_norm": 3.2316322326660156,
      "learning_rate": 6.635295095043373e-06,
      "loss": 0.1565,
      "num_input_tokens_seen": 104032896,
      "step": 48205
    },
    {
      "epoch": 7.864600326264274,
      "grad_norm": 29.43779182434082,
      "learning_rate": 6.630466194448906e-06,
      "loss": 0.0824,
      "num_input_tokens_seen": 104044128,
      "step": 48210
    },
    {
      "epoch": 7.865415986949429,
      "grad_norm": 0.30524954199790955,
      "learning_rate": 6.625638782982058e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 104055808,
      "step": 48215
    },
    {
      "epoch": 7.866231647634584,
      "grad_norm": 0.0591411255300045,
      "learning_rate": 6.620812861034159e-06,
      "loss": 0.126,
      "num_input_tokens_seen": 104066816,
      "step": 48220
    },
    {
      "epoch": 7.867047308319739,
      "grad_norm": 4.141035079956055,
      "learning_rate": 6.615988428996426e-06,
      "loss": 0.123,
      "num_input_tokens_seen": 104078624,
      "step": 48225
    },
    {
      "epoch": 7.867862969004894,
      "grad_norm": 0.016146790236234665,
      "learning_rate": 6.611165487259946e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 104090592,
      "step": 48230
    },
    {
      "epoch": 7.868678629690049,
      "grad_norm": 0.14641910791397095,
      "learning_rate": 6.6063440362157e-06,
      "loss": 0.2131,
      "num_input_tokens_seen": 104101376,
      "step": 48235
    },
    {
      "epoch": 7.869494290375204,
      "grad_norm": 5.0318450927734375,
      "learning_rate": 6.601524076254534e-06,
      "loss": 0.1209,
      "num_input_tokens_seen": 104111968,
      "step": 48240
    },
    {
      "epoch": 7.870309951060359,
      "grad_norm": 0.09114504605531693,
      "learning_rate": 6.5967056077671785e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 104123616,
      "step": 48245
    },
    {
      "epoch": 7.871125611745514,
      "grad_norm": 0.19192178547382355,
      "learning_rate": 6.591888631144244e-06,
      "loss": 0.0098,
      "num_input_tokens_seen": 104134176,
      "step": 48250
    },
    {
      "epoch": 7.871941272430669,
      "grad_norm": 0.16777630150318146,
      "learning_rate": 6.587073146776221e-06,
      "loss": 0.0134,
      "num_input_tokens_seen": 104144704,
      "step": 48255
    },
    {
      "epoch": 7.872756933115824,
      "grad_norm": 0.10116895288228989,
      "learning_rate": 6.582259155053472e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 104155648,
      "step": 48260
    },
    {
      "epoch": 7.873572593800979,
      "grad_norm": 0.3359253406524658,
      "learning_rate": 6.577446656366248e-06,
      "loss": 0.0092,
      "num_input_tokens_seen": 104166080,
      "step": 48265
    },
    {
      "epoch": 7.874388254486134,
      "grad_norm": 0.10709737986326218,
      "learning_rate": 6.572635651104672e-06,
      "loss": 0.1907,
      "num_input_tokens_seen": 104177152,
      "step": 48270
    },
    {
      "epoch": 7.875203915171289,
      "grad_norm": 0.183837428689003,
      "learning_rate": 6.56782613965875e-06,
      "loss": 0.0903,
      "num_input_tokens_seen": 104187104,
      "step": 48275
    },
    {
      "epoch": 7.876019575856444,
      "grad_norm": 0.24258825182914734,
      "learning_rate": 6.56301812241836e-06,
      "loss": 0.1851,
      "num_input_tokens_seen": 104198176,
      "step": 48280
    },
    {
      "epoch": 7.876835236541599,
      "grad_norm": 3.056492805480957,
      "learning_rate": 6.558211599773273e-06,
      "loss": 0.2398,
      "num_input_tokens_seen": 104208896,
      "step": 48285
    },
    {
      "epoch": 7.877650897226753,
      "grad_norm": 0.07631750404834747,
      "learning_rate": 6.55340657211313e-06,
      "loss": 0.0541,
      "num_input_tokens_seen": 104220512,
      "step": 48290
    },
    {
      "epoch": 7.878466557911908,
      "grad_norm": 3.429745674133301,
      "learning_rate": 6.5486030398274444e-06,
      "loss": 0.2371,
      "num_input_tokens_seen": 104230816,
      "step": 48295
    },
    {
      "epoch": 7.879282218597064,
      "grad_norm": 1.1050548553466797,
      "learning_rate": 6.543801003305619e-06,
      "loss": 0.0085,
      "num_input_tokens_seen": 104242240,
      "step": 48300
    },
    {
      "epoch": 7.880097879282219,
      "grad_norm": 0.06108357384800911,
      "learning_rate": 6.539000462936931e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 104253376,
      "step": 48305
    },
    {
      "epoch": 7.8809135399673735,
      "grad_norm": 0.15179091691970825,
      "learning_rate": 6.534201419110536e-06,
      "loss": 0.0255,
      "num_input_tokens_seen": 104264864,
      "step": 48310
    },
    {
      "epoch": 7.881729200652528,
      "grad_norm": 2.2326877117156982,
      "learning_rate": 6.529403872215467e-06,
      "loss": 0.0546,
      "num_input_tokens_seen": 104275712,
      "step": 48315
    },
    {
      "epoch": 7.882544861337683,
      "grad_norm": 0.08912886679172516,
      "learning_rate": 6.524607822640638e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 104287328,
      "step": 48320
    },
    {
      "epoch": 7.883360522022839,
      "grad_norm": 0.3407483696937561,
      "learning_rate": 6.519813270774835e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 104297344,
      "step": 48325
    },
    {
      "epoch": 7.884176182707994,
      "grad_norm": 0.053344812244176865,
      "learning_rate": 6.515020217006745e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 104308992,
      "step": 48330
    },
    {
      "epoch": 7.8849918433931485,
      "grad_norm": 0.1516156941652298,
      "learning_rate": 6.510228661724907e-06,
      "loss": 0.1034,
      "num_input_tokens_seen": 104319616,
      "step": 48335
    },
    {
      "epoch": 7.885807504078303,
      "grad_norm": 0.0773768424987793,
      "learning_rate": 6.5054386053177515e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 104331520,
      "step": 48340
    },
    {
      "epoch": 7.886623164763458,
      "grad_norm": 0.17040883004665375,
      "learning_rate": 6.500650048173582e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 104341664,
      "step": 48345
    },
    {
      "epoch": 7.887438825448614,
      "grad_norm": 4.182097434997559,
      "learning_rate": 6.495862990680585e-06,
      "loss": 0.1941,
      "num_input_tokens_seen": 104352640,
      "step": 48350
    },
    {
      "epoch": 7.888254486133769,
      "grad_norm": 0.11969134956598282,
      "learning_rate": 6.4910774332268195e-06,
      "loss": 0.1603,
      "num_input_tokens_seen": 104362496,
      "step": 48355
    },
    {
      "epoch": 7.8890701468189235,
      "grad_norm": 0.05024494603276253,
      "learning_rate": 6.486293376200234e-06,
      "loss": 0.1065,
      "num_input_tokens_seen": 104372192,
      "step": 48360
    },
    {
      "epoch": 7.889885807504078,
      "grad_norm": 0.12330567091703415,
      "learning_rate": 6.481510819988645e-06,
      "loss": 0.0069,
      "num_input_tokens_seen": 104383168,
      "step": 48365
    },
    {
      "epoch": 7.890701468189233,
      "grad_norm": 0.037438564002513885,
      "learning_rate": 6.47672976497975e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 104393632,
      "step": 48370
    },
    {
      "epoch": 7.891517128874388,
      "grad_norm": 0.08976734429597855,
      "learning_rate": 6.471950211561125e-06,
      "loss": 0.2618,
      "num_input_tokens_seen": 104404096,
      "step": 48375
    },
    {
      "epoch": 7.892332789559543,
      "grad_norm": 0.10741182416677475,
      "learning_rate": 6.46717216012023e-06,
      "loss": 0.063,
      "num_input_tokens_seen": 104414592,
      "step": 48380
    },
    {
      "epoch": 7.8931484502446985,
      "grad_norm": 0.9168614745140076,
      "learning_rate": 6.462395611044383e-06,
      "loss": 0.0085,
      "num_input_tokens_seen": 104424736,
      "step": 48385
    },
    {
      "epoch": 7.893964110929853,
      "grad_norm": 0.20880509912967682,
      "learning_rate": 6.457620564720815e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 104434976,
      "step": 48390
    },
    {
      "epoch": 7.894779771615008,
      "grad_norm": 0.07101184874773026,
      "learning_rate": 6.452847021536609e-06,
      "loss": 0.1205,
      "num_input_tokens_seen": 104445920,
      "step": 48395
    },
    {
      "epoch": 7.895595432300163,
      "grad_norm": 0.09273168444633484,
      "learning_rate": 6.44807498187873e-06,
      "loss": 0.244,
      "num_input_tokens_seen": 104455584,
      "step": 48400
    },
    {
      "epoch": 7.896411092985318,
      "grad_norm": 0.051877886056900024,
      "learning_rate": 6.443304446134024e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 104466304,
      "step": 48405
    },
    {
      "epoch": 7.897226753670473,
      "grad_norm": 0.043426841497421265,
      "learning_rate": 6.438535414689215e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 104476096,
      "step": 48410
    },
    {
      "epoch": 7.898042414355628,
      "grad_norm": 0.26901328563690186,
      "learning_rate": 6.4337678879309055e-06,
      "loss": 0.0567,
      "num_input_tokens_seen": 104486656,
      "step": 48415
    },
    {
      "epoch": 7.898858075040783,
      "grad_norm": 4.49504280090332,
      "learning_rate": 6.4290018662455764e-06,
      "loss": 0.1719,
      "num_input_tokens_seen": 104497600,
      "step": 48420
    },
    {
      "epoch": 7.899673735725938,
      "grad_norm": 0.18512462079524994,
      "learning_rate": 6.424237350019582e-06,
      "loss": 0.0659,
      "num_input_tokens_seen": 104508384,
      "step": 48425
    },
    {
      "epoch": 7.900489396411093,
      "grad_norm": 0.2118251919746399,
      "learning_rate": 6.419474339639161e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 104519136,
      "step": 48430
    },
    {
      "epoch": 7.901305057096248,
      "grad_norm": 0.22650903463363647,
      "learning_rate": 6.414712835490428e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 104529728,
      "step": 48435
    },
    {
      "epoch": 7.902120717781403,
      "grad_norm": 0.09416618198156357,
      "learning_rate": 6.409952837959374e-06,
      "loss": 0.0758,
      "num_input_tokens_seen": 104541888,
      "step": 48440
    },
    {
      "epoch": 7.902936378466558,
      "grad_norm": 0.12530404329299927,
      "learning_rate": 6.405194347431864e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 104552736,
      "step": 48445
    },
    {
      "epoch": 7.903752039151713,
      "grad_norm": 3.2568163871765137,
      "learning_rate": 6.400437364293655e-06,
      "loss": 0.1064,
      "num_input_tokens_seen": 104562464,
      "step": 48450
    },
    {
      "epoch": 7.904567699836868,
      "grad_norm": 0.12182630598545074,
      "learning_rate": 6.395681888930361e-06,
      "loss": 0.0293,
      "num_input_tokens_seen": 104573600,
      "step": 48455
    },
    {
      "epoch": 7.9053833605220225,
      "grad_norm": 0.12934081256389618,
      "learning_rate": 6.390927921727494e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 104584800,
      "step": 48460
    },
    {
      "epoch": 7.906199021207177,
      "grad_norm": 0.8263911604881287,
      "learning_rate": 6.386175463070429e-06,
      "loss": 0.1427,
      "num_input_tokens_seen": 104596192,
      "step": 48465
    },
    {
      "epoch": 7.907014681892333,
      "grad_norm": 2.735842227935791,
      "learning_rate": 6.3814245133444196e-06,
      "loss": 0.1951,
      "num_input_tokens_seen": 104606976,
      "step": 48470
    },
    {
      "epoch": 7.907830342577488,
      "grad_norm": 6.71162223815918,
      "learning_rate": 6.376675072934618e-06,
      "loss": 0.153,
      "num_input_tokens_seen": 104618304,
      "step": 48475
    },
    {
      "epoch": 7.908646003262643,
      "grad_norm": 0.39882490038871765,
      "learning_rate": 6.371927142226028e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 104627840,
      "step": 48480
    },
    {
      "epoch": 7.9094616639477975,
      "grad_norm": 0.19895535707473755,
      "learning_rate": 6.367180721603541e-06,
      "loss": 0.0083,
      "num_input_tokens_seen": 104639296,
      "step": 48485
    },
    {
      "epoch": 7.910277324632952,
      "grad_norm": 24.118324279785156,
      "learning_rate": 6.3624358114519275e-06,
      "loss": 0.1069,
      "num_input_tokens_seen": 104648992,
      "step": 48490
    },
    {
      "epoch": 7.911092985318108,
      "grad_norm": 0.043225184082984924,
      "learning_rate": 6.3576924121558246e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 104659680,
      "step": 48495
    },
    {
      "epoch": 7.911908646003263,
      "grad_norm": 0.02157299593091011,
      "learning_rate": 6.352950524099774e-06,
      "loss": 0.1461,
      "num_input_tokens_seen": 104670688,
      "step": 48500
    },
    {
      "epoch": 7.912724306688418,
      "grad_norm": 0.05414982885122299,
      "learning_rate": 6.348210147668165e-06,
      "loss": 0.1177,
      "num_input_tokens_seen": 104681600,
      "step": 48505
    },
    {
      "epoch": 7.9135399673735725,
      "grad_norm": 0.39657047390937805,
      "learning_rate": 6.343471283245283e-06,
      "loss": 0.0792,
      "num_input_tokens_seen": 104692672,
      "step": 48510
    },
    {
      "epoch": 7.914355628058727,
      "grad_norm": 0.06217885762453079,
      "learning_rate": 6.33873393121528e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 104703296,
      "step": 48515
    },
    {
      "epoch": 7.915171288743883,
      "grad_norm": 0.033534884452819824,
      "learning_rate": 6.33399809196219e-06,
      "loss": 0.0851,
      "num_input_tokens_seen": 104715648,
      "step": 48520
    },
    {
      "epoch": 7.915986949429038,
      "grad_norm": 4.821717262268066,
      "learning_rate": 6.329263765869925e-06,
      "loss": 0.1039,
      "num_input_tokens_seen": 104724864,
      "step": 48525
    },
    {
      "epoch": 7.916802610114193,
      "grad_norm": 3.659557819366455,
      "learning_rate": 6.324530953322275e-06,
      "loss": 0.131,
      "num_input_tokens_seen": 104734592,
      "step": 48530
    },
    {
      "epoch": 7.917618270799347,
      "grad_norm": 6.6302056312561035,
      "learning_rate": 6.319799654702904e-06,
      "loss": 0.0903,
      "num_input_tokens_seen": 104744992,
      "step": 48535
    },
    {
      "epoch": 7.918433931484502,
      "grad_norm": 0.024048956111073494,
      "learning_rate": 6.315069870395354e-06,
      "loss": 0.1195,
      "num_input_tokens_seen": 104756512,
      "step": 48540
    },
    {
      "epoch": 7.919249592169657,
      "grad_norm": 6.27457332611084,
      "learning_rate": 6.310341600783049e-06,
      "loss": 0.2202,
      "num_input_tokens_seen": 104767840,
      "step": 48545
    },
    {
      "epoch": 7.920065252854813,
      "grad_norm": 0.27394771575927734,
      "learning_rate": 6.305614846249283e-06,
      "loss": 0.122,
      "num_input_tokens_seen": 104778944,
      "step": 48550
    },
    {
      "epoch": 7.920880913539968,
      "grad_norm": 0.05479596182703972,
      "learning_rate": 6.300889607177229e-06,
      "loss": 0.0889,
      "num_input_tokens_seen": 104789088,
      "step": 48555
    },
    {
      "epoch": 7.921696574225122,
      "grad_norm": 0.11633160710334778,
      "learning_rate": 6.296165883949947e-06,
      "loss": 0.1402,
      "num_input_tokens_seen": 104800032,
      "step": 48560
    },
    {
      "epoch": 7.922512234910277,
      "grad_norm": 1.6217732429504395,
      "learning_rate": 6.291443676950357e-06,
      "loss": 0.0125,
      "num_input_tokens_seen": 104810720,
      "step": 48565
    },
    {
      "epoch": 7.923327895595432,
      "grad_norm": 0.1704300045967102,
      "learning_rate": 6.286722986561272e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 104820896,
      "step": 48570
    },
    {
      "epoch": 7.924143556280587,
      "grad_norm": 0.15686701238155365,
      "learning_rate": 6.282003813165368e-06,
      "loss": 0.1773,
      "num_input_tokens_seen": 104831808,
      "step": 48575
    },
    {
      "epoch": 7.924959216965743,
      "grad_norm": 0.061979733407497406,
      "learning_rate": 6.2772861571452125e-06,
      "loss": 0.0268,
      "num_input_tokens_seen": 104841920,
      "step": 48580
    },
    {
      "epoch": 7.925774877650897,
      "grad_norm": 0.050359781831502914,
      "learning_rate": 6.272570018883236e-06,
      "loss": 0.0933,
      "num_input_tokens_seen": 104853408,
      "step": 48585
    },
    {
      "epoch": 7.926590538336052,
      "grad_norm": 0.05893663316965103,
      "learning_rate": 6.26785539876176e-06,
      "loss": 0.1893,
      "num_input_tokens_seen": 104862880,
      "step": 48590
    },
    {
      "epoch": 7.927406199021207,
      "grad_norm": 0.32345038652420044,
      "learning_rate": 6.2631422971629605e-06,
      "loss": 0.2086,
      "num_input_tokens_seen": 104873952,
      "step": 48595
    },
    {
      "epoch": 7.928221859706362,
      "grad_norm": 0.36849120259284973,
      "learning_rate": 6.2584307144689245e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 104884128,
      "step": 48600
    },
    {
      "epoch": 7.9290375203915175,
      "grad_norm": 0.05867064371705055,
      "learning_rate": 6.25372065106159e-06,
      "loss": 0.0709,
      "num_input_tokens_seen": 104895296,
      "step": 48605
    },
    {
      "epoch": 7.929853181076672,
      "grad_norm": 0.05543160438537598,
      "learning_rate": 6.249012107322774e-06,
      "loss": 0.0085,
      "num_input_tokens_seen": 104905760,
      "step": 48610
    },
    {
      "epoch": 7.930668841761827,
      "grad_norm": 0.09976672381162643,
      "learning_rate": 6.244305083634181e-06,
      "loss": 0.0156,
      "num_input_tokens_seen": 104916864,
      "step": 48615
    },
    {
      "epoch": 7.931484502446982,
      "grad_norm": 0.11526015400886536,
      "learning_rate": 6.239599580377381e-06,
      "loss": 0.1241,
      "num_input_tokens_seen": 104927104,
      "step": 48620
    },
    {
      "epoch": 7.932300163132137,
      "grad_norm": 17.803361892700195,
      "learning_rate": 6.234895597933832e-06,
      "loss": 0.0556,
      "num_input_tokens_seen": 104937632,
      "step": 48625
    },
    {
      "epoch": 7.933115823817292,
      "grad_norm": 3.604686737060547,
      "learning_rate": 6.2301931366848555e-06,
      "loss": 0.297,
      "num_input_tokens_seen": 104948448,
      "step": 48630
    },
    {
      "epoch": 7.933931484502447,
      "grad_norm": 0.1886691451072693,
      "learning_rate": 6.225492197011654e-06,
      "loss": 0.0221,
      "num_input_tokens_seen": 104958688,
      "step": 48635
    },
    {
      "epoch": 7.934747145187602,
      "grad_norm": 0.12465963512659073,
      "learning_rate": 6.220792779295326e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 104969408,
      "step": 48640
    },
    {
      "epoch": 7.935562805872757,
      "grad_norm": 0.08143021911382675,
      "learning_rate": 6.216094883916815e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 104979968,
      "step": 48645
    },
    {
      "epoch": 7.936378466557912,
      "grad_norm": 0.1900239884853363,
      "learning_rate": 6.211398511256966e-06,
      "loss": 0.1017,
      "num_input_tokens_seen": 104990688,
      "step": 48650
    },
    {
      "epoch": 7.937194127243067,
      "grad_norm": 4.787083625793457,
      "learning_rate": 6.206703661696484e-06,
      "loss": 0.0626,
      "num_input_tokens_seen": 105002048,
      "step": 48655
    },
    {
      "epoch": 7.938009787928221,
      "grad_norm": 0.132158562541008,
      "learning_rate": 6.20201033561596e-06,
      "loss": 0.0542,
      "num_input_tokens_seen": 105013056,
      "step": 48660
    },
    {
      "epoch": 7.938825448613377,
      "grad_norm": 3.366595983505249,
      "learning_rate": 6.197318533395858e-06,
      "loss": 0.2686,
      "num_input_tokens_seen": 105021856,
      "step": 48665
    },
    {
      "epoch": 7.939641109298532,
      "grad_norm": 2.141418695449829,
      "learning_rate": 6.192628255416519e-06,
      "loss": 0.2362,
      "num_input_tokens_seen": 105031936,
      "step": 48670
    },
    {
      "epoch": 7.940456769983687,
      "grad_norm": 5.5656023025512695,
      "learning_rate": 6.18793950205816e-06,
      "loss": 0.3258,
      "num_input_tokens_seen": 105044000,
      "step": 48675
    },
    {
      "epoch": 7.941272430668842,
      "grad_norm": 3.6275901794433594,
      "learning_rate": 6.183252273700879e-06,
      "loss": 0.1159,
      "num_input_tokens_seen": 105054272,
      "step": 48680
    },
    {
      "epoch": 7.942088091353996,
      "grad_norm": 0.09550347924232483,
      "learning_rate": 6.178566570724642e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 105065984,
      "step": 48685
    },
    {
      "epoch": 7.942903752039152,
      "grad_norm": 4.608447551727295,
      "learning_rate": 6.1738823935092975e-06,
      "loss": 0.1095,
      "num_input_tokens_seen": 105077440,
      "step": 48690
    },
    {
      "epoch": 7.943719412724307,
      "grad_norm": 1.4750423431396484,
      "learning_rate": 6.16919974243457e-06,
      "loss": 0.0965,
      "num_input_tokens_seen": 105087104,
      "step": 48695
    },
    {
      "epoch": 7.944535073409462,
      "grad_norm": 0.16015009582042694,
      "learning_rate": 6.164518617880058e-06,
      "loss": 0.0268,
      "num_input_tokens_seen": 105097824,
      "step": 48700
    },
    {
      "epoch": 7.945350734094617,
      "grad_norm": 0.10820517688989639,
      "learning_rate": 6.159839020225231e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 105108064,
      "step": 48705
    },
    {
      "epoch": 7.946166394779771,
      "grad_norm": 0.08304480463266373,
      "learning_rate": 6.155160949849453e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 105118784,
      "step": 48710
    },
    {
      "epoch": 7.946982055464927,
      "grad_norm": 0.127269446849823,
      "learning_rate": 6.150484407131945e-06,
      "loss": 0.1073,
      "num_input_tokens_seen": 105128608,
      "step": 48715
    },
    {
      "epoch": 7.947797716150082,
      "grad_norm": 0.13916225731372833,
      "learning_rate": 6.145809392451815e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 105138720,
      "step": 48720
    },
    {
      "epoch": 7.948613376835237,
      "grad_norm": 0.12183847278356552,
      "learning_rate": 6.141135906188039e-06,
      "loss": 0.0196,
      "num_input_tokens_seen": 105150368,
      "step": 48725
    },
    {
      "epoch": 7.9494290375203915,
      "grad_norm": 0.19774994254112244,
      "learning_rate": 6.136463948719475e-06,
      "loss": 0.0217,
      "num_input_tokens_seen": 105160544,
      "step": 48730
    },
    {
      "epoch": 7.950244698205546,
      "grad_norm": 0.22997406125068665,
      "learning_rate": 6.131793520424859e-06,
      "loss": 0.1491,
      "num_input_tokens_seen": 105170784,
      "step": 48735
    },
    {
      "epoch": 7.951060358890701,
      "grad_norm": 0.17153584957122803,
      "learning_rate": 6.1271246216827945e-06,
      "loss": 0.1337,
      "num_input_tokens_seen": 105180960,
      "step": 48740
    },
    {
      "epoch": 7.951876019575856,
      "grad_norm": 0.09357471764087677,
      "learning_rate": 6.122457252871769e-06,
      "loss": 0.094,
      "num_input_tokens_seen": 105191200,
      "step": 48745
    },
    {
      "epoch": 7.952691680261012,
      "grad_norm": 6.854122161865234,
      "learning_rate": 6.117791414370141e-06,
      "loss": 0.0299,
      "num_input_tokens_seen": 105202176,
      "step": 48750
    },
    {
      "epoch": 7.9535073409461665,
      "grad_norm": 1.0626060962677002,
      "learning_rate": 6.113127106556149e-06,
      "loss": 0.2266,
      "num_input_tokens_seen": 105212352,
      "step": 48755
    },
    {
      "epoch": 7.954323001631321,
      "grad_norm": 0.368046373128891,
      "learning_rate": 6.108464329807903e-06,
      "loss": 0.1021,
      "num_input_tokens_seen": 105222464,
      "step": 48760
    },
    {
      "epoch": 7.955138662316476,
      "grad_norm": 0.19278091192245483,
      "learning_rate": 6.1038030845033956e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 105233984,
      "step": 48765
    },
    {
      "epoch": 7.955954323001631,
      "grad_norm": 0.07780694961547852,
      "learning_rate": 6.0991433710204885e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 105244768,
      "step": 48770
    },
    {
      "epoch": 7.956769983686787,
      "grad_norm": 0.050757069140672684,
      "learning_rate": 6.0944851897369206e-06,
      "loss": 0.2335,
      "num_input_tokens_seen": 105255680,
      "step": 48775
    },
    {
      "epoch": 7.9575856443719415,
      "grad_norm": 0.022331800311803818,
      "learning_rate": 6.0898285410303015e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 105268608,
      "step": 48780
    },
    {
      "epoch": 7.958401305057096,
      "grad_norm": 3.4827897548675537,
      "learning_rate": 6.085173425278137e-06,
      "loss": 0.1042,
      "num_input_tokens_seen": 105278464,
      "step": 48785
    },
    {
      "epoch": 7.959216965742251,
      "grad_norm": 0.14530052244663239,
      "learning_rate": 6.080519842857787e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 105289856,
      "step": 48790
    },
    {
      "epoch": 7.960032626427406,
      "grad_norm": 0.05929364264011383,
      "learning_rate": 6.075867794146497e-06,
      "loss": 0.0576,
      "num_input_tokens_seen": 105300160,
      "step": 48795
    },
    {
      "epoch": 7.960848287112562,
      "grad_norm": 0.08685368299484253,
      "learning_rate": 6.071217279521382e-06,
      "loss": 0.0626,
      "num_input_tokens_seen": 105312352,
      "step": 48800
    },
    {
      "epoch": 7.9616639477977165,
      "grad_norm": 0.15442676842212677,
      "learning_rate": 6.06656829935944e-06,
      "loss": 0.0215,
      "num_input_tokens_seen": 105324544,
      "step": 48805
    },
    {
      "epoch": 7.962479608482871,
      "grad_norm": 0.07632876187562943,
      "learning_rate": 6.061920854037531e-06,
      "loss": 0.0954,
      "num_input_tokens_seen": 105335072,
      "step": 48810
    },
    {
      "epoch": 7.963295269168026,
      "grad_norm": 0.16926124691963196,
      "learning_rate": 6.0572749439324146e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 105344960,
      "step": 48815
    },
    {
      "epoch": 7.964110929853181,
      "grad_norm": 0.22158217430114746,
      "learning_rate": 6.052630569420706e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 105355264,
      "step": 48820
    },
    {
      "epoch": 7.964926590538336,
      "grad_norm": 0.06808106601238251,
      "learning_rate": 6.047987730878904e-06,
      "loss": 0.0749,
      "num_input_tokens_seen": 105366624,
      "step": 48825
    },
    {
      "epoch": 7.9657422512234906,
      "grad_norm": 0.026783689856529236,
      "learning_rate": 6.043346428683375e-06,
      "loss": 0.1196,
      "num_input_tokens_seen": 105377888,
      "step": 48830
    },
    {
      "epoch": 7.966557911908646,
      "grad_norm": 0.04408380389213562,
      "learning_rate": 6.0387066632103695e-06,
      "loss": 0.0098,
      "num_input_tokens_seen": 105388192,
      "step": 48835
    },
    {
      "epoch": 7.967373572593801,
      "grad_norm": 8.979509353637695,
      "learning_rate": 6.034068434836013e-06,
      "loss": 0.3422,
      "num_input_tokens_seen": 105400256,
      "step": 48840
    },
    {
      "epoch": 7.968189233278956,
      "grad_norm": 1.1973680257797241,
      "learning_rate": 6.029431743936298e-06,
      "loss": 0.1211,
      "num_input_tokens_seen": 105411328,
      "step": 48845
    },
    {
      "epoch": 7.969004893964111,
      "grad_norm": 2.606157064437866,
      "learning_rate": 6.024796590887105e-06,
      "loss": 0.063,
      "num_input_tokens_seen": 105421408,
      "step": 48850
    },
    {
      "epoch": 7.9698205546492655,
      "grad_norm": 0.08284850418567657,
      "learning_rate": 6.020162976064178e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 105431872,
      "step": 48855
    },
    {
      "epoch": 7.970636215334421,
      "grad_norm": 0.16866230964660645,
      "learning_rate": 6.0155308998431415e-06,
      "loss": 0.1046,
      "num_input_tokens_seen": 105443776,
      "step": 48860
    },
    {
      "epoch": 7.971451876019576,
      "grad_norm": 0.1879003942012787,
      "learning_rate": 6.0109003625994975e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 105453088,
      "step": 48865
    },
    {
      "epoch": 7.972267536704731,
      "grad_norm": 0.08856741338968277,
      "learning_rate": 6.006271364708621e-06,
      "loss": 0.0213,
      "num_input_tokens_seen": 105463744,
      "step": 48870
    },
    {
      "epoch": 7.973083197389886,
      "grad_norm": 0.10097008943557739,
      "learning_rate": 6.0016439065457595e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 105474752,
      "step": 48875
    },
    {
      "epoch": 7.9738988580750405,
      "grad_norm": 0.07519751787185669,
      "learning_rate": 5.997017988486039e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 105486016,
      "step": 48880
    },
    {
      "epoch": 7.974714518760196,
      "grad_norm": 0.04896562919020653,
      "learning_rate": 5.992393610904465e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 105496800,
      "step": 48885
    },
    {
      "epoch": 7.975530179445351,
      "grad_norm": 0.03369162604212761,
      "learning_rate": 5.987770774175905e-06,
      "loss": 0.0125,
      "num_input_tokens_seen": 105506528,
      "step": 48890
    },
    {
      "epoch": 7.976345840130506,
      "grad_norm": 3.082150459289551,
      "learning_rate": 5.983149478675113e-06,
      "loss": 0.0669,
      "num_input_tokens_seen": 105517312,
      "step": 48895
    },
    {
      "epoch": 7.977161500815661,
      "grad_norm": 0.021023431792855263,
      "learning_rate": 5.978529724776713e-06,
      "loss": 0.205,
      "num_input_tokens_seen": 105526976,
      "step": 48900
    },
    {
      "epoch": 7.9779771615008155,
      "grad_norm": 0.16616760194301605,
      "learning_rate": 5.97391151285521e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 105537408,
      "step": 48905
    },
    {
      "epoch": 7.97879282218597,
      "grad_norm": 0.19769129157066345,
      "learning_rate": 5.969294843284978e-06,
      "loss": 0.0757,
      "num_input_tokens_seen": 105547712,
      "step": 48910
    },
    {
      "epoch": 7.979608482871125,
      "grad_norm": 0.08259648829698563,
      "learning_rate": 5.964679716440258e-06,
      "loss": 0.1356,
      "num_input_tokens_seen": 105557760,
      "step": 48915
    },
    {
      "epoch": 7.980424143556281,
      "grad_norm": 5.66948938369751,
      "learning_rate": 5.9600661326951916e-06,
      "loss": 0.2767,
      "num_input_tokens_seen": 105568800,
      "step": 48920
    },
    {
      "epoch": 7.981239804241436,
      "grad_norm": 0.044336091727018356,
      "learning_rate": 5.955454092423773e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 105578944,
      "step": 48925
    },
    {
      "epoch": 7.9820554649265905,
      "grad_norm": 6.642369747161865,
      "learning_rate": 5.950843595999877e-06,
      "loss": 0.0746,
      "num_input_tokens_seen": 105589664,
      "step": 48930
    },
    {
      "epoch": 7.982871125611745,
      "grad_norm": 0.09263117611408234,
      "learning_rate": 5.946234643797252e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 105601056,
      "step": 48935
    },
    {
      "epoch": 7.9836867862969,
      "grad_norm": 0.8972885012626648,
      "learning_rate": 5.941627236189526e-06,
      "loss": 0.0703,
      "num_input_tokens_seen": 105611808,
      "step": 48940
    },
    {
      "epoch": 7.984502446982056,
      "grad_norm": 0.0643182098865509,
      "learning_rate": 5.9370213735501974e-06,
      "loss": 0.1625,
      "num_input_tokens_seen": 105622944,
      "step": 48945
    },
    {
      "epoch": 7.985318107667211,
      "grad_norm": 0.13953441381454468,
      "learning_rate": 5.9324170562526345e-06,
      "loss": 0.1258,
      "num_input_tokens_seen": 105633920,
      "step": 48950
    },
    {
      "epoch": 7.986133768352365,
      "grad_norm": 0.21682924032211304,
      "learning_rate": 5.927814284670097e-06,
      "loss": 0.1442,
      "num_input_tokens_seen": 105645280,
      "step": 48955
    },
    {
      "epoch": 7.98694942903752,
      "grad_norm": 0.08131765574216843,
      "learning_rate": 5.923213059175709e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 105656032,
      "step": 48960
    },
    {
      "epoch": 7.987765089722675,
      "grad_norm": 0.0924699455499649,
      "learning_rate": 5.918613380142463e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 105666208,
      "step": 48965
    },
    {
      "epoch": 7.988580750407831,
      "grad_norm": 14.537064552307129,
      "learning_rate": 5.914015247943233e-06,
      "loss": 0.3662,
      "num_input_tokens_seen": 105677600,
      "step": 48970
    },
    {
      "epoch": 7.989396411092986,
      "grad_norm": 0.18694168329238892,
      "learning_rate": 5.909418662950769e-06,
      "loss": 0.0375,
      "num_input_tokens_seen": 105688256,
      "step": 48975
    },
    {
      "epoch": 7.99021207177814,
      "grad_norm": 0.08439765870571136,
      "learning_rate": 5.904823625537695e-06,
      "loss": 0.0259,
      "num_input_tokens_seen": 105698560,
      "step": 48980
    },
    {
      "epoch": 7.991027732463295,
      "grad_norm": 0.05119810253381729,
      "learning_rate": 5.900230136076504e-06,
      "loss": 0.112,
      "num_input_tokens_seen": 105709792,
      "step": 48985
    },
    {
      "epoch": 7.99184339314845,
      "grad_norm": 0.06547822058200836,
      "learning_rate": 5.895638194939568e-06,
      "loss": 0.1206,
      "num_input_tokens_seen": 105720608,
      "step": 48990
    },
    {
      "epoch": 7.992659053833605,
      "grad_norm": 0.11921960115432739,
      "learning_rate": 5.891047802499136e-06,
      "loss": 0.0024,
      "num_input_tokens_seen": 105732192,
      "step": 48995
    },
    {
      "epoch": 7.993474714518761,
      "grad_norm": 0.10682287812232971,
      "learning_rate": 5.886458959127328e-06,
      "loss": 0.254,
      "num_input_tokens_seen": 105742688,
      "step": 49000
    },
    {
      "epoch": 7.994290375203915,
      "grad_norm": 11.609454154968262,
      "learning_rate": 5.881871665196137e-06,
      "loss": 0.1577,
      "num_input_tokens_seen": 105754208,
      "step": 49005
    },
    {
      "epoch": 7.99510603588907,
      "grad_norm": 0.07879647612571716,
      "learning_rate": 5.877285921077433e-06,
      "loss": 0.0279,
      "num_input_tokens_seen": 105765920,
      "step": 49010
    },
    {
      "epoch": 7.995921696574225,
      "grad_norm": 0.12344497442245483,
      "learning_rate": 5.872701727142963e-06,
      "loss": 0.1149,
      "num_input_tokens_seen": 105776992,
      "step": 49015
    },
    {
      "epoch": 7.99673735725938,
      "grad_norm": 0.05152774602174759,
      "learning_rate": 5.868119083764337e-06,
      "loss": 0.1549,
      "num_input_tokens_seen": 105788576,
      "step": 49020
    },
    {
      "epoch": 7.997553017944535,
      "grad_norm": 0.06692085415124893,
      "learning_rate": 5.863537991313047e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 105800384,
      "step": 49025
    },
    {
      "epoch": 7.99836867862969,
      "grad_norm": 11.177392959594727,
      "learning_rate": 5.858958450160473e-06,
      "loss": 0.0913,
      "num_input_tokens_seen": 105810528,
      "step": 49030
    },
    {
      "epoch": 7.999184339314845,
      "grad_norm": 0.01515423133969307,
      "learning_rate": 5.854380460677847e-06,
      "loss": 0.0335,
      "num_input_tokens_seen": 105821440,
      "step": 49035
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.07321982085704803,
      "learning_rate": 5.849804023236285e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 105830544,
      "step": 49040
    },
    {
      "epoch": 8.0,
      "eval_loss": 0.18560057878494263,
      "eval_runtime": 132.9521,
      "eval_samples_per_second": 20.496,
      "eval_steps_per_second": 5.13,
      "num_input_tokens_seen": 105830544,
      "step": 49040
    },
    {
      "epoch": 8.000815660685156,
      "grad_norm": 0.23461724817752838,
      "learning_rate": 5.845229138206776e-06,
      "loss": 0.0165,
      "num_input_tokens_seen": 105843312,
      "step": 49045
    },
    {
      "epoch": 8.00163132137031,
      "grad_norm": 0.2039797604084015,
      "learning_rate": 5.8406558059601825e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 105851248,
      "step": 49050
    },
    {
      "epoch": 8.002446982055465,
      "grad_norm": 16.558528900146484,
      "learning_rate": 5.836084026867244e-06,
      "loss": 0.1154,
      "num_input_tokens_seen": 105862352,
      "step": 49055
    },
    {
      "epoch": 8.00326264274062,
      "grad_norm": 0.07633313536643982,
      "learning_rate": 5.831513801298572e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 105873264,
      "step": 49060
    },
    {
      "epoch": 8.004078303425775,
      "grad_norm": 0.03318179398775101,
      "learning_rate": 5.82694512962465e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 105884272,
      "step": 49065
    },
    {
      "epoch": 8.00489396411093,
      "grad_norm": 4.506711006164551,
      "learning_rate": 5.822378012215837e-06,
      "loss": 0.2231,
      "num_input_tokens_seen": 105894832,
      "step": 49070
    },
    {
      "epoch": 8.005709624796085,
      "grad_norm": 0.055067844688892365,
      "learning_rate": 5.81781244944237e-06,
      "loss": 0.0018,
      "num_input_tokens_seen": 105905520,
      "step": 49075
    },
    {
      "epoch": 8.00652528548124,
      "grad_norm": 0.1632113754749298,
      "learning_rate": 5.813248441674357e-06,
      "loss": 0.09,
      "num_input_tokens_seen": 105917424,
      "step": 49080
    },
    {
      "epoch": 8.007340946166394,
      "grad_norm": 3.296776533126831,
      "learning_rate": 5.8086859892817755e-06,
      "loss": 0.0764,
      "num_input_tokens_seen": 105928560,
      "step": 49085
    },
    {
      "epoch": 8.00815660685155,
      "grad_norm": 0.16078858077526093,
      "learning_rate": 5.804125092634485e-06,
      "loss": 0.0067,
      "num_input_tokens_seen": 105939120,
      "step": 49090
    },
    {
      "epoch": 8.008972267536704,
      "grad_norm": 0.19128017127513885,
      "learning_rate": 5.799565752102207e-06,
      "loss": 0.008,
      "num_input_tokens_seen": 105949424,
      "step": 49095
    },
    {
      "epoch": 8.00978792822186,
      "grad_norm": 0.05958623066544533,
      "learning_rate": 5.795007968054555e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 105959280,
      "step": 49100
    },
    {
      "epoch": 8.010603588907015,
      "grad_norm": 0.0718739703297615,
      "learning_rate": 5.790451740861005e-06,
      "loss": 0.0067,
      "num_input_tokens_seen": 105968272,
      "step": 49105
    },
    {
      "epoch": 8.01141924959217,
      "grad_norm": 2.842352867126465,
      "learning_rate": 5.7858970708909056e-06,
      "loss": 0.0921,
      "num_input_tokens_seen": 105978736,
      "step": 49110
    },
    {
      "epoch": 8.012234910277325,
      "grad_norm": 0.05983918905258179,
      "learning_rate": 5.78134395851348e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 105989872,
      "step": 49115
    },
    {
      "epoch": 8.013050570962479,
      "grad_norm": 0.06373894214630127,
      "learning_rate": 5.7767924040978275e-06,
      "loss": 0.0103,
      "num_input_tokens_seen": 105999792,
      "step": 49120
    },
    {
      "epoch": 8.013866231647635,
      "grad_norm": 0.035480622202157974,
      "learning_rate": 5.772242408012921e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 106010256,
      "step": 49125
    },
    {
      "epoch": 8.01468189233279,
      "grad_norm": 7.287117004394531,
      "learning_rate": 5.767693970627597e-06,
      "loss": 0.2716,
      "num_input_tokens_seen": 106020496,
      "step": 49130
    },
    {
      "epoch": 8.015497553017944,
      "grad_norm": 0.08558224886655807,
      "learning_rate": 5.763147092310592e-06,
      "loss": 0.0279,
      "num_input_tokens_seen": 106031248,
      "step": 49135
    },
    {
      "epoch": 8.0163132137031,
      "grad_norm": 0.05756988376379013,
      "learning_rate": 5.758601773430489e-06,
      "loss": 0.0021,
      "num_input_tokens_seen": 106041232,
      "step": 49140
    },
    {
      "epoch": 8.017128874388254,
      "grad_norm": 0.13379822671413422,
      "learning_rate": 5.7540580143557564e-06,
      "loss": 0.1162,
      "num_input_tokens_seen": 106051696,
      "step": 49145
    },
    {
      "epoch": 8.01794453507341,
      "grad_norm": 0.06322108209133148,
      "learning_rate": 5.7495158154547365e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 106062672,
      "step": 49150
    },
    {
      "epoch": 8.018760195758565,
      "grad_norm": 0.3233621120452881,
      "learning_rate": 5.744975177095638e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 106072784,
      "step": 49155
    },
    {
      "epoch": 8.01957585644372,
      "grad_norm": 0.05716429650783539,
      "learning_rate": 5.740436099646551e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 106083952,
      "step": 49160
    },
    {
      "epoch": 8.020391517128875,
      "grad_norm": 0.3119354844093323,
      "learning_rate": 5.735898583475438e-06,
      "loss": 0.118,
      "num_input_tokens_seen": 106095632,
      "step": 49165
    },
    {
      "epoch": 8.021207177814029,
      "grad_norm": 0.0634540542960167,
      "learning_rate": 5.731362628950129e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 106106160,
      "step": 49170
    },
    {
      "epoch": 8.022022838499185,
      "grad_norm": 0.17123360931873322,
      "learning_rate": 5.726828236438334e-06,
      "loss": 0.1068,
      "num_input_tokens_seen": 106116784,
      "step": 49175
    },
    {
      "epoch": 8.022838499184338,
      "grad_norm": 0.07622867077589035,
      "learning_rate": 5.722295406307632e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 106127088,
      "step": 49180
    },
    {
      "epoch": 8.023654159869494,
      "grad_norm": 0.3558162450790405,
      "learning_rate": 5.71776413892548e-06,
      "loss": 0.0909,
      "num_input_tokens_seen": 106136912,
      "step": 49185
    },
    {
      "epoch": 8.02446982055465,
      "grad_norm": 0.19106322526931763,
      "learning_rate": 5.713234434659203e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 106148816,
      "step": 49190
    },
    {
      "epoch": 8.025285481239804,
      "grad_norm": 0.04668969660997391,
      "learning_rate": 5.708706293876004e-06,
      "loss": 0.0022,
      "num_input_tokens_seen": 106158672,
      "step": 49195
    },
    {
      "epoch": 8.02610114192496,
      "grad_norm": 5.1401238441467285,
      "learning_rate": 5.7041797169429536e-06,
      "loss": 0.022,
      "num_input_tokens_seen": 106169744,
      "step": 49200
    },
    {
      "epoch": 8.026916802610113,
      "grad_norm": 0.06998219341039658,
      "learning_rate": 5.699654704227003e-06,
      "loss": 0.2022,
      "num_input_tokens_seen": 106179824,
      "step": 49205
    },
    {
      "epoch": 8.02773246329527,
      "grad_norm": 0.04609766975045204,
      "learning_rate": 5.695131256094971e-06,
      "loss": 0.2098,
      "num_input_tokens_seen": 106190640,
      "step": 49210
    },
    {
      "epoch": 8.028548123980425,
      "grad_norm": 0.10465165972709656,
      "learning_rate": 5.6906093729135495e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 106200112,
      "step": 49215
    },
    {
      "epoch": 8.029363784665579,
      "grad_norm": 0.07307170331478119,
      "learning_rate": 5.6860890550493095e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 106211440,
      "step": 49220
    },
    {
      "epoch": 8.030179445350734,
      "grad_norm": 0.617223858833313,
      "learning_rate": 5.681570302868688e-06,
      "loss": 0.1075,
      "num_input_tokens_seen": 106222224,
      "step": 49225
    },
    {
      "epoch": 8.030995106035888,
      "grad_norm": 0.04321765899658203,
      "learning_rate": 5.677053116737999e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 106233968,
      "step": 49230
    },
    {
      "epoch": 8.031810766721044,
      "grad_norm": 0.4132688045501709,
      "learning_rate": 5.67253749702342e-06,
      "loss": 0.2008,
      "num_input_tokens_seen": 106244944,
      "step": 49235
    },
    {
      "epoch": 8.0326264274062,
      "grad_norm": 4.113018989562988,
      "learning_rate": 5.668023444091025e-06,
      "loss": 0.1212,
      "num_input_tokens_seen": 106255664,
      "step": 49240
    },
    {
      "epoch": 8.033442088091354,
      "grad_norm": 0.09752390533685684,
      "learning_rate": 5.663510958306739e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 106266672,
      "step": 49245
    },
    {
      "epoch": 8.03425774877651,
      "grad_norm": 0.09980696439743042,
      "learning_rate": 5.659000040036366e-06,
      "loss": 0.1372,
      "num_input_tokens_seen": 106278288,
      "step": 49250
    },
    {
      "epoch": 8.035073409461663,
      "grad_norm": 4.008538722991943,
      "learning_rate": 5.654490689645589e-06,
      "loss": 0.0063,
      "num_input_tokens_seen": 106289744,
      "step": 49255
    },
    {
      "epoch": 8.035889070146819,
      "grad_norm": 0.23308169841766357,
      "learning_rate": 5.649982907499951e-06,
      "loss": 0.0445,
      "num_input_tokens_seen": 106298480,
      "step": 49260
    },
    {
      "epoch": 8.036704730831975,
      "grad_norm": 0.031971950083971024,
      "learning_rate": 5.645476693964874e-06,
      "loss": 0.0855,
      "num_input_tokens_seen": 106310800,
      "step": 49265
    },
    {
      "epoch": 8.037520391517129,
      "grad_norm": 3.505389451980591,
      "learning_rate": 5.640972049405666e-06,
      "loss": 0.2035,
      "num_input_tokens_seen": 106322000,
      "step": 49270
    },
    {
      "epoch": 8.038336052202284,
      "grad_norm": 0.046675022691488266,
      "learning_rate": 5.636468974187492e-06,
      "loss": 0.1079,
      "num_input_tokens_seen": 106331632,
      "step": 49275
    },
    {
      "epoch": 8.039151712887438,
      "grad_norm": 0.07738251984119415,
      "learning_rate": 5.631967468675392e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 106343344,
      "step": 49280
    },
    {
      "epoch": 8.039967373572594,
      "grad_norm": 0.04473774880170822,
      "learning_rate": 5.627467533234282e-06,
      "loss": 0.1477,
      "num_input_tokens_seen": 106354288,
      "step": 49285
    },
    {
      "epoch": 8.040783034257748,
      "grad_norm": 0.04963133484125137,
      "learning_rate": 5.622969168228947e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 106365168,
      "step": 49290
    },
    {
      "epoch": 8.041598694942904,
      "grad_norm": 0.052206624299287796,
      "learning_rate": 5.61847237402405e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 106375440,
      "step": 49295
    },
    {
      "epoch": 8.04241435562806,
      "grad_norm": 4.268579483032227,
      "learning_rate": 5.613977150984123e-06,
      "loss": 0.1155,
      "num_input_tokens_seen": 106386032,
      "step": 49300
    },
    {
      "epoch": 8.043230016313213,
      "grad_norm": 0.07151418924331665,
      "learning_rate": 5.609483499473575e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 106398160,
      "step": 49305
    },
    {
      "epoch": 8.044045676998369,
      "grad_norm": 0.0778755322098732,
      "learning_rate": 5.604991419856678e-06,
      "loss": 0.0722,
      "num_input_tokens_seen": 106408432,
      "step": 49310
    },
    {
      "epoch": 8.044861337683523,
      "grad_norm": 0.1610373854637146,
      "learning_rate": 5.600500912497586e-06,
      "loss": 0.3244,
      "num_input_tokens_seen": 106419440,
      "step": 49315
    },
    {
      "epoch": 8.045676998368679,
      "grad_norm": 0.12459628283977509,
      "learning_rate": 5.596011977760324e-06,
      "loss": 0.0707,
      "num_input_tokens_seen": 106430000,
      "step": 49320
    },
    {
      "epoch": 8.046492659053834,
      "grad_norm": 0.08844204992055893,
      "learning_rate": 5.591524616008784e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 106441392,
      "step": 49325
    },
    {
      "epoch": 8.047308319738988,
      "grad_norm": 0.9334814548492432,
      "learning_rate": 5.587038827606736e-06,
      "loss": 0.007,
      "num_input_tokens_seen": 106452688,
      "step": 49330
    },
    {
      "epoch": 8.048123980424144,
      "grad_norm": 0.9453881978988647,
      "learning_rate": 5.582554612917823e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 106462352,
      "step": 49335
    },
    {
      "epoch": 8.048939641109298,
      "grad_norm": 9.040762901306152,
      "learning_rate": 5.578071972305554e-06,
      "loss": 0.2573,
      "num_input_tokens_seen": 106472208,
      "step": 49340
    },
    {
      "epoch": 8.049755301794454,
      "grad_norm": 3.1483774185180664,
      "learning_rate": 5.57359090613331e-06,
      "loss": 0.329,
      "num_input_tokens_seen": 106481840,
      "step": 49345
    },
    {
      "epoch": 8.05057096247961,
      "grad_norm": 0.07931210845708847,
      "learning_rate": 5.569111414764363e-06,
      "loss": 0.2614,
      "num_input_tokens_seen": 106492912,
      "step": 49350
    },
    {
      "epoch": 8.051386623164763,
      "grad_norm": 0.08259209990501404,
      "learning_rate": 5.564633498561839e-06,
      "loss": 0.2622,
      "num_input_tokens_seen": 106503440,
      "step": 49355
    },
    {
      "epoch": 8.052202283849919,
      "grad_norm": 0.1957506388425827,
      "learning_rate": 5.560157157888735e-06,
      "loss": 0.0165,
      "num_input_tokens_seen": 106512880,
      "step": 49360
    },
    {
      "epoch": 8.053017944535073,
      "grad_norm": 0.04240275174379349,
      "learning_rate": 5.555682393107928e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 106525296,
      "step": 49365
    },
    {
      "epoch": 8.053833605220229,
      "grad_norm": 7.280547618865967,
      "learning_rate": 5.551209204582167e-06,
      "loss": 0.2064,
      "num_input_tokens_seen": 106535536,
      "step": 49370
    },
    {
      "epoch": 8.054649265905383,
      "grad_norm": 0.2381419986486435,
      "learning_rate": 5.54673759267407e-06,
      "loss": 0.1138,
      "num_input_tokens_seen": 106546160,
      "step": 49375
    },
    {
      "epoch": 8.055464926590538,
      "grad_norm": 0.05607723817229271,
      "learning_rate": 5.542267557746128e-06,
      "loss": 0.1253,
      "num_input_tokens_seen": 106558064,
      "step": 49380
    },
    {
      "epoch": 8.056280587275694,
      "grad_norm": 0.1966191530227661,
      "learning_rate": 5.537799100160704e-06,
      "loss": 0.0864,
      "num_input_tokens_seen": 106568400,
      "step": 49385
    },
    {
      "epoch": 8.057096247960848,
      "grad_norm": 0.2882792055606842,
      "learning_rate": 5.533332220280038e-06,
      "loss": 0.0096,
      "num_input_tokens_seen": 106578192,
      "step": 49390
    },
    {
      "epoch": 8.057911908646004,
      "grad_norm": 0.08140349388122559,
      "learning_rate": 5.5288669184662325e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 106587312,
      "step": 49395
    },
    {
      "epoch": 8.058727569331158,
      "grad_norm": 0.12514647841453552,
      "learning_rate": 5.524403195081271e-06,
      "loss": 0.1624,
      "num_input_tokens_seen": 106598672,
      "step": 49400
    },
    {
      "epoch": 8.059543230016313,
      "grad_norm": 0.07031819224357605,
      "learning_rate": 5.519941050487007e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 106608208,
      "step": 49405
    },
    {
      "epoch": 8.060358890701469,
      "grad_norm": 0.0699918270111084,
      "learning_rate": 5.515480485045152e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 106619984,
      "step": 49410
    },
    {
      "epoch": 8.061174551386623,
      "grad_norm": 3.1284658908843994,
      "learning_rate": 5.511021499117322e-06,
      "loss": 0.1468,
      "num_input_tokens_seen": 106631792,
      "step": 49415
    },
    {
      "epoch": 8.061990212071779,
      "grad_norm": 0.015321805141866207,
      "learning_rate": 5.5065640930649725e-06,
      "loss": 0.1166,
      "num_input_tokens_seen": 106643600,
      "step": 49420
    },
    {
      "epoch": 8.062805872756933,
      "grad_norm": 3.460632801055908,
      "learning_rate": 5.502108267249448e-06,
      "loss": 0.0934,
      "num_input_tokens_seen": 106654672,
      "step": 49425
    },
    {
      "epoch": 8.063621533442088,
      "grad_norm": 5.855128765106201,
      "learning_rate": 5.497654022031959e-06,
      "loss": 0.0693,
      "num_input_tokens_seen": 106665456,
      "step": 49430
    },
    {
      "epoch": 8.064437194127244,
      "grad_norm": 0.20783241093158722,
      "learning_rate": 5.493201357773589e-06,
      "loss": 0.2219,
      "num_input_tokens_seen": 106676880,
      "step": 49435
    },
    {
      "epoch": 8.065252854812398,
      "grad_norm": 0.14572155475616455,
      "learning_rate": 5.488750274835291e-06,
      "loss": 0.1397,
      "num_input_tokens_seen": 106688528,
      "step": 49440
    },
    {
      "epoch": 8.066068515497554,
      "grad_norm": 0.29012978076934814,
      "learning_rate": 5.4843007735778996e-06,
      "loss": 0.0727,
      "num_input_tokens_seen": 106698992,
      "step": 49445
    },
    {
      "epoch": 8.066884176182707,
      "grad_norm": 0.12157758325338364,
      "learning_rate": 5.4798528543620965e-06,
      "loss": 0.0553,
      "num_input_tokens_seen": 106709168,
      "step": 49450
    },
    {
      "epoch": 8.067699836867863,
      "grad_norm": 0.16166500747203827,
      "learning_rate": 5.475406517548476e-06,
      "loss": 0.194,
      "num_input_tokens_seen": 106719920,
      "step": 49455
    },
    {
      "epoch": 8.068515497553017,
      "grad_norm": 0.06889856606721878,
      "learning_rate": 5.47096176349747e-06,
      "loss": 0.0778,
      "num_input_tokens_seen": 106729872,
      "step": 49460
    },
    {
      "epoch": 8.069331158238173,
      "grad_norm": 3.6685426235198975,
      "learning_rate": 5.466518592569391e-06,
      "loss": 0.1442,
      "num_input_tokens_seen": 106741168,
      "step": 49465
    },
    {
      "epoch": 8.070146818923329,
      "grad_norm": 0.1276254802942276,
      "learning_rate": 5.4620770051244275e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 106751376,
      "step": 49470
    },
    {
      "epoch": 8.070962479608482,
      "grad_norm": 0.37503746151924133,
      "learning_rate": 5.457637001522636e-06,
      "loss": 0.1265,
      "num_input_tokens_seen": 106761904,
      "step": 49475
    },
    {
      "epoch": 8.071778140293638,
      "grad_norm": 6.915079593658447,
      "learning_rate": 5.453198582123947e-06,
      "loss": 0.1625,
      "num_input_tokens_seen": 106772240,
      "step": 49480
    },
    {
      "epoch": 8.072593800978792,
      "grad_norm": 0.09333750605583191,
      "learning_rate": 5.448761747288161e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 106783216,
      "step": 49485
    },
    {
      "epoch": 8.073409461663948,
      "grad_norm": 0.2951674163341522,
      "learning_rate": 5.444326497374949e-06,
      "loss": 0.0985,
      "num_input_tokens_seen": 106794896,
      "step": 49490
    },
    {
      "epoch": 8.074225122349104,
      "grad_norm": 1.0948500633239746,
      "learning_rate": 5.439892832743856e-06,
      "loss": 0.2702,
      "num_input_tokens_seen": 106805360,
      "step": 49495
    },
    {
      "epoch": 8.075040783034257,
      "grad_norm": 5.78016996383667,
      "learning_rate": 5.435460753754296e-06,
      "loss": 0.1262,
      "num_input_tokens_seen": 106816144,
      "step": 49500
    },
    {
      "epoch": 8.075856443719413,
      "grad_norm": 0.3723219037055969,
      "learning_rate": 5.431030260765557e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 106827088,
      "step": 49505
    },
    {
      "epoch": 8.076672104404567,
      "grad_norm": 0.06229551509022713,
      "learning_rate": 5.426601354136799e-06,
      "loss": 0.0437,
      "num_input_tokens_seen": 106837488,
      "step": 49510
    },
    {
      "epoch": 8.077487765089723,
      "grad_norm": 0.041666120290756226,
      "learning_rate": 5.4221740342270455e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 106849520,
      "step": 49515
    },
    {
      "epoch": 8.078303425774878,
      "grad_norm": 0.05304932966828346,
      "learning_rate": 5.4177483013952065e-06,
      "loss": 0.0637,
      "num_input_tokens_seen": 106860528,
      "step": 49520
    },
    {
      "epoch": 8.079119086460032,
      "grad_norm": 2.4275498390197754,
      "learning_rate": 5.413324156000046e-06,
      "loss": 0.1433,
      "num_input_tokens_seen": 106869936,
      "step": 49525
    },
    {
      "epoch": 8.079934747145188,
      "grad_norm": 3.568582534790039,
      "learning_rate": 5.408901598400212e-06,
      "loss": 0.0892,
      "num_input_tokens_seen": 106881936,
      "step": 49530
    },
    {
      "epoch": 8.080750407830342,
      "grad_norm": 0.1210571676492691,
      "learning_rate": 5.40448062895422e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 106891376,
      "step": 49535
    },
    {
      "epoch": 8.081566068515498,
      "grad_norm": 0.16298231482505798,
      "learning_rate": 5.400061248020452e-06,
      "loss": 0.1194,
      "num_input_tokens_seen": 106901168,
      "step": 49540
    },
    {
      "epoch": 8.082381729200652,
      "grad_norm": 0.21729084849357605,
      "learning_rate": 5.395643455957172e-06,
      "loss": 0.2148,
      "num_input_tokens_seen": 106912432,
      "step": 49545
    },
    {
      "epoch": 8.083197389885807,
      "grad_norm": 15.05525016784668,
      "learning_rate": 5.391227253122502e-06,
      "loss": 0.0354,
      "num_input_tokens_seen": 106924592,
      "step": 49550
    },
    {
      "epoch": 8.084013050570963,
      "grad_norm": 0.0854477807879448,
      "learning_rate": 5.386812639874439e-06,
      "loss": 0.0096,
      "num_input_tokens_seen": 106934704,
      "step": 49555
    },
    {
      "epoch": 8.084828711256117,
      "grad_norm": 0.37851154804229736,
      "learning_rate": 5.382399616570869e-06,
      "loss": 0.0083,
      "num_input_tokens_seen": 106945936,
      "step": 49560
    },
    {
      "epoch": 8.085644371941273,
      "grad_norm": 15.90233325958252,
      "learning_rate": 5.377988183569521e-06,
      "loss": 0.0422,
      "num_input_tokens_seen": 106957264,
      "step": 49565
    },
    {
      "epoch": 8.086460032626427,
      "grad_norm": 0.1077587828040123,
      "learning_rate": 5.3735783412280134e-06,
      "loss": 0.1699,
      "num_input_tokens_seen": 106968816,
      "step": 49570
    },
    {
      "epoch": 8.087275693311582,
      "grad_norm": 0.0675574466586113,
      "learning_rate": 5.36917008990383e-06,
      "loss": 0.2308,
      "num_input_tokens_seen": 106981680,
      "step": 49575
    },
    {
      "epoch": 8.088091353996738,
      "grad_norm": 0.11761905997991562,
      "learning_rate": 5.364763429954317e-06,
      "loss": 0.0504,
      "num_input_tokens_seen": 106992048,
      "step": 49580
    },
    {
      "epoch": 8.088907014681892,
      "grad_norm": 0.19656305015087128,
      "learning_rate": 5.360358361736714e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 107004016,
      "step": 49585
    },
    {
      "epoch": 8.089722675367048,
      "grad_norm": 3.6156342029571533,
      "learning_rate": 5.3559548856081135e-06,
      "loss": 0.0982,
      "num_input_tokens_seen": 107015024,
      "step": 49590
    },
    {
      "epoch": 8.090538336052202,
      "grad_norm": 0.1276438683271408,
      "learning_rate": 5.351553001925486e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 107025008,
      "step": 49595
    },
    {
      "epoch": 8.091353996737357,
      "grad_norm": 17.190284729003906,
      "learning_rate": 5.347152711045664e-06,
      "loss": 0.0235,
      "num_input_tokens_seen": 107035408,
      "step": 49600
    },
    {
      "epoch": 8.092169657422513,
      "grad_norm": 2.734891414642334,
      "learning_rate": 5.342754013325363e-06,
      "loss": 0.0422,
      "num_input_tokens_seen": 107046960,
      "step": 49605
    },
    {
      "epoch": 8.092985318107667,
      "grad_norm": 5.170504570007324,
      "learning_rate": 5.338356909121159e-06,
      "loss": 0.1617,
      "num_input_tokens_seen": 107058736,
      "step": 49610
    },
    {
      "epoch": 8.093800978792823,
      "grad_norm": 0.4141773581504822,
      "learning_rate": 5.3339613987895084e-06,
      "loss": 0.1956,
      "num_input_tokens_seen": 107069872,
      "step": 49615
    },
    {
      "epoch": 8.094616639477977,
      "grad_norm": 0.05458392947912216,
      "learning_rate": 5.329567482686729e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 107080400,
      "step": 49620
    },
    {
      "epoch": 8.095432300163132,
      "grad_norm": 0.1334652304649353,
      "learning_rate": 5.325175161169019e-06,
      "loss": 0.0111,
      "num_input_tokens_seen": 107091696,
      "step": 49625
    },
    {
      "epoch": 8.096247960848286,
      "grad_norm": 0.24071387946605682,
      "learning_rate": 5.320784434592438e-06,
      "loss": 0.1279,
      "num_input_tokens_seen": 107101168,
      "step": 49630
    },
    {
      "epoch": 8.097063621533442,
      "grad_norm": 0.18897297978401184,
      "learning_rate": 5.316395303312921e-06,
      "loss": 0.0832,
      "num_input_tokens_seen": 107112464,
      "step": 49635
    },
    {
      "epoch": 8.097879282218598,
      "grad_norm": 0.2674519419670105,
      "learning_rate": 5.3120077676862754e-06,
      "loss": 0.0082,
      "num_input_tokens_seen": 107123280,
      "step": 49640
    },
    {
      "epoch": 8.098694942903752,
      "grad_norm": 4.2298102378845215,
      "learning_rate": 5.307621828068177e-06,
      "loss": 0.1066,
      "num_input_tokens_seen": 107134800,
      "step": 49645
    },
    {
      "epoch": 8.099510603588907,
      "grad_norm": 0.09429102391004562,
      "learning_rate": 5.303237484814169e-06,
      "loss": 0.0733,
      "num_input_tokens_seen": 107145904,
      "step": 49650
    },
    {
      "epoch": 8.100326264274061,
      "grad_norm": 3.282207489013672,
      "learning_rate": 5.2988547382796735e-06,
      "loss": 0.1514,
      "num_input_tokens_seen": 107156464,
      "step": 49655
    },
    {
      "epoch": 8.101141924959217,
      "grad_norm": 0.05307849869132042,
      "learning_rate": 5.294473588819968e-06,
      "loss": 0.0103,
      "num_input_tokens_seen": 107167824,
      "step": 49660
    },
    {
      "epoch": 8.101957585644373,
      "grad_norm": 1.4988816976547241,
      "learning_rate": 5.2900940367902245e-06,
      "loss": 0.0117,
      "num_input_tokens_seen": 107178928,
      "step": 49665
    },
    {
      "epoch": 8.102773246329527,
      "grad_norm": 0.04693633317947388,
      "learning_rate": 5.2857160825454635e-06,
      "loss": 0.0963,
      "num_input_tokens_seen": 107191120,
      "step": 49670
    },
    {
      "epoch": 8.103588907014682,
      "grad_norm": 0.039586059749126434,
      "learning_rate": 5.28133972644059e-06,
      "loss": 0.0864,
      "num_input_tokens_seen": 107201872,
      "step": 49675
    },
    {
      "epoch": 8.104404567699836,
      "grad_norm": 2.518397808074951,
      "learning_rate": 5.276964968830367e-06,
      "loss": 0.1285,
      "num_input_tokens_seen": 107214192,
      "step": 49680
    },
    {
      "epoch": 8.105220228384992,
      "grad_norm": 0.31964388489723206,
      "learning_rate": 5.272591810069438e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 107225616,
      "step": 49685
    },
    {
      "epoch": 8.106035889070148,
      "grad_norm": 0.03488241881132126,
      "learning_rate": 5.268220250512315e-06,
      "loss": 0.0101,
      "num_input_tokens_seen": 107236208,
      "step": 49690
    },
    {
      "epoch": 8.106851549755302,
      "grad_norm": 0.058267757296562195,
      "learning_rate": 5.263850290513373e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 107247632,
      "step": 49695
    },
    {
      "epoch": 8.107667210440457,
      "grad_norm": 0.07162422686815262,
      "learning_rate": 5.259481930426869e-06,
      "loss": 0.0017,
      "num_input_tokens_seen": 107258288,
      "step": 49700
    },
    {
      "epoch": 8.108482871125611,
      "grad_norm": 0.031090060248970985,
      "learning_rate": 5.2551151706069225e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 107268944,
      "step": 49705
    },
    {
      "epoch": 8.109298531810767,
      "grad_norm": 0.12256871908903122,
      "learning_rate": 5.250750011407521e-06,
      "loss": 0.1123,
      "num_input_tokens_seen": 107279472,
      "step": 49710
    },
    {
      "epoch": 8.11011419249592,
      "grad_norm": 7.657137870788574,
      "learning_rate": 5.246386453182533e-06,
      "loss": 0.1078,
      "num_input_tokens_seen": 107290064,
      "step": 49715
    },
    {
      "epoch": 8.110929853181077,
      "grad_norm": 0.08682429790496826,
      "learning_rate": 5.242024496285683e-06,
      "loss": 0.0024,
      "num_input_tokens_seen": 107300560,
      "step": 49720
    },
    {
      "epoch": 8.111745513866232,
      "grad_norm": 0.09137769788503647,
      "learning_rate": 5.237664141070583e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 107310512,
      "step": 49725
    },
    {
      "epoch": 8.112561174551386,
      "grad_norm": 0.29633578658103943,
      "learning_rate": 5.2333053878907e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 107321296,
      "step": 49730
    },
    {
      "epoch": 8.113376835236542,
      "grad_norm": 0.16949453949928284,
      "learning_rate": 5.228948237099379e-06,
      "loss": 0.1126,
      "num_input_tokens_seen": 107331920,
      "step": 49735
    },
    {
      "epoch": 8.114192495921696,
      "grad_norm": 0.373892605304718,
      "learning_rate": 5.224592689049832e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 107342768,
      "step": 49740
    },
    {
      "epoch": 8.115008156606851,
      "grad_norm": 0.3301897346973419,
      "learning_rate": 5.220238744095137e-06,
      "loss": 0.058,
      "num_input_tokens_seen": 107352304,
      "step": 49745
    },
    {
      "epoch": 8.115823817292007,
      "grad_norm": 0.01428622379899025,
      "learning_rate": 5.215886402588255e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 107362448,
      "step": 49750
    },
    {
      "epoch": 8.116639477977161,
      "grad_norm": 0.029162505641579628,
      "learning_rate": 5.211535664882003e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 107372368,
      "step": 49755
    },
    {
      "epoch": 8.117455138662317,
      "grad_norm": 0.17050567269325256,
      "learning_rate": 5.207186531329075e-06,
      "loss": 0.0826,
      "num_input_tokens_seen": 107383120,
      "step": 49760
    },
    {
      "epoch": 8.11827079934747,
      "grad_norm": 0.08083541691303253,
      "learning_rate": 5.202839002282037e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 107394576,
      "step": 49765
    },
    {
      "epoch": 8.119086460032626,
      "grad_norm": 0.043164532631635666,
      "learning_rate": 5.198493078093311e-06,
      "loss": 0.169,
      "num_input_tokens_seen": 107404752,
      "step": 49770
    },
    {
      "epoch": 8.119902120717782,
      "grad_norm": 0.10148000717163086,
      "learning_rate": 5.194148759115214e-06,
      "loss": 0.1128,
      "num_input_tokens_seen": 107414288,
      "step": 49775
    },
    {
      "epoch": 8.120717781402936,
      "grad_norm": 0.08053573966026306,
      "learning_rate": 5.189806045699913e-06,
      "loss": 0.0664,
      "num_input_tokens_seen": 107425648,
      "step": 49780
    },
    {
      "epoch": 8.121533442088092,
      "grad_norm": 0.2354675829410553,
      "learning_rate": 5.185464938199449e-06,
      "loss": 0.0611,
      "num_input_tokens_seen": 107437936,
      "step": 49785
    },
    {
      "epoch": 8.122349102773246,
      "grad_norm": 0.5793733596801758,
      "learning_rate": 5.181125436965739e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 107448336,
      "step": 49790
    },
    {
      "epoch": 8.123164763458401,
      "grad_norm": 0.08733253180980682,
      "learning_rate": 5.176787542350558e-06,
      "loss": 0.0689,
      "num_input_tokens_seen": 107458416,
      "step": 49795
    },
    {
      "epoch": 8.123980424143557,
      "grad_norm": 0.06490445882081985,
      "learning_rate": 5.172451254705559e-06,
      "loss": 0.0483,
      "num_input_tokens_seen": 107469872,
      "step": 49800
    },
    {
      "epoch": 8.124796084828711,
      "grad_norm": 0.10985083132982254,
      "learning_rate": 5.1681165743822676e-06,
      "loss": 0.1949,
      "num_input_tokens_seen": 107479760,
      "step": 49805
    },
    {
      "epoch": 8.125611745513867,
      "grad_norm": 0.035494428128004074,
      "learning_rate": 5.1637835017320726e-06,
      "loss": 0.0651,
      "num_input_tokens_seen": 107490096,
      "step": 49810
    },
    {
      "epoch": 8.12642740619902,
      "grad_norm": 0.09943348914384842,
      "learning_rate": 5.159452037106236e-06,
      "loss": 0.1107,
      "num_input_tokens_seen": 107500368,
      "step": 49815
    },
    {
      "epoch": 8.127243066884176,
      "grad_norm": 0.0746140256524086,
      "learning_rate": 5.155122180855884e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 107511600,
      "step": 49820
    },
    {
      "epoch": 8.12805872756933,
      "grad_norm": 0.059252284467220306,
      "learning_rate": 5.150793933332024e-06,
      "loss": 0.1505,
      "num_input_tokens_seen": 107522672,
      "step": 49825
    },
    {
      "epoch": 8.128874388254486,
      "grad_norm": 0.12755051255226135,
      "learning_rate": 5.146467294885518e-06,
      "loss": 0.0026,
      "num_input_tokens_seen": 107534096,
      "step": 49830
    },
    {
      "epoch": 8.129690048939642,
      "grad_norm": 3.337573766708374,
      "learning_rate": 5.142142265867112e-06,
      "loss": 0.1877,
      "num_input_tokens_seen": 107543472,
      "step": 49835
    },
    {
      "epoch": 8.130505709624796,
      "grad_norm": 3.7103707790374756,
      "learning_rate": 5.137818846627409e-06,
      "loss": 0.2132,
      "num_input_tokens_seen": 107554224,
      "step": 49840
    },
    {
      "epoch": 8.131321370309951,
      "grad_norm": 0.11919867992401123,
      "learning_rate": 5.13349703751689e-06,
      "loss": 0.125,
      "num_input_tokens_seen": 107564240,
      "step": 49845
    },
    {
      "epoch": 8.132137030995105,
      "grad_norm": 0.4787959158420563,
      "learning_rate": 5.129176838885905e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 107574800,
      "step": 49850
    },
    {
      "epoch": 8.132952691680261,
      "grad_norm": 0.0798761248588562,
      "learning_rate": 5.124858251084666e-06,
      "loss": 0.1976,
      "num_input_tokens_seen": 107585520,
      "step": 49855
    },
    {
      "epoch": 8.133768352365417,
      "grad_norm": 0.036203090101480484,
      "learning_rate": 5.120541274463264e-06,
      "loss": 0.1533,
      "num_input_tokens_seen": 107596112,
      "step": 49860
    },
    {
      "epoch": 8.13458401305057,
      "grad_norm": 13.261408805847168,
      "learning_rate": 5.116225909371649e-06,
      "loss": 0.0112,
      "num_input_tokens_seen": 107607120,
      "step": 49865
    },
    {
      "epoch": 8.135399673735726,
      "grad_norm": 0.05671172961592674,
      "learning_rate": 5.111912156159657e-06,
      "loss": 0.18,
      "num_input_tokens_seen": 107618000,
      "step": 49870
    },
    {
      "epoch": 8.13621533442088,
      "grad_norm": 16.111249923706055,
      "learning_rate": 5.107600015176975e-06,
      "loss": 0.122,
      "num_input_tokens_seen": 107629232,
      "step": 49875
    },
    {
      "epoch": 8.137030995106036,
      "grad_norm": 0.17345628142356873,
      "learning_rate": 5.103289486773169e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 107640784,
      "step": 49880
    },
    {
      "epoch": 8.137846655791192,
      "grad_norm": 3.813416004180908,
      "learning_rate": 5.098980571297673e-06,
      "loss": 0.1718,
      "num_input_tokens_seen": 107650576,
      "step": 49885
    },
    {
      "epoch": 8.138662316476346,
      "grad_norm": 0.14174208045005798,
      "learning_rate": 5.094673269099781e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 107660368,
      "step": 49890
    },
    {
      "epoch": 8.139477977161501,
      "grad_norm": 0.0553726889193058,
      "learning_rate": 5.090367580528679e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 107670096,
      "step": 49895
    },
    {
      "epoch": 8.140293637846655,
      "grad_norm": 13.390634536743164,
      "learning_rate": 5.086063505933403e-06,
      "loss": 0.1784,
      "num_input_tokens_seen": 107680240,
      "step": 49900
    },
    {
      "epoch": 8.141109298531811,
      "grad_norm": 0.0556836873292923,
      "learning_rate": 5.081761045662861e-06,
      "loss": 0.0079,
      "num_input_tokens_seen": 107691504,
      "step": 49905
    },
    {
      "epoch": 8.141924959216965,
      "grad_norm": 3.9966859817504883,
      "learning_rate": 5.077460200065834e-06,
      "loss": 0.2348,
      "num_input_tokens_seen": 107701968,
      "step": 49910
    },
    {
      "epoch": 8.14274061990212,
      "grad_norm": 3.627897262573242,
      "learning_rate": 5.073160969490967e-06,
      "loss": 0.232,
      "num_input_tokens_seen": 107713136,
      "step": 49915
    },
    {
      "epoch": 8.143556280587276,
      "grad_norm": 9.396637916564941,
      "learning_rate": 5.068863354286779e-06,
      "loss": 0.0747,
      "num_input_tokens_seen": 107724080,
      "step": 49920
    },
    {
      "epoch": 8.14437194127243,
      "grad_norm": 3.8547706604003906,
      "learning_rate": 5.064567354801658e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 107735792,
      "step": 49925
    },
    {
      "epoch": 8.145187601957586,
      "grad_norm": 0.5412003993988037,
      "learning_rate": 5.060272971383862e-06,
      "loss": 0.1378,
      "num_input_tokens_seen": 107747280,
      "step": 49930
    },
    {
      "epoch": 8.14600326264274,
      "grad_norm": 1.3821433782577515,
      "learning_rate": 5.055980204381508e-06,
      "loss": 0.1476,
      "num_input_tokens_seen": 107758448,
      "step": 49935
    },
    {
      "epoch": 8.146818923327896,
      "grad_norm": 0.20223121345043182,
      "learning_rate": 5.051689054142594e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 107769712,
      "step": 49940
    },
    {
      "epoch": 8.147634584013051,
      "grad_norm": 0.023549934849143028,
      "learning_rate": 5.047399521014984e-06,
      "loss": 0.0973,
      "num_input_tokens_seen": 107780624,
      "step": 49945
    },
    {
      "epoch": 8.148450244698205,
      "grad_norm": 0.10849035531282425,
      "learning_rate": 5.043111605346404e-06,
      "loss": 0.0709,
      "num_input_tokens_seen": 107790800,
      "step": 49950
    },
    {
      "epoch": 8.149265905383361,
      "grad_norm": 0.05483119934797287,
      "learning_rate": 5.03882530748446e-06,
      "loss": 0.0826,
      "num_input_tokens_seen": 107802736,
      "step": 49955
    },
    {
      "epoch": 8.150081566068515,
      "grad_norm": 0.07685652375221252,
      "learning_rate": 5.034540627776618e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 107813552,
      "step": 49960
    },
    {
      "epoch": 8.15089722675367,
      "grad_norm": 0.048805173486471176,
      "learning_rate": 5.030257566570215e-06,
      "loss": 0.1742,
      "num_input_tokens_seen": 107825264,
      "step": 49965
    },
    {
      "epoch": 8.151712887438826,
      "grad_norm": 0.058466944843530655,
      "learning_rate": 5.025976124212461e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 107836880,
      "step": 49970
    },
    {
      "epoch": 8.15252854812398,
      "grad_norm": 0.09701434522867203,
      "learning_rate": 5.0216963010504295e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 107849392,
      "step": 49975
    },
    {
      "epoch": 8.153344208809136,
      "grad_norm": 3.819746732711792,
      "learning_rate": 5.017418097431059e-06,
      "loss": 0.0969,
      "num_input_tokens_seen": 107860144,
      "step": 49980
    },
    {
      "epoch": 8.15415986949429,
      "grad_norm": 0.054632510989904404,
      "learning_rate": 5.013141513701173e-06,
      "loss": 0.0978,
      "num_input_tokens_seen": 107869520,
      "step": 49985
    },
    {
      "epoch": 8.154975530179446,
      "grad_norm": 0.19069510698318481,
      "learning_rate": 5.008866550207447e-06,
      "loss": 0.1492,
      "num_input_tokens_seen": 107879792,
      "step": 49990
    },
    {
      "epoch": 8.1557911908646,
      "grad_norm": 0.7505430579185486,
      "learning_rate": 5.004593207296434e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 107890672,
      "step": 49995
    },
    {
      "epoch": 8.156606851549755,
      "grad_norm": 0.06662391871213913,
      "learning_rate": 5.000321485314552e-06,
      "loss": 0.105,
      "num_input_tokens_seen": 107900528,
      "step": 50000
    },
    {
      "epoch": 8.15742251223491,
      "grad_norm": 0.20353369414806366,
      "learning_rate": 4.9960513846080885e-06,
      "loss": 0.2116,
      "num_input_tokens_seen": 107911248,
      "step": 50005
    },
    {
      "epoch": 8.158238172920065,
      "grad_norm": 0.11364521086215973,
      "learning_rate": 4.991782905523196e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 107922800,
      "step": 50010
    },
    {
      "epoch": 8.15905383360522,
      "grad_norm": 8.014961242675781,
      "learning_rate": 4.987516048405905e-06,
      "loss": 0.0122,
      "num_input_tokens_seen": 107934032,
      "step": 50015
    },
    {
      "epoch": 8.159869494290374,
      "grad_norm": 0.06340456008911133,
      "learning_rate": 4.983250813602103e-06,
      "loss": 0.0084,
      "num_input_tokens_seen": 107946128,
      "step": 50020
    },
    {
      "epoch": 8.16068515497553,
      "grad_norm": 0.0635066032409668,
      "learning_rate": 4.978987201457555e-06,
      "loss": 0.0737,
      "num_input_tokens_seen": 107956912,
      "step": 50025
    },
    {
      "epoch": 8.161500815660686,
      "grad_norm": 13.93176555633545,
      "learning_rate": 4.97472521231789e-06,
      "loss": 0.0521,
      "num_input_tokens_seen": 107968112,
      "step": 50030
    },
    {
      "epoch": 8.16231647634584,
      "grad_norm": 0.04283421114087105,
      "learning_rate": 4.9704648465286e-06,
      "loss": 0.0223,
      "num_input_tokens_seen": 107978352,
      "step": 50035
    },
    {
      "epoch": 8.163132137030995,
      "grad_norm": 0.055563513189554214,
      "learning_rate": 4.966206104435064e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 107989264,
      "step": 50040
    },
    {
      "epoch": 8.16394779771615,
      "grad_norm": 17.561492919921875,
      "learning_rate": 4.961948986382511e-06,
      "loss": 0.0892,
      "num_input_tokens_seen": 107999888,
      "step": 50045
    },
    {
      "epoch": 8.164763458401305,
      "grad_norm": 0.06243748217821121,
      "learning_rate": 4.957693492716048e-06,
      "loss": 0.0994,
      "num_input_tokens_seen": 108009680,
      "step": 50050
    },
    {
      "epoch": 8.16557911908646,
      "grad_norm": 0.10302278399467468,
      "learning_rate": 4.953439623780643e-06,
      "loss": 0.0101,
      "num_input_tokens_seen": 108020368,
      "step": 50055
    },
    {
      "epoch": 8.166394779771615,
      "grad_norm": 0.44655972719192505,
      "learning_rate": 4.949187379921136e-06,
      "loss": 0.0458,
      "num_input_tokens_seen": 108030736,
      "step": 50060
    },
    {
      "epoch": 8.16721044045677,
      "grad_norm": 0.038387883454561234,
      "learning_rate": 4.9449367614822384e-06,
      "loss": 0.0027,
      "num_input_tokens_seen": 108041232,
      "step": 50065
    },
    {
      "epoch": 8.168026101141924,
      "grad_norm": 0.08896858245134354,
      "learning_rate": 4.940687768808525e-06,
      "loss": 0.0988,
      "num_input_tokens_seen": 108053264,
      "step": 50070
    },
    {
      "epoch": 8.16884176182708,
      "grad_norm": 2.4326529502868652,
      "learning_rate": 4.936440402244441e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 108063088,
      "step": 50075
    },
    {
      "epoch": 8.169657422512234,
      "grad_norm": 0.07476741820573807,
      "learning_rate": 4.932194662134298e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 108072688,
      "step": 50080
    },
    {
      "epoch": 8.17047308319739,
      "grad_norm": 2.7072935104370117,
      "learning_rate": 4.92795054882228e-06,
      "loss": 0.0825,
      "num_input_tokens_seen": 108085296,
      "step": 50085
    },
    {
      "epoch": 8.171288743882545,
      "grad_norm": 0.5258146524429321,
      "learning_rate": 4.9237080626524294e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 108096816,
      "step": 50090
    },
    {
      "epoch": 8.1721044045677,
      "grad_norm": 0.061414800584316254,
      "learning_rate": 4.919467203968675e-06,
      "loss": 0.1125,
      "num_input_tokens_seen": 108107280,
      "step": 50095
    },
    {
      "epoch": 8.172920065252855,
      "grad_norm": 0.07064393162727356,
      "learning_rate": 4.915227973114797e-06,
      "loss": 0.0509,
      "num_input_tokens_seen": 108117552,
      "step": 50100
    },
    {
      "epoch": 8.173735725938009,
      "grad_norm": 0.07800977677106857,
      "learning_rate": 4.910990370434449e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 108129264,
      "step": 50105
    },
    {
      "epoch": 8.174551386623165,
      "grad_norm": 0.19258436560630798,
      "learning_rate": 4.906754396271152e-06,
      "loss": 0.01,
      "num_input_tokens_seen": 108139472,
      "step": 50110
    },
    {
      "epoch": 8.17536704730832,
      "grad_norm": 0.9658310413360596,
      "learning_rate": 4.902520050968293e-06,
      "loss": 0.025,
      "num_input_tokens_seen": 108149712,
      "step": 50115
    },
    {
      "epoch": 8.176182707993474,
      "grad_norm": 0.07832033932209015,
      "learning_rate": 4.898287334869134e-06,
      "loss": 0.0234,
      "num_input_tokens_seen": 108160464,
      "step": 50120
    },
    {
      "epoch": 8.17699836867863,
      "grad_norm": 3.9769275188446045,
      "learning_rate": 4.8940562483168005e-06,
      "loss": 0.1665,
      "num_input_tokens_seen": 108171312,
      "step": 50125
    },
    {
      "epoch": 8.177814029363784,
      "grad_norm": 0.09289734810590744,
      "learning_rate": 4.889826791654281e-06,
      "loss": 0.1265,
      "num_input_tokens_seen": 108181616,
      "step": 50130
    },
    {
      "epoch": 8.17862969004894,
      "grad_norm": 0.02215544320642948,
      "learning_rate": 4.8855989652244415e-06,
      "loss": 0.2234,
      "num_input_tokens_seen": 108191792,
      "step": 50135
    },
    {
      "epoch": 8.179445350734095,
      "grad_norm": 0.12375655770301819,
      "learning_rate": 4.8813727693700104e-06,
      "loss": 0.1492,
      "num_input_tokens_seen": 108203216,
      "step": 50140
    },
    {
      "epoch": 8.18026101141925,
      "grad_norm": 0.08403193950653076,
      "learning_rate": 4.877148204433582e-06,
      "loss": 0.0063,
      "num_input_tokens_seen": 108214000,
      "step": 50145
    },
    {
      "epoch": 8.181076672104405,
      "grad_norm": 0.08748190850019455,
      "learning_rate": 4.872925270757623e-06,
      "loss": 0.1165,
      "num_input_tokens_seen": 108224528,
      "step": 50150
    },
    {
      "epoch": 8.181892332789559,
      "grad_norm": 0.26224926114082336,
      "learning_rate": 4.868703968684466e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 108235088,
      "step": 50155
    },
    {
      "epoch": 8.182707993474715,
      "grad_norm": 0.2896566092967987,
      "learning_rate": 4.86448429855631e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 108245616,
      "step": 50160
    },
    {
      "epoch": 8.18352365415987,
      "grad_norm": 3.337801694869995,
      "learning_rate": 4.860266260715221e-06,
      "loss": 0.1663,
      "num_input_tokens_seen": 108256144,
      "step": 50165
    },
    {
      "epoch": 8.184339314845024,
      "grad_norm": 0.033951111137866974,
      "learning_rate": 4.856049855503139e-06,
      "loss": 0.126,
      "num_input_tokens_seen": 108267248,
      "step": 50170
    },
    {
      "epoch": 8.18515497553018,
      "grad_norm": 0.1714457869529724,
      "learning_rate": 4.8518350832618655e-06,
      "loss": 0.1017,
      "num_input_tokens_seen": 108278128,
      "step": 50175
    },
    {
      "epoch": 8.185970636215334,
      "grad_norm": 0.08282965421676636,
      "learning_rate": 4.847621944333064e-06,
      "loss": 0.1976,
      "num_input_tokens_seen": 108290352,
      "step": 50180
    },
    {
      "epoch": 8.18678629690049,
      "grad_norm": 0.6131508350372314,
      "learning_rate": 4.8434104390582855e-06,
      "loss": 0.0069,
      "num_input_tokens_seen": 108301776,
      "step": 50185
    },
    {
      "epoch": 8.187601957585644,
      "grad_norm": 8.647669792175293,
      "learning_rate": 4.839200567778932e-06,
      "loss": 0.0964,
      "num_input_tokens_seen": 108312208,
      "step": 50190
    },
    {
      "epoch": 8.1884176182708,
      "grad_norm": 4.240434169769287,
      "learning_rate": 4.834992330836274e-06,
      "loss": 0.1322,
      "num_input_tokens_seen": 108322480,
      "step": 50195
    },
    {
      "epoch": 8.189233278955955,
      "grad_norm": 0.07553743571043015,
      "learning_rate": 4.8307857285714545e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 108333392,
      "step": 50200
    },
    {
      "epoch": 8.190048939641109,
      "grad_norm": 0.057867687195539474,
      "learning_rate": 4.826580761325475e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 108343856,
      "step": 50205
    },
    {
      "epoch": 8.190864600326265,
      "grad_norm": 0.07736869156360626,
      "learning_rate": 4.822377429439223e-06,
      "loss": 0.1104,
      "num_input_tokens_seen": 108355728,
      "step": 50210
    },
    {
      "epoch": 8.191680261011419,
      "grad_norm": 0.054170429706573486,
      "learning_rate": 4.818175733253438e-06,
      "loss": 0.2302,
      "num_input_tokens_seen": 108366736,
      "step": 50215
    },
    {
      "epoch": 8.192495921696574,
      "grad_norm": 6.451676368713379,
      "learning_rate": 4.813975673108731e-06,
      "loss": 0.0176,
      "num_input_tokens_seen": 108376912,
      "step": 50220
    },
    {
      "epoch": 8.19331158238173,
      "grad_norm": 0.06798990815877914,
      "learning_rate": 4.809777249345576e-06,
      "loss": 0.1984,
      "num_input_tokens_seen": 108387440,
      "step": 50225
    },
    {
      "epoch": 8.194127243066884,
      "grad_norm": 0.13363580405712128,
      "learning_rate": 4.8055804623043235e-06,
      "loss": 0.3014,
      "num_input_tokens_seen": 108397232,
      "step": 50230
    },
    {
      "epoch": 8.19494290375204,
      "grad_norm": 0.15777772665023804,
      "learning_rate": 4.801385312325182e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 108408016,
      "step": 50235
    },
    {
      "epoch": 8.195758564437194,
      "grad_norm": 6.402616500854492,
      "learning_rate": 4.7971917997482376e-06,
      "loss": 0.0928,
      "num_input_tokens_seen": 108419024,
      "step": 50240
    },
    {
      "epoch": 8.19657422512235,
      "grad_norm": 0.10083062946796417,
      "learning_rate": 4.79299992491343e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 108430320,
      "step": 50245
    },
    {
      "epoch": 8.197389885807505,
      "grad_norm": 0.05472959205508232,
      "learning_rate": 4.788809688160581e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 108441104,
      "step": 50250
    },
    {
      "epoch": 8.198205546492659,
      "grad_norm": 0.05518600344657898,
      "learning_rate": 4.784621089829366e-06,
      "loss": 0.1861,
      "num_input_tokens_seen": 108452656,
      "step": 50255
    },
    {
      "epoch": 8.199021207177815,
      "grad_norm": 3.555570602416992,
      "learning_rate": 4.780434130259339e-06,
      "loss": 0.1197,
      "num_input_tokens_seen": 108462992,
      "step": 50260
    },
    {
      "epoch": 8.199836867862969,
      "grad_norm": 0.02412850223481655,
      "learning_rate": 4.7762488097899154e-06,
      "loss": 0.0082,
      "num_input_tokens_seen": 108473744,
      "step": 50265
    },
    {
      "epoch": 8.200652528548124,
      "grad_norm": 0.2734960913658142,
      "learning_rate": 4.772065128760375e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 108485136,
      "step": 50270
    },
    {
      "epoch": 8.201468189233278,
      "grad_norm": 0.1171422004699707,
      "learning_rate": 4.767883087509872e-06,
      "loss": 0.0686,
      "num_input_tokens_seen": 108496720,
      "step": 50275
    },
    {
      "epoch": 8.202283849918434,
      "grad_norm": 4.667135238647461,
      "learning_rate": 4.763702686377425e-06,
      "loss": 0.0875,
      "num_input_tokens_seen": 108508016,
      "step": 50280
    },
    {
      "epoch": 8.20309951060359,
      "grad_norm": 3.5564005374908447,
      "learning_rate": 4.759523925701914e-06,
      "loss": 0.1315,
      "num_input_tokens_seen": 108517872,
      "step": 50285
    },
    {
      "epoch": 8.203915171288743,
      "grad_norm": 0.11621157079935074,
      "learning_rate": 4.7553468058220915e-06,
      "loss": 0.1008,
      "num_input_tokens_seen": 108527280,
      "step": 50290
    },
    {
      "epoch": 8.2047308319739,
      "grad_norm": 0.081459179520607,
      "learning_rate": 4.751171327076579e-06,
      "loss": 0.136,
      "num_input_tokens_seen": 108538192,
      "step": 50295
    },
    {
      "epoch": 8.205546492659053,
      "grad_norm": 4.702872276306152,
      "learning_rate": 4.746997489803853e-06,
      "loss": 0.1771,
      "num_input_tokens_seen": 108546960,
      "step": 50300
    },
    {
      "epoch": 8.206362153344209,
      "grad_norm": 0.1000506579875946,
      "learning_rate": 4.7428252943422794e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 108556912,
      "step": 50305
    },
    {
      "epoch": 8.207177814029365,
      "grad_norm": 0.04609856382012367,
      "learning_rate": 4.738654741030074e-06,
      "loss": 0.006,
      "num_input_tokens_seen": 108568144,
      "step": 50310
    },
    {
      "epoch": 8.207993474714518,
      "grad_norm": 0.059388112276792526,
      "learning_rate": 4.734485830205318e-06,
      "loss": 0.0026,
      "num_input_tokens_seen": 108578160,
      "step": 50315
    },
    {
      "epoch": 8.208809135399674,
      "grad_norm": 0.03651902452111244,
      "learning_rate": 4.730318562205965e-06,
      "loss": 0.0806,
      "num_input_tokens_seen": 108589296,
      "step": 50320
    },
    {
      "epoch": 8.209624796084828,
      "grad_norm": 0.17606563866138458,
      "learning_rate": 4.7261529373698404e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 108600240,
      "step": 50325
    },
    {
      "epoch": 8.210440456769984,
      "grad_norm": 0.2413724958896637,
      "learning_rate": 4.721988956034626e-06,
      "loss": 0.0703,
      "num_input_tokens_seen": 108611056,
      "step": 50330
    },
    {
      "epoch": 8.21125611745514,
      "grad_norm": 0.13980406522750854,
      "learning_rate": 4.717826618537874e-06,
      "loss": 0.161,
      "num_input_tokens_seen": 108620400,
      "step": 50335
    },
    {
      "epoch": 8.212071778140293,
      "grad_norm": 0.16237755119800568,
      "learning_rate": 4.713665925217009e-06,
      "loss": 0.0288,
      "num_input_tokens_seen": 108632432,
      "step": 50340
    },
    {
      "epoch": 8.21288743882545,
      "grad_norm": 5.084513187408447,
      "learning_rate": 4.709506876409317e-06,
      "loss": 0.283,
      "num_input_tokens_seen": 108642896,
      "step": 50345
    },
    {
      "epoch": 8.213703099510603,
      "grad_norm": 9.8268461227417,
      "learning_rate": 4.705349472451942e-06,
      "loss": 0.132,
      "num_input_tokens_seen": 108652816,
      "step": 50350
    },
    {
      "epoch": 8.214518760195759,
      "grad_norm": 0.09373872727155685,
      "learning_rate": 4.70119371368192e-06,
      "loss": 0.079,
      "num_input_tokens_seen": 108663792,
      "step": 50355
    },
    {
      "epoch": 8.215334420880913,
      "grad_norm": 0.08155173808336258,
      "learning_rate": 4.697039600436132e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 108674896,
      "step": 50360
    },
    {
      "epoch": 8.216150081566068,
      "grad_norm": 0.20419591665267944,
      "learning_rate": 4.6928871330513296e-06,
      "loss": 0.2251,
      "num_input_tokens_seen": 108685584,
      "step": 50365
    },
    {
      "epoch": 8.216965742251224,
      "grad_norm": 0.18128938972949982,
      "learning_rate": 4.6887363118641335e-06,
      "loss": 0.1375,
      "num_input_tokens_seen": 108696560,
      "step": 50370
    },
    {
      "epoch": 8.217781402936378,
      "grad_norm": 0.2932167947292328,
      "learning_rate": 4.68458713721103e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 108706512,
      "step": 50375
    },
    {
      "epoch": 8.218597063621534,
      "grad_norm": 4.024229049682617,
      "learning_rate": 4.680439609428372e-06,
      "loss": 0.1881,
      "num_input_tokens_seen": 108717808,
      "step": 50380
    },
    {
      "epoch": 8.219412724306688,
      "grad_norm": 0.6397793292999268,
      "learning_rate": 4.676293728852379e-06,
      "loss": 0.006,
      "num_input_tokens_seen": 108729040,
      "step": 50385
    },
    {
      "epoch": 8.220228384991843,
      "grad_norm": 0.0787268728017807,
      "learning_rate": 4.6721494958191395e-06,
      "loss": 0.115,
      "num_input_tokens_seen": 108739536,
      "step": 50390
    },
    {
      "epoch": 8.221044045676999,
      "grad_norm": 13.2654447555542,
      "learning_rate": 4.6680069106646014e-06,
      "loss": 0.0751,
      "num_input_tokens_seen": 108750640,
      "step": 50395
    },
    {
      "epoch": 8.221859706362153,
      "grad_norm": 0.34706342220306396,
      "learning_rate": 4.663865973724591e-06,
      "loss": 0.0423,
      "num_input_tokens_seen": 108760368,
      "step": 50400
    },
    {
      "epoch": 8.222675367047309,
      "grad_norm": 0.12569265067577362,
      "learning_rate": 4.659726685334786e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 108772336,
      "step": 50405
    },
    {
      "epoch": 8.223491027732463,
      "grad_norm": 4.785270690917969,
      "learning_rate": 4.655589045830735e-06,
      "loss": 0.1103,
      "num_input_tokens_seen": 108782800,
      "step": 50410
    },
    {
      "epoch": 8.224306688417618,
      "grad_norm": 0.02386106364428997,
      "learning_rate": 4.651453055547872e-06,
      "loss": 0.1051,
      "num_input_tokens_seen": 108794000,
      "step": 50415
    },
    {
      "epoch": 8.225122349102774,
      "grad_norm": 0.12528622150421143,
      "learning_rate": 4.647318714821469e-06,
      "loss": 0.0025,
      "num_input_tokens_seen": 108805392,
      "step": 50420
    },
    {
      "epoch": 8.225938009787928,
      "grad_norm": 1.0077322721481323,
      "learning_rate": 4.643186023986681e-06,
      "loss": 0.0081,
      "num_input_tokens_seen": 108817616,
      "step": 50425
    },
    {
      "epoch": 8.226753670473084,
      "grad_norm": 0.08704919368028641,
      "learning_rate": 4.639054983378521e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 108829360,
      "step": 50430
    },
    {
      "epoch": 8.227569331158238,
      "grad_norm": 2.131197929382324,
      "learning_rate": 4.634925593331876e-06,
      "loss": 0.1934,
      "num_input_tokens_seen": 108839696,
      "step": 50435
    },
    {
      "epoch": 8.228384991843393,
      "grad_norm": 0.19904682040214539,
      "learning_rate": 4.630797854181495e-06,
      "loss": 0.0344,
      "num_input_tokens_seen": 108851184,
      "step": 50440
    },
    {
      "epoch": 8.229200652528547,
      "grad_norm": 0.16750061511993408,
      "learning_rate": 4.626671766261992e-06,
      "loss": 0.2497,
      "num_input_tokens_seen": 108861744,
      "step": 50445
    },
    {
      "epoch": 8.230016313213703,
      "grad_norm": 0.1557835340499878,
      "learning_rate": 4.622547329907848e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 108872816,
      "step": 50450
    },
    {
      "epoch": 8.230831973898859,
      "grad_norm": 0.040223173797130585,
      "learning_rate": 4.618424545453409e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 108884688,
      "step": 50455
    },
    {
      "epoch": 8.231647634584013,
      "grad_norm": 14.621254920959473,
      "learning_rate": 4.6143034132328955e-06,
      "loss": 0.0825,
      "num_input_tokens_seen": 108896208,
      "step": 50460
    },
    {
      "epoch": 8.232463295269168,
      "grad_norm": 0.15322542190551758,
      "learning_rate": 4.610183933580381e-06,
      "loss": 0.1539,
      "num_input_tokens_seen": 108907536,
      "step": 50465
    },
    {
      "epoch": 8.233278955954322,
      "grad_norm": 0.017540767788887024,
      "learning_rate": 4.606066106829815e-06,
      "loss": 0.0825,
      "num_input_tokens_seen": 108917360,
      "step": 50470
    },
    {
      "epoch": 8.234094616639478,
      "grad_norm": 0.05903009697794914,
      "learning_rate": 4.601949933315009e-06,
      "loss": 0.2029,
      "num_input_tokens_seen": 108927952,
      "step": 50475
    },
    {
      "epoch": 8.234910277324634,
      "grad_norm": 0.2780179977416992,
      "learning_rate": 4.597835413369639e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 108939248,
      "step": 50480
    },
    {
      "epoch": 8.235725938009788,
      "grad_norm": 0.2921516001224518,
      "learning_rate": 4.593722547327248e-06,
      "loss": 0.1172,
      "num_input_tokens_seen": 108950832,
      "step": 50485
    },
    {
      "epoch": 8.236541598694943,
      "grad_norm": 0.1570664346218109,
      "learning_rate": 4.589611335521249e-06,
      "loss": 0.0138,
      "num_input_tokens_seen": 108961488,
      "step": 50490
    },
    {
      "epoch": 8.237357259380097,
      "grad_norm": 0.3389540910720825,
      "learning_rate": 4.585501778284912e-06,
      "loss": 0.1036,
      "num_input_tokens_seen": 108971888,
      "step": 50495
    },
    {
      "epoch": 8.238172920065253,
      "grad_norm": 0.07394928485155106,
      "learning_rate": 4.581393875951387e-06,
      "loss": 0.0891,
      "num_input_tokens_seen": 108983856,
      "step": 50500
    },
    {
      "epoch": 8.238988580750409,
      "grad_norm": 0.04614948108792305,
      "learning_rate": 4.577287628853677e-06,
      "loss": 0.3118,
      "num_input_tokens_seen": 108995408,
      "step": 50505
    },
    {
      "epoch": 8.239804241435563,
      "grad_norm": 0.04369710758328438,
      "learning_rate": 4.5731830373246574e-06,
      "loss": 0.2083,
      "num_input_tokens_seen": 109006736,
      "step": 50510
    },
    {
      "epoch": 8.240619902120718,
      "grad_norm": 0.19078224897384644,
      "learning_rate": 4.5690801016970655e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 109016528,
      "step": 50515
    },
    {
      "epoch": 8.241435562805872,
      "grad_norm": 0.15958210825920105,
      "learning_rate": 4.564978822303498e-06,
      "loss": 0.0959,
      "num_input_tokens_seen": 109028688,
      "step": 50520
    },
    {
      "epoch": 8.242251223491028,
      "grad_norm": 0.040741484612226486,
      "learning_rate": 4.560879199476442e-06,
      "loss": 0.1921,
      "num_input_tokens_seen": 109039792,
      "step": 50525
    },
    {
      "epoch": 8.243066884176184,
      "grad_norm": 0.07980673760175705,
      "learning_rate": 4.5567812335482244e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 109050416,
      "step": 50530
    },
    {
      "epoch": 8.243882544861338,
      "grad_norm": 0.1288427859544754,
      "learning_rate": 4.5526849248510475e-06,
      "loss": 0.0647,
      "num_input_tokens_seen": 109060816,
      "step": 50535
    },
    {
      "epoch": 8.244698205546493,
      "grad_norm": 0.0687427669763565,
      "learning_rate": 4.548590273716979e-06,
      "loss": 0.0029,
      "num_input_tokens_seen": 109071920,
      "step": 50540
    },
    {
      "epoch": 8.245513866231647,
      "grad_norm": 0.36271336674690247,
      "learning_rate": 4.5444972804779525e-06,
      "loss": 0.2503,
      "num_input_tokens_seen": 109082512,
      "step": 50545
    },
    {
      "epoch": 8.246329526916803,
      "grad_norm": 0.02701878547668457,
      "learning_rate": 4.540405945465767e-06,
      "loss": 0.2103,
      "num_input_tokens_seen": 109093744,
      "step": 50550
    },
    {
      "epoch": 8.247145187601957,
      "grad_norm": 0.27419909834861755,
      "learning_rate": 4.536316269012086e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 109104368,
      "step": 50555
    },
    {
      "epoch": 8.247960848287113,
      "grad_norm": 3.8257272243499756,
      "learning_rate": 4.532228251448439e-06,
      "loss": 0.2266,
      "num_input_tokens_seen": 109115920,
      "step": 50560
    },
    {
      "epoch": 8.248776508972268,
      "grad_norm": 0.47641947865486145,
      "learning_rate": 4.528141893106225e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 109127088,
      "step": 50565
    },
    {
      "epoch": 8.249592169657422,
      "grad_norm": 0.05158061906695366,
      "learning_rate": 4.5240571943167e-06,
      "loss": 0.065,
      "num_input_tokens_seen": 109138736,
      "step": 50570
    },
    {
      "epoch": 8.250407830342578,
      "grad_norm": 0.043185245245695114,
      "learning_rate": 4.519974155410992e-06,
      "loss": 0.1124,
      "num_input_tokens_seen": 109149488,
      "step": 50575
    },
    {
      "epoch": 8.251223491027732,
      "grad_norm": 0.44439175724983215,
      "learning_rate": 4.515892776720096e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 109160656,
      "step": 50580
    },
    {
      "epoch": 8.252039151712887,
      "grad_norm": 0.0924413651227951,
      "learning_rate": 4.5118130585748655e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 109171312,
      "step": 50585
    },
    {
      "epoch": 8.252854812398043,
      "grad_norm": 0.20112235844135284,
      "learning_rate": 4.507735001306024e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 109182384,
      "step": 50590
    },
    {
      "epoch": 8.253670473083197,
      "grad_norm": 0.03435681015253067,
      "learning_rate": 4.503658605244163e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 109191792,
      "step": 50595
    },
    {
      "epoch": 8.254486133768353,
      "grad_norm": 0.17052924633026123,
      "learning_rate": 4.499583870719728e-06,
      "loss": 0.12,
      "num_input_tokens_seen": 109202960,
      "step": 50600
    },
    {
      "epoch": 8.255301794453507,
      "grad_norm": 0.034675903618335724,
      "learning_rate": 4.495510798063046e-06,
      "loss": 0.0059,
      "num_input_tokens_seen": 109214288,
      "step": 50605
    },
    {
      "epoch": 8.256117455138662,
      "grad_norm": 0.029451005160808563,
      "learning_rate": 4.4914393876042984e-06,
      "loss": 0.092,
      "num_input_tokens_seen": 109225328,
      "step": 50610
    },
    {
      "epoch": 8.256933115823816,
      "grad_norm": 0.1021382138133049,
      "learning_rate": 4.48736963967353e-06,
      "loss": 0.1346,
      "num_input_tokens_seen": 109236016,
      "step": 50615
    },
    {
      "epoch": 8.257748776508972,
      "grad_norm": 0.2410012185573578,
      "learning_rate": 4.483301554600655e-06,
      "loss": 0.2154,
      "num_input_tokens_seen": 109247344,
      "step": 50620
    },
    {
      "epoch": 8.258564437194128,
      "grad_norm": 0.19051925837993622,
      "learning_rate": 4.479235132715462e-06,
      "loss": 0.0198,
      "num_input_tokens_seen": 109258864,
      "step": 50625
    },
    {
      "epoch": 8.259380097879282,
      "grad_norm": 0.040067970752716064,
      "learning_rate": 4.4751703743475895e-06,
      "loss": 0.0454,
      "num_input_tokens_seen": 109270192,
      "step": 50630
    },
    {
      "epoch": 8.260195758564437,
      "grad_norm": 0.09321948140859604,
      "learning_rate": 4.47110727982655e-06,
      "loss": 0.0029,
      "num_input_tokens_seen": 109280016,
      "step": 50635
    },
    {
      "epoch": 8.261011419249591,
      "grad_norm": 0.054231688380241394,
      "learning_rate": 4.467045849481716e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 109290256,
      "step": 50640
    },
    {
      "epoch": 8.261827079934747,
      "grad_norm": 0.44992342591285706,
      "learning_rate": 4.462986083642329e-06,
      "loss": 0.0834,
      "num_input_tokens_seen": 109300208,
      "step": 50645
    },
    {
      "epoch": 8.262642740619903,
      "grad_norm": 0.11483272165060043,
      "learning_rate": 4.4589279826374955e-06,
      "loss": 0.1073,
      "num_input_tokens_seen": 109310960,
      "step": 50650
    },
    {
      "epoch": 8.263458401305057,
      "grad_norm": 0.13658924400806427,
      "learning_rate": 4.454871546796182e-06,
      "loss": 0.0845,
      "num_input_tokens_seen": 109321040,
      "step": 50655
    },
    {
      "epoch": 8.264274061990212,
      "grad_norm": 0.11663892865180969,
      "learning_rate": 4.4508167764472254e-06,
      "loss": 0.1995,
      "num_input_tokens_seen": 109331056,
      "step": 50660
    },
    {
      "epoch": 8.265089722675366,
      "grad_norm": 0.16184276342391968,
      "learning_rate": 4.446763671919321e-06,
      "loss": 0.1159,
      "num_input_tokens_seen": 109340816,
      "step": 50665
    },
    {
      "epoch": 8.265905383360522,
      "grad_norm": 0.14844438433647156,
      "learning_rate": 4.442712233541046e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 109351984,
      "step": 50670
    },
    {
      "epoch": 8.266721044045678,
      "grad_norm": 0.35642722249031067,
      "learning_rate": 4.438662461640825e-06,
      "loss": 0.0729,
      "num_input_tokens_seen": 109361616,
      "step": 50675
    },
    {
      "epoch": 8.267536704730832,
      "grad_norm": 0.12397274374961853,
      "learning_rate": 4.4346143565469485e-06,
      "loss": 0.0027,
      "num_input_tokens_seen": 109372240,
      "step": 50680
    },
    {
      "epoch": 8.268352365415987,
      "grad_norm": 0.10330452769994736,
      "learning_rate": 4.430567918587583e-06,
      "loss": 0.0067,
      "num_input_tokens_seen": 109383120,
      "step": 50685
    },
    {
      "epoch": 8.269168026101141,
      "grad_norm": 0.07007979601621628,
      "learning_rate": 4.42652314809075e-06,
      "loss": 0.0456,
      "num_input_tokens_seen": 109391952,
      "step": 50690
    },
    {
      "epoch": 8.269983686786297,
      "grad_norm": 0.09421925991773605,
      "learning_rate": 4.4224800453843394e-06,
      "loss": 0.041,
      "num_input_tokens_seen": 109402576,
      "step": 50695
    },
    {
      "epoch": 8.270799347471453,
      "grad_norm": 5.5964508056640625,
      "learning_rate": 4.418438610796105e-06,
      "loss": 0.1264,
      "num_input_tokens_seen": 109412976,
      "step": 50700
    },
    {
      "epoch": 8.271615008156607,
      "grad_norm": 5.199951171875,
      "learning_rate": 4.414398844653666e-06,
      "loss": 0.0822,
      "num_input_tokens_seen": 109425040,
      "step": 50705
    },
    {
      "epoch": 8.272430668841762,
      "grad_norm": 0.1465538889169693,
      "learning_rate": 4.410360747284508e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 109436592,
      "step": 50710
    },
    {
      "epoch": 8.273246329526916,
      "grad_norm": 3.4535036087036133,
      "learning_rate": 4.406324319015978e-06,
      "loss": 0.1077,
      "num_input_tokens_seen": 109447184,
      "step": 50715
    },
    {
      "epoch": 8.274061990212072,
      "grad_norm": 0.17683878540992737,
      "learning_rate": 4.4022895601752905e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 109457616,
      "step": 50720
    },
    {
      "epoch": 8.274877650897226,
      "grad_norm": 0.07014454901218414,
      "learning_rate": 4.398256471089518e-06,
      "loss": 0.0019,
      "num_input_tokens_seen": 109466896,
      "step": 50725
    },
    {
      "epoch": 8.275693311582382,
      "grad_norm": 22.82807159423828,
      "learning_rate": 4.394225052085613e-06,
      "loss": 0.0455,
      "num_input_tokens_seen": 109478800,
      "step": 50730
    },
    {
      "epoch": 8.276508972267537,
      "grad_norm": 0.24154672026634216,
      "learning_rate": 4.390195303490377e-06,
      "loss": 0.0285,
      "num_input_tokens_seen": 109488272,
      "step": 50735
    },
    {
      "epoch": 8.277324632952691,
      "grad_norm": 0.15540587902069092,
      "learning_rate": 4.3861672256304835e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 109499248,
      "step": 50740
    },
    {
      "epoch": 8.278140293637847,
      "grad_norm": 0.19240988790988922,
      "learning_rate": 4.382140818832467e-06,
      "loss": 0.1734,
      "num_input_tokens_seen": 109509584,
      "step": 50745
    },
    {
      "epoch": 8.278955954323001,
      "grad_norm": 5.04918909072876,
      "learning_rate": 4.378116083422732e-06,
      "loss": 0.2013,
      "num_input_tokens_seen": 109520656,
      "step": 50750
    },
    {
      "epoch": 8.279771615008157,
      "grad_norm": 0.4211946427822113,
      "learning_rate": 4.374093019727541e-06,
      "loss": 0.0063,
      "num_input_tokens_seen": 109531600,
      "step": 50755
    },
    {
      "epoch": 8.280587275693312,
      "grad_norm": 0.3730059564113617,
      "learning_rate": 4.370071628073025e-06,
      "loss": 0.1024,
      "num_input_tokens_seen": 109541936,
      "step": 50760
    },
    {
      "epoch": 8.281402936378466,
      "grad_norm": 0.16246718168258667,
      "learning_rate": 4.366051908785177e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 109552336,
      "step": 50765
    },
    {
      "epoch": 8.282218597063622,
      "grad_norm": 0.14463460445404053,
      "learning_rate": 4.3620338621898575e-06,
      "loss": 0.2052,
      "num_input_tokens_seen": 109563376,
      "step": 50770
    },
    {
      "epoch": 8.283034257748776,
      "grad_norm": 8.738451957702637,
      "learning_rate": 4.35801748861279e-06,
      "loss": 0.1017,
      "num_input_tokens_seen": 109575216,
      "step": 50775
    },
    {
      "epoch": 8.283849918433932,
      "grad_norm": 0.09802068769931793,
      "learning_rate": 4.354002788379558e-06,
      "loss": 0.0492,
      "num_input_tokens_seen": 109585904,
      "step": 50780
    },
    {
      "epoch": 8.284665579119087,
      "grad_norm": 0.14776861667633057,
      "learning_rate": 4.34998976181562e-06,
      "loss": 0.2586,
      "num_input_tokens_seen": 109597936,
      "step": 50785
    },
    {
      "epoch": 8.285481239804241,
      "grad_norm": 0.05964956060051918,
      "learning_rate": 4.345978409246287e-06,
      "loss": 0.081,
      "num_input_tokens_seen": 109609424,
      "step": 50790
    },
    {
      "epoch": 8.286296900489397,
      "grad_norm": 0.0592818446457386,
      "learning_rate": 4.341968730996743e-06,
      "loss": 0.0111,
      "num_input_tokens_seen": 109621488,
      "step": 50795
    },
    {
      "epoch": 8.28711256117455,
      "grad_norm": 1.5230180025100708,
      "learning_rate": 4.337960727392032e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 109631696,
      "step": 50800
    },
    {
      "epoch": 8.287928221859707,
      "grad_norm": 1.5851444005966187,
      "learning_rate": 4.333954398757054e-06,
      "loss": 0.272,
      "num_input_tokens_seen": 109642544,
      "step": 50805
    },
    {
      "epoch": 8.28874388254486,
      "grad_norm": 0.3063996434211731,
      "learning_rate": 4.329949745416598e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 109652752,
      "step": 50810
    },
    {
      "epoch": 8.289559543230016,
      "grad_norm": 2.3268768787384033,
      "learning_rate": 4.325946767695297e-06,
      "loss": 0.0708,
      "num_input_tokens_seen": 109664496,
      "step": 50815
    },
    {
      "epoch": 8.290375203915172,
      "grad_norm": 0.16275164484977722,
      "learning_rate": 4.321945465917646e-06,
      "loss": 0.1863,
      "num_input_tokens_seen": 109676752,
      "step": 50820
    },
    {
      "epoch": 8.291190864600326,
      "grad_norm": 0.11027777194976807,
      "learning_rate": 4.317945840408019e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 109687792,
      "step": 50825
    },
    {
      "epoch": 8.292006525285482,
      "grad_norm": 0.08528872579336166,
      "learning_rate": 4.313947891490638e-06,
      "loss": 0.0579,
      "num_input_tokens_seen": 109698416,
      "step": 50830
    },
    {
      "epoch": 8.292822185970635,
      "grad_norm": 0.08500778675079346,
      "learning_rate": 4.309951619489597e-06,
      "loss": 0.1109,
      "num_input_tokens_seen": 109709488,
      "step": 50835
    },
    {
      "epoch": 8.293637846655791,
      "grad_norm": 0.14026764035224915,
      "learning_rate": 4.3059570247288624e-06,
      "loss": 0.1589,
      "num_input_tokens_seen": 109720912,
      "step": 50840
    },
    {
      "epoch": 8.294453507340947,
      "grad_norm": 0.07855894416570663,
      "learning_rate": 4.301964107532255e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 109731600,
      "step": 50845
    },
    {
      "epoch": 8.2952691680261,
      "grad_norm": 0.10529959201812744,
      "learning_rate": 4.297972868223457e-06,
      "loss": 0.2531,
      "num_input_tokens_seen": 109742064,
      "step": 50850
    },
    {
      "epoch": 8.296084828711257,
      "grad_norm": 0.0845145657658577,
      "learning_rate": 4.293983307126018e-06,
      "loss": 0.0251,
      "num_input_tokens_seen": 109752848,
      "step": 50855
    },
    {
      "epoch": 8.29690048939641,
      "grad_norm": 0.08708660304546356,
      "learning_rate": 4.289995424563353e-06,
      "loss": 0.0059,
      "num_input_tokens_seen": 109763728,
      "step": 50860
    },
    {
      "epoch": 8.297716150081566,
      "grad_norm": 0.2024458646774292,
      "learning_rate": 4.286009220858742e-06,
      "loss": 0.288,
      "num_input_tokens_seen": 109774864,
      "step": 50865
    },
    {
      "epoch": 8.298531810766722,
      "grad_norm": 0.178129643201828,
      "learning_rate": 4.282024696335324e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 109785296,
      "step": 50870
    },
    {
      "epoch": 8.299347471451876,
      "grad_norm": 3.1330602169036865,
      "learning_rate": 4.278041851316106e-06,
      "loss": 0.0069,
      "num_input_tokens_seen": 109795856,
      "step": 50875
    },
    {
      "epoch": 8.300163132137031,
      "grad_norm": 3.9104366302490234,
      "learning_rate": 4.274060686123959e-06,
      "loss": 0.1223,
      "num_input_tokens_seen": 109806480,
      "step": 50880
    },
    {
      "epoch": 8.300978792822185,
      "grad_norm": 0.0885840579867363,
      "learning_rate": 4.270081201081613e-06,
      "loss": 0.0029,
      "num_input_tokens_seen": 109817680,
      "step": 50885
    },
    {
      "epoch": 8.301794453507341,
      "grad_norm": 0.03894002363085747,
      "learning_rate": 4.2661033965116695e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 109829616,
      "step": 50890
    },
    {
      "epoch": 8.302610114192497,
      "grad_norm": 0.11713149398565292,
      "learning_rate": 4.2621272727365875e-06,
      "loss": 0.0605,
      "num_input_tokens_seen": 109839632,
      "step": 50895
    },
    {
      "epoch": 8.30342577487765,
      "grad_norm": 0.26046085357666016,
      "learning_rate": 4.2581528300786906e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 109851344,
      "step": 50900
    },
    {
      "epoch": 8.304241435562806,
      "grad_norm": 0.04204858839511871,
      "learning_rate": 4.2541800688601696e-06,
      "loss": 0.1063,
      "num_input_tokens_seen": 109861072,
      "step": 50905
    },
    {
      "epoch": 8.30505709624796,
      "grad_norm": 0.10443831980228424,
      "learning_rate": 4.250208989403073e-06,
      "loss": 0.1402,
      "num_input_tokens_seen": 109872144,
      "step": 50910
    },
    {
      "epoch": 8.305872756933116,
      "grad_norm": 0.14044144749641418,
      "learning_rate": 4.2462395920293215e-06,
      "loss": 0.0688,
      "num_input_tokens_seen": 109883152,
      "step": 50915
    },
    {
      "epoch": 8.30668841761827,
      "grad_norm": 0.03933459892868996,
      "learning_rate": 4.242271877060691e-06,
      "loss": 0.0094,
      "num_input_tokens_seen": 109894672,
      "step": 50920
    },
    {
      "epoch": 8.307504078303426,
      "grad_norm": 0.3973071873188019,
      "learning_rate": 4.238305844818827e-06,
      "loss": 0.1753,
      "num_input_tokens_seen": 109905456,
      "step": 50925
    },
    {
      "epoch": 8.308319738988581,
      "grad_norm": 4.436182022094727,
      "learning_rate": 4.234341495625233e-06,
      "loss": 0.3647,
      "num_input_tokens_seen": 109915024,
      "step": 50930
    },
    {
      "epoch": 8.309135399673735,
      "grad_norm": 0.1660321056842804,
      "learning_rate": 4.230378829801282e-06,
      "loss": 0.1378,
      "num_input_tokens_seen": 109923728,
      "step": 50935
    },
    {
      "epoch": 8.309951060358891,
      "grad_norm": 0.21303358674049377,
      "learning_rate": 4.226417847668201e-06,
      "loss": 0.0951,
      "num_input_tokens_seen": 109934128,
      "step": 50940
    },
    {
      "epoch": 8.310766721044045,
      "grad_norm": 0.03064492531120777,
      "learning_rate": 4.222458549547101e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 109945072,
      "step": 50945
    },
    {
      "epoch": 8.3115823817292,
      "grad_norm": 0.3193550705909729,
      "learning_rate": 4.218500935758935e-06,
      "loss": 0.1199,
      "num_input_tokens_seen": 109955184,
      "step": 50950
    },
    {
      "epoch": 8.312398042414356,
      "grad_norm": 0.0818767175078392,
      "learning_rate": 4.214545006624526e-06,
      "loss": 0.1143,
      "num_input_tokens_seen": 109965232,
      "step": 50955
    },
    {
      "epoch": 8.31321370309951,
      "grad_norm": 0.04161890223622322,
      "learning_rate": 4.210590762464564e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 109976944,
      "step": 50960
    },
    {
      "epoch": 8.314029363784666,
      "grad_norm": 0.04804445430636406,
      "learning_rate": 4.206638203599597e-06,
      "loss": 0.1408,
      "num_input_tokens_seen": 109987504,
      "step": 50965
    },
    {
      "epoch": 8.31484502446982,
      "grad_norm": 0.11469604074954987,
      "learning_rate": 4.202687330350044e-06,
      "loss": 0.0958,
      "num_input_tokens_seen": 109998160,
      "step": 50970
    },
    {
      "epoch": 8.315660685154976,
      "grad_norm": 0.058986835181713104,
      "learning_rate": 4.1987381430361735e-06,
      "loss": 0.0617,
      "num_input_tokens_seen": 110008464,
      "step": 50975
    },
    {
      "epoch": 8.31647634584013,
      "grad_norm": 5.525466442108154,
      "learning_rate": 4.194790641978141e-06,
      "loss": 0.1832,
      "num_input_tokens_seen": 110020080,
      "step": 50980
    },
    {
      "epoch": 8.317292006525285,
      "grad_norm": 0.0891265943646431,
      "learning_rate": 4.1908448274959436e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 110029104,
      "step": 50985
    },
    {
      "epoch": 8.318107667210441,
      "grad_norm": 0.3449834883213043,
      "learning_rate": 4.186900699909446e-06,
      "loss": 0.0549,
      "num_input_tokens_seen": 110040528,
      "step": 50990
    },
    {
      "epoch": 8.318923327895595,
      "grad_norm": 0.04940696805715561,
      "learning_rate": 4.182958259538386e-06,
      "loss": 0.0297,
      "num_input_tokens_seen": 110049744,
      "step": 50995
    },
    {
      "epoch": 8.31973898858075,
      "grad_norm": 0.1019030436873436,
      "learning_rate": 4.179017506702351e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 110061680,
      "step": 51000
    },
    {
      "epoch": 8.320554649265905,
      "grad_norm": 0.11362658441066742,
      "learning_rate": 4.1750784417208065e-06,
      "loss": 0.0953,
      "num_input_tokens_seen": 110072880,
      "step": 51005
    },
    {
      "epoch": 8.32137030995106,
      "grad_norm": 0.036861419677734375,
      "learning_rate": 4.171141064913061e-06,
      "loss": 0.1668,
      "num_input_tokens_seen": 110084464,
      "step": 51010
    },
    {
      "epoch": 8.322185970636216,
      "grad_norm": 0.1589759886264801,
      "learning_rate": 4.16720537659831e-06,
      "loss": 0.0877,
      "num_input_tokens_seen": 110095536,
      "step": 51015
    },
    {
      "epoch": 8.32300163132137,
      "grad_norm": 0.09122513979673386,
      "learning_rate": 4.1632713770955956e-06,
      "loss": 0.149,
      "num_input_tokens_seen": 110104496,
      "step": 51020
    },
    {
      "epoch": 8.323817292006526,
      "grad_norm": 0.020270230248570442,
      "learning_rate": 4.159339066723827e-06,
      "loss": 0.0373,
      "num_input_tokens_seen": 110115888,
      "step": 51025
    },
    {
      "epoch": 8.32463295269168,
      "grad_norm": 0.06882867217063904,
      "learning_rate": 4.155408445801779e-06,
      "loss": 0.0067,
      "num_input_tokens_seen": 110125104,
      "step": 51030
    },
    {
      "epoch": 8.325448613376835,
      "grad_norm": 0.05521862953901291,
      "learning_rate": 4.151479514648085e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 110135728,
      "step": 51035
    },
    {
      "epoch": 8.326264274061991,
      "grad_norm": 0.1626533567905426,
      "learning_rate": 4.147552273581248e-06,
      "loss": 0.0091,
      "num_input_tokens_seen": 110146832,
      "step": 51040
    },
    {
      "epoch": 8.327079934747145,
      "grad_norm": 0.17563822865486145,
      "learning_rate": 4.143626722919619e-06,
      "loss": 0.1437,
      "num_input_tokens_seen": 110157520,
      "step": 51045
    },
    {
      "epoch": 8.3278955954323,
      "grad_norm": 0.06099024415016174,
      "learning_rate": 4.139702862981443e-06,
      "loss": 0.1121,
      "num_input_tokens_seen": 110168688,
      "step": 51050
    },
    {
      "epoch": 8.328711256117455,
      "grad_norm": 0.0700126588344574,
      "learning_rate": 4.135780694084793e-06,
      "loss": 0.033,
      "num_input_tokens_seen": 110179056,
      "step": 51055
    },
    {
      "epoch": 8.32952691680261,
      "grad_norm": 0.08973924815654755,
      "learning_rate": 4.131860216547623e-06,
      "loss": 0.0089,
      "num_input_tokens_seen": 110190160,
      "step": 51060
    },
    {
      "epoch": 8.330342577487766,
      "grad_norm": 0.04193677008152008,
      "learning_rate": 4.127941430687751e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 110201424,
      "step": 51065
    },
    {
      "epoch": 8.33115823817292,
      "grad_norm": 0.06412708014249802,
      "learning_rate": 4.1240243368228485e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 110212304,
      "step": 51070
    },
    {
      "epoch": 8.331973898858076,
      "grad_norm": 0.16163991391658783,
      "learning_rate": 4.120108935270459e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 110223216,
      "step": 51075
    },
    {
      "epoch": 8.33278955954323,
      "grad_norm": 0.03767704963684082,
      "learning_rate": 4.11619522634798e-06,
      "loss": 0.3622,
      "num_input_tokens_seen": 110234160,
      "step": 51080
    },
    {
      "epoch": 8.333605220228385,
      "grad_norm": 0.2308705896139145,
      "learning_rate": 4.11228321037268e-06,
      "loss": 0.1111,
      "num_input_tokens_seen": 110245264,
      "step": 51085
    },
    {
      "epoch": 8.33442088091354,
      "grad_norm": 0.10310493409633636,
      "learning_rate": 4.108372887661688e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 110256560,
      "step": 51090
    },
    {
      "epoch": 8.335236541598695,
      "grad_norm": 0.3172813653945923,
      "learning_rate": 4.10446425853199e-06,
      "loss": 0.2201,
      "num_input_tokens_seen": 110267920,
      "step": 51095
    },
    {
      "epoch": 8.33605220228385,
      "grad_norm": 0.053970981389284134,
      "learning_rate": 4.100557323300444e-06,
      "loss": 0.294,
      "num_input_tokens_seen": 110278032,
      "step": 51100
    },
    {
      "epoch": 8.336867862969005,
      "grad_norm": 0.13745872676372528,
      "learning_rate": 4.096652082283764e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 110289232,
      "step": 51105
    },
    {
      "epoch": 8.33768352365416,
      "grad_norm": 0.6048405766487122,
      "learning_rate": 4.092748535798527e-06,
      "loss": 0.0346,
      "num_input_tokens_seen": 110299664,
      "step": 51110
    },
    {
      "epoch": 8.338499184339314,
      "grad_norm": 0.15191605687141418,
      "learning_rate": 4.088846684161177e-06,
      "loss": 0.1041,
      "num_input_tokens_seen": 110310480,
      "step": 51115
    },
    {
      "epoch": 8.33931484502447,
      "grad_norm": 0.08844520896673203,
      "learning_rate": 4.0849465276880105e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 110322640,
      "step": 51120
    },
    {
      "epoch": 8.340130505709626,
      "grad_norm": 0.06009729579091072,
      "learning_rate": 4.081048066695209e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 110334992,
      "step": 51125
    },
    {
      "epoch": 8.34094616639478,
      "grad_norm": 0.10387653857469559,
      "learning_rate": 4.077151301498791e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 110345616,
      "step": 51130
    },
    {
      "epoch": 8.341761827079935,
      "grad_norm": 0.2893862724304199,
      "learning_rate": 4.073256232414649e-06,
      "loss": 0.0067,
      "num_input_tokens_seen": 110356752,
      "step": 51135
    },
    {
      "epoch": 8.34257748776509,
      "grad_norm": 0.06195401772856712,
      "learning_rate": 4.069362859758541e-06,
      "loss": 0.0081,
      "num_input_tokens_seen": 110367568,
      "step": 51140
    },
    {
      "epoch": 8.343393148450245,
      "grad_norm": 0.12342137843370438,
      "learning_rate": 4.065471183846079e-06,
      "loss": 0.0098,
      "num_input_tokens_seen": 110378576,
      "step": 51145
    },
    {
      "epoch": 8.3442088091354,
      "grad_norm": 0.15264281630516052,
      "learning_rate": 4.061581204992742e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 110389072,
      "step": 51150
    },
    {
      "epoch": 8.345024469820554,
      "grad_norm": 0.1557842642068863,
      "learning_rate": 4.057692923513867e-06,
      "loss": 0.1976,
      "num_input_tokens_seen": 110400464,
      "step": 51155
    },
    {
      "epoch": 8.34584013050571,
      "grad_norm": 0.06787768751382828,
      "learning_rate": 4.0538063397246725e-06,
      "loss": 0.1965,
      "num_input_tokens_seen": 110411728,
      "step": 51160
    },
    {
      "epoch": 8.346655791190864,
      "grad_norm": 0.08586647361516953,
      "learning_rate": 4.049921453940214e-06,
      "loss": 0.0123,
      "num_input_tokens_seen": 110422480,
      "step": 51165
    },
    {
      "epoch": 8.34747145187602,
      "grad_norm": 0.09865080565214157,
      "learning_rate": 4.046038266475421e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 110433680,
      "step": 51170
    },
    {
      "epoch": 8.348287112561174,
      "grad_norm": 0.14197717607021332,
      "learning_rate": 4.0421567776450895e-06,
      "loss": 0.0137,
      "num_input_tokens_seen": 110442512,
      "step": 51175
    },
    {
      "epoch": 8.34910277324633,
      "grad_norm": 0.08103757351636887,
      "learning_rate": 4.038276987763864e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 110454032,
      "step": 51180
    },
    {
      "epoch": 8.349918433931485,
      "grad_norm": 4.606961727142334,
      "learning_rate": 4.034398897146269e-06,
      "loss": 0.1384,
      "num_input_tokens_seen": 110464432,
      "step": 51185
    },
    {
      "epoch": 8.350734094616639,
      "grad_norm": 0.07641594856977463,
      "learning_rate": 4.0305225061066735e-06,
      "loss": 0.1245,
      "num_input_tokens_seen": 110476176,
      "step": 51190
    },
    {
      "epoch": 8.351549755301795,
      "grad_norm": 0.10787976533174515,
      "learning_rate": 4.026647814959325e-06,
      "loss": 0.1265,
      "num_input_tokens_seen": 110486800,
      "step": 51195
    },
    {
      "epoch": 8.352365415986949,
      "grad_norm": 0.13637679815292358,
      "learning_rate": 4.022774824018321e-06,
      "loss": 0.2487,
      "num_input_tokens_seen": 110498448,
      "step": 51200
    },
    {
      "epoch": 8.353181076672104,
      "grad_norm": 0.05383555591106415,
      "learning_rate": 4.018903533597629e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 110510064,
      "step": 51205
    },
    {
      "epoch": 8.35399673735726,
      "grad_norm": 26.989179611206055,
      "learning_rate": 4.015033944011071e-06,
      "loss": 0.0466,
      "num_input_tokens_seen": 110522352,
      "step": 51210
    },
    {
      "epoch": 8.354812398042414,
      "grad_norm": 0.055170685052871704,
      "learning_rate": 4.011166055572338e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 110534640,
      "step": 51215
    },
    {
      "epoch": 8.35562805872757,
      "grad_norm": 0.13349996507167816,
      "learning_rate": 4.007299868594983e-06,
      "loss": 0.0754,
      "num_input_tokens_seen": 110546608,
      "step": 51220
    },
    {
      "epoch": 8.356443719412724,
      "grad_norm": 0.11645912379026413,
      "learning_rate": 4.003435383392415e-06,
      "loss": 0.0661,
      "num_input_tokens_seen": 110556656,
      "step": 51225
    },
    {
      "epoch": 8.35725938009788,
      "grad_norm": 0.5707449316978455,
      "learning_rate": 3.999572600277912e-06,
      "loss": 0.1064,
      "num_input_tokens_seen": 110567184,
      "step": 51230
    },
    {
      "epoch": 8.358075040783035,
      "grad_norm": 0.15371613204479218,
      "learning_rate": 3.995711519564607e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 110576848,
      "step": 51235
    },
    {
      "epoch": 8.358890701468189,
      "grad_norm": 0.08689006417989731,
      "learning_rate": 3.991852141565503e-06,
      "loss": 0.0489,
      "num_input_tokens_seen": 110587920,
      "step": 51240
    },
    {
      "epoch": 8.359706362153345,
      "grad_norm": 0.08142328262329102,
      "learning_rate": 3.987994466593456e-06,
      "loss": 0.0912,
      "num_input_tokens_seen": 110599184,
      "step": 51245
    },
    {
      "epoch": 8.360522022838499,
      "grad_norm": 0.19214679300785065,
      "learning_rate": 3.9841384949611924e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 110610768,
      "step": 51250
    },
    {
      "epoch": 8.361337683523654,
      "grad_norm": 0.4272608160972595,
      "learning_rate": 3.980284226981299e-06,
      "loss": 0.251,
      "num_input_tokens_seen": 110621136,
      "step": 51255
    },
    {
      "epoch": 8.362153344208808,
      "grad_norm": 0.11546285450458527,
      "learning_rate": 3.976431662966209e-06,
      "loss": 0.0334,
      "num_input_tokens_seen": 110631088,
      "step": 51260
    },
    {
      "epoch": 8.362969004893964,
      "grad_norm": 0.2972230613231659,
      "learning_rate": 3.972580803228249e-06,
      "loss": 0.0172,
      "num_input_tokens_seen": 110643184,
      "step": 51265
    },
    {
      "epoch": 8.36378466557912,
      "grad_norm": 3.7936272621154785,
      "learning_rate": 3.96873164807958e-06,
      "loss": 0.1162,
      "num_input_tokens_seen": 110653712,
      "step": 51270
    },
    {
      "epoch": 8.364600326264274,
      "grad_norm": 0.05377197265625,
      "learning_rate": 3.964884197832236e-06,
      "loss": 0.1022,
      "num_input_tokens_seen": 110663600,
      "step": 51275
    },
    {
      "epoch": 8.36541598694943,
      "grad_norm": 3.144099235534668,
      "learning_rate": 3.96103845279811e-06,
      "loss": 0.0973,
      "num_input_tokens_seen": 110674192,
      "step": 51280
    },
    {
      "epoch": 8.366231647634583,
      "grad_norm": 0.0675184354186058,
      "learning_rate": 3.957194413288956e-06,
      "loss": 0.1611,
      "num_input_tokens_seen": 110684880,
      "step": 51285
    },
    {
      "epoch": 8.367047308319739,
      "grad_norm": 0.0853884220123291,
      "learning_rate": 3.953352079616387e-06,
      "loss": 0.0214,
      "num_input_tokens_seen": 110695120,
      "step": 51290
    },
    {
      "epoch": 8.367862969004895,
      "grad_norm": 0.12090042978525162,
      "learning_rate": 3.949511452091898e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 110704912,
      "step": 51295
    },
    {
      "epoch": 8.368678629690049,
      "grad_norm": 0.08174553513526917,
      "learning_rate": 3.945672531026817e-06,
      "loss": 0.0104,
      "num_input_tokens_seen": 110715952,
      "step": 51300
    },
    {
      "epoch": 8.369494290375204,
      "grad_norm": 0.08900509774684906,
      "learning_rate": 3.941835316732348e-06,
      "loss": 0.1461,
      "num_input_tokens_seen": 110727312,
      "step": 51305
    },
    {
      "epoch": 8.370309951060358,
      "grad_norm": 0.037323251366615295,
      "learning_rate": 3.9379998095195606e-06,
      "loss": 0.0391,
      "num_input_tokens_seen": 110738448,
      "step": 51310
    },
    {
      "epoch": 8.371125611745514,
      "grad_norm": 3.7181408405303955,
      "learning_rate": 3.9341660096993725e-06,
      "loss": 0.1128,
      "num_input_tokens_seen": 110750960,
      "step": 51315
    },
    {
      "epoch": 8.37194127243067,
      "grad_norm": 0.10436207801103592,
      "learning_rate": 3.9303339175825736e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 110761584,
      "step": 51320
    },
    {
      "epoch": 8.372756933115824,
      "grad_norm": 0.0793977677822113,
      "learning_rate": 3.926503533479817e-06,
      "loss": 0.002,
      "num_input_tokens_seen": 110771792,
      "step": 51325
    },
    {
      "epoch": 8.37357259380098,
      "grad_norm": 3.5391643047332764,
      "learning_rate": 3.922674857701608e-06,
      "loss": 0.1114,
      "num_input_tokens_seen": 110783472,
      "step": 51330
    },
    {
      "epoch": 8.374388254486133,
      "grad_norm": 0.13794291019439697,
      "learning_rate": 3.918847890558322e-06,
      "loss": 0.0848,
      "num_input_tokens_seen": 110793584,
      "step": 51335
    },
    {
      "epoch": 8.375203915171289,
      "grad_norm": 0.11728204041719437,
      "learning_rate": 3.915022632360188e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 110803440,
      "step": 51340
    },
    {
      "epoch": 8.376019575856443,
      "grad_norm": 0.1369055062532425,
      "learning_rate": 3.911199083417305e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 110813808,
      "step": 51345
    },
    {
      "epoch": 8.376835236541599,
      "grad_norm": 1.1726471185684204,
      "learning_rate": 3.9073772440396285e-06,
      "loss": 0.0797,
      "num_input_tokens_seen": 110824016,
      "step": 51350
    },
    {
      "epoch": 8.377650897226754,
      "grad_norm": 0.16108562052249908,
      "learning_rate": 3.903557114536973e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 110834672,
      "step": 51355
    },
    {
      "epoch": 8.378466557911908,
      "grad_norm": 0.060831084847450256,
      "learning_rate": 3.899738695219024e-06,
      "loss": 0.0019,
      "num_input_tokens_seen": 110845456,
      "step": 51360
    },
    {
      "epoch": 8.379282218597064,
      "grad_norm": 0.15530067682266235,
      "learning_rate": 3.89592198639531e-06,
      "loss": 0.1605,
      "num_input_tokens_seen": 110856080,
      "step": 51365
    },
    {
      "epoch": 8.380097879282218,
      "grad_norm": 0.2525686025619507,
      "learning_rate": 3.8921069883752465e-06,
      "loss": 0.1055,
      "num_input_tokens_seen": 110867056,
      "step": 51370
    },
    {
      "epoch": 8.380913539967374,
      "grad_norm": 23.300146102905273,
      "learning_rate": 3.88829370146809e-06,
      "loss": 0.04,
      "num_input_tokens_seen": 110878352,
      "step": 51375
    },
    {
      "epoch": 8.38172920065253,
      "grad_norm": 0.14016051590442657,
      "learning_rate": 3.884482125982969e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 110889584,
      "step": 51380
    },
    {
      "epoch": 8.382544861337683,
      "grad_norm": 0.37360477447509766,
      "learning_rate": 3.880672262228863e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 110900464,
      "step": 51385
    },
    {
      "epoch": 8.383360522022839,
      "grad_norm": 0.07993663847446442,
      "learning_rate": 3.876864110514622e-06,
      "loss": 0.1006,
      "num_input_tokens_seen": 110911568,
      "step": 51390
    },
    {
      "epoch": 8.384176182707993,
      "grad_norm": 7.038994312286377,
      "learning_rate": 3.8730576711489555e-06,
      "loss": 0.0923,
      "num_input_tokens_seen": 110922640,
      "step": 51395
    },
    {
      "epoch": 8.384991843393149,
      "grad_norm": 0.09898694604635239,
      "learning_rate": 3.86925294444043e-06,
      "loss": 0.1509,
      "num_input_tokens_seen": 110933872,
      "step": 51400
    },
    {
      "epoch": 8.385807504078304,
      "grad_norm": 0.07565029710531235,
      "learning_rate": 3.8654499306974765e-06,
      "loss": 0.0961,
      "num_input_tokens_seen": 110945648,
      "step": 51405
    },
    {
      "epoch": 8.386623164763458,
      "grad_norm": 0.04033258929848671,
      "learning_rate": 3.86164863022839e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 110956528,
      "step": 51410
    },
    {
      "epoch": 8.387438825448614,
      "grad_norm": 4.092426300048828,
      "learning_rate": 3.857849043341316e-06,
      "loss": 0.0125,
      "num_input_tokens_seen": 110966224,
      "step": 51415
    },
    {
      "epoch": 8.388254486133768,
      "grad_norm": 0.13605235517024994,
      "learning_rate": 3.854051170344278e-06,
      "loss": 0.0566,
      "num_input_tokens_seen": 110976912,
      "step": 51420
    },
    {
      "epoch": 8.389070146818923,
      "grad_norm": 0.10848422348499298,
      "learning_rate": 3.8502550115451425e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 110988080,
      "step": 51425
    },
    {
      "epoch": 8.38988580750408,
      "grad_norm": 0.08110389113426208,
      "learning_rate": 3.846460567251648e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 110999472,
      "step": 51430
    },
    {
      "epoch": 8.390701468189233,
      "grad_norm": 0.13696368038654327,
      "learning_rate": 3.8426678377713884e-06,
      "loss": 0.0685,
      "num_input_tokens_seen": 111011184,
      "step": 51435
    },
    {
      "epoch": 8.391517128874389,
      "grad_norm": 0.05284971371293068,
      "learning_rate": 3.8388768234118275e-06,
      "loss": 0.0027,
      "num_input_tokens_seen": 111022480,
      "step": 51440
    },
    {
      "epoch": 8.392332789559543,
      "grad_norm": 0.08694755285978317,
      "learning_rate": 3.8350875244802855e-06,
      "loss": 0.1245,
      "num_input_tokens_seen": 111032912,
      "step": 51445
    },
    {
      "epoch": 8.393148450244698,
      "grad_norm": 4.483850955963135,
      "learning_rate": 3.831299941283936e-06,
      "loss": 0.0805,
      "num_input_tokens_seen": 111044272,
      "step": 51450
    },
    {
      "epoch": 8.393964110929852,
      "grad_norm": 14.72894287109375,
      "learning_rate": 3.827514074129823e-06,
      "loss": 0.0379,
      "num_input_tokens_seen": 111054704,
      "step": 51455
    },
    {
      "epoch": 8.394779771615008,
      "grad_norm": 0.12237244099378586,
      "learning_rate": 3.823729923324848e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 111065040,
      "step": 51460
    },
    {
      "epoch": 8.395595432300164,
      "grad_norm": 0.025830551981925964,
      "learning_rate": 3.819947489175771e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 111077200,
      "step": 51465
    },
    {
      "epoch": 8.396411092985318,
      "grad_norm": 0.20611804723739624,
      "learning_rate": 3.816166771989218e-06,
      "loss": 0.1059,
      "num_input_tokens_seen": 111088016,
      "step": 51470
    },
    {
      "epoch": 8.397226753670473,
      "grad_norm": 0.1460016667842865,
      "learning_rate": 3.812387772071668e-06,
      "loss": 0.0431,
      "num_input_tokens_seen": 111098320,
      "step": 51475
    },
    {
      "epoch": 8.398042414355627,
      "grad_norm": 0.05462673678994179,
      "learning_rate": 3.808610489729472e-06,
      "loss": 0.008,
      "num_input_tokens_seen": 111108528,
      "step": 51480
    },
    {
      "epoch": 8.398858075040783,
      "grad_norm": 0.06848473101854324,
      "learning_rate": 3.804834925268838e-06,
      "loss": 0.1276,
      "num_input_tokens_seen": 111120080,
      "step": 51485
    },
    {
      "epoch": 8.399673735725939,
      "grad_norm": 0.0810776874423027,
      "learning_rate": 3.801061078995827e-06,
      "loss": 0.0591,
      "num_input_tokens_seen": 111130512,
      "step": 51490
    },
    {
      "epoch": 8.400489396411093,
      "grad_norm": 0.10097594559192657,
      "learning_rate": 3.7972889512163656e-06,
      "loss": 0.202,
      "num_input_tokens_seen": 111141520,
      "step": 51495
    },
    {
      "epoch": 8.401305057096248,
      "grad_norm": 0.04713840037584305,
      "learning_rate": 3.7935185422362433e-06,
      "loss": 0.0203,
      "num_input_tokens_seen": 111152112,
      "step": 51500
    },
    {
      "epoch": 8.402120717781402,
      "grad_norm": 0.49332499504089355,
      "learning_rate": 3.7897498523611104e-06,
      "loss": 0.0081,
      "num_input_tokens_seen": 111162896,
      "step": 51505
    },
    {
      "epoch": 8.402936378466558,
      "grad_norm": 0.20173262059688568,
      "learning_rate": 3.7859828818964716e-06,
      "loss": 0.222,
      "num_input_tokens_seen": 111173616,
      "step": 51510
    },
    {
      "epoch": 8.403752039151712,
      "grad_norm": 0.02554556354880333,
      "learning_rate": 3.7822176311477027e-06,
      "loss": 0.0211,
      "num_input_tokens_seen": 111183856,
      "step": 51515
    },
    {
      "epoch": 8.404567699836868,
      "grad_norm": 0.07756847888231277,
      "learning_rate": 3.7784541004200287e-06,
      "loss": 0.1304,
      "num_input_tokens_seen": 111193936,
      "step": 51520
    },
    {
      "epoch": 8.405383360522023,
      "grad_norm": 0.04575493559241295,
      "learning_rate": 3.774692290018542e-06,
      "loss": 0.0981,
      "num_input_tokens_seen": 111204912,
      "step": 51525
    },
    {
      "epoch": 8.406199021207177,
      "grad_norm": 0.09417518228292465,
      "learning_rate": 3.770932200248195e-06,
      "loss": 0.0824,
      "num_input_tokens_seen": 111215408,
      "step": 51530
    },
    {
      "epoch": 8.407014681892333,
      "grad_norm": 0.03317214548587799,
      "learning_rate": 3.7671738314137978e-06,
      "loss": 0.1031,
      "num_input_tokens_seen": 111227056,
      "step": 51535
    },
    {
      "epoch": 8.407830342577487,
      "grad_norm": 4.939886569976807,
      "learning_rate": 3.7634171838200253e-06,
      "loss": 0.0707,
      "num_input_tokens_seen": 111238512,
      "step": 51540
    },
    {
      "epoch": 8.408646003262643,
      "grad_norm": 0.10969651490449905,
      "learning_rate": 3.75966225777141e-06,
      "loss": 0.012,
      "num_input_tokens_seen": 111249392,
      "step": 51545
    },
    {
      "epoch": 8.409461663947798,
      "grad_norm": 4.015342712402344,
      "learning_rate": 3.7559090535723427e-06,
      "loss": 0.2732,
      "num_input_tokens_seen": 111259600,
      "step": 51550
    },
    {
      "epoch": 8.410277324632952,
      "grad_norm": 0.07740370184183121,
      "learning_rate": 3.7521575715270817e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 111270032,
      "step": 51555
    },
    {
      "epoch": 8.411092985318108,
      "grad_norm": 0.08327538520097733,
      "learning_rate": 3.748407811939736e-06,
      "loss": 0.2263,
      "num_input_tokens_seen": 111279504,
      "step": 51560
    },
    {
      "epoch": 8.411908646003262,
      "grad_norm": 3.396444320678711,
      "learning_rate": 3.7446597751142844e-06,
      "loss": 0.1199,
      "num_input_tokens_seen": 111291504,
      "step": 51565
    },
    {
      "epoch": 8.412724306688418,
      "grad_norm": 0.05676591768860817,
      "learning_rate": 3.7409134613545587e-06,
      "loss": 0.1198,
      "num_input_tokens_seen": 111303152,
      "step": 51570
    },
    {
      "epoch": 8.413539967373573,
      "grad_norm": 6.9310150146484375,
      "learning_rate": 3.7371688709642555e-06,
      "loss": 0.0769,
      "num_input_tokens_seen": 111313232,
      "step": 51575
    },
    {
      "epoch": 8.414355628058727,
      "grad_norm": 0.03313552960753441,
      "learning_rate": 3.7334260042469232e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 111324240,
      "step": 51580
    },
    {
      "epoch": 8.415171288743883,
      "grad_norm": 0.03135775774717331,
      "learning_rate": 3.7296848615059913e-06,
      "loss": 0.1245,
      "num_input_tokens_seen": 111335056,
      "step": 51585
    },
    {
      "epoch": 8.415986949429037,
      "grad_norm": 7.66591739654541,
      "learning_rate": 3.725945443044729e-06,
      "loss": 0.0955,
      "num_input_tokens_seen": 111345200,
      "step": 51590
    },
    {
      "epoch": 8.416802610114193,
      "grad_norm": 0.11677058041095734,
      "learning_rate": 3.722207749166273e-06,
      "loss": 0.02,
      "num_input_tokens_seen": 111354640,
      "step": 51595
    },
    {
      "epoch": 8.417618270799348,
      "grad_norm": 0.06249038875102997,
      "learning_rate": 3.7184717801736186e-06,
      "loss": 0.0382,
      "num_input_tokens_seen": 111365008,
      "step": 51600
    },
    {
      "epoch": 8.418433931484502,
      "grad_norm": 0.1685234159231186,
      "learning_rate": 3.7147375363696168e-06,
      "loss": 0.0441,
      "num_input_tokens_seen": 111376080,
      "step": 51605
    },
    {
      "epoch": 8.419249592169658,
      "grad_norm": 0.15778842568397522,
      "learning_rate": 3.7110050180569985e-06,
      "loss": 0.1111,
      "num_input_tokens_seen": 111387376,
      "step": 51610
    },
    {
      "epoch": 8.420065252854812,
      "grad_norm": 0.035594161599874496,
      "learning_rate": 3.707274225538332e-06,
      "loss": 0.0121,
      "num_input_tokens_seen": 111398224,
      "step": 51615
    },
    {
      "epoch": 8.420880913539968,
      "grad_norm": 3.7639381885528564,
      "learning_rate": 3.7035451591160535e-06,
      "loss": 0.2597,
      "num_input_tokens_seen": 111408624,
      "step": 51620
    },
    {
      "epoch": 8.421696574225122,
      "grad_norm": 0.0966758206486702,
      "learning_rate": 3.699817819092463e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 111419664,
      "step": 51625
    },
    {
      "epoch": 8.422512234910277,
      "grad_norm": 0.11365343630313873,
      "learning_rate": 3.6960922057697163e-06,
      "loss": 0.0726,
      "num_input_tokens_seen": 111430480,
      "step": 51630
    },
    {
      "epoch": 8.423327895595433,
      "grad_norm": 0.019891297444701195,
      "learning_rate": 3.6923683194498295e-06,
      "loss": 0.207,
      "num_input_tokens_seen": 111439984,
      "step": 51635
    },
    {
      "epoch": 8.424143556280587,
      "grad_norm": 0.10552123188972473,
      "learning_rate": 3.6886461604346807e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 111451472,
      "step": 51640
    },
    {
      "epoch": 8.424959216965743,
      "grad_norm": 0.32801109552383423,
      "learning_rate": 3.6849257290260066e-06,
      "loss": 0.0199,
      "num_input_tokens_seen": 111462544,
      "step": 51645
    },
    {
      "epoch": 8.425774877650896,
      "grad_norm": 0.11163122206926346,
      "learning_rate": 3.6812070255254043e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 111474800,
      "step": 51650
    },
    {
      "epoch": 8.426590538336052,
      "grad_norm": 0.1100960224866867,
      "learning_rate": 3.677490050234331e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 111485776,
      "step": 51655
    },
    {
      "epoch": 8.427406199021208,
      "grad_norm": 0.057948071509599686,
      "learning_rate": 3.6737748034541054e-06,
      "loss": 0.2664,
      "num_input_tokens_seen": 111497616,
      "step": 51660
    },
    {
      "epoch": 8.428221859706362,
      "grad_norm": 0.2203397899866104,
      "learning_rate": 3.670061285485901e-06,
      "loss": 0.156,
      "num_input_tokens_seen": 111508848,
      "step": 51665
    },
    {
      "epoch": 8.429037520391518,
      "grad_norm": 0.12745971977710724,
      "learning_rate": 3.6663494966307553e-06,
      "loss": 0.2073,
      "num_input_tokens_seen": 111520816,
      "step": 51670
    },
    {
      "epoch": 8.429853181076671,
      "grad_norm": 5.879717826843262,
      "learning_rate": 3.662639437189566e-06,
      "loss": 0.1333,
      "num_input_tokens_seen": 111531984,
      "step": 51675
    },
    {
      "epoch": 8.430668841761827,
      "grad_norm": 12.864153861999512,
      "learning_rate": 3.65893110746309e-06,
      "loss": 0.1447,
      "num_input_tokens_seen": 111542800,
      "step": 51680
    },
    {
      "epoch": 8.431484502446983,
      "grad_norm": 0.07838033139705658,
      "learning_rate": 3.655224507751934e-06,
      "loss": 0.0621,
      "num_input_tokens_seen": 111553040,
      "step": 51685
    },
    {
      "epoch": 8.432300163132137,
      "grad_norm": 0.08984548598527908,
      "learning_rate": 3.6515196383565873e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 111563120,
      "step": 51690
    },
    {
      "epoch": 8.433115823817293,
      "grad_norm": 0.0729716420173645,
      "learning_rate": 3.6478164995773807e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 111573776,
      "step": 51695
    },
    {
      "epoch": 8.433931484502446,
      "grad_norm": 0.05781647562980652,
      "learning_rate": 3.644115091714509e-06,
      "loss": 0.2378,
      "num_input_tokens_seen": 111584080,
      "step": 51700
    },
    {
      "epoch": 8.434747145187602,
      "grad_norm": 7.219149589538574,
      "learning_rate": 3.640415415068027e-06,
      "loss": 0.0125,
      "num_input_tokens_seen": 111594480,
      "step": 51705
    },
    {
      "epoch": 8.435562805872756,
      "grad_norm": 5.097474098205566,
      "learning_rate": 3.6367174699378476e-06,
      "loss": 0.1759,
      "num_input_tokens_seen": 111604656,
      "step": 51710
    },
    {
      "epoch": 8.436378466557912,
      "grad_norm": 0.09925854206085205,
      "learning_rate": 3.6330212566237477e-06,
      "loss": 0.0974,
      "num_input_tokens_seen": 111616304,
      "step": 51715
    },
    {
      "epoch": 8.437194127243067,
      "grad_norm": 48.08366394042969,
      "learning_rate": 3.6293267754253566e-06,
      "loss": 0.0383,
      "num_input_tokens_seen": 111627920,
      "step": 51720
    },
    {
      "epoch": 8.438009787928221,
      "grad_norm": 0.07857240736484528,
      "learning_rate": 3.6256340266421747e-06,
      "loss": 0.0059,
      "num_input_tokens_seen": 111638864,
      "step": 51725
    },
    {
      "epoch": 8.438825448613377,
      "grad_norm": 0.15507225692272186,
      "learning_rate": 3.6219430105735476e-06,
      "loss": 0.0081,
      "num_input_tokens_seen": 111649296,
      "step": 51730
    },
    {
      "epoch": 8.439641109298531,
      "grad_norm": 0.4046974182128906,
      "learning_rate": 3.6182537275186947e-06,
      "loss": 0.0351,
      "num_input_tokens_seen": 111659152,
      "step": 51735
    },
    {
      "epoch": 8.440456769983687,
      "grad_norm": 0.6124518513679504,
      "learning_rate": 3.614566177776682e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 111669744,
      "step": 51740
    },
    {
      "epoch": 8.441272430668842,
      "grad_norm": 0.16115950047969818,
      "learning_rate": 3.6108803616464376e-06,
      "loss": 0.0998,
      "num_input_tokens_seen": 111680304,
      "step": 51745
    },
    {
      "epoch": 8.442088091353996,
      "grad_norm": 5.681840896606445,
      "learning_rate": 3.6071962794267667e-06,
      "loss": 0.2223,
      "num_input_tokens_seen": 111692400,
      "step": 51750
    },
    {
      "epoch": 8.442903752039152,
      "grad_norm": 0.052736107259988785,
      "learning_rate": 3.603513931416311e-06,
      "loss": 0.0227,
      "num_input_tokens_seen": 111702160,
      "step": 51755
    },
    {
      "epoch": 8.443719412724306,
      "grad_norm": 0.05737826228141785,
      "learning_rate": 3.5998333179135783e-06,
      "loss": 0.107,
      "num_input_tokens_seen": 111713712,
      "step": 51760
    },
    {
      "epoch": 8.444535073409462,
      "grad_norm": 0.02551724575459957,
      "learning_rate": 3.596154439216942e-06,
      "loss": 0.0286,
      "num_input_tokens_seen": 111724112,
      "step": 51765
    },
    {
      "epoch": 8.445350734094617,
      "grad_norm": 0.07931575924158096,
      "learning_rate": 3.5924772956246273e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 111735504,
      "step": 51770
    },
    {
      "epoch": 8.446166394779771,
      "grad_norm": 0.04812389984726906,
      "learning_rate": 3.5888018874347257e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 111744912,
      "step": 51775
    },
    {
      "epoch": 8.446982055464927,
      "grad_norm": 0.2566978931427002,
      "learning_rate": 3.5851282149451798e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 111756080,
      "step": 51780
    },
    {
      "epoch": 8.447797716150081,
      "grad_norm": 0.056581467390060425,
      "learning_rate": 3.5814562784538012e-06,
      "loss": 0.0023,
      "num_input_tokens_seen": 111766928,
      "step": 51785
    },
    {
      "epoch": 8.448613376835237,
      "grad_norm": 0.06746412813663483,
      "learning_rate": 3.5777860782582523e-06,
      "loss": 0.2365,
      "num_input_tokens_seen": 111775760,
      "step": 51790
    },
    {
      "epoch": 8.449429037520392,
      "grad_norm": 0.0598878487944603,
      "learning_rate": 3.5741176146560558e-06,
      "loss": 0.099,
      "num_input_tokens_seen": 111787184,
      "step": 51795
    },
    {
      "epoch": 8.450244698205546,
      "grad_norm": 0.08283038437366486,
      "learning_rate": 3.570450887944601e-06,
      "loss": 0.1214,
      "num_input_tokens_seen": 111798064,
      "step": 51800
    },
    {
      "epoch": 8.451060358890702,
      "grad_norm": 0.4032798409461975,
      "learning_rate": 3.5667858984211323e-06,
      "loss": 0.1477,
      "num_input_tokens_seen": 111808016,
      "step": 51805
    },
    {
      "epoch": 8.451876019575856,
      "grad_norm": 0.17475727200508118,
      "learning_rate": 3.5631226463827492e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 111819568,
      "step": 51810
    },
    {
      "epoch": 8.452691680261012,
      "grad_norm": 0.6154772639274597,
      "learning_rate": 3.5594611321264125e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 111829840,
      "step": 51815
    },
    {
      "epoch": 8.453507340946166,
      "grad_norm": 0.05333033576607704,
      "learning_rate": 3.5558013559489457e-06,
      "loss": 0.2209,
      "num_input_tokens_seen": 111841136,
      "step": 51820
    },
    {
      "epoch": 8.454323001631321,
      "grad_norm": 0.05515044555068016,
      "learning_rate": 3.5521433181470306e-06,
      "loss": 0.1007,
      "num_input_tokens_seen": 111852208,
      "step": 51825
    },
    {
      "epoch": 8.455138662316477,
      "grad_norm": 0.03507401421666145,
      "learning_rate": 3.5484870190171994e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 111862608,
      "step": 51830
    },
    {
      "epoch": 8.455954323001631,
      "grad_norm": 0.025188079103827477,
      "learning_rate": 3.5448324588558566e-06,
      "loss": 0.096,
      "num_input_tokens_seen": 111873424,
      "step": 51835
    },
    {
      "epoch": 8.456769983686787,
      "grad_norm": 4.245416641235352,
      "learning_rate": 3.541179637959255e-06,
      "loss": 0.2786,
      "num_input_tokens_seen": 111884016,
      "step": 51840
    },
    {
      "epoch": 8.45758564437194,
      "grad_norm": 0.22885818779468536,
      "learning_rate": 3.537528556623515e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 111895440,
      "step": 51845
    },
    {
      "epoch": 8.458401305057096,
      "grad_norm": 0.07951080054044724,
      "learning_rate": 3.5338792151446087e-06,
      "loss": 0.0023,
      "num_input_tokens_seen": 111907120,
      "step": 51850
    },
    {
      "epoch": 8.459216965742252,
      "grad_norm": 0.08638417720794678,
      "learning_rate": 3.530231613818372e-06,
      "loss": 0.1129,
      "num_input_tokens_seen": 111916464,
      "step": 51855
    },
    {
      "epoch": 8.460032626427406,
      "grad_norm": 0.118630051612854,
      "learning_rate": 3.526585752940495e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 111927376,
      "step": 51860
    },
    {
      "epoch": 8.460848287112562,
      "grad_norm": 4.591496467590332,
      "learning_rate": 3.522941632806534e-06,
      "loss": 0.1518,
      "num_input_tokens_seen": 111937392,
      "step": 51865
    },
    {
      "epoch": 8.461663947797716,
      "grad_norm": 0.19774377346038818,
      "learning_rate": 3.519299253711897e-06,
      "loss": 0.0566,
      "num_input_tokens_seen": 111948848,
      "step": 51870
    },
    {
      "epoch": 8.462479608482871,
      "grad_norm": 0.04803958162665367,
      "learning_rate": 3.515658615951856e-06,
      "loss": 0.0023,
      "num_input_tokens_seen": 111959664,
      "step": 51875
    },
    {
      "epoch": 8.463295269168025,
      "grad_norm": 0.06117786839604378,
      "learning_rate": 3.5120197198215356e-06,
      "loss": 0.1215,
      "num_input_tokens_seen": 111969712,
      "step": 51880
    },
    {
      "epoch": 8.464110929853181,
      "grad_norm": 0.2503572404384613,
      "learning_rate": 3.508382565615928e-06,
      "loss": 0.0076,
      "num_input_tokens_seen": 111981168,
      "step": 51885
    },
    {
      "epoch": 8.464926590538337,
      "grad_norm": 26.263456344604492,
      "learning_rate": 3.5047471536298697e-06,
      "loss": 0.0785,
      "num_input_tokens_seen": 111991952,
      "step": 51890
    },
    {
      "epoch": 8.46574225122349,
      "grad_norm": 0.07542380690574646,
      "learning_rate": 3.5011134841580805e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 112003024,
      "step": 51895
    },
    {
      "epoch": 8.466557911908646,
      "grad_norm": 3.779695749282837,
      "learning_rate": 3.4974815574951135e-06,
      "loss": 0.2171,
      "num_input_tokens_seen": 112013360,
      "step": 51900
    },
    {
      "epoch": 8.4673735725938,
      "grad_norm": 2.3623011112213135,
      "learning_rate": 3.4938513739353973e-06,
      "loss": 0.1671,
      "num_input_tokens_seen": 112023664,
      "step": 51905
    },
    {
      "epoch": 8.468189233278956,
      "grad_norm": 0.3942054808139801,
      "learning_rate": 3.4902229337732074e-06,
      "loss": 0.0985,
      "num_input_tokens_seen": 112034928,
      "step": 51910
    },
    {
      "epoch": 8.469004893964112,
      "grad_norm": 13.913886070251465,
      "learning_rate": 3.4865962373026805e-06,
      "loss": 0.2255,
      "num_input_tokens_seen": 112046128,
      "step": 51915
    },
    {
      "epoch": 8.469820554649266,
      "grad_norm": 0.012509414926171303,
      "learning_rate": 3.4829712848178293e-06,
      "loss": 0.0021,
      "num_input_tokens_seen": 112056944,
      "step": 51920
    },
    {
      "epoch": 8.470636215334421,
      "grad_norm": 0.037642624229192734,
      "learning_rate": 3.4793480766124986e-06,
      "loss": 0.3158,
      "num_input_tokens_seen": 112067472,
      "step": 51925
    },
    {
      "epoch": 8.471451876019575,
      "grad_norm": 0.05478547513484955,
      "learning_rate": 3.4757266129804093e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 112076400,
      "step": 51930
    },
    {
      "epoch": 8.47226753670473,
      "grad_norm": 2.140247106552124,
      "learning_rate": 3.4721068942151324e-06,
      "loss": 0.1839,
      "num_input_tokens_seen": 112087312,
      "step": 51935
    },
    {
      "epoch": 8.473083197389887,
      "grad_norm": 7.122751712799072,
      "learning_rate": 3.4684889206101025e-06,
      "loss": 0.1102,
      "num_input_tokens_seen": 112098544,
      "step": 51940
    },
    {
      "epoch": 8.47389885807504,
      "grad_norm": 0.10864073783159256,
      "learning_rate": 3.464872692458612e-06,
      "loss": 0.0878,
      "num_input_tokens_seen": 112108304,
      "step": 51945
    },
    {
      "epoch": 8.474714518760196,
      "grad_norm": 0.05819137021899223,
      "learning_rate": 3.4612582100538082e-06,
      "loss": 0.0015,
      "num_input_tokens_seen": 112119600,
      "step": 51950
    },
    {
      "epoch": 8.47553017944535,
      "grad_norm": 0.07787571847438812,
      "learning_rate": 3.4576454736887003e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 112130192,
      "step": 51955
    },
    {
      "epoch": 8.476345840130506,
      "grad_norm": 0.10811415314674377,
      "learning_rate": 3.4540344836561546e-06,
      "loss": 0.1739,
      "num_input_tokens_seen": 112142288,
      "step": 51960
    },
    {
      "epoch": 8.477161500815662,
      "grad_norm": 0.08363020420074463,
      "learning_rate": 3.4504252402488974e-06,
      "loss": 0.043,
      "num_input_tokens_seen": 112154032,
      "step": 51965
    },
    {
      "epoch": 8.477977161500815,
      "grad_norm": 0.09594326466321945,
      "learning_rate": 3.446817743759512e-06,
      "loss": 0.0021,
      "num_input_tokens_seen": 112165520,
      "step": 51970
    },
    {
      "epoch": 8.478792822185971,
      "grad_norm": 4.208395481109619,
      "learning_rate": 3.443211994480439e-06,
      "loss": 0.1565,
      "num_input_tokens_seen": 112176560,
      "step": 51975
    },
    {
      "epoch": 8.479608482871125,
      "grad_norm": 0.13062646985054016,
      "learning_rate": 3.4396079927039804e-06,
      "loss": 0.0023,
      "num_input_tokens_seen": 112187248,
      "step": 51980
    },
    {
      "epoch": 8.48042414355628,
      "grad_norm": 0.08347542583942413,
      "learning_rate": 3.436005738722292e-06,
      "loss": 0.0792,
      "num_input_tokens_seen": 112198128,
      "step": 51985
    },
    {
      "epoch": 8.481239804241435,
      "grad_norm": 3.741190195083618,
      "learning_rate": 3.432405232827396e-06,
      "loss": 0.1506,
      "num_input_tokens_seen": 112209680,
      "step": 51990
    },
    {
      "epoch": 8.48205546492659,
      "grad_norm": 0.0963752344250679,
      "learning_rate": 3.428806475311164e-06,
      "loss": 0.0102,
      "num_input_tokens_seen": 112220688,
      "step": 51995
    },
    {
      "epoch": 8.482871125611746,
      "grad_norm": 26.20519256591797,
      "learning_rate": 3.4252094664653316e-06,
      "loss": 0.1882,
      "num_input_tokens_seen": 112231632,
      "step": 52000
    },
    {
      "epoch": 8.4836867862969,
      "grad_norm": 0.08957038819789886,
      "learning_rate": 3.4216142065814806e-06,
      "loss": 0.228,
      "num_input_tokens_seen": 112240592,
      "step": 52005
    },
    {
      "epoch": 8.484502446982056,
      "grad_norm": 0.0813271775841713,
      "learning_rate": 3.418020695951077e-06,
      "loss": 0.0098,
      "num_input_tokens_seen": 112251440,
      "step": 52010
    },
    {
      "epoch": 8.48531810766721,
      "grad_norm": 1.235237717628479,
      "learning_rate": 3.414428934865421e-06,
      "loss": 0.1309,
      "num_input_tokens_seen": 112262192,
      "step": 52015
    },
    {
      "epoch": 8.486133768352365,
      "grad_norm": 0.02408054657280445,
      "learning_rate": 3.4108389236156806e-06,
      "loss": 0.0021,
      "num_input_tokens_seen": 112273808,
      "step": 52020
    },
    {
      "epoch": 8.486949429037521,
      "grad_norm": 27.396228790283203,
      "learning_rate": 3.4072506624928808e-06,
      "loss": 0.0425,
      "num_input_tokens_seen": 112284304,
      "step": 52025
    },
    {
      "epoch": 8.487765089722675,
      "grad_norm": 4.579169750213623,
      "learning_rate": 3.4036641517878997e-06,
      "loss": 0.2476,
      "num_input_tokens_seen": 112295152,
      "step": 52030
    },
    {
      "epoch": 8.48858075040783,
      "grad_norm": 0.04967943951487541,
      "learning_rate": 3.400079391791483e-06,
      "loss": 0.11,
      "num_input_tokens_seen": 112305648,
      "step": 52035
    },
    {
      "epoch": 8.489396411092985,
      "grad_norm": 0.20065897703170776,
      "learning_rate": 3.3964963827942257e-06,
      "loss": 0.0088,
      "num_input_tokens_seen": 112316144,
      "step": 52040
    },
    {
      "epoch": 8.49021207177814,
      "grad_norm": 0.1346825212240219,
      "learning_rate": 3.3929151250865903e-06,
      "loss": 0.082,
      "num_input_tokens_seen": 112328560,
      "step": 52045
    },
    {
      "epoch": 8.491027732463296,
      "grad_norm": 0.754371166229248,
      "learning_rate": 3.389335618958886e-06,
      "loss": 0.1196,
      "num_input_tokens_seen": 112338640,
      "step": 52050
    },
    {
      "epoch": 8.49184339314845,
      "grad_norm": 0.17561429738998413,
      "learning_rate": 3.385757864701286e-06,
      "loss": 0.0105,
      "num_input_tokens_seen": 112349232,
      "step": 52055
    },
    {
      "epoch": 8.492659053833606,
      "grad_norm": 0.09084920585155487,
      "learning_rate": 3.3821818626038198e-06,
      "loss": 0.0706,
      "num_input_tokens_seen": 112360400,
      "step": 52060
    },
    {
      "epoch": 8.49347471451876,
      "grad_norm": 0.05691864714026451,
      "learning_rate": 3.378607612956386e-06,
      "loss": 0.075,
      "num_input_tokens_seen": 112371408,
      "step": 52065
    },
    {
      "epoch": 8.494290375203915,
      "grad_norm": 0.056467730551958084,
      "learning_rate": 3.375035116048722e-06,
      "loss": 0.0652,
      "num_input_tokens_seen": 112382544,
      "step": 52070
    },
    {
      "epoch": 8.49510603588907,
      "grad_norm": 0.7548681497573853,
      "learning_rate": 3.371464372170438e-06,
      "loss": 0.085,
      "num_input_tokens_seen": 112392688,
      "step": 52075
    },
    {
      "epoch": 8.495921696574225,
      "grad_norm": 0.08632798492908478,
      "learning_rate": 3.3678953816109916e-06,
      "loss": 0.1076,
      "num_input_tokens_seen": 112404240,
      "step": 52080
    },
    {
      "epoch": 8.49673735725938,
      "grad_norm": 0.25324541330337524,
      "learning_rate": 3.3643281446597092e-06,
      "loss": 0.131,
      "num_input_tokens_seen": 112414768,
      "step": 52085
    },
    {
      "epoch": 8.497553017944535,
      "grad_norm": 0.031284745782613754,
      "learning_rate": 3.3607626616057624e-06,
      "loss": 0.1112,
      "num_input_tokens_seen": 112425424,
      "step": 52090
    },
    {
      "epoch": 8.49836867862969,
      "grad_norm": 0.08141515403985977,
      "learning_rate": 3.3571989327381923e-06,
      "loss": 0.0529,
      "num_input_tokens_seen": 112436720,
      "step": 52095
    },
    {
      "epoch": 8.499184339314844,
      "grad_norm": 0.0655307024717331,
      "learning_rate": 3.3536369583458905e-06,
      "loss": 0.2946,
      "num_input_tokens_seen": 112446896,
      "step": 52100
    },
    {
      "epoch": 8.5,
      "grad_norm": 0.06876910477876663,
      "learning_rate": 3.3500767387176114e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 112458064,
      "step": 52105
    },
    {
      "epoch": 8.5,
      "eval_loss": 0.19772686064243317,
      "eval_runtime": 568.5108,
      "eval_samples_per_second": 4.793,
      "eval_steps_per_second": 1.2,
      "num_input_tokens_seen": 112458064,
      "step": 52105
    },
    {
      "epoch": 8.500815660685156,
      "grad_norm": 0.2995106875896454,
      "learning_rate": 3.3465182741419547e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 112469200,
      "step": 52110
    },
    {
      "epoch": 8.50163132137031,
      "grad_norm": 6.758239269256592,
      "learning_rate": 3.3429615649074013e-06,
      "loss": 0.3036,
      "num_input_tokens_seen": 112479888,
      "step": 52115
    },
    {
      "epoch": 8.502446982055465,
      "grad_norm": 0.08517052233219147,
      "learning_rate": 3.3394066113022706e-06,
      "loss": 0.4295,
      "num_input_tokens_seen": 112490896,
      "step": 52120
    },
    {
      "epoch": 8.50326264274062,
      "grad_norm": 0.1996101438999176,
      "learning_rate": 3.335853413614745e-06,
      "loss": 0.1195,
      "num_input_tokens_seen": 112503152,
      "step": 52125
    },
    {
      "epoch": 8.504078303425775,
      "grad_norm": 0.113855741918087,
      "learning_rate": 3.332301972132862e-06,
      "loss": 0.1288,
      "num_input_tokens_seen": 112513936,
      "step": 52130
    },
    {
      "epoch": 8.50489396411093,
      "grad_norm": 0.1581876277923584,
      "learning_rate": 3.3287522871445263e-06,
      "loss": 0.0837,
      "num_input_tokens_seen": 112523536,
      "step": 52135
    },
    {
      "epoch": 8.505709624796085,
      "grad_norm": 0.044547755271196365,
      "learning_rate": 3.3252043589374866e-06,
      "loss": 0.1824,
      "num_input_tokens_seen": 112534512,
      "step": 52140
    },
    {
      "epoch": 8.50652528548124,
      "grad_norm": 0.07674051076173782,
      "learning_rate": 3.3216581877993564e-06,
      "loss": 0.2144,
      "num_input_tokens_seen": 112546032,
      "step": 52145
    },
    {
      "epoch": 8.507340946166394,
      "grad_norm": 4.138210296630859,
      "learning_rate": 3.3181137740176118e-06,
      "loss": 0.252,
      "num_input_tokens_seen": 112556016,
      "step": 52150
    },
    {
      "epoch": 8.50815660685155,
      "grad_norm": 0.3108614385128021,
      "learning_rate": 3.3145711178795753e-06,
      "loss": 0.0322,
      "num_input_tokens_seen": 112565520,
      "step": 52155
    },
    {
      "epoch": 8.508972267536706,
      "grad_norm": 0.062382735311985016,
      "learning_rate": 3.3110302196724368e-06,
      "loss": 0.0231,
      "num_input_tokens_seen": 112575824,
      "step": 52160
    },
    {
      "epoch": 8.50978792822186,
      "grad_norm": 0.420159250497818,
      "learning_rate": 3.3074910796832363e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 112587056,
      "step": 52165
    },
    {
      "epoch": 8.510603588907015,
      "grad_norm": 0.11795055866241455,
      "learning_rate": 3.303953698198875e-06,
      "loss": 0.0024,
      "num_input_tokens_seen": 112598192,
      "step": 52170
    },
    {
      "epoch": 8.51141924959217,
      "grad_norm": 0.1338420957326889,
      "learning_rate": 3.300418075506112e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 112609584,
      "step": 52175
    },
    {
      "epoch": 8.512234910277325,
      "grad_norm": 8.539312362670898,
      "learning_rate": 3.296884211891563e-06,
      "loss": 0.1122,
      "num_input_tokens_seen": 112620720,
      "step": 52180
    },
    {
      "epoch": 8.513050570962479,
      "grad_norm": 0.13929800689220428,
      "learning_rate": 3.293352107641698e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 112631696,
      "step": 52185
    },
    {
      "epoch": 8.513866231647635,
      "grad_norm": 0.12888842821121216,
      "learning_rate": 3.2898217630428523e-06,
      "loss": 0.1069,
      "num_input_tokens_seen": 112641872,
      "step": 52190
    },
    {
      "epoch": 8.51468189233279,
      "grad_norm": 0.21990133821964264,
      "learning_rate": 3.2862931783812083e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 112652240,
      "step": 52195
    },
    {
      "epoch": 8.515497553017944,
      "grad_norm": 0.10042712092399597,
      "learning_rate": 3.282766353942815e-06,
      "loss": 0.0082,
      "num_input_tokens_seen": 112663824,
      "step": 52200
    },
    {
      "epoch": 8.5163132137031,
      "grad_norm": 1.5217121839523315,
      "learning_rate": 3.279241290013568e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 112675216,
      "step": 52205
    },
    {
      "epoch": 8.517128874388254,
      "grad_norm": 0.08880387246608734,
      "learning_rate": 3.275717986879237e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 112684464,
      "step": 52210
    },
    {
      "epoch": 8.51794453507341,
      "grad_norm": 0.09611815214157104,
      "learning_rate": 3.2721964448254345e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 112695120,
      "step": 52215
    },
    {
      "epoch": 8.518760195758565,
      "grad_norm": 0.02544640563428402,
      "learning_rate": 3.268676664137635e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 112705776,
      "step": 52220
    },
    {
      "epoch": 8.51957585644372,
      "grad_norm": 0.04137982428073883,
      "learning_rate": 3.2651586451011657e-06,
      "loss": 0.2269,
      "num_input_tokens_seen": 112716368,
      "step": 52225
    },
    {
      "epoch": 8.520391517128875,
      "grad_norm": 0.1521427482366562,
      "learning_rate": 3.2616423880012153e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 112727440,
      "step": 52230
    },
    {
      "epoch": 8.521207177814029,
      "grad_norm": 3.602827548980713,
      "learning_rate": 3.2581278931228363e-06,
      "loss": 0.2515,
      "num_input_tokens_seen": 112736336,
      "step": 52235
    },
    {
      "epoch": 8.522022838499185,
      "grad_norm": 0.1897648721933365,
      "learning_rate": 3.254615160750926e-06,
      "loss": 0.1008,
      "num_input_tokens_seen": 112747312,
      "step": 52240
    },
    {
      "epoch": 8.522838499184338,
      "grad_norm": 11.034748077392578,
      "learning_rate": 3.2511041911702483e-06,
      "loss": 0.2812,
      "num_input_tokens_seen": 112758160,
      "step": 52245
    },
    {
      "epoch": 8.523654159869494,
      "grad_norm": 0.34525004029273987,
      "learning_rate": 3.247594984665417e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 112768080,
      "step": 52250
    },
    {
      "epoch": 8.52446982055465,
      "grad_norm": 0.06467463821172714,
      "learning_rate": 3.244087541520907e-06,
      "loss": 0.0746,
      "num_input_tokens_seen": 112778416,
      "step": 52255
    },
    {
      "epoch": 8.525285481239804,
      "grad_norm": 3.4724481105804443,
      "learning_rate": 3.24058186202105e-06,
      "loss": 0.1234,
      "num_input_tokens_seen": 112788944,
      "step": 52260
    },
    {
      "epoch": 8.52610114192496,
      "grad_norm": 0.06703388690948486,
      "learning_rate": 3.2370779464500317e-06,
      "loss": 0.0018,
      "num_input_tokens_seen": 112799664,
      "step": 52265
    },
    {
      "epoch": 8.526916802610113,
      "grad_norm": 0.11705297976732254,
      "learning_rate": 3.2335757950919003e-06,
      "loss": 0.2958,
      "num_input_tokens_seen": 112809840,
      "step": 52270
    },
    {
      "epoch": 8.52773246329527,
      "grad_norm": 0.18924038112163544,
      "learning_rate": 3.230075408230557e-06,
      "loss": 0.096,
      "num_input_tokens_seen": 112820240,
      "step": 52275
    },
    {
      "epoch": 8.528548123980425,
      "grad_norm": 0.06967607140541077,
      "learning_rate": 3.2265767861497597e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 112830704,
      "step": 52280
    },
    {
      "epoch": 8.529363784665579,
      "grad_norm": 2.9614319801330566,
      "learning_rate": 3.2230799291331244e-06,
      "loss": 0.074,
      "num_input_tokens_seen": 112842352,
      "step": 52285
    },
    {
      "epoch": 8.530179445350734,
      "grad_norm": 0.10939770191907883,
      "learning_rate": 3.219584837464126e-06,
      "loss": 0.1415,
      "num_input_tokens_seen": 112853904,
      "step": 52290
    },
    {
      "epoch": 8.530995106035888,
      "grad_norm": 0.18360036611557007,
      "learning_rate": 3.2160915114260947e-06,
      "loss": 0.0145,
      "num_input_tokens_seen": 112863504,
      "step": 52295
    },
    {
      "epoch": 8.531810766721044,
      "grad_norm": 0.12482015788555145,
      "learning_rate": 3.212599951302214e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 112873808,
      "step": 52300
    },
    {
      "epoch": 8.5326264274062,
      "grad_norm": 3.226783037185669,
      "learning_rate": 3.2091101573755306e-06,
      "loss": 0.1789,
      "num_input_tokens_seen": 112883696,
      "step": 52305
    },
    {
      "epoch": 8.533442088091354,
      "grad_norm": 0.1161847934126854,
      "learning_rate": 3.2056221299289423e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 112893424,
      "step": 52310
    },
    {
      "epoch": 8.53425774877651,
      "grad_norm": 0.06892212480306625,
      "learning_rate": 3.20213586924521e-06,
      "loss": 0.0437,
      "num_input_tokens_seen": 112904016,
      "step": 52315
    },
    {
      "epoch": 8.535073409461663,
      "grad_norm": 1.1102094650268555,
      "learning_rate": 3.1986513756069426e-06,
      "loss": 0.0104,
      "num_input_tokens_seen": 112915376,
      "step": 52320
    },
    {
      "epoch": 8.535889070146819,
      "grad_norm": 0.03022949770092964,
      "learning_rate": 3.1951686492966094e-06,
      "loss": 0.1211,
      "num_input_tokens_seen": 112925424,
      "step": 52325
    },
    {
      "epoch": 8.536704730831975,
      "grad_norm": 0.09329638630151749,
      "learning_rate": 3.1916876905965483e-06,
      "loss": 0.1195,
      "num_input_tokens_seen": 112936656,
      "step": 52330
    },
    {
      "epoch": 8.537520391517129,
      "grad_norm": 0.07435731589794159,
      "learning_rate": 3.188208499788936e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 112947600,
      "step": 52335
    },
    {
      "epoch": 8.538336052202284,
      "grad_norm": 0.038756757974624634,
      "learning_rate": 3.184731077155817e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 112957840,
      "step": 52340
    },
    {
      "epoch": 8.539151712887438,
      "grad_norm": 0.11953333765268326,
      "learning_rate": 3.1812554229790848e-06,
      "loss": 0.0206,
      "num_input_tokens_seen": 112969488,
      "step": 52345
    },
    {
      "epoch": 8.539967373572594,
      "grad_norm": 0.11215139180421829,
      "learning_rate": 3.1777815375404944e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 112979216,
      "step": 52350
    },
    {
      "epoch": 8.540783034257748,
      "grad_norm": 13.117161750793457,
      "learning_rate": 3.17430942112166e-06,
      "loss": 0.0316,
      "num_input_tokens_seen": 112990736,
      "step": 52355
    },
    {
      "epoch": 8.541598694942904,
      "grad_norm": 6.0222249031066895,
      "learning_rate": 3.170839074004045e-06,
      "loss": 0.2536,
      "num_input_tokens_seen": 113001296,
      "step": 52360
    },
    {
      "epoch": 8.54241435562806,
      "grad_norm": 0.06390171498060226,
      "learning_rate": 3.1673704964689743e-06,
      "loss": 0.0097,
      "num_input_tokens_seen": 113011312,
      "step": 52365
    },
    {
      "epoch": 8.543230016313213,
      "grad_norm": 1.7926284074783325,
      "learning_rate": 3.1639036887976286e-06,
      "loss": 0.0108,
      "num_input_tokens_seen": 113020688,
      "step": 52370
    },
    {
      "epoch": 8.544045676998369,
      "grad_norm": 0.10173257440328598,
      "learning_rate": 3.1604386512710387e-06,
      "loss": 0.0845,
      "num_input_tokens_seen": 113031888,
      "step": 52375
    },
    {
      "epoch": 8.544861337683523,
      "grad_norm": 0.18404307961463928,
      "learning_rate": 3.1569753841701106e-06,
      "loss": 0.0089,
      "num_input_tokens_seen": 113042800,
      "step": 52380
    },
    {
      "epoch": 8.545676998368679,
      "grad_norm": 0.0896589383482933,
      "learning_rate": 3.1535138877755887e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 113052176,
      "step": 52385
    },
    {
      "epoch": 8.546492659053834,
      "grad_norm": 0.023683076724410057,
      "learning_rate": 3.1500541623680795e-06,
      "loss": 0.1377,
      "num_input_tokens_seen": 113062608,
      "step": 52390
    },
    {
      "epoch": 8.547308319738988,
      "grad_norm": 0.09791173785924911,
      "learning_rate": 3.1465962082280474e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 113075248,
      "step": 52395
    },
    {
      "epoch": 8.548123980424144,
      "grad_norm": 0.06262429058551788,
      "learning_rate": 3.1431400256358073e-06,
      "loss": 0.0331,
      "num_input_tokens_seen": 113086256,
      "step": 52400
    },
    {
      "epoch": 8.548939641109298,
      "grad_norm": 0.1712462455034256,
      "learning_rate": 3.1396856148715375e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 113096752,
      "step": 52405
    },
    {
      "epoch": 8.549755301794454,
      "grad_norm": 0.021678127348423004,
      "learning_rate": 3.13623297621527e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 113106480,
      "step": 52410
    },
    {
      "epoch": 8.550570962479608,
      "grad_norm": 0.05082330107688904,
      "learning_rate": 3.1327821099468915e-06,
      "loss": 0.1551,
      "num_input_tokens_seen": 113117424,
      "step": 52415
    },
    {
      "epoch": 8.551386623164763,
      "grad_norm": 4.563812732696533,
      "learning_rate": 3.1293330163461503e-06,
      "loss": 0.1964,
      "num_input_tokens_seen": 113127728,
      "step": 52420
    },
    {
      "epoch": 8.552202283849919,
      "grad_norm": 4.3056321144104,
      "learning_rate": 3.125885695692646e-06,
      "loss": 0.1261,
      "num_input_tokens_seen": 113138000,
      "step": 52425
    },
    {
      "epoch": 8.553017944535073,
      "grad_norm": 0.13502757251262665,
      "learning_rate": 3.122440148265829e-06,
      "loss": 0.2172,
      "num_input_tokens_seen": 113148752,
      "step": 52430
    },
    {
      "epoch": 8.553833605220229,
      "grad_norm": 0.0628582164645195,
      "learning_rate": 3.1189963743450235e-06,
      "loss": 0.0363,
      "num_input_tokens_seen": 113159600,
      "step": 52435
    },
    {
      "epoch": 8.554649265905383,
      "grad_norm": 0.2686457931995392,
      "learning_rate": 3.115554374209395e-06,
      "loss": 0.0767,
      "num_input_tokens_seen": 113170416,
      "step": 52440
    },
    {
      "epoch": 8.555464926590538,
      "grad_norm": 0.1591261923313141,
      "learning_rate": 3.1121141481379735e-06,
      "loss": 0.0202,
      "num_input_tokens_seen": 113179984,
      "step": 52445
    },
    {
      "epoch": 8.556280587275694,
      "grad_norm": 0.36520224809646606,
      "learning_rate": 3.1086756964096327e-06,
      "loss": 0.0954,
      "num_input_tokens_seen": 113189680,
      "step": 52450
    },
    {
      "epoch": 8.557096247960848,
      "grad_norm": 0.2917357087135315,
      "learning_rate": 3.105239019303116e-06,
      "loss": 0.0105,
      "num_input_tokens_seen": 113200112,
      "step": 52455
    },
    {
      "epoch": 8.557911908646004,
      "grad_norm": 0.06772247701883316,
      "learning_rate": 3.101804117097018e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 113210416,
      "step": 52460
    },
    {
      "epoch": 8.558727569331158,
      "grad_norm": 0.03786826878786087,
      "learning_rate": 3.0983709900697903e-06,
      "loss": 0.0021,
      "num_input_tokens_seen": 113221392,
      "step": 52465
    },
    {
      "epoch": 8.559543230016313,
      "grad_norm": 0.09939499199390411,
      "learning_rate": 3.0949396384997357e-06,
      "loss": 0.3358,
      "num_input_tokens_seen": 113232528,
      "step": 52470
    },
    {
      "epoch": 8.560358890701469,
      "grad_norm": 0.04357518255710602,
      "learning_rate": 3.0915100626650206e-06,
      "loss": 0.1614,
      "num_input_tokens_seen": 113243056,
      "step": 52475
    },
    {
      "epoch": 8.561174551386623,
      "grad_norm": 37.47199249267578,
      "learning_rate": 3.0880822628436613e-06,
      "loss": 0.2143,
      "num_input_tokens_seen": 113254576,
      "step": 52480
    },
    {
      "epoch": 8.561990212071779,
      "grad_norm": 0.06570349633693695,
      "learning_rate": 3.0846562393135352e-06,
      "loss": 0.0669,
      "num_input_tokens_seen": 113263472,
      "step": 52485
    },
    {
      "epoch": 8.562805872756933,
      "grad_norm": 0.3499777019023895,
      "learning_rate": 3.0812319923523706e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 113276112,
      "step": 52490
    },
    {
      "epoch": 8.563621533442088,
      "grad_norm": 0.045446496456861496,
      "learning_rate": 3.077809522237754e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 113285520,
      "step": 52495
    },
    {
      "epoch": 8.564437194127244,
      "grad_norm": 0.18569421768188477,
      "learning_rate": 3.0743888292471322e-06,
      "loss": 0.008,
      "num_input_tokens_seen": 113296304,
      "step": 52500
    },
    {
      "epoch": 8.565252854812398,
      "grad_norm": 0.02989482879638672,
      "learning_rate": 3.0709699136578006e-06,
      "loss": 0.0936,
      "num_input_tokens_seen": 113307632,
      "step": 52505
    },
    {
      "epoch": 8.566068515497554,
      "grad_norm": 0.16343659162521362,
      "learning_rate": 3.0675527757469124e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 113318352,
      "step": 52510
    },
    {
      "epoch": 8.566884176182707,
      "grad_norm": 0.049167994409799576,
      "learning_rate": 3.064137415791485e-06,
      "loss": 0.2914,
      "num_input_tokens_seen": 113329552,
      "step": 52515
    },
    {
      "epoch": 8.567699836867863,
      "grad_norm": 0.2270321398973465,
      "learning_rate": 3.0607238340683713e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 113339216,
      "step": 52520
    },
    {
      "epoch": 8.568515497553017,
      "grad_norm": 1.0442709922790527,
      "learning_rate": 3.057312030854306e-06,
      "loss": 0.1237,
      "num_input_tokens_seen": 113349904,
      "step": 52525
    },
    {
      "epoch": 8.569331158238173,
      "grad_norm": 6.7521162033081055,
      "learning_rate": 3.0539020064258682e-06,
      "loss": 0.0847,
      "num_input_tokens_seen": 113358832,
      "step": 52530
    },
    {
      "epoch": 8.570146818923329,
      "grad_norm": 0.033351555466651917,
      "learning_rate": 3.0504937610594837e-06,
      "loss": 0.1551,
      "num_input_tokens_seen": 113370480,
      "step": 52535
    },
    {
      "epoch": 8.570962479608482,
      "grad_norm": 0.11367560178041458,
      "learning_rate": 3.0470872950314476e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 113380592,
      "step": 52540
    },
    {
      "epoch": 8.571778140293638,
      "grad_norm": 4.26261568069458,
      "learning_rate": 3.043682608617898e-06,
      "loss": 0.1709,
      "num_input_tokens_seen": 113392112,
      "step": 52545
    },
    {
      "epoch": 8.572593800978792,
      "grad_norm": 0.1217903420329094,
      "learning_rate": 3.0402797020948446e-06,
      "loss": 0.0707,
      "num_input_tokens_seen": 113403152,
      "step": 52550
    },
    {
      "epoch": 8.573409461663948,
      "grad_norm": 0.13792067766189575,
      "learning_rate": 3.0368785757381418e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 113413424,
      "step": 52555
    },
    {
      "epoch": 8.574225122349104,
      "grad_norm": 0.20448121428489685,
      "learning_rate": 3.033479229823502e-06,
      "loss": 0.1303,
      "num_input_tokens_seen": 113424848,
      "step": 52560
    },
    {
      "epoch": 8.575040783034257,
      "grad_norm": 3.6465656757354736,
      "learning_rate": 3.030081664626494e-06,
      "loss": 0.3299,
      "num_input_tokens_seen": 113434608,
      "step": 52565
    },
    {
      "epoch": 8.575856443719413,
      "grad_norm": 0.014260239899158478,
      "learning_rate": 3.0266858804225388e-06,
      "loss": 0.1935,
      "num_input_tokens_seen": 113446224,
      "step": 52570
    },
    {
      "epoch": 8.576672104404567,
      "grad_norm": 0.09066756814718246,
      "learning_rate": 3.0232918774869194e-06,
      "loss": 0.2235,
      "num_input_tokens_seen": 113457296,
      "step": 52575
    },
    {
      "epoch": 8.577487765089723,
      "grad_norm": 0.020743852481245995,
      "learning_rate": 3.0198996560947657e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 113468720,
      "step": 52580
    },
    {
      "epoch": 8.578303425774878,
      "grad_norm": 5.239649295806885,
      "learning_rate": 3.016509216521074e-06,
      "loss": 0.1828,
      "num_input_tokens_seen": 113479152,
      "step": 52585
    },
    {
      "epoch": 8.579119086460032,
      "grad_norm": 3.68194317817688,
      "learning_rate": 3.0131205590406886e-06,
      "loss": 0.1205,
      "num_input_tokens_seen": 113490608,
      "step": 52590
    },
    {
      "epoch": 8.579934747145188,
      "grad_norm": 0.0550539493560791,
      "learning_rate": 3.0097336839283118e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 113501968,
      "step": 52595
    },
    {
      "epoch": 8.580750407830342,
      "grad_norm": 0.18661192059516907,
      "learning_rate": 3.0063485914584995e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 113512944,
      "step": 52600
    },
    {
      "epoch": 8.581566068515498,
      "grad_norm": 0.050649698823690414,
      "learning_rate": 3.0029652819056646e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 113524304,
      "step": 52605
    },
    {
      "epoch": 8.582381729200652,
      "grad_norm": 5.219583988189697,
      "learning_rate": 2.9995837555440748e-06,
      "loss": 0.1349,
      "num_input_tokens_seen": 113536400,
      "step": 52610
    },
    {
      "epoch": 8.583197389885807,
      "grad_norm": 3.2164628505706787,
      "learning_rate": 2.9962040126478548e-06,
      "loss": 0.2181,
      "num_input_tokens_seen": 113546992,
      "step": 52615
    },
    {
      "epoch": 8.584013050570963,
      "grad_norm": 0.12123799324035645,
      "learning_rate": 2.992826053490985e-06,
      "loss": 0.0788,
      "num_input_tokens_seen": 113557808,
      "step": 52620
    },
    {
      "epoch": 8.584828711256117,
      "grad_norm": 0.0507858544588089,
      "learning_rate": 2.9894498783473e-06,
      "loss": 0.0652,
      "num_input_tokens_seen": 113568336,
      "step": 52625
    },
    {
      "epoch": 8.585644371941273,
      "grad_norm": 0.19948065280914307,
      "learning_rate": 2.986075487490486e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 113579280,
      "step": 52630
    },
    {
      "epoch": 8.586460032626427,
      "grad_norm": 0.20050232112407684,
      "learning_rate": 2.982702881194091e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 113591024,
      "step": 52635
    },
    {
      "epoch": 8.587275693311582,
      "grad_norm": 0.148842453956604,
      "learning_rate": 2.9793320597315154e-06,
      "loss": 0.0901,
      "num_input_tokens_seen": 113602032,
      "step": 52640
    },
    {
      "epoch": 8.588091353996738,
      "grad_norm": 0.08197193592786789,
      "learning_rate": 2.975963023376008e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 113612688,
      "step": 52645
    },
    {
      "epoch": 8.588907014681892,
      "grad_norm": 0.06971848756074905,
      "learning_rate": 2.9725957724006936e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 113622224,
      "step": 52650
    },
    {
      "epoch": 8.589722675367048,
      "grad_norm": 3.8267710208892822,
      "learning_rate": 2.9692303070785325e-06,
      "loss": 0.106,
      "num_input_tokens_seen": 113631920,
      "step": 52655
    },
    {
      "epoch": 8.590538336052202,
      "grad_norm": 0.08621606975793839,
      "learning_rate": 2.9658666276823427e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 113642672,
      "step": 52660
    },
    {
      "epoch": 8.591353996737357,
      "grad_norm": 0.09965116530656815,
      "learning_rate": 2.9625047344848082e-06,
      "loss": 0.0872,
      "num_input_tokens_seen": 113653456,
      "step": 52665
    },
    {
      "epoch": 8.592169657422513,
      "grad_norm": 0.7939848899841309,
      "learning_rate": 2.959144627758453e-06,
      "loss": 0.1169,
      "num_input_tokens_seen": 113664304,
      "step": 52670
    },
    {
      "epoch": 8.592985318107667,
      "grad_norm": 3.727461099624634,
      "learning_rate": 2.955786307775671e-06,
      "loss": 0.1173,
      "num_input_tokens_seen": 113675024,
      "step": 52675
    },
    {
      "epoch": 8.593800978792823,
      "grad_norm": 0.09197498857975006,
      "learning_rate": 2.9524297748087014e-06,
      "loss": 0.23,
      "num_input_tokens_seen": 113685200,
      "step": 52680
    },
    {
      "epoch": 8.594616639477977,
      "grad_norm": 0.5910281538963318,
      "learning_rate": 2.949075029129644e-06,
      "loss": 0.006,
      "num_input_tokens_seen": 113696528,
      "step": 52685
    },
    {
      "epoch": 8.595432300163132,
      "grad_norm": 0.039027437567710876,
      "learning_rate": 2.945722071010443e-06,
      "loss": 0.0739,
      "num_input_tokens_seen": 113707568,
      "step": 52690
    },
    {
      "epoch": 8.596247960848288,
      "grad_norm": 0.16804245114326477,
      "learning_rate": 2.9423709007229184e-06,
      "loss": 0.0024,
      "num_input_tokens_seen": 113719280,
      "step": 52695
    },
    {
      "epoch": 8.597063621533442,
      "grad_norm": 3.382453203201294,
      "learning_rate": 2.9390215185387287e-06,
      "loss": 0.235,
      "num_input_tokens_seen": 113729072,
      "step": 52700
    },
    {
      "epoch": 8.597879282218598,
      "grad_norm": 29.68398094177246,
      "learning_rate": 2.93567392472939e-06,
      "loss": 0.032,
      "num_input_tokens_seen": 113740784,
      "step": 52705
    },
    {
      "epoch": 8.598694942903752,
      "grad_norm": 0.04309482127428055,
      "learning_rate": 2.932328119566277e-06,
      "loss": 0.077,
      "num_input_tokens_seen": 113751088,
      "step": 52710
    },
    {
      "epoch": 8.599510603588907,
      "grad_norm": 4.468674659729004,
      "learning_rate": 2.928984103320617e-06,
      "loss": 0.3409,
      "num_input_tokens_seen": 113762672,
      "step": 52715
    },
    {
      "epoch": 8.600326264274061,
      "grad_norm": 0.09823358058929443,
      "learning_rate": 2.9256418762634936e-06,
      "loss": 0.0076,
      "num_input_tokens_seen": 113773072,
      "step": 52720
    },
    {
      "epoch": 8.601141924959217,
      "grad_norm": 0.04677252098917961,
      "learning_rate": 2.922301438665842e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 113784080,
      "step": 52725
    },
    {
      "epoch": 8.601957585644373,
      "grad_norm": 0.4189244210720062,
      "learning_rate": 2.9189627907984576e-06,
      "loss": 0.0166,
      "num_input_tokens_seen": 113794704,
      "step": 52730
    },
    {
      "epoch": 8.602773246329527,
      "grad_norm": 0.14216023683547974,
      "learning_rate": 2.9156259329319867e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 113806512,
      "step": 52735
    },
    {
      "epoch": 8.603588907014682,
      "grad_norm": 0.46621978282928467,
      "learning_rate": 2.9122908653369335e-06,
      "loss": 0.1245,
      "num_input_tokens_seen": 113817872,
      "step": 52740
    },
    {
      "epoch": 8.604404567699836,
      "grad_norm": 0.5484855771064758,
      "learning_rate": 2.908957588283656e-06,
      "loss": 0.0204,
      "num_input_tokens_seen": 113829808,
      "step": 52745
    },
    {
      "epoch": 8.605220228384992,
      "grad_norm": 0.037593141198158264,
      "learning_rate": 2.9056261020423582e-06,
      "loss": 0.0094,
      "num_input_tokens_seen": 113840656,
      "step": 52750
    },
    {
      "epoch": 8.606035889070148,
      "grad_norm": 0.0696600005030632,
      "learning_rate": 2.9022964068831204e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 113851664,
      "step": 52755
    },
    {
      "epoch": 8.606851549755302,
      "grad_norm": 0.1309024840593338,
      "learning_rate": 2.898968503075858e-06,
      "loss": 0.0082,
      "num_input_tokens_seen": 113862288,
      "step": 52760
    },
    {
      "epoch": 8.607667210440457,
      "grad_norm": 0.06248234957456589,
      "learning_rate": 2.895642390890349e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 113873456,
      "step": 52765
    },
    {
      "epoch": 8.608482871125611,
      "grad_norm": 0.036878447979688644,
      "learning_rate": 2.8923180705962226e-06,
      "loss": 0.2432,
      "num_input_tokens_seen": 113883536,
      "step": 52770
    },
    {
      "epoch": 8.609298531810767,
      "grad_norm": 0.103336863219738,
      "learning_rate": 2.888995542462969e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 113893520,
      "step": 52775
    },
    {
      "epoch": 8.61011419249592,
      "grad_norm": 0.05401739478111267,
      "learning_rate": 2.885674806759925e-06,
      "loss": 0.2536,
      "num_input_tokens_seen": 113904592,
      "step": 52780
    },
    {
      "epoch": 8.610929853181077,
      "grad_norm": 0.07610581070184708,
      "learning_rate": 2.8823558637562893e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 113916272,
      "step": 52785
    },
    {
      "epoch": 8.611745513866232,
      "grad_norm": 0.04994501173496246,
      "learning_rate": 2.8790387137211105e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 113928336,
      "step": 52790
    },
    {
      "epoch": 8.612561174551386,
      "grad_norm": 0.11280269175767899,
      "learning_rate": 2.8757233569232933e-06,
      "loss": 0.1173,
      "num_input_tokens_seen": 113939920,
      "step": 52795
    },
    {
      "epoch": 8.613376835236542,
      "grad_norm": 0.12876686453819275,
      "learning_rate": 2.8724097936316004e-06,
      "loss": 0.08,
      "num_input_tokens_seen": 113949616,
      "step": 52800
    },
    {
      "epoch": 8.614192495921696,
      "grad_norm": 0.04100547358393669,
      "learning_rate": 2.8690980241146415e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 113959408,
      "step": 52805
    },
    {
      "epoch": 8.615008156606851,
      "grad_norm": 0.05942099168896675,
      "learning_rate": 2.8657880486408884e-06,
      "loss": 0.2803,
      "num_input_tokens_seen": 113970736,
      "step": 52810
    },
    {
      "epoch": 8.615823817292007,
      "grad_norm": 0.2773951590061188,
      "learning_rate": 2.862479867478665e-06,
      "loss": 0.097,
      "num_input_tokens_seen": 113981008,
      "step": 52815
    },
    {
      "epoch": 8.616639477977161,
      "grad_norm": 0.12048663944005966,
      "learning_rate": 2.859173480896149e-06,
      "loss": 0.0705,
      "num_input_tokens_seen": 113992816,
      "step": 52820
    },
    {
      "epoch": 8.617455138662317,
      "grad_norm": 0.11280005425214767,
      "learning_rate": 2.85586888916137e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 114005392,
      "step": 52825
    },
    {
      "epoch": 8.61827079934747,
      "grad_norm": 0.09502550214529037,
      "learning_rate": 2.852566092542211e-06,
      "loss": 0.007,
      "num_input_tokens_seen": 114015600,
      "step": 52830
    },
    {
      "epoch": 8.619086460032626,
      "grad_norm": 4.35811710357666,
      "learning_rate": 2.8492650913064274e-06,
      "loss": 0.2785,
      "num_input_tokens_seen": 114026448,
      "step": 52835
    },
    {
      "epoch": 8.619902120717782,
      "grad_norm": 21.104766845703125,
      "learning_rate": 2.8459658857216074e-06,
      "loss": 0.0709,
      "num_input_tokens_seen": 114037552,
      "step": 52840
    },
    {
      "epoch": 8.620717781402936,
      "grad_norm": 0.09337308257818222,
      "learning_rate": 2.8426684760551993e-06,
      "loss": 0.1148,
      "num_input_tokens_seen": 114049072,
      "step": 52845
    },
    {
      "epoch": 8.621533442088092,
      "grad_norm": 0.05295537784695625,
      "learning_rate": 2.83937286257451e-06,
      "loss": 0.1148,
      "num_input_tokens_seen": 114059824,
      "step": 52850
    },
    {
      "epoch": 8.622349102773246,
      "grad_norm": 0.098955437541008,
      "learning_rate": 2.8360790455466996e-06,
      "loss": 0.1785,
      "num_input_tokens_seen": 114070960,
      "step": 52855
    },
    {
      "epoch": 8.623164763458401,
      "grad_norm": 0.09193826466798782,
      "learning_rate": 2.8327870252387727e-06,
      "loss": 0.0114,
      "num_input_tokens_seen": 114082224,
      "step": 52860
    },
    {
      "epoch": 8.623980424143557,
      "grad_norm": 0.0787716954946518,
      "learning_rate": 2.829496801917611e-06,
      "loss": 0.1159,
      "num_input_tokens_seen": 114091760,
      "step": 52865
    },
    {
      "epoch": 8.624796084828711,
      "grad_norm": 2.387537956237793,
      "learning_rate": 2.826208375849931e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 114100912,
      "step": 52870
    },
    {
      "epoch": 8.625611745513867,
      "grad_norm": 0.061954595148563385,
      "learning_rate": 2.8229217473023094e-06,
      "loss": 0.1305,
      "num_input_tokens_seen": 114111856,
      "step": 52875
    },
    {
      "epoch": 8.62642740619902,
      "grad_norm": 0.1902577131986618,
      "learning_rate": 2.8196369165411767e-06,
      "loss": 0.1039,
      "num_input_tokens_seen": 114121872,
      "step": 52880
    },
    {
      "epoch": 8.627243066884176,
      "grad_norm": 0.38934525847435,
      "learning_rate": 2.8163538838328176e-06,
      "loss": 0.1306,
      "num_input_tokens_seen": 114132304,
      "step": 52885
    },
    {
      "epoch": 8.62805872756933,
      "grad_norm": 0.4867192804813385,
      "learning_rate": 2.8130726494433684e-06,
      "loss": 0.1491,
      "num_input_tokens_seen": 114142160,
      "step": 52890
    },
    {
      "epoch": 8.628874388254486,
      "grad_norm": 0.07474728673696518,
      "learning_rate": 2.8097932136388285e-06,
      "loss": 0.1089,
      "num_input_tokens_seen": 114153392,
      "step": 52895
    },
    {
      "epoch": 8.629690048939642,
      "grad_norm": 0.16183558106422424,
      "learning_rate": 2.8065155766850425e-06,
      "loss": 0.0882,
      "num_input_tokens_seen": 114165648,
      "step": 52900
    },
    {
      "epoch": 8.630505709624796,
      "grad_norm": 0.5834521651268005,
      "learning_rate": 2.8032397388477098e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 114175248,
      "step": 52905
    },
    {
      "epoch": 8.631321370309951,
      "grad_norm": 3.95804500579834,
      "learning_rate": 2.799965700392393e-06,
      "loss": 0.3783,
      "num_input_tokens_seen": 114186256,
      "step": 52910
    },
    {
      "epoch": 8.632137030995105,
      "grad_norm": 0.2856743037700653,
      "learning_rate": 2.7966934615844957e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 114197264,
      "step": 52915
    },
    {
      "epoch": 8.632952691680261,
      "grad_norm": 0.13744287192821503,
      "learning_rate": 2.793423022689284e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 114208944,
      "step": 52920
    },
    {
      "epoch": 8.633768352365417,
      "grad_norm": 0.08259768038988113,
      "learning_rate": 2.7901543839718795e-06,
      "loss": 0.0328,
      "num_input_tokens_seen": 114219504,
      "step": 52925
    },
    {
      "epoch": 8.63458401305057,
      "grad_norm": 0.20659682154655457,
      "learning_rate": 2.7868875456972534e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 114229072,
      "step": 52930
    },
    {
      "epoch": 8.635399673735726,
      "grad_norm": 0.15795865654945374,
      "learning_rate": 2.783622508130229e-06,
      "loss": 0.0446,
      "num_input_tokens_seen": 114238768,
      "step": 52935
    },
    {
      "epoch": 8.63621533442088,
      "grad_norm": 0.023627901449799538,
      "learning_rate": 2.7803592715354877e-06,
      "loss": 0.0017,
      "num_input_tokens_seen": 114250256,
      "step": 52940
    },
    {
      "epoch": 8.637030995106036,
      "grad_norm": 5.866702556610107,
      "learning_rate": 2.7770978361775667e-06,
      "loss": 0.16,
      "num_input_tokens_seen": 114259696,
      "step": 52945
    },
    {
      "epoch": 8.63784665579119,
      "grad_norm": 0.08152702450752258,
      "learning_rate": 2.7738382023208526e-06,
      "loss": 0.1679,
      "num_input_tokens_seen": 114270288,
      "step": 52950
    },
    {
      "epoch": 8.638662316476346,
      "grad_norm": 0.060370657593011856,
      "learning_rate": 2.770580370229589e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 114280784,
      "step": 52955
    },
    {
      "epoch": 8.639477977161501,
      "grad_norm": 8.027362823486328,
      "learning_rate": 2.7673243401678704e-06,
      "loss": 0.0089,
      "num_input_tokens_seen": 114290832,
      "step": 52960
    },
    {
      "epoch": 8.640293637846655,
      "grad_norm": 0.32207539677619934,
      "learning_rate": 2.7640701123996445e-06,
      "loss": 0.0645,
      "num_input_tokens_seen": 114301648,
      "step": 52965
    },
    {
      "epoch": 8.641109298531811,
      "grad_norm": 0.05956178158521652,
      "learning_rate": 2.7608176871887242e-06,
      "loss": 0.0966,
      "num_input_tokens_seen": 114311728,
      "step": 52970
    },
    {
      "epoch": 8.641924959216965,
      "grad_norm": 0.08997718244791031,
      "learning_rate": 2.7575670647987606e-06,
      "loss": 0.2073,
      "num_input_tokens_seen": 114322512,
      "step": 52975
    },
    {
      "epoch": 8.64274061990212,
      "grad_norm": 0.316913366317749,
      "learning_rate": 2.7543182454932705e-06,
      "loss": 0.1184,
      "num_input_tokens_seen": 114332592,
      "step": 52980
    },
    {
      "epoch": 8.643556280587276,
      "grad_norm": 3.2485742568969727,
      "learning_rate": 2.751071229535615e-06,
      "loss": 0.241,
      "num_input_tokens_seen": 114343696,
      "step": 52985
    },
    {
      "epoch": 8.64437194127243,
      "grad_norm": 0.04890444874763489,
      "learning_rate": 2.7478260171890175e-06,
      "loss": 0.0979,
      "num_input_tokens_seen": 114354928,
      "step": 52990
    },
    {
      "epoch": 8.645187601957586,
      "grad_norm": 0.46548447012901306,
      "learning_rate": 2.744582608716548e-06,
      "loss": 0.0733,
      "num_input_tokens_seen": 114365392,
      "step": 52995
    },
    {
      "epoch": 8.64600326264274,
      "grad_norm": 0.14548994600772858,
      "learning_rate": 2.741341004381129e-06,
      "loss": 0.0863,
      "num_input_tokens_seen": 114374992,
      "step": 53000
    },
    {
      "epoch": 8.646818923327896,
      "grad_norm": 0.029076898470520973,
      "learning_rate": 2.7381012044455535e-06,
      "loss": 0.0026,
      "num_input_tokens_seen": 114385808,
      "step": 53005
    },
    {
      "epoch": 8.647634584013051,
      "grad_norm": 0.27490729093551636,
      "learning_rate": 2.73486320917245e-06,
      "loss": 0.0554,
      "num_input_tokens_seen": 114397072,
      "step": 53010
    },
    {
      "epoch": 8.648450244698205,
      "grad_norm": 0.08811581134796143,
      "learning_rate": 2.7316270188243064e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 114408624,
      "step": 53015
    },
    {
      "epoch": 8.649265905383361,
      "grad_norm": 0.056794021278619766,
      "learning_rate": 2.728392633663468e-06,
      "loss": 0.191,
      "num_input_tokens_seen": 114418512,
      "step": 53020
    },
    {
      "epoch": 8.650081566068515,
      "grad_norm": 0.9381062984466553,
      "learning_rate": 2.7251600539521248e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 114427792,
      "step": 53025
    },
    {
      "epoch": 8.65089722675367,
      "grad_norm": 0.1318144053220749,
      "learning_rate": 2.7219292799523316e-06,
      "loss": 0.3686,
      "num_input_tokens_seen": 114439888,
      "step": 53030
    },
    {
      "epoch": 8.651712887438826,
      "grad_norm": 0.10454489290714264,
      "learning_rate": 2.718700311925987e-06,
      "loss": 0.0994,
      "num_input_tokens_seen": 114451280,
      "step": 53035
    },
    {
      "epoch": 8.65252854812398,
      "grad_norm": 0.026726340875029564,
      "learning_rate": 2.715473150134848e-06,
      "loss": 0.0681,
      "num_input_tokens_seen": 114463376,
      "step": 53040
    },
    {
      "epoch": 8.653344208809136,
      "grad_norm": 0.05666026845574379,
      "learning_rate": 2.7122477948405277e-06,
      "loss": 0.0106,
      "num_input_tokens_seen": 114474896,
      "step": 53045
    },
    {
      "epoch": 8.65415986949429,
      "grad_norm": 0.07485075294971466,
      "learning_rate": 2.7090242463044896e-06,
      "loss": 0.0167,
      "num_input_tokens_seen": 114486000,
      "step": 53050
    },
    {
      "epoch": 8.654975530179446,
      "grad_norm": 0.038886334747076035,
      "learning_rate": 2.7058025047880466e-06,
      "loss": 0.0082,
      "num_input_tokens_seen": 114497456,
      "step": 53055
    },
    {
      "epoch": 8.655791190864601,
      "grad_norm": 0.10611975938081741,
      "learning_rate": 2.702582570552373e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 114508368,
      "step": 53060
    },
    {
      "epoch": 8.656606851549755,
      "grad_norm": 0.4545610547065735,
      "learning_rate": 2.699364443858493e-06,
      "loss": 0.0965,
      "num_input_tokens_seen": 114519888,
      "step": 53065
    },
    {
      "epoch": 8.65742251223491,
      "grad_norm": 0.1703397035598755,
      "learning_rate": 2.6961481249672765e-06,
      "loss": 0.1311,
      "num_input_tokens_seen": 114529712,
      "step": 53070
    },
    {
      "epoch": 8.658238172920065,
      "grad_norm": 0.07904315739870071,
      "learning_rate": 2.69293361413947e-06,
      "loss": 0.0991,
      "num_input_tokens_seen": 114540528,
      "step": 53075
    },
    {
      "epoch": 8.65905383360522,
      "grad_norm": 0.15389762818813324,
      "learning_rate": 2.6897209116356457e-06,
      "loss": 0.1363,
      "num_input_tokens_seen": 114552016,
      "step": 53080
    },
    {
      "epoch": 8.659869494290374,
      "grad_norm": 0.07232028245925903,
      "learning_rate": 2.6865100177162484e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 114562672,
      "step": 53085
    },
    {
      "epoch": 8.66068515497553,
      "grad_norm": 0.04487145319581032,
      "learning_rate": 2.6833009326415663e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 114574320,
      "step": 53090
    },
    {
      "epoch": 8.661500815660686,
      "grad_norm": 0.04890063777565956,
      "learning_rate": 2.680093656671745e-06,
      "loss": 0.1374,
      "num_input_tokens_seen": 114585456,
      "step": 53095
    },
    {
      "epoch": 8.66231647634584,
      "grad_norm": 0.0658997967839241,
      "learning_rate": 2.6768881900667787e-06,
      "loss": 0.1267,
      "num_input_tokens_seen": 114596400,
      "step": 53100
    },
    {
      "epoch": 8.663132137030995,
      "grad_norm": 0.5864744782447815,
      "learning_rate": 2.673684533086526e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 114607120,
      "step": 53105
    },
    {
      "epoch": 8.66394779771615,
      "grad_norm": 0.7246565222740173,
      "learning_rate": 2.6704826859906858e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 114617808,
      "step": 53110
    },
    {
      "epoch": 8.664763458401305,
      "grad_norm": 0.18121881783008575,
      "learning_rate": 2.667282649038816e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 114627664,
      "step": 53115
    },
    {
      "epoch": 8.66557911908646,
      "grad_norm": 0.06088462844491005,
      "learning_rate": 2.6640844224903318e-06,
      "loss": 0.1073,
      "num_input_tokens_seen": 114638064,
      "step": 53120
    },
    {
      "epoch": 8.666394779771615,
      "grad_norm": 0.026404354721307755,
      "learning_rate": 2.660888006604498e-06,
      "loss": 0.1699,
      "num_input_tokens_seen": 114647760,
      "step": 53125
    },
    {
      "epoch": 8.66721044045677,
      "grad_norm": 7.807608127593994,
      "learning_rate": 2.6576934016404264e-06,
      "loss": 0.1214,
      "num_input_tokens_seen": 114658192,
      "step": 53130
    },
    {
      "epoch": 8.668026101141924,
      "grad_norm": 0.01555652916431427,
      "learning_rate": 2.654500607857091e-06,
      "loss": 0.0022,
      "num_input_tokens_seen": 114669392,
      "step": 53135
    },
    {
      "epoch": 8.66884176182708,
      "grad_norm": 0.2186608761548996,
      "learning_rate": 2.651309625513318e-06,
      "loss": 0.0641,
      "num_input_tokens_seen": 114681168,
      "step": 53140
    },
    {
      "epoch": 8.669657422512234,
      "grad_norm": 0.9420276284217834,
      "learning_rate": 2.648120454867778e-06,
      "loss": 0.0112,
      "num_input_tokens_seen": 114692688,
      "step": 53145
    },
    {
      "epoch": 8.67047308319739,
      "grad_norm": 1.198050856590271,
      "learning_rate": 2.6449330961790116e-06,
      "loss": 0.0428,
      "num_input_tokens_seen": 114703472,
      "step": 53150
    },
    {
      "epoch": 8.671288743882545,
      "grad_norm": 0.06262125819921494,
      "learning_rate": 2.641747549705395e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 114714704,
      "step": 53155
    },
    {
      "epoch": 8.6721044045677,
      "grad_norm": 0.08199375122785568,
      "learning_rate": 2.638563815705167e-06,
      "loss": 0.0028,
      "num_input_tokens_seen": 114725200,
      "step": 53160
    },
    {
      "epoch": 8.672920065252855,
      "grad_norm": 0.22733455896377563,
      "learning_rate": 2.635381894436417e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 114735568,
      "step": 53165
    },
    {
      "epoch": 8.673735725938009,
      "grad_norm": 0.05899897217750549,
      "learning_rate": 2.63220178615709e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 114746448,
      "step": 53170
    },
    {
      "epoch": 8.674551386623165,
      "grad_norm": 0.08533771336078644,
      "learning_rate": 2.629023491124971e-06,
      "loss": 0.1314,
      "num_input_tokens_seen": 114758704,
      "step": 53175
    },
    {
      "epoch": 8.67536704730832,
      "grad_norm": 0.09958028048276901,
      "learning_rate": 2.6258470095977262e-06,
      "loss": 0.1571,
      "num_input_tokens_seen": 114769744,
      "step": 53180
    },
    {
      "epoch": 8.676182707993474,
      "grad_norm": 2.5879385471343994,
      "learning_rate": 2.6226723418328437e-06,
      "loss": 0.0075,
      "num_input_tokens_seen": 114780112,
      "step": 53185
    },
    {
      "epoch": 8.67699836867863,
      "grad_norm": 3.7749695777893066,
      "learning_rate": 2.6194994880876843e-06,
      "loss": 0.1072,
      "num_input_tokens_seen": 114790384,
      "step": 53190
    },
    {
      "epoch": 8.677814029363784,
      "grad_norm": 0.06751300394535065,
      "learning_rate": 2.616328448619454e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 114801168,
      "step": 53195
    },
    {
      "epoch": 8.67862969004894,
      "grad_norm": 3.2267613410949707,
      "learning_rate": 2.613159223685213e-06,
      "loss": 0.1381,
      "num_input_tokens_seen": 114812560,
      "step": 53200
    },
    {
      "epoch": 8.679445350734095,
      "grad_norm": 0.16413560509681702,
      "learning_rate": 2.609991813541876e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 114823664,
      "step": 53205
    },
    {
      "epoch": 8.68026101141925,
      "grad_norm": 34.654502868652344,
      "learning_rate": 2.6068262184462066e-06,
      "loss": 0.0904,
      "num_input_tokens_seen": 114835056,
      "step": 53210
    },
    {
      "epoch": 8.681076672104405,
      "grad_norm": 0.064839206635952,
      "learning_rate": 2.6036624386548277e-06,
      "loss": 0.2715,
      "num_input_tokens_seen": 114845744,
      "step": 53215
    },
    {
      "epoch": 8.681892332789559,
      "grad_norm": 0.3497132956981659,
      "learning_rate": 2.6005004744242082e-06,
      "loss": 0.0927,
      "num_input_tokens_seen": 114857072,
      "step": 53220
    },
    {
      "epoch": 8.682707993474715,
      "grad_norm": 0.24611487984657288,
      "learning_rate": 2.597340326010675e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 114868304,
      "step": 53225
    },
    {
      "epoch": 8.68352365415987,
      "grad_norm": 0.45386627316474915,
      "learning_rate": 2.5941819936704053e-06,
      "loss": 0.0986,
      "num_input_tokens_seen": 114877360,
      "step": 53230
    },
    {
      "epoch": 8.684339314845024,
      "grad_norm": 0.04542342573404312,
      "learning_rate": 2.5910254776594256e-06,
      "loss": 0.0318,
      "num_input_tokens_seen": 114888496,
      "step": 53235
    },
    {
      "epoch": 8.68515497553018,
      "grad_norm": 0.3080884516239166,
      "learning_rate": 2.587870778233625e-06,
      "loss": 0.1251,
      "num_input_tokens_seen": 114900656,
      "step": 53240
    },
    {
      "epoch": 8.685970636215334,
      "grad_norm": 0.07289087027311325,
      "learning_rate": 2.584717895648739e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 114912208,
      "step": 53245
    },
    {
      "epoch": 8.68678629690049,
      "grad_norm": 5.651989459991455,
      "learning_rate": 2.5815668301603537e-06,
      "loss": 0.1794,
      "num_input_tokens_seen": 114924144,
      "step": 53250
    },
    {
      "epoch": 8.687601957585644,
      "grad_norm": 10.517882347106934,
      "learning_rate": 2.5784175820239094e-06,
      "loss": 0.1098,
      "num_input_tokens_seen": 114936496,
      "step": 53255
    },
    {
      "epoch": 8.6884176182708,
      "grad_norm": 0.11167328804731369,
      "learning_rate": 2.575270151494702e-06,
      "loss": 0.257,
      "num_input_tokens_seen": 114946928,
      "step": 53260
    },
    {
      "epoch": 8.689233278955955,
      "grad_norm": 0.11391818523406982,
      "learning_rate": 2.5721245388278805e-06,
      "loss": 0.1043,
      "num_input_tokens_seen": 114957392,
      "step": 53265
    },
    {
      "epoch": 8.690048939641109,
      "grad_norm": 0.03107847273349762,
      "learning_rate": 2.5689807442784404e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 114968816,
      "step": 53270
    },
    {
      "epoch": 8.690864600326265,
      "grad_norm": 0.26515620946884155,
      "learning_rate": 2.5658387681012337e-06,
      "loss": 0.0102,
      "num_input_tokens_seen": 114979408,
      "step": 53275
    },
    {
      "epoch": 8.691680261011419,
      "grad_norm": 2.514280080795288,
      "learning_rate": 2.5626986105509677e-06,
      "loss": 0.0753,
      "num_input_tokens_seen": 114989776,
      "step": 53280
    },
    {
      "epoch": 8.692495921696574,
      "grad_norm": 0.1590150147676468,
      "learning_rate": 2.5595602718821916e-06,
      "loss": 0.1038,
      "num_input_tokens_seen": 115000976,
      "step": 53285
    },
    {
      "epoch": 8.69331158238173,
      "grad_norm": 0.30125266313552856,
      "learning_rate": 2.5564237523493295e-06,
      "loss": 0.1155,
      "num_input_tokens_seen": 115010608,
      "step": 53290
    },
    {
      "epoch": 8.694127243066884,
      "grad_norm": 0.1378786861896515,
      "learning_rate": 2.553289052206634e-06,
      "loss": 0.0111,
      "num_input_tokens_seen": 115021904,
      "step": 53295
    },
    {
      "epoch": 8.69494290375204,
      "grad_norm": 4.505702972412109,
      "learning_rate": 2.5501561717082204e-06,
      "loss": 0.0923,
      "num_input_tokens_seen": 115032240,
      "step": 53300
    },
    {
      "epoch": 8.695758564437194,
      "grad_norm": 0.0683271661400795,
      "learning_rate": 2.547025111108056e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 115043024,
      "step": 53305
    },
    {
      "epoch": 8.69657422512235,
      "grad_norm": 3.7424323558807373,
      "learning_rate": 2.5438958706599623e-06,
      "loss": 0.1452,
      "num_input_tokens_seen": 115053584,
      "step": 53310
    },
    {
      "epoch": 8.697389885807503,
      "grad_norm": 0.07999517768621445,
      "learning_rate": 2.540768450617609e-06,
      "loss": 0.2054,
      "num_input_tokens_seen": 115064368,
      "step": 53315
    },
    {
      "epoch": 8.698205546492659,
      "grad_norm": 0.830254077911377,
      "learning_rate": 2.537642851234523e-06,
      "loss": 0.0569,
      "num_input_tokens_seen": 115073712,
      "step": 53320
    },
    {
      "epoch": 8.699021207177815,
      "grad_norm": 9.407503128051758,
      "learning_rate": 2.5345190727640828e-06,
      "loss": 0.0531,
      "num_input_tokens_seen": 115085616,
      "step": 53325
    },
    {
      "epoch": 8.699836867862969,
      "grad_norm": 3.5490689277648926,
      "learning_rate": 2.5313971154595135e-06,
      "loss": 0.1871,
      "num_input_tokens_seen": 115096080,
      "step": 53330
    },
    {
      "epoch": 8.700652528548124,
      "grad_norm": 0.08478696644306183,
      "learning_rate": 2.5282769795738987e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 115107344,
      "step": 53335
    },
    {
      "epoch": 8.701468189233278,
      "grad_norm": 62.311641693115234,
      "learning_rate": 2.5251586653601722e-06,
      "loss": 0.1402,
      "num_input_tokens_seen": 115117712,
      "step": 53340
    },
    {
      "epoch": 8.702283849918434,
      "grad_norm": 3.5588507652282715,
      "learning_rate": 2.522042173071121e-06,
      "loss": 0.2631,
      "num_input_tokens_seen": 115128048,
      "step": 53345
    },
    {
      "epoch": 8.70309951060359,
      "grad_norm": 0.06941116601228714,
      "learning_rate": 2.518927502959384e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 115139152,
      "step": 53350
    },
    {
      "epoch": 8.703915171288743,
      "grad_norm": 6.748194694519043,
      "learning_rate": 2.5158146552774486e-06,
      "loss": 0.2439,
      "num_input_tokens_seen": 115150032,
      "step": 53355
    },
    {
      "epoch": 8.7047308319739,
      "grad_norm": 0.0572577528655529,
      "learning_rate": 2.512703630277663e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 115159536,
      "step": 53360
    },
    {
      "epoch": 8.705546492659053,
      "grad_norm": 0.04039865359663963,
      "learning_rate": 2.5095944282122226e-06,
      "loss": 0.1804,
      "num_input_tokens_seen": 115170576,
      "step": 53365
    },
    {
      "epoch": 8.706362153344209,
      "grad_norm": 0.018278568983078003,
      "learning_rate": 2.50648704933317e-06,
      "loss": 0.1922,
      "num_input_tokens_seen": 115182160,
      "step": 53370
    },
    {
      "epoch": 8.707177814029365,
      "grad_norm": 0.0602726973593235,
      "learning_rate": 2.5033814938924095e-06,
      "loss": 0.1083,
      "num_input_tokens_seen": 115190960,
      "step": 53375
    },
    {
      "epoch": 8.707993474714518,
      "grad_norm": 0.10237548500299454,
      "learning_rate": 2.500277762141692e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 115201680,
      "step": 53380
    },
    {
      "epoch": 8.708809135399674,
      "grad_norm": 9.355179786682129,
      "learning_rate": 2.49717585433262e-06,
      "loss": 0.0525,
      "num_input_tokens_seen": 115212496,
      "step": 53385
    },
    {
      "epoch": 8.709624796084828,
      "grad_norm": 2.9365639686584473,
      "learning_rate": 2.4940757707166474e-06,
      "loss": 0.1061,
      "num_input_tokens_seen": 115223888,
      "step": 53390
    },
    {
      "epoch": 8.710440456769984,
      "grad_norm": 0.06928296387195587,
      "learning_rate": 2.490977511545092e-06,
      "loss": 0.1105,
      "num_input_tokens_seen": 115233616,
      "step": 53395
    },
    {
      "epoch": 8.71125611745514,
      "grad_norm": 0.10657060146331787,
      "learning_rate": 2.4878810770691096e-06,
      "loss": 0.0438,
      "num_input_tokens_seen": 115244496,
      "step": 53400
    },
    {
      "epoch": 8.712071778140293,
      "grad_norm": 1.0137279033660889,
      "learning_rate": 2.48478646753971e-06,
      "loss": 0.1113,
      "num_input_tokens_seen": 115255216,
      "step": 53405
    },
    {
      "epoch": 8.71288743882545,
      "grad_norm": 0.08584178239107132,
      "learning_rate": 2.4816936832077615e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 115264720,
      "step": 53410
    },
    {
      "epoch": 8.713703099510603,
      "grad_norm": 0.23638053238391876,
      "learning_rate": 2.478602724323981e-06,
      "loss": 0.2156,
      "num_input_tokens_seen": 115276080,
      "step": 53415
    },
    {
      "epoch": 8.714518760195759,
      "grad_norm": 0.1255992203950882,
      "learning_rate": 2.4755135911389364e-06,
      "loss": 0.016,
      "num_input_tokens_seen": 115286576,
      "step": 53420
    },
    {
      "epoch": 8.715334420880914,
      "grad_norm": 0.11487893015146255,
      "learning_rate": 2.472426283903048e-06,
      "loss": 0.0873,
      "num_input_tokens_seen": 115297200,
      "step": 53425
    },
    {
      "epoch": 8.716150081566068,
      "grad_norm": 0.05824054032564163,
      "learning_rate": 2.4693408028665878e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 115308240,
      "step": 53430
    },
    {
      "epoch": 8.716965742251224,
      "grad_norm": 0.1160576343536377,
      "learning_rate": 2.4662571482796797e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 115319600,
      "step": 53435
    },
    {
      "epoch": 8.717781402936378,
      "grad_norm": 0.394844651222229,
      "learning_rate": 2.4631753203923052e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 115330736,
      "step": 53440
    },
    {
      "epoch": 8.718597063621534,
      "grad_norm": 0.030199168249964714,
      "learning_rate": 2.460095319454289e-06,
      "loss": 0.2393,
      "num_input_tokens_seen": 115342672,
      "step": 53445
    },
    {
      "epoch": 8.719412724306688,
      "grad_norm": 0.010499256663024426,
      "learning_rate": 2.4570171457153123e-06,
      "loss": 0.2822,
      "num_input_tokens_seen": 115354320,
      "step": 53450
    },
    {
      "epoch": 8.720228384991843,
      "grad_norm": 3.4621529579162598,
      "learning_rate": 2.4539407994249088e-06,
      "loss": 0.3082,
      "num_input_tokens_seen": 115364880,
      "step": 53455
    },
    {
      "epoch": 8.721044045676999,
      "grad_norm": 0.05974346771836281,
      "learning_rate": 2.450866280832456e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 115375824,
      "step": 53460
    },
    {
      "epoch": 8.721859706362153,
      "grad_norm": 0.049211133271455765,
      "learning_rate": 2.4477935901872e-06,
      "loss": 0.0997,
      "num_input_tokens_seen": 115386992,
      "step": 53465
    },
    {
      "epoch": 8.722675367047309,
      "grad_norm": 0.07548020035028458,
      "learning_rate": 2.4447227277382244e-06,
      "loss": 0.1105,
      "num_input_tokens_seen": 115397008,
      "step": 53470
    },
    {
      "epoch": 8.723491027732463,
      "grad_norm": 0.15559165179729462,
      "learning_rate": 2.441653693734472e-06,
      "loss": 0.1576,
      "num_input_tokens_seen": 115407888,
      "step": 53475
    },
    {
      "epoch": 8.724306688417618,
      "grad_norm": 0.1762966513633728,
      "learning_rate": 2.438586488424727e-06,
      "loss": 0.1806,
      "num_input_tokens_seen": 115417936,
      "step": 53480
    },
    {
      "epoch": 8.725122349102774,
      "grad_norm": 0.4542035162448883,
      "learning_rate": 2.435521112057637e-06,
      "loss": 0.071,
      "num_input_tokens_seen": 115429200,
      "step": 53485
    },
    {
      "epoch": 8.725938009787928,
      "grad_norm": 0.06879319995641708,
      "learning_rate": 2.432457564881699e-06,
      "loss": 0.1011,
      "num_input_tokens_seen": 115439920,
      "step": 53490
    },
    {
      "epoch": 8.726753670473084,
      "grad_norm": 1.1868400573730469,
      "learning_rate": 2.429395847145252e-06,
      "loss": 0.0097,
      "num_input_tokens_seen": 115450608,
      "step": 53495
    },
    {
      "epoch": 8.727569331158238,
      "grad_norm": 0.3320519030094147,
      "learning_rate": 2.4263359590965042e-06,
      "loss": 0.0099,
      "num_input_tokens_seen": 115461680,
      "step": 53500
    },
    {
      "epoch": 8.728384991843393,
      "grad_norm": 0.23768679797649384,
      "learning_rate": 2.4232779009835006e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 115472752,
      "step": 53505
    },
    {
      "epoch": 8.729200652528547,
      "grad_norm": 0.17958806455135345,
      "learning_rate": 2.420221673054143e-06,
      "loss": 0.1738,
      "num_input_tokens_seen": 115483536,
      "step": 53510
    },
    {
      "epoch": 8.730016313213703,
      "grad_norm": 0.049967117607593536,
      "learning_rate": 2.417167275556187e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 115494736,
      "step": 53515
    },
    {
      "epoch": 8.730831973898859,
      "grad_norm": 0.06931749731302261,
      "learning_rate": 2.4141147087372336e-06,
      "loss": 0.125,
      "num_input_tokens_seen": 115505552,
      "step": 53520
    },
    {
      "epoch": 8.731647634584013,
      "grad_norm": 0.07650643587112427,
      "learning_rate": 2.4110639728447433e-06,
      "loss": 0.1952,
      "num_input_tokens_seen": 115515920,
      "step": 53525
    },
    {
      "epoch": 8.732463295269168,
      "grad_norm": 3.825303792953491,
      "learning_rate": 2.4080150681260212e-06,
      "loss": 0.2009,
      "num_input_tokens_seen": 115526896,
      "step": 53530
    },
    {
      "epoch": 8.733278955954322,
      "grad_norm": 2.012338161468506,
      "learning_rate": 2.4049679948282305e-06,
      "loss": 0.1668,
      "num_input_tokens_seen": 115536912,
      "step": 53535
    },
    {
      "epoch": 8.734094616639478,
      "grad_norm": 0.3810151517391205,
      "learning_rate": 2.401922753198377e-06,
      "loss": 0.0103,
      "num_input_tokens_seen": 115547728,
      "step": 53540
    },
    {
      "epoch": 8.734910277324634,
      "grad_norm": 0.1658923178911209,
      "learning_rate": 2.398879343483329e-06,
      "loss": 0.0388,
      "num_input_tokens_seen": 115559184,
      "step": 53545
    },
    {
      "epoch": 8.735725938009788,
      "grad_norm": 0.03131018951535225,
      "learning_rate": 2.3958377659297983e-06,
      "loss": 0.1574,
      "num_input_tokens_seen": 115569552,
      "step": 53550
    },
    {
      "epoch": 8.736541598694943,
      "grad_norm": 5.776989936828613,
      "learning_rate": 2.392798020784348e-06,
      "loss": 0.1313,
      "num_input_tokens_seen": 115578928,
      "step": 53555
    },
    {
      "epoch": 8.737357259380097,
      "grad_norm": 0.09969309717416763,
      "learning_rate": 2.3897601082934013e-06,
      "loss": 0.017,
      "num_input_tokens_seen": 115589904,
      "step": 53560
    },
    {
      "epoch": 8.738172920065253,
      "grad_norm": 0.2182675451040268,
      "learning_rate": 2.3867240287032214e-06,
      "loss": 0.006,
      "num_input_tokens_seen": 115599440,
      "step": 53565
    },
    {
      "epoch": 8.738988580750409,
      "grad_norm": 0.33458808064460754,
      "learning_rate": 2.3836897822599317e-06,
      "loss": 0.1146,
      "num_input_tokens_seen": 115610064,
      "step": 53570
    },
    {
      "epoch": 8.739804241435563,
      "grad_norm": 3.599860191345215,
      "learning_rate": 2.380657369209502e-06,
      "loss": 0.2455,
      "num_input_tokens_seen": 115621360,
      "step": 53575
    },
    {
      "epoch": 8.740619902120718,
      "grad_norm": 0.0940905213356018,
      "learning_rate": 2.3776267897977543e-06,
      "loss": 0.1025,
      "num_input_tokens_seen": 115632272,
      "step": 53580
    },
    {
      "epoch": 8.741435562805872,
      "grad_norm": 0.0359291210770607,
      "learning_rate": 2.374598044270365e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 115643280,
      "step": 53585
    },
    {
      "epoch": 8.742251223491028,
      "grad_norm": 0.03600488603115082,
      "learning_rate": 2.3715711328728575e-06,
      "loss": 0.0069,
      "num_input_tokens_seen": 115654128,
      "step": 53590
    },
    {
      "epoch": 8.743066884176184,
      "grad_norm": 0.08532227575778961,
      "learning_rate": 2.3685460558506097e-06,
      "loss": 0.0088,
      "num_input_tokens_seen": 115664528,
      "step": 53595
    },
    {
      "epoch": 8.743882544861338,
      "grad_norm": 5.841217517852783,
      "learning_rate": 2.3655228134488505e-06,
      "loss": 0.0946,
      "num_input_tokens_seen": 115675024,
      "step": 53600
    },
    {
      "epoch": 8.744698205546493,
      "grad_norm": 0.0855046808719635,
      "learning_rate": 2.362501405912651e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 115687056,
      "step": 53605
    },
    {
      "epoch": 8.745513866231647,
      "grad_norm": 0.10552932322025299,
      "learning_rate": 2.3594818334869568e-06,
      "loss": 0.1236,
      "num_input_tokens_seen": 115697840,
      "step": 53610
    },
    {
      "epoch": 8.746329526916803,
      "grad_norm": 0.3092189133167267,
      "learning_rate": 2.3564640964165386e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 115708848,
      "step": 53615
    },
    {
      "epoch": 8.747145187601957,
      "grad_norm": 0.2367440164089203,
      "learning_rate": 2.353448194946037e-06,
      "loss": 0.3288,
      "num_input_tokens_seen": 115719568,
      "step": 53620
    },
    {
      "epoch": 8.747960848287113,
      "grad_norm": 0.07751596719026566,
      "learning_rate": 2.350434129319928e-06,
      "loss": 0.0903,
      "num_input_tokens_seen": 115730768,
      "step": 53625
    },
    {
      "epoch": 8.748776508972268,
      "grad_norm": 0.11400056630373001,
      "learning_rate": 2.347421899782551e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 115741840,
      "step": 53630
    },
    {
      "epoch": 8.749592169657422,
      "grad_norm": 0.02843559719622135,
      "learning_rate": 2.3444115065780953e-06,
      "loss": 0.0128,
      "num_input_tokens_seen": 115752208,
      "step": 53635
    },
    {
      "epoch": 8.750407830342578,
      "grad_norm": 0.24501417577266693,
      "learning_rate": 2.341402949950594e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 115762992,
      "step": 53640
    },
    {
      "epoch": 8.751223491027732,
      "grad_norm": 11.36231803894043,
      "learning_rate": 2.338396230143941e-06,
      "loss": 0.2064,
      "num_input_tokens_seen": 115774224,
      "step": 53645
    },
    {
      "epoch": 8.752039151712887,
      "grad_norm": 2.922313690185547,
      "learning_rate": 2.335391347401872e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 115785776,
      "step": 53650
    },
    {
      "epoch": 8.752854812398043,
      "grad_norm": 6.933698654174805,
      "learning_rate": 2.3323883019679805e-06,
      "loss": 0.2152,
      "num_input_tokens_seen": 115796656,
      "step": 53655
    },
    {
      "epoch": 8.753670473083197,
      "grad_norm": 6.864072799682617,
      "learning_rate": 2.3293870940857084e-06,
      "loss": 0.093,
      "num_input_tokens_seen": 115808272,
      "step": 53660
    },
    {
      "epoch": 8.754486133768353,
      "grad_norm": 0.2145785391330719,
      "learning_rate": 2.326387723998347e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 115820144,
      "step": 53665
    },
    {
      "epoch": 8.755301794453507,
      "grad_norm": 0.023491421714425087,
      "learning_rate": 2.3233901919490404e-06,
      "loss": 0.3007,
      "num_input_tokens_seen": 115831888,
      "step": 53670
    },
    {
      "epoch": 8.756117455138662,
      "grad_norm": 0.1304338425397873,
      "learning_rate": 2.3203944981807835e-06,
      "loss": 0.0839,
      "num_input_tokens_seen": 115843024,
      "step": 53675
    },
    {
      "epoch": 8.756933115823816,
      "grad_norm": 3.298250675201416,
      "learning_rate": 2.3174006429364263e-06,
      "loss": 0.118,
      "num_input_tokens_seen": 115854288,
      "step": 53680
    },
    {
      "epoch": 8.757748776508972,
      "grad_norm": 0.06621116399765015,
      "learning_rate": 2.314408626458664e-06,
      "loss": 0.1373,
      "num_input_tokens_seen": 115865296,
      "step": 53685
    },
    {
      "epoch": 8.758564437194128,
      "grad_norm": 0.6509382724761963,
      "learning_rate": 2.311418448990041e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 115876304,
      "step": 53690
    },
    {
      "epoch": 8.759380097879282,
      "grad_norm": 22.45487403869629,
      "learning_rate": 2.3084301107729633e-06,
      "loss": 0.0732,
      "num_input_tokens_seen": 115887632,
      "step": 53695
    },
    {
      "epoch": 8.760195758564437,
      "grad_norm": 0.30147695541381836,
      "learning_rate": 2.3054436120496736e-06,
      "loss": 0.059,
      "num_input_tokens_seen": 115898640,
      "step": 53700
    },
    {
      "epoch": 8.761011419249591,
      "grad_norm": 0.5400973558425903,
      "learning_rate": 2.302458953062275e-06,
      "loss": 0.0063,
      "num_input_tokens_seen": 115909392,
      "step": 53705
    },
    {
      "epoch": 8.761827079934747,
      "grad_norm": 0.04615411162376404,
      "learning_rate": 2.2994761340527195e-06,
      "loss": 0.0599,
      "num_input_tokens_seen": 115920080,
      "step": 53710
    },
    {
      "epoch": 8.762642740619903,
      "grad_norm": 0.14032778143882751,
      "learning_rate": 2.2964951552628096e-06,
      "loss": 0.0025,
      "num_input_tokens_seen": 115930640,
      "step": 53715
    },
    {
      "epoch": 8.763458401305057,
      "grad_norm": 0.09734567999839783,
      "learning_rate": 2.293516016934202e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 115941392,
      "step": 53720
    },
    {
      "epoch": 8.764274061990212,
      "grad_norm": 1.4439170360565186,
      "learning_rate": 2.2905387193083965e-06,
      "loss": 0.088,
      "num_input_tokens_seen": 115952176,
      "step": 53725
    },
    {
      "epoch": 8.765089722675366,
      "grad_norm": 15.759501457214355,
      "learning_rate": 2.287563262626749e-06,
      "loss": 0.0239,
      "num_input_tokens_seen": 115962896,
      "step": 53730
    },
    {
      "epoch": 8.765905383360522,
      "grad_norm": 0.1682003140449524,
      "learning_rate": 2.2845896471304667e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 115972944,
      "step": 53735
    },
    {
      "epoch": 8.766721044045678,
      "grad_norm": 0.09591232240200043,
      "learning_rate": 2.2816178730606012e-06,
      "loss": 0.1179,
      "num_input_tokens_seen": 115983600,
      "step": 53740
    },
    {
      "epoch": 8.767536704730832,
      "grad_norm": 0.06589756160974503,
      "learning_rate": 2.2786479406580658e-06,
      "loss": 0.1541,
      "num_input_tokens_seen": 115994832,
      "step": 53745
    },
    {
      "epoch": 8.768352365415987,
      "grad_norm": 0.05247294902801514,
      "learning_rate": 2.2756798501636146e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 116004848,
      "step": 53750
    },
    {
      "epoch": 8.769168026101141,
      "grad_norm": 7.89129638671875,
      "learning_rate": 2.272713601817855e-06,
      "loss": 0.0208,
      "num_input_tokens_seen": 116016016,
      "step": 53755
    },
    {
      "epoch": 8.769983686786297,
      "grad_norm": 0.05102195218205452,
      "learning_rate": 2.269749195861251e-06,
      "loss": 0.0758,
      "num_input_tokens_seen": 116027088,
      "step": 53760
    },
    {
      "epoch": 8.770799347471453,
      "grad_norm": 0.08562790602445602,
      "learning_rate": 2.266786632534107e-06,
      "loss": 0.1945,
      "num_input_tokens_seen": 116038192,
      "step": 53765
    },
    {
      "epoch": 8.771615008156607,
      "grad_norm": 0.23722116649150848,
      "learning_rate": 2.2638259120765864e-06,
      "loss": 0.1114,
      "num_input_tokens_seen": 116048272,
      "step": 53770
    },
    {
      "epoch": 8.772430668841762,
      "grad_norm": 0.18164660036563873,
      "learning_rate": 2.2608670347286947e-06,
      "loss": 0.0102,
      "num_input_tokens_seen": 116060112,
      "step": 53775
    },
    {
      "epoch": 8.773246329526916,
      "grad_norm": 0.03893053159117699,
      "learning_rate": 2.257910000730304e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 116071216,
      "step": 53780
    },
    {
      "epoch": 8.774061990212072,
      "grad_norm": 0.2040170133113861,
      "learning_rate": 2.2549548103211222e-06,
      "loss": 0.0421,
      "num_input_tokens_seen": 116081360,
      "step": 53785
    },
    {
      "epoch": 8.774877650897226,
      "grad_norm": 0.6495329737663269,
      "learning_rate": 2.2520014637407076e-06,
      "loss": 0.0762,
      "num_input_tokens_seen": 116091216,
      "step": 53790
    },
    {
      "epoch": 8.775693311582382,
      "grad_norm": 0.07339529693126678,
      "learning_rate": 2.2490499612284777e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 116101168,
      "step": 53795
    },
    {
      "epoch": 8.776508972267537,
      "grad_norm": 0.22962583601474762,
      "learning_rate": 2.246100303023696e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 116112272,
      "step": 53800
    },
    {
      "epoch": 8.777324632952691,
      "grad_norm": 3.569058895111084,
      "learning_rate": 2.2431524893654743e-06,
      "loss": 0.2096,
      "num_input_tokens_seen": 116123184,
      "step": 53805
    },
    {
      "epoch": 8.778140293637847,
      "grad_norm": 0.17319542169570923,
      "learning_rate": 2.2402065204927797e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 116133776,
      "step": 53810
    },
    {
      "epoch": 8.778955954323001,
      "grad_norm": 2.4589014053344727,
      "learning_rate": 2.237262396644421e-06,
      "loss": 0.1882,
      "num_input_tokens_seen": 116144688,
      "step": 53815
    },
    {
      "epoch": 8.779771615008157,
      "grad_norm": 0.08419797569513321,
      "learning_rate": 2.2343201180590745e-06,
      "loss": 0.0986,
      "num_input_tokens_seen": 116154288,
      "step": 53820
    },
    {
      "epoch": 8.780587275693312,
      "grad_norm": 0.09047245234251022,
      "learning_rate": 2.2313796849752516e-06,
      "loss": 0.057,
      "num_input_tokens_seen": 116165584,
      "step": 53825
    },
    {
      "epoch": 8.781402936378466,
      "grad_norm": 0.1527445912361145,
      "learning_rate": 2.2284410976313174e-06,
      "loss": 0.0334,
      "num_input_tokens_seen": 116176848,
      "step": 53830
    },
    {
      "epoch": 8.782218597063622,
      "grad_norm": 0.05455156788229942,
      "learning_rate": 2.2255043562654926e-06,
      "loss": 0.0829,
      "num_input_tokens_seen": 116186800,
      "step": 53835
    },
    {
      "epoch": 8.783034257748776,
      "grad_norm": 0.07151231914758682,
      "learning_rate": 2.2225694611158366e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 116196656,
      "step": 53840
    },
    {
      "epoch": 8.783849918433932,
      "grad_norm": 0.3095216453075409,
      "learning_rate": 2.2196364124202756e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 116207760,
      "step": 53845
    },
    {
      "epoch": 8.784665579119086,
      "grad_norm": 2.7524521350860596,
      "learning_rate": 2.2167052104165724e-06,
      "loss": 0.092,
      "num_input_tokens_seen": 116216752,
      "step": 53850
    },
    {
      "epoch": 8.785481239804241,
      "grad_norm": 6.77707576751709,
      "learning_rate": 2.2137758553423454e-06,
      "loss": 0.112,
      "num_input_tokens_seen": 116227568,
      "step": 53855
    },
    {
      "epoch": 8.786296900489397,
      "grad_norm": 0.18915368616580963,
      "learning_rate": 2.2108483474350626e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 116238736,
      "step": 53860
    },
    {
      "epoch": 8.78711256117455,
      "grad_norm": 3.433610200881958,
      "learning_rate": 2.207922686932046e-06,
      "loss": 0.232,
      "num_input_tokens_seen": 116249584,
      "step": 53865
    },
    {
      "epoch": 8.787928221859707,
      "grad_norm": 0.03905387595295906,
      "learning_rate": 2.2049988740704604e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 116259664,
      "step": 53870
    },
    {
      "epoch": 8.78874388254486,
      "grad_norm": 4.9666547775268555,
      "learning_rate": 2.202076909087328e-06,
      "loss": 0.1455,
      "num_input_tokens_seen": 116270352,
      "step": 53875
    },
    {
      "epoch": 8.789559543230016,
      "grad_norm": 0.8719344139099121,
      "learning_rate": 2.199156792219517e-06,
      "loss": 0.0247,
      "num_input_tokens_seen": 116281232,
      "step": 53880
    },
    {
      "epoch": 8.790375203915172,
      "grad_norm": 0.11355309933423996,
      "learning_rate": 2.1962385237037445e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 116291728,
      "step": 53885
    },
    {
      "epoch": 8.791190864600326,
      "grad_norm": 3.836967706680298,
      "learning_rate": 2.193322103776585e-06,
      "loss": 0.1835,
      "num_input_tokens_seen": 116302480,
      "step": 53890
    },
    {
      "epoch": 8.792006525285482,
      "grad_norm": 3.569146156311035,
      "learning_rate": 2.1904075326744543e-06,
      "loss": 0.1813,
      "num_input_tokens_seen": 116314192,
      "step": 53895
    },
    {
      "epoch": 8.792822185970635,
      "grad_norm": 0.04698646813631058,
      "learning_rate": 2.1874948106336254e-06,
      "loss": 0.1143,
      "num_input_tokens_seen": 116325616,
      "step": 53900
    },
    {
      "epoch": 8.793637846655791,
      "grad_norm": 0.033510949462652206,
      "learning_rate": 2.1845839378902167e-06,
      "loss": 0.0024,
      "num_input_tokens_seen": 116335344,
      "step": 53905
    },
    {
      "epoch": 8.794453507340947,
      "grad_norm": 0.059157855808734894,
      "learning_rate": 2.181674914680196e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 116346480,
      "step": 53910
    },
    {
      "epoch": 8.7952691680261,
      "grad_norm": 0.3026077449321747,
      "learning_rate": 2.178767741239382e-06,
      "loss": 0.2606,
      "num_input_tokens_seen": 116358576,
      "step": 53915
    },
    {
      "epoch": 8.796084828711257,
      "grad_norm": 0.14737290143966675,
      "learning_rate": 2.1758624178034537e-06,
      "loss": 0.1134,
      "num_input_tokens_seen": 116369872,
      "step": 53920
    },
    {
      "epoch": 8.79690048939641,
      "grad_norm": 0.30312269926071167,
      "learning_rate": 2.1729589446079252e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 116380432,
      "step": 53925
    },
    {
      "epoch": 8.797716150081566,
      "grad_norm": 0.05973745882511139,
      "learning_rate": 2.1700573218881694e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 116390544,
      "step": 53930
    },
    {
      "epoch": 8.798531810766722,
      "grad_norm": 0.0973372831940651,
      "learning_rate": 2.167157549879406e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 116402512,
      "step": 53935
    },
    {
      "epoch": 8.799347471451876,
      "grad_norm": 0.11348201334476471,
      "learning_rate": 2.1642596288166976e-06,
      "loss": 0.1278,
      "num_input_tokens_seen": 116413488,
      "step": 53940
    },
    {
      "epoch": 8.800163132137031,
      "grad_norm": 0.046488139778375626,
      "learning_rate": 2.1613635589349756e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 116424976,
      "step": 53945
    },
    {
      "epoch": 8.800978792822185,
      "grad_norm": 0.1353694200515747,
      "learning_rate": 2.1584693404690076e-06,
      "loss": 0.105,
      "num_input_tokens_seen": 116435856,
      "step": 53950
    },
    {
      "epoch": 8.801794453507341,
      "grad_norm": 0.15112315118312836,
      "learning_rate": 2.155576973653409e-06,
      "loss": 0.3163,
      "num_input_tokens_seen": 116447472,
      "step": 53955
    },
    {
      "epoch": 8.802610114192497,
      "grad_norm": 0.11449641734361649,
      "learning_rate": 2.1526864587226525e-06,
      "loss": 0.1872,
      "num_input_tokens_seen": 116458064,
      "step": 53960
    },
    {
      "epoch": 8.80342577487765,
      "grad_norm": 0.1149986982345581,
      "learning_rate": 2.14979779591106e-06,
      "loss": 0.0391,
      "num_input_tokens_seen": 116468464,
      "step": 53965
    },
    {
      "epoch": 8.804241435562806,
      "grad_norm": 0.020535923540592194,
      "learning_rate": 2.1469109854527993e-06,
      "loss": 0.1241,
      "num_input_tokens_seen": 116477488,
      "step": 53970
    },
    {
      "epoch": 8.80505709624796,
      "grad_norm": 0.09510662406682968,
      "learning_rate": 2.1440260275818856e-06,
      "loss": 0.102,
      "num_input_tokens_seen": 116488368,
      "step": 53975
    },
    {
      "epoch": 8.805872756933116,
      "grad_norm": 0.1702425479888916,
      "learning_rate": 2.1411429225321965e-06,
      "loss": 0.1228,
      "num_input_tokens_seen": 116498704,
      "step": 53980
    },
    {
      "epoch": 8.80668841761827,
      "grad_norm": 0.1370113044977188,
      "learning_rate": 2.138261670537445e-06,
      "loss": 0.0079,
      "num_input_tokens_seen": 116510000,
      "step": 53985
    },
    {
      "epoch": 8.807504078303426,
      "grad_norm": 0.06765083223581314,
      "learning_rate": 2.1353822718312016e-06,
      "loss": 0.1072,
      "num_input_tokens_seen": 116521552,
      "step": 53990
    },
    {
      "epoch": 8.808319738988581,
      "grad_norm": 0.1415599137544632,
      "learning_rate": 2.132504726646883e-06,
      "loss": 0.006,
      "num_input_tokens_seen": 116532208,
      "step": 53995
    },
    {
      "epoch": 8.809135399673735,
      "grad_norm": 0.2205033153295517,
      "learning_rate": 2.1296290352177644e-06,
      "loss": 0.0219,
      "num_input_tokens_seen": 116542608,
      "step": 54000
    },
    {
      "epoch": 8.809951060358891,
      "grad_norm": 0.10757742077112198,
      "learning_rate": 2.1267551977769555e-06,
      "loss": 0.2081,
      "num_input_tokens_seen": 116552656,
      "step": 54005
    },
    {
      "epoch": 8.810766721044045,
      "grad_norm": 0.1324695497751236,
      "learning_rate": 2.123883214557429e-06,
      "loss": 0.0659,
      "num_input_tokens_seen": 116565008,
      "step": 54010
    },
    {
      "epoch": 8.8115823817292,
      "grad_norm": 0.07865919172763824,
      "learning_rate": 2.1210130857920034e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 116574576,
      "step": 54015
    },
    {
      "epoch": 8.812398042414356,
      "grad_norm": 0.02997676655650139,
      "learning_rate": 2.1181448117133408e-06,
      "loss": 0.0074,
      "num_input_tokens_seen": 116584944,
      "step": 54020
    },
    {
      "epoch": 8.81321370309951,
      "grad_norm": 0.05799383670091629,
      "learning_rate": 2.115278392553963e-06,
      "loss": 0.0945,
      "num_input_tokens_seen": 116595920,
      "step": 54025
    },
    {
      "epoch": 8.814029363784666,
      "grad_norm": 0.10029251128435135,
      "learning_rate": 2.112413828546231e-06,
      "loss": 0.0311,
      "num_input_tokens_seen": 116606704,
      "step": 54030
    },
    {
      "epoch": 8.81484502446982,
      "grad_norm": 0.1537550836801529,
      "learning_rate": 2.1095511199223676e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 116618032,
      "step": 54035
    },
    {
      "epoch": 8.815660685154976,
      "grad_norm": 0.07866183668375015,
      "learning_rate": 2.106690266914435e-06,
      "loss": 0.1524,
      "num_input_tokens_seen": 116628464,
      "step": 54040
    },
    {
      "epoch": 8.81647634584013,
      "grad_norm": 0.18933890759944916,
      "learning_rate": 2.10383126975435e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 116639408,
      "step": 54045
    },
    {
      "epoch": 8.817292006525285,
      "grad_norm": 0.031053628772497177,
      "learning_rate": 2.1009741286738745e-06,
      "loss": 0.35,
      "num_input_tokens_seen": 116649808,
      "step": 54050
    },
    {
      "epoch": 8.818107667210441,
      "grad_norm": 0.07190708070993423,
      "learning_rate": 2.098118843904626e-06,
      "loss": 0.1437,
      "num_input_tokens_seen": 116661584,
      "step": 54055
    },
    {
      "epoch": 8.818923327895595,
      "grad_norm": 0.044518325477838516,
      "learning_rate": 2.0952654156780686e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 116671664,
      "step": 54060
    },
    {
      "epoch": 8.81973898858075,
      "grad_norm": 0.0714629516005516,
      "learning_rate": 2.09241384422551e-06,
      "loss": 0.1103,
      "num_input_tokens_seen": 116683248,
      "step": 54065
    },
    {
      "epoch": 8.820554649265905,
      "grad_norm": 2.834881067276001,
      "learning_rate": 2.089564129778121e-06,
      "loss": 0.0947,
      "num_input_tokens_seen": 116693936,
      "step": 54070
    },
    {
      "epoch": 8.82137030995106,
      "grad_norm": 0.06168261170387268,
      "learning_rate": 2.0867162725669077e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 116703856,
      "step": 54075
    },
    {
      "epoch": 8.822185970636216,
      "grad_norm": 0.03860916569828987,
      "learning_rate": 2.0838702728227356e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 116713136,
      "step": 54080
    },
    {
      "epoch": 8.82300163132137,
      "grad_norm": 0.07390040904283524,
      "learning_rate": 2.081026130776309e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 116723696,
      "step": 54085
    },
    {
      "epoch": 8.823817292006526,
      "grad_norm": 2.8101046085357666,
      "learning_rate": 2.078183846658199e-06,
      "loss": 0.0877,
      "num_input_tokens_seen": 116735280,
      "step": 54090
    },
    {
      "epoch": 8.82463295269168,
      "grad_norm": 4.103845119476318,
      "learning_rate": 2.075343420698811e-06,
      "loss": 0.1353,
      "num_input_tokens_seen": 116746416,
      "step": 54095
    },
    {
      "epoch": 8.825448613376835,
      "grad_norm": 0.0914025604724884,
      "learning_rate": 2.0725048531284015e-06,
      "loss": 0.1435,
      "num_input_tokens_seen": 116757232,
      "step": 54100
    },
    {
      "epoch": 8.826264274061991,
      "grad_norm": 0.024981753900647163,
      "learning_rate": 2.0696681441770836e-06,
      "loss": 0.0075,
      "num_input_tokens_seen": 116768304,
      "step": 54105
    },
    {
      "epoch": 8.827079934747145,
      "grad_norm": 9.234810829162598,
      "learning_rate": 2.066833294074813e-06,
      "loss": 0.0957,
      "num_input_tokens_seen": 116779216,
      "step": 54110
    },
    {
      "epoch": 8.8278955954323,
      "grad_norm": 0.2372504025697708,
      "learning_rate": 2.064000303051397e-06,
      "loss": 0.0026,
      "num_input_tokens_seen": 116790160,
      "step": 54115
    },
    {
      "epoch": 8.828711256117455,
      "grad_norm": 0.20221616327762604,
      "learning_rate": 2.061169171336494e-06,
      "loss": 0.0067,
      "num_input_tokens_seen": 116801936,
      "step": 54120
    },
    {
      "epoch": 8.82952691680261,
      "grad_norm": 0.20247258245944977,
      "learning_rate": 2.058339899159606e-06,
      "loss": 0.1894,
      "num_input_tokens_seen": 116812816,
      "step": 54125
    },
    {
      "epoch": 8.830342577487766,
      "grad_norm": 0.10631213337182999,
      "learning_rate": 2.0555124867500915e-06,
      "loss": 0.1848,
      "num_input_tokens_seen": 116821840,
      "step": 54130
    },
    {
      "epoch": 8.83115823817292,
      "grad_norm": 0.06082427129149437,
      "learning_rate": 2.05268693433715e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 116832912,
      "step": 54135
    },
    {
      "epoch": 8.831973898858076,
      "grad_norm": 0.13098251819610596,
      "learning_rate": 2.049863242149844e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 116842896,
      "step": 54140
    },
    {
      "epoch": 8.83278955954323,
      "grad_norm": 7.468482971191406,
      "learning_rate": 2.0470414104170694e-06,
      "loss": 0.0903,
      "num_input_tokens_seen": 116853744,
      "step": 54145
    },
    {
      "epoch": 8.833605220228385,
      "grad_norm": 0.12263431400060654,
      "learning_rate": 2.044221439367583e-06,
      "loss": 0.11,
      "num_input_tokens_seen": 116863824,
      "step": 54150
    },
    {
      "epoch": 8.83442088091354,
      "grad_norm": 0.03963104262948036,
      "learning_rate": 2.0414033292299823e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 116875056,
      "step": 54155
    },
    {
      "epoch": 8.835236541598695,
      "grad_norm": 0.06285598129034042,
      "learning_rate": 2.0385870802327176e-06,
      "loss": 0.1412,
      "num_input_tokens_seen": 116886768,
      "step": 54160
    },
    {
      "epoch": 8.83605220228385,
      "grad_norm": 3.2838828563690186,
      "learning_rate": 2.0357726926040875e-06,
      "loss": 0.0909,
      "num_input_tokens_seen": 116898032,
      "step": 54165
    },
    {
      "epoch": 8.836867862969005,
      "grad_norm": 0.1277763992547989,
      "learning_rate": 2.0329601665722453e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 116909008,
      "step": 54170
    },
    {
      "epoch": 8.83768352365416,
      "grad_norm": 0.15885736048221588,
      "learning_rate": 2.030149502365186e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 116920464,
      "step": 54175
    },
    {
      "epoch": 8.838499184339314,
      "grad_norm": 0.05947447568178177,
      "learning_rate": 2.027340700210753e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 116931568,
      "step": 54180
    },
    {
      "epoch": 8.83931484502447,
      "grad_norm": 3.056044578552246,
      "learning_rate": 2.0245337603366472e-06,
      "loss": 0.0993,
      "num_input_tokens_seen": 116943312,
      "step": 54185
    },
    {
      "epoch": 8.840130505709626,
      "grad_norm": 0.056010182946920395,
      "learning_rate": 2.0217286829704115e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 116954480,
      "step": 54190
    },
    {
      "epoch": 8.84094616639478,
      "grad_norm": 3.2493302822113037,
      "learning_rate": 2.018925468339436e-06,
      "loss": 0.0349,
      "num_input_tokens_seen": 116964880,
      "step": 54195
    },
    {
      "epoch": 8.841761827079935,
      "grad_norm": 0.08195551484823227,
      "learning_rate": 2.01612411667097e-06,
      "loss": 0.1324,
      "num_input_tokens_seen": 116974896,
      "step": 54200
    },
    {
      "epoch": 8.84257748776509,
      "grad_norm": 0.3906518816947937,
      "learning_rate": 2.013324628192101e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 116985072,
      "step": 54205
    },
    {
      "epoch": 8.843393148450245,
      "grad_norm": 0.08634142577648163,
      "learning_rate": 2.0105270031297725e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 116995856,
      "step": 54210
    },
    {
      "epoch": 8.844208809135399,
      "grad_norm": 3.831986904144287,
      "learning_rate": 2.0077312417107695e-06,
      "loss": 0.3087,
      "num_input_tokens_seen": 117007152,
      "step": 54215
    },
    {
      "epoch": 8.845024469820554,
      "grad_norm": 2.8840560913085938,
      "learning_rate": 2.0049373441617363e-06,
      "loss": 0.1635,
      "num_input_tokens_seen": 117017040,
      "step": 54220
    },
    {
      "epoch": 8.84584013050571,
      "grad_norm": 0.3361341059207916,
      "learning_rate": 2.002145310709155e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 117028304,
      "step": 54225
    },
    {
      "epoch": 8.846655791190864,
      "grad_norm": 0.07720833271741867,
      "learning_rate": 1.9993551415793647e-06,
      "loss": 0.1723,
      "num_input_tokens_seen": 117039120,
      "step": 54230
    },
    {
      "epoch": 8.84747145187602,
      "grad_norm": 0.06056373566389084,
      "learning_rate": 1.9965668369985507e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 117050032,
      "step": 54235
    },
    {
      "epoch": 8.848287112561174,
      "grad_norm": 0.3385069966316223,
      "learning_rate": 1.993780397192749e-06,
      "loss": 0.1919,
      "num_input_tokens_seen": 117058448,
      "step": 54240
    },
    {
      "epoch": 8.84910277324633,
      "grad_norm": 0.12695370614528656,
      "learning_rate": 1.9909958223878424e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 117068880,
      "step": 54245
    },
    {
      "epoch": 8.849918433931485,
      "grad_norm": 4.353246688842773,
      "learning_rate": 1.988213112809559e-06,
      "loss": 0.2479,
      "num_input_tokens_seen": 117079312,
      "step": 54250
    },
    {
      "epoch": 8.850734094616639,
      "grad_norm": 14.258933067321777,
      "learning_rate": 1.9854322686834766e-06,
      "loss": 0.2642,
      "num_input_tokens_seen": 117089776,
      "step": 54255
    },
    {
      "epoch": 8.851549755301795,
      "grad_norm": 0.03966104984283447,
      "learning_rate": 1.982653290235034e-06,
      "loss": 0.263,
      "num_input_tokens_seen": 117100304,
      "step": 54260
    },
    {
      "epoch": 8.852365415986949,
      "grad_norm": 0.0750674456357956,
      "learning_rate": 1.979876177689505e-06,
      "loss": 0.1308,
      "num_input_tokens_seen": 117110064,
      "step": 54265
    },
    {
      "epoch": 8.853181076672104,
      "grad_norm": 28.39978790283203,
      "learning_rate": 1.9771009312720164e-06,
      "loss": 0.1608,
      "num_input_tokens_seen": 117121872,
      "step": 54270
    },
    {
      "epoch": 8.85399673735726,
      "grad_norm": 0.1714479923248291,
      "learning_rate": 1.9743275512075417e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 117133040,
      "step": 54275
    },
    {
      "epoch": 8.854812398042414,
      "grad_norm": 2.3563148975372314,
      "learning_rate": 1.9715560377209093e-06,
      "loss": 0.1389,
      "num_input_tokens_seen": 117145488,
      "step": 54280
    },
    {
      "epoch": 8.85562805872757,
      "grad_norm": 0.08297800272703171,
      "learning_rate": 1.9687863910367866e-06,
      "loss": 0.0536,
      "num_input_tokens_seen": 117156464,
      "step": 54285
    },
    {
      "epoch": 8.856443719412724,
      "grad_norm": 0.19360771775245667,
      "learning_rate": 1.9660186113796996e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 117166128,
      "step": 54290
    },
    {
      "epoch": 8.85725938009788,
      "grad_norm": 0.3158252239227295,
      "learning_rate": 1.963252698974019e-06,
      "loss": 0.0311,
      "num_input_tokens_seen": 117176272,
      "step": 54295
    },
    {
      "epoch": 8.858075040783035,
      "grad_norm": 0.14304521679878235,
      "learning_rate": 1.96048865404396e-06,
      "loss": 0.2595,
      "num_input_tokens_seen": 117187440,
      "step": 54300
    },
    {
      "epoch": 8.858890701468189,
      "grad_norm": 0.05150622874498367,
      "learning_rate": 1.9577264768135927e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 117198416,
      "step": 54305
    },
    {
      "epoch": 8.859706362153345,
      "grad_norm": 0.029016217216849327,
      "learning_rate": 1.9549661675068303e-06,
      "loss": 0.1438,
      "num_input_tokens_seen": 117209104,
      "step": 54310
    },
    {
      "epoch": 8.860522022838499,
      "grad_norm": 0.06012250483036041,
      "learning_rate": 1.9522077263474436e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 117220592,
      "step": 54315
    },
    {
      "epoch": 8.861337683523654,
      "grad_norm": 0.0844341441988945,
      "learning_rate": 1.94945115355904e-06,
      "loss": 0.0067,
      "num_input_tokens_seen": 117231280,
      "step": 54320
    },
    {
      "epoch": 8.86215334420881,
      "grad_norm": 3.6612021923065186,
      "learning_rate": 1.946696449365082e-06,
      "loss": 0.0855,
      "num_input_tokens_seen": 117243216,
      "step": 54325
    },
    {
      "epoch": 8.862969004893964,
      "grad_norm": 0.191433846950531,
      "learning_rate": 1.943943613988883e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 117252976,
      "step": 54330
    },
    {
      "epoch": 8.86378466557912,
      "grad_norm": 0.10980988293886185,
      "learning_rate": 1.9411926476535976e-06,
      "loss": 0.0156,
      "num_input_tokens_seen": 117264016,
      "step": 54335
    },
    {
      "epoch": 8.864600326264274,
      "grad_norm": 3.5337917804718018,
      "learning_rate": 1.938443550582239e-06,
      "loss": 0.2485,
      "num_input_tokens_seen": 117274864,
      "step": 54340
    },
    {
      "epoch": 8.86541598694943,
      "grad_norm": 0.14542673528194427,
      "learning_rate": 1.935696322997657e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 117285552,
      "step": 54345
    },
    {
      "epoch": 8.866231647634583,
      "grad_norm": 6.901078224182129,
      "learning_rate": 1.932950965122554e-06,
      "loss": 0.0518,
      "num_input_tokens_seen": 117295728,
      "step": 54350
    },
    {
      "epoch": 8.867047308319739,
      "grad_norm": 0.24575941264629364,
      "learning_rate": 1.930207477179491e-06,
      "loss": 0.1869,
      "num_input_tokens_seen": 117307408,
      "step": 54355
    },
    {
      "epoch": 8.867862969004895,
      "grad_norm": 4.659783840179443,
      "learning_rate": 1.9274658593908647e-06,
      "loss": 0.1681,
      "num_input_tokens_seen": 117318832,
      "step": 54360
    },
    {
      "epoch": 8.868678629690049,
      "grad_norm": 1.5721524953842163,
      "learning_rate": 1.9247261119789252e-06,
      "loss": 0.1441,
      "num_input_tokens_seen": 117329712,
      "step": 54365
    },
    {
      "epoch": 8.869494290375204,
      "grad_norm": 4.646288871765137,
      "learning_rate": 1.9219882351657696e-06,
      "loss": 0.1658,
      "num_input_tokens_seen": 117339728,
      "step": 54370
    },
    {
      "epoch": 8.870309951060358,
      "grad_norm": 0.1199723333120346,
      "learning_rate": 1.9192522291733434e-06,
      "loss": 0.0715,
      "num_input_tokens_seen": 117351408,
      "step": 54375
    },
    {
      "epoch": 8.871125611745514,
      "grad_norm": 6.661080837249756,
      "learning_rate": 1.9165180942234435e-06,
      "loss": 0.4362,
      "num_input_tokens_seen": 117361904,
      "step": 54380
    },
    {
      "epoch": 8.87194127243067,
      "grad_norm": 0.06170143187046051,
      "learning_rate": 1.913785830537712e-06,
      "loss": 0.0782,
      "num_input_tokens_seen": 117371536,
      "step": 54385
    },
    {
      "epoch": 8.872756933115824,
      "grad_norm": 0.0556231290102005,
      "learning_rate": 1.911055438337639e-06,
      "loss": 0.1225,
      "num_input_tokens_seen": 117379984,
      "step": 54390
    },
    {
      "epoch": 8.87357259380098,
      "grad_norm": 0.03875993564724922,
      "learning_rate": 1.9083269178445636e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 117390416,
      "step": 54395
    },
    {
      "epoch": 8.874388254486133,
      "grad_norm": 4.18616247177124,
      "learning_rate": 1.9056002692796698e-06,
      "loss": 0.0878,
      "num_input_tokens_seen": 117401136,
      "step": 54400
    },
    {
      "epoch": 8.875203915171289,
      "grad_norm": 0.110075943171978,
      "learning_rate": 1.9028754928640008e-06,
      "loss": 0.1132,
      "num_input_tokens_seen": 117411344,
      "step": 54405
    },
    {
      "epoch": 8.876019575856443,
      "grad_norm": 0.05919650197029114,
      "learning_rate": 1.9001525888184407e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 117421680,
      "step": 54410
    },
    {
      "epoch": 8.876835236541599,
      "grad_norm": 0.12138954550027847,
      "learning_rate": 1.8974315573637185e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 117432048,
      "step": 54415
    },
    {
      "epoch": 8.877650897226754,
      "grad_norm": 0.07046827673912048,
      "learning_rate": 1.8947123987204135e-06,
      "loss": 0.0811,
      "num_input_tokens_seen": 117444080,
      "step": 54420
    },
    {
      "epoch": 8.878466557911908,
      "grad_norm": 0.16570985317230225,
      "learning_rate": 1.8919951131089602e-06,
      "loss": 0.1182,
      "num_input_tokens_seen": 117454864,
      "step": 54425
    },
    {
      "epoch": 8.879282218597064,
      "grad_norm": 0.04637861251831055,
      "learning_rate": 1.88927970074963e-06,
      "loss": 0.1158,
      "num_input_tokens_seen": 117464944,
      "step": 54430
    },
    {
      "epoch": 8.880097879282218,
      "grad_norm": 0.07738249003887177,
      "learning_rate": 1.8865661618625491e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 117475344,
      "step": 54435
    },
    {
      "epoch": 8.880913539967374,
      "grad_norm": 6.084794044494629,
      "learning_rate": 1.883854496667692e-06,
      "loss": 0.2447,
      "num_input_tokens_seen": 117486288,
      "step": 54440
    },
    {
      "epoch": 8.88172920065253,
      "grad_norm": 3.370311975479126,
      "learning_rate": 1.8811447053848796e-06,
      "loss": 0.0346,
      "num_input_tokens_seen": 117496432,
      "step": 54445
    },
    {
      "epoch": 8.882544861337683,
      "grad_norm": 0.1351543515920639,
      "learning_rate": 1.878436788233781e-06,
      "loss": 0.0549,
      "num_input_tokens_seen": 117506960,
      "step": 54450
    },
    {
      "epoch": 8.883360522022839,
      "grad_norm": 0.1686161607503891,
      "learning_rate": 1.8757307454339095e-06,
      "loss": 0.209,
      "num_input_tokens_seen": 117517360,
      "step": 54455
    },
    {
      "epoch": 8.884176182707993,
      "grad_norm": 0.39949020743370056,
      "learning_rate": 1.8730265772046396e-06,
      "loss": 0.145,
      "num_input_tokens_seen": 117527376,
      "step": 54460
    },
    {
      "epoch": 8.884991843393149,
      "grad_norm": 0.07697786390781403,
      "learning_rate": 1.8703242837651818e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 117537104,
      "step": 54465
    },
    {
      "epoch": 8.885807504078304,
      "grad_norm": 0.2855949401855469,
      "learning_rate": 1.8676238653345945e-06,
      "loss": 0.1307,
      "num_input_tokens_seen": 117546896,
      "step": 54470
    },
    {
      "epoch": 8.886623164763458,
      "grad_norm": 0.10021426528692245,
      "learning_rate": 1.8649253221317914e-06,
      "loss": 0.0806,
      "num_input_tokens_seen": 117557328,
      "step": 54475
    },
    {
      "epoch": 8.887438825448614,
      "grad_norm": 0.127748504281044,
      "learning_rate": 1.8622286543755252e-06,
      "loss": 0.0226,
      "num_input_tokens_seen": 117567568,
      "step": 54480
    },
    {
      "epoch": 8.888254486133768,
      "grad_norm": 0.22251491248607635,
      "learning_rate": 1.8595338622844072e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 117578032,
      "step": 54485
    },
    {
      "epoch": 8.889070146818923,
      "grad_norm": 0.10934317857027054,
      "learning_rate": 1.8568409460768848e-06,
      "loss": 0.0878,
      "num_input_tokens_seen": 117589456,
      "step": 54490
    },
    {
      "epoch": 8.88988580750408,
      "grad_norm": 4.995783805847168,
      "learning_rate": 1.8541499059712641e-06,
      "loss": 0.0089,
      "num_input_tokens_seen": 117599952,
      "step": 54495
    },
    {
      "epoch": 8.890701468189233,
      "grad_norm": 0.062409330159425735,
      "learning_rate": 1.8514607421856928e-06,
      "loss": 0.1433,
      "num_input_tokens_seen": 117610288,
      "step": 54500
    },
    {
      "epoch": 8.891517128874389,
      "grad_norm": 0.21258576214313507,
      "learning_rate": 1.8487734549381686e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 117620528,
      "step": 54505
    },
    {
      "epoch": 8.892332789559543,
      "grad_norm": 0.07005354762077332,
      "learning_rate": 1.8460880444465367e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 117631984,
      "step": 54510
    },
    {
      "epoch": 8.893148450244698,
      "grad_norm": 0.034182339906692505,
      "learning_rate": 1.843404510928487e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 117642800,
      "step": 54515
    },
    {
      "epoch": 8.893964110929852,
      "grad_norm": 0.1602908968925476,
      "learning_rate": 1.8407228546015648e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 117654544,
      "step": 54520
    },
    {
      "epoch": 8.894779771615008,
      "grad_norm": 0.08445174992084503,
      "learning_rate": 1.8380430756831574e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 117664688,
      "step": 54525
    },
    {
      "epoch": 8.895595432300164,
      "grad_norm": 0.027813170105218887,
      "learning_rate": 1.835365174390502e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 117675664,
      "step": 54530
    },
    {
      "epoch": 8.896411092985318,
      "grad_norm": 0.15260443091392517,
      "learning_rate": 1.8326891509406808e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 117685744,
      "step": 54535
    },
    {
      "epoch": 8.897226753670473,
      "grad_norm": 0.1360047608613968,
      "learning_rate": 1.8300150055506254e-06,
      "loss": 0.0072,
      "num_input_tokens_seen": 117697104,
      "step": 54540
    },
    {
      "epoch": 8.898042414355627,
      "grad_norm": 3.5753955841064453,
      "learning_rate": 1.8273427384371127e-06,
      "loss": 0.0972,
      "num_input_tokens_seen": 117708272,
      "step": 54545
    },
    {
      "epoch": 8.898858075040783,
      "grad_norm": 2.59792423248291,
      "learning_rate": 1.8246723498167806e-06,
      "loss": 0.0975,
      "num_input_tokens_seen": 117718832,
      "step": 54550
    },
    {
      "epoch": 8.899673735725939,
      "grad_norm": 0.17426784336566925,
      "learning_rate": 1.8220038399060973e-06,
      "loss": 0.1208,
      "num_input_tokens_seen": 117729744,
      "step": 54555
    },
    {
      "epoch": 8.900489396411093,
      "grad_norm": 11.262162208557129,
      "learning_rate": 1.8193372089213872e-06,
      "loss": 0.1129,
      "num_input_tokens_seen": 117741104,
      "step": 54560
    },
    {
      "epoch": 8.901305057096248,
      "grad_norm": 0.047902703285217285,
      "learning_rate": 1.816672457078819e-06,
      "loss": 0.0027,
      "num_input_tokens_seen": 117751280,
      "step": 54565
    },
    {
      "epoch": 8.902120717781402,
      "grad_norm": 0.10405686497688293,
      "learning_rate": 1.8140095845944117e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 117761200,
      "step": 54570
    },
    {
      "epoch": 8.902936378466558,
      "grad_norm": 0.1675781011581421,
      "learning_rate": 1.811348591684031e-06,
      "loss": 0.098,
      "num_input_tokens_seen": 117771536,
      "step": 54575
    },
    {
      "epoch": 8.903752039151712,
      "grad_norm": 0.1404532641172409,
      "learning_rate": 1.8086894785633969e-06,
      "loss": 0.0085,
      "num_input_tokens_seen": 117781808,
      "step": 54580
    },
    {
      "epoch": 8.904567699836868,
      "grad_norm": 0.041269417852163315,
      "learning_rate": 1.806032245448061e-06,
      "loss": 0.1004,
      "num_input_tokens_seen": 117793296,
      "step": 54585
    },
    {
      "epoch": 8.905383360522023,
      "grad_norm": 0.03334139287471771,
      "learning_rate": 1.8033768925534378e-06,
      "loss": 0.2275,
      "num_input_tokens_seen": 117803568,
      "step": 54590
    },
    {
      "epoch": 8.906199021207177,
      "grad_norm": 0.09672871977090836,
      "learning_rate": 1.8007234200947826e-06,
      "loss": 0.1079,
      "num_input_tokens_seen": 117813648,
      "step": 54595
    },
    {
      "epoch": 8.907014681892333,
      "grad_norm": 0.1703348308801651,
      "learning_rate": 1.7980718282871982e-06,
      "loss": 0.0073,
      "num_input_tokens_seen": 117824816,
      "step": 54600
    },
    {
      "epoch": 8.907830342577487,
      "grad_norm": 0.07169262319803238,
      "learning_rate": 1.7954221173456382e-06,
      "loss": 0.2201,
      "num_input_tokens_seen": 117835312,
      "step": 54605
    },
    {
      "epoch": 8.908646003262643,
      "grad_norm": 0.05825048312544823,
      "learning_rate": 1.7927742874848997e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 117845616,
      "step": 54610
    },
    {
      "epoch": 8.909461663947798,
      "grad_norm": 0.1675950139760971,
      "learning_rate": 1.7901283389196278e-06,
      "loss": 0.0888,
      "num_input_tokens_seen": 117856656,
      "step": 54615
    },
    {
      "epoch": 8.910277324632952,
      "grad_norm": 6.598090648651123,
      "learning_rate": 1.7874842718643203e-06,
      "loss": 0.4248,
      "num_input_tokens_seen": 117867056,
      "step": 54620
    },
    {
      "epoch": 8.911092985318108,
      "grad_norm": 4.078062534332275,
      "learning_rate": 1.7848420865333172e-06,
      "loss": 0.3353,
      "num_input_tokens_seen": 117877552,
      "step": 54625
    },
    {
      "epoch": 8.911908646003262,
      "grad_norm": 0.1207270622253418,
      "learning_rate": 1.782201783140805e-06,
      "loss": 0.087,
      "num_input_tokens_seen": 117888624,
      "step": 54630
    },
    {
      "epoch": 8.912724306688418,
      "grad_norm": 4.053979873657227,
      "learning_rate": 1.779563361900824e-06,
      "loss": 0.1236,
      "num_input_tokens_seen": 117897520,
      "step": 54635
    },
    {
      "epoch": 8.913539967373573,
      "grad_norm": 0.5061556696891785,
      "learning_rate": 1.7769268230272557e-06,
      "loss": 0.007,
      "num_input_tokens_seen": 117907920,
      "step": 54640
    },
    {
      "epoch": 8.914355628058727,
      "grad_norm": 0.7372855544090271,
      "learning_rate": 1.774292166733832e-06,
      "loss": 0.0862,
      "num_input_tokens_seen": 117919088,
      "step": 54645
    },
    {
      "epoch": 8.915171288743883,
      "grad_norm": 4.361306667327881,
      "learning_rate": 1.7716593932341319e-06,
      "loss": 0.1022,
      "num_input_tokens_seen": 117928848,
      "step": 54650
    },
    {
      "epoch": 8.915986949429037,
      "grad_norm": 0.05198827013373375,
      "learning_rate": 1.7690285027415792e-06,
      "loss": 0.0897,
      "num_input_tokens_seen": 117939984,
      "step": 54655
    },
    {
      "epoch": 8.916802610114193,
      "grad_norm": 0.040379416197538376,
      "learning_rate": 1.7663994954694508e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 117952784,
      "step": 54660
    },
    {
      "epoch": 8.917618270799348,
      "grad_norm": 0.07739093899726868,
      "learning_rate": 1.7637723716308646e-06,
      "loss": 0.2355,
      "num_input_tokens_seen": 117963504,
      "step": 54665
    },
    {
      "epoch": 8.918433931484502,
      "grad_norm": 0.051868923008441925,
      "learning_rate": 1.7611471314387867e-06,
      "loss": 0.0803,
      "num_input_tokens_seen": 117973808,
      "step": 54670
    },
    {
      "epoch": 8.919249592169658,
      "grad_norm": 0.055537253618240356,
      "learning_rate": 1.7585237751060357e-06,
      "loss": 0.0777,
      "num_input_tokens_seen": 117984464,
      "step": 54675
    },
    {
      "epoch": 8.920065252854812,
      "grad_norm": 0.21607373654842377,
      "learning_rate": 1.7559023028452748e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 117994928,
      "step": 54680
    },
    {
      "epoch": 8.920880913539968,
      "grad_norm": 0.0954468846321106,
      "learning_rate": 1.7532827148690145e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 118004304,
      "step": 54685
    },
    {
      "epoch": 8.921696574225122,
      "grad_norm": 0.3734135329723358,
      "learning_rate": 1.75066501138961e-06,
      "loss": 0.0083,
      "num_input_tokens_seen": 118015152,
      "step": 54690
    },
    {
      "epoch": 8.922512234910277,
      "grad_norm": 0.23452778160572052,
      "learning_rate": 1.7480491926192638e-06,
      "loss": 0.0132,
      "num_input_tokens_seen": 118027344,
      "step": 54695
    },
    {
      "epoch": 8.923327895595433,
      "grad_norm": 3.485624313354492,
      "learning_rate": 1.7454352587700284e-06,
      "loss": 0.1744,
      "num_input_tokens_seen": 118038672,
      "step": 54700
    },
    {
      "epoch": 8.924143556280587,
      "grad_norm": 4.423807621002197,
      "learning_rate": 1.7428232100538067e-06,
      "loss": 0.0318,
      "num_input_tokens_seen": 118050576,
      "step": 54705
    },
    {
      "epoch": 8.924959216965743,
      "grad_norm": 0.05518285557627678,
      "learning_rate": 1.7402130466823373e-06,
      "loss": 0.0565,
      "num_input_tokens_seen": 118061520,
      "step": 54710
    },
    {
      "epoch": 8.925774877650896,
      "grad_norm": 0.1752781867980957,
      "learning_rate": 1.7376047688672182e-06,
      "loss": 0.096,
      "num_input_tokens_seen": 118072080,
      "step": 54715
    },
    {
      "epoch": 8.926590538336052,
      "grad_norm": 0.08497949689626694,
      "learning_rate": 1.734998376819888e-06,
      "loss": 0.0049,
      "num_input_tokens_seen": 118082416,
      "step": 54720
    },
    {
      "epoch": 8.927406199021208,
      "grad_norm": 0.1496218591928482,
      "learning_rate": 1.732393870751639e-06,
      "loss": 0.1595,
      "num_input_tokens_seen": 118092880,
      "step": 54725
    },
    {
      "epoch": 8.928221859706362,
      "grad_norm": 0.9150623679161072,
      "learning_rate": 1.729791250873597e-06,
      "loss": 0.0094,
      "num_input_tokens_seen": 118105136,
      "step": 54730
    },
    {
      "epoch": 8.929037520391518,
      "grad_norm": 0.03522399812936783,
      "learning_rate": 1.7271905173967513e-06,
      "loss": 0.0405,
      "num_input_tokens_seen": 118114736,
      "step": 54735
    },
    {
      "epoch": 8.929853181076671,
      "grad_norm": 0.5155400633811951,
      "learning_rate": 1.7245916705319276e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 118126064,
      "step": 54740
    },
    {
      "epoch": 8.930668841761827,
      "grad_norm": 0.08923421800136566,
      "learning_rate": 1.7219947104897994e-06,
      "loss": 0.0978,
      "num_input_tokens_seen": 118137456,
      "step": 54745
    },
    {
      "epoch": 8.931484502446983,
      "grad_norm": 0.12209942936897278,
      "learning_rate": 1.7193996374808924e-06,
      "loss": 0.1175,
      "num_input_tokens_seen": 118147376,
      "step": 54750
    },
    {
      "epoch": 8.932300163132137,
      "grad_norm": 0.051897816359996796,
      "learning_rate": 1.7168064517155747e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 118158096,
      "step": 54755
    },
    {
      "epoch": 8.933115823817293,
      "grad_norm": 0.07628517597913742,
      "learning_rate": 1.714215153404064e-06,
      "loss": 0.0791,
      "num_input_tokens_seen": 118168688,
      "step": 54760
    },
    {
      "epoch": 8.933931484502446,
      "grad_norm": 0.0787888765335083,
      "learning_rate": 1.7116257427564259e-06,
      "loss": 0.2482,
      "num_input_tokens_seen": 118179792,
      "step": 54765
    },
    {
      "epoch": 8.934747145187602,
      "grad_norm": 0.2817718982696533,
      "learning_rate": 1.7090382199825672e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 118190736,
      "step": 54770
    },
    {
      "epoch": 8.935562805872756,
      "grad_norm": 0.03979673609137535,
      "learning_rate": 1.7064525852922424e-06,
      "loss": 0.0974,
      "num_input_tokens_seen": 118201584,
      "step": 54775
    },
    {
      "epoch": 8.936378466557912,
      "grad_norm": 0.12454798817634583,
      "learning_rate": 1.7038688388950675e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 118212624,
      "step": 54780
    },
    {
      "epoch": 8.937194127243067,
      "grad_norm": 0.12224704772233963,
      "learning_rate": 1.7012869810004856e-06,
      "loss": 0.1946,
      "num_input_tokens_seen": 118222800,
      "step": 54785
    },
    {
      "epoch": 8.938009787928221,
      "grad_norm": 0.11173922568559647,
      "learning_rate": 1.698707011817799e-06,
      "loss": 0.0836,
      "num_input_tokens_seen": 118232592,
      "step": 54790
    },
    {
      "epoch": 8.938825448613377,
      "grad_norm": 0.2910785377025604,
      "learning_rate": 1.6961289315561512e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 118244464,
      "step": 54795
    },
    {
      "epoch": 8.939641109298531,
      "grad_norm": 0.5948049426078796,
      "learning_rate": 1.6935527404245366e-06,
      "loss": 0.0901,
      "num_input_tokens_seen": 118253552,
      "step": 54800
    },
    {
      "epoch": 8.940456769983687,
      "grad_norm": 0.06635866314172745,
      "learning_rate": 1.6909784386317906e-06,
      "loss": 0.1438,
      "num_input_tokens_seen": 118263824,
      "step": 54805
    },
    {
      "epoch": 8.941272430668842,
      "grad_norm": 4.436750411987305,
      "learning_rate": 1.688406026386602e-06,
      "loss": 0.1401,
      "num_input_tokens_seen": 118275216,
      "step": 54810
    },
    {
      "epoch": 8.942088091353996,
      "grad_norm": 0.1608649641275406,
      "learning_rate": 1.6858355038975038e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 118286224,
      "step": 54815
    },
    {
      "epoch": 8.942903752039152,
      "grad_norm": 0.09187374264001846,
      "learning_rate": 1.6832668713728711e-06,
      "loss": 0.1731,
      "num_input_tokens_seen": 118296240,
      "step": 54820
    },
    {
      "epoch": 8.943719412724306,
      "grad_norm": 0.043479304760694504,
      "learning_rate": 1.6807001290209374e-06,
      "loss": 0.0103,
      "num_input_tokens_seen": 118305680,
      "step": 54825
    },
    {
      "epoch": 8.944535073409462,
      "grad_norm": 0.04699716717004776,
      "learning_rate": 1.6781352770497694e-06,
      "loss": 0.143,
      "num_input_tokens_seen": 118316880,
      "step": 54830
    },
    {
      "epoch": 8.945350734094617,
      "grad_norm": 0.14820754528045654,
      "learning_rate": 1.6755723156672925e-06,
      "loss": 0.0888,
      "num_input_tokens_seen": 118328304,
      "step": 54835
    },
    {
      "epoch": 8.946166394779771,
      "grad_norm": 0.0861411988735199,
      "learning_rate": 1.6730112450812685e-06,
      "loss": 0.0244,
      "num_input_tokens_seen": 118339728,
      "step": 54840
    },
    {
      "epoch": 8.946982055464927,
      "grad_norm": 0.14119026064872742,
      "learning_rate": 1.6704520654993145e-06,
      "loss": 0.0925,
      "num_input_tokens_seen": 118349872,
      "step": 54845
    },
    {
      "epoch": 8.947797716150081,
      "grad_norm": 0.07659178227186203,
      "learning_rate": 1.6678947771288866e-06,
      "loss": 0.2229,
      "num_input_tokens_seen": 118359888,
      "step": 54850
    },
    {
      "epoch": 8.948613376835237,
      "grad_norm": 0.03260839357972145,
      "learning_rate": 1.665339380177297e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 118370544,
      "step": 54855
    },
    {
      "epoch": 8.949429037520392,
      "grad_norm": 0.09648166596889496,
      "learning_rate": 1.6627858748516912e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 118381584,
      "step": 54860
    },
    {
      "epoch": 8.950244698205546,
      "grad_norm": 0.11531396955251694,
      "learning_rate": 1.6602342613590754e-06,
      "loss": 0.0707,
      "num_input_tokens_seen": 118392624,
      "step": 54865
    },
    {
      "epoch": 8.951060358890702,
      "grad_norm": 0.04172559082508087,
      "learning_rate": 1.6576845399062985e-06,
      "loss": 0.025,
      "num_input_tokens_seen": 118402768,
      "step": 54870
    },
    {
      "epoch": 8.951876019575856,
      "grad_norm": 16.108680725097656,
      "learning_rate": 1.6551367107000503e-06,
      "loss": 0.0799,
      "num_input_tokens_seen": 118413584,
      "step": 54875
    },
    {
      "epoch": 8.952691680261012,
      "grad_norm": 0.14586836099624634,
      "learning_rate": 1.6525907739468689e-06,
      "loss": 0.1229,
      "num_input_tokens_seen": 118424400,
      "step": 54880
    },
    {
      "epoch": 8.953507340946166,
      "grad_norm": 0.0429881252348423,
      "learning_rate": 1.6500467298531414e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 118435696,
      "step": 54885
    },
    {
      "epoch": 8.954323001631321,
      "grad_norm": 0.16560575366020203,
      "learning_rate": 1.6475045786251059e-06,
      "loss": 0.2786,
      "num_input_tokens_seen": 118446544,
      "step": 54890
    },
    {
      "epoch": 8.955138662316477,
      "grad_norm": 3.961792469024658,
      "learning_rate": 1.6449643204688364e-06,
      "loss": 0.1506,
      "num_input_tokens_seen": 118457616,
      "step": 54895
    },
    {
      "epoch": 8.955954323001631,
      "grad_norm": 0.10775754600763321,
      "learning_rate": 1.6424259555902627e-06,
      "loss": 0.1212,
      "num_input_tokens_seen": 118468816,
      "step": 54900
    },
    {
      "epoch": 8.956769983686787,
      "grad_norm": 0.05108390003442764,
      "learning_rate": 1.6398894841951561e-06,
      "loss": 0.1589,
      "num_input_tokens_seen": 118480176,
      "step": 54905
    },
    {
      "epoch": 8.95758564437194,
      "grad_norm": 0.1063050851225853,
      "learning_rate": 1.6373549064891358e-06,
      "loss": 0.0035,
      "num_input_tokens_seen": 118491024,
      "step": 54910
    },
    {
      "epoch": 8.958401305057096,
      "grad_norm": 0.10395338386297226,
      "learning_rate": 1.6348222226776705e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 118502096,
      "step": 54915
    },
    {
      "epoch": 8.959216965742252,
      "grad_norm": 0.2134932279586792,
      "learning_rate": 1.6322914329660655e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 118512464,
      "step": 54920
    },
    {
      "epoch": 8.960032626427406,
      "grad_norm": 0.059857435524463654,
      "learning_rate": 1.629762537559487e-06,
      "loss": 0.1359,
      "num_input_tokens_seen": 118523728,
      "step": 54925
    },
    {
      "epoch": 8.960848287112562,
      "grad_norm": 0.10444968938827515,
      "learning_rate": 1.6272355366629327e-06,
      "loss": 0.1108,
      "num_input_tokens_seen": 118533680,
      "step": 54930
    },
    {
      "epoch": 8.961663947797716,
      "grad_norm": 0.09019356220960617,
      "learning_rate": 1.6247104304812604e-06,
      "loss": 0.1299,
      "num_input_tokens_seen": 118544176,
      "step": 54935
    },
    {
      "epoch": 8.962479608482871,
      "grad_norm": 0.06750141829252243,
      "learning_rate": 1.6221872192191651e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 118554384,
      "step": 54940
    },
    {
      "epoch": 8.963295269168025,
      "grad_norm": 4.921962261199951,
      "learning_rate": 1.6196659030811938e-06,
      "loss": 0.133,
      "num_input_tokens_seen": 118566064,
      "step": 54945
    },
    {
      "epoch": 8.964110929853181,
      "grad_norm": 0.10307639092206955,
      "learning_rate": 1.6171464822717337e-06,
      "loss": 0.1893,
      "num_input_tokens_seen": 118575952,
      "step": 54950
    },
    {
      "epoch": 8.964926590538337,
      "grad_norm": 0.07564139366149902,
      "learning_rate": 1.6146289569950208e-06,
      "loss": 0.0843,
      "num_input_tokens_seen": 118587408,
      "step": 54955
    },
    {
      "epoch": 8.96574225122349,
      "grad_norm": 0.030069146305322647,
      "learning_rate": 1.612113327455142e-06,
      "loss": 0.2027,
      "num_input_tokens_seen": 118598320,
      "step": 54960
    },
    {
      "epoch": 8.966557911908646,
      "grad_norm": 0.837108314037323,
      "learning_rate": 1.6095995938560288e-06,
      "loss": 0.1128,
      "num_input_tokens_seen": 118609488,
      "step": 54965
    },
    {
      "epoch": 8.9673735725938,
      "grad_norm": 0.14493216574192047,
      "learning_rate": 1.6070877564014514e-06,
      "loss": 0.3061,
      "num_input_tokens_seen": 118621040,
      "step": 54970
    },
    {
      "epoch": 8.968189233278956,
      "grad_norm": 0.35746636986732483,
      "learning_rate": 1.6045778152950357e-06,
      "loss": 0.0958,
      "num_input_tokens_seen": 118632720,
      "step": 54975
    },
    {
      "epoch": 8.969004893964112,
      "grad_norm": 0.06156148388981819,
      "learning_rate": 1.6020697707402472e-06,
      "loss": 0.0043,
      "num_input_tokens_seen": 118643024,
      "step": 54980
    },
    {
      "epoch": 8.969820554649266,
      "grad_norm": 8.342617988586426,
      "learning_rate": 1.599563622940406e-06,
      "loss": 0.2614,
      "num_input_tokens_seen": 118653712,
      "step": 54985
    },
    {
      "epoch": 8.970636215334421,
      "grad_norm": 0.11663028597831726,
      "learning_rate": 1.5970593720986666e-06,
      "loss": 0.3216,
      "num_input_tokens_seen": 118664400,
      "step": 54990
    },
    {
      "epoch": 8.971451876019575,
      "grad_norm": 0.11002062261104584,
      "learning_rate": 1.5945570184180413e-06,
      "loss": 0.1151,
      "num_input_tokens_seen": 118675504,
      "step": 54995
    },
    {
      "epoch": 8.97226753670473,
      "grad_norm": 0.08915618062019348,
      "learning_rate": 1.5920565621013822e-06,
      "loss": 0.0066,
      "num_input_tokens_seen": 118686192,
      "step": 55000
    },
    {
      "epoch": 8.973083197389887,
      "grad_norm": 0.06287045776844025,
      "learning_rate": 1.5895580033513908e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 118697776,
      "step": 55005
    },
    {
      "epoch": 8.97389885807504,
      "grad_norm": 0.08433584868907928,
      "learning_rate": 1.5870613423706103e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 118708720,
      "step": 55010
    },
    {
      "epoch": 8.974714518760196,
      "grad_norm": 0.4887405037879944,
      "learning_rate": 1.5845665793614322e-06,
      "loss": 0.0047,
      "num_input_tokens_seen": 118720112,
      "step": 55015
    },
    {
      "epoch": 8.97553017944535,
      "grad_norm": 3.545954465866089,
      "learning_rate": 1.5820737145260972e-06,
      "loss": 0.3707,
      "num_input_tokens_seen": 118731184,
      "step": 55020
    },
    {
      "epoch": 8.976345840130506,
      "grad_norm": 0.12726637721061707,
      "learning_rate": 1.5795827480666852e-06,
      "loss": 0.1355,
      "num_input_tokens_seen": 118741392,
      "step": 55025
    },
    {
      "epoch": 8.977161500815662,
      "grad_norm": 5.345801830291748,
      "learning_rate": 1.5770936801851322e-06,
      "loss": 0.0954,
      "num_input_tokens_seen": 118751984,
      "step": 55030
    },
    {
      "epoch": 8.977977161500815,
      "grad_norm": 0.9549118280410767,
      "learning_rate": 1.5746065110832124e-06,
      "loss": 0.0093,
      "num_input_tokens_seen": 118763600,
      "step": 55035
    },
    {
      "epoch": 8.978792822185971,
      "grad_norm": 0.08882303535938263,
      "learning_rate": 1.5721212409625485e-06,
      "loss": 0.0072,
      "num_input_tokens_seen": 118774512,
      "step": 55040
    },
    {
      "epoch": 8.979608482871125,
      "grad_norm": 0.07915233075618744,
      "learning_rate": 1.5696378700246094e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 118785072,
      "step": 55045
    },
    {
      "epoch": 8.98042414355628,
      "grad_norm": 0.07184240221977234,
      "learning_rate": 1.5671563984707066e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 118795984,
      "step": 55050
    },
    {
      "epoch": 8.981239804241435,
      "grad_norm": 0.030804220587015152,
      "learning_rate": 1.564676826502004e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 118804656,
      "step": 55055
    },
    {
      "epoch": 8.98205546492659,
      "grad_norm": 0.8717603087425232,
      "learning_rate": 1.5621991543195103e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 118814224,
      "step": 55060
    },
    {
      "epoch": 8.982871125611746,
      "grad_norm": 0.03816085681319237,
      "learning_rate": 1.5597233821240732e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 118825552,
      "step": 55065
    },
    {
      "epoch": 8.9836867862969,
      "grad_norm": 0.06686482578516006,
      "learning_rate": 1.557249510116393e-06,
      "loss": 0.1324,
      "num_input_tokens_seen": 118836752,
      "step": 55070
    },
    {
      "epoch": 8.984502446982056,
      "grad_norm": 0.04887497425079346,
      "learning_rate": 1.5547775384970154e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 118846160,
      "step": 55075
    },
    {
      "epoch": 8.98531810766721,
      "grad_norm": 0.04898509755730629,
      "learning_rate": 1.5523074674663296e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 118857168,
      "step": 55080
    },
    {
      "epoch": 8.986133768352365,
      "grad_norm": 0.08052734285593033,
      "learning_rate": 1.549839297224573e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 118868432,
      "step": 55085
    },
    {
      "epoch": 8.986949429037521,
      "grad_norm": 0.03552599251270294,
      "learning_rate": 1.5473730279718296e-06,
      "loss": 0.014,
      "num_input_tokens_seen": 118879120,
      "step": 55090
    },
    {
      "epoch": 8.987765089722675,
      "grad_norm": 0.027884574607014656,
      "learning_rate": 1.5449086599080204e-06,
      "loss": 0.0991,
      "num_input_tokens_seen": 118889040,
      "step": 55095
    },
    {
      "epoch": 8.98858075040783,
      "grad_norm": 0.027779938653111458,
      "learning_rate": 1.5424461932329298e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 118898608,
      "step": 55100
    },
    {
      "epoch": 8.989396411092985,
      "grad_norm": 0.09449734538793564,
      "learning_rate": 1.5399856281461734e-06,
      "loss": 0.003,
      "num_input_tokens_seen": 118909264,
      "step": 55105
    },
    {
      "epoch": 8.99021207177814,
      "grad_norm": 0.1730055809020996,
      "learning_rate": 1.5375269648472162e-06,
      "loss": 0.1187,
      "num_input_tokens_seen": 118920048,
      "step": 55110
    },
    {
      "epoch": 8.991027732463294,
      "grad_norm": 0.19182473421096802,
      "learning_rate": 1.5350702035353716e-06,
      "loss": 0.0072,
      "num_input_tokens_seen": 118931120,
      "step": 55115
    },
    {
      "epoch": 8.99184339314845,
      "grad_norm": 0.4815743565559387,
      "learning_rate": 1.5326153444097934e-06,
      "loss": 0.0793,
      "num_input_tokens_seen": 118940464,
      "step": 55120
    },
    {
      "epoch": 8.992659053833606,
      "grad_norm": 0.07847863435745239,
      "learning_rate": 1.5301623876694898e-06,
      "loss": 0.1133,
      "num_input_tokens_seen": 118951120,
      "step": 55125
    },
    {
      "epoch": 8.99347471451876,
      "grad_norm": 2.0929036140441895,
      "learning_rate": 1.5277113335133097e-06,
      "loss": 0.1297,
      "num_input_tokens_seen": 118961616,
      "step": 55130
    },
    {
      "epoch": 8.994290375203915,
      "grad_norm": 0.070279061794281,
      "learning_rate": 1.5252621821399443e-06,
      "loss": 0.1001,
      "num_input_tokens_seen": 118973296,
      "step": 55135
    },
    {
      "epoch": 8.99510603588907,
      "grad_norm": 38.61929702758789,
      "learning_rate": 1.5228149337479347e-06,
      "loss": 0.0552,
      "num_input_tokens_seen": 118983344,
      "step": 55140
    },
    {
      "epoch": 8.995921696574225,
      "grad_norm": 0.033813487738370895,
      "learning_rate": 1.5203695885356694e-06,
      "loss": 0.0651,
      "num_input_tokens_seen": 118994544,
      "step": 55145
    },
    {
      "epoch": 8.99673735725938,
      "grad_norm": 0.29596346616744995,
      "learning_rate": 1.5179261467013817e-06,
      "loss": 0.0592,
      "num_input_tokens_seen": 119005264,
      "step": 55150
    },
    {
      "epoch": 8.997553017944535,
      "grad_norm": 4.728759765625,
      "learning_rate": 1.5154846084431463e-06,
      "loss": 0.2518,
      "num_input_tokens_seen": 119016688,
      "step": 55155
    },
    {
      "epoch": 8.99836867862969,
      "grad_norm": 5.094381809234619,
      "learning_rate": 1.513044973958888e-06,
      "loss": 0.15,
      "num_input_tokens_seen": 119027856,
      "step": 55160
    },
    {
      "epoch": 8.999184339314844,
      "grad_norm": 4.91703987121582,
      "learning_rate": 1.5106072434463742e-06,
      "loss": 0.1511,
      "num_input_tokens_seen": 119038000,
      "step": 55165
    },
    {
      "epoch": 9.0,
      "grad_norm": 0.021419517695903778,
      "learning_rate": 1.5081714171032186e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 119047920,
      "step": 55170
    },
    {
      "epoch": 9.0,
      "eval_loss": 0.19297294318675995,
      "eval_runtime": 568.4632,
      "eval_samples_per_second": 4.794,
      "eval_steps_per_second": 1.2,
      "num_input_tokens_seen": 119047920,
      "step": 55170
    },
    {
      "epoch": 9.000815660685156,
      "grad_norm": 7.182825088500977,
      "learning_rate": 1.5057374951268883e-06,
      "loss": 0.1803,
      "num_input_tokens_seen": 119058096,
      "step": 55175
    },
    {
      "epoch": 9.00163132137031,
      "grad_norm": 3.8748457431793213,
      "learning_rate": 1.5033054777146838e-06,
      "loss": 0.128,
      "num_input_tokens_seen": 119068080,
      "step": 55180
    },
    {
      "epoch": 9.002446982055465,
      "grad_norm": 4.433473587036133,
      "learning_rate": 1.5008753650637585e-06,
      "loss": 0.2174,
      "num_input_tokens_seen": 119078576,
      "step": 55185
    },
    {
      "epoch": 9.00326264274062,
      "grad_norm": 0.3918544352054596,
      "learning_rate": 1.4984471573711105e-06,
      "loss": 0.1319,
      "num_input_tokens_seen": 119089808,
      "step": 55190
    },
    {
      "epoch": 9.004078303425775,
      "grad_norm": 0.045883674174547195,
      "learning_rate": 1.4960208548335825e-06,
      "loss": 0.203,
      "num_input_tokens_seen": 119100208,
      "step": 55195
    },
    {
      "epoch": 9.00489396411093,
      "grad_norm": 0.216802179813385,
      "learning_rate": 1.4935964576478584e-06,
      "loss": 0.0988,
      "num_input_tokens_seen": 119111632,
      "step": 55200
    },
    {
      "epoch": 9.005709624796085,
      "grad_norm": 0.16552449762821198,
      "learning_rate": 1.4911739660104785e-06,
      "loss": 0.0147,
      "num_input_tokens_seen": 119122608,
      "step": 55205
    },
    {
      "epoch": 9.00652528548124,
      "grad_norm": 0.10056468844413757,
      "learning_rate": 1.4887533801178188e-06,
      "loss": 0.0041,
      "num_input_tokens_seen": 119132976,
      "step": 55210
    },
    {
      "epoch": 9.007340946166394,
      "grad_norm": 2.736128568649292,
      "learning_rate": 1.486334700166106e-06,
      "loss": 0.1052,
      "num_input_tokens_seen": 119143984,
      "step": 55215
    },
    {
      "epoch": 9.00815660685155,
      "grad_norm": 0.6527325510978699,
      "learning_rate": 1.483917926351408e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 119154256,
      "step": 55220
    },
    {
      "epoch": 9.008972267536704,
      "grad_norm": 0.30277392268180847,
      "learning_rate": 1.4815030588696432e-06,
      "loss": 0.0959,
      "num_input_tokens_seen": 119165712,
      "step": 55225
    },
    {
      "epoch": 9.00978792822186,
      "grad_norm": 0.04763905704021454,
      "learning_rate": 1.4790900979165717e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 119176496,
      "step": 55230
    },
    {
      "epoch": 9.010603588907015,
      "grad_norm": 0.08349156379699707,
      "learning_rate": 1.4766790436878035e-06,
      "loss": 0.1009,
      "num_input_tokens_seen": 119186640,
      "step": 55235
    },
    {
      "epoch": 9.01141924959217,
      "grad_norm": 5.5012922286987305,
      "learning_rate": 1.4742698963787854e-06,
      "loss": 0.2143,
      "num_input_tokens_seen": 119197680,
      "step": 55240
    },
    {
      "epoch": 9.012234910277325,
      "grad_norm": 0.1282111555337906,
      "learning_rate": 1.4718626561848193e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 119208080,
      "step": 55245
    },
    {
      "epoch": 9.013050570962479,
      "grad_norm": 4.5648908615112305,
      "learning_rate": 1.469457323301046e-06,
      "loss": 0.134,
      "num_input_tokens_seen": 119218896,
      "step": 55250
    },
    {
      "epoch": 9.013866231647635,
      "grad_norm": 4.804664134979248,
      "learning_rate": 1.4670538979224547e-06,
      "loss": 0.4,
      "num_input_tokens_seen": 119230032,
      "step": 55255
    },
    {
      "epoch": 9.01468189233279,
      "grad_norm": 0.05063999816775322,
      "learning_rate": 1.4646523802438805e-06,
      "loss": 0.011,
      "num_input_tokens_seen": 119240112,
      "step": 55260
    },
    {
      "epoch": 9.015497553017944,
      "grad_norm": 0.05710785835981369,
      "learning_rate": 1.4622527704599986e-06,
      "loss": 0.3593,
      "num_input_tokens_seen": 119251184,
      "step": 55265
    },
    {
      "epoch": 9.0163132137031,
      "grad_norm": 0.09515246003866196,
      "learning_rate": 1.4598550687653394e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 119263216,
      "step": 55270
    },
    {
      "epoch": 9.017128874388254,
      "grad_norm": 0.18920369446277618,
      "learning_rate": 1.4574592753542698e-06,
      "loss": 0.0086,
      "num_input_tokens_seen": 119272112,
      "step": 55275
    },
    {
      "epoch": 9.01794453507341,
      "grad_norm": 0.7133920192718506,
      "learning_rate": 1.4550653904210038e-06,
      "loss": 0.1259,
      "num_input_tokens_seen": 119282352,
      "step": 55280
    },
    {
      "epoch": 9.018760195758565,
      "grad_norm": 0.07079581916332245,
      "learning_rate": 1.4526734141596026e-06,
      "loss": 0.1497,
      "num_input_tokens_seen": 119293488,
      "step": 55285
    },
    {
      "epoch": 9.01957585644372,
      "grad_norm": 10.122812271118164,
      "learning_rate": 1.4502833467639725e-06,
      "loss": 0.1082,
      "num_input_tokens_seen": 119304592,
      "step": 55290
    },
    {
      "epoch": 9.020391517128875,
      "grad_norm": 8.083893775939941,
      "learning_rate": 1.4478951884278669e-06,
      "loss": 0.1842,
      "num_input_tokens_seen": 119315088,
      "step": 55295
    },
    {
      "epoch": 9.021207177814029,
      "grad_norm": 3.10662579536438,
      "learning_rate": 1.4455089393448778e-06,
      "loss": 0.1553,
      "num_input_tokens_seen": 119326416,
      "step": 55300
    },
    {
      "epoch": 9.022022838499185,
      "grad_norm": 0.252075731754303,
      "learning_rate": 1.4431245997084425e-06,
      "loss": 0.006,
      "num_input_tokens_seen": 119337296,
      "step": 55305
    },
    {
      "epoch": 9.022838499184338,
      "grad_norm": 0.2267480194568634,
      "learning_rate": 1.4407421697118617e-06,
      "loss": 0.0845,
      "num_input_tokens_seen": 119347824,
      "step": 55310
    },
    {
      "epoch": 9.023654159869494,
      "grad_norm": 0.1325957477092743,
      "learning_rate": 1.438361649548256e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 119359216,
      "step": 55315
    },
    {
      "epoch": 9.02446982055465,
      "grad_norm": 0.06241501495242119,
      "learning_rate": 1.4359830394106071e-06,
      "loss": 0.1319,
      "num_input_tokens_seen": 119370800,
      "step": 55320
    },
    {
      "epoch": 9.025285481239804,
      "grad_norm": 3.341966152191162,
      "learning_rate": 1.4336063394917333e-06,
      "loss": 0.2197,
      "num_input_tokens_seen": 119380944,
      "step": 55325
    },
    {
      "epoch": 9.02610114192496,
      "grad_norm": 0.05169983580708504,
      "learning_rate": 1.4312315499843077e-06,
      "loss": 0.101,
      "num_input_tokens_seen": 119390608,
      "step": 55330
    },
    {
      "epoch": 9.026916802610113,
      "grad_norm": 0.06638845056295395,
      "learning_rate": 1.428858671080835e-06,
      "loss": 0.007,
      "num_input_tokens_seen": 119401328,
      "step": 55335
    },
    {
      "epoch": 9.02773246329527,
      "grad_norm": 0.13590192794799805,
      "learning_rate": 1.4264877029736778e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 119411472,
      "step": 55340
    },
    {
      "epoch": 9.028548123980425,
      "grad_norm": 0.03334615379571915,
      "learning_rate": 1.424118645855041e-06,
      "loss": 0.0096,
      "num_input_tokens_seen": 119423152,
      "step": 55345
    },
    {
      "epoch": 9.029363784665579,
      "grad_norm": 0.11990103870630264,
      "learning_rate": 1.4217514999169678e-06,
      "loss": 0.0053,
      "num_input_tokens_seen": 119434992,
      "step": 55350
    },
    {
      "epoch": 9.030179445350734,
      "grad_norm": 0.11351681500673294,
      "learning_rate": 1.4193862653513524e-06,
      "loss": 0.2898,
      "num_input_tokens_seen": 119445520,
      "step": 55355
    },
    {
      "epoch": 9.030995106035888,
      "grad_norm": 0.11845037341117859,
      "learning_rate": 1.4170229423499353e-06,
      "loss": 0.0168,
      "num_input_tokens_seen": 119456080,
      "step": 55360
    },
    {
      "epoch": 9.031810766721044,
      "grad_norm": 0.11720054596662521,
      "learning_rate": 1.4146615311042972e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 119467504,
      "step": 55365
    },
    {
      "epoch": 9.0326264274062,
      "grad_norm": 2.3431005477905273,
      "learning_rate": 1.4123020318058649e-06,
      "loss": 0.2547,
      "num_input_tokens_seen": 119478896,
      "step": 55370
    },
    {
      "epoch": 9.033442088091354,
      "grad_norm": 0.11629412323236465,
      "learning_rate": 1.4099444446459138e-06,
      "loss": 0.1221,
      "num_input_tokens_seen": 119489840,
      "step": 55375
    },
    {
      "epoch": 9.03425774877651,
      "grad_norm": 0.06350281089544296,
      "learning_rate": 1.4075887698155599e-06,
      "loss": 0.0078,
      "num_input_tokens_seen": 119499472,
      "step": 55380
    },
    {
      "epoch": 9.035073409461663,
      "grad_norm": 0.09608178585767746,
      "learning_rate": 1.4052350075057673e-06,
      "loss": 0.1319,
      "num_input_tokens_seen": 119510768,
      "step": 55385
    },
    {
      "epoch": 9.035889070146819,
      "grad_norm": 0.07565252482891083,
      "learning_rate": 1.4028831579073448e-06,
      "loss": 0.1231,
      "num_input_tokens_seen": 119521968,
      "step": 55390
    },
    {
      "epoch": 9.036704730831975,
      "grad_norm": 0.009342601522803307,
      "learning_rate": 1.4005332212109424e-06,
      "loss": 0.1157,
      "num_input_tokens_seen": 119532912,
      "step": 55395
    },
    {
      "epoch": 9.037520391517129,
      "grad_norm": 0.1403772234916687,
      "learning_rate": 1.3981851976070603e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 119544208,
      "step": 55400
    },
    {
      "epoch": 9.038336052202284,
      "grad_norm": 0.0709967240691185,
      "learning_rate": 1.395839087286041e-06,
      "loss": 0.0075,
      "num_input_tokens_seen": 119554960,
      "step": 55405
    },
    {
      "epoch": 9.039151712887438,
      "grad_norm": 0.05165521055459976,
      "learning_rate": 1.3934948904380712e-06,
      "loss": 0.1353,
      "num_input_tokens_seen": 119566224,
      "step": 55410
    },
    {
      "epoch": 9.039967373572594,
      "grad_norm": 0.09342238306999207,
      "learning_rate": 1.3911526072531795e-06,
      "loss": 0.0979,
      "num_input_tokens_seen": 119577424,
      "step": 55415
    },
    {
      "epoch": 9.040783034257748,
      "grad_norm": 0.1905154436826706,
      "learning_rate": 1.3888122379212527e-06,
      "loss": 0.1214,
      "num_input_tokens_seen": 119588752,
      "step": 55420
    },
    {
      "epoch": 9.041598694942904,
      "grad_norm": 0.0606345497071743,
      "learning_rate": 1.3864737826320058e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 119600048,
      "step": 55425
    },
    {
      "epoch": 9.04241435562806,
      "grad_norm": 18.631378173828125,
      "learning_rate": 1.3841372415750093e-06,
      "loss": 0.2015,
      "num_input_tokens_seen": 119608944,
      "step": 55430
    },
    {
      "epoch": 9.043230016313213,
      "grad_norm": 0.08189153671264648,
      "learning_rate": 1.381802614939673e-06,
      "loss": 0.0864,
      "num_input_tokens_seen": 119618512,
      "step": 55435
    },
    {
      "epoch": 9.044045676998369,
      "grad_norm": 0.09821552783250809,
      "learning_rate": 1.3794699029152563e-06,
      "loss": 0.0163,
      "num_input_tokens_seen": 119629040,
      "step": 55440
    },
    {
      "epoch": 9.044861337683523,
      "grad_norm": 0.04215510934591293,
      "learning_rate": 1.3771391056908555e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 119639248,
      "step": 55445
    },
    {
      "epoch": 9.045676998368679,
      "grad_norm": 0.07656820118427277,
      "learning_rate": 1.3748102234554222e-06,
      "loss": 0.0091,
      "num_input_tokens_seen": 119648848,
      "step": 55450
    },
    {
      "epoch": 9.046492659053834,
      "grad_norm": 1.7920234203338623,
      "learning_rate": 1.372483256397744e-06,
      "loss": 0.0094,
      "num_input_tokens_seen": 119660624,
      "step": 55455
    },
    {
      "epoch": 9.047308319738988,
      "grad_norm": 2.013300895690918,
      "learning_rate": 1.3701582047064592e-06,
      "loss": 0.0095,
      "num_input_tokens_seen": 119671120,
      "step": 55460
    },
    {
      "epoch": 9.048123980424144,
      "grad_norm": 0.06278388947248459,
      "learning_rate": 1.3678350685700447e-06,
      "loss": 0.2559,
      "num_input_tokens_seen": 119680720,
      "step": 55465
    },
    {
      "epoch": 9.048939641109298,
      "grad_norm": 0.18598142266273499,
      "learning_rate": 1.3655138481768303e-06,
      "loss": 0.079,
      "num_input_tokens_seen": 119690896,
      "step": 55470
    },
    {
      "epoch": 9.049755301794454,
      "grad_norm": 0.1875404268503189,
      "learning_rate": 1.3631945437149823e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 119699824,
      "step": 55475
    },
    {
      "epoch": 9.05057096247961,
      "grad_norm": 0.9875158667564392,
      "learning_rate": 1.3608771553725168e-06,
      "loss": 0.1063,
      "num_input_tokens_seen": 119709552,
      "step": 55480
    },
    {
      "epoch": 9.051386623164763,
      "grad_norm": 0.09905755519866943,
      "learning_rate": 1.3585616833372894e-06,
      "loss": 0.0125,
      "num_input_tokens_seen": 119721136,
      "step": 55485
    },
    {
      "epoch": 9.052202283849919,
      "grad_norm": 0.09456333518028259,
      "learning_rate": 1.3562481277970108e-06,
      "loss": 0.1033,
      "num_input_tokens_seen": 119731408,
      "step": 55490
    },
    {
      "epoch": 9.053017944535073,
      "grad_norm": 0.03764314204454422,
      "learning_rate": 1.3539364889392281e-06,
      "loss": 0.2635,
      "num_input_tokens_seen": 119743728,
      "step": 55495
    },
    {
      "epoch": 9.053833605220229,
      "grad_norm": 0.17601527273654938,
      "learning_rate": 1.3516267669513305e-06,
      "loss": 0.0888,
      "num_input_tokens_seen": 119754992,
      "step": 55500
    },
    {
      "epoch": 9.054649265905383,
      "grad_norm": 0.10600312799215317,
      "learning_rate": 1.3493189620205572e-06,
      "loss": 0.0077,
      "num_input_tokens_seen": 119765936,
      "step": 55505
    },
    {
      "epoch": 9.055464926590538,
      "grad_norm": 0.029680950567126274,
      "learning_rate": 1.3470130743339914e-06,
      "loss": 0.0018,
      "num_input_tokens_seen": 119776688,
      "step": 55510
    },
    {
      "epoch": 9.056280587275694,
      "grad_norm": 2.44942569732666,
      "learning_rate": 1.3447091040785619e-06,
      "loss": 0.0859,
      "num_input_tokens_seen": 119787088,
      "step": 55515
    },
    {
      "epoch": 9.057096247960848,
      "grad_norm": 0.09405805170536041,
      "learning_rate": 1.342407051441033e-06,
      "loss": 0.0126,
      "num_input_tokens_seen": 119798576,
      "step": 55520
    },
    {
      "epoch": 9.057911908646004,
      "grad_norm": 3.5197908878326416,
      "learning_rate": 1.3401069166080278e-06,
      "loss": 0.082,
      "num_input_tokens_seen": 119809424,
      "step": 55525
    },
    {
      "epoch": 9.058727569331158,
      "grad_norm": 0.5775467753410339,
      "learning_rate": 1.3378086997660077e-06,
      "loss": 0.0104,
      "num_input_tokens_seen": 119821232,
      "step": 55530
    },
    {
      "epoch": 9.059543230016313,
      "grad_norm": 0.0336451530456543,
      "learning_rate": 1.3355124011012744e-06,
      "loss": 0.1223,
      "num_input_tokens_seen": 119832688,
      "step": 55535
    },
    {
      "epoch": 9.060358890701469,
      "grad_norm": 23.774309158325195,
      "learning_rate": 1.3332180207999783e-06,
      "loss": 0.2499,
      "num_input_tokens_seen": 119842736,
      "step": 55540
    },
    {
      "epoch": 9.061174551386623,
      "grad_norm": 0.11912717670202255,
      "learning_rate": 1.3309255590481129e-06,
      "loss": 0.011,
      "num_input_tokens_seen": 119853808,
      "step": 55545
    },
    {
      "epoch": 9.061990212071779,
      "grad_norm": 0.08176933974027634,
      "learning_rate": 1.3286350160315181e-06,
      "loss": 0.2745,
      "num_input_tokens_seen": 119864976,
      "step": 55550
    },
    {
      "epoch": 9.062805872756933,
      "grad_norm": 0.09938140958547592,
      "learning_rate": 1.3263463919358759e-06,
      "loss": 0.0142,
      "num_input_tokens_seen": 119875312,
      "step": 55555
    },
    {
      "epoch": 9.063621533442088,
      "grad_norm": 0.1812557876110077,
      "learning_rate": 1.3240596869467158e-06,
      "loss": 0.08,
      "num_input_tokens_seen": 119886704,
      "step": 55560
    },
    {
      "epoch": 9.064437194127244,
      "grad_norm": 0.31800466775894165,
      "learning_rate": 1.3217749012494062e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 119897424,
      "step": 55565
    },
    {
      "epoch": 9.065252854812398,
      "grad_norm": 18.320484161376953,
      "learning_rate": 1.3194920350291657e-06,
      "loss": 0.0201,
      "num_input_tokens_seen": 119909520,
      "step": 55570
    },
    {
      "epoch": 9.066068515497554,
      "grad_norm": 0.03986978158354759,
      "learning_rate": 1.3172110884710541e-06,
      "loss": 0.1012,
      "num_input_tokens_seen": 119919056,
      "step": 55575
    },
    {
      "epoch": 9.066884176182707,
      "grad_norm": 0.23460790514945984,
      "learning_rate": 1.314932061759977e-06,
      "loss": 0.1249,
      "num_input_tokens_seen": 119930448,
      "step": 55580
    },
    {
      "epoch": 9.067699836867863,
      "grad_norm": 0.07698375731706619,
      "learning_rate": 1.3126549550806832e-06,
      "loss": 0.0044,
      "num_input_tokens_seen": 119939824,
      "step": 55585
    },
    {
      "epoch": 9.068515497553017,
      "grad_norm": 0.6116136908531189,
      "learning_rate": 1.310379768617767e-06,
      "loss": 0.0445,
      "num_input_tokens_seen": 119951152,
      "step": 55590
    },
    {
      "epoch": 9.069331158238173,
      "grad_norm": 8.246176719665527,
      "learning_rate": 1.308106502555667e-06,
      "loss": 0.0944,
      "num_input_tokens_seen": 119961104,
      "step": 55595
    },
    {
      "epoch": 9.070146818923329,
      "grad_norm": 3.7423102855682373,
      "learning_rate": 1.3058351570786665e-06,
      "loss": 0.0187,
      "num_input_tokens_seen": 119971920,
      "step": 55600
    },
    {
      "epoch": 9.070962479608482,
      "grad_norm": 0.016244083642959595,
      "learning_rate": 1.3035657323708927e-06,
      "loss": 0.1457,
      "num_input_tokens_seen": 119980944,
      "step": 55605
    },
    {
      "epoch": 9.071778140293638,
      "grad_norm": 0.15408752858638763,
      "learning_rate": 1.3012982286163129e-06,
      "loss": 0.1199,
      "num_input_tokens_seen": 119991792,
      "step": 55610
    },
    {
      "epoch": 9.072593800978792,
      "grad_norm": 0.6007954478263855,
      "learning_rate": 1.2990326459987434e-06,
      "loss": 0.0087,
      "num_input_tokens_seen": 120002064,
      "step": 55615
    },
    {
      "epoch": 9.073409461663948,
      "grad_norm": 0.48222196102142334,
      "learning_rate": 1.296768984701846e-06,
      "loss": 0.1087,
      "num_input_tokens_seen": 120012560,
      "step": 55620
    },
    {
      "epoch": 9.074225122349104,
      "grad_norm": 0.0767175629734993,
      "learning_rate": 1.2945072449091212e-06,
      "loss": 0.0098,
      "num_input_tokens_seen": 120023568,
      "step": 55625
    },
    {
      "epoch": 9.075040783034257,
      "grad_norm": 0.09177295118570328,
      "learning_rate": 1.292247426803922e-06,
      "loss": 0.2351,
      "num_input_tokens_seen": 120034448,
      "step": 55630
    },
    {
      "epoch": 9.075856443719413,
      "grad_norm": 0.23538443446159363,
      "learning_rate": 1.2899895305694408e-06,
      "loss": 0.1497,
      "num_input_tokens_seen": 120046096,
      "step": 55635
    },
    {
      "epoch": 9.076672104404567,
      "grad_norm": 0.14989183843135834,
      "learning_rate": 1.2877335563887095e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 120055344,
      "step": 55640
    },
    {
      "epoch": 9.077487765089723,
      "grad_norm": 0.11194797605276108,
      "learning_rate": 1.2854795044446116e-06,
      "loss": 0.0068,
      "num_input_tokens_seen": 120065808,
      "step": 55645
    },
    {
      "epoch": 9.078303425774878,
      "grad_norm": 0.3517356514930725,
      "learning_rate": 1.2832273749198708e-06,
      "loss": 0.0977,
      "num_input_tokens_seen": 120075472,
      "step": 55650
    },
    {
      "epoch": 9.079119086460032,
      "grad_norm": 0.037554457783699036,
      "learning_rate": 1.2809771679970522e-06,
      "loss": 0.0675,
      "num_input_tokens_seen": 120085072,
      "step": 55655
    },
    {
      "epoch": 9.079934747145188,
      "grad_norm": 0.11872192472219467,
      "learning_rate": 1.2787288838585793e-06,
      "loss": 0.1116,
      "num_input_tokens_seen": 120096304,
      "step": 55660
    },
    {
      "epoch": 9.080750407830342,
      "grad_norm": 0.09016989171504974,
      "learning_rate": 1.2764825226867005e-06,
      "loss": 0.0033,
      "num_input_tokens_seen": 120105968,
      "step": 55665
    },
    {
      "epoch": 9.081566068515498,
      "grad_norm": 0.07086058706045151,
      "learning_rate": 1.2742380846635231e-06,
      "loss": 0.1139,
      "num_input_tokens_seen": 120117360,
      "step": 55670
    },
    {
      "epoch": 9.082381729200652,
      "grad_norm": 0.3153351843357086,
      "learning_rate": 1.2719955699709907e-06,
      "loss": 0.1097,
      "num_input_tokens_seen": 120128464,
      "step": 55675
    },
    {
      "epoch": 9.083197389885807,
      "grad_norm": 0.13674883544445038,
      "learning_rate": 1.2697549787908908e-06,
      "loss": 0.0048,
      "num_input_tokens_seen": 120139440,
      "step": 55680
    },
    {
      "epoch": 9.084013050570963,
      "grad_norm": 0.11322315782308578,
      "learning_rate": 1.267516311304856e-06,
      "loss": 0.0101,
      "num_input_tokens_seen": 120149392,
      "step": 55685
    },
    {
      "epoch": 9.084828711256117,
      "grad_norm": 4.931153774261475,
      "learning_rate": 1.265279567694369e-06,
      "loss": 0.359,
      "num_input_tokens_seen": 120160848,
      "step": 55690
    },
    {
      "epoch": 9.085644371941273,
      "grad_norm": 0.08744145929813385,
      "learning_rate": 1.2630447481407486e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 120172368,
      "step": 55695
    },
    {
      "epoch": 9.086460032626427,
      "grad_norm": 0.0977356880903244,
      "learning_rate": 1.2608118528251611e-06,
      "loss": 0.0542,
      "num_input_tokens_seen": 120183184,
      "step": 55700
    },
    {
      "epoch": 9.087275693311582,
      "grad_norm": 0.06535228341817856,
      "learning_rate": 1.2585808819286172e-06,
      "loss": 0.0083,
      "num_input_tokens_seen": 120194480,
      "step": 55705
    },
    {
      "epoch": 9.088091353996738,
      "grad_norm": 0.07237150520086288,
      "learning_rate": 1.2563518356319664e-06,
      "loss": 0.0734,
      "num_input_tokens_seen": 120205616,
      "step": 55710
    },
    {
      "epoch": 9.088907014681892,
      "grad_norm": 0.06289654970169067,
      "learning_rate": 1.2541247141159119e-06,
      "loss": 0.189,
      "num_input_tokens_seen": 120216496,
      "step": 55715
    },
    {
      "epoch": 9.089722675367048,
      "grad_norm": 3.9390320777893066,
      "learning_rate": 1.2518995175609949e-06,
      "loss": 0.3441,
      "num_input_tokens_seen": 120226896,
      "step": 55720
    },
    {
      "epoch": 9.090538336052202,
      "grad_norm": 0.6136237382888794,
      "learning_rate": 1.2496762461475992e-06,
      "loss": 0.006,
      "num_input_tokens_seen": 120236400,
      "step": 55725
    },
    {
      "epoch": 9.091353996737357,
      "grad_norm": 0.3308112919330597,
      "learning_rate": 1.2474549000559527e-06,
      "loss": 0.0056,
      "num_input_tokens_seen": 120247920,
      "step": 55730
    },
    {
      "epoch": 9.092169657422513,
      "grad_norm": 0.07058186829090118,
      "learning_rate": 1.245235479466131e-06,
      "loss": 0.0082,
      "num_input_tokens_seen": 120258512,
      "step": 55735
    },
    {
      "epoch": 9.092985318107667,
      "grad_norm": 21.97412872314453,
      "learning_rate": 1.2430179845580537e-06,
      "loss": 0.0963,
      "num_input_tokens_seen": 120270032,
      "step": 55740
    },
    {
      "epoch": 9.093800978792823,
      "grad_norm": 0.10977151989936829,
      "learning_rate": 1.24080241551148e-06,
      "loss": 0.0042,
      "num_input_tokens_seen": 120280816,
      "step": 55745
    },
    {
      "epoch": 9.094616639477977,
      "grad_norm": 0.09765518456697464,
      "learning_rate": 1.2385887725060135e-06,
      "loss": 0.1115,
      "num_input_tokens_seen": 120291664,
      "step": 55750
    },
    {
      "epoch": 9.095432300163132,
      "grad_norm": 0.8600424528121948,
      "learning_rate": 1.236377055721108e-06,
      "loss": 0.0672,
      "num_input_tokens_seen": 120302320,
      "step": 55755
    },
    {
      "epoch": 9.096247960848286,
      "grad_norm": 0.07710446417331696,
      "learning_rate": 1.234167265336053e-06,
      "loss": 0.0632,
      "num_input_tokens_seen": 120312688,
      "step": 55760
    },
    {
      "epoch": 9.097063621533442,
      "grad_norm": 0.11269769817590714,
      "learning_rate": 1.2319594015299862e-06,
      "loss": 0.0187,
      "num_input_tokens_seen": 120322928,
      "step": 55765
    },
    {
      "epoch": 9.097879282218598,
      "grad_norm": 0.1080789640545845,
      "learning_rate": 1.2297534644818891e-06,
      "loss": 0.0774,
      "num_input_tokens_seen": 120332080,
      "step": 55770
    },
    {
      "epoch": 9.098694942903752,
      "grad_norm": 0.05265503376722336,
      "learning_rate": 1.227549454370583e-06,
      "loss": 0.1184,
      "num_input_tokens_seen": 120344400,
      "step": 55775
    },
    {
      "epoch": 9.099510603588907,
      "grad_norm": 0.17311835289001465,
      "learning_rate": 1.225347371374741e-06,
      "loss": 0.0172,
      "num_input_tokens_seen": 120355344,
      "step": 55780
    },
    {
      "epoch": 9.100326264274061,
      "grad_norm": 0.08054272830486298,
      "learning_rate": 1.2231472156728707e-06,
      "loss": 0.007,
      "num_input_tokens_seen": 120366640,
      "step": 55785
    },
    {
      "epoch": 9.101141924959217,
      "grad_norm": 0.02693832479417324,
      "learning_rate": 1.2209489874433294e-06,
      "loss": 0.0046,
      "num_input_tokens_seen": 120377584,
      "step": 55790
    },
    {
      "epoch": 9.101957585644373,
      "grad_norm": 0.06729228049516678,
      "learning_rate": 1.2187526868643162e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 120388368,
      "step": 55795
    },
    {
      "epoch": 9.102773246329527,
      "grad_norm": 0.01973656751215458,
      "learning_rate": 1.2165583141138748e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 120399312,
      "step": 55800
    },
    {
      "epoch": 9.103588907014682,
      "grad_norm": 3.6598472595214844,
      "learning_rate": 1.2143658693698933e-06,
      "loss": 0.2122,
      "num_input_tokens_seen": 120409616,
      "step": 55805
    },
    {
      "epoch": 9.104404567699836,
      "grad_norm": 0.15364094078540802,
      "learning_rate": 1.212175352810102e-06,
      "loss": 0.1852,
      "num_input_tokens_seen": 120421424,
      "step": 55810
    },
    {
      "epoch": 9.105220228384992,
      "grad_norm": 0.4004209041595459,
      "learning_rate": 1.2099867646120754e-06,
      "loss": 0.0997,
      "num_input_tokens_seen": 120431088,
      "step": 55815
    },
    {
      "epoch": 9.106035889070148,
      "grad_norm": 0.09255687147378922,
      "learning_rate": 1.20780010495323e-06,
      "loss": 0.0032,
      "num_input_tokens_seen": 120442320,
      "step": 55820
    },
    {
      "epoch": 9.106851549755302,
      "grad_norm": 2.7680530548095703,
      "learning_rate": 1.2056153740108295e-06,
      "loss": 0.1717,
      "num_input_tokens_seen": 120453104,
      "step": 55825
    },
    {
      "epoch": 9.107667210440457,
      "grad_norm": 0.04503974691033363,
      "learning_rate": 1.2034325719619794e-06,
      "loss": 0.0051,
      "num_input_tokens_seen": 120463536,
      "step": 55830
    },
    {
      "epoch": 9.108482871125611,
      "grad_norm": 0.22261156141757965,
      "learning_rate": 1.2012516989836242e-06,
      "loss": 0.0845,
      "num_input_tokens_seen": 120474672,
      "step": 55835
    },
    {
      "epoch": 9.109298531810767,
      "grad_norm": 1.0107585191726685,
      "learning_rate": 1.1990727552525588e-06,
      "loss": 0.0063,
      "num_input_tokens_seen": 120484208,
      "step": 55840
    },
    {
      "epoch": 9.11011419249592,
      "grad_norm": 6.194400310516357,
      "learning_rate": 1.196895740945425e-06,
      "loss": 0.1166,
      "num_input_tokens_seen": 120494896,
      "step": 55845
    },
    {
      "epoch": 9.110929853181077,
      "grad_norm": 7.5901947021484375,
      "learning_rate": 1.194720656238696e-06,
      "loss": 0.0399,
      "num_input_tokens_seen": 120505936,
      "step": 55850
    },
    {
      "epoch": 9.111745513866232,
      "grad_norm": 0.11343776434659958,
      "learning_rate": 1.1925475013086968e-06,
      "loss": 0.0034,
      "num_input_tokens_seen": 120516944,
      "step": 55855
    },
    {
      "epoch": 9.112561174551386,
      "grad_norm": 0.29412412643432617,
      "learning_rate": 1.190376276331598e-06,
      "loss": 0.0055,
      "num_input_tokens_seen": 120527344,
      "step": 55860
    },
    {
      "epoch": 9.113376835236542,
      "grad_norm": 0.13387349247932434,
      "learning_rate": 1.1882069814834057e-06,
      "loss": 0.0037,
      "num_input_tokens_seen": 120537520,
      "step": 55865
    },
    {
      "epoch": 9.114192495921696,
      "grad_norm": 0.1723552793264389,
      "learning_rate": 1.186039616939974e-06,
      "loss": 0.1019,
      "num_input_tokens_seen": 120549072,
      "step": 55870
    },
    {
      "epoch": 9.115008156606851,
      "grad_norm": 21.81083106994629,
      "learning_rate": 1.1838741828770039e-06,
      "loss": 0.055,
      "num_input_tokens_seen": 120559888,
      "step": 55875
    },
    {
      "epoch": 9.115823817292007,
      "grad_norm": 0.06051301956176758,
      "learning_rate": 1.1817106794700327e-06,
      "loss": 0.1739,
      "num_input_tokens_seen": 120571504,
      "step": 55880
    },
    {
      "epoch": 9.116639477977161,
      "grad_norm": 15.309614181518555,
      "learning_rate": 1.1795491068944453e-06,
      "loss": 0.0321,
      "num_input_tokens_seen": 120581456,
      "step": 55885
    },
    {
      "epoch": 9.117455138662317,
      "grad_norm": 0.07540477812290192,
      "learning_rate": 1.1773894653254736e-06,
      "loss": 0.1269,
      "num_input_tokens_seen": 120592080,
      "step": 55890
    },
    {
      "epoch": 9.11827079934747,
      "grad_norm": 0.14149914681911469,
      "learning_rate": 1.1752317549381857e-06,
      "loss": 0.0064,
      "num_input_tokens_seen": 120602128,
      "step": 55895
    },
    {
      "epoch": 9.119086460032626,
      "grad_norm": 2.3565332889556885,
      "learning_rate": 1.1730759759074978e-06,
      "loss": 0.0081,
      "num_input_tokens_seen": 120614352,
      "step": 55900
    },
    {
      "epoch": 9.119902120717782,
      "grad_norm": 0.06479676812887192,
      "learning_rate": 1.1709221284081666e-06,
      "loss": 0.0039,
      "num_input_tokens_seen": 120625424,
      "step": 55905
    },
    {
      "epoch": 9.120717781402936,
      "grad_norm": 4.313600063323975,
      "learning_rate": 1.1687702126147976e-06,
      "loss": 0.0142,
      "num_input_tokens_seen": 120635536,
      "step": 55910
    },
    {
      "epoch": 9.121533442088092,
      "grad_norm": 0.020870855078101158,
      "learning_rate": 1.1666202287018313e-06,
      "loss": 0.1353,
      "num_input_tokens_seen": 120646064,
      "step": 55915
    },
    {
      "epoch": 9.122349102773246,
      "grad_norm": 0.04624507948756218,
      "learning_rate": 1.1644721768435617e-06,
      "loss": 0.0036,
      "num_input_tokens_seen": 120657072,
      "step": 55920
    },
    {
      "epoch": 9.123164763458401,
      "grad_norm": 0.15288259088993073,
      "learning_rate": 1.1623260572141137e-06,
      "loss": 0.101,
      "num_input_tokens_seen": 120668816,
      "step": 55925
    },
    {
      "epoch": 9.123980424143557,
      "grad_norm": 0.2106817364692688,
      "learning_rate": 1.16018186998747e-06,
      "loss": 0.0071,
      "num_input_tokens_seen": 120679248,
      "step": 55930
    },
    {
      "epoch": 9.124796084828711,
      "grad_norm": 0.14101427793502808,
      "learning_rate": 1.1580396153374446e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 120690096,
      "step": 55935
    },
    {
      "epoch": 9.125611745513867,
      "grad_norm": 0.4781815707683563,
      "learning_rate": 1.1558992934376982e-06,
      "loss": 0.0057,
      "num_input_tokens_seen": 120702352,
      "step": 55940
    },
    {
      "epoch": 9.12642740619902,
      "grad_norm": 0.09064584970474243,
      "learning_rate": 1.1537609044617398e-06,
      "loss": 0.1309,
      "num_input_tokens_seen": 120713072,
      "step": 55945
    },
    {
      "epoch": 9.127243066884176,
      "grad_norm": 0.033964842557907104,
      "learning_rate": 1.1516244485829193e-06,
      "loss": 0.0884,
      "num_input_tokens_seen": 120725200,
      "step": 55950
    },
    {
      "epoch": 9.12805872756933,
      "grad_norm": 4.713009834289551,
      "learning_rate": 1.1494899259744258e-06,
      "loss": 0.0967,
      "num_input_tokens_seen": 120735696,
      "step": 55955
    },
    {
      "epoch": 9.128874388254486,
      "grad_norm": 3.6822562217712402,
      "learning_rate": 1.147357336809296e-06,
      "loss": 0.1994,
      "num_input_tokens_seen": 120747344,
      "step": 55960
    },
    {
      "epoch": 9.129690048939642,
      "grad_norm": 0.591336190700531,
      "learning_rate": 1.1452266812604056e-06,
      "loss": 0.1115,
      "num_input_tokens_seen": 120758416,
      "step": 55965
    },
    {
      "epoch": 9.130505709624796,
      "grad_norm": 6.977701187133789,
      "learning_rate": 1.1430979595004777e-06,
      "loss": 0.2717,
      "num_input_tokens_seen": 120769552,
      "step": 55970
    },
    {
      "epoch": 9.131321370309951,
      "grad_norm": 0.14338502287864685,
      "learning_rate": 1.1409711717020794e-06,
      "loss": 0.1445,
      "num_input_tokens_seen": 120780656,
      "step": 55975
    },
    {
      "epoch": 9.132137030995105,
      "grad_norm": 0.12120555341243744,
      "learning_rate": 1.1388463180376175e-06,
      "loss": 0.0778,
      "num_input_tokens_seen": 120792048,
      "step": 55980
    },
    {
      "epoch": 9.132952691680261,
      "grad_norm": 0.11915405839681625,
      "learning_rate": 1.1367233986793429e-06,
      "loss": 0.1098,
      "num_input_tokens_seen": 120802224,
      "step": 55985
    },
    {
      "epoch": 9.133768352365417,
      "grad_norm": 0.12654858827590942,
      "learning_rate": 1.1346024137993516e-06,
      "loss": 0.0612,
      "num_input_tokens_seen": 120812720,
      "step": 55990
    },
    {
      "epoch": 9.13458401305057,
      "grad_norm": 10.58918571472168,
      "learning_rate": 1.1324833635695808e-06,
      "loss": 0.157,
      "num_input_tokens_seen": 120824880,
      "step": 55995
    },
    {
      "epoch": 9.135399673735726,
      "grad_norm": 0.07049047201871872,
      "learning_rate": 1.13036624816181e-06,
      "loss": 0.105,
      "num_input_tokens_seen": 120835632,
      "step": 56000
    },
    {
      "epoch": 9.13621533442088,
      "grad_norm": 0.06958387047052383,
      "learning_rate": 1.1282510677476655e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 120847984,
      "step": 56005
    },
    {
      "epoch": 9.137030995106036,
      "grad_norm": 0.36804723739624023,
      "learning_rate": 1.126137822498613e-06,
      "loss": 0.152,
      "num_input_tokens_seen": 120858832,
      "step": 56010
    },
    {
      "epoch": 9.137846655791192,
      "grad_norm": 0.10517661273479462,
      "learning_rate": 1.1240265125859628e-06,
      "loss": 0.0054,
      "num_input_tokens_seen": 120869872,
      "step": 56015
    },
    {
      "epoch": 9.138662316476346,
      "grad_norm": 0.3555206060409546,
      "learning_rate": 1.1219171381808696e-06,
      "loss": 0.1546,
      "num_input_tokens_seen": 120879920,
      "step": 56020
    },
    {
      "epoch": 9.139477977161501,
      "grad_norm": 0.030262192711234093,
      "learning_rate": 1.11980969945433e-06,
      "loss": 0.2091,
      "num_input_tokens_seen": 120891088,
      "step": 56025
    },
    {
      "epoch": 9.140293637846655,
      "grad_norm": 3.5948376655578613,
      "learning_rate": 1.1177041965771823e-06,
      "loss": 0.0912,
      "num_input_tokens_seen": 120901968,
      "step": 56030
    },
    {
      "epoch": 9.141109298531811,
      "grad_norm": 1.0993356704711914,
      "learning_rate": 1.1156006297201093e-06,
      "loss": 0.0855,
      "num_input_tokens_seen": 120912176,
      "step": 56035
    },
    {
      "epoch": 9.141924959216965,
      "grad_norm": 0.12425486743450165,
      "learning_rate": 1.1134989990536387e-06,
      "loss": 0.0052,
      "num_input_tokens_seen": 120923216,
      "step": 56040
    },
    {
      "epoch": 9.14274061990212,
      "grad_norm": 0.30227938294410706,
      "learning_rate": 1.1113993047481369e-06,
      "loss": 0.1335,
      "num_input_tokens_seen": 120934448,
      "step": 56045
    },
    {
      "epoch": 9.143556280587276,
      "grad_norm": 0.09625578671693802,
      "learning_rate": 1.1093015469738177e-06,
      "loss": 0.0106,
      "num_input_tokens_seen": 120944880,
      "step": 56050
    },
    {
      "epoch": 9.14437194127243,
      "grad_norm": 3.6121058464050293,
      "learning_rate": 1.107205725900734e-06,
      "loss": 0.1135,
      "num_input_tokens_seen": 120956304,
      "step": 56055
    },
    {
      "epoch": 9.145187601957586,
      "grad_norm": 5.352141857147217,
      "learning_rate": 1.105111841698786e-06,
      "loss": 0.073,
      "num_input_tokens_seen": 120966320,
      "step": 56060
    },
    {
      "epoch": 9.14600326264274,
      "grad_norm": 2.422102212905884,
      "learning_rate": 1.1030198945377128e-06,
      "loss": 0.0069,
      "num_input_tokens_seen": 120977648,
      "step": 56065
    },
    {
      "epoch": 9.146818923327896,
      "grad_norm": 0.14568936824798584,
      "learning_rate": 1.1009298845871013e-06,
      "loss": 0.015,
      "num_input_tokens_seen": 120987728,
      "step": 56070
    },
    {
      "epoch": 9.147634584013051,
      "grad_norm": 0.10506874322891235,
      "learning_rate": 1.098841812016374e-06,
      "loss": 0.0639,
      "num_input_tokens_seen": 120998640,
      "step": 56075
    },
    {
      "epoch": 9.148450244698205,
      "grad_norm": 0.078071728348732,
      "learning_rate": 1.096755676994804e-06,
      "loss": 0.1179,
      "num_input_tokens_seen": 121008912,
      "step": 56080
    },
    {
      "epoch": 9.149265905383361,
      "grad_norm": 0.15917930006980896,
      "learning_rate": 1.0946714796915032e-06,
      "loss": 0.0067,
      "num_input_tokens_seen": 121019184,
      "step": 56085
    },
    {
      "epoch": 9.150081566068515,
      "grad_norm": 2.8171603679656982,
      "learning_rate": 1.092589220275425e-06,
      "loss": 0.133,
      "num_input_tokens_seen": 121031184,
      "step": 56090
    },
    {
      "epoch": 9.15089722675367,
      "grad_norm": 17.829561233520508,
      "learning_rate": 1.0905088989153712e-06,
      "loss": 0.2337,
      "num_input_tokens_seen": 121042672,
      "step": 56095
    },
    {
      "epoch": 9.151712887438826,
      "grad_norm": 1.4083698987960815,
      "learning_rate": 1.0884305157799785e-06,
      "loss": 0.2532,
      "num_input_tokens_seen": 121053520,
      "step": 56100
    },
    {
      "epoch": 9.15252854812398,
      "grad_norm": 0.061346959322690964,
      "learning_rate": 1.0863540710377373e-06,
      "loss": 0.0061,
      "num_input_tokens_seen": 121064688,
      "step": 56105
    },
    {
      "epoch": 9.153344208809136,
      "grad_norm": 0.05480477958917618,
      "learning_rate": 1.0842795648569688e-06,
      "loss": 0.2609,
      "num_input_tokens_seen": 121075088,
      "step": 56110
    },
    {
      "epoch": 9.15415986949429,
      "grad_norm": 0.09183188527822495,
      "learning_rate": 1.0822069974058464e-06,
      "loss": 0.1282,
      "num_input_tokens_seen": 121085424,
      "step": 56115
    },
    {
      "epoch": 9.154975530179446,
      "grad_norm": 0.22845900058746338,
      "learning_rate": 1.0801363688523858e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 121097072,
      "step": 56120
    },
    {
      "epoch": 9.1557911908646,
      "grad_norm": 0.09074364602565765,
      "learning_rate": 1.0780676793644362e-06,
      "loss": 0.005,
      "num_input_tokens_seen": 121108880,
      "step": 56125
    },
    {
      "epoch": 9.156606851549755,
      "grad_norm": 0.3043750822544098,
      "learning_rate": 1.0760009291097022e-06,
      "loss": 0.0171,
      "num_input_tokens_seen": 121120432,
      "step": 56130
    },
    {
      "epoch": 9.15742251223491,
      "grad_norm": 0.32757917046546936,
      "learning_rate": 1.0739361182557194e-06,
      "loss": 0.3645,
      "num_input_tokens_seen": 121131952,
      "step": 56135
    },
    {
      "epoch": 9.158238172920065,
      "grad_norm": 0.04966404289007187,
      "learning_rate": 1.071873246969876e-06,
      "loss": 0.0075,
      "num_input_tokens_seen": 121140304,
      "step": 56140
    },
    {
      "epoch": 9.15905383360522,
      "grad_norm": 0.04576646909117699,
      "learning_rate": 1.0698123154193967e-06,
      "loss": 0.1015,
      "num_input_tokens_seen": 121150608,
      "step": 56145
    },
    {
      "epoch": 9.159869494290374,
      "grad_norm": 0.050863754004240036,
      "learning_rate": 1.0677533237713533e-06,
      "loss": 0.0795,
      "num_input_tokens_seen": 121161296,
      "step": 56150
    },
    {
      "epoch": 9.16068515497553,
      "grad_norm": 0.09488074481487274,
      "learning_rate": 1.0656962721926539e-06,
      "loss": 0.1087,
      "num_input_tokens_seen": 121172976,
      "step": 56155
    },
    {
      "epoch": 9.161500815660686,
      "grad_norm": 0.039611224085092545,
      "learning_rate": 1.063641160850054e-06,
      "loss": 0.0031,
      "num_input_tokens_seen": 121183024,
      "step": 56160
    },
    {
      "epoch": 9.16231647634584,
      "grad_norm": 3.2668161392211914,
      "learning_rate": 1.0615879899101567e-06,
      "loss": 0.1267,
      "num_input_tokens_seen": 121194032,
      "step": 56165
    },
    {
      "epoch": 9.163132137030995,
      "grad_norm": 0.11332713067531586,
      "learning_rate": 1.0595367595393978e-06,
      "loss": 0.0045,
      "num_input_tokens_seen": 121205296,
      "step": 56170
    },
    {
      "epoch": 9.16394779771615,
      "grad_norm": 0.09239339828491211,
      "learning_rate": 1.0574874699040643e-06,
      "loss": 0.0025,
      "num_input_tokens_seen": 121217168,
      "step": 56175
    },
    {
      "epoch": 9.164763458401305,
      "grad_norm": 4.365166664123535,
      "learning_rate": 1.0554401211702787e-06,
      "loss": 0.4424,
      "num_input_tokens_seen": 121228176,
      "step": 56180
    },
    {
      "epoch": 9.16557911908646,
      "grad_norm": 0.03770161792635918,
      "learning_rate": 1.0533947135040106e-06,
      "loss": 0.0038,
      "num_input_tokens_seen": 121237968,
      "step": 56185
    },
    {
      "epoch": 9.166394779771615,
      "grad_norm": 0.14600355923175812,
      "learning_rate": 1.0513512470710695e-06,
      "loss": 0.01,
      "num_input_tokens_seen": 121248048,
      "step": 56190
    },
    {
      "epoch": 9.16721044045677,
      "grad_norm": 0.13089050352573395,
      "learning_rate": 1.0493097220371117e-06,
      "loss": 0.072,
      "num_input_tokens_seen": 121259088,
      "step": 56195
    },
    {
      "epoch": 9.168026101141924,
      "grad_norm": 0.10141665488481522,
      "learning_rate": 1.0472701385676326e-06,
      "loss": 0.2774,
      "num_input_tokens_seen": 121270416,
      "step": 56200
    },
    {
      "epoch": 9.16884176182708,
      "grad_norm": 0.10410164296627045,
      "learning_rate": 1.045232496827972e-06,
      "loss": 0.097,
      "num_input_tokens_seen": 121281424,
      "step": 56205
    },
    {
      "epoch": 9.169657422512234,
      "grad_norm": 0.17221102118492126,
      "learning_rate": 1.043196796983309e-06,
      "loss": 0.0065,
      "num_input_tokens_seen": 121292496,
      "step": 56210
    },
    {
      "epoch": 9.17047308319739,
      "grad_norm": 0.06364672631025314,
      "learning_rate": 1.0411630391986698e-06,
      "loss": 0.0023,
      "num_input_tokens_seen": 121303280,
      "step": 56215
    },
    {
      "epoch": 9.171288743882545,
      "grad_norm": 0.3799787163734436,
      "learning_rate": 1.03913122363892e-06,
      "loss": 0.0966,
      "num_input_tokens_seen": 121315248,
      "step": 56220
    },
    {
      "epoch": 9.1721044045677,
      "grad_norm": 0.17793439328670502,
      "learning_rate": 1.0371013504687692e-06,
      "loss": 0.0834,
      "num_input_tokens_seen": 121326256,
      "step": 56225
    },
    {
      "epoch": 9.172920065252855,
      "grad_norm": 0.18157732486724854,
      "learning_rate": 1.0350734198527696e-06,
      "loss": 0.2261,
      "num_input_tokens_seen": 121337264,
      "step": 56230
    },
    {
      "epoch": 9.173735725938009,
      "grad_norm": 0.2199692577123642,
      "learning_rate": 1.033047431955317e-06,
      "loss": 0.1282,
      "num_input_tokens_seen": 121347600,
      "step": 56235
    },
    {
      "epoch": 9.174551386623165,
      "grad_norm": 0.1965482532978058,
      "learning_rate": 1.0310233869406437e-06,
      "loss": 0.161,
      "num_input_tokens_seen": 121358800,
      "step": 56240
    },
    {
      "epoch": 9.17536704730832,
      "grad_norm": 0.4902295470237732,
      "learning_rate": 1.0290012849728358e-06,
      "loss": 0.0069,
      "num_input_tokens_seen": 121369712,
      "step": 56245
    },
    {
      "epoch": 9.176182707993474,
      "grad_norm": 0.10036962479352951,
      "learning_rate": 1.0269811262158092e-06,
      "loss": 0.0804,
      "num_input_tokens_seen": 121380784,
      "step": 56250
    },
    {
      "epoch": 9.17699836867863,
      "grad_norm": 0.0895148441195488,
      "learning_rate": 1.024962910833327e-06,
      "loss": 0.0062,
      "num_input_tokens_seen": 121392752,
      "step": 56255
    },
    {
      "epoch": 9.177814029363784,
      "grad_norm": 0.0164335947483778,
      "learning_rate": 1.022946638989003e-06,
      "loss": 0.0784,
      "num_input_tokens_seen": 121402928,
      "step": 56260
    },
    {
      "epoch": 9.17862969004894,
      "grad_norm": 0.08182208240032196,
      "learning_rate": 1.0209323108462816e-06,
      "loss": 0.004,
      "num_input_tokens_seen": 121413008,
      "step": 56265
    },
    {
      "epoch": 9.179445350734095,
      "grad_norm": 0.0456121489405632,
      "learning_rate": 1.018919926568457e-06,
      "loss": 0.0522,
      "num_input_tokens_seen": 121422896,
      "step": 56270
    },
    {
      "epoch": 9.18026101141925,
      "grad_norm": 0.07641737908124924,
      "learning_rate": 1.0169094863186623e-06,
      "loss": 0.2069,
      "num_input_tokens_seen": 121434512,
      "step": 56275
    },
    {
      "epoch": 9.181076672104405,
      "grad_norm": 0.0399506501853466,
      "learning_rate": 1.0149009902598706e-06,
      "loss": 0.2618,
      "num_input_tokens_seen": 121445840,
      "step": 56280
    },
    {
      "epoch": 9.181892332789559,
      "grad_norm": 0.06404338777065277,
      "learning_rate": 1.0128944385549038e-06,
      "loss": 0.0735,
      "num_input_tokens_seen": 121457264,
      "step": 56285
    },
    {
      "epoch": 9.182707993474715,
      "grad_norm": 0.10424448549747467,
      "learning_rate": 1.0108898313664267e-06,
      "loss": 0.0026,
      "num_input_tokens_seen": 121467728,
      "step": 56290
    },
    {
      "epoch": 9.18352365415987,
      "grad_norm": 0.0442383848130703,
      "learning_rate": 1.0088871688569397e-06,
      "loss": 0.2079,
      "num_input_tokens_seen": 121478608,
      "step": 56295
    },
    {
      "epoch": 9.184339314845024,
      "grad_norm": 0.13860155642032623,
      "learning_rate": 1.006886451188785e-06,
      "loss": 0.0058,
      "num_input_tokens_seen": 121489488,
      "step": 56300
    },
    {
      "epoch": 9.18515497553018,
      "grad_norm": 0.09070900827646255,
      "learning_rate": 1.0048876785241578e-06,
      "loss": 0.1313,
      "num_input_tokens_seen": 121500400,
      "step": 56305
    },
    {
      "epoch": 9.185970636215334,
      "grad_norm": 0.12147332727909088,
      "learning_rate": 1.0028908510250846e-06,
      "loss": 0.1195,
      "num_input_tokens_seen": 121511088,
      "step": 56310
    },
    {
      "epoch": 9.18678629690049,
      "grad_norm": 0.374837189912796,
      "learning_rate": 1.000895968853438e-06,
      "loss": 0.118,
      "num_input_tokens_seen": 121523088,
      "step": 56315
    },
    {
      "epoch": 9.187601957585644,
      "grad_norm": 0.039751630276441574,
      "learning_rate": 9.989030321709336e-07,
      "loss": 0.0051,
      "num_input_tokens_seen": 121533808,
      "step": 56320
    },
    {
      "epoch": 9.1884176182708,
      "grad_norm": 19.650569915771484,
      "learning_rate": 9.969120411391308e-07,
      "loss": 0.1255,
      "num_input_tokens_seen": 121544240,
      "step": 56325
    },
    {
      "epoch": 9.189233278955955,
      "grad_norm": 5.189046382904053,
      "learning_rate": 9.949229959194313e-07,
      "loss": 0.2986,
      "num_input_tokens_seen": 121554992,
      "step": 56330
    },
    {
      "epoch": 9.190048939641109,
      "grad_norm": 0.09287931025028229,
      "learning_rate": 9.929358966730696e-07,
      "loss": 0.16,
      "num_input_tokens_seen": 121565424,
      "step": 56335
    },
    {
      "epoch": 9.190864600326265,
      "grad_norm": 0.11300577968358994,
      "learning_rate": 9.909507435611365e-07,
      "loss": 0.1685,
      "num_input_tokens_seen": 121575504,
      "step": 56340
    },
    {
      "epoch": 9.191680261011419,
      "grad_norm": 0.10862802714109421,
      "learning_rate": 9.889675367445589e-07,
      "loss": 0.0042,
      "num_input_tokens_seen": 121586320,
      "step": 56345
    },
    {
      "epoch": 9.192495921696574,
      "grad_norm": 0.10327637195587158,
      "learning_rate": 9.869862763841026e-07,
      "loss": 0.2934,
      "num_input_tokens_seen": 121596976,
      "step": 56350
    },
    {
      "epoch": 9.19331158238173,
      "grad_norm": 0.07766059041023254,
      "learning_rate": 9.85006962640378e-07,
      "loss": 0.0132,
      "num_input_tokens_seen": 121607344,
      "step": 56355
    },
    {
      "epoch": 9.194127243066884,
      "grad_norm": 0.06651419401168823,
      "learning_rate": 9.8302959567384e-07,
      "loss": 0.0667,
      "num_input_tokens_seen": 121617424,
      "step": 56360
    },
    {
      "epoch": 9.19494290375204,
      "grad_norm": 7.505390644073486,
      "learning_rate": 9.810541756447855e-07,
      "loss": 0.0799,
      "num_input_tokens_seen": 121629744,
      "step": 56365
    },
    {
      "epoch": 9.195758564437194,
      "grad_norm": 3.8123583793640137,
      "learning_rate": 9.790807027133446e-07,
      "loss": 0.2384,
      "num_input_tokens_seen": 121641008,
      "step": 56370
    },
    {
      "epoch": 9.19657422512235,
      "grad_norm": 0.07194401323795319,
      "learning_rate": 9.77109177039509e-07,
      "loss": 0.0053,
      "num_input_tokens_seen": 121652528,
      "step": 56375
    },
    {
      "epoch": 9.197389885807505,
      "grad_norm": 0.610551655292511,
      "learning_rate": 9.751395987830924e-07,
      "loss": 0.0187,
      "num_input_tokens_seen": 121663856,
      "step": 56380
    },
    {
      "epoch": 9.198205546492659,
      "grad_norm": 8.918259620666504,
      "learning_rate": 9.731719681037616e-07,
      "loss": 0.1054,
      "num_input_tokens_seen": 121676112,
      "step": 56385
    },
    {
      "epoch": 9.199021207177815,
      "grad_norm": 0.08791600912809372,
      "learning_rate": 9.712062851610222e-07,
      "loss": 0.098,
      "num_input_tokens_seen": 121688400,
      "step": 56390
    },
    {
      "epoch": 9.199836867862969,
      "grad_norm": 4.068906307220459,
      "learning_rate": 9.692425501142217e-07,
      "loss": 0.0957,
      "num_input_tokens_seen": 121698928,
      "step": 56395
    },
    {
      "epoch": 9.200652528548124,
      "grad_norm": 8.034584045410156,
      "learning_rate": 9.672807631225521e-07,
      "loss": 0.2178,
      "num_input_tokens_seen": 121708496,
      "step": 56400
    },
    {
      "epoch": 9.201468189233278,
      "grad_norm": 41.141563415527344,
      "learning_rate": 9.65320924345045e-07,
      "loss": 0.0419,
      "num_input_tokens_seen": 121720240,
      "step": 56405
    },
    {
      "epoch": 9.202283849918434,
      "grad_norm": 0.07082147151231766,
      "learning_rate": 9.633630339405731e-07,
      "loss": 0.099,
      "num_input_tokens_seen": 121732080,
      "step": 56410
    },
    {
      "epoch": 9.20309951060359,
      "grad_norm": 0.12724247574806213,
      "learning_rate": 9.614070920678536e-07,
      "loss": 0.0115,
      "num_input_tokens_seen": 121742096,
      "step": 56415
    },
    {
      "epoch": 9.203915171288743,
      "grad_norm": 0.16074572503566742,
      "learning_rate": 9.59453098885446e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 121752496,
      "step": 56420
    },
    {
      "epoch": 9.2047308319739,
      "grad_norm": 0.1263059824705124,
      "learning_rate": 9.575010545517487e-07,
      "loss": 0.1537,
      "num_input_tokens_seen": 121763024,
      "step": 56425
    },
    {
      "epoch": 9.205546492659053,
      "grad_norm": 0.044610343873500824,
      "learning_rate": 9.5555095922501e-07,
      "loss": 0.1087,
      "num_input_tokens_seen": 121774352,
      "step": 56430
    },
    {
      "epoch": 9.206362153344209,
      "grad_norm": 0.02099367417395115,
      "learning_rate": 9.53602813063309e-07,
      "loss": 0.0054,
      "num_input_tokens_seen": 121785296,
      "step": 56435
    },
    {
      "epoch": 9.207177814029365,
      "grad_norm": 0.11674931645393372,
      "learning_rate": 9.516566162245749e-07,
      "loss": 0.1457,
      "num_input_tokens_seen": 121796752,
      "step": 56440
    },
    {
      "epoch": 9.207993474714518,
      "grad_norm": 13.21556282043457,
      "learning_rate": 9.49712368866576e-07,
      "loss": 0.0238,
      "num_input_tokens_seen": 121807440,
      "step": 56445
    },
    {
      "epoch": 9.208809135399674,
      "grad_norm": 0.03066655993461609,
      "learning_rate": 9.477700711469223e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 121818640,
      "step": 56450
    },
    {
      "epoch": 9.209624796084828,
      "grad_norm": 0.02105654403567314,
      "learning_rate": 9.458297232230684e-07,
      "loss": 0.0798,
      "num_input_tokens_seen": 121828560,
      "step": 56455
    },
    {
      "epoch": 9.210440456769984,
      "grad_norm": 5.688864707946777,
      "learning_rate": 9.438913252523024e-07,
      "loss": 0.0776,
      "num_input_tokens_seen": 121839088,
      "step": 56460
    },
    {
      "epoch": 9.21125611745514,
      "grad_norm": 0.07022027671337128,
      "learning_rate": 9.41954877391768e-07,
      "loss": 0.0042,
      "num_input_tokens_seen": 121849744,
      "step": 56465
    },
    {
      "epoch": 9.212071778140293,
      "grad_norm": 0.03331270068883896,
      "learning_rate": 9.400203797984397e-07,
      "loss": 0.1969,
      "num_input_tokens_seen": 121859696,
      "step": 56470
    },
    {
      "epoch": 9.21288743882545,
      "grad_norm": 0.18326084315776825,
      "learning_rate": 9.380878326291392e-07,
      "loss": 0.003,
      "num_input_tokens_seen": 121869840,
      "step": 56475
    },
    {
      "epoch": 9.213703099510603,
      "grad_norm": 0.12050646543502808,
      "learning_rate": 9.361572360405246e-07,
      "loss": 0.0027,
      "num_input_tokens_seen": 121880432,
      "step": 56480
    },
    {
      "epoch": 9.214518760195759,
      "grad_norm": 0.13081873953342438,
      "learning_rate": 9.342285901891068e-07,
      "loss": 0.0263,
      "num_input_tokens_seen": 121889264,
      "step": 56485
    },
    {
      "epoch": 9.215334420880913,
      "grad_norm": 0.3042246401309967,
      "learning_rate": 9.323018952312273e-07,
      "loss": 0.0051,
      "num_input_tokens_seen": 121900016,
      "step": 56490
    },
    {
      "epoch": 9.216150081566068,
      "grad_norm": 0.041951294988393784,
      "learning_rate": 9.303771513230752e-07,
      "loss": 0.0053,
      "num_input_tokens_seen": 121911312,
      "step": 56495
    },
    {
      "epoch": 9.216965742251224,
      "grad_norm": 0.16849592328071594,
      "learning_rate": 9.284543586206784e-07,
      "loss": 0.0035,
      "num_input_tokens_seen": 121923088,
      "step": 56500
    },
    {
      "epoch": 9.217781402936378,
      "grad_norm": 2.54909348487854,
      "learning_rate": 9.265335172799094e-07,
      "loss": 0.1501,
      "num_input_tokens_seen": 121932720,
      "step": 56505
    },
    {
      "epoch": 9.218597063621534,
      "grad_norm": 0.19697806239128113,
      "learning_rate": 9.246146274564798e-07,
      "loss": 0.0434,
      "num_input_tokens_seen": 121943248,
      "step": 56510
    },
    {
      "epoch": 9.219412724306688,
      "grad_norm": 2.1499359607696533,
      "learning_rate": 9.226976893059458e-07,
      "loss": 0.0835,
      "num_input_tokens_seen": 121953712,
      "step": 56515
    },
    {
      "epoch": 9.220228384991843,
      "grad_norm": 0.06755515933036804,
      "learning_rate": 9.207827029837052e-07,
      "loss": 0.0033,
      "num_input_tokens_seen": 121965456,
      "step": 56520
    },
    {
      "epoch": 9.221044045676999,
      "grad_norm": 0.14053942263126373,
      "learning_rate": 9.188696686449949e-07,
      "loss": 0.0049,
      "num_input_tokens_seen": 121975120,
      "step": 56525
    },
    {
      "epoch": 9.221859706362153,
      "grad_norm": 11.012677192687988,
      "learning_rate": 9.169585864448965e-07,
      "loss": 0.2189,
      "num_input_tokens_seen": 121987344,
      "step": 56530
    },
    {
      "epoch": 9.222675367047309,
      "grad_norm": 0.3366019129753113,
      "learning_rate": 9.150494565383305e-07,
      "loss": 0.005,
      "num_input_tokens_seen": 121998448,
      "step": 56535
    },
    {
      "epoch": 9.223491027732463,
      "grad_norm": 0.04674627259373665,
      "learning_rate": 9.13142279080062e-07,
      "loss": 0.3208,
      "num_input_tokens_seen": 122009104,
      "step": 56540
    },
    {
      "epoch": 9.224306688417618,
      "grad_norm": 20.346275329589844,
      "learning_rate": 9.112370542246978e-07,
      "loss": 0.1621,
      "num_input_tokens_seen": 122019952,
      "step": 56545
    },
    {
      "epoch": 9.225122349102774,
      "grad_norm": 0.092558853328228,
      "learning_rate": 9.093337821266784e-07,
      "loss": 0.0986,
      "num_input_tokens_seen": 122031280,
      "step": 56550
    },
    {
      "epoch": 9.225938009787928,
      "grad_norm": 0.23202751576900482,
      "learning_rate": 9.074324629403025e-07,
      "loss": 0.0039,
      "num_input_tokens_seen": 122041584,
      "step": 56555
    },
    {
      "epoch": 9.226753670473084,
      "grad_norm": 5.594674110412598,
      "learning_rate": 9.055330968196912e-07,
      "loss": 0.0708,
      "num_input_tokens_seen": 122051312,
      "step": 56560
    },
    {
      "epoch": 9.227569331158238,
      "grad_norm": 0.14860931038856506,
      "learning_rate": 9.036356839188243e-07,
      "loss": 0.01,
      "num_input_tokens_seen": 122061552,
      "step": 56565
    },
    {
      "epoch": 9.228384991843393,
      "grad_norm": 0.10265801846981049,
      "learning_rate": 9.017402243915091e-07,
      "loss": 0.1881,
      "num_input_tokens_seen": 122071184,
      "step": 56570
    },
    {
      "epoch": 9.229200652528547,
      "grad_norm": 3.455475330352783,
      "learning_rate": 8.998467183914061e-07,
      "loss": 0.0723,
      "num_input_tokens_seen": 122083344,
      "step": 56575
    },
    {
      "epoch": 9.230016313213703,
      "grad_norm": 0.05952727794647217,
      "learning_rate": 8.97955166072012e-07,
      "loss": 0.0022,
      "num_input_tokens_seen": 122094832,
      "step": 56580
    },
    {
      "epoch": 9.230831973898859,
      "grad_norm": 0.07937181740999222,
      "learning_rate": 8.960655675866653e-07,
      "loss": 0.1006,
      "num_input_tokens_seen": 122106032,
      "step": 56585
    },
    {
      "epoch": 9.231647634584013,
      "grad_norm": 3.776068687438965,
      "learning_rate": 8.941779230885433e-07,
      "loss": 0.1298,
      "num_input_tokens_seen": 122116304,
      "step": 56590
    },
    {
      "epoch": 9.232463295269168,
      "grad_norm": 0.0652937963604927,
      "learning_rate": 8.92292232730671e-07,
      "loss": 0.0735,
      "num_input_tokens_seen": 122128176,
      "step": 56595
    },
    {
      "epoch": 9.233278955954322,
      "grad_norm": 0.027470342814922333,
      "learning_rate": 8.904084966659121e-07,
      "loss": 0.1113,
      "num_input_tokens_seen": 122139696,
      "step": 56600
    },
    {
      "epoch": 9.234094616639478,
      "grad_norm": 0.22010153532028198,
      "learning_rate": 8.885267150469723e-07,
      "loss": 0.2322,
      "num_input_tokens_seen": 122151024,
      "step": 56605
    },
    {
      "epoch": 9.234910277324634,
      "grad_norm": 0.056948062032461166,
      "learning_rate": 8.866468880263961e-07,
      "loss": 0.0115,
      "num_input_tokens_seen": 122161648,
      "step": 56610
    },
    {
      "epoch": 9.235725938009788,
      "grad_norm": 0.05341499298810959,
      "learning_rate": 8.847690157565758e-07,
      "loss": 0.1356,
      "num_input_tokens_seen": 122172656,
      "step": 56615
    },
    {
      "epoch": 9.236541598694943,
      "grad_norm": 0.05375171825289726,
      "learning_rate": 8.828930983897366e-07,
      "loss": 0.1139,
      "num_input_tokens_seen": 122183056,
      "step": 56620
    },
    {
      "epoch": 9.237357259380097,
      "grad_norm": 0.24702052772045135,
      "learning_rate": 8.810191360779513e-07,
      "loss": 0.0217,
      "num_input_tokens_seen": 122194224,
      "step": 56625
    },
    {
      "epoch": 9.238172920065253,
      "grad_norm": 0.04193101450800896,
      "learning_rate": 8.791471289731346e-07,
      "loss": 0.0308,
      "num_input_tokens_seen": 122204784,
      "step": 56630
    },
    {
      "epoch": 9.238988580750409,
      "grad_norm": 0.053464144468307495,
      "learning_rate": 8.7727707722704e-07,
      "loss": 0.0023,
      "num_input_tokens_seen": 122215280,
      "step": 56635
    },
    {
      "epoch": 9.239804241435563,
      "grad_norm": 0.9686161875724792,
      "learning_rate": 8.75408980991263e-07,
      "loss": 0.0977,
      "num_input_tokens_seen": 122226128,
      "step": 56640
    },
    {
      "epoch": 9.240619902120718,
      "grad_norm": 4.109722137451172,
      "learning_rate": 8.735428404172408e-07,
      "loss": 0.1423,
      "num_input_tokens_seen": 122237552,
      "step": 56645
    },
    {
      "epoch": 9.241435562805872,
      "grad_norm": 0.08541485667228699,
      "learning_rate": 8.716786556562495e-07,
      "loss": 0.2074,
      "num_input_tokens_seen": 122248880,
      "step": 56650
    },
    {
      "epoch": 9.242251223491028,
      "grad_norm": 0.3115135133266449,
      "learning_rate": 8.698164268594155e-07,
      "loss": 0.1094,
      "num_input_tokens_seen": 122259760,
      "step": 56655
    },
    {
      "epoch": 9.243066884176184,
      "grad_norm": 5.633404731750488,
      "learning_rate": 8.679561541776959e-07,
      "loss": 0.0973,
      "num_input_tokens_seen": 122271120,
      "step": 56660
    },
    {
      "epoch": 9.243882544861338,
      "grad_norm": 0.1086176410317421,
      "learning_rate": 8.660978377618951e-07,
      "loss": 0.0678,
      "num_input_tokens_seen": 122281424,
      "step": 56665
    },
    {
      "epoch": 9.244698205546493,
      "grad_norm": 0.23381710052490234,
      "learning_rate": 8.64241477762659e-07,
      "loss": 0.0081,
      "num_input_tokens_seen": 122292784,
      "step": 56670
    },
    {
      "epoch": 9.245513866231647,
      "grad_norm": 0.07667340338230133,
      "learning_rate": 8.62387074330473e-07,
      "loss": 0.0459,
      "num_input_tokens_seen": 122302992,
      "step": 56675
    },
    {
      "epoch": 9.246329526916803,
      "grad_norm": 3.798373222351074,
      "learning_rate": 8.605346276156611e-07,
      "loss": 0.0729,
      "num_input_tokens_seen": 122314320,
      "step": 56680
    },
    {
      "epoch": 9.247145187601957,
      "grad_norm": 0.08883180469274521,
      "learning_rate": 8.586841377683951e-07,
      "loss": 0.0029,
      "num_input_tokens_seen": 122324848,
      "step": 56685
    },
    {
      "epoch": 9.247960848287113,
      "grad_norm": 0.037743981927633286,
      "learning_rate": 8.568356049386827e-07,
      "loss": 0.0028,
      "num_input_tokens_seen": 122335760,
      "step": 56690
    },
    {
      "epoch": 9.248776508972268,
      "grad_norm": 0.11212898045778275,
      "learning_rate": 8.549890292763819e-07,
      "loss": 0.0862,
      "num_input_tokens_seen": 122346416,
      "step": 56695
    },
    {
      "epoch": 9.249592169657422,
      "grad_norm": 0.07642631232738495,
      "learning_rate": 8.531444109311781e-07,
      "loss": 0.094,
      "num_input_tokens_seen": 122358608,
      "step": 56700
    },
    {
      "epoch": 9.250407830342578,
      "grad_norm": 0.07463725656270981,
      "learning_rate": 8.513017500526105e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 122369264,
      "step": 56705
    },
    {
      "epoch": 9.251223491027732,
      "grad_norm": 0.4207376539707184,
      "learning_rate": 8.49461046790051e-07,
      "loss": 0.3118,
      "num_input_tokens_seen": 122380528,
      "step": 56710
    },
    {
      "epoch": 9.252039151712887,
      "grad_norm": 0.08353845775127411,
      "learning_rate": 8.476223012927193e-07,
      "loss": 0.0027,
      "num_input_tokens_seen": 122391856,
      "step": 56715
    },
    {
      "epoch": 9.252854812398043,
      "grad_norm": 0.053139425814151764,
      "learning_rate": 8.457855137096682e-07,
      "loss": 0.0053,
      "num_input_tokens_seen": 122402672,
      "step": 56720
    },
    {
      "epoch": 9.253670473083197,
      "grad_norm": 0.13485592603683472,
      "learning_rate": 8.439506841898037e-07,
      "loss": 0.0875,
      "num_input_tokens_seen": 122412368,
      "step": 56725
    },
    {
      "epoch": 9.254486133768353,
      "grad_norm": 24.39397430419922,
      "learning_rate": 8.421178128818624e-07,
      "loss": 0.0173,
      "num_input_tokens_seen": 122423248,
      "step": 56730
    },
    {
      "epoch": 9.255301794453507,
      "grad_norm": 0.08826201409101486,
      "learning_rate": 8.402868999344283e-07,
      "loss": 0.0068,
      "num_input_tokens_seen": 122433040,
      "step": 56735
    },
    {
      "epoch": 9.256117455138662,
      "grad_norm": 0.04894102364778519,
      "learning_rate": 8.384579454959185e-07,
      "loss": 0.0032,
      "num_input_tokens_seen": 122442928,
      "step": 56740
    },
    {
      "epoch": 9.256933115823816,
      "grad_norm": 0.3541920483112335,
      "learning_rate": 8.366309497146063e-07,
      "loss": 0.0129,
      "num_input_tokens_seen": 122453488,
      "step": 56745
    },
    {
      "epoch": 9.257748776508972,
      "grad_norm": 0.07416754215955734,
      "learning_rate": 8.348059127385926e-07,
      "loss": 0.0055,
      "num_input_tokens_seen": 122463632,
      "step": 56750
    },
    {
      "epoch": 9.258564437194128,
      "grad_norm": 0.09072282910346985,
      "learning_rate": 8.329828347158231e-07,
      "loss": 0.0029,
      "num_input_tokens_seen": 122474736,
      "step": 56755
    },
    {
      "epoch": 9.259380097879282,
      "grad_norm": 0.07462501525878906,
      "learning_rate": 8.311617157940904e-07,
      "loss": 0.0051,
      "num_input_tokens_seen": 122485776,
      "step": 56760
    },
    {
      "epoch": 9.260195758564437,
      "grad_norm": 0.04590676724910736,
      "learning_rate": 8.293425561210183e-07,
      "loss": 0.0034,
      "num_input_tokens_seen": 122496176,
      "step": 56765
    },
    {
      "epoch": 9.261011419249591,
      "grad_norm": 0.03457889333367348,
      "learning_rate": 8.275253558440776e-07,
      "loss": 0.0037,
      "num_input_tokens_seen": 122507760,
      "step": 56770
    },
    {
      "epoch": 9.261827079934747,
      "grad_norm": 19.248991012573242,
      "learning_rate": 8.257101151105839e-07,
      "loss": 0.0902,
      "num_input_tokens_seen": 122518768,
      "step": 56775
    },
    {
      "epoch": 9.262642740619903,
      "grad_norm": 11.731621742248535,
      "learning_rate": 8.23896834067689e-07,
      "loss": 0.0843,
      "num_input_tokens_seen": 122531216,
      "step": 56780
    },
    {
      "epoch": 9.263458401305057,
      "grad_norm": 17.964866638183594,
      "learning_rate": 8.220855128623805e-07,
      "loss": 0.0988,
      "num_input_tokens_seen": 122541552,
      "step": 56785
    },
    {
      "epoch": 9.264274061990212,
      "grad_norm": 0.7958881258964539,
      "learning_rate": 8.202761516415025e-07,
      "loss": 0.0886,
      "num_input_tokens_seen": 122553712,
      "step": 56790
    },
    {
      "epoch": 9.265089722675366,
      "grad_norm": 0.15371784567832947,
      "learning_rate": 8.184687505517236e-07,
      "loss": 0.1292,
      "num_input_tokens_seen": 122564976,
      "step": 56795
    },
    {
      "epoch": 9.265905383360522,
      "grad_norm": 0.05780799314379692,
      "learning_rate": 8.166633097395626e-07,
      "loss": 0.1017,
      "num_input_tokens_seen": 122576272,
      "step": 56800
    },
    {
      "epoch": 9.266721044045678,
      "grad_norm": 2.097024917602539,
      "learning_rate": 8.148598293513804e-07,
      "loss": 0.097,
      "num_input_tokens_seen": 122586576,
      "step": 56805
    },
    {
      "epoch": 9.267536704730832,
      "grad_norm": 7.795865058898926,
      "learning_rate": 8.130583095333739e-07,
      "loss": 0.0138,
      "num_input_tokens_seen": 122598000,
      "step": 56810
    },
    {
      "epoch": 9.268352365415987,
      "grad_norm": 0.16377630829811096,
      "learning_rate": 8.112587504315844e-07,
      "loss": 0.1037,
      "num_input_tokens_seen": 122606896,
      "step": 56815
    },
    {
      "epoch": 9.269168026101141,
      "grad_norm": 4.764657974243164,
      "learning_rate": 8.094611521918927e-07,
      "loss": 0.0071,
      "num_input_tokens_seen": 122617200,
      "step": 56820
    },
    {
      "epoch": 9.269983686786297,
      "grad_norm": 0.371650367975235,
      "learning_rate": 8.076655149600237e-07,
      "loss": 0.2225,
      "num_input_tokens_seen": 122626512,
      "step": 56825
    },
    {
      "epoch": 9.270799347471453,
      "grad_norm": 0.23115122318267822,
      "learning_rate": 8.058718388815362e-07,
      "loss": 0.1342,
      "num_input_tokens_seen": 122636848,
      "step": 56830
    },
    {
      "epoch": 9.271615008156607,
      "grad_norm": 2.2169458866119385,
      "learning_rate": 8.040801241018386e-07,
      "loss": 0.099,
      "num_input_tokens_seen": 122646928,
      "step": 56835
    },
    {
      "epoch": 9.272430668841762,
      "grad_norm": 1.1989107131958008,
      "learning_rate": 8.022903707661761e-07,
      "loss": 0.1356,
      "num_input_tokens_seen": 122656752,
      "step": 56840
    },
    {
      "epoch": 9.273246329526916,
      "grad_norm": 0.11020349711179733,
      "learning_rate": 8.005025790196325e-07,
      "loss": 0.0815,
      "num_input_tokens_seen": 122668656,
      "step": 56845
    },
    {
      "epoch": 9.274061990212072,
      "grad_norm": 0.1726074516773224,
      "learning_rate": 7.987167490071362e-07,
      "loss": 0.0086,
      "num_input_tokens_seen": 122679728,
      "step": 56850
    },
    {
      "epoch": 9.274877650897226,
      "grad_norm": 8.0415678024292,
      "learning_rate": 7.969328808734577e-07,
      "loss": 0.1705,
      "num_input_tokens_seen": 122691536,
      "step": 56855
    },
    {
      "epoch": 9.275693311582382,
      "grad_norm": 0.1015799269080162,
      "learning_rate": 7.951509747632063e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 122701488,
      "step": 56860
    },
    {
      "epoch": 9.276508972267537,
      "grad_norm": 0.15888768434524536,
      "learning_rate": 7.933710308208275e-07,
      "loss": 0.0042,
      "num_input_tokens_seen": 122712784,
      "step": 56865
    },
    {
      "epoch": 9.277324632952691,
      "grad_norm": 0.11911559104919434,
      "learning_rate": 7.915930491906198e-07,
      "loss": 0.1025,
      "num_input_tokens_seen": 122723184,
      "step": 56870
    },
    {
      "epoch": 9.278140293637847,
      "grad_norm": 0.11030533164739609,
      "learning_rate": 7.898170300167096e-07,
      "loss": 0.111,
      "num_input_tokens_seen": 122733808,
      "step": 56875
    },
    {
      "epoch": 9.278955954323001,
      "grad_norm": 0.9021227955818176,
      "learning_rate": 7.880429734430706e-07,
      "loss": 0.0076,
      "num_input_tokens_seen": 122745552,
      "step": 56880
    },
    {
      "epoch": 9.279771615008157,
      "grad_norm": 0.10199274867773056,
      "learning_rate": 7.862708796135182e-07,
      "loss": 0.0093,
      "num_input_tokens_seen": 122755632,
      "step": 56885
    },
    {
      "epoch": 9.280587275693312,
      "grad_norm": 0.09611979871988297,
      "learning_rate": 7.845007486717099e-07,
      "loss": 0.0684,
      "num_input_tokens_seen": 122767696,
      "step": 56890
    },
    {
      "epoch": 9.281402936378466,
      "grad_norm": 0.17125804722309113,
      "learning_rate": 7.827325807611391e-07,
      "loss": 0.1048,
      "num_input_tokens_seen": 122779376,
      "step": 56895
    },
    {
      "epoch": 9.282218597063622,
      "grad_norm": 0.1180923730134964,
      "learning_rate": 7.80966376025144e-07,
      "loss": 0.0117,
      "num_input_tokens_seen": 122790832,
      "step": 56900
    },
    {
      "epoch": 9.283034257748776,
      "grad_norm": 0.10925772786140442,
      "learning_rate": 7.792021346068989e-07,
      "loss": 0.1719,
      "num_input_tokens_seen": 122801456,
      "step": 56905
    },
    {
      "epoch": 9.283849918433932,
      "grad_norm": 0.050430506467819214,
      "learning_rate": 7.774398566494201e-07,
      "loss": 0.0705,
      "num_input_tokens_seen": 122813520,
      "step": 56910
    },
    {
      "epoch": 9.284665579119087,
      "grad_norm": 0.15304403007030487,
      "learning_rate": 7.756795422955737e-07,
      "loss": 0.0806,
      "num_input_tokens_seen": 122823344,
      "step": 56915
    },
    {
      "epoch": 9.285481239804241,
      "grad_norm": 0.0694902166724205,
      "learning_rate": 7.739211916880595e-07,
      "loss": 0.2018,
      "num_input_tokens_seen": 122834128,
      "step": 56920
    },
    {
      "epoch": 9.286296900489397,
      "grad_norm": 0.06205080822110176,
      "learning_rate": 7.721648049694108e-07,
      "loss": 0.0031,
      "num_input_tokens_seen": 122845456,
      "step": 56925
    },
    {
      "epoch": 9.28711256117455,
      "grad_norm": 0.03544263914227486,
      "learning_rate": 7.704103822820164e-07,
      "loss": 0.0061,
      "num_input_tokens_seen": 122856272,
      "step": 56930
    },
    {
      "epoch": 9.287928221859707,
      "grad_norm": 0.021660730242729187,
      "learning_rate": 7.686579237680957e-07,
      "loss": 0.1363,
      "num_input_tokens_seen": 122866544,
      "step": 56935
    },
    {
      "epoch": 9.28874388254486,
      "grad_norm": 8.081960678100586,
      "learning_rate": 7.669074295697132e-07,
      "loss": 0.1119,
      "num_input_tokens_seen": 122876432,
      "step": 56940
    },
    {
      "epoch": 9.289559543230016,
      "grad_norm": 0.17294886708259583,
      "learning_rate": 7.651588998287717e-07,
      "loss": 0.0653,
      "num_input_tokens_seen": 122888432,
      "step": 56945
    },
    {
      "epoch": 9.290375203915172,
      "grad_norm": 0.13838863372802734,
      "learning_rate": 7.634123346870165e-07,
      "loss": 0.1269,
      "num_input_tokens_seen": 122900016,
      "step": 56950
    },
    {
      "epoch": 9.291190864600326,
      "grad_norm": 0.11037831753492355,
      "learning_rate": 7.616677342860312e-07,
      "loss": 0.0072,
      "num_input_tokens_seen": 122910096,
      "step": 56955
    },
    {
      "epoch": 9.292006525285482,
      "grad_norm": 0.08414048701524734,
      "learning_rate": 7.599250987672446e-07,
      "loss": 0.0041,
      "num_input_tokens_seen": 122920336,
      "step": 56960
    },
    {
      "epoch": 9.292822185970635,
      "grad_norm": 0.7040512561798096,
      "learning_rate": 7.581844282719213e-07,
      "loss": 0.2497,
      "num_input_tokens_seen": 122931632,
      "step": 56965
    },
    {
      "epoch": 9.293637846655791,
      "grad_norm": 0.02372356317937374,
      "learning_rate": 7.564457229411709e-07,
      "loss": 0.1279,
      "num_input_tokens_seen": 122943728,
      "step": 56970
    },
    {
      "epoch": 9.294453507340947,
      "grad_norm": 0.2222600281238556,
      "learning_rate": 7.547089829159415e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 122953264,
      "step": 56975
    },
    {
      "epoch": 9.2952691680261,
      "grad_norm": 0.4200578033924103,
      "learning_rate": 7.529742083370206e-07,
      "loss": 0.0649,
      "num_input_tokens_seen": 122965296,
      "step": 56980
    },
    {
      "epoch": 9.296084828711257,
      "grad_norm": 7.137514591217041,
      "learning_rate": 7.512413993450373e-07,
      "loss": 0.2838,
      "num_input_tokens_seen": 122974384,
      "step": 56985
    },
    {
      "epoch": 9.29690048939641,
      "grad_norm": 2.8665337562561035,
      "learning_rate": 7.495105560804627e-07,
      "loss": 0.0794,
      "num_input_tokens_seen": 122985296,
      "step": 56990
    },
    {
      "epoch": 9.297716150081566,
      "grad_norm": 0.11218998581171036,
      "learning_rate": 7.477816786836122e-07,
      "loss": 0.004,
      "num_input_tokens_seen": 122996272,
      "step": 56995
    },
    {
      "epoch": 9.298531810766722,
      "grad_norm": 0.11332866549491882,
      "learning_rate": 7.460547672946294e-07,
      "loss": 0.0908,
      "num_input_tokens_seen": 123007440,
      "step": 57000
    },
    {
      "epoch": 9.299347471451876,
      "grad_norm": 3.5079095363616943,
      "learning_rate": 7.443298220535106e-07,
      "loss": 0.3637,
      "num_input_tokens_seen": 123018096,
      "step": 57005
    },
    {
      "epoch": 9.300163132137031,
      "grad_norm": 0.09769842028617859,
      "learning_rate": 7.426068431000882e-07,
      "loss": 0.2011,
      "num_input_tokens_seen": 123029072,
      "step": 57010
    },
    {
      "epoch": 9.300978792822185,
      "grad_norm": 1.6165767908096313,
      "learning_rate": 7.408858305740368e-07,
      "loss": 0.1306,
      "num_input_tokens_seen": 123040816,
      "step": 57015
    },
    {
      "epoch": 9.301794453507341,
      "grad_norm": 0.28688180446624756,
      "learning_rate": 7.391667846148697e-07,
      "loss": 0.0069,
      "num_input_tokens_seen": 123051376,
      "step": 57020
    },
    {
      "epoch": 9.302610114192497,
      "grad_norm": 3.106509208679199,
      "learning_rate": 7.374497053619423e-07,
      "loss": 0.1187,
      "num_input_tokens_seen": 123061232,
      "step": 57025
    },
    {
      "epoch": 9.30342577487765,
      "grad_norm": 0.052906475961208344,
      "learning_rate": 7.357345929544485e-07,
      "loss": 0.2703,
      "num_input_tokens_seen": 123071632,
      "step": 57030
    },
    {
      "epoch": 9.304241435562806,
      "grad_norm": 2.5314533710479736,
      "learning_rate": 7.340214475314244e-07,
      "loss": 0.0949,
      "num_input_tokens_seen": 123083600,
      "step": 57035
    },
    {
      "epoch": 9.30505709624796,
      "grad_norm": 0.0716528445482254,
      "learning_rate": 7.323102692317452e-07,
      "loss": 0.1539,
      "num_input_tokens_seen": 123093392,
      "step": 57040
    },
    {
      "epoch": 9.305872756933116,
      "grad_norm": 0.08974530547857285,
      "learning_rate": 7.306010581941275e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 123104784,
      "step": 57045
    },
    {
      "epoch": 9.30668841761827,
      "grad_norm": 0.23656074702739716,
      "learning_rate": 7.288938145571328e-07,
      "loss": 0.207,
      "num_input_tokens_seen": 123115344,
      "step": 57050
    },
    {
      "epoch": 9.307504078303426,
      "grad_norm": 0.40098971128463745,
      "learning_rate": 7.271885384591503e-07,
      "loss": 0.0064,
      "num_input_tokens_seen": 123125520,
      "step": 57055
    },
    {
      "epoch": 9.308319738988581,
      "grad_norm": 0.11197123676538467,
      "learning_rate": 7.25485230038428e-07,
      "loss": 0.0041,
      "num_input_tokens_seen": 123136016,
      "step": 57060
    },
    {
      "epoch": 9.309135399673735,
      "grad_norm": 0.10504305362701416,
      "learning_rate": 7.237838894330412e-07,
      "loss": 0.1315,
      "num_input_tokens_seen": 123147536,
      "step": 57065
    },
    {
      "epoch": 9.309951060358891,
      "grad_norm": 0.10544651001691818,
      "learning_rate": 7.220845167809076e-07,
      "loss": 0.0087,
      "num_input_tokens_seen": 123157904,
      "step": 57070
    },
    {
      "epoch": 9.310766721044045,
      "grad_norm": 0.08824353665113449,
      "learning_rate": 7.203871122197891e-07,
      "loss": 0.006,
      "num_input_tokens_seen": 123168720,
      "step": 57075
    },
    {
      "epoch": 9.3115823817292,
      "grad_norm": 0.19445312023162842,
      "learning_rate": 7.186916758872841e-07,
      "loss": 0.0123,
      "num_input_tokens_seen": 123180144,
      "step": 57080
    },
    {
      "epoch": 9.312398042414356,
      "grad_norm": 0.1056525856256485,
      "learning_rate": 7.169982079208326e-07,
      "loss": 0.0042,
      "num_input_tokens_seen": 123190896,
      "step": 57085
    },
    {
      "epoch": 9.31321370309951,
      "grad_norm": 0.20880796015262604,
      "learning_rate": 7.153067084577192e-07,
      "loss": 0.0068,
      "num_input_tokens_seen": 123201424,
      "step": 57090
    },
    {
      "epoch": 9.314029363784666,
      "grad_norm": 2.59110689163208,
      "learning_rate": 7.13617177635062e-07,
      "loss": 0.1177,
      "num_input_tokens_seen": 123211536,
      "step": 57095
    },
    {
      "epoch": 9.31484502446982,
      "grad_norm": 0.04983392357826233,
      "learning_rate": 7.119296155898236e-07,
      "loss": 0.0978,
      "num_input_tokens_seen": 123222896,
      "step": 57100
    },
    {
      "epoch": 9.315660685154976,
      "grad_norm": 3.8871772289276123,
      "learning_rate": 7.102440224588086e-07,
      "loss": 0.2656,
      "num_input_tokens_seen": 123235280,
      "step": 57105
    },
    {
      "epoch": 9.31647634584013,
      "grad_norm": 0.08402803540229797,
      "learning_rate": 7.085603983786576e-07,
      "loss": 0.1197,
      "num_input_tokens_seen": 123246416,
      "step": 57110
    },
    {
      "epoch": 9.317292006525285,
      "grad_norm": 0.32142174243927,
      "learning_rate": 7.068787434858532e-07,
      "loss": 0.0052,
      "num_input_tokens_seen": 123257200,
      "step": 57115
    },
    {
      "epoch": 9.318107667210441,
      "grad_norm": 0.09484074264764786,
      "learning_rate": 7.051990579167195e-07,
      "loss": 0.2763,
      "num_input_tokens_seen": 123268816,
      "step": 57120
    },
    {
      "epoch": 9.318923327895595,
      "grad_norm": 0.14092539250850677,
      "learning_rate": 7.035213418074227e-07,
      "loss": 0.1005,
      "num_input_tokens_seen": 123279408,
      "step": 57125
    },
    {
      "epoch": 9.31973898858075,
      "grad_norm": 0.3421323895454407,
      "learning_rate": 7.018455952939651e-07,
      "loss": 0.167,
      "num_input_tokens_seen": 123289232,
      "step": 57130
    },
    {
      "epoch": 9.320554649265905,
      "grad_norm": 0.09815683960914612,
      "learning_rate": 7.001718185121908e-07,
      "loss": 0.2351,
      "num_input_tokens_seen": 123299984,
      "step": 57135
    },
    {
      "epoch": 9.32137030995106,
      "grad_norm": 0.09825440496206284,
      "learning_rate": 6.98500011597783e-07,
      "loss": 0.017,
      "num_input_tokens_seen": 123310768,
      "step": 57140
    },
    {
      "epoch": 9.322185970636216,
      "grad_norm": 0.09648661315441132,
      "learning_rate": 6.96830174686272e-07,
      "loss": 0.003,
      "num_input_tokens_seen": 123321232,
      "step": 57145
    },
    {
      "epoch": 9.32300163132137,
      "grad_norm": 0.06510291993618011,
      "learning_rate": 6.951623079130192e-07,
      "loss": 0.0617,
      "num_input_tokens_seen": 123332432,
      "step": 57150
    },
    {
      "epoch": 9.323817292006526,
      "grad_norm": 0.10147050023078918,
      "learning_rate": 6.934964114132303e-07,
      "loss": 0.0987,
      "num_input_tokens_seen": 123344240,
      "step": 57155
    },
    {
      "epoch": 9.32463295269168,
      "grad_norm": 13.28692626953125,
      "learning_rate": 6.918324853219527e-07,
      "loss": 0.2151,
      "num_input_tokens_seen": 123354672,
      "step": 57160
    },
    {
      "epoch": 9.325448613376835,
      "grad_norm": 0.053556740283966064,
      "learning_rate": 6.901705297740729e-07,
      "loss": 0.2067,
      "num_input_tokens_seen": 123365168,
      "step": 57165
    },
    {
      "epoch": 9.326264274061991,
      "grad_norm": 0.11643710732460022,
      "learning_rate": 6.885105449043138e-07,
      "loss": 0.0099,
      "num_input_tokens_seen": 123375696,
      "step": 57170
    },
    {
      "epoch": 9.327079934747145,
      "grad_norm": 0.0685681700706482,
      "learning_rate": 6.868525308472484e-07,
      "loss": 0.2615,
      "num_input_tokens_seen": 123387344,
      "step": 57175
    },
    {
      "epoch": 9.3278955954323,
      "grad_norm": 0.09857043623924255,
      "learning_rate": 6.851964877372802e-07,
      "loss": 0.006,
      "num_input_tokens_seen": 123398992,
      "step": 57180
    },
    {
      "epoch": 9.328711256117455,
      "grad_norm": 2.408092737197876,
      "learning_rate": 6.835424157086573e-07,
      "loss": 0.2,
      "num_input_tokens_seen": 123409168,
      "step": 57185
    },
    {
      "epoch": 9.32952691680261,
      "grad_norm": 0.1575390100479126,
      "learning_rate": 6.818903148954642e-07,
      "loss": 0.0059,
      "num_input_tokens_seen": 123421488,
      "step": 57190
    },
    {
      "epoch": 9.330342577487766,
      "grad_norm": 0.11237485706806183,
      "learning_rate": 6.802401854316298e-07,
      "loss": 0.0151,
      "num_input_tokens_seen": 123432816,
      "step": 57195
    },
    {
      "epoch": 9.33115823817292,
      "grad_norm": 0.7823300957679749,
      "learning_rate": 6.78592027450925e-07,
      "loss": 0.015,
      "num_input_tokens_seen": 123444048,
      "step": 57200
    },
    {
      "epoch": 9.331973898858076,
      "grad_norm": 0.07418625801801682,
      "learning_rate": 6.769458410869595e-07,
      "loss": 0.1235,
      "num_input_tokens_seen": 123454320,
      "step": 57205
    },
    {
      "epoch": 9.33278955954323,
      "grad_norm": 0.06946953386068344,
      "learning_rate": 6.753016264731738e-07,
      "loss": 0.003,
      "num_input_tokens_seen": 123465424,
      "step": 57210
    },
    {
      "epoch": 9.333605220228385,
      "grad_norm": 0.12592476606369019,
      "learning_rate": 6.736593837428639e-07,
      "loss": 0.0917,
      "num_input_tokens_seen": 123475856,
      "step": 57215
    },
    {
      "epoch": 9.33442088091354,
      "grad_norm": 0.23111547529697418,
      "learning_rate": 6.720191130291514e-07,
      "loss": 0.4095,
      "num_input_tokens_seen": 123484816,
      "step": 57220
    },
    {
      "epoch": 9.335236541598695,
      "grad_norm": 0.016037456691265106,
      "learning_rate": 6.703808144650076e-07,
      "loss": 0.0044,
      "num_input_tokens_seen": 123495920,
      "step": 57225
    },
    {
      "epoch": 9.33605220228385,
      "grad_norm": 0.08093839138746262,
      "learning_rate": 6.687444881832455e-07,
      "loss": 0.2162,
      "num_input_tokens_seen": 123507344,
      "step": 57230
    },
    {
      "epoch": 9.336867862969005,
      "grad_norm": 0.8716356158256531,
      "learning_rate": 6.67110134316512e-07,
      "loss": 0.0896,
      "num_input_tokens_seen": 123517808,
      "step": 57235
    },
    {
      "epoch": 9.33768352365416,
      "grad_norm": 0.0805845558643341,
      "learning_rate": 6.654777529972928e-07,
      "loss": 0.2977,
      "num_input_tokens_seen": 123529456,
      "step": 57240
    },
    {
      "epoch": 9.338499184339314,
      "grad_norm": 9.077909469604492,
      "learning_rate": 6.638473443579179e-07,
      "loss": 0.1592,
      "num_input_tokens_seen": 123538960,
      "step": 57245
    },
    {
      "epoch": 9.33931484502447,
      "grad_norm": 3.288137912750244,
      "learning_rate": 6.622189085305597e-07,
      "loss": 0.2394,
      "num_input_tokens_seen": 123549200,
      "step": 57250
    },
    {
      "epoch": 9.340130505709626,
      "grad_norm": 0.2037453055381775,
      "learning_rate": 6.605924456472262e-07,
      "loss": 0.0729,
      "num_input_tokens_seen": 123560176,
      "step": 57255
    },
    {
      "epoch": 9.34094616639478,
      "grad_norm": 0.06806585192680359,
      "learning_rate": 6.589679558397648e-07,
      "loss": 0.1231,
      "num_input_tokens_seen": 123571280,
      "step": 57260
    },
    {
      "epoch": 9.341761827079935,
      "grad_norm": 0.15784721076488495,
      "learning_rate": 6.573454392398648e-07,
      "loss": 0.1468,
      "num_input_tokens_seen": 123581296,
      "step": 57265
    },
    {
      "epoch": 9.34257748776509,
      "grad_norm": 0.09280826151371002,
      "learning_rate": 6.557248959790596e-07,
      "loss": 0.1565,
      "num_input_tokens_seen": 123592016,
      "step": 57270
    },
    {
      "epoch": 9.343393148450245,
      "grad_norm": 0.050861794501543045,
      "learning_rate": 6.541063261887137e-07,
      "loss": 0.0804,
      "num_input_tokens_seen": 123602992,
      "step": 57275
    },
    {
      "epoch": 9.3442088091354,
      "grad_norm": 0.4141773283481598,
      "learning_rate": 6.524897300000388e-07,
      "loss": 0.0277,
      "num_input_tokens_seen": 123614384,
      "step": 57280
    },
    {
      "epoch": 9.345024469820554,
      "grad_norm": 14.976688385009766,
      "learning_rate": 6.508751075440856e-07,
      "loss": 0.0174,
      "num_input_tokens_seen": 123625744,
      "step": 57285
    },
    {
      "epoch": 9.34584013050571,
      "grad_norm": 3.195305347442627,
      "learning_rate": 6.49262458951741e-07,
      "loss": 0.2904,
      "num_input_tokens_seen": 123636816,
      "step": 57290
    },
    {
      "epoch": 9.346655791190864,
      "grad_norm": 3.126220464706421,
      "learning_rate": 6.476517843537395e-07,
      "loss": 0.1124,
      "num_input_tokens_seen": 123648656,
      "step": 57295
    },
    {
      "epoch": 9.34747145187602,
      "grad_norm": 0.684742271900177,
      "learning_rate": 6.46043083880643e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 123658672,
      "step": 57300
    },
    {
      "epoch": 9.348287112561174,
      "grad_norm": 1.4990553855895996,
      "learning_rate": 6.44436357662867e-07,
      "loss": 0.1757,
      "num_input_tokens_seen": 123669968,
      "step": 57305
    },
    {
      "epoch": 9.34910277324633,
      "grad_norm": 12.546418190002441,
      "learning_rate": 6.428316058306571e-07,
      "loss": 0.0582,
      "num_input_tokens_seen": 123680880,
      "step": 57310
    },
    {
      "epoch": 9.349918433931485,
      "grad_norm": 3.4454562664031982,
      "learning_rate": 6.412288285141066e-07,
      "loss": 0.1173,
      "num_input_tokens_seen": 123691536,
      "step": 57315
    },
    {
      "epoch": 9.350734094616639,
      "grad_norm": 0.20921696722507477,
      "learning_rate": 6.396280258431391e-07,
      "loss": 0.0072,
      "num_input_tokens_seen": 123701904,
      "step": 57320
    },
    {
      "epoch": 9.351549755301795,
      "grad_norm": 0.11067357659339905,
      "learning_rate": 6.38029197947529e-07,
      "loss": 0.0066,
      "num_input_tokens_seen": 123712528,
      "step": 57325
    },
    {
      "epoch": 9.352365415986949,
      "grad_norm": 0.15110088884830475,
      "learning_rate": 6.364323449568804e-07,
      "loss": 0.0124,
      "num_input_tokens_seen": 123723184,
      "step": 57330
    },
    {
      "epoch": 9.353181076672104,
      "grad_norm": 0.04171011224389076,
      "learning_rate": 6.348374670006485e-07,
      "loss": 0.0042,
      "num_input_tokens_seen": 123733968,
      "step": 57335
    },
    {
      "epoch": 9.35399673735726,
      "grad_norm": 0.12456182390451431,
      "learning_rate": 6.332445642081214e-07,
      "loss": 0.0531,
      "num_input_tokens_seen": 123743120,
      "step": 57340
    },
    {
      "epoch": 9.354812398042414,
      "grad_norm": 0.09891777485609055,
      "learning_rate": 6.316536367084236e-07,
      "loss": 0.0162,
      "num_input_tokens_seen": 123754992,
      "step": 57345
    },
    {
      "epoch": 9.35562805872757,
      "grad_norm": 0.024394137784838676,
      "learning_rate": 6.300646846305241e-07,
      "loss": 0.0053,
      "num_input_tokens_seen": 123765488,
      "step": 57350
    },
    {
      "epoch": 9.356443719412724,
      "grad_norm": 4.904883861541748,
      "learning_rate": 6.28477708103234e-07,
      "loss": 0.1155,
      "num_input_tokens_seen": 123776944,
      "step": 57355
    },
    {
      "epoch": 9.35725938009788,
      "grad_norm": 2.6884372234344482,
      "learning_rate": 6.268927072552028e-07,
      "loss": 0.0686,
      "num_input_tokens_seen": 123789168,
      "step": 57360
    },
    {
      "epoch": 9.358075040783035,
      "grad_norm": 0.09686939418315887,
      "learning_rate": 6.253096822149113e-07,
      "loss": 0.2114,
      "num_input_tokens_seen": 123799536,
      "step": 57365
    },
    {
      "epoch": 9.358890701468189,
      "grad_norm": 0.16365379095077515,
      "learning_rate": 6.237286331106984e-07,
      "loss": 0.1283,
      "num_input_tokens_seen": 123811536,
      "step": 57370
    },
    {
      "epoch": 9.359706362153345,
      "grad_norm": 0.18582594394683838,
      "learning_rate": 6.221495600707227e-07,
      "loss": 0.0113,
      "num_input_tokens_seen": 123820848,
      "step": 57375
    },
    {
      "epoch": 9.360522022838499,
      "grad_norm": 0.1327890008687973,
      "learning_rate": 6.205724632229987e-07,
      "loss": 0.0052,
      "num_input_tokens_seen": 123832112,
      "step": 57380
    },
    {
      "epoch": 9.361337683523654,
      "grad_norm": 0.05526406317949295,
      "learning_rate": 6.18997342695371e-07,
      "loss": 0.1159,
      "num_input_tokens_seen": 123842544,
      "step": 57385
    },
    {
      "epoch": 9.362153344208808,
      "grad_norm": 3.5303916931152344,
      "learning_rate": 6.174241986155238e-07,
      "loss": 0.1894,
      "num_input_tokens_seen": 123852816,
      "step": 57390
    },
    {
      "epoch": 9.362969004893964,
      "grad_norm": 0.06925785541534424,
      "learning_rate": 6.158530311109884e-07,
      "loss": 0.1415,
      "num_input_tokens_seen": 123863312,
      "step": 57395
    },
    {
      "epoch": 9.36378466557912,
      "grad_norm": 0.1369050145149231,
      "learning_rate": 6.142838403091322e-07,
      "loss": 0.0222,
      "num_input_tokens_seen": 123873936,
      "step": 57400
    },
    {
      "epoch": 9.364600326264274,
      "grad_norm": 13.540993690490723,
      "learning_rate": 6.127166263371592e-07,
      "loss": 0.2189,
      "num_input_tokens_seen": 123885648,
      "step": 57405
    },
    {
      "epoch": 9.36541598694943,
      "grad_norm": 0.09398102015256882,
      "learning_rate": 6.111513893221149e-07,
      "loss": 0.0838,
      "num_input_tokens_seen": 123896528,
      "step": 57410
    },
    {
      "epoch": 9.366231647634583,
      "grad_norm": 0.09548982232809067,
      "learning_rate": 6.095881293908867e-07,
      "loss": 0.0808,
      "num_input_tokens_seen": 123907728,
      "step": 57415
    },
    {
      "epoch": 9.367047308319739,
      "grad_norm": 0.11356570571660995,
      "learning_rate": 6.080268466702011e-07,
      "loss": 0.004,
      "num_input_tokens_seen": 123918128,
      "step": 57420
    },
    {
      "epoch": 9.367862969004895,
      "grad_norm": 0.31677040457725525,
      "learning_rate": 6.064675412866233e-07,
      "loss": 0.0081,
      "num_input_tokens_seen": 123929392,
      "step": 57425
    },
    {
      "epoch": 9.368678629690049,
      "grad_norm": 0.34641584753990173,
      "learning_rate": 6.049102133665552e-07,
      "loss": 0.1761,
      "num_input_tokens_seen": 123940688,
      "step": 57430
    },
    {
      "epoch": 9.369494290375204,
      "grad_norm": 0.041574303060770035,
      "learning_rate": 6.033548630362457e-07,
      "loss": 0.0078,
      "num_input_tokens_seen": 123950704,
      "step": 57435
    },
    {
      "epoch": 9.370309951060358,
      "grad_norm": 0.26669496297836304,
      "learning_rate": 6.018014904217801e-07,
      "loss": 0.0069,
      "num_input_tokens_seen": 123961552,
      "step": 57440
    },
    {
      "epoch": 9.371125611745514,
      "grad_norm": 0.10131517052650452,
      "learning_rate": 6.002500956490798e-07,
      "loss": 0.0067,
      "num_input_tokens_seen": 123971408,
      "step": 57445
    },
    {
      "epoch": 9.37194127243067,
      "grad_norm": 0.13997958600521088,
      "learning_rate": 5.987006788439109e-07,
      "loss": 0.0047,
      "num_input_tokens_seen": 123981264,
      "step": 57450
    },
    {
      "epoch": 9.372756933115824,
      "grad_norm": 0.17554813623428345,
      "learning_rate": 5.971532401318758e-07,
      "loss": 0.133,
      "num_input_tokens_seen": 123992944,
      "step": 57455
    },
    {
      "epoch": 9.37357259380098,
      "grad_norm": 0.18611003458499908,
      "learning_rate": 5.956077796384185e-07,
      "loss": 0.0985,
      "num_input_tokens_seen": 124004464,
      "step": 57460
    },
    {
      "epoch": 9.374388254486133,
      "grad_norm": 0.10970862954854965,
      "learning_rate": 5.940642974888195e-07,
      "loss": 0.0104,
      "num_input_tokens_seen": 124017072,
      "step": 57465
    },
    {
      "epoch": 9.375203915171289,
      "grad_norm": 0.06436091661453247,
      "learning_rate": 5.925227938082034e-07,
      "loss": 0.1005,
      "num_input_tokens_seen": 124027792,
      "step": 57470
    },
    {
      "epoch": 9.376019575856443,
      "grad_norm": 3.337602138519287,
      "learning_rate": 5.909832687215317e-07,
      "loss": 0.0926,
      "num_input_tokens_seen": 124039440,
      "step": 57475
    },
    {
      "epoch": 9.376835236541599,
      "grad_norm": 3.4454894065856934,
      "learning_rate": 5.894457223536071e-07,
      "loss": 0.1433,
      "num_input_tokens_seen": 124050608,
      "step": 57480
    },
    {
      "epoch": 9.377650897226754,
      "grad_norm": 0.0916593074798584,
      "learning_rate": 5.879101548290716e-07,
      "loss": 0.0908,
      "num_input_tokens_seen": 124061296,
      "step": 57485
    },
    {
      "epoch": 9.378466557911908,
      "grad_norm": 0.2612755298614502,
      "learning_rate": 5.863765662724036e-07,
      "loss": 0.3228,
      "num_input_tokens_seen": 124071920,
      "step": 57490
    },
    {
      "epoch": 9.379282218597064,
      "grad_norm": 0.09947580844163895,
      "learning_rate": 5.848449568079228e-07,
      "loss": 0.115,
      "num_input_tokens_seen": 124082608,
      "step": 57495
    },
    {
      "epoch": 9.380097879282218,
      "grad_norm": 0.14681026339530945,
      "learning_rate": 5.83315326559794e-07,
      "loss": 0.1171,
      "num_input_tokens_seen": 124094416,
      "step": 57500
    },
    {
      "epoch": 9.380913539967374,
      "grad_norm": 2.713966131210327,
      "learning_rate": 5.817876756520125e-07,
      "loss": 0.0842,
      "num_input_tokens_seen": 124105168,
      "step": 57505
    },
    {
      "epoch": 9.38172920065253,
      "grad_norm": 0.03372839093208313,
      "learning_rate": 5.80262004208415e-07,
      "loss": 0.0769,
      "num_input_tokens_seen": 124114992,
      "step": 57510
    },
    {
      "epoch": 9.382544861337683,
      "grad_norm": 2.9920592308044434,
      "learning_rate": 5.787383123526891e-07,
      "loss": 0.0075,
      "num_input_tokens_seen": 124125936,
      "step": 57515
    },
    {
      "epoch": 9.383360522022839,
      "grad_norm": 0.09146337956190109,
      "learning_rate": 5.772166002083467e-07,
      "loss": 0.0833,
      "num_input_tokens_seen": 124137616,
      "step": 57520
    },
    {
      "epoch": 9.384176182707993,
      "grad_norm": 3.874807119369507,
      "learning_rate": 5.756968678987451e-07,
      "loss": 0.1147,
      "num_input_tokens_seen": 124147856,
      "step": 57525
    },
    {
      "epoch": 9.384991843393149,
      "grad_norm": 0.15907549858093262,
      "learning_rate": 5.741791155470854e-07,
      "loss": 0.0058,
      "num_input_tokens_seen": 124159824,
      "step": 57530
    },
    {
      "epoch": 9.385807504078304,
      "grad_norm": 0.1259309947490692,
      "learning_rate": 5.726633432764e-07,
      "loss": 0.0044,
      "num_input_tokens_seen": 124169168,
      "step": 57535
    },
    {
      "epoch": 9.386623164763458,
      "grad_norm": 0.3803000748157501,
      "learning_rate": 5.711495512095682e-07,
      "loss": 0.0052,
      "num_input_tokens_seen": 124179952,
      "step": 57540
    },
    {
      "epoch": 9.387438825448614,
      "grad_norm": 0.07788604497909546,
      "learning_rate": 5.696377394693003e-07,
      "loss": 0.1984,
      "num_input_tokens_seen": 124189648,
      "step": 57545
    },
    {
      "epoch": 9.388254486133768,
      "grad_norm": 0.09048943966627121,
      "learning_rate": 5.681279081781593e-07,
      "loss": 0.0955,
      "num_input_tokens_seen": 124200336,
      "step": 57550
    },
    {
      "epoch": 9.389070146818923,
      "grad_norm": 0.638272762298584,
      "learning_rate": 5.66620057458539e-07,
      "loss": 0.1214,
      "num_input_tokens_seen": 124209264,
      "step": 57555
    },
    {
      "epoch": 9.38988580750408,
      "grad_norm": 0.030949106439948082,
      "learning_rate": 5.651141874326666e-07,
      "loss": 0.0039,
      "num_input_tokens_seen": 124219184,
      "step": 57560
    },
    {
      "epoch": 9.390701468189233,
      "grad_norm": 0.049306951463222504,
      "learning_rate": 5.636102982226221e-07,
      "loss": 0.0085,
      "num_input_tokens_seen": 124230800,
      "step": 57565
    },
    {
      "epoch": 9.391517128874389,
      "grad_norm": 0.09985353797674179,
      "learning_rate": 5.621083899503138e-07,
      "loss": 0.0032,
      "num_input_tokens_seen": 124242224,
      "step": 57570
    },
    {
      "epoch": 9.392332789559543,
      "grad_norm": 4.070796012878418,
      "learning_rate": 5.606084627374969e-07,
      "loss": 0.1413,
      "num_input_tokens_seen": 124252592,
      "step": 57575
    },
    {
      "epoch": 9.393148450244698,
      "grad_norm": 0.39609283208847046,
      "learning_rate": 5.591105167057631e-07,
      "loss": 0.0078,
      "num_input_tokens_seen": 124263760,
      "step": 57580
    },
    {
      "epoch": 9.393964110929852,
      "grad_norm": 0.14552120864391327,
      "learning_rate": 5.576145519765402e-07,
      "loss": 0.004,
      "num_input_tokens_seen": 124274704,
      "step": 57585
    },
    {
      "epoch": 9.394779771615008,
      "grad_norm": 0.24937283992767334,
      "learning_rate": 5.561205686711035e-07,
      "loss": 0.0061,
      "num_input_tokens_seen": 124284784,
      "step": 57590
    },
    {
      "epoch": 9.395595432300164,
      "grad_norm": 0.12961940467357635,
      "learning_rate": 5.546285669105589e-07,
      "loss": 0.0408,
      "num_input_tokens_seen": 124294512,
      "step": 57595
    },
    {
      "epoch": 9.396411092985318,
      "grad_norm": 0.31368333101272583,
      "learning_rate": 5.531385468158595e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 124306096,
      "step": 57600
    },
    {
      "epoch": 9.397226753670473,
      "grad_norm": 4.694768905639648,
      "learning_rate": 5.516505085077895e-07,
      "loss": 0.1096,
      "num_input_tokens_seen": 124317104,
      "step": 57605
    },
    {
      "epoch": 9.398042414355627,
      "grad_norm": 0.1727198362350464,
      "learning_rate": 5.501644521069799e-07,
      "loss": 0.0124,
      "num_input_tokens_seen": 124327792,
      "step": 57610
    },
    {
      "epoch": 9.398858075040783,
      "grad_norm": 0.04438969865441322,
      "learning_rate": 5.486803777338956e-07,
      "loss": 0.0879,
      "num_input_tokens_seen": 124340112,
      "step": 57615
    },
    {
      "epoch": 9.399673735725939,
      "grad_norm": 0.183514803647995,
      "learning_rate": 5.471982855088459e-07,
      "loss": 0.098,
      "num_input_tokens_seen": 124351184,
      "step": 57620
    },
    {
      "epoch": 9.400489396411093,
      "grad_norm": 0.06302642822265625,
      "learning_rate": 5.457181755519763e-07,
      "loss": 0.066,
      "num_input_tokens_seen": 124361456,
      "step": 57625
    },
    {
      "epoch": 9.401305057096248,
      "grad_norm": 6.795217514038086,
      "learning_rate": 5.442400479832715e-07,
      "loss": 0.0512,
      "num_input_tokens_seen": 124372848,
      "step": 57630
    },
    {
      "epoch": 9.402120717781402,
      "grad_norm": 16.359466552734375,
      "learning_rate": 5.427639029225551e-07,
      "loss": 0.0202,
      "num_input_tokens_seen": 124383664,
      "step": 57635
    },
    {
      "epoch": 9.402936378466558,
      "grad_norm": 0.09848824888467789,
      "learning_rate": 5.412897404894896e-07,
      "loss": 0.0053,
      "num_input_tokens_seen": 124394160,
      "step": 57640
    },
    {
      "epoch": 9.403752039151712,
      "grad_norm": 0.05733660236001015,
      "learning_rate": 5.398175608035821e-07,
      "loss": 0.1236,
      "num_input_tokens_seen": 124405584,
      "step": 57645
    },
    {
      "epoch": 9.404567699836868,
      "grad_norm": 12.071624755859375,
      "learning_rate": 5.38347363984168e-07,
      "loss": 0.0749,
      "num_input_tokens_seen": 124416560,
      "step": 57650
    },
    {
      "epoch": 9.405383360522023,
      "grad_norm": 0.1648532748222351,
      "learning_rate": 5.368791501504378e-07,
      "loss": 0.0059,
      "num_input_tokens_seen": 124426864,
      "step": 57655
    },
    {
      "epoch": 9.406199021207177,
      "grad_norm": 0.15438251197338104,
      "learning_rate": 5.354129194214103e-07,
      "loss": 0.0054,
      "num_input_tokens_seen": 124438384,
      "step": 57660
    },
    {
      "epoch": 9.407014681892333,
      "grad_norm": 0.1673191785812378,
      "learning_rate": 5.339486719159404e-07,
      "loss": 0.0133,
      "num_input_tokens_seen": 124449936,
      "step": 57665
    },
    {
      "epoch": 9.407830342577487,
      "grad_norm": 0.059828829020261765,
      "learning_rate": 5.324864077527331e-07,
      "loss": 0.2073,
      "num_input_tokens_seen": 124461040,
      "step": 57670
    },
    {
      "epoch": 9.408646003262643,
      "grad_norm": 0.28742295503616333,
      "learning_rate": 5.310261270503214e-07,
      "loss": 0.1346,
      "num_input_tokens_seen": 124472144,
      "step": 57675
    },
    {
      "epoch": 9.409461663947798,
      "grad_norm": 0.21105064451694489,
      "learning_rate": 5.295678299270884e-07,
      "loss": 0.0846,
      "num_input_tokens_seen": 124483792,
      "step": 57680
    },
    {
      "epoch": 9.410277324632952,
      "grad_norm": 0.1553102731704712,
      "learning_rate": 5.281115165012479e-07,
      "loss": 0.1999,
      "num_input_tokens_seen": 124495152,
      "step": 57685
    },
    {
      "epoch": 9.411092985318108,
      "grad_norm": 3.6352155208587646,
      "learning_rate": 5.266571868908582e-07,
      "loss": 0.1038,
      "num_input_tokens_seen": 124505360,
      "step": 57690
    },
    {
      "epoch": 9.411908646003262,
      "grad_norm": 0.22404715418815613,
      "learning_rate": 5.252048412138111e-07,
      "loss": 0.3082,
      "num_input_tokens_seen": 124515184,
      "step": 57695
    },
    {
      "epoch": 9.412724306688418,
      "grad_norm": 0.12199053168296814,
      "learning_rate": 5.237544795878457e-07,
      "loss": 0.1025,
      "num_input_tokens_seen": 124526416,
      "step": 57700
    },
    {
      "epoch": 9.413539967373573,
      "grad_norm": 4.170020580291748,
      "learning_rate": 5.22306102130532e-07,
      "loss": 0.1867,
      "num_input_tokens_seen": 124537648,
      "step": 57705
    },
    {
      "epoch": 9.414355628058727,
      "grad_norm": 0.051957301795482635,
      "learning_rate": 5.208597089592871e-07,
      "loss": 0.1987,
      "num_input_tokens_seen": 124547920,
      "step": 57710
    },
    {
      "epoch": 9.415171288743883,
      "grad_norm": 0.1836085319519043,
      "learning_rate": 5.194153001913588e-07,
      "loss": 0.0062,
      "num_input_tokens_seen": 124556624,
      "step": 57715
    },
    {
      "epoch": 9.415986949429037,
      "grad_norm": 3.2332894802093506,
      "learning_rate": 5.179728759438368e-07,
      "loss": 0.2768,
      "num_input_tokens_seen": 124566448,
      "step": 57720
    },
    {
      "epoch": 9.416802610114193,
      "grad_norm": 0.09326938539743423,
      "learning_rate": 5.165324363336582e-07,
      "loss": 0.0992,
      "num_input_tokens_seen": 124576624,
      "step": 57725
    },
    {
      "epoch": 9.417618270799348,
      "grad_norm": 0.11804001778364182,
      "learning_rate": 5.150939814775852e-07,
      "loss": 0.0694,
      "num_input_tokens_seen": 124587184,
      "step": 57730
    },
    {
      "epoch": 9.418433931484502,
      "grad_norm": 3.470707893371582,
      "learning_rate": 5.136575114922299e-07,
      "loss": 0.1025,
      "num_input_tokens_seen": 124598448,
      "step": 57735
    },
    {
      "epoch": 9.419249592169658,
      "grad_norm": 6.755199909210205,
      "learning_rate": 5.122230264940409e-07,
      "loss": 0.1097,
      "num_input_tokens_seen": 124610576,
      "step": 57740
    },
    {
      "epoch": 9.420065252854812,
      "grad_norm": 0.10873281955718994,
      "learning_rate": 5.107905265993001e-07,
      "loss": 0.006,
      "num_input_tokens_seen": 124620656,
      "step": 57745
    },
    {
      "epoch": 9.420880913539968,
      "grad_norm": 0.06490042805671692,
      "learning_rate": 5.09360011924137e-07,
      "loss": 0.066,
      "num_input_tokens_seen": 124631344,
      "step": 57750
    },
    {
      "epoch": 9.421696574225122,
      "grad_norm": 8.283732414245605,
      "learning_rate": 5.079314825845144e-07,
      "loss": 0.098,
      "num_input_tokens_seen": 124641872,
      "step": 57755
    },
    {
      "epoch": 9.422512234910277,
      "grad_norm": 0.10572148114442825,
      "learning_rate": 5.065049386962395e-07,
      "loss": 0.1146,
      "num_input_tokens_seen": 124652464,
      "step": 57760
    },
    {
      "epoch": 9.423327895595433,
      "grad_norm": 0.04828275740146637,
      "learning_rate": 5.050803803749532e-07,
      "loss": 0.0065,
      "num_input_tokens_seen": 124662896,
      "step": 57765
    },
    {
      "epoch": 9.424143556280587,
      "grad_norm": 0.14845052361488342,
      "learning_rate": 5.036578077361381e-07,
      "loss": 0.0052,
      "num_input_tokens_seen": 124674480,
      "step": 57770
    },
    {
      "epoch": 9.424959216965743,
      "grad_norm": 0.08852400630712509,
      "learning_rate": 5.022372208951131e-07,
      "loss": 0.0095,
      "num_input_tokens_seen": 124684240,
      "step": 57775
    },
    {
      "epoch": 9.425774877650896,
      "grad_norm": 0.06310316920280457,
      "learning_rate": 5.008186199670389e-07,
      "loss": 0.0028,
      "num_input_tokens_seen": 124696336,
      "step": 57780
    },
    {
      "epoch": 9.426590538336052,
      "grad_norm": 0.19328726828098297,
      "learning_rate": 4.994020050669152e-07,
      "loss": 0.1241,
      "num_input_tokens_seen": 124708432,
      "step": 57785
    },
    {
      "epoch": 9.427406199021208,
      "grad_norm": 0.09310528635978699,
      "learning_rate": 4.979873763095805e-07,
      "loss": 0.0279,
      "num_input_tokens_seen": 124721072,
      "step": 57790
    },
    {
      "epoch": 9.428221859706362,
      "grad_norm": 0.09432114660739899,
      "learning_rate": 4.965747338097099e-07,
      "loss": 0.0948,
      "num_input_tokens_seen": 124731696,
      "step": 57795
    },
    {
      "epoch": 9.429037520391518,
      "grad_norm": 0.17877328395843506,
      "learning_rate": 4.951640776818228e-07,
      "loss": 0.136,
      "num_input_tokens_seen": 124742416,
      "step": 57800
    },
    {
      "epoch": 9.429853181076671,
      "grad_norm": 0.1892382949590683,
      "learning_rate": 4.937554080402695e-07,
      "loss": 0.1272,
      "num_input_tokens_seen": 124752784,
      "step": 57805
    },
    {
      "epoch": 9.430668841761827,
      "grad_norm": 5.177409648895264,
      "learning_rate": 4.923487249992476e-07,
      "loss": 0.2363,
      "num_input_tokens_seen": 124763248,
      "step": 57810
    },
    {
      "epoch": 9.431484502446983,
      "grad_norm": 0.07287146896123886,
      "learning_rate": 4.909440286727879e-07,
      "loss": 0.3061,
      "num_input_tokens_seen": 124774576,
      "step": 57815
    },
    {
      "epoch": 9.432300163132137,
      "grad_norm": 4.4076056480407715,
      "learning_rate": 4.895413191747633e-07,
      "loss": 0.1904,
      "num_input_tokens_seen": 124784368,
      "step": 57820
    },
    {
      "epoch": 9.433115823817293,
      "grad_norm": 0.03391724079847336,
      "learning_rate": 4.881405966188801e-07,
      "loss": 0.1205,
      "num_input_tokens_seen": 124793904,
      "step": 57825
    },
    {
      "epoch": 9.433931484502446,
      "grad_norm": 0.044700562953948975,
      "learning_rate": 4.867418611186974e-07,
      "loss": 0.0039,
      "num_input_tokens_seen": 124804848,
      "step": 57830
    },
    {
      "epoch": 9.434747145187602,
      "grad_norm": 0.28951773047447205,
      "learning_rate": 4.853451127875968e-07,
      "loss": 0.1278,
      "num_input_tokens_seen": 124817840,
      "step": 57835
    },
    {
      "epoch": 9.435562805872756,
      "grad_norm": 0.12186767905950546,
      "learning_rate": 4.839503517388072e-07,
      "loss": 0.0055,
      "num_input_tokens_seen": 124828080,
      "step": 57840
    },
    {
      "epoch": 9.436378466557912,
      "grad_norm": 0.10500957816839218,
      "learning_rate": 4.825575780853964e-07,
      "loss": 0.005,
      "num_input_tokens_seen": 124839632,
      "step": 57845
    },
    {
      "epoch": 9.437194127243067,
      "grad_norm": 0.3825618326663971,
      "learning_rate": 4.811667919402685e-07,
      "loss": 0.0913,
      "num_input_tokens_seen": 124852176,
      "step": 57850
    },
    {
      "epoch": 9.438009787928221,
      "grad_norm": 0.18801839649677277,
      "learning_rate": 4.797779934161667e-07,
      "loss": 0.161,
      "num_input_tokens_seen": 124862768,
      "step": 57855
    },
    {
      "epoch": 9.438825448613377,
      "grad_norm": 6.414939880371094,
      "learning_rate": 4.78391182625676e-07,
      "loss": 0.1748,
      "num_input_tokens_seen": 124873936,
      "step": 57860
    },
    {
      "epoch": 9.439641109298531,
      "grad_norm": 0.04732104018330574,
      "learning_rate": 4.770063596812146e-07,
      "loss": 0.0807,
      "num_input_tokens_seen": 124884912,
      "step": 57865
    },
    {
      "epoch": 9.440456769983687,
      "grad_norm": 0.07606936246156693,
      "learning_rate": 4.7562352469504855e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 124895632,
      "step": 57870
    },
    {
      "epoch": 9.441272430668842,
      "grad_norm": 0.059153228998184204,
      "learning_rate": 4.7424267777927414e-07,
      "loss": 0.0026,
      "num_input_tokens_seen": 124906640,
      "step": 57875
    },
    {
      "epoch": 9.442088091353996,
      "grad_norm": 0.04881925880908966,
      "learning_rate": 4.728638190458323e-07,
      "loss": 0.0048,
      "num_input_tokens_seen": 124917392,
      "step": 57880
    },
    {
      "epoch": 9.442903752039152,
      "grad_norm": 0.13776183128356934,
      "learning_rate": 4.7148694860649765e-07,
      "loss": 0.005,
      "num_input_tokens_seen": 124927408,
      "step": 57885
    },
    {
      "epoch": 9.443719412724306,
      "grad_norm": 0.10099656879901886,
      "learning_rate": 4.701120665728892e-07,
      "loss": 0.0964,
      "num_input_tokens_seen": 124938160,
      "step": 57890
    },
    {
      "epoch": 9.444535073409462,
      "grad_norm": 0.04308278486132622,
      "learning_rate": 4.687391730564594e-07,
      "loss": 0.0055,
      "num_input_tokens_seen": 124947376,
      "step": 57895
    },
    {
      "epoch": 9.445350734094617,
      "grad_norm": 0.10519593209028244,
      "learning_rate": 4.673682681684999e-07,
      "loss": 0.0677,
      "num_input_tokens_seen": 124958576,
      "step": 57900
    },
    {
      "epoch": 9.446166394779771,
      "grad_norm": 0.17816418409347534,
      "learning_rate": 4.6599935202014943e-07,
      "loss": 0.2961,
      "num_input_tokens_seen": 124968944,
      "step": 57905
    },
    {
      "epoch": 9.446982055464927,
      "grad_norm": 0.6837834715843201,
      "learning_rate": 4.646324247223749e-07,
      "loss": 0.0211,
      "num_input_tokens_seen": 124980496,
      "step": 57910
    },
    {
      "epoch": 9.447797716150081,
      "grad_norm": 18.661529541015625,
      "learning_rate": 4.6326748638598485e-07,
      "loss": 0.157,
      "num_input_tokens_seen": 124990256,
      "step": 57915
    },
    {
      "epoch": 9.448613376835237,
      "grad_norm": 0.057231463491916656,
      "learning_rate": 4.619045371216324e-07,
      "loss": 0.1044,
      "num_input_tokens_seen": 125001840,
      "step": 57920
    },
    {
      "epoch": 9.449429037520392,
      "grad_norm": 0.04664487764239311,
      "learning_rate": 4.605435770398042e-07,
      "loss": 0.0549,
      "num_input_tokens_seen": 125012272,
      "step": 57925
    },
    {
      "epoch": 9.450244698205546,
      "grad_norm": 8.235623359680176,
      "learning_rate": 4.591846062508232e-07,
      "loss": 0.1696,
      "num_input_tokens_seen": 125023024,
      "step": 57930
    },
    {
      "epoch": 9.451060358890702,
      "grad_norm": 0.13234318792819977,
      "learning_rate": 4.578276248648594e-07,
      "loss": 0.0821,
      "num_input_tokens_seen": 125035184,
      "step": 57935
    },
    {
      "epoch": 9.451876019575856,
      "grad_norm": 0.6127078533172607,
      "learning_rate": 4.5647263299191113e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 125045424,
      "step": 57940
    },
    {
      "epoch": 9.452691680261012,
      "grad_norm": 9.226028442382812,
      "learning_rate": 4.5511963074182653e-07,
      "loss": 0.2673,
      "num_input_tokens_seen": 125055504,
      "step": 57945
    },
    {
      "epoch": 9.453507340946166,
      "grad_norm": 0.32815080881118774,
      "learning_rate": 4.5376861822428176e-07,
      "loss": 0.1972,
      "num_input_tokens_seen": 125065744,
      "step": 57950
    },
    {
      "epoch": 9.454323001631321,
      "grad_norm": 0.5989218354225159,
      "learning_rate": 4.524195955488031e-07,
      "loss": 0.1054,
      "num_input_tokens_seen": 125076720,
      "step": 57955
    },
    {
      "epoch": 9.455138662316477,
      "grad_norm": 0.24300257861614227,
      "learning_rate": 4.5107256282474196e-07,
      "loss": 0.3522,
      "num_input_tokens_seen": 125087504,
      "step": 57960
    },
    {
      "epoch": 9.455954323001631,
      "grad_norm": 0.14050297439098358,
      "learning_rate": 4.4972752016129995e-07,
      "loss": 0.0249,
      "num_input_tokens_seen": 125099408,
      "step": 57965
    },
    {
      "epoch": 9.456769983686787,
      "grad_norm": 56.48374557495117,
      "learning_rate": 4.483844676675092e-07,
      "loss": 0.1118,
      "num_input_tokens_seen": 125109904,
      "step": 57970
    },
    {
      "epoch": 9.45758564437194,
      "grad_norm": 0.13741572201251984,
      "learning_rate": 4.4704340545224934e-07,
      "loss": 0.0132,
      "num_input_tokens_seen": 125120144,
      "step": 57975
    },
    {
      "epoch": 9.458401305057096,
      "grad_norm": 3.443633794784546,
      "learning_rate": 4.457043336242306e-07,
      "loss": 0.1897,
      "num_input_tokens_seen": 125131056,
      "step": 57980
    },
    {
      "epoch": 9.459216965742252,
      "grad_norm": 2.717332601547241,
      "learning_rate": 4.443672522920078e-07,
      "loss": 0.1386,
      "num_input_tokens_seen": 125141072,
      "step": 57985
    },
    {
      "epoch": 9.460032626427406,
      "grad_norm": 0.04186335578560829,
      "learning_rate": 4.4303216156396933e-07,
      "loss": 0.0061,
      "num_input_tokens_seen": 125152656,
      "step": 57990
    },
    {
      "epoch": 9.460848287112562,
      "grad_norm": 0.09012457728385925,
      "learning_rate": 4.416990615483396e-07,
      "loss": 0.115,
      "num_input_tokens_seen": 125164880,
      "step": 57995
    },
    {
      "epoch": 9.461663947797716,
      "grad_norm": 0.19651956856250763,
      "learning_rate": 4.4036795235319617e-07,
      "loss": 0.0072,
      "num_input_tokens_seen": 125175792,
      "step": 58000
    },
    {
      "epoch": 9.462479608482871,
      "grad_norm": 0.5376098155975342,
      "learning_rate": 4.390388340864415e-07,
      "loss": 0.0346,
      "num_input_tokens_seen": 125187152,
      "step": 58005
    },
    {
      "epoch": 9.463295269168025,
      "grad_norm": 0.05495506897568703,
      "learning_rate": 4.377117068558201e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 125197072,
      "step": 58010
    },
    {
      "epoch": 9.464110929853181,
      "grad_norm": 5.769318580627441,
      "learning_rate": 4.363865707689152e-07,
      "loss": 0.0105,
      "num_input_tokens_seen": 125208144,
      "step": 58015
    },
    {
      "epoch": 9.464926590538337,
      "grad_norm": 0.17950724065303802,
      "learning_rate": 4.350634259331465e-07,
      "loss": 0.0974,
      "num_input_tokens_seen": 125218448,
      "step": 58020
    },
    {
      "epoch": 9.46574225122349,
      "grad_norm": 2.7524445056915283,
      "learning_rate": 4.33742272455781e-07,
      "loss": 0.2416,
      "num_input_tokens_seen": 125229168,
      "step": 58025
    },
    {
      "epoch": 9.466557911908646,
      "grad_norm": 0.229463130235672,
      "learning_rate": 4.324231104439136e-07,
      "loss": 0.0923,
      "num_input_tokens_seen": 125240496,
      "step": 58030
    },
    {
      "epoch": 9.4673735725938,
      "grad_norm": 0.1134629175066948,
      "learning_rate": 4.3110594000448365e-07,
      "loss": 0.1163,
      "num_input_tokens_seen": 125251408,
      "step": 58035
    },
    {
      "epoch": 9.468189233278956,
      "grad_norm": 0.019398687407374382,
      "learning_rate": 4.29790761244267e-07,
      "loss": 0.1041,
      "num_input_tokens_seen": 125262128,
      "step": 58040
    },
    {
      "epoch": 9.469004893964112,
      "grad_norm": 0.08468424528837204,
      "learning_rate": 4.2847757426988097e-07,
      "loss": 0.0041,
      "num_input_tokens_seen": 125272816,
      "step": 58045
    },
    {
      "epoch": 9.469820554649266,
      "grad_norm": 0.3412284553050995,
      "learning_rate": 4.271663791877767e-07,
      "loss": 0.1515,
      "num_input_tokens_seen": 125283792,
      "step": 58050
    },
    {
      "epoch": 9.470636215334421,
      "grad_norm": 4.713772773742676,
      "learning_rate": 4.258571761042468e-07,
      "loss": 0.1922,
      "num_input_tokens_seen": 125294960,
      "step": 58055
    },
    {
      "epoch": 9.471451876019575,
      "grad_norm": 0.0300301481038332,
      "learning_rate": 4.2454996512542033e-07,
      "loss": 0.054,
      "num_input_tokens_seen": 125304816,
      "step": 58060
    },
    {
      "epoch": 9.47226753670473,
      "grad_norm": 0.27174460887908936,
      "learning_rate": 4.2324474635727085e-07,
      "loss": 0.0666,
      "num_input_tokens_seen": 125314448,
      "step": 58065
    },
    {
      "epoch": 9.473083197389887,
      "grad_norm": 0.14359116554260254,
      "learning_rate": 4.219415199056026e-07,
      "loss": 0.0037,
      "num_input_tokens_seen": 125325584,
      "step": 58070
    },
    {
      "epoch": 9.47389885807504,
      "grad_norm": 0.27878323197364807,
      "learning_rate": 4.2064028587606163e-07,
      "loss": 0.0049,
      "num_input_tokens_seen": 125337328,
      "step": 58075
    },
    {
      "epoch": 9.474714518760196,
      "grad_norm": 0.22504855692386627,
      "learning_rate": 4.19341044374133e-07,
      "loss": 0.0062,
      "num_input_tokens_seen": 125348336,
      "step": 58080
    },
    {
      "epoch": 9.47553017944535,
      "grad_norm": 0.06329631060361862,
      "learning_rate": 4.180437955051436e-07,
      "loss": 0.1611,
      "num_input_tokens_seen": 125359152,
      "step": 58085
    },
    {
      "epoch": 9.476345840130506,
      "grad_norm": 0.08869102597236633,
      "learning_rate": 4.167485393742482e-07,
      "loss": 0.0036,
      "num_input_tokens_seen": 125371792,
      "step": 58090
    },
    {
      "epoch": 9.477161500815662,
      "grad_norm": 0.058582451194524765,
      "learning_rate": 4.1545527608645163e-07,
      "loss": 0.0945,
      "num_input_tokens_seen": 125382960,
      "step": 58095
    },
    {
      "epoch": 9.477977161500815,
      "grad_norm": 4.101239204406738,
      "learning_rate": 4.1416400574659233e-07,
      "loss": 0.485,
      "num_input_tokens_seen": 125392752,
      "step": 58100
    },
    {
      "epoch": 9.478792822185971,
      "grad_norm": 0.1705857366323471,
      "learning_rate": 4.128747284593448e-07,
      "loss": 0.0626,
      "num_input_tokens_seen": 125404080,
      "step": 58105
    },
    {
      "epoch": 9.479608482871125,
      "grad_norm": 2.991602659225464,
      "learning_rate": 4.1158744432922835e-07,
      "loss": 0.1067,
      "num_input_tokens_seen": 125415568,
      "step": 58110
    },
    {
      "epoch": 9.48042414355628,
      "grad_norm": 0.2598673105239868,
      "learning_rate": 4.1030215346059e-07,
      "loss": 0.0563,
      "num_input_tokens_seen": 125426832,
      "step": 58115
    },
    {
      "epoch": 9.481239804241435,
      "grad_norm": 3.7980761528015137,
      "learning_rate": 4.090188559576269e-07,
      "loss": 0.105,
      "num_input_tokens_seen": 125436656,
      "step": 58120
    },
    {
      "epoch": 9.48205546492659,
      "grad_norm": 0.20062430202960968,
      "learning_rate": 4.077375519243698e-07,
      "loss": 0.0139,
      "num_input_tokens_seen": 125447440,
      "step": 58125
    },
    {
      "epoch": 9.482871125611746,
      "grad_norm": 7.75474214553833,
      "learning_rate": 4.0645824146468834e-07,
      "loss": 0.1233,
      "num_input_tokens_seen": 125458928,
      "step": 58130
    },
    {
      "epoch": 9.4836867862969,
      "grad_norm": 0.10313055664300919,
      "learning_rate": 4.0518092468228297e-07,
      "loss": 0.0067,
      "num_input_tokens_seen": 125469520,
      "step": 58135
    },
    {
      "epoch": 9.484502446982056,
      "grad_norm": 0.1118944063782692,
      "learning_rate": 4.0390560168070966e-07,
      "loss": 0.0042,
      "num_input_tokens_seen": 125480016,
      "step": 58140
    },
    {
      "epoch": 9.48531810766721,
      "grad_norm": 0.06879203021526337,
      "learning_rate": 4.026322725633441e-07,
      "loss": 0.0038,
      "num_input_tokens_seen": 125490608,
      "step": 58145
    },
    {
      "epoch": 9.486133768352365,
      "grad_norm": 0.12433100491762161,
      "learning_rate": 4.0136093743341485e-07,
      "loss": 0.0892,
      "num_input_tokens_seen": 125501392,
      "step": 58150
    },
    {
      "epoch": 9.486949429037521,
      "grad_norm": 0.09972847253084183,
      "learning_rate": 4.000915963939783e-07,
      "loss": 0.1006,
      "num_input_tokens_seen": 125510448,
      "step": 58155
    },
    {
      "epoch": 9.487765089722675,
      "grad_norm": 0.1359276920557022,
      "learning_rate": 3.988242495479383e-07,
      "loss": 0.1287,
      "num_input_tokens_seen": 125521968,
      "step": 58160
    },
    {
      "epoch": 9.48858075040783,
      "grad_norm": 0.08717767894268036,
      "learning_rate": 3.9755889699802926e-07,
      "loss": 0.1284,
      "num_input_tokens_seen": 125533360,
      "step": 58165
    },
    {
      "epoch": 9.489396411092985,
      "grad_norm": 0.10583186894655228,
      "learning_rate": 3.962955388468248e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 125544400,
      "step": 58170
    },
    {
      "epoch": 9.49021207177814,
      "grad_norm": 0.04262809455394745,
      "learning_rate": 3.950341751967457e-07,
      "loss": 0.0052,
      "num_input_tokens_seen": 125555568,
      "step": 58175
    },
    {
      "epoch": 9.491027732463296,
      "grad_norm": 0.1110820323228836,
      "learning_rate": 3.9377480615003794e-07,
      "loss": 0.2439,
      "num_input_tokens_seen": 125566224,
      "step": 58180
    },
    {
      "epoch": 9.49184339314845,
      "grad_norm": 0.1875482201576233,
      "learning_rate": 3.9251743180879483e-07,
      "loss": 0.0306,
      "num_input_tokens_seen": 125576976,
      "step": 58185
    },
    {
      "epoch": 9.492659053833606,
      "grad_norm": 0.2513222396373749,
      "learning_rate": 3.9126205227494605e-07,
      "loss": 0.0071,
      "num_input_tokens_seen": 125587856,
      "step": 58190
    },
    {
      "epoch": 9.49347471451876,
      "grad_norm": 6.3517632484436035,
      "learning_rate": 3.900086676502601e-07,
      "loss": 0.1194,
      "num_input_tokens_seen": 125598832,
      "step": 58195
    },
    {
      "epoch": 9.494290375203915,
      "grad_norm": 0.08826222270727158,
      "learning_rate": 3.8875727803634186e-07,
      "loss": 0.1277,
      "num_input_tokens_seen": 125610832,
      "step": 58200
    },
    {
      "epoch": 9.49510603588907,
      "grad_norm": 0.5476817488670349,
      "learning_rate": 3.8750788353463243e-07,
      "loss": 0.0902,
      "num_input_tokens_seen": 125621104,
      "step": 58205
    },
    {
      "epoch": 9.495921696574225,
      "grad_norm": 0.0348920039832592,
      "learning_rate": 3.862604842464201e-07,
      "loss": 0.0948,
      "num_input_tokens_seen": 125632752,
      "step": 58210
    },
    {
      "epoch": 9.49673735725938,
      "grad_norm": 0.1593414843082428,
      "learning_rate": 3.8501508027281865e-07,
      "loss": 0.0044,
      "num_input_tokens_seen": 125642832,
      "step": 58215
    },
    {
      "epoch": 9.497553017944535,
      "grad_norm": 0.04932843893766403,
      "learning_rate": 3.8377167171479154e-07,
      "loss": 0.0838,
      "num_input_tokens_seen": 125653840,
      "step": 58220
    },
    {
      "epoch": 9.49836867862969,
      "grad_norm": 0.16821761429309845,
      "learning_rate": 3.825302586731333e-07,
      "loss": 0.091,
      "num_input_tokens_seen": 125665104,
      "step": 58225
    },
    {
      "epoch": 9.499184339314844,
      "grad_norm": 0.13988712430000305,
      "learning_rate": 3.8129084124848e-07,
      "loss": 0.0044,
      "num_input_tokens_seen": 125675824,
      "step": 58230
    },
    {
      "epoch": 9.5,
      "grad_norm": 3.213824510574341,
      "learning_rate": 3.800534195413069e-07,
      "loss": 0.1249,
      "num_input_tokens_seen": 125686064,
      "step": 58235
    },
    {
      "epoch": 9.5,
      "eval_loss": 0.19253920018672943,
      "eval_runtime": 568.4597,
      "eval_samples_per_second": 4.794,
      "eval_steps_per_second": 1.2,
      "num_input_tokens_seen": 125686064,
      "step": 58235
    },
    {
      "epoch": 9.500815660685156,
      "grad_norm": 3.043863296508789,
      "learning_rate": 3.788179936519226e-07,
      "loss": 0.1592,
      "num_input_tokens_seen": 125697296,
      "step": 58240
    },
    {
      "epoch": 9.50163132137031,
      "grad_norm": 0.08366536349058151,
      "learning_rate": 3.775845636804776e-07,
      "loss": 0.0057,
      "num_input_tokens_seen": 125708688,
      "step": 58245
    },
    {
      "epoch": 9.502446982055465,
      "grad_norm": 0.06130436807870865,
      "learning_rate": 3.7635312972696404e-07,
      "loss": 0.0042,
      "num_input_tokens_seen": 125719792,
      "step": 58250
    },
    {
      "epoch": 9.50326264274062,
      "grad_norm": 0.10451815277338028,
      "learning_rate": 3.751236918912021e-07,
      "loss": 0.0058,
      "num_input_tokens_seen": 125729680,
      "step": 58255
    },
    {
      "epoch": 9.504078303425775,
      "grad_norm": 0.03382956609129906,
      "learning_rate": 3.7389625027285936e-07,
      "loss": 0.171,
      "num_input_tokens_seen": 125739472,
      "step": 58260
    },
    {
      "epoch": 9.50489396411093,
      "grad_norm": 0.05498937889933586,
      "learning_rate": 3.726708049714367e-07,
      "loss": 0.0024,
      "num_input_tokens_seen": 125751952,
      "step": 58265
    },
    {
      "epoch": 9.505709624796085,
      "grad_norm": 19.4271183013916,
      "learning_rate": 3.714473560862797e-07,
      "loss": 0.0378,
      "num_input_tokens_seen": 125762992,
      "step": 58270
    },
    {
      "epoch": 9.50652528548124,
      "grad_norm": 0.08409254997968674,
      "learning_rate": 3.702259037165617e-07,
      "loss": 0.2106,
      "num_input_tokens_seen": 125774480,
      "step": 58275
    },
    {
      "epoch": 9.507340946166394,
      "grad_norm": 10.254300117492676,
      "learning_rate": 3.690064479613009e-07,
      "loss": 0.0538,
      "num_input_tokens_seen": 125784944,
      "step": 58280
    },
    {
      "epoch": 9.50815660685155,
      "grad_norm": 4.610869407653809,
      "learning_rate": 3.67788988919357e-07,
      "loss": 0.2103,
      "num_input_tokens_seen": 125795120,
      "step": 58285
    },
    {
      "epoch": 9.508972267536706,
      "grad_norm": 10.56911849975586,
      "learning_rate": 3.665735266894177e-07,
      "loss": 0.1609,
      "num_input_tokens_seen": 125806640,
      "step": 58290
    },
    {
      "epoch": 9.50978792822186,
      "grad_norm": 18.840635299682617,
      "learning_rate": 3.653600613700209e-07,
      "loss": 0.036,
      "num_input_tokens_seen": 125817712,
      "step": 58295
    },
    {
      "epoch": 9.510603588907015,
      "grad_norm": 0.27261725068092346,
      "learning_rate": 3.6414859305952955e-07,
      "loss": 0.2522,
      "num_input_tokens_seen": 125829008,
      "step": 58300
    },
    {
      "epoch": 9.51141924959217,
      "grad_norm": 3.41131329536438,
      "learning_rate": 3.629391218561512e-07,
      "loss": 0.1678,
      "num_input_tokens_seen": 125840016,
      "step": 58305
    },
    {
      "epoch": 9.512234910277325,
      "grad_norm": 0.09870193898677826,
      "learning_rate": 3.6173164785794076e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 125850608,
      "step": 58310
    },
    {
      "epoch": 9.513050570962479,
      "grad_norm": 9.098423957824707,
      "learning_rate": 3.605261711627728e-07,
      "loss": 0.0976,
      "num_input_tokens_seen": 125863312,
      "step": 58315
    },
    {
      "epoch": 9.513866231647635,
      "grad_norm": 0.09959303587675095,
      "learning_rate": 3.593226918683745e-07,
      "loss": 0.0034,
      "num_input_tokens_seen": 125874960,
      "step": 58320
    },
    {
      "epoch": 9.51468189233279,
      "grad_norm": 0.18624134361743927,
      "learning_rate": 3.5812121007230414e-07,
      "loss": 0.006,
      "num_input_tokens_seen": 125885264,
      "step": 58325
    },
    {
      "epoch": 9.515497553017944,
      "grad_norm": 0.0898330882191658,
      "learning_rate": 3.569217258719587e-07,
      "loss": 0.1345,
      "num_input_tokens_seen": 125896624,
      "step": 58330
    },
    {
      "epoch": 9.5163132137031,
      "grad_norm": 3.7100512981414795,
      "learning_rate": 3.557242393645771e-07,
      "loss": 0.1631,
      "num_input_tokens_seen": 125907056,
      "step": 58335
    },
    {
      "epoch": 9.517128874388254,
      "grad_norm": 0.24952083826065063,
      "learning_rate": 3.5452875064723445e-07,
      "loss": 0.2797,
      "num_input_tokens_seen": 125917616,
      "step": 58340
    },
    {
      "epoch": 9.51794453507341,
      "grad_norm": 0.2295496165752411,
      "learning_rate": 3.5333525981683937e-07,
      "loss": 0.2073,
      "num_input_tokens_seen": 125927984,
      "step": 58345
    },
    {
      "epoch": 9.518760195758565,
      "grad_norm": 4.289388656616211,
      "learning_rate": 3.521437669701422e-07,
      "loss": 0.106,
      "num_input_tokens_seen": 125938736,
      "step": 58350
    },
    {
      "epoch": 9.51957585644372,
      "grad_norm": 9.160137176513672,
      "learning_rate": 3.5095427220373513e-07,
      "loss": 0.2638,
      "num_input_tokens_seen": 125948816,
      "step": 58355
    },
    {
      "epoch": 9.520391517128875,
      "grad_norm": 0.06078990548849106,
      "learning_rate": 3.497667756140438e-07,
      "loss": 0.0036,
      "num_input_tokens_seen": 125959504,
      "step": 58360
    },
    {
      "epoch": 9.521207177814029,
      "grad_norm": 0.1819821298122406,
      "learning_rate": 3.4858127729733015e-07,
      "loss": 0.0325,
      "num_input_tokens_seen": 125971504,
      "step": 58365
    },
    {
      "epoch": 9.522022838499185,
      "grad_norm": 0.08073264360427856,
      "learning_rate": 3.473977773496978e-07,
      "loss": 0.0682,
      "num_input_tokens_seen": 125981840,
      "step": 58370
    },
    {
      "epoch": 9.522838499184338,
      "grad_norm": 0.015772346407175064,
      "learning_rate": 3.462162758670895e-07,
      "loss": 0.1356,
      "num_input_tokens_seen": 125992432,
      "step": 58375
    },
    {
      "epoch": 9.523654159869494,
      "grad_norm": 3.9206371307373047,
      "learning_rate": 3.4503677294527857e-07,
      "loss": 0.3046,
      "num_input_tokens_seen": 126002576,
      "step": 58380
    },
    {
      "epoch": 9.52446982055465,
      "grad_norm": 0.09783914685249329,
      "learning_rate": 3.438592686798886e-07,
      "loss": 0.0091,
      "num_input_tokens_seen": 126011920,
      "step": 58385
    },
    {
      "epoch": 9.525285481239804,
      "grad_norm": 0.06690984964370728,
      "learning_rate": 3.4268376316636816e-07,
      "loss": 0.0039,
      "num_input_tokens_seen": 126022864,
      "step": 58390
    },
    {
      "epoch": 9.52610114192496,
      "grad_norm": 0.031618401408195496,
      "learning_rate": 3.4151025650001056e-07,
      "loss": 0.0023,
      "num_input_tokens_seen": 126034960,
      "step": 58395
    },
    {
      "epoch": 9.526916802610113,
      "grad_norm": 0.13895615935325623,
      "learning_rate": 3.4033874877595074e-07,
      "loss": 0.0151,
      "num_input_tokens_seen": 126046672,
      "step": 58400
    },
    {
      "epoch": 9.52773246329527,
      "grad_norm": 0.06268240511417389,
      "learning_rate": 3.3916924008915163e-07,
      "loss": 0.0068,
      "num_input_tokens_seen": 126058000,
      "step": 58405
    },
    {
      "epoch": 9.528548123980425,
      "grad_norm": 0.09085869044065475,
      "learning_rate": 3.3800173053442354e-07,
      "loss": 0.003,
      "num_input_tokens_seen": 126068496,
      "step": 58410
    },
    {
      "epoch": 9.529363784665579,
      "grad_norm": 0.11956500262022018,
      "learning_rate": 3.3683622020640736e-07,
      "loss": 0.0331,
      "num_input_tokens_seen": 126078544,
      "step": 58415
    },
    {
      "epoch": 9.530179445350734,
      "grad_norm": 0.46510857343673706,
      "learning_rate": 3.356727091995859e-07,
      "loss": 0.318,
      "num_input_tokens_seen": 126089712,
      "step": 58420
    },
    {
      "epoch": 9.530995106035888,
      "grad_norm": 0.032259371131658554,
      "learning_rate": 3.3451119760828374e-07,
      "loss": 0.0792,
      "num_input_tokens_seen": 126100656,
      "step": 58425
    },
    {
      "epoch": 9.531810766721044,
      "grad_norm": 0.1584005057811737,
      "learning_rate": 3.333516855266533e-07,
      "loss": 0.2102,
      "num_input_tokens_seen": 126111376,
      "step": 58430
    },
    {
      "epoch": 9.5326264274062,
      "grad_norm": 0.07590462267398834,
      "learning_rate": 3.321941730486916e-07,
      "loss": 0.0088,
      "num_input_tokens_seen": 126122640,
      "step": 58435
    },
    {
      "epoch": 9.533442088091354,
      "grad_norm": 0.33584555983543396,
      "learning_rate": 3.3103866026823473e-07,
      "loss": 0.0088,
      "num_input_tokens_seen": 126132432,
      "step": 58440
    },
    {
      "epoch": 9.53425774877651,
      "grad_norm": 0.05497417226433754,
      "learning_rate": 3.2988514727895217e-07,
      "loss": 0.0028,
      "num_input_tokens_seen": 126142576,
      "step": 58445
    },
    {
      "epoch": 9.535073409461663,
      "grad_norm": 0.08528464287519455,
      "learning_rate": 3.287336341743524e-07,
      "loss": 0.0036,
      "num_input_tokens_seen": 126153552,
      "step": 58450
    },
    {
      "epoch": 9.535889070146819,
      "grad_norm": 0.0931963175535202,
      "learning_rate": 3.275841210477887e-07,
      "loss": 0.1672,
      "num_input_tokens_seen": 126165072,
      "step": 58455
    },
    {
      "epoch": 9.536704730831975,
      "grad_norm": 0.11097992956638336,
      "learning_rate": 3.264366079924419e-07,
      "loss": 0.007,
      "num_input_tokens_seen": 126175696,
      "step": 58460
    },
    {
      "epoch": 9.537520391517129,
      "grad_norm": 0.05439147725701332,
      "learning_rate": 3.252910951013349e-07,
      "loss": 0.0553,
      "num_input_tokens_seen": 126187088,
      "step": 58465
    },
    {
      "epoch": 9.538336052202284,
      "grad_norm": 0.1262970119714737,
      "learning_rate": 3.2414758246733234e-07,
      "loss": 0.1925,
      "num_input_tokens_seen": 126198480,
      "step": 58470
    },
    {
      "epoch": 9.539151712887438,
      "grad_norm": 0.46631112694740295,
      "learning_rate": 3.2300607018312944e-07,
      "loss": 0.0052,
      "num_input_tokens_seen": 126209904,
      "step": 58475
    },
    {
      "epoch": 9.539967373572594,
      "grad_norm": 0.08133397996425629,
      "learning_rate": 3.2186655834126335e-07,
      "loss": 0.0042,
      "num_input_tokens_seen": 126219792,
      "step": 58480
    },
    {
      "epoch": 9.540783034257748,
      "grad_norm": 0.1267090141773224,
      "learning_rate": 3.207290470341101e-07,
      "loss": 0.0105,
      "num_input_tokens_seen": 126230256,
      "step": 58485
    },
    {
      "epoch": 9.541598694942904,
      "grad_norm": 0.06662649661302567,
      "learning_rate": 3.1959353635388214e-07,
      "loss": 0.1066,
      "num_input_tokens_seen": 126241136,
      "step": 58490
    },
    {
      "epoch": 9.54241435562806,
      "grad_norm": 0.4028434753417969,
      "learning_rate": 3.1846002639263074e-07,
      "loss": 0.0069,
      "num_input_tokens_seen": 126252048,
      "step": 58495
    },
    {
      "epoch": 9.543230016313213,
      "grad_norm": 0.09449943155050278,
      "learning_rate": 3.17328517242238e-07,
      "loss": 0.1113,
      "num_input_tokens_seen": 126262320,
      "step": 58500
    },
    {
      "epoch": 9.544045676998369,
      "grad_norm": 4.737239837646484,
      "learning_rate": 3.161990089944389e-07,
      "loss": 0.3361,
      "num_input_tokens_seen": 126272464,
      "step": 58505
    },
    {
      "epoch": 9.544861337683523,
      "grad_norm": 0.17989328503608704,
      "learning_rate": 3.150715017407907e-07,
      "loss": 0.0036,
      "num_input_tokens_seen": 126283984,
      "step": 58510
    },
    {
      "epoch": 9.545676998368679,
      "grad_norm": 0.09410925954580307,
      "learning_rate": 3.1394599557269534e-07,
      "loss": 0.0155,
      "num_input_tokens_seen": 126295024,
      "step": 58515
    },
    {
      "epoch": 9.546492659053834,
      "grad_norm": 0.03972543403506279,
      "learning_rate": 3.128224905813965e-07,
      "loss": 0.0909,
      "num_input_tokens_seen": 126305296,
      "step": 58520
    },
    {
      "epoch": 9.547308319738988,
      "grad_norm": 0.05499481409788132,
      "learning_rate": 3.1170098685796565e-07,
      "loss": 0.0051,
      "num_input_tokens_seen": 126317264,
      "step": 58525
    },
    {
      "epoch": 9.548123980424144,
      "grad_norm": 0.32111480832099915,
      "learning_rate": 3.1058148449331914e-07,
      "loss": 0.0893,
      "num_input_tokens_seen": 126326640,
      "step": 58530
    },
    {
      "epoch": 9.548939641109298,
      "grad_norm": 0.15600375831127167,
      "learning_rate": 3.09463983578212e-07,
      "loss": 0.007,
      "num_input_tokens_seen": 126337840,
      "step": 58535
    },
    {
      "epoch": 9.549755301794454,
      "grad_norm": 15.249553680419922,
      "learning_rate": 3.0834848420323305e-07,
      "loss": 0.1863,
      "num_input_tokens_seen": 126347792,
      "step": 58540
    },
    {
      "epoch": 9.550570962479608,
      "grad_norm": 0.20458751916885376,
      "learning_rate": 3.0723498645880976e-07,
      "loss": 0.0073,
      "num_input_tokens_seen": 126359344,
      "step": 58545
    },
    {
      "epoch": 9.551386623164763,
      "grad_norm": 0.1945304274559021,
      "learning_rate": 3.061234904352089e-07,
      "loss": 0.1916,
      "num_input_tokens_seen": 126370064,
      "step": 58550
    },
    {
      "epoch": 9.552202283849919,
      "grad_norm": 0.18202432990074158,
      "learning_rate": 3.0501399622253344e-07,
      "loss": 0.0049,
      "num_input_tokens_seen": 126380080,
      "step": 58555
    },
    {
      "epoch": 9.553017944535073,
      "grad_norm": 0.618599534034729,
      "learning_rate": 3.0390650391072527e-07,
      "loss": 0.0072,
      "num_input_tokens_seen": 126390864,
      "step": 58560
    },
    {
      "epoch": 9.553833605220229,
      "grad_norm": 0.07423996925354004,
      "learning_rate": 3.028010135895598e-07,
      "loss": 0.0576,
      "num_input_tokens_seen": 126402224,
      "step": 58565
    },
    {
      "epoch": 9.554649265905383,
      "grad_norm": 0.09729321300983429,
      "learning_rate": 3.016975253486598e-07,
      "loss": 0.0036,
      "num_input_tokens_seen": 126412464,
      "step": 58570
    },
    {
      "epoch": 9.555464926590538,
      "grad_norm": 0.2405720204114914,
      "learning_rate": 3.0059603927747313e-07,
      "loss": 0.0124,
      "num_input_tokens_seen": 126423600,
      "step": 58575
    },
    {
      "epoch": 9.556280587275694,
      "grad_norm": 0.2697793245315552,
      "learning_rate": 2.9949655546529785e-07,
      "loss": 0.064,
      "num_input_tokens_seen": 126434928,
      "step": 58580
    },
    {
      "epoch": 9.557096247960848,
      "grad_norm": 0.12202146649360657,
      "learning_rate": 2.9839907400125986e-07,
      "loss": 0.0051,
      "num_input_tokens_seen": 126444944,
      "step": 58585
    },
    {
      "epoch": 9.557911908646004,
      "grad_norm": 0.0535927414894104,
      "learning_rate": 2.973035949743269e-07,
      "loss": 0.1052,
      "num_input_tokens_seen": 126456208,
      "step": 58590
    },
    {
      "epoch": 9.558727569331158,
      "grad_norm": 0.14638705551624298,
      "learning_rate": 2.9621011847330293e-07,
      "loss": 0.0034,
      "num_input_tokens_seen": 126467312,
      "step": 58595
    },
    {
      "epoch": 9.559543230016313,
      "grad_norm": 0.13557329773902893,
      "learning_rate": 2.951186445868337e-07,
      "loss": 0.0099,
      "num_input_tokens_seen": 126476656,
      "step": 58600
    },
    {
      "epoch": 9.560358890701469,
      "grad_norm": 0.12493931502103806,
      "learning_rate": 2.940291734034012e-07,
      "loss": 0.2067,
      "num_input_tokens_seen": 126487920,
      "step": 58605
    },
    {
      "epoch": 9.561174551386623,
      "grad_norm": 0.0669429674744606,
      "learning_rate": 2.929417050113181e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 126499216,
      "step": 58610
    },
    {
      "epoch": 9.561990212071779,
      "grad_norm": 0.11747512966394424,
      "learning_rate": 2.918562394987445e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 126509936,
      "step": 58615
    },
    {
      "epoch": 9.562805872756933,
      "grad_norm": 4.202183723449707,
      "learning_rate": 2.907727769536683e-07,
      "loss": 0.1327,
      "num_input_tokens_seen": 126521776,
      "step": 58620
    },
    {
      "epoch": 9.563621533442088,
      "grad_norm": 0.12939080595970154,
      "learning_rate": 2.8969131746392763e-07,
      "loss": 0.0051,
      "num_input_tokens_seen": 126532464,
      "step": 58625
    },
    {
      "epoch": 9.564437194127244,
      "grad_norm": 0.08323273807764053,
      "learning_rate": 2.886118611171884e-07,
      "loss": 0.0843,
      "num_input_tokens_seen": 126542992,
      "step": 58630
    },
    {
      "epoch": 9.565252854812398,
      "grad_norm": 3.485731363296509,
      "learning_rate": 2.875344080009529e-07,
      "loss": 0.1096,
      "num_input_tokens_seen": 126553680,
      "step": 58635
    },
    {
      "epoch": 9.566068515497554,
      "grad_norm": 0.17163684964179993,
      "learning_rate": 2.8645895820257065e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 126563440,
      "step": 58640
    },
    {
      "epoch": 9.566884176182707,
      "grad_norm": 0.3211774230003357,
      "learning_rate": 2.8538551180921913e-07,
      "loss": 0.1721,
      "num_input_tokens_seen": 126573904,
      "step": 58645
    },
    {
      "epoch": 9.567699836867863,
      "grad_norm": 3.841701030731201,
      "learning_rate": 2.8431406890792045e-07,
      "loss": 0.2517,
      "num_input_tokens_seen": 126584560,
      "step": 58650
    },
    {
      "epoch": 9.568515497553017,
      "grad_norm": 0.06320230662822723,
      "learning_rate": 2.8324462958552735e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 126595632,
      "step": 58655
    },
    {
      "epoch": 9.569331158238173,
      "grad_norm": 0.043477654457092285,
      "learning_rate": 2.821771939287371e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 126606800,
      "step": 58660
    },
    {
      "epoch": 9.570146818923329,
      "grad_norm": 19.05152702331543,
      "learning_rate": 2.811117620240833e-07,
      "loss": 0.102,
      "num_input_tokens_seen": 126617808,
      "step": 58665
    },
    {
      "epoch": 9.570962479608482,
      "grad_norm": 0.1006511002779007,
      "learning_rate": 2.800483339579274e-07,
      "loss": 0.2518,
      "num_input_tokens_seen": 126628720,
      "step": 58670
    },
    {
      "epoch": 9.571778140293638,
      "grad_norm": 0.11229012906551361,
      "learning_rate": 2.789869098164838e-07,
      "loss": 0.0066,
      "num_input_tokens_seen": 126639664,
      "step": 58675
    },
    {
      "epoch": 9.572593800978792,
      "grad_norm": 0.061839509755373,
      "learning_rate": 2.779274896857947e-07,
      "loss": 0.0032,
      "num_input_tokens_seen": 126650736,
      "step": 58680
    },
    {
      "epoch": 9.573409461663948,
      "grad_norm": 9.058664321899414,
      "learning_rate": 2.768700736517416e-07,
      "loss": 0.16,
      "num_input_tokens_seen": 126662224,
      "step": 58685
    },
    {
      "epoch": 9.574225122349104,
      "grad_norm": 0.0872805118560791,
      "learning_rate": 2.7581466180004454e-07,
      "loss": 0.0052,
      "num_input_tokens_seen": 126673040,
      "step": 58690
    },
    {
      "epoch": 9.575040783034257,
      "grad_norm": 0.10303211212158203,
      "learning_rate": 2.747612542162603e-07,
      "loss": 0.0087,
      "num_input_tokens_seen": 126683824,
      "step": 58695
    },
    {
      "epoch": 9.575856443719413,
      "grad_norm": 0.07268112152814865,
      "learning_rate": 2.737098509857816e-07,
      "loss": 0.0068,
      "num_input_tokens_seen": 126693488,
      "step": 58700
    },
    {
      "epoch": 9.576672104404567,
      "grad_norm": 3.7061920166015625,
      "learning_rate": 2.726604521938458e-07,
      "loss": 0.2255,
      "num_input_tokens_seen": 126703888,
      "step": 58705
    },
    {
      "epoch": 9.577487765089723,
      "grad_norm": 0.11714824289083481,
      "learning_rate": 2.716130579255155e-07,
      "loss": 0.1504,
      "num_input_tokens_seen": 126715440,
      "step": 58710
    },
    {
      "epoch": 9.578303425774878,
      "grad_norm": 0.09520381689071655,
      "learning_rate": 2.7056766826570045e-07,
      "loss": 0.2826,
      "num_input_tokens_seen": 126725680,
      "step": 58715
    },
    {
      "epoch": 9.579119086460032,
      "grad_norm": 9.55907917022705,
      "learning_rate": 2.6952428329914956e-07,
      "loss": 0.0382,
      "num_input_tokens_seen": 126736144,
      "step": 58720
    },
    {
      "epoch": 9.579934747145188,
      "grad_norm": 0.10738665610551834,
      "learning_rate": 2.684829031104397e-07,
      "loss": 0.0664,
      "num_input_tokens_seen": 126747504,
      "step": 58725
    },
    {
      "epoch": 9.580750407830342,
      "grad_norm": 0.08543211966753006,
      "learning_rate": 2.6744352778399204e-07,
      "loss": 0.0955,
      "num_input_tokens_seen": 126757904,
      "step": 58730
    },
    {
      "epoch": 9.581566068515498,
      "grad_norm": 0.12505745887756348,
      "learning_rate": 2.6640615740406436e-07,
      "loss": 0.006,
      "num_input_tokens_seen": 126768400,
      "step": 58735
    },
    {
      "epoch": 9.582381729200652,
      "grad_norm": 6.947787284851074,
      "learning_rate": 2.6537079205475323e-07,
      "loss": 0.2567,
      "num_input_tokens_seen": 126779440,
      "step": 58740
    },
    {
      "epoch": 9.583197389885807,
      "grad_norm": 3.702768564224243,
      "learning_rate": 2.6433743181998316e-07,
      "loss": 0.1049,
      "num_input_tokens_seen": 126790672,
      "step": 58745
    },
    {
      "epoch": 9.584013050570963,
      "grad_norm": 0.11286371201276779,
      "learning_rate": 2.633060767835316e-07,
      "loss": 0.0074,
      "num_input_tokens_seen": 126799824,
      "step": 58750
    },
    {
      "epoch": 9.584828711256117,
      "grad_norm": 1.5042606592178345,
      "learning_rate": 2.6227672702900106e-07,
      "loss": 0.0064,
      "num_input_tokens_seen": 126811248,
      "step": 58755
    },
    {
      "epoch": 9.585644371941273,
      "grad_norm": 2.318598508834839,
      "learning_rate": 2.61249382639836e-07,
      "loss": 0.2699,
      "num_input_tokens_seen": 126822256,
      "step": 58760
    },
    {
      "epoch": 9.586460032626427,
      "grad_norm": 0.4412936270236969,
      "learning_rate": 2.6022404369931976e-07,
      "loss": 0.1239,
      "num_input_tokens_seen": 126833840,
      "step": 58765
    },
    {
      "epoch": 9.587275693311582,
      "grad_norm": 0.15247473120689392,
      "learning_rate": 2.592007102905719e-07,
      "loss": 0.0034,
      "num_input_tokens_seen": 126846160,
      "step": 58770
    },
    {
      "epoch": 9.588091353996738,
      "grad_norm": 0.17033638060092926,
      "learning_rate": 2.581793824965484e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 126857904,
      "step": 58775
    },
    {
      "epoch": 9.588907014681892,
      "grad_norm": 0.09159134328365326,
      "learning_rate": 2.5716006040004123e-07,
      "loss": 0.0447,
      "num_input_tokens_seen": 126869232,
      "step": 58780
    },
    {
      "epoch": 9.589722675367048,
      "grad_norm": 0.021363073959946632,
      "learning_rate": 2.5614274408368444e-07,
      "loss": 0.0054,
      "num_input_tokens_seen": 126880784,
      "step": 58785
    },
    {
      "epoch": 9.590538336052202,
      "grad_norm": 10.857181549072266,
      "learning_rate": 2.5512743362994527e-07,
      "loss": 0.0184,
      "num_input_tokens_seen": 126891600,
      "step": 58790
    },
    {
      "epoch": 9.591353996737357,
      "grad_norm": 0.3296675980091095,
      "learning_rate": 2.541141291211302e-07,
      "loss": 0.1118,
      "num_input_tokens_seen": 126903056,
      "step": 58795
    },
    {
      "epoch": 9.592169657422513,
      "grad_norm": 0.08449359238147736,
      "learning_rate": 2.5310283063938457e-07,
      "loss": 0.0585,
      "num_input_tokens_seen": 126914032,
      "step": 58800
    },
    {
      "epoch": 9.592985318107667,
      "grad_norm": 0.07261674106121063,
      "learning_rate": 2.5209353826668726e-07,
      "loss": 0.2212,
      "num_input_tokens_seen": 126923920,
      "step": 58805
    },
    {
      "epoch": 9.593800978792823,
      "grad_norm": 0.1694912612438202,
      "learning_rate": 2.510862520848589e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 126935216,
      "step": 58810
    },
    {
      "epoch": 9.594616639477977,
      "grad_norm": 0.10771431773900986,
      "learning_rate": 2.500809721755509e-07,
      "loss": 0.0042,
      "num_input_tokens_seen": 126944752,
      "step": 58815
    },
    {
      "epoch": 9.595432300163132,
      "grad_norm": 9.962686538696289,
      "learning_rate": 2.490776986202592e-07,
      "loss": 0.0914,
      "num_input_tokens_seen": 126955856,
      "step": 58820
    },
    {
      "epoch": 9.596247960848288,
      "grad_norm": 0.11950484663248062,
      "learning_rate": 2.480764315003159e-07,
      "loss": 0.0081,
      "num_input_tokens_seen": 126966064,
      "step": 58825
    },
    {
      "epoch": 9.597063621533442,
      "grad_norm": 0.024678900837898254,
      "learning_rate": 2.470771708968866e-07,
      "loss": 0.116,
      "num_input_tokens_seen": 126977424,
      "step": 58830
    },
    {
      "epoch": 9.597879282218598,
      "grad_norm": 3.2517993450164795,
      "learning_rate": 2.4607991689097607e-07,
      "loss": 0.1095,
      "num_input_tokens_seen": 126987728,
      "step": 58835
    },
    {
      "epoch": 9.598694942903752,
      "grad_norm": 0.09872573614120483,
      "learning_rate": 2.4508466956343066e-07,
      "loss": 0.1991,
      "num_input_tokens_seen": 126999536,
      "step": 58840
    },
    {
      "epoch": 9.599510603588907,
      "grad_norm": 0.10038405656814575,
      "learning_rate": 2.4409142899492474e-07,
      "loss": 0.1479,
      "num_input_tokens_seen": 127010384,
      "step": 58845
    },
    {
      "epoch": 9.600326264274061,
      "grad_norm": 0.21252353489398956,
      "learning_rate": 2.4310019526597726e-07,
      "loss": 0.0055,
      "num_input_tokens_seen": 127020176,
      "step": 58850
    },
    {
      "epoch": 9.601141924959217,
      "grad_norm": 0.216371089220047,
      "learning_rate": 2.4211096845694336e-07,
      "loss": 0.0673,
      "num_input_tokens_seen": 127030192,
      "step": 58855
    },
    {
      "epoch": 9.601957585644373,
      "grad_norm": 0.5040737390518188,
      "learning_rate": 2.411237486480145e-07,
      "loss": 0.0073,
      "num_input_tokens_seen": 127041808,
      "step": 58860
    },
    {
      "epoch": 9.602773246329527,
      "grad_norm": 0.09158005565404892,
      "learning_rate": 2.4013853591922097e-07,
      "loss": 0.089,
      "num_input_tokens_seen": 127052400,
      "step": 58865
    },
    {
      "epoch": 9.603588907014682,
      "grad_norm": 0.07197082042694092,
      "learning_rate": 2.391553303504296e-07,
      "loss": 0.1119,
      "num_input_tokens_seen": 127063056,
      "step": 58870
    },
    {
      "epoch": 9.604404567699836,
      "grad_norm": 0.12315355986356735,
      "learning_rate": 2.3817413202134041e-07,
      "loss": 0.094,
      "num_input_tokens_seen": 127073840,
      "step": 58875
    },
    {
      "epoch": 9.605220228384992,
      "grad_norm": 0.031999364495277405,
      "learning_rate": 2.3719494101149543e-07,
      "loss": 0.079,
      "num_input_tokens_seen": 127084656,
      "step": 58880
    },
    {
      "epoch": 9.606035889070148,
      "grad_norm": 0.05075620487332344,
      "learning_rate": 2.3621775740027553e-07,
      "loss": 0.008,
      "num_input_tokens_seen": 127095984,
      "step": 58885
    },
    {
      "epoch": 9.606851549755302,
      "grad_norm": 0.030352383852005005,
      "learning_rate": 2.3524258126689235e-07,
      "loss": 0.0222,
      "num_input_tokens_seen": 127105776,
      "step": 58890
    },
    {
      "epoch": 9.607667210440457,
      "grad_norm": 0.060310106724500656,
      "learning_rate": 2.3426941269040213e-07,
      "loss": 0.1309,
      "num_input_tokens_seen": 127117424,
      "step": 58895
    },
    {
      "epoch": 9.608482871125611,
      "grad_norm": 3.352529764175415,
      "learning_rate": 2.3329825174969455e-07,
      "loss": 0.0962,
      "num_input_tokens_seen": 127128240,
      "step": 58900
    },
    {
      "epoch": 9.609298531810767,
      "grad_norm": 0.132798433303833,
      "learning_rate": 2.3232909852349273e-07,
      "loss": 0.0089,
      "num_input_tokens_seen": 127140176,
      "step": 58905
    },
    {
      "epoch": 9.61011419249592,
      "grad_norm": 0.14223125576972961,
      "learning_rate": 2.3136195309036435e-07,
      "loss": 0.0106,
      "num_input_tokens_seen": 127149392,
      "step": 58910
    },
    {
      "epoch": 9.610929853181077,
      "grad_norm": 0.08127401024103165,
      "learning_rate": 2.303968155287134e-07,
      "loss": 0.0066,
      "num_input_tokens_seen": 127160496,
      "step": 58915
    },
    {
      "epoch": 9.611745513866232,
      "grad_norm": 1.1695650815963745,
      "learning_rate": 2.294336859167745e-07,
      "loss": 0.0062,
      "num_input_tokens_seen": 127170544,
      "step": 58920
    },
    {
      "epoch": 9.612561174551386,
      "grad_norm": 0.03738000616431236,
      "learning_rate": 2.2847256433262686e-07,
      "loss": 0.1002,
      "num_input_tokens_seen": 127182992,
      "step": 58925
    },
    {
      "epoch": 9.613376835236542,
      "grad_norm": 0.07341515272855759,
      "learning_rate": 2.2751345085418042e-07,
      "loss": 0.1272,
      "num_input_tokens_seen": 127194320,
      "step": 58930
    },
    {
      "epoch": 9.614192495921696,
      "grad_norm": 0.09440329670906067,
      "learning_rate": 2.265563455591896e-07,
      "loss": 0.1013,
      "num_input_tokens_seen": 127205776,
      "step": 58935
    },
    {
      "epoch": 9.615008156606851,
      "grad_norm": 5.581656455993652,
      "learning_rate": 2.2560124852523955e-07,
      "loss": 0.1296,
      "num_input_tokens_seen": 127217264,
      "step": 58940
    },
    {
      "epoch": 9.615823817292007,
      "grad_norm": 0.07413234561681747,
      "learning_rate": 2.246481598297573e-07,
      "loss": 0.1324,
      "num_input_tokens_seen": 127229584,
      "step": 58945
    },
    {
      "epoch": 9.616639477977161,
      "grad_norm": 0.14779232442378998,
      "learning_rate": 2.2369707955000318e-07,
      "loss": 0.126,
      "num_input_tokens_seen": 127239824,
      "step": 58950
    },
    {
      "epoch": 9.617455138662317,
      "grad_norm": 0.09638264775276184,
      "learning_rate": 2.2274800776307946e-07,
      "loss": 0.082,
      "num_input_tokens_seen": 127250672,
      "step": 58955
    },
    {
      "epoch": 9.61827079934747,
      "grad_norm": 0.26495882868766785,
      "learning_rate": 2.2180094454591903e-07,
      "loss": 0.1262,
      "num_input_tokens_seen": 127260496,
      "step": 58960
    },
    {
      "epoch": 9.619086460032626,
      "grad_norm": 0.05389063432812691,
      "learning_rate": 2.2085588997529938e-07,
      "loss": 0.1803,
      "num_input_tokens_seen": 127272336,
      "step": 58965
    },
    {
      "epoch": 9.619902120717782,
      "grad_norm": 0.4798796772956848,
      "learning_rate": 2.1991284412782864e-07,
      "loss": 0.0042,
      "num_input_tokens_seen": 127284368,
      "step": 58970
    },
    {
      "epoch": 9.620717781402936,
      "grad_norm": 0.043308209627866745,
      "learning_rate": 2.18971807079954e-07,
      "loss": 0.0773,
      "num_input_tokens_seen": 127294032,
      "step": 58975
    },
    {
      "epoch": 9.621533442088092,
      "grad_norm": 0.06999657303094864,
      "learning_rate": 2.1803277890796447e-07,
      "loss": 0.1303,
      "num_input_tokens_seen": 127304080,
      "step": 58980
    },
    {
      "epoch": 9.622349102773246,
      "grad_norm": 7.49120569229126,
      "learning_rate": 2.170957596879797e-07,
      "loss": 0.0539,
      "num_input_tokens_seen": 127315888,
      "step": 58985
    },
    {
      "epoch": 9.623164763458401,
      "grad_norm": 0.24277690052986145,
      "learning_rate": 2.1616074949595832e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 127326864,
      "step": 58990
    },
    {
      "epoch": 9.623980424143557,
      "grad_norm": 0.12107644230127335,
      "learning_rate": 2.1522774840770087e-07,
      "loss": 0.011,
      "num_input_tokens_seen": 127336368,
      "step": 58995
    },
    {
      "epoch": 9.624796084828711,
      "grad_norm": 13.174209594726562,
      "learning_rate": 2.1429675649883575e-07,
      "loss": 0.0277,
      "num_input_tokens_seen": 127347824,
      "step": 59000
    },
    {
      "epoch": 9.625611745513867,
      "grad_norm": 3.4110028743743896,
      "learning_rate": 2.1336777384484141e-07,
      "loss": 0.1964,
      "num_input_tokens_seen": 127356656,
      "step": 59005
    },
    {
      "epoch": 9.62642740619902,
      "grad_norm": 0.1018814742565155,
      "learning_rate": 2.1244080052101879e-07,
      "loss": 0.2314,
      "num_input_tokens_seen": 127367120,
      "step": 59010
    },
    {
      "epoch": 9.627243066884176,
      "grad_norm": 0.8009899854660034,
      "learning_rate": 2.115158366025133e-07,
      "loss": 0.0108,
      "num_input_tokens_seen": 127378800,
      "step": 59015
    },
    {
      "epoch": 9.62805872756933,
      "grad_norm": 3.683324098587036,
      "learning_rate": 2.1059288216431217e-07,
      "loss": 0.1045,
      "num_input_tokens_seen": 127390000,
      "step": 59020
    },
    {
      "epoch": 9.628874388254486,
      "grad_norm": 0.12710148096084595,
      "learning_rate": 2.0967193728123334e-07,
      "loss": 0.0032,
      "num_input_tokens_seen": 127400144,
      "step": 59025
    },
    {
      "epoch": 9.629690048939642,
      "grad_norm": 0.03041822463274002,
      "learning_rate": 2.0875300202793101e-07,
      "loss": 0.0743,
      "num_input_tokens_seen": 127411568,
      "step": 59030
    },
    {
      "epoch": 9.630505709624796,
      "grad_norm": 0.14042043685913086,
      "learning_rate": 2.0783607647889837e-07,
      "loss": 0.0704,
      "num_input_tokens_seen": 127421520,
      "step": 59035
    },
    {
      "epoch": 9.631321370309951,
      "grad_norm": 0.14955595135688782,
      "learning_rate": 2.0692116070847035e-07,
      "loss": 0.1032,
      "num_input_tokens_seen": 127433424,
      "step": 59040
    },
    {
      "epoch": 9.632137030995105,
      "grad_norm": 0.15251396596431732,
      "learning_rate": 2.0600825479080986e-07,
      "loss": 0.0048,
      "num_input_tokens_seen": 127443408,
      "step": 59045
    },
    {
      "epoch": 9.632952691680261,
      "grad_norm": 6.020075798034668,
      "learning_rate": 2.0509735879992442e-07,
      "loss": 0.0084,
      "num_input_tokens_seen": 127454256,
      "step": 59050
    },
    {
      "epoch": 9.633768352365417,
      "grad_norm": 0.0663226842880249,
      "learning_rate": 2.041884728096549e-07,
      "loss": 0.1546,
      "num_input_tokens_seen": 127464624,
      "step": 59055
    },
    {
      "epoch": 9.63458401305057,
      "grad_norm": 0.07296096533536911,
      "learning_rate": 2.0328159689368133e-07,
      "loss": 0.0904,
      "num_input_tokens_seen": 127474192,
      "step": 59060
    },
    {
      "epoch": 9.635399673735726,
      "grad_norm": 2.504868984222412,
      "learning_rate": 2.0237673112551704e-07,
      "loss": 0.0083,
      "num_input_tokens_seen": 127483056,
      "step": 59065
    },
    {
      "epoch": 9.63621533442088,
      "grad_norm": 0.22530889511108398,
      "learning_rate": 2.0147387557851727e-07,
      "loss": 0.1021,
      "num_input_tokens_seen": 127493360,
      "step": 59070
    },
    {
      "epoch": 9.637030995106036,
      "grad_norm": 0.17570574581623077,
      "learning_rate": 2.005730303258735e-07,
      "loss": 0.1274,
      "num_input_tokens_seen": 127503728,
      "step": 59075
    },
    {
      "epoch": 9.63784665579119,
      "grad_norm": 0.11782146990299225,
      "learning_rate": 1.9967419544060784e-07,
      "loss": 0.0038,
      "num_input_tokens_seen": 127514672,
      "step": 59080
    },
    {
      "epoch": 9.638662316476346,
      "grad_norm": 0.03208833560347557,
      "learning_rate": 1.987773709955898e-07,
      "loss": 0.0049,
      "num_input_tokens_seen": 127525200,
      "step": 59085
    },
    {
      "epoch": 9.639477977161501,
      "grad_norm": 0.13830067217350006,
      "learning_rate": 1.9788255706351678e-07,
      "loss": 0.1041,
      "num_input_tokens_seen": 127535824,
      "step": 59090
    },
    {
      "epoch": 9.640293637846655,
      "grad_norm": 0.050570495426654816,
      "learning_rate": 1.9698975371693075e-07,
      "loss": 0.0148,
      "num_input_tokens_seen": 127546960,
      "step": 59095
    },
    {
      "epoch": 9.641109298531811,
      "grad_norm": 0.31174710392951965,
      "learning_rate": 1.9609896102820157e-07,
      "loss": 0.0097,
      "num_input_tokens_seen": 127557680,
      "step": 59100
    },
    {
      "epoch": 9.641924959216965,
      "grad_norm": 0.3156566321849823,
      "learning_rate": 1.9521017906954654e-07,
      "loss": 0.0218,
      "num_input_tokens_seen": 127567696,
      "step": 59105
    },
    {
      "epoch": 9.64274061990212,
      "grad_norm": 0.08258449286222458,
      "learning_rate": 1.9432340791301073e-07,
      "loss": 0.1024,
      "num_input_tokens_seen": 127578736,
      "step": 59110
    },
    {
      "epoch": 9.643556280587276,
      "grad_norm": 7.100406169891357,
      "learning_rate": 1.9343864763048392e-07,
      "loss": 0.0236,
      "num_input_tokens_seen": 127590224,
      "step": 59115
    },
    {
      "epoch": 9.64437194127243,
      "grad_norm": 0.13057167828083038,
      "learning_rate": 1.925558982936865e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 127601328,
      "step": 59120
    },
    {
      "epoch": 9.645187601957586,
      "grad_norm": 0.20511381328105927,
      "learning_rate": 1.916751599741806e-07,
      "loss": 0.0041,
      "num_input_tokens_seen": 127612464,
      "step": 59125
    },
    {
      "epoch": 9.64600326264274,
      "grad_norm": 0.11308741569519043,
      "learning_rate": 1.90796432743362e-07,
      "loss": 0.0048,
      "num_input_tokens_seen": 127622864,
      "step": 59130
    },
    {
      "epoch": 9.646818923327896,
      "grad_norm": 6.885258674621582,
      "learning_rate": 1.8991971667246533e-07,
      "loss": 0.0504,
      "num_input_tokens_seen": 127634000,
      "step": 59135
    },
    {
      "epoch": 9.647634584013051,
      "grad_norm": 0.5002412796020508,
      "learning_rate": 1.8904501183256152e-07,
      "loss": 0.0101,
      "num_input_tokens_seen": 127645264,
      "step": 59140
    },
    {
      "epoch": 9.648450244698205,
      "grad_norm": 0.28994885087013245,
      "learning_rate": 1.8817231829455773e-07,
      "loss": 0.0049,
      "num_input_tokens_seen": 127655376,
      "step": 59145
    },
    {
      "epoch": 9.649265905383361,
      "grad_norm": 0.05022817477583885,
      "learning_rate": 1.8730163612920015e-07,
      "loss": 0.2531,
      "num_input_tokens_seen": 127665584,
      "step": 59150
    },
    {
      "epoch": 9.650081566068515,
      "grad_norm": 0.08764076977968216,
      "learning_rate": 1.8643296540707121e-07,
      "loss": 0.1174,
      "num_input_tokens_seen": 127675920,
      "step": 59155
    },
    {
      "epoch": 9.65089722675367,
      "grad_norm": 0.13821817934513092,
      "learning_rate": 1.855663061985896e-07,
      "loss": 0.005,
      "num_input_tokens_seen": 127686928,
      "step": 59160
    },
    {
      "epoch": 9.651712887438826,
      "grad_norm": 0.11403993517160416,
      "learning_rate": 1.8470165857401023e-07,
      "loss": 0.0042,
      "num_input_tokens_seen": 127697296,
      "step": 59165
    },
    {
      "epoch": 9.65252854812398,
      "grad_norm": 0.06376594305038452,
      "learning_rate": 1.8383902260342422e-07,
      "loss": 0.1246,
      "num_input_tokens_seen": 127707312,
      "step": 59170
    },
    {
      "epoch": 9.653344208809136,
      "grad_norm": 0.2896687388420105,
      "learning_rate": 1.8297839835676456e-07,
      "loss": 0.0052,
      "num_input_tokens_seen": 127719056,
      "step": 59175
    },
    {
      "epoch": 9.65415986949429,
      "grad_norm": 0.22857573628425598,
      "learning_rate": 1.8211978590379486e-07,
      "loss": 0.092,
      "num_input_tokens_seen": 127730864,
      "step": 59180
    },
    {
      "epoch": 9.654975530179446,
      "grad_norm": 10.07347297668457,
      "learning_rate": 1.8126318531412056e-07,
      "loss": 0.0627,
      "num_input_tokens_seen": 127740944,
      "step": 59185
    },
    {
      "epoch": 9.655791190864601,
      "grad_norm": 0.03637094795703888,
      "learning_rate": 1.8040859665718057e-07,
      "loss": 0.0032,
      "num_input_tokens_seen": 127752016,
      "step": 59190
    },
    {
      "epoch": 9.656606851549755,
      "grad_norm": 0.07079402357339859,
      "learning_rate": 1.795560200022528e-07,
      "loss": 0.0033,
      "num_input_tokens_seen": 127762480,
      "step": 59195
    },
    {
      "epoch": 9.65742251223491,
      "grad_norm": 0.0924726203083992,
      "learning_rate": 1.7870545541845418e-07,
      "loss": 0.095,
      "num_input_tokens_seen": 127773232,
      "step": 59200
    },
    {
      "epoch": 9.658238172920065,
      "grad_norm": 0.10256163775920868,
      "learning_rate": 1.7785690297473234e-07,
      "loss": 0.0082,
      "num_input_tokens_seen": 127784112,
      "step": 59205
    },
    {
      "epoch": 9.65905383360522,
      "grad_norm": 0.09814684092998505,
      "learning_rate": 1.770103627398767e-07,
      "loss": 0.1416,
      "num_input_tokens_seen": 127795600,
      "step": 59210
    },
    {
      "epoch": 9.659869494290374,
      "grad_norm": 6.013543605804443,
      "learning_rate": 1.7616583478251013e-07,
      "loss": 0.1384,
      "num_input_tokens_seen": 127805648,
      "step": 59215
    },
    {
      "epoch": 9.66068515497553,
      "grad_norm": 1.4875679016113281,
      "learning_rate": 1.7532331917109457e-07,
      "loss": 0.014,
      "num_input_tokens_seen": 127816560,
      "step": 59220
    },
    {
      "epoch": 9.661500815660686,
      "grad_norm": 0.11682610213756561,
      "learning_rate": 1.7448281597393368e-07,
      "loss": 0.1847,
      "num_input_tokens_seen": 127828784,
      "step": 59225
    },
    {
      "epoch": 9.66231647634584,
      "grad_norm": 0.35796356201171875,
      "learning_rate": 1.736443252591563e-07,
      "loss": 0.0053,
      "num_input_tokens_seen": 127839088,
      "step": 59230
    },
    {
      "epoch": 9.663132137030995,
      "grad_norm": 0.2201497107744217,
      "learning_rate": 1.7280784709473862e-07,
      "loss": 0.0703,
      "num_input_tokens_seen": 127848016,
      "step": 59235
    },
    {
      "epoch": 9.66394779771615,
      "grad_norm": 0.05457500368356705,
      "learning_rate": 1.719733815484903e-07,
      "loss": 0.0819,
      "num_input_tokens_seen": 127858320,
      "step": 59240
    },
    {
      "epoch": 9.664763458401305,
      "grad_norm": 0.2189728021621704,
      "learning_rate": 1.7114092868805443e-07,
      "loss": 0.2399,
      "num_input_tokens_seen": 127869936,
      "step": 59245
    },
    {
      "epoch": 9.66557911908646,
      "grad_norm": 0.20777510106563568,
      "learning_rate": 1.7031048858091313e-07,
      "loss": 0.008,
      "num_input_tokens_seen": 127881040,
      "step": 59250
    },
    {
      "epoch": 9.666394779771615,
      "grad_norm": 0.15500149130821228,
      "learning_rate": 1.6948206129439037e-07,
      "loss": 0.2172,
      "num_input_tokens_seen": 127891792,
      "step": 59255
    },
    {
      "epoch": 9.66721044045677,
      "grad_norm": 3.3623745441436768,
      "learning_rate": 1.6865564689564074e-07,
      "loss": 0.2944,
      "num_input_tokens_seen": 127903376,
      "step": 59260
    },
    {
      "epoch": 9.668026101141924,
      "grad_norm": 0.15240967273712158,
      "learning_rate": 1.6783124545165785e-07,
      "loss": 0.0104,
      "num_input_tokens_seen": 127914576,
      "step": 59265
    },
    {
      "epoch": 9.66884176182708,
      "grad_norm": 0.05557018890976906,
      "learning_rate": 1.6700885702926882e-07,
      "loss": 0.0044,
      "num_input_tokens_seen": 127924784,
      "step": 59270
    },
    {
      "epoch": 9.669657422512234,
      "grad_norm": 0.7313627600669861,
      "learning_rate": 1.6618848169514533e-07,
      "loss": 0.0936,
      "num_input_tokens_seen": 127934800,
      "step": 59275
    },
    {
      "epoch": 9.67047308319739,
      "grad_norm": 0.08608107268810272,
      "learning_rate": 1.6537011951578974e-07,
      "loss": 0.1346,
      "num_input_tokens_seen": 127945584,
      "step": 59280
    },
    {
      "epoch": 9.671288743882545,
      "grad_norm": 0.16615554690361023,
      "learning_rate": 1.645537705575406e-07,
      "loss": 0.1261,
      "num_input_tokens_seen": 127956272,
      "step": 59285
    },
    {
      "epoch": 9.6721044045677,
      "grad_norm": 0.07968626916408539,
      "learning_rate": 1.6373943488657562e-07,
      "loss": 0.0086,
      "num_input_tokens_seen": 127966672,
      "step": 59290
    },
    {
      "epoch": 9.672920065252855,
      "grad_norm": 4.198758125305176,
      "learning_rate": 1.6292711256891134e-07,
      "loss": 0.321,
      "num_input_tokens_seen": 127978032,
      "step": 59295
    },
    {
      "epoch": 9.673735725938009,
      "grad_norm": 0.4226551949977875,
      "learning_rate": 1.6211680367039793e-07,
      "loss": 0.008,
      "num_input_tokens_seen": 127989072,
      "step": 59300
    },
    {
      "epoch": 9.674551386623165,
      "grad_norm": 0.027734851464629173,
      "learning_rate": 1.6130850825672173e-07,
      "loss": 0.0023,
      "num_input_tokens_seen": 127999568,
      "step": 59305
    },
    {
      "epoch": 9.67536704730832,
      "grad_norm": 0.19730962812900543,
      "learning_rate": 1.6050222639340807e-07,
      "loss": 0.0055,
      "num_input_tokens_seen": 128010256,
      "step": 59310
    },
    {
      "epoch": 9.676182707993474,
      "grad_norm": 0.058019958436489105,
      "learning_rate": 1.5969795814581856e-07,
      "loss": 0.0036,
      "num_input_tokens_seen": 128020176,
      "step": 59315
    },
    {
      "epoch": 9.67699836867863,
      "grad_norm": 0.13013321161270142,
      "learning_rate": 1.5889570357915108e-07,
      "loss": 0.0796,
      "num_input_tokens_seen": 128032496,
      "step": 59320
    },
    {
      "epoch": 9.677814029363784,
      "grad_norm": 0.06711160391569138,
      "learning_rate": 1.5809546275843968e-07,
      "loss": 0.0038,
      "num_input_tokens_seen": 128043632,
      "step": 59325
    },
    {
      "epoch": 9.67862969004894,
      "grad_norm": 0.03470923751592636,
      "learning_rate": 1.572972357485575e-07,
      "loss": 0.1232,
      "num_input_tokens_seen": 128054608,
      "step": 59330
    },
    {
      "epoch": 9.679445350734095,
      "grad_norm": 0.10599584132432938,
      "learning_rate": 1.5650102261421107e-07,
      "loss": 0.0508,
      "num_input_tokens_seen": 128065360,
      "step": 59335
    },
    {
      "epoch": 9.68026101141925,
      "grad_norm": 0.054533615708351135,
      "learning_rate": 1.557068234199488e-07,
      "loss": 0.1016,
      "num_input_tokens_seen": 128076464,
      "step": 59340
    },
    {
      "epoch": 9.681076672104405,
      "grad_norm": 6.840791702270508,
      "learning_rate": 1.5491463823014697e-07,
      "loss": 0.0148,
      "num_input_tokens_seen": 128086320,
      "step": 59345
    },
    {
      "epoch": 9.681892332789559,
      "grad_norm": 0.11100902408361435,
      "learning_rate": 1.5412446710902917e-07,
      "loss": 0.0075,
      "num_input_tokens_seen": 128096208,
      "step": 59350
    },
    {
      "epoch": 9.682707993474715,
      "grad_norm": 0.47080928087234497,
      "learning_rate": 1.5333631012064698e-07,
      "loss": 0.1193,
      "num_input_tokens_seen": 128107376,
      "step": 59355
    },
    {
      "epoch": 9.68352365415987,
      "grad_norm": 0.08434358239173889,
      "learning_rate": 1.5255016732889648e-07,
      "loss": 0.0066,
      "num_input_tokens_seen": 128118800,
      "step": 59360
    },
    {
      "epoch": 9.684339314845024,
      "grad_norm": 0.2086295783519745,
      "learning_rate": 1.5176603879750173e-07,
      "loss": 0.2222,
      "num_input_tokens_seen": 128131152,
      "step": 59365
    },
    {
      "epoch": 9.68515497553018,
      "grad_norm": 0.1977551281452179,
      "learning_rate": 1.509839245900313e-07,
      "loss": 0.0075,
      "num_input_tokens_seen": 128142480,
      "step": 59370
    },
    {
      "epoch": 9.685970636215334,
      "grad_norm": 1.4557702541351318,
      "learning_rate": 1.5020382476988726e-07,
      "loss": 0.0072,
      "num_input_tokens_seen": 128152816,
      "step": 59375
    },
    {
      "epoch": 9.68678629690049,
      "grad_norm": 0.22527346014976501,
      "learning_rate": 1.4942573940030791e-07,
      "loss": 0.0064,
      "num_input_tokens_seen": 128163632,
      "step": 59380
    },
    {
      "epoch": 9.687601957585644,
      "grad_norm": 0.10114660859107971,
      "learning_rate": 1.4864966854437056e-07,
      "loss": 0.0034,
      "num_input_tokens_seen": 128174640,
      "step": 59385
    },
    {
      "epoch": 9.6884176182708,
      "grad_norm": 6.902266502380371,
      "learning_rate": 1.4787561226498048e-07,
      "loss": 0.1922,
      "num_input_tokens_seen": 128184144,
      "step": 59390
    },
    {
      "epoch": 9.689233278955955,
      "grad_norm": 0.05543965846300125,
      "learning_rate": 1.4710357062489577e-07,
      "loss": 0.0826,
      "num_input_tokens_seen": 128192976,
      "step": 59395
    },
    {
      "epoch": 9.690048939641109,
      "grad_norm": 0.04904047027230263,
      "learning_rate": 1.4633354368669694e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 128204144,
      "step": 59400
    },
    {
      "epoch": 9.690864600326265,
      "grad_norm": 0.5166468024253845,
      "learning_rate": 1.4556553151280628e-07,
      "loss": 0.0083,
      "num_input_tokens_seen": 128215600,
      "step": 59405
    },
    {
      "epoch": 9.691680261011419,
      "grad_norm": 0.11157098412513733,
      "learning_rate": 1.447995341654851e-07,
      "loss": 0.1073,
      "num_input_tokens_seen": 128226672,
      "step": 59410
    },
    {
      "epoch": 9.692495921696574,
      "grad_norm": 3.7563095092773438,
      "learning_rate": 1.4403555170682816e-07,
      "loss": 0.1257,
      "num_input_tokens_seen": 128237936,
      "step": 59415
    },
    {
      "epoch": 9.69331158238173,
      "grad_norm": 0.1909668743610382,
      "learning_rate": 1.4327358419876646e-07,
      "loss": 0.1063,
      "num_input_tokens_seen": 128248688,
      "step": 59420
    },
    {
      "epoch": 9.694127243066884,
      "grad_norm": 0.1627446413040161,
      "learning_rate": 1.4251363170307008e-07,
      "loss": 0.0057,
      "num_input_tokens_seen": 128259440,
      "step": 59425
    },
    {
      "epoch": 9.69494290375204,
      "grad_norm": 0.10831242799758911,
      "learning_rate": 1.4175569428134527e-07,
      "loss": 0.0613,
      "num_input_tokens_seen": 128268880,
      "step": 59430
    },
    {
      "epoch": 9.695758564437194,
      "grad_norm": 0.07561370730400085,
      "learning_rate": 1.4099977199503178e-07,
      "loss": 0.0046,
      "num_input_tokens_seen": 128277744,
      "step": 59435
    },
    {
      "epoch": 9.69657422512235,
      "grad_norm": 17.639007568359375,
      "learning_rate": 1.4024586490540837e-07,
      "loss": 0.094,
      "num_input_tokens_seen": 128288176,
      "step": 59440
    },
    {
      "epoch": 9.697389885807503,
      "grad_norm": 0.12820494174957275,
      "learning_rate": 1.3949397307359557e-07,
      "loss": 0.0041,
      "num_input_tokens_seen": 128298672,
      "step": 59445
    },
    {
      "epoch": 9.698205546492659,
      "grad_norm": 0.12352553755044937,
      "learning_rate": 1.3874409656054189e-07,
      "loss": 0.0053,
      "num_input_tokens_seen": 128309264,
      "step": 59450
    },
    {
      "epoch": 9.699021207177815,
      "grad_norm": 0.5875067114830017,
      "learning_rate": 1.3799623542703478e-07,
      "loss": 0.1419,
      "num_input_tokens_seen": 128320656,
      "step": 59455
    },
    {
      "epoch": 9.699836867862969,
      "grad_norm": 0.15500937402248383,
      "learning_rate": 1.3725038973370076e-07,
      "loss": 0.0062,
      "num_input_tokens_seen": 128332016,
      "step": 59460
    },
    {
      "epoch": 9.700652528548124,
      "grad_norm": 4.622019290924072,
      "learning_rate": 1.3650655954100532e-07,
      "loss": 0.009,
      "num_input_tokens_seen": 128342448,
      "step": 59465
    },
    {
      "epoch": 9.701468189233278,
      "grad_norm": 0.28856053948402405,
      "learning_rate": 1.3576474490924195e-07,
      "loss": 0.0933,
      "num_input_tokens_seen": 128352368,
      "step": 59470
    },
    {
      "epoch": 9.702283849918434,
      "grad_norm": 8.22559928894043,
      "learning_rate": 1.3502494589855142e-07,
      "loss": 0.3689,
      "num_input_tokens_seen": 128361488,
      "step": 59475
    },
    {
      "epoch": 9.70309951060359,
      "grad_norm": 0.09286815673112869,
      "learning_rate": 1.3428716256889962e-07,
      "loss": 0.215,
      "num_input_tokens_seen": 128371632,
      "step": 59480
    },
    {
      "epoch": 9.703915171288743,
      "grad_norm": 0.10720881074666977,
      "learning_rate": 1.3355139498009706e-07,
      "loss": 0.2728,
      "num_input_tokens_seen": 128380688,
      "step": 59485
    },
    {
      "epoch": 9.7047308319739,
      "grad_norm": 0.21968218684196472,
      "learning_rate": 1.3281764319179046e-07,
      "loss": 0.205,
      "num_input_tokens_seen": 128390480,
      "step": 59490
    },
    {
      "epoch": 9.705546492659053,
      "grad_norm": 0.3598906397819519,
      "learning_rate": 1.320859072634628e-07,
      "loss": 0.2241,
      "num_input_tokens_seen": 128401712,
      "step": 59495
    },
    {
      "epoch": 9.706362153344209,
      "grad_norm": 0.06498785316944122,
      "learning_rate": 1.3135618725442778e-07,
      "loss": 0.0063,
      "num_input_tokens_seen": 128413040,
      "step": 59500
    },
    {
      "epoch": 9.707177814029365,
      "grad_norm": 0.0967356488108635,
      "learning_rate": 1.3062848322384357e-07,
      "loss": 0.1098,
      "num_input_tokens_seen": 128421744,
      "step": 59505
    },
    {
      "epoch": 9.707993474714518,
      "grad_norm": 0.4873754382133484,
      "learning_rate": 1.2990279523069916e-07,
      "loss": 0.1651,
      "num_input_tokens_seen": 128432528,
      "step": 59510
    },
    {
      "epoch": 9.708809135399674,
      "grad_norm": 0.04376742243766785,
      "learning_rate": 1.291791233338252e-07,
      "loss": 0.0069,
      "num_input_tokens_seen": 128443664,
      "step": 59515
    },
    {
      "epoch": 9.709624796084828,
      "grad_norm": 0.03959466516971588,
      "learning_rate": 1.2845746759188314e-07,
      "loss": 0.0029,
      "num_input_tokens_seen": 128454256,
      "step": 59520
    },
    {
      "epoch": 9.710440456769984,
      "grad_norm": 0.044352855533361435,
      "learning_rate": 1.277378280633762e-07,
      "loss": 0.0193,
      "num_input_tokens_seen": 128466768,
      "step": 59525
    },
    {
      "epoch": 9.71125611745514,
      "grad_norm": 0.2385054975748062,
      "learning_rate": 1.27020204806641e-07,
      "loss": 0.1477,
      "num_input_tokens_seen": 128477264,
      "step": 59530
    },
    {
      "epoch": 9.712071778140293,
      "grad_norm": 11.07192325592041,
      "learning_rate": 1.2630459787985326e-07,
      "loss": 0.1438,
      "num_input_tokens_seen": 128489040,
      "step": 59535
    },
    {
      "epoch": 9.71288743882545,
      "grad_norm": 0.10194293409585953,
      "learning_rate": 1.2559100734102214e-07,
      "loss": 0.0041,
      "num_input_tokens_seen": 128500176,
      "step": 59540
    },
    {
      "epoch": 9.713703099510603,
      "grad_norm": 0.15271548926830292,
      "learning_rate": 1.248794332479958e-07,
      "loss": 0.1518,
      "num_input_tokens_seen": 128510096,
      "step": 59545
    },
    {
      "epoch": 9.714518760195759,
      "grad_norm": 0.10981497168540955,
      "learning_rate": 1.2416987565845861e-07,
      "loss": 0.0938,
      "num_input_tokens_seen": 128520976,
      "step": 59550
    },
    {
      "epoch": 9.715334420880914,
      "grad_norm": 0.0959286168217659,
      "learning_rate": 1.2346233462992852e-07,
      "loss": 0.2037,
      "num_input_tokens_seen": 128532080,
      "step": 59555
    },
    {
      "epoch": 9.716150081566068,
      "grad_norm": 0.12493567168712616,
      "learning_rate": 1.2275681021976515e-07,
      "loss": 0.154,
      "num_input_tokens_seen": 128544176,
      "step": 59560
    },
    {
      "epoch": 9.716965742251224,
      "grad_norm": 0.1780611276626587,
      "learning_rate": 1.220533024851589e-07,
      "loss": 0.0787,
      "num_input_tokens_seen": 128555472,
      "step": 59565
    },
    {
      "epoch": 9.717781402936378,
      "grad_norm": 0.04792303219437599,
      "learning_rate": 1.213518114831419e-07,
      "loss": 0.0298,
      "num_input_tokens_seen": 128565968,
      "step": 59570
    },
    {
      "epoch": 9.718597063621534,
      "grad_norm": 4.539177417755127,
      "learning_rate": 1.206523372705798e-07,
      "loss": 0.3807,
      "num_input_tokens_seen": 128576528,
      "step": 59575
    },
    {
      "epoch": 9.719412724306688,
      "grad_norm": 0.1179857924580574,
      "learning_rate": 1.199548799041772e-07,
      "loss": 0.0076,
      "num_input_tokens_seen": 128586928,
      "step": 59580
    },
    {
      "epoch": 9.720228384991843,
      "grad_norm": 34.01148223876953,
      "learning_rate": 1.1925943944047225e-07,
      "loss": 0.0785,
      "num_input_tokens_seen": 128599280,
      "step": 59585
    },
    {
      "epoch": 9.721044045676999,
      "grad_norm": 0.12337164580821991,
      "learning_rate": 1.1856601593583928e-07,
      "loss": 0.005,
      "num_input_tokens_seen": 128609424,
      "step": 59590
    },
    {
      "epoch": 9.721859706362153,
      "grad_norm": 0.12772507965564728,
      "learning_rate": 1.1787460944649443e-07,
      "loss": 0.0034,
      "num_input_tokens_seen": 128620368,
      "step": 59595
    },
    {
      "epoch": 9.722675367047309,
      "grad_norm": 0.09744003415107727,
      "learning_rate": 1.1718522002848175e-07,
      "loss": 0.0339,
      "num_input_tokens_seen": 128631792,
      "step": 59600
    },
    {
      "epoch": 9.723491027732463,
      "grad_norm": 0.20774321258068085,
      "learning_rate": 1.164978477376899e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 128641584,
      "step": 59605
    },
    {
      "epoch": 9.724306688417618,
      "grad_norm": 0.044551603496074677,
      "learning_rate": 1.1581249262984096e-07,
      "loss": 0.327,
      "num_input_tokens_seen": 128651824,
      "step": 59610
    },
    {
      "epoch": 9.725122349102774,
      "grad_norm": 0.12571464478969574,
      "learning_rate": 1.1512915476049325e-07,
      "loss": 0.2225,
      "num_input_tokens_seen": 128661616,
      "step": 59615
    },
    {
      "epoch": 9.725938009787928,
      "grad_norm": 0.07284899055957794,
      "learning_rate": 1.1444783418503857e-07,
      "loss": 0.0051,
      "num_input_tokens_seen": 128671952,
      "step": 59620
    },
    {
      "epoch": 9.726753670473084,
      "grad_norm": 0.2515014708042145,
      "learning_rate": 1.1376853095871332e-07,
      "loss": 0.0033,
      "num_input_tokens_seen": 128682928,
      "step": 59625
    },
    {
      "epoch": 9.727569331158238,
      "grad_norm": 4.521650314331055,
      "learning_rate": 1.1309124513657899e-07,
      "loss": 0.3166,
      "num_input_tokens_seen": 128692592,
      "step": 59630
    },
    {
      "epoch": 9.728384991843393,
      "grad_norm": 5.1966423988342285,
      "learning_rate": 1.124159767735472e-07,
      "loss": 0.0091,
      "num_input_tokens_seen": 128703280,
      "step": 59635
    },
    {
      "epoch": 9.729200652528547,
      "grad_norm": 0.18649420142173767,
      "learning_rate": 1.1174272592435197e-07,
      "loss": 0.0031,
      "num_input_tokens_seen": 128713904,
      "step": 59640
    },
    {
      "epoch": 9.730016313213703,
      "grad_norm": 0.287245512008667,
      "learning_rate": 1.1107149264357186e-07,
      "loss": 0.0097,
      "num_input_tokens_seen": 128723696,
      "step": 59645
    },
    {
      "epoch": 9.730831973898859,
      "grad_norm": 0.11232049763202667,
      "learning_rate": 1.1040227698562445e-07,
      "loss": 0.006,
      "num_input_tokens_seen": 128735120,
      "step": 59650
    },
    {
      "epoch": 9.731647634584013,
      "grad_norm": 0.04480063170194626,
      "learning_rate": 1.0973507900475521e-07,
      "loss": 0.1014,
      "num_input_tokens_seen": 128745232,
      "step": 59655
    },
    {
      "epoch": 9.732463295269168,
      "grad_norm": 0.13746652007102966,
      "learning_rate": 1.0906989875505425e-07,
      "loss": 0.0428,
      "num_input_tokens_seen": 128755856,
      "step": 59660
    },
    {
      "epoch": 9.733278955954322,
      "grad_norm": 0.21175557374954224,
      "learning_rate": 1.0840673629044228e-07,
      "loss": 0.0038,
      "num_input_tokens_seen": 128766768,
      "step": 59665
    },
    {
      "epoch": 9.734094616639478,
      "grad_norm": 12.107427597045898,
      "learning_rate": 1.0774559166467912e-07,
      "loss": 0.0743,
      "num_input_tokens_seen": 128778448,
      "step": 59670
    },
    {
      "epoch": 9.734910277324634,
      "grad_norm": 0.08206488937139511,
      "learning_rate": 1.0708646493135799e-07,
      "loss": 0.049,
      "num_input_tokens_seen": 128788176,
      "step": 59675
    },
    {
      "epoch": 9.735725938009788,
      "grad_norm": 0.2523277997970581,
      "learning_rate": 1.0642935614391392e-07,
      "loss": 0.0056,
      "num_input_tokens_seen": 128799280,
      "step": 59680
    },
    {
      "epoch": 9.736541598694943,
      "grad_norm": 0.05426796153187752,
      "learning_rate": 1.0577426535561541e-07,
      "loss": 0.0036,
      "num_input_tokens_seen": 128808752,
      "step": 59685
    },
    {
      "epoch": 9.737357259380097,
      "grad_norm": 0.173715740442276,
      "learning_rate": 1.0512119261956999e-07,
      "loss": 0.0432,
      "num_input_tokens_seen": 128819728,
      "step": 59690
    },
    {
      "epoch": 9.738172920065253,
      "grad_norm": 0.21712566912174225,
      "learning_rate": 1.0447013798871308e-07,
      "loss": 0.0774,
      "num_input_tokens_seen": 128831440,
      "step": 59695
    },
    {
      "epoch": 9.738988580750409,
      "grad_norm": 2.1026999950408936,
      "learning_rate": 1.0382110151582469e-07,
      "loss": 0.0072,
      "num_input_tokens_seen": 128841200,
      "step": 59700
    },
    {
      "epoch": 9.739804241435563,
      "grad_norm": 0.03553512692451477,
      "learning_rate": 1.0317408325352107e-07,
      "loss": 0.0052,
      "num_input_tokens_seen": 128852112,
      "step": 59705
    },
    {
      "epoch": 9.740619902120718,
      "grad_norm": 0.10768935084342957,
      "learning_rate": 1.0252908325425192e-07,
      "loss": 0.0048,
      "num_input_tokens_seen": 128864592,
      "step": 59710
    },
    {
      "epoch": 9.741435562805872,
      "grad_norm": 0.013585901819169521,
      "learning_rate": 1.018861015703032e-07,
      "loss": 0.0045,
      "num_input_tokens_seen": 128876272,
      "step": 59715
    },
    {
      "epoch": 9.742251223491028,
      "grad_norm": 0.31762149930000305,
      "learning_rate": 1.0124513825379989e-07,
      "loss": 0.0062,
      "num_input_tokens_seen": 128887056,
      "step": 59720
    },
    {
      "epoch": 9.743066884176184,
      "grad_norm": 3.006070613861084,
      "learning_rate": 1.0060619335669764e-07,
      "loss": 0.1659,
      "num_input_tokens_seen": 128898128,
      "step": 59725
    },
    {
      "epoch": 9.743882544861338,
      "grad_norm": 0.08332651108503342,
      "learning_rate": 9.996926693079945e-08,
      "loss": 0.0051,
      "num_input_tokens_seen": 128909040,
      "step": 59730
    },
    {
      "epoch": 9.744698205546493,
      "grad_norm": 0.043319880962371826,
      "learning_rate": 9.93343590277307e-08,
      "loss": 0.1583,
      "num_input_tokens_seen": 128918288,
      "step": 59735
    },
    {
      "epoch": 9.745513866231647,
      "grad_norm": 19.856128692626953,
      "learning_rate": 9.870146969896688e-08,
      "loss": 0.0231,
      "num_input_tokens_seen": 128929584,
      "step": 59740
    },
    {
      "epoch": 9.746329526916803,
      "grad_norm": 0.1288125216960907,
      "learning_rate": 9.807059899580861e-08,
      "loss": 0.0092,
      "num_input_tokens_seen": 128940208,
      "step": 59745
    },
    {
      "epoch": 9.747145187601957,
      "grad_norm": 6.128967761993408,
      "learning_rate": 9.744174696939834e-08,
      "loss": 0.1955,
      "num_input_tokens_seen": 128950224,
      "step": 59750
    },
    {
      "epoch": 9.747960848287113,
      "grad_norm": 0.06756886094808578,
      "learning_rate": 9.681491367071193e-08,
      "loss": 0.004,
      "num_input_tokens_seen": 128961072,
      "step": 59755
    },
    {
      "epoch": 9.748776508972268,
      "grad_norm": 0.10933032631874084,
      "learning_rate": 9.619009915056987e-08,
      "loss": 0.0045,
      "num_input_tokens_seen": 128972528,
      "step": 59760
    },
    {
      "epoch": 9.749592169657422,
      "grad_norm": 0.08776938915252686,
      "learning_rate": 9.556730345961773e-08,
      "loss": 0.3243,
      "num_input_tokens_seen": 128982480,
      "step": 59765
    },
    {
      "epoch": 9.750407830342578,
      "grad_norm": 0.15287411212921143,
      "learning_rate": 9.494652664834292e-08,
      "loss": 0.0811,
      "num_input_tokens_seen": 128992304,
      "step": 59770
    },
    {
      "epoch": 9.751223491027732,
      "grad_norm": 6.290401458740234,
      "learning_rate": 9.432776876707183e-08,
      "loss": 0.1513,
      "num_input_tokens_seen": 129003344,
      "step": 59775
    },
    {
      "epoch": 9.752039151712887,
      "grad_norm": 8.094135284423828,
      "learning_rate": 9.371102986595881e-08,
      "loss": 0.0618,
      "num_input_tokens_seen": 129014224,
      "step": 59780
    },
    {
      "epoch": 9.752854812398043,
      "grad_norm": 0.020119240507483482,
      "learning_rate": 9.309630999500551e-08,
      "loss": 0.0683,
      "num_input_tokens_seen": 129024016,
      "step": 59785
    },
    {
      "epoch": 9.753670473083197,
      "grad_norm": 0.09110990911722183,
      "learning_rate": 9.248360920404154e-08,
      "loss": 0.0056,
      "num_input_tokens_seen": 129033840,
      "step": 59790
    },
    {
      "epoch": 9.754486133768353,
      "grad_norm": 0.28636953234672546,
      "learning_rate": 9.187292754273269e-08,
      "loss": 0.0074,
      "num_input_tokens_seen": 129044848,
      "step": 59795
    },
    {
      "epoch": 9.755301794453507,
      "grad_norm": 0.13501136004924774,
      "learning_rate": 9.126426506058938e-08,
      "loss": 0.0106,
      "num_input_tokens_seen": 129055664,
      "step": 59800
    },
    {
      "epoch": 9.756117455138662,
      "grad_norm": 0.07010824978351593,
      "learning_rate": 9.06576218069527e-08,
      "loss": 0.2729,
      "num_input_tokens_seen": 129066288,
      "step": 59805
    },
    {
      "epoch": 9.756933115823816,
      "grad_norm": 0.11294228583574295,
      "learning_rate": 9.005299783099441e-08,
      "loss": 0.0641,
      "num_input_tokens_seen": 129076016,
      "step": 59810
    },
    {
      "epoch": 9.757748776508972,
      "grad_norm": 0.10147662460803986,
      "learning_rate": 8.945039318173365e-08,
      "loss": 0.0031,
      "num_input_tokens_seen": 129086704,
      "step": 59815
    },
    {
      "epoch": 9.758564437194128,
      "grad_norm": 0.13516418635845184,
      "learning_rate": 8.884980790801745e-08,
      "loss": 0.0098,
      "num_input_tokens_seen": 129096560,
      "step": 59820
    },
    {
      "epoch": 9.759380097879282,
      "grad_norm": 3.2705533504486084,
      "learning_rate": 8.825124205853463e-08,
      "loss": 0.079,
      "num_input_tokens_seen": 129107984,
      "step": 59825
    },
    {
      "epoch": 9.760195758564437,
      "grad_norm": 0.07122524827718735,
      "learning_rate": 8.76546956818075e-08,
      "loss": 0.0819,
      "num_input_tokens_seen": 129118352,
      "step": 59830
    },
    {
      "epoch": 9.761011419249591,
      "grad_norm": 0.20543478429317474,
      "learning_rate": 8.706016882619461e-08,
      "loss": 0.0078,
      "num_input_tokens_seen": 129129072,
      "step": 59835
    },
    {
      "epoch": 9.761827079934747,
      "grad_norm": 0.13996347784996033,
      "learning_rate": 8.646766153989072e-08,
      "loss": 0.01,
      "num_input_tokens_seen": 129140496,
      "step": 59840
    },
    {
      "epoch": 9.762642740619903,
      "grad_norm": 0.15542426705360413,
      "learning_rate": 8.587717387092686e-08,
      "loss": 0.0041,
      "num_input_tokens_seen": 129151952,
      "step": 59845
    },
    {
      "epoch": 9.763458401305057,
      "grad_norm": 0.19635987281799316,
      "learning_rate": 8.528870586717308e-08,
      "loss": 0.1406,
      "num_input_tokens_seen": 129165008,
      "step": 59850
    },
    {
      "epoch": 9.764274061990212,
      "grad_norm": 3.898505687713623,
      "learning_rate": 8.470225757633565e-08,
      "loss": 0.1201,
      "num_input_tokens_seen": 129176272,
      "step": 59855
    },
    {
      "epoch": 9.765089722675366,
      "grad_norm": 0.15890909731388092,
      "learning_rate": 8.411782904594879e-08,
      "loss": 0.0254,
      "num_input_tokens_seen": 129187504,
      "step": 59860
    },
    {
      "epoch": 9.765905383360522,
      "grad_norm": 0.06403691321611404,
      "learning_rate": 8.3535420323394e-08,
      "loss": 0.0052,
      "num_input_tokens_seen": 129198928,
      "step": 59865
    },
    {
      "epoch": 9.766721044045678,
      "grad_norm": 0.14090733230113983,
      "learning_rate": 8.295503145588357e-08,
      "loss": 0.0036,
      "num_input_tokens_seen": 129211024,
      "step": 59870
    },
    {
      "epoch": 9.767536704730832,
      "grad_norm": 5.934608459472656,
      "learning_rate": 8.237666249046593e-08,
      "loss": 0.0678,
      "num_input_tokens_seen": 129222096,
      "step": 59875
    },
    {
      "epoch": 9.768352365415987,
      "grad_norm": 3.298285484313965,
      "learning_rate": 8.180031347402583e-08,
      "loss": 0.2141,
      "num_input_tokens_seen": 129232880,
      "step": 59880
    },
    {
      "epoch": 9.769168026101141,
      "grad_norm": 0.13178037106990814,
      "learning_rate": 8.122598445328699e-08,
      "loss": 0.0047,
      "num_input_tokens_seen": 129244560,
      "step": 59885
    },
    {
      "epoch": 9.769983686786297,
      "grad_norm": 0.1279880404472351,
      "learning_rate": 8.065367547480384e-08,
      "loss": 0.0919,
      "num_input_tokens_seen": 129255824,
      "step": 59890
    },
    {
      "epoch": 9.770799347471453,
      "grad_norm": 0.9681717753410339,
      "learning_rate": 8.008338658497538e-08,
      "loss": 0.0922,
      "num_input_tokens_seen": 129267248,
      "step": 59895
    },
    {
      "epoch": 9.771615008156607,
      "grad_norm": 5.033680438995361,
      "learning_rate": 7.95151178300313e-08,
      "loss": 0.0105,
      "num_input_tokens_seen": 129277616,
      "step": 59900
    },
    {
      "epoch": 9.772430668841762,
      "grad_norm": 0.08865257352590561,
      "learning_rate": 7.894886925603473e-08,
      "loss": 0.0049,
      "num_input_tokens_seen": 129288112,
      "step": 59905
    },
    {
      "epoch": 9.773246329526916,
      "grad_norm": 0.807728111743927,
      "learning_rate": 7.838464090889342e-08,
      "loss": 0.2076,
      "num_input_tokens_seen": 129298064,
      "step": 59910
    },
    {
      "epoch": 9.774061990212072,
      "grad_norm": 0.055666111409664154,
      "learning_rate": 7.782243283434299e-08,
      "loss": 0.0025,
      "num_input_tokens_seen": 129307408,
      "step": 59915
    },
    {
      "epoch": 9.774877650897226,
      "grad_norm": 0.2182152420282364,
      "learning_rate": 7.726224507795809e-08,
      "loss": 0.1182,
      "num_input_tokens_seen": 129318128,
      "step": 59920
    },
    {
      "epoch": 9.775693311582382,
      "grad_norm": 0.09525197744369507,
      "learning_rate": 7.67040776851552e-08,
      "loss": 0.0035,
      "num_input_tokens_seen": 129328080,
      "step": 59925
    },
    {
      "epoch": 9.776508972267537,
      "grad_norm": 0.36281150579452515,
      "learning_rate": 7.614793070117865e-08,
      "loss": 0.1037,
      "num_input_tokens_seen": 129339120,
      "step": 59930
    },
    {
      "epoch": 9.777324632952691,
      "grad_norm": 9.132209777832031,
      "learning_rate": 7.559380417111184e-08,
      "loss": 0.1949,
      "num_input_tokens_seen": 129349648,
      "step": 59935
    },
    {
      "epoch": 9.778140293637847,
      "grad_norm": 0.17699457705020905,
      "learning_rate": 7.504169813987716e-08,
      "loss": 0.0053,
      "num_input_tokens_seen": 129360400,
      "step": 59940
    },
    {
      "epoch": 9.778955954323001,
      "grad_norm": 3.648254871368408,
      "learning_rate": 7.449161265223048e-08,
      "loss": 0.1654,
      "num_input_tokens_seen": 129373168,
      "step": 59945
    },
    {
      "epoch": 9.779771615008157,
      "grad_norm": 0.08858548104763031,
      "learning_rate": 7.394354775276391e-08,
      "loss": 0.1162,
      "num_input_tokens_seen": 129382736,
      "step": 59950
    },
    {
      "epoch": 9.780587275693312,
      "grad_norm": 0.2097587138414383,
      "learning_rate": 7.339750348590857e-08,
      "loss": 0.1244,
      "num_input_tokens_seen": 129394160,
      "step": 59955
    },
    {
      "epoch": 9.781402936378466,
      "grad_norm": 0.1020960733294487,
      "learning_rate": 7.285347989592628e-08,
      "loss": 0.0699,
      "num_input_tokens_seen": 129405744,
      "step": 59960
    },
    {
      "epoch": 9.782218597063622,
      "grad_norm": 0.2437201291322708,
      "learning_rate": 7.231147702692065e-08,
      "loss": 0.0067,
      "num_input_tokens_seen": 129416368,
      "step": 59965
    },
    {
      "epoch": 9.783034257748776,
      "grad_norm": 0.17274110019207,
      "learning_rate": 7.177149492282876e-08,
      "loss": 0.0042,
      "num_input_tokens_seen": 129426928,
      "step": 59970
    },
    {
      "epoch": 9.783849918433932,
      "grad_norm": 3.8773226737976074,
      "learning_rate": 7.123353362742391e-08,
      "loss": 0.168,
      "num_input_tokens_seen": 129438896,
      "step": 59975
    },
    {
      "epoch": 9.784665579119086,
      "grad_norm": 0.029082465916872025,
      "learning_rate": 7.069759318431567e-08,
      "loss": 0.0073,
      "num_input_tokens_seen": 129449808,
      "step": 59980
    },
    {
      "epoch": 9.785481239804241,
      "grad_norm": 0.15257468819618225,
      "learning_rate": 7.016367363694986e-08,
      "loss": 0.0472,
      "num_input_tokens_seen": 129461232,
      "step": 59985
    },
    {
      "epoch": 9.786296900489397,
      "grad_norm": 4.954206943511963,
      "learning_rate": 6.963177502861129e-08,
      "loss": 0.1873,
      "num_input_tokens_seen": 129471984,
      "step": 59990
    },
    {
      "epoch": 9.78711256117455,
      "grad_norm": 0.19747355580329895,
      "learning_rate": 6.910189740241269e-08,
      "loss": 0.0062,
      "num_input_tokens_seen": 129483088,
      "step": 59995
    },
    {
      "epoch": 9.787928221859707,
      "grad_norm": 0.11289242655038834,
      "learning_rate": 6.857404080131691e-08,
      "loss": 0.0078,
      "num_input_tokens_seen": 129492144,
      "step": 60000
    },
    {
      "epoch": 9.78874388254486,
      "grad_norm": 0.0487457811832428,
      "learning_rate": 6.804820526810917e-08,
      "loss": 0.0402,
      "num_input_tokens_seen": 129502128,
      "step": 60005
    },
    {
      "epoch": 9.789559543230016,
      "grad_norm": 0.14821499586105347,
      "learning_rate": 6.75243908454165e-08,
      "loss": 0.1213,
      "num_input_tokens_seen": 129512848,
      "step": 60010
    },
    {
      "epoch": 9.790375203915172,
      "grad_norm": 2.631269931793213,
      "learning_rate": 6.700259757570216e-08,
      "loss": 0.0793,
      "num_input_tokens_seen": 129522992,
      "step": 60015
    },
    {
      "epoch": 9.791190864600326,
      "grad_norm": 0.058667488396167755,
      "learning_rate": 6.648282550126562e-08,
      "loss": 0.0076,
      "num_input_tokens_seen": 129533328,
      "step": 60020
    },
    {
      "epoch": 9.792006525285482,
      "grad_norm": 0.10266431421041489,
      "learning_rate": 6.59650746642454e-08,
      "loss": 0.0035,
      "num_input_tokens_seen": 129542768,
      "step": 60025
    },
    {
      "epoch": 9.792822185970635,
      "grad_norm": 0.1482907086610794,
      "learning_rate": 6.544934510660794e-08,
      "loss": 0.1718,
      "num_input_tokens_seen": 129554448,
      "step": 60030
    },
    {
      "epoch": 9.793637846655791,
      "grad_norm": 5.6294684410095215,
      "learning_rate": 6.493563687016424e-08,
      "loss": 0.009,
      "num_input_tokens_seen": 129566736,
      "step": 60035
    },
    {
      "epoch": 9.794453507340947,
      "grad_norm": 0.07283642888069153,
      "learning_rate": 6.442394999655599e-08,
      "loss": 0.0081,
      "num_input_tokens_seen": 129577392,
      "step": 60040
    },
    {
      "epoch": 9.7952691680261,
      "grad_norm": 0.25296127796173096,
      "learning_rate": 6.391428452726389e-08,
      "loss": 0.0058,
      "num_input_tokens_seen": 129588080,
      "step": 60045
    },
    {
      "epoch": 9.796084828711257,
      "grad_norm": 0.08539183437824249,
      "learning_rate": 6.340664050360767e-08,
      "loss": 0.1011,
      "num_input_tokens_seen": 129599312,
      "step": 60050
    },
    {
      "epoch": 9.79690048939641,
      "grad_norm": 3.745346784591675,
      "learning_rate": 6.29010179667322e-08,
      "loss": 0.1297,
      "num_input_tokens_seen": 129609488,
      "step": 60055
    },
    {
      "epoch": 9.797716150081566,
      "grad_norm": 4.010807991027832,
      "learning_rate": 6.239741695763246e-08,
      "loss": 0.0944,
      "num_input_tokens_seen": 129620944,
      "step": 60060
    },
    {
      "epoch": 9.798531810766722,
      "grad_norm": 0.1772913634777069,
      "learning_rate": 6.189583751712857e-08,
      "loss": 0.0069,
      "num_input_tokens_seen": 129631920,
      "step": 60065
    },
    {
      "epoch": 9.799347471451876,
      "grad_norm": 0.0517205074429512,
      "learning_rate": 6.139627968588524e-08,
      "loss": 0.0051,
      "num_input_tokens_seen": 129643696,
      "step": 60070
    },
    {
      "epoch": 9.800163132137031,
      "grad_norm": 0.1302974820137024,
      "learning_rate": 6.089874350439506e-08,
      "loss": 0.1105,
      "num_input_tokens_seen": 129655184,
      "step": 60075
    },
    {
      "epoch": 9.800978792822185,
      "grad_norm": 0.10189558565616608,
      "learning_rate": 6.040322901299245e-08,
      "loss": 0.0031,
      "num_input_tokens_seen": 129665008,
      "step": 60080
    },
    {
      "epoch": 9.801794453507341,
      "grad_norm": 0.028916558250784874,
      "learning_rate": 5.990973625184526e-08,
      "loss": 0.0027,
      "num_input_tokens_seen": 129675344,
      "step": 60085
    },
    {
      "epoch": 9.802610114192497,
      "grad_norm": 3.744006633758545,
      "learning_rate": 5.9418265260960394e-08,
      "loss": 0.2226,
      "num_input_tokens_seen": 129687120,
      "step": 60090
    },
    {
      "epoch": 9.80342577487765,
      "grad_norm": 0.07970011234283447,
      "learning_rate": 5.892881608017819e-08,
      "loss": 0.0097,
      "num_input_tokens_seen": 129697936,
      "step": 60095
    },
    {
      "epoch": 9.804241435562806,
      "grad_norm": 0.13731755316257477,
      "learning_rate": 5.844138874917526e-08,
      "loss": 0.0071,
      "num_input_tokens_seen": 129707312,
      "step": 60100
    },
    {
      "epoch": 9.80505709624796,
      "grad_norm": 0.0961083248257637,
      "learning_rate": 5.795598330746721e-08,
      "loss": 0.0053,
      "num_input_tokens_seen": 129718256,
      "step": 60105
    },
    {
      "epoch": 9.805872756933116,
      "grad_norm": 0.39364707469940186,
      "learning_rate": 5.747259979440034e-08,
      "loss": 0.0087,
      "num_input_tokens_seen": 129728336,
      "step": 60110
    },
    {
      "epoch": 9.80668841761827,
      "grad_norm": 0.16876554489135742,
      "learning_rate": 5.699123824916275e-08,
      "loss": 0.0075,
      "num_input_tokens_seen": 129738960,
      "step": 60115
    },
    {
      "epoch": 9.807504078303426,
      "grad_norm": 3.1732735633850098,
      "learning_rate": 5.6511898710776e-08,
      "loss": 0.1868,
      "num_input_tokens_seen": 129748752,
      "step": 60120
    },
    {
      "epoch": 9.808319738988581,
      "grad_norm": 0.09069392085075378,
      "learning_rate": 5.603458121809513e-08,
      "loss": 0.0052,
      "num_input_tokens_seen": 129758864,
      "step": 60125
    },
    {
      "epoch": 9.809135399673735,
      "grad_norm": 15.149872779846191,
      "learning_rate": 5.555928580981418e-08,
      "loss": 0.1611,
      "num_input_tokens_seen": 129768112,
      "step": 60130
    },
    {
      "epoch": 9.809951060358891,
      "grad_norm": 0.11818689107894897,
      "learning_rate": 5.5086012524466216e-08,
      "loss": 0.0076,
      "num_input_tokens_seen": 129779440,
      "step": 60135
    },
    {
      "epoch": 9.810766721044045,
      "grad_norm": 0.18144717812538147,
      "learning_rate": 5.4614761400414996e-08,
      "loss": 0.0782,
      "num_input_tokens_seen": 129790320,
      "step": 60140
    },
    {
      "epoch": 9.8115823817292,
      "grad_norm": 0.2294866293668747,
      "learning_rate": 5.414553247586329e-08,
      "loss": 0.1056,
      "num_input_tokens_seen": 129799792,
      "step": 60145
    },
    {
      "epoch": 9.812398042414356,
      "grad_norm": 7.488786220550537,
      "learning_rate": 5.367832578884735e-08,
      "loss": 0.18,
      "num_input_tokens_seen": 129811472,
      "step": 60150
    },
    {
      "epoch": 9.81321370309951,
      "grad_norm": 0.09118147194385529,
      "learning_rate": 5.3213141377245205e-08,
      "loss": 0.0047,
      "num_input_tokens_seen": 129822512,
      "step": 60155
    },
    {
      "epoch": 9.814029363784666,
      "grad_norm": 0.48315927386283875,
      "learning_rate": 5.2749979278762794e-08,
      "loss": 0.0061,
      "num_input_tokens_seen": 129834608,
      "step": 60160
    },
    {
      "epoch": 9.81484502446982,
      "grad_norm": 0.08426214009523392,
      "learning_rate": 5.228883953094788e-08,
      "loss": 0.1253,
      "num_input_tokens_seen": 129845008,
      "step": 60165
    },
    {
      "epoch": 9.815660685154976,
      "grad_norm": 3.5015861988067627,
      "learning_rate": 5.182972217118165e-08,
      "loss": 0.1164,
      "num_input_tokens_seen": 129854928,
      "step": 60170
    },
    {
      "epoch": 9.81647634584013,
      "grad_norm": 7.503712177276611,
      "learning_rate": 5.137262723668712e-08,
      "loss": 0.0292,
      "num_input_tokens_seen": 129865904,
      "step": 60175
    },
    {
      "epoch": 9.817292006525285,
      "grad_norm": 0.10699860006570816,
      "learning_rate": 5.0917554764515206e-08,
      "loss": 0.1108,
      "num_input_tokens_seen": 129878288,
      "step": 60180
    },
    {
      "epoch": 9.818107667210441,
      "grad_norm": 0.04898899421095848,
      "learning_rate": 5.0464504791553066e-08,
      "loss": 0.0695,
      "num_input_tokens_seen": 129890512,
      "step": 60185
    },
    {
      "epoch": 9.818923327895595,
      "grad_norm": 0.100347600877285,
      "learning_rate": 5.001347735453521e-08,
      "loss": 0.1822,
      "num_input_tokens_seen": 129900560,
      "step": 60190
    },
    {
      "epoch": 9.81973898858075,
      "grad_norm": 0.02564556896686554,
      "learning_rate": 4.95644724900185e-08,
      "loss": 0.0029,
      "num_input_tokens_seen": 129912272,
      "step": 60195
    },
    {
      "epoch": 9.820554649265905,
      "grad_norm": 0.06976073980331421,
      "learning_rate": 4.91174902344016e-08,
      "loss": 0.1059,
      "num_input_tokens_seen": 129923120,
      "step": 60200
    },
    {
      "epoch": 9.82137030995106,
      "grad_norm": 0.08213542401790619,
      "learning_rate": 4.867253062391941e-08,
      "loss": 0.0031,
      "num_input_tokens_seen": 129935280,
      "step": 60205
    },
    {
      "epoch": 9.822185970636216,
      "grad_norm": 0.016145285218954086,
      "learning_rate": 4.822959369464586e-08,
      "loss": 0.0025,
      "num_input_tokens_seen": 129946032,
      "step": 60210
    },
    {
      "epoch": 9.82300163132137,
      "grad_norm": 0.3220311403274536,
      "learning_rate": 4.7788679482485556e-08,
      "loss": 0.0115,
      "num_input_tokens_seen": 129956784,
      "step": 60215
    },
    {
      "epoch": 9.823817292006526,
      "grad_norm": 0.16524042189121246,
      "learning_rate": 4.734978802318213e-08,
      "loss": 0.0046,
      "num_input_tokens_seen": 129967504,
      "step": 60220
    },
    {
      "epoch": 9.82463295269168,
      "grad_norm": 0.40293651819229126,
      "learning_rate": 4.69129193523099e-08,
      "loss": 0.1095,
      "num_input_tokens_seen": 129977840,
      "step": 60225
    },
    {
      "epoch": 9.825448613376835,
      "grad_norm": 0.07762454450130463,
      "learning_rate": 4.6478073505290544e-08,
      "loss": 0.0092,
      "num_input_tokens_seen": 129988080,
      "step": 60230
    },
    {
      "epoch": 9.826264274061991,
      "grad_norm": 0.15140385925769806,
      "learning_rate": 4.6045250517370854e-08,
      "loss": 0.0033,
      "num_input_tokens_seen": 129999728,
      "step": 60235
    },
    {
      "epoch": 9.827079934747145,
      "grad_norm": 0.12313911318778992,
      "learning_rate": 4.561445042363666e-08,
      "loss": 0.0932,
      "num_input_tokens_seen": 130010352,
      "step": 60240
    },
    {
      "epoch": 9.8278955954323,
      "grad_norm": 0.23499034345149994,
      "learning_rate": 4.518567325901279e-08,
      "loss": 0.1038,
      "num_input_tokens_seen": 130021776,
      "step": 60245
    },
    {
      "epoch": 9.828711256117455,
      "grad_norm": 0.03810376301407814,
      "learning_rate": 4.475891905825758e-08,
      "loss": 0.0047,
      "num_input_tokens_seen": 130032304,
      "step": 60250
    },
    {
      "epoch": 9.82952691680261,
      "grad_norm": 0.076126828789711,
      "learning_rate": 4.4334187855968326e-08,
      "loss": 0.1281,
      "num_input_tokens_seen": 130041840,
      "step": 60255
    },
    {
      "epoch": 9.830342577487766,
      "grad_norm": 0.05613238364458084,
      "learning_rate": 4.391147968657028e-08,
      "loss": 0.0048,
      "num_input_tokens_seen": 130051216,
      "step": 60260
    },
    {
      "epoch": 9.83115823817292,
      "grad_norm": 17.134361267089844,
      "learning_rate": 4.3490794584336024e-08,
      "loss": 0.0249,
      "num_input_tokens_seen": 130061872,
      "step": 60265
    },
    {
      "epoch": 9.831973898858076,
      "grad_norm": 0.05618003383278847,
      "learning_rate": 4.307213258336606e-08,
      "loss": 0.0035,
      "num_input_tokens_seen": 130072944,
      "step": 60270
    },
    {
      "epoch": 9.83278955954323,
      "grad_norm": 20.586030960083008,
      "learning_rate": 4.2655493717597137e-08,
      "loss": 0.1361,
      "num_input_tokens_seen": 130082704,
      "step": 60275
    },
    {
      "epoch": 9.833605220228385,
      "grad_norm": 0.5808132886886597,
      "learning_rate": 4.224087802080778e-08,
      "loss": 0.0055,
      "num_input_tokens_seen": 130093936,
      "step": 60280
    },
    {
      "epoch": 9.83442088091354,
      "grad_norm": 0.12000591307878494,
      "learning_rate": 4.182828552660722e-08,
      "loss": 0.0049,
      "num_input_tokens_seen": 130103888,
      "step": 60285
    },
    {
      "epoch": 9.835236541598695,
      "grad_norm": 0.11502066254615784,
      "learning_rate": 4.141771626844093e-08,
      "loss": 0.005,
      "num_input_tokens_seen": 130114864,
      "step": 60290
    },
    {
      "epoch": 9.83605220228385,
      "grad_norm": 0.12205211073160172,
      "learning_rate": 4.100917027959617e-08,
      "loss": 0.1241,
      "num_input_tokens_seen": 130125136,
      "step": 60295
    },
    {
      "epoch": 9.836867862969005,
      "grad_norm": 0.12927758693695068,
      "learning_rate": 4.0602647593185325e-08,
      "loss": 0.0072,
      "num_input_tokens_seen": 130136240,
      "step": 60300
    },
    {
      "epoch": 9.83768352365416,
      "grad_norm": 0.07437407970428467,
      "learning_rate": 4.0198148242168163e-08,
      "loss": 0.1205,
      "num_input_tokens_seen": 130146864,
      "step": 60305
    },
    {
      "epoch": 9.838499184339314,
      "grad_norm": 0.05137110501527786,
      "learning_rate": 3.979567225933234e-08,
      "loss": 0.0067,
      "num_input_tokens_seen": 130156560,
      "step": 60310
    },
    {
      "epoch": 9.83931484502447,
      "grad_norm": 0.12136679887771606,
      "learning_rate": 3.939521967730731e-08,
      "loss": 0.2939,
      "num_input_tokens_seen": 130167664,
      "step": 60315
    },
    {
      "epoch": 9.840130505709626,
      "grad_norm": 0.06857368350028992,
      "learning_rate": 3.8996790528555985e-08,
      "loss": 0.1693,
      "num_input_tokens_seen": 130178992,
      "step": 60320
    },
    {
      "epoch": 9.84094616639478,
      "grad_norm": 0.08583206683397293,
      "learning_rate": 3.860038484537476e-08,
      "loss": 0.1008,
      "num_input_tokens_seen": 130188944,
      "step": 60325
    },
    {
      "epoch": 9.841761827079935,
      "grad_norm": 0.19351406395435333,
      "learning_rate": 3.820600265989904e-08,
      "loss": 0.0969,
      "num_input_tokens_seen": 130199408,
      "step": 60330
    },
    {
      "epoch": 9.84257748776509,
      "grad_norm": 0.27275794744491577,
      "learning_rate": 3.78136440040977e-08,
      "loss": 0.0939,
      "num_input_tokens_seen": 130211152,
      "step": 60335
    },
    {
      "epoch": 9.843393148450245,
      "grad_norm": 0.14788496494293213,
      "learning_rate": 3.742330890978141e-08,
      "loss": 0.0939,
      "num_input_tokens_seen": 130222192,
      "step": 60340
    },
    {
      "epoch": 9.844208809135399,
      "grad_norm": 0.17083828151226044,
      "learning_rate": 3.703499740859151e-08,
      "loss": 0.0094,
      "num_input_tokens_seen": 130232912,
      "step": 60345
    },
    {
      "epoch": 9.845024469820554,
      "grad_norm": 0.0933980867266655,
      "learning_rate": 3.6648709532002835e-08,
      "loss": 0.0043,
      "num_input_tokens_seen": 130242704,
      "step": 60350
    },
    {
      "epoch": 9.84584013050571,
      "grad_norm": 0.12900620698928833,
      "learning_rate": 3.6264445311334774e-08,
      "loss": 0.0423,
      "num_input_tokens_seen": 130254224,
      "step": 60355
    },
    {
      "epoch": 9.846655791190864,
      "grad_norm": 0.12694202363491058,
      "learning_rate": 3.588220477773463e-08,
      "loss": 0.1178,
      "num_input_tokens_seen": 130265392,
      "step": 60360
    },
    {
      "epoch": 9.84747145187602,
      "grad_norm": 0.23511680960655212,
      "learning_rate": 3.5501987962191505e-08,
      "loss": 0.0045,
      "num_input_tokens_seen": 130276048,
      "step": 60365
    },
    {
      "epoch": 9.848287112561174,
      "grad_norm": 0.042450256645679474,
      "learning_rate": 3.5123794895522425e-08,
      "loss": 0.0026,
      "num_input_tokens_seen": 130287472,
      "step": 60370
    },
    {
      "epoch": 9.84910277324633,
      "grad_norm": 0.5879544615745544,
      "learning_rate": 3.4747625608391735e-08,
      "loss": 0.0049,
      "num_input_tokens_seen": 130298832,
      "step": 60375
    },
    {
      "epoch": 9.849918433931485,
      "grad_norm": 0.1835554540157318,
      "learning_rate": 3.4373480131288936e-08,
      "loss": 0.0038,
      "num_input_tokens_seen": 130308432,
      "step": 60380
    },
    {
      "epoch": 9.850734094616639,
      "grad_norm": 0.15611405670642853,
      "learning_rate": 3.400135849454811e-08,
      "loss": 0.1483,
      "num_input_tokens_seen": 130318544,
      "step": 60385
    },
    {
      "epoch": 9.851549755301795,
      "grad_norm": 0.11268353462219238,
      "learning_rate": 3.363126072833123e-08,
      "loss": 0.2281,
      "num_input_tokens_seen": 130328176,
      "step": 60390
    },
    {
      "epoch": 9.852365415986949,
      "grad_norm": 11.306412696838379,
      "learning_rate": 3.326318686264485e-08,
      "loss": 0.1049,
      "num_input_tokens_seen": 130338896,
      "step": 60395
    },
    {
      "epoch": 9.853181076672104,
      "grad_norm": 0.05320761352777481,
      "learning_rate": 3.2897136927323436e-08,
      "loss": 0.0933,
      "num_input_tokens_seen": 130350160,
      "step": 60400
    },
    {
      "epoch": 9.85399673735726,
      "grad_norm": 0.08861350268125534,
      "learning_rate": 3.253311095204048e-08,
      "loss": 0.004,
      "num_input_tokens_seen": 130360976,
      "step": 60405
    },
    {
      "epoch": 9.854812398042414,
      "grad_norm": 0.07633858174085617,
      "learning_rate": 3.2171108966308486e-08,
      "loss": 0.0029,
      "num_input_tokens_seen": 130371184,
      "step": 60410
    },
    {
      "epoch": 9.85562805872757,
      "grad_norm": 0.1442979872226715,
      "learning_rate": 3.1811130999473415e-08,
      "loss": 0.0056,
      "num_input_tokens_seen": 130381168,
      "step": 60415
    },
    {
      "epoch": 9.856443719412724,
      "grad_norm": 0.22508582472801208,
      "learning_rate": 3.145317708071194e-08,
      "loss": 0.0039,
      "num_input_tokens_seen": 130390544,
      "step": 60420
    },
    {
      "epoch": 9.85725938009788,
      "grad_norm": 0.05943860858678818,
      "learning_rate": 3.1097247239048057e-08,
      "loss": 0.0036,
      "num_input_tokens_seen": 130401232,
      "step": 60425
    },
    {
      "epoch": 9.858075040783035,
      "grad_norm": 0.08935709297657013,
      "learning_rate": 3.074334150333091e-08,
      "loss": 0.0031,
      "num_input_tokens_seen": 130412560,
      "step": 60430
    },
    {
      "epoch": 9.858890701468189,
      "grad_norm": 0.06829848140478134,
      "learning_rate": 3.039145990225145e-08,
      "loss": 0.0875,
      "num_input_tokens_seen": 130423472,
      "step": 60435
    },
    {
      "epoch": 9.859706362153345,
      "grad_norm": 0.30248787999153137,
      "learning_rate": 3.0041602464334076e-08,
      "loss": 0.0062,
      "num_input_tokens_seen": 130433008,
      "step": 60440
    },
    {
      "epoch": 9.860522022838499,
      "grad_norm": 4.402347087860107,
      "learning_rate": 2.9693769217942203e-08,
      "loss": 0.1105,
      "num_input_tokens_seen": 130445776,
      "step": 60445
    },
    {
      "epoch": 9.861337683523654,
      "grad_norm": 3.2330546379089355,
      "learning_rate": 2.9347960191269952e-08,
      "loss": 0.2335,
      "num_input_tokens_seen": 130456720,
      "step": 60450
    },
    {
      "epoch": 9.86215334420881,
      "grad_norm": 0.17700782418251038,
      "learning_rate": 2.900417541235323e-08,
      "loss": 0.0091,
      "num_input_tokens_seen": 130468176,
      "step": 60455
    },
    {
      "epoch": 9.862969004893964,
      "grad_norm": 0.05328657478094101,
      "learning_rate": 2.8662414909058634e-08,
      "loss": 0.0154,
      "num_input_tokens_seen": 130479344,
      "step": 60460
    },
    {
      "epoch": 9.86378466557912,
      "grad_norm": 0.07495290786027908,
      "learning_rate": 2.8322678709094553e-08,
      "loss": 0.0159,
      "num_input_tokens_seen": 130490672,
      "step": 60465
    },
    {
      "epoch": 9.864600326264274,
      "grad_norm": 0.06120337173342705,
      "learning_rate": 2.7984966839997294e-08,
      "loss": 0.0036,
      "num_input_tokens_seen": 130502160,
      "step": 60470
    },
    {
      "epoch": 9.86541598694943,
      "grad_norm": 0.04699868708848953,
      "learning_rate": 2.7649279329142185e-08,
      "loss": 0.0029,
      "num_input_tokens_seen": 130514160,
      "step": 60475
    },
    {
      "epoch": 9.866231647634583,
      "grad_norm": 0.05974680557847023,
      "learning_rate": 2.7315616203749118e-08,
      "loss": 0.1149,
      "num_input_tokens_seen": 130524336,
      "step": 60480
    },
    {
      "epoch": 9.867047308319739,
      "grad_norm": 0.057295531034469604,
      "learning_rate": 2.6983977490860345e-08,
      "loss": 0.171,
      "num_input_tokens_seen": 130535440,
      "step": 60485
    },
    {
      "epoch": 9.867862969004895,
      "grad_norm": 0.044656090438365936,
      "learning_rate": 2.6654363217362698e-08,
      "loss": 0.1006,
      "num_input_tokens_seen": 130546832,
      "step": 60490
    },
    {
      "epoch": 9.868678629690049,
      "grad_norm": 0.09445744752883911,
      "learning_rate": 2.632677340997647e-08,
      "loss": 0.0058,
      "num_input_tokens_seen": 130557168,
      "step": 60495
    },
    {
      "epoch": 9.869494290375204,
      "grad_norm": 0.26121169328689575,
      "learning_rate": 2.6001208095258188e-08,
      "loss": 0.0071,
      "num_input_tokens_seen": 130567664,
      "step": 60500
    },
    {
      "epoch": 9.870309951060358,
      "grad_norm": 0.3801686465740204,
      "learning_rate": 2.5677667299597863e-08,
      "loss": 0.0044,
      "num_input_tokens_seen": 130577520,
      "step": 60505
    },
    {
      "epoch": 9.871125611745514,
      "grad_norm": 0.1790764480829239,
      "learning_rate": 2.5356151049221734e-08,
      "loss": 0.0052,
      "num_input_tokens_seen": 130589328,
      "step": 60510
    },
    {
      "epoch": 9.87194127243067,
      "grad_norm": 5.141573905944824,
      "learning_rate": 2.5036659370197836e-08,
      "loss": 0.0888,
      "num_input_tokens_seen": 130599760,
      "step": 60515
    },
    {
      "epoch": 9.872756933115824,
      "grad_norm": 0.644772469997406,
      "learning_rate": 2.4719192288424896e-08,
      "loss": 0.008,
      "num_input_tokens_seen": 130610576,
      "step": 60520
    },
    {
      "epoch": 9.87357259380098,
      "grad_norm": 0.12314771860837936,
      "learning_rate": 2.440374982963789e-08,
      "loss": 0.063,
      "num_input_tokens_seen": 130621168,
      "step": 60525
    },
    {
      "epoch": 9.874388254486133,
      "grad_norm": 0.024125229567289352,
      "learning_rate": 2.409033201940525e-08,
      "loss": 0.0063,
      "num_input_tokens_seen": 130630736,
      "step": 60530
    },
    {
      "epoch": 9.875203915171289,
      "grad_norm": 0.2039998471736908,
      "learning_rate": 2.3778938883139977e-08,
      "loss": 0.0713,
      "num_input_tokens_seen": 130641840,
      "step": 60535
    },
    {
      "epoch": 9.876019575856443,
      "grad_norm": 0.05761853605508804,
      "learning_rate": 2.3469570446080223e-08,
      "loss": 0.0062,
      "num_input_tokens_seen": 130652816,
      "step": 60540
    },
    {
      "epoch": 9.876835236541599,
      "grad_norm": 0.11537981778383255,
      "learning_rate": 2.3162226733305925e-08,
      "loss": 0.0074,
      "num_input_tokens_seen": 130663248,
      "step": 60545
    },
    {
      "epoch": 9.877650897226754,
      "grad_norm": 0.06408042460680008,
      "learning_rate": 2.2856907769736037e-08,
      "loss": 0.0311,
      "num_input_tokens_seen": 130675184,
      "step": 60550
    },
    {
      "epoch": 9.878466557911908,
      "grad_norm": 3.4832675457000732,
      "learning_rate": 2.255361358011465e-08,
      "loss": 0.0079,
      "num_input_tokens_seen": 130686576,
      "step": 60555
    },
    {
      "epoch": 9.879282218597064,
      "grad_norm": 0.15520964562892914,
      "learning_rate": 2.2252344189033213e-08,
      "loss": 0.0038,
      "num_input_tokens_seen": 130696240,
      "step": 60560
    },
    {
      "epoch": 9.880097879282218,
      "grad_norm": 3.5604052543640137,
      "learning_rate": 2.1953099620911076e-08,
      "loss": 0.0101,
      "num_input_tokens_seen": 130706064,
      "step": 60565
    },
    {
      "epoch": 9.880913539967374,
      "grad_norm": 0.12187746912240982,
      "learning_rate": 2.165587990000939e-08,
      "loss": 0.0063,
      "num_input_tokens_seen": 130716720,
      "step": 60570
    },
    {
      "epoch": 9.88172920065253,
      "grad_norm": 0.27157095074653625,
      "learning_rate": 2.1360685050419994e-08,
      "loss": 0.1446,
      "num_input_tokens_seen": 130726288,
      "step": 60575
    },
    {
      "epoch": 9.882544861337683,
      "grad_norm": 0.08113939315080643,
      "learning_rate": 2.106751509607374e-08,
      "loss": 0.0055,
      "num_input_tokens_seen": 130736368,
      "step": 60580
    },
    {
      "epoch": 9.883360522022839,
      "grad_norm": 0.10311729460954666,
      "learning_rate": 2.0776370060737737e-08,
      "loss": 0.1237,
      "num_input_tokens_seen": 130746160,
      "step": 60585
    },
    {
      "epoch": 9.884176182707993,
      "grad_norm": 0.1000794768333435,
      "learning_rate": 2.0487249968012546e-08,
      "loss": 0.0123,
      "num_input_tokens_seen": 130755728,
      "step": 60590
    },
    {
      "epoch": 9.884991843393149,
      "grad_norm": 0.04848021641373634,
      "learning_rate": 2.020015484133497e-08,
      "loss": 0.0051,
      "num_input_tokens_seen": 130767088,
      "step": 60595
    },
    {
      "epoch": 9.885807504078304,
      "grad_norm": 0.1445668786764145,
      "learning_rate": 1.9915084703980845e-08,
      "loss": 0.0079,
      "num_input_tokens_seen": 130776944,
      "step": 60600
    },
    {
      "epoch": 9.886623164763458,
      "grad_norm": 0.26100313663482666,
      "learning_rate": 1.9632039579053907e-08,
      "loss": 0.0172,
      "num_input_tokens_seen": 130786480,
      "step": 60605
    },
    {
      "epoch": 9.887438825448614,
      "grad_norm": 0.058276962488889694,
      "learning_rate": 1.935101948950524e-08,
      "loss": 0.0161,
      "num_input_tokens_seen": 130797968,
      "step": 60610
    },
    {
      "epoch": 9.888254486133768,
      "grad_norm": 10.515925407409668,
      "learning_rate": 1.9072024458113847e-08,
      "loss": 0.0844,
      "num_input_tokens_seen": 130808272,
      "step": 60615
    },
    {
      "epoch": 9.889070146818923,
      "grad_norm": 0.08804245293140411,
      "learning_rate": 1.8795054507494967e-08,
      "loss": 0.1519,
      "num_input_tokens_seen": 130817488,
      "step": 60620
    },
    {
      "epoch": 9.88988580750408,
      "grad_norm": 0.04314444214105606,
      "learning_rate": 1.852010966010287e-08,
      "loss": 0.0036,
      "num_input_tokens_seen": 130828432,
      "step": 60625
    },
    {
      "epoch": 9.890701468189233,
      "grad_norm": 0.06376510858535767,
      "learning_rate": 1.8247189938225274e-08,
      "loss": 0.129,
      "num_input_tokens_seen": 130838416,
      "step": 60630
    },
    {
      "epoch": 9.891517128874389,
      "grad_norm": 0.2041330337524414,
      "learning_rate": 1.7976295363988927e-08,
      "loss": 0.0052,
      "num_input_tokens_seen": 130847920,
      "step": 60635
    },
    {
      "epoch": 9.892332789559543,
      "grad_norm": 4.164166450500488,
      "learning_rate": 1.7707425959348488e-08,
      "loss": 0.2132,
      "num_input_tokens_seen": 130859184,
      "step": 60640
    },
    {
      "epoch": 9.893148450244698,
      "grad_norm": 0.05904705449938774,
      "learning_rate": 1.744058174610319e-08,
      "loss": 0.0067,
      "num_input_tokens_seen": 130869808,
      "step": 60645
    },
    {
      "epoch": 9.893964110929852,
      "grad_norm": 0.11198320239782333,
      "learning_rate": 1.7175762745885727e-08,
      "loss": 0.0046,
      "num_input_tokens_seen": 130880912,
      "step": 60650
    },
    {
      "epoch": 9.894779771615008,
      "grad_norm": 6.100296974182129,
      "learning_rate": 1.6912968980162257e-08,
      "loss": 0.0115,
      "num_input_tokens_seen": 130892336,
      "step": 60655
    },
    {
      "epoch": 9.895595432300164,
      "grad_norm": 0.1114816814661026,
      "learning_rate": 1.665220047023519e-08,
      "loss": 0.0078,
      "num_input_tokens_seen": 130900752,
      "step": 60660
    },
    {
      "epoch": 9.896411092985318,
      "grad_norm": 0.062797412276268,
      "learning_rate": 1.639345723724872e-08,
      "loss": 0.1296,
      "num_input_tokens_seen": 130911600,
      "step": 60665
    },
    {
      "epoch": 9.897226753670473,
      "grad_norm": 0.08252307027578354,
      "learning_rate": 1.6136739302169412e-08,
      "loss": 0.0045,
      "num_input_tokens_seen": 130922640,
      "step": 60670
    },
    {
      "epoch": 9.898042414355627,
      "grad_norm": 14.04305362701416,
      "learning_rate": 1.588204668581672e-08,
      "loss": 0.0549,
      "num_input_tokens_seen": 130933136,
      "step": 60675
    },
    {
      "epoch": 9.898858075040783,
      "grad_norm": 0.21598735451698303,
      "learning_rate": 1.5629379408832468e-08,
      "loss": 0.02,
      "num_input_tokens_seen": 130944336,
      "step": 60680
    },
    {
      "epoch": 9.899673735725939,
      "grad_norm": 0.047149281948804855,
      "learning_rate": 1.537873749169749e-08,
      "loss": 0.1066,
      "num_input_tokens_seen": 130952848,
      "step": 60685
    },
    {
      "epoch": 9.900489396411093,
      "grad_norm": 0.06601323932409286,
      "learning_rate": 1.513012095473443e-08,
      "loss": 0.1251,
      "num_input_tokens_seen": 130963536,
      "step": 60690
    },
    {
      "epoch": 9.901305057096248,
      "grad_norm": 0.13944591581821442,
      "learning_rate": 1.4883529818096598e-08,
      "loss": 0.23,
      "num_input_tokens_seen": 130973840,
      "step": 60695
    },
    {
      "epoch": 9.902120717781402,
      "grad_norm": 0.1133158877491951,
      "learning_rate": 1.4638964101773568e-08,
      "loss": 0.0043,
      "num_input_tokens_seen": 130984304,
      "step": 60700
    },
    {
      "epoch": 9.902936378466558,
      "grad_norm": 0.11855597794055939,
      "learning_rate": 1.4396423825588367e-08,
      "loss": 0.0057,
      "num_input_tokens_seen": 130995504,
      "step": 60705
    },
    {
      "epoch": 9.903752039151712,
      "grad_norm": 3.568680763244629,
      "learning_rate": 1.4155909009205826e-08,
      "loss": 0.1106,
      "num_input_tokens_seen": 131005968,
      "step": 60710
    },
    {
      "epoch": 9.904567699836868,
      "grad_norm": 0.0719594955444336,
      "learning_rate": 1.3917419672124233e-08,
      "loss": 0.0043,
      "num_input_tokens_seen": 131018160,
      "step": 60715
    },
    {
      "epoch": 9.905383360522023,
      "grad_norm": 0.3674751818180084,
      "learning_rate": 1.368095583367257e-08,
      "loss": 0.0835,
      "num_input_tokens_seen": 131029008,
      "step": 60720
    },
    {
      "epoch": 9.906199021207177,
      "grad_norm": 0.3260965645313263,
      "learning_rate": 1.3446517513021617e-08,
      "loss": 0.0512,
      "num_input_tokens_seen": 131038960,
      "step": 60725
    },
    {
      "epoch": 9.907014681892333,
      "grad_norm": 0.09662462770938873,
      "learning_rate": 1.321410472917839e-08,
      "loss": 0.0058,
      "num_input_tokens_seen": 131050160,
      "step": 60730
    },
    {
      "epoch": 9.907830342577487,
      "grad_norm": 11.745221138000488,
      "learning_rate": 1.2983717500977822e-08,
      "loss": 0.0256,
      "num_input_tokens_seen": 131060048,
      "step": 60735
    },
    {
      "epoch": 9.908646003262643,
      "grad_norm": 5.630153656005859,
      "learning_rate": 1.275535584710219e-08,
      "loss": 0.1848,
      "num_input_tokens_seen": 131069360,
      "step": 60740
    },
    {
      "epoch": 9.909461663947798,
      "grad_norm": 0.08159533888101578,
      "learning_rate": 1.2529019786061691e-08,
      "loss": 0.0065,
      "num_input_tokens_seen": 131079888,
      "step": 60745
    },
    {
      "epoch": 9.910277324632952,
      "grad_norm": 0.3844931721687317,
      "learning_rate": 1.2304709336205533e-08,
      "loss": 0.0058,
      "num_input_tokens_seen": 131090512,
      "step": 60750
    },
    {
      "epoch": 9.911092985318108,
      "grad_norm": 0.07297611981630325,
      "learning_rate": 1.2082424515713619e-08,
      "loss": 0.1885,
      "num_input_tokens_seen": 131101680,
      "step": 60755
    },
    {
      "epoch": 9.911908646003262,
      "grad_norm": 0.09410180151462555,
      "learning_rate": 1.1862165342607645e-08,
      "loss": 0.0615,
      "num_input_tokens_seen": 131112336,
      "step": 60760
    },
    {
      "epoch": 9.912724306688418,
      "grad_norm": 0.14489737153053284,
      "learning_rate": 1.1643931834745548e-08,
      "loss": 0.0077,
      "num_input_tokens_seen": 131123088,
      "step": 60765
    },
    {
      "epoch": 9.913539967373573,
      "grad_norm": 0.6937752366065979,
      "learning_rate": 1.1427724009813179e-08,
      "loss": 0.0498,
      "num_input_tokens_seen": 131132368,
      "step": 60770
    },
    {
      "epoch": 9.914355628058727,
      "grad_norm": 0.08811195194721222,
      "learning_rate": 1.1213541885340962e-08,
      "loss": 0.0106,
      "num_input_tokens_seen": 131143600,
      "step": 60775
    },
    {
      "epoch": 9.915171288743883,
      "grad_norm": 0.07232557237148285,
      "learning_rate": 1.1001385478692783e-08,
      "loss": 0.0036,
      "num_input_tokens_seen": 131153872,
      "step": 60780
    },
    {
      "epoch": 9.915986949429037,
      "grad_norm": 0.056486278772354126,
      "learning_rate": 1.0791254807063223e-08,
      "loss": 0.0977,
      "num_input_tokens_seen": 131164848,
      "step": 60785
    },
    {
      "epoch": 9.916802610114193,
      "grad_norm": 0.14884202182292938,
      "learning_rate": 1.0583149887488653e-08,
      "loss": 0.0057,
      "num_input_tokens_seen": 131175280,
      "step": 60790
    },
    {
      "epoch": 9.917618270799348,
      "grad_norm": 0.3710552752017975,
      "learning_rate": 1.0377070736838912e-08,
      "loss": 0.0042,
      "num_input_tokens_seen": 131185776,
      "step": 60795
    },
    {
      "epoch": 9.918433931484502,
      "grad_norm": 12.366984367370605,
      "learning_rate": 1.017301737182008e-08,
      "loss": 0.0479,
      "num_input_tokens_seen": 131196944,
      "step": 60800
    },
    {
      "epoch": 9.919249592169658,
      "grad_norm": 0.05777287483215332,
      "learning_rate": 9.970989808974485e-09,
      "loss": 0.0048,
      "num_input_tokens_seen": 131208752,
      "step": 60805
    },
    {
      "epoch": 9.920065252854812,
      "grad_norm": 0.15197744965553284,
      "learning_rate": 9.77098806467791e-09,
      "loss": 0.007,
      "num_input_tokens_seen": 131217360,
      "step": 60810
    },
    {
      "epoch": 9.920880913539968,
      "grad_norm": 0.06549471616744995,
      "learning_rate": 9.573012155145166e-09,
      "loss": 0.003,
      "num_input_tokens_seen": 131228304,
      "step": 60815
    },
    {
      "epoch": 9.921696574225122,
      "grad_norm": 0.08914361149072647,
      "learning_rate": 9.37706209642175e-09,
      "loss": 0.0324,
      "num_input_tokens_seen": 131239568,
      "step": 60820
    },
    {
      "epoch": 9.922512234910277,
      "grad_norm": 3.147571563720703,
      "learning_rate": 9.183137904397732e-09,
      "loss": 0.1684,
      "num_input_tokens_seen": 131250576,
      "step": 60825
    },
    {
      "epoch": 9.923327895595433,
      "grad_norm": 0.07605230063199997,
      "learning_rate": 8.991239594788315e-09,
      "loss": 0.0946,
      "num_input_tokens_seen": 131261616,
      "step": 60830
    },
    {
      "epoch": 9.924143556280587,
      "grad_norm": 0.07768086344003677,
      "learning_rate": 8.801367183153276e-09,
      "loss": 0.1195,
      "num_input_tokens_seen": 131272400,
      "step": 60835
    },
    {
      "epoch": 9.924959216965743,
      "grad_norm": 0.1955137997865677,
      "learning_rate": 8.61352068488308e-09,
      "loss": 0.1882,
      "num_input_tokens_seen": 131283280,
      "step": 60840
    },
    {
      "epoch": 9.925774877650896,
      "grad_norm": 0.11185144633054733,
      "learning_rate": 8.427700115207216e-09,
      "loss": 0.005,
      "num_input_tokens_seen": 131293808,
      "step": 60845
    },
    {
      "epoch": 9.926590538336052,
      "grad_norm": 0.0732840970158577,
      "learning_rate": 8.243905489185855e-09,
      "loss": 0.0054,
      "num_input_tokens_seen": 131304016,
      "step": 60850
    },
    {
      "epoch": 9.927406199021208,
      "grad_norm": 0.13591451942920685,
      "learning_rate": 8.062136821723742e-09,
      "loss": 0.1467,
      "num_input_tokens_seen": 131314032,
      "step": 60855
    },
    {
      "epoch": 9.928221859706362,
      "grad_norm": 0.18943621218204498,
      "learning_rate": 7.882394127550763e-09,
      "loss": 0.1236,
      "num_input_tokens_seen": 131324112,
      "step": 60860
    },
    {
      "epoch": 9.929037520391518,
      "grad_norm": 0.09203652292490005,
      "learning_rate": 7.704677421238593e-09,
      "loss": 0.1006,
      "num_input_tokens_seen": 131334352,
      "step": 60865
    },
    {
      "epoch": 9.929853181076671,
      "grad_norm": 0.5394014120101929,
      "learning_rate": 7.528986717195152e-09,
      "loss": 0.2849,
      "num_input_tokens_seen": 131344656,
      "step": 60870
    },
    {
      "epoch": 9.930668841761827,
      "grad_norm": 0.402437686920166,
      "learning_rate": 7.355322029661826e-09,
      "loss": 0.0323,
      "num_input_tokens_seen": 131355984,
      "step": 60875
    },
    {
      "epoch": 9.931484502446983,
      "grad_norm": 0.08685159683227539,
      "learning_rate": 7.183683372719019e-09,
      "loss": 0.0141,
      "num_input_tokens_seen": 131367056,
      "step": 60880
    },
    {
      "epoch": 9.932300163132137,
      "grad_norm": 2.5764238834381104,
      "learning_rate": 7.0140707602805995e-09,
      "loss": 0.0777,
      "num_input_tokens_seen": 131378192,
      "step": 60885
    },
    {
      "epoch": 9.933115823817293,
      "grad_norm": 0.03751058876514435,
      "learning_rate": 6.846484206091131e-09,
      "loss": 0.2534,
      "num_input_tokens_seen": 131389040,
      "step": 60890
    },
    {
      "epoch": 9.933931484502446,
      "grad_norm": 0.20254509150981903,
      "learning_rate": 6.6809237237425156e-09,
      "loss": 0.0498,
      "num_input_tokens_seen": 131399664,
      "step": 60895
    },
    {
      "epoch": 9.934747145187602,
      "grad_norm": 0.194112628698349,
      "learning_rate": 6.517389326651801e-09,
      "loss": 0.0091,
      "num_input_tokens_seen": 131409680,
      "step": 60900
    },
    {
      "epoch": 9.935562805872756,
      "grad_norm": 0.3252966105937958,
      "learning_rate": 6.3558810280778254e-09,
      "loss": 0.006,
      "num_input_tokens_seen": 131420240,
      "step": 60905
    },
    {
      "epoch": 9.936378466557912,
      "grad_norm": 1.819018840789795,
      "learning_rate": 6.196398841112893e-09,
      "loss": 0.0073,
      "num_input_tokens_seen": 131429488,
      "step": 60910
    },
    {
      "epoch": 9.937194127243067,
      "grad_norm": 0.1418514996767044,
      "learning_rate": 6.038942778685553e-09,
      "loss": 0.1406,
      "num_input_tokens_seen": 131439792,
      "step": 60915
    },
    {
      "epoch": 9.938009787928221,
      "grad_norm": 0.05585205927491188,
      "learning_rate": 5.883512853557816e-09,
      "loss": 0.0762,
      "num_input_tokens_seen": 131448976,
      "step": 60920
    },
    {
      "epoch": 9.938825448613377,
      "grad_norm": 0.11939923465251923,
      "learning_rate": 5.730109078330714e-09,
      "loss": 0.0109,
      "num_input_tokens_seen": 131459088,
      "step": 60925
    },
    {
      "epoch": 9.939641109298531,
      "grad_norm": 0.08607755601406097,
      "learning_rate": 5.578731465444298e-09,
      "loss": 0.0796,
      "num_input_tokens_seen": 131470096,
      "step": 60930
    },
    {
      "epoch": 9.940456769983687,
      "grad_norm": 3.4439589977264404,
      "learning_rate": 5.429380027163755e-09,
      "loss": 0.1152,
      "num_input_tokens_seen": 131479376,
      "step": 60935
    },
    {
      "epoch": 9.941272430668842,
      "grad_norm": 0.13130417466163635,
      "learning_rate": 5.2820547755988434e-09,
      "loss": 0.0122,
      "num_input_tokens_seen": 131491792,
      "step": 60940
    },
    {
      "epoch": 9.942088091353996,
      "grad_norm": 0.037019938230514526,
      "learning_rate": 5.1367557226927875e-09,
      "loss": 0.0021,
      "num_input_tokens_seen": 131502512,
      "step": 60945
    },
    {
      "epoch": 9.942903752039152,
      "grad_norm": 3.488393783569336,
      "learning_rate": 4.9934828802250535e-09,
      "loss": 0.2115,
      "num_input_tokens_seen": 131514224,
      "step": 60950
    },
    {
      "epoch": 9.943719412724306,
      "grad_norm": 0.09522642195224762,
      "learning_rate": 4.852236259805798e-09,
      "loss": 0.096,
      "num_input_tokens_seen": 131525840,
      "step": 60955
    },
    {
      "epoch": 9.944535073409462,
      "grad_norm": 0.09075871109962463,
      "learning_rate": 4.7130158728925236e-09,
      "loss": 0.0079,
      "num_input_tokens_seen": 131536208,
      "step": 60960
    },
    {
      "epoch": 9.945350734094617,
      "grad_norm": 3.328155994415283,
      "learning_rate": 4.575821730765095e-09,
      "loss": 0.1114,
      "num_input_tokens_seen": 131547344,
      "step": 60965
    },
    {
      "epoch": 9.946166394779771,
      "grad_norm": 3.054372549057007,
      "learning_rate": 4.440653844545173e-09,
      "loss": 0.0839,
      "num_input_tokens_seen": 131558544,
      "step": 60970
    },
    {
      "epoch": 9.946982055464927,
      "grad_norm": 0.0750637948513031,
      "learning_rate": 4.307512225196209e-09,
      "loss": 0.1256,
      "num_input_tokens_seen": 131569136,
      "step": 60975
    },
    {
      "epoch": 9.947797716150081,
      "grad_norm": 0.23475918173789978,
      "learning_rate": 4.176396883504019e-09,
      "loss": 0.1248,
      "num_input_tokens_seen": 131580848,
      "step": 60980
    },
    {
      "epoch": 9.948613376835237,
      "grad_norm": 0.05277956649661064,
      "learning_rate": 4.047307830101765e-09,
      "loss": 0.0022,
      "num_input_tokens_seen": 131591664,
      "step": 60985
    },
    {
      "epoch": 9.949429037520392,
      "grad_norm": 0.23914973437786102,
      "learning_rate": 3.9202450754533e-09,
      "loss": 0.1028,
      "num_input_tokens_seen": 131602032,
      "step": 60990
    },
    {
      "epoch": 9.950244698205546,
      "grad_norm": 3.7414486408233643,
      "learning_rate": 3.79520862985594e-09,
      "loss": 0.1077,
      "num_input_tokens_seen": 131613520,
      "step": 60995
    },
    {
      "epoch": 9.951060358890702,
      "grad_norm": 0.11124800145626068,
      "learning_rate": 3.6721985034515738e-09,
      "loss": 0.0221,
      "num_input_tokens_seen": 131624624,
      "step": 61000
    },
    {
      "epoch": 9.951876019575856,
      "grad_norm": 0.16320769488811493,
      "learning_rate": 3.5512147062072287e-09,
      "loss": 0.0057,
      "num_input_tokens_seen": 131634416,
      "step": 61005
    },
    {
      "epoch": 9.952691680261012,
      "grad_norm": 0.25731217861175537,
      "learning_rate": 3.4322572479345005e-09,
      "loss": 0.0025,
      "num_input_tokens_seen": 131645232,
      "step": 61010
    },
    {
      "epoch": 9.953507340946166,
      "grad_norm": 0.12489716708660126,
      "learning_rate": 3.3153261382729008e-09,
      "loss": 0.1842,
      "num_input_tokens_seen": 131655824,
      "step": 61015
    },
    {
      "epoch": 9.954323001631321,
      "grad_norm": 0.24719005823135376,
      "learning_rate": 3.2004213867009582e-09,
      "loss": 0.0044,
      "num_input_tokens_seen": 131666640,
      "step": 61020
    },
    {
      "epoch": 9.955138662316477,
      "grad_norm": 0.08298831433057785,
      "learning_rate": 3.0875430025362197e-09,
      "loss": 0.0048,
      "num_input_tokens_seen": 131675120,
      "step": 61025
    },
    {
      "epoch": 9.955954323001631,
      "grad_norm": 0.08427087217569351,
      "learning_rate": 2.9766909949296983e-09,
      "loss": 0.08,
      "num_input_tokens_seen": 131686096,
      "step": 61030
    },
    {
      "epoch": 9.956769983686787,
      "grad_norm": 0.05212075263261795,
      "learning_rate": 2.8678653728658746e-09,
      "loss": 0.0041,
      "num_input_tokens_seen": 131697968,
      "step": 61035
    },
    {
      "epoch": 9.95758564437194,
      "grad_norm": 0.06858661770820618,
      "learning_rate": 2.761066145168245e-09,
      "loss": 0.1066,
      "num_input_tokens_seen": 131709168,
      "step": 61040
    },
    {
      "epoch": 9.958401305057096,
      "grad_norm": 0.09093804657459259,
      "learning_rate": 2.656293320490999e-09,
      "loss": 0.2007,
      "num_input_tokens_seen": 131720656,
      "step": 61045
    },
    {
      "epoch": 9.959216965742252,
      "grad_norm": 0.3559302091598511,
      "learning_rate": 2.5535469073301176e-09,
      "loss": 0.0086,
      "num_input_tokens_seen": 131731536,
      "step": 61050
    },
    {
      "epoch": 9.960032626427406,
      "grad_norm": 0.1537299007177353,
      "learning_rate": 2.4528269140150497e-09,
      "loss": 0.0055,
      "num_input_tokens_seen": 131742928,
      "step": 61055
    },
    {
      "epoch": 9.960848287112562,
      "grad_norm": 0.0762186124920845,
      "learning_rate": 2.354133348711485e-09,
      "loss": 0.0622,
      "num_input_tokens_seen": 131752496,
      "step": 61060
    },
    {
      "epoch": 9.961663947797716,
      "grad_norm": 0.09247704595327377,
      "learning_rate": 2.2574662194158047e-09,
      "loss": 0.0055,
      "num_input_tokens_seen": 131764048,
      "step": 61065
    },
    {
      "epoch": 9.962479608482871,
      "grad_norm": 0.3819979727268219,
      "learning_rate": 2.1628255339689596e-09,
      "loss": 0.0688,
      "num_input_tokens_seen": 131774352,
      "step": 61070
    },
    {
      "epoch": 9.963295269168025,
      "grad_norm": 0.5185974836349487,
      "learning_rate": 2.0702113000425903e-09,
      "loss": 0.0081,
      "num_input_tokens_seen": 131783696,
      "step": 61075
    },
    {
      "epoch": 9.964110929853181,
      "grad_norm": 0.06221344321966171,
      "learning_rate": 1.979623525141805e-09,
      "loss": 0.0128,
      "num_input_tokens_seen": 131795696,
      "step": 61080
    },
    {
      "epoch": 9.964926590538337,
      "grad_norm": 0.1394263356924057,
      "learning_rate": 1.891062216610728e-09,
      "loss": 0.0053,
      "num_input_tokens_seen": 131805616,
      "step": 61085
    },
    {
      "epoch": 9.96574225122349,
      "grad_norm": 5.402780055999756,
      "learning_rate": 1.804527381629728e-09,
      "loss": 0.1396,
      "num_input_tokens_seen": 131815056,
      "step": 61090
    },
    {
      "epoch": 9.966557911908646,
      "grad_norm": 0.06656648963689804,
      "learning_rate": 1.7200190272126382e-09,
      "loss": 0.0044,
      "num_input_tokens_seen": 131825392,
      "step": 61095
    },
    {
      "epoch": 9.9673735725938,
      "grad_norm": 5.79808235168457,
      "learning_rate": 1.6375371602123103e-09,
      "loss": 0.0852,
      "num_input_tokens_seen": 131836144,
      "step": 61100
    },
    {
      "epoch": 9.968189233278956,
      "grad_norm": 0.22231650352478027,
      "learning_rate": 1.5570817873122868e-09,
      "loss": 0.0615,
      "num_input_tokens_seen": 131846672,
      "step": 61105
    },
    {
      "epoch": 9.969004893964112,
      "grad_norm": 3.6045851707458496,
      "learning_rate": 1.4786529150379036e-09,
      "loss": 0.3261,
      "num_input_tokens_seen": 131858544,
      "step": 61110
    },
    {
      "epoch": 9.969820554649266,
      "grad_norm": 0.045092545449733734,
      "learning_rate": 1.4022505497424122e-09,
      "loss": 0.097,
      "num_input_tokens_seen": 131869296,
      "step": 61115
    },
    {
      "epoch": 9.970636215334421,
      "grad_norm": 0.08182383328676224,
      "learning_rate": 1.3278746976236322e-09,
      "loss": 0.0165,
      "num_input_tokens_seen": 131879376,
      "step": 61120
    },
    {
      "epoch": 9.971451876019575,
      "grad_norm": 0.1155458465218544,
      "learning_rate": 1.255525364710075e-09,
      "loss": 0.0055,
      "num_input_tokens_seen": 131890768,
      "step": 61125
    },
    {
      "epoch": 9.97226753670473,
      "grad_norm": 0.12539751827716827,
      "learning_rate": 1.1852025568637183e-09,
      "loss": 0.0039,
      "num_input_tokens_seen": 131901104,
      "step": 61130
    },
    {
      "epoch": 9.973083197389887,
      "grad_norm": 0.03399152681231499,
      "learning_rate": 1.116906279791108e-09,
      "loss": 0.2061,
      "num_input_tokens_seen": 131912208,
      "step": 61135
    },
    {
      "epoch": 9.97389885807504,
      "grad_norm": 5.785745620727539,
      "learning_rate": 1.0506365390211547e-09,
      "loss": 0.1843,
      "num_input_tokens_seen": 131923472,
      "step": 61140
    },
    {
      "epoch": 9.974714518760196,
      "grad_norm": 4.4891533851623535,
      "learning_rate": 9.863933399328895e-10,
      "loss": 0.3352,
      "num_input_tokens_seen": 131934384,
      "step": 61145
    },
    {
      "epoch": 9.97553017944535,
      "grad_norm": 0.08644963800907135,
      "learning_rate": 9.241766877304825e-10,
      "loss": 0.1085,
      "num_input_tokens_seen": 131944784,
      "step": 61150
    },
    {
      "epoch": 9.976345840130506,
      "grad_norm": 3.782694101333618,
      "learning_rate": 8.639865874571218e-10,
      "loss": 0.0992,
      "num_input_tokens_seen": 131955088,
      "step": 61155
    },
    {
      "epoch": 9.977161500815662,
      "grad_norm": 0.06311924755573273,
      "learning_rate": 8.058230439950132e-10,
      "loss": 0.1173,
      "num_input_tokens_seen": 131966384,
      "step": 61160
    },
    {
      "epoch": 9.977977161500815,
      "grad_norm": 0.0469924658536911,
      "learning_rate": 7.496860620570534e-10,
      "loss": 0.0584,
      "num_input_tokens_seen": 131977200,
      "step": 61165
    },
    {
      "epoch": 9.978792822185971,
      "grad_norm": 0.06673408299684525,
      "learning_rate": 6.955756461951568e-10,
      "loss": 0.0748,
      "num_input_tokens_seen": 131988848,
      "step": 61170
    },
    {
      "epoch": 9.979608482871125,
      "grad_norm": 8.973872184753418,
      "learning_rate": 6.434918007947044e-10,
      "loss": 0.066,
      "num_input_tokens_seen": 132000752,
      "step": 61175
    },
    {
      "epoch": 9.98042414355628,
      "grad_norm": 0.1769551783800125,
      "learning_rate": 5.934345300773192e-10,
      "loss": 0.0055,
      "num_input_tokens_seen": 132010544,
      "step": 61180
    },
    {
      "epoch": 9.981239804241435,
      "grad_norm": 0.06666838377714157,
      "learning_rate": 5.454038381008664e-10,
      "loss": 0.0827,
      "num_input_tokens_seen": 132021552,
      "step": 61185
    },
    {
      "epoch": 9.98205546492659,
      "grad_norm": 0.1354232281446457,
      "learning_rate": 4.993997287622287e-10,
      "loss": 0.2025,
      "num_input_tokens_seen": 132032752,
      "step": 61190
    },
    {
      "epoch": 9.982871125611746,
      "grad_norm": 0.06022237241268158,
      "learning_rate": 4.554222057889801e-10,
      "loss": 0.0046,
      "num_input_tokens_seen": 132043536,
      "step": 61195
    },
    {
      "epoch": 9.9836867862969,
      "grad_norm": 0.4063657522201538,
      "learning_rate": 4.1347127274493635e-10,
      "loss": 0.0163,
      "num_input_tokens_seen": 132053840,
      "step": 61200
    },
    {
      "epoch": 9.984502446982056,
      "grad_norm": 0.08528797328472137,
      "learning_rate": 3.735469330301555e-10,
      "loss": 0.0111,
      "num_input_tokens_seen": 132064400,
      "step": 61205
    },
    {
      "epoch": 9.98531810766721,
      "grad_norm": 0.09119515120983124,
      "learning_rate": 3.356491898837133e-10,
      "loss": 0.1199,
      "num_input_tokens_seen": 132073552,
      "step": 61210
    },
    {
      "epoch": 9.986133768352365,
      "grad_norm": 0.06901831179857254,
      "learning_rate": 2.997780463753763e-10,
      "loss": 0.0062,
      "num_input_tokens_seen": 132084400,
      "step": 61215
    },
    {
      "epoch": 9.986949429037521,
      "grad_norm": 0.12367760390043259,
      "learning_rate": 2.659335054139289e-10,
      "loss": 0.0056,
      "num_input_tokens_seen": 132095856,
      "step": 61220
    },
    {
      "epoch": 9.987765089722675,
      "grad_norm": 0.05554777756333351,
      "learning_rate": 2.341155697471731e-10,
      "loss": 0.0772,
      "num_input_tokens_seen": 132108368,
      "step": 61225
    },
    {
      "epoch": 9.98858075040783,
      "grad_norm": 0.19484113156795502,
      "learning_rate": 2.043242419452751e-10,
      "loss": 0.1241,
      "num_input_tokens_seen": 132120432,
      "step": 61230
    },
    {
      "epoch": 9.989396411092985,
      "grad_norm": 0.039668235927820206,
      "learning_rate": 1.7655952443129675e-10,
      "loss": 0.024,
      "num_input_tokens_seen": 132131568,
      "step": 61235
    },
    {
      "epoch": 9.99021207177814,
      "grad_norm": 0.15951573848724365,
      "learning_rate": 1.5082141945343963e-10,
      "loss": 0.083,
      "num_input_tokens_seen": 132142992,
      "step": 61240
    },
    {
      "epoch": 9.991027732463294,
      "grad_norm": 0.0305704977363348,
      "learning_rate": 1.2710992909892306e-10,
      "loss": 0.0029,
      "num_input_tokens_seen": 132153520,
      "step": 61245
    },
    {
      "epoch": 9.99184339314845,
      "grad_norm": 0.09939110279083252,
      "learning_rate": 1.0542505528565727e-10,
      "loss": 0.0069,
      "num_input_tokens_seen": 132164368,
      "step": 61250
    },
    {
      "epoch": 9.992659053833606,
      "grad_norm": 0.0912177562713623,
      "learning_rate": 8.57667997788969e-11,
      "loss": 0.0032,
      "num_input_tokens_seen": 132174992,
      "step": 61255
    },
    {
      "epoch": 9.99347471451876,
      "grad_norm": 0.19032734632492065,
      "learning_rate": 6.813516416626087e-11,
      "loss": 0.0041,
      "num_input_tokens_seen": 132186320,
      "step": 61260
    },
    {
      "epoch": 9.994290375203915,
      "grad_norm": 0.11631111800670624,
      "learning_rate": 5.2530149877161315e-11,
      "loss": 0.0359,
      "num_input_tokens_seen": 132198000,
      "step": 61265
    },
    {
      "epoch": 9.99510603588907,
      "grad_norm": 0.09231505542993546,
      "learning_rate": 3.895175818002805e-11,
      "loss": 0.0831,
      "num_input_tokens_seen": 132209840,
      "step": 61270
    },
    {
      "epoch": 9.995921696574225,
      "grad_norm": 0.0757303237915039,
      "learning_rate": 2.7399990173981872e-11,
      "loss": 0.0034,
      "num_input_tokens_seen": 132220848,
      "step": 61275
    },
    {
      "epoch": 9.99673735725938,
      "grad_norm": 3.7855224609375,
      "learning_rate": 1.7874846797161228e-11,
      "loss": 0.2133,
      "num_input_tokens_seen": 132230384,
      "step": 61280
    },
    {
      "epoch": 9.997553017944535,
      "grad_norm": 0.0497620664536953,
      "learning_rate": 1.0376328818395564e-11,
      "loss": 0.1674,
      "num_input_tokens_seen": 132241200,
      "step": 61285
    },
    {
      "epoch": 9.99836867862969,
      "grad_norm": 4.879983901977539,
      "learning_rate": 4.90443684553199e-12,
      "loss": 0.2153,
      "num_input_tokens_seen": 132251664,
      "step": 61290
    },
    {
      "epoch": 9.999184339314844,
      "grad_norm": 0.039522819221019745,
      "learning_rate": 1.4591713254352712e-12,
      "loss": 0.0041,
      "num_input_tokens_seen": 132262576,
      "step": 61295
    },
    {
      "epoch": 10.0,
      "grad_norm": 0.16737177968025208,
      "learning_rate": 4.053253843672167e-14,
      "loss": 0.0941,
      "num_input_tokens_seen": 132272272,
      "step": 61300
    },
    {
      "epoch": 10.0,
      "eval_loss": 0.1923980414867401,
      "eval_runtime": 568.4728,
      "eval_samples_per_second": 4.794,
      "eval_steps_per_second": 1.2,
      "num_input_tokens_seen": 132272272,
      "step": 61300
    },
    {
      "epoch": 10.0,
      "num_input_tokens_seen": 132272272,
      "step": 61300,
      "total_flos": 5.956371922343952e+18,
      "train_loss": 0.11942618750450527,
      "train_runtime": 66936.7162,
      "train_samples_per_second": 3.663,
      "train_steps_per_second": 0.916
    }
  ],
  "logging_steps": 5,
  "max_steps": 61300,
  "num_input_tokens_seen": 132272272,
  "num_train_epochs": 10,
  "save_steps": 3065,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.956371922343952e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}